Finding contacts of homeless tuberculosis patients in New York City.

OBJECTIVE: To determine factors associated with no contacts identified for homeless patients in New York City. DESIGN: Culture-confirmed pulmonary tuberculosis cases in persons >18 years old diagnosed in 1997-1999 were included. Demographic and clinical characteristics of tuberculosis patients associated with the number of contacts identified according to homeless status were analyzed using unconditional logistic regression. RESULTS: Homeless patients (n = 152) had a significantly lower median number of contacts than non-homeless patients (n = 2836) (1 vs. 4, P < 0.001). Among homeless patients, having AFB smear-positive sputum with cavitary lesions reduced the likelihood of having no contacts identified. Homeless patients who lived on the street at the time of diagnosis were more likely to have no contacts identified compared to those with contacts identified (61.4% vs. 56.1%); however, the difference was not statistically significant (P = 0.506). Unlike non-homeless patients, being hospitalized at the time of tuberculosis diagnosis was not associated with having contacts identified in homeless patients. CONCLUSIONS: Homelessness independently predicted the likelihood of having no contacts identified. Strategies such as interviews that focus on location rather than persons may be more effective for identifying contacts. Furthermore, being homeless at the time of diagnosis should be used as an indicator for prioritizing prompt contact evaluation.