Abstract
We studied a cohort of 15,620 adults who had experienced at least 1 jail incarceration and 1 homeless shelter stay in 2001-2003 in New York City to identify trajectories of these events and tested whether a particular trajectory was associated with all-cause, drug-related, or human immunodeficiency virus (HIV)-related mortality risk in 2004-2005. Using matched data on jail time, homeless shelter stays, and vital statistics, we performed sequence analysis and assessed mortality risk using standardized mortality ratios (SMRs) and marginal structural modeling. We identified 6 trajectories. Sixty percent of the cohort members had a temporary pattern, which was characterized by sporadic experiences of brief incarceration and homelessness, whereas the rest had the other 5 patterns, which reflected experiences of increasing, decreasing, or persistent jail or shelter stays. Mortality risk among individuals with a temporary pattern was significantly higher than those of adults who had not been incarcerated or stayed in a homeless shelter during the study period (all-cause SMR: 1.35, 95% confidence interval (CI): 1.14, 1.59; drug-related SMR: 4.60, 95% CI: 3.17, 6.46; HIV-related SMR: 1.54, 95% CI: 1.03, 2.21); all-cause and HIV-related SMRs in other patterns were not statistically significantly different. When we compared all 6 trajectories, the temporary pattern was more strongly associated with higher mortality risk than was the continuously homelessness pattern. Institutional interventions to reduce recurrent cycles of incarceration and homelessness are needed to augment behavioral interventions to reduce mortality risk.