Objectives: To determine the risk factors for HIV seroconversion and a
ssess the association between contraceptive use and HIV infection amon
g women attending three large family planning clinics in Dar es Salaam
, Tanzania. Design: Prospective cohort study. Methods: Between 1992 an
d 1995, 2471 HIV-negative women were followed prospectively. Informati
on about sociodemographic characteristics, sexual behavior, contracept
ive use and other risk factors was collected at recruitment and update
d at follow-up visits. At the end of the study, specimens were collect
ed for HIV testing and laboratory diagnosis of sexually transmitted di
seases. Results: The overall HIV incidence was 3.4 per 100 person-year
s [95% confidence interval (CI), 2.6-4.1]. The risk of HIV seroconvers
ion decreased with increasing age (P = 0.04, test for trend). Women re
porting three or more sex partners during the follow-up period had the
highest risk of HIV [age-adjusted relative risk (RR), 4.89; 95% CI, 2
.61-9.17]. Having an uncircumcised husband was associated with a signi
ficantly increased risk of HIV (age-adjusted RR, 3.60; 95% CI, 1.12-11
.59). The risk of HIV was also significantly increased among women wit
h gonorrhoea (age-adjusted RR, 3.51; 95% CI, 1.60-7.71) and candidiasi
s at baseline (age-adjusted RR, 1.98; 95% CI, 1.17-3.33) and among wom
en reporting alcohol consumption during the follow-up period. After co
ntrolling for other risk factors, the risk of HIV infection amongst us
ers of oral contraceptive, intrauterine device and injectable contrace
ptive was not significantly increased. Similarly, there was no signifi
cant trend associated with increasing duration of use of any of these
contraceptive methods. Conclusion: These findings confirm that a large
number of new HIV infections continue to occur in this population. Re
assuringly, no significant association was observed between HIV and us
e of specific contraceptive methods. Interventions to reduce further s
pread of HIV are still urgently needed.