comprehensive systematic review for advancedlghttp.48653.nexcesscdn.net › 80223cf ›...

This is a sample from COMPREHENSIVE SYSTEMATIC REVIEW FOR ADVANCED PRACTICE NURSING, SECOND EDITION

Visit This Book’s Web Page / Buy Now / Request an Exam/Review Copy

© Springer Publishing Company

Comprehensive Systematic Review for Advanced Practice Nursing

http://www.springerpub.com/product/9780826131850




Cheryl Holly, EdD, RN, ANEF, FNAP, is professor, associate dean, and co-director of the Northeast Insti-tute of Evidence Synthesis and Translation at Rutgers School of Nursing, a Joanna Briggs Institute (JBI) Center of Excellence. She teaches DNP program courses in population health and implementation science, as well as works with DNP students in their capstone residency course and with BS honors students. Dr. Holly holds a BS in Nursing from Pace University, Leinhard School of Nursing, Pleasantville, New York; both an MEd in Adult Health and Physical Illness and an EdD in Research and Evaluation in Curriculum/Teaching from Columbia University. In addition, Dr. Holly has completed postgraduate work in Comprehensive Meta-Analysis and Advanced Meta-Analysis in the Statistics Institute. Among many other positions held, Dr. Holly previously was director of Nursing Research and Informatics at New York University Langone Medical Center in New York City, senior vice president of Quality Clinical Resource Management at the Westchester Medical Center, and associate dean, Columbia University School of Nursing. She is certified as a train-the-trainer in a comprehensive systematic review conducted by the JBI and has offered workshops on com-prehensive systematic reviews across the United States. Dr. Holly is the coordinator of the Eastern Nursing Research Society’s Research Interest Group on Comprehensive Systematic Review and Knowledge Trans-lation, and she is a member of the Committee of Directors of the JBI of Nursing and Midwifery, the Evidence Translation Group, the Cochrane Nursing Care Field, the Cochrane Injuries Group, and the JBI Scientific Methodology Group on Umbrella Reviews. She serves as manuscript reviewer for several journals, includ-ing Nursing Outlook, Nursing Education Perspectives, American Journal of Nursing, and The International Journal of Evidence-Based Healthcare. She is also a founding member of the Implementation Science and Synthesis Network of the United States. Dr. Holly has been principal or co-principal investigator or project director of 10 funded research projects. She has published extensively and presented both nationally and internationally in the areas of evidence-based practice (EBP), systematic review (SR), knowledge transla-tion, and critical care nursing. She is a fellow in the Academy of Nursing Education.

Susan Salmond, EdD, RN, FAAN, is executive vice dean, professor, a Distinguished Scholar and Fellow of the National Academies of Practice, and co-director of the Northeast Institute of Evidence Synthesis and Translation, a Joanna Briggs Institute (JBI) Center of Excellence, at Rutgers University School of Nursing, Newark, New Jersey. Dr. Salmond received a BSN from Villanova University, an MSN in Adult Health and Chronic Illness from Seton Hall University, and an EdD in Nursing Administration from Columbia Univer-sity. She is certified in nursing administration, nursing education, and transcultural nursing. In addition, Dr. Salmond has completed postgraduate work in Comprehensive Meta-Analysis and Advanced Meta-Analysis in the Statistics Institute. She is certified as a train-the-trainer in a comprehensive systematic review conducted by the JBI and has offered workshops on comprehensive systematic review across the United States. She serves on the JBI scientific methodology committees for mixed methods and qualitative synthesis, and is a member of the Committee of Directors of the JBI for Nursing and Midwifery, the editorial board for the JBI Database of Systematic Reviews and Implementation Reports, and the Cochrane Nursing Care Field. She is the research editor of Orthopedic Nursing and a member of the editorial board for the American Journal of Nursing. She has served as a research, educational, and clinical management consul-tant. Dr. Salmond has coauthored three editions of Orthopedic Nursing; authored two other books, numer-ous book chapters, and peer-reviewed journal articles. She is a highly sought-after conference presenter and workshop leader, both nationally and internationally, in the areas of cultural competence, qualitative research, evidence-based practice (EBP), and systematic review (SR). She is a founding member of the Implementation Science and Synthesis Network of the United States. She is a fellow of the American Acad-emy of Nursing and the Academy of Nursing Education.

Maria Saimbert, PharmD, MSN, MLIS, RN, is a nursing (pediatrics) and pharmacy (acute care/community) professional, with research interests in medication management during transfer of patient care to health professionals/facilities, integration, and use of electronic technologies to support evidence-based practice (EBP) and lifelong learning. She holds a PharmD in Pharmacy (2015) from the University of Florida, both an MSN in Nursing Informatics (2008) and an MS in Library and Information Science (2003) from Rutgers University, and a BS in both Nursing (1994) and English Literature (1994) from Bloomfield College, Bloomfield, New Jersey. Dr. Saimbert has spent more than 8 years working with health science students and clinicians, assisting in design of search strategies, navigation of online search resources, and manage-ment of literature for systematic reviews (SRs). She supports research librarians and information specialists in context as visible, co-reviewers on SR research, with accountability for influencing patient care through work on review search methods. She has published several refereed articles. Dr. Saimbert has spent more than 5 years facilitating SRs and searching for skills alongside health science students and health professionals.





Comprehensive Systematic Review for Advanced Practice NursingSecond Edition

Cheryl Holly, EdD, RN, ANEF, FNAPSusan Salmond, EdD, RN, FAANMaria Saimbert, PharmD, MSN, MLIS, RNEditors





Copyright © 2017 Springer Publishing Company, LLC

All rights reserved.

No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, elec-tronic, mechanical, photocopying, recording, or otherwise, without the prior permission of Springer Publishing Company, LLC, or authorization through payment of the appropriate fees to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400, fax 978-646-8600, [email protected] or on the Web at www.copyright.com.

Springer Publishing Company, LLC11 West 42nd StreetNew York, NY 10036www.springerpub.com

Acquisitions Editor: Margaret ZuccariniCompositor: Westchester Publishing Company

ISBN: 978-0-8261-3185-0e-book ISBN: 978-0-8261-3186-7Instructor’s PowerPoints: 978-0-8261-3212-3

Instructor’s Materials: Qualified instructors may request supplements by emailing textbook@springer pub.com

16 17 18 19 20 / 5 4 3 2 1

The author and the publisher of this Work have made every effort to use sources believed to be reliable to provide information that is accurate and compatible with the standards generally accepted at the time of publication. Because medical science is continually advancing, our knowledge base continues to expand. Therefore, as new information becomes available, changes in procedures become necessary. We recommend that the reader always consult current research and specific institutional policies before performing any clinical procedure. The author and publisher shall not be liable for any special, consequential, or exemplary damages resulting, in whole or in part, from the readers’ use of, or reliance on, the information contained in this book. The pub-lisher has no responsibility for the persistence or accuracy of URLs for external or third-party Internet websites referred to in this publication and does not guarantee that any content on such websites is, or will remain, accurate or appropriate.

Library of Congress Cataloging-in-Publication Data

Names: Holly, Cheryl, author. | Salmond, Susan Warner, author. | Saimbert, Maria, 1971– , author.Title: Comprehensive systematic review for advanced practice nursing / Cheryl Holly, Susan Salmond, Maria Saimbert.Other titles: Comprehensive systematic review for advanced nursing practice.Description: Second edition. | New York, NY : Springer Publishing Company, LLC, [2016] | Preceded by Comprehensive

systematic review for advanced nursing practice / Cheryl Holly, Susan W. Salmond, Marie K. Saimbert. c2012. | Includes bibliographical references and index.

Identifiers: LCCN 2016005672 | ISBN 9780826131850 (hard copy : alk. paper) | ISBN 9780826131867 (ebook)Subjects: | MESH: Advanced Practice Nursing | Review Literature as Topic | Meta-Analysis as Topic | Evidence-Based

Nursing—methodsClassification: LCC RT82.8 | NLM WY 128 | DDC 610.73—dc23 LC record available at http://lccn.loc.gov/2016005672

Special discounts on bulk quantities of our books are available to corporations, professional associations, pharmaceutical companies, health care organizations, and other qualifying groups. If you are interested in a custom book, including chapters from more than one of our titles, we can provide that service as well.

For details, please contact:Special Sales Department, Springer Publishing Company, LLC11 West 42nd Street, 15th Floor, New York, NY 10036-8002Phone: 877-687-7476 or 212-431-4370; Fax: 212-941-7842E-mail: [email protected]

Printed in the United States of America by McNaughton & Gunn.

mailto:[email protected]

http://www.copyright.com

http://www.springerpub.com

http://lccn.loc.gov/2016005672

mailto:[email protected]

http://[email protected]





For Grace, who peeked over my shoulder at the computer to make sure I was working on this book.

—Cheryl Holly

For family, friends, and students, who ground me and motivate me to do more.—Susan Salmond

For those who reach for and help sustain the good in every interaction between others and themselves, through the tunnel, toward the light.

—Maria Saimbert





vii

Contents

CONTRIBUTORS ix

FOREWORD Zoe Jordan, PhD xi

PREFACE xiii

ACKNOWLEDGMENTS xv

PART I: INTRODUCTION

CHAPTER 1 SYSTEMATIC REVIEW AS A BASIS FOR EVIDENCE-BASED PRACTICE 3Susan Salmond and Cheryl Holly

CHAPTER 2 STEPS IN THE SYSTEMATIC REVIEW PROCESS 17Susan Salmond and Adam C. Cooper

PART II: A FRAMEWORK FOR CONDUCTING SYSTEMATIC REVIEWS

CHAPTER 3 ORGANIZING AND PLANNING A SYSTEMATIC REVIEW 41Cheryl Holly

CHAPTER 4 DEVELOPING CLINICAL QUESTIONS FOR SYSTEMATIC REVIEW 79Maria Saimbert, Jenny Pierce, and Pam Hargwood

PART III: SEARCHING AND APPRAISING THE LITERATURE

CHAPTER 5 KEY PRINCIPLES FOR SEARCHING THE LITERATURE 105Maria Saimbert

CHAPTER 6 SEARCH RESOURCES AND TECHNIQUES TO MAXIMIZE SEARCH EFFORTS 139Maria Saimbert, Susan A. Fowler, Jenny Pierce, and Pam Hargwood

CHAPTER 7 CRITICAL APPRAISAL 173Susan Salmond and Sallie Porter

PART IV: METHODS FOR SYSTEMATIC REVIEWS

CHAPTER 8 SYSTEMATIC REVIEW OF EXPERIMENTAL EVIDENCE: META-ANALYSIS OF INTERVENTIONS 193Cheryl Holly

CHAPTER 9 SYSTEMATIC REVIEW OF OBSERVATIONAL AND DESCRIPTIVE EVIDENCE 225Cheryl Holly





viii Contents

CHAPTER 10 SYSTEMATIC REVIEW OF QUALITATIVE EVIDENCE 249Susan Salmond and Daphne Stannard

CHAPTER 11 SYSTEMATIC REVIEW OF ECONOMIC EVIDENCE 279Cheryl Holly and Yuri T. Jadotte

PART V: NEW METHODS OF REVIEW

CHAPTER 12 UMBRELLA REVIEWS 301Cheryl Holly

CHAPTER 13 MIXED-METHOD REVIEWS 309Susan Salmond

CHAPTER 14 OTHER TYPES OF REVIEWS: RAPID, SCOPING, INTEGRATED, AND REVIEWS OF TEXT AND OPINION 321Cheryl Holly

PART VI: USING SYSTEMATIC REVIEWS IN PRACTICE

CHAPTER 15 CLINICAL GUIDELINES: USING SYSTEMATIC REVIEWS AT THE POINT OF CARE 339Ronell Kirkley and Aleksandr Nevelev

CHAPTER 16 FORMULATING EVIDENCE-BASED POLICY 361David Anthony (Tony) Forrester, Cheryl Holly, Rita M. Musanti, and Patricia Polansky

CHAPTER 17 FUTURE DEVELOPMENT OF SYSTEMATIC REVIEWS AND EVIDENCE-BASED MEDICINE 379Jos Kleijnen

CHAPTER 18 RESOURCES FOR SYSTEMATIC REVIEW 387Cheryl Holly and Maria Saimbert

PART VII: EXAMPLES OF SYSTEMATIC REVIEWS

CHAPTER 19 THE EFFECTIVENESS OF NON-PHARMACOLOGICAL MULTICOMPONENT INTERVENTIONS FOR THE PREVENTION OF DELIRIUM IN NON–INTENSIVE CARE UNIT, OLDER ADULT, HOSPITALIZED PATIENTS: A SYSTEMATIC REVIEW 401Elizabeth Thomas, Jane E. Smith, Yuri T. Jadotte, and Cheryl Holly

CHAPTER 20 INTERPROFESSIONAL COLLABORATION AND HEALTH OUTCOMES: A SYSTEMATIC REVIEW AND META-SYNTHESIS 425Yuri T. Jadotte, Cheryl Holly, Sabrina M. Chase, Arthur Powell, and Marian Passannante

APPENDIX A PLANNING FOR A SYSTEMATIC REVIEW 449

APPENDIX B GUIDELINES SOURCES 453

APPENDIX C SOCIAL SCIENCE AND BIOMEDICAL SCIENCE GREY LITERATURE 455

APPENDIX D DISSERTATIONS AND THESES DATABASES 457

APPENDIX E ANSWER KEYS TO CHAPTER EXERCISES 459

INDEX 467


ix

Contributors

Sabrina M. Chase, PhD, Assistant Professor and Director, PhD Program in Urban Systems/Health, Rutgers University, Newark, New Jersey

Adam C. Cooper, MSN, RN-BC, Manager, Institute for Nursing Excellence, Deputy Director, University of California–San Francisco (UCSF), UCSF Joanna Briggs Institute Center for Synthesis and Implementation, UCSF Medical Center and Benioff Children’s Hospital, San Francisco, California

David Anthony (Tony) Forrester, PhD, RN, ANEF, FAAN, Professor, Rutgers University School of Nursing, Newark, New Jersey

Susan A. Fowler, MLS, Information Librarian, Rutgers University, Newark, New Jersey

Pam Hargwood, MLIS, Information Librarian, Rutgers University, Newark, New Jersey

Cheryl Holly, EdD, RN, ANEF, FNAP, Professor, Associate Dean, and Co-director, Northeast Institute of Evidence Synthesis and Translation, Rutgers University School of Nursing, Newark, New Jersey

Yuri T. Jadotte, MD, PhD, Assistant Professor and Assistant Director, North-east Institute for Evidence Synthesis and Translation, Rutgers University School of Nursing, Newark, New Jersey

Ronell Kirkley, DNP, CRNA, APN-Acute Care, Chief Nurse Anesthetist, Department of Anesthesiology, New York Methodist Hospital, Brooklyn, New York

Jos Kleijnen, MD, PhD, Director, Kleijnen Systematic Reviews Ltd., London, England

Rita M. Musanti, PhD, APN-C, AOCNP, Assistant Professor, Rutgers University School of Nursing, Newark, New Jersey

Aleksandr Nevelev, BSN, BA, RN, CCRN, Staff Nurse, Intensive Care Unit, New York Methodist Hospital, Brooklyn, New York





x Contr ibutors

Lisa M. Paplanus, DNP, RN-C, CCRN, ACNP-BC, ANP-BC, Vascular Nurse Practitioner, Langone Medical Center, New York University Medical Center, New York, New York

Marian Passannante, PhD, Professor and Associate Dean for Education, School of Public Health, Rutgers University, Newark, New Jersey

Jenny Pierce, MS, Research Librarian, Rutgers University School of Nursing, Newark, New Jersey

Patricia Polansky, MS, RN, Co-director, Center to Champion Nursing in America, AARP, Washington, DC

Sallie Porter, RN, DNP, PhD, Assistant Professor, Rutgers University School of Nursing, Newark, New Jersey

Arthur Powell, PhD, Professor, Department of Urban Education, New Jersey Institute of Technology, Rutgers University, Newark, New Jersey

Maria Saimbert, PharmD, MSN, MLIS, RN, Nursing (Pediatrics) and Phar-macy (Acute Care and Community) Professional, Plainsboro, New Jersey

Susan Salmond, EdD, RN, FAAN, Executive Vice Dean, Professor, and Co-director, Northeast Institute of Evidence Synthesis and Translation, Rutgers University School of Nursing, Newark, New Jersey

Jane E. Smith, DNP, APN-BC, ONC, Clinical Nurse Specialist, Atlantic Health System, Morristown, Orthopedics, Morristown, New Jersey

Daphne Stannard, PhD, RN-BC, CNS, Director and Chief Nurse Researcher, University of California–San Francisco (UCSF) Institute for Nursing Excellence, Director, UCSF Joanna Briggs Institute Center for Synthesis and Implemen-tation, San Francisco, California

Elizabeth Thomas, DNP, MSN, Nurse Practitioner, Atlantic Health System, Arthritis Center, Morristown, New Jersey








xi

Foreword

A wise man proportions his beliefs to the evidence—David Hume

It is a great honor and privilege to have been asked to introduce this valuable book. The connection between research and practice is as important today as it has ever been and yet, at times, it seems just as elusive as it ever was. Of course, the best available research evidence forms only one component of the clinical decision-making process, and there are indeed other important elements that must be taken into account: “the context in which care is delivered; client prefer-ence; and the professional judgement of the health professional” (Pearson, Wie-chula, Court, & Lockwood, 2005, p. 209). However, as Hume suggested, evidence forms a critical component that should not be left out of the equation.

Goldacre (2013, p. 16) has noted that the creation of an “information archi-tecture” is no small feat and can indeed take decades to achieve. It is, of course, not as simple as undertaking good quality research and then using it in prac-tice. There are a multitude of steps that need to occur before, during, and after the research itself is conducted in order to ensure that successful “knowledge translation” occurs.

Unmet knowledge needs are prolific across health care. There are still many questions related to various aspects of care provision that remain unanswered. These questions will remain unanswered as long as researchers fail to con-sult with key stakeholders about their knowledge requirements. These key stakeholders may include patients, the general community, clinicians (health professionals working at the point of care), governments, and other policy orga-nizations. The relationship between these groups and the scientists and research-ers during the “discovery process” has never been more important than it is today, and it is a vital component addressed by very few groups internationally (Pearson, Weeks, & Stern, 2011, p. 14).

Understanding that the questions being asked by health professionals do not only relate to effectiveness is another important factor in the transla-tional cycle. The nature of evidence is broad and can come from a range of sources, both research and nonresearch. Feasibility, appropriateness, mean-ingfulness, and effectiveness are all important elements of the clinical decision-making process and it is important to acknowledge that evidence does exist to answer important clinical questions (Pearson, 2004, p. 47). Therefore, the systematic reviews of evidence should encompass the best available evidence in relation to each of these.





xii Foreword

As previously suggested, although systematic reviews are important docu-ments, they do not and cannot stand alone. They are but one piece of informa-tion within the “information architecture” that Goldacre referred to. Clinical decisions are made within complex systems that are influenced by many things. It is, thus, just as important to ensure that the evidence is supported by orga-nizational structures and resources for implementation at the point of care.

