CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2003;1:322–327 EVIDENCE-BASED MEDICINE SERIES An Evidence-Based Approach to Clinical Practice Guidelines: Diagnosis and Treatment of Irritable Bowel Syndrome PHILIP SCHOENFELD*,‡ and WILLIAM D. CHEY* *Division of Gastroenterology, University of Michigan School of Medicine, and ‡Veterans Affairs Center for Excellence in Health Services Research, Ann Arbor, Michigan T he Institute of Medicine defines clinical practice guidelines as “systematically developed statements to assist practitioner and patient decisions about appropriate health care for specific clinical circumstances.”1 In other words, clinical practice guidelines may offer clear recommendations about the utility of diagnostic tests and the efficacy of different treatments for specific disorders. Guidelines have been promoted as tools that enhance the quality of medical care by reducing practice variations, minimizing spending on unnecessary diagnostic tests, and preventing the use of ineffective treatments.2 During the past decade, the number of published guidelines has increased exponentially. The Agency of Health Care Research and Quality established a National Guidelines Clearinghouse website, which lists hundreds of guidelines. Although it is in vogue to state that a guideline is evidence-based, most guidelines are developed from the consensus opinion of experts. An evidence-based guideline usually consists of multiple systematic reviews about the diagnosis and treatment of a disorder. These systematic reviews should meet rigorous criteria for the critical appraisal of clinical research on which the guidelines for diagnosis and treatment are based.3– 6 Nevertheless, expert opinion is still important in evidence-based guidelines because these experts are called on to render an opinion when there simply are no data. Experts also provide guidance about the implementation of evidence-based recommendations! This article, the fourth in the evidence-based medicine (EBM) series, provides an EBM format for critically appraising a clinical practice guideline (Table 1). This article follows the format of other EBM articles in the Journal of the American Medical Association and the User’s Guide to the Medical Literature.3– 6 The EBM concepts have been expanded from previous articles, and examples from gastroenterology literature have been used. Case Scenario: Part One You are a gastroenterologist in a large, multispecialty internal medicine practice. Irritable bowel syn- drome (IBS) is one of the most common disorders managed by your fellow physicians, but the clinic administrator noted significant variability in the use of diagnostic tests and choice of medical therapies. Therefore, you are asked to develop a guideline for the management of these IBS patients. You recall that an evidence-based guideline about the management of IBS was recently published.7 You plan to review this guideline to determine whether it meets criteria for an evidence-based guideline and to see whether the recommendations will be helpful in your multispecialty internal medicine practice. Assessing the Design of a Clinical Practice Guideline Before examining guideline recommendations, you should assess the methodology used to develop the guideline. If the guideline is based solely on the opinion of experts and does not perform systematic critical appraisals of currently available research, the guideline might provide inaccurate or biased recommendations that are not supported by good science. EBM provides frameworks for the systematic and critical appraisal of the methodology of a clinical practice guideline (Table 1). This approach reduces the tension between the reader’s concern that an expert’s opinion might be subjective and the desire to base clinical practice on published data whose quality and veracity can be appraised by the reader if he/she wishes to do so. Did the guideline clearly specify appropriate questions? Evidence-based guidelines require clearly fo- cused questions about specific patient populations, specific interventions, and distinct outcomes. Clearly focused questions might be developed about issues of Abbreviations used in this paper: EBM, evidence-based medicine; IBS, irritable bowel syndrome. © 2003 by the American Gastroenterological Association 1542-3565/03/$30.00 doi:10.1016/S1542-3565(03)00139-3 July 2003 Table 1. Guides for the Evaluation of an Article About a Clinical Practice Guideline Assessing the Design of a Clinical Practice Guideline Did the guideline clearly specify appropriate questions? Is there a systematic review for each question? Did the systematic review use explicit study selection criteria? Was a comprehensive literature search performed for each question? Was the methodology of individual studies assessed? Is there a reproducible assessment of study methodology and study results? Is there a transparent link between evidence from systematic reviews and guideline recommendations? Interpreting the Results How did the authors rate levels of evidence? How do authors grade recommendations? Can these results be applied to my patients? Are your patients similar to the patients described in the clinical practice guideline? How were