Developing screening indexes to predict HIV infection in men who have sex with men and injection drug users in the US
D.K. Smith1, S.L. Pals1, S.H. Mehta1, A.C. Voetsch1, J.H. Herbst2, J.W. Carey2
1Centers for Disease Control and Prevention (CDC), Division of HIV/AIDS Prevention, NCHHSTP, Atlanta, United States, 2Centers for Disease Control and Prevention (CDC), Division of HIV/AIDS Prevention, Atlanta, United States
Background: The cost-effective use of biomedical HIV prevention interventions (e.g., preexposure prophylaxis) will depend on identifying persons at highest risk fo infection. Healthcare providers need rapid, simple, and valid HIV risk screening tools similar to those currently used to screen for alcohol abuse and suicide risk. For men who have sex with men (MSM) and injection drug users (IDU), we developed brief, provider-administrated screening indexes.
Methods: For MSM, we used data from the VAX004 study, a three-year phase III multicenter HIV vaccine trial; and for IDU, 10 years of data from the ALIVE study, a community-based cohort study. Participants in each study were HIV-negative at baseline and received biannual risk behavior and HIV testing. We used generalized estimating equations (GEE) with a bootstrap and backward elimination method to identify significant predictors of incident HIV infection, then assigned point values to variables based on regression coefficients, and summed to create a risk index score. Using the total score, we computed the area under the reviever operating characteristic (ROC) curve and assessed the sensitivity and specificity of varying cut-points.
|Index variables||Area under ROC curve||Score cutoff/total||HIV incident cases >cutoff|
|MSM||Age; prior 6 months: total # sex partners, total # HIV-positive partners, # receptive anal sex acts with HIV-positive partner, # receptive anal sex acts with unknown status partner, amphetamine use, popper use||0.75||10/46||80%|
|IDU||Age; prior 6 months: injected cocaine, visited a shooting gallery, shared a cooker or cotton, on methadone treatment, reported any sexually transmitted disease||0.76||22/49||83%|
[Risk Index Characteristics]
Conclusions: Using variables that predict incident HIV infection in large prospective cohorts of MSM and IDU in the Unites States, we developed two risk screening indexes to prioritize patients for further evaluation for biomedical or other intensive HIV prevention. Next, we will pilot test these two instruments with providers to assess ease of administration/scoring and will develop and index for those at risk by heterosexual behaviors.