American Journal of Infection Control

Volume 35, Issue 4, Pages 207-288 (May 2007)

1. Table of Contents Pages A2, A4

2. Editorial Board Page A6

3. Information for Readers Page A9

Major Articles

4. Survey of isolation practices at a tertiary care pediatric hospital Pages 207-211 Joseph V. Vayalumkal, Laurie Streitenberger, Rick Wray, Carol Goldman, Renee Freeman, Steven Drews and Anne Matlow

5. Emergence of resistant Acinetobacter baumannii in critically ill patients within an acute care teaching hospital and a long-term acute care hospital Pages 212-215 Claudester Stephens, Stephen J. Francis, Virginia Abell, Joseph R. DiPersio and Patricia Wells

6. Validation of surgical site infection surveillance in orthopedic procedures Pages 216-221 Kaisa Huotari, Niina Agthe and Outi Lyytikäinen

7. On the move? Page 221

8. Surveillance of multidrug-resistant gram-negative bacilli in a neonatal intensive care unit: prominent role of cross transmission Pages 222-230 Caterina Mammina, Paola Di Carlo, Domenico Cipolla, Mario Giuffrè, Alessandra Casuccio, Vincenzo Di Gaetano, Maria Rosa Anna Plano, Emma D'Angelo, Lucina Titone and Giovanni Corsello

9. Antimicriobial resistance patterns of colonizing flora on nurses' hands in the neonatal intensive care unit Pages 231-236 Heather A. Cook, Jeannie P. Cimiotti, Phyllis Della-Latta, Lisa Saiman and Elaine L. Larson

10. Implications of the changing face of Clostridium difficile disease for health care practitioners Pages 237-253 Lynne V. McFarland, Henry W. Beneda, Jill E. Clarridge and Gregory J. Raugi

11. Screening for pulmonary tuberculosis using chest radiography in new employees in an industrial park in Taiwan Pages 254-259 Shih-Bin Su, Chien-Fang Chiu, Cheng-Ta Chang, Kow-Tong Chen, Ching-Yih Lin and How-Ran Guo

12. Prescription of prophylactic antibiotics for neurosurgical procedures in teaching hospitals in Iran Pages 260-262 Mehrdad Askarian, Ali Reza Moravveji and Ojan Assadian

Brief Reports

13. Bacterial contamination of stethoscopes with antimicrobial diaphragm covers Pages 263-266 Mark W. Wood, Ryan C. Lund and Kurt B. Stevenson

14. Occupationally acquired infectious diseases among health care workers in Brazil: Use of Internet tools to improve management, prevention, and surveillance Pages 267-270 Cristiane Rapparini, Paulo Feijó Barroso, Valéria Saraceni, Alcyone Artioli Machado and Guilherme Côrtes Fernandes

Practice Forum

15. The infection control audit: The standardized audit as a tool for change Pages 271-283 Elizabeth Ann Bryce, Sydney Scharf, Moira Walker and Anne Walsh

Letters to the Editor

16. Of viruses, gloves, and crêpes Page 284 Arnaud Tarantola

17. In response to Silvestri Pages 284-285 Scott A. Flanders, Sanjay Saint and Harold R. Collard

18. Lack of occurrence of methicillin-resistant Staphylococcus aureus on municipal public telephones Pages 285-286 Anna O'Connor, Anne Loughrey, B. Cherie Millar, Colm J. Lowery, James S.G. Dooley, Colin E. Goldsmith, Paul J. Rooney and John E. Moore

19. Assessing vaccination rates at an emergency room Pages 286-287 V.M.M. Miranda and C.M. Nascimento-Carvalho

20. Phlebitis associated with peripheral intravenous catheters Page 287 Joan Webster and Sonya Osborne

21. In response to Webster and Osborne Pages 287-288 Ziona Jerassy, Bernard Rudensky, David Raveh and Amos M. Yinnon

May 2007

MAJOR ARTICLES

Survey of isolation practices at a tertiary care pediatric hospital 207JV Vayalumkal, L Streitenberger, R Wray, C Goldman, R Freeman, S Drews, and A Matlow

Emergence of resistant Acinetobacter baumannii in critically ill patients within an

acute care teaching hospital and a long-term acute care hospital 212C Stephens, SJ Francis, V Abell, JR DiPersio, and P Wells

Validation of surgical site infection surveillance in orthopedic procedures 216K Huotari, N Agthe, and O Lyytikainen

Surveillance of multidrug-resistant gram-negative bacilli in a neonatal intensive

care unit: prominent role of cross transmission 222C Mammina, P Di Carlo, D Cipolla, M Giuffre, A Casuccio, V Di Gaetano, MRA Plano,

E D’Angelo, L Titone, and G Corsello

Antimicriobial resistance patterns of colonizing flora on nurses’ hands in the

neonatal intensive care unit 231HA Cook, JP Cimiotti, P Della-Latta, L Saiman, and EL Larson

Implications of the changing face of Clostridium difficile disease for

health care practitioners 237LV McFarland, HW Beneda, JE Clarridge, and GJ Raugi

Screening for pulmonary tuberculosis using chest radiography in new employees

in an industrial park in Taiwan 254S-B Su, C-F Chiu, C-T Chang, K-T Chen, C-Y Lin, and H-R Guo

Prescription of prophylactic antibiotics for neurosurgical procedures in teaching

hospitals in Iran 260M Askarian, A Reza Moravveji, and O Assadian

BRIEF REPORTS

Bacterial contamination of stethoscopes with antimicrobial diaphragm covers 263MW Wood, RC Lund, and KB Stevenson

Occupationally acquired infectious diseases among health care workers in Brazil:

Use of Internet tools to improve management, prevention, and surveillance 267C Rapparini, P Feijo Barroso, V Saraceni, A Artioli Machado, and G Cortes Fernandes

AJIC: American Journal of Infection Control (ISSN: 0196-6553) is published 10 times per year by Elsevier Inc., 360 Park Avenue South, New

York, NY 10010-1710. Months of publication are: February, March, April, May, June, August, September, October, November, and

December. Business and Editorial Offices: 1600 John F. Kennedy Blvd., Suite 1800, Philadelphia, PA 19103-2899. Customer Service Office:

6277 Sea Harbor Drive, Orlando, FL 32887-4800. Periodicals postage paid at New York, NY and additional mailing offices.

POSTMASTER: Send address changes to AJIC: American Journal of Infection Control, Elsevier Periodicals Customer Service, 6277 Sea Harbor

Drive, Orlando, FL 32887-4800.

2A May 2007

PRACTICE FORUM

The infection control audit: The standardized audit as a tool for change 271EA Bryce, S Scharf, M Walker, and A Walsh

LETTERS TO THE EDITOR

Of viruses, gloves, and crepes 284A Tarantola

In response to Silvestri 284SA Flanders, S Saint, and HR Collard

Lack of occurrence of methicillin-resistant Staphylococcus aureus on municipal

public telephones 285A O’Connor, A Loughrey, BC Millar, CJ Lowery, JSG Dooley, CE Goldsmith,

PJ Rooney, and JE Moore

Assessing vaccination rates at an emergency room 286VMM Miranda and CM Nascimento-Carvalho

Phlebitis associated with peripheral intravenous catheters 287J Webster and S Osborne

In response to Webster and Osborne 287Z Jerassy, B Rudensky, D Raveh, and AM Yinnon

READER SERVICES

Information for authors www.ajicjournal.org and December 2006, pages 9A-11AInformation for readers 9AChange of address 221

Full-text online access to the American Journal of Infection Control isnow available for all print subscribers. See page 211 for details.

American Journal of Infection Control (AJIC) is the official scientific publicationof the Association for Professionals in Infection Control and Epidemiology, Inc(APIC), a multidisciplinary international organization of health care profes-sionals. The mission of AJIC is to improve health care by reducing risks ofinfection and related adverse outcomes by critical review, selection, anddissemination of new and relevant information in the fields of infectionprevention and control and health care epidemiology in all health care settingsand the community.

4A May 2007

EditorElaine L. Larson, RN, PhD, FAAN, CIC New York, NY

Associate EditorsZiad Memish, Riyadh, Saudi ArabiaRussell N. Olmsted, MPH, CIC Ann Arbor, MichBarbara M. Soule, RN, MPH, CIC Olympia, Wash

SECTION EDITORS

GLOBAL PERSPECTIVES

Barbara M. Soule, RN, MPH, CIC Olympia, WashZiad Memish, MD Riyadh, Saudi Arabia

NEW AND NOTEWORTHY

Russell N. Olmsted, MPH, CIC Ann Arbor, Mich

EDITORIAL BOARD 2007

ASSOCIATION FOR PROFESSIONALS IN INFECTION CONTROL AND EPIDEMIOLOGY, INC

BOARD OF DIRECTORS 2007

Denise Murphy, RN, BSN, MPH, CIC PRESIDENT

Janet E. Frain, RN, CIC, CPHQ, CPHRM PRESIDENT-ELECT

James Marx, RN, MS, CIC TREASURER

Patti Grant, RN, BSN, MS, CIC SECRETARY

Kathleen Arias, MS, MT, SM, CIC IMMEDIATE PAST PRESIDENT

EXECUTIVE DIRECTOR Kathy L. Warye1275 K St NW, Suite 1000Washington, DC 20005-4006202-454-2601; 202-789-1899 (fax); kwarye@apic.org; http://www.apic.org/

MANAGING EDITOR

Manuel Cortazal, BA New York, NY

CURRENT METHODOLOGICAL CONCEPTS

Sanjay Saint, MD, MPHTeresa Horan, MPH, CIC

Allison E. Aiello, PhD Ann Arbor, Mich

Antoine Andremont, MD Paris, France

Judene Bartley, MS, MPH, CIC Beverly Hills, Mich

Michael T. Brady, MD Columbus, Ohio

Jeannie Cimiotti, DNSc, RN, CRNP Philadelphia, Pa

Sean Clarke, RN, PhD, FAAN Philadelphia, Pa

Paula Denlick, MPH, SM(ASCP), CIC Santee, Calif

Charles E. Edmiston, Jr, PhD, CIC Milwaukee, Wis

Judith English, RN, MSN, CIC Bethesda, Md

Joan Faoagali, MD Brisbane, Australia

Janet Franck, RN, MBA, CIC St. Louis, Mo

Candace Friedman, MT(ASCP), MPH, CIC

Ann Arbor, Mich

Barbara Goldrick, RN, MPH, PhD Chatham, Mass

Dinah Gould, Mphil, PhD, International Liaison

London, UK

Patti S. Grant, RN, MS, CIC Dallas, Tex

Christopher Griffith, PhD Cardiff, Wales

Joan Hebden, RN, MS, CIC Baltimore, Md

Teresa C. Horan, MPH, CIC Atlanta, Ga

Denise Korniewicz, DNSc, RN, FAAN

Coral Gables, Fla

Joan Kramer, PhD, RN, BC, CIC Baltimore, Md

James Lee, MD, PhD Saint Paul, Minn

Gavin Leslie, RN, PhD Perth, Australia

Robert Longfield, MD, FACP San Antonio, Tex

Frank Lowy, MD New York, NY

Farrin A. Manian, MD, MPH, FIDSA St. Louis, Mo

Michael Massanari, MD Detroit, Mich

Gerardo Maupome, CD, MSc, DDPH, RCS(E),

PhD Portland, Ore

Jennie Mayfield, MPH, CIC St. Louis, Mo

Mary-Louise McLaws, PhD Sydney, Australia

Geeta Mehta, MD, MAMS, FIAMS New Delhi, India

Hanako Misao, MS, RN San Francisco, Calif

John Molinari, PhD Detroit, Mich

Cathryn Murphy, MPH, PhD Sydney, Australia

Folasade T. Ogunsola, PhD Lagos, Nigeria

Daryl S. Paulson, MBA, PhD Bozeman, Mont

Teresa Pi-Sunyer, RN, MHM Barcelona, Spain

Didier Pittet, MD Geneva, Switzerland

Gregg Pullen, MT(ASCP)M, RM(NRM)

Madera, Calif

Larry Reimer, MD Salt Lake City, Utah

Vıctor D. Rosenthal, MD, CIC, MSc Buenos Aires,

Argentina

Henning Ruden, MD Berlin, Germany

Sanjay Saint, MD, MPH Ann Arbor, Mich

Steven J. Schweon, RN, MPH, CIC Bethlehem, Pa

Keith St. John, MS, MT(ASCP), CIC Philadelphia, Pa

Syed Sattar, PhD Ottawa, Canada

Fran Slater-Feltovich, RN, MBA, CIC, CPHQ

Houston, Tex

Richard A. Stein, MD, PhD New York, NY

Larry J. Strausbaugh, MD Portland, Ore

Thelma D. van der Merwe, RN, PhD South Africa

Gertie van Knippenberg-Gordebeke, Editor,

International Federation of Infection Control Venlo,

The Netherlands

Marilyn Wagener, MPH Pittsburgh, Pa

Pair-Dong Wang, MD Taipei, Taiwan

David Woodard, MSc Los Angeles, Calif

Cosmina Zeana, MD New York, NY

Mary L. Andrus, RN, BA, CICLillian Burns, MT, MPH, CICLinda J. Burton, RN, BSN, CICRuth Carrico, PhD, RN, CIC

Michelle Farber, RN, CICGail E. Harris, RN, MS, MA, CIC, CPHRMCathryn Murphy, RN, MPH, PhD, CICShannon Oriola, RN, CIC, COHN

Patricia Rosenbaum, RN, CICJoseph K. Whiting, BA, MBA, FACHE

6A May 2007

May 2007

CUSTOMER SERVICE (orders, claims, online, change of address)

Elsevier Periodicals Customer Service, 6277 Sea Harbor Drive, Orlando,FL 32887-4800. Tel: (800) 654-2452 (US and Canada); (407) 345-4000(outside US and Canada). Fax: (800) 225-6030 (US and Canada); (407)363-9661 (outside US and Canada). E-mail: elspcs@elsevier.com

Address changes must be submitted four weeks in advance.

YEARLY SUBSCRIPTION RATES

United States and possessions: Individual $133.00; Institution $415.00.All other countries (prices include airspeed delivery): Individual$161.00; Institution $447.00. Single Issues $45.00. Current prices arein effect for back volumes and back issues.

Further information on this journal is available from the Publisher orfrom this journal’s Web site (http://www.ajicjournal.org). Informationon other Elsevier products is available through Elsevier’s Web site(http://www.elsevier.com).

Advertising information. Advertising orders and inquiries can besent to: USA, Canada, and South America, Jeffrey S. Berman, BeMedia Partners, LLC, 415 King Road, Doylestown, PA 18901-5538;phone 215-249-3060; fax 215-249-4741; e-mail jberman@bemediapartners.com. Classified advertising orders and inquiries can be sentto Jim Maloney; phone 215-239-3130; e-mail j.maloney@elsevier.com. Europe and the rest of the world, Julie Toop; phone +44(0)1865 843016; fax +44(0) 1865 843976; e-mail media@elsevier.com.

AUTHOR INQUIRIES

For inquiries relating to the submission of articles (including electronicsubmission where available), please visit www.elsevier.com/authors. Thissite also provides the facility to track accepted articles and set up e-mailalerts to inform you of when an article’s status has changed, as well asdetailed artwork guidelines, copyright information, frequently askedquestions, and more. Please see Information for Authors for individualjournal requirements. Contact details for questions arising afteracceptance of an article, especially those relating to proofs, are providedafter registration of an article for publication.

�‘ The paper used in this publication meets the requirements ofANSI/NISO Z39.48-1992 (Permanence of Paper).

Reprints. For queries about author reprints, e-mail authorsupport@elsevier.com. To order 100 or more reprints for educational, commer-cial, or promotional use, contact the Commercial Reprints Department,Elsevier Inc., 360 Park Avenue South, New York, NY 10010-1710. Fax212-462-1935; e-mail reprints@elsevier.com. Reprints of single articlesavailable online may be obtained by purchasing Pay-Per-View accessfor $10 per article on the journal Web site, www.ajicjournal.org.

� 2007 by the Association for Professionals in Infection Controland Epidemiology, Inc. All rights reserved. This journal and theindividual contributions contained in it are protected under copyrightby the Association for Professionals in Infection Control andEpidemiology, Inc, and the following terms and conditions apply totheir use:

Photocopying. Single photocopies of single articles may be made forpersonal use as allowed by national copyright laws. Permission of thePublisher and payment of a fee is required for all other photocopying,including multiple or systematic copying, copying for advertising or

May 2007

promotional purposes, resale, and all forms of documentdelivery. Special rates are available for educational institutionsthat wish to make photocopies for nonprofit educationalclassroom use.

Permissions may be sought directly from Elsevier’s RightsDepartment in Philadelphia, PA, USA: phone 215-239-3804, fax215-239-3805, e-mail healthpermissions@elsevier.com. Re-quests may also be completed online via the Elsevier homepage(http://www.elsevier.com/locate/permissions).

In the USA, users may clear permissions and make paymentsthrough the Copyright Clearance Center, Inc., 222 RosewoodDrive, Danvers, MA 01923, USA; phone: (978) 750-8400, fax:(978) 750-4744, and in the UK through the Copyright LicensingAgency Rapid Clearance Service (CLARCS), 90 Tottenham CourtRoad, London W1P 0LP, UK; phone: (+44) 20 7631 5555; fax:(+44) 20 7631 5500. Other countries may have a localreprographic rights agency for payments.

Derivative Works. Subscribers may reproduce tables of contentsor prepare lists of articles including abstracts for internalcirculation within their institutions. Permission of the Publisheris required for resale or distribution outside the institution.

Permission of the Publisher is required for all other derivativeworks, including compilations and translations.

Electronic Storage or Usage. Permission of the Publisher isrequired to store or use electronically any material contained inthis journal, including any article or part of an article.

Except as outlined above, no part of this publication may bereproduced, stored in a retrieval system or transmitted in anyform or by any means, electronic, mechanical, photocopying,recording or otherwise, without prior written permission of thePublisher.

Address permissions requests to: Elsevier Rights Department, atthe fax and e-mail addresses noted above.

Notice. No responsibility is assumed by the Publisher or theAssociation for Professionals in Infection Control and Epide-miology, Inc for any injury and/or damage to persons or pro-perty as a matter of products liability, negligence or otherwise,or from any use or operation of any methods, products,instructions or ideas contained in the material herein. Becauseof rapid advances in the medical sciences, in particular,independent verification of diagnoses and drug dosages shouldbe made.

Although all advertising material is expected to conform toethical (medical) standards, inclusion in this publication doesnot constitute a guarantee or endorsement of the quality or valueof such product or of the claims made of it by its manufacturer.

Indexed or Abstracted in Index Medicus, Current Contents/Clinical Medicine, International Nursing Index, CumulativeIndex to Nursing & Allied Health Literature, and MEDLINE.

9A

Survey of isolation practices ata tertiary care pediatric hospitalJoseph V. Vayalumkal, BSc, MD,a Laurie Streitenberger, RN, BSc, CIC,a,b Rick Wray, RN, BHS, CIC,a,b Carol Goldman, RN,BScN, CIC,a,b Renee Freeman, RN, BScN, CIC,a,b Steven Drews, PhD,d and Anne Matlow, MD, FRCPCa,b,c,d

Toronto, Ontario, Canada

Background: Although isolation precautions are an important aspect of hospital infection control, current rates of isolation in apediatric hospital and rates of compliance with established precautions are unknown. We therefore initiated hospital-wide pointprevalence studies to determine unit-specific rates of patient isolation and compliance with proper isolation requirements focusingon communication of isolation status and availability of personal protective equipment. In this report, we present data from thefirst 14 months of the study.Methods: This was a prospective observational study. Twice monthly, between January 2004 and February 2005, infection controlprofessionals reviewed the types and appropriateness of isolation of all hospitalized patients, except for those on the psychiatryunit.Results: Seventeen percent of patients in the hospital during the study period were isolated, most frequently for community-acquired infections. Droplet isolation precautions were the most common form of isolation. Overall, only 74.6% of patientswere isolated appropriately. The solid organ transplantation, hematology/oncology, and bone marrow transplantation unitswere those with the highest rates of inappropriate isolation.Conclusion: At our hospital, community-acquired infections, in particular respiratory infections, were the most common reasonsfor patient isolation. Monitoring of the appropriateness of isolation precautions offers the opportunity to reduce health care-related transmission of infection and identify specific target areas for improvement. (Am J Infect Control 2007;35:207-11.)

Isolation precautions are an important aspect ofhospital infection control programs and are parti-cularly important in pediatric settings given the highadmission rates for viral respiratory (VRI) and gastroin-testinal (VGI) infections. Factors such as diapering ofpatients and inability of young patients to adhere toproper respiratory etiquette help facilitate transmis-sion of infectious pathogens. The high prevalence ofhealth care-associated VRI and VGI reported in pediat-ric institutions underscores the importance of compli-ance with isolation protocols. Education and periodicevaluation of adherence to precautions are recommen-ded administrative controls to optimize isolation prac-tices in hospitals.1

From the Division of Infectious Diseases,a the Department of Pediatrics,b

and the Department of Pediatric Laboratory Medicine, The Hospital ForSick Children, Toronto, Canadac; and the Department of LaboratoryMedicine and Pathobiology,d University of Toronto, Toronto, Canada.

Address correspondence to Anne Matlow, MD, FRCPC, Director Infec-tion Prevention and Control, The Hospital for Sick Children, 555University Avenue, Toronto, Ontario M5G 1X8, Canada. E-mail: anne.matlow@sickkids.ca.

0196-6553/$32.00

Copyright ª 2007 by the Association for Professionals in InfectionControl and Epidemiology, Inc.

doi:10.1016/j.ajic.2006.03.012

As the first step of a performance improvement ini-tiative to improve isolation practices, we examined theepidemiology of patient isolation in the hospital andinitiated regular evaluation of the appropriateness ofisolation practices at our hospital. In this paper, we re-port our first 14 months of data.

METHODS

Setting

The Hospital for Sick Children is a 300-bed, univer-sity-affiliated tertiary care pediatric hospital in Toronto,Canada. All inpatient units in the hospital except forthe psychiatry unit were included in this study.

Data collection

The study took place from January 2004 to February2005 in the form of repeated point prevalence evalua-tions. Twice monthly, for a total of 28 observations, acertified infection control professional visited a unitto review the types (as per the Hospital Infection Con-trol Practices Advisory Committee [HICPAC] guide-lines1) and appropriateness of isolation precautionsin place at the time of the visit, focusing on communi-cation of isolation status (signage, computer entry) andavailability of personal protective equipment. Timesfor the audit were based on convenience for the

207

208 Vol. 35 No. 4 Vayalumkal et al

Table 1. Isolation rates stratified by source of infection

Unit

Isolation rate (total isolated/

total census 3 100)

Isolation rate for HAIC (total isolated

for HAIC/total isolated 3 100) Isolation rate for CAIC

PICU 21.9% (77/351) 31.2% (24/77) 68.8% (53/77)

NICU 5.7% (25/436) 80% (20/25) 20% (5/25)

Cardiology and cardiac surgery 7.9% (21/265) 38.1% (8/21) 61.9% (13/21)

Surgical wards 6.9% (62/901) 25.8% (16/62) 74.2% (46/62)

SOT, gastroenterology, and rheumatology 18.8% (55/293) 18.2% (10/55) 81.8% (45/55)

Neurology 6.7% (8/120) 0 (0/8) 100% (8/8)

General pediatrics and respirology 39.2% (320/816) 2.5% (8/320) 97.5% (312/320)

Hematology/oncology/HSCT 12.1% (55/454) 43.6% (24/55) 66.4% (31/55)

Total 17.1% (622/3636) 17.7% (110/622) 82.3% (512/622)

HAIC, Hospital-associated infection/colonization; HSCT, hematopoetic stem cell transplant; PICU, pediatric intensive care unit; NICU, neonatal intensive care unit; CAIC, community-

associated infection/colonization; SOT, solid organ transplant.

Table 2. Distribution of isolation precautions by category

Unit

Contact (total No. in

contact isolation/total

No. isolated) (%)

Droplet/contact (total No. in

droplet/contact isolated/total

No. isolated) (%)

Airborne (total No. in

airborne isolation/total

No. isolated) (%)

MDRO (total No. in

MDRO isolation/total

No. isolated) (%)

PICU 8/77 (10.4) 48/77 (62.3) 2/77 (2.5) 19/77 (24.7)

NICU 1/25 (4) 1/25 (4) 2/25 (8) 21/25 (84)

Cardiology and cardiac surgery 9/21 (42.9) 10/21 (47.6) 0/21 (0) 2/21 (9.5)

Surgical wards 35/62 (56.5) 22/62 (35.5) 0/62 (0) 5/62 (8.1)

SOT, gastroenterology,

and rheumatology

26/55 (47.3) 21/55 (38.2) 0/55 (0) 8/55 (14.5)

Neurology 3/8 (37.5) 5/8 (62.5) 0/8 (0) 0/8 (0)

General pediatrics and respirology 82/320 (25.6) 180/320 (56.3) 35/320 (10.9) 23/320 (7.2)

Hematology/oncology/HSCT 30/55 (54.5) 25/55 (45.5) 0/55 (0) 0/55 (0)

Total 194/623 (31.1) 312/623 (50.1) 39/623 (6.3) 77/623 (12.5)

MDRO, Multidrug-resistant organisms; PICU, pediatric intensive care unit; NICU, neonatal intensive care unit: SOT, solid organ transplant; HSCT, hematopoetic stem cell transplant.

practitioner and were approximately 2 weeks apart.The practitioner initially reviewed all patient diagnosesand any symptoms that would necessitate isolation (eg,cough, diarrhea). Patients were discussed with the clin-ical leader/charge nurse of the unit and walk-aroundrounds on the unit were conducted to review each pa-tient’s isolation status. Appropriate isolation was de-fined as (1) type of isolation consistent with HICPACrecommendations, (2) isolation duration consistentwith the Centers for Disease Control and Prevention(CDC) guidelines, (3) proper isolation sign on door, (4)proper personal protective equipment available out-side patient room, and (5) proper documentation of iso-lation requirements in the computerized patient recordsystem (KIDCOM), excluding the pediatric intensivecare unit (PICU) and neonatal intensive care unit(NICU), which do not have access to KIDCOM. Inappro-priate isolation was defined as (1) type of isolation notconsistent with HICPAC recommendations (eg, dropletinstead of airborne precautions), (2) isolation durationnot consistent with CDC guidelines (ie, too long or tooshort), (3) proper isolation sign not on door, (4) properpersonal protective equipment not available outsidepatient room (this depended on type of precautionsrequired, ie, gloves and gowns available for contact

isolation), and (5) proper documentation of isolationrequirements not entered in the computerized patientrecord system (KIDCOM). The computerized recordalerts other health care workers about the isolation sta-tus of the patient in case the patient is transported toanother part of the hospital for diagnostic tests, proce-dures, or treatments.

Data management

All information was entered and stored in an elec-tronic database (Microsoft Excel; Microsoft Corp, Red-mond, WA).

Statistical analysis

Testing was performed using the x2 test on GraphPadPrism 4 for Windows (Hearne Scientific Software, NewZealand).

RESULTS

Seventeen percent (623/3636) of all patients hospi-talized during the study period were isolated, primarilyfor community-associated rather than health care-associated infection or multidrug-resistant organism

Vayalumkal et al May 2007 209

(MDRO) colonization (14.1 vs 3, respectively) (Table 1).The general pediatrics ward had the highest rate of pa-tient isolation. The NICU, followed by the hematology/oncology/human stem cell transplantation ward hadthe highest percentage of patients under isolation pre-cautions for health care-associated infection/MDROcolonizations. The distribution of isolation require-ments is shown in Table 2. Overall, droplet precautionswere the most common category of isolation precau-tions (50.2%), followed by contact precautions (31.2%).Isolation for MDROs accounted for 12.3% of patients inisolation.

One hundred sixty-seven of the 623 isolated pa-tients (26.4%) were isolated inappropriately. The ma-jority of those deemed inappropriately isolated lackedeither documentation in the computerized medical rec-ord or had no or inappropriate signage. Patients were4 times more likely to be under isolated than overisolated (Table 3). A complete breakdown of the appro-priateness of isolation is provided in Table 4. A signifi-cant association between inappropriate isolation andclinical unit was noted, with the solid organ transplan-tation unit together with the hematology/oncology/hematopoeitic stem cell transplantation wards havinghigher rates of isolation than the other units combined(P , .001) (Table 4).

DISCUSSION

A report from our institution close to 2 decades agoquantified hospital isolation in terms of extent of us-age, seasonal variation, isolation category, and type ofinfection (nosocomial vs community acquired).2 Theauthors reported that 15% of the beds were occupiedby patients requiring isolation and that the number ofpatients requiring isolation exceeded the number ofisolation rooms on 32% of the days of the year. Twentyyears later, the percentage of patients requiring isola-tion is similar at 17.1%. In another report from our hos-pital, multibed rooms were more likely to be associatedwith nosocomial transmission of VGI than single-bedrooms.3 An increased need for single-bed rooms in

Table 3. Distribution of inappropriate events by category

Criterion for

inappropriateness

Number of

inappropriate

over-isolation

events (%)

Number of

inappropriate

under-isolation

events (%)

Incorrect sign 12/167 (7.2) 13/167 (7.8)

No sign 0/167 (0) 32/167 (19.2)

No KIDCOM change 0/167 (0) 81/167 (48.5)

Isolated too long 22/167 (13.2) 0/167 (0)

No observed PPE 0/167 (0) 7/167 (4.1)

Total 34/167 (20) 133/167 (80)

PPE, Personal protective equipment.

hospitals for isolation purposes has previously beenidentified4; we are fortunate that 85% of the beds inour current hospital are in single-bed rooms.

In the current study, droplet precautions and contactprecautions were the most common types of isolationprecautions used, primarily for VRI and VGI, respec-tively (data not shown). The high rate of isolation onthe general pediatrics ward for community-associatedinfection/MDRO colonization is consistent with the pre-dominance of precautions for VRI and VGI at our centerand at others.4,5 Although the NICU had the lowest rateof isolation overall, it had the highest rate of isolationfor health care-associated infections/MDRO coloniza-tion, with colonizations with MDROs accounting forthe majority of isolations. The hematology/oncology/hematopoeitic stem cell transplantation ward followed,with a high rate of patients isolated for health care-associated infections/MDRO colonization, in this case,primarily contact isolation for gastrointestinal infec-tion including VGI and Clostridium difficile infection.Isolation for MDROs accounted for 12.3% of patientsin isolation.

Few studies have reported on actual compliancewith isolation precautions. Compliance with infectioncontrol procedures for tuberculosis was evaluated in2 pediatric hospitals, and significant breeches in infec-tion control precautions were noted.6 In another study,widespread noncompliance in an isolation bay of asurgical intensive care unit was reported.7 In a studyconducted at a children’s hospital in Nebraska, therewas significant improvement in a number of infectioncontrol practices including adherence to isolation pre-cautions as a result of surveillance rounds combinedwith ongoing education.8 In our study, in which thefocus of the audit was the appropriateness of commu-nication of isolation status (ie, signage, computer entry)and availability of personal protective equipment (asopposed to compliance with hand hygiene before and

Table 4. Distribution of inappropriate isolation byclinical unit

Clinical unit

Number

inappropriately

isolated/total

number isolated Percentage

PICU 10/77 13.1

NICU 0/25 0

Cardiology and cardiac surgery 1/21 4.8

Surgical wards 14/62 22.6

SOT, gastroenterology,

and rheumatology

25/55 45.5

Neurology 2/8 25.0

General pediatrics and respirology 96/320 30.0

Hematology/oncology and HSCT 19/55 34.5

PICU, pediatric intensive care unit; NICU, neonatal intensive care unit; SOT, solid organ

transplant; HSCT, hematopoetic stem cell transplant.

210 Vol. 35 No. 4 Vayalumkal et al

after patient contact, or proper use of barriers/personalprotective equipment), we found that approximately25% of patients were inappropriately and usuallyunder isolated, suggesting that here too there is greatopportunity for improvement.

Although appropriate patient isolation is considereda standard of care, the numerous reports of health care-acquired infections and transmission of MDROs implythat it is often inadequately performed. There are neg-ative consequences to both over isolation and underisolation For example, a nationwide hospital surveyinvestigating infection control practices in Thailanddemonstrated that staff in a high percentage of hospitalsperformed unnecessary infection control procedures,including wearing protective gowns in intensive careunits and overusing sterile gloves.9 Such practices con-sume valuable resources unnecessarily. Furthermore,evidence is emerging that some patients in isolationmay receive fewer visits by health care workers, lead-ing to suboptimal care and decreased patient satis-faction.10-12 Donning precautions required for thecare of an intensive care unit patient with MDROshas been reported to consume 2 hours of time, whichmay be considered a deterrent to patient care.13

Conversely, under isolation of patients facilitatescross transmission of infection, as well as acquisitionof infection by health care personnel. A recent studyusing pulsed-field gel electrophoresis to type bacteriaisolated from clinical infections estimated that at least37.5% of all nosocomial infections identified in a sur-gical intensive care unit were due to cross transmis-sion.14 Reports of staff acquisition of infections suchas tuberculosis and pertussis further attests to theimportance of appropriate isolation practices.

In this study, we found the 2 main reasons for inap-propriate isolation to be wrong or lack of signage andfailure to update the computerized medical record(KIDCOM). The wrong signage was usually an airborneisolation sign instead of one for droplet precautionsor inappropriate use of an MDRO sign. Thirty-two pa-tients who should have been isolated had no sign out-side their room.

The computerized medical record is of particularimportance in relaying isolation information at transi-tions in care (eg, transfer to the operating room). Hand-offs are known to be vulnerable times for patientsafety.15-17 In a multicenter, pediatric performanceimprovement collaborative study of patient transfersfrom the emergency department to an inpatient unit,isolation requirements were relayed only 60% of thetime. Use of a checklist increased isolation informationtransfer to 98%.18 Although the current study did notassess the modes of information transfer betweenhealth care workers, it is generally accepted thatbuilt-in redundancies are important features of highly

reliable organizations. As such, we feel justified tohave included failure-to-enter isolation requirementson our computerized record as inappropriate isolation.

In summary, in this study, we have revisited isola-tion patterns and practices at our hospital. During thestudy period, close to one fifth of all patients requiredisolation, approximating data reported from our institu-tion close to 2 decades ago. Consistent with other re-ports, the bulk of the burden of communicable diseasewas due to community-acquired VRI and VGI. Suchinformation may be useful to other institutions plan-ning new or renovating existing pediatric facilities.The study has identified units in our hospital that wouldbenefit from focused interventions to improve isolationpractices. Frequent surveillance by infection controlprofessionals with real-time feedback to front-lineworkers is one way to improve compliance with isola-tion precautions. The resultant effect should be a de-crease in infection transmission and saving of resources.

References

1. Garner JS, Hospital Infection Control Practices Advisory Committee.

Guideline for isolation precautions in hospitals. Infect Control Hosp

Epidemiol 1996;17:53-80.

2. Kim MH, Mindorff C, Patrick ML, Gold R, Ford-Jones EL. Isolation

usage in a pediatric hospital. Infect Control 1987;8:195-9.

3. Ford-Jones EL, Mindorff CM, Gold R, Petric M. The incidence of viral-

associated diarrhea after admission to a pediatric hospital. Am J Epide-

miol 1990;131:711-8.

4. Langley JM, Hanakowski M, Bortolussi R. Demand for isolation beds in

a pediatric hospital. Am J Infect Control 1994;22:207-11.

5. Siegel J. Controversies in isolation and general infection control prac-

tices in pediatrics. Semin Pediatr Infect Dis 2002;13:48-54.

6. Kellerman S, Saiman L, San Gabriel P, Besser R, Jarvis WR. Observational

study of the use of infection control interventions for Mycobacterium

tuberculosis in pediatric facilities. Pediatr Infect Dis J 2001;20:566-70.

7. Pettinger A, Nettleman M. Epidemiology of isolation precautions.

Infect Control Hosp Epidemiol 1991;12:303-7.

8. Chatterjee A, Heybrock B, Plummer S, Eischen K. Impact of surveillance

rounds on adherence to infection control policies and procedures at

a children’s hospital. Infect Control Hosp Epidemiol 2004;25:786-8.

9. Kunaratanapruk S, Silpapojakul K. Unnecessary hospital infection con-

trol practices in Thailand: a survey. J Hosp Infect 1998;40:55-9.

10. Stelfox H, Bates D, Redelmeier D. Safety of patients isolated for infec-

tion control. JAMA 2003;290:1899-905.

11. McKeever P, O’Neill S, Miller K-L. Managing space and marking time:

mothering severely ill infants in hospital isolation. Qual Health Res

2002;12:1020-32.

12. Evans HL, Shaffer MM, Hughes MG, Smith RL, Chong TW, Raymond

DP, et al. Contact isolation in surgical patients: a barrier to care?

Surgery 2003;134:180-8.

13. Saulnier FF, Hubert H, Onimus TM, Beague S, Nseir S, Grandbastien B,

et al. Assessing excess nurse workload generated by multiresistant

nosocomial bacteria in intensive care. Infect Control Hosp Epidemiol

2001;22:273-8.

14. Patterson ES, Roth EM, Woords DD, Chow R, Gomes JO. Handoff

strategies in settings with high consequences for failure: lessons for

health care operations. Int J Qual Health Care 2004;16:125-32.

15. Weist K, Pollege K, Schulz I, Ruden H, Gastmeier P. How many

nosocomial infections are associated with cross-transmission?

Vayalumkal et al May 2007 211

A prospective cohort study in a surgical intensive care unit. Infect

Control Hosp Epidemiol 2002;23:127-32.

16. Warren J, Fromm RE, Orr RA, Rotello LC, Horst HM. Guidelines for

the inter- and intra-hospital transport of critically ill patients. Crit

Care Med 2004;32:256-62.

17. Beckmann U, Gillies DM, Berenholtz SM, Wu AW, Pronovost P.

Incidents relating to the intra-hospital transfer of critically ill

patients. An analysis of the reports submitted to the Australian

Incident Monitoring Study in Intensive Care. Intensive Care Med

2004;30:1579-85.

18. Streitenberger K, Matlow A. Check and cross-check: using a standard-

ized checklist to reduce adverse events during patient transfer.

Presented at the National Institute for Child Health Quality Annual

Meeting, Coronado, California, January 2005.

Emergence of resistant Acinetobacterbaumannii in critically ill patients withinan acute care teaching hospital anda long-term acute care hospitalClaudester Stephens, MT, MPH, CIC, Stephen J. Francis, MD, Virginia Abell, RN, BA, CIC,Joseph R. DiPersio, PhD, Diplomate, ABMM, and Patricia Wells, RN, CIC

Akron, Ohio

Background: Acinetobacter baumannii is a gram-negative, coccobacillus found in water and is a significant nosocomial pathogen inhospitals. This report chronicles the appearance in June 2003 of a multidrug-resistant A baumannii (MDR-AB) strain, its dissemi-nation, and interventions used to control it in an acute care hospital (ACH) and long-term acute care facility (LTAC).Methods: Molecular typing using pulsed-field gel electrophoresis (PFGE) showed that 88 of 99 strains (89%) gave an identical band-ing designated as clone A. Eight additional isolates were variants of clone A, and 3 isolates were unrelated.Results: A baumannii was isolated from 229 patients between January 2003 and December 2004. Of these patients, 151 (66%)were colonized/infected with MDR-AB. Most isolates were resistant to antibiotics except for imipenem and ampicillin/sulbactam.Isolates included 108 (72%) in the respiratory tract, 32 (21%) in wounds, 6 (4%) in blood, and 5 (3%) in urine. Most isolates werefound in the LTAC (70 isolates), ICU step-down (27 isolates), and ICU (26 isolates).Conclusion: This epidemiologic history illustrates (1) epidemic clonal spread, (2) target populations, (3) variable monthly preva-lence, and (4) intervention outcomes. With intervention, the number of new isolates in the ACH decreased by dedicating an infec-tion control professional to critical care, daily surveillance, isolation of positive MDR-AB patients, universal gloving, and routinelyreporting results. (Am J Infect Control 2007;35:212-5.)

Acinetobacter baumannii (AB) is a nonlactose-fermenting, gram-negative coccobacillus that is wide-spread in the environment and is increasingly foundin medical facilities.1 AB has been implicated in skinand soft tissue infections, pneumonia, bacteremia,and urinary tract infections. Studies have shown thatthe increase in nosocomial infections caused bygram-negative multidrug-resistant organisms is relatedto the escalated use of broad-spectrum antibiotics.2

AB is an organism that can colonize or cause infectionand has the ability to survive for long periods of time inthe hospital environment, which increases the risk ofhorizontal transmission via a health care worker oran environmental source.3

This study examined AB isolates from patientswithin a large, acute care (teaching) hospital (ACH)

From the Summa Health System, Akron, OH.

Address correspondence to Patricia Wells, RN, CIC, Infection ControlDepartment, 525 E. Market Street, Akron, OH 44309-2090. E-mail:wellsp@summa-health.org.

Supported by MERCK and Co., Inc. with a research grant.

0196-6553/$32.00

Copyright ª 2007 by the Association for Professionals in InfectionControl and Epidemiology, Inc.

doi:10.1016/j.ajic.2006.04.208

212

and a long-term acute care (LTAC) facility containedwithin the same building complex located in northeastcentral Ohio. The ACH facility has 4 intensive careunits (ICUs) including a surgical/trauma ICU, a surgicalcardiovascular ICU, a coronary care unit (CCU), and amedical ICU. The LTAC is an independent 40-bed hos-pital based within the physical building of the ACH.The LTAC admits critically ill long-term patients fromthe ACH facility and outlying hospitals.

In May 2003, the microbiology laboratory detectedan increase in the number of multidrug-resistant Acine-tobacter baumannii (MDR-AB) isolates. The patient withthe initial resistant isolate (a sacral wound ulcer) onMarch 28, 2003, was admitted to the LTAC on January28, 2006, from the ACH and may have been exposedto several other patients with a relatively more sensi-tive Acinetobacter baumannii on the LTAC. The first 3isolates from April to June of 2003 were identified inthe LTAC. Initial review of the room transfers of patientswith MDR-AB indicates spread to the ACH units viareadmissions to the ACH from the LTAC. The majorityof isolates were from respiratory specimens frompatients located in the ACH ICUs and the LTAC. The larg-est numbers of MDR-AB isolates were seen in June2003 (n 5 15). Initial DNA fingerprinting studies indi-cated that the majority of resistant isolates belongedto 1 DNA fingerprint pattern. A search of computer

Stephens et al May 2007 213

records indicted that the resistant phenotype firstappeared in late March 2003 in the LTAC unit. The orga-nism next appeared in the ACH ICU. Of 229 patientsidentified as having at least 1 culture positive for AB,151 had MDR-AB. This report outlines the epidemio-logic study that was undertaken, the results of mole-cular testing, and the interventions taken to reducetransmission and prevalence of MDR-AB.

MATERIALS AND METHODS

Organism identification

Using standard commercial identification systems,229 AB isolates were identified in the clinical microbi-ology laboratory between March 2003 and December2004. Of the 229 isolates, 151 (66%) were (MDR-AB).

Susceptibility testing

Initial susceptibility testing was performed by eitherdisk diffusion or broth microdilution using a commer-cially prepared panel (DADE Microscan, West Sacra-mento, CA). Isolates were considered to be MDR-AB ifthey were sensitive to 3 or less antibiotics. A notablecharacteristic was that all MDR-AB isolates were sensi-tive to imipenem. The index case isolate was resistantto all antibiotics tested except ampicillin/sulbactamand imipenem. Resistance was noted with penicillins,aminoglycosides, and fluoroquinolones.

Molecular epidemiology

Pulsed-field gel electrophoresis (PFGE) of SpeI digestsof bacterial chromosomal DNA was performed in houseon 103 AB isolates (99 MDR-AB and 4 non-MDR-AB) us-ing the GenePath system (Bio-Rad Laboratories, Hercu-les, CA). Analysis of fingerprint patterns was performedby visual inspection. MDR-AB isolates showing identicalDNA banding patterns were considered to be clonal. Iso-lates that differed by 1 to 3 bands were considered pos-sibly related, whereas isolates showing greater than 3band differences were thought to be unrelated.

Data analysis

Statistical data was collected by review of patientmedical charts, discharge summaries, pharmacy rec-ords, and microbiology laboratory reports.