values incorporated into the optimal course of action? Were important subgroups examined? diagnosis, treatment, or prognosis of a disorder. For example, the IBS guideline7 asked clearly focused questions about IBS treatments: “Are anti-spasmodic agents (intervention) more effective than placebo for global IBS symptom improvement (outcome) among IBS patients (patient population)?” By starting with clearly focused questions, appropriate systematic reviews can be performed to provide the foundation of evidence for guideline recommendations. Appropriately designed guidelines consider all relevant patient populations, interventions, and clinical outcomes. For example, multiple options are available for the treatment of IBS. If a guideline only made recommendations about the efficacy of antispasmodic agents, it would be an inadequate guideline. However, if recommendations were made on the efficacy of bulking agents, antidiarrheal agents, antispasmodic agents, serotonin receptor agonists and antagonists, antidepressants, and behavioral therapy, you would be reassured that most IBS treatment options had been reviewed by the authors of the guideline. Furthermore, the guideline should consider multiple clinical outcomes. For example, the ROME II committee8 stated that global IBS symptom improvement should be the primary outcome for therapy trials in patients with functional gastrointestinal disorders. However, IBS is a disorder characterized by multiple symptoms, including abdominal discomfort, bloating, and altered bowel habits. Therefore, an IBS guideline should also report on the efficacy of different treatments for the improvement of these individual IBS symptoms. A comprehensive clinical practice guideline might also report on improvements in quality of life, decrease in utilization EVIDENCE-BASED IRRITABLE BOWEL SYNDROME 323 of health care resources, or decrease in the number of days missed from work because of illness. Finally, appropriately designed guidelines should also consider all relevant patient groups. Is there a systematic review for each question? After developing clearly focused questions, the authors of evidence-based guidelines should systematically gather and analyze the clinical research surrounding each question. A future article in this EBM series will describe an EBM approach to systematic reviews and meta-analysis (a quantitative systematic review). In this article, we will provide a brief description of the methodology of a systematic review.9,10 Did the systematic review use explicit study selection criteria? Classic review articles or narrative re- views usually do not provide explicit criteria for the inclusion of studies in the review. This allows the authors of a narrative review to selectively reference studies that support their viewpoint while ignoring others that contradict their conclusions. In a systematic review, clearly focused questions specify the patient population, intervention, and outcome to be explored. Therefore, authors of systematic reviews explicitly state that they will include any study that examines a specific patient population, intervention, and outcome identified in the clearly focused question. The EBM approach also notes that poorly designed studies might produce biased or inaccurate data.11 Therefore, EBM criteria suggest that only randomized controlled trials should be used in systematic reviews about therapies. For issues of diagnosis, systematic reviews should rely on studies that performed a blinded comparison of a new diagnostic test to a gold standard diagnostic test, and these studies should be performed in a uniform group of patients who meet the symptom-based criteria for a disorder.12 Was a comprehensive literature search performed for each question? Authors of a systematic re- view must ensure that all relevant studies are identified and included in the review. To identify these articles, comprehensive searching of the medical literature must be performed. This requires searches of multiple medical databases, including PUBMED, MEDLINE, and EMBASE. Authors should be cautious to include nonEnglish language articles because all relevant studies are not always published in English. Traditionally, search terms used in the literature search are included to allow the reader to determine whether a comprehensive search was performed. Although comprehensive searching of the medical databases will identify many appropriate studies, these searches might not provide a truly comprehensive repre- 324 SCHOENFELD AND CHEY sentation of studies. “File drawer” bias refers to the well-known phenomenon that negative studies (i.e., studies that do not show a difference between treatment and placebo) are less likely to be published than “positive” studies (i.e., studies that do show a statistically significant difference between treatment and placebo).13 Therefore, a truly comprehensive literature search will include hand searching of abstracts from national and international medical meetings, because negative studies are more likely to be published as abstracts than as full articles.13 Finally, authors might write to pharmaceutical companies and experts to identify other relevant unpublished trials. Was the methodology