An Outbreak of Hepatitis A among

Health Care Workers: Risk Factorsfor Transmission

Bradley N. Doebbeling, MD, MS, Ning Li MB, MS, and Richard P.Wenzel, MD, MSc

IntroductionNosocomial outbreaks of hepatitis A

are relatively uncommon. Several food-related epidemics have been reported.",2Most nosocomial outbreaks have in-volved exposure either to an asymptom-atic infant or to a young child, typicallyafter the transfusion of blood from anasymptomatic donor.3-8 Occasionally,outbreak clusters have been reported aftercontact with an older child or adult withvomiting, diarrhea, or fecal incon-tinence.9-'4 Several large outbreaks inneonatal intensive-care units have beenreported recently.4,5,7'8 Despite extensiveinvestigations, most previous reportsfailed to identify significant risk factors fornosocomial transmission, other than asso-ciated care of the infected individual.Rosenblum and colleagues8 recently re-ported an outbreak of hepatitisA in a neo-natal intensive-care unit, implicating thebehaviors of drinking beverages in the unitand failure to wear gloves routinely as riskfactors for transmission to health careworkers.

We report here the first nosocomialoutbreak of hepatitisA in abum treatmentcenter. The outbreak occurred betweenJanuary and March 1990 in a tertiary-carehospital. A 32-year-old man and his8-month-old son were admitted to thebum treatment center on December 9,1989, after each had sustained large bodysurface area bums in a house fire, 70% and40%, respectively. On January 8, 1990,the father became jaundiced, and bothwere found to have acute hepatitis A in-fections. Despite institution of appropriateisolation precautions,15 11 health care

workers and 1 other bum patient becameclinically ill with hepatitis A. We under-took an epidemiologic investigation to

identify important risk factors for trans-

mission in the hospital. Our investigationis the subject of this report.

BackgroundThe hospital is a 900-bed referral cen-

ter with a 16-bed bum treatment center.The typical nursing staff to patient ratio onthe burn treatment center is approxi-mately 1:3, although it varies dependingon the acuity of the patient and the time ofday. Health care workers regulariy as-signed to the bum treatment center ofteneat in the staff lounge or in a chartingroom, located approximately 15-50 feetfrom most of the patient rooms (Figure 1).Staff food is stored in a refrigerator in adesignated staff food room, the stafflounge, or occasionally in the patientnourishment room. Infants and youngchildren with bums are allowed to play ina playpen and eat beside the desks in thecenter of the bum treatment area. Addi-tionally, outpatients with bums are rou-tinely seen in follow-up on the ward andoften bring in snack foods to share withthe bum treatment personnel. Two sinksare located at the entrance to the center,and hand-washing is strongly encouraged

Bradley N. Doebbeling and Richard P. Wenzelare with the Department of Internal Medicine,University of Iowa College of Medicine, IowaCity, Iowa. Richard P. Wenzel is also with theProgram of Epidemiology, The University ofIowa Hospitals and Clinics. Ning Li is with theDepartment of Preventive Medicine and Envi-ronmental Health, Division of Biostatistics,The University of Iowa College of Medicine.

Requests for reprints should be sent toBradley N. Doebbeling, MD, MS, Departmentof Internal Medicine, C41L GH, The Univer-sity of Iowa College of Medicine, 200 HawkinsDr, Iowa City, IA 52242-1081.

This paper was accepted April 26, 1993.Editor's Note. See related editorial by

Mosley (p 1664) in this issue.

American Journal of Public Health 1679

Do_bing et aL

on entering and leaving the ward with a

display of prominent signs.

MdhodsEpidemic Investigation

Selected burn treatment personneland the burn patient with the secondarycase of hepatitis were interviewed to de-termine potential common exposures andrisk factors. The burn treatment center'scharting room, staff lounge, refrigerators,sinks, ice machine, nourishment room,

hydrotherapy room, rest rooms, watersupply, and all patient care areas were m-

spected by members of the hospital epi-demiology staff.

