Brief Report: Incidence and Correlates of Pregnancy in HIV-Positive Kenyan Sex Workers

Introduction: The incidence of pregnancy in female sex workers (FSWs) living with HIV is not well characterized. Methods: Current and former FSWs living with HIV and enrolled in a prospective cohort study in Mombasa, Kenya, were followed monthly to ascertain sexual behavior and underwent quarterly pregnancy testing. Pregnancies were considered planned, mistimed, or unwanted according to fertility desires and pregnancy intentions. Cox proportional hazards models were used to estimate hazard ratios (HRs) of the association between characteristics and incident pregnancy. Results: Two hundred seventy-nine FSWs were eligible (October 2012-April 2017). Most women had a nonpaying, regular partner (83.2%, 232/279), were not using modern nonbarrier contraception (69.5%, 194/279), and did not desire additional children (70.6%, 197/279). Of 34 first incident pregnancies [5.8/100 person-years (p-y); 95% confidence interval (CI): 4.1 to 8.0], 91.2% (n = 31) were unintended. The incidences of planned (5.7/100 p-y, n = 3), mistimed (10.4/100 p-y, n = 9), and unwanted pregnancies (4.9/100 p-y, n = 22) were similar (P= 0.15). In univariable analysis, oral contraceptive pill use (versus no contraception), having a nonpaying, regular partner, transactional sex, vaginal washing, condomless sex, and higher sex frequency were associated with an increased pregnancy risk. Older age was associated with a lower pregnancy risk. In multivariable analysis, having a nonpaying, regular partner (adjusted HR 4.0, 95% CI: 1.2 to 14.1) and age >= 40 years (aHR 0.2, 95% CI: 0.0 to 0.9) remained significantly associated with a higher and lower pregnancy risk, respectively. Conclusion: In this cohort of HIV-positive FSWs, most pregnancies were unintended. Identifying FSWs' fertility desires and pregnancy intentions could facilitate efforts to increase contraceptive use and implement safer conception strategies.