Helping Physicians Make Evidence-Based Decisions
Am Fam Physician. 2001 Jun 1;63(11):2130-2136.
In the face of exploding medical information, family physicians continue to look for better ways to keep up and to provide patients with the best care based on the most reliable and up-to-date evidence. This issue of American Family Physician introduces a valuable tool in that battle—systematic reviews from the Cochrane Library. What exactly is this new tool, and how should readers of AFP plan to use it?
The Cochrane Library is produced by a relatively new international organization—the Cochrane Collaboration. The name of the collaboration is derived from that of Archie Cochrane, a British statistician who believed passionately in the use of randomized controlled trials (RCTs) to determine the effectiveness of medical interventions and who dreamed of a day when all existing RCTs would be summarized into topical reviews. It is this huge task that the collaboration has undertaken. Starting with a few reviews on perinatal care, the Cochrane Library has grown rapidly and now contains more than 850 reviews on a wide variety of topics from “acellular pertussis vaccines” to “zinc for the comXmon cold.”
Primary care physicians have played an important role in the Cochrane Collaboration since its inception. Chris Silagy, an Australian general practitioner and a former chairperson of the Collaboration's steering group, formed the Cochrane Primary Health Care Field to promote the production of Cochrane Reviews with relevance to primary care practice. His efforts have paid off magnificently. Each quarterly issue of the Cochrane Library contains new and updated reviews on primary care topics, with an increasing number of family physicians and general practitioners serving as review authors. Abstracts of all of the current Cochrane reviews are available free of charge online at http://www.cochrane.org.au/, with a search engine that allows easy access to reviews on any specific topic. Full text versions of Cochrane reviews are available by subscription on the Internet or on CD-ROMs that are distributed quarterly. Most medical school libraries are subscribers, and many are making Internet access available to clinical teachers associated with their institutions. In some states, such as Oregon, medical schools have provided free access to Cochrane reviews to all physicians in the state.
Readers who peruse “Cochrane for Clinicians” in this issue or who have encountered the Cochrane Library in other settings will note significant differences in format and emphasis from traditional review articles. Cochrane reviews begin with a rather narrow, tightly focused clinical question and attempt to find all relevant RCTs that address the question. Articles are uncovered in an exhaustive search that often includes non-English sources and are then assessed for quality. Data from high-quality studies are combined in a series of charts and tables that give an overview of the results when all studies are taken into account. The statistical method used is termed “meta-analysis,” and it can uncover important and statistically significant effects that may not be noticed if individual trials are looked at in isolation.
Systematic reviews of this sort are appearing with increasing frequency in the medical literature. This change has come about because of significant problems with the timeliness and validity of traditional review articles.
The rapid pace of scientific discoveries in medicine has increased the challenge of incorporating the most up-to-date information into a review. Difficulties arise when original research studies reach conflicting conclusions or contain such small study samples that a clear benefit cannot be shown for the new procedure, test or treatment under investigation. Meta-analysis can be particularly helpful here because it allows the results of several small studies to be mathematically combined.
The Cochrane logo illustrates a systemic review of data from seven RCTs, comparing one health care treatment with placebo. The diagram fundamental to the logo tells the story of one of the first systemic reviews rendered by the Cochrane Collaboration: a review of a short, inexpensive course of corticosteroids given to women about to give birth prematurely. Although corticosteroids are now routinely used to accelerate fetal lung maturity in infants at risk of premature delivery, this practice was considered to be controversial in the early days of the Collaboration. Seven small RCTs had been completed, but the results were conflicting. Most of the studies did not show a statistically significant benefit from corticosteroid therapy. When the studies were combined in a Cochrane review, however, it became clear that corticosteroid therapy has a positive overall effect in infants at risk of preterm delivery.