Screening. AU health care workersassigned to the center in the prior 6weeks(n = 154), allinpatients at the time ofiden-tification of the outbreak on February 9,1990 (n = 12), and all patients dischargedin the previous month (n = 45) were con-

tacted and offered the option of screeningwith liver function tests and a serologictest for hepatitis A virus imnmunoglobulinM antibody (IgM anti-HAV test) (enzymeimmunoassay, Abbott Laboratories, Chi-cago, Ill). Additionally, all food prepara-tionworkers in the hospital and the dietaryservices staff working on the burn center(n = 119) were contacted and similarlyscreened. AU exposed employees of the

burn treatment centerwhowere IgM anti-

HAV negative were tested for total anti-HAV antilody (anti-HAV, Abbott). A re-

peat IgM anti-HAV test was obtained forall consenting exposed health care work-ers regularly assigned to the bum treat-ment center 6 weeks later (n = 37) to de-termine whether seroconversion hadoccurred despite prophylaxis.

Case definition. A health careworkerwas defined as exposed if he or she haddirect patient contacton thebum center inthe 6 weeks (Decmber 28, 1989, to Feb-ruary 7, 1990) before the outbreak. A pre-

liminary case definition required elevationof serum levels of one of the liver tran-saminases to twice the upper limit of nor-mal or symptoms consistent with acuteHAV infection. Final inclusion as a case

subject required a positive IgM anti-HAVtiter, with orwithout clinical symptoms.Asusceptible control subject was an ex-

posed health care workerwho tested neg-ative for anti-HAV andwhowas in one ofthe same occupational categories as thehealth-care-worker case subjects (n = 60).Individuals with incomplete test results orprior immunity were excluded from fur-ther analysis.

Immunwprophykaxis. All asymptom-atic health care workers in the burn treat-ment center who had contact with the in-dex cases, all current patients, and all

patients discharged in the previous monthwere contacted and offered immunoglob-ulin immunoprophylaxis. Health careworkerswere informed that the efficacy ofimmunoglobulin immunoprophylaxis hasbeen demonstrated only when it is givenwithin 2 weeks of exposure. Additionally,regional exchange nurses temporarily as-signed to the burn treatment center tocover staffing shortages and all new pa-tients admitted for the subsequent 6weekswere offered inmunoglobulin. All closecontacts of those persons who were clin-ically ill or who had elevated liver tran-saminases were contacted to be screenedand offered immunogiobulin. Consentingexposed individuals were given a singledose of immunoglobulin 0.02 ml/kg (Gam-astan, Cutter Biologicals, West Haven,Conn) intramuscularly in the deltoid mus-cle. AU case subjects were contacted bytelephone 4 months later to determinewhether any symptomatic secondarycases had occurred among their familymembers or close contacts.

Four bum treatment center nurseshad been "pulled" to staff four other in-tensive-care units in the 48 hours beforethey became clinically ill. The patients forwhom they cared and health care workersin close contact with the infected nurseswere contacted and offered immunoglob-ulin prophylaxis.

Inde cases investaton The sourceof the initial index cases was investigatedby interviewing the adult indexcase patientand his familymembers regarding potentialrisk factors and any possible commonsources of exposure to hepatitis A beforeadmission to the hospital. The Iowa StateHealth Department inspected a single fam-ily day-care home that cared for the indexinfant and sixother children before the out-break. Serum was obtained for an anti-HAV test from five of the six children, thebabysitter, and family members of thebabysitter. However, a sample could notbe obtained from one of the children. Ad-ditionally, blood bank records were re-viewed to determine the individuals donat-ing blood to the two index case patients.None ofthe donors hadgivenblood tobothpatients; all were contacted and asked toreport any clinical illness in the 2 monthsbefore their donations.