RESULTS

Age and sex demographics show a total of 86 menranging in age from 22 to 85 years and 65 women rang-ing in age from 34 to 84 years. Sources of isolates wererespiratory tract, 108 (72%); wounds, 32 (21%); blood,6 (4%); and urine, 5 (3%). An epidemiologic review ofMDR-AB isolates revealed that 20% of the ventilator-

associated pneumonias (VAPs) reported in 2003 and22% of the VAPs in 2004 were associated with MDR-AB. A low percentage of bloodstream infections (BSIs)and surgical site infections (SSIs) were associatedwith MDR-AB. The infection rates were 0% for BSIsand 1% for SSIs with MDR-AB in 2003 and 2004.

The index case appeared in March 2003. Thenumber of isolates from the LTAC peaked in May of2003, and the number of isolates from ACH peakedin June of 2003. The highest monthly occurrence ofMDR-AB occurred June 2003 (see Fig 1). The LTAChad the highest rate of MDR-AB, and the ACH had thehighest number of isolates. There were 70 (46.4%)isolates on the LTAC, 27 (17.9%) isolates on the medicalICU, and 26 (17.2%) isolates on the surgical/traumaICU. Of the 28 (18.5%) remaining isolates, none wereconcentrated on any one particular unit (includingthe surgical cardiovascular ICU and the CCU). Figure 2shows the incidence rate of MDR-AB per 1000 patient-days.

Molecular typing of chromosomal DNA digest usingPFGE showed that 88 of 99 MDR-AB strains tested(89%) gave an identical banding pattern, which wasdesignated as clone A (see Fig 3). Eight additional iso-lates were variants of clone A, and 3 isolates wereunrelated to clone A. These 3 isolates gave similar sus-ceptibility patterns to the index MDR-AB isolate. The4 non-MDR-AB isolates tested gave DNA fingerprintpatterns distinct from all MDR-AB isolates tested.

DISCUSSION

To control the spread of this organism in the ACH,a 3-fold team approach was initiated. This approachincluded (1) the Microbiology Department’s rapidpreliminary identification and notification to the Infec-tion Control Department, (2) the placement of all new

Fig 1. MDR-AB isolates 2003-2004 in ACHand LTAC.

214 Vol. 35 No. 4 Stephens et al

patients with positive isolates into contact precaution,4

and (3) the notification of attending physicians of theincreased occurrence of MDR-AB.

The original infection control precautions—isolation of patients into private rooms, contact pre-cautions, and STOP signs at the door—failed to controlcompletely the spread of this organism. Additional mea-sures were taken beginning the first week of January2004. Education of staff was provided on a case-by-case basis.

Timely notification of readmissions and transfers ofaffected patients were improved so that isolation pre-cautions could be instituted immediately. The infectioncontrol staff instituted extended contact precautions.These precautions included universal gloving for theentire nursing unit in which these patients were admit-ted. Universal gloving is the use of disposable glovesfor any patient contact regardless of the patient’smicrobiologic status.5 The Infection Control Depart-ment increased the investigation to include room trans-fers of affected patients and comparisons of thesensitivities of the organisms to assess for horizontaltransmission.

The infection control staff at the ACH worked collab-oratively with the infection control nurse at the LTAC.Equivalent procedural revisions, as discussed above,were initiated at the LTAC. Several conditions madecompliance more difficult for the LTAC: nurse/patientratio, high-risk observation rooms, and turnover inthe infection control position. There are more patientsper nurse in the LTAC. The LTAC high-risk observationrooms are multibed. This demanded prioritizationbetween close observation of respiratory status andthe need for isolation. The change in the infection

Fig 2. MDR-AB incidence rate per 1000 patient-days comparison ACH and LTAC 2003-2004.

control nurse left the LTAC without adequate infectioncontrol expertise for several weeks. The newly hirednurse was without infection control experience. Thesefactors created an environment less conducive to fullcompliance of the newest procedures. As time pro-gressed, LTAC infection control compliance improved,including appropriate use of extended contact precau-tions and universal gloving.

CONCLUSION

MDR-AB, in agreement with the experience ofothers,6 is primarily a respiratory organism, with themajority of the isolates found to be colonizers ratherthan causing disease. The organism is very resistantto most antibiotics, but we do not believe it to be highlypathogenic. The incidence of new cases was as highas 15 isolates in June of 2003 and as low as 3 isolatesin each of the months of March, April, and June of2004. The LTAC appears to have been an ongoing reser-voir for multidrug-resistant organisms.7

Continuing surveillance of MDR-AB indicates thatthis organism has become entrenched in both theACH and the LTAC facilities. Regardless of our stringentinfection control measures, we have been unable tototally eradicate the organism. Surveillance of newadmissions from local nursing homes are now showingvarious diagnostic isolates colonized with MDR-AB,indicating the presence of this organism in other facil-ities. The number of isolates has remained low in theACH (1 or 2 per month) since the end of this study,

Fig 3. PFGE patterns of SpeI-digested genomicDNA of A baumannii nosocomial isolates (lanes 2-4and 6-9). Control Staphylococcus aureus in lanes 1, 5,

and 10.

Stephens et al May 2007 215

but the number of isolates in the LTAC continuesto remain elevated (2 or 3 per month) with a high of8 in June 2005 and a low of 0 in July 2005).

The infection control professional dedicated to thecritical care units in the ACH maintains daily surveil-lance of AB occurrence in the LTAC and ACH, evalu-ates sensitivity patterns for increased resistance,and reviews antibiotic usage. The use of extendedcontact precautions, continuing education of medicalstaff in affected units, monitoring for readmissionsand transfers of affected patients, and reporting re-sults to the individual nursing units on a routine ba-sis has kept transmission of MDR-AB controlled inthe ACH. The Infection Control Department fromthe ACH continues to monitor for all emerging multi-drug-resistant organisms and works with the infec-tion control professional at the LTAC. Additionalactions could include aggressive environmental pro-cedures and practice revisions including respiratoryprecautions.6,8

The authors thank L. B. Fields and S. K. Bullock for helping with the preparation of themanuscript.

References

1. Abbo A, Navon-Venezia S, Hammer-Muntz O, Krichali T, Siegman-Igra

Y, Carmeli Y. Multidrug-resistant Acinetobacter baumannii. Emerg Infect

Dis 2005;11:1-13.

2. Busk K. Beta-lactamases of increasing clinical importance. Curr Pharm

Des 1999;5:839-45.

3. Maslow JN, Glaze T, Adams P, Lataillade M. Concurrent outbreak of

multidrug-resistant and susceptible subclones of Acinetobacter baumannii

affecting different wards of a single hospital. Infect Control Hosp Epide-

miol 2005;26:69-75.

4. Guidelines for isolation precautions in hospitals. Hospital Infection Control

Advisory Committee. Publication date: January 1, 1996. Available at: http:

//wonder.cdc.gov/wonder/pervguid/p0000419/p0000419.asp. Accessed

January 1, 2004.

5. Weinstein RA. Controlling antimicrobial resistance in hospitals: infec-

tion control and use of antibiotics. Emerg Infect Dis 2001;7:188-91.

6. Urban C, Segal-Maurer S, Rahal JJ. Considerations in control and treat-

ment of nosocomial infections due to multidrug-resistant Acinetobacter

baumannii. Clin Pract 2003;36:1268-74.

7. Dipersio JR, Deshpande LM, Biedenbach DJ, Toleman MA, Walsh TR,

Jones RN. Evolution and dissemination of extended-spectrum

b-lactamase-producing Klebsiella pneumoniae: epidemiology and molec-

ular report from the SENTRY Antimicrobial Surveillance program

(1997-2003). Diagn Microbiol Infect Dis 2005;51:1-7.

8. Brooks SE, Walczak MA, Hameed R. Are we doing enough to contain

acinetobacter infections? Infect Control Epidemiol 2000;21:304.

Validation of surgical site infectionsurveillance in orthopedic proceduresKaisa Huotari, MD,a,b Niina Agthe, RN,a and Outi Lyytikainen, MD, PhDa

Helsinki, Finland

Background: Valid data are essential for a national surveillance system of nosocomial infections.Methods: In 8 hospitals conducting surgical site infection (SSI) surveillance for orthopedic procedures, a validation team performeda blinded retrospective chart review (10 operations with reported infections, 40 without) and interviewed infection control nurses.Results: In total, 397 patient charts were reviewed. Positive and negative predictive values for routine surveillance were 94% (95%CI: 89%-99%) and 99% (95% CI: 99%-100%), respectively. When these results were applied to the aggregated surveillance data (403infections, 10,068 noninfections), sensitivity was 75% (95% CI: 56%-93%) and specificity 100% (95% CI: 97%-100%). The follow-ing case finding methods were used: ward visits (in 7/8 hospitals), microbiology reports (5/8), ward notifications by link nurses (8/8),and other nursing (7/8) and medical (5/8) staff. The wound culture rate ranged from 9 to 67 per 1000 patient-days. All hospitalscarried out postdischarge surveillance on readmission and all but 1 at follow-up visits and by an additional questionnaire.Conclusion: Most SSIs reported by the hospitals were true infections, showing that, when an SSI was reported, the definitions werecorrectly implemented. Some SSIs were missed, which might be due to weaknesses in case finding. Variation in diagnostic prac-tices may also affect SSI rates. (Am J Infect Control 2007;35:216-21.)

Since the Study on the Efficacy of Nosocomial Infec-tion Control (SENIC) showed that hospitals with inten-sive infection surveillance and control activities wereeffective in reducing their nosocomial infection rates,1

surveillance has been an essential part of well-de-signed infection control programs. Efficacy of sur-veillance with feedback to practising surgeons inpreventing surgical site infections (SSIs) has also beendemonstrated in several other studies.2-5 To facilitatethe prevention of SSI, national or regional surveillancenetworks for nosocomial infections have been widelyintroduced in the United States and Europe.

There are demands for interhospital comparison ofSSI rates as a measure of the quality of patient care.Data should therefore be collected on accuracy andconsistency.6,7 In surveillance networks, validationstudies are essential to ensure credibility and validityof data.8,9

From the National Public Health Institutea and Department of Medicine,Division of Infectious Diseases, Helsinki University Central Hospital,b

Helsinki, Finland.

Address correspondence to Kaisa Huotari, MD, National Public HealthInstitute, Mannerheimintie 166, 00300 Helsinki, Finland. E-mail: kaisa.huotari@ktl.fi.

Supported in part by a grant from the Research Foundation of OrionCorporation.

Abstract presented at the 15th Annual Scientific Meeting of the Societyfor Healthcare Epidemiology of America, April 9-12, 2005, Philadelphia,PA.

0196-6553/$32.00

Copyright ª 2007 by the Association for Professionals in InfectionControl and Epidemiology, Inc.

doi:10.1016/j.ajic.2006.01.009

216

SSIs in orthopedic prosthetic surgery are among themost serious health care-associated infections: pros-thetic joint infections may prolong length of hospitalstay, lead to reoperations, and increase health carecosts.10,11 Orthopedic operations were the first proce-dures under surveillance in the Finnish Hospital Infec-tion Program (SIRO), which has been operating since1999, and more hospitals perform SSI surveillancefor these operations than any other procedure. OurSSI rates for hip and knee prosthetic procedureshave been relatively high compared with rates re-ported by other surveillance systems in Europe andthe United States.12 Our previous study showed awide variation in these rates between participatinghospitals. The variation was due not only to differ-ences in postdischarge surveillance intensity but alsoto differences in in-hospital surveillance. We thereforefocused our first validation study on surveillance datain orthopedic surgery.

METHODS

Surveillance system

Nine hospitals (3 tertiary care, 3 secondary care, and3 other hospitals) conducted prospective surveillancefor SSIs in orthopedic surgery on a continuous basisduring 1999-2003. Procedures under surveillancewere hip and knee joint replacements and open reduc-tions of femur fracture.

In-hospital surveillance was conducted by using theCenters for Disease Control and Prevention (CDC) defi-nitions13 and the National Nosocomial Infection Sur-veillance System (NNIS) methodology.14,15 A writtenprotocol with CDC definitions translated into Finnish

Huotari, Agthe, and Lyytikainen May 2007 217

was provided. For case finding, the infection controlnurses (ICNs) responsible for surveillance are recom-mended to visit wards once a week and obtain addi-tional information from microbiology laboratoryreports, patient charts, and medical/nursing staff. Afterdischarge, all hospitals conducted surveillance on read-mission and at follow-up visits and, in 7 hospitals, alsoby an additional postdischarge questionnaire, whichwas given to each patient at discharge. The standard-ized questionnaire included the following variables:purulent drainage, other signs or symptoms of infec-tion (pain, tenderness, swelling, redness, or heat), inci-sion opened by a surgeon, wound culture taken, andantimicrobial treatment given to SSI. If a patient hadclinical signs or symptoms in the wound area and con-tacted the health care system, the questionnaire wasfilled out by a health care professional (nurse or physi-cian). The training on surveillance methodology orga-nized by the SIRO for local ICNs included a site visitat the beginning of the surveillance, meetings at leastonce a year, and an opportunity to consult the SIROteam by phone whenever needed.

For each infection, local ICNs collected and recordedmanually the following data on a form: the patient’snational identity code (includes age and sex), date ofsurgery, procedure code, date and type of SSI, microor-ganism if identified, and location of detection. Whenthe SSI was detected after discharge, the location of de-tection was recorded, ie, on readmission to the hospi-tal, at follow-up visit, or in outpatient setting by thepostdischarge questionnaire. For each patient undersurveillance, the following data were uploaded fromhospital databases and sent in electronic form to thenational center: the patient’s national identity code,date of surgery, procedure code, American Society ofAnesthesiologists (ASA) score, wound contaminationclass, duration and urgency of operation, date of ad-mission and discharge, and discharge status. In the na-tional surveillance database, the infection reports werelinked to the uploaded data by using the patient’s na-tional identity code and the date and code of the proce-dure. At regular intervals, the infection reports notcombined in the automatic process were examined.These reports were first checked for possible errorsin data entering, and, second, the local ICNs were con-tacted, if needed. The errors found were manuallycorrected.

Feedback was given through the project Web site,which was only accessible by use of a password givento authorized persons from the participating hospitals.Each hospital had access to its own data, and all hospi-tals had access to the aggregated data. Feedback in-cluded several report tables generated on-line fromthe database according to the users’ search criteria(hospital, time period, and procedure group).

Validation study

All 9 hospitals were invited to participate in thevoluntary validation study. During 1-day visits, a retro-spective chart review and a structured interview withthe ICN were carried out.

At each hospital, a validation team retrospectivelyreviewed a sample of patient charts, including all clin-ical data and laboratory and radiology reports. Thesample of charts contained 10 orthopedic operationswith and 40 without infection. The surveyors wereblinded to the patient’s infection status as recordedby the hospital ICN. The validation team, which wasconsidered the gold standard for validation, consistedof 2 SIRO team members: an experienced ICN (N.A.)and an infectious disease specialist-in-training (K.H.),supported by the SIRO project leader (O.L.). If an SSIwas assessed to be present, the CDC criteria that theSSI fulfilled and the type and date of the SSI wererecorded. After the review, discrepant files werediscussed with the ICN to determine sources ofdiscordance.

Exhaustiveness (completeness) of the denominatorwas examined by checking the proportion of infectionreports not combined in the automatic process (as aresult of, eg, an error in the patient’s national identitycode or missing denominator data). The proportionsof missing values in certain important fields (ASA score,wound contamination class, and duration of operation)were also evaluated.

In every hospital, the ICN responsible for quality ofsurveillance data was interviewed. The structured in-terview included questions about the process of datacollection, interpretation of the case definition, andmethods of postdischarge surveillance.

To calculate the rate of wound cultures per 1000patient-days in orthopedic wards of each hospital, thenumber of wound cultures, both surgical and otherwounds, taken by orthopedic wards in 2003 was ob-tained from local microbiology laboratories and thepatient-days from hospital administration.

Statistical methods

From the sample of charts reviewed, a positive pre-dictive value (PPV) was determined by calculating theproportion of true infections among all infections iden-tified by routine surveillance and a negative predictivevalue (NPV) by the proportion of true negative casesamong all patients identified as not infected by routinesurveillance. Because the prevalence of SSI differedbetween the sample of reviewed charts and the aggre-gated SIRO data (20.7% vs 3.8%, respectively), theresults of the chart review were applied to the aggre-gated surveillance data by multiplying the numberof all SSIs detected by routine surveillance during

218 Vol. 35 No. 4 Huotari, Agthe, and Lyytikainen

1999-2003 by the positive predictive value to obtainan approximation of the number of true infections.The same procedure was performed for negative caseswith the negative predictive value. This allowed sensi-tivity and specificity for the aggregated surveillancedata to be determined. Confidence intervals were cal-culated by the asymptotic normal theory with the deltamethod.

RESULTS

Eight of the 9 invited hospitals participated in thestudy. In total, the validation team reviewed 397 patientcharts. Results of the chart review are presented inTable 1 Routine surveillance had identified 83 SSIs,78 of which were identified as SSIs also by the valida-tion team (ie, true positive infections). Thus, the posi-tive predictive value was 94.0%, with a 95% CI of88.9% to 99.1%. Among the charts reviewed, no infec-tions during routine surveillance had been reported af-ter 314 operations, and the validation team confirmed310 of these negative reports (ie, true negative cases),yielding a negative predictive value of 98.7%, with a95% CI of 97.5% to 100%.

When the results of the validation study wereapplied to the aggregated surveillance data (during1999-2003: 592 infections and 14,551 noninfections),a sensitivity of 75.0%, with a 95% CI of 56.7% to93.4%, and a specificity of 99.8%, with a 95% CI of99.5% to 100%, were yielded (Table 2). One hospitalwas responsible for most of the infections missed(3/4) and for 1 false-positive infection (1/5). When theresults of this hospital were excluded, the positive pre-dictive value for the remaining 7 hospitals was 94.6%,with a 95% CI of 89.6% to 99.8%, and the negative pre-dictive value 99.6%, with a 95% CI of 98.9% to 100%.When these predictive values were applied to the ag-gregated surveillance data, without the excluded hospi-tal, the sensitivity for routine surveillance increased to91.5% (95% CI: 76.4%-100%) and the specificity to99.8% (95% CI: 99.6%-100%).

The reasons for missing 4 (2 superficial and 2 deepincisional) SSIs during routine surveillance were thatan ICN had not received information about the SSIs

Table 1. Results of the chart review*

Chart review by validation team

Infection 1 Infection 2 Total

Routine surveillance 1 78 5 83

Routine surveillance 2 4 310 314

Total 82 315 397

*Positive predictive value (PPV) 5 78 / 83 5 94.0 %. Negative predictive value (NPV) 5

310 / 314 5 98.7 %.

from an outpatient department (2) or from a ward (2)Explanations for overreporting SSIs were related tointerpretation of the case definition (5). Most (3/5)false-positive SSIs were superficial incisional withoutappropriate clinical signs or symptoms. One organ/space SSI with clinical onset 2 years after the operationwas reported.

Of the 78 SSIs classified as true infections by thevalidation team, 49 (63%) were superficial incisional,8 (10%) deep incisional, and 21 (27%) organ/space in-fections. The types of SSIs reported by routine surveil-lance differed from these (Table 3). Problems withclassification most commonly occurred in the organ/space SSI: half of which were misclassified as deep. Ifthe types of SSIs are given in 2 categories, superficialand severe (ie, deep incisional and organ/space), lessmisclassifications were found. Only 2 (4%) superficialSSIs were reported as severe and 4 (14%) severe SSIsas superficial in routine surveillance.

In the routine surveillance data for 1999-2003, theproportion of infection reports that was not combinedin the automatic process of all infection reports was4.4% (18/403): 8 with a typing error in the patient’snational identity code and 10 with denominator datamissing. The typing errors were corrected, and infec-tion and procedure data were then successfully com-bined. The proportions of missing values in the ASAscore, wound contamination class, and duration ofoperation were 3.7%, 1.4%, and 1.1%, respectively.

According to the structured interviews, the mostcommon case finding methods for prospective SSI sur-veillance were ICN ward visits (mostly once a week toonce a month, in 1 hospital only when needed), wardnotifications by link nurses and by other nursing staff,and postdischarge questionnaires, but microbiologylaboratory reports were also used (Table 4). If a caseof an SSI was found by ward notification, an ICNreviewed patient charts of every suspected SSI in 3hospitals and selectively in 5 hospitals. Some incom-pleteness in medical records was reported by ICNs (7/8);most ICNs (6/8) particularly described a need for more

Table 2. Chart review results applied to total SIROsurveillance data*

All operations

under surveillance during 1999-2003

Infection 1 Infection 2 Total

Surveillance 1 556y 36 592

Surveillance 2 185 14,366z 14,551

Total 741 14,402 15,143

*Sensitivity 5 556/741 5 75.0 %. Specificity 5 14,366/14,402 5 99.8 %.yPPV 3 592 5 556.zNPV 3 14,551 5 14,366.

Huotari, Agthe, and Lyytikainen May 2007 219

accurate documenting of wound status. One ICN re-ported that antimicrobial agents used were not alwaysdocumented (mostly prescribed at discharge).

Most of the hospitals conducted postdischarge sur-veillance at follow-up visits (7/8), on readmission (8/8), and in outpatient settings by an additional question-naire (7/8). The postdischarge questionnaire was givento each patient under surveillance at discharge. Fivehospitals used the standard questionnaire providedby the SIRO, which included all criteria of the CDC def-inition. Two hospitals used their own postdischargequestionnaires: one with no criteria and the otherwith the most important criteria. All questionnaireswere requested to be returned by 4 hospitals: the re-sponse rate in these hospitals varied between 46%and 70%. In 3 hospitals, the questionnaires were re-quested to be returned only if an SSI was identified.Three ICNs noted that, also, local/regional settings tak-ing care of follow-up treatment and rehabilitation weretrained in postdischarge surveillance.

The ICNs experienced difficulties in the interpreta-tion of SSI case definition and date of SSI (onset of signsand symptoms vs date of wound culture). Five ICNs hadproblems with distinguishing deep incisional infec-tions from organ/space infections after orthopedic pro-cedures. Two ICNs were uncertain of how to interpretprolonged serosal drainage with positive microbial cul-ture. Six ICNs responded that they had also reportedSSIs found after the time limit imposed by the CDC def-inition (30 days for superficial incisional infections and1 year for deep and organ/space infections).

The rate of wound cultures was calculable for 7 hos-pitals (1 hospital did not have a separate orthopedicward). The rate in the orthopedic wards of participatinghospitals varied between 9 and 67 per 1000 patient-days, but did not correlate with SSI rates by hospitals(P 5 .38).

DISCUSSION

Our findings suggest that most SSIs reported to thenational surveillance system by participating hospitalswere true infections. Thus, when an SSI case was re-ported, the criteria of the case definition were correctly

Table 3. Types of true surgical site infections reported byroutine surveillance and by the validation team

Validation team, n (%)

Routine

surveillance

Superficial

incisional

Deep

incisional

Organ/

space

Superficial incisional 47 (96) 2 (25) 2 (10)

Deep incisional 2 (4) 5 (63) 9 (43)

Organ/space 0 (0) 1 (13) 10 (48)

All SSIs 49 (100) 8 (100) 21 (100)

interpreted. Some SSIs were missed, which might bedue to weaknesses in case finding. Variation in diag-nostic practices may also have an effect on SSI rates.

When the sensitivity, specificity, and predictivevalues of different validation studies are compared,it is important to take into account the study designand the calculation method used. The results arestraight applicable to the aggregated surveillance datawhen the study sample includes all operations duringa certain time period,9,16 but, if not, as in our study,they need further calculations. The positive predictivevalue (94.0%), sensitivity (74.8%), and specificity(99.8%) calculated for routine SIRO surveillance werefavorable compared with those reported in the few val-idation studies of national surveillance systems for SSIavailable.4,16,17 Our indicators were higher than in thefirst validation study of the NNIS system, in which pos-itive predictive value, sensitivity, and specificity forSSI were 72%, 67%, and 97.7%, respectively.17 In thisstudy, the gold standard was also the experts’ retro-spective chart review. In the Dutch national surveil-lance system, a validation study of SSI surveillanceconfirmed high accuracy for routine surveillance: pos-itive and negative predictive values were as high as 100%and 99%, respectively.4 Sensitivity and specificity forthe aggregated data were not, however, published.When our results are compared with single-hospitalvalidation studies, they are again quite similar or a littlepoorer. For example, in a validation study carried out ina single tertiary care hospital with experts’ prospectivedaily wound examination and chart reviews, accuracyfor standard SSI surveillance was slightly higher thanours.16 During the first study period, sensitivity andspecificity for routine surveillance were 83.8% and99.8%, respectively, and, during the second period,sensitivity was 92.3%. However, in our study, the sen-sitivity and specificity increased markedly (to 91.4%

Table 4. Results of the interview concerning case findingmethods in 8 hospitals participating in the validation study

Case finding method Number of hospitals (%)

Ward visits 7 (88)

Reports by medical/nursing staff 8 (100)

Method

Information technology-based reporting 7 (88)

Telephone calls 1 (13)

Person responsible for reporting

Link nurse from the ward 8 (100)

Other nurses from the ward 7 (88)

Training for case finding

Link nurses 8 (100)

Other nurses 4 (50)

SSI definitions available in the orthopedic ward 8 (100)

Microbiology reports 6 (75)

Infectious disease consultant 5 (63)

Postdischarge questionnaire 7 (88)

220 Vol. 35 No. 4 Huotari, Agthe, and Lyytikainen

and 99.8%, respectively) when the surveillance data ofthe hospital with the most missed infections wereexcluded.

In our validation sample, some underreporting byhospitals was detected. Patients with missed SSIs hadpassed through the SIRO hospitals, either through anoutpatient clinic or a ward, so that an ICN did not obtaininformation about them. Thus, as in the NNIS valida-tion study, inadequate case finding explained theunderreporting of SSIs.17 In the NNIS system, reviewof both the microbiology reports and the patient chartsis mandatory, although case finding is not outlined indetail because the most suitable sources of informationcan vary locally. In our written protocol, all methodsavailable for case finding are recommended to be used.

Overreporting of infections was not a significantproblem. When it did occur, it was related to slightvariations in interpretation of the case definition; mostfalse-positive infections were superficial incisional SSIswithout adequate clinical signs or symptoms–at leastwhat was documented in charts. In addition, 1 organ/space SSI was reported 2 years after an operation;the patient had been completely symptom free at the1-year follow-up visit. Even when the CDC definitionis correctly interpreted, it is not fully objective becauseof 2 criteria: the surgeon’s diagnosis and wound cul-ture.13,18,19 Variability in the clinician’s diagnosis and7-fold variability in wound culturing activity is likelyto affect SSI rates, as well as having an impact on the de-cisions made by the validation team. Although valida-tion studies enable the accuracy of case finding andthe surveillance process to be evaluated, SSI rates evenafter validation may be affected by differences in diag-nostic practices between hospitals and surgeons. Al-though the positive culture is only one of the CDCcriteria, SSIs without positive culture are more difficultto find.

The interpretation of the CDC definition was also ex-amined by comparing the types of SSIs detected duringroutine surveillance and by the validation team. Dis-crepancies were less common when the infectiontypes were analyzed in 2 categories instead of 3. Thismight be related to the difficulty of distinguishing be-tween infection types after orthopedic surgery. Mostsevere SSIs misclassified as superficial had begun asskin infections that had progressed. The SSI was re-ported at the beginning of the infection, but no furtherreport had been given, even when indicated. Thus, theimportance of reporting these possible prosthetic jointinfections should be emphasized.

According to the ICN interviews, an important casefinding method in routine surveillance is informationtechnology-based reporting of SSIs by link nurses insurgical wards. Most of the link nurses were trainedin case finding and definitions, and the CDC criteria

for SSIs were available in all wards. ICNs confirmedthe reported SSIs by reviewing patient charts either al-ways or when needed. Most ICNs visited surgical wardswith a frequency from once a week to once a month,which is very likely to cause variation in sensitivityof case finding. ICNs carried out further case findingby reviewing microbiology reports, cooperating withinfectious disease specialists, and following up theresults of a postdischarge questionnaire.

A limitation of this study was the small number ofcharts reviewed. An optimal sample size would havebeen 2 to 3 times larger. The retrospective study designmay also be considered a limitation, although this isoften the only method available for validation of anational surveillance system with several hospitals.9,17

Despite these methodologic limitations, the results canbe used to plan future training for ICNs and to improvethe reliability of the surveillance system and the usageof surveillance data in preventing SSIs.

The authors thank infection control nurses in participating hospitals: Irma Merio-Hietaniemi, Erja Tenhunen (Helsinki University Central Hospital), Liisa Eliin (JorviHospital), Merja Fellman (Kanta-Hame Central Hospital), Anne-Mar Kimmo (PeijasHospital), Anne Reiman (Paijat-Hame Central Hospital), Sirpa Harle (Orton Hospital),Irma Teirila (Oulu University Hospital), and Nina Elomaa (Vaasa Central Hospital).We also thank Petri Ruutu for valuable comments of the manuscript and Jukka Ollgrenand Teemu Mottonen for data management and statistical advice.

References

1. Haley RW, Culver DH, White JW, Morgan WM, Emori TG, Munn VP,

et al. The efficacy of infection surveillance and control programs in

preventing nosocomial infections in US hospitals. Am J Epidemiol

1985;121:182-205.

2. Cruse PJ, Foord R. The epidemiology of wound infection: a 10-year pro-

spective study of 62,939 wounds. Surg Clin North Am 1980;60:27-40.

3. Olson MM, Lee JT Jr. Continuous, 10-year wound infection surveil-

lance: results, advantages, and unanswered questions. Arch Surg 1990;

125:794-803.

4. Geubbels ELPE. Prevention of surgical site infections through surveil-

lance. Utrecht: Utrecht University; 2002.

5. Gastmeier P, Brauer H, Forster D, Dietz E, Daschner F, Ruden H.

A quality management project in 8 selected hospitals to reduce noso-

comial infections: a prospective, controlled study. Infect Control Hosp

Epidemiol 2002;23:91-7.

6. Archibald LK, Gaynes RP. Hospital-acquired infections in the United

States: the importance of interhospital comparisons. Infect Dis Clin

North Am 1997;11:245-55.

7. Gaynes RP, Culver DH, Horan TC, Edwards JR, Richards C, Tolson JS.

Surgical site infection (SSI) rates in the United States, 1992-1998: the

National Nosocomial Infections Surveillance System basic SSI risk

index. Clin Infect Dis 2001;33:S69-77.

8. Broderick A, Mori M, Nettleman MD, Streed SA, Wenzel RP. Nosoco-

mial infections: validation of surveillance and computer modeling to

identify patients at risk. Am J Epidemiol 1990;131:734-42.

9. Gastmeier P, Kampf G, Hauer T, Schlingmann J, Schumacher M, Dasch-

ner F, et al. Experience with two validation methods in a prevalence

survey on nosocomial infections. Infect Control Hosp Epidemiol 1998;

19:668-73.

10. Whitehouse JD, Friedman ND, Kirkland KB, Richardson WJ, Sexton

DJ. The impact of surgical-site infections following orthopedic surgery

at a community hospital and a university hospital: adverse quality of

Huotari, Agthe, and Lyytikainen May 2007 221

life, excess length of stay, and extra cost. Infect Control Hosp Epide-

miol 2002;23:183-9.

11. Darouiche RO. Treatment of infections associated with surgical im-

plants. N Engl J Med 2004;350:1422-9.

12. National Nosocomial Infections Surveillance (NNIS) System Report,

data summary from January 1992 through June 2004, issued October

2004. Am J Infect Control 2004;32:470-85.

13. Horan TC, Gaynes RP, Martone WJ, Jarvis WR, Emori TG. CDC def-

initions of nosocomial surgical site infections, 1992: a modification of

CDC definitions of surgical wound infections. Infect Control Hosp

Epidemiol 1992;13:606-8.

14. Emori TG, Culver DH, Horan TC, Jarvis WR, White JW, Olson DR,

et al. National nosocomial infections surveillance system (NNIS):

description of surveillance methods. Am J Infect Control 1991;19:

19-35.

15. National Nosocomial Infections Surveillance System CfDCaP. NNIS

manual. Atlanta, GA: US Department of Health and Human Services,

Public Health Service, CDC; 1994.

16. Cardo DM, Falk PS, Mayhall CG. Validation of surgical wound surveil-

lance. Infect Control Hosp Epidemiol 1993;14:211-5.

17. Emori TG, Edwards JR, Culver DH, Sartor C, Stroud LA, Gaunt EE,

et al. Accuracy of reporting nosocomial infections in intensive-care-

unit patients to the National Nosocomial Infections Surveillance Sys-

tem: a pilot study. Infect Control Hosp Epidemiol 1998;19:308-16.

18. Wilson AP, Gibbons C, Reeves BC, Hodgson B, Liu M, Plummer D, et al.

Surgical wound infection as a performance indicator: agreement of com-

mon definitions of wound infection in 4773 patients. BMJ 2004;329:720.

19. Taylor G, McKenzie M, Kirkland T, Wiens R. Effect of surgeon’s diag-

nosis on surgical wound infection rates. Am J Infect Control 1990;18:

295-9.

Surveillance of multidrug-resistantgram-negative bacilli in a neonatalintensive care unit: prominentrole of cross transmissionCaterina Mammina, MD,a Paola Di Carlo, MD,b Domenico Cipolla, MD,c Mario Giuffre, MD,c Alessandra Casuccio, BSc,d

Vincenzo Di Gaetano, MD,a Maria Rosa Anna Plano, MD,a Emma D’Angelo, MD,a Lucina Titone, MD,b

and Giovanni Corsello, MDc

Palermo, Italy

Background: Multidrug-resistant gram-negative bacilli (MDRGN) are an important cause of nosocomial infections in neonatalintensive care units (NICUs). We conducted a 1-year prospective surveillance study in an NICU to assess the epidemiology ofMDRGN among newborns and the relative importance of acquisition routes.Methods: Neonates admitted at the NICU of the Dipartimento Materno-Infantile, University Hospital, Palermo, Italy, from January 7,2003, to January 6, 2004, were included in the study. Colonization of patients with MDRGN was assessed by cultures of rectal swabssampled twice a week. Pulsed-field gel electrophoresis was used to determine relatedness among MDRGN isolates. Extended-spectrum b-lactamases (ESBL) and metallo-b-lactamases (MBL) production was investigated. The association between risk factorsat admission and during the NICU stay was analyzed by multivariate logistic regression analysis.Results: During the 12-month period January 7, 2003, through January 6, 2004, 1021 rectal swabs were cultured from 210 infants.One hundred sixteen infants (55.2%) were colonized by MDRGN. The monthly incidence of acquisition of MDRGN ranged between12 and 53 cases per 1000 patient-days. Eighty-four (72.4%) of the 116 patients were cross colonized. Exclusive feeding by formulawas significantly associated with cross transmission (RR 5 1.8, P 5 .02). Fifty-seven (49.1%) of the 116 infants were colonized byESBL-producing Enterobacteriaceae. Feeding by formula was significantly associated with colonization by ESBL-producing Enter-obacteriaceae (RR 5 1.6, P 5 .007), whereas breastfeeding proved to be protective (RR 5 0.5, P 5 .001). Ninety-two (43.8%) of the210 infants received antibiotics during the NICU stay, but exposure to those most frequently administered, ampicillin-sulbactamand gentamicin, was not significantly associated with MDRGN colonization.Conclusion: The emerging picture of this study is that spread of MDRGN in an NICU may be the result of diffuse cross transmissionand, consequently, of poor infection control procedures. (Am J Infect Control 2007;35:222-30.)

Nosocomial infections are an important problem inneonatal intensive care units (NICUs), in which envi-ronmental and host factors often contribute to higherrates of infection than pediatric and adult ICUs.1-3

Since the 1980s, NICUs more and more frequently pro-vide care for high-risk infants, including very-low-birth-weight (VLBW) and chronically ill infants, and theneed of prolonged stay along with large use of invasivelife support measures increases the opportunity foracquisition of late-onset nosocomial infections.3

From the Dipartimento di Igiene e Microbiologia ‘‘G. D’Alessandro’’a;Istituto di Patologia Infettiva e Virologiab; Dipartimento Materno-Infantilec; and Di. N.O.O.P.—Sez. Oftalmologia,d Universita degli Studi,Palermo, Italy.

Address correspondence to Caterina Mammina, MD, Dipartimento diIgiene e Microbiologia ‘‘D’Alessandro,’’ Universita degli Studi, Palermo,Via del Vespro 133, I-90127 Palermo, Italy. E-mail: diptigmi@unipa.it.

0196-6553/$32.00

Copyright ª 2007 by the Association for Professionals in InfectionControl and Epidemiology, Inc.

doi:10.1016/j.ajic.2006.04.210

222

Infections with gram-negative bacilli that are resis-tant to many commonly used antibacterial drugs areincreasingly reported in NICUs.4-10 The major endoge-nous reservoir of multidrug-resistant gram-negativebacilli (MDRGN) is the intestinal tract of hospitalizedinfants.11 The large use of antimicrobial drugs andcross transmission via the hands of caregivers, contam-inated equipment, or inanimate objects play a promi-nent role in selection and dissemination of MDRGN,and failures in infection control practices can be re-sponsible for diffuse horizontal spread.4-8,10 The inci-dence of infections caused by organisms resistant tothe b-lactam agents has also sharply increased inrecent years and has often been associated to clonaloutbreaks.5,8

There have been several attempts to characterize theepidemiology of nosocomial MDRGN by identifyingrisk factors for their acquisition and using moleculartyping to trace horizontal transmission between new-borns.4,6,8-10 However, in ‘‘endemic’’ contexts, dynam-ics of transmission of MDRGN are expected to bewidely variable, depending on interaction betweenhuman and environmental ecology of NICUs. Indeed,

Mammina et al May 2007 223

surveillance studies have yielded quite different find-ings: drug-resistant bacilli clonally related on the basisof molecular analysis accounted for proportions vary-ing between 12% and more than 50% of isolates fromcolonized patients.4,7-10

A prospective study was performed in the NICUof the Dipartimento Materno-Infantile, University Hos-pital, Palermo, Italy, to address the epidemiology ofMDRGN among newborns and determine by usingpulsed-field gel electrophoresis (PFGE) the frequencyof cross transmission. Risk factors for acquiringMDRGN in the NICU were also identified.

METHODS

Setting

We conducted a 1-year prospective surveillancestudy in the NICU of the University Hospital ofPalermo, Italy. The unit is part of the DipartmentoMaterno-Infantile, which is a reference center forcongenital malformation in Sicily. The departmentincludes also units of infertility and assisted reproduc-tion and materno-fetal medicine. In the NICU understudy, approximately 200 patients are usually admittedannually, including approximately 15% of VLBW new-borns (birthweight ,1500 g). The NICU is divided into 2rooms connected by sliding doors. Sinks are availablein each room. Gloves are used routinely in aseptic pro-cedures. The average patient-to-nurse ratio is 3:1 and2:1 in the intermediate care and intensive care section,respectively.

Prematurity is the most common reason for admis-sion to the NICU (approximately 70%). Other diagnosesfor admission include respiratory distress, congenitalmalformation, surgical conditions, and infections.First-line therapy for suspected early-onset sepsis is acombination of ampicillin-sulbactam plus gentamicin.

Patients

All patients admitted from January 7, 2003, to Janu-ary 6, 2004, who remained hospitalized for at least 48hours in the NICU and from whom at least 1 rectal swabculture could be obtained were enrolled in our study.Demographic, clinical, and microbiologic data wereprospectively collected and entered into an AccessDatabase (Microsoft Corp, Redmond, WA). The follow-ing parameters were recorded for all patients: demo-graphic characteristics, gestational age, birth weight,inborn or outborn condition, delivery type, APGARscore, length of stay, comorbidity conditions. Duringthe NICU stay, data regarding use and duration of thefollowing devices and procedures that could be con-sidered as possible risk factors were also collected:antimicrobial therapy, central venous and peripheral

catheterization, nasogastric and endotracheal tubeinsertion, and type of feeding (ie, parenteral, breastmilk, formula or mixed, breast milk plus formula).

A rectal swab culture yielding gram-negative bacilliin absence of clinical symptoms and signs was con-sidered evidence of colonization. Nosocomial infectionwas identified according to the recommendationsof the Centers for Disease Control and Prevention,Atlanta, GA.12 Diagnosis of infection was based on thepresence of clinical symptoms and signs, laboratoryfindings (blood C-reactive protein level, white cell for-mula), and culture results. Congenital anomalies weredefined as conditions listed under the InternationalClassification of Disease (ICD-9) codes 740 to 759.

Microbiologic surveillance

Rectal swabs were collected twice a week everyTuesday and Friday from all infants throughout theirNICU stay. These were inoculated onto 1 MacConkeyagar plate per swab to obtain a continuous lawn afterovernight incubation in ambient air at 378C. Four anti-biotic disks, containing gentamicin (10 mg), amoxi-cillin-clavulanic acid (20-10 mg), imipenem (30 mg),ceftazidime (30 mg), were placed on each plate beforeincubation, as previously described.13 Selective mediawere not used so that colonization by gram-negativebacteria susceptible to antimicrobial drugs could bedetected. After incubation, plates were examined and,colonies growing into each antibiotic inhibition halowere Gram’s stained and subcultured for purity. Next,after biochemical identification by API20E or API20NE(BioMerieux, Marcy-l’Etoile, France), susceptibility to apanel of 10 antimicrobial substances was assessed bydisk diffusion on Mueller-Hinton agar plates, accordingto the National Committee for Clinical Laboratory Stan-dards guidelines.14 The following antimicrobials weretested: ampicillin (30 mg), amoxicillin-clavulanic acid(30 mg), carbenicillin (10 mg), cefotaxime (30 mg), cefta-zidime 30 mg), ceftriaxone (30 mg), cefepime (30 mg),imipenem (30 mg), aztreonam (30 mg), and gentamicin(10 mg).

For the purpose of our study, a gram-negativeorganism resistant to at least 3 different groups of anti-microbial agents (penicillins, cephalosporins, amino-glycosides, carbapenems) was defined as MDRGN, anda gram-negative organism resistant to 2 or less groupsof antimicrobial agents was defined as susceptible.Extended-spectrum b-lactamase (ESBL) productionwas detected by decreased susceptibility or resistanceto third-generation cephalosporins and the synergybetween disks containing cefotaxime, ceftazidime,cefepime, and aztreonam and a disk containing amoxi-cillin-clavulanic acid.15 Metallo-b-lactamases (MBL) pro-duction by imipenem-resistant strains of Pseudomonas

224 Vol. 35 No. 4 Mammina et al

aeruginosa was detected by the EDTA-imipenem disksynergy test.16

Molecular methods

To evaluate the frequency of cross transmission,PFGE was performed on all available isolates associ-ated with colonization, except for multiple isolates ofthe same species obtained from the same infant thatdid not exhibit differences in $1 biochemical reactionand/or in susceptibility to $1 antimicrobial agent.

Chromosomal DNA was digested using XbaI endonu-clease and resolved by PFGE in a CHEF-Mapper appara-tus (Bio-Rad, Hercules, CA). Restriction fragment bandswere compared and analyzed using the Diversity Data-base software (Bio-Rad). Strains were interpreted asundistinguishable, closely related, possibly related, orunrelated strains according to the criteria of Tenoveret al.17 Polymerase chain reaction (PCR) amplificationof CTX-M and SHV-5 type ESBLs and VIM type MBL se-quences were performed as previously described.18,19

Statistical analysis

The association between potential risk factors andMDRGN colonization was analyzed for variables presentat admission and during the NICU stay. Time-dependentvariables, ie, length of stay, nutrition, invasive procedures,and antimicrobial use, were also measured as daysbetween admission and discharge or cumulative daysof exposure because patients were assumed to be atrisk of further colonization by a different species or strainof MDRGN after the first isolation of a similar organism.

The association was tested by the Pearson x2 testand the Fisher exact test for frequency analysis. Therelative risk (RR) and the 95% confidence intervals(CI) for the risk factors were also calculated. One-wayanalysis of variance (ANOVA) and the Mann-WhitneyU statistic test were used for parametric and nonpara-metric analysis, respectively, to evaluate differencesbetween the variables considered. A multivariate modelwas analyzed by stepwise logistic regression. All Pvalues were 2-sided, and P values less than .05 wereconsidered statistically significant. Data were analyzedby the Epi Info software (version 6.0, Centers for DiseaseControl and Prevention) and the Systat Software 8.0version (SPSS, Inc., Chicago, IL).

