Will microbicide trials yield unbiased estimates of microbicide efficacy?

Objective: To evaluate the effect of adherence and condom use on apparent efficacy of microbicides. Design: Hypothetical trial designs and scenarios. Methods: Mathematical calculations of effectiveness. Results: In a randomized trial of a candidate microbicide and a placebo, nonuse of the microbicide will result in underestimation of microbicide efficacy, with the magnitude of this difference between effectiveness and efficacy increasing directly with file level of microbicide nonuse. Adding condoms to the trial will not change this expected result as long as use of condoms and microbicide is independent, and microbicide use is the same in the trials with and without condoms. However, if microbicide use is lower in the trial with condoms, then effectiveness will be lower than in the trial without condoms, with the magnitude of the difference between effectiveness and efficacy being even greater. Moreover, condom and microbicide use may not be independent. If participants tend to use condoms rather than nothing, the trial result will more closely approximate microbicide efficacy. If, however, participants substitute condom use for microbicide use, then the expected estimate of effectiveness will less closely approximate microbicide efficacy and could be closer to or further away from the expected estimate of effectiveness than in the trial without condoms. In another trial design, where there is either simultaneous use of microbicide and condoms or no use of either (guaranteed when condoms are packaged with either a microbicidal or a placebo gel), expected effectiveness will fall short of microbicide efficacy. If nonuse is the same in a trial without condoms and a trial with microbicidally lubricated condoms, then the trial with condoms will produce an estimate of effectiveness that less closely approximates microbicide efficacy than would the trial without condoms. If there is less nonuse in the trial with condoms, then the expected estimate of effectiveness will more closely approximate microbicide efficacy and could be closer to or further away from the expected estimate of effectiveness than in the trial without condoms. Conclusions: Nonuse of a microbicide and use of condoms may seriously impair our ability to identify an effective microbicide. (c) 2005 Elsevier Inc. All rights reserved.