Patient care study. A retrospectivecohort study of health care workers wasperformed to evaluate the risk of infectionassociatedwith caringforthe patientswiththe indexcases. Nursing care assignmentsfor the 6 weeks before the period of thepeak incidence were examined to deter-mine which nursing personnel had been

1680 American Journal of Public Health

Xj4fMoOv- r - a

LI.= HE

Z Food umPO-l Food Pmpwdm03 PduWtAmm

Mote. Staff cons edfood (hthed patten) in the saflounge, chaig room, andoccahlon ind alWOices (OF). Food and prpatoan (doed paen)forthe staffm erInduedt sWood ro (SFR), p --ntorsm ooml (PNR)Iandestlounge. Patetcare s(sded)lncudeshbin eech shn and doub p atroom (PR and DPR, respecleL. OMer abbrvios inldude dirty utiy roonm,Ien n room (MR), housekeqei (, plcih FT), spplrooms (SfB, andon-cd rooms (OCR).

FIGURE 1-Foor plan of the bum tament center.

December 1993, Vol. 83, No. 12

Hepatitis A Outeak

exposed to the index case patients. Thestudy population included nurses andnursing studentsworking in the burn treat-ment center during the 6-week period. Se-rosusceptible nurses caring for either ofthe index case patients were classified asexposed if assignment to one of the indexpatients occurred at least 2 weeks beforethe development of illness by the healthcare worker. Adjusted infection rates byweek of exposure were calculated andwere stratified by the number of shiftsworked with either or both of the patients.Cumulative incidence ratios (relative riskratios) were also calculated by week ofexposure to either index patient.

Risk factor study. A case-controlstudy was perfonned to evaluate behav-ioral risk factors and common exposuresamong health care workers for develop-ment of hepatitis A infection. A 50-itemself-administered questionnaire was devel-oped to assess potential risk factors forhepatitis A infection such as patient carepractices (e.g., hand-washing, glove use,providing hydrotherapy, specific types ofcare of infants and adults), personal habitswhile working (e.g., eating, drinking,smoking, biting fingemails), any group ex-posures (e.g., potlucks, shared meals, din-ing out), environmental exposures, andspecific food items eaten while at work.The questionnaire was pretested, revised,and then mailed to all 154 health carework-ers assigned to work in the bum treatmentcenter in the 6weeks before the outbreak'speak. All questionnaires were identifiedonlybya confidential studynumberknownto the investigators. Three sequential mail-ings of the questionnaire, which were fol-lowed by telephone contacts, were used toimprove the response rate.

Statistical MethodsThe statistical analyses included the

use of the Mantel-Haenszel chi-square sta-tistic, Fisher's two-tailed exact test, andodds ratios with 95% confidence intervalsfor all significance testing.16 For the tableswith zero entries, 0.5 was added to eachcell; the Mantel-Haenszel estimator of theodds ratio and the test-based confidenceintervalwere determined. Stepwise logisticregression was conducted by using SAS17to evaluate allvariables thatwere identifiedby the univariate approach (P < .10). Thesignificance level was set at .05. All Pval-ues reported are two-tailed.

Remz&

The IgM anti-HAV test was positivein the adult index case patient 1 month

| Patients - Staff

5-.

Ca

a

4

2-

I

K~Index Cases

__.. 1.. _

___IL__Iv l- - l-1 I I I

0 1. 21 17Desembwr

II

Administration ofImmune Globulin

I

I I1 I I I I I I I V. ..3 t I .F I -1 I I I

2 * 14 20 26 1 .i819 26 8 eJanuary Feruary Marh

Akte. AIlo0 the aed dae of onset in Xth in is both nde c are_"deas Noatlehedeofonsofclnkl ineessinthealtThel pinby thecdk-bar-inFbusges te only seonay p c. Not shown are thea omIncI motherw of the ift and an anormcho ln dofro fthednumeahoa Woc is unea. Notet Mncubal mn peod of ap-FiGURE -EpIdemSic cuvel%of'an otbarea ohlsA

FIGURE 2-Epkie curve of an ouexak of hepVts A on a bum bnearn cener.

after admission. His asymptomatic8-month-old son was also IgM anti-HAVpositive at the same time. Serologic testsfor hepatitisA (IgM-specific and total anti-HAV antibody) and liver function tests forboth patients had been negative on admis-sion. At the time of the investigation, noadditional blood samples were availablefor antibody testing from the first month ofthe patients' hospitalization. Neither pa-tient had vomiting or diarrhea. Inspectionof the burn treatment center revealedsinks and dispensers for hand-cleansingagents located in each room, and all werefound tobe ingood repair. Boxes ofglovesand gowns and masks were readily avail-able in each patient's room. All burn treat-ment personnelwere asked to refrain fromeating or drinking on the ward after iden-tification of the outbreak.