RESULTS

General

During the 12-month period January 7, 2003,through January 6, 2004, a total of 221 neonates wereadmitted to the NICU for at least 48 hours. Amongthese, only 11 missed rectal swab cultures and wereexcluded from further analysis.

A total of 1021 rectal swab cultures were sequen-tially obtained from 210 infants. The characteristicsof the study population are summarized in Table 1.The average length of stay was 22.3 days (median,13 days; range, 3-140). Ninety-two (43.8%) infantsreceived antibiotics during the NICU stay. Criteria forinfection were met by 25 (11.9%) patients, of whom2 died. Association between infection and potentialrisk factors at admission and during the NICU staywere evaluated. No significant differences were notedbetween infected and noninfected patients, except fora higher frequency of malformation within the infec-tion cases (20.8% vs 8.1%, P 5 .04), for the followingvariables: gestational age, gender, outborn/inborn con-dition, birth weight, twin birth, type of delivery, andAPGAR score.

Culture results

On average, patients were monitored for a mean of14.5 days (median, 7 days; range, 1-87). One hundredsixteen (55.2%) of 210 subjects were colonized withMDRGN at some point over the study period. Thirty-nine (18.6%) subjects were colonized by susceptiblebacteria, whereas 55 (26.2%) had no positive culturefor gram-negative bacilli (GN). Twenty-eight subjectswere colonized with MDRGN at the time of entry intothe study (ie, when their first rectal swab was taken atthe start of the study). Patients colonized by MDRGNwere monitored for a mean of 20.5 days (median,11 days; range, 1-87), whereas patients colonized bydrug susceptible GN were monitored for a mean of8.4 days (median, 7 days; range 1-28) and those notcolonized at any point over the study period for a meanof 4.8 days (median, 1 days; range, 1-85, P , .0005;MDRGN vs not colonized, P , .0005; MDRGN vs sus-ceptible GN, P 5 .001; not colonized vs susceptibleGN, P 5 .61)

The time interval between admission and first rectalswab for the patients colonized by MDRGN (mean, 5.7days; median, 3 days, range, 1-107) was higher but notsignificantly (P 5 .12) than that of the infants colonizedby susceptible GN (mean, 2.9 days; median, 2 days;range, 1-18) and not colonized at all (mean, 2.3 days;median, 2 days, range, 1-6). The interval between ad-mission and first rectal swab for the 28 patients colo-nized with MDRGN at the time of entry into the studywas significantly higher (mean, 10.7 days; median,4 days; range, 1-107) than that of the infants whose firstrectal samples were culture negative (mean, 4.1 days;median, 2 days; range, 1-81; P 5 .032).

Eighty-two (70.7%) colonized patients were still pos-itive when their last rectal swab was taken before thedischarge. Of the 116 colonized patients, 56 (48.3%)had 1 culture-positive rectal swab, 15 (12.9%) had 2,

Mammina et al May 2007 225

13 (11.2%) had 3, 11 (9.5%) had 4, and 20 (17.2%) had5 or more. Thirty-nine patients (33.6 %) were colonizedsimultaneously or subsequently with .1 MDRGN dur-ing follow-up. For patients not found to be colonizedat the first rectal sampling, colonization was first de-tected on average 15.7 days after admission (median,12 days; range 4-84). Fourteen (12.1%), 3 (7.7%), and8 (14.5%) infection cases were included among thepatients colonized by MDRGN, colonized by susceptibleGN, and culture negative, respectively (P 5 .35)

The monthly incidence of acquisition of MDRGNvaried over the 12-month period, with a minimum of12 cases per 1000 patient-days and 3 clusters in March,May, and September (Fig 1). This last month was asso-ciated with an overcrowding of the NICU because ofthe contemporary admission of 5 groups of pretermmultiple-birth neonates.

Risk factors for colonization with MDRGN

Selected clinical parameters at admission and dur-ing the NICU stay were tested by univariate analysisfor their association with no growth of GN, with coloni-zation with drug susceptible GN, and with at least1 strain of MDRGN (Tables 2 and 3). Statistical analysisindicated that the characteristics at admission moststrongly associated with resistant colonization were in-born status, twin birth, early gestational age, and lowbirth weight (Table 2).

Ampicillin-sulbactam and gentamicin were the mostfrequently used antimicrobials, being administered tomore than 10% of patients. The mean time of use afteradmission was 9.2 days (median, 7.5 days; range, 1-35)

Table 1. Characteristics of patients at the time ofadmission to the neonatal intensive care unit

Population No. infants (%)

Sex (male) 106 (50.5)

Birth weight, g

#500 2 (0.9)

501-1000 11 (5.2)

1001-1500 17 (8.1)

1501-2000 45 (21.4)

2001-2500 36 (17.1)

.2500 99 (47.3)

Gestational age, wk

24-29 13 (6.2)

30-36 99 (47.1)

.36 98 (46.7)

Inborn 109 (51.9)

Age at admission .24 h 16 (7.6)

Twin birth 25 (11.9)

Cesarean delivery* 153 (75.4)

APGAR score at 5 min ,5y 2 (1.1)

Malformation 20 (9.5)

*Information about delivery type was available for 203 of the 210 infants.yInformation about APGAR score was available for 189 of the 210 infants.

for ampicillin-sulbactam and 6.8 days (median, 6.0days; range 1-23) for gentamicin. However, when theiruse was assessed as a risk factor, no significant asso-ciation with MDRGN acquisition was found (Table 3).On the other hand, prolonged length of NICU stayand total exposure to antimicrobial drugs were stronglyassociated with MDRGN colonization (Table 3) In the 21infants who were treated with third-generation cepha-losporins, colonization by MDRGN (81.0%) was signifi-cantly (P 5 .04) more frequent than colonization by GN(4.8%) or negative culture for GN (14.3%).

Within the procedures used during the NICU stay,insertion of endotracheal tubes and nasogastric tubesproved to be significantly associated with MDRGN col-onization by frequency analysis, although no signifi-cant differences were detected in terms of days of thedevice use (Table 3). On the contrary, insertion of cen-tral venous catheters was significantly associated withMDRGN colonization by the number of device-days(Table 3). More days of breast milk feeding were signif-icantly associated with colonization by susceptible GN(Table 3). When these risk factors were assessed inmultivariate analysis, 3 major risk factors—early ges-tational age, low birth weight, and length of NICUstay—remained significantly associated with MDRGNcolonization (P , .005).

Colonization by multiple MDRGN species or geneti-cally discordant strains, simultaneously or in subsequentrectal swabs, was significantly associated (P , .001)with the length of stay; for the multicolonized patients,the average length was 47.0 days (median, 35.0 days;range 18-158) in comparison with that of the mono-colonized patients, 21.8 days (median, 14.0 days; range3-74). The use of central venous catheters (RR 5 1.6;

Fig 1. Incidence density of colonization by allbacterial species and the most frequently identified

bacterial clones.

226 Vol. 35 No. 4 Mammina et al

Table 2. Risk factors for colonization with MDRGN at admission

Colonized with

MDRGN n 5 116

Colonized with susceptible

GN n 5 39

Not colonized

n 5 55 P value

Gender (male)* 57 (49.1) 24 (61.5) 25 (45.5) NS

Inborn* 69 (59.5) 13 (33.3) 27 (49.1) ,.05z

Caesarean delivery* 86 (74.1) 26 (66.6) 46 (83.6) NS

Twin birth* 21 (18.1) 4 (10.3) 0 ,.01z

Malformation* 11 (9.5) 5 (12.8) 4 (7.3) NS

Gestational age (wk)y 34.2 (4.9) 36.9 (2.6) 37.1 (2.6) ,.001§

Birth weight (g)y 2135.2 (851.4) 2558.5 (757.5) 2788.5 (657.3) ,.001§

MDRGN, multidrug-resistant gram-negative bacilli; GN, gram-negative bacilli; NS, not significant.

*Expressed as yes frequency (%).yExpressed as mean (SD).zx2 test.§ANOVA test.

95% CI: 1.2-2.3; P , .001) and nasogastric tubes (RR 5

1.8; 95% CI: 1.2-2.7; P , .001) was also significantlyassociated with multiple colonization. By contrast, theuse of endotracheal tubes (RR 5 1.3; 95% CI: 0.9-1.8;P 5 .40), peripheral catheters (RR 5 1.0; 95% CI: 0.8-1.4; P 5 .36), parenteral nutrition (RR 5 1.1; 95% CI:0.8-1.4; P 5 .26), feeding by formula only (RR 5 1.2;95% CI: 0.9-1.6; P 5 .07), and administration of ampicil-lin-sulbactam (RR 5 1.2; 95% CI: 0.9-1.6; P 5 .07) orgentamicin (RR 5 1.1; 95% CI: 0.9-1.5; P 5 .16) atadmission was not associated with multiple MDRGNcolonization. When these risk factors were assessed inmultivariate analysis, early gestational age, low birthweight, and length of NICU stay were significantly asso-ciated with colonization by multiple species or strains(P , .0005).

Eighty-four (72.4%) of the 116 infants positive forMDRGN were colonized with genetically indistinguish-able organisms. These patients were not significantlydifferent from those colonized by unique strains bylength of stay (mean, 32.2 days; median, 22 days; range4-158 vs mean, 25.3 days, median, 14 days; range3-90, P 5 .25), even though a positive trend wasevident. Exclusive feeding by formula was the only indi-vidual factor significantly associated with cross trans-mission (RR 5 1.8; 95% CI: 1.1-3.5; P 5 .02). The useof central venous catheters (RR 5 0.9; 95% CI: 0.4-2.0;P 5 .36), nasogastric tubes (RR 5 1.4; 95% CI: 0.7-3.0;P 5 .17), endotracheal tubes (RR 5 0.7; 95% CI: 0.4-1.4; P 5 .20), peripheral catheters (RR 5 0.9; 95% CI:0.5-1.7; P 5 .38), parenteral nutrition (RR 5 0.5; 95%CI: 0.2-1.2; P 5 .06), and administration of ampicillin-sulbactam (RR 5 1.1; 95% CI: 0.6-2.0; P 5 .40) or genta-micin (RR 5 1.0; 95% CI: 0.6-1.8; P 5 .49) at admissionwere not associated with cross transmission.

Fifty-seven (49.1%) of the 116 MDRGN-positiveinfants were colonized by ESBL-producing Enterobac-teriaceae. These patients significantly differed fromthose who were colonized by non-ESBL-producingstrains by birth weight (mean, 1964.9 g; median, 1850

g; range, 600-3450 vs mean, 2299.7 g; median, 2340 g;range, 330-4500; P 5 .02) and length of stay (mean,38.7 days; median, 30 days; range, 5-158 vs mean,22.1 days; median, 13.0 days; range 3-74; P 5 .002).By univariate analysis, feeding by formula was theonly factor significantly associated with colonizationby ESBL-producing Enterobacteriaceae (RR 5 1.6,95% CI, 1.1-2.3, P 5 .007). On the contrary, feedingby breast milk proved to be protective (RR 5 0.5; 95%CI: 0.4-0.8; P 5 .001). Administration of gentamicin(RR 5 1.1; 95% CI: 0.8-1.6; P 5 .30) and ampicillin-sulbactam (RR 5 1.1; 95% CI: 0.7-1.5; P 5 .43) did notincrease significantly the risk of colonization by ESBL-producing strains.

Colonizing strains, PFGE results, andcharacteristics of the clusters

The species identified were as follows: Enterobactercloacae, 41 patients (35.3%); Pseudomonas aeruginosa,36 patients (31.0%); Klebsiella oxytoca, 25 patients(21.6%), Escherichia coli, 22 patients (19%); K pneu-moniae, 21 patients (18.1%); Serratia marcescens,9 patients (7.8%); and others (Citrobacter freundii,Morganella morganii, E hermanni, Acinetobacter lwoffi,Stenotrophomonas maltophilia), 15 patients (12.9%).

The 6 most frequent MDRGN species associated withneonatal colonization during the study period wereE cloacae, P aeruginosa, K oxytoca, E coli, K pneumoniae,and S marcescens (Table 4). Among the other species,strains of S maltophilia, C freundii, and A lwoffi wereisolated from 3, 2, and 2 infants, respectively, butthey were genetically unrelated. Of the 153 strainsthat were biochemically attributed to the 6 mostfrequently identified species and submitted to PFGEanalysis, 101 (66.0%) were isolated from at least 2patients. In 11 instances, the cluster included 2 or 3patients. The remaining 7 clusters involved from 4 to21 patients, the largest being caused by P aeruginosa(21 patients) and K oxytoca (21 patients) (Table 4). The

Mammina et al May 2007 227

Table 3. Risk factors for colonization with MDRGN during NICU stay

Colonized with

MDRGN n 5 116

Colonized with susceptible

GN n 5 39

Not colonized

n 5 55 P value

Endotracheal tube* 33 (28.4) 2 (5.1) 9 (16.4) ,.01z

Central venous catheter* 44 (37.9) 8 (20.5) 15 (27.3) NS

Intravenous catheter* 77 (66.4) 19 (48.7) 34 (61.8) NS

Nasogastric tube* 33 (28.4) 8 (20.5) 3 (5.5) ,.01z

Parenteral nutrition* 81 (69.8) 23 (59.0) 37 (67.3) NS

Breast milk feeding* 56 (48.3) 22 (56.4) 35 (63.6) NS

Formula only* 61 (52.6) 20 (51.3) 22 (40.0) NS

Endotracheal tube, daysy 7.6 (9.0) 2.0 (1.4) 13.2 (28.5) NS

Central venous catheter, daysy 12.1 (10.8) 6.0 (3.9) 5.2 (2.2) ,.05§

Intravenous catheter, daysy 6.2 (6.9) 5.6 (2.7) 7.7 (14.5) NS

Nasogastric tube, daysy 12.7 (12.6) 14.7 (11.7) 16.7 (13.0) NS

Parenteral nutrition, daysy 9.6 (10.2) 7.3 (11.1) 8.3 (16.7) NS

Breast milk feeding, daysy 6.6 (5.2) 10.4 (9.8) 5.9 (4.7) ,.05§

Formula only, daysy 12.1 (14.4) 15.5 (9.6) 9.4 (9.9) NS

Length of stay, daysy 29.7 (27.5) 15.9 (10.8) 11.0 (14.7) ,.001§

Total exposure to antibiotics, daysy 8.0 (13.5) 2.3 (6.2) 5.5 (13.8) ,.01§

Ampicillin-sulbactam, daysy 4.7 (7.0) 1.7 (3.3) 3.2 (4.5) NS

Gentamicin, daysy 3.2 (4.3) 1.6 (3.4) 2.3 (3.2) NS

MDRGN, multidrug-resistant gram-negative; GN, gram-negative; NS, not significant.

*Expressed as yes frequency (%).yExpressed as mean (SD).zx2 test.§ANOVA test

mean number of days that a cluster- related MDRorganism could be cultured was 60.4 days. In mostinstances, genetically indistinguishable organismswere isolated from infants whose NICU stay overlappedin time. However, in 3 clusters involving E cloacae (pul-sotype C), P aeruginosa (pulsotype A), and S marcescens(pulsotype A), indistinguishable organisms were identi-fied several weeks after the previous positive cases hadbeen discharged.

Pattern of spread

The monthly incidence of acquisition of MDRGNand the 6 most frequently identified strains over the12-month study period are illustrated in Fig 1. Thepeaks of the overall incidence density of colonizationappear to be largely attributable to the cumulativeeffect of the epidemic cross-transmission phases ofthe 6 most frequent bacterial clones. Eighteen patientsout of 28, which were colonized with MDRGN at thetime of entry into the study, proved to be colonizedby strains with a PFGE profile previously detected inthe NICU, whereas 9 were positive for unique P aerugi-nosa strains genetically unrelated to each other and tothe prevalent one (pulsotype A).

DISCUSSION

MDRGN organisms are becoming the prevalentcausal agents of infection in NICUs.20 Epidemiologicfeatures of circulation of these organisms have been

described by several authors, who have identified riskfactors for colonization and infection and appliedmolecular tracing as a tool for assessing transmissionpathways and cross-transmission burden and address-ing drug-resistance control strategies.4,6,8-10

Our study showed a colonization rate higher thanthat reported by some previous studies9; indeed,more than 50% of patients proved to be colonized byMDRGN, and approximately one fourth of these wereprecociously colonized, probably as a consequence ofa high prevalence of already colonized patients, a var-iable previously defined as ‘‘colonization pressure.’’21

Moreover, in our study, most patients were colonizedby the same strain for prolonged intervals of timeand often up to their hospital discharge. Multiplecolonization was also more frequent than previousfindings.9 Of particular interest, a very high cross-transmission rate was detected in our study by theapplication of molecular typing, with several clustersof small size including 2 to 3 patients but also largersize clusters involving up to 21 patients. Moreover,some large size clusters were traced over a long period,suggesting the probable role of a common environ-mental reservoir interacting with inadequate hygienicpractices by the health care providers.4,10,22 This find-ing seems to be in sharp contrast with those reportedby some previous studies, which suggest that shortduration and low incidence of cross transmission arethe prominent features of colonization in nonepidemicperiods in NICUs.9 Different hygienic practices and

228 Vol. 35 No. 4 Mammina et al

Table 4. Characteristics of clusters of MDRGN strains*

Species

No. of cross-colonizing

isolates/total

No. of patients/cluster

(duration of cluster, days)y

Resistance mechanism

of the predominant strains

Enterobacter cloacae 31/43 12 (293), 8 (68), 3 (26), 2 (26), 2 (1), 2 (1), 2 (1) ESBL (SHV-5)

Pseudomonas aeruginosa 25/34 21 (280), 2 (36), 2 (4) MBL (VIM-11)z

Klebsiella oxytoca 24/28 21 (121), 3 (18) ESBL (SHV-5)

Escherichia coli 19/24 15 (60), 2 (18), 2 (4) ESBL (CTX-M)

Klebsiella pneumoniae 6/17 4 (28), 2 (15)

Serratia marcescens 6/7 6 (88)

MDRGN, multidrug-resistant gram-negative.

*Only isolates available for PFGE analysis are considered.yEach number indicates the number of colonized infants in each cluster. The duration of clusters is calculated on the basis of the isolation dates.zGenBank accession number AY635904.

predominant circulation of gram-negative bacilli withan enteric ecology could have significantly influencedthe peculiar behavior of MDRGN in our study.10,11

Furthermore, overcrowding and consequent relativeunderstaffing, by amplifying the chances of crosscontamination via hands of health care workers, havelikely unleashed the incidence peak of colonizationwe observed in September. As outlined by Harbarthet al,7 understaffing, overcrowding, and poor hygienepractices are issues of major concern in NICUs andmay switch on epidemic chains of transmission, espe-cially in our age of cost containment in healthcare.10,23 However, the concurrent role in promotingepidemic cross transmission of some microorganismproperties, like the tenacious resistance to hostile envi-ronmental conditions of P aeruginosa or the attitudetoward explosive spread of some ‘‘particularly fit’’ESBL-producing strains, should not be overlooked.6,22

Furthermore, our study identified a high percentageof ESBL- and MBL-producing isolates. Noteworthy, acarbapenem-resistant strain of Pseudomonas aerugi-nosa proved to produce a new metallo-b-lactamaseidentified as VIM11. Such a finding appears alarmingbecause of the increased mortality and cost of hospital-ization associated with infections because of theseorganisms and the poor sensitivity of routine labora-tory testing in their detection.

However, in contrast with the very high frequency ofMDRGN colonization, infection occurred in only 12%of patients, accordingly with other studies.9 In our set-ting, the relatively low proportion of VLBW infants, inwhom health care-associated infections are most prob-able to occur after colonization, likely contributed tominimize the risk of infection. Infected and uninfectedinfants did not significantly differ in their characteris-tics at admission, except for more frequent malfor-mation in the infected ones. However, it is knownthat congenital malformation, especially the majorones that often require surgery and very prolongedhospital stay, may play a significant role in neonatalinfections.1,3

Both MDRGN colonization and multicolonizationproved to be associated with early gestational age,low birth weight, and length of stay, whereas the roleof invasive life support procedures as independentrisk factors was not confirmed by multivariate analysis.This is consistent with previous observations that, insome cases, have actually described a negative asso-ciation between the use of invasive equipment andacquisition of MDRGN as a result of more accuratehandling and manipulations reserved to more sus-ceptible newborns.8 Alternatively, it is probable that, inthe NICU under study, the widespread colonization am-plified by cross transmission has negatively influencedthe strength of association. Although prematurity andlow birth weight are not modifiable risk factors, it isimperative to promote among caregivers awarenessof risk of acquiring MDRGN isolates among intensivecare infants.

Newborn nutrition has been recently emphasized asa critical issue.24-27 In the analysis of risk factors, wefound that feeding by formula was consistently associ-ated with cross transmission and colonization by ESBL-producing strains, whereas breast-feeding proved tobe protective against acquisition of ESBL-producingbacilli. Milk formulas are a well-known vehicle of nos-ocomial infection in NICUs, and their contaminationmay be also responsible for cross transmission as aresult of inadequate manipulation procedures andhandwashing.24,27 On the other hand, the absolute pro-tective role of breast-feeding, because of its peculiarnutritional and immunologic properties, is confirmedby several authors.25,26

As expected, MDRGN colonization was associatedwith total exposure to antimicrobial drugs, but theuse as empiric therapy of early onset infections ofampicillin-sulbactam and gentamicin did not resultin increased colonization. Indeed, many newborns inour study acquired MDRGN colonization in an earlyphase of their NICU stay, before induction or selectionof resistance mechanisms would likely occur after an-tibiotic exposure. In other studies, previous antibiotic

Mammina et al May 2007 229

exposure has been variably associated with coloniza-tion by MDRGN.4,8,9 However, epidemiologic featuresof these organisms may be peculiar of a given unit,and, in our setting, the strength of association withantecedent antibiotic exposure might be weakenedby the extensive cross transmission.

Our study has some potential limitations. The lim-ited data on the organisms responsible for clinicalinfection cases did not allow for a reliable assessmentof the ratio of colonized to infected patients and thecause-effect relationship between risk factors andinfection. The NICU patient is a unique host ecosystembecause the intestinal microflora develops afteradmission to the unit and colonization by MDR orga-nisms from the nosocomial environment could be con-sidered unavoidable.9 However, most experts agree inbelieving that colonization is a prerequisite for infec-tion by gram-negative bacilli and that knowledge oftheir transmission routes in nonepidemic situationsmay be the key to identify the prevalent mode of trans-mission in an NICU.28 Moreover, rectal samples werenot obtained at admission and discharge, so time ofacquisition and clearing of MDRGN isolates was notthoroughly available. Examination of the formula andenvironmental flora was not performed during theperiod under study. Finally, it should be noted thatsome peculiarities of our setting, such as the highrate of caesarean section and the low frequency ofVLBW infants, make problematic a generalization ofthe results to other NICUs.

The antibiotic usage pattern in ICUs largely contrib-utes to emergence of drug resistance, especially in anNICU in which the spectrum of available antibiotics ismore restricted. Variable effects of antibiotic controlpolicies in decreasing the reservoir of resistant gram-negative bacilli have been reported by some au-thors.29,30 In our study, the large proportion of MDRGNcolonization acquired through cross transmission iden-tifies infection control instead of antibiotic control asthe key issue and urges the identification and applic-ation of targeted infection control policies. Assumingthat MDRGN are transmitted to the infants mainly viathe hands of the caregivers, the importance of limitingovercrowding-understaffing and implementing strin-gent hand cleansing procedures must be emphasized,along with the adoption of compliance-enhancingstrategies and measuring tools.7,23,31 The frequentassociation between ESBL/MBL production and epi-demic clones demands that adequate measures beimplemented to control spread of these resistant orga-nisms in NICUs. At present, adoption of targeted infec-tion control strategies and antibiotic stewardship arethe approaches most supported by sound scientificevidence to discourage the emergence and diffusionof multidrug-resistant organisms.30,31 A greater

understanding of the epidemiology of nosocomial or-ganisms and the peculiar features of health care unitsmay greatly contribute to contain the reservoir ofMDRGN organisms.

References

1. Gaynes RP, Edwards JR, Jarvis WR, Culver DH, Tolson JS, Martone WJ.

Nosocomial infections among neonates in high-risk nurseries in the

United States. Pediatrics 1996;98:357-61.

2. Stoll BJ, Hansen N, Fanaroff AA, Wright LL, Carlo WA, Ehrenkranz

RA, et al. Late-onset sepsis in very low birth weight neonates: the ex-

perience of the NICHD Neonatal Research Network. Pediatrics 2002;

110:285-91.

3. Baltimore RS. Neonatal sepsis: epidemiology and management. Pae-

diatr Drugs 2003;5:723-40.

4. Almuneef MA, Baltimore RS, Farrel PA, Reagan-Cirincione P, Dembry

LM. Molecular typing demonstrating transmission of gram-negative

rods in a neonatal intensive care unit in the absence of a recognized

epidemic. Clin Infect Dis 2001;32:220-7.

5. Bisson G, Fishman NO, Patel JB, Edelstein PH, Lautenbach E.

Extended-spectrum b-lactamase-producing Escherichia coli and Kleb-

siella species: risk factors for colonization and impact of antimicrobial

formulary interventions on colonization prevalence. Infect Control

Hosp Epidemiol 2002;23:254-60.

6. D’Agata EM, Venkataraman L, DeGirolami P, Samore M. Molecular

epidemiology of ceftazidime-resistant gram-negative bacilli on inani-

mate surfaces and their role in cross-transmission during nonoutbreak

periods. J Clin Microbiol 1999;37:3065-7.

7. Harbarth S, Sudre P, Dharan S, Cadenas M, Pittet D. Outbreak of

Enterobacter cloacae related to understaffing, overcrowding, and poor

hygiene practices. Infect Control Hosp Epidemiol 1999;20:598-603.

8. Pessoa-Silva CL, Meurer Moreira B, Camara Almeida V, Flannery B,

Almeida Lins MC, Mello Sampaio JL, et al. Extended-spectrum

b-lactamase-producing Klebsiella pneumoniae in a neonatal intensive

care unit: risk factors for infection and colonization. J Hosp Infect

2003;53:198-206.

9. Toltzis P, Dul MJ, Hoyen C, Salvator A, Walsh M, Zetts L, et al.

Molecular epidemiology of antibiotic resistant gram-negative bacilli

in a neonatal intensive care unit during a nonoutbreak period.

Pediatrics 2001;108:1143-8.

10. Waters V, Larson E, Wu F, San Gabriel P, Haas J, Cimiotti J, et al.

Molecular epidemiology of gram-negative bacilli from infected neo-

nates and health care workers’ hands in neonatal intensive care

units. Clin Infect Dis 2004;38:1682-7.

11. Donskey CJ. The role of the intestinal tract as a reservoir and source

for transmission of nosocomial pathogens. Clin Infect Dis 2004;39:

219-26.

12. Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. CDC defini-

tions for nosocomial infections, 1988. Am J Infect Control 1988;16:

128-40.

13. Lee PY, Holliman RE, Davies EG. Surveillance cultures on neonatal

intensive care units. J Hosp Infect 1995;29:233-6.

14. NCCLS. Performance standards for antimicrobial susceptibility test-

ing, 12th informational supplement. Document M100-S12. Wayne,

PA: NCCLS; 2003.

15. MacKenzie FM, Miller CA, Gould IM. Comparison of screening

methods for TEM- and SHV-derived extended-spectrum b-lactamase

detection. Clin Microbiol Infect 2002;8:715-24.

16. Lee K, Chong Y, Shin HB, Kim YA, Yong D, Yum JH. Modified Hodge

and EDTA-disk sinergy tests to screen metallo-b-lactamase-producing

strains of Pseudomonas and Acinetobacter strains. Clin Microbiol Infect

2001;7:88-91.

17. Tenover FC, Arbeit RD, Goering RV, Mickelsen PA, Murray BE, Persing

DH, et al. Interpreting chromosomal DNA restriction patterns

230 Vol. 35 No. 4 Mammina et al

produced by pulsed-field gel electrophoresis: criteria for bacterial

strain typing. J Clin Microbiol 1995;33:2233-9.

18. Tzelepi E, Giakkoupi P, Sofianou D, Loukova V, Kemeroglou A, Tsakris

A. Detection of extended-spectrum b-lactamases in clinical isolates of

Enterobacter cloacae and Enterobacter aerogenes. J Clin Microbiol 2000;

38:542-6.

19. Walsh TR, Toleman MA, Hryniewicz W, Bennett PM, Jones RN.

Evolution of an integron carrying blaVIM-2 in Eastern Europe: report

from the SENTRY Antimicrobial Surveillance Program. J Antimicrob

Chemother 2003;52:116-9.

20. Nambiar S, Singh N. Change in epidemiology of health care-associated

infections in a neonatal intensive care unit. Pediatr Infect Dis J 2002;21:

839-42.

21. Bonten MJ, Slaughter S, Ambergen AW, Hayden MK, van Voorhis J,

Nathan C, et al. The role of ‘‘colonization pressure’’ in the spread

of vancomycin-resistant enterococci: an important infection control

variable. Arch Intern Med 1998;158:1127-32.

22. Grundmann H, Kropec A, Hartung D, Berner R, Daschner F. Pseu-

domonas aeruginosa in a neonatal intensive care unit: reservoirs

and ecology of the nosocomial pathogen. J Infect Dis 1993;168:

943-7.

23. Singh N. Large infection problems in small patients merit a renewed

emphasis on prevention. Infect Control Hosp Epidemiol 2004;25:

714-6.

24. Drudy D, Mullane NR, Quinn T, Wall PG, Fanning S. Enterobacter saka-

zakii: an emerging pathogen in powdered infant formula. Clin Infect

Dis 2006;42:996-1002.

25. Hylander MA, Strobino DM, Dhanireddy R. Human milk feeding and in-

fection among very-low-birth-weight infants. Pediatrics 1998;102:e38.

26. de Silva A, Jones PW, Spencer SA. Does human milk reduce infection

rates in preterm infants? A systematic review. Arch Dis Child Fetal

Neonatal Ed 2004;89:F509-13.

27. Fleisch F, Zimmermann-Baer U, Zbinden R, Bischoff G, Arlettaz R,

Waldvogel K, et al. Three consecutive outbreaks of Serratia marcescens

in a neonatal intensive care unit. Clin Infect Dis 2002;34:767-73.

28. Singh N, Patel KM, Leger MM, Short B, Sprague BM, Kalu N, et al.

Risk of resistant infections with Enterobacteriaceae in hospitalized

neonates. Pediatr Infect Dis J 2002;21:1029-33.

29. de Man P, Verhoeven BA, Verbrugh HA, Vos MC, van den Anker JN.

An antibiotic policy to prevent emergence of resistant bacilli. Lancet

2000;355:973-8.

30. Toltzis P, Dul MJ, Hoyen C, Salvator A, Walsh M, Zetts L, et al. The

effect of antibiotic rotation on colonization with antibiotic-resistant

bacilli in a neonatal intensive care unit. Pediatrics 2002;110:707-11.

31. Won SP, Chou HC, Hsieh WS, Chen CY, Huang SM, Tsou KI, et al.

Handwashing program for the prevention of nosocomial infections

in a neonatal intensive care unit. Infect Control Hosp Epidemiol

2004;25:742-6.

Antimicriobial resistance patterns ofcolonizing flora on nurses’ hands inthe neonatal intensive care unitHeather A. Cook, MPH,a Jeannie P. Cimiotti, DNS, RN,b Phyllis Della-Latta, PhD,c Lisa Saiman, MD, MPH,d

and Elaine L. Larson, RN, PhDa,e

New York, New York, and Philadelphia, Pennsylvania

Background: The Centers for Disease Control and Prevention recommends the use of an alcohol-based handrub for health careworker hand hygiene. The purpose of this study was to examine effects of hand hygiene product and skin condition on theantimicrobial resistance patterns of colonizing hand flora among nurses.Methods: Colonizing hand flora of 119 nurses working in 2 neonatal intensive care units was compared during a 22-month cross-over study using alcohol handrub or antiseptic soap.Results: Altogether, 1442 isolates from 834 hand cultures (mean, 7 cultures/nurse) were obtained. In 3 of 9 regression analysesmodeling for resistant staphylococcal flora, the use of antiseptic soap was a significant predictor of resistance, and nurses withdamaged skin were 2.79 times more likely to carry Staphylococcus warneri isolates resistant to gentamicin.Conclusion: Hand hygiene product and skin condition may influence resistance patterns of hand flora of care providers.(Am J Infect Control 2007;35:231-6.)

The hands of health care workers may serve as res-ervoirs for organisms causing health care-associatedinfections, including infections caused by multidrug-resistant strains.1,2 In 2002, the Centers for DiseaseControl and Prevention revised the recommendationsfor hand hygiene to include the use of alcohol-basedproducts for standard hand hygiene.3 However, thepossible effects of hand hygiene products on the skinflora of personnel hands, including antibiotic resis-tance patterns, have not been extensively studied. Werecently reported that the hands of a small number ofnew graduate nurses initially harbored methicillin-sus-ceptible strains of staphylococci, which were replaced

From the Department of Epidemiology,a Mailman School of PublicHealth, Columbia University; School of Nursing,b University of Pennsyl-vania, Philadelphia, Pennsylvania; Department of Pathology,c ColumbiaUniversity; Department of Pediatrics,d College of Physicians and Sur-geons, Columbia University; and School of Nursing,e Columbia Univer-sity, New York, New York.

Address correspondence to Elaine L. Larson, RN, PhD, ColumbiaUniversity School of Nursing, 630 W. 168th St, New York, NY 10032.E-mail: ELL23@columbia.edu.

Supported by National Institutes of Health grant, ‘‘Effect of Staff HandHygiene on Infections in Neonates,’’ 5 R01NR05197; National Instituteof Nursing Research, Bethesda, MD; and by the Center for Interdiscipli-nary Research on Antimicrobial Resistance, P20 RR020616, NationalCenter for Research Resources.

There are no conflicts of interest for any author.

0196-6553/$32.00

Copyright ª 2007 by the Association for Professionals in InfectionControl and Epidemiology, Inc.

doi:10.1016/j.ajic.2006.05.291

by methicillin-resistant strains within a few monthsof employment in the 2 study units.4 The purposes ofthis study were to describe the types and antimicrobialresistance patterns of hand flora among nurses work-ing in 2 neonatal intensive care units (NICUs) over a22-month period and to examine risk factors, includingthe type of hand hygiene product, for antimicrobialresistance in the hand flora.

METHODS

Sample and setting

A substudy was conducted of a larger clinical trialin which a crossover design was used to evaluate theimpact of 2 hand hygiene products,5 a detergent-basedantiseptic containing 2% chlorhexidine gluconate(Bactoshield; Steris Corporation, St. Louis, MO) and a61% alcohol-based handrub (Avagard; 3M HealthCare,St. Paul, MN), on health care-associated infections in ne-onates. The study was conducted in 2 Manhattan NICUsbetween March 2001 and January 2003. The 2 NICUsare part of the New York Presbyterian Hospital system:a 43-bed unit (NICU 1) and a 50-bed unit (NICU 2).

All full-time nurses working in these units were askedto volunteer for the study; 119 of 155 nurses (76.8%)agreed to participate. Nurses were selected for this studybecause they were the primary staff members perma-nently assigned to the study units and they had themost frequent contact with the neonates.6 Throughoutthe study, all staff on each unit used the same handhygiene product, including a moisturizer provided bythe hospital, and artificial fingernails were prohibited.Nonstoned rings were allowed. The study was approved

231

232 Vol. 35 No. 4 Cook et al

by the institutional review boards of both study sites,and each of the participating nurses signed a consentform. During the 22 study months, the condition ofeach nurse’s hands was assessed monthly, and a handculture was obtained quarterly, as described below.

Skin assessment

Nurses’ skin condition was recorded monthly using 2assessment tools. First, trained research personnel ratedthe skin on the hands of the participating nurses under33 magnification on a scale of 0 to 5, with 0 being exten-sive cracking of the skin and widespread redness and 5being normal skin with no observable irritation. Previousstudies have confirmed that the ratings generated fromthis observer assessment tool were consistent with der-matologist ratings of skin condition and had an interraterreliability of $90% over a range of skin conditions.7-9

The second assessment tool was a self-rated scalewith which participating nurses scored the skin ontheir hands using 4 criteria: appearance, intactness,moisture content, and sensation. The scale rangedfrom 4 to 28, with healthier skin having a higher score.Previous studies have demonstrated that this self-ratingtool was significantly correlated with other measuresof skin condition.10-13 The observer-rated and self-rated skin condition scores were collected indepen-dently of one another.

Hand culturing and species identification

Hand cultures were obtained every 3 months using amodified glove-juice technique.14 Nurses cleaned theirhands with the available product immediately prior tosampling so that we could identify resident rather thantransient flora. Their dominant hand was inserted intoa sterile polyethylene bag containing 50 mL sampling so-lution (0.075 mol/L phosphate buffer, pH 7.9, containing0.1% polysorbate 80 and 0.1% sodium thiosulfate), andthe hand was massaged through the bag for 1 minute.The Clinical Microbiology Service of New York-Presbyte-rian Hospital, Columbia Campus, performed all micro-biologic testing. Undiluted, 10- and 100-fold dilutedaliquots of sampling solution were plated onto 5% sheepblood agar plates (BBL; Becton Dickinson MicrobiologySystems, Cockeysville, MD). Organisms were identifiedby the MicroScan Walk Away 96 system (Dade BehringInc, Newark, DE). Using the clinical laboratory’s standardprotocols, a colony of each morphologically distinctcolony type was selected for identification.

Antimicrobial susceptibility patterns

Antimicrobial susceptibilities were determined bythe MicroScan Walk Away 96 system (Dade BehringInc). In accordance with the Clinical and LaboratoryStandards Institute,15 the isolates were dichotomized

as either susceptible or nonsusceptible based onminimum inhibitory concentration values, with iso-lates determined to be ‘‘intermediate’’ or ‘‘resistant’’categorized as nonsusceptible. Further analyses wereconducted to examine changes in the staphylococcalflora (Staphylococcus warneri, Staphylococcus epidermi-dis, Staphylococcus aureus) of each nurse during thestudy period. For each nurse, we assessed changes insusceptibility of their S epidermidis isolates to oxacillin,rifampin, and gentamicin at each sampling period. Wedefined a potentially clinically relevant change in resis-tance as a change from susceptible to resistant or fromresistant to susceptible to 2 or more of these agents.

Data analysis

Initially, x2 tests were used to examine differencesin the relative proportions of the type of organism (ie,gram negative, gram positive, or fungi) by study unit(NICU 1 or 2) and hand hygiene product in use at thetime of the culture (antiseptic soap or alcohol-basedrub) as well as the difference in the proportion of iso-lates susceptible or nonsusceptible to antibiotics dur-ing the 2 hand hygiene product periods. Next, logisticregression models were used to examine potential pre-dictors of carrying S warneri, S epidermidis, or S aureusisolates resistant to oxacillin, rifampin, and gentami-cin. Models to predict vancomycin resistance werenot included because there were too few isolates resis-tant to vancomycin. Separate regression models werefit for each species and antibiotic combination. Ineach model, the dependent variable was the presenceor absence of resistance, and predictor variablesincluded hand product used (alcohol-based rub or anti-septic soap), number of years worked in the NICU, site(NICU 1 or 2), and observer-rated skin condition. Onlyobserver-rated skin condition was used in the modelsbecause observer and self-rating assessments of skincondition were significantly correlated (P , .01). Astepwise procedure with a variable entering the modelat P 5 .1 and retained in the model at P 5 .05 was used(SAS software; SAS Institute, Cary, NC).

RESULTS

There were 834 hand cultures available for this anal-ysis: 417 during the alcohol-based handrub period and417 during the antiseptic soap period. An average of7.0 (range, 1-8) cultures were obtained from each nurseparticipant (7.1/nurse in NICU 1 and 6.4/nurse in NICU2). Only 3 of the 119 nurses were male, and the averageage of participating nurses was 41.1 years.5

Types of flora

During the study period, 1442 isolates were collectedfrom the hands of 119 nurses (Table 1). Most were

Cook et al May 2007 233

gram-positive bacteria (88.8%, 1281/1442), 6.6% weregram-negative bacteria (95/1442), and 4.6% were fungi(66/1442). Of all the isolates, 85.7% (1236/1442) werestaphylococci, with S epidermidis and S warneri account-ing for 36.3% (524/1442) and 36.1% (520/1442) of iso-lates, respectively. Because there were no differences inthe types of organisms found on the hands of nursesfrom the 2 units (P 5 .13), the data from the 2 siteswere combined for the subsequent multivariate analyses.

Antibiotic susceptibility patterns

The antibiotic susceptibility profiles of staphylococ-cal species for which $10 isolates were identified dur-ing the study period are shown in Table 2. In general,S epidermidis and S warneri isolates were not suscep-tible to b-lactam agents, including oxacillin, but weresusceptible to quinolone agents.

During the 22-month study period, there were 81.5%(97/119) of nurses from whom S epidermidis was iso-lated more than once, and 70.6% (84/119) of nurseshad S epidermidis isolated from at least half of the cul-tures obtained. The antimicrobial susceptibility profilesof S epidermidis remained the same in 69.2% (63/91) ofnurses from whom more than 1 isolate was identified;in 16.5% (15/91), there was 1 change, and, in 14.3%(13/91), there were 2 or more changes in the antibioticresistance profile over time. However, changes in theresistance patterns of S epidermidis isolates from these13 nurses had no distinguishable pattern with time.

Hand product, skin condition,and antibiotic resistance

We explored 9 regression models to identify predic-tors of resistance to oxacillin, rifampin, or gentamicinin S warneri, S epidermidis, and S aureus. In 6 of thesemodels, none of the variables assessed were significantpredictors of resistance. In 3 models, the hand productwas the only significant predictor of antibiotic resis-tance; during the period when antiseptic soap wasused, there was a significant increase in S epidermidisisolates resistant to oxacillin (relative risk, 1.92; 95%confidence intervals (CI): 1.08-3.43) and gentamicin(relative risk, 1.50; 95% CI: 1.00-2.27) and a 7.22 timesincreased risk of rifampin resistance among S warneriisolates (95% CI: 2.97-17.56). Additionally, the relativerisk of resistance to gentamicin among S warneri iso-lates from nurses with damaged skin compared withnurses with healthy skin was 2.79 (95% CI: 1.35-5.81;Table 3).

DISCUSSION

To our knowledge, this is the largest longitudinalstudy of nurses’ hand flora conducted to date. New

and clinically relevant information was identified re-garding the types and antimicrobial resistance patternsof nurses’ hand flora, the relationship between handhygiene product and flora, and the potential role ofskin health on antimicrobial resistance of the coloniz-ing flora.

Types and antimicrobial susceptibilitiesof nurses’ hand flora

As expected, S epidermidis was the most commonorganism isolated from nurses’ hands. In this sam-ple, however, there was also a higher prevalence of

Table 1. Gram positive, gram negative, and fungal speciesisolated from NICU nurses’ hands*

Isolates Nurses

Species n % n %y

Gram-positive organisms

Enterococcus faecalis 20 1.4 18 15.1

Enterococcus faecium 4 0.3 4 3.4

Micrococcus species 10 0.7 9 7.6

Staphylococcus aureus 41 2.8 26 21.8

Staphylococcus auricularis 15 1.0 13 10.9

Staphylococcus capitis 14 0.9 12 10.1

Staphylococcus capitis-ureolyticus 35 2.4 24 20.2

Staphylococcus epidermidis 524 36.3 118 99.2

Staphylococcus haemolyticus 27 1.9 20 16.8

Staphylococcus hominis 27 1.9 22 18.5

Staphylococcus simulans 16 1.1 13 10.9

Staphylococcus warneri 520 36.1 116 97.5

Staphylococcus species (other) 17 1.2 15 12.6

Streptococcus species 3 0.2 3 2.5

Other gram-positive organisms 8 0.6 8 6.7

Subtotal 1281 88.8 421

Gram-negative organisms

Acinetobacter baumanii 8 0.6 5 4.2

Acinetobacter lwoffi 17 1.2 16 13.4

Enterobacter cloacae 10 0.7 9 7.6

Enterobacter species (other) 10 0.7 10 8.4

Escherichia coli 3 0.2 3 2.5

Klebsiella oxytoca 10 0.7 9 7.6

Klebsiella pneumoniae 11 0.8 9 7.6

Pseudomonas aeruginosa 2 0.1 2 1.7

Psuedomonas species (other) 7 0.5 7 5.9

Serratia marcescens 4 0.3 4 3.4

Other gram-negative organisms 13 0.9 11 9.2

Subtotal 95 6.6 85

Other

Candida albicans 8 0.6 5 4.2

Candida parapsilosis 41 2.8 27 22.7

Candida species (other) 12 0.8 11 9.2

Other 5 0.3 5 4.2

Subtotal 66 4.6 48

Total 1442 100.0 554

*N 5 119 nurses.ySome nurses carried .1 isolate of a given species, but were only counted once/

species. Thispercentage is the percentage of nurses from whom this organism was

isolated at least once.

