Targeted HIV testing programs that rely on prison inmates' self-report of conventional HIV risk behaviors are likely to underestimate HIV prevalence, according to a study of data from the North Carolina Department of Corrections.1 Among the male and female prisoners who were voluntarily tested, the overall HIV prevalence was 3%. Forty-four percent of HIV-positive prisoners did not report any of the conventional HIV risk behaviors, such as history of multiple partners or sharing needles. In general, HIV risk behaviors were moderately associated with the likelihood of HIV infection, suggesting that testing programs based on prisoners' self-report of such risk factors may underestimate the prevalence of infection. An estimated 24–61% of HIV cases were not detected because not all male prisoners agreed to be tested.
To examine associations between incarcerated individuals' characteristics and their risk of HIV, researchers used data collected from male and female prisoners entering the North Carolina Department of Corrections between January 1, 2004, and May 30, 2006. Upon intake, most male and some female prisoners were screened for conventional HIV risk behaviors—ever having shared needles or had a blood transfusion between 1978 and 1985, having had multiple sex partners, having had sex as or with a sex worker, having had sex with men (for males only) and having had sex with an injection-drug user or with a male partner who has sex with men (both for females only). Inmates were offered voluntary HIV testing. Multivariate logistic regression analyses were conducted to determine which characteristics were linked to HIV infection. Researchers estimated the number of untested male prisoners living with HIV using age-, gender- and race-specific prevalences for prisoners and for the general male population of the state.
Of the 54,644 inmates aged 18 or older who entered the prison system during the study period, 39%—15,461 males and 5,958 females—were tested for HIV. Of these, 1% of men had had sex with men, and 3% of women had had sex with a male partner who has sex with men. Eighteen percent of men reported having multiple sex partners, and 7% had had sex with or as a sex worker; among women, those proportions were 27% and 17%, respectively. Three percent of men and 9% of women had ever shared needles; 16% of women had had sex with someone who shares needles. The proportions of males and females who had had a blood transfusion were 1% and 3%, respectively.
Some 718 prisoners tested for HIV were positive, for an overall HIV prevalence of 3% (4% for males and 3% for females). Among HIV-positive prisoners, 44% reported no HIV risk behaviors, and 34% reported at least one; 22% had no data. HIV prevalence was greater among those who reported at least one HIV risk behavior than among those who did not (7% vs. 3% for males, and 4% vs. 2% for females).
In multivariate analyses controlling for social and demographic characteristics, details about the crime and the sentence, and sexual and drug-use behaviors, having had a same-sex partner was strongly associated with HIV infection among males (odds ratio, 8.0); however, other conventional risk behaviors—multiple partners, sex with or as a sex worker, and sharing needles—were more modestly associated with infection (1.4–2.1). For women, having had multiple partners and having had a blood transfusion were associated with increased odds of HIV infection (2.2 and 2.6, respectively).
During the study period, 32,241 male prisoners were not tested for HIV; of those, the researchers estimated that between 223 and 1,101 were HIV-positive. Thus, an estimated 24–61% of all HIV cases were undetected.
While acknowledging that their HIV prevalence estimates are greater than those for neighboring South Carolina's prison system, the general population of North Carolina and the United States, the researchers express confidence in their estimates for female prisoners—most of whom underwent testing. However, in regard to male prisoners, they comment that "elective testing combined with a moderate testing rate likely resulted in a high estimate of seropositivity that is unrepresentative of the general male prison population." The researchers note that targeted testing based on prisoners' self-reported HIV risk behaviors is likely to underestimate HIV prevalence because of the modest associations between risk factors and infection, and because of prisoners' unwillingness to disclose sensitive information. They suggest that "testing should be widely encouraged among all prisoners," but warn that "prisoners should maintain their right to accept or decline HIV testing."
1. Rosen DL et al., Characteristics and behaviors associated with HIV infection among inmates in the North Carolina prison system, American Journal of Public Health, 2009, 99(6):1123–1130.