A total of 154 health care workerswere identified as having been exposed topatients on the burn treatment center inthe 6weeks before the health careworkersbecame ill. Serologic tests for anti-HAVantibody in all consenting exposed work-ers demonstrated that only8.7% ofthe 126workers screened had detectable immu-noglobulin G antibody to the hepatitis Avirus. Three health careworkerswith clin-ical hepatitis A infection and the second-ary case patient were initially seronega-tive for IgM anti-HAV antibody despitesymptoms and elevated hepatic transam-

inases, but subsequently became seropos-itive in 5 to 7 days.

Eventually a total of 11 health careworkers developed IgM anti-HAV anti-body, and all became clinically ill (Figure2). Two required hospitalization for dehy-dration. The infection rates among sero-logically defined susceptible individualsranged from 14% (7 of 50) among staffnurses and nursing assistants to 33% (2 of6) among ward clerks and 67% (2 of 3)among nursing students. All case subjectshad contact with the infant, whereas only8 of 11 case subjects (73%) also had con-tact with the father. The 32-year-oldmother of the index infant was also sero-positive for IgM anti-HAV antibody, butremained asymptomatic. No symptom-atic secondary cases occurred in closecontacts or family members of the in-fected case subjects. The mother of thechild was seropositive when checked inearly February, although she was asymp-tomatic. One asymptomatic seroconver-sion occurred after prophylaxis with im-munoglobuiin in the fiance of one of thenurse case subjects. None of 37 seroneg-ative health care workers whose serologywas rechecked at 4 to 6 weeks developedanti-HAV antibody.

Importantly, four cases of hepatitisAoccurred in the county of residence of theinfant and his father during November1989, followed by 8 and 17 cases, respec-

American Journal of Public Health 1681December 1993, Vol. 83, No. 12

as 41

Do_eing et aL

tively, during the next 2 months. No otherclustering of hepatitis A infections was

noted at restaurants or other facilities inthe county of origin during the period of

the outbreak. Anti-HAV serologywas ob-tained from the two donors reportingsymptomatic illness coincident with the

blood donations.

Charactef7stics ofthe StudySamples

Ninety-two percent (n = 132) of thequestionnaires from potentially eligiblecontrols (n = 154) were retumed; all casesubjects retumed the questionnaire. Con-trols were required to be both serosuscep-tible (negative fortotal anti-HAVantibody)and to have returned the questionnaire(n = 91, Figure 3). The control group wasfurther refined by limiting the group to thesame job classifications (i.e., those withmaxiwmal patient exposure) as those of thehealth-care-worker case subjects (e.g.,nurses, nursing assistants, nursing stu-dents, and ward clerks; n = 60). The re-ported rates of hand-washing after patientcare, hand-washing after diaper changing,and wearing gloves when contact was an-ticipated with blood and body fluids or se-cretionswere similar forboth groups. Afterpatient care, 82% of the case patients and75% of the controls reported that theywashed their hands routinely (relative risk[RR] = 1.09, 95% confidence interval[95%CI] = 0.77, 1.55).Afterdiaperchang-ing, 91% of the case patients and 100%o ofthe controls reported always washing theirhands (RR = 0.95, 95% Cl = 0.85, 1.06).Similarly, 82% of the case and 63% of thecontrol subjects reported consistent use ofgloves when contact with blood or bodyfluids was anticipated (RR = 1.29, 95%CI = 0.85, 1.98).

