RESUMO
A 2014 U.S. Department of Justice Best Practices Report advocates that states eliminate HIV-specific criminal penalties except under 2 conditions: when a human immunodeficiency virus (HIV)-positive person intentionally commits a sex crime or transmits the virus by engaging in behavior that poses a significant risk of transmission, regardless of actual transmission. We assess the premise of these exceptions to understand whether these best practices are based on scientific evidence about the population at risk of infection and the risk of sexual violence by HIV-positive individuals. We employ nationally representative, cross-sectional survey data from the Current Population Survey (CPS), the Survey of Inmates in State, Federal, and Local Jails (SISFLJ), and the National Health and Nutrition Survey (NHANES). Data from the CPS, SISFLJ, and NHANES are weighted and combined to analyze bias in the population at risk of HIV. Linear probability models are employed to estimate the likelihood that HIV-positive inmates are incarcerated for violent or sexual offenses, net of socioeconomic factors. We find significant measurement bias in HIV prevalence rates. The selection of national surveys for population denominators distorts contemporary estimates of HIV prevalence by 7% to 20%. Our findings also illustrate that HIV-positive inmates are 10 percentage-points less likely to be incarcerated for violent offenses than HIV-negative inmates. National best practice guidelines may undermine effective social policy that aims to curtail stigma within HIV-positive communities because scientific evidence neither include inmates into prevalence denominators (as a measure of the population at risk) nor assess the likelihood that HIV-positive inmates commit violent or sexual crimes.