Latent tuberculosis infection: screening and treatment in an urban setting.

Despite its benefit for treating active tuberculosis, directly observed therapy (DOT) for latent tuberculosis infection (LTBI) has been largely understudied among challenging inner city populations. Utilizing questionnaire data from a comprehensive mobile healthcare clinic in New Haven, CT from 2003 to July 2011, a total of 2,523 completed tuberculin skin tests (TSTs) resulted in 356 new LTBIs. Multivariate logistic regression correlated covariates of the two outcomes (a) initiation of isoniazid preventative therapy (IPT) and (b) completion of 9 months of IPT. Of the 357 newly positive TSTs, 86.3 % (n = 308) completed screening chest radiographs (CXRs): 90.3 % (n = 278) were normal, and 0.3 % (n = 1) had active tuberculosis. Of those completing CXR screening, 44.0 % (n = 135) agreed to IPT: 69.6 % (n = 94) selected DOT, and 30.4 % (n = 41) selected self-administered therapy (SAT). Initiating IPT was correlated with undocumented status (AOR = 3.43; p < 0.001) and being born in a country of highest and third highest tuberculosis prevalence (AOR = 14.09; p = 0.017 and AOR = 2.25; p = 0.005, respectively). Those selecting DOT were more likely to be Hispanic (83.0 vs 53.7 %; p < 0.0001), undocumented (57.4 vs 41.5 %; p = 0.012), employed (p < 0.0001), uninsured (p = 0.014), and have stable housing (p = 0.002), no prior cocaine or crack use (p = 0.013) and no recent incarceration (p = 0.001). Completing 9 months of IPT was correlated with no recent incarceration (AOR 5.95; p = 0.036) and younger age (AOR 1.03; p = 0.031). SAT and DOT participants did not significantly differ for IPT duration (6.54 vs 5.68 months; p = 0.216) nor 9-month completion (59.8 vs 46.3 %; p = 0.155). In an urban mobile healthcare sample, screening completion for LTBI was high with nearly half initiating IPT. Undocumented, Hispanic immigrants from high prevalence tuberculosis countries were more likely to self-select DOT at the mobile outreach clinic, potentially because of more culturally, linguistically, and logistically accessible services and self-selection optimization phenomena. Within a diverse, urban environment, DOT and SAT IPT models for LTBI treatment resulted in similar outcomes, yet outcomes were hampered by differential measurement bias between DOT and SAT participants.