of individual studies assessed? There are many issues that might affect the validity of a study (i.e., the likelihood that the study will produce accurate and unbiased results). For example, a randomized controlled trial that is double-blind is more likely to produce accurate and unbiased results than a similar trial that is not blinded.14 Therefore, appropriately designed systematic reviews also analyze the methodology of individual studies included in the review. It is particularly important to conduct this exercise in an evidence-based guideline because the strength of guideline recommendations is based on the methodologic quality of individual studies. The rigor with which the analysis is performed is one factor that differentiates an evidence-based systematic review from a classic narrative review article. Another factor might be the “Achilles heel” of both evidence-based systematic and narrative reviews: poor quality of the published data. For example, data about the efficacy of antispasmodic agents for IBS are based on poorly designed randomized controlled trials. Although these studies meet certain methodologic quality criteria (e.g., randomization, placebo-controlled), results from these trials are questionable because of the inadequate sample size, short duration of trials, and unclear inclusion criteria in each study.15 Therefore, in addition to EBM frameworks to assess the design of studies about therapy or diagnostic tests,11,12,16,17 other sources might also be used for the qualitative assessment of study methodology. For example, the ROME II committee developed standard criteria for the design of therapy trials in patients with functional gastrointestinal disorders.8 The authors of these criteria noted that unique circumstances might bias the results of a therapy trial conducted in patients with a functional gastrointestinal disorder. The IBS guideline7 used ROME II criteria to grade the quality of each therapy trial included in the guideline. It is paradoxical that many of the recommen- CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 1, No. 4 dations of the ROME II committee are based on consensus expert opinion and are not truly based on results of epidemiologic studies. Thus, expert opinion still plays a role in the development of evidence-based criteria for the qualitative assessment of study methodology! Hopefully, an evidence-based guideline should present the criteria for assessing study methodology, allowing the reader to rate the strength of evidence underlying each recommendation. Is there a reproducible assessment of study methodology and study results? To limit bias when extracting data from individual studies, more than one investigator should independently extract data from the studies. After data extraction is completed, the 2 investigators might compare their results and report on the frequency of agreement between the 2 investigators. Thus, each investigator must ensure that he/she has performed a comprehensive and accurate data extraction. If this does not occur, there will be frequent disagreement between the data extracted by the 2 investigators. Optimally, study results might be summarized quantitatively in a meta-analysis. Meta-analysis combines data from multiple studies to provide a summary statistic of the outcome (e.g., odds ratio or relative risk). However, a meta-analysis might only be performed when very similar populations, interventions, outcomes, and study methodology have been used. Unfortunately, many studies use different methods, which minimizes the opportunity to combine study results quantitatively. If this occurs, the authors of systematic reviews should present data extraction in a tabular format and should point out the potential pitfalls in the conclusions from individual studies. Is there a transparent link between evidence from systematic reviews and guideline recommendations? For the reader to assess the evidence supporting guideline recommendations, there should be a transparent and explicit link between systematic review evidence and guideline recommendations. This process can be accomplished by numbering each section of the systematic review and linking it to a corresponding number in the guideline recommendations. For example, in the IBS guideline,7 each recommendation in the evidence-based guideline is numbered to a corresponding section in the IBS systematic review. Case Scenario: Part Two The IBS guideline7 starts with an evidence-based position statement that lists multiple clearly focused questions about the diagnosis and treatment of IBS. The guideline is followed by a series of systematic reviews July 2003 EVIDENCE-BASED IRRITABLE BOWEL SYNDROME 325 Table 2. Levels of Evidence and Grading of Recommendations Level I evidence: Randomized controlled trials with P values ⬍ 0.05, adequate sample sizes, and appropriate methodology Level II evidence: Randomized controlled trials with P values ⬎ 0.05, and inadequate sample sizes and/or inappropriate methodology Level III evidence: Nonrandomized trials with contemporaneous controls Level IV evidence: Nonrandomized trials with historical controls Level V evidence: Case series Grade A recommendations: Recommendations based on