Do loss to follow-up and death rates from ART care vary across primary health care facilities and hospitals in south Ethiopia? A retrospective follow-up study

BACKGROUND: Decentralization and task shifting has significantly improved access to antiretroviral therapy (ART). Many studies conducted to determine the attrition rate in Ethiopia have not compared attrition rates between hospitals and health centers in a relatively recent cohort of patients. This study compared death and loss to follow-up (LTFU) rates among ART patients in hospitals and health centers in south Ethiopia. METHODS: Data routinely collected from patients aged older than 15 years who started ART between July 2011 and August 2012 in 20 selected health facilities (12 being hospitals) were analyzed. The outcomes of interest were LTFU and death. The data were entered, cleaned, and analyzed using Statistical Package for the Social Sciences version 20.0 and Stata version 12.0. Competing-risk regression models were used. RESULTS: The service years of the facilities were similar (median 8 and 7.5 for hospitals and health centers, respectively). The mean patient age was 33.7±9.6 years. The median baseline CD4 count was 179 (interquartile range 93–263) cells/mm(3). A total of 2,356 person-years of observation were made with a median follow-up duration of 28 (interquartile range 22–31) months; 24.6% were either dead or LTFU, resulting in a retention rate of 75.4%. The death rates were 3.0 and 1.5 and the LTFU rate were 9.0 and 10.9 per 100 person-years of observation in health centers and hospitals, respectively. The competing-risk regression model showed that the gap between testing and initiation of ART, body mass index, World Health Organization clinical stage, isoniazid prophylaxis, age, facility type, and educational status were independently associated with LTFU. Moreover, baseline tuberculous disease, poor functional status, and follow-up at a health center were associated with an elevated probability of death. CONCLUSION: We observed a higher death rate and a lower LTFU rate in health centers than in hospitals. Most of the associated variables were also previously documented. Higher LTFU was noticed for patients with a smaller gap between testing and initiation of treatment.