Biologic Versus Behavioral Endpoints-The Duet Continues

EDITORS LOVE CONTROVERSY. Publishing papers that are contentious and raise issues that lead directly to further research to help solve a problem can be … delicious. About five years ago, Zenilman and colleagues published a study performed at sexually transmitted disease (STD) clinics in Baltimore, that questioned the validity of self-reports of condom use.1 In that study, higher STD infection rates were reported among those who had said that they always used condoms, than among those who reported less condom use. Immediately after that paper appeared one could hear the collective moaning of behavioral scientists who felt that their field had been gored. I was asked, “How could you publish that paper?” (How could we not? It was a well-done study with real results.) Or, “Who could have reviewed a paper that was so flawed from a behavioral viewpoint?” Others raised the issue, “Why is Zenilman, who is not a behavioral scientist, publishing papers dealing with behavioral issues?” Clearly a case of the message is bad-shoot the messenger. That paper lead to a dramatic polarization in planning and discussing behavioral intervention studies aimed at reducing STD/HIV incidence. On one side there was an absolute defense of behavioral endpoints. The opposite side felt that only biologic endpoints should be used as outcomes. As is typical, both extremes were probably wrong. However, the controversy led to research with a much finer focus on exactly what is meant by condom use; as a result, we have learned much. For example, there are questions on use-effectiveness with condoms-simply saying that one wears a condom doesn't answer the question of whether it is being used correctly. As is pointed out in the Peterman paper and the Fishbein editorial, a focus on when condoms are being used (or with which partner) is also important.2,3 There may be a tendency toward regular condom use with low-risk partners and irregular condom use with higher-risk partners as a function of the setting in which the partners are contacted. Certainly, experts would cite other new findings, but as a minimum these two points have important ramifications for future studies and for control efforts. Thus, the action and reaction following the publication of the Zenilman paper has had a salutary effect in moving the field forward. It is clear that to make subsequent analyses more robust, more research is needed on how to validate responses to questions dealing with behaviors. The choice of appropriate outcome measures for behavioral interventions to reduce STD/HIV incidence is a complicated issue.4 But it is also clear that it is inappropriate to demand biologic endpoints for behavioral interventions in their initial evaluation. The prevalence of STDs, and their transmission rates, are likely to be low enough in virtually all settings, so that risky behaviors need not result in acquisition of STDs. The sample size and associated costs of piloting behavioral intervention studies using biologic endpoints would be enormous and would restrict our ability to do behavioral research. It is equally true however, that before an intervention is put into practice, it should be validated by a biologic endpoint (i.e., a reduction in STDs). This should be true whether the ultimate goal is reduction of STDs generically (a most laudatory goal, in and of itself), or HIV, specifically. The fact that bacterial STDs might not be a perfect model for HIV infection does not mean they should not be used in operational research to assess efforts to reduce sexual transmission of HIV. HIV incidence is too low in most populations for the intervention to be readily assessed and an insistence on having a perfect model before accepting a result is too rigid. In the effort to reduce the incidence of HIV, an intervention that reduced the incidence of bacterial STDs would be considered an encouraging result worthy of further exploration, while an intervention that failed to reduce the incidence of bacterial STDs should certainly raise a red flag about its likely success in reducing HIV incidence. The controversy about behavioral and biologic endpoints in intervention studies is far from over. How to validate behavioral responses, when they will be sufficient end-points, and when and which biologic outcomes should be accepted as endpoints will continue to be the focus of much research. Recent research on diagnosis of chlamydial and gonococcal infection has mitigated previous concerns about the validity of biologic endpoints.5 Biologic outcomes for these bacterial STDs are far more easily obtained with the combination of noninvasive specimen collection (urine or vaginal/introital swabs) and nucleic acid amplification tests providing diagnostic methods that are more sensitive and specific than previous nonculture tests and that allow efficient screening of asymptomatic individuals.6,7,8 It is clear that STD/HIV control will depend on a combination of behavioral and biologic studies. Thus, advances in both areas are extremely important and the dialog has been useful. These pages will be open for further discussion on the subject.