Nwrsing asyignet. The infectionrates for nurses assigned to either patientwere the highest during the exposureweek of January 2 through January 8,1990, just before the onset ofsymptoms inthe father. During that week, the infectionrates among nurses assigned to two ormore shifts with the infant were higherthan those for nurses assigned to a singleshift (4 of 5 [80%] vs 1 of 6 [17%], respec-tively). Similarly, infection rates werehigher for nurses assigned to two or moreshifts with the infant's father than fornurses assigned to a single shift (3 of 3[100%o] vs 1 of 13 [8%], respectively). Thecumulative incidence risk ratio was ele-vated for nurses caringfor either the infant(RR = 14.6, 95% CI = 3.4, 63.2) or hisfather (RR = 10.6, 95% CI = 2.2, 50.2)during that same week. The only twonurses assigned to care for both patientsduring the same week also became ill.

Riskfactorassessment The univariateanalysis (Table 1) identified the followingsi int sk factors for hepatitisA infec-tion: eating on the ward (odds ratio [OR]=14.2, 95% CI = 2.5, 80.5), bringing snackfoods to share with others (OR = 8.8, 95%

1682 American Journal of Public HealthD

Exposed HCWs 154

HCWs consenting to serologic screening 126

Seronegative HCWs (Anti-HAV Ab) 115

HCWs returning questionnaire 91

HCWs in same occupational class as cases 60Noe. Note the selecto of serosuscepbble control subjects whose occupatons were similar

tothose of the case subjects. The number of control subjects remaining after each selectionis shown. HCWs = healfth care workers.

FIGURE 3-Selcon of dudy cool

December 1993, Vol. 83, No. 12

Hep A Outbeak

Cl = 2.4, 31.3), and providing hydrother-apy treatments (OR = 5.9, 95% CI = 1.3,26.1). Of 12 different types of patient caregiven to adults or infants, none appeared topreipose workers to hepatitis A infec-tion. No other risk factors (e.g., smoking,chewing on pens or pencils, eating specificsnack foods at work, eating out with co-workers) were statistically sigficant riskfactors. After correction for confoundingvariables with a stepwise logistic regres-sion analysis, the single important risk fac-tor for the development of infection wasfrequent eating on the ward (OR = 14.2,P = .0029). No additional variables metthe criteria for retention in the model(P < .05) or improved the single variablemodel.

DiscussionHealth care workers caring for ther-

mally injured patients are potentially ex-posed to a variety of blood-borne patho-gens daily through the open burnwoundsthat must be debrided with hydrotherapyand frequent dressing changes. In ourburn treatment center, both case andcontrol subjects reported a high level ofprior compliance with routine hand-washing, glove use, and other barrierforms of universal precautions. Never-theless, an important outbreak of hepati-tis A occurred in a burn treatment centerinvolving 11 health care workers and 1patient with a secondary case. Once theepidemic was recognized, reemphasis ofgood hygienic and isolation practices andthe administration of immunoglobulinapparently prevented other secondarycases. Although the relative contribu-tions of each of these measures in termi-nating the outbreak cannot be assessed inthe current study, we would recommendthe use of a similar approach for controlof a nosocomial outbreak.

We were unable to determine thesource of hepatitis A infection in the twoindex patients. The usual incubation pe-riod of 10-50 days (mean = 1 month) isconsistent with either nosocomial or com-munity-acquired disease.?8 The elevatedcumulative incidence ratios among nursesexposed to either case in the week beforeonset of symptoms in the adult index pa-tient suggest that both index patients weremost infectious during that period, whichcorresponds to the time of greatest fecalshedding of hepatitis A virus. We alsohave epidemiologic evidence of an expo-sure dose-response relationship: the in-fection rates were higher for nurses as-

signed to more than one shift than for

nurses assigned to only one shift with ei-ther patient and highest among nurses as-signed to care forboth patients. However,the infant son may have been infected be-fore his aival at the hospital, explainingthe exposure to and timing of the diseasein his father. Infants usually remainasymptomatic, liver enzymes may not beelevated, and secretion ofvirus in the stoolmay continue for weeks.8

Similar exposure dose-response re-lationships have been reported in the lit-erature. In one multistate outbreak of 55cases, the 16% infection rate among sus-ceptible nurseswas four times higher thanthat among physicians with considerablyless exposure to case patients.4 Moreover,nurses working 12-hour shifts were signif-icantly more likely to become infectedthan those working 8-hour shifts. In an-other epidemic, nurses assigned to two ormore shifts with an infected infant with anileostomy had an infection rate 4.6 timeshigher than those assigned to a singleshift.5

