Name-Based Surveillance for HIV-Infected Persons

TO THE EDITOR:

Osmond and colleagues (1) found that HIV-infected persons who tested anonymously reported that they personally notified as many sex and needle-sharing partners (mean, 3.9) as persons who tested confidentially and notified partners with the help of health department workers (mean, 3.8). From these results, Osmond and colleagues argue that patient referral may be as effective as health department–assisted referral for notifying partners. A methodologic flaw in their study precludes this conclusion.

The authors did not validate the retrospective claims of their patients with AIDS about how many partners were notified after HIV diagnosis, an event likely to have occurred several years before AIDS diagnosis and the study survey. In contrast, Landis and colleagues (2) conducted a randomized trial comparing patient referral to health department–assisted referral. They measured the number of partners notified in each group directly from health department contact-tracing records, finding that seven times more partners were notified with health department–assisted referral than with patient referral. Many reports indicate that persons with HIV infection are unlikely to notify their partners. In one study (3), for example, 50% of HIV-infected persons had not notified any past sex partner 12 months after initial HIV diagnosis; in another (4), 90% had not notified any partner 3 months after initial HIV diagnosis. Even after 2 years of repeated counseling to notify partners, 30% of participants in a third study (5) still had not notified any past sex partner. By extension, if one were to assume that, say, 50% of Osmond and colleagues' patients who chose to test anonymously did not in fact notify their partners, the other 50% must have notified a mean of 7.8 partners, a highly implausible scenario. Given the countervailing evidence, Osmond and colleagues' study reminds us that unvalidated data cannot be trusted in evaluating partner notification approaches.