Objective: To evaluate the effectiveness of supervised therapy for tub
erculosis (TB) in patients with HIV infection. Design: Retrospective,
chart review. Patients: Patients with TB and HIV infection. Setting: U
rban, public TB clinic. Main measures and results: A total of 107 pati
ents with TB and HIV infection were studied. Most were men (78%), Afri
can American (91%), uninsured or on Medicaid (88%), and 67% were injec
ting drug users. TB was diagnosed before AIDS in 31% of subjects, at t
he time of AIDS in 32%, and after AIDS in 37%. Clinical features varie
d by stage of HIV disease. Sixteen patients received no therapy and di
ed before TB was diagnosed, 10 died during the first 8 weeks of treatm
ent. Seventy-eight patients received > 8 weeks therapy, of whom 48 (62
%) were given directly observed therapy twice weekly and 30 (38%) rece
ived self-administered daily therapy. Patients who received directly o
bserved therapy were more likely to complete 6 months of therapy (96 v
ersus 76%, P = 0.02) and more likely to survive after therapy ended (8
5 versus 57%, P = 0.01). By logistic regression, directly observed the
rapy, AIDS diagnosed before TB, and age were significantly associated
with survival outcome. Conclusion: Directly observed therapy for TB in
patients with HIV infection is highly effective and associated with b
etter adherence to therapy and survival.