Because the index patients did nothave direct contact with each other untilafter the onset of illness in the father, anosocomial source for their infections ap-pears unlikely. Nevertheless, a possiblenosocomial source would be a contam-inated blood product given to both pa-tients. Transfusion-associated hepatitis inneonates has been recognized as thesource of several nosocomial outbreaksinvolving health care workers.-6 How-ever, in our center, none of the donorsgave blood to both patients. Moreover, ofthe 33 donors contributing blood to eitherpatient, only 2 reported feeling ill near thetime of their donation. Hepatitis A sero-logic tests from both of the latter donorswere negative.

The outbreak occurred in the settingof a national increase in the incidence ofhepatitis A in the United States.19 At thetime of the outbreak, the incidence ofcommunity-acquired hepatitis A in thestate of Iowawas fivefold higher than thatin the previousyear, with one fourth ofthecases occurring in the same county of or-

igin as the index cases.Up to one third of community-ac-

quired cases of hepatitis A have beenlinked to day-care centers.20}-= However,in our study, none ofthe children tested orfamilymembers of the babysitter from thesmall day-care home that had previouslycared for the index infant were seroposi-tive. Nevertheless, one of the six childrencared for in the center could not be con-

tacted for testing. Similarly, no apparentcommon source of exposure could be

identified for the index family in the 2months before admission.

A case-control study using seroneg-ative, susceptible health care workers ascontrols convincingly demonstrated thatfrequent eating on the ward was the singlemost important risk factor for develop-mentofhepatitisA infection. Importantly,previous investigations ofnosocomial out-breaks of hepatitis A have typically failedto implicate a particular mode oftransmis-sion. The finding that hepatitisA infectionis clearly associated with eating on thehospitalward is not only biologically plau-sible, but also is entirely consistent withthe acknowledged fecal-oral route oftransmission of hepatitis A infections.

Our data indicating nonroutine hand-washing by approximately 25% of healthcare workers and nonroutine use ofglovesby 37% of health care workers are self-reported, retrospective, and conservativeestimates ofcompliancewith standard hy-gienic procedures. These results supportthe previous findings ofRosenblum et al.,8who found a significantly increased risk ofinfection during an outbreak in a neonatalintensive-care unit among health careworkers who did not wear gloves whentaping an intravenous line.

We speculate that hygiene was sub-optimal in the burn center before the out-break and that hand contamination prob-ably occurred. Hand contamination mayhave occurred after providing hydrother-apy (9 of 11 cases) or after other patientcare activities. Subsequently, noso-comial infection likely occurred eitherthrough eating snacks or sharing food onthe ward. The strategic placement ofsinks at the entrance to the center mayhave resulted in improved hand-washingon leaving the burn treatment center andexplain the lack of association of illnesswith eating in the hospital cafeteria. Afood-related outbreak was also unlikelybecause there was geographic clusteringof the cases on a single patient unit, se-rologic tests for hepatitis A among di-etary workers were negative, and therewas no supportive evidence in the case-control study. Importantly, none of thesnack foods available on the center or thespecific food items at the potluck dinnerwere associated with illness. However,we are unable to exclude absolutely thepossibility that food contamination ratherthan, or in addition to, contamination ofhealth care workers' hands was the ap-parent mode of transmission.

Previously reported outbreaks have

been linked to patients with vomiting, di-arrhea, or fecal incontinence.914 An out-

American Journal of Public Health 1683December 1993, Vol. 83, No. 12

Doeelng et aL

break of hepatitis A in a pediatric inten-sive-care unit in 1981 involving 10secondary cases suggested that food wasa potentially important risk factor.9 Theapparent index patient in that outbreakhad been incontinent of feces for severaldays before his death. His family had fre-quently brought in food that they sharedwith the boy and members of the staff.Sharing food with patients or their fami-lies, dking coffee at work, sharing cig-arettes, and eating in the nurses' office in-creased the likelihood of infection,although none of the variables reachedstatistical significance.9 Moreover, in an-other study, case patients were almostthree times more likely to smoke thanwere controls (36% vs 13%), suggestingthe importance of the fecal-oral route oftransmission.5 Similarly, Rosenblum andcolleagues8 reported that dfinking bever-ages in the intensive-care unit was a sig-nificant risk factor for infection.

