Recent increases in male rectal gonorrhea and sexual risk behavior among men who have sex with men (MSM) in San Francisco, USA  raises the worrisome prospect that the incidence of HIV may also be increasing. Unfortunately, traditional methods (i.e. cohort studies) to detect HIV seroconversions and estimate the incidence of HIV are logistically complex, expensive, time-consuming, and prone to bias through selective loss to follow-up. Data documenting HIV seroconversions are scant, and current HIV and AIDS surveillance information is likely to lag behind actual transmission by several years.
We recently reported a method of estimating incidence using HIV counselling and testing data. The method is based on identifying seroconversions among repeat testers using self-reported dates of previous HIV-negative tests. The validity of the method is supported by consistency with results by other methods and by the observation that MSM test on a frequent, regular basis. The method provides a simple means of estimating local HIV incidence using data readily available in most health jurisdictions. The method can also be expanded to track HIV seroconversions over time and estimate recent HIV incidence in a timely fashion. We applied the method to HIV counselling and testing data collected between 1995 and 1998.
During the study period, 31 500 anonymous HIV tests were performed in San Francisco, USA, of which 13 125 (42%) were among MSM with at least one previous test. We excluded testers whose date of last test was missing (n = 578), or whose last test results were not recorded (n = 246), were HIV positive (n = 79), inconclusive (n = 18), or not known to the tester (n = 272). We also excluded persons whose last test was more than 5 years ago (n = 606) or was within the previous 2 months (n = 107). A seroconverter was defined as an individual who reported their last test to be HIV negative and whose current test was HIV positive. The period of observation, in person-years (py), was approximated as the interval of time between the two tests. When only the year of the last test was reported, we assumed the month to be June (n = 531). Incidence estimates were calculated separately for each of the four calendar years from 1995 to 1998.
The number of HIV seroconversions identified was highest in 1995 (n = 110) and lowest in 1997 (n = 41). Figure 1 displays HIV incidence estimated by year. HIV incidence declined from a high of 2.8 per 100 py in 1995 to a low of 1.2 per 100 py in 1997. The overall decline in HIV incidence among repeat testers from 1995 to 1998 was significant (chi-square test for trend, P = 0.002). However, the increase in HIV incidence from 1997 to 1998, the most recent year of data available, was also significant (chi-square test, P = 0.013).
We recognize several limitations of the data and methods employed. First, HIV seroconversion and the exposure period are self-reported and are therefore vulnerable to imprecision and potential biases not present in traditional cohort studies. A second limitation is that we assigned HIV seroconversions to the year they were detected. Actual transmission may have occurred earlier. However, 50% of subjects reported previous tests within the preceding 12 months, and less than 5% reported previous tests beyond the past 5 years. Moreover, this limitation applies equally to all years in our study and thus should not affect the temporal trend. Third, trends in HIV incidence estimated by our method may reflect changes in testing patterns as well as changes in the underlying incidence in the target population. Finally, data represent MSM who repeatedly seek publicly funded anonymous testing, and do not include MSM who do not test, test only one time, or test in non-public venues.
Nevertheless, publicly funded anonymous programmes test the largest number of persons at risk and diagnose more HIV infections than any other source in the country. Because of the difficulties in estimating the incidence of HIV, no one data source can be assumed to be accurate. Nonetheless, the upturn in HIV incidence in 1998 among repeat anonymous testers in San Francisco is worrisome because it is consistent with an increase in rectal gonorrhea and sexual risk behavior reported among MSM.
1. Centers for Disease Control and Prevention. Increases in unsafe sex and rectal gonorrhea among men who have sex with men - San Francisco, California, 1994-1997. MMWR
2. McFarland W, Kellogg TA, Dilley J, Katz M. Estimation of human immunodeficiency virus (HIV) seroincidence among repeat anonymous testers in San Francisco. Am J Epidemiol
3. Kalichman SC, Schaper PE, Belcher L, et al
. It‚s like a regular part of gay life: repeat HIV antibody testing among gay and bisexual men. AIDS Educ Prevent