234 Vol. 35 No. 4 Cook et al

Table 2. Proportion of staphylococcal isolates susceptible to selected antimicrobial agents

Percentage (number of isolates)

Total No. isolates Oxacillin Rifampin Gentamicin Levofloxacin

Staphylococcus aureus 41 85.4 (35) 92.7 (38) 85.4 (35) 92.7 (38)

Staphylococcus auricularis 15 53.3 (8) 100 (15) 86.7 (13) 80.0 (12)

Staphylococcus capitis-capitis 14 92.9 (13) 92.9 (13) 85.7 (12) 92.9 (13)

Staphylococcus capitis-ureolyticus 35 82.9 (29) 97.1 (34) 82.9 (29) 94.3 (33)

Staphylococcus epidermidis 524 12 (63) 96.6 (506) 27.5 (144) 93.3 (489)

Staphylococcus haemolyticus 27 19.2 (5) 92.6 (25) 55.6 (15) 66.7 (18)

Staphylococcus hominis 27 48.1 (13) 100 (27) 66.7 (18) 96.2 (25)

Staphylococcus simulans 16 60.0 (9) 100 (16) 75.0 (12) 93.8 (15)

Staphylococcus warneri 520 14.6 (76) 88.6 (459) 21.4 (111) 93.8 (486)

Vancomycin not included in table because most isolates were susceptible.

Table 3. Predictors of resistance in staphylococcal hand flora of nurses, as analyzed by logistic regression*

Organism

Antibiotic

(No. isolates tested) Significant variables

Relative risk

(95% CI)

Staphylococcus epidermidis Oxacillin (n 5 478) Hand product

Alcohol Referent

Antiseptic soap 1.92 (1.08-3.43)

Gentamicin (n 5 472) Hand product

Alcohol Referent

Antiseptic soap 1.50 (1.00-2.27)

Staphylococcus warneri Gentamicin (n 5 447) Skin Condition

Healthy Referent

Damaged 2.79 (1.35-5.81)

Rifampin (n 5 463) Hand product

Alcohol Referent

Antiseptic soap 7.22 (2.97-17.56)

*Predictor variables included in each model: hand product, skin condition, years worked on unit, and site (NICU 1 or 2).

S warneri than has been noted in other studies,9,16,17

despite the fact that it was not known to be a prevalentorganism in the study hospital. This finding is impor-tant for 2 reasons. First, it may indicate that the colo-nizing flora of health care personnel can becomemodified over time to reflect individual differences bypractice site. This was further confirmed by our findingthat, although these nurses often retained the samestrain of S epidermidis during the study period, theyalso shared a single clone among other each otherand among their patients, as previously reported,18

evidence that this clone was associated with this spe-cific critical care environment.

Our group and others have recently reporteddecreased susceptibility to vancomycin among neo-nates.19,20 Although S warneri is not known to be ofmajor clinical significance, cases of health care-associ-ated infection with this organism have been reportedamong neonates.21-24 Hence, this ‘‘normal flora’’ maybe emerging as yet another multidrug-resistant oppor-tunistic pathogen among high-risk, low-birth-weightneonates.

The majority of the S epidermidis and S warneriisolates from the nurses’ hands were resistant toantibiotics commonly used on the NICU. In previousstudies conducted more than a decade ago, rates ofoxacillin resistance among coagulase-negative staphy-lococci from nurses’ hands ranged from 26% to60%.16,23 In our study, 79% of these isolates wereresistant to oxacillin. Patient care personnel havebeen found to be colonized with higher rates of antimi-crobial-resistant S aureus isolates than nonmedicalpersonnel.25 In contrast with the high rates of oxacillinresistance noted in coagulase-negative staphylococci,generally low rates of methicillin-resistant S aureushave been isolated from nurses’ hands in previousstudies: 1 of 129 (0.8%)22 and 3 of 79 (3.8%) nurses.25

These rates are slightly lower than the rate noted in ourstudy (5/119, 4.2%), although in 1 survey of severalhospitals in which MRSA was endemic, the majorityof S aureus isolates from staff were methicillin resis-tant.26 Clearly, personnel hands are a potential reser-voir of resistance that can be transmitted to patientsin high-risk environments.

Cook et al May 2007 235

Factors associated with antimicrobial resistance

We found in this study that antimicrobial resistancewas significantly more common for several antibioticswhen nurses were using the chlorhexidine-containingsoap or when the skin on their hands was damaged.In a 6-year study, Gordin et al27 reported a decreasein the number of health care-associated methicillin-resistant S aureus and vancomycin-resistant enterococ-cus acquired by patients when an alcohol productwas used for staff hand hygiene, but we did not exam-ine the rates of clinical infections caused by antibiotic-resistant organisms among neonates in our study.

Although others have reported a higher prevalenceof certain potential pathogens such as S aureus andgram-negative bacteria among nurses with damagedskin,9 this may be the first report of a possible relation-ship between hand hygiene product, skin condition,and antimicrobial resistance. Irritant contact dermatitisand other skin problems are prevalent among healthcare providers, and the detergent base in soaps is oneof the primary causes.7,28

To our knowledge, there has been no biologic mech-anism identified that could explain our observed as-sociation between chlorhexidine use and antibioticresistance. It is possible that the link may be an inter-mediary one, ie, that the soap is associated with moreskin damage and that it is the damaged skin that facil-itates the colonization of hands with hospital-associ-ated resistant flora. We have previously reported, infact, that skin condition was significantly better whenthe alcohol product as compared with the antisepticsoap was used among these nurses.5 An alternativeexplanation is that microorganisms may develop resis-tance mechanisms to the chlorhexidine itself, as hasbeen described with triclosan, a bisphenol ingredientused in some antiseptic products.29 The potential asso-ciation between the antiseptic soap and increasedantibiotic resistance clearly needs further study. In themeantime, the Centers for Disease Control and Preven-tion’s recommendations3 regarding use of alcoholhand hygiene products and attention to skin healthamong staff hands may be important for reducing thepossibility that the hands of health care personnelcould become reservoirs of antimicrobial resistance.

This study had several limitations that must be con-sidered when assessing the relevance of our findings.First, we only obtained cultures every 3 months anddid not sample every nurse participant at every sam-pling interval, although the average number of sampleswas 7/nurse. Clearly, this represents only a snapshot ofchanges in skin flora that might occur over time. Sec-ond, only a single colony-forming unit of each mor-phologic type was speciated from each hand sample.Hence, it is possible that several strains or different

clones of any organism, in particular the coagulase-negative staphylococci, were present but not detected.Furthermore, we were unable to study multidrug-resis-tant, gram-negative bacteria on nurses’ hands becausethere were too few isolates. Finally, there may havebeen unmeasured factors that changed during theperiods when the 2 hand hygiene products were usedthat could have affected resistance patterns.

CONCLUSION

In summary, it appears that the normal skin flora ofnurses has become increasingly resistant when com-pared with earlier studies. Furthermore, in this studyhand hygiene product and skin condition were associ-ated with antibiotic resistance in the staphylococcalflora; for some antibiotics, an increased risk of resis-tance was found when nurses used an antiseptic soaprather than an alcohol product and when their skinwas damaged. The potential impact of various hand hy-giene products on the numbers, types, and resistancepatterns of staff hand flora warrants further research.

References

1. Aiello AE, Cimiotti J, Della-Latta P, Larson EL. A comparison of the

bacteria found on the hands of ‘‘homemakers’’ and neonatal intensive

care unit nurses. J Hosp Infect 2003;54:310-5.

2. Larson E. Skin hygiene and infection prevention: more of the same or

different approaches? Clin Infect Dis 1999;29:1287-94.

3. Boyce JM, Pittet D. Healthcare Infection Control Practices Advisory

Committee; HICPAC/SHEA/APIC/IDSA Hand Hygiene Task Force.

Guideline for hand hygiene in health-care settings. Recommendations

of the Healthcare Infection Control Practices Advisory Committee

and the HIPAC/SHEA/APIC/IDSA Hand Hygiene Task Force. Am J

Infect Control 2002;30:S1-46.

4. Cimiotti JP, Wu F, Della-Latta P, Nesin M, Larson E. Emergence of

resistant staphylococci on the hands of new graduate nurses. Infect

Control Hosp Epidemiol 2004;25:431-5.

5. Larson E, Cimiotti J, Haas J, Parides M, Nesin M, Della-Latta P, et al.

Effect of antiseptic handwashing vs alcohol sanitizer on health-care

associated infections in neonatal intensive care units. Arch Pediatr

Adoles Med 2005;159:377-83.

6. Cohen B, Saiman L, Cimiotti J, Larson E. Factors associated with hand

hygiene practices in two neonatal intensive care units. Pediatr Infect

Dis J 2003;22:494-9.

7. Larson E, Friedman C, Cohran J, Treston-Aurand J, Green S. Preva-

lence and correlates of skin damage on the hands of nurses. Heart

Lung 1997;26:404-12.

8. Larson E, Silberger M, Jakob K, Whittier S, Lai L, Della-Latta P, et al.

Assessment of alternative hand hygiene regimens to improve skin

health among neonatal intensive care unit nurses. Heart Lung 2000;

29:136-42.

9. Larson EL, Hughes CA, Pyrek JD, Sparks SM, Cagatay EU, Bartkus JM.

Changes in bacterial flora associated with skin damage on hands of

health care personnel. Am J Infect Control 1998;26:513-21.

10. Grove GL, Zerweck CR, Heilman JM, Pyrek JD. Methods for evaluat-

ing changes in skin condition due to the effects of antimicrobial hand

cleansers: two studies comparing a new waterless chlorhexidine

gluconate/ethanol-emollient antiseptic preparation with a conventional

water-applied product. Am J Infect Control 2001;29:361-9.

236 Vol. 35 No. 4 Cook et al

11. Larson E, Leyden JJ, McKinley KJ, Grove GL, Talbot GH. Physiologic

and microbiologic changes in skin related to frequent handwashing.

Infect Control 1986;7:59-63.

12. Larson E, McKinley KJ, Grove GL, Leyden JJ, Talbot GH. Physiologic,

microbiologic, and seasonal effects of handwashing on the skin of

health care personnel. Am J Infect Control 1986;14:51-9.

13. Simion FA, Rhein LD, Morrison BM, Scala DD, Salko DM, Kligman AM,

et al. Self-perceived sensory responses to soap and synthetic deter-

gent bars correlate with clinical signs of irritation. J Am Acad Derma-

tol 1995;32:205-11.

14. American Society for Testing and Materials. Standard test method

for evaluation of healthcare personnel handwash formulation. West

Conshohocken, PA: American Society Testing Materials; 1994.

15. National Committee for Clinical Laboratory Standards. Performance

standards for antimicrobial susceptibility testing. Wayne, PA: National

Committee for Clinical Laboratory Standards; 2002.

16. Horn WA, Larson EL, McGinley KJ, Leyden JJ. Microbial flora on the

hands of health care personnel: differences in composition and anti-

bacterial resistance. Infect Control Hosp Epidemiol 1988;9:189-93.

17. Kloos WE, Musselwhite MS. Distribution and persistence of Staphylo-

coccus and Micrococcus species and other aerobic bacteria on human

skin. Appl Microbiol 1975;30:381-5.

18. Milisavljevic V, Wu F, Cimiotti J, Haas J, San Gabriel P, Della-Latta P,

et al. Genetic relatedness of Staphylococcus epidermidis from infected

infants and staff in the neonatal intensive care unit. Am J Infect Control

2005;33:341-7.

19. Cimiotti JP, Haas JP, Della-Latta P, Wu F, Saiman L, Larson EL. Preva-

lence and clinical relevance of Staphylococcus warneri in the neonatal

ICU. Infect Control Hosp Epidemiol 2007;38:326-30.

20. Center KJ, Reboli AC, Hubler R, Rodgers GL, Long SS. Decreased

vancomycin susceptibility of coagulase-negative staphylococci in a ne-

onatal intensive care unit: evidence of spread of Staphylococcus warneri.

J Clin Microbiol 2003;41:4660-5.

21. Kacica MA, Horgan MJ, Preston KE, Lepow M, Venezia RA. Relat-

edness of coagulase-negative staphylococci causing bacteremia in

low-birth-weight infants. Infect Control Hosp Epidemiol 1994;15:

658-62.

22. Mehr SS, Sadowsky JL, Doyle LW, Carr J. Sepsis in neonatal intensive

care in the late 1990s. J Pediatr Child Health 2002;38:246-51.

23. Lee YL, Cesario T, Lee R, Nothvogel S, Nasser J, Farsad N, et al.

Colonization by Staphylococcus species resistant to methicillin or

quinolone on hands of medical personnel in a skilled-nursing facility.

Am J Infect Control 1994;22:346-51.

24. Cespedes C, Miller M, Quagiarello B, Vavagiakis P, Klein RS, Lowy

FD. Differences between Staphylococcus aureus isolates from medi-

cal and nonmedical hospital personnel. J Clin Microbiol 2002;40:

2594-7.

25. Larson E, Bobo L, Bennett R, Murphy S, Seng ST, Choo JT, et al. Lack

of caregiver hand contamination with endemic bacterial pathogens in

a nursing home. Am J Infect Control 1992;20:11-5.

26. Opal SM, Mayer KH, Stenberg MJ, Blazek JE, Mikolich DJ, Dicken-

sheets DL, et al. Frequent acquisition of multiple strains of methicillin-

resistant Staphylococcus aureus by healthcare workers in an endemic

hospital environment. Infect Control Hosp Epidemiol 1990;11:

479-85.

27. Gordin FM, Schultz ME, Huber RA, Gill JA. Reduction in nosoco-

mial transmission of drug-resistant bacteria after introduction of

an alcohol-based handrub. Infect Control Hosp Epidemiol ;26:

650-3.

28. Ananthapadmanabhan KP, Moore DJ, Moore DJ, Subramanyan K,

Misra M, Meyer F. Cleansing without compromise: the impact of

cleansers on the skin barrier and the technology of mild cleansing.

Dermatol Ther 2004;17(Suppl 1):16-25.

29. Aiello AE, Larson E. Antibacterial cleaning and hygiene products as an

emerging risk factor for antibiotic resistance in the community. Lancet

2003;3:501-10.

Implications of the changing face ofClostridium difficile disease for healthcare practitionersLynne V. McFarland, PhD,a,b Henry W. Beneda, MPH,c Jill E. Clarridge, PhD,d,e and Gregory J. Raugi, MD, PhDf,g

Seattle, Washington

Recent reported outbreaks of Clostridium difficile-associated disease in Canada have changed the profile of C difficile infections.Historically, C difficile disease was thought of mainly as a nosocomial disease associated with broad-spectrum antibiotics, andthe disease was usually not life threatening. The emergence of an epidemic strain, BI/NAP1/027, which produces a binary toxinin addition to the 2 classic C difficile toxins A and B and is resistant to some fluoroquinolones, was associated with large numbersof cases with high rates of mortality. Recently, C difficile has been reported more frequently in nonhospital-based settings, such ascommunity-acquired cases. The C difficile disease is also being reported in populations once considered of low risk (children andyoung healthy women). In addition, poor response to metronidazole treatment is increasing. Faced with an increasing incidence ofC difficile infections and the changing profile of patients who become infected, this paper will reexamine the current concepts onthe epidemiology and treatment of C difficile-associated disease, present new hypotheses for risk factors, examine the role ofspores in the transmission of C difficile, and provide recommendations that may enhance infection control practices. (Am J InfectControl 2007;35:237-53.)

Recent events have refocused attention on Clos-tridium difficile infections and caused the medicalcommunity to reevaluate its assumptions about thevirulence, known risk factors, and possible modes oftransmission of this important health care-associatedpathogen.

Clostridium difficile–associated disease (CDAD) istraditionally thought of as a nosocomial pathogenthat may cause limited, small outbreaks. The pathogen-esis of CDAD involves a triad of factors: (1) disruptionof normal intestinal flora (usually by broad-spectrumantibiotics), (2) exposure to C difficile (usually duringhospitalization), and (3) host factors (comorbidity andadvanced age or impaired immune status). The CDAD

From the Department of Health Services Research and Development,a

Veterans Administration Puget Sound Health Care System; Departmentof Medicinal Chemistry,b School of Pharmacy, University of Washington;Department of Infection Control,c Veterans Administration PugetSound Health Care System; Laboratory Service,d Veterans Administra-tion Puget Sound Health Care System; Department of Laboratory Med-icine,e University of Washington; Dermatology Section,f Primary andSpecialty Services, Veterans Administration Puget Sound Health CareSystem; and Department of Medicine,g Division of Dermatology, Uni-versity of Washington, Seattle, WA.

Address correspondence to Lynne V. McFarland, PhD, Department ofHealth Services Research and Development, VA Puget Sound HealthCare System, S152, 1100 Olive Way No. 1400, Seattle, WA 98101.E-mail: Lynne.McFarland@va.gov.

0196-6553/$32.00

Copyright ª 2007 by the Association for Professionals in InfectionControl and Epidemiology, Inc.

doi:10.1016/j.ajic.2006.06.004

is not a trivial disease because C difficile may havelife-threatening complications, such as ileal perfora-tion,1 fulminant colitis,2 toxic megacolon,3 or brainempyema.4 Clostridium difficile-associated disease hasalso been shown to extend hospital stays a mean of 4to 14 days,5-7 increase the risk of other nosocomial in-fections,8 and increase health care costs.5-7,9,10 The esti-mated cost of health care associated with CDAD is $1.1billion/year in the United States.7,11 Although most pa-tients with an initial episode of CDAD respond well to ei-ther metronidazole or vancomycin, 20% developrecurrent episodes of CDAD that occur episodicallyover years, despite repeated antibiotic treatments.9

After more than 20 years of research and experiencewith C difficile, health care providers became com-placent that methods of transmission, risk factors,effective infection control practices, and effective ther-apies were well understood. However, the incidence ofCDAD continues to rise despite continued focus onmethods to reduce the number of hospital cases, toeradicate the spores of C difficile from the environment,and to find effective strategies for treating patients withrecurrent CDAD.

In 2002, hospitals in Quebec, Canada, reported out-breaks of high numbers of cases of serious CDADinfections that were health care associated and had ahigh mortality rate. These outbreaks challenged practi-tioners to reevaluate diagnosis, treatment, and infec-tion control strategies in light of the changing profileof CDAD. The purpose of this paper is to summarizeand integrate these new findings into our understand-ing of CDAD and to discuss how these new data shouldinfluence infection control practices.

237

238 Vol. 35 No. 4 McFarland et al

INCIDENCE IN NORTH AMERICA

In the United States, rates of CDAD have continued toincrease since the 1990s despite established infectioncontrol programs and effective treatments (Table 1).Higher CDAD rates are typically reported in teachinghospitals,25 on medical and surgical services or inten-sive care units, and among elderly patients.15

In December 2002, CDAD outbreaks of an unexpect-edly large number of cases were reported in 30 hospi-tals in the providence of Quebec, Canada. From 2003through 2005, outbreaks spread to adjacent hospi-tals.12 These outbreaks were characterized by a 4.5-fold higher incidence (156.3/100,000) over historicalrates (35.6/100,000 in 1991), nearly a 5-fold increasein mortality (4.5% in 1991 to 22% in 2004), and a2.5-fold increase in complicated CDAD cases (7.1%in 1991 to 18.2% in 2003).12,26 An estimated 14,000cases, including 2000 deaths (28% mortality rate),were reported in Quebec during this outbreak period.27

C difficile was the primary cause of mortality in 3.2% ofthe 2000 deaths. It was proposed that possible causesfor these outbreaks included the emergence of a hy-pervirulent strain of C difficile, suboptimal infectioncontrol practices, or a poor response to the standardtreatment for CDAD (metronidazole).12,19,28

CHARACTERIZATION OF THE BI/NAP1/027STRAIN

The designation of the strain associated with theCanadian outbreaks, BI/NAP1/027, combines the re-sults of several strain typing methods: restriction-endo-nuclease analysis (REA) group BI, pulsed-field gelelectrophoresis (PFGE) type NAP1 (North AmericanPFGE type 1), and ribotype 027. This strain is of toxino-type III (presence of binary toxin CDT and contains an18-base pair [bp] tcdC deletion). As a result of thetcdC deletion, toxin A and B production is not in-hibited.29 Strains with the tcdC deletion have beenfound to produce 16 times more toxin A and 23 timesmore toxin B than other C difficile strains that do nothave this gene deletion.30 Another novel characteristicof this strain is that it is resistant to gatifloxacin andmoxifloxacin. This is a recent development becauseolder isolates of BI/NAP1 were not resistant to fluoro-quinolones.31 As fluoroquinolones have become oneof the most prescribed groups of antibiotics, the devel-opment of antibiotic resistance in health care-associ-ated strains of CDAD has become a focus of clinicalconcern.

Because of the unprecedented increase in the num-ber of cases in Canada caused by this strain, historicalC difficile isolate collections were reexamined for thepresence of the BI/NAP1 strain. McDonald et al found

that 96 (51%) of 187 isolates from 8 United States hospi-tals collected between 2000 and 2003 were typed asBI/NAP1.31 This strain was found in 2 hospitals in theUnited States before the appearance of the Canadianoutbreaks. This strain is not new because it was first iso-lated in 1984, although infrequently (14 of 6000 isolatescollected before 2001). The difference is that older iso-lates of BI/NAP1 were not associated with outbreaks orresistant to fluoroquinolones.31 Since the Canadian out-breaks, the BI/NAP1 strain has been found in the UnitedStates,32 the United Kingdom,33 and The Netherlands.34

Implications for infection control: Examination of in-dividual clinical cases and incidence rates of CDAD maynot reveal the presence of an outbreak caused by hyper-toxigenic-producing strains. Although few hospitallaboratories currently assay for C difficile by cultureand strain type C difficile isolates, these would be a valu-able tool to track the occurrence of specific strains ofC difficile and the transmission through an institution.Strain typing methods such as ribotyping or pulsed-fieldgel electrophoresis are not readily available at most hos-pital laboratories.

VIRULENCE OF C DIFFICILE STRAINS

The main virulence factors for C difficile are anenterotoxin designated toxin A and a cytotoxindesignated toxin B. Both toxins disrupt the actin cyto-skeleton of intestinal epithelial cells by the UDP-glucose-dependent glycosylation of Rho and Rasproteins. Usually C difficile isolates from CDAD patientsproduce both toxin A and B, but nontoxigenic strains(A2B2) or atypical toxin strains (A2B1) may alsocause symptoms.4,35,36 Of 153 isolates from CDADcases at the Chicago Veterans Administration (VA) Med-ical Center, 123 (80%) were A1B1, 13 (9%) were non-toxigenic (A2B2), and 17 (11%) were toxin variants(A2B1).37 The frequency of toxin variants (A2B1) iso-lates from CDAD cases was found to be higher in otherpopulations (39% of 77 isolates) in Japan.38 Strains thatproduce only toxin A and not B (A1B2) have only beenreported in 1 patient with recurrent CDAD.39 As shownin Fig 1, the pathogenicity locus (PaLoc) of C difficilecontains 2 genes encoding toxin A (tcdA) and toxinB (tcdB) and 3 other important genes: a positive regu-lator for toxin production (tcdR), a gene producinga holin-like protein involved in membrane disruption(tcdE), and a negative regulator for toxin production(tcdC).29,40

A third toxin, a binary toxin designated CDT (actin-specific ADP-ribosyltransferease), was described in1988 from the C difficile strain CD196.41 The binarytoxin is composed of 2 independent proteins: cdtA (en-zymic component that catalyzes the ADP-ribosylationof actin and causes cytoskeleton disorganization) and

McFarland et al May 2007 239

Table 1. Changes in the frequency of Clostridium difficile-associated disease in various populations

Population CDAD rate (yr) References (No.)

Secular trends

Hospitalized patients, Quebec, Canada 35.6/100,000 (1991) Pepin et al, 2004 (12)

156.3/100,000 (2003)

Veterans Administration Facilities, United States 5.1/1000 (1994) Perlin, 2005 (13)

13.5/1000 (2004)

NNISS,* United States 4.7/10,000 (1994) Archibald et al, 2004 (14)

5.8/10,000 (2000)

NHDS,y United States 31/100,000 (1996) McDonald et al, 2006 (15)

61/100,000 (2003)

Community acquired

Community acquired, United States 7.6/100,000 (2004-2005) CDC, 2005 (16)

Community acquired, managed care, United States 3.2/100,000 (1993-1997) Frost et al, 1999 (17)

Community managed care, United States 12/100,000 (1992-1994) Levy et al, 2000 (18)

Community acquired, United Kingdom 22/100,000 (1994-2004) Dial et al, 2000 (19)

Community acquired, Sweden 25/100,000 (1999-2000) Noren et al, 2004 (20)

High-risk populations

Elderly patients, United Kingdom 39.5/100 Settle et al, 1998 (21)

Adult oncology, France 13.2/100 Blot et al, 2003 (22)

General medicine ward, United States 7.8/100 McFarland et al, 1989 (23)

Burn ICU, Kuwait 5.9/100 Rotimi et al, 2002 (24)

Elderly patients with NAP1 strain 865/100,000 Pepin et al, 2004 (12)

CDAD, Clostridium difficile-associated disease; NNISS, National Nosocomial Infections Surveillance system; NHDS, National Hospital Discharge Survey; ICU, intensive care unit.

*NNISS incidence density, cases per 10,000 patient days.yNHDS cases per discharges.

cdtB (binding component that recognizes cell-surfacereceptors and causes the enzymic component to be in-ternalized).42 The location of the binary genes has notbeen determined, but it is not located within the PaLoclocus. The prevalence of binary toxin in C difficile iso-lates from human sources ranges from 1% to16%.36,37,43-45 In some outbreaks, binary toxin-posi-tive strains of C difficile are more frequently isolated.Binary toxin was detected in 65% of the C difficileisolates among different REA strain types from a hospi-tal in Pennsylvania during 2001-2002.32 Most of thebinary toxin isolates also produce toxins A and B,43

but binary toxin isolates were also found in 16% of non-toxigenic (A2B2) C difficile isolates37 and in 11% oftoxin variant (A2B1) strains.36 Recently, binary toxinstrains of C difficile have been found to be resistant tofluoroquinolones, although there is no direct linkbetween the development of antibiotic resistance andthe carriage of binary toxin. The association betweenbinary toxin and pathogeneity is still unclear. Binarytoxin-positive isolates are cytopathic in cell cultureand, when tested in CDAD animal models, A2B2CDT1

strains were found to cause fluid accumulation in rab-bit ileal loops but not diarrhea or death in hamstermodels.46 To determine whether strains carrying bi-nary toxins were associated with increased virulencein humans, 26 CDAD cases of patients who were toxinA1B1CDT1 were compared with 42 CDAD controlswho were toxin A1B1CDT2.42 The binary-positivecases reported more abdominal pain (64% vs 39%,

respectively) and reported long (.7 days) course ofdiarrhea (56% vs 29%, respectively). Unfortunately,no data on the strain types were reported in this study.To analyze the influence of binary toxin alone, withoutthe confounding presence of toxin A or B, it wouldbe necessary to compare the severity of CDAD in pa-tients who are A2B2CDT1 with patients who areA2B2CDT2, but this has not been reported.

The presence of 2 major toxins (A or B) is usuallymore predictive of severe CDAD cases rather than in-fection with a specific strain or the presence of binarytoxin. One study found that the fecal toxin concentra-tion was associated with increasing CDAD severity (0.5U/g in mild CDAD, 6.8 U/g in moderate, and 149 U/g insevere CDAD (P , .001), but the strain type was not as-sociated with disease severity.47 This was in contrastto an older study that did not find significant differ-ences in toxin presence, toxin amount, or immunoblotstrain type in CDAD cases compared with asymptom-atic carriers.48 The new BI/NAP1 strain of C difficileappears to be an exception to the rule that straintype does not predict the severity of CDAD. A studyin the United States linked an increase in colectomiesbecause of fulminant cases caused by the BI/NAP1strain.31 It is still unclear whether the apparent in-creased severity of CDAD cases caused by the BI/NAP1 strain is due to increased toxin A and B produc-tion, presence of binary toxin, resistance to gatifloxa-cin and moxifloxacin, or another undescribedvirulence factor.28,30,31

240 Vol. 35 No. 4 McFarland et al

Fig 1. Schematic of Clostridium difficile toxin genes. Nineteen-kilobase region pathogenicity locus (PaLoc) containstcdR (positive regulator of toxin A and B production), tcdB (cytotoxin B), tcdE (holin-like protein), tcdA (enterotoxinA), and tcdC (negative regulator of toxin A and B). The binary toxin is controlled by cdtA (enzymic component) and

cdtB (binding component). Adapted from Rupnik et al29 and Barbut et al.42

Implications for infection control: Surveillance forthe BI/NAP strain of C difficile in health care facilitiesmay be of value to predict a possible increase in severecases of CDAD and define potential outbreaks; there-fore, early intervention of infection control measurescould prevent potential outbreaks. However, severecases of CDAD have been reported associated withother strains besides BI/NAP/027. The role of binarytoxin and virulence of CDAD is unclear.

COMMUNITY-ACQUIRED CDAD

The preconception that CDAD is mainly a nosoco-mial disease has been challenged recently by theincreasing frequency of community-acquired cases.In the past, rates of community-acquired CDADhave been low (Table 1), ranging from 3.2 to 16.2/100,000.17,18 Unfortunately, the definition of commu-nity acquired varies. Some studies of community-acquired CDAD in outpatients failed to determinewhether there had been recent hospitalization anddid not culture for C difficile at admission. One study,which did culture newly admitted patients at admis-sion, found that of the 112 patients who had positiveC difficile assays during the study, 74% were negativeon admission but acquired C difficile during their stayon the study ward; 21% were admitted with a positiveC difficile assay and had a history of hospitalization(#3 months); and only 5% were communityacquired (positive at admission with no history ofhospitalization).23

Currently, more community-acquired cases of CDADare being reported (Table 1).49 Patients admitted into 2hospitals over 4.5 years in British Columbia with a pri-mary diagnosis of CDAD were reviewed, and 67 (44%)had community-acquired C difficile.50 In France, signif-icantly more of the 26 C difficile isolates with binarytoxins were community acquired (65% vs 36%, respec-tively, P , .05), however, this study defined communityacquired as ‘‘onset of symptoms within 3 days after ad-mission, regardless of previous hospitalizations,’’ whichprobably overestimated the frequency of community-

acquired cases by including cases that may had beenpreviously hospitalized.42 In the United States, commu-nity-acquired cases of the BI/NAP1 strain have beenreported in 4 states.16 These 23 community-acquiredCDAD patients were young (48% #18 years of age),35% had no history of recent antibiotic exposure(in the prior 3 months), and 30% were thought to haveacquired C difficile from close contacts with diarrhealdisease. Of the 23 community-acquired cases, 26% re-quired hospitalization and 35% relapsed.

Implications for infection control: Community-acquired CDAD is becoming a greater problem. Theprofile of affected patients may be different than healthcare-associated CDAD cases and should prompt ap-propriate laboratory investigation in young peoplewith diarrheal disease. If community-acquired casesof CDAD continue to become more frequent, diagnosticalgorithms and standard definitions may need to beexpanded from including only patients with a recenthospitalization to include the community-acquiredcases as CDAD.51,52

SOURCES OF C DIFFICILE

Although the majority of cases of CDAD are reportedin the hospital setting, CDAD should be considereda health care-associated disease because it occurs inlong-term care facilities, day care facilities, and outpa-tient clinics.16,52-55 The question of whether the orga-nism is acquired in the hospital or is already presentat admission as part of the patient’s indigenous micro-flora has not been fully answered. Asymptomatic car-riers of C difficile may become symptomatic afterexposures to specific risk factors in the hospital leadto overgrowth and toxin production, triggering diar-rheal symptoms. Unfortunately, few investigators haveboth cultured patients at admission to a health carefacility and asked about prior history of recent hospi-talization.23 It has been accepted that if the newly admit-ted patient has a negative assay for C difficile and hashad no history of prior hospitalization, then if CDADdevelops during the hospital stay, the case can be

McFarland et al May 2007 241

considered as nosocomial. However, as the specificityof C difficile toxin assays range from 88% to 99% andC difficile can be frequently detected after an initial neg-ative C difficile toxin assay, false-negative results mayoccur.55,56 A recent study using a more sensitive poly-merase chain reaction (PCR) method detected C difficilein 53% of healthy, community-based adults (with nohistory of hospitalization or antibiotic exposure) thatconventional PCR methods or conventional toxinassays did not detect.57 Therefore, low-level, endemiccarriage of C difficile may be more common than previ-ously thought. Only after patients are exposed to agentsthat disrupt normal intestinal microflora (such as anti-biotics, surgery, medications) that allows the upsurgein the numbers of C difficile organisms do laboratoryassays detect C difficile frequently. Other evidencefrom patients who have multiple recurrences of CDADsupports endogenous carriage. A study using PCR ribo-typing of 89 C difficile isolates reported that most recur-rences (90%) were due to the endogenous strain ofC difficile acquisition, whereas only 10% of the recur-rences were due to a new strain.20 The source of theoriginal C difficile strain is still in question becausemost (84%) had a prior history of hospitalization inthis study. In another study of 18 patients with recurrentCDAD, 67% of the isolates from the recurrence wereidentical to the isolates from the initial episode, andonly 33% had been reinfected with a new strain of C dif-ficile.58 Reliance on strain typing to determine epidemi-ologic spread and source of infection was recentlycalled into doubt by a study that found simultaneouscarriage of multiple types of strains in patients withCDAD. Of 23 patients with a primary episode ofCDAD, 2 (8.7%) carried 2 distinct PCR ribotypes, and,of 23 patients with recurrent CDAD, 6 (26%) had differ-ent C difficile strains isolated from the same fecalsample.59

Sources of community-based or nonhealth care set-tings of C difficile may include exposure to spores in thesoil, carriage by pets (dogs, cats, horses), contaminatedfoods, or exposure to household contacts with diar-rhea.60,61 In 102 health care visitation dogs in Canada,58% were positive for C difficile, 71% of those isolatesproduced toxins, and 1 was typed as the hypervirulentBI/NAP1 strain.62,63 Although few reports of animal tohuman transmission of C difficile have been reported,these types of sources may be overlooked, especiallyin community-acquired CDAD.16

Implications for infection control: As community-acquired cases are more frequently reported, thesearch for the source of C difficile infections needs tobe expanded to nonhospital vectors. Further researchwith more sensitive C difficile assays is needed to deter-mine the prevalence of endogenous carriage of C diffi-cile to correctly target the preventive measures.

RISK FACTORS

Emerging populations at risk

Reports of CDAD in what has been considered as‘‘low risk’’ populations (young healthy women andchildren) have increased recently. Children are knownto develop CDAD, but reported outbreaks are less com-mon than in adult patients.54 Reports from Canada dur-ing 2000-2003 found that 200 young children (meanage, 5.4 years) developed CDAD, of whom 23% were suf-ficiently serious to require hospitalization, and 31% hadat least 1 recurrence.64 Whether the high rate of hospi-talization was due to a more virulent C difficile strainor shifts in other risk factors in the pediatric populationwas not reported. A high frequency (18%) of CDAD wasdetected in 250 children aged 5 to 12 years on antibi-otics in a hospital in India.65 Clostridium difficile-associ-ated disease was recently found in another populationtraditionally considered of low risk (young peripartumwomen). Ten women developed CDAD, of whom 40%were sufficiently severe to require hospitalization, and50% had at least 1 recurrence of disease.16 Of these 10cases, 64% had isolates of the recent epidemic strain(BI/NAP1/027) that carries the binary toxin and is consid-ered more virulent than other C difficile strains.30

Proton pump inhibitors

Association of CDAD with antacid therapy or gastricacid-lowering medications remains controversial; sev-eral studies found a positive association,8,66 but an-other study found no significant association.67

Recently, 3 studies have shown that proton pumpinhibitors (PPI) were associated with higher rates ofCDAD,19,68,69 whereas 2 studies did not find a signifi-cant association between CDAD and PPIs.70,71 At thetime of the Canadian outbreaks, a cohort study of1187 inpatients at a Montreal teaching hospital fromAugust 2002 to May 2003 found CDAD in 81 (68%) ofthe patients, and those who used PPIs had twice therisk of CDAD (odds ratio [OR], 2.1; 95% CI: 1.2-3.5).19

Another study in community-acquired CDAD casesfound that people taking PPI had nearly 3 times therisk for developing CDAD.68 In another study in along-term facility (347 beds) in New York, significantlymore (60%) of CDAD cases used PPI compared withonly 32% of control residents.69 The PPIs are fre-quently used in hospitalized patients (49%-68%), butit is apparent that more research is needed before lim-iting PPI use in hospitals, as was recently suggested.19

Fluoroquinolone therapy

Fluoroquinolones became the most common class ofantibiotics prescribed to adults in the United States.72

Historically, fluoroquinolone use was associated with

242 Vol. 35 No. 4 McFarland et al

a 2- to 13-fold risk of CDAD but, because this typeof antibiotic was not frequently prescribed, theabsolute numbers of quinolone-associated CDAD werelow.10,73-75

One of the proposed theories behind the largereported outbreaks in Canada was that a fluoroquino-lone-resistant strain (BI/NAP/027) was circulating atthe same time when fluoroquinolone use was commonin Canadian hospitals. Two studies in Canada founda significantly higher risk of CDAD for patients pre-scribed fluoroquinolones. In a study of 12 hospitalsin Quebec, Canada, during 2004, CDAD case patientshad nearly 4 times higher rates of fluoroquinoloneuse (OR, 3.9; 95% CI: 2.3-6.6) compared with matchedcontrols.28 Ciprofloxacin, gatifloxacin, and moxifloxa-cin were significantly associated with CDAD, whereaslevofloxacin was not. A retrospective cohort study per-formed of 5619 patients admitted to 1 hospital from2003 to 2004 in Quebec found that use of fluoroquino-lones was high (29%). Patients given fluoroquinoloneshave 3.4 times the risk of developing CDAD than thosepatients not given this class of antibiotic.70

An opportune time to study the effect of an antibi-otic on health care-associated pathogen rates is whena hospital changes its formulary recommendations.Two such studies found a higher incidence of CDADassociated with the switch to gatifloxacin. A formularyreplacement of levofloxacin by gatifloxacin occurredat the Atlanta VA long-term care facility in October2001.53 Before the formulary change, the rate ofCDAD was 0.4/1000 patient-days and after the switch,the CDAD rate increased to 1.3/1000. After levofloxacinwas reintroduced on July 2002, the rate of CDAD fell topregatifloxacin levels (0.3/1000 patient-days). Theseresults were confounded by a single facility-widehypochlorite disinfection performed in June 2002.However, because CDAD rates continued to be low formonths following the single disinfection process, theseresults favor a central causative role for the formularyantibiotic change for the increase in CDAD rates. Anested case-control study of CDAD cases at this institu-tion found that gatifloxacin use was significantly morecommon in CDAD cases than in controls (67% vs 25%,respectively, P , .05).53 Another study at a 600-bed ter-tiary-care teaching medical center in Pittsburgh docu-mented an increased rate of CDAD after a formularychange from ciprofloxacin to levofloxacin.76 The for-mulary change occurred March 1999 when the rateof CDAD was 2.7/1000 discharges, and, 9 months afterthe switch to levofloxacin, the rate of CDAD increasedto 6.8/1000 discharges. A nested case-control study of203 CDAD cases and 203 matched (admission date,type of medical service, and length of stay [LOS]) con-trols found that levofloxacin use was associated withan increased risk of CDAD (OR, 2.0; 95% CI: 1.2-3.3).

Implications for infection control: Fluoroquinoloneuse is a major risk factor for CDAD. Although data arepreliminary regarding the precise risk for individualmembers of this antibiotic class, it is clear that changesin hospital formulary preferences among fluoroquino-lones have a profound effect on CDAD rates. The role ofPPIs as a risk factor for CDAD is not as clear.

INFECTION CONTROL PRACTICES

Historically, the control of C difficile outbreaks haverelied on multiple strategies, including (1) interruptingthe routes of transmission by handwashing, patientisolation, enteric precautions; (2) reducing the riskof exposure to C difficile spores using environmentaldisinfectants or disposable medical equipment; (3) re-ducing the pool of susceptible patients by antibioticcontrol policies that limit broad-spectrum antibioticuse; and (4) reducing infectious cases by promptdiagnosis and effective treatment.51,55 Although thestrengths of the evidence-based effectiveness of thesedifferent strategies vary (Table 2), infection control pol-icies have relied on a combination of these methods tocontrol CDAD.

With the upsurge in reported CDAD outbreaks andthe emergence of an epidemic, hypervirulent strainof C difficile, reliance on standard infection controlpractices may become more important. One explana-tion for the reported outbreaks in Canada was substan-dard infection control practices.113 The decline ofCDAD cases in 2005 (from .40 cases/1000 admissionsto ,12 cases/1000 admissions) was associated withmore stringent infection control and cleaning practicesthat began in June 2004.113 During this same time pe-riod, no changes in antibiotic use occurred.114 Otherclinicians claim that changes in infection control poli-cies did not significantly reduce CDAD rates.115 TheQuebec provincial surveillance system reported thatCDAD rates in late 2004 and early 2005 were similarto the time of preinfection control practice changes.116

Regardless of the cause of the upsurge in reportedcases in Canada, recent research has focused (Table 3)on the effective control of CDAD in health care settingsin 6 strategic areas: (1) surveillance programs, (2) inter-rupting transmission, (3) reducing exposure to C diffi-cile spores, (4) antibiotic stewardship programs, (5)preventive vaccines, and (6) effective treatments. How-ever, the most effective infection control programs areintegrated programs that combine all 6 strategies.51,55,112

Surveillance

Active surveillance and reporting of CDAD is notrequired in the United States, but mandatory reportingwas started in the United Kingdom in 2004.120 The Vet-erans Health Administration (VHA) is the single largest

McFarland et al May 2007 243

Table 2. Historical evidence rating for studies on strategies to limit the transmission of Clostridium difficile

Studies by evidence-based rating, reference (No.)*

Recommended strategies A B C

Surveillance

Secular trend tracking Frost et al, 1999 (17)

Rexach et al, 2005 (77)

Archibald et al, 2004 (14)

Van den Berg et al, 2004 (78)

Typing of C difficile strains McCoubrey et al, 2003 (79)

Wullt et al, 2003 (80)

Noren et al, 2004 (20)

Geric et al, 2004 (37)

Kato et al, 2005 (81)

Outbreak investigations to

reveal weak areas

McFarland et al, 1989 (23)

Struelens et al, 1991 (82)

Prevent the spread of

C difficile organism

Handwashing with chlorohexidine

or use of gloves

Johnson and Gerding,

1990 (83)

Pittet et al, 2000 (84) McFarland et al, 1989 (23)

Bettin et al, 1994 (86) Johnson and Gerding, 1998 (85)

Boyce and Pittet, 2002 (87)

Salemi et al, 2002 (88)

Environmental disinfectants Wilcox et al, 2004 (89) Kaatz et al, 1998 (90) Rutala et al, 1993 (91)

Struelens et al, 1991 (82) Wilcox and Fawley, 2000 (92)

Mayfield et al, 2000 (93) Hota, 2004 (94)

Use of disposable

medical equipment

Jernigan et al, 1998 (95) Brooks et al, 1992 (96)

Patient isolation/cohorting Boone et al, 1998 (97) McFarland et al, 1989 (23)

Verity et al, 2001 (98)

Treatment of asymptomatic

carriers not effective

Johnson et al, 1992 (99) Bender et al, 1986 (100)

Kerr et al, 1990 (101)

Reduce clinical burden of disease

Antibiotic restriction programs Pear et al, 1994 (102) Climo et al, 1998 (103)

Ho et al, 1996 (104)

McNulty et al, 1997 (105)

Settle et al, 1998 (21)

Ludlam et al, 1999 (106)

Carling et al, 2003 (107)

Khan and Cheesbrough,

2003 (108)

Thomas et al, 2003 (109)

Integrated infection control programs

Combinations of the above Stone et al, 1998 (110) Struelens et al, 1991 (82)

Zafar et al, 1998 (111)

Apisarnthanarak et al,

2004 (112)

*A, Evidence from randomized, controlled trials in patients with C difficile-associated disease; B, evidence from open trials, pre- and postintervention trials, crossover trials in

healthy volunteers or uncontrolled trials; C, observational studies, expert opinion, case reports, anecdotal evidence only.

provider of health care in the United States, operating163 hospitals and 130 long-term care facilities. Withinthe VHA, 93% of the facilities operate periodic or con-tinuous surveillance for C difficile.121 Hospital surveil-lance programs have shown to reduce rates ofmethicillin-resistant Staphylococccus aureus but havenot been studied specifically to observe its effect onCDAD rates.122 The value of an active surveillanceprogram includes the documentation of secular trends,

the detection of emergent antibiotic-resistant strains,the investigation of outbreaks, and the ability to gatherdata on the effectiveness of infection controlprograms.14,35,77,78,123,124

Interruption of transmission routes

Clostridium difficile is thought to be transmitted ei-ther from person to person by contaminated hands

244 Vol. 35 No. 4 McFarland et al

Table 3. Recent studies of interventions to limit the transmission of Clostridium difficile

Strategy

Outcomes

References (No.)

