PURPOSE: To evaluate the measures of community human immunodeficiency virus (HIV) viral load (VL) and the association with HIV incidence among people who inject drugs (PWID).
METHODS: Data were from 1986 to 1999 Urban Health Study conducted among PWID in the San Francisco Bay Area. Extant measures of community VL use mean VL among HIV + study participants, not accounting for the proportion of HIV- individuals. We compared the strength of the associations between HIV incidence and the traditionally measured mean community VL and a new prevalence-adjusted community VL, calculated by dividing the sum of VL among HIV + participants by the total participants irrespective of HIV status.
RESULTS: Mean community VL was not correlated with HIV incidence in this sample of PWID (rs = 0.32, P = .28). However, prevalence-adjusted community VL was strongly correlated with HIV incidence (rs = 0.69, P = .009). Nested complimentary log-log linear models indicated that increases in community VL and prevalence-adjusted community VL were both associated with HIV incidence, but prevalence-adjusted community VL was a more sensitive measure (hazard ratio = 1.28, P = .038 and hazard ratio = 3.29, P < .001, respectively).
CONCLUSIONS: The effect of community VL on HIV incidence may be stronger than previously reported. Future studies of community VL surveillance should consider accounting for the prevalence of HIV using a prevalence-adjusted community VL measure.