This book represents an important resource for the next generation of health professionals and potential leaders in evidence-based health care. Arming yourselves, as readers of this book, with skill sets to identify, retrieve, appraise, and synthesize evidence for use at the point of care will be empowering and will change the way in which you approach decision making. It will ultimately improve your interactions with colleagues and patients alike and result in better health outcomes.

Zoe Jordan, PhDExecutive Director

The Joanna Briggs InstituteAssociate Professor, Faculty of Health Sciences

The University of AdelaideAdelaide, South Australia

■■ References

Goldacre, B. (2013). Building evidence into education. Retrieved from https://educationen dowmentfoundation.org.uk/news/eef-welcomes-ben-goldacre-report

Pearson, A. (2004). Balancing the evidence: Incorporating the synthesis of qualitative data into systematic reviews. JBI Reports, 2(2), 45–64.

Pearson, A., Weeks, S., & Stern, C. (2011). Translation science and the JBI model of evidence-based healthcare. Philadelphia, PA: Lippincott, Williams, and Wilkins.

Pearson, A., Wiechula, R., Court, A., & Lockwood, C. (2005). The JBI model of evidence-based healthcare. International Journal of Evidence-Based Healthcare, 3(8), 207–215.





xiii

Preface

Evidence-based health care is grounded in the premise that although practi-tioners may be expert in the art of their discipline, the explosion of scien-tific information, the demands of a more educated health care consumer, and spiraling health care costs have created a situation where the practitioner’s knowledge of the science of the discipline is often either not current or not enough. This book presents a research method to address this issue. It is based on our belief that published articles may reflect the predispositions of their authors, journals tend to accept articles with only positive findings (known as publication bias), and books date rapidly; hence, there is the need for a system-atic review (SR) that looks at all findings on a particular topic and incorporates them to the best extent possible into research and experiential and contextual evidence. This book has evolved from our work in SR and knowledge translation over the past 10 years, and it is based on both craft and personal expertise over those years. Our motivation to write a book about SR, and then this expanded second edition, is based on our belief that establishing a culture where evidence is routinely used in daily practice will improve outcomes and increase profes-sional independence.

This perspective does not challenge individual judgment or the need to incorporate patient values and preferences; rather, it puts judgment in the fore-ground, hugely improved. There are challenges, though, in reviewing, synthe-sizing, and transferring knowledge as reflected in the works of the 18th-century Scottish naval surgeon James Lind, who wrote in his work on scurvy:

As it is no easy matter to root out prejudices . . . it became requisite to exhibit a full and impartial view of what had hitherto been published on the scurvy, and that in a chronological order, by which the sources of these mistakes may be detected. Indeed, before the subject could be set in a clear and proper light, it was necessary to remove a great deal of rubbish. (Chalmers, Hedges, & Cooper, 2002, p. 14)

In other words, we need to learn more from what we already know. SRs can inform us about what is known, what is not known, and what needs to be known. In that way, SRs can inform discovery. An SR is based in population health, where every person is different and unique, but similar enough so that we can determine what works best, overall and under what circumstances, to achieve the best outcomes.





xiv Preface

Historically, an SR that is aimed at providing a comprehensive and unbiased summary synthesis of evidence on a single topic by pooling multiple studies in a single report grounded in the research process. Emerging methods of SR, however, have extended that reach given the recent finding that there are more than 11 SRs published each day. This new explosion of information has given rise to new methods to create synthesis, and we have added chapters to address these.

The book is divided into seven sections. Part I is an introduction to an SR and deals with how to plan and organize a review. Part II provides a framework for the SR process and the development of a focused question by using the PICO (population, intervention, comparison, outcome) formula. Part III pro-vides information on searching and appraising the literature. Part IV is about the primary methods of an SR: qualitative, interventional, observational, and economic. Part V presents the emerging methods of an SR: umbrella reviews, mixed-method reviews, and other reviews (scoping reviews, integrated reviews, rapid reviews, and the review of text and opinion). Part VI is based on using the findings of SRs in practice and policy situations, and Part VII presents examples of completed reviews. We have updated and expanded the chapter on resources for an SR to serve as a toolkit for this work. Qualified instructors can obtain a PowerPoint slide presentation of the material found in the chapters by emailing Springer Publishing Company at textbook@springer pub.com.

As the field of the SR is rapidly emerging, we need the help of friends and colleagues who are statisticians, physicians, information specialists such as research librarians, and practice experts and who use evidence at the point of care and in policy decisions to address the growing complexity of this method of research to assist us in achieving our aim of making this second edition useful to practitioners. We hope we have succeeded.

Cheryl HollySusan SalmondMaria Saimbert

■■ Reference

Chalmers, I., Hedges, L. V., & Cooper, H. (2002). A brief history of research synthesis. Evaluation and the Health Professions, 25(1), 12–37.







xv

Acknowledgments

A very special thank you to all of our contributors who made this second edi-tion a reality. Your hard work and dedication to the advancement of systematic review methodology is praiseworthy. —Cheryl Holly

A heartfelt thank you to all of those who see the value of systematic review in advancing nursing practice. —Susan Salmond

Thank you to my mom and dad, Anne and Joseph Saimbert, and my brother, John Saimbert, for believing in me and for the prayers of so many, which sup-port me always. —Maria Saimbert





6 Search Resources and Techniques to Maximize Search Efforts

Maria Saimbert, Susan A. Fowler, Jenny Pierce, and Pam Hargwood

OBJECTIVESAt the end of this chapter, the reader will be able to:

O Use health and social sciences online search resources for retrieving cita-tions to answer systematic review (SR) questions

O Define point-of-care resources and identify their role in researching practice-based questions

O Define grey literature and explore its role in SR research O Use various search techniques such as citation chasing and hand searching

CHAPTER HIGHLIGHTS O A librarian should be a part of the SR team’s plans; reviewers should consult

and/or collaborate (e.g., librarians as co-reviewers of an SR) with research librarians or information science professionals who are trained in search work for SR research.

O Searches to support SR construction are iterative, often performed more than once in several search resources to gain as much relevant literature as possible to answer foreground questions. Searches supporting foreground question research are systematic, as one or two best search strategies, master searches, are repeated throughout all relevant databases, and search resources for the review topic. One should keep in mind that depending on the features of a search resource and/or search retrieval system, slight variations in the master search strategies may be needed.

O Searches to support SR construction are comprehensive, making use of natural language terms (e.g., text words/phrases) and a database’s controlled vocabulary terminology (e.g., subject headings), allowing researchers a chance for retrieving more citations for appraisal, and forming conclusions related to foreground questions.

O Preliminary searches in filtered search resources can assist reviewers in revising, expanding, and tailoring foreground questions.

O Pearl-growing and preliminary searches allow reviewers a chance to figure out which text words from a diagrammed research question and which





140 I I I . Searching and Appra is ing the L i terature

database-controlled vocabulary terms (e.g., subject headings) best represent and allow for best retrieval of citations on a topic.

O Although it is advisable to use personal accounts provided by each data-base’s search retrieval system (e.g., EBSCOhost personal account, Ovid personal account) to save search strategies, all literature searches, natural language terms, and vocabulary or database index terms used in searches should be documented on paper or in a computer document for use at a later time by the SR team or by other researchers who wish to duplicate, revise, or add to the body of research on a review topic.

O Setting up search alerts in searched databases, as available, can facilitate retrieval and appraisal of the latest research that is relevant to an ongoing SR research. Use of search alerts contributes to keeping an SR as up to date as possible before publication.

O All significant results from literature searches should be collected and organized by using a bibliographic citation management tool; examples include EndNote (Thomson Reuters), Mendeley (Elsevier B.V.), or RefWorks (ProQuest).

O Book appendices related to this chapter are Appendix A (Planning for a Systematic Review), Appendix B (Guidelines Sources), Appendix C (Social Science and Biomedical Science Grey Literature), and Appendix D (Disser-tations and Theses Databases).

When conducting an SR, the search for relevant articles and other pertinent sources should be iterative, comprehensive, and exhaustive. Although this is daunting, it is a main part of SR research toward answering foreground ques-tions, which eventually will touch patients/families. As with other research, planning for a review is important and this should include plans for the litera-ture search process. The best practice is to work with an information science professional (e.g., research librarian) who is trained and actively involved in SR literature searching. The use of a librarian or an information specialist in plan-ning search strategies for comprehensive SR searches is a part of the standards for SRs (Institute of Medicine [IOM], 2011). In line with working with an infor-mation professional to plan and conduct literature searches, this chapter is a complementary offering of details on select search resources and search tech-niques that are useful to SR research teams in the health sciences.

Literature resources that are used for gathering the evidence for SRs can be divided into two broad categories: filtered and unfiltered. Filtered resources are also known as synthesized resources, and unfiltered resources may be referred to as raw resources.

Filtered or synthesized resources typically offer some interpretation and appraisal of included literature. Each piece of filtered literature covers many individual studies or publications. Examples include SRs, meta-analyses,





6. Search Resources and Techniques to Maximize Search Ef for ts 141

meta-syntheses, clinical/evidence summaries, and guidelines. In contrast, unfil-tered literature sources provide access to individual studies or publications. The term raw is fitting for this kind of research. Unfiltered resources typically con-tain large quantities of individual academic articles and many different types of materials, including conference proceedings, dissertations, editorials and commentaries, randomized controlled trials (RCTs), epidemiologic studies, qualitative studies, traditional literature review articles, and some highly struc-tured, filtered literature such as SRs. Electronic biomedical databases, such as Medical Literature Analysis and Retrieval System Online (MEDLINE) from the United States National Library of Medicine (NLM) and Cumulative Index to Nursing and Allied Health Literature (CINAHL) from EBSCO Publishing, are examples of where to find raw literature.

Raw or unfiltered resources form the body of evidence from which filtered resources draw and synthesize research findings. This means that many items found using filtered resources could ultimately be found by directly searching unfiltered resources. Using filtered resources in concert with unfiltered resources carries important benefits. One is that these filtered resources are tremendously valuable discovery tools that can help reviewers quickly find the most impor-tant or influential research on a given topic. Finding one very relevant article via a filtered resource can open the gateway to literally hundreds of other related citations. See Figure 6.1 for examples of resources and search techniques for locating filtered and unfiltered literature.

Before proceeding to review search resources in this chapter, here is some background information on online search resources such as databases. For each database, information sheets can be located by searching an Internet search engine (e.g., Google.com) or looking at the Help section within a database’s search retrieval system (e.g., EBSCOhost, ProQuest). Information included in the database information sheets includes history and description of the data-base, dates of literature coverage, and types of literature included in the database. Databases are dynamic, with database producers adding older and newer content according to their working standards/policies. There are also details on how to search the database on specific search retrieval systems. Specific search features from a particular search retrieval system that can be used to search a specific database effectively are reviewed, often with examples for how to use those features in search strategies.

Some information about search retrieval systems includes the following. Similar to databases, search retrieval systems are dynamic platforms where the look and search features available may change as time goes on. Therefore, it is best to consult with librarians who are familiar with electronic search resources and their corresponding search retrieval systems. In addition, one should review, as needed, the Help section of search retrieval systems for important updates related to searching a database on that particular system. Search engines such as Google.com also include a Help section to assist researchers in using their search features.






Ideal search start:

• Formulate clinical research questions(background and foreground).

• Generate and refine search strategies.

Run searches in filtered, summary resources:

• Campbell Library of Systematic Reviews• Cochrane Database of Systematic Reviews• Database of Abstracts of Reviews of Effects • Dynamed• Essential Evidence Plus • Joanna Briggs Institute Evidence-Based Summaries• Joanna Briggs Institute Systematic Reviews• Nursing Reference Center Evidence Summaries

• PROSPERO• PubMed Special Queries• Guidelines of evidence• Textbooks of evidence (e.g., British Medical Journal,

section Clinical Evidence)

Run searches or employ search techniques in unfiltered, individual-study resources:

• Author searching• Citation chasing• Data sets• Individual trials• Individual citations from bibliographic databases:

(Cumulative Index to Nursing and Allied Health Literature [CINAHL]; Cochrane Central Register of Controlled Trials; Embase; Health Services, Technology, Administrative and, Research; ProQuest Dissertations and Theses Database; ProQuest Nursing and Allied Health Source; PsycINFO; PubMed; Scopus; Web of Science) Gray literatureHand searching

••• Internet searches (web searches, journal

portals, blogs, listservs)

Items identified in systematic review

process

Relevant itemsselected and

appraised for the systematic review

(Retain included and excluded citations).

FIGURE 6.1Searching filtered and unfiltered sources.Source: John T. Oliver and Maria Saimbert.






■■ Filtered Evidence Resources

One search strategy when proposing and conducting an SR should include a look for existing reviews of filtered evidence on a clinical question. Some of the resource portals include the Campbell Collaboration Library of Systematic Reviews from the Campbell Collaboration, the Cochrane Database of System-atic Reviews (CDSR) from Cochrane, the Database of Abstracts of Reviews of Effects (DARE) available through Cochrane or the Center for Reviews and Dis-semination (CRD), and the Joanna Briggs Library of Systematic Reviews from the Joanna Briggs Institute (JBI).

CAMPBELL COLLABORATION LIBRARY OF SYSTEMATIC REVIEWS

The Campbell Collaboration (C2) is an international research network that pro-duces SRs on the effects of social interventions on policies and services. Started in 2000, C2 covers the subject areas of education, crime and justice, social wel-fare, and international development. Records are placed in the Campbell Library by using vocabulary or thesaurus terms and subject areas, as deemed by Camp-bell producers. All reviews are available free on the Internet at www.campbell-collaboration.org/library.php, and many are co-registered and searchable through the Cochrane Library. For example, a C2 review on children and aggres-sion may also be indexed or found as a part of reviews from the Developmen-tal, Psychosocial, and Learning Problems Group within the Cochrane Library database of SRs.

COCHRANE LIBRARY

The Cochrane Library is named after Archie Cochrane, who is often consid-ered the father of evidence-based medicine, and it includes databases produced by the Cochrane Collaboration and CRD. Started in 1993, Cochrane is a non-profit organization based in England with centers all over the world.

The Cochrane Library consists of six evidence-based medicine databases and a database with information about review groups. Information in the Cochrane Library will help researchers determine the quality of studies or the state of research regarding a particular topic. Access to the Cochrane Library depends on where you live. In the United States, with the exception of Wyo-ming, you will need paid subscription access through John Wiley and Sons, Ltd. to the Wiley Online Library. Check with your institution or a local academic medical library. A search in Cochrane Library retrieves records from all six data-bases and Cochrane review groups. You can search the Cochrane Library by using a combination of Medical Subject Headings (MeSH) terms and plain lan-guage text words by using Boolean and proximity operators just as you would in MEDLINE, though the formatting/syntax is different. See the following example of using the operator NEAR.

http://www.campbellcollaboration.org/library.php

http://www.campbellcollaboration.org/library.php






Proximity Operators in CochraneNEAR/# = # represents the number of words within which Cochrane will locate each term within each other. For example, nursing NEAR/4 students locates the term nursing within four words of students.NEXT = will find each term in the order in which you entered it. This is a useful operator for searching for phrases with truncation. For example, nursing NEXT student* locates the term nursing when it appears right before student or students.

For more guidance on searching Cochrane, go to onlinelibrary.wiley .com/cochranelibrary/search/advanced# and click Search Help.

Cochrane Database of Systematic ReviewsCochrane reviews are considered the “gold standard” of SRs. Each review is prepared according to the methodology set out in the Cochrane Handbook for Systematic Reviews of Interventions or the Cochrane Handbook for Diagnos-tic Test Accuracy Reviews. As such, their rigor makes them good models for examples of well-constructed SRs. CDSR contains current protocols and SRs, which are categorized under a Cochrane review group area (e.g., Anesthesia, Critical, and Emergency Care; Dementia and Cognitive Improvement; Oral Health). An SR protocol is a proposal for an upcoming SR research, and it includes information on why the review is needed, objectives, and plans for review methods.

A look at SRs within a Cochrane review group area may reveal methods and search strategies that can be adopted/adapted for an upcoming review. If an existing search strategy from a published SR is utilized, this should be refer-enced accordingly in the manuscript detailing the new review.

DATABASE OF ABSTRACTS OF REVIEWS OF EFFECTS (DARE)

DARE was produced and funded by the National Institute for Health Research (NIHR) CRD at the University of York, the United Kingdom. “NIHR funding to produce DARE ceased at the end of March 2015,” and the database was last updated in May 2015 (tinyurl.com/pbj3f8d). DARE contains abstracts of non-Cochrane SRs that include a summary of the review and an assessment of the quality of the methods used in the review process. Two independent researchers screened identified citations by using the following criteria:

1. Were inclusion/exclusion criteria reported? 2. Was the search adequate? 3. Were the included studies synthesized? 4. Was the validity of the included studies assessed? 5. Are sufficient details about the individual included studies presented?






DARE records do not open up to a full-text article since the point is the qual-ity assessment. DARE covers a broader range of health care reviews than CDSR, because it combs through literature from many sources. Although an SR is assessed in DARE, that does not imply a recommendation of the quality of that SR; however, DARE does alert researchers to important points about a review’s methodology and other components—raising questions that can assist in appraisal of SRs.

COCHRANE CENTRAL REGISTER OF CONTROLLED TRIALS

Cochrane Central Register of Controlled Trials (CENTRAL) includes reports of RCTs and semi-RCTs located via MEDLINE and EMBASE databases, and hand searches by the Cochrane review group members of both published and unpublished literature sources. The unpublished sources are what make CEN-TRAL a valuable resource, since some included trials may not be found by searchers anywhere else. CENTRAL records include the title, citation informa-tion, and a summary, but not the full text of the article.

Other Useful Cochrane ResourcesThe following are links to useful tools for navigating through the Cochrane Library and writing SRs:

■■ Cochrane Collaboration Glossary of Research Termswww.cochrane.org/glossary/5

■■ Cochrane Collaboration Handbook for Systematic Reviews of Interventionswww.cochrane.org/resources/handbook

■■ Cochrane Community—Information for Practitioners, Providers, and Policy Makerswww.cochrane.org/information-practitioners

■■ Cochrane Library User Guide, Tutorials, and Live Online Workshopsolabout.wiley.com/WileyCDA/Section/id-390244.html

DYNAMED

DynaMed is a point-of-care resource created by physicians to assist health pro-fessionals in answering background and foreground clinical questions. DynaMed is a useful tool for both qualitative and quantitative SR research, but it may prove more useful for writing quantitative reviews such as reviews of therapeutic interventions. DynaMed can assist in accumulating background information for context/support of foreground clinical questions, making it useful for writing the background or introduction section of an SR. It can help researchers locate citations of individual studies for possible inclusion or exclusion in an SR and discover information that is relevant to SR research and that is not found in jour-nal articles. DynaMed can also be used to locate reviews, guidelines, and grey literature.

http://www.cochrane.org/glossary/5

http://www.cochrane.org/resources/handbook

http://www.cochrane.org/information-practitioners

http://olabout.wiley.com/WileyCDA/Section/id-390244.html






DynaMed includes searchable evidence-based summaries covering diseases or conditions and drugs. Each topic summary is broken into common sections—Epidemiology, Etiology and Pathogenesis, History and Physical, Diagnosis, Treat-ment, Complications and Prognosis, Prevention and Screening, Guidelines and Resources, and References. Browsing DynaMed by category will reveal sec-tions on Differential Diagnosis, Diagnostic Testing, and Quality Improvement. DynaMed editors monitor journals, journal review services, SR collections, guideline collections, drug information sources, and other relevant sources to facilitate updating topic summaries. Updating is a process that involves a criti-cal appraisal system to review and rate new evidence. DynaMed editors con-centrate on revealing ratings for therapy-based content, although there are ratings of content from question domains such as Etiology.