CDAD incidence

before intervention

CDAD incidence

after intervention

Interrupting routes of transmission

Alcohol hand sanitizers 2.7/1000 patient-days 6.8/1000 patient-days* Muto et al, 2005 (76)

Alcohol hand gel 3.2/10,000 patient-days 3.4/10,000 patient-days, ns Gordin et al, 2005 (117)

Reducing exposure to C difficile spores

Hypochlorite disinfectant 8.9/100 admissions 5.3/100 admissions* Wilcox et al, 2004 (89)

Antibiotic stewardship programs

Reduce cephalosporins 3.9/100 admissions 1.2/100 admissions* O’Connor et al, 2004 (118)

Reduce ceftaximes 4.5/100 admissions 2.2/100 admissions* Wilcox et al, 2004 (119)

Integrated infection control program 8.0/1000 patient-days 4.2/1000 patient-days* Apisarntharak et al, 2004 (112)

CDAD, Clostridium difficile-associated disease; ns, not significant.

*P , .05.

or indirectly through contaminated environmental sur-faces. Clostridium difficile has been cultured from thehands of health care personnel, patients, and visitorsin rooms of CDAD patients, and national guidelines rec-ommend thorough handwashing and/or use of dispos-able gloves.23,85,87 In an effort to reduce CDAD rates atseveral hospitals, alcohol hand sanitizers were installed,but this infection control practice did not have any sig-nificant impact on the rates of CDAD.76,125 Alcoholhand gels were introduced at the Veterans Affairs Medi-cal Center in Washington, DC, but had no effect onCDAD rates (3.2/10,000 patient-days before and 3.4/10,000 after).117 The lack of effectiveness for alcohol inthese 2 studies is not surprising because alcohol is usedto stimulate C difficile spore germination in culturebroths.14 Therefore, handwashing using soap or chloro-hexidine or use ofdisposable gloves is still recommended.

Reducing environmental contamination

The persistence of C difficile in health care settingsis largely due to its ability to shed durable spores ona wide variety of environmental sites.23,126 The Societyof Hospital Epidemiologists of America (SHEA) and theCenters for Disease Control and Prevention (CDC) con-tinues to recommend the use of unbuffered 1:10 hypo-chlorite solution rather than detergents to disinfectsurfaces.52,94,127 Hypochlorite solutions have beenfound to be more effective than other types of environ-mental disinfectants, including detergents,93,128 gluta-radehyde,91 combination products of acetic acid andhydrogen peroxide.129 Acidified bleach, regular bleach,and hydrogen peroxide were found to inactivateC difficile spores within 15 minutes, but are all strongoxidizers and not recommended for routine use.130

Wilcox et al recently confirmed the usefulness of hypo-chlorite disinfectants on a ward that showed a signifi-cant reduction in CDAD rates (8.9/100 admissionsbefore and 5.3/100 after, P , .05), but a ward using

detergents did not show a change in CDAD rates.89

The continued development of an environmental disin-fectant that is sporicidal yet sufficiently nontoxic forroutine use is challenging.

The role of spores

Spores are part of the natural life cycle of C difficile.Humans may ingest, inhale, or swallow the spores,which are then passed through the stomach. Fordtranet al showed that, at normal gastric pH (,4.0), vegeta-tive cells of C difficile die, but spores are not affected.131

In the acidic environment, the spores are not killed di-rectly, but germinant binding may be inhibited. Germi-nants are factors that initiate spore germination andinclude heat, free bile acids, and lactate.132-134 ThePPIs, which have been associated with higher CDADrates, raise gastric pH, allowing germinants to bindmore effectively.135 This may be the mechanism be-hind the observation that patients given PPI have in-creased susceptibility to CDAD. Gastric acid acts as anatural defense mechanism, not by directly killingspores, but by inhibiting spore germination. Once thespores reache the small intestine, most germinatewithin 1 hour.136 Vegetative cells of C difficile mayflourish in the intestines only if the normal microbialflora have been disrupted. Normal flora has the abilityto inhibit the colonization of opportunistic pathogensthrough a complex mechanism termed ‘‘colonizationresistance.’’137 Once this normal microflora barrier isdisrupted (for example, by broad-spectrum antibiotics,surgery, chemotherapy, or other medications), C diffi-cile can colonize, reproduce, and produce toxins. Thesetoxins (toxins A and B) attach to enterocytes, causingcellular disruption and fluid loss, resulting in diarrhea.As vegetative C difficile cells die, new spores are formedand released in the stool into the environment. Somespores may remain lodged in the intestines, acting asseeds for future CDAD episodes.

McFarland et al May 2007 245

Although vegetative forms of C difficile die rapidlyoutside the anaerobic environment of the intestines,C difficile spores are stable at a wide range of tempera-tures from 2208C to 908C,138 are resistant to desicca-tion, are sensitive to only high heat and alkalinepH,139 and can survive brief heating at 1008C (2-35minutes).140 Spores facilitate transmission of CDADwithin health care settings, and their persistence canlead to subsequent infections and outbreaks. Whenenvironmental surfaces are assayed for C difficilespores in a room with either an asymptomatic carrieror a symptomatic patient, spores have been found onnearly every type of environmental surface, especiallynear the patient areas.23,126 Several studies have shownpersistence of spores of C difficile in the environmentfor 5 months.141 In 1 study, environmental sites werecultured in rooms with and without CDAD patients.23

Once a CDAD patient was in the room, 45% of environ-mental samples were positive for C difficile spores, and82% of those environmental isolates were of the iden-tical strain type as the patient’s (Fig 2). One year later,C difficile spores were still detected in the room, and36% of the isolates were of the original patient’s straintype. As time progressed, not only did the patient’sstrain persist, but other types of C difficile strains werefound on environmental surfaces as other patientstransitioned through the room.

Some strains of C difficile are hyperproducers ofspores, which enables wide spore dissemination.92

The emerging BI/NAP1/027 strain is one of these hyper-producing spore strains, which might help to explainwhy outbreaks caused by this strain resulted in agreater number of patients than outbreaks caused byother C difficile strains.142

Efforts to reduce the dissemination of spores intothe environment have been focused on 2 fronts: (1) ef-fective therapies that might reduce spore carriage and(2) environmental disinfectants. Among 163 patientswith recurrent CDAD, vancomycin or metronidazoletherapy cleared vegetative cells of C difficile effectively,but neither antimicrobial was effective in clearingspores by the end of therapy (43% and 56% positivefor spores, respectively).143 To date, there is no effectivetherapy directed at C difficile spore eradication. Thesearch continues for an environmental disinfectantthat would be sporicidal for C difficile and practical fora wide variety of surfaces and yet be sufficiently non-toxic for routine use by patients and health care workers.

Antibiotic stewardship programs

One current strategy to retard the increasing inci-dence of antibiotic-resistant pathogens (such as theBI/NAP1 strain of C difficile) focuses on antibiotic utili-zation programs. Overuse of antibiotics has led to the

development of antibiotic-resistant strains by severalpathogens, and C difficile is no exception to this trend.Clindamycin-resistant strains of C difficile have beenreported in Sweden,144 Poland,145 and the UnitedStates.146 The frequency of macrolide (MLSb)-resistantstrains of C difficile also has been increasing inPoland4,123 and in Germany.147 Antibiotic stewardshipprograms, multidisciplinary programs involving physi-cians, microbiologists, pharmacists, and infection con-trol practitioners, set protocols for prior authorizationand concurrently review and give feedback on antibi-otic use.148 There are many reports in the literaturethat antibiotic restriction has reduced CDADrates,52,103,108 and a recent review found that 60% ofstudies found a significant reduction of CDAD withthis strategy.149 Recent investigations have confirmedthese earlier findings. A reduction in cephalosporinuse at a hospital in Ireland led to a reduction in riskof acquiring CDAD by one third.118 Cefotaxime usewas reduced on a geriatric medical unit in a UnitedKingdom hospital in favor of piperacillin-tazobactam,resulting in a 52% decrease in CDAD rates (Table 3).119

Vaccines

Vaccines against C difficile may be another strategyto reduce the number of susceptible patients and limit

Fig 2. Persistence of Clostridium difficile spores inthe hospital environment. Sequential assay results

over 1 year of environmental sites from 22 roomsnegative for C difficile before a CDAD-positive

patient was admitted to room. (Previouslyunpublished data, used by permission of author,

Lynne McFarland.)

246 Vol. 35 No. 4 McFarland et al

health care-associated outbreaks. Early research intovaccines has shown promise. A toxoid vaccine wastested in 30 healthy volunteers and showed an in-creased level of antitoxin A immunoglobulin G.150 Aparenteral C difficile vaccine containing toxoid A andtoxoid B was tested in 3 patients with recurrentCDAD.151 Two of the 3 patients showed significant in-creases in serum IgG for antitoxin A and B, and nonereported further recurrences. Currently, clinical trialsare ongoing to determine the value of vaccines forthe prevention of CDAD.

Effective treatments

The last strategy to control CDAD rates is the prompttreatment of infected cases so that transmission ofspores into the environment is shortened and potentialtransmission to other susceptible patients is limited.Treatments for CDAD include antibiotic therapy, pro-biotics, prebiotics, immunoglobulin treatment, andtoxin-binding polymers.

Antibiotic therapy

Treatment of infected patients is another strategy toreduce health care associated CDAD. Oral vancomycinis the only FDA-approved treatment for CDAD, butmetronidazole has also been used. These 2 antibioticshave been considered the ‘‘standard’’ treatments forCDAD152 and have been considered of equivocal effi-cacy for decades, but recent data place this opinioninto question. In 80% to 90% of patients with CDAD,diarrhea symptoms resolve within a typical 10-daytreatment, regardless of antibiotic type. After antibi-otics have been discontinued, 20% to 60% of patientsmay develop at least 1 recurrence of CDAD within 2to 4 weeks, necessitating additional antibiotic treat-ments.143 Some patients may develop ‘‘recurrentCDAD’’ or repeated episodes that occur over a periodof years. In these types of patients, vancomycin is usu-ally given for the recurrences, along with another typeof adjunctive therapy. In 163 cases of recurrent CDAD,tapering or pulsed doses of vancomycin were signifi-cantly more effective than metronidazole.143 By theend of therapy, vancomycin was more effective atclearing C difficile culture and/or toxins (89%) thanmetronidazole (59%, P , .001).

Recently, the effectiveness of metronidazole hasbeen questioned because several studies reportedhigh failure rates.71 Typically, once metronidazole isgiven, over 95% of patients report that their diarrhealsymptoms resolved within 3 to 4 days.55,152 Severalrecent studies have reported higher than expectedmetronidazole failure rates. Of 207 patients withCDAD treated with metronidazole at the HoustonVeterans Administration Medical Center from 2003 to

2004, 46 (22%) did not respond to the initial metroni-dazole treatment, and 58 (28%) recurred within 90days.153 Musher et al153 examined a small number(18 isolates) from this institution and found that all iso-lates were sensitive to metronidazole and were ofmixed PFGE strain types. In 845 patients treated withmetronidazole in Quebec during 2003-2004, 26%failed initial treatment, and 338 (47.2%) recurredwithin 60 days of treatment.154 The recurrence rate re-mained high even after adjusting for the aging popula-tion. A retrospective study of 99 cases of CDAD atTemple Hospital in Pennsylvania during 2000-2001found that 38% failed to respond to metronidazoletreatment.155 Risk factors that predicted metronidazolefailure were low albumin level (,2.5 g/L) and stay in anintensive care unit. No recurrence rates were given inthis paper. A prospective study at the same hospitalin 2005 found that of 27 CDAD patients treated withmetronidazole, 17 (59%) did not respond to treat-ment.156 All of the 17 patients were still taking the an-tibiotic that induced CDAD, whereas all the patientsthat responded well to metronidazole discontinuedthe inducing antibiotic.

One possible reason why a higher rate of metronida-zole treatment failure was observed is antibiotic resis-tance. Previously, metronidazole resistance has beeninfrequent in C difficile isolates. No resistance to eithermetronidazole or vancomycin was found in 186 iso-lates from patients in the United Kingdom157 or in140 isolates from patients in Poland.145 Of 49 C difficilestrains isolated from patients in Israel, none were resis-tant to vancomycin, but 1 (2%) was resistant to metro-nidazole.158 Of 415 C difficile isolates from patients inSpain, 6% of isolates were resistant to metronidazole,and 3% had reduced susceptibility to vancomycin.159

Several studies analyzed a small number (10-20) ofmetronidazole failures and did not find any metronida-zole-resistant isolates associated with the fail-ures.160,161 None of the isolates from patients withmetronidazole treatment failure in the previous 4 stud-ies were checked for metronidazole resistance nor hasit been reported whether the BI/NAP1 strain has devel-oped metronidazole resistance. The reason behind theapparent increase in metronidazole failures has notbeen explained but may be due to other factors relatedto the host, such as poor immune status, comorbidities,or intestinal conditions. As recurrent CDAD continuesto persist as a clinical problem and the frequency ofmetronidazole failure increases, the search for moreeffective treatment strategies continue.162

Probiotics

Probiotics (nonpathogenic microbes administeredto improve intestinal balance and restore disrupted

McFarland et al May 2007 247

Table 4. Proposed changes to established strategies for health care practitioners for Clostridium difficile-associated disease

Previous recommendations* Current proposed recommendations

Infection control Use of alcohol-based antiseptics or gloves Discontinue alcohol-based antiseptics

Patient isolation precautions if incontinent Contact precautions, per facility isolation

policy guidelines Patient isolation or

cohorting during outbreaks for

incontinent cases

Hypochlorite disinfection of surfaces of room

with CDAD patient

Hypochlorite disinfectant effectiveness

confirmed

Antibiotic restriction programs

Surveillance for C difficile isolates rates,

review of severe CDAD cases, identification

of specific strains

Educational programs stressing transmission,

handwashing procedures, and

environmental disinfection

Antibiotic stewardship programs established

More research for methods to destroy

C difficile spores

Integrated infection control programs during

outbreaks heightened (more stringent

handwashing, cohorting, contact

precautions, environmental disinfection)

Diagnosis Culture or toxin assay for diarrhea stools only Consider both C difficile culture and toxin

assays, rapid assays during outbreaksDo not assay asymptomatic carriers

Inclusion of ‘‘low-risk’’ populations when

diagnosing C difficile cases (community

acquired, children, peripartum women,

recent use of fluoroquinolones)

Treatments Prompt treatment of CDAD Prompt treatment of CDAD cases confirmed

Oral metronidazole is treatment of choice Combine standard antibiotic treatment

(vancomycin) with alternative treatments

for cases of recurrent C difficile diseaseRetreat recurrences with second course of

metronidazoleMetronidazole failures followed until

diarrhea resolved

Discontinue offending antibiotic if possible

CDAD, Clostridium difficile-associated disease.

*Reference: 2002 Society of Hospital Epidemiologists of America (SHEA) Position Paper.52

normal microflora) are moving into mainstream for thetreatment of gastrointestinal disorders.162,163 In 1 study,patients with CDAD were randomized to a combinationtreatment of oral vancomycin (2 g/day for 10 days) andSaccharomyces boulardii (1 g/day for 4 weeks) or vanco-mycin and placebo. Patients treated with vancomycinand the probiotic had significantly decreased recur-rence rates (16.7%) compared with vancomycin andplacebo (50%).164 A metaanalysis of 6 randomized con-trolled trials using probiotics combined with 1 of the 2standard antibiotics to treat CDAD found thatprobiotics significantly reduced the risk of CDAD(combined relative risk, 0.59; 95% CI: 0.41-0.85;P 5 .005).165

Prebiotics

Prebiotics are nondigestible food components(starch or fiber) that stimulate the growth of bifidobac-teria, a type of bacterium thought to play a major rolein inhibiting the establishment of opportunistic patho-gens in the intestine. Prebiotics are given alone, with-out a probiotic organism. A randomized trial of 142

CDAD cases treated with standard antibiotics withor without an adjunctive prebiotic (oligofructose) wasconducted. Significantly fewer (8.3%) patients in theprebiotic-treated group had a recurrence within 60days compared with those on placebo (34.3%, P ,

.001).166

Immunoglobulin treatment

The host immune response is an important predic-tor of recurrences of CDAD, and the production of an-titoxins A and B antibodies are found in patients whodo not develop recurrent CDAD.162 Unfortunately, norandomized, placebo-controlled trials have been re-ported. Evidence from uncontrolled trials or case seriesstudies show that increasing the immune responsemay be beneficial.167 In a small pilot study of patientswith CDAD (9 had recurrent CDAD), all were givenbovine antibody-enriched whey for 2 weeks after treat-ment with standard therapy (usually metronida-zole).168 This enriched whey product has high levelsof IgA antitoxin antibodies and has been shown to beprotective of CDAD in hamster models. Of the 16

248 Vol. 35 No. 4 McFarland et al

patients, 15 cleared C difficile toxins by the end of treat-ment, and none reported further recurrences (medianfollow-up, 333 days). In another small case series, 5 pa-tients with recurrent CDAD were treated with standardantibiotics (metronidazole and/or vancomycin) and in-travenous immunoglobulin (400 mg/kg) given either insingle or multiple doses.169 Two patients died of unre-lated causes during the study, 2 patients reported nofurther recurrences, and 1 patient recurred after 6weeks. Another study treated 14 patients with refractoryCDAD with intravenous immunoglobulin (150-400 mg/kg), and 64% responded within 10 days.170 This ap-proach seems promising, but requires controlled clinicaltrials to prove that immunoglobulin is an effective treat-ment for recurrent CDAD.

Toxin-binding polymer

Another approach is to give an agent that would bindC difficile toxins within the intestinal lumen. Braunlinet al found that tolevamer, a sodium salt of styrene sul-fonate polymer, binds C difficile toxin A extensively.171

In a phase 2 trial of 222 patients with CDAD, a highdose (6 g/day) of tolevamer was found to be equivalentto vancomycin treatment (500 mg/day).172 Of thosetreated with vancomycin, 91% were cured by day 10compared with 83% treated with 6 g/day of tolevamer,and only 67% of those treated with 3 g/day of tolevamerresponded. Controlled clinical studies in humans with ahigher dose (9 g/day tolevamer) are ongoing.162

Implications for infection control: Although newerinfection control strategies have recently been tested(alcohol handrubs, different types of disinfectants,vaccines, new treatments), only antibiotic stewardshipprograms have reduced CDAD rates. Current infectioncontrol practices remain in good standing but areonly effective if practiced. More research is needed toreduce transmission by C difficile spores, to find effec-tive disinfectants and treatments, and to find methodsto prevent CDAD outbreaks.

CONCLUSION

The emergence of a fluroroquinolone-resistant, hy-pervirulent strain of C difficile resulted in high numbersof patients with severe CDAD and caused several largeoutbreaks in Canada. These events have renewed focuson infection control practices for this disease. Severalrecommendations for infection control practitionersare suggested (Table 4). The occurrence of this strainis not unique to North America, and it should be ex-pected to appear at other health care-associated institu-tions. The need to determine the specific strain type ofC difficile isolates may be valuable in detecting thearrival of hypervirulent strains and tracking its trans-mission through health care facilities. However, this

requires that hospital laboratories isolate C difficile byculture and have strain-typing methods available.These strain-typing assays are not widely available,and greater accessibility is recommended. An increasein the frequency of severe cases of CDAD may heraldthe arrival of the BI/NAP1/027 strain at an institution.Infection control practitioners and clinicians need tobe vigilant for other strains of C difficile because a dif-ferent profile of patients with CDAD is becomingmore frequent (community-acquired cases, children,fluoroquinolone-associated cases) that may not be as-sociated with the emergent BI/NAP1/027 strain. Re-search into more effective treatments for CDAD andrecurrent CDAD is warranted. The increasing fre-quency of metronidazole failures is concerning, andrandomized clinical trials are needed to determinewhether this observation is due to a failure of the met-ronidazole itself or to confounders present in the ob-servational studies. It would be prudent for infectioncontrol practitioners to have established strategies inplace to control CDAD infections. Integrated infectioncontrol programs that involve educational programs,handwashing/gloving protocols, appropriate environ-mental disinfectants, and antibiotic stewardship pro-grams offer the most effective benefits.

References

1. Hayetian FD, Read TE, Brozovich M, Garvin RP, Caushaj PF. Ileal per-

foration secondary to Clostridium difficile enteritis: Report of 2 cases.

Arch Surg 2006;141:97-9.

2. Koss K, Clark MA, Sanders DS, Morton D, Keighley MR, Goh J. The

outcome of surgery in fulminant Clostridium difficile colitis. Colorectal

Dis 2006;8:149-54.

3. Wolf PL, Kasyan A. Images in clinical medicine: Pseudomembranous

colitis associated with Clostridium difficile. N Engl J Med 2005;353:

2491.

4. Gravisse J, Barnaud G, Hanau-Bercot B, Raskine L, Riahi J, Gaillard JL,

et al. Clostridium difficile brain empyema after prolonged intestinal

carriage. J Clin Microbiol 2003;41:509-11.

5. Al Eidan FA, McElnay JC, Scott MG, Kearney MP. Clostridium difficile-

associated diarrhoea in hospitalised patients. J Clin Pharm Ther 2000;

25:101-9.

6. Miller MA, Hyland M, Ofner-Agostini M, Gourdeau M, Ishak M.

Morbidity, mortality, and healthcare burden of nosocomial Clostridium

difficile-associated diarrhea in Canadian hospitals. Infect Control

Hosp Epidemiol 2002;23:137-40.

7. Kyne L, Hamel MB, Polavaram R, Kelly CP. Health care costs and

mortality associated with nosocomial diarrhea due to Clostridium

difficile. Clin Infect Dis 2002;34:346-53.

8. McFarland LV, Surawicz CM, Stamm WE. Risk factors for Clostridium

difficile carriage and C difficile-associated diarrhea in a cohort of hos-

pitalized patients. J Infect Dis 1990;162:678-84.

9. McFarland LV, Surawicz CM, Rubin M, Fekety R, Elmer GW, Green-

berg RN. Recurrent Clostridium difficile disease: Epidemiology and

clinical characteristics. Infect Control Hosp Epidemiol 1999;20:43-50.

10. Changela U, Cannon JP, Aneziokoro C, Shah PS, Thottapurathu L,

Lentino J. Risk factors and mortality associated with Clostridium diffi-

cile-associated diarrhoea at a VA hospital. Int J Antimicrob Agents

2004;24:562-6.

McFarland et al May 2007 249

11. Oldfield EC III. Clostridium difficile-associated diarrhea: Risk factors,

diagnostic methods, and treatment. Rev Gastroenterol Disord 2004;

4:186-95.

12. Pepin J, Valiquette L, Alary ME, Villemure P, Pelletier A, Forget K, et al.

Clostridium difficile-associated diarrhea in a region of Quebec from

1991 to 2003: A changing pattern of disease severity. CMAJ 2004;171:

466-72.

13. Perlin JB. Undersecretary for Health’s Information Letter. Clostridium

difficile. September 8, 2005. Available at: vaww1.va.gov/

vhapulbications/viewPublication.asp?pub_ID51305. Accessed Sep-

tember 2, 2005.

14. Archibald LK, Banerjee SN, Jarvis WR. Secular trends in hospital-

acquired Clostridium difficile disease in the United States, 1987-2001.

J Infect Dis 2004;189:1585-9.

15. McDonald LC, Owings M, Jernigan DB. Clostridium difficile infection in

patients discharged from US short-stay hospitals, 1996-2003. Emerg

Infect Dis 2006;3:409-15.

16. Centers for Disease Control and Prevention. Severe Clostridium diffi-

cile-associated disease in populations previously at low risk—Four

states, 2005. MMWR Morb Mortal Wkly Rep 2005;54:1201-5.

17. Frost F, Hurley JS, Petersen HV, Casciano RN. Estimated incidence of

Clostridium difficile infection. Emerg Infect Dis 1999;5:303-4.

18. Levy DG, Stergachis A, McFarland LV, Van Vorst K, Graham DJ, John-

son ES, et al. Antibiotics and Clostridium difficile diarrhea in the ambu-

latory care setting. Clin Ther 2000;22:91-102.

19. Dial S, Delaney JA, Barkun AN, Suissa S. Use of gastric acid-

suppressive agents and the risk of community-acquired Clostridium

difficile-associated disease. JAMA 2005;294:2989-95.

20. Noren T, Akerlund E, Back E, Sjoberg L, Persson I, Alriksson I, et al.

Molecular epidemiology of hospital-associated and community-

acquired Clostridium difficile infection in a Swedish country. J Clin

Microbiol 2004;8:3635-43.

21. Settle CD, Wilcox MH, Fawley WN, Corrado OJ, Hawkey PM.

Prospective study of the risk of Clostridium difficile diarrhoea in

elderly patients following treatment with cefotaxime or piperacillin-

tazobactam. Aliment Pharmacol Ther 1998;12:1217-23.

22. Blot E, Escande MC, Besson D, Barbut F, Granpeix C, Asselain B,

et al. Outbreak of Clostridium difficile-related diarrhoea in an adult on-

cology unit: Risk factors and microbiological characteristics. J Hosp

Infect 2003;53:187-92.

23. McFarland LV, Mulligan ME, Kwok RY, Stamm WE. Nosocomial

acquisition of Clostridium difficile infection. N Engl J Med 1989;320:

204-10.

24. Rotimi VO, Mokaddas EM, Jamal WY, Verghese TL, el Din K, Junaid

TA. Hospital-acquired Clostridium difficile infection amongst ICU and

burn patients in Kuwait. Med Princ Pract 2002;11:23-8.

25. Agarwal R, Gupta D, Ray P, Aggarwal AN, Jindal SK. Epidemiology,

risk factors and outcome of nosocomial infections in a respiratory

intensive care unit in North India. J Infect 2006;53:98-105.

26. Pepin J, Valiquette L, Cossette B. Mortality attributable to nosocomial

Clostridium difficile-associated disease during an epidemic caused by

a hypervirulent strain in Quebec. CMAJ 2005;173:1037-42.

27. Eggertson L. C difficile may have killed 2000 in Quebec: Study. CMAJ

2005;173:1020-1.

28. Loo VG, Poirier L, Miller MA, Oughton M, Libman MD, Michaud S,

et al. A predominantly clonal multi-institutional outbreak of Clostrid-

ium difficile-associated diarrhea with high morbidity and mortality.

N Engl J Med 2005;353:2442-9.

29. Rupnik M, Dupuy B, Fairweather NF, Gerding DN, Johnson S, Just I,

et al. Revised nomenclature of Clostridium difficile toxins and associ-

ated genes. J Med Microbiol 2005;54:113-7.

30. Warny M, Pepin J, Fang A, Killgore G, Thompson A, Brazier J, et al.

Toxin production by an emerging strain of Clostridium difficile associ-

ated with outbreaks of severe disease in North America and Europe.

Lancet 2005;366:1079-84.

31. McDonald LC, Killgore GE, Thompson A, Owens RC Jr, Kazakova

SV, Sambol SP, et al. An epidemic, toxin gene-variant strain of

Clostridium difficile. N Engl J Med 2005;353:2433-41.

32. McEllistrem MC, Carman RJ, Gerding DN, Genheimer CW, Zheng L.

A hospital outbreak of Clostridium difficile disease associated with iso-

lates carrying binary toxin genes. Clin Infect Dis 2005;40:265-72.

33. Ross E. British authorities probe hospital superbug. 2005. New York:

ABC News. Available at: www.abcnews.gocom/Health/print?id5

828758. Accessed April 28, 2006.

34. Krausz S, Bessems M, Boermeester MA, Kuijper EJ, Visser CE, Speel-

man P. Life-threatening infections with a new strain of Clostridium dif-

ficile. Ned Tijdschr Geneeskd 2005;149:2081-6.

35. Martirosian G, Szczesny A, Cohen SH, Silva J Jr. Analysis of Clostrid-

ium difficile-associated diarrhea among patients hospitalized in ter-

tiary care academic hospital. Diagn Microbiol Infect Dis 2005;52:

153-5.

36. Rupnik M, Kato N, Grabnar M, Kato H. New types of toxin A-neg-

ative, toxin B-positive strains among Clostridium difficile isolates from

Asia. J Clin Microbiol 2003;41:1118-25.

37. Geric B, Rupnik M, Gerding DN, Grabnar M, Johnson S. Distribution

of Clostridium difficile variant toxinotypes and strains with binary toxin

genes among clinical isolates in an American hospital. J Med Micro-

biol 2004;53:887-94.

38. Komatsu M, Kato H, Aihara M, Shimakawa K, Iwasaki M, Nagasaka Y,

et al. High frequency of antibiotic-associated diarrhea due to toxin

A-negative, toxin B-positive Clostridium difficile in a hospital in

Japan and risk factors for infection. Eur J Clin Microbiol Infect Dis

2003;22:525-9.

39. Cohen SH, Tang YJ, Hansen B, Silva J Jr. Isolation of a toxin B-defi-

cient mutant strain of Clostridium difficile in a case of recurrent

C difficile-associated diarrhea. Clin Infect Dis 1998;26:410-2.

40. Goh S, Chang BJ, Riley TV. Effect of phage infection on toxin produc-

tion by Clostridium difficile. J Med Microbiol 2005;54:129-35.

41. Popoff MR, Rubin EJ, Gill DM, Boquet P. Actin-specific ADP-ribosyl-

transferase produced by a Clostridium difficile strain. Infect Immun

1988;56:2299-306.

42. Barbut F, Decre D, Lalande V, Burghoffer B, Noussair L, Gigandon A,

et al. Clinical features of Clostridium difficile-associated diarrhoea due

to binary toxin (actin-specific ADP-ribosyltransferase)-producing

strains. J Med Microbiol 2005;54:181-5.

43. Goncalves C, Decre D, Barbut F, Burghoffer B, Petit JC. Prevalence

and characterization of a binary toxin (actin-specific ADP-ribo-

syltransferase) from Clostridium difficile. J Clin Microbiol 2004;42:

1933-9.

44. Alonso R, Martin A, Pelaez T, Marin M, Rodriguez-Creixems M,

Bouza E. Toxigenic status of Clostridium difficile in a large Spanish

teaching hospital. J Med Microbiol 2005;54:159-62.

45. Pituch H, Rupnik M, Obuch-Woszczatynski P, Grubesic A, Meisel-

Mikolajczyk F, Luczak M. Detection of binary-toxin genes (cdtA

and cdtB) among Clostridium difficile strains isolated from patients

with C difficile-associated diarrhoea (CDAD) in Poland. J Med

Microbiol 2005;54:143-7.

46. Geric B, Carman RJ, Rupnik M, Genheimer CW, Sambol SP, Lyerly

DM, et al. Binary toxin-producing, large clostridial toxin-negative

Clostridium difficile strains are enterotoxic but do not cause disease

in hamsters. J Infect Dis 2006;193:1143-50.

47. Akerlund T, Svenungsson B, Lagergren A, Burman LG. Correlation of

disease severity with fecal toxin levels in patients with Clostridium dif-

ficile-associated diarrhea and distribution of PCR ribotypes and toxin

yields in vitro of corresponding isolates. J Clin Microbiol 2006;44:

353-8.

48. McFarland LV, Elmer GW, Stamm WE, Mulligan ME. Correlation of

immunoblot type, enterotoxin production, and cytotoxin production

with clinical manifestations of Clostridium difficile infection in a cohort

of hospitalized patients. Infect Immun 1991;59:2456-62.

250 Vol. 35 No. 4 McFarland et al

49. Terhes G, Urban E, Soki J, Hamid KA, Nagy E. Community-acquired

Clostridium difficile diarrhea caused by binary toxin, toxin A, and toxin

B gene-positive isolates in Hungary. J Clin Microbiol 2004;42:4316-8.

50. Andrews CN, Raboud J, Kassen BO, Enns R. Clostridium difficile-

associated diarrhea: Predictors of severity in patients presenting to

the emergency department. Can J Gastroenterol 2003;17:369-73.

51. Berrington A. National Clostridium difficile Standards Group: Report

to the Department of Health. J Hosp Infect 2004;56(Suppl 1):1-38.

52. Simor AE, Bradley SF, Strausbaugh LJ, Crossley K, Nicolle LE. SHEA

Position Paper. Clostridium difficile in long-term-care facilities for the

elderly. Infect Control Hosp Epidemiol 2002;23:696-703.

53. Gaynes R, Rimland D, Killum E, Lowery HK, Johnson TM, Killgore G,

et al. Outbreak of Clostridium difficile infection in a long-term care facil-

ity: Association with gatifloxacin use. Clin Infect Dis 2004;38:640-5.

54. McFarland LV, Brandmarker SA, Guandalini S. Pediatric Clostridium dif-

ficile: A phantom menace or clinical reality? J Pediatr Gastroenterol

Nutr 2000;31:220-31.

55. McFarland LV, Surawicz CM. Clostridium difficile disease. In: McDonald

JWD, Burroughs AK, editors. Evidence-based gastroenterology and

child health. London: BMJ Books; 2004. p. 285-301.

56. Mohan SS, McDermott BP, Parchuri S, Cunha BA. Lack of value of repeat

stool testing for Clostridium difficile toxin. Am J Med 2006;119:356-8.

57. Iizuka M, Konno S, Itou H, Chihara J, Toyoshima I, Horie Y, et al.

Novel evidence suggesting Clostridium difficile is present in human

gut microbiota more frequently than previously suspected. Microbiol

Immunol 2004;48:889-92.

58. Tang-Feldman Y, Mayo S, Silva JJ Jr, Cohen SH. Molecular analysis of

Clostridium difficile strains isolated from 18 cases of recurrent Clostrid-

ium difficile-associated diarrhea. J Clin Microbiol 2003;41:3413-4.

59. van den Berg RJ, Ameen HA, Furusawa T, Claas EC, van der Vorm

ER, Kuijper EJ. Coexistence of multiple PCR-ribotype strains of Clos-

tridium difficile in faecal samples limits epidemiological studies. J Med

Microbiol 2005;54:173-9.

60. Arroyo LG, Kruth SA, Willey BM, Staempfli HR, Low DE, Weese JS.

PCR ribotyping of Clostridium difficile isolates originating from human

and animal sources. J Med Microbiol 2005;54:163-6.

61. Madewell BR, Bea JK, Kraegel SA, Winthrop M, Tang YJ, Silva J Jr.

Clostridium difficile: A survey of fecal carriage in cats in a veterinary

medical teaching hospital. J Vet Diagn Invest 1999;11:50-4.

62. Lefebvre SL, Waltner-Toews D, Peregrine AS, Reid-Smith R, Hodge L,

Arroyo LG, et al. Prevalence of zoonotic agents in dogs visiting hos-

pitalized people in Ontario: Implications for infection control. J Hosp

Infect 2006;62:458-66.

63. Lefebvre SL, Arroyo LG, Weese JS. Epidemic Clostridium difficile strain

in hospital visitation dog. Emerg Infect Dis 2006;12:1036-7.

64. Morinville V, McDonald J. Clostridium difficile-associated diarrhea in

200 Canadian children. Can J Gastroenterol 2005;19:497-501.

65. Gogate A, De A, Nanivadekar R, Mathur M, Saraswathi K, Jog A,

et al. Diagnostic role of stool culture and toxin detection in antibiotic

associated diarrhoea due to Clostridium difficile in children. Indian J

Med Res 2005;122:518-24.

66. CunninghamR,DaleB,UndyB,GauntN.Protonpump inhibitors as a risk

factor for Clostridium difficile diarrhoea. J Hosp Infect 2003;54:243-5.

67. Shah S, Lewis A, Leopold D, Dunstan F, Woodhouse K. Gastric acid

suppression does not promote clostridial diarrhoea in the elderly.

QJM 2000;93:175-81.

68. Dial S, Alrasadi K, Manoukian C, Huang A, Menzies D. Risk of Clostridium

difficile diarrhea among hospital inpatients prescribed proton pump

inhibitors: Cohort and case-control studies. CMAJ 2004;171:33-8.

69. Al Tureihi FI, Hassoun A, Wolf-Klein G, Isenberg H. Albumin, length

of stay, and proton pump inhibitors: Key factors in Clostridium difficile-

associated disease in nursing home patients. J Am Med Dir Assoc

2005;6:105-8.

70. Pepin J, Saheb N, Coulombe MA, Alary ME, Corriveau MP, Authier S,

et al. Emergence of fluoroquinolones as the predominant risk factor

for Clostridium difficile-associated diarrhea: A cohort study during an

epidemic in Quebec. Clin Infect Dis 2005;41:1254-60.

71. Modena S, Bearelly D, Swartz K, Friedenberg FK. Clostridium difficile

among hospitalized patients receiving antibiotics: A case-control

study. Infect Control Hosp Epidemiol 2005;26:685-90.

72. Linder JA, Huang ES, Steinman MA, Gonzales R, Stafford RS.

Fluoroquinolone prescribing in the United States: 1995 to 2002.

Am J Med 2005;118:259-68.

73. McFarland LV, Bauwens JE, Melcher SA, Surawicz CM, Greenberg

RN, Elmer GW. Ciprofloxacin-associated Clostridium difficile disease.

Lancet 1995;346:977-8.

74. Yip C, Loeb M, Salama S, Moss L, Olde J. Quinolone use as a risk fac-

tor for nosocomial Clostridium difficile-associated diarrhea. Infect

Control Hosp Epidemiol 2001;22:572-5.

75. McCusker ME, Harris AD, Perencevich E, Roghmann MC. Fluoroqui-

nolone use and Clostridium difficile-associated diarrhea. Emerg

Infect Dis 2003;9:730-3.

76. Muto CA, Pokrywka M, Shutt K, Mendelsohn AB, Nouri K, Posey K,

et al. A large outbreak of Clostridium difficile-associated disease with

an unexpected proportion of deaths and colectomies at a teaching

hospital following increased fluoroquinolone use. Infect Control

Hosp Epidemiol 2005;26:273-80.

77. Rexach CE, Yajarayma J, Tang-Feldman MS, Cohen SH. Spatial and

temporal analysis of Clostridium difficile infection in patients at a pedi-

atric hospital in California. Infect Control Hosp Epidemiol 2005;26:

691-6.

78. van den Berg RJ, Claas EC, Oyib DH, Klaassen CH, Dijkshoorn L,

Brazier JS, et al. Characterization of toxin A-negative, toxin B-posi-

tive Clostridium difficile isolates from outbreaks in different countries

by amplified fragment length polymorphism and PCR ribotyping.

J Clin Microbiol 2004;42:1035-41.

79. McCoubrey J, Starr J, Martin H, Poxton IR. Clostridium difficile in a ger-

iatric unit: A prospective epidemiological study employing a novel

S-layer typing method. J Med Microbiol 2003;52:573-8.

80. Wullt M, Burman LG, Laurell MH, Akerlund T. Comparison of AP-

PCR typing and PCR-ribotyping for estimation of nosocomial

transmission of Clostridium difficile. J Hosp Infect 2003;55:

124-30.

81. Kato H, Yokoyama T, Arakawa Y. Typing by sequencing the slpA gene

of Clostridium difficile strains causing multiple outbreaks in Japan.

J Med Microbiol 2005;54:167-71.

82. Struelens MJ, Maas A, Nonhoff C, Deplano A, Rost F, et al. Control

of nosocomial transmission of Clostridium difficile based on sporadic

case surveillance. Am J Med 1991;91:S138-44.

83. Johnson S, Gerding DN, Olson MM, Weiler MD, Hughes RA, et al.

Prospective, controlled study of vinyl glove use to interrupt

Clostridium difficile nosocomial transmission. Am J Med 1990;88:

137-40.

84. Pittet D, Hugonnet S, Harbarth S, Mourouga P, Sauvan V, et al. Effec-

tiveness of a hospital-wide programme to improve compliance

with hand hygiene. Infection Control Programme. Lancet 2000;356:

2196.

85. Johnson S, Gerding DN. Clostridium difficile-associated diarrhea.

Clin Infect Dis 1998;26:1027-34.

86. Bettin K, Clabots C, Mathie P, Willard K, Gerding DN. Effectiveness

of liquid soap vs. chlorohexidine gluconate for the removal of Clos-

tridium difficile from bare hands and gloved hands. Infect Cont

Hosp Epidemiol 1994;15:697-702.

87. Boyce JM, Pittet D. Guideline for hand hygiene in health-care

settings: Recommendations of the Healthcare Infection Control

Practices Advisory Committee and the HICPAC/SHEA/APIC/IDSA

Hand Hygiene Task Force. Infect Control Hosp Epidemiol 2002;

23(Suppl 12):S3-40.

88. Salemi C, Canola MT, Eck EK. Handwashing and physicians How to

get them together. Infect Cont Hosp Epidemiol 2002;23:32-5.

McFarland et al May 2007 251

89. Wilcox MH, Fawley WN, Wigglesworth N, Parnell P, Verity P,

Freeman J. Detergent versus hypochlorite cleaning and Clostridium

difficile infection. J Hosp Infect 2004;56:331.

90. Kaatz GW, Gitlin SD, Schaberg DR, Wilson KH, Kauffman CA, et al.

Acquisition of Clostridium difficile from the hospital environment. Am

J Epidemiol 1988;127:1289-94.

91. Rutala WA, Gergen MF, Weber DJ. Inactivation of Clostridium difficile

spores by disinfectants. Infect Control Hosp Epidemiol 1993;14:36-9.

92. Wilcox MH, Fawley WN. Hospital disinfectants and spore formation

by Clostridium difficile. Lancet 2000;356:1324.

93. Mayfield JL, Leet T, Miller J, Mundy LM. Environmental control to reduce

transmission of Clostridium difficile. Clin Infect Dis 2000;31:995-1000.

94. Hota B. Contamination, disinfection, and cross-colonization: Are

hospital surfaces reservoirs for nosocomial infection? Clin Infect

Dis 2004;39:1182-9.

95. Jernigan JA, Siegman-Igra Y, Guerrant RC, Farr BM. A randomized

crossover study of disposable thermometers for prevention of Clos-

tridium difficile and other nosocomial infections. Infect Control Hosp

Epidemiol 1998;19:494-9.

96. Brooks SE, Veal RO, Kramer M, Dore L, Schupf N, et al. Reduction in

the incidence of Clostridium difficile-associated diarrhea in an acute

care hospital and a skilled nursing facility following replacement

of electronic thermometers with single-use disposables. Infect

Control Hosp Epidemiol 1992;13:98-103.

97. Boone N, Eagan J, Gillern P, Armstrong D, Sepkovitz KA. Evaluation

of an interdisciplinary re-isolation policy for patients with previous

Clostridium difficile diarrhea. Am J Infect Control 1998;26:584-7.

98. Verity P, Wilcox MH, Fawley W, Parnell P. Prospective evaluation of

environmental contamination by Clostridium difficile in isolation side

rooms. J Hosp Infect 2001;49:204-9.

99. Johnson S, Homann SR, Bettin KM, Quick JN, Clabots CR, et al.

Treatment of asymptomatic Clostridium difficile carriers (fecal excre-

tors) with vancomycin or metronidazole: A randomized, placebo-

controlled trial. Ann Intern Med 1992;117:297-302.