consistent evidence from multiple level I trials Grade B recommendations: Recommendations based on conflicting evidence from multiple level I trials or recommendations supported by consistent evidence from multiple level II trials Grade C recommendations: Recommendations based on Level III–V trials Data from Cook D, Guyatt G, Laupacis A, Sackett D. Chest 1992;102:305S. about the prognosis, diagnosis, and treatment of IBS. The methods section of the systematic review details explicit study selection criteria, comprehensive literature search techniques, and a quantitative assessment of individual study methodology. You also note that there is an explicit numbering system linking recommendations in the evidence-based position statement to specific sections in the systematic review. However, independent data extraction by 2 or more investigators was not performed. On the basis of your overall assessment of the strengths and weaknesses of the guideline development process, you decide that the design of the guideline appears appropriate for the question you are seeking to explore. Interpreting the Results If the methodology of the guideline is appropriate, the guideline recommendations should be credible. However, different systems have been proposed for rating levels of evidence and for grading recommendations in an evidence-based guideline. In this section, we provide guidance about systems used to rate levels of evidence and to grade recommendations. How did the authors rate levels of evidence? Although several systems have been proposed for rating levels of evidence, some common themes are present in most of these grading systems.18 In almost all systems, level I evidence refers to randomized controlled trials that demonstrate appropriate study methodology (e.g., use of double-blinding, use of concealed allocation, complete follow-up of the patients), have enrolled an adequate sample size of patients, and produce statistically significant results (Table 2). Thus, level I evidence represents evidence from highquality randomized controlled trials with statistically significant results and few limitations in their study design. Therefore, positive results from these trials should not be inappropriately inflated as a result of a biased study design (i.e., type I error), and negative results from these trials should not be due to enrollment of an inadequate number of patients (i.e., type II error). Level II evidence refers to randomized controlled trials that have important limitations in their study methodology (e.g., lack of adequate blinding) or did not enroll an adequate number of patients. These trials do have important limitations in their study design that could produce inflated estimates of treatment benefit (i.e., type I error) or could fail to demonstrate statistically significant results because of an inadequate sample size of patients (i.e., type II error). Inflated estimates of treatment benefit are potentially relevant to the appraisal of meta-analyses of IBS treatments, given the important limitations in study design of IBS therapeutic trials published before 1999. Level III–V evidence comes from nonrandomized trials with contemporaneous controls or historical controls or case studies (Table 2). Because randomized controlled trials are rarely used to assess the utility of diagnostic tests or to assess the natural history of a specific disorder, level III-V evidence usually is reserved to make recommendations about the prognosis and diagnostic approach to a specific disorder. Nevertheless, recommendations based on level III-V evidence are relatively weaker than the recommendations that are based on level I or level II evidence. How do authors grade recommendations? The strength of graded guideline recommendations should encompass the quality of study design and consistency of results across studies. Several systems have been proposed to grade recommendations in an evidence-based guideline. Table 2 provides one proposed system, which shares similarities with many of the proposed systems for grading guideline recommendations.18 Grade A recommendations are usually supported by level I evidence, appear to be accurate and unbiased, and should be implemented in patient care. Grade B recommendations are usually supported by level II evidence. These are recommendations supported by (at best) intermediate quality evidence, which might have important limitations. These recommendations could change in the future, especially if high-quality (level I) evidence from new studies be- 326 SCHOENFELD AND CHEY comes available. Grade C recommendations are usually based on level III-V evidence. These are recommendations based on observational studies, and the strength of evidence supporting these recommendations is quite limited and might only be appropriate for questions of prognosis or diagnostic approach, because no other data are typically available. Case Scenario: Part Three When reviewing the IBS guideline,7 you note a standard system for rating levels of evidence and for grading recommendations (Table 2). You note that all recommendations are linked with a specific grade. You also note that the guideline provides specific recommendations about the prognosis of