FDA CLINICAL GUIDELINE ON NDA STATISTICAL SECTION ALLOWS BRIEF REPORTS OF EFFICACY DATA FOR FLAWED STUDIES; FULL SAFETY REPORTS REQUIRED

NDA sponsors should report efficacy data for flawed or unsupportive studies in an abbreviated fashion, according to FDA's just-released final guideline on the Format and Content of the Clinical and Statistical Sections of NDSs. "For completed studies or ongoing studies with interim analysis that on their face do not support effectiveness, for favorable studies that the sponsor considers flawed or unreliable, and for incomplete studies no longer active, a briefer description of the results related to effectiveness should be provided in addition to the brief summary of the study design and conduct," the guideline states. The final guideline combines the draft versions for the statistical section of an NDA, published in 1985, and the draft guideline for the clinical section, published in 1986. Under the draft guidelines, FDA required full reports of unsupportive and flawed studies. Although FDA will allow efficacy reports on unsupportive or flawed studies to be abbreviated, the agency noted that the description of results should be "sufficient to convey the outcome of the study quantitatively, even if no statistical analysis was performed." The guideline also notes that "the reason for providing only a brief report should be given, if not apparent (e.g., the flaws that prevent the study from being relied upon)." In an Oct. 7 Federal Register notice announcing availability of the guideline, FDA explained that it was giving "greater discretion to applicants in providing full reports of controlled studies that are either non-supportive or flawed" because of the questionable benefits of requiring a full report. "While there are circumstances in which a close review of an obviously negative study is useful (e.g., a detailed analysis of procedures, study population, or concomitant therapy may provide an explanation of why the study was non-supportive), this should be a matter of discretion because the search for such explanations is often fruitless." Unsupportive or flawed studies should be "discussed and analyzed" in the section summarizing efficacy data, according to the guideline. In addition, "there should be exploration of the reasons for such 'negative' or non-supportive studies, as they may suggest influences of study populations, dosages, study designs or other features that need to be understood," FDA stated. The guideline, however, requires that sponsors submit a full report of safety data from all trials, including those that are "flawed" or unsupportive. "Safety-related data should be presented thoroughly . . . for all these studies, even where a full analysis of effectiveness is not included," FDA said. In an introductory section discussing principles that have shaped the guideline, FDA pointed out that "interrelationships among data from different studies should be examined." The agency noted that "one consequence of this [principle] is rejection, for the purposes of format, of the concept of the 'pivotol' study, which has all to often been nothing more than the one or two of a group similar studies that worked out best." Studies "should be grouped by design and, within design, by other relevant features, such as whether case records are available and where the study was carried out," FDA explained. "The implications of all studies, successful, non-supportive, terminated, etc., need to be considered." A second consequence, FDA continued, "is that summaries are called for at a variety of levels, ranging from the broadest (overall summary, integrated summary of effectiveness data, integrated summary of safety information) to more narrow summaries (summary of clinical pharmacology studies)." The guideline is available from FDA's Legislative, Professional and Consumer Affairs Branch (HFD-365).