ESSENTIAL EVIDENCE PLUS

Essential Evidence Plus, previously known as InfoRetriever, was started in 1994 from a project created by Dr. Mark Ebell. Essential Evidence is a filtered, evi-dence-based, clinical decision support tool that allows medical providers to find quick answers to clinical questions for more than 13,000 topics in the exam room or at the bedside. More than 5,000 Patient-Oriented Evidence That Matters (POEMs) are available in Essential Evidence. POEMs are evidence-based summaries of an original article or review. A citation of the article or material that is the focus of the POEM is included. (It is possible to sign up for daily POEMs for delivery by e-mail, and for podcasts for Essential Evidence.) Essen-tial Evidence Plus also has abstracts of Cochrane reviews, evidence-based medicine guidelines, and decision support tools such as interactive diagnostic test calculators.

Essential Evidence Plus consists of 10 databases and/or interactive tools: Essential Evidence Topics, Cochrane SRs, Evidence-Based Medical Guidelines, Evidence Summaries of Guidelines, POEMs Research Summaries, DERM Expert Image Viewer, Decision Support Tools, History and Physical Exam Calculators, Diagnostic Test Calculators, and Expert, Evaluation, and Management (E/M) Codes from Medicare. Topics in Essential Evidence Plus are periodically updated as new information (e.g., SRs, guidelines) becomes available.

JBI FOR NURSING AND MIDWIFERY

The JBI for Nursing and Midwifery began in 1996 at the Royal Adelaide Hospi-tal in Australia and is named after the first matron (chief nursing officer) of that hospital. JBI is an international organization with evidence-based health care resources for allied health, medicine, nursing, and midwifery professionals (JBI, 2014). JBI resources include review protocols, completed SRs, best prac-tice information sheets, evidence summaries, and tools for assisting in appraisal of literature for SRs. Tools within the System for the Unified Management, Assessment, and Review of Information (SUMARI) include Comprehensive






Review Management System (CReMS) software to assist reviewers in the man-agement and assessment of search results. Modules are also a part of SUMARI, and they facilitate critical appraisal of specific types of literature.

■■ QARI—Qualitative Assessment and Review Instrument■■ MASTARI—Meta-Analysis of Statistics Assessment and Review Instrument■■ NOTARI—Narrative, Opinion and Text Assessment, and Review Instrument■■ ACTUARI—Analysis of Cost, Technology and Utilization Assessment, and

Review Instrument

From the JBI website, researchers can access a reviewer’s manual with details for the JBI SR process, as well as details on JBI tools to assist in that process.

The JBI evidence-based practice (EBP) database can be searched by using the JBI website (site registration and subscription required for access to certain materials) or through the Ovid search retrieval system (check your academic library for access to the JBI EBP database). When using the JBI EBP database on Ovid, searches can be narrowed down to specific publication types (e.g., SR protocol, SRs, evidence summaries) and/or subject areas (e.g., Aged Care, Car-diovascular Care, Health Management and Assessment, Renal Care).

Additionally, content from the referred journal, JBI Database of Systematic Reviews and Implementation Reports, is indexed or included in CINAHL, EMBASE, MEDLINE, and SCOPUS databases.

PROSPERO

PROSPERO is an international, open access (free) registry for SR protocols. PROSPERO is funded by the NIHR CRD at the University of York, the United Kingdom. The aim of PROSPERO is to make SR research transparent and to prevent waste by alerting reviewers of projects already in process. As of Octo-ber 2014, PROSPERO includes SR protocols that have been submitted to the CDSR (www.crd.york.ac.uk/PROSPERO). It should be noted that protocols included in CDSR before October 2014 are not included in PROSPERO, so reviewers wanting to find SR protocols should search both CDSR and PROS-PERO databases.

Protocols include the rationale for conducting an SR, hypotheses, and detailed plans for how the reviewers plan to carry out their review. Reviewers can search PROSPERO and benefit from reading through the protocols of studies on topics similar to theirs. Sometimes, protocols include full search strategies. If reviewers decide to use methods or searches from protocols they find in PROSPERO, even if they alter the searches to fit their needs, they should specify that they have done so and cite the original protocol in their manuscript.

Reviewers may also wish to register their own SR protocols in PROSPERO. PROSPERO will include SR protocols on any health-related outcome. It is recommended that reviewers follow the Preferred Reporting Items for Systematic

http://www.crd.york.ac.uk/PROSPERO






Review and Meta-Analysis Protocols (PRISMA-P) when writing and reporting their protocols (www.prisma-statement.org/Extensions/Protocols.aspx).

■■ Unfiltered Evidence Resources

CUMULATIVE INDEX TO NURSING AND ALLIED HEALTH LITERATURE (CINAHL)

CINAHL, originally a print resource in the 1950s, has been an electronic data-base since 1983.

Some citations in CINAHL may also be in other biomedical databases such as MEDLINE and the British Nursing Index (BNI), a resource from partner libraries at Bournemouth University, Poole Hospital National Health Service (NHS) Foundation Trust, the Royal College of Nursing, and Salisbury NHS Foun-dation Trust, but CINAHL does offer unique content. CINAHL is useful for locating qualitative studies dealing with nursing topics. Search CINAHL, mak-ing use of CINAHL headings as well as natural language or text words. There are two ways to locate CINAHL headings. Searchers can search the CINAHL headings database directly by clicking on CINAHL headings at the top of the page or, from the Advanced Search page, they can check a box that says, “Sug-gest Subject Terms.” The CINAHL database indexes nursing, allied health, bio-medical, complementary and alternative medicine, consumer health, and health sciences librarianship literature. Types of literature indexed include commer-cially published literature such as journal articles, book chapters, articles with care plans and critical pathways, nursing dissertations, audiovisuals, confer-ence proceedings, and accredited continuing education articles. Furthermore, CINAHL Plus and CINAHL Complete index records for research instruments. Contact your librarian to determine which version of CINAHL you can access. As of this writing, there are five versions of CINAHL: CINAHL, CINAHL Plus, CINAHL with Full Text, CINAHL Plus with Full Text, and CINAHL Complete. The difference among these is not insignificant (see www.ebscohost.com/pro moMaterials/CINAHL_NRC_Chart_06202013.pdf). Cited reference searching in CINAHL Plus and CINAHL Complete dates back further and includes more journals than CINAHL. All versions of CINAHL offer a save searches/alerts option for saving searches and running periodic alerts so that new search results are e-mailed or available via a Really Simple Syndication (RSS) feed.

As with most bibliographic databases, CINAHL offers many search features— including advanced search modes. But regardless of which mode is being used, CINAHL and other traditional electronic databases, such as PubMed, have a list of stop words that, when placed in a search box, will be ignored by the database. The list is periodically updated and currently includes and, are, as, at, be, because, been, but, by, for, however, if, not, of, on, or, so, the, there, was, were, whatever, whether, and would. Tips for using basic and advanced searches are included in the succeeding sections.

Searchers can use Boolean and proximity operators in CINAHL as well as truncation. CINAHL differentiates Boolean operators from stop words when

http://www.prisma-statement.org/Extensions/Protocols.aspx

http://www.ebscohost.com/promoMaterials/CINAHL_NRC_Chart_06202013.pdf

http://www.ebscohost.com/promoMaterials/CINAHL_NRC_Chart_06202013.pdf






Boolean operators are typed in all caps. Use the asterisk (*) for truncation. The syntax for proximity searching is N#, where # is the number of possible words between entered search terms.

In the advanced search mode, reviewers can choose to search for key terms or search for controlled vocabulary known as CINAHL headings. To search for CINAHL headings from the Advanced Search page, check the box that says “Suggest Subject Terms.” The advanced search feature in CINAHL allows search-ers to build their search, placing CINAHL headings and plain language terms that represent one concept in one or more search boxes and connecting those terms with the Boolean operator OR. Avoid trying to place all possible key terms for all concepts in a search question in the search box at once. This is the equivalent of placing all ingredients for a stew in a pot at once. The final prod-uct may or may not be to one’s liking. Running searches for each concept sepa-rately allows searchers to catch mistakes or terms retrieving undesirable results and saves searchers time in the end. Searchers can then combine each concept search from the Search History page. The best practice is to use both controlled vocabulary and natural language terms in SR search strategies. CINAHL offers several ways for limiting searches; however, the best practice for creating SR search strategies is to avoid using database supplied limits.

Each library may choose a different look for CINAHL’s basic and advanced search modes. In addition, the library may make decisions on CINAHL limits options or ways in which database researchers can narrow down search results, as well as decide which other CINAHL features to include or exclude, such as the inclusion of a visual search mode in CINAHL. Researchers should consult with a librarian at their institution regarding which version of the CINAHL database is available and on library changes or customizations to CINAHL that may affect their searches.

EMBASE

The EMBASE database includes records dating back to 1947 and selective records dating back to 1902. EMBASE includes all of MEDLINE and an addi-tional 2,600-plus journal titles that are not indexed in MEDLINE. EMBASE covers biomedical, life sciences (note that the term life sciences includes nurs-ing literature), and pharmaceutical literature worldwide. Although both MED-LINE and EMBASE include literature from all over the world, EMBASE is known to be more inclusive of literature outside the United States. What really sets EMBASE apart, however, is its index of controlled vocabulary called Emtree. MEDLINE includes around 28,000 MeSH terms, whereas Emtree has more than 72,000. For this reason, searchers will often find MEDLINE records in EMBASE that they could not locate in MEDLINE.

EMBASE can be searched on different search retrieval systems or platforms such as those provided by Dialog, LLC (Dialog and DataStar platforms), Ovid Technologies (Ovid platform), and Elsevier B.V. (Embase.com). All the exam-ples given next are from Embase.com.






Under the search tab at Embase.com are six options: Quick, Advanced, Drug, Device, Disease, and Article. In Quick Search, a searcher can type text words/phrases in a search box. There is a check box for broadening a search (mapping, explosion as key word) that is check marked by default and, if left checked, searchers will broaden their search, because EMBASE will apply auto-matic mapping. Automatic mapping means that words typed in the search box will be mapped to Emtree terms and automatically added to the entered text words with the Boolean operator, OR. The best way of starting your search in EMBASE, however, is to click on Browse and choose Emtree. From here, you can discover the Emtree terms that match your concepts, find additional synonyms, and begin to build your search.

Proximity operators in EMBASE are NEAR/# and NEXT/#, where # is the number of words between the search terms. NEAR is interpreted as meaning within the number of words of each other in any order, and NEXT means the same but the words must be in the same order in the citation record as they are in the search strategy. Use of the truncation feature allows for retrieving word variations, one or more letters, or a single letter. To retrieve results with the dif-ference of one or more letters, use an asterisk (*); for a single letter, use a question mark (?).

EMBASE includes features for managing search results such as save search, e-mail alerts, RSS feeds, and a way to export citations into bibliographic citation managers.

HealthSTAR

The Health Services, Technology, Administration, and Research (HealthSTAR) database contains materials from 1966 to the present on health services research, including clinical and administrative aspects involved in health care delivery. There is information on evaluation of patient outcomes, effectiveness of procedures, programs, products, services, processes, health care administra-tion, economics, planning, and policy.

Citation records include journal articles, monographs, technical reports, meeting abstracts and papers, book chapters, government documents, and newspaper articles. Reviewers can use MEDLINE MeSH subject terms to retrieve citations from HealthSTAR, as records are indexed via MeSH.

The HealthSTAR database is available on the Ovid search retrieval system and, similar to other databases on Ovid, researchers can use Ovid’s basic and advanced search features with HealthSTAR. A basic search in most Ovid data-bases, such as HealthSTAR, allows searchers to enter text words/phrases that are connected by Boolean logic operators, such as AND, OR, or NOT. The Ovid system then processes the search terms, and results ranked by relevance are presented. Default fields that search for the text words/phrases that are entered into the search box are abstract, name of substance word, title, subject heading word, and original title. Reviewers using advanced search in Ovid HealthSTAR can directly identify potential MeSH terms for text words that they enter into






a search box if they leave the “Map Term to Subject Heading” box checked before processing a search. Ovid advanced search is best explored through a tutorial, library database workshop, or through a visit with a librarian who is involved in reference work or literature research activities. Besides MeSH, advanced search allows for use of features to narrow down—center or focus and/or expand or explode—search terms. Use of the focus feature retrieves citations where a searcher’s term is the primary or major focus of an article. Use of the explode feature retrieves not only general articles on a search concept but also articles in subcategories that are represented under a concept (see the section “Subject Heading Features” in Chapter 5 for more details on explode and focus).

Even though reviewers may search Ovid HealthSTAR in the advanced search mode, it is still possible to incorporate some basic search features in the advanced mode, as well as to make use of special search features such as truncation and wildcards (see Chapter 5 for a review of truncation and wildcards).

The HealthSTAR database on the Ovid search retrieval system includes simi-lar search/citation management features that are also available through other search retrieval systems (e.g., EBSCOhost, ProQuest) for databases. Researchers can create personal accounts on Ovid to hold saved searches and new citation alerts. Citation records can be exported directly into reference management software.

MEDLINE

MEDLINE is the NLM original database with biomedical journal citations from more than 80 countries. The subject scope of MEDLINE is biomedical and health sciences literature, including medicine, nursing, allied health, dentistry, veterinary medicine, life sciences, behavioral sciences, chemical sci-ences, bioengineering, and biophysics. MEDLINE is available through several different search retrieval systems, including the NLM’s National Center for Biotechnology Information (NCBI) Entrez retrieval system (PubMed), which is free, Wolters Kluwer Health’s Ovid system, and EBSCO Publishing’s EBSCO-host system. PubMed also contains many citations from PubMed Central (PMC) database, which includes free access to manuscripts that are not available else-where. Most medical libraries provide link-outs to their journal subscriptions from PubMed, enhancing full-text access for researchers. Ovid and EBSCO require a paid subscription to access MEDLINE, but they allow for proximity searching that is not available on PubMed. Contact your local medical library to find out what access options for MEDLINE are available to researchers. Reference or research librarians will most likely be able to assist you with an advanced MEDLINE search.

Searching MEDLINE Through PubMedMEDLINE uses MeSH, a controlled vocabulary. As previously mentioned, con-trolled vocabulary is used by database producers to place citations in categories or under subject headings to facilitate maximum and comprehensive retrieval of






relevant articles by researchers. PubMed automatically “maps” or closely matches terms typed in the basic search box to a MeSH term before processing a search.

The default search in PubMed appears as a basic search box on the PubMed home page, but be aware that this is no ordinary simple search, because natural language text words entered into the search box are mapped to MeSH terms and additional plain text words automatically. Although this is a useful feature, it is a good idea to check the search details found on the right side and toward the bottom of the page. PubMed may have mapped the text word to MeSH terms and plain language terms that you had not intended. For example, PubMed maps nursing to terms related to the profession and to breastfeeding. Learn how to search PubMed via the MeSH database and gain more control of your search. Build specific, focused search queries that capitalize on the way articles are grouped in PubMed—by MeSH.

To explore searching PubMed by using the MeSH database, select MeSH from the drop-down list next to the search box. At this point, you are not search-ing for articles; you are actually searching for something analogous to a dic-tionary or thesaurus. Once you have finished choosing MeSH terms, then you can use those to search PubMed for articles. In the search box next to MeSH, type a text word. The MeSH database will retrieve a list of controlled vocabu-lary terms or subject headings for review. Click on the MeSH term (if any) that best describes the original concept. At this point, you are in the actual record for the MeSH term. In that record, you can read the definition of the term to be sure that is, in fact, what you intended. You can see a list of synonyms for that term, options to explode or to search for it as a main focus, and also see the hierarchy (tree) of MeSH terms. The hierarchy can be useful in locating potential MeSH terms for your topic. Terms listed above the entry MeSH term are broader, and those listed below are narrower. If the entry term is the one you want, click on the option “Send To,” which places the chosen MeSH term in the search box. Note that MeSH terms are exploded automatically. This means that all citation records that have been indexed under the more narrow terms will also be retrieved. To unexplode MeSH terms before sending your entry term to the search box, you can check a box in the MeSH record that reads, “Do not include MeSH terms found below this term in the MeSH hierarchy.” Usually, searching MeSH terms that are exploded is the best. Repeat this process as many times as needed with other subject headings by using the provided Boolean operator choices under the search box. Text words can and should be added to the final query before the search is processed. A PubMed MeSH Database tuto-rial (search the Internet to find one from the NLM) or a workshop session with a reference librarian will prove invaluable to learning how to search MEDLINE within PubMed.

Still another way to search PubMed is through the advanced search page, a cool way to search, allowing researchers a preview of the number of search results for each query as they continue building their search. In PubMed Advanced Search, it is easy to see options for limiting searches to a certain author, journal,






publication date, or date range, as well as a host of other options. Researchers can see all the searches they have run in their session and combine searches from the advanced search page.

PubMed’s Clinical Queries include a search for clinical studies by the fore-ground question category, SRs, and citations on medical genetics. After a reviewer enters his or her search statement in the Clinical Queries search box and selects a question category, along with the scope for the citations to be retrieved (broad or narrow), the queries page processes the search, noting the number of primary studies, SRs, and medical genetics citations on the search statement. The search filters developed by Dr. R. B. Haynes are behind the Clinical Queries search algorithms (www.ncbi.nlm.nih.gov/entrez/query/static/clin icaltable.html) and ensure that a reviewer will retrieve evidence-based citations on his or her topic—bypassing nonevidence that otherwise would have been part of the result set.

One tip for managing PubMed searches is to set up a group account that will be accessible by all reviewers and/or individual accounts by using the My NCBI account sign in option. A My NCBI account will facilitate saving original searches and collections of selected search results (from original searches, related article searches, author chasing, and results of hand searches of journals indexed on PubMed). Search results can be set up so that reviewers will receive an automatic e-mail when new citations are available for their SR topic.

Under a researcher’s personal or group My NCBI account, there are cus-tomizations that facilitate reviewing of citations in a more efficient manner. Use the My NCBI’s Preferences option to choose a highlighting color. This will mean that the next time a search is performed in PubMed and researchers are logged in to that PubMed NCBI account, search results will show original key words in a highlighting color as researchers review citations. My NCBI also allows for search filters to be applied to searches each time a researcher is signed into his or her My NCBI account, selects frequently requested search filters, searches for specific filters, or browses for additional filters.

Searching MEDLINE on OvidThe basic search option in Ovid MEDLINE behaves just like entering search terms in PubMed. Ovid will automatically map text words entered into a search box to MeSH terms and additional plain language terms and run the search. These basic searches in MEDLINE are appropriate early in the reviewers’ research process, but more detailed and comprehensive searches will be needed for researching and writing SRs.