IBS and the use of diagnostic tests and treatments. Furthermore, you see that each specific recommendation is followed by a short explanation about the evidence supporting the recommendation and a discussion about how to implement the recommendation. Can These Results Be Applied to Your Patients? Although an evidence-based guideline might be properly designed and might produce appropriate graded recommendations, guidelines must be implemented in the context of good clinical judgment. These recommendations are a guide, not a “cookbook” for medical management. Each individual patient represents a unique situation, which must be carefully considered before implementing guideline recommendations. Are your patients similar to the patients described in the guideline? An evidence-based guideline might be valid for a specific patient population. However, the recommendations might not be generalizable to all patients with a disorder. For example, the IBS guideline7 used a North American perspective, so this guideline only performed systematic reviews of medications available in the United States. Therefore, the only antispasmodic agents included in the guideline were anticholinergic agents, including hyoscyamine (Levsin) and dicyclomine (Bentyl). Three randomized controlled trials about these antispasmodic agents were poorly designed, showed minimal to no improvement versus placebo, and demonstrated significant adverse events.7 Because of the limited data, the IBS guideline7 states “insufficient data to make a recommendation about the effectiveness of the antispasmodic agents available in the United States (Grade B recommendation).” However, antispasmodic agents that directly affect intestinal smooth muscle, including mebeverine, otilinium, and pinaverium, are available CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 1, No. 4 outside the United States, and there are data from randomized controlled trials19,20 to suggest that these specific antispasmodic agents might be effective in the treatment of abdominal pain associated with IBS. Therefore, the recommendations of this guideline7 might not be generalizable to patients in countries in which these antispasmodic agents are available. How were values incorporated into the optimal course of action? The discussion of a graded recommen- dation should encompass many factors, including the quality of study design, the magnitude of positive treatment effect, consistency of effect across studies, frequency of adverse events, and costs of the medication. No formal system is available to incorporate these disparate factors into a graded recommendation. Therefore, the authors of a guideline should describe how these different factors were valued when making recommendations. For example, the IBS guideline7 provides straightforward statements about the effectiveness of different IBS treatments compared to placebo. Alosetron received a Grade A recommendation for being “more effective than placebo at relieving global IBS symptoms in female IBS patients with diarrhea.” The Grade A recommendation is based on results from well-designed studies that consistently showed that patients who used alosetron had significant improvement in multiple individual IBS symptoms and global IBS symptoms compared to patients who used placebo in 5 randomized controlled trials.21 However, alosetron has been associated with complications from severe constipation. Furthermore, the effectiveness of alosetron has only been demonstrated definitively among female IBS patients. Therefore, the authors of this guideline provided context for their recommendation and noted that alosetron use should be limited to “women with severe diarrhea-predominant IBS who have failed to respond to conventional therapy.” Through this process, a specific graded recommendation is followed by a description of how different values, including magnitude of treatment benefit and frequency of adverse events, are incorporated into an optimal course of management. Were important subgroups examined? Subgroups of patients might respond differently to similar therapies. Therefore, evidence-based guidelines should account for these differences. For example, loperamide did not improve global IBS symptoms, abdominal discomfort, or bloating among IBS patients, but it did improve stool frequency and consistency. Therefore, the authors of the IBS guideline7 stated that “Loperamide is not more effective than placebo at relieving global IBS symptoms (Grade B recommendation). . . . Loperamide is effective for diarrhea.” The authors of the guideline July 2003 also noted that tegaserod has only demonstrated consistent efficacy in women with constipation-predominant IBS and alosteron has only demonstrated consistent efficacy in women with diarrhea-predominant IBS. Therefore, the graded recommendations were limited to these subgroups of patients. Case Scenario: Part Four—Resolution Because your practice is in the United States, you determine that these recommendations will be applicable to your patients. You recognize that this guideline provides many specific recommendations that can be used by your fellow physicians for the diagnosis and treatment of IBS patients. Nevertheless, when you advise your clinic administrator to distribute this guideline to your colleagues, you add a cover letter emphasizing that each individual patient is unique and an evidence-based guideline only provides broad principles for the management of these patients. You also emphasize that each patient’s unique preferences about the diagnosis and treatment of their condition can only be fully explored within the context of a therapeutic patientphysician relationship. References 1. Field MJ, Lohr KN, eds. Clinical practice guidelines: directions for a new program. Washington, DC: National Academy Press, 1990. 