What Do I Do in My Practice That Isn't Evidence Based?

Although this question might be considered to be provocative by someone who spends his academic life conducting randomized trials, I was pleased when David Sachar invited me to discuss this topic. My perception is that although most gastroenterologists are aware of evidence-based medicine (EBM), some misconceptions about the subject still exist. In this article, I will describe the origins of the EBM movement and attempt to provide some insight into how clinicians should use EBM in their daily practice. Incidentally, my answer to David's question is an emphatic `Nothing!' (I hope that after reading the article you won't find this response to be so self-righteous!) Evidence-based medicine is simply the application of the most valid scientific information to patient care. Physicians who treat patients with inflammatory bowel disease should practice medicine using the best available evidence, and this requires the ability to access and evaluate scientific information. Although it has been argued that physicians have always practiced according to this doctrine, considerable evidence exists that this is not the case; for example, large variances have been documented in practice patterns despite the presence of strong scientific evidence for a particular standard of care. This point is illustrated by a study performed by Bernstein and colleagues (1), who surveyed gastroenterologists regarding the management dysplasia in patients with ulcerative colitis. Only 19% of respondents understood the implications of finding dysplasia on a surveillance biopsy. More importantly, when asked to specify the appropriate management for these patients, a substantial proportion of the physicians inappropriately recommended continued surveillance rather than colectomy. This example illustrates the importance of motivating clinicians to obtain and use the best available evidence. EBM began in the late 1940s coincident with the performance of the first randomized, controlled trial (RCT). In these early days, a handful of epidemiologists and statisticians, notably Sir Bradford Hill and Sir Richard Doll in England, recognized the need to apply scientific principles to the evaluation of new therapies. The performance of the British Medical Research Council trial of streptomycin (2) for the treatment of tuberculosis was a landmark event that forever changed the practice of clinical medicine. A novel principle, that clinical decision-making should be guided by scientific data rather than exclusively by physician experience, was firmly established. The IBD community can take pride that our own Sidney Truelove was quick to appreciate the value of the RCT. In 1955, the Oxford study of hydrocortisone (3) for the treatment of active ulcerative colitis provided valuable new information to clinicians and greatly influenced practice. The results of this RCT were of tremendous benefit to patients, as shown by subsequent decreases in mortality and morbidity. Over the next decade, clinical trials provided a wealth of new scientific knowledge that busy clinicians were required to assimilate. Consequently, Archie Cochrane recognized the need for a systematic approach to compiling the world's medical literature. His efforts culminated in the creation of the Cochrane Collaboration, an organization consisting of volunteers who identify, assess, and summarize topic-specific information in all medical specialties. Cochrane Reviews are now readily available on a wide range of topics (4). Several other clinical epidemiologists were instrumental in bringing evidence-based medicine to the bedside. David Sackett et al. at McMaster University developed principles of critical appraisal that could be used by clinicians to evaluate the rapidly growing medical literature. Sackett's classic text Clinical Epidemiology: A Basic Science for Clinical Medicine (5) introduced many physicians to the principles of EBM. Among gastroenterologists, Thomas Chalmers (6) was a strong advocate of EBM who emphasized the value of evidence from controlled trials. Similarly, Alvan Feinstein (7) informed clinicians regarding the scientific interpretation of diagnostic tests. These leaders found many willing and enthusiastic followers. Today, the Cochrane IBD group, chaired by Dr. Jack McDonald, generates high-quality reviews that are available in electronic form through the Cochrane Library (8). Every clinician who treats patients with IBD can use this valuable resource as a guide to practice. Given the availability of this and other sources of high-quality information (9), why wouldn't a good clinician practice EBM? Several misconceptions exist about the EBM movement. The first is the notion that practitioners of EBM rely exclusively or even slavishly on the results of RCTs to make clinical decisions. Perhaps some clarification is necessary. There is no doubt that the RCT is the scientific gold standard for the evaluation of new therapies. Randomization offers two powerful advantages. Foremost is the capacity to distribute confounding variables equally among experimental groups. Many factors can influence the outcome of treatment in patients with IBD. For example, patients with active Crohn's disease who require corticosteroid therapy to control symptoms have a worse prognosis than those who do not. In a placebo-controlled trial of a new drug randomization distributes patients receiving corticosteroids equally among the experimental groups. This minimizes the effects of corticosteroid therapy as a confounding variable. Second, randomization reduces bias (systematic deviation from the truth). Many of the outcomes used in IBD trials, such as the Crohn's Disease Activity Index, are relatively subjective and thus sensitive to bias (10). Random allocation allows concealment of treatment from investigators and patients and reduces the possibility of bias. The importance of this issue was first demonstrated by Chalmers (11), who showed that nonrandomized studies consistently overestimate the benefit of a new treatment in comparison to RCTs. Frequently, new treatments for IBD that show promising results in open studies are found to be ineffective when a methodologically rigorous RCT is performed. Nevertheless, in some situations data from RCTs is either difficult or impossible to obtain and other kinds of evidence must be used. For example, in studies of causation it is often problematic to perform randomized experiments. Nevertheless, high-quality observational evidence can be used to guide clinical decision-making. To illustrate this point, we can revisit the previous topic of dysplasia screening in patients with ulcerative colitis. I unabashedly admit that I perform surveillance colonoscopy despite the absence of any RCT that has definitively shown that this practice is effective. In this circumstance my decision is based on the best available evidence—high-quality observational data (12,13). Another common misconception is that data generated from RCTs have only limited relevance to the “real world” of clinical medicine because of the unique features of each clinical encounter. The nidus for this argument is that RCTs are experiments performed on unique populations of patients who are intrinsically different than those encountered in your own practice. The question then becomes whether the results of an internally valid trial is also equally valid in the context of a specific practice setting. This issue, known as generalizability in EBM-speak, is dear to the hearts of critics of EBM. It is certainly true that the volunteers who participate in RCTs are healthier and have a better prognosis than unselected patients (14,15). Furthermore, participants also are selected on the basis of inclusion and exclusion criteria that tends both to increase the likelihood of successful treatment and to minimize the possibility of adverse events. Therefore, it is reasonable to speculate that some treatments found to be effective by RCTs might not translate well in the setting of usual practice. Unfortunately, this supposition has led to underutilization of many highly effective therapies. The best examples come from cardiology, where underutilization of beta blockers after myocardial infarction and clinicians' reluctance to prescribe thrombolytics to elderly patients are well documented (16,–19). In these instances, clinicians likely concluded, incorrectly, that these interventions would be tolerated poorly in older patients under usual practice conditions and elected not to offer them to patients. Unfortunately, this misapplication of `experience-based medicine' undoubtedly had a negative effect on patients. A similar situation may exist in the world of IBD where many gastroenterologists have been reluctant to prescribe antimetabolite therapy for patients with Crohn's disease, despite the presence of favorable efficacy and safety data derived from RCTs (20,–24). My perspective on this issue is that the results of an RCT are generalizable to my patient unless strong evidence exists that this is not the case. Too often, clinicians deny patients access to effective therapies based on an incorrect clinical instinct. Finally, one of the most troubling misconceptions about EBM is that it is a separate entity from traditional clinical skills. Use of the principles of EBM in the management of patients is only complementary to the qualities possessed by expert clinicians and will never supersede careful observation, sound judgment, and compassion for the patient. It is important to recognize that good clinicians intrinsically have used some of the fundamental concepts of EBM by espousing principles such as `do the last test first' and `go where the money is.' Knowledge of EBM enables clinicians to understand why these basic principles of clinical medicine are valid through the use of a quantitative approach to decision-making. Making use of the knowledge provided by EBM can in no way be considered detrimental to the doctor-patient relationship. Most of us would like to believe that we practice EBM; however, my perception is that there are several areas, particularly with respect to the treatment of IBD, where this is not the case. Here are two examples. First, consider the use of 5-aminosalycylic acid (ASA) therapy for Crohn's disease. Both the National Collaborative Crohn's Disease (NCCDS) (25) and European Collaborative Crohn's Disease (ECCDS) (26) trials failed to demonstrate a benefit for sulfasalazine maintenance therapy. Subsequently, with the development of new formulations of 5-ASA that allowed delivery of higher concentrations of mesalamine with a low risk of adverse events, multiple (n = 17) RCTs were performed that evaluated these newer agents. These studies consistently failed to demonstrate efficacy for 5-ASA maintenance therapy. Although meta-analyses (27,–29) suggested there may be a benefit for some patients, particularly following a surgically induced remission, strong evidence to support this intervention is not available. Nevertheless, 5-ASA is the most frequently prescribed maintenance therapy. A similar situation exists regarding the use of 5-ASA for induction of remission. Both the NCCDS (25) and ECCDS (26) trials showed a modest benefit for sulfasalazine induction therapy. Based on 1) the results of a relatively large RCT that showed a small improvement in remission rates in comparison to placebo (30) and 2) a superior adverse event profile, most clinicians abandoned sulfasalazine for the newer formulations of 5-ASA. It was recently confirmed, however, that two additional large RCTs of 5-ASA never were reported (31). These studies showed no benefit of high-dose 5-ASA over placebo for induction of remission. My opinion, after reviewing these data, is that the use of 5-ASA by gastroenterologists is not evidence based. Moreover, sulfasalazine has been unfairly discarded as an induction therapy. It should be interesting to observe if the prescribing behavior of physicians will change now that new evidence has become available on this subject. Another example concerns the use of infliximab. Following successful development as a treatment of Crohn's disease, it was inevitable that infliximab would be evaluated for ulcerative colitis. My sense is that off-label use of infliximab for this indication is relatively common in North America. This is surprising, considering that until recently no RCTs have been performed that establish the safety and efficacy of this intervention. At this year's Digestive Disease Week conference, Probert et al. described the results of a study in which 42 patients with active ulcerative colitis were randomized to either infliximab therapy or placebo (32). No benefit was demonstrated. Although this trial was relatively small and may not have detected a benefit because of a type 2 statistical error, these results should diminish enthusiasm for the empiric use of a drug that can cause serious morbidity and mortality (33,34). Similarly, 6-thioguanine (6-TG) has been advocated (35) as an alternative treatment of patients with Crohn's disease who are either refractory or intolerant to the purine antimetabolites. Again, given that no controlled data are available to support efficacy and that 6-TG may be associated with venoocclusive disease (36), the wisdom of this practice is questionable. In the past 50 years, the EBM movement has gained acceptance among physicians. Misconceptions regarding EBM are persistent, however, and in some instances, clinical practices continue that are not evidence based. If you believe that patient care should be based on the best data available, it is difficult not to buy into the concept of EBM. I hope that when I write for the readers of Inflammatory Bowel Diseases, a journal committed to publishing high-quality evidence, I am preaching to the converted.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.