To locate MeSH terms in MEDLINE on Ovid, use the advanced search mode. Check the box that reads, “Map Term to Subject Heading” and enter your natural language term or text words/phrases. Ovid will present you with a list of possible MeSH terms. Choose the terms that best match your intended concept, and Ovid will guide you from there. If a researcher is unsure of the meaning for particular MeSH terms, he or she can click on the “Scope Note” to read the definition for each term.

http://www.ncbi.nlm.nih.gov/entrez/query/static/clinicaltable.html






Ovid MEDLINE allows for truncation, wildcards, and proximity opera-tors. The symbols for unlimited truncation are the dollar sign ($) or colon (:). “Unlimited” means that you will retrieve all versions of the root word; for exam-ple, entering nurs* will retrieve records that contain nurse, nursing, nursed, and so forth. If a searcher would like his or her key word expanded with a single extra character, then he or she should consider the use of the question mark (?) for wildcard truncation (e.g., an?esthesia or p?ediatrics). This may prove useful when a researcher is trying to retrieve documents with both American (e.g., anesthe-sia, pediatrics) and British spellings (e.g., anaesthesia, paediatrics). The proxim-ity operator for Ovid is adj#, where # is the number of possible words between search terms. Similar to the NLM, the producer of PubMed, Ovid Technologies offers a variety of tutorials (www.ovid.com/site/support/training .jsp) in several formats to help searchers become proficient in searching databases such as MEDLINE on its platform.

ProQuest DISSERTATIONS AND THESES DATABASE

The ProQuest Dissertations and Theses (PQDT) database is derived from doc-toral dissertations and master’s theses from graduate schools and universities internationally. Full text is available for many dissertations and theses, depend-ing on the type of subscription that a library has with ProQuest. New entries are added to the PQDT database and indexed or organized by using a unique ProQuest vocabulary of terms. Some records are tagged or organized by using terms supplied by the author of a dissertation or an abstract.

PQDT can be searched through basic and advanced search modes. As with other online search databases, selected citation records can be automati-cally downloaded into citation management software. PQDT includes the My Research tab that provides researchers with a way to save searches/alerts. See Appendix D for additional guidance on searching for dissertations and theses.

PsycINFO

PsycINFO is a database from the American Psychological Association (APA) that provides systematic coverage of the behavioral sciences and mental health literature from the late 16th century to the present. The database includes cita-tion records from European literature, U.S. dissertations, and book chapters.

Similar to some databases, PsycINFO is available to be searched on more than one search retrieval system through vendors such as APA PsycNET, DIALOG/DataStar, EBSCOhost, Ovid Technologies, and ProQuest. Search features (described in subsequent paragraphs) are available when a reviewer searches PsycINFO on the Ovid search retrieval system. For information on searching PsycINFO on other search retrieval systems, see the APA website for database training search guides.

These features apply to both basic and advanced searches in PsycINFO: phrase searching, truncation with the dollar sign ($), and wildcard truncation

http://www.ovid.com/site/support/training.jsp






with the number sign or question mark (# or ?) for retrieving specialized plural forms of a word. For example, searching for the term wom#n retrieves results that contain both woman and women. A wildcard search for colo?r retrieves results that contain both color and colour.

A basic search in PsycINFO allows searchers to enter a topic by using natural language terms (key words), including those that are in the form of a question. The database system then parses the question into a search strategy and ranks results by relevance. Default fields searched include the abstract, heading word, table of contents (TOC) titles/headings, tests and measures, key phrases, and title. Key phrases concisely summarizes a document’s subject content and is in addi-tion to subject headings. The word heading searches the words and phrases in subject headings, and the TOC searches the TOC of books. Tests and measures mentioned in documents are identified by using a key words search.

Materials indexed in PsycINFO can be accessed via natural language searches and/or searches using controlled vocabulary (e.g., subject headings). If search-ing PsycINFO on the Ovid search retrieval system, available features for the application to choose subject headings include explode and focus (see the sec-tion “Subject Heading Features” in Chapter 5 for more details on explode and focus). PsycINFO has two areas with a controlled vocabulary/classification: the-saurus of psychological index terms and classification categories and codes. The thesaurus of psychological index terms can be considered the APA’s version of MeSH. During an advanced search in PsycINFO, researchers can select appro-priate thesaurus terms as shared by the database to increase retrieval of desired search results. PsycINFO offers a cited reference search, a technique described later in this chapter.

PsycEXTRA AND PsycARTICLES

PsycINFO is only one of the databases from the APA. PsycARTICLES is a full-text database of publications from the APA Educational Publishing Foundation, the Canadian Psychological Association, and Hogrefe Publishing Group. It has journal coverage from approximately 1894 to the present. All full-text citation records in PsycARTICLES are indexed as citations in PsycINFO.

Another APA database that may be of interest is PsycEXTRA, a grey litera-ture database. PsycEXTRA content is written for professionals, is disseminated outside of peer-reviewed journals, and includes abstracts and citations. To date, there is no coverage overlap between PsycINFO and PsycEXTRA.

SCOPUS

Scopus is a relatively new database, less than 20 years old, including both peer-reviewed biomedical and social science literature (books, conference pro-ceedings, scientific articles). Many different journal citations are represented in Scopus that may not be in EMBASE, MEDLINE, or Web of Science (WoS; Gavel & Iselid, 2008). Scopus is set up to allow researchers to retrieve most






EMBASE database records in searches. The EMBASE search via Scopus allows for searching EMBASE via a text words search compared with searching EMBASE directly by using text words and/or the Emtree-controlled vocabu-lary. This means that EMBASE records using different text words or those that are indexed under a specific Emtree-controlled vocabulary term may be missed. For example, a Scopus search for the text words heart attack misses EMBASE records mentioning myocardial infarction or indexed under the Emtree vocab-ulary term heart infarction (www.elsevier.com/solutions/embase-biomedical-research/learn-and-support).

Use Boolean logic operators in advanced searches in Scopus. There are also field codes such as TITLE-ABS-KEY, which are used in advanced searches to locate text words/phrases in the title, abstract, or key word fields of a citation record. A feature of Scopus is Refine Results, which allows a researcher to limit to or exclude search results by selecting one or more categories (e.g., year, sub-ject, area, author name, document type). For narrowing down a search to more relevant citations, there is also the Search Within Results box, where a researcher can type additional words/phrases to extract specific citations from original search results.

Regarding viewing and saving search results, there is a point that may have not been discussed in other chapters, but which should prove valuable. After performing searches in a database such as Scopus, it is wise to use a feature (e.g., Outline Query) to view, copy, and save search strategies used. Recall that researchers can save searches in personal accounts on database search retrieval systems and re-execute those searches later. However, it is also advisable that researchers copy/paste the outline of entire search strategies as an addition to their personal SR search worksheets or trackers. Examples of one or more search strategy outlines can be found in most SR appendices.

Both Scopus and WoS are useful for performing cited reference searches. With a cited reference search, a reviewer is looking for the names of other authors who had cited or included an original author’s paper as part of their work. This is actually a form of the search technique citation chasing.

WEB OF SCIENCE

WoS is available through the Institute for Scientific Information (ISI) Web of Knowledge platform provided by Thomson Reuters. It consists of seven data-bases. They are Science Citation Index (SCI) Expanded, Social Sciences Citation Index (SSCI), Arts and Humanities Citation Index (A&HCI), Conference Pro-ceedings Citation Index–Science (CPCI–S), Conference Proceedings Citation Index–Social Sciences and Humanities (CPCI–SSH), Index Chemicus, and Cur-rent Chemical Reactions. The first three databases allow researchers to perform cited reference searching. The conference proceedings indexes include pub-lished literature of the significant conferences, symposia, seminars, colloquia, workshops, and conventions in various disciplines. Many nursing researchers

http://www.elsevier.com/solutions/embase-biomedical-research/learn-and-support

http://www.elsevier.com/solutions/embase-biomedical-research/learn-and-support






look for materials for SR work in SCI Expanded, SSCI, and the conference pro-ceedings indexes.

WoS covers scientific topics from agriculture to zoology, including behav-ioral sciences, biochemistry, biomedical sciences, chemistry, medicine, micro-biology, and pharmacology, and it dates back to 1900.

Some of the disciplines covered in SSCI are psychiatry, psychology, public health, social work, and substance abuse, and this database goes back to 1956.

Records from WoS are not indexed via a searchable list of controlled vocab-ulary words, so searchers need to search WoS by using synonyms for their search terms (e.g., heart attack or myocardial infarction or cardiovascular stroke).

The truncation symbols used in WoS are the asterisk (*), the question mark (?), and the dollar sign ($). The asterisk is used to replace any group of characters. This is useful for truncating journal or publication titles. The question mark represents any single character. If a researcher were unsure of the ending of an author’s last name, the question mark would be a useful truncation symbol. The dollar sign can represent a single character such as “a” in the British spelling for p?ediatrics.

WoS researchers can create a personal ISI Web of Knowledge profile. This personal account allows for saving searches (any search up to 20 sets) and for creating search alerts, including alerts of when an article on a citation alerts list has been cited by a new article. Alerts are active for 1 year, but they can be renewed as needed. WoS allows researchers to perform cited reference searches. Search for a cited reference by searching for the author, the work’s source, and the publication year.

Most libraries or institutions have WoS set up to open to its basic search page, which includes three search boxes for researchers to type in search terms. WoS allows for addition of more search boxes as needed. For each search box, there is a corresponding drop-down menu to choose the search field represent-ing the location in which the search term can be found (e.g., in the topic, title, document type). Terms in each search box are under a searcher’s control for linkage with either selection of Boolean logic operators (AND, OR, or NOT ) to connect words from one search box to another. Refine search results by apply-ing limits from one or more of the following categories: subject areas, document types, authors, source titles, publication years, institutions, funding agencies, languages, or countries/territories.

When searching WoS, it is important to remember that capitalization of text words is not necessary, and the use of Boolean operators varies, depending on which search fields a researcher is selecting for a search box entry. For exam-ple, use the Boolean operator AND when searching for a topic but not when searching for a publication name. Use quotation marks when searching for an exact phrase as part of a topic or title.

There is an advanced search option in WoS that allows investigators to search for records by using field tags (e.g., TI = title, CF = conference) and to






combine sets of results. The following link is from WoS producers and includes brief videos on searching WoS products: wokinfo.com/training_support/train ing/recordedtraining/#wos.

■■ Grey Literature

Grey literature was defined in 2004 at the International Conference on Grey Literature as government, academics, business, and industry information in electronic and print formats that is not controlled by commercial publishing (GreyNet International, 2009). Grey literature should be searched as applicable to avoid publication bias in SRs. Elements involved in publication bias may include limiting searches for SR articles to “English language,” omitting nega-tive studies, studies with less favorable results, or those showing no difference in effect. More successful studies or studies with more positive outcomes may already overwhelm the literature; therefore, when performing searches for stud-ies to be included in an SR, researchers should ensure to identify and transfer the full text of all qualified studies for possible inclusion into an SR. Include studies with opposing views in SRs if they meet the SR inclusion criteria.

Government or organization reports, conference proceedings, and trials where the results were not published in a traditional venue can all show evi-dence that should be included in an SR. Statistical analysis of the research cited in published reviews has shown that the lack of grey literature can lead to over-estimation of the impact of an intervention (Hopewell, McDonald, Clarke, & Egger, 2007; McAuley, Pham, Tugwell, & Moher, 2000).

SRs of therapies or interventions are not the only types of SRs that should include grey literature. Benzies, Premji, Hayden, and Serrett (2006) created a list of questions that can be asked when deciding whether to include grey literature in an SR.

Finding grey literature in traditional electronic biomedical databases such as PubMed or CINAHL is unlikely. Grey literature is not peer reviewed and is less likely to reference sources of information. Blackhall (2007) identified six ways to locate grey literature: search CENTRAL, hand searching, review data-bases of unpublished and ongoing studies, search the Internet, comb through reference lists, and seek and consult subject experts.

Another source of grey literature in the health sciences field is the Grey Lit-erature Report from the New York Academy of Medicine (NYAM; www.greylit .org). The report is bimonthly, alerting readers to new grey literature publications in public health and health services research topics (NYAM Library, 2009).

Grey literature can also come from websites on laws, drug-regulating body documents such as those from the U.S. Food and Drug Administration (FDA), consultations with professionals at departments of health and related depart-ments, and nursing centers housing health research and policy data. See Appendix C for additional select Social Science and Biomedical Science Grey Literature.

http://www.greylit.org

http://www.greylit.org






Grey literature can be an important addition in SRs, even though such litera-ture does not usually meet the same strict guidelines or provide a high level of evidence similar to RCTs. Document the form of grey literature used, date/time span of the literature, and other details to keep track of what has been reviewed.

CONFERENCE PROCEEDINGS

It is a good idea to add conference proceedings, congresses, and meeting abstracts to reviewers’ lists of places to search for information related to an SR topic. These may be pieces of literature addressing new angles or putting forth new solutions for a foreground question, which may not yet be part of an article indexed in traditional resources. Searching proceedings and the like help decrease publication bias in an SR, as reviewers are attempting to get all pos-sible literature on their foreground question. Some traditional search resources or databases may include information on meetings too, such as the resources BioMed Central (BioMed Central Ltd.; free resource), BIOSIS Previews (Thomson Reuters), EMBASE, MEDLINE, NLM Gateway (free resource), ProceedingsFirst (OCLC), WoS, and WorldCAT (OCLC).

DATABASES OF ONGOING RESEARCH

Clinical TrialsClinical trials may fit into the category of a type of continuing or ongoing research. Trials in this category include those currently enrolling subjects, active trials beyond the subject enrollment phase, and trials where data analysis is in process. For trials where the results are never published, these resources may be the only online record of the work.

In 2000, the NLM created ClinicalTrials.gov, the first trial registry. In 2007, the FDA Amendments Act was passed; it required phase II trials and beyond to be registered at ClinicalTrials.gov and, regardless of publication, results of the trial must be listed in the registry, generally within 1 year of the trial’s com-pletion. In addition, the International Committee of Medical Journal Editors (ICMJE) noted that from 2005 onward, there would be a commitment to pub-lish research where a trial had been registered. As a result, trial registration in the United States increased and new registries were created (The Evolution of Trial Registration, 2009; Mathieu, Boutron, Moher, Altman, & Ravaud, 2009). The good news is that there are clinical trial registries, and the bad news is that there is not just one trial registry in each country or worldwide. This means that searches for information may have to be performed in multiple registry databases. Sometimes a website such as Current Controlled Trials (www .controlled-trials.com) is available and such a site offers a metaRegister of Controlled Trials (mRCT; http://mrctcenter.org), allowing researchers to search multiple registries at once.

The rate of publication of registered trials has been reviewed in the literature, and it has been noted that not all trials are published and that inconclusive and

http://mrctcenter.org

http://www.controlled-trials.com

http://ClinicalTrials.gov







negative results are less likely to be published. Further, discrepancies are present in some published results that are noted in some registries (Mathieu et al., 2009; Ross, Mulvey, Hines, Nissen, & Krumholz, 2009). The group Consolidated Stan-dards of Reporting Trials (CONSORT; www.consort-statement .org) advocates for transparent reporting of RCTs.

ClinicalTrials.govAs previously mentioned, ClinicalTrials.gov is the largest and oldest of all trial registries, and it includes both government and privately funded trials. There are more than 80,000 trials, and the trials are from the United States and about 170 other countries. The ClinicalTrials.gov database is updated daily to accom-modate newly received trial data.

On the basic search page for ClinicalTrials.gov, enter important text words/phrases to locate specific studies. These may include words describing diseases, interventions, and/or trial locations such as South America (the ClinicalTrials .gov database does allow for browsing lists of trials by conditions, drug inter-vention, sponsor, and location). A search for two or more terms in ClinicalTri als.gov is processed by using AND as the default Boolean operator between the terms typed in the search box. Searchers can choose to exercise more control over searches in ClinicalTrials.gov by making use of the following Boolean operators: AND, OR, or NOT. Consider using quotation marks (“ ”) for terms that the database should interpret as a phrase or searching a Boolean operator as a word (e.g., “peanut butter and jelly”). Parentheses can also be used to control the way the database interprets a search statement; for example, (heart disease OR heart attack) AND (stroke OR clot). It is not necessary for researchers to brainstorm every synonym for a term and to include it in parentheses when searching ClinicalTrials.gov. The ClinicalTrials.gov database does try to deter-mine synonyms for search terms when possible. For example, a search for heart attack will also find occurrences of myocardial infarction.

For more precise results, use the ClinicalTrials.gov advanced search screen and use, when appropriate, the optional additional fields provided. Only fill in the fields appropriate for specific search needs. Boolean logic operators can be used in any field that is free text. Place the logic operators in all capital letters (see Table 6.1).

With either basic or advanced search, results will display a list of studies found. Click on the refine search option to modify search results by using the same fields offered in advanced search.

Searching for Registered Trials in PubMed and Ovid MEDLINEWhen attempting to locate registered trials in PubMed, enter ClinicalTri als.gov [SI] into the PubMed basic search box. The abbreviation SI stands for secondary source ID. The SI field contains information pertaining to the reg-istration of several types of data discussed in MEDLINE articles, including, as of summer 2005, clinical trial numbers. The SI field is composed of the source (e.g., ClinicalTrials.gov), followed by a slash, and then followed by an ID

http://www.consort-statement.org





















(e.g., ClinicalTrials.gov trial identifier number). Results from the SI field can be combined with text words to limit and focus the number of results.

To locate trials from ClinicalTrials.gov in MEDLINE on Ovid, click on the search fields option in Ovid MEDLINE and follow these steps:

■■ Step 1: Type ClinicalTrials.gov in the search box.■■ Step 2: Place a check mark in front of the box for SI.■■ Step 3: Click on the search button near the search box to run the search.

Review the results. Similar to searching PubMed, results from an SI search of MEDLINE on Ovid can be manipulated to retrieve a more focused set of clinical trial citations.

WHO International Clinical Trials Registry PlatformAnother ICMJE-approved registry, the searchable World Health Organization (WHO) registry focuses on bringing as many international registries together

TABLE 6.1 CLINICALTRIALS.GOV ADVANCED SEARCH—TIPS ON SEARCHING SOME FIELDS

ClinicalTrials.gov Advanced Search Field

Search Tip

Recruitment Use the provided drop-down menu to choose among all studies, open studies, or closed studies.

Study results Use the provided drop-down menus to choose among all studies, studies with results, or studies without results.

The study results field is relatively new, and not many studies include results.

Study type Use the provided drop-down menus to choose among all studies: interventional, observational, or expanded access studies.

Conditions, interventions, outcomes measures, lead sponsors, sponsors, and study IDs

All are fields allowing a researcher to type his or her search terms.

Lead sponsors, sponsors Check mark a box for exact match to have the database locate exactly what was typed in either the lead sponsors or sponsors option.

State/country Use the provided drop-down menu to choose state/country.

