Given that the full text has already appeared in print, why publish a shorter version of this guideline in this issue [2]?

First, the whole point of creating and distributing clinical guidelines is to bring the widest possible consistency and rationality to the diagnosis and management of disease. This stands in marked contrast to the purpose of original research papers, which is to make highly selective, incremental contributions to the body of underlying scientific knowledge and debate. Looked at this way, publication of the ITP guideline in Annals seems logical and useful because a broad spectrum of regular Annals readers (including many generalists, most of whom do not regularly read Blood) are involved in the diagnosis and management of ITP. Second, much of the material in the full guideline concerns ITP in children and pregnant women. The version of the guideline published in this issue contains only the material from the full document that is most directly relevant to the practice of internal medicine, further modified and adapted in response to external Annals peer review. Interestingly, material from the full document that is pertinent to pediatrics and obstetrics is being excerpted and prepared for publication in journals of these disciplines, a new departure in collaborative development and distribution of guidelines that seems both responsible and efficient.

The publication of this guideline at this time reflects the convergence of several current trends in medicine. First, clinical guidelines and related statements on diagnosis and management are increasingly common as physicians attempt to define standards of practice with a goal of improving the quality of patient care [3]. Second, medical specialty societies are increasingly aware of their role as sources of guidance for clinical care. Adoption of this stance has had a positive influence because clinical guidelines issued by a specialty society gain greater acceptance among the society's physician members than do guidelines from other sources [4]. Third, interest in clinical guidelines comes at a time when the science of evaluating, reporting, and using clinical data is developing rapidly. Rigorously designed clinical trials are supplementing and succeeding anecdotal observations and descriptions of case series [5], the reporting of clinical trials is becoming more standardized [6], and analyses that use sophisticated methods to find and synthesize published data are beginning to supersede review articles representing personal opinion [7].

The process by which the practice guideline on ITP was developed, and the criticism the guideline has received, may be as important and as educational as the recommendations themselves. Idiopathic thrombocytopenic purpura was chosen because it is clearly identified with the specialty of hematology, is a relatively common condition, affects both children and adults, and encompasses important unresolved issues in diagnosis and management that have major cost implications. Such diagnostic procedures as platelet antibody studies and bone marrow aspiration are of uncertain value, indications for hospitalization and specific treatments are unclear, and even the goals of treatment are undefined. To address these issues, a panel was assembled that represented internal medicine and pediatric hematologists who work in both university and private practice settings. This composition ensured a diverse panel membership and provided expertise from physicians who have devoted research careers to the study of ITP as well as from physicians who have gained practical insights from clinical practice. As the work of the panel progressed, the value of this diversity of perspective became increasingly obvious.

Although the panel was initially committed to developing an evidence-based guideline, it was soon clear that evidence was insufficient to support strong or definitive recommendations. This itself is an important revelation and a powerful criticism of decades of clinical research and publication. Although this conclusion may be viewed as "negative," it is important for clinicians to appreciate that the diagnosis and management of ITP, as presented in current textbooks and reviews, are based not on firm data but rather on a compilation of case series that lacked controls, the weakest form of clinical evidence. As difficult as it is to believe, no firm data are available on the incidence of ITP, the natural history of the condition, the mortality rate and frequency of major bleeding episodes associated with ITP, and the relative risks and benefits of different treatments. And ITP is a relatively common and easily recognized disease! How can this be, in 1997? Part of the explanation may lie in the relatively benign nature of the disease. Despite the occurrence of (often) frighteningly low platelet counts, mortality is thought to be low and many treatments appear to have some efficacy, at least as measured by transient increases in the platelet count. The revelation that no firm data are available provides insight and, paradoxically, some comfort for physicians who feel insecure about their decisions on the diagnosis and management of patients with ITP. Physicians must recognize that, at this time, there are no clear, correct answers to most important management questions.

Having come to this realization, the panel faced a critical issue: Should it admit that the task of developing a clinical guideline on ITP is impossible and defer any recommendations until firm evidence is available? Can recommendations based merely on opinion be of real clinical value? Could opinion-based recommendations jeopardize the decisions of physicians whose opinions differ from those of the panel? The panel decided: It resolved to describe the state of the literature, emphasize the absence of evidence and the limitations of opinion-based recommendations, issue recommendations based on an explicit method of assessing opinion that clearly documented the strength and variance of opinion, and define priorities for future research. In this phase of its work, the panel faced the vivid reality that the absence of data bears no relation to the strength of opinion; varieties of clinical experience forge very strong opinions indeed.

The initial positive responses to the ITP clinical guideline have focused on the objectivity of the review and interpretation of the literature. Physicians have gained support for doing fewer diagnostic tests and for not treating asymptomatic patients. The initial negative responses have focused on recommendations in the guideline that differ from the practices of other experienced and respected physicians and on the lack of recommendations for such management dilemmas as the treatment of patients who have persistent, symptomatic thrombocytopenia after splenectomy. But these issues could not be resolved with evidence by the panel and cannot now be resolved by debate. However, even these negative responses support the importance of this guideline effort. Management dilemmas in ITP can only be resolved by well-designed clinical trials assessing the most important clinical outcomes: bleeding and death. The American Society of Hematology hopes that the clinical guideline can serve as a preliminary effort that defines important questions for future research and that this research will bring more clarity and confidence to the management of patients with ITP.

Dr. Davidoff: Annals of Internal Medicine, American College of Physicians, Independence Mall West, Sixth Street at Race, Philadelphia, PA 19106.