100. Bender BS, Bennett R, Laughon BE, Greenough WE III, Gaydos C, et al. Is

Clostridium difficile endemic in chronic-care facilties? Lancet 1986;2:

11-3.

101. Kerr RB, McLaughlin DI, Sonnenberg LW. Control of Clostridium dif-

ficile colitis outbreak by treating asymptomatic carriers with metro-

nidazole. Am J Infect Control 1990;18:332-5.

102. Pear SM, Williamson TH, Bettin KM, Gerding DN, Galgiani JN.

Decrease in nosocomial Clostridium difficile-associated diarrhea by

restricting clindamycin use. Ann Intern Med 1994;120:272-7.

103. Climo MW, Israel DS, Wong ES, Williams D, Coudron P, Markovitz

SM. Hospital-wide restriction of clindamycin: Effect on the incidence

of Clostridium difficile-associated diarrhea and cost. Ann Intern Med

1998;128:989-95.

104. Ho M, Yang D, Wyle FA, Mulligan ME. Increased incidence of Clostrid-

ium difficile-associated diarrhea following decreased restriction of

antibiotic use. Clin Infect Dis 1996;23:S102-6.

105. McNulty C, Logan M, Donald IP, Ennis D, Taylor D, Balwin RN, et al. Suc-

cessful control of Clostridium difficile infection in an elderly care unit

through use of a restrictive antibiotic policy. J Anitmicrob Chemother

1997;40:707-11.

106. Ludlam H, Brown N, Sule O, Redpath C, Coni N, Owen G. An an-

tibiotic policy associated with reduced risk of Clostridium difficile-asso-

ciated diarrhoea. Age Ageing 1999;28:578-80.

107. Carling P, Fung T, Killion AN, Terrin N, Barza M. Favorable impact of

a multidisciplinary antibiotic management program conducted during

7 years. Infect Control Hosp Epidemiol 2003;24:699-706.

108. Khan R, Cheesbrough J. Impact of changes in antibiotic policy on Clos-

tridium difficile-associated diarrhoea (CDAD) over a five-year period

in a district general hospital. J Hosp Infect 2003;54:104-8.

109. Thomas C, Stevenson M, Williamson DJ, Riley TV. Clostridium difficile-

associated diarrhea: Epidemiological data from western Australia

associated with a modified antibiotic policy. Clin Infect Dis 2002;

35:1457-62.

110. Stone SP, Beric V, Quick A, Balestrini AA, Kibbler CC. The effect of

an enhanced infection-control policy on the incidence of Clostridium

difficile infection and methicillin-resistant Staphylococcus aureus coloni-

zation in acute elderly medical patients. Age Ageing 1998;27:561-8.

111. Zafar AB, Gaydos LA, Furlong WB, Nguyen MH, Mennonna PA.

Effectiveness of infection control program in controlling nosocomial

Clostridium difficile. Am J Infect Control 1998;26:588-93.

112. Apisarnthanarak A, Zack JE, Mayfield JL, Freeman J, Dunne WM, Little JR,

et al. Effectiveness of environmental and infection control programs to

reduce transmission of Clostridium difficile. Clin Infect Dis 2004;39:601-2.

113. Beaulieu M, Thirion DJ, Williamson D, Pichette G. Clostridium difficile-

associated diarrhea outbreaks: The name of the game is isolation and

cleaning. Clin Infect Dis 2006;42:725-9.

114. Weiss K. Poor infection control, not fluoroquinolones, likely to be

primary cause of Clostridium difficile-associated diarrhea outbreaks

in Quebec. Clin Infect Dis 2006;42:725-7.

115. Pepin J, Routhier S, Gagnon S, Brazeau I. Management and outcomes

of a first recurrence of Clostridium difficile-associated disease in

Quebec, Canada. Clin Infect Dis 2006;42:758-64.

116. InstitutNationalDeSantePubliqueDuQuebec. Surveillancedesdiarrhees

associees a Clostridium difficile au Quebec. March 31, 2005. Available

at: http://www.inspq.qc.ca/publications/defaultlien.asp?E5e&submit51&

NumPublication5389. Accessed January 24, 2006.

117. Gordin FM, Schultz ME, Huber RA, Gill JA. Reduction in nosocomial

transmission of drug-resistant bacteria after introduction of an alco-

hol-based handrub. Infect Control Hosp Epidemiol 2005;26:650-3.

118. O’Connor KA, Kingston M, O’Donovan M, Cryan B, Twomey C,

O’Mahony D. Antibiotic prescribing policy and Clostridium difficile

diarrhoea. QJM 2004;97:423-9.

119. Wilcox MH, Freeman J, Fawley W, MacKinlay S, Brown A, Donaldson

K, et al. Long-term surveillance of cefotaxime and piperacillin-tazo-

bactam prescribing and incidence of Clostridium difficile diarrhoea.

J Antimicrob Chemother 2004;54:168-72.

120. Barrett S. Clostridium difficile—the next step in mandatory reporting.

J Hosp Infect 2004;56:83-4.

121. Tsan L, Hojlo C, Kearns MA, Davis C, Langberg R, Claggett M, et al.

Infection surveillance and control programs in the Department of

Veterans Affairs nursing home care units: A preliminary assessment.

Am J Infect Control 2006;34:80-3.

122. Zoutman DE, Ford BD. The relationship between hospital infection

surveillance and control activities and antibiotic-resistant pathogen

rates. Am J Infect Control 2005;33:1-5.

123. Pituch H, Brazier JS, Obuch-Woszczatynski P, Wultanska D, Meisel-

Mikolajczyk F, Luczak M. Prevalence and association of PCR ribotypes

of Clostridium difficile isolated from symptomatic patients from War-

saw with macrolide-lincosamide-streptogramin B (MLSB) type resis-

tance. J Med Microbiol 2006;55:207-13.

124. Samore MH, Venkataraman L, DeGirolami PC, Merrigan MM,

Johnson S, Gerding DN, et al. Genotypic and phenotypic analysis

of Clostridium difficile correlated with previous antibiotic exposure.

Microb Drug Resist 2006;12:23-8.

125. Boyce JM, Ligi C, Kohan C, Dumigan D, Havill NL. Lack of assciation

between the increased incidence of Clostridium difficile-associated

disease and the increasing use of alcohol-based hand rubs. Infect

Control Hosp Epidemiol 2006;27:479-83.

126. Fawley WN, Parnell P, Verity P, Freeman J, Wilcox MH. Molecular

epidemiology of endemic Clostridium difficile infection and the sig-

nificance of subtypes of the United Kingdom epidemic strain (PCR

ribotype 1). J Clin Microbiol 2005;43:2685-96.

127. Sehulster L, Chinn RY. Guidelines for environmental infection control

in health-care facilities: Recommendations of CDC and the Health-

care Infection Control Practices Advisory Committee (HICPAC).

MMWR Recomm Rep 2003;52(RR-10):1-42.

252 Vol. 35 No. 4 McFarland et al

128. Wilcox MH, Fawley WN, Wigglesworth N, Parnell P, Verity P, Free-

man J. Comparison of the effect of detergent versus hypochlorite

cleaning on environmental contamination and incidence of Clostridium

difficile infection. J Hosp Infect 2003;54:109-14.

129. Block C. The effect of Perasafe and sodium dichloroisocyanurate

(NaDCC) against spores of Clostridium difficile and Bacillus atrophaeus

on stainless steel and polyvinyl chloride surfaces. J Hosp Infect 2004;

57:144-8.

130. Perez J, Springthorpe VS, Sattar SA. Activity of selected oxidizing

microbicides against the spores of Clostridium difficile: Relevance to

environmental control. Am J Infect Control 2005;33:320-5.

131. Fordtran JS. Colitis due to Clostridium difficile toxins: Underdiagnosed,

highly virulent, and nosocomial. Proc (Bayl Univ Med Cent) 2006;19:

3-12.

132. Peck MW, Granum PE, Gould GW, Mainil JG. Chapter 4. Foodborne

clostridia and sporulation. 2004. Available at: www.genusclostridium.

net. Accessed April 28, 2006.

133. Wilson KH, Kennedy MJ, Fekety FR. Use of sodium taurocholate

to enhance spore recovery on a medium selective for Clostridium

difficile. J Clin Microbiol 1982;15:443-6.

134. Mukherjee K, Karlsson S, Burman LG, Akerlund T. Proteins released

during high toxin production in Clostridium difficile. Microbiology

2002;148:2245-53.

135. Blocher JC, Busta FF. Inhibition of germinant binding by bacterial

spores in acidic environments. Appl Environ Microbiol 1985;50:

274-9.

136. Wilson KH, Sheagren JN, Freter R. Population dynamics of ingested

Clostridium difficile in the gastrointestinal tract of the Syrian hamster.

J Infect Dis 1985;151:355-61.

137. McFarland LV. Normal flora: Diversity and functions. Microb Ecology

Health Dis 2000;12:193-207.

138. Freeman J, Wilcox MH. The effects of storage conditions on viability

of Clostridium difficile vegetative cells and spores and toxin activity

in human faeces. J Clin Pathol 2003;56:126-8.

139. Kamiya S, Yamakawa K, Ogura H, Nakamura S. Recovery of spores

of Clostridium difficile altered by heat or alkali. J Med Microbiol 1989;

28:217-21.

140. Nakamura S, Yamakawa K, Izumi J, Nakashio S, Nishida S. Germina-

bility and heat resistance of spores of Clostridium difficile strains.

Microbiol Immunol 1985;29:113-8.

141. Kim KH, Fekety R, Batts DH, Brown D, Cudmore M, Silva J Jr, et al.

Isolation of Clostridium difficile from the environment and contacts of

patients with antibiotic-associated colitis. J Infect Dis 1981;143:42-50.

142. Underwood S, Stephenson K, Fawley WN, Freeman J, Baines S,

Owens RC, et al. Effects of hospital cleaning agents on spore forma-

tion by US and UK outbreak Clostridium difficile strains (abstract).

158th Meeting Society for General Microbiology April 3, 2006. Avail-

able at: www.sgm.ac.uk. Accessed May 1, 2006.

143. McFarland LV, Elmer GW, Surawicz CM. Breaking the cycle: Treat-

ment strategies for 163 cases of recurrent Clostridium difficile disease.

Am J Gastroenterol 2002;97:1769-75.

144. Noren T, Tang-Feldman YJ, Cohen SH, Silva J Jr, Olcen P. Clindamycin

resistant strains of Clostridium difficile isolated from cases of C difficile

associated diarrhea (CDAD) in a hospital in Sweden. Diagn Microbiol

Infect Dis 2002;42:149-51.

145. Pituch H, Obuch-Woszczatynski P, Glinka D, Lazinska B, Meisel-

Mikolajczyk F, Luczak M. Assessment of susceptibility to metroni-

dazole and vancomycin of Clostridium difficile strains isolated

between 1998-2002. Med Dosw Mikrobiol 2003;55:253-8.

146. Johnson S, Samore MH, Farrow KA, Killgore GE, Tenover FC, Lyras

D, et al. Epidemics of diarrhea caused by a clindamycin-resistant

strain of Clostridium difficile in four hospitals. N Engl J Med 1999;

341:1645-51.

147. Ackermann G, Tang YJ, Kueper R, Heisig P, Rodloff AC, Silva J Jr, et al.

Resistance to moxifloxacin in toxigenic Clostridium difficile isolates is

associated with mutations in gyrA. Antimicrob Agents Chemother

2001;45:2348-53.

148. Fishman N. Antimicrobial stewardship. Am J Med 2006;119:S53-61.

149. Davey P, Brown E, Fenelon L, Finch R, Gould I, Hartman G, et al.

Interventions to improve antibiotic prescribing practices for hospital

inpatients. Cochrane Database Syst Rev 2005;4:CD003543.

150. Aboudola S, Kotloff KL, Kyne L, et al. Clostridium difficile vaccine

and serum immunoglobulin G antibody response to toxin A. Infect

Immun 2003;71:1608-10.

151. Sougioultzis S, Kyne L, Drudy D, Keates S, Maroo S, Pothoulakis C,

et al. Clostridium difficile toxoid vaccine in recurrent C difficile-associ-

ated diarrhea. Gastroenterology 2005;128:764-70.

152. Gerding DN. Metronidazole for Clostridium difficile-associated dis-

ease: Is it okay for mom? Clin Infect Dis 2005;40:1598-600.

153. Musher DM, Aslam S, Logan N, Nallacheru S, Bhaila I, Borchert F,

et al. Relatively poor outcome after treatment of Clostridium

difficile colitis with metronidazole. Clin Infect Dis 2005;40:

1586-1590.

154. Pepin J, Alary ME, Valiquette L, Raiche E, Ruel J, Fulop K, et al. In-

creasing risk of relapse after treatment of Clostridium difficile colitis

in Quebec, Canada. Clin Infect Dis 2005;40:1591-7.

155. Fernandez A, Anand G, Friedenberg F. Factors associated with failure

of metronidazole in Clostridium difficile-associated disease. J Clin Gas-

troenterol 2004;38:414-8.

156. Modena S, Gollamudi S, Friedenberg F. Continuation of antibiotics

is associated with failure of metronidazole for Clostridium difficile-

associated diarrhea. J Clin Gastroenterol 2006;40:49-54.

157. Drummond LJ, McCoubrey J, Smith DG, Starr JM, Poxton IR.

Changes in sensitivity patterns to selected antibiotics in Clostridium

difficile in geriatric in-patients over an 18-month period. J Med Micro-

biol 2003;52:259-63.

158. Bishara J, Bloch Y, Garty M, Behor J, Samra Z. Antimicrobial resis-

tance of Clostridium difficile isolates in a tertiary medical center, Israel.

Diagn Microbiol Infect Dis 2006;54:141-4.

159. Pelaez T, Alcala L, Alonso R, Rodriguez-Creixems M, Garcia-Lechuz

JM, Bouza E. Reassessment of Clostridium difficile susceptibility to met-

ronidazole and vancomycin. Antimicrob Agents Chemother 2002;46:

1647-50.

160. Johnson S, Sanchez JL, Gerding DN. Metronidazole resistance in

Clostridium difficile. Clin Infect Dis 2000;31:625-6.

161. Freeman J, Stott J, Baines SD, Fawley WN, Wilcox MH. Surveillance

for resistance to metronidazole and vancomycin in genotypically dis-

tinct and UK epidemic Clostridium difficile isolates in a large teaching

hospital. J Antimicrob Chemother 2005;56:988-9.

162. McFarland LV. Alternative treatments for Clostridium difficile disease:

What really works? J Med Microbiol 2005;54:101-1.

163. Fedorak RN, Madsen KL. Probiotics and prebiotics in gastrointestinal

disorders. Curr Opin Gastroenterol 2004;20:146-55.

164. Surawicz CM, McFarland LV, Greenberg RN, Rubin M, Fekety R,

Mulligan ME, et al. The search for a better treatment for recurrent

Clostridium difficile disease: Use of high-dose vancomycin combined

with Saccharomyces boulardii. Clin Infect Dis 2000;31:1012-7.

165. McFarland LV. Meta-analysis of probiotics for the prevention of anti-

biotic associated diarrhea and the treatment of Clostridium difficile

disease. Am J Gastroenterol 2006;101:812-22.

166. Lewis S, Burmeister S, Brazier J. Effect of the prebiotic oligofructose

on relapse of Clostridium difficile-associated diarrhea: A randomized,

controlled study. Clin Gastroenterol Hepatol 2005;3:442-8.

167. Beales IL. Intravenous immunoglobulin for recurrent Clostridium diffi-

cile diarrhoea. Gut 2002;51:456.

168. van Dissel JT, de Groot N, Hensgens CM, Numan S, Kuijper EJ,

Veldkamp P, et al. Bovine antibody-enriched whey to aid in the

prevention of a relapse of Clostridium difficile-associated diarrhoea:

Preclinical and preliminary clinical data. J Med Microbiol 2005;54:

197-205.

McFarland et al May 2007 253

169. Wilcox MH. Descriptive study of intravenous immunoglobulin for

the treatment of recurrent Clostridium difficile diarrhoea. J Antimicrob

Chemother 2004;53:882-4.

170. McPherson S, Rees CJ, Ellis R, Soo S, Panter SJ. Intravenous im-

munoglobulin for the treatment of severe, refractory, and recur-

rent Clostridium difficile diarrhea. Dis Colon Rectum 2006;49:

640-5.

171. Braunlin W, Xu Q, Hook P, Fitzpatrick R, Klinger JD, Burrier R, et al.

Toxin binding of tolevamer, a polyanionic drug that protects against

antibiotic-associated diarrhea. Biophys J 2004;87:534-9.

172. Davidson D, Peppe J, Louie T. A phase 2 study of the toxin binding

polymer Tolevamer in patients with Clostridium difficile-associated

diarrhea. First International Clostridium difficile Symposium, Slovenia,

May 5-8, 2004.

Screening for pulmonary tuberculosisusing chest radiography in newemployees in an industrial parkin TaiwanShih-Bin Su, MD, MS,a,b Chien-Fang Chiu, MD,a Cheng-Ta Chang, MD,c Kow-Tong Chen, MD, PhD,d

Ching-Yih Lin, MD,e and How-Ran Guo, MD, MPH, ScDb,f

Tainan, Taiwan

Background: Pulmonary tuberculosis (TB) is prevalent in Taiwan, but it is suspected that its occurrence has been underestimatedby the National TB Surveillance Program. A pre-employment health examination is mandated by law in Taiwan, providing a mech-anism to assess the occurrence of TB more accurately.Methods: A pre-employment TB screening program of an industrial park was used to evaluate the performance of the National TBSurveillance Program in Taiwan. The yields of the pre-employment TB screening, using chest radiography from July 2004 to June2005, were compared with corresponding results of the National TB Surveillance Program.Results: A total of 17,105 new employees with an even gender distribution (men:women ratio, 50.2%:49.8%) underwent screeningduring the study period. Among the participants, 22 (128.6 per 100,000) new patients with pulmonary TB were diagnosed, and7 (31.8%) of the patients had positive bacteriology findings. Compared with the results of the National TB Surveillance Program,the pre-employment screening had a much higher yield (128 vs. 47 per 100,000, P , .001).Conclusion: The yield of the active surveillance program through mandatory screening was much higher than that of the NationalTB Surveillance Program, which is a passive reporting system. The results of this study highlight the need for more active TB sur-veillance efforts in endemic areas like Taiwan. (Am J Infect Control 2007;35:254-9.)

Pulmonary tuberculosis (TB) has been prevalent inTaiwan for a substantial period of time. The mortalityrelated to TB was reported to be as high as 294.4 per100,000 individuals in 19471 but has declined gradu-ally and reached a historical low of 4.23 per 100,000in 2003.2 Despite this decline in the mortality, TB wasstill listed as the thirteenth leading cause of death inTaiwan as of 2004.2 A nationwide survey revealedthat the prevalence of bacteriologically confirmed pul-monary TB was 1.02% in 1947, and this figure declinedto 0.06% in 1993. On the other hand, owing to the

From the Tainan Science-Based Industrial Park Clinic,a Chi-Mei MedicalCenter; Department of Environmental and Occupational Health,b

College of Medicine, National Cheng Kung University; Chest Hospital,Department of Healthc; Graduate Institute of Public Health,d Collegeof Medicine, National Cheng Kung University; Department of InternalMedicine,e Chi-Mei Medical Center; and Department of Occupationaland Environmental Medicine,f National Cheng Kung University Hospital,Tainan, Taiwan.

Address correspondence to How-Ran Guo, MD, MPH, ScD, Depart-ment of Environmental and Occupational Health, National ChengKung University College of Medicine, 138 Sheng-Li Road, Tainan70428, Taiwan. E-mail: hrguo@mail.ncku.edu.tw.

0196-6553/$32.00

Copyright ª 2007 by the Association for Professionals in InfectionControl and Epidemiology, Inc.

doi:10.1016/j.ajic.2006.04.211

254

improvement in the notification mechanism, the num-ber of TB cases reported to the National TB SurveillanceProgram, which is operated by the Department ofHealth of the central government, has increased in re-cent years. Such an improvement was achieved by theimplementation of the ‘‘no-report-no-reimbursement’’and ‘‘notification-fee’’ policies.3 In 1997, the number ofnewly diagnosed cases reached 15,386, with a reportedincidence rate of 71.1 per 100,000 individuals, and thenumber of newly diagnosed cases remained at approx-imately 15,000 for each year since that time.4

Even so, it has been suspected that the National TBSurveillance Program has underestimated the occurrenceof TB in Taiwan because the program assesses the occur-rence on the basis of people who participate in varioushealth checkups and people who seek medical help,mostly because of clinical symptoms and signs. In fact,2 studies have been conducted to evaluate the occurrenceof pulmonary TB among certain Taiwanese subpopula-tions using active case-ascertainment approaches: 1 onprison inmates5 and the other on military conscripts.6

Whereas both studies reported higher estimates of theoccurrence of TB than the results from the National TBSurveillance Program for corresponding time periods,they were both conducted among very specific subpopu-lations; inmates are at a higher risk of acquiring pulmo-nary TB because prisons are crowded confined spaces,and all military conscripts are 19-year-old men.

Su et al May 2007 255

Therefore, further studies are needed to evaluate the pos-sible underestimation for the whole population.

A preemployment health examination for everynew employee is mandated by law in Taiwan. Whereasthe main purpose of this examination is to determinewhether the newly employed worker is fit for the job,the examination canalso serve as a screening forcommu-nicable diseases to prevent the spread of such diseaseswithin the workplace. Chest radiography (CXR) is a re-quired item of the examination and can serve as screen-ing for pulmonary TB. In an industrial park in Taiwan,new employees receive the examinations at the clinic lo-cated within the park, and such a scenario offers a uniqueopportunity to assess the occurrence of pulmonary TB ina working population, which can be adopted as a modelto evaluate whether the National TB Surveillance Pro-gram has underestimated the occurrence of pulmonaryTB. We analyzed the data from the preemploymentscreening and compared those with the correspondingdata obtainedfrom the National TB Surveillance Program.

METHODS

We included all new employees who underwent thestandard preemployment health examination duringthe period between July 2004 and June 2005 in anindustrial park in Tainan, Taiwan. All the examinationswere carried out at the clinic in the park, which is theonly health care institute located in the park. Accordingto certain regulations of the Taiwanese government,every new employee of a company must undergo astandard health examination prior to commencingwork with that company. This requisite health exami-nation includes a standard 14 3 14-inch CXR film,routine urine sample, complete blood cell count, bloodbiochemistry, and a general physical examination.Whereas the main purpose of the examination is to de-termine whether the worker is fit for the job, the CXRalso serves as screening for pulmonary TB so as to pre-vent the spread of this disease within the workplace. Asis the case for the National TB Surveillance Program,participants with suspicious lesions on the CXR filmswere referred for further evaluation to a specialist hos-pital, the Chest Hospital of the Department of Health(also known as the Tainan TB Center) in this case. Thediagnosis of TB was made according to the guidelinesimplemented by the Taiwan Center for Disease Control(CDC) of the Department of Health, which include fol-low-up CXR, sputum smears for acid-fast bacilli (AFB),and culture for Mycobacterium tuberculosis.1 The finaldiagnosis was made during the weekly staff conferenceafter a review of results of the examinations and wasconfirmed by the TB specialists of the Taiwan CDC.The Taiwan CDC has adopted the World Health Organi-zation’s case definitions of TB in general, although,

because Taiwan is an endemic area of TB and becausemany physicians are experienced in the diagnosis ofTB, a more aggressive approach to the diagnosis ofTB is applied to minimize the spread of the diseasethroughout Taiwan. Therefore, the diagnosis of TB isconfirmed for a culture-negative patient who has CXRfindings compatible with active TB but fails to respondclinically to a 1-week-treatment regimen with broad-spectrum antibiotics and if the physician deems itappropriate to administer a full course of anti-TB treat-ment after the treatment with antibiotics.1 In addition,cases of nontuberculous mycobacteria were excludedas cases of TB.

The extent of the patient’s disease was classifiedas ‘‘minimal’’ (pulmonary involvement less than thelung parenchyma area demarcated by a horizontalline extending from along the lower end of the secondrib), ‘‘moderately advanced’’ (pulmonary involvementof less than a single lung parenchyma), ‘‘far-advanced’’(pulmonary involvement of more than a single lungparenchyma), and ‘‘cavitary’’ (cavity formation in lungparenchyma).1 All detected cases of TB were reportedto the National TB Surveillance Program and treatedwith a rifampin-based, short-course anti-TB regimen.All suspicious cases underwent follow-up examinationsand were required to complete the treatment course. Toprevent the introduction of TB into the workplace, newemployees with TB were not allowed to enter the work-place before the completion of their treatment.

The yield of a screening program was calculatedusing the following equation7:

yield 5 number of new TB patients=

number of new employees examined:

Patients who were being treated for TB at the timeof screening and participants who had a history of TBwere excluded from the study. In addition to the overallyield, we also calculated sex- and age-specific yields.

To evaluate the possible underestimation of TB oc-currence by the National Surveillance Program, we ob-tained data from the program for the year 2003, whichwas the most updated published data available at thetime of study. This program has been and still is oper-ated by the CDC of the Department of Health as apartof the National Notifiable Disease Surveillance Systemand receives reports of new cases identified throughscreening programs at all hospitals in Taiwan as wellas those identified through mobile CXR screeningbuses. We compared the gender- and age-specific yieldsderiving from the national surveillance program to thecorresponding data observed in our study and evalu-ated the differences in the results of the 2 programs us-ing the x2 test. All statistical analyses were conductedusing the Statistical Package for the Social Sciences

256 Vol. 35 No. 4 Su et al

(SPSS for Windows, Version 12.0, Chicago, IL), and allstatistical tests were performed at a 2-tailed significancelevel of 05.

To protect patient confidentiality, the database cre-ated by our study was password protected, and all per-sonal identifiers were removed from the database. Theprotocol of this study was determined as not requiringethical review by the Human Experiment Review Boardof the National Cheng Kung University.

RESULTS

During the study period, 17,105 new employees,comprising 8592 women and 8513 men, underwentpreemployment health examinations at the clinic.The study participants were aged from 15 to 68 years,with a mean of 27.3 years. Among those, a male patientwith previously diagnosed pulmonary TB, who had al-ready commenced treatment for this condition at thetime of screening, was identified and excluded fromfurther analyses. Of the remaining 17,104 participants,pulmonary TB was diagnosed for 22, resulting in anoverall yield of 128.6 per 100,000 individuals. Halfof the 22 newly diagnosed patients were between 25and 34 years of age, and half of the patients revealed‘‘minimal’’ disease involvement. Seven patients (31.8%)had at least 1 sputum culture that was positive forMycobacteriun tuberculosis, and 2 patients (9.1%) had 1,

Table 1. Characteristics of 22 new patients of pulmonarytuberculosis identified through preemployment screeningin an industrial park in Tainan, Taiwan, from July 2004to June 2005

Disease characteristics Number %

Sex

Women 11 50.0

Men 11 50.0

Age (yr)

15-24 8 36.4

25-34 11 50.0

$35 3 13.6

Extent of disease

Minimal 11 50.0

Moderately advanced 8 36.4

Far advanced 1 4.5

Cavities 2 9.1

Sputum smear acid-fast stain

Positive 2 9.1

Negative 20 90.9

Bacteriologically confirmed*

Positive 7 31.8

Negative 15 68.2

Family history of TB

Yes 1 4.5

No 21 95.5

*Sputum smear positive on acid-fast stain and/or culture positive for Mycobacterium

tuberculosis.

or more than 1, sputum smear exhibiting the presenceof AFB as seen under microscopy. Both of the 2 patientswith positive sputum AFB smears also had positive spu-tum cultures, such that the yield of bacteriologicallyconfirmed pulmonary TB was 40.9 per 100,000 individ-uals (Table 1).

The overall yield of the National TB SurveillanceProgram was 47.2 per 100,000 individuals, which wassignificantly lower than the corresponding figure foundin our screening program (Table 2). Furthermore, ascompared with the results from the National TB Sur-veillance Program, our screening program had signifi-cantly higher yields for all age groups and for bothmen and women. However, whereas our program hadsimilar yields between the 2 sexes, the National TB sur-veillance Program had a more than 2-fold higher yieldfor men as compared with women (Table 2).

DISCUSSION

It is generally believed that providing prompt andeffective treatment to patients suffering from TB maybring the disease under control globally,8 but the prob-lem regarding how to identify effectively such patientsamong an apparently healthy population (the screen-ing for TB) remains a great challenge to both publicheath practitioners and clinicians. In areas with a rela-tively low prevalence, the policy of preemploymentmass CXR screening for pulmonary TB remains contro-versial even for health care workers,9,10 who are at ahigher risk.11 TB can arise and spread within a regularworkplace,12 and certain mass CXR screening pro-grams have been applied in areas with a relativelylow prevalence, this being especially the case whenthe goal is the elimination of the disease.13

The main purpose of the mandated preemploymenthealth examination in Taiwan is to determine whetherthe worker is fit for the job, although it may be seen toalso serve the purpose of screening for communicablediseases. Because this compulsory examination in-cludes a CXR, it can serve effectively as screening forpulmonary TB The mandatory nature of the health ex-amination offers a unique opportunity to evaluate theperformance of the National TB Surveillance Program,which is a combination of various screening programs,including the complimentary Adult Health Examina-tion offered by the National Health Insurance, the pre-recruitment health examination for the military servicefor 19-year-old men,6 the screening using mobile CXRbuses, and others. Whereas the reporting to the CDCof all cases of TB as detected by various health careinstitutes and all physicians is mandated by law inTaiwan, the participation in some of the above-mentioned programs, such as the Adult Health Exami-nation and the mobile bus screening, is voluntary.

Su et al May 2007 257

Table 2. Comparisons between results of the preemployment screening at an industrial park and those of theNational TB Surveillance Program in Taiwan

Characteristics

Participants Patients

Yield of the

preemployment screening

Yield of the National TB

Surveillance Program

Relative risk* P valueNumber % Number % (per 100,000 individuals) (per 100,000 individuals)

Sex

Women 8592 50.2 11 50.0 128.0 40.7 3.15 ,.001

Men 8512 49.8 11 50.0 129.2 91.7 1.41 ,.001

Age (yr)

15-24 5548 32.4 8 36.4 144.2 25.6 5.64 ,.001

25-34 10,152 59.4 11 50.0 108.4 31.3 3.46 ,.001

$35 1404 8.2 3 13.6 213.7 63.3 3.34 .025

Total 17,105 100.0 22 100.0 128.6 47.2 2.72 ,.001

*Relative risk of being detected by the preemployment screening vs by the National TB Surveillance Program; P values are for x2 tests.

Therefore, some previous studies have speculated thatthe occurrence of pulmonary TB in Taiwan may havebeen underestimated by the National TB SurveillanceProgram,5,6 as has been the cases for a number of otherpassive public health surveillance systems.14

In fact, a mass TB survey of all prisons in Taiwanin 1999 screened a total of 51,496 inmates using 70 3

70-mm miniature films and referred 1148 (2%) of theseindividuals to undergo follow-up examinations using14 3 14-inch films.5 Among the 922 individuals whocompleted follow-up examinations, pulmonary TB wasdiagnosed for 107. On the basis of the assumptionthat the prevalence of TB among those who failed toundergo follow-up examinations was equivalent tothat among those who did undergo follow-up examina-tions, the study obtained an estimated prevalence of258.7 per 100,000 inmates. Among these 107 patients,88 (82.2%) were newly diagnosed, resulting in a yieldof 212.9 per 100,000 individuals under the similar as-sumption that the yield of new TB cases among thosewho failed to undergo follow-up examinations wasequal to the yield for those who did undergo follow-up examinations. Contrasting such figures, however,the National TB Surveillance Program’s yield wasonly 61.3 per 100,000 in 1999.1 Likewise, anothersurvey of TB applied similar approaches to militaryconscripts in 1997 and 1998 in Taiwan and notified2969 of the 305,140 19-year-old men (1%) initiallyexamined to undergo follow-up examinations.6 Amongthe 2782 men (94%) who completed the follow-up ex-aminations, pulmonary TB was diagnosed for 237, ofwhom 212 (89%) were newly diagnosed. Accordingly,the estimated yield from this military conscript studywas 74.3 per 100,000 individuals, which was similarto the yield of 82 per 100,000 of the National TB Sur-veillance Program for 19-year-old men conducted in1998.6 For that same year, however, if the 99 casesdetected by the military screening were excluded, theyield of the National TB Surveillance Program for

19-year-old men was only 32.7 per 100,000, a figuresimilar to the corresponding yields for 22-year-oldmen (31.8 per 100,000) and 23-year-old men (30.6per 100,000) for the same year. Comparing the dataon men with those of the National TB SurveillanceProgram on women (23 per 100,000 for 19 year olds,22 per 100,000 for 20 year olds, 25 per 100,000 for 21year olds, 23 per 100,000 for 22 year olds, and 25 per100,000 for 23 year olds), the researchers concludedthat the peak in the yield for 19-year-old men revealedby the data from the National TB Surveillance Programwas due to the military screening and that the relativelylow yields for 20-year-old men (25.3 per 100,000) and21-year-old men (16 per 100,000) might be due to a‘‘post-screening effect’’ following early case detectionby the military screening that resulted in a decrease indetectable cases subsequent to this screening.6 In otherwords, the result of this military conscript study sup-ported the suspicion raised by the prison inmate studythat the National TB Surveillance Program tends to un-derestimate the occurrence of pulmonary TB in Taiwan.

The mandatory preemployment health examinationin Taiwan provides a potential mechanism to assess theoccurrence of pulmonary TB in a working population,but, to the best of our knowledge, no study using suchan approach, as was the case in our study, has beenconducted and published in the relevant literature.One previous study reported that, at the time of study,there existed 65 industrial parks in Taiwan, but only 10of those had an on-site common health care unit.15

Although all of the other 9 parks were more than 10years old, the clinic at which we conducted the currentstudy was the most recently established and was stillhiring large numbers of new employees; such a sce-nario, therefore, offered a unique opportunity to assessthe occurrence of certain diseases in the general popu-lation. In fact, a previous study on the workers in thissame park found that the seroprevalence of rubellaamong employees was compatible with corresponding

258 Vol. 35 No. 4 Su et al

figures observed in studies using representative sam-ples of the general population.16 Using the data fromthe preemployment screening, the results of the cur-rent study confirmed the suspected underestimationof the occurrence of TB by the National TB SurveillanceProgram.

Of the 22 new pulmonary TB patients identified inour study, half (11 patients; 50.0%) had only ‘‘minimal’’disease involvement, an outcome that is compatible tothe corresponding finding of 51% (120 in 237 indi-viduals) in the military conscript study.6 Both studiesapplied an active surveillance approach instead of apassive reporting approach, and the results of bothhave demonstrated that most patients identified bythis approach were in the early stage of the disease,an outcome that illustrates the relative effectivenessof such an approach for early diagnosis.

Although HIV infection is a well-known risk factorfor TB, none of the patients identified in our studyhad HIV infection. This is not surprising because HIVis not as prevalent as TB in Taiwan. Because none ofour study participants were health care workers, we be-lieve the individuals identified as TB patients acquiredthe disease in the general community, which is not atall uncommon for endemic area of TB like Taiwan. Al-though lung malignancy should always be consideredas a differential diagnosis for pulmonary TB, no caseof lung malignancy was identified among the TB pa-tients diagnosed in our study.

The National TB Surveillance Program had a morethan 2-fold higher yield for men as compared withwomen (91.7 vs 40.7, respectively, per 100,000), whereasthe current study has observed similar yields for bothsexes (129.2 vs 128.0, respectively, per 100,000). Thismay be attributable, at least in part, to the fact thatmen have more opportunities to receive mandatory TBscreenings, including the military recruitment screeningand preemployment screenings. This finding indicatesthe need for more active TB surveillance for women inthe general population. All of the 22 new patients iden-tified in our study were reported to the National TB Sur-veillance Program as required by law, and each of thesepatients received anti-TB therapy. In addition, their start-ing dates for going to work were postponed to preventthe transmission of TB to their coworkers.

A substantial proportion of the TB patients (31.8%)identified in the current study had positive sputum cul-ture, which indicated that these patients were able totransmit the diseases to others, and they are importantcases to both public health and occupational health.17

In particular, patients who are sputum smear-positivefor AFB are highly infectious and thus considered asthe most important source of the transmission of TB.For the modern industry, almost all of the workerswork in air-conditioned buildings, and the presence

of such patients in the workplace poses a threat tothe heath and well-being of other workers. Our studyresults demonstrate the importance of preemploymentscreening for TB in endemic areas like Taiwan. The pro-portion of bacteriologically confirmed pulmonary TBcases constituted approximately 61.4% of all tubercu-losis cases reported to the National TB SurveillanceProgram.1 It should be noted that a large proportionof cases reported to the program are identified throughvoluntary participation in various screenings, and manyof those individuals are in fact symptomatic or suspectthat they have contracted the disease. However, in thecurrent study, participation in the screening was man-dated by law, which may explain the relatively smallerproportion of bacteriologically confirmed cases. In theprevious study on military conscripts, in which partic-ipation in the screening was also mandated by law,the proportion of bacteriologically confirmed caseswas even lower (8.9%) than that in our study.6 There-fore, the yields in our study were not likely to beunderestimations.

The preemployment health examination is not de-signed for screening for TB exclusively, and, therefore,the tuberculin skin test and sputum smear are notincluded. However, all the participants with suspectedpulmonary TB were referred to the Chest Hospital forfurther diagnostic procedures, although none of themunderwent bronchoscopy for diagnosis. Whereas thediagnosis of TB in some of the culture-negative patientsmight have been inaccurate, such an outcome wouldnot affect the conclusions of our study because thesame diagnostic criteria were applied to both the pre-employment screening of workers and the NationalTB Surveillance Program. Because new employees con-stitute an active working population, they may tend tobe somewhat healthier than other people of the sameage and thus may not be truly representative of thegeneral population. If that happened to be the casehere, the rates of TB might have been underestimatedin our study. Nonetheless, because our main findingwas that the preemployment screening revealed higherrates of TB infection than the National TB SurveillanceProgram, if the preemployment screening did underes-timate the rates, such an outcome would not changeour main conclusion. Still, it should be recognizedthat the accurate diagnosis of pulmonary TB usingCXR alone is difficult, and, therefore, other TB screen-ing efforts in addition to the preemployment healthexamination are needed to identify new cases.

Our study has revealed that the National TB Surveil-lance Program, which is a passive reporting system, un-derestimates the occurrence of TB in Taiwan, this beingparticularly so for women. It has been argued previouslythat proper coordination between occupational andcommunity health services will give the best results,

Su et al May 2007 259

and logically, this should lead to a rapid decline of thedisease.18 The results of our study support this argu-ment and demonstrate the importance of preemploy-ment TB screening in endemic areas like Taiwan.

References

1. Center for Disease Control. Tuberculosis in Taiwan. Available at:

http://www.cdc.gov.tw. Accessed October 25, 2005.

2. Department of Health. Summary of statistics on causes of death in

Taiwan, 2004. Available at: http://www.doh.gov.tw. Accessed October 25,

2005.

3. Chiang C-Y, Enarson DA, Yang S-L, Suo J, Lin T-P. The impact of

national health insurance on the notification of tuberculosis in Taiwan.

Int J Tuberc Lung Dis 2002;6:974-9.

4. Chronic Disease Control Bureau, Department of Health. TB statis-

tics—1997. Taipei: Chronic Disease Control Bureau; 1998.

5. Chiang C-Y, Hsu C-J, Hsu P-K, Suo J, Lin T-P. Pulmonary tuberculosis

in the Taiwanese prison population. J Formos Med Assoc 2002;101:

537-41.

6. Chiang C-Y, Yu M-C, Yang S-L, Lin T-P. Screening for pulmonary

tuberculosis among military conscripts in Taiwan. J Formos Med Assoc

2002;101:841-5.

7. Last JM. A dictionary of epidemiology. 4th ed. Oxford: Oxford

University Press; 2001.

8. Brewer TF, Heymann SJ. To control and beyond: moving towards elimi-

nating the global tuberculosis threat. J Epidemiol Commun Health 2004;

58:822-5.

9. Chaturvedi N, Cockcroft A. Tuberculosis screening in health service

employees: who needs chest X-rays? Occup Med 1992;42:179-82.

10. Lamden K, Cheesbrough J, Madi S. Detecting tuberculosis in new

arrivals to UK: occupational health screening of doctors must be

improved. Br Med J 2000;321:569.

11. Meredith S, Watson JM, Citron KM, Cockcroft A, Darbyshire JH. Are

healthcare workers in England and Wales at increased risk of tubercu-

losis? Br Med J 1996;313:522-5.

12. Shirai M, Sato A, Chida K, Kishimoto H, Wada R, Takizawa S. Five

cases of tuberculosis occurring in an accounting office during 5-year

period. Kekkaku 1991;66:1-6.

13. Pang SC. Chest radiography and tuberculosis case-finding in Western

Australia. Resp Med 1998;92:198-202.

14. Heipel D, Ober J, Edmond M, Bearman G. Surgical site infection

surveillance for neurosurgical procedures: a comparison of passive

surveillance by surgeons to active surveillance by infection control

professional. Am J Infect Control 2005;33:e21.

15. Su S-B, Guo H-R, Huang P-C. Common health-care units in the indus-

trial parks in Taiwan. J Occup Health 2001;43:382-7.

16. Su S-B, Guo H-R. Seroprevalence of rubella among women of child-

bearing age in Taiwan after nationwide vaccination. Am J Trop Med

Hyg 2002;67:549-53.

17. Klitzman S, Kellner P. Control of tuberculosis in the workplace:

toward an integration of occupational health and public health. Occup

Med 1994;9:723-34.

18. Khogali M. Tuberculosis among immigrants in the United Kingdom:

the role of occupational health services. J Epidemiol Commun Health

1979;33:134-7.

Prescription of prophylactic antibioticsfor neurosurgical proceduresin teaching hospitals in IranMehrdad Askarian, MD, MPH,a Ali Reza Moravveji, MD, MPH,a and Ojan Assadian, MD, DTMHb

Shiraz, Iran, and Vienna, Austria

Background: To assess the appropriateness of surgical antibiotic prophylaxis in neurosurgical procedures, using the AmericanSociety of Health-System Pharmacists (ASHP) guideline as reference, 110 patients were prospectively evaluated. Monitoring surgicalantibiotic prophylaxis is crucial in ensuring appropriate use of antimicrobial agents in this setting. This will minimize the conse-quences of antibiotic misuse such as increased drug antibiotic resistance, adverse events, and higher costs to the institution.Methods: We recruited 110 consecutive patients undergoing clean neurosurgical treatment in 2 hospitals. Data were collectedprospectively from patients’ medical records between February 2004 and April 2004. The data collection forms for each patientincluded hospital name, patient demographics, type of surgery, and type of antimicrobial prophylaxis regimen (drug name,dose, interval, route of administration, number of doses and time administered, and duration of administration).Results: Discrepancies about antibiotic selection, duration, and start time of prophylaxis were seen between current administra-tion and the ASHP guideline. The direct cost of prophylactic antibiotics for the 110 procedures was 14 times greater than what itwould have cost to administer prophylactic antibiotics adhering to the ASHP guideline (US $802 vs US $59; US $7.29 vs US $0.54per patient, respectively). This is equivalent to US $6.75 of extra costs per procedure and patient.Conclusion: This study indicates the need for interventions to improve the rational use of antibiotic prophylaxis in Iran to prevent thecomplications of inappropriate administration of antimicrobials and decrease unnecessary costs. (Am J Infect Control 2007;35:260-2.)

Surgical site infections (SSIs) are the most commonand serious complication among surgically treatedpatients, resulting in increased rates of morbidity andmortality, length of stay in hospital, and cost.1

Postoperative infection following neurosurgeryoccurs at a reported incidence of 4%, accounting fora mortality of approximately 14% of the total postoper-ative deaths.2 Postoperative infection in clean neuro-surgery is even lower, with reported frequencies up to1%.3 Gram-positive Staphylococci are the most com-mon causative pathogens for postoperative infectionsin neurosurgery patients. Infection can occur in super-ficial tissues such as skin surrounding the wound, boneflap, or in the deeper tissues; even superficial woundinfections may cause complications such as osteomye-litis, meningitis, encephalitis, abscess, or death.4

From the Department of Community Medicine,a Shiraz University ofMedical Sciences, Shiraz, Iran; and Department of Hygiene and MedicalMicrobiology,b Division of Hospital Hygiene, Medical University ofVienna, Vienna General Hospital, Waehringer Guertel, Vienna, Austria.