Location terms Location terms is a free text field allowing researchers to type free text to search a city or facility, as in Mayo Clinic. Not all studies include this level of detail, but if they do, this will find them.

Gender Select from the provided drop-down menu.

Age group, study phase, and funded by

More than one box can be checked for age group, study phase, and funded by.

Date Options include first received and last updated

http://CLINICALTRIALS.GOV

http://CLINICALTRIALS.GOV









as possible. According to the website, the WHO registry network is composed of WHO primary registries, partner registries, and registries working with the International Clinical Trials Registry Platform (ICTRP) toward becoming WHO primary registries.

Any registry that enters clinical trials into its database prospectively, that is, before the first participant is recruited, and meets the WHO registry crite-ria or that is working with ICTRP toward becoming a WHO primary registry, can be part of the WHO registry network. Partner registries meet the same criteria as primary registries, except that they do not need to be national or regional in scope, have the support of a government, be managed by a not-for-profit agency, or be open to all prospective registrants, as in disease-specific registries. Registry data sets from providers are updated every Tuesday evening.

The standard WHO ICTRP search is very sensitive, finding as many records as possible. The resource looks for terms (words or phrases) in the fields: title, primary sponsor, health condition(s), intervention(s), countries, main source ID(s), and SI(s). Do not enclose a phrase in double quotes; the system is set to consider two or more words as a phrase. The standard search automatically includes synonyms for terms.

Use Boolean logic operators such as AND, OR, or NOT in searches on WHO ICTRP. Do not use parentheses; the WHO ICTRP database does not process searches according to what is in parentheses first. The resource does process searches by paying attention to terms connected with the Boolean operator NOT first, then attends to terms joined by AND, and finally focuses on terms joined with OR. The asterisk (*) can be used as a wildcard for truncation in searches, but realize that this will keep the database from coming up with syn-onyms for words typed by a searcher. Truncation such as the (*) does not work if placed in the middle of a word or phrase; it works only when placed at the end of a word or phrase.

Advanced search in WHO ICTRP also allows for use of Boolean logic oper-ators in a similar fashion to that previously described in the basic search section. Researchers can still have more control in combining search terms in advanced search, because one or more fields can be searched and limits can be added at the beginning of a search. Fields that can be searched include title, condition, intervention, recruitment status, primary sponsor, SI, countries of recruitment, and date of trial registration.

It is possible to search for trials registered in certain WHO primary regis-tries in the following languages: Chinese, Dutch, German, Japanese, and Persian. Searches are sensitive to punctuation (anti-depressants vs. antidepressants). Consider searching for both terms by using OR. Any number of terms can be combined. However, some combinations of AND, OR, and NOT can lead to ambiguity. Care must be taken when constructing complex searches.

Research Portfolio Online Reporting Tool Expenditures and ResultsThe Research Portfolio Online Reporting Tool (RePORT) provides access to reports, data, and analyses of National Institutes of Health (NIH) research






activities, including information on NIH expenditures and the results of NIH-supported research. To search RePORT, use NIH RePORT Expenditures and Results (NIH RePORTER); this replaced Computer Retrieval of Information on Scientific Projects (CRISP). RePORTER has more information on research, providing links to PMC, PubMed, and the U.S. Patent and Trademark Office Patent Full-Text and Image Database.

RePORTER is limited by the fact that the information provided goes back 25 years; downloading is sort of all or nothing, and results go into Microsoft Excel. On the positive side, if articles have been published from the research, they are listed, contact information is easy to find, and it is possible to limit to projects that are funded by a specific agency or center, including the National Institute of Nursing Research (NINR).

The results of a search are returned in a project listing that can be down-loaded. Information includes project number and title, contact information for the principal investigator and organization, year(s) of funding, funding insti-tutes, and total amount of funding by year. The project number, subproject ID, and the project title are linked to more detailed information on a series of tabs: description, the science being conducted; detail, administrative and budget details on the project; results, the publications and patents associated with the project; history, past years’ project information; and subprojects (for multipro-ject grants only), all of the component subprojects.

RePORTER is updated on a weekly basis. Each update includes not only the addition of newly funded projects but also revisions to prior awards (e.g., change of grantee institution or revised award amounts).

Use one or more term or words to search RePORTER and place quota-tion marks around terms to be interpreted exactly as typed by the RePORTER database. Use Boolean logic operators AND or OR, but not both. AND is the default Boolean operator in RePORTER. RePORTER returns results (projects) in which the search terms are found within the title, abstract, or scientific terms. There is also a long list of stop words (referred by RePORTER as reserved words) that can be found at projectreporter.nih.gov/reporter_help.html.

Most fields have strictly controlled vocabulary terminology. Free text fields include terms such as principal investigator, organization, and request for applications (RFA)/program announcement (PA), which comes from RFA, PA, or notice (NOT) and Public Health Relevance. Not all fields are available for all years.

■■ Search Techniques

Depending on the foreground question, reviewers may also consider search methods or techniques such as the ones discussed in the subsequent sections to decrease bias and increase the number of citations in an SR. Many authors have weighed in on this. Some have taken a close look at SR research methodo-logy and search results, highlighting search flaws or missteps, as well as reviews

http://projectreporter.nih.gov/reporter_help.html






omitting or missing relevant articles on a foreground question (Bassler et al., 2007; Goodman, 2008; Honest & Khan, 2002; Lundh & Gøtzsche, 2008; Sterne, Egger, & Smith, 2001; Trowman, Dumville, Torgerson, & Cranny, 2007).

AUTHOR SEARCHING

Author searching is useful to find out who the prolific authors are in a topic area. This form of searching can be undertaken in several ways. Before electronic databases, most researchers looked for authors by checking the references at the end of the articles retrieved from a literature search. If an author was men-tioned often in a reference list at the end of an article or had written much on a topic, researchers would use available tools to “chase” or locate more articles from that same author. Today, most databases have some sort of an author-searching tool. For example, Ovid MEDLINE has a search aid that allows researchers to narrow down their search by authors. After performing a search, click on authors in this search aid to view a list of the top five authors who have written on a search topic.

Another resource for author searching is the Virginia Henderson Interna-tional Nursing Library. This is a database of nursing research abstracts from research studies, conference presentations, practice innovations, and evidence-based projects. It also provides contact information for the authors of the abstracts.

CITATION CHASING

Citation chasing, also referred to as pearling, is a method of searching that is best used at the beginning of your SR project. This method allows searchers to maximize the discovery of “one good article.” This is done by looking back in time at references cited in a bibliography to see which authors, publications, including journals, are reporting on a topic of interest (Bates, 1989). Researchers can look forward in time, discovering articles that have cited an initial article. Initial articles that are “highly cited” may be significant to the SR team’s research.

To find additional articles that have mentioned that “one good or highly cited article,” you can enter the title or author of the initial article in various resources, including Scopus, WoS, Google Scholar, CINAHL, and PsycINFO.

Researchers need some details to maximize their search yield from a citation-chasing search in specific search resources and databases. For instance, more than one source has suggested that if an article was published after 1995, performing a search in Scopus is sufficient, because Scopus includes MED-LINE, EMBASE, as well as additional content, making it a resource with more content than WoS, CINAHL, or PsycINFO. For searches of articles pub-lished before 1995, a search of both WoS and Scopus is recommended (Falagas, Pitsouni, Malietzis, & Pappas, 2008). Google Scholar is a useful source for locating books, journal articles, web pages, and other web ephemera, but the






resource has limited quality control, and its producers offer little information or transparency on how search results are processed compared with the infor-mation one receives when searching traditional bibliographic databases such as CINAHL and MEDLINE. Google Scholar should not be the sole source used by a researcher for citation chasing (Jacsó, 2009).

WoS Cited Reference SearchingWoS is a search resource consisting of seven databases. It is possible to do a single search in all the databases that your institution subscribes to at the same time. However, when you select Cited Reference Search from the main page, you will be searching the core collection of WoS, which consists of SCI Expanded, SSCI, and A&HCI. You should uncheck the last database before searching unless you have a good reason for wanting Arts and Humanities articles. SCI Expanded includes content from about 8,060 major journals across 150 disciplines, dating back to 1900, and SSCI covers 697 journals across 50 social science disciplines, as well as items from 3,500 of the world’s leading scientific and technical jour-nals, from 1900 to the present.

The cited reference search page on WoS includes a link to brief recorded tuto-rials (wokinfo.com/training_support/training/web-of-science/#re cor dedtrain ing) on cited reference searching. Use WoS to search for a cited author or a cited work (article). When using WoS cited reference searches, look for a cited author, a cited work, and/or cited year(s) for a work. Only the first cited author of a publication is indexed and will be displayed with the reference or citation. Sec-ondary authors are not indexed but are searchable. Therefore, if Ben Johnston is the first or primary author for an article, look for cited references to the article by Ben Johnston et al., by typing his last name in the cited author search box in WoS.

Once articles are located by citing an original author’s work, search for more information on those articles, and see those articles’ reference lists and other works that cite the authors who had cited the original author. Sometimes, a researcher discovers a cited reference and it does not offer a clickable hyper-link, just plain black text. In that case, the item a searcher is interested in may be a reference to a book or another document not indexed in WoS, an article that is not a part of their institution’s journal subscription, or the item of inter-est was not cited correctly in a publication.

Scopus Cited Reference SearchingScopus is a relatively new database that includes peer-reviewed citations from biomedical and social sciences literature. Scopus has fully indexed or catego-rized article citations, including their references. This makes it possible to browse the cited references for Scopus article citations. To access cited refer-ences, use the appropriate link within a Scopus record. Along with accessing a key cited reference is the option to look for additional documents related to the author(s) and/or key words from that reference.






Google Scholar’s Cited by Number (#)Google Scholar is an Internet resource from Google that boasts inclusion of articles, theses, books, abstracts, court opinions, repositories, and other schol-arly literature from various research areas. One thing that searchers may have noticed when searching Google Scholar is the possible inclusion of multiple versions of an article in a search results list. There may be several reasons for this, including the possibility of a different version of an article. Each article that appears in Google Scholar has a link called cited by #. Clicking on the link cited by # reveals other articles that Google Scholar has determined to contain some version of the source article information. Choose Google Scholar from the Google toolbar under more or type www.scholar.google.com directly into the computer browser.

Once in Scholar, enter information on an author and/or title of a source article. Use the available option to pick articles excluding patents and then click the search button. Google Scholar also has a set of preferences (usually under Settings) that can be selected before researchers start searching it. It allows for selection of language(s) for search results, adding libraries that a researcher may be affiliated with to facilitate full-text access of retrieved search records, and a way to set up parameters for exporting references to a citation management product (e.g., EndNote, RefWorks) after a search.

If using advanced Scholar search, consider limiting search results by date, journal, and author. When typing search terms in the advanced Scholar search box, keep in mind the way Google processes searches. For instance, if a researcher is used to typing asthma AND therapy in a search box, note that Google auto-matically places AND in between words when more than one word is typed in a search box. Therefore, a search for chronic asthma is processed by Google search engines as chronic AND asthma. Depending on what an investigator is looking for, the way Google processes this search may be either a good or a bad thing. Similar to a search performed in Google.com, searches on Google prod-ucts (e.g., Google.com, Google Scholar) can be performed by using advanced operators or search connector words or symbols such as those in the examples given next. Visit support.google.com/websearch/answer/ 134 479?hl=en for “How to Search on Google.”

■■ Minus sign (−) operator excludes all results that include a particular search term, as in (flowers –author: flowers).

■■ Phrase search only returns results that include this exact phrase, for exam-ple, “as you like it.”

■■ OR operator returns results that include several search terms, for example, telenursing OR telehealth.

■■ Intitle: operator returns results that include search terms in the document’s title, for example, intitle: “barcode medication verification.”

http://www.scholar.google.com






EvidenceUpdates

EvidenceUpdates is a collaboration of British Medical Journal (BMJ) and McMaster University’s Health Information Research Unit (HIRU). This resource at plus.mcmaster.ca/EvidenceUpdates may be useful as a part of pearl-growing searches or searches aimed at generating ideas for SRs. EvidenceUpdates can also be used at later stages in the SR search process to locate specific evidence on a foreground research question. Complete an online registration to search the EvidenceUpdates database.

HAND SEARCHING

Hand searching is a method of physically searching the entire contents of a jour-nal or journals that are considered relevant to a research topic. This was more common in the pre-digital age. However, it is still a useful searching technique today, because sometimes studies are missed during the electronic searching process due to incorrect, incomplete, or missed indexing. The area where hand searching is the most useful is when reviewers are looking at journal supple-ments and special issues of journals, because these are not always indexed in the databases. Searching the bibliography of a useful article is a type of hand searching called snowballing. The Cochrane Collaboration has a manual for hand searching available at tinyurl.com/nq7glaz; see also www.cochrane .org/training/handsearchers-tscs.

■■ Summary

In this chapter, we reviewed many search resources, traditional search databases such as MEDLINE and CINAHL, nontraditional resources such as grey litera-ture, and Internet resources. Search tips and search techniques to maximize search time in each resource were explored. The best practices for conducting SR searches include combining controlled vocabulary terms with plain language text words, using Boolean logic and proximity operators, and maintaining a thorough record of your final search strategy in a separate document such that it can be repeated by researchers at a later date to test the methods or to con-duct an update. Reviewers should also keep track of all search histories and research notes by using a worksheet (in conjunction with saving actual search steps within each database). See Exhibit 6.1 for a Comprehensive SR Search Strategies Worksheet.

■■ Practice Activity

1. Complete Exhibit 6.1 (Comprehensive Systematic Review Search Strategies Worksheet) and conduct a search of the databases listed. Be sure to keep accurate records of your search findings.






EXHIBIT 6.1 COMPREHENSIVE SYSTEMATIC REVIEW SEARCH STRATEGIES WORKSHEET

Foreground Search Question (Use the most appropriate question framework [others available besides PICO and SPICE ] to diagram your research question.)

P =I =C =O =

S =P =I =C =E =

Harvesting Search Text Words, Controlled Vocabulary Terms (e.g., Subject Headings)

Text Words:

Subject Headings:

Search Inclusion Criteria (informs natural language terms or text word selections and ways to limit search results after searching)

Equivalent Database Search Limits:

FILTERED EVIDENCE DATABASES (As applicable, include dates represented by search resources)

Date(s) Searches Performed

Search Text Words and Phrases

Name(s) for Saved Searches and Alerts

Research Notes

Campbell Collaboration Library of Systematic Reviews

Cochrane Database of Systematic Reviews

Clinical Evidence Electronic Textbook

Dynamed Useful for background questions

Essential Evidence Plus

JBI Evidence-Based Database

EBSCO’s Nursing Reference Center Evidence-Based Summaries

TRIP Database

(continued)






EXHIBIT 6.1 COMPREHENSIVE SYSTEMATIC REVIEW SEARCH STRATEGIES WORKSHEET (continued )

GUIDELINES (As applicable, include dates represented by search resources)




Research Notes

National Guideline Clearinghouse

National Institute for Health and Clinical Excellence

UNFIILTERED EVIDENCE DATABASES (As applicable, include dates represented by search resources)




Research Notes

CINAHL

Current Contents

EMBASE

Health Source: Nursing/Academic Edition

HealthSTAR

MEDLINE: 1946 to November (Week 3) 2015

Search A: December 2, 2015

Search B:

Searched through the database search retrieval system: Ovid

ProQuest Dissertations and Theses

ProQuest Nursing and Allied Health Source

PsycINFO Includes citations from PsycARTICLES

PubMed

Scopus

Web of Science

OTHER SEARCH RESOURCES (As applicable, include dates represented by search resources)




Research Notes

My university’s dissertations and theses repository:

Listing of Complete Search Strategies per Search Resource: JBI, Joanna Briggs Institute.CINAHL, Cumulative Index to Nursing and Allied Health Literature; PICO, patient, intervention, comparison outcome; SPICE, Setting, Perspective, Intervention, Comparison, Evaluation; TRIP, Turning Research into Practice.






■■ Suggested Reading

Balshem, H., Stevens, A., Ansari, M., Norris, S., Kansagara, D., Shamliyan, T., . . . Dickersin, K. (2008). AHRQ methods for effective health care. Finding grey literature evidence and assessing for outcome and analysis reporting biases when comparing medical interventions: AHRQ and the effective health care program. In Methods guide for effectiveness and comparative effectiveness reviews (Contract Nos. 290-2007-10057-I and 290-10059-I; AHRQ Publication No. 13(14)-EHC096-EF). Rockville, MD: Agency for Healthcare Research and Quality (US).

Chou, R. (2008). Using evidence in pain practice: Part I: Assessing quality of systematic reviews and clinical practice guidelines. Pain Medicine, 9(5), 518–530.

Egger, M., Juni, P., Bartlett, C., Holenstein, F., & Sterne, J. (2003). How important are com-prehensive literature searches and the assessment of trial quality in systematic reviews? Empirical study. Health Technology Assessment, 7(1), 1–76.

Goodfellow, L. M. (2009). Electronic theses and dissertations: A review of this valuable resource for nurse scholars worldwide. International Nursing Review, 56(2), 159–165.

GRADE Working Group. (2005–2009). GRADE Working Group. Retrieved from http://www.gradeworkinggroup.org/index.htm

Haynes, B. (2007). Of studies, syntheses, synopses, summaries, and systems: The “5S” evo-lution of information services for evidence-based healthcare decisions. Evidence-Based Nursing, 10(1), 6–7. doi:10.1136/ebn.10.1.6

Khan, K. S., Daya, S., & Jadad, A. (1996). The importance of quality of primary studies in producing unbiased systematic reviews. Archives of Internal Medicine, 156(6), 661–666.

Laupacis, A. (1997). Methodological studies of systematic reviews: Is there publication bias? Archives of Internal Medicine, 157(3), 357–358.

Martin, J. L., Pérez, V., Sacristán, M., & Alvarez, E. (2005). Is grey literature essential for a better control of publication bias in psychiatry? An example from three meta-analyses of schizophrenia. European Psychiatry, 20(8), 550–553. doi:10.1016/j.eurpsy .2005 .03.011

McGowan, J., & Sampson, M. (2005). Systematic reviews need systematic searchers. Journal of the Medical Library Association, 93(1), 74–80.

McKibbon, A., & Wilczynski, N. (2009). PDQ: Pretty Darned Quick. Evidence-based prin-ciples and practice (2nd ed.). Shelton, CT: People’s Medical Publishing House.

