Human immunodeficiency virus transmission and the role of other sexually transmitted diseases. Measures of association and study design.

BACKGROUND

Studies have reported that infection with certain sexually transmitted diseases (STD), such as chancroid, gonorrhea, and genital herpes, enhances the probability of human immunodeficiency virus (HIV) transmission by sexual contact with an infected partner. Epidemiologic studies vary in design from longitudinal to cross-sectional, with varying periods of follow-up or retrospective history of exposure to STD. The major difficulty in assessing the results of past work centers on the validity and precision of widely used measures of association, such as relative risk (RR) and odds ratio (OR), in situations in which common behavior (e.g., different facets of sexual behavior) underpins the acquisition of both the STD cofactor and HIV.

GOAL

To evaluate the quality of the cumulative incidence ratio (CIR), the hazard rate ratio (HRR), and the prevalence ratio (PR) as measures of association to estimate and test the increase in HIV transmission probabilities caused by the STD.

STUDY DESIGN

The study is based on a proportional hazard stochastic model of concomitant HIV and STD cofactor transmission. Analysis was performed using Monte-Carlo simulation.

RESULTS

Estimates of the HIV-STD association by the CIR, HRR, and PR, adjusted and nonadjusted for sexual activity, are shown to have poor validity and great variability. The adjusted CIR, HRR, and PR tend to underestimate the strength of the true association (specified in the model) in both longitudinal and cross-sectional designs. In the absence an HIV-STD association, the PR tends to overestimate the magnitude, whereas the CIR and HRR may either underestimate or overestimate it in longitudinal studies. These results have direct consequences on the reliability of the test of association showing both a lack of specificity (empirical type I error) and sensitivity (empirical power). Some reasons contributing to the bias in the estimates of the measures of association are the presence of confounding variables, namely the frequency of change of sex partner and the mixing pattern between sexual activity classes, as well as the adopted definition of exposure to the STD cofactor and the prevalence of both HIV and the STD cofactor.

CONCLUSIONS

The precision of estimates and reliability of the test of HIV-STD association could be improved through longitudinal studies using more careful definition and measurement of exposure to the STD cofactor and larger sample sizes permitting finer stratification of sexual behavior and a sufficient number of persons per stratum.