BACKGROUND

In 2004, the Mozambican Ministry of Health began a national scale-up of antiretroviral therapy (ART) using a vertical model of HIV clinics colocated within large urban hospitals. In 2006, the ministry expanded access by integrating ART into primary health care clinics.

METHODS

We conducted a retrospective cohort study including adult ART-naive patients initiating ART between January 2006 and June 2008 in public sector clinics in Manica and Sofala provinces. Cox proportional hazards models with robust variances were used to estimate the association between clinic model (vertical/integrated), clinic location (urban/rural), and clinic experience (first 6 months/post first 6 months) and attrition occurring in early patient follow-up (≤ 6 months) and attrition occurring in late patient follow-up (>6 months), while controlling for age, sex, education, pre-ART CD4 count, World Health Organization stage and pharmacy staff burden.

RESULTS

A total of 11,775 patients from 17 clinics were studied. The overall attrition rate was 37 per 100 person-years. Patients attending integrated clinics had a higher risk of attrition in late follow-up [hazard ratio (HR) = 1.75; 95% confidence interval (CI): 1.04 to 2.94], and patients attending urban clinics (HR = 0.57; 95% CI: 0.35 to 0.91) had a lower risk of attrition in late follow-up. Though not statistically significant, clinics open for longer than 6 months (HR = 0.71; 95% CI: 0.49 to 1.04) had a lower risk of attrition in early follow-up.

CONCLUSIONS

Patients attending vertical clinics had a lower risk of attrition. Utilizing primary health clinics to implement ART is necessary to reach higher levels of coverage; however, further implementation strategies should be developed to improve patient retention in these settings.