■■ References

Bassler, D., Ferreira-Gonzalez, I., Briel, M., Cook, D. J., Devereaux, P. J., Heels-Ansdell, D., . . . Guyatt, G. H. (2007). Systematic reviewers neglect bias that results from trials stopped early for benefit. Journal of Clinical Epidemiology, 60(9), 869–873. doi:10 .1016/j.jclin epi.2006.12.006

Bates, M. J. (1989). The design of browsing and berrypicking techniques for the online search interface. Online Review, 13(5), 407–424. Retrieved from https://pages.gseis.ucla .edu/faculty/bates/berrypicking.html

Benzies, K. M., Premji, S., Hayden, K. A., & Serrett, K. (2006). State-of-the-evidence reviews: Advantages and challenges of including grey literature. Worldviews on Evidence Based Nursing, 3(2), 55–61. doi:10.1111/j.1741-6787.2006.00051.x

Blackhall, K. (2007). Finding studies for inclusion in systematic reviews of interventions for injury prevention the importance of grey and unpublished literature. Injury Pre-vention, 13(5), 359. doi:10.1136/ip.2007.017020

https://pages.gseis.ucla.edu/faculty/bates/berrypicking.html

https://pages.gseis.ucla.edu/faculty/bates/berrypicking.html






Centre for Reviews and Dissemination. (2010). Help section: About DARE. Retrieved from http://community-archive.cochrane.org/editorial-and-publishing-policy-resource/database-abstracts-reviews-effects-dare

Falagas, M. E., Pitsouni, E. I., Malietzis, G. A., & Pappas, G. (2008). Comparison of PubMed, Scopus, Web of Science, and Google Scholar: Strengths and weaknesses. Federation of American Societies for Experimental Biology Journal, 22(2), 338–342. doi:10.1096/fj.07-9492LSF

Gavel, Y., & Iselid, L. (2008). Web of Science and Scopus: A journal title overlap study. Online Information Review, 32(1), 8–21. doi:10.1108/14684520810865958

Goodman, S. N. (2008). Systematic reviews are not biased by results from trials stopped early for benefit. Journal of Clinical Epidemiology, 61(1), 95–96. doi:10.1016/j.jc linepi.2007.06.012

GreyNet International. (2009). Grey literature network service. About GreyNet. Retrieved from http://www.greynet.org

Honest, H., & Khan, K. S. (2002). Reporting of measures of accuracy in systematic reviews of diagnostic literature. BioMed Central Health Services Research, 2(1), 4.

Hopewell, S., McDonald, S., Clarke, M., & Egger, M. (2007). Grey literature in meta- analyses of randomized trials of health care interventions. Cochrane Database Systematic Reviews (2), MR000010. doi:10.1002/14651858.MR000010.pub3

Institute of Medicine. (2011). Standards for systematic review. Washington, DC: Author. Retrieved from https://iom.nationalacademies.org/Reports/2011/Finding-What -Works-in-Health-Care-Standards-for-Systematic-Reviews/Standards.aspx?page=2

Jacsó, P. (2009). The h-index for countries in web of science and scopus. Online Information Review, 33(4), 831–837. doi:10.1108/14684520910985756

Joanna Briggs Institute. (2014). About us. Retrieved from http://joannabriggs.org/about .html

Lundh, A., & Gøtzsche, P. C. (2008). Recommendations by Cochrane Review Groups for assessment of the risk of bias in studies. BioMed Central Medical Research Methodol-ogy, 8, 22.

Mathieu, S., Boutron, I., Moher, D., Altman, D. G., & Ravaud, P. (2009). Comparison of registered and published primary outcomes in randomized controlled trials. Journal of the American Medical Association, 302(9), 977–984. doi:10.1001/jama.2009.1242

McAuley, L., Pham, B., Tugwell, P., & Moher, D. (2000). Does the inclusion of grey literature influence estimates of intervention effectiveness reported in meta-analyses? Lan-cet, 356(9237), 1228–1231. doi:10.1016/S0140-6736(00)02786-0

New York Academy of Medicine Library. (2009). Grey literature report. Retrieved from http://www.nyam.org

Ross, J. S., Mulvey, G. K., Hines, E. M., Nissen, S. E., & Krumholz, H. M. (2009). Trial pub-lication after registration in ClinicalTrials.Gov: A cross-sectional analysis. PLoS Medicine, 6(9), e1000144. doi:10.1371/journal.pmed.1000144

Sterne, J. A., Egger, M., & Smith, G. D. (2001). Systematic reviews in health care: Investi-gating and dealing with publication and other biases in meta-analysis. British Medical Journal, 323(7304), 101–105. Retrieved from http://www.nyam.org

The Evolution of Trial Registration. (2009). Editorial. Nature Reviews Drug Discovery, 8, 755. Retrieved from http://www.nature.com/nrd/journal/v8/n10/full/nrd3021.html

Trowman, R., Dumville, J. C., Torgerson, D. J., & Cranny, G. (2007). The impact of trial baseline imbalances should be considered in systematic reviews: A methodological case study. Journal of Clinical Epidemiology, 60(12), 1229–1233. doi:10.1016/j .jclinepi .2007.03.014

http://community-archive.cochrane.org/editorial-and-publishing-policy-resource/database-abstracts-reviews-effects-dare

http://www.greynet.org

https://iom.nationalacademies.org/Reports/2011/Finding-What-Works-in-Health-Care-Standards-for-Systematic-Reviews/Standards.aspx?page=2

http://joannabriggs.org/about.html

http://www.nyam.org

http://www.nyam.org

http://www.nature.com/nrd/journal/v8/n10/full/nrd3021.html

http://community-archive.cochrane.org/editorial-and-publishing-policy-resource/database-abstracts-reviews-effects-dare

https://iom.nationalacademies.org/Reports/2011/Finding-What-Works-in-Health-Care-Standards-for-Systematic-Reviews/Standards.aspx?page=2

http://joannabriggs.org/about.html

http://ClinicalTrials.Gov





9 Systematic Review of Observational and Descriptive Evidence

Cheryl Holly

OBJECTIVESAt the end of this chapter, the reader will be able to:

O Describe the research designs used for observational systematic review (SR) O Compare and contrast the benefits of observational and interventional

meta-analysis O Explain the various types of bias and confounding variables found in

observational research studies O Identify topics amenable to SR by using observational studies

CHAPTER HIGHLIGHTS O Observation is about collecting information from a primary source. O In an observational study, a researcher simply notes behavior without

influencing or interfering with the behavior. O The evidence available from randomized controlled trials (RCTs) will not

answer all clinical questions. O Due to differences in patient groups and settings used in observational

studies, heterogeneity may present a challenge. O The use of observational data is almost always necessary to assess harm

adequately and may provide the only data for assessing either benefit or harm in minority or vulnerable populations.

O SRs of observational data will always have confounding variables and bias, as the interventions studied are not deliberately chosen and randomized.

An observation is a factual description of a phenomenon. Observations can be either qualitative or quantitative. Qualitative observations use the senses (sight, smell, touch, taste, and hearing) to observe. Quantitative observations are made with instruments such as rulers, laboratory tests, or thermometers, and they are measurable. According to Dahnke (2011), observation is a focused, intentional





226 IV. Methods for Systemat ic Rev iews

recognition; part of a formalized effort to acquire knowledge. An SR of observa-tional data meets this characterization, with the ultimate goal of using acquired knowledge to determine best practice. An SR of observational data is conducted for essentially the same reason as a meta-analysis of interventions: to create esti-mates of performance based on all available and relevant evidence and to account for variations of findings both between and among studies (Deeks, 2007).

■■ Observational Versus Interventional Studies

Although the RCT is considered the gold standard and principal research design for evaluation of interventions, procedures, and services, there are many aspects of care that cannot be tested or evaluated in a randomized fashion. Apart from this, RCTs are often time and labor intensive, as well as expensive.

Consider, for example, a study that found that nursing students who ran 1 mile each morning and then ate breakfast had a higher grade point average (GPA) than those who did not run (3.72 vs. 2.98) but did eat breakfast. Among the stu-dents who ran 1 mile each morning, 22% had all As compared with only 3% of the non-running students. This is a retrospective observational study. Existing data were collected and compared; however, they cannot prove that running improves grades.

In comparison, an experimental study would assign students at random to one of two groups. One would include coaching and training on running, whereas the other would not. Both groups would eat breakfast. The running group (experimental group) would be required to run at least 1 mile each day. After a period, the GPAs of both groups would be compared. This study would be a prospective one.

Often, the RCT will confirm what has been found in previous observational studies but findings may also differ significantly, as in the inconsistent results found with regard to cardiac risk in studies on the drug Avandia (rosiglitazone). Avandia is used for those with type 2 diabetes to become more sensitive to insu-lin, helping them control and maintain safe glucose levels. Avandia was severely restricted by the Food and Drug Administration (FDA) in 2010 (the restric-tions were lifted in 2013) after it was linked to thousands of strokes and heart attacks. The drug company clinical trials, however, showed that the drug pro-duced only mild side effects, such as upper respiratory tract infection, head-ache, back pain, hyperglycemia, fatigue, and sinusitis (Harris, 2010; www.drug watch.com/avandia). Contradictory findings such as these lend themselves to an SR, which pools all available evidence to determine effect size. Although it can be called a meta-analysis due to this pooling of data across studies, the results of an observational meta-analysis are less precise than those of a meta-analysis of interventions.

Much health care research is conducted by using observational designs. Included among these are epidemiological studies on risk factors and prog-nostic factors for the prediction of future events, studies of etiology for an

http://www.drugwatch.com/avandia

http://www.drugwatch.com/avandia





9. Systemat ic Rev iew of Observat ional and Descr ipt ive Ev idence 227

understanding of associations, or studies of diagnostic and screening accuracy. Observational studies are designed to understand the many different conditions and exposures experienced by people that are not limited to treatment or inter-vention. In an observational study, the investigator takes a much less active role than in an RCT. For example, questions about the consequences of choice such as Does eating a diet high in unsaturated fat cause breast cancer? cannot be tested in a randomized experiment (Egger, Smith, & Schneider, 2001).

Observational studies are designed to:

1. Ask a question about a naturally occurring phenomenon 2. Make observations about that phenomenon 3. Construct an explanation for the phenomenon (a hypothesis) 4. Test the hypothesis 5. Formulate a conclusion based on the data gathered (inference)

For example, to study the clinical profile and response to treatment of chil-dren and adolescents referred to a psychiatric service, Grover et al. (2012) used an observational, retrospective medical record review. The subjects must have been diagnosed as having acute delirium by using International Classification of Diseases-10 (ICD-10) during their inpatient stay. By reviewing existing data from the medical record, they found that the most common underlying pathol-ogy for acute delirium was infection of various types, followed by neoplasms. All subjects exhibited sleep–wake cycle disturbance and impaired orientation. Most of the patients were treated with low-dose haloperidol, and they responded well. Thus, it can be concluded that sleep–wake cycle disturbances and cogni-tive dysfunction are common in children and adolescents with delirium.

For the purposes of this chapter, observational studies are defined as those in which the investigators do not assign an intervention. These studies include descriptive observational studies (prevalence surveys, case series, case reports) and analytical observational studies (cohort, case–control, cross-sectional), studies of diagnostic and prognostic test accuracy, and studies of etiology for causes and factors associated with a particular condition. The investigator’s role in these studies is to observe, record, and analyze results and to make inferences based on the findings. See Table 9.1 for a depiction of these approaches.

SIGNIFICANCE IN OBSERVATIONAL STUDIES

Significance in research terms has nothing to do with the importance or use-fulness of findings. Significance refers to the probability that a relationship could be caused by chance; in other words, it is probably true or not true, but it is not necessarily important. Statistical tests allow an estimate to be made of the prob-ability of the observed degree of association between variables, and from this the statistical significance can be expressed, commonly in terms of the p value, where p is understood to mean probability. In calculating a p value, it is assumed that there really is no difference. A p value of less than .05 is considered small,





228

TABLE 9.1 RESEARCH DESIGNS FOR OBSERVATIONAL STUDIES

Study Design Description Data Collection Advantages Disadvantages

Prospective cohort study

A longitudinal study of a group of people who share a common characteristic, such as year of high school graduation, taking the same drug

Medical records

Environmental or lifestyle questionnaires

Physiological measurement

Interview data

Allows investigation of rare diseases or diseases with a long latency

Can measure risk

Large cohorts can minimize selection bias

Aids in understanding causal associations

Can be hypothesis testing

Subjects are not randomly assigned

High attrition rates

Expensive to conduct

Long time to obtain useful data for analysis

Retrospective cohort studies

A historical study of a group of people who share a common characteristic, such as smoking and heart disease or working in a coal mine

Medical records Less expensive and faster than a prospective cohort study


Lack of follow-up affects study validity.

Case–control studies

A study of “cases” who have a condition or exposure in comparison with a “control,” cases who do not have the condition or exposure

Medical records

Registry data

Death certificates

Population surveys

Allows understanding of risk factors

Inexpensive

Few personnel needed

Useful for studying outcomes that take a long time to develop, for example, cancer


There is a better recall of information among exposed groups (recall bias)

Odds ratios are used as a proxy measure for relative risk

Case series/case report

A tracking study of people given a similar treatment, such as the use of an automated external defibrillator for out-of-hospital cardiac arrest and survival

Medical records

Detailed patient reports

Adverse events and side effects are identified.

Can be hypothesis generating

Cross-sectional studies

A descriptive study of a group of people at one point in time; a “snapshot”

Also called a prevalence study

Survey methods

Secondary data

Provides a foundation for the stronger cohort or RCT designs

Can be hypothesis generating

No control group

Difficult to determine when the outcome of interest began, as everything is measured at one time

Descriptive studies

Descriptive studies identify descriptive characteristics; they are an important first step in the search for determinants or risk factors that can be altered or eliminated to reduce or prevent disease

Hypothesis generating

Can be useful in establishing trends or patterns

RCT, randomized controlled trial.






that is, there is a less than one in 20, or five in 100, chance of something hap-pening. When p values are this small, the results of the study are interpreted to be significant (unlikely to have resulted through chance). Even smaller p values at the level of .01 are called “highly significant”; p values this small indicate that the results could occur less than once in 100 times (Davies & Crombie, 2003).

Confidence intervals (CIs) can also be used to determine the significance of study results. A CI is the range of values that encompass a population param-eter. Since it is impossible to measure a total population, the study sample is used to calculate a range within which the population value will most likely fall. This is known as the 95% CI. The values at either end of the range are known as the confidence limits—the wider the CI, the less the accuracy, that is, there is less certainty that the strength of the association has been precisely estimated (Davies & Crombie, 2003).

The two definitions of statistical significance are compatible. If you have a p value of less than .05, it is the same as getting a 95% CI that does not overlap zero. On the contrary, a p value greater than .05 equates to a 95% CI that includes 1 (the null value), meaning that there is no effect (Davies & Crombie, 2003).

Significance of study results or lack thereof, however, is less a concern in SR than in primary research. SRs are concerned with the combined or pooled results of individual primary studies. The process of SRs involves searching for all available and relevant studies regardless of significance. Inclusion of studies that have significant or insignificant results changes the focus from statistical significance to the direction and magnitude of the result, which is important to the development of best practice recommendations. In an observational SR, more information may be gained from an in-depth review of identified variance than from statistical computation (Egger, Smith, & Schneider, 2001).

■■ Descriptive and Analytic Observational Studies

An observational study is one in which the researcher does not intervene. Observational studies can be either descriptive or analytic. Descriptive studies include case series/case reports and prevalence surveys, whereas analytic stud-ies include cohort studies, case–control studies, and cross-sectional studies. Studies of diagnostic accuracy and tests used for prognosis are also considered observational studies. Recently, SR methodology has been expanded to include a review of associations, or etiology, to better understand the burden of disease (Moola et al., 2015). A study of associations examines the relationship between an exposure and an outcome.

Descriptive studies describe the occurrence of an outcome. Analytic stud-ies describe associations between an exposure and an outcome. In other words, descriptive studies examine the frequency to which diseases occur, whereas analytic studies evaluate the relationship between disease and different expo-sures. Descriptive studies are useful for identifying hypotheses to test in analytic studies. Case–control studies are then usually applied to evaluate whether the






identified hypothesis is related to the outcome of interest. Subsequently, cohort or longitudinal studies can be used to further describe the occurrence for the development of the outcome.

The most commonly used observational study is the descriptive study. Descriptive studies are observational studies that describe patterns related to person, place, and time. Often, a descriptive study is the first step into research in a new area. Descriptive studies simply describe, explain, or explore a topic; they do not establish cause and effect. Descriptive studies are valuable, as they can highlight associations between variables or between or among variables. Descriptive studies do not have a comparison (control) group, which means that they do not allow for inferences to be drawn about associations (Heffernan, 2010). For example, Aznar et al. (2011) studied patterns of physical activity in Spanish children. Their findings indicated that few children achieved the exer-cise levels recommended for health; adolescent girls were at particular risk. They concluded that more effort needs to be devoted to promoting appropriate phys-ical activity opportunities for Spanish girls.

Descriptive study designs include case reports, case series, incidence stud-ies, and ecologic studies. The case report is the most elementary study design. It generally describes an occurrence to one or two individuals that has been identified in a medical setting. A case report is an anecdotal yet detailed report of the symptoms, signs, diagnosis, treatment, and follow-up of an individual patient (National Cancer Institute Dictionary [NCI], 2010). Typically, case reports are constructed to describe an unexpected observation or association or an unexpected event or to share findings on rare occurrences or adverse effects (NCI, 2010). For example, Fraser, Estabrooks, Allen, and Strang (2010) provided a case example to illustrate how the case manager balances and weighs the factors that influence his or her resource allocation decisions and discusses home care case manager resource allocation decisions.

The case series design is an extension of the case report. In a case series, a number of events are described. These events usually have been observed over a set period (such as 1 year) and are identified from one reporting source (e.g., a hospital). A case series (also known as a clinical series) tracks patients with a known exposure who are given similar treatment or reviews their medical records for exposure and outcome. Thus, this study can be retrospective or pro-spective and usually involves a smaller number of patients than case–control studies (Heffernan, 2010). For example, Shellmer et al. (2011) reported on patients who completed cognitive and adaptive functioning testing pre- and 1 and 3 years post-liver transplantation for maple syrup urine disease, which is associated with central nervous system damage, developmental delays, and neurocogni-tive deficits. Findings showed either no significant change (N = 8) or improve-ment (N = 5) in IQ scores pre- to post-liver transplantation.

Analytic observational studies are designed to address a specific research hypothesis, usually about the effect of a particular exposure on a particular outcome (e.g., the effect of low-dose aspirin on cardiovascular disease). In con-trast, descriptive studies measure and report data without addressing a specific






hypo thesis. Analytic studies use either group or individual data, except for eco-logic studies in which only group data are used. An ecologic study is an exami-nation of the effects of risk-modifying factors on health or other outcomes in a geographically or temporally defined population. Grant (2014), for example, studied dietary patterns while looking for risk factors for cancer in multiple countries. He found that animal products (meat, milk, fish, and eggs), sugar, and some fats appear to be risk factors for cancer, whereas cereals/grains and veg-etables appear to reduce the risk of cancer.

Individual data are used in cohort, case–control, and cross-sectional stud-ies. A cohort study is a longitudinal observational study, the focus of which is a clearly defined exposure or risk with a control group. The word cohort is used to mean a group of individuals followed forward over a period to examine an outcome of interest and to map its long-term effects. The study starts with cohorts of well individuals, and it follows them until events occur. Cohort stud-ies follow two or more groups from exposure to outcome, and they then com-pare the experience of the exposed group with that of the nonexposed group. If the exposed group has a higher or lower rate on the outcome of study interest, then an association between exposure and outcome is suggested (Heffernan, 2010). For example, Chan et al. (2010) followed a group of adults who experi-enced in-hospital cardiac arrest to determine their survival rate to discharge after use of an automated external defibrillator (AED). Results indicated that for cardiac arrests due to shockable rhythms, AED use was not associated with sur-vival (adjusted relative risk [RR], 1.00; 95% CI [0.88, 1.13]; p = .99).