Address correspondence to Mehrdad Askarian, MD, MPH, AssociateProfessor of Community Medicine, Shiraz University of Medical Sci-ences, PO Box 71345-1737, Shiraz-Iran. E-mail: askariam@sums.ac.ir.

Supported by the Deputy for Research at the Shiraz University ofMedical Science (Grant no. 83-2168).

0196-6553/$32.00

Copyright ª 2007 by the Association for Professionals in InfectionControl and Epidemiology, Inc.

doi:10.1016/j.ajic.2006.04.214

260

Although the rate of infection in clean neurosurgicaloperations is known to be low, the routine use of pro-phylactic antibiotics has been adopted by many neu-rosurgeons because of the potentially devastatingconsequences of infectious complications.5 The effective-ness of antibiotic prophylaxis in preventing postoperativewound infections has been proven in several studies.Therefore, surgical antibiotic prophylaxis accounts forover 30% of antibiotic prescriptions in general hospitals.However, in surgical centers it can be as high as 95%.Monitoring surgical antibiotic prophylaxis is crucial inensuring appropriate use of antimicrobial agents in thissetting. This will minimize the consequences of antibioticmisuse such as increased drug antibiotic resistance,adverse events, and higher costs to the institution.6

The choice of antimicrobial agent, the starting timeof administration, and the duration of prophylaxis arefactors that can affect the appropriate use of surgicalantibiotic prophylaxis. In this study, we used therecommendations made by the American Society ofHealth-System Pharmacists (ASHP)7 to assess theappropriateness of antibiotic prophylaxis for surgicalprocedures in 2 teaching hospitals affiliated withShiraz University of Medical Sciences in Shiraz, Iran.

METHODS

We specifically chose to review adherence to anti-biotic prophylaxis in neurosurgical procedures. Werecruited 110 consecutive patients undergoing cleanneurosurgical treatment from 2 hospitals (57 cases

Askarian, Moravveji, and Assadian May 2007 261

from hospital ‘‘A’’ and 53 from hospital ‘‘B’’) in Shiraz,the capital of the province Fars. These hospitals are thesole referral centers for neurosurgical treatment for apopulation of 4 million inhabitants living in this prov-ince. Data were collected prospectively from patients’medical records between February 2004 and April2004. The data collection forms for each patientincluded hospital name, patient demographics, typeof surgery, and type of antimicrobial prophylaxisregimen (drug name, dose, interval, route of admini-stration, number of doses and time administered, andduration of administration). Because of therapeuticuse of antibiotics instead of prophylactic, the proce-dures classified as dirty-infected wound were excludedfrom the study. Cases for which it was not possible todetermine prophylactic or therapeutic use of antibi-otics were also excluded. Adherence to the ASHP rec-ommendations was assessed by a clinician. Data wererecorded into EPI-INFO 2002 program (Centers for Dis-ease Control and Prevention, Atlanta, GA) and analyzedusing SPSS for Windows 11.5 (SPSS Inc., Chicago, IL).Statistical evaluation was conducted by descriptiveanalysis including frequencies.

RESULTS

Records for 110 patients were collected andreviewed. Sixty-six percent were male. Patients’ ageranged between 14 and 76 years (mean, 40.6 years;SD 6 16.3); 15.5% of operations were emergencyprocedures, and wound classes were clean in 100%.

Only 1 patient (0.9%) received prophylactic anti-biotics in accordance with the entire set of recommen-dations given by the ASHP guidelines. In 81 patients(74.3%), prophylaxis was started at the time of induc-tion of anesthesia as recommended by the ASHP.Prophylaxis was continued in patients for severaldays (mean, 5 days; SD 6 4.8) rather than the ASHPrecommendation of only a single-dose antibiotic.

A total of 9 different antibiotics or combinationsthereof were prescribed for 109 patients. One patientreceived only 1 substance (cefazolin), and up to 5 typesof antibiotics were administered for others. The mostfrequently used antibiotics, alone or in combination,were intravenous gentamicin (in 108 cases; 98%) andcefazolin (in 107 cases; 97%). Some oral agents suchas cephalexin, cloxacillin, and metronidazol were usedsubsequently after intravenous agents in 32 patients(29%). Vancomycin was prescribed for 3 patients(2.7%) without any evidence of b-lactam allergy.

The direct cost of prophylactic antibiotics for the 110procedures was 13.6 times greater than what it wouldhave cost to administer prophylactic antibiotics adher-ing to the ASHP guideline (US $802 vs US $59 or US$7.29 vs US $0.54 per patient, respectively).

DISCUSSION

This study showed discrepancies between currentadministration of antibiotics for surgical prophylaxisand the ASHP guideline. The major mistakes were anti-biotic selection, duration, and start time of prophylaxis.

The ASHP recommends a single dose of 1000-mgcefazolin at the time of induction of anesthesia forprevention of SSI. However, most patients receivedcefazolin but in combination with other antibiotics, in-cluding gentamicin or oral antimicrobial compounds.In general, it is debatable whether the concomitantuse of oral and parenteral antibiotics has greater effi-cacy in prevention of postsurgical infections.8

The question of why some patients received subse-quent oral antibiotics regardless of any postoperativeinfection in our study was unclear and needs furtherinvestigation. The use of vancomycin for prophylaxisis appropriate only when there is a true type I hyper-sensitivity to b-lactam or when there is a highincidence of SSI because of methicillin-resistant staph-ylococci.9 Vancomycin was used in 3% of our patientswithout any of these situations present.

Aminoglycosides such as gentamicin, which wereused for most patients, may be nephrotoxic and oto-toxic, even with precise therapeutic doses, and the useof these in prophylaxis may predispose for the emer-gence of methicillin-resistant staphylococci. This mayinitiate an unfavorable cascade in which treatmentwith vancomycin may select for the emergence ofvancomycin-resistant enterococci. Therefore, amino-glycosides are not endorsed for systemic prophylaxis.10

In light of the ASHP’s recommendations of only1 time, single-dose antibiotic prophylaxis for neurosur-gical procedures, none of the patients in our studyreceived optimal duration of prophylaxis. In fact, theunwarranted prolonged use of antibiotics increasesthe likelihood of developing infections by antibiotic-resistant bacteria, exposes patients to more adversedrug effects, and increases the overall medical cost.11

The initiation time of antibiotic therapy was appro-priate for three fourths of the patients. Antibiotic pro-phylaxis must be initiated early enough to allow theantibiotic to achieve effective tissue concentration.Administration of the antibiotic too early may compro-mise antibiotic efficacy.12

The results of this study show that the direct cost ofcurrent surgical antibiotic prophylaxis was approxi-mately 14-fold greater than the cost of a suggested reg-imen by guideline. This is equivalent to US $6.75 ofextra cost per procedure and patient or US $743 forthe 110 investigated patients. To put these excess costsin a socioeconomic perspective, it should be noted thatthe minimum salary of a worker in Shiraz that year wasUS $120 per month.

262 Vol. 35 No. 4 Askarian, Moravveji, and Assadian

Our study has several important limitations. Al-though most Iranian neurosurgical departments sharea close communication and standards are similar, ourresults represent the situation in the province of Farsand cannot be generalized to the Iranian situation. Sec-ond, we have no documented evidence that the ASHPpractices would have been any more effective or moreeconomic. We decided to use recommendations pub-lished by the ASHP to measure against rational andevidence-based international standards. However, thepossibility exists that recommendations given by theASHP guidelines were not feasible in our patients orfor the situation in Iran.

In conclusion, this study indicates the need for inter-ventions to improve the rational use of antibiotic pro-phylaxis to prevent the complications of inappropriateadministration of antimicrobials and decrease unneces-sary costs. We suggest that the mistakes in prescribingpractice for surgical antibiotic prophylaxis in our studywere due to lack of adherence to guidelines among sur-gical teams.

References

1. Erman T, Demirhindi H, Gocer AI, Tuna M, Ildan F, Boyar B. Risk fac-

tors for surgical site infection in neurosurgery patients with antibiotic

prophylaxis. Surg Neurol 2005;63:107-12.

2. Korinek AM. Risk factors for neurosurgical site infections after crani-

otomy: a prospective multicenter study of 2944 patients. The French

Study Group of Neurosurgical Infections, the SEHP, and the C-CLIN

Paris-Nord. Service Epidemiologie Hygiene et Prevention. Neurosur-

gery 1998;41:1073-9.

3. Mastronardi L, Tatta C. Intraoperative antibiotic prophylaxis in clean

spinal surgery: a retrospective analysis in a consecutive series of 973

cases. Surg Neurol 2004;61:129-35.

4. Zhao JZ, Wang S, Li JS, Yang J, Zang JT, He Q, et al. The perioperative

use of Ceftriaxone as infection prophylaxis in neurosurgery. Clin Neu-

rol Neurosurg 1995;97:285-9.

5. Zhu XL, Wong WK, Yeung WM, Mo P, Tsang CS, Pang KH, et al.

A randomized, double-blind comparison of Ampicillin/sulbactam and

Ceftriaxone in the prevention of surgical site infection after neurosur-

gery. Clin Ther 2001;23:1281-91.

6. Alerany C, Campany D, Monterde J, Semeraro C. Impact of lo-

cal guidelines and an integrated dispensing system on antibiotic

prophylaxis quality in a surgical centre. J Hosp Infect 2005;60:

111-7.

7. American Society of Health-System Pharmacists. ASHP therapeutic

guidelines on antimicrobial prophylaxis in surgery. Am J Health Syst

Pharm 1999;56:1839-88.

8. Silver A, Eichorn A, Kral J, Pickett G, Barie P, Pryor V. Timeliness and

use of antibiotic prophylaxis in selected inpatient surgical procedures.

Am J Surg 1996;171:548-52.

9. Osmon DR. Antimicrobial prophylaxis in adults. Mayo Clin Proc 2000;

75:98-109.

10. Polk HC Jr, Christmas AB. Prophylactic antibiotics in surgery and

surgical wound infections. Am Surg 2000;66:105-11.

11. Mastronardi L, Tatta C. Intraoperative antibiotic prophylaxis in clean

spinal surgery: a retrospective analysis in a consecutive series of 973

cases. Surg Neurol 2004;61:129-35.

12. Martin C. Antimicrobial prophylaxis in surgery: general concepts

and clinical guidelines. Infect Control Hosp Epidemiol 1994;15:

463-71.

Bacterial contamination ofstethoscopes with antimicrobialdiaphragm coversMark W. Wood, RRT, BS,a Ryan C. Lund, MPH, CIC,b and Kurt B. Stevenson, MD, MPHb

Boise, Idaho

Antimicrobial stethoscope covers impregnated with silver ions have been developed to prevent surface contamination and poten-tial transmission of bacterial pathogens to patients. To test their practical utility, covers were distributed with the manufacturers’recommendations to a mixed group of health care professionals in a medical/surgical intensive care unit and an emergency de-partment. Seventy-four clinicians were selected from a convenience sample for surface cultures and standard questioning regard-ing cleaning and cover use. Surface colony counts were significantly lower for uncovered stethoscope diaphragms (mean, 71.4colonies) compared with covers used #1 week (mean, 246.5 colonies) and those .1 week old (mean, 335.6 colonies). Aftercontrolling for type of clinician, frequency of stethoscope cleaning, and method of stethoscope cleaning, only the presence ofa stethoscope cover was associated with higher colony counts (P , .0001). We question the practical utility of the antimicrobialdiaphragm covers evaluated in this study for reducing the surface colonization of potentially harmful microorganisms. (Am J InfectControl 2007;35:263-6.)

Indirect contact with contaminated patient environ-ment or medical devices has been associated withtransmission of nosocomial pathogens and providesthe basis for the implementation of current transmis-sion-based isolation guidelines.1 Many nosocomial in-fections have been demonstrated to be due to crosstransmission.2 Many of these infections could poten-tially be prevented through adequate application ofinfection control practices.

Medical devices including stethoscopes and oto-scopes have been demonstrated to be contaminatedwith bacterial species and have been implicated as po-tential vectors of cross transmission.3-10 Other inani-mate objects such as pagers,11 toys in the neonatalintensive care unit,12 and hospital surfaces13 have like-wise been implicated. The stethoscope represents per-haps the most commonly used medical device in thehospital. Numerous studies have examined the dia-phragm surface and earpieces for bacterial contamina-tion and have determined that it occurs at a veryhigh rate (66%-100%).3-10 Among isolated pathogens,

staphylococcus species including methicillin-resistantStaphylococcus aureus (MRSA) have been reported.3-8,10,14

Cleaning the diaphragm with alcohol has beendemonstrated to reduce surface bacterial counts.3,6-8

Although cleaning does decrease bacterial load,studies show that health professionals may not becommitted to the process of routine cleaning. In onestudy, only 48% of health care providers cleaned theirstethoscopes daily or weekly, 37% monthly, and 7%yearly; and 7% had never cleaned their stethoscopes.6

Covering the surface of the stethoscope diaphragmwith an antimicrobial cover may prevent contamina-tion and cross transmission of nosocomial pathogens.Diaphragm covers impregnated with silver ions havethe potential to suppress growth of bacteria and fungiand thus prevent transmission of nosocomial patho-gens. We report our experience of the practical utilityof antimicrobial stethoscope diaphragm covers bycomparing cultures from the surfaces of covered anduncovered stethoscopes and evaluating clinician’spractices regarding stethoscope cleaning and care. We

From the Respiratory Therapy Department,a and Infection Control De-partment,b Saint Alphonsus Regional Medical Center, Boise, ID.

Address correspondence to Mark Wood, RRT, BS, Respiratory TherapyDepartment, 1055 N. Curtis Rd., Boise, ID, 83706. E-mail: markwood@sarmc.org.

Dr. Stevenson is currently at the Division of Infectious Diseases andDepartment of Clinical Epidemiology at The Ohio State University Med-ical Center, Columbus, OH.

0196-6553/$32.00

Copyright ª 2007 by the Association for Professionals in InfectionControl and Epidemiology, Inc.

doi:10.1016/j.ajic.2006.09.004

263

264 Vol. 35 No. 4 Wood, Lund, and Stevenson

discuss the implications of our findings on the poten-tial impact of these antimicrobial covers and stetho-scope care in infection prevention.

METHODS

During the fall of 2003, antimicrobial diaphragmcovers ($2 per individual) with the manufacturers’ rec-ommendations about cleaning and routine changingof covers were distributed to a mixed group of healthcare professionals working in a large medical/surgi-cal/trauma intensive care unit (ICU) and a regionaltrauma emergency department (ED). All clinicianswere instructed that they may be randomly selectedto participate in a culture study and to complete a sur-vey regarding their stethoscope care.

During the study period, health care workers wereselected from a convenience sample of the ICU andED staffs as randomly as possible for culture of stetho-scope diaphragms. These included equal number ofindividuals with and without antimicrobial covers. Cul-tures were obtained by pressing the diaphragm surfacefor 5 seconds on a blood agar plate and incubating theculture aerobically for 48 hours. After incubation, thetotal number of colonies of bacterial growth wascounted for each sample, but the genus and speciesof the cultured organisms were not specifically identi-fied. An example of a positive culture is illustrated inFig 1.

At the time of culture, each health care worker wasasked to complete a short written survey. Each clinicianindicated whether they were using an antimicrobial

Fig 1. Example of a culture plate demonstratingbacterial growth from the surface of a contaminated

stethoscope.

cover and whether it was being used beyond the1-week time frame recommended by the manufacturer.Additionally, clinicians were categorized by role andtraining (physician, nurse, respiratory therapist, orother), were queried about how often they cleanedthe stethoscope (between every patient, several timesa day, once daily, once monthly, or rarely if ever), andwere asked to identify which cleaning agent was used(alcohol wipes, soap and water, antiseptic wipes, orthe alcohol-based handrub). The mean number of col-onies was compared between those without coversand those with short-term (,1 week) or long-termcover use (.1 week old). Multiple linear regressionmodeling was performed to determine predictors ofthe number of colonies cultured from the diaphragmsurface with presence or absence of antimicrobial dia-phragm cover, type of clinician, frequency of stetho-scope cleaning, and method of stethoscope cleaningas covariates in the model. Surface colony counts, thedependent variables, were transformed to the naturallogarithm to satisfy the assumption of homoscedascityin the linear regression model. All statistical tests were2-sided, and significance was set at P , .05, with allanalyses performed using STATA, version 8 (Stata Cor-poration, College Station, TX).

RESULTS

One hundred percent of the diaphragms culturedwere contaminated. Colony counts were significantlylower for uncovered stethoscope diaphragms (mean,71.4 colonies) compared with covers less than or equalto 1 week old, as recommended by the manufacturer(mean, 246.5 colonies) and those greater than 1 weekold (mean, 335.6 colonies). Health care workers whowere surveyed were generally reflective of the distribu-tion of types of clinicians within the institution (36nurses, 21 respiratory therapists, 14 physicians, and3 nurse assistants). Mean colony counts by cliniciantype were highest among nurses (203.1 colonies), fol-lowed by nurse assistants (197.7 colonies), physicians(193.1 colonies), and respiratory therapists (169.5 col-onies). Viewed differently, the majority of uncoveredstethoscopes (91.9%) had colony counts in the 0 to199 range compared with only 35.1% of covered steth-oscopes (Table 1). Most of the covered stethoscopes had.200 colonies (23, 62.1%) with 14 (37.8%) having 400to 599 colonies.

Only 11 of 74 (15%) health care workers cleanedtheir stethoscope between every patient. One third(25/74, 34%) cleaned several times a day, with slightlyless (22/74, 30%) cleaning their stethoscopes daily.Fewer (9/74, 12.2%) of those surveyed rarely if evercleaned their stethoscopes, and 7 of 74 (9.5%) did soonce a month. Most surveyed health care workers

Wood, Lund, and Stevenson May 2007 265

Table 2. Stethoscope surface colony counts with or without cover stratified by stethoscope care

Physician Nurse

Respiratory

Therapist Other Totals

Cover

No. (%)

No Cover

No. (%)

Cover

No. (%)

No Cover

No. (%)

Cover

No. (%)

No Cover

No. (%)

Cover

No. (%)

No Cover

No. (%)

Cover

No. (%)

No Cover

No. (%)

Frequency of stethoscope

cleaning

Between Every Patient 0 (0) 1 (50.0) 5 (55.6) 0 (0) 4 (44.4) 1 (50.0) 0 (0) 0 (0) 9 (100) 2 (100)

Several Times a Day 1 (6.3) 2 (22.2) 8 (50.0) 4 (44.4) 7 (43.8) 3 (33.3) 0 (0) 0 (0) 16 (100) 9 (100)

Once Daily 5 (71.4) 2 (13.3) 1 (14.3) 7 (46.7) 0 (0) 4 (26.7) 1 (14.3) 2 (13.3) 7 (100) 15 (100)

Once Monthly 0 (0) 1 (16.7) 1 (100) 4 (66.7) 0 (0) 1 (16.7) 0 (0) 0 (0) 1 (100) 6 (100)

Rarely if Ever 0 (0) 2 (40.0) 3 (75.0) 3 (60.0) 1 (25.0) 0 (0) 0 (0) 0 (0) 4 (100) 5 (100)

Totals 6 (16.2) 8 (21.6) 18 (48.7) 18 (48.7) 12 (32.4) 9 (24.3) 1 (2.7) 2 (5.4) 37 (100) 37 (100)

Method of stethoscope

cleaning

Alcohol Wipes 1 (3.5) 6 (22.2) 17 (58.6) 13 (48.2) 10 (34.5) 6 (22.2) 1 (3.5) 2 (7.4) 29 (100) 27 (100)

Antiseptic wipes 4 (66.7) 2 (22.2) 1 (16.7) 4 (44.4) 1 (16.7) 3 (33.3) 0 (0) 0 (0) 6 (100) 9 (100)

Alcohol Hand Gel 0 (0) 0 (0) 0 (0) 1 (100) 1 (100) 0 (0) 0 (0) 0 (0) 1 (100) 1 (100)

Soap and Water 1 (100) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 1 (100) 0 (0)

Totals 6 (16.2) 8 (21.6) 18 (48.7) 18 (48.7) 12 (32.4) 9 (24.3) 1 (2.7) 2 (5.4) 37 (100) 37 (100)

Table 1. Stethoscope colony counts with or without cover stratified by job description

Colony counts from culture of stethoscope diaphraghm surface

0-199 200-399 400-599 .600 Totals

Job Description

Cover

No. (%)

No Cover

No. (%)

Cover

No. (%)

No Cover

No. (%)

Cover

No. (%)

No Cover

No. (%)

Cover

No. (%)

No Cover

No. (%)

Cover

No. (%)

No Cover

No. (%)

Physician 2 (33.3) 7 (87.5) 0 (0) 1 (12.5) 4 (66.7) 0 (0) 0 (0) 0 (0) 6 (100) 8 (100)

Nurse 6 (33.3) 16 (88.9) 4 (22.2) 2 (11.1) 7 (38.9) 0 (0) 1 (5.6) 0 (0) 18 (100) 18 (100)

Respiratory Therapist 5 (41.7) 9 (100) 5 (41.7) 0 (0) 2 (16.7) 0 (0) 0 (0) 0 (0) 12 (100) 9 (100)

Other 0 (0) 2 (100) 0 (0) 0 (0) 1 (100) 0 (0) 0 (0) 0 (0) 1 (100) 2 (100)

Totals 13 (35.1) 34 (91.9) 9 (24.3) 3 (8.1) 14 (37.8) 0 (0) 1 (2.7) 0 (0) 37 (100) 37 (100)

Table 3. Linear regression model evaluating the predictors of stethoscope surface colony counts*

Variable

Regression

coefficient* p value

95% Confidence

Intervals

Job Description

Physician Reference - -

Nurse 20.0054001 0.987 20.6426102, 0.6318101

Respiratory Therapist 0.0157252 0.964 20.6848798, 0.7163302

Other 0.1543865 0.809 21.118059, 1.426832

Frequency of Cleaning

Rarely, if ever Reference - -

Between every patient 20.4117073 0.363 21.310265, 0.4868503

Once daily 0.0023183 0.995 20.7746493, 0.779286

Once monthly 0.3188067 0.503 20.6277348, 1.265348

Several times a day 20.2414011 0.519 20.9859584, 0.5031561

Method of Cleaning

Soap and water Reference - -

Alcohol wipes 20.4439758 0.660 22.450394, 1.562442

Antiseptic wipes 20.1818037 0.856 22.172516, 1.808909

Alcohol hand gel 20.5350889 0.664 22.988912, 1.918734

Stethoscope Cover

No cover Reference - -

Cover used ,1 week 1.66603 ,0.0001 0.8939513, 32.438108

Cover used .1 week 1.615923 ,0.0001 1.108545, 2.123301

*Natural logarithmic transformation of colony counts evaluated in this regression model.

266 Vol. 35 No. 4 Wood, Lund, and Stevenson

(56/74, 76%) used alcohol wipes to clean their stetho-scope compared with only 15 of 74 (20%) who usedantiseptic wipes. The mean colony counts were notdramatically different between these types of clean-ing methods (188.5 and 201.3, respectively). Oneindividual used soap and water as a cleaning method.Data on method of stethoscope cleaning are stratifiedby clinician type in Table 2.

The relationship between surface colony counts andthe presence or absence of stethoscope cover, type ofclinician, and frequency and method of cleaning wasdefinitively determined using multiple linear regres-sion modeling (Table 3). When controlling for thetype of health care worker and the frequency andmethod of cleaning the stethoscope, only the presenceof an antimicrobial stethoscope cover was associatedwith higher colony counts (P , .0001).

DISCUSSION

All stethoscope diaphragm surfaces in this studydemonstrated bacterial contamination consistent withthe findings of other investigators.3-10 Surprisingly,the surface colony counts were higher on the surfacesof the antimicrobial covers, especially if they were be-ing used beyond the manufacturer’s recommendedtime frame of 1 week. It is speculated that the addedsurface area in conjunction with the embossing mayprovide a surface that protects microbes from cleaningagents, thus creating higher colony counts. Further-more, many of the cultures of the covered diaphragmshad colonies in the shape of the embossed lettering onthe covers. The method and frequency of cleaning,however, did not appear to impact independently thesurface colonization; only the use of the cover was as-sociated with higher colony counts. In addition, at thetime of this study, the investigators were not aware ofan antimicrobial stethoscope cover impregnated withsilver ions that did not have embossing. Therefore,one was not examined.

The practical utility of these covers in preventingsurface colonization and potential transmission oforganisms to patients is seriously questioned basedon the results of this small pilot study. The validity ofthis study should be confirmed by a larger randomizedtrial. Using these types of devices may give a falsesense of security to health care workers; prolongeduse of covers appeared to result in even higher colonycounts. Their use may also induce clinicians to cleantheir stethoscopes less frequently. Although this studyfailed to show a relationship between cleaning andreduced colony counts, other investigators havedemonstrated that cleaning the diaphragm withalcohol reduces surface bacterial counts.3,6-8

This study has several limitations. Clinicians,although selected in random fashion from a largerconvenience sample of ICU staff and ED staff, werenot prospectively randomized to either study arm.This may have introduced selection bias. Another lim-itation was the failure to identify the organisms.This study was initially conceived as an internal per-formance improvement project, and budgetary restric-tions limited extensive microbiologic studies. It ispossible that the higher colony counts on the coveredstethoscopes were skin organisms representing littlerisk to patients and that the lower counts on uncoveredscopes were serious pathogens. Regardless, it isconcerning that silver ion impregnated covers wouldgenerally demonstrate significantly higher coloniza-tion rates than uncovered stethoscope surfaces. Thisobservation should be considered by others attemptingto use these antimicrobial devices as an infection con-trol intervention, and these results should be validatedby larger studies.

References

1. Centers for Disease Control and Prevention. Guideline for isolation

precautions in hospitals. Part II. Recommendations for isolation pre-

cautions in hospitals. Hospital Infection Control Practices Advisory

Committee. Am J Infect Control 1996;24:32-52.

2. Weist K, Pollege K, Schulz I, Ruden H, Gastmeier P. How many nosoco-

mial infections are associated with cross-transmission? A prospective

cohort study in a surgical intensive care unit. Infect Control Hosp Epide-

miol 2002;23:127-32.

3. Sood P, Mishra B, Mandal A. Potential infection hazards of stetho-

scopes. J Indian Med Assoc 2000;98:368-70.

4. Bernard L, Kereveur A, Durand D, Gonot J, Goldstein F, Mainardi JL,

et al. Bacterial contamination of hospital physicians’ stethoscopes.

Infect Control Hosp Epidemiol 1999;20:626-8.

5. Smith MA, Mathewson JJ, Ulert IA, Scerpella EG, Ericsson CD.

Contaminated stethoscopes revisited. Arch Intern Med 1996;156:82-4.

6. Jones JS, Hoerle D, Riekse R. Stethoscopes: a potential vector of infec-

tion? Ann Emerg Med 1995;26:296-9.

7. Cohen HA, Amir J, Matalon A, Mayan R, Beni S, Barzilai A. Stetho-

scopes and otoscopes—a potential vector of infection? Fam Pract

1997;14:446-9.

8. Marinella MA, Pierson C, Chenoweth C. The stethoscope: a poten-

tial source of nosocomial infection? Arch Intern Med 1997;157:

786-90.

9. Zuliani Maluf ME, Maldonado AF, Bercial ME, Pedrosa SA. Stethoscope:

a friend or an enemy? Sao Paulo Med J 2002;120:13-5.

10. Breathnach AS, Jenkins DR, Pedler SJ. Stethoscopes as possible vec-

tors of infection by staphylococci. BMJ 1992;305:1573-4.

11. Singh D, Kaur H, Gardner WG, Treen LB. Bacterial contamination of

hospital pagers. Infect Control Hosp Epidemiol 2002;23:274-6.

12. Hanrahan KS, Lofgren M. Evidence-based practice: examining the risk

of toys in the microenvironment of infants in the neonatal intensive

care unit. Adv Neonatal Care 2004;4:184-201.

13. Hota B. Contamination, disinfection, and cross-colonization: are

hospital surfaces reservoirs for nosocomial infection? Clin Infect Dis

2004;39:1182-9.

14. Nunez S, Moreno A, Green K, Villar J. The stethoscope in the emer-

gency department: a vector of infection? Epidemiol Infect 2000;124:

233-7.

Occupationally acquired infectiousdiseases among health care workersin Brazil: Use of Internet toolsto improve management, prevention,and surveillanceCristiane Rapparini, MD, MSc,a,b Paulo Feijo Barroso, MD, PhD,a Valeria Saraceni, MD, PhD,c

Alcyone Artioli Machado, MD, PhD,d and Guilherme Cortes Fernandes, MD, MSce

Rio de Janeiro, Sao Paulo, and Juiz de Fora, Brazil

Background: Education is a major component of prevention strategies to reduce the risk of occupational transmission of blood-borne pathogens to health care personnel.Methods: This study describes the results of an Internet-based project, ‘‘Projeto Riscobiologico.org’’, for which the main objectivesare to disseminate information in the Portuguese language about occupational exposures to bloodborne pathogens through a mail-ing list and Web site as well as to increase case reports of these events in Brazil.Results: The mailing list expanded quickly with a total of 2078 participants (from 337 different cities and all Brazilian states), and5613 messages were exchanged over a 5-year period. Mean length of participation was almost 2 years (697 days). Most of the par-ticipants (74%) reported that they frequently manage occupational exposures. Nevertheless, results showed an important lack ofbasic knowledge regarding this issue. In contrast with the high participation in the mailing list, a small number of institutionsstarted to participate in the voluntary surveillance system.Conclusion: The Internet can be used as a tool to increase knowledge and improve practices in the prevention of occupationalbloodborne pathogen exposures. In addition, it may represent a unique opportunity to implement a national surveillance system.(Am J Infect Control 2007;35:267-70.)

Health care workers (HCWs) are at increased risk ofacquiring infections by bloodborne pathogens in thehealth care setting.1,2 A series of prevention strategiescan be implemented to reduce this risk. Educationalprograms for HCWs are extremely important and areamong the major components of these strategies.3

The Internet is a worldwide electronic network pro-viding access to millions of resources. The estimatednumber of active Web sites worldwide was greaterthan 74 million as of October 2005.4 The use of elec-tronic mail (e-mail) has grown exponentially in recentyears. It is estimated that a total of 1.2 billion e-mailmailboxes are in use in 2005 and that the number of

person-to-person e-mails sent on an average day ex-ceeds 130 billion.5 Mailing lists are currently used tofacilitate online discussions on health issues amonghealth care workers, students, and patients.6,7

‘‘Projeto Riscobiologico.org’’ (www.riscobiologico.org)was implemented in Brazil in 2000. A Portuguese lan-guage mailing list and a Web site were created to allowthe discussion of the management and prevention ofinfectious occupationalexposures. In addition, a softwareprogram to increase surveillance of occupational expo-sures to bloodborne pathogens was developed. The pur-pose of this article is to describe the results of theimplementation of ‘‘Projeto Riscobiologico.org.’’

From the Infectious Diseases Service,a Department of PreventiveMedicine, Hospital Universitario Clementino Fraga Filho, School of Medicine,Universidade Federal do Rio de Janeiro, Rio de Janeiro STD/AIDS Programof Rio de Janeiro City Health Department,b Rio de Janeiro Transmissible Dis-eases Coordination,c Rio de Janeiro City Health Department, Rio de JaneiroMedicine Faculty of Ribeirao Preto,d Department of Medical Clinic, Universi-dade de Sao Paulo (USP), Sao Paulo and Internal Medicine Department,e SantaCasa de Misericordia de Juiz de Fora, Juiz de Fora, Brazil.

Address correspondence to Cristiane Rapparini, MD, MSc, RuaGetulio das Neves 22, Jardim Botanico, Rio de Janeiro, RJ, Brazil,22461.210. E-mail: rapparini@riscobiologico.org.

Members of Riscobiologico.org Project: M. R. Resende, MD, PhD; R. A. M.Lopes, RN, MSc; C. M. Rosa, MD; R. M. Figueiredo, RN, PhD; E. Gir, RN,PhD; S. M. C. Chaves, MD; and V. Vellozo, PhD.

‘‘Projeto Riscobiologico.org’’ was partially supported by a grant from Becton,Dickinson and Co; by a grant from Coordenacxao de Aperfeicxoamento de Pes-soal de Nıvel Superior (CAPES; to C. R.); and by grants from Fogarty Interna-tional Center/USNIH (2D43TW00010) and from Fundacxao Carlos Chagas deAmparo a Pesquisa do Estado do Rio de Janeiro (FAPERJ; to F. B.).

0196-6553/$32.00

Copyright ª 2007 by the Association for Professionals in InfectionControl and Epidemiology, Inc.

doi:10.1016/j.ajic.2006.06.001

267

268 Vol. 35 No. 4 Rapparini et al

MATERIALS AND METHODS

Description

‘‘Projeto Riscobiologico.org’’ was initiated inBrazil in 2000. Initially, an e-mail discussion list(listadediscussao@riscobiologico.org) and a Web site(www.riscobiologico.org) were implemented. Themain objective was to provide information and to dis-cuss occupational infections among HCWs in thePortuguese language. In addition, software (PSBio)was developed to establish a voluntary surveillancesystem of occupational exposures to bloodborne path-ogens in Brazilian health services.

Invitations to participate in the list and advertise-ments of the Web site were sent by e-mail to approxi-mately 800 HCWs (infection control professionals[90%] and/or occupational health practitioners [10%])on August 5, 2000. Simultaneously, the Web site wasregistered in 40 Web search engines (eg, Google, Yahoo,Lycos). The project was approved by the InstitutionalReview Board of the School of Medicine/HospitalUniversitario Clementino Fraga Filho of the Univer-sidade Federal do Rio de Janeiro.

Mailing list

Participants subscribed to the list through a formavailable on the Web site. Variables collected includedname, e-mail address, occupation, information abouthow did he/she learn about the list, experience withoccupational exposures to infectious agents, and spe-cialty. E-mail messages are sent to a central address,and a moderator reviewed and approved messagesbefore distribution to subscribers to eliminate adver-tisements, junk e-mails, and spam. Only registered par-ticipants are allowed to send questions and answersthrough the mailing list. A maximum of 12 messagesare sent per day. Health and outbreak alerts are distrib-uted anytime. Since June 2002, the number of e-mailstransferred by the server has been measured. Thismeasurement represents the number of different mes-sages sent per day multiplied by the number of partic-ipants subscribed in the period.

E-mails sent are classified in 7 major topics: generalinformation and newsletter; health care-related infec-tions and issues; occupational health and infections;medical waste; legislation; emergent disease, health,outbreak alerts; and miscellaneous. All messages arearchived on the Web site.

Web site

The Web site was established with approximately 100pages containing information about occupational expo-sures to infectious agents and prevention strategies. Ed-ucational materials such as international and Brazilian

guidelines as well as topic-related articles were included.The Web site, which is regularly updated, includes ques-tions and surveys about several health-related subjects.Web statistics and web log analyses have been collectedmonthly through WebTrends Software since 2000 (Web-Trends Inc., Portland, OR). Data analyzed include pageviews (refers to the sum total of what a user sees in abrowser window), number of visits (a visit beginswhen a visitors enters the site with an entry click andviews a succession of 1 or more page views; the endof a visit is signaled by an exit click or a 10-minuteperiod of inactivity), and duration of the visit.

Surveillance software-PSBio

A surveillance software based on NaSH (NationalSurveillance System for Health Care Workers) fromthe Centers for Disease Control and Prevention (CDC,Atlanta, GA)3 was developed within this project. Themain objective was to implement a voluntary surveil-lance system of occupational exposures to bloodbornepathogens among Brazilian health services. Healthprofessionals and institutions may request the softwarePSBio, which is distributed free of charge, through aform posted on the Web site. At request, informationon occupation, experience with occupational expo-sures to infectious agents, and specialty is collected.

RESULTS

Mailing list: participants

We present data collected between August 2000and June 2005. As of June 2005, a total of 2078 partic-ipants had joined the mailing list, and 1389 (67%)were still participating. The mean length of participa-tion was 697 days (standard deviation [SD] 6 592). Forthose participants who remained on the list until theend of the evaluation period, the mean length of par-ticipation was 887 days (SD 6 576), whereas, for thosewho dropped from the list, it was 270 days (SD 6

355).One hundred sixty-five out of 800 (21%) HCWs who

were invited to participate by e-mail in August 2000joined the list, and 103 (62%) of those were still partici-pating as of June 2005. Other users joined the systemafter being informed by colleagues (46%) and throughsearch engines on the Web (34%). From 2000 to 2005,these last 2 categories increased from 14% to 41% and5% to 39%, respectively.

Participants from 337 different Brazilian citiesjoined the list. The mailing list reached 607 HCWsfrom all 27 Brazilian states within 5 months of itsimplementation. In addition, participants from 6 othercountries (Cuba, 1; Uruguay, 2; Chile, 1; Peru, 2; Portu-gal, 11; and the United States, 3) joined the list.

Rapparini et al May 2007 269

Most of the participants were nurses (37%), followedby physicians (19%), occupational health technicians(8%), pharmacists (7%), and biologists (5%). Five hun-dred fifty-nine participants (559/2078; 27%) were frominfection control, and 540 (540/2078; 26%) were fromoccupational health services. Most of the participants(74%) reported that they frequently managed HCWsexposed to blood and body fluids.

Messages content

During the evaluation period, 5613 messages weresent through the mailing list, with a mean of 4.8 mes-sages/day (SD 6 2.8). From June 2002 to June 2005,4106 messages sent by participants translated to4,987,948 e-mails generated by the server.

The main topics of discussion in the list are shownin Table 1. Among 5613 messages, 1312 (23%) were clas-sified as ‘‘general information and newsletter,’’ 1301(23%) as ‘‘health care-related infections and issues,’’880 (16%) as ‘‘occupational health and infections,’’817 (15%) as ‘‘medical waste management,’’ 397 (7%)as ‘‘legislation,’’ 170 (3%) as ‘‘emergent diseases and out-breaks alerts,’’ and 736 (13%) as ‘‘miscellaneous.’’

Web site

The mean monthly number of visits in year 2000was 787. This number increased to 12,280 in 2005.The mean length of time visitors spend connected tothe site ranged from 8 to 11 minutes. During the studyperiod, the number of visitors increased from a meanof 5115 in 2000 to 34,480 page views/month in 2005.

As of June 2005, the Web site had almost 500 pages.These pages included information about occupationalexposures among HCWs to several infectious agents.Additionally, a mean of 12 breaking news/monthhad been included in the first page of the Web sitesince August 2003. They included mostly health andoutbreak alerts (22%) and announcements from theBrazilian Ministry of Health and scientific societies(24%).

An interactive ‘‘questions and quiz’’ tool was also in-cluded in the Web site: 3556 answers were sent to 28questions; 59% of all responses were correct. Questionsabout basic concepts of management and prevention ofoccupational infections showed a low proportion ofcorrect answers. For example, only 386 of 646 (60%)participants knew what tests were recommended forscreening source patients; 36 of 68 (53%) knew aboutthe safety for use of hepatitis B virus vaccine and hepa-titis B immune globulin (HBIG) during pregnancy, and230 of 498 (46%) knew about interpretation of serologichepatitis B virus tests (hepatitis B surface antigen, anti-body to hepatitis B surface antigen, and antibody tohepatitis B core antigen).

Surveillance software

Since January 2002, 1574 professionals requestedthe PSBio software. Most of the requests were fromnurses (38%), followed by physicians (20%). MostHCWs were from occupational health services (432/1358; 32%) or were from infection control services(399/1358; 29%). This information was not availablefor all respondents.

One hundred fifty-eight health services have com-pleted registration for the software, but, as of June2005, only 4 different health care services actuallyhave started sending data to this voluntary surveillancesystem. Difficulties were identified with the use of soft-ware and complexity of data forms. A new version ofPSBio software was developed, showing better prelim-inary results with higher adherence to voluntary partic-ipation by the end of 2005.

DISCUSSION

This study describes the results of an Internet-basedproject, ‘‘Projeto Riscobiologico.org,’’ which includeda mailing list, a Web site, and surveillance softwarefor occupational exposures to bloodborne pathogens.To our knowledge, this is the first mailing list focusedon infectious occupational exposures among HCWsin the Portuguese language. A great proportion ofBrazilian HCWs do not understand English, the mainlanguage in scientific medical literature and Web sites.

Table 1. Messages topics of Riscobiologico.org mailinglist-August 2000 to June 2005

N Percentage

General information and newsletter 1312 23

Information about courses, events,

and scientific meetings

713 13

Bibliography and educative material 438 8

General reports and contacts 161 3

Health care-related infections and issues 1301 23

Sterilization, disinfection, reuse 545 10

PPE and isolation precautions 304 5

Support areas (eg, laundry, pharmacy,

housekeeping)

246 4

Other 206 4

Occupational health and infections 880 16

Concepts and follow-up procedures 243 4

Occupational HIV/AIDS Infections 215 4

Immunizations for HCWs 144 3

Surveillance systems 71 1

Other 207 4

Medical waste 817 15

Legislation 397 7

Emergent diseases, health, and outbreak alerts 170 3

Miscellaneous 736 13

Total 5613 100.0

PPE, Personal protection equipment.

270 Vol. 35 No. 4 Rapparini et al

Brazil and 7 other countries have Portuguese as theofficial language, with more than 220 million peoplespeaking Portuguese worldwide.8 This fact supportsthe need of up-to-date scientific information in Portu-guese with easy access via the World Wide Web.

E-mail discussion groups on the Internet are newacademic forums in which knowledge and experiencecan be shared.6,7,9-11 Lists may be particularly usefulto disseminate health and outbreak alerts and also asan early warning system for emerging diseases, suchas the example of ProMED-mail from InternationalSociety for Infectious Diseases, an electronic outbreakreporting system that monitors infectious diseasesglobally.12 Information about emergent diseases, health,and outbreak alerts increased from less than 0.5% ofthe subjects of the mailing list messages in 2002 toalmost 9% in 2005.

Participants subscribed to this list for long periods oftime, with a mean duration of almost 2 years, suggest-ing a high interest and quality of the subjects discussed.Using e-mail to reach the target audience was an effi-cient strategy at the begging of the project; more than20% of invited professionals joined the list, and morethan 60% of those were still participating after 5 years.

This project was initially implemented to discussand provide information about management and pre-vention of occupational exposures and infectious dis-eases among HCWs, but other areas of interest werefrequently discussed, such as health care-related infec-tions and medical waste. This may represent a lack ofother forums to discuss these issues and the need ofadditional sources of information for HCWs involvedwith infection control. Overall, occupational healthand infections represented only 16% of messages sentin the list.

Although most of the participants (74%) reportedthat they frequently manage HCWs exposed to bloodand body fluids, the messages in the mailing list andanswers to the quizzes showed an important lack ofbasic knowledge regarding this subject. It was alarmingthat only nearly half of HCWs were able to give correctanswers about serologic screening recommended forsource patient testing and about interpretation of sero-logic HBV tests. Brazil has resources such as antiretro-viral drugs for postexposure prophylaxis, guidelines,and standardized protocols to manage exposures inthe health care setting. Nevertheless, our results dem-onstrated an important lack of knowledge amongHCWs who manage these exposures. The project ex-panded quickly, reaching HCWs from several differentcities and all Brazilian states in a short period of time.This finding indicates the prospect of using the WorldWide Web as an efficient tool to disseminate health-related information and the opportunity to monitorknowledge and practices in specific issues and to target

better the topics regarding educational programs.13 Inaddition, this model can be expanded to other settingsat which Portuguese language infection preventionmaterials are needed and access to health informationis limited. Furthermore, even beyond Portuguese-speak-ing nations, the experience reported in this study maybe adapted for users in countries without existinginfection prevention program infrastructure.

This project also aimed to implement a voluntarysurveillance system. Although this kind of surveillancesystem may lead to a biased sample that may notrepresent the national situation, results obtained fromsimilar systems implemented in developed countrieshave been useful to help the development of preventionstrategies.3 Presently, no national surveillance systemfor occupational infections among HCWs is in place inBrazil. In contrast with the high participation in themailing list, only 4 institutions began to participate inthe voluntary surveillance system. In conclusion, thisproject was successful in functioning as an online dis-cussion group. In addition, several strategies will betested to increase adherence to this voluntary surveil-lance system.

The authors thank Dr. Denise Cardo for invaluable assistance in the preparation ofthis article and for her contribution and participation in ‘‘Projeto Riscobiologico.org’’since its implementation in 2000 and Dr. Keith Sabin for his comments.

