LETTERS TO THE EDITOR
The authors respond:
The airing of the complex issues surrounding the design of vaginal microbicide trials for HIV prevention is important, and we appreciate the comments provided by van de Wijgert and colleagues 1 on our recent discussion of the choice of control arms in these studies.
The authors bypass one important issue we raise, namely, the lack of economy in using different placebos in the several large trials concurrently in development. However, they do clearly set out several of the points we make against the inclusion of a “condoms–only” control arm. Our prime concern is with the validity of the results of a randomized controlled trial of experimental microbicides, and we aimed to show that the introduction of the condoms-only arm undermines the well-established principles of a valid randomized and blinded study. Our critics do not address the main burden of our critique; indeed, they ignore altogether this fundamental issue. Instead, on the basis of possible ancillary results, they defend the inclusion of a “condoms-only” control arm that we show to be both inappropriate and confusing if it is to address the central hypothesis of the trial regarding HIV prevention. Although secondary substudies may be of importance, these can be addressed using more appropriate and parsimonious alternate designs.
We find particular cause for concern with the final argument of van de Wijgert and colleagues, 1 as exemplified by the following two points. First, they surmise that a possible result of microbicide use could be “the migration”—or better, the emigration—from condom use to reliance on the untested gel as an alternative, with a consequent reduction in safety from infection. Such secondary, if potentially important, questions are better answered by small inexpensive observational studies designed for that specific purpose.
Second, were a randomized but unblinded comparison of microbicide compared with condoms-only to suggest a lower incidence of HIV in the microbicide arm, the authors posit that “few would disagree that it is a worthwhile investment for public health policy makers.” We do disagree, and expect that many others would as well. This argument of the authors flouts the prerequisite of internal validity in epidemiologic studies; they seem ready to mingle any evidence to support the desired result regardless of its quality. In our view, the mere hope of a subsidiary result of possible value to public health is no justification for deploying the skilled personnel and the large-scale resources involved in mounting a relatively large and expensive but faulty randomized controlled trial.
On the other hand, there is certainly justification for such a trial given rigorous design. Designs should always be fitted to address the question posed. The traditional, individual randomized controlled trial is not always the most appropriate and feasible design to address every research question. Our critics, in justifying the inclusion of a condoms-only arm in a microbicide trial, make subsidiary aims paramount. For the purposes of evaluating the effectiveness of microbicides for HIV prevention, a comparison based on a condoms-only control is simply an impostor that assumes the cloak of rigor associated with the randomized double-blind trial.
1. Jones H, van d Wijgert J, Kelvin E. The need for a “condoms-only” control group in microbicide trials [letter]. Epidemiology. 2003; 14: 505.