Objective: In Tanzania during the past 6 years reported tuberculosis (
TB) cases have nearly doubled, with proportionately much greater incre
ases in smear-negative and extrapulmonary cases compared with smear-po
sitive cases. At the same time, HIV infection has become widespread th
roughout the country. This survey was undertaken in order to study the
association of TB and HIV and to determine the impact of HIV on prese
nt and future TB cases in Tanzania. Methods: The survey design provide
d for HIV testing of a representative country-wide sample of approxima
tely one-sixth of all new and relapse cases registered between January
1991 and December 1993, with linkage to demographic, clinical and bac
teriological data for these cases. HIV surveillance data were used for
comparison purposes. Results: A total of 6928 TB cases from all of th
e country's 20 mainland regions were tested. The overall HIV seropreva
lence was 32%. Both crude and adjusted odds ratios (OR) for HIV infect
ion were higher in women, those aged 25-44 years, urban residents, cas
es of smear-negative and extrapulmonary disease, and persons with a ba
cille Calmette-Guerin (BCC) vaccination scar; The age-and sex-adjusted
relative risk for HIV infection in TB patients compared to blood dono
rs in the same regions was 7.1 (95% confidence interval, 6.6-7.5), and
was significantly higher among those aged 25-34 years. Of 3360 patien
ts with bacteriological culture results 46% were culture-positive for
Mycobacterium tuberculosis. Drug susceptibility tests were performed o
n 1164 isolates with the overall rate of drug resistance of 6.2%. Rate
s of initial resistance were low in both HIV-positive (4%) and HIV-neg
ative (5.8%) patients. Rates of acquired resistance were higher (19% o
verall) and did not vary significantly by HIV serostatus. Initial comb
ined resistance to both isoniazid and rifampicin was uncommon (0.4%) a
s was monoresistance to rifampicin (0.3%). Conclusions: The higher OR
for women and young adults reflect the higher rates of HIV infection i
n those populations. The finding that smear-positive relapse cases wer
e no more likely to have HIV infection than new smear-positive cases s
uggests that the treatment regimen for new cases is effective in HIV-a
ssociated TB. The low rates of both initial and acquired drug resistan
ce in HIV-positive patients is further evidence of adequacy of treatme
nt. The higher relative risk for HIV infection among patients aged 25-
34 years suggests increased HIV-related TB transmission. Finally, it i
s estimated that approximately two-thirds of the increase in the rate
of smear-positive tuberculosis in the country can be directly attribut
ed to HIV infection.