ANTIBIOTIC LABEL INDICATIONS FOR PROPHYLACTIC USE IN SURGERY REQUIRE STATISTICALLY SIGNIFICANT STUDY RESULTS DEMONSTRATING EFFICACY, FDA CMTE. SAYS

Claims for prophylactic use of antibiotics in surgery should be allowed as labeled indications only if studies show statistically significant efficacy in preventing infection after specified surgical procedures, FDA's Anti-Infective Drugs Advisory Cmte. agreed at its Oct. 29 meeting. The cmte. made its recommendation in response to questions from FDA on the use of antibiotics as prophylactics in surgery. Specifically, FDA asked the cmte.: "If the positive control studies for antibiotics in surgical prophylaxis are not large enough to achieve statistical significance, should we give the claim without an efficacy study if the antibiotic has been proved to be effective in the treatment of the expected organisms?" FDA Anti-Infective Drugs Div. Director Edward Tabor, MD, told the cmte. that the division has "not been extrapolating ]prophylactic efficacy[ from therapeutic efficacy, but . . . wanted to discuss the issue." He explained that "when you're doing a positive controlled study for prophylaxis in a situation where you have a very los rate of infection, you really need a large number of patients in each arm of the study to achieve statistical significance." Sometimes that's very difficult, Tabor said, "however, if you feel it is not reasonable to extrapolate from the therapeutic efficacy of the antibiotic, that's what we want to hear." The cmte. agreed that a labeled indication for prophylaxis should require statistically significant studies and that prophylactic efficacy can not be inferred from therapeutic efficacy. Cmte. Chairman Sarah Cheeseman, MD, University of Massachusetts Medical School, stated that it is "a very significant intellectual error extrapolating ]prophylactic efficacy[ from therapeutic efficacy." As an example Cheeseman cited "the collerectal surgery situation, ]in which[ metronidazole is not effective against many of the organisms isolated from wound infection ]but[ is effective in surgical prophylaxis." Similarly, Cheeseman continued, "Cephalothin has a spectrum of action almost identical to cefazolin, and yet one has been effective and one not in prophylaxis of collerectal surgery. I think that demonstrates that prophylaxis is a special form of use of antimicrobials in which success cannot necessarily be predicted on the basis of in vitro microbiological . . . data and the organisms expected." That, she explained, "is my hang-up with inferring . . . efficacy for prophylaxis," from therapeutic efficacy data. The cmte. was also asked whether prophylaxis in catastrophic surgery (e.g., open heart surgery) should "be allowed as a claim with no data, if the drug works in treatment of the expected organisms?" The cmte. unanimously voted against allowing such a claim. As a preface to the question, Tabor explained that "most of the studies submitted involve a relatively small number of patients, positive control -- because of the perceived ethics of not using any antibiotic -- and many of the studies really don't achieve statistical significance." In addition, he noted, "they involve comparisons with drugs for which there's never really been a real placebo controlled study" showing statistical significance. Expanding on Tabor's comments, FDA Anti-Infective Drug Products Div. Group Leader George Stanley, MD, stated that the first approval of an indication for surgical prophylaxis, Lilly's Keflin, was based on data which did not reach statistical significance. "Everything since then has been built on this house authority and we're left with -- how do we practically handle this?" Stanley said. "I mean," he added, "if you come in now and say we have to have statistical significance everything that has gone before really needs to get thrown out the window." Stanley also reminded the cmte. that antibiotic prophylactic surgical labeling states that "drug X may be effective prophylactically," whereas "every other claim we approve says drug X is effective in the following." When the prophylactic approvals began in the late 1970s, Stanley said, "the cmte. designed that ]labeling statement[ intentionally because there was nobody proving at the time that it was effective -- and we've lived with that labeling ever since." While the cmte. did not suggest that existing prophylaxis indications be revoked, it did recommend that statistical significance be the standard for subsequent approvals of prophylaxis indications. Speaking for the cmte., Cheeseman said: "We understand as a practical point that you can't take back very much that's been given. But, I think we're supporting you in saying that you should not be giving indications that don't have a trial sufficient to prove ]prophylactic efficacy[, with reasonable power and the appropriate statistical comparisons and all of the considerations on sample size, that is at least as good as what you have already approved." Cheeseman emphasized that a labeled indication does not necessarily dictate medical practice. Requiring such studies for approval, she said "does not stand in the way of medical practice which may want to use an antibiotic for some procedure which is not feasible to do a study on -- but it just won't say it in a labeled indication."