BACKGROUND. Death from all causes during tuberculosis (TB) treatment has remained about 10% in San Francisco County. Estimating risk factors for all-cause mortality was this study's aim.

METHODS. A retrospective cohort study design was used to examine 1,730 subjects with TB older than 24 years in San Francisco from 1995 to 2004. The observation period was from the first day of TB treatment to the last day of treatment or death from any cause. Mean follow-up was 35.8 (± 20) weeks. Host characteristics including age, gender, race, immigration, and language; social-environmental factors including residence, incarceration, and unemployment; TB disease characteristics including anatomic site, culture, chest radiograph, resistance, previous TB; comorbid conditions including HIV coinfection and substance use; health care delivery factors including directly observed (DOT) or self-administered therapy (SAT), and public or private TB medical provider were tested for their predictive value on all-cause mortality.

RESULTS. All-cause mortality was 10.6% (184/1730); 50% of deaths (92/184) occurred < 8 weeks of treatment. Statistically significant predictors of all-cause mortality were age 45-64 (OR 2.2, 95% CI: 1.3, 3.9), age ≥ 65 (OR 2.5, 95% CI: 1.3, 4.7); male gender (OR 1.6, 95% CI: 1.1, 2.4); homelessness (OR 2.1, 95% CI: 1.2, 3.5); not working (OR 5.9, 95% CI: 3.0, 11.5); HIV coinfection (OR 8.1, 95% CI: 3.8, 17.0); and private TB medical provider (OR 3.8, 95%CI: 2.6, 5.7). Previous TB was associated with protection from all-cause mortality (OR 0.6, 95% CI: 0.3, 0.9); concomitant extrapulmonary TB was not associated with all-cause mortality (OR 1.3, 95% CI 0.9, 1.8). Subset analysis of patients receiving public TB care showed that SAT (OR 0.5, 95% CI 0.2, 0.9) and a combination of DOT and SAT were associated with a protective effect when compared with DOT only (OR 0.3, 95% CI: 0.1, 0.8).

CONCLUSION. Middle and older age, male gender, homelessness, not working, positive TB culture, and HIV coinfection predict all-cause mortality. Private TB medical provider was predictive of mortality, and the combination of DOT and SAT is beneficial, suggesting that public-private, TB-care partnerships need to be strengthened and that supervision of therapy should be flexible.