Validity of reported retention in antiretroviral therapy after roll-out to peripheral facilities in Mozambique: Results of a retrospective national cohort analysis

BACKGROUND: Retention in anti-retroviral therapy (ART) presents a challenge in sub-Saharan Africa. In Mozambique, after roll-out to peripheral facilities, the 12-month retention rate was reported mostly from sites with an electronic patient tracking system (EPTS), representing only 65% of patients. We conducted a nationally representative study, compared 12-month retention at EPTS and non-EPTS sites, and its predictors. METHODS: Applying a proportionate to population size sampling strategy, we obtained a nationally representative sample of patients who initiated ART between January 2013 and June 2014. We calculated weighted proportions of the patients’ status at 12 months after ART initiation, and 12-month incidence of lost to follow-up (LTFU) and death. We assessed determinants of LTFU and death by calculating adjusted hazard ratios (AHR) through multivariate cox-proportional hazard models. RESULTS: Among 19,297 patients sampled, 54.3% were still active, 33.1% LTFU, 2.0% dead, 2.6% transferred-out and 8.0% had unknown status, 12 months after ART initiation. Total attrition rate (LTFU or dead) was 45.5/100PY, higher at facilities without EPTS (51.8/100PY) than with EPTS (37.7/100PY). Clinical stage IV (AHR = 1.7), CD4 count ≤150 (AHR = 1.3) and being pregnant (AHR = 1.6) were significantly associated with LTFU. Clinical stage III or IV (AHR = 2.1 and 3.8), CD4 count ≤150 (AHR = 3.0), not being pregnant (AHR = 3.0), and ART regimens with stavudine (AHR = 4.28) were significantly associated with deaths. Patients enrolled in adherence support groups were 4.6 times less likely to be LTFU, but the number (n = 174) was too small to be significant (p = 0.273). CONCLUSION: Retention in ART was substantially lower at non-EPTS sites. EPTS should be expanded to all ART sites to facilitate comprehensive routine monitoring of retention in care. Retention in Mozambique is low and needs to be improved, especially among pregnant women and patients with advanced disease at ART initiation. The effect of ART adherence support groups needs to be further monitored.