HIV Infection and Smoking Behavior

To the Editor. —Halsey et al¹ reported an association between cigarette smoking and the risk of HIV-1 infection in Haitian women that remained after adjustment for "all known risk factors." The authors suggest two alternative explanations for their results: First, that there is a "hidden" or "unrecognized" confounder that has not been accounted for and, second, that there is a biologic effect of cigarette smoking on the risk of HIV transmission.

We would like to suggest a third possibility, namely, that the association is not due to an unknown confounder, but rather the inability to properly measure a well-known confounder—frequency of unprotected sexual intercourse with an HIV-infected man or men. The dichotomous variable "≥3 lifetime sexual partners" that was used in the study is only a poor proxy measure of this true confounder and thus considerably underestimates its association with risk of HIV infection.^2-4 The underestimation of the