Cohort studies can be prospective (future) or retrospective (past). Pro-spective design allows exposure to risk factors to be assessed directly over time. A retrospective design is effective for diseases with a long development time. Retrospective cohort studies should not be confused with case–control studies, which are also retrospective. Cohort studies track people forward in time from exposure to outcome, whereas case–control studies trace them back-ward from outcome to exposure. Risk in cohort studies is measured by RR, absolute risk, and attributable risk (see Table 9.2; Davies & Crombie, 2003; Heffernan, 2010).

A case–control study is an observational study with a control group. A par-ticular disease or condition, such as food poisoning, is the starting point—the

TABLE 9.2 MEASURES FOR COHORT STUDIES

Measure Description

RR The likelihood or risk that those exposed to a condition will get the disease as opposed to the nonexposed group

AR The incidence rate of exposure

ATR The difference in incidence between the exposed and nonexposed groups

AR, absolute risk; ATR, attributable risk; RR, relative risk.






case—and the study examines backward (retrospectively) to try and determine a cause and any associated risk factors. The cases are known and they exhibit symptoms of the condition under study; for example, the diarrhea that might be expected from a case of food poisoning. Controls must be similar to the cases, except that that they do not have the outcome in question, that is, no diarrhea. Results of a case–control study are reported by using an odds ratio (OR): the ratio of exposed to nonexposed in the case group divided by the same ratio in the control group. For example, Suarez et al. (2011), using cases and controls from the large, multistate, population-based National Birth Defects Preven-tion Study, examined the relationship between neural tube defects and mater-nal exposures to cigarette smoke, including passive smoke exposure. Results indicated that compared with nonsmokers (and no exposure to passive smok-ing), mothers exposed only to passive smoke had an increased neural tube defect OR (1.7; 95% CI [1.4, 2.0]), adjusted for race, ethnicity, and study center, suggesting that maternal exposure to passive smoking increases the incidence of neural tube defects in infants.

A cross-sectional study is a descriptive study that gathers information on both the condition and the exposure at the same time. Cross-sectional studies provide a “snapshot” of one point in time. This type of data can be used to assess the prevalence of acute or chronic conditions in a population. A cross-sectional study is sometimes called a prevalence study (Heffernan, 2010). For example, Chen, Liu, Liu, and Tsai (2010) examined the level of awareness of hyperten-sion guidelines and associated factors among nurses in 10 hospitals in Taiwan by using survey methods. Among the seven dimensions of the Hypertension Management Questionnaire, the definition of hypertension, methods for blood pressure measurements, and impact of high blood pressure on cardiovascular disease had the lowest rates of correct answers. Multiple regression analysis revealed that the nurse’s clinical experience, educational level, work setting, in-service education training on hypertension, and level of the hospital (R = 35.4%, F = 52.89, p < .001) independently predicted the nurse’s level of awareness. The authors concluded that a large proportion of the nurses in northern Taiwan had insufficient knowledge of the hypertension guidelines.

■■ Studies of Diagnostic Accuracy

Studies of diagnostic accuracy are conducted to determine the exactness of tests used to diagnose conditions. Without an accurate diagnosis, treatment may be ineffective and wasteful or even harmful. An accurate diagnosis is essential to effective treatment. Accuracy is plotted against a standard reference test. A diagnostic test is any test that is used in making a diagnosis based on present-ing signs and symptoms or monitoring the progression of a disease or condition (Deeks, 2007). Diagnostic tests include blood tests, urine tests, psychological evaluations, or imaging studies, among others. However, not only diagnostic






tests are used in health care. There are many educational tests used to diagnose learning disabilities, intellectual delays, and learning styles. The results of a diag-nostic test are compared with the results of a reference standard in the same person or patient. The reference standard can be a test, a series of tests, a com-bination of different tests, or the clinical follow-up of patients used to determine the presence or absence of the target condition in patients. For example, a urine culture and sensitivity is the reference standard for a urine dipstick and a chest x-ray is the standard reference for tuberculin testing. Diagnostic test accu-racy studies differ from prognostic studies in that prognostic accuracy stud-ies test information used to identify patients who may experience a future event, such as disease recurrence or sudden death (Bossuyt & Leeflang, 2008). “What’s the prognosis?” is a commonly asked question in health care. Although evidence of accuracy is essential for a diagnostic test to be implemented in clinical practice, evidence of effectiveness, safety, cost, and reliability are also needed to fully inform clinical practice and ensure optimal patient outcomes (Campbell, 2015).

A good diagnostic test is defined by its sensitivity and specificity. Sensitiv-ity (positive results) and specificity (negative results) results are not static. They can vary based on the population or setting. According to the Joanna Briggs Institute (JBI, 2015) the fact that an ideal range of normal values does not exist for human beings. A diagnostic test can have one of four outcomes: true positive (TP), true negative (TN), false positive (FP), or false negative (FN). Sensitivity and specificity can be calculated by using the formula TP/(TP + FN). Specific-ity can be calculated by using TN/(TN + FP) (see Table 9.3). There is no defin-itive answer on which of these is the most important, as the possible harm caused by an FP or FN result varies based on the patient’s presentation and the treatment chosen. “This highlights the necessity of systematic reviews that syn-thesize the best available evidence to assess the accuracy of diagnostic tests and provide high quality evidence for this important area of clinical practice” (Campbell, 2015).

These measures along with a likelihood ratio (LR) and a receiver operating characteristic (ROC) curve are used to determine the accuracy of a diagnostic test. Studies that test the accuracy of a diagnostic test using these measures are suitable for an SR. LR is a ratio of the probability that a particular result will occur in a patient with the disease compared with the probability that someone

TABLE 9.3 SENSITIVITY AND SPECIFICITY

Disease Present Disease Absent

Positive test result TP FP

Negative test result FN TN

FN, false negative; FP, false positive; TN, true negative; TP, true positive.






without the disease would have the same result. It is calculated by using the sen-sitivity and specificity results as follows (JBI, 2009):

Sensitivity/(1-specificity) is a positive LR(1-Sensitivity)/specificity is a negative LR

The ROC curves are plots of the sensitivity and specificity data. These curves are used to determine the overall diagnostic accuracy of a test. An inaccurate test will have a line close to the rising diagonal. In a perfect test, the line will rise steeply and pass close to the top of the left-hand corner. Sensitivity is plotted on the vertical axis, and 1-specificity is plotted on the horizontal axis. Once a curve has been developed, the area between the curve and the x axis can be determined and is called the area under the curve (AUC). This represents the accuracy of the diagnostic test. An AUC of 1 represents a perfect test, and an AUC of 0.5 represents a useless test. A quick way to determine the utility of the AUC results is the traditional school grading system, where 0.9 to 1 is excellent (an A), 0.8 to 0.9 is good (a B), and so on until 0.6 to 0.5, which is failing (an F) (JBI, 2009). For further information on the use of ROC curves and AUC in determining diagnostic accuracy, see Zou, O’Malley, and Mauri (2007).

■■ Studies of Prognosis

The primary purpose of a study of prognosis is to predict outcomes with or without establishing causality (Glasziou, Irwig, Bain, & Colditz, 2001). For example, low slung ears may be predictive of Down syndrome, but clearly, it is not a cause. According to Glasziou et al. (2001), there are two primary rea-sons to be interested in making predictions about the future status of patients. The first concerns the patient. Patients with a specific condition are interested in their future. They want to make plans and determine whether any adapta-tion to lifestyle and/or living arrangements might be necessary. The second concerns the physician. Physicians want to be able to classify patients as high or low risk for development of related symptoms and to appropriately deter-mine therapy.

SRs of prognosis are designed to answer one of the following three ques-tions:

1. What is the most likely course of this disease or condition? 2. What factors are associated with the outcome? 3. Does everyone at risk experience the same outcome?






■■ Studies of Etiology or Association

Etiology is studied via observational research that is conducted to infer corre-lations between two variables, such as a risk factor and a disease outcome. Etiological studies are important in that they provide a foundation for the for-mation of hypotheses about the risk or progress of a disease (Moola et al., 2015). These studies address two questions:

1. What is the association between an exposure and an outcome? 2. What factors are associated with the disease?

Etiological studies summarize associations between variables, but they do not make a direct inference about the etiology or the effect of a condition. Results from etiological studies, although not causal, are important and useful in generating hypotheses regarding risk and potential preventive factors in dis-ease development and progression (Moola et al., 2015).

■■ Bias and Confounding in Observational Studies

A meta-analysis of interventions is predicated on an unbiased effect of the inter-vention, with random variation held in check by randomization and the inten-tion to treat (ITT) principle. An observational study, on the other hand, can yield estimates that may prove misleading due to confounding factors, biases, or both. Eggers, Smith, and Schneider (2007) report, for example, that any review assessing the effect of coffee on myocardial infarction must adjust for smoking since smoking and drinking large amounts of coffee have been found to be asso-ciated, and smoking is a cause of coronary heart disease (CHD). However, even with adjustment for confounding factors such as these, the issue of bias always arises. Bias in an observational study is a deviation of a measurement from the truth, leading to either an under- or overestimation of significance (Glasziou et al., 2000), and it can be related to the way in which the study was designed, conducted, or interpreted. Bias may result from poor diagnosis or poor diagnos-tic criteria, poor case choice, poor choice of controls, or variations in the way that risk exposure data are collected or measured in cases and controls. The internal validity of the study is affected by any bias present (Jepsen, Johnsen, Gillma, & Sorensen, 2004).

Two potential biases that may limit the suitability of an observational study are selection bias and information bias (Jepsen et al., 2004). Table 9.4 pres-ents questions to be asked to determine whether a bias is present in an obser-vational study. Selection bias refers to differences between the sample studied and the larger population. In other words, the sample is not representative of the larger population. Selection bias is more common in case–control stud-ies than in cohort studies. Examples of selection bias include self-selection,






volunteerism, failure to follow the ITT principle, attrition, and lack of follow-up (Jepsen et al., 2004).

Information bias can happen in a cohort design. Information bias, also known as classification or measurement bias, refers to an error in measuring the exposure or outcome. Information bias can occur in a variety of ways, includ-ing differences in the way information is gathered (in person vs. by telephone), coded, entered, or interpreted. For example, an interviewer may ask unclear questions, fail to precisely follow the research protocol, or transcribe informa-tion inaccurately. In case–control studies that rely on memory of past experi-ences, information bias is referred to as recall bias, where those in the exposed group may have better recall or better information than the healthier controls (Table 9.5; Grimes & Schultz, 2002; Jepsen et al., 2004).

A confounding is a clouding of results such that the outcomes of a study cannot be clearly determined. Confounding variables are perplexing, as the design of the study does not take these variables into account a priori, causing an incorrect interpretation of results. Bias involves errors in the measurement of a variable, whereas confounding involves errors in the interpretation of the measurement (JBI, 2009). According to Mann (2003), a confounding variable is independently associated with both the variable of interest and the outcome of interest. For example, lung cancer (outcome) is less common in people with asthma (variable). However, it is impro bable that asthma provides protection against lung cancer. It is more likely that the occurrence of lung cancer is lower in people with asthma because fewer asthmatics smoke cigarettes (confounding variable; Mann, 2003). The only way to eliminate the possibility of a confound-ing variable is via a prospective randomized controlled study. In this type of study, each type of exposure is assigned by chance and so confounding vari-ables are equally present across all groups (Mann, 2003).

TABLE 9.4 QUESTIONS TO ASK TO DETERMINE BIAS IN OBSERVATIONAL RESEARCH

1. To determine whether a selection bias is present, ask:

In a cohort study, are participants in the exposed and unexposed groups alike in all key aspects except for the exposure?

In a case–control study, are cases and controls alike in all key aspects except for the condition under study?

2. To determine whether an information bias is present, ask:

In a cohort study, is information about outcome obtained in exactly the same way for both exposed and nonexposed groups?

In a case–control study, is information about exposure collected in the same way for cases and controls?

3. To determine whether a confounding is present, ask:

Could the results be accounted for by the presence of a factor that was not considered a priori, for example, smoking, diet, activity level?

Source: Grimes and Schultz (2002).






■■ Observational SR

An observational SR is conducted when there are gaps in the RCT evidence and there are observational data available that can provide valid and useful infor-mation to answer the review question. Gaps in the RCT evidence can be iden-tified at a number of points during the review: when scoping the review, when reviewing titles and abstracts, or when appraising the results of the RCT. Apart from this, trial data may be insufficient for a number of reasons as presented in Table 9.6. Other reasons for conducting an SR of observational data include developing a more precise understanding of etiology, particularly those with rate outcome, evaluating future outcomes, and reviewing public and commu-nity health services.

An SR of observational studies is useful to quantify sources of variability in results across studies. It may also be the only available method for assessing the

TABLE 9.5 TYPES OF BIAS

Type of Bias

Explanation of Bias Crit ical Appraisal for Bias

Selection bias

■■ Definition: Results from errors in the way that research participants were selected into the study from the target population or as a result of factors that influence whether research participants remain in a study.

■■ The intervention group is different from the control/comparison group in baseline characteristics.

■■ The participants are not representative of the population of all possible participants.

■■ Nonrandom samples are at the greatest risk for selection bias.

Randomization and allocation concealment are key to minimizing selection bias.

Evaluate whether:■■ Randomization was used■■ The allocation sequence was

appropriate and adequately concealed

Performance bias

■■ Definition: Systematic differences in care provided to the participants in the intervention and control/comparison group.

■■ More likely to occur if the caregiver is aware of whether a patient is in a control or treatment group.

Was there blinding of subject?

Was there blinding of researcher/clinician?

Attrition bias

■■ Definition: Differences between control and treatment groups in terms of patients dropping out of a study or not being followed.

■■ Although dropouts will occur, the researcher wants to be assured that missing outcome data are balanced in numbers across groups with similar reasons for missing data across groups.

Was loss to follow-up (dropout, nonresponse, withdrawal, protocol deviators) reported?

Did researchers apply the concept of intention to treat?

Detection (assessor or ascertain-ment) bias

■■ Definition: Detection bias occurs if outcomes are measured differently for patients depending on whether they are in the control or treatment group.

■■ A detection bias generally occurs when the assessor (the one determining the outcome results) knows whether the subject is in the control or treatment group.

Was blinding of the assessor carried out?






efficacy and effectiveness of some interventions; for example, fluid resuscita-tion in the burn patient (Al-Shahi & Warlow, 2001).

An SR review of diagnostic evidence is conducted to draw conclusions about the accuracy of tests used to make a diagnosis. The test that is being evaluated is known as the index test; several index tests can be evaluated within the same review. The index test is evaluated against a comparator (a reference standard) for a target condition, that is, the condition of interest. The reference standard is usually the test representing the best available method of detecting the target condition, for example, a chest x-ray is considered a clinical reference test for the diagnosis of tuberculosis. It is sometimes referred to as the “gold standard” (Smidt, Deeks, & Moore, 2008).

Prognostic SRs pose some challenges for the reviewers. According to Altman (2001), in most prognostic studies, the outcome of primary interest is the time to an event, often death, making meta-analysis difficult. Second, in many studies, more than one prognostic indicator is studied and it is often difficult to isolate individual outcomes for each of the indicators. However, consideration must be given to the fact that the study of many variables simultaneously could result in development of a prognostic model for predicting the course of a disease or condition (JBI, 2009). Finally, Altman (2001) relates that many prognostic fea-tures are continuous variables and researchers use a wide variety of methods of analysis, making pooling difficult. However, despite these shortcomings, there are several reasons to conduct an SR of prognostic evidence. They are as fol-lows: to guide clinical decision making, improve understanding of the disease progress, define risk based on prognosis, and make accurate predictions about future events and outcomes (Altman, 2001).

■■ Conducting the Review

The first step in conducting an SR of observational evidence is to develop a focused clinical question. For descriptive and analytic observational review, the question formulation should follow the typical PICO parts of patient or population (P), intervention or exposure (I), comparison (C), and outcome (O); for example, What are the effects of economic and nutritional status on pre-term births?

TABLE 9.6 DANGERS INHERENT IN OVERRELIANCE ON THE RCT

1. RCTs may be of short duration and may not be able to assess harm. 2. RCTs do not often include vulnerable or minority populations. Populations are homogeneous,

as women, elderly, and minority groups are often excluded. 3. RCTs report efficacy (worth), not effectiveness. 4. RCT settings may not represent typical situations (e.g., research university hospital vs. community hospital). 5. Assessment of harm is a secondary consideration in an RCT. 6. Patients who are more susceptible to harm are often underrepresented.

RCT, randomized controlled trial.






Since an SR of diagnostic test accuracy is used to ascertain how well a test, or a series of tests, is able to correctly identify patients with the target condition (Bossuyt & Leeflang, 2008), the review question will vary slightly, as patient/condition of focus (P), index test (I), reference standard (C), and outcomes (O). For example, Doust, Glasziou, Pietztrak, and Dobson (2004) assessed the diag-nostic accuracy of brain natriuretic peptide (BNP), including a comparison with atrial natriuretic peptide (ANP), in patients with heart failure, finding that BNP is an accurate marker of heart failure.

Prognostic questions generally have only two parts: the population (P) and outcome (O). For example, Al-Shahi and Warlow (2001) were interested in the frequency and clinical course of arteriovenous malformations (AVM) in the brain of adults.

Etiology questions serve to establish the degree of relationship between two or more quantifiable variables, for example, the consumption of coffee and heart disease. According to Moola et al. (2015), questions for an etiologic SR should outline the exposure, disease, condition, symptom of interest, the population at risk, the context (e.g., geographical, temporal, or cultural elements as appropri-ate), the period (e.g., seasonal variations), and the length of time (e.g., length of a pregnancy). For example, the question is: Are children who are exposed to secondhand tobacco smoke in vitro at risk for asthma? In this example, the dura-tion of exposure is included to study the long-term effects of secondhand smoke exposure in utero (Moola et al., 2015).

SEARCH STRATEGY

The search strategy for an SR of observational evidence follows the same stages as presented in previous chapters with some minor variation. Reviewers may perform an initial search while looking for both observational studies and RCTs, or searches can be performed sequentially, searching first for an RCT and then for observational studies. In any review of quantitative evidence, RCTs are always included in the search. With observational evidence used as the default in the event, there are either no or limited numbers of RCTs found. The use of clinical filters is important for finding diagnostic and prognostic studies, and one way to accomplish this is by using PubMed’s clinical inquiry fields.

Searching for studies to conduct an SR of diagnostic test accuracy or prog-nosis is more challenging than searching for the RCTs that are necessary to an SR of interventions. There is no unequivocal key word or indexing term ran-domized controlled trial to use. The Medical Subject Heading (MeSH) “sensi-tivity and specificity” used in diagnostic test accuracy is not consistently applied in electronic bibliographic databases. Leeflang, Deeks, Gatsonis, and Bossuyt (2008) suggest that diagnostic test review authors search for tests with a clear name that are used for a single purpose; searching for publications in which those tests are mentioned may be helpful. For other reviews, it may be necessary to add the patient description, although this is also often poorly indexed. Until






indexing systems properly code studies of test accuracy, searching for them will remain challenging, and footnote chasing may be necessary.

