Background:  US Centers for Disease Control and Prevention (CDC) guidelines recommend annual HIV testing for men who have sex with men (MSM). We sought to evaluate adherence to these guidelines and to determine their effectiveness at reducing the number of MSM with undiagnosed HIV infection.

Methods:  We analyzed data collected from MSM during 2003 to 2005 in the National HIV Behavioral Surveillance system (NHBS), a multi-city survey of HIV risk behaviors. We restricted our analysis to a subset of men in 5 cities who were offered HIV testing at the time of interview. We examined 2 testing strategies for reducing the number of MSM with undiagnosed HIV infection:  annual HIV testing for those who had tested previously, and first-time testing for those who had never tested before. For each testing strategy, the number of HIV-infected men who were undiagnosed at the time of interview was estimated to be the sum of the number who did not obtain their last test result (estimated to be 4.8% of those testing based on NHBS data) and the number who became newly infected with HIV after their last HIV^– test (estimated to be half the annual HIV incidence in those testing).

Results:  Of 1749 MSM interviewed, 1578 (90%) had previously tested for HIV; 258 (16%) reported that they were HIV⁺, 1250 (79%) HIV^–, and 70 (4%) had not obtained their last test result. Excluding the 258 men who reported that they were HIV⁺, 651 (44%) of 1491 had not tested within the past year. HIV tests conducted at the time of interview on all 1749 participants showed that 495 (28%) were actually HIV⁺; 237 (48%) of whom were newly diagnosed. Of the 237 newly diagnosed men, 182 (77%) had previously tested for HIV and 55 (23%) had not. If the men who had previously tested for HIV had tested annually, the proportion of HIV-infected men who were undiagnosed at the time of interview would have been 16% instead of 48% (χ² p <0.001). In comparison, a strategy of testing those who had never tested before would have only reduced the proportion of men with undiagnosed HIV infection from 48% to 37% (χ² p = 0.001).

Conclusions:  Despite CDC recommendations, nearly half the MSM who were at risk for HIV infection had not been tested within the past year. If they had, the large proportion with undiagnosed HIV infection would have been greatly reduced. Regular HIV testing is critical for reducing undiagnosed HIV infection and should be emphasized in testing campaigns for MSM.