In other cases, the lack of a systematic review has delayed the adoption of effective treatments or prolonged the use of ineffective interventions. For example, traditional reviews and textbook articles failed to recommend the use of thrombolytic agents for acute myocardial infarction until 12 years after the effectiveness of this therapy could have been documented by a meta-analysis.1
Through the efforts of the Cochrane Collaboration, dozens of new systematic reviews are produced and published every quarter, making delays of this sort less likely in the future. Writers of traditional medical reviews often do not have an opportunity to update their articles, and information may quickly become outdated after publication. The Cochrane Collaboration has addressed this issue with an expectation that producers of Cochrane reviews will continuously survey ongoing medical research and update their reviews as new information becomes available. Each review is updated at least once every two years.
In scientific endeavors, validity is related to reproducibility—could the work be reproduced by a colleague with the same results? Traditional review articles have never been required to meet this criterion. Authors are given broad margin to offer opinions based on their own eclectic selection of supporting evidence, with the assumption that authors with the highest academic credentials would produce the most valid reviews. Unfortunately, this is not a model assumption. Empiric studies have shown that the quality of a review article is often superior when it is written by a nonexpert.2 The junior faculty member, when handed such a task, may be more likely to do an exhaustive literature review to find and include the most recent studies.
Regardless of expertise, authors of traditional reviews may do an incomplete literature search, thinking that they already know the field quite well. Moreover, they may have their own agenda to further, or they may be influenced by a variety of external factors. For example, one study3 showed that authors who received funding from pharmaceutical companies were more likely to minimize the potential harm of calcium channel blockers. The problem is not that these biases exist—we are all human, after all—but that the reader of a traditional review has no way to assess biases or determine how such biases may have affected the conclusions of the review. Systematic reviews minimize the potential for bias by using a clearly defined method for identifying, evaluating and summarizing evidence.
Because of the focus on quality and reproducibility, the format of Cochrane abstracts and articles will seem unfamiliar compared with traditional review articles. Cochrane reviews contain information on the methods and mechanics of how the review was carried out—which databases were searched, how articles were selected for inclusion into the review, how data were abstracted from each article, etc. In a similar fashion, Cochrane reviews provide a great deal of detail about the methods used in the studies. The inclusion of this material allows readers of Cochrane reviews to evaluate their validity. The graphic depiction of data used in Cochrane reviews, while it may appear unfamiliar at first, contains a wealth of information about the overall findings when studies are combined, as well as the results of individual studies.
The Cochrane Library, with its growing collection of systematic reviews, is revolutionizing medicine. Cochrane reviews are increasingly seen as the gold standard of evidence-based summaries. Although much of the content of the library is relevant to family physicians and their patients, many physicians are not aware of the Cochrane Library or would have difficulty accessing it. AFP's new “Cochrane for Clinicians” series will address these issues by introducing readers to the content of the Cochrane Library, giving examples and summaries of key reviews and providing readers with tools to help them understand the format and data presented in Cochrane reviews.
LORNE BECKER, M.D., is professor and chair of the Department of Family Medicine at State University of New York Upstate Medical University in Syracuse and coordinator of the Cochrane Primary Health Care Field.
Address correspondence to Lorne Becker, M.D., Department of Family Medicine, SUNY Upstate Medical University, Syracuse, N.Y. 13210.
1. Antman EM, Lau J, Kupelnick B, Mosteller F, Chalmers TC. A comparison of results of meta-analyses of randomized control trials and recommendations of clinical experts. Treatments for myocardial infarction. JAMA. 1992:268;240–8.
2. Oxman AD, Guyatt GH. The science of reviewing research. Ann N Y Acad Sci. 1993;703:125–33.
3. Stelfox HT, Chua G, O'Rourke K, Detsky AS. Conflict of interest in the debate over calcium-channel antagonists. N Engl J Med. 1998;338:101–6.
Copyright © 2001 by the American Academy of Family Physicians.
This content is owned by the AAFP. A person viewing it online may make one printout of the material and may use that printout only for his or her personal, non-commercial reference. This material may not otherwise be downloaded, copied, printed, stored, transmitted or reproduced in any medium, whether now known or later invented, except as authorized in writing by the AAFP. Contact firstname.lastname@example.org for copyright questions and/or permission requests.
Want to use this article elsewhere? Get Permissions