Recent data demonstrated that hep-atitis A virus survives well on the fin-gerpads, despite drying, for 4 hours.23Additionally, the virus was transferred ef-fectively between fingerpads and from fin-gerpads to clean surfaces despite pro-longed drying. After suspensions of thehepatitis A virus were dried for 20 min-utes, an inoculum of the virus was trans-ferred from an inanimate surface to thefingerpads. Both pressure and friction sig-nificantly increased the transfer of hepa-titis A virus. The results of this experi-mental study support the importance ofhuman hands in the direct spread of hep-atitis A observed in the current outbreak.

The diagnosis of acute hepatitis A isusually confirmed easily by the presenceofa positive IgM anti-HAV serologic test,which is typically positive very early incases of acute infection.'8 24 Four of ourpatients developed nausea, vomiting, andelevated alanine aminotransferase levelsbut were initially seronegative. Each ofthese case patients, however, developedIgM anti-HAV antibody over the next 5 to7 days. Our findings confirm those of Za-choval and colleagues,25 who recently re-ported four patients with early acute hep-atitis A infection who were initiallyseronegative for IgM anti-HAV antibody.Therefore, patients suspected of havingearly hepatitisA infection but who have anegative IgM anti-HAV test should havethe test repeated several days later.25

Our results confirm the empiric ob-servations of Drusin and colleagues9 and

Reed et al.10 and are consistent with pre-vious guidelines26 suggesting that eatingon the hospital ward when hygiene is sub-optimal predisposes health care workersand patients to hepatitis A. Moreover, thecombination of a low level of populationimmunity and an increase in the incidenceof hepatitis A infection in the UnitedStates suggests that other hospitals maysoon experience similar clusters of infec-tion if health care workers continue to eaton the hospital ward. [

AcknowledgmentsThis paper was presented in part at the 30thInterscience Conference on AntimicrobialAgents and Chemotherapy (ICAAC), Atlanta,Ga, in October 1991 (abstract 1022).

We thank Jeanne Lyons and Jackie Oma-hen for their assistance and Drs Jack Stapleton,Douglas LaBrecque, and Loreen Herwaldt fortheir advice during the investigation of the out-break.

References1. Eisenstein AB, Aach RD, Jacobsohn W,

Goldman A. An epidemic of infectious hep-atitis in a general hospital: probable trans-mission by contaminated orange juice.JAMA. 1963;185:171-174.

2. Meyers JD, Romm FJ, Tihen WS, BryanJA. Food-borne hepatitisA in a general hos-pital: epidemiologic study ofan outbreak at-tributed to sandwiches. JAMA 1975;231:1049-1053.

3. Seeberg S, Brandberg A, Hermodsson S,Larsson P, Lundgren S. Hospital outbreakof hepatitis A secondary to blood exchangeinababy.Lancet. 1981;1:1155-1156. Letter.

4. Noble RC, Kane MA, Reeves SA, RoeckelI. Posttransfusion hepatitis A in a neonatalintensive care unit. JAMA. 1984;252:2711-2715.

5. Azimi PH, Roberto RR, Guralnik J, et al.Transfusion-acquired hepatitis A in a pre-mature infant with secondary nosocomialspread in an intensive care nursery. Am JDis Chik 1986;140:23-27.

6. Giacoia GP, Kasprisin DO. Transfusion-acquired hepatitis A. South MedJ. 1989;82:1357-1360.

7. Klein BS, Michaels JA, Rytel MW, BergKG, Davis JP. Nosocomial hepatitis A: amultinursery outbreak in Wisconsin.JAMA 1984;252:2716-2721.

