Background The use of antiretroviral (ARV) therapy has reduced mortality and morbidity globally among people living with HIV. We sought to assess patient’s characteristics at the commencement of ARV therapy and their association with increased CD4 count following initiation of treatment with ARV.

Methods A retrospective cohort study of medical records patients who had receiving ARV therapy between 2002 and 2012 at Kerti Praja Foundation in Bali. We included all patients >15 years old with CD4 at start of treatment of ≤350 cells/mm³ and at least one follow-up CD4 test result. Variables included in the analyses were; sex, age, education, occupation, risk behaviours, opportunistic Infection, supervision of ART, CD4 count, body weight, and haemoglobin level also starting dates ARV therapy. Kaplan Meier and univariate and multivariate Cox proportional Hazard Model were used to assess predictors of achieving a CD4 count >350 cells/mm³. Patients were followed up to time of reaching CD4 >350 cells or to date of death or last clinic visit.

Result From the 311 patients, we found 46% had a CD4 count that increased to above 350 cell/mm³. Median time to achieving  this end point was 1.4 years (IQR = 0.7–3.0). In multivariate analysis, a CD4 count at baseline above 200 cells/mm³ was associated with achieving the end point (HR = 3.83; 95% CI = 2.59–5.68). Patients with a history of injecting drugs were significantly less likely to achieve the endpoint compared to patients reporting heterosexual contact (HR = 0.54; 95% CI = 0.36–0.82).

Conclusion Patients with CD4 count >200 cell/mm³ and without a history of injecting drugs were more likely to achieve a CD4 count >350 cell/mm³. Our findings demonstrate the benefit of starting HIV treatment earlier when CD4 counts is higher and support national recommendations encouraging early initiation ARV therapy. These data also suggests that close attention and further assessment are needed for patients starting ART who have a history of injecting drugs.

Disclosure of interest statement None.