BackgroundDespite a call for differentiated care, there are limited data from sub-Saharan Africa on comprehensive community-based HIV care for key populations (KP), including commercial sex workers (CSW), men who have sex with men (MSM), and people who inject drugs (PWID). In Nigeria, a programme was implemented that liaised with community-based organizations and offered HIV testing, same-day ART initiation, and ART follow-up to KP. Here we characterize KP and their partners enrolled on ART. Our objective is to assess the early treatment outcomes and to estimate predictors of attrition among KP.MethodThis is a retrospective cohort study of routinely collected data in a community-based HIV program for KP in Nasarawa state, Nigeria from August 2016 to November 2017. Variables of interest were socio-demographic, KP types, treatment outcomes, ART adherence, WHO stage, TB status and viral load. Summary statistics, logistic and Cox proportional hazard regression were used to describe the characteristics of KP and estimate predictors of attrition (patients either lost to follow-up (LTFU) or dead).ResultSeven hundred and ten (710) KP and their partners were enrolled into this study, 77.3% (549) of study participants were female and the median age was 30 years (IQR: 24–35). Respectively, 74.2%, 4.5%, 1.1% and 20% were FSW, MSM, PWID and their partners. Of 710 KP who started ART, 13.9% (99/710) discontinued after the first visit. After a median follow-up time of 7 months on ART 73.2% of patients were retained, 23.4% were LTFU, and 3.4% were dead. Lack of formal education (aHR 1.8; 95% CI 1.3–2.6) and unemployment (aHR 1.8; 95% CI 1.2–2.6) were significantly associated with attrition.ConclusionComprehensive community-based HIV care, including HIV testing and same-day ART is feasible. However, ART initiation on the same day of confirmatory HIV testing resulted in a high uptake of ART, but possibly inflated early attrition on ART. To mitigate early attrition among KP after same-day ART initiation, the psychosocial readiness of clients should be assessed better. We strongly recommend further studies to understand factors contributing to high attrition among the KP.