Younger age and virological failure predict attrition within the paediatric HIV programme, findings from South Africa’s National Health Laboratory Services’ Data Warehouse: A descriptive analysis (Preprint)

Abstract

A number of cohort studies in South Africa have described low viral suppression and high attrition rates within the paediatric HIV treatment programme.

Using routine laboratory data, we evaluated HIV viral load (VL) monitoring, including mobility and over-due VL (OVL) testing, within five priority districts in South Africa.

Retrospective descriptive analysis of National Health Laboratory Service (NHLS) data for children and adolescents 1-<15 years with HIV VL testing between 1st May 2019 - 30th April 2020 from 152 facilities within City of Johannesburg, City of Tshwane, eThekwini, uMgungundlovu, and Zululand. Test-level data were de-duplicated to patient-level data using the NHLS CDW probabilistic record-linking algorithm and then further manually de-duplicated. An OVL was defined as no subsequent VL within 18 months of the last test. Variables associated with last VL result, including age, sex, VL result, district type and facility type are described. A multivariate logistic regression analysis was performed to identify whether these variables were associated with an OVL.

Among 21 338 children and adolescents 1–<15 years of age who had an HIV VL test, 72.70% (15 512) had a follow-up VL test within 18 months. 13.33% (2 194) followed up at a different facility, of which 3.79% (624) were in a different district and 1.71% (281) in a different province. Among patients with a VL ≥1 000 cps/ml, median time to subsequent test was 6 months [IQR: 4–10 months]. The younger the age of the patient, the greater the proportion with an OVL, ranging from a high of 52% among one year olds to a low of 21% among 14 year olds. On multivariate analysis, two consecutive HIV VL results ≥ 1 000 cps/ml were associated with an increased adjusted odds ratio (AOR) of having an OVL (AOR 2.07; 95% CI 1.71-2.51). Conversely, patients seen at a hospital (AOR 0.86; 95% CI 0.77-0.96), patients with ≥2 previous tests (AOR 0.78; 95% CI 0.70-0.86), patients seen in a rural district (AOR 0.63; 95% CI 0.54-0.73) and older age groups, 5-9 years (AOR 0.56; 95% CI 0.47-0.65); 10-14 years (AOR 0.51; 95% CI 0.44-0.59), compared with 1-4 years were associated with a significantly decreased odds of having an OVL test.

Considerable attrition occurs within the paediatric HIV treatment programme, with younger children and those with virological failure at greatest risk.