The search for diagnostic studies should focus on cross-sectional studies, whereas a prognostic review should focus on retrieval of cohort studies. When conducting a review for either of these, a target condition should be specified. A target condition describes patients with a particular clinical history, exami-nation, and test results (Bossuyt & Leeflang, 2008).

Studies of etiology are derived from any type of observational study. Prospec-tive cohort studies, though, generally provide stronger evidence for etiology questions and these should be at the foundation of the search strategy.

APPRAISING STUDIES

As discussed in previous chapters, appraisal of retrieved studies and the deci-sion on whether to include or not include a study in an SR is a crucial decision for reviewers. Any SR should include those available studies that are of the high-est quality. Appraisal of each paper by two reviewers independently accom-plishes this goal. The JBI (2008) has developed appraisal questions for cohort, cross-sectional, and descriptive studies as follows:

For case–control and cohort studies:

1. Is the sample representative of patients in the population as a whole? 2. Are the patients at a similar point in the course of their condition/illness? 3. Has bias been minimized in relation to selection of cases and controls? 4. Are confounding factors identified and strategies to deal with them stated? 5. Are outcomes assessed using objective criteria? 6. Was follow-up carried out over a sufficient period? 7. Were the outcomes of people who withdrew described and included in the

analysis? 8. Were outcomes measured in a reliable way? 9. Was appropriate statistical analysis used?

For descriptive/case series/case report studies:

1. Was the study based on a random or pseudorandom sample? 2. Were the criteria for inclusion in the sample clearly defined? 3. Were confounding factors identified and strategies to deal with them

stated? 4. Were outcomes assessed using objective criteria? 5. If comparisons were being made, was there sufficient description of groups? 6. Was follow-up carried out over a sufficient period? 7. Were the outcomes of people who withdrew described and included in the

analysis?






8. Were outcomes measured in a reliable way? 9. Was appropriate statistical analysis used?

A more detailed description can be found in the JBI 2015 Reviewers Manual (www.joannabriggs.edu.au).

For a diagnostic SR, tools such as the Standards for Reporting Diagnostic Accuracy Studies (STARD; www.stard-statement.org) statement can be used. The STARD checklist comprises 25 items for evaluation related to the design of the study, study participants, test methods and results, statistical analysis, and an evaluation of the study discussion. The aim of the STARD initiative is to improve the accuracy and completeness of reporting of studies of diagnostic accuracy, to allow readers to assess the potential for bias in the study (internal validity), and to evaluate its generalizability (external validity). An easy-to- follow flow diagram that is useful in describing the study and the flow of patients is also available on the STARD website.

The Quality Assessment of Diagnostic Accuracy Studies (QUADAS) check-list can also be used for appraisal of diagnostic studies when doing an SR. The tool is organized into a checklist of 14 questions, which are answered “yes,” “no,” or “unclear” (see Table 9.7). The majority of items included in QUADAS relate to bias (items 3, 4, 5, 6, 7, 10, 11, 12, and 14), with only several items relating to variability (items 1 and 2) and reporting (items 8, 9, and 13). A more detailed description of each item together with a guide on how to score each of the ques-tions can be found at Whiting et al. (2003).

TABLE 9.7 THE QUADAS TOOL

1. Was the spectrum of patients representative of the patients who will receive the test in practice? 2. Were selection criteria clearly described? 3. Is the reference standard likely to correctly classify the target condition? 4. Is the period between reference standard and index test short enough to be reasonably sure that

the target condition did not change between the two tests? 5. Did the whole sample or a random selection of the sample receive verification by using a reference

standard of diagnosis? 6. Did patients receive the same reference standard regardless of the index test result? 7. Was the reference standard independent of the index test (i.e., the index test did not form part of

the reference standard)? 8. Was the execution of the index test described in sufficient detail to permit replication of the test? 9. Was the execution of the reference standard described in sufficient detail to permit its

replication? 10. Were the index test results interpreted without knowledge of the results of the reference

standard? 11. Were the reference standard results interpreted without knowledge of the results of the index

test? 12. Were the same clinical data available when test results were interpreted as would be available when

the test is used in practice? 13. Were uninterpretable/intermediate test results reported? 14. Were withdrawals from the study explained?

QUADAS, Quality Assessment of Diagnostic Accuracy Studies.Source: Whiting et al. (2003).

http://www.stard-statement.org






Notably, the following additional questions should be asked when appraising diagnostic studies to be sure that only the highest quality studies are included in the review (JBI, 2015):

■■ What is the clinical spectrum of the patients in the study?■■ Was there blinded interpretation of the test and reference standard results?■■ If a prospective design, was there consecutive patient sampling?■■ Is there adequate description of the index test, the reference standard, and

the study population and setting?

The appraisal of studies of prognosis is a bit different. Establishing an accu-rate prognosis is one of the most common tasks for the health care provider. Prognosis is also important in making inferences about quality of care when comparing outcomes in populations from different contexts and institutions (Carneiro, 2002). According to Carneiro (2002), there are three components of an appraisal: (a) a qualitative element (What are the possible outcomes?); (b) a quantitative element (What is the probability of these outcomes occur-ring?); and (c) a temporal element (Over what period might they occur?). Cardarelli and Oberdorfer (2007) refine this approach by suggesting specific areas for assessment:

■■ Representativeness of the population■■ Association between study subjects and patient population■■ Point in the course of the disease ■■ Length of the follow-up period ■■ Blinding■■ Attention to prognostic factors■■ Long-term effects of projected outcomes■■ Association among immediate and long-term outcomes■■ Clinical significance of the outcomes to the practitioner’s patient population

DATA EXTRACTION

It is important in an SR of observational evidence to extract and code data according to a prespecified plan (see Chapter 8 for a detailed description of coding and extracting data). Specific data to be extracted for an observational review are presented in Table 9.8.

DATA SYNTHESIS

Although there is no consensus as to how to synthesize data for an observa-tional SR, the usual procedure is to use the same methods used in an SR of interventions (see Chapter 8) while keeping in mind that there may be greater






degrees of confounding, bias, and heterogeneity in a review of observational evidence. Caution needs to be taken in interpreting the results of an obser-vational review so as not to fall victim to plausible but spurious results. The statistical combination of data should not, therefore, be a prominent feature of an observational review. More may be gained from a careful review of hetero-geneity (Eggers, Smith, & Schneider, 2007).

WRITING THE REPORT

Standards for reporting the results of an observational SR must be maintained to allow proper evaluation of the quality and completeness of the meta-analysis (Stroup, Berlin, & Morton, 2000). The Meta-Analysis of Observational Studies in Epidemiology (MOOSE) Group, convened by the Centers for Disease Control and Prevention (CDC), developed the MOOSE reporting guidelines to address this issue. The guidelines were developed to address the problem of increasing diversity and variability that exist in reporting meta-analyses of observational studies in an easy-to-follow checklist format (see Table 9.9).

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA), formerly QUOROM, statement can also be used. The PRISMA statement provides an evidence-based set of 27 items in a checklist format. Table 9.10 presents an outline of the essential elements of the PRISMA statement. Although it focuses on randomized trials, the PRISMA statement can also be used as a basis for reporting any SRs.

TABLE 9.8 DATA TO BE EXTRACTED FOR A SYSTEMATIC REVIEW OF OBSERVATIONAL DATA

Descriptive studies

Mean

Standard deviation

Confidence interval

Odds ratio

Relative risk

Correlations

Diagnostic studies

Results from both the index test and comparator

Sensitivity results

Specificity results

Likelihood ratio

Receiver operator characteristic

Prognostic studies

A numerical summary of prognostic strength, such as a hazard ratio, a survival curve, or an odds ratio






TABLE 9.9 ESSENTIAL ELEMENTS OF A MOOSE REPORT

■■ Description of the problem■■ Description of the population under study■■ Search strategy, including time frame, key words, and phrases■■ Data coding schema used■■ Quality appraisal of selected studies■■ Methods of data analysis and statistical methods used■■ Description of any heterogeneity■■ Table of evidence ■■ Description of any subgroup analysis■■ Presence of any publication bias■■ Discussion of results using appropriate tables or graphs■■ Discussion of conclusions within the domain of the review of the literature■■ Suggestions for future research

MOOSE, Meta-Analysis of Observational Studies in Epidemiology.Adapted from Stroup et al. (2000).

TABLE 9.10 OUTLINE OF PRISMA STATEMENT

Title

Identify the review as a systematic review

Abstract

Introduction

Rationale for the review

Objectives of the review

Methods

Protocol and registration

Eligibility criteria

Information sources

Eligibility criteria

Search

Study selection

Data collection process

Data items

Risk of bias in individual studies

Summary measures

Synthesis of results

Selection/topic

Risk of bias across studies

Additional analysis

Results

Summary of evidence

Studies included

Study characteristics

(continued)






■■ Practice Activities

Think about a practice problem you have recently encountered, for example, an increase in fall rates or urinary tract infections, high staff turnover rate, and increased rates of physical restraints.

1. Write a PICO to study this problem. 2. Rewrite the PICO as a research question. 3. What type of research design would you use to study this problem? 4. What sampling strategy would you use? 5. What data would you collect and how would you analyze them?

■■ Suggested Reading

Grimes, D. A., & Schultz, K. F. (2002). Cohort studies: Marching towards outcomes. Lancet, 359, 341–345.

Grimes, D. A., & Schultz, K. F. (2002). Descriptive studies: What they can and cannot do. Lancet, 359, 145–149.

Hayden, J. A., Cote, P., & Bombardier, C. (2006). Evaluation of the quality of prognosis studies in systematic reviews. Annals of Internal Medicine, 144(6), 427–437.

Moola, S., Munn, Z., Sears, K., Sfetcu, R., Currie, M., Lisy, K., . . . Mu, P. (2015). Conduct-ing systematic reviews of association (etiology): The Joanna Briggs Approach. Inter-national Journal of Evidence Based Healthcare, 13, 163–169.

Schultz, K. F., & Grimes, D. A. (2002). Case-control studies: Research in reverse. Lancet, 359, 431–434.

Shamliyan, T., Kane, R. L., & Dickinson, S. (2008). A systematic review of tools used to assess the quality of observational studies that examine incidence or prevalence and risk factors for diseases. Journal of Clinical Epidemiology, 63(10), 1061–1070.

Sjöström, L., Narbro, K., Sjöström, C. D., Karason, K., Larsson, B., Wedel, H., . . . Carlsson, L. M. (2007). Swedish Obese Subjects Study: Effects of bariatric surgery on mortality in Swedish obese subjects. New England Journal of Medicine, 357, 741–752.

Risk of bias within studies

Results of individual studies

Additional analysis

Discussion

Limitations

Conclusions

Funding

PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.Source: www.prisma-statement.org/statement.htm.

TABLE 9.10 OUTLINE OF PRISMA STATEMENT (continued )

http://www.prisma-statement.org/statement.htm






Zou, K. H., O’Malley, A. J., & Mauri, L. (2007). Receiver-operating characteristic anal-ysis for evaluating diagnostic tests and predictive models. Circulation, 115(5), 654–657.

■■ References

Al-Shahi, R., & Warlow, C. (2001). A systematic review of the frequency and prognosis of arteriovenous malformation of the brain in adults. Brain, 24(10), 1900–1926.

Altman, D. (2001). Systematic reviews of evaluations of prognostic variables. British Med-ical Journal, 323(7306), 224–228. doi:10.1136/bmj.323.7306.224

Aznar, S., Naylor, P. J., Silva, P., Pérez, M., Angulo, T., Laguna, M., . . . López-Chicharro, J. (2011). Patterns of physical activity in Spanish children: A descriptive pilot study. Child: Care, Health and Development, 37(3), 322–328. doi:10.1111/j.1365-2214 .2010 .01175.x

Bossuyt, P. M., & Leeflang, M. M. (2008). Chapter 6: Developing criteria for including studies. In J. P. T. Higgins & S. Green (Eds.), Cochrane Handbook for Systematic Reviews of Interventions, version 5.1.0 [updated March 2011]. The Cochrane Collaboration, 2011. Retrieved from www.cochrane-handbook.org

Campbell, J. (2015). Does a systematic review of diagnostic tests begin and end with accu-racy? JBI Database of Systematic Reviews & Implementation Reports, 13(10), 1–2.

Cardarelli, R., & Oberdorfer, J. R. (2007). Evidence based medicine, part 5: An introduc-tion to critical appraisal of articles on prognosis. Journal of the American Osteopathic Association, 107(8), 315–319.

Carneiro, A. V. (2002). Critical appraisal of prognostic evidence: Practical rules. Revista Portuguesa de Cardiologia, 21(7–8), 891–900.

Chan, P. S., Krumholz, H. M., Spertus, J. A., Jones., P. G., Cram, P., Berg, R. A., . . . American Heart Association National Registry of Cardiopulmonary Resuscitation (NRCPR) Investigators. (2010). Automated external defibrillators and survival after in-hospital cardiac arrest. Journal of the American Medical Association, 304(19), 2129–2136.

Chen, H. L., Liu, P. F., Liu, P. W., & Tsai, P. S. (2010). Awareness of hypertension guidelines in Taiwanese nurses: A questionnaire survey. Journal of Cardiovascular Nursing, 26(2), 129–136.

Dahnke, M. (2011). Observation: The scientific gaze. In M. Dahnke & H. M. Dreher (Eds.), Philosophy of science for nursing practice: Concepts and application (pp. 153–172). New York, NY: Springer.

Davies, H. T. O., & Crombie, I. K. (2003). What is a systematic review? “What is . . .?” Bulletins, 1(5), 1–6. Retrieved from www.evidence-based medicine.co.uk

Deeks, J. (2007). Systematic reviews of evaluations of diagnostic and screening tests. In M. Egger, G. D. Smith, & D. Altman (Eds.), Systematic reviews in health care: Meta-analysis in context (pp. 110–117). London, UK: BMJ Publishing.

Doust, J. A., Glasziou, P. P., Pietztrak, E., & Dobson, A. (2004). A systematic review of the diagnostic accuracy of natriuretic peptides for heart failure. Archives of Internal Med-icine, 164, 1978–1984.

Egger, M., Davey Smith, G., & Schneider, M. (2001). Systematic reviews of observa-tional studies. In M. Egger, G. Davey Smith, & D. G. Altman (Eds.), Systematic reviews in health care: Meta-analysis in context (pp. 211–227). London, UK: BMJ Publishing.

Fraser, K. D., Estabrooks, C., Allen, M., & Strang, V. (2010). The relational nature of case manager resource allocation decision making: An illustrated case. Care Management Journals, 11(3), 151–156.






Glasziou, P. P., Irwig, L., Bain, C. J., & Colditz, G. A. (2000). How to review the evidence: Systematic identification and review of the scientific literature. Canberra, Australia: National Health & Medical Research Council.

Glasziou, P., Irwig, L., Bain, C., & Colditz, G. (2001). Systematic reviews in health care. A practical guide. New York, NY: Cambridge University Press.

Grant, W. (2014). A multicountry ecological study of cancer incidence rates in 2008 with respect to various risk-modifying factors. Nutrients, 6(1), 163–189.

Grimes, D. A., & Schultz, K. F. (2002). Bias and causal relationships in observational research. Lancet, 359, 248–252.

Grover, S., Kate, N., Malhotra, S., Chakrabarti, S., Mattoo, S. K., & Avasthi, A. (2012). Symptom profile of delirium in children and adolescent—Does it differ from adults and elderly? General Hospital Psychiatry, 34(6), 626–632.

Harris, G. (2010, July 13). Diabetes drug maker hid test data, files indicate. The New York Times, Health News. Retrieved from http://www.nytimes.com

Heffernan, C. (2010). Ask Dr. Cath. Retrieved from http://www.drcath .net/index.htmlJepsen, P., Johnsen, S. P., Gillma, M. W., & Sorensen, H. T. (2004). Interpretation of obser-

vational studies. Heart, 90(8), 956–960.Joanna Briggs Institute. (2008). Reviewers manual. Adelaide, Australia: University of

Adelaide.Joanna Briggs Institute. (2015). Reviewers manual. Adelaide, Australia: University of

Adelaide.Joanna Briggs Institute for Nursing and Midwifery. (2009). Study guide for module 2. Ade-

laide, Australia: Author.Leeflang, M., Deeks, J., Gatsonis, C., & Bossuyt, P. (2008). Systematic reviews of diagnostic

test accuracy. Annals of Internal Medicine, 149(12), 889–897.Mann, C. (2003). Observational research methods. Emergency Medicine Journal, 20(1),

54–60.Moola, S., Munn, Z., Sears, K., Sfetcu, R., Currie, M., Lisy, K., . . . Mu, P. (2015). Conducting

systematic reviews of association (etiology): The Joanna Briggs Approach. Interna-tional Journal of Evidence Based Healthcare, 13, 163–169.

National Cancer Institute Dictionary. Retrieved from http://www.cancer .gov/publications/dictionaries/cancer-terms

Shellmer, D. A., DeVito Dabbs, A., Dew, M. A., Noll, R. B., Feldman, H., Strauss, K. A., . . . Mazariegos, G. V. (2011). Cognitive and adaptive functioning after liver transplantation for maple syrup urine disease: A case series. Pediatric Transplantation, 15(1), 58–64. doi:10.1111/j.1399-3046.2010.01411.x

Smidt, M., Deeks, J., & Moore, T. (2008). Guide to a Cochrane review and protocol for diagnostic accuracy. In J. P. T. Higgins & S. Green (Eds.), Handbook for systematic reviews of diagnostic test accuracy version 0.4 [updated September 2008]. The Cochrane Collaboration. Retrieved from www.cochrane-handbook.org

Stroup, D., Berlin, J., & Morton, S. (2000). Meta-analysis of observational studies in epide-miology: A proposal for reporting. Journal of the American Medical Association, 238(15), 2008–2013.

Suarez, L., Ramadhani, T., Felkner, M., Canfield, M. A., Brender, J. D., Romitti, P. A., & Sun, L. (2011). Maternal smoking, passive tobacco smoke, and neural tube defects. Birth Defects Research Part A: Clinical and Molecular Teratology, 91, 29–33. doi:10 .1002/bdra.20743

Whiting, P., Rutjes, A. W. S., Reitsma, J. B., Bossuyt, P. M. M., & Kleijnen, J. (2003). The development of QUADAS: A tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews. BMC Medical Research Methodology, 3, 25. doi:10.1186/1471-2288-3-25

Zou, K. H., O’Malley, A. J., & Mauri, L. (2007). Receiver-operating characteristic analysis for evaluating diagnostic tests and predictive models. Circulation, 115(5), 654–657.

http://www.nytimes.com

http://www.drcath.net/index.html

http://www.cancer.gov/publications/dictionaries/cancer-terms

http://www.cancer.gov/publications/dictionaries/cancer-terms


comprehensive systematic review for advancedlghttp.48653.nexcesscdn.net › 80223cf ›...

Documents