8. Rosenblum LS, Villarino ME, Nainan OV,et al. Hepatitis A outbreak in a neonatal in-tensive care unit: risk factors for transmis-sion and evidence of prolonged viral excre-tion among preterm infants. J Infect Dis.1991;164:476-482.

9. Drusin LM, Sohmer M, Groshen SL, Spir-itos MD, Senterfit LB, Christenson WN.Nosocomial hepatitis A infection in a pae-diatric intensive care unit. Arch Dis ChikL1987;62:690-695.

10. Reed CM, Gustafson TL, Siegel J, Duer P.Nosocomial transmission of hepatitis Afrom a hospital-acquired case. Pediatr In-

fectDis. 1984;3:300-303.11. Krober MS, Bass JW, Brown JD, Lemon

SM, Rupert KJ. Hospital outbreak of hep-atitis A: risk factors for spread. Pediatr In-fect Dis. 1984;3:296-299.

12. OrensteinWA,WuE,WilkinsJ, etal. Hos-pital-acquired hepatitis A: report of an out-break. Pediatrics. 1981;67:494-497.

13. EdgarWM, CampbellAD. Nosocomial in-fection with hepatitis A. JInfect. 1985;10:43-47.

14. Goodman RA, Carder CC, Alien JR, Oren-stein WA, Finton RJ. Nosocomial hepatitisA transmission by an adult patient with di-arrhea.Am JMed. 1982;73:220-226.

15. American Public Health Association. Viralhepatitis A. In: Benenson AS, ed. Controlof Communicable Diseases in Man. 15thed. Washington, DC: American PublicHealth Association; 1990:197-200.

16. Schlesselman JJ. Case-Control Studies.New York: Oxford University Press; 1982:177-211.

17. SAS Institute Inc. SUGISupplmental Li-brary User's Guide, Version 5. Cary, NC:SAS Institute; 1986.

18. Hollinger FB, Glombicki AP. Hepatitis Avirus. In: Mandell GL, Douglas RG Jr,BennettJE, eds.Piinc@les andPracticeofInfectious Diseases. New York: ChurchillLivingstone; 1990:1383-1399.

19. Centers for Disease Control. FoodborneHepatitis A-Alaska, Florida, North Caro-lina, Washington. MMWRf 1990;39:228-232.

20. Benenson MW, Takafuji ET, BancroftWH, Lemon SM, Callahan MC, LeachDA. A military community outbreak ofhepatitis typeA related to transmission in achild care facility. Am J EpidemioL 1980;112:471-481.

21. Hadler SC, Webster HM, Erben JJ, Swan-son JE, Maynard JE. Hepatitis A in day-care centers. A community-wide assess-ment. NEngIJMed 1980;302:1222-1227.

22. Vernon AA, Schable C, Francis D. A largeoutbreak of hepatitis A in a day-care cen-ter: association with non-toilet-trained chil-dren and persistence of IgM antibody tohepatitis A virus. Am J EpidemioL 1982;115:325-331.

23. Mbithi JN, Springthorpe S, Boulet JR, Sat-tar SA. Survival of hepatitis Avirus on hu-man hands and its transfer on contact withanimate and inanimate surfaces.J Clin Mi-crobioL 1992;30:757-763.

24. FroesnerGG, Deinhardt F, RoggendorfM,Scheid R, Bayerl B, Zachoval R. Labora-tory diagnosis of hepatitis A infection. In:Bianchi L, Gerok W, Sickinger K, StalderGA, eds. Vru and the Liver. Lancaster,England: MTP Press; 1980:17-26.

25. Zachoval R, Kroener M, Brommer M,Deinhardt F. Serology and interferon pro-duction during the earlyphase ofacute hep-atitis A. J Infect Dis. 1990;161:353-354.Letter.

26. Doebbeling BN, Wenzel RP. NosocomialViral Hepatitis. In: Mandell GL, DouglasRG Jr, Bennett JE, eds. Princtples andPractice of Infectious Diseases. 3rd ed.New York: Churchill i vingstone; 1990:2215-2221.

1684 American Journal of Public Health December 1993, Vol. 83, No. 12