Impacts of disaggregating programmatic outcomes by documentation of citizenship in South Africa

Abstract 

Background: In many resource-limited settings, people in rural areas often migrate to urban hubs in search of w ork. As such, public-sector HIV clinics in urban areas of South Africa often include local residents and patients from outside of the province or other African countries. This analysis compares programmatic outcomes by documentation of citizenship status in an urban resouce-limited setting. Methods: We included all adult ART-naïve patients w ho initiated treatment from April 2004-May 2012 at a public-sector HIV clinic in Johannesburg. Patients w ere follow ed from ART initiation until death, transfer, lost to follow -up (LTF; ≥3 months late for a scheduled visit), or completion of 12 months of follow -up. We describe attrition as mortality and LTF using proportions stratified by possession of a SA national ID number, as an indicator of SA citizenship. An ID number also allows linkage w ith the National Vital Registration System to improve mortality ascertainment. We modeled the risk of attrition using a log-binomial model. Results: Of the 18237 patients w ho initiated ART betw een April 2004-May 2012, 66.2% had an ID, 61.2% w ere female, median (IQR) age w as 36.8 (31.4-43.4), and median (IQR) baseline CD4 count was 100 (37-174) cells/mm3. Baseline clinical characteristics w ere similar across ID groups.  Patients w ere follow ed for a median (range) of 12 (0.03-12) months each. After 1 year of follow -up, 13.6% of patients with an ID had either died or left care compared to 31.1% of patients without an ID. Patients w ith an ID w ere slightly more likely to have died (9.1%) than patients w ithout an ID (6.6%), likely due to better ascertainment of mortality among patients w ith an ID. However, patients without an ID w ere far more likely to be LTF (24.4%) compared to patients with an ID (4.6%), resulting in an overall increased risk of attrition (aRR:1.83; 95% CI:1.69-1.97). Conclusions: Patients w ithout an ID w ere more likely to drop out of care within 12 months of ART initiation than patients w ith an ID. Further research is needed to determine w hether such patients self-transfer to another HIV clinic for further care as such high rates of attrition pose challenges for the success of potential test-and-treat programs.

Conference: International AIDS conferece 2014, Melbourne, Australia