2. Woolf SH, Gorl R, Hutchinson A, Eccles M, Grimshaw J. Potential benefits, limitations, and arms of clinical guidelines. BMJ 1999; 318:527–530. 3. Guyatt G, Sinclair J, Cook D, Glasziou P. User’s guide to the medical literature: how to use a treatment recommendation. JAMA 1999;281:1836 –1843. 4. Guyatt G, Rennie D. User’s guide to the medical literature. Chicago, IL, American Medical Association Press, 2002. 5. Hayward R, Wilson M, Tunes S, Bass E, Guyatt G. User’s guide to the medical literature: how to use clinical practice guidelines—a. are the recommendations valid? JAMA 1995;274:570 –574. 6. Wilson M, Hayward R, Tunes S, Bass E, Guyatt G. User’s guide to the medical literature: how to use clinical practice guidelines— b. what are the recommendations and will they help you in caring for your patients? JAMA 1995;274:1630 –1632. 7. Brandt LJ, Bjorkman D, Fennerty MB, Locke GR, Olden K, Peterson W, Quigley E, Schoenfeld P, Schuster M, Talley N. Systematic review on the management of irritable bowel syndrome in North America. Am J Gastroenterol 2002;97:S7–S26. 8. Veldhuizen van Zanten S, Talley N, Bytzer P. Design of treatment trials for functional gastrointestinal disorders. Gut 1999; 45(suppl 2):II69 –77. 9. Oxman A, Cook D, Guyatt G. User’s guides to the medical literature: how to use an overview. JAMA 1994;272:1367–1371. EVIDENCE-BASED IRRITABLE BOWEL SYNDROME 327 10. Egger M, Smith G. Principles of and procedures for systematic reviews. In: Egger M, Smith G, Altman D (eds): Systematic reviews in health care. 2nd ed. London: BMJ, 2001;23– 43. 11. Schoenfeld P, Scheiman J. An evidence-based approach to studies of gastrointestinal therapies. Clin Gastroenterol Hepatol 2003;1:57– 63. 12. Schoenfeld P, Guyatt G, Hamilton F. An evidence-based approach to gastroenterology diagnosis. Gastroenterology 1999;116: 1230 –1237. 13. Eloubeidi MA, Wade SB, Provenzale D. Factors associated with acceptance and full publication of GI endoscopic research originally published in abstract form. Gastrointest Endosc 2001;53: 275–282. 14. Jadad AR, Moore RA, Carroll D, Jenkinson C, Reynolds DJ, Gavaghan DJ, McQuay HJ. Assessing the quality of reports of randomized clinical trials: is blinding necessary? Control Clin Trials 1996;17:1–12. 15. Thompson WG. The treatment of irritable bowel syndrome. Aliment Pharmacol Ther 2002;16:1395–1406. 16. Cash B, Schoenfeld P, Rex D. An evidence-based medicine approach to studies of diagnostic tests: assessing the validity of virtual colonoscopy. Clin Gastroenterol Hepatol 2003;1:136–144. 17. Lewis J, Schoenfeld P, Lichtenstein G. An evidence-based medicine approach to studies of the natural history of gastrointestinal diseases: recurrence of symptomatic Crohn’s disease after surgery. Clin Gastroenterol Hepatol 2003;1:229 –236. 18. Cook DJ, Guyatt G, Laupacis A, Sackett D. Rules of evidence and clinical recommendations on the use of anti-thrombotic agents. Chest 1992;102(suppl):S305–S311. 19. Poynard T, Reginbeau M, Behamou Y. Meta-analysis of smooth muscle relaxants in the treatment of irritable bowel syndrome. Aliment Pharmacol Ther 2001;15:355–361. 20. Jailwala J, Imperiale T, Kroenke K. Pharmacologic treatment of the irritable bowel syndrome: a systematic review of randomized controlled trials. Ann Intern Med 2000;133:136 –147. 21. Cremonini F, Delgado-Aros S, Camilleri M. Efficacy of alosetron in irritable bowel syndrome: a meta-analysis of randomized controlled trials. Neurogastroenterol Motil 2003;15:79 – 86. Address requests for reprints to: Philip Schoenfeld, M.D., MSEd, MSc (Epi), Division of Gastroenterology (111D), VA Ann Arbor Healthcare System, 2215 Fuller Road, Ann Arbor, Michigan 48105. e-mail: pschoenf@umich.edu; fax: (734) 761-5260. The opinions and assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the Department of Veterans Affairs. Dr. Schoenfeld is a consultant for Novartis Pharmaceuticals Corporation, Proctor and Gamble, and AstraZeneca LP. Dr. Schoenfeld is on the Speakers Bureau for AstraZeneca LP, Wyeth Laboratories, TAP Pharmaceuticals, Novartis Pharmaceuticals Corporation, Boehringer Ingleheim, and Merck. Dr. Chey is a consultant for Novartis Pharmaceuticals Corporation, TAP Pharmaceuticals, and AstraZeneca LP. Dr. Chey has received research support from Novartis Pharmaceuticals Corporation, TAP Pharmaceuticals, AstraZeneca LP, and Janssen. Dr. Chey is on the Speakers Bureau for Novartis Pharmaceutical Corporation, AstraZeneca LP, TAP Pharmaceuticals, and Janssen. The authors thank Michael Camilleri, M.D., and Loren Laine, M.D., for contributing ideas to the content of this article and for editorial support.