SETTING: Despite worldwide scale-up of human immunodeficiency virus (HIV) care services, relatively few countries have implemented isoniazid preventive therapy (IPT). Among other programmatic concerns, IPT completion tends to be low, especially when not fully integrated into HIV care clinics.

OBJECTIVE: To estimate non-completion of 6-month IPT and its predictors among HIV-positive adults aged 16 years.

DESIGN: A prospective cohort study nested within a cluster-randomised trial of TB prevention was conducted between February 2012 and June 2014. IPT for 6 months was provided with pyridoxine at study clinics. Non-completion was defined as loss to follow-up (LTFU), death, active/presumptive TB or stopping IPT for any other reason. Random-effects logistic regression was used to determine predictors of non-completion.

RESULTS: Of 1284 HIV-positive adults initiated on IPT, 885/1280 (69.1%) were female; the median CD4 count was 337 cells/μl (IQR 199–511); 320 (24.9%) did not complete IPT. After controlling for antiretroviral treatment status, IPT initiation year, age and sex, non-completion of IPT was associated with World Health Organization stage 3/4 (aOR 1.76, 95%CI 1.22–2.55), CD4 count 100–349 cells/μl (aOR 1.93, 95%CI 1.10–3.38) and any reported side effects (aOR 22.00, 95%CI 9.45–46.71).

CONCLUSION: Completion of IPT was suboptimal. Interventions to further improve retention should target immunosuppressed HIV-positive adults and address side effects.