References

1. Pruss-Ustun A, Rapiti E, Hutin Y. Sharps injuries: Global burden of

disease from sharps injuries to health-care workers. Environmental

Burden of disease series No.3, WHO 2003:1-39.

2. Sepkowitz KA. Occupationally acquired infections in health care

workers. Part II. Ann Intern Med 1996;125:917-28.

3. Centers for Diseases Control and Prevention. Workbook for design-

ing, implementing, and evaluating a sharps injury prevention program.

2004.Available at: http://www.cdc.gov/sharpssafety/.Accessed July 1, 2005.

4. Netcraft. October 2005 Web server survey. 2005. Available at: http://

news.netcraft.com/archives/web_server_survey.html. Accessed Octo-

ber 25, 2005.

5. The Radicati Group Inc. Worldwide E-mail traffic, 2005-2009. Avail-

able at: http://www.radicati.com. Accessed July 10, 2005.

6. Burton PF. Electronic mail as an academic discussion forum. J Doc

1994;50:99-110.

7. Prasad S. Internet mailing lists: A primer. J R Coll Surg Edinb 2000;45:

122-6.

8. Ethnologue. Languages of the World, 2004. Available at: http://

www.ethnologue.com/home.asp. Accessed January 29, 2006.

9. Matzat U. Informal academic communication and scientific usage of

internet discussion groups. Available at: http://sosig.ac.uk/iriss/

papers/paper19.htm. Accessed July 10, 2005.

10. Worth ER, Patrick TB. Do electronic mail discussion lists act as virtual

colleagues? Proc AMIA Annu Fall Symp 1997;8280(97):325-9.

11. Culver JD, GerrF, FrumkinH.Medical informationon the Internet:A study

of an electronic bulletin board. J Gen Intern Med 1997;12:466-70.

12. Madoff LC. ProMED-mail: An early warning system for emerging

diseases. Clin Infect Dis 2004;39:227-32.

13. Canady JW, Cram A, Thompson SA, Barnes I, Ricciardelli E, Chang P.

Using Internet as an educational/survey tool. Plast Reconstr Surg

1996;98:908-9.

The infection control audit: Thestandardized audit as a tool for changeElizabeth Ann Bryce, MD, FRCPC, Sydney Scharf, RN, CIC, Moira Walker, RN, CIC, and Anne Walsh, RN

Vancouver, British Columbia, Canada

Background: Health care workers’ compliance with infection control practices and principles is vital in preventing the spread ofdisease. One tool to assess infection control practice in clinical areas is the infection control audit; however, many institutions donot approach this in a systematic fashion.Methods: Key features of the infection control audit were identified by the infection control team and developed into a standard-ized format for review of clinical areas. The audit incorporates a review of the physical layout, protocols and policies, knowledge ofbasic infection control principles, and workplace practice review.Results: Over the last 13 years, the infection control unit has completed 17 audits involving 1525 employees. Four-hundred-onestaff members have filled out questionnaires that assessed their understanding of standard precautions. A total of 257 recommen-dations have been made, and 95% of these have been implemented. The majority of recommendations address separation of cleanand dirty supplies, hand hygiene compliance, hand hygiene signage, proper use of barriers, and environmental cleaning.Conclusion: The infection control audit is an opportunity to implement changes and to introduce remedial measures in collabo-ration with various departments and services. A standardized approach to the audit allows benchmarking of practices across theinstitution and enhances standards of care. (Am J Infect Control 2007;35:271-83.)

Consistent adherence to infection control principlesis the means by which health care workers can protectthemselves and their patients. The infection control au-dit is an ideal vehicle to assess consistency of approachto infection prevention, and it has proven to be a usefulpart of infection control programs.1 The audit is an or-ganized examination of ward or service practices andprocedures that provides an opportunity to simulta-neously review safety in the workplace and identifyand remedy deficiencies. It is also an ideal time toreinforce and acknowledge those procedures and prac-tices that meet high standards of care. The purpose ofthis paper is to present our institution’s developmentof a standardized audit form to ensure the consistentand thorough application of key infection controlprinciples.

From the Division of Medical Microbiology and Infection Control,Vancouver General Hospital, British Columbia, Canada.

Address correspondence to Elizabeth Ann Bryce, MD, FRCPC, JPN1111, Division of Medical Microbiology and Infection Control, 899West 12th Ave., Vancouver, British Columbia, Canada. E-mail: elizabeth.bryce@vch.ca.

0196-6553/$32.00

Copyright ª 2007 by the Association for Professionals in InfectionControl and Epidemiology, Inc.

doi:10.1016/j.ajic.2006.05.293

MATERIALS AND METHODS

Background

Vancouver General Hospital is a 700-bed adulttertiary care facility for British Columbia, Canada, ad-mitting an average of 22,000 patients a year. The hospi-tal is the provincial transplantation center as well as thereferral institution for burns, neurosurgery, trauma,and spinal cord injury. The hospital employs, on aver-age, 7000 full-time employees (FTE), and the infectioncontrol service is composed of 3.5 FTE medical micro-biologists, one of whom is the infection control officer(ICO), and 3.75 FTE infection control professionals(ICP).

The audit structure

The audit tool was designed and developed over aseries of meetings with members of the infection con-trol team, consisting of ICPs, ICOs, and medical micro-biologists. Audit items were selected based on reviewof standards and guidelines from the Public HealthAgency of Canada (PHAC), the Association for Profes-sionals in Infection Control and Epidemiology, Inc(APIC), and the Hospital Infection Society (HIS) aswell as on practical experience. The audit tool wasshared with the Community Hospital Infection ControlAssociation-Canada (CHICA-Canada) members, andfeedback was requested. Prior to the audit, the patient

271

272 Vol. 35 No. 4 Bryce et al

Fig 1. Continued.

Bryce et al May 2007 273

Fig 1. Continued.

274 Vol. 35 No. 4 Bryce et al

Fig 1. Continued.

Bryce et al May 2007 275

Fig 1. Continued.

276 Vol. 35 No. 4 Bryce et al

Fig 1. Infection control audit form.

services coordinator and the medical director are ad-vised in writing that a formal audit of their work areais to be conducted, and a meeting is arranged to reviewthe audit process. It is stressed that confidentiality willbe respected and that a standardized audit form (Fig 1)will be used during the inspection and review processto ensure fairness and objectivity.

The audit includes an inspection of the physicalplant, a review of workplace infection control prac-tices, an assessment of health care workers’ knowledgeand application of infection control principles, and areport documenting deficiencies and required inter-ventions. All documented deficiencies are followed byrecommendations that are summarized in a worksheetformat with a completion date acceptable to both theaudited unit and the infection control team. The re-view team involves at least 2 ICPs and 1 medicalmicrobiologist.

Physical environment

Inspection of the physical environment consistsof a general examination of the layout of the unitwith emphasis on the soiled and clean utility rooms,medication room, and patient/resident rooms. Specialequipment such as oximeters, endoscopes, and gluc-ometers are checked for the presence of a regularcleaning schedule and for compliance. Design flawsthat may inhibit good infection control practice are

noted. The process requires several visits to assessproperly the levels of cleanliness and consistency incleaning practices.

Workplace practice review

The infection control practice of all staff is observedand evaluated using a standardized form to recordlapses in accepted practice (Fig 1). The appropriateuse of isolation rooms, proper hand hygiene, barrierprecautions, and waste disposal are observed anddocumented on at least 3 occasions, more frequentlyif deficiencies are initially noted. Policy and proceduremanuals are reviewed to determine that all policiesconform to current infection control standards. Thestandard audit form also includes appropriate use ofbarrier protection, specimen handling, decontamina-tion of body fluid spills, correct use of isolation proto-cols, and general staff appearance and attire. Otherpractices assessed when applicable include intravascu-lar line insertion, suctioning, urinary catheter care andinsertion, skin and wound care, and unit-related proce-dures. When practices are unique to a ward/depart-ment (eg, the morgue), these are covered with anaddendum to the standard form.

Over several visits, a minimum of 75 hand hygieneevents are observed by an ICP. Standardized criteriafor when hand hygiene is required are used. Episodesare recorded, using personal digital assistant software

Bryce et al May 2007 277

Fig 2. Hand hygiene survey.

(Pendragon Forms 3.1, Pendragon Software Corpora-tion, Buffalo Grove, IL) onto a hand hygiene auditform (Fig 2). No more than 3 events are recorded foran individual, and only the profession is documentedto maintain anonymity. The ICP is discreet duringauditing and generally is on the ward as part of theirdaily rounds. Sampling is opportunistic because notevery individual working on the unit is observed. The

intention of the hand hygiene audit is to primarilyprovide feedback in a positive manner to the unit andis not intended as a research endeavor.

Assessing knowledge and its application

Prior to any documented observations in the work-place, a questionnaire assessing routine precautions

278 Vol. 35 No. 4 Bryce et al

Fig 3. Continued.

Bryce et al May 2007 279

Fig 3. Infection control standard precautions survey.

280 Vol. 35 No. 4 Bryce et al

(Fig 3), frequency of education sessions and health careworker comprehension, and application of infectioncontrol procedures in the workplace is filled in anony-mously by staff. The questionnaire contains standardquestions as well as questions suitable for that specificwork environment. After completion of the audit, edu-cation sessions are scheduled to discuss the responsesto the knowledge and workplace practice assessmentquestions.

Report and recommendations

The audit report has a structured format consistingof a description of the unit, number of staff, numberof beds, and number of monthly admissions. The re-port follows the same flow as the review process:(1) description of the unit with its activities, (2) physicalenvironment, (3) workplace practices, and (4) infectioncontrol knowledge and its application. Findings aresummarized as a list of recommendations that are sup-ported with references and guidelines whenever possi-ble. A checklist of responsibilities and action dates areappended, and a meeting with the clinical area’s multi-disciplinary team is arranged to discuss the findingsand the recommendations. Typically, this will includea medical director, patient services manager, nurserepresentative, unit educator, respiratory therapist, andother ward/department specific staff (eg, morgue atten-dants). The meeting is an opportunity to discuss anycontentious issues prior to the final recommendations.The ICPs then visit the unit at regular intervals (generally3 and 6 months) to document progress in implementingthe various recommendations.

RESULTS

The infection control unit has completed 17 auditsat this institution within the last 13 years. The wardsand services reviewed included intensive care,

Table 1. Demographic information*

Demographic Result

Age range 20-65 yr

Years in health care ,5 to .30 yr (average 14.5)

Sex

Female 234 (60%)

Male 89 (23%)

Unknown 67 (17%)

Occupation

Nurse 187 (48%)

Physician 34 (9%)

Technologist/therapist 89 (23%)

Emergency support staff 43 (11%)

Other 37 (9%)

*N 5 390.

emergency department, orthopedic surgery, solid or-gan transplantation, outpatient clinics, spinal cordinjury unit, pathology and clinical laboratory services,morgue, food and nutrition services, occupational ther-apy kitchens, respiratory services, operating room, ra-diology, and hemodialysis. On average, each auditoccurs over 3 to 6 months and entails approximately50 to 100 ICP hours. Approximately, 1525 staff mem-bers have been contacted during the audit periods.Questionnaires on routine precautions have been com-pleted by 401 staff.

There were 390 forms available for demographicanalysis of which 60% were submitted by nursing staff(Table 1). The age range was 20 to 65 years with an av-erage of 14.5 years in health care. Approximately halfof respondents had attended an infection control edu-cation session; however, the majority felt that theyunderstood basic principles of infection control, partic-ularly barrier use (Table 2). When formal knowledgewas assessed, however, deficiencies were noted inapplication of routine precautions, barrier use for air-borne infections, and hand hygiene practice (Table 3).

A total of 257 recommendations have been made,and 95% have been acted on (Table 4). The most com-mon recommendations focused on proper environ-mental cleaning (11%), proper equipment cleaningprotocols (11%), correct use of personal protectiveequipment (11%), hand hygiene procedures (8%),and separation of clean and soiled supplies (8%). In

Table 2. Infection control education*

Parameter

No. and percentage

responding affirmatively

Attended infection control session 197 (49%)

Sufficient information received 143 (36%)

Personal understanding

of infection control principles

317 (79%)

Personal understanding

of correct barrier use

334 (83%)

*N 5 401 respondents.

Table 3. Infection control knowledge assessment*

Area of assessment Correct response, n (%)

Sharps injury policy 321 (85)

Sharps exposure management 302 (80)

Sharps handling and disposal 310 (82)

Routine precautions

appropriately applied

177 (47)

Personal protection

requirements for tuberculosis

185 (49)

Body fluid spill cleanup procedure 311 (83)

Hand hygiene practice 257 (68)

*N 5 377.

Bryce et al May 2007 281

many instances, once a problem or deficiency wasidentified, it was corrected prior to completion of theaudit.

Large scale improvements resulting directly fromthe audit process include the introduction of standard-ized orders for antimicrobial prophylaxis (arising fromthe first operating room audit in 1996); the addition ofalcohol handrub dispensers throughout the institutionbeginning in 1995 (to address observed lack of handhygiene stations); the introduction of a template in1998 to address additional costs associated with en-hanced infection control measures in the tenderingprocess for new construction and renovations (a re-sponse to physical plant design constraints observed inthe audits); and, more specifically, the addition of 102negative-pressure rooms in the new acute care tower(20% of all beds). Other improvements arising fromaudit recommendations include use of the hands-freetechnique for passing sharps (2000); enhancedpersonal protective equipment and new safety pro-tocols in the autopsy suites (1999); revised protocolsfor cleaning of dialysis machines; and introduction ofsyndromic surveillance for gastrointestinal andrespiratory infections in the emergency department(2003).

DISCUSSION

Audits in infection control have received relativelylittle attention, although the area should be an idealsubject because of its focus on patient and healthcare worker safety, the availability of standards bywhich to measure the quality of care, and the abilityto document improvement in practice. Unfortunately,most audits involving infection control focus on envi-ronmental cleanliness rather than encompassing unitprocedures as they apply to the practice of infectionprevention.2 To be truly effective, an audit must consistof a topic, appropriate practice standards, observationand testing against the selected standards, identifica-tion of areas for improvement, and subsequentinterventions and demonstration of improvement inpractice.3-5

Following completion of the first 3 infection controlaudits in the early 1990s, it was apparent that there wasa need for a more consistent and organized review pro-cess. The audit form had been shared with members ofCHICA-Canada and feedback requested, but this wasnot a formal verification or validation process.6 Thus,the audit tool has not been verified, and this is anacknowledged limitation. The infection control ser-vices have recently been regionalized, and plans areunderway to review the audit document, weight theobservations, and develop a scoring system. This willaddress the urgency of a particular action and the

risk to the patient and/or staff. Construct validity andinterrater reliability will also be assessed as part ofthis process.

It must be emphasized that the infection controlaudit presents an opportunity to promote infectionprevention and control improvement activities in part-nership with an organization’s multidisciplinary teams.Issues that influence the prioritization of the ward/department to be audited include acuity of patientcare, central venous catheter-associated bloodstreaminfections, surgical wound infection rates and highrates of Clostridium difficile or antibiotic-resistant orga-nisms, and date of last audit.

The audit is comprehensive and includes inspectionof the physical plant, review of workplace infectioncontrol practices, and assessment of health careworkers’ knowledge and application of infectioncontrol principles. The observational period is thenfollowed by identification of areas for improvement,involvement of staff in the report writing process,and recommendations for further intervention. Al-though initially the process may seem daunting, theprolonged period of time for the audit is designed toaccount for the busy schedule of the ICP and the needto incorporate the audit as part of their routine for thenext several months. It is similar to planning a calendarof educational sessions, but instead, unit or ward visitsare scheduled into the ICP’s day. In this staged man-ner, the audit becomes less intimidating, particularlybecause the ICPs can incorporate some of the observa-tional tasks as part of their daily rounds. The prolongedperiod of observation likely more accurately reflectstrue unit practices compared with a set of observationsat a single point in time.

An assessment of the physical layout of any unitis necessary to determine the ease or difficulty withwhich staff can maintain a safe, clean environment

Table 4. Frequency of infection controlrecommendations*

Category of recommendation

Frequency,

n (%)

Environmental cleaning procedures 29 (11)

Cleaning of equipment 28 (11)

Use of protective equipment 27 (11)

Hand hygiene practice 21 (8)

Separation of clean and soiled supplies 21 (8)

Accessible protective equipment 16 (6)

Handling and disposal of sharps 13 (5)

Multidose vial procedures 7 (3)

Prohibition of food and drink in work area 6 (2)

Isolation practices 5 (2)

Other (eg, traffic flow, untidiness, isolation signage) 84 (33)

*N 5 257.

282 Vol. 35 No. 4 Bryce et al

and prevent cross infection. Good or bad design canaffect one’s ability to maintain a clean environment,and clinical areas must be designed to facilitate goodwork habits. Recommendations for changes to thephysical environment are consistent with The Ameri-can Institute of Architects Academy of Architecturefor Health guidelines and are used to assist with futurerenovations.7 Consultation between facility planningand infection control is now a regular outcome of theaudit process.

Review of workplace practices is evaluated againstexisting standards and current facility-specific infec-tion control policies and procedures. The most com-monly found deficiencies in the areas reviewedinclude incorrect storage of clean and dirty supplies,poor compliance with handwashing, incorrect use ofbarriers, inadequate cleaning of shared equipment,and consumption of food and beverages in the workarea. Ways in which deficiencies are addressed includereview of work practice with staff, reorganization ofservice areas to improve infection control compliance,and revision of protocols or procedures. For example,confusion over cleaning of electronic equipment suchas oximeters and glucometers led to changes in clean-ing protocols and assignment of housekeeping person-nel to the task. Identification of poor aseptic techniquewhen accessing multidose lidocaine vials on certainwards led to the immediate withdrawal of this itemfrom unit stock. Documentation of inappropriatemask use and review of exposure risk to airborne dis-ease in our institution led to increased promotion ofrespirators for high-risk respiratory care and trainingin their use (prior to severe acute respiratory syn-drome). The latter intervention was felt to be a highpriority because the region accounted for 38% of allnewly diagnosed tuberculosis cases in the provinceand our institution admitted an average of 64 cases ofactive respiratory tuberculosis a year.8 These examplesof early successes were a direct outcome of the auditprocess and may have contributed to the facility’ssuccessful infection control management of severeacute respiratory syndrome cases.9 All of these specificexamples were documented as completed on 3- or 6-month follow-up to ‘‘close the loop’’ between observa-tion, recommendation, and implementation.

A recent addition to the audit process is the observa-tion of opportunities for hand hygiene during a 1-hourperiod on 2 or 3 occasions. Compliance rates for handhygiene practices vary from 28% to 60% across all pro-fessions. Varying compliance rates and hand hygienepractices are highlighted during the postaudit meet-ings, and staff are invited to provide feedback and sug-gestions for improving compliance. This componentof the audit allows for the opportunity to comparethe practice of other units within the same institution

and is excellent material for postaudit education ses-sions. It is difficult to determine whether the handhygiene audits and feedback have improved healthcare worker compliance, partly because of the numer-ous barriers and facilitators that affect intent to complywith this simple yet effective measure and partlybecause the intent of the audits was for feedback ratherthan as independent evaluation of compliance overtime.

A major component of infection control is educationof the health care worker on routine precautions. Com-prehension of infection control principles is vital to theprotection of both staff and patients. The purpose ofthe knowledge and practice survey as part of the auditprocess is to ascertain the level of infection controlknowledge, to determine whether perception of knowl-edge is genuine, and to evaluate whether knowledge isapplied in the workplace setting. The questionnaire hasbeen particularly useful in detecting areas that requirefurther attention. Infection control arranges educationsessions to review the correct responses with unitpersonnel and circulates FAQ sheets for each questionfollowing collection of all the survey forms.

In retrospect, the behavioral and knowledge compo-nents in the knowledge assessment form would havebeen better if items had been rated on a scale ratherthan as a simple yes or no question. This is a limitationof the form from an analytical standpoint; however,it must be emphasized that this particular audit toolserves mainly to inform the infection control team ofgaps in knowledge and deficiencies in practice, whichthen can be communicated back to the ward or unit.Future plans for the knowledge assessment forminclude a factor analysis to verify construct validityand revision to allow for further assessment of changesin knowledge over time. This then would allow formore focused intervention with detailed observationsand research questions.

One illustration of the ability of the knowledge andpractice assessment form to identify areas for improve-ment was the disappointing results regarding atten-dance at infection control education sessions andcomments regarding the subject matter presented inthe early audits. It clearly documented a need to revisethe infection control unit’s approach to content deliv-ery and provided the impetus to create an on-lineinfection control education module and infection con-trol manual. The latter has recently received funding toassess infection control knowledge retention over time.

Involvement of health care workers from the onsetof the audit and review of the audit findings with feed-back by staff prior to the final audit draft should bestressed. Participation in the process by the clinicalareas facilitates acceptance and completion of recom-mendations in a timely fashion. The nursing unit

Bryce et al May 2007 283

manager, the medical director, the nurse educator, andthe infection control designate on the ward are in-volved at the beginning of the audit process and arekey participants in the development of action plans.

The goal of the infection control team is to revieweach clinical area on a 5- to 7-year cycle, similar tothe hospital accreditation process. Documentationand consistency of approach should allow the teamto focus on previous deficiencies and to compare infec-tion control practice over time. This then would com-plete the audit cycle. The concept of regular reviewhopefully will also emphasize the importance of incor-poration of infection control practices as routine.

The infection control audit can be a daunting task.A standardized protocol for the audit process providesa template for an impartial, organized, structured, andthorough review. Uniformity of approach, front-linepresence of the ICP on the unit, involvement of theclinical area under review, and documentation ofresults have been effective in promoting and improvingstandards of infection control in this institution.

References

1. Millward S, Barnett J, Thomlinson D. Evaluation of the objectivity of an

infection control audit tool. J Hosp Infect 1995;31:229-33.

2. Malik RE, Cooper RA, Griffith CJ. Use of audit tools to evaluate the

efficacy of cleaning systems in hospitals. Am J Infect Control 2003;31:

181-7.

3. Hay A. Audit in infection control. J Hosp Infect 2006;62:270-7.

4. French GL. Closing the loop: audit in infection control. J Hosp Infect

1993;24:301-8.

5. Millward S, Barnett J, Thomlinson D. A clinical infection control audit

programme: evaluation of an audit tool used by infection control nurses

to monitor standards and assess effective staff training. J Hosp Infect

1993;24:219-32.

6. Infection Control Practitioners’ audit form for patient/resident service

units. Can J Infect Control 2002;17:23-6.

7. The American Institute of Architects Academy of Architecture for

Health. Guidelines for design and construction of hospital and health

care facilities. The Facilities Guidelines Institute, 2001 edition. Washing-

ton, DC.

8. B. C. Centre for Disease Control. Annual report division of tuber-

culosis control, 1999-2000. Ministry of Health, British Columbia,

Canada.

9. Skowronski DM, Petric M, Daly P, Ra Parker, Bryce E, Doyle P, et al.

Coordinated response to SARS, Vancouver, Canada. Emerg Infect Dis

2006;12:155-8.

Of viruses, gloves, and crepesTo the Editor:

Dr. V. Wiwanitkit should be commended for hisattempt to estimate some bloodborne viral pathogens’(HIV, hepatitis B virus, hepatitis C virus, and hepatitis Dvirus) ability to pass through gloves, based on a consider-ation at the nanostructure level (AJIC 2006,34;(6):400).Unfortunately, the size of pre-use glove pores in relationto that of the viruses is far from being the main para-meter involved in gloves providing adequate protectionto personnel and patients during health care.

Most gloves that are used throughout the world,especially sterile ones, are made of latex (others aremade of vinyl or nitrile). Latex gloves are manufacturedby applying the thinnest possible coating of latex ontohand-shaped forms before drying. Making pancakes–or rather crepes as the French would have it–providesa suitable illustration: the thinner the desired crepe(glove layer), the more liquid the crepe batter (poly-mer) needs to be, and the more likely it is that holeswill appear spontaneously in the crepe’s structure whenthe dough is distributed on the surface of the pan sothat it is entirely covered. Furthermore, additional mi-cropores may appear during the polymerization stage.

Therefore, gloves are permeable even prior to use, ina varying proportion that depends on their composi-tion, the manufacturer, and the quality of the glove.1,2

The physical properties of gloves are further alteredwith exposure to body heat, humidity, and mechanicalconstraints, whether in the surgical setting or whileexamining patients. Studies on surgeons’ gloves haveshown that this proportion increases with use, evenin the absence of documented needlestick.3,4 Thus,pre-existing holes or in-use tears in the clinical settingmay be significantly larger than are the viruses de-scribed by V. Wiwanitkit.

In the clinical, nonsurgical setting, wearing a singlelayer of quality gloves on intact skin and followed byhand hygiene procedures can be considered to confera high degree of protection. For the reasons describedabove, double gloving must be the rule in the surgicalsetting. Ideally, both layers must be changed at leastevery hour, to provide protection against leakage, expo-sure to blood if holes appear spontaneously, and incase of needlestick.5-7

Having access to a sufficient and uninterruptedsupply of gloves is a considerable challenge in resource-

constrained health care settings.8 Unfortunately, manyof these are located in high-prevalence areas, and thescarcity of health care resources further concentrateshighly viremic patients in hospital wards. Implementingbasic, well-understood, and sustainable infection con-trol measures—including gloving when necessary—should be considered a priority in all developing countriesto protect health care workers and patients alike fromthese pathogens as well as undiscovered ones.9,10

Arnaud Tarantola, MD, MScDepartement International et TropicalInstitut de Veille Sanitaire12 rue du Val d’Osne94415 Saint-Maurice cedex, France

E-mail: a.tarantola@invs.sante.fr

References

1. Nelson JR, Roming TA, Bennett JK. A whole-glove method for the

evaluation of surgical gloves as barriers to viruses. Am J Contact

Dermat 1999;10:183-9.

2. Korniewicz DM, El-Masri MM, Broyles JM, Martin CD, O’Connell KP.

A laboratory-based study to assess the performance of surgical gloves.

AORN J 2003;77:772-9.

3. Gerberding JL, Littell C, Tarkington A, Brown A, Schecter WP. Risk

of exposure of surgical personnel to patients’ blood during surgery

at San Francisco General Hospital. N Engl J Med 1990;322:1788-93.

4. Wright JG, McGeer AJ, Chyatte D, Ransohoff DF. Mechanisms of glove

tear and sharps injuries among surgical personnel. JAMA 1991;266:

1668-71.

5. Bennett B, Duff P. The effect of double gloving on frequency of glove

perforations. Obstet Gynecol 1991;78:1019-21.

6. Bennett NT, Howard RJ. Quantity of blood inoculated in a needlestick

injury from suture needles. J Am Coll Surg 1994;178:107-10.

7. Tanner J, Parkinson H. Double gloving to reduce surgical cross-infec-

tion. Cochrane Database Syst Rev 2002;(3):CD003087.

8. Danchaivijijr S, Tangtrakool T, Chokloikaew S, Thanilikitkul V. Univer-

sal precautions: costs for protective equipment. Am J Infect Control

2001;25:44-50.

9. Tarantola A, Rachline A. Surveillance and training, not postexposure

prophylaxis, are the basis for the prevention of occupational infection

by blood-borne pathogens in developing countries. J Hosp Infect

2005;60:91-2.

10. Tarantola A, Koumare A, Rachline A, Sow PS, Diallo MB, Doumbia S,

et al. A descriptive, retrospective study of 567 accidental blood expo-

sures in healthcare workers in three West African countries. J Hosp

Infect 2005;60:276-82.

doi:10.1016/j.ajic.2006.08.004

0196-6553/$32.00

Copyright ª 2007 by the Association for Professionals inInfection Control and Epidemiology, Inc.

284

In response to SilvestriTo the Editor:

We would like to thank Dr. Silvestri and colleagues fortheir letter and their contributions to the literature re-lated to selective decontamination of the digestive tract

Municipal public telephones (n 5 100) were selectedin the inner Belfast area during the summer periodof 2006. Two swabs were taken of each telephone, bypremoistening each sterile cotton swab in sterilephosphate-buffered saline solution and by thoroughlyswabbing the ear- and mouthpiece of each telephoneapparatus. Swabs were transported to a laboratorywith 2 hours of sampling, after which they were en-riched in both salt-cooked meat broth (Oxoid CM0094;Oxoid Ltd, Basingstoke, United Kingdom) and in MASTMRSA ID broth (MAST Diagnostics Ltd., Merseyside,United Kingdom) for 24 hours at 378C, in accordancewith the manufacturer’s instructions. After this time,a 20-mL inoculum from the salt-cooked meat broth en-richment was plated onto colomycin selective agar forthe detection of presumptive MRSA colonies. Appropri-ate positive and negative controls were established. NoMRSA was detected in any public telephone examinedby either of the enrichment methods employed.

There have been relatively few studies on thecontamination of public telephones with MRSA.Previously, a report from Turkey5 identified severalS aureus isolates on a study of 50 phones, but themethods employed were not able to determinewhether or not these isolates were resistant to methicil-lin. Additionally, a report on bacterial contaminationon mobile telephones among health care workers6

Letters to the Editor May 2007 285

(SDD). We agree with their assertion that use of SDD,when both enteral and parenteral antibiotics arecombined, is associated with lower rates of ventilator-associated pneumonia and improved mortality. Wealso acknowledge that SDD remains controversialbecause of the concerns regarding the potential for fur-ther emergence of resistant organisms. We believe thatindividual clinicians and hospitals can make up theirown mind on the use of SDD based on their criticalappraisal of the literature. However, in our opinion,there are insufficient long-term data on the emergenceof resistant organisms to recommend widespread androutine use of selective digestive decontamination. Wesuspect that, in time, data such as those presented byDr. Silvestri and colleagues may help resolve this issue.

Scott A. Flanders, MDUniversity of MichiganAnn Arbor, MI

Sanjay Saint, MD, MPHAnn Arbor VA Medical CenterUniversity of MichiganAnn Arbor, MI

Harold R. Collard, MDUniversity of California, San Francisco

doi:10.1016/j.ajic.2006.10.009

Lack of occurrence of methicillin-resistant Staphylococcus aureuson municipal public telephonesTo the Editor:

Nosocomially acquired methicillin-resistant Staphy-lococcus aureus (MRSA) is recognized as a bacterialpathogen with a significant increase in morbidity andmortality.1 More recently, community-acquired MRSAhas emerged as a significant pathogen in the commu-nity, particularly associated with infection in younghealthy adult populations, for example, players on foot-ball and soccer teams.2 Although these organisms havebeen associated with a variety of clinical infections,asymptomatic MRSA carriage rates of 0.84% (UnitedStates data)3 and 1.5% (United Kingdom data)4 in other-wise healthy individuals in the community have beenreported. From time to time, as health care microbiolo-gists, we are asked by patients and members of the pub-lic, with an anxiety of acquiring MRSA, whether publictelephones are a reservoir of this organism? To date,there have been relatively few studies examining theoccurrence, persistence, and survival kinetics of MRSAin the environment. It was therefore the aim of thisstudy to examine the occurrence of MRSA on publictelephones in and around the inner Belfast area.

identified 2 phones (2/105; 1.9%) with MRSA; however,given that mobile phones are used personally and arenot commonly shared among workers or from healthcare workers to patients, the risk of transmission ofMRSA from person to person is less likely, which isnot the case with municipal public telephones.

In conclusion, although negative in its findings, thisstudy demonstrated that, within the community exam-ined, MRSA contamination of public telephones is notsignificant. The significance and impact of this studythus helps us in our epidemiologic understandingthat public phones are not important reservoirs orvehicles for the transmission of these organisms withinthe community. There is relatively limited informationon the reservoirs and survival kinetics of nosocomiallyacquired MRSA and, more importantly, community-acquired MRSA in the community outside the healthcare environment. Further work is urgently requiredto identify other environmental reservoirs of MRSA inthe community, their routes of transmission, and thesurvival kinetics of these organisms in the communityso that we have a better understanding of how theybehave/survive occultly in the community.

Anna O’Connor,a,b

Anne Loughrey, MBa

B. Cherie Millar, PhDa

Colm J. Lowery, DPhilb

James S. G. Dooley, PhDb

Colin E. Goldsmith, MBa

Paul J. Rooney, MBa

John E. Moore, PhDa

Northern Ireland Public Health Laboratory,a

Department of Bacteriology, Belfast City Hospital,Lisburn Road, Belfast, Northern Ireland, BT9 7AD,United Kingdom; School of Biomedical Sciences,b

University of Ulster, Cromore Road, Coleraine, Co.Londonderry, Northern Ireland, BT52 1SA, UnitedKingdom

E-mail: jemoore@niphl.dnet.co.uk

Supported financially by a Wellcome Trust Vacation Scholarship(to A.O.C).

References

1. Brown DF, Edwards DI, Hawkey PM, Morrison D, Ridgway GL, Towner

KJ, et al, and the Joint Working Party of the British Society for Antimi-

crobial Chemotherapy; Hospital Infection Society; Infection Control

Nurses Association. Guidelines for the laboratory diagnosis and sus-

ceptibility testing of methicillin-resistant Staphylococcus aureus (MRSA).

J Antimicrob Chemother 2005;56:1000-18.

2. Cohen PR. Cutaneous community-acquired methicillin-resistant Staph-

ylococcus aureus infection in participants of athletic activities. South

Med J 2005;98:596-602.

3. Mainous AG III, Hueston WJ, Everett CJ, Diaz VA. Nasal carriage of

Staphylococcus aureus and methicillin-resistant S aureus in the United

States, 2001-2002. Ann Fam Med 2006;4:132-7.

4. Abudu L, Blair I, Fraise A, Cheng KK. Methicillin-resistant Staphylococcus

aureus (MRSA): a community-based prevalence survey. Epidemiol Infect

2001;126:351-6.

5. Tunc K, Olgun U. Microbiology of public telephones. J Infect 2006;53:140-3.

6. Brady RR, Wasson A, Stirling I, McAllister C, Damani NN. Is your phone

bugged? The incidence of bacteria known to cause nosocomial infection

on healthcare workers’ mobile phones. J Hosp Infect 2006;62:123-5.

doi:10.1016/j.ajic.2006.09.015

286 Vol. 35 No. 4 Letters to the Editor

Assessing vaccination rates atan emergency roomTo the Editor:

Vaccination is among the most cost-effective tools todecrease childhood morbidity-mortality. Nonetheless,this effectiveness is dependent on vaccination cover-age. Missed opportunities (MO) for vaccinations occurwhen a vaccine-eligible child does not receive theneeded vaccines, and their contribution to under im-munization is significant.1

To assess the vaccination rates among children ad-mitted to the emergency room of a university hospi-tal in Salvador, Brazil, between November 2001 andNovember 2003, a retrospective cross-sectional studywas conducted. The medical records were reviewed.The patients had been evaluated by medical studentssupervised by the same professor (C. M. Nascimento-Carvalho). A standardized form was used, and inquiryregarding vaccine use was included in the form. TheBrazilian Ministry of Health vaccine schedule was

used as standard. The study was approved by the Insti-tutional Review Board.

Out of 267 patients, 122 (45.7%) showed the vac-cines card, 87 (32.6%) informed on vaccines use, and58 (21.7%) did not provide any information; overall,31.6% (66/209) had the vaccination schedule incom-plete (42.6% with and 16.1% without card). The me-dian age (years) was 1.7 (mean, 2.6 6 2.6 years), and53.9% were males. Overall, association between youngage (years) with incomplete vaccination schedule wasidentified (1.6 6 1.6 vs 2.7 6 2.7 years, respectively,95% CI mean difference, 0.6-1.8; P 5 .006). Thisassociation was also found by analyzing vaccine card-holders or vaccine-use reports separately. The medianfamily income was US $227.30 (mean US, $276.40 6

$166.82). The family income (US $) was greater amongchildren with complete vaccine schedule, by analyzingcardholders ($302.73 6 $167.73 vs $230.45 6 $160.00,respectively, 95% confidence interval mean difference,8.6-127.3; P 5 .02). The frequency of delayed vaccineswere MMR, 20.8%; Haemophilus influenzae type b firstdose, 17.3%; BCG, 11.5%; hepatitis B second dose, poliofirst dose, HIB second dose, tetra first dose, tetra seconddose, 9.6% each; and hepatitis B first dose, polio seconddose, and DPT first booster, 7.7% each.

It is noteworthy that almost half (45.7%) of thepatients included in this study presented the vaccinescard, even looking for health care at an emergencyroom, and this made accurate assessment for vaccinecoverage possible. Another study analyzing data fromdifferent settings has reported similar rate for incom-plete vaccination schedules: 40.4%.2 Different ratesof vaccination coverage have been found by consider-ing vaccine documentation or patients report, with atrend for underestimation for patient report,2 as itwas observed herein. The association of young agewith incomplete vaccine schedule highlights howimportant it is to decrease MO because the youngerthe child the greater is morbidity-mortality, includingvaccine-preventable diseases.1 In general, the lowerthe vaccination coverage and the higher the burdenof vaccine-preventable diseases in a population, thegreater the need to improve coverage; urban, low-socioeconomic-status populations are particularly vul-nerable to vaccine-preventable diseases,3 which is inaccordance with the association of lower income andincomplete vaccine schedule found in this study. Im-proving coverage in impoverished urban communitiesshould be a priority.3

MO was one of the identified barriers that made itdifficult to vaccinate fully the young children and con-tributed to the measles epidemics of 1989 and 1990 inthe United States.4 Specific recommendations to elim-inate MO have been published in 1993 and includedthe use of all clinical encounters, including visits for

skin integrity, which occurs more frequently with 72-hour changes, may contribute to this result. We havesupported their conclusions in a recent randomizedcontrolled trial, in which the incidence of phlebitiswas similar in the 3-day change group and the changewhen clinically indicated group.4 Among those whohad their peripheral catheter removed for phlebitis,the mean length of time that the catheter was in situwas 48.7 hours.

We believe that, if patients are not matched forrisk factors that may influence outcomes, incorrectconclusions may be drawn. This could have con-siderable patient care and economic implications.Consequently, it is important to use the correct studydesign when trying to understand significant healthcare questions.

Joan Webster, RN, BANursing Director, ResearchRoyal Brisbane and Women’s HospitalHerston, QLD, Australia

E-mail: joan_webster@health.qld.gov.au

Sonya Osborne, RN, BN, MNLecturerQueensland University of TechnologyVictoria Park RoadKelvin Grove, Queensland, Australia

E-mail: s.osborne@qut.edu.au

References

1. Malach T, Jerassy Z, Rudensky B, Schlesinger Y, Broide E, Olsha O, et al.

Prospective surveillance of phlebitis associated with peripheral intrave-

nous catheters. Am J Infect Control 2006;34:308-12.

2. Bregenzer T, Conen D, Sakmann P, Widmer AF. Is routine replacement

of peripheral intravenous catheters necessary? Arch Intern Med 1998;

158:151-6.

3. Homer LD, Holmes KR. Risks associated with 72- and 96-hour

Letters to the Editor May 2007 287

mild illnesses, for the provision of needed immuni-zation.5 In a nationwide study, there was no differencein a state preference for receiving a needed immu-nization during an illness visit.6 Expanding access isstrongly recommended on the basis that it improvesvaccination coverage.3 The authors propose the screen-ing of the immunization status during health careassistance and the promotion of vaccine use duringmild illnesses at emergency rooms as a strategy toreduce MO.

V. M. M. Miranda, MDC. M. Nascimento-Carvalho, MD, PhD

Department of Pediatrics, School of Medicine, FederalUniversity of Bahia, Salvador, Bahia, Brazil

E-mail: nascimentocarvalho@hotmail.com

References

1. Sabnis SS, Pomeranz AJ, Lye PS, Amateau MM. Do missed opportunities

stay missed? A 6-month follow-up of missed vaccine opportunities in

inner city Milwaukee children. Pediatrics 1998;101:E5.

2. Tugumisirize F, Tumwine JK, Mworozi EA. Missed opportunities and

caretaker constraints to childhood vaccination in a rural area in Uganda.

East Afr Med J 2002;79:347-54.

3. Task Force on Community Preventive Services. Recommendations

regarding interventions to improve vaccination coverage in children,

adolescents, and adults. Am J Prev Med 2000;18:92-6.

4. The National Vaccine Advisory Committee. The measles epidemic:

the problems, barriers, and recommendations. JAMA 1991;266:

1547-52.

5. Ad Hoc Working Group for the Development of Standards of Pediatric

Immunization Practices. Standards for pediatric immunization practices.

JAMA 1993;269:1817-22.

6. Taylor JA, Darden PM, Brooks DA, Hendricks JW, Wasserman RC,

Bocian AB. Association between parents’ preferences and perceptions

of barriers to vaccination and the immunization status of their children:

a study from Pediatric Research in Office Setting and the National

Medical Association. Pediatrics 2002;110:1110-6.

doi:10.1016/j.ajic.2006.10.008

Phlebitis associated with peripheralintravenous cathetersTo the Editor:

Malach et al1 suggest that presence of an intrave-nous peripheral catheter longer than 3 days is a riskfactor for phlebitis. A point-prevalence research designwas used in their study whereby patients with phlebitiswere compared with an unmatched control group ofpatients who did not have phlebitis. There are signifi-cant problems with drawing strong conclusions fromsuch a design, which the authors themselves acknowl-edge. Other prospective, longitudinal studies havefound that it is within the first 2 days following periph-eral catheter insertion that the patient is at highest riskfor infection.2,3 These authors surmise that breaching

peripheral intravenous catheter dwell times. J Intraven Nurs 1998;

21:301-5.

4. Webster J, Lloyd S, Hopkins T, Osborne S, Yaxley M. Developing a

Research base for intravenous peripheral cannula re-sites (DRIP trial).

A randomized controlled trial of hospital in-patients. Int J Nurs Stud

2006. In press.

doi:10.1016/j.ajic.2006.09.013

In response to Webster and OsborneTo the Editor:

We appreciate the comments made by Webster andOsborne. Rather than establish the optimal duration ofperipheral intravenous catheters, the purpose of ourstudy was to determine the rate of phlebitis and associ-ated risk factors and, through educational intervention,reduce the rate of phlebitis.1 These aims were achieved.

288 Vol. 35 No. 4 Letters to the Editor

Duration of an intravenous catheter of more than3 days was found to be a risk factor for phlebitis,therefore suggesting that routine replacement beforethis time limit could help reduce the rate of phlebitis.We agree with Webster and Osborne’s comment thatprospective longitudinal studies are superior for deter-mination of the optimal duration of intravenous cath-eters. Their elegant study2 seems to confirm thefinding of other studies3 that replacement when indi-cated is clinically safe and cost-effective. However,their study evidently involved a phlebotomy team,the use of which is associated with a lower rate ofcomplications than elsewhere, such as in our hospital,in which house staff insert intravenous catheters.4,5

In such settings, daily examination of the site andscrupulous observation of the 72-hour limit forreplacement may be called for, especially if point-prevalence surveys indicate a relatively high rate ofphlebitis.

Ziona Jerassy, RN, MABernard Rudensky, PhDDavid Raveh, MDAmos M. Yinnon, MD

Correspondence should be addressed to: Prof. A. M.Yinnon, Infectious Disease Unit, Shaare ZedekMedical Center, POB 3235, Jerusalem 91031, Israel

E-mail: Yinnon@SZMC.org.il

References

1. Malach T, Jerassy Z, Rudensky B, Schlesinger Y, Broide E, Olsha O, et al.

Prospective surveillance of phlebitis associated with peripheral intrave-

nous catheters. Am J Infect Control 2006;34:308-12.

2. Webster J, Lloyd S, Hopkins T, Osborne S, Yaxley M. Developing a

research base for intravenous peripheral cannula re-sites (DRIP trial):

a randomized controlled trial of hospital in-patients. Int J Nurs Stud

2006. Available at: www.sciencedirect.com; www.elsevier.com/locate/

ijnurstu. Accessed.

3. Bregenzer T, Conen D, Sakmann P, Widmer AF. Is routine replacement

of peripheral intravenous catheters necessary? Arch Intern Med 1998;

158:151-6.

4. Bekeris LG, Tworek JA, Walsh MK, Valenstein PN. Trends in blood cul-

ture contamination: a College of American Pathologists Q-tracks study

of 356 institutions. Arch Pathol Lab Med 2005;129:1222-5.

5. Surdulescu S, Utamsingh D, Shekar R. Clin Perform Qual Health care

1998;6:60-2.

doi:10.1016/j.ajic.2006.11.005