We suggest four factors that could explain the diminishing proportion of HIV/AIDS borne by the populations moving through CFs; these four factors could contribute either alone or in combination. First, with increased life expectancy for persons with HIV with the advent of better therapeutics, infected persons are aging out of the crime-prone years, generally considered to be between the ages of 15 and 24. Second, while a person may have been infected in his or her crime-prone years, with HAART, prison AIDS mortality has fallen 
, and that person is more likely to survive incarceration, be released, and stay out. Third, the past decade has seen a decline in the number and proportion of HIV/AIDS cases among injection drug users 
, probably due to interventions to reduce the harm associated with parenteral drug use.
Finally, considerable effort has been made by prison systems, and some jails, to enhance discharge planning for HIV-infected persons 
. The effects of these programs are difficult to assess. A controlled but non-randomized trial has been conducted in North Carolina 
. Preliminary results show a non-significant trend in the efficacy of discharge planning, but final results have yet to be published as of submission of this manuscript. Observational studies seem to show that those CFs with adequate discharge planning do better than those without 
. A program in Rhode Island compared the recidivism of HIV-infected women to historical controls and found a significant difference in return rates 
. Enhanced discharge planning of HIV-infected inmates beyond that usually received by prisoners may be contributing to lower recidivism rates in the HIV-infected population in Rhode Island.
One limitation of this study was the need to make numerous assumptions about the proportion of persons who would move from a jail to a prison and the number of releasees who would return to the same type of facility in the space of 1 y. The sensitivity analysis shows that even with faulty estimates, the general trend is probably accurate. We emphasize the unlikelihood of all jail and prison releasees representing unique persons. A second limitation is that our recidivism data were based on rates of return to the same institution. The assumption that recidivists would be reincarcerated in the identical jurisdiction is not far-fetched, given that many releasees would be on probation or parole, which often stipulates residency in the same jurisdiction. Third, HIV/AIDS prevalence in prisons and jails was largely derived from self-reported data, and an estimate of the prevalence in untested individuals was applied. We believe this would bias our estimate of inmates with HIV/AIDS slightly downwards, if at all.
Because certain groups at high risk for HIV/AIDS come together in correctional facilities, seroprevalence was high early in the epidemic; 16.2% of men and 25.1% of women tested for HIV in 1989 at the New York City jail on Rikers Island were found to be HIV-positive 
. The decline in HIV prevalence among correctional populations has been offset by the growing number of inmates. Although the proportional share of HIV/AIDS borne by those passing through CFs has declined since 1997, the total number of HIV infected persons who are in this flow has remained steady at roughly 150,000 individuals, an estimate that is only marginally perturbed by an assumption that each detainee is incarcerated 1.5 rather than 1.4 times per year.
As the HIV epidemic has matured, the share borne by releasees has decreased, but the total number of persons with HIV released from CFs is unchanged. This steady size of the target population leads us to conclude that CFs still represent a rich focus for public health interventions. Interventions in CFs may have the greatest impact on the HIV epidemic among minority men, given the disproportionate incarceration rates in the US criminal justice system. The proportion of minorities diagnosed with HIV late in the course of disease, less than 12 mo before a diagnosis of AIDS, continues to lag behind whites 
, so reaching minority populations is a public health priority. Jail and prison inmates represent a captive, and still very important, audience for HIV testing, counseling, and prevention messages. After diagnosis, enabling HIV-infected releasees to link to community care is of utmost importance. Because virtually all persons entering CFs return to the community, effective interventions benefit not only CF populations but also the communities to which releasees return.