Return
Using data science to investigate rising HIV low-level viraemia results at Groote Schuur laboratory in South Africa
M
K
L
S
M
D
DOI:10.4102/ajlm.v14i1.2953.png)
Abstract
En 中文
Background: Following a major service disruption across the National Health Laboratory Service, backlogged HIV viral load (VL) specimens from Limpopo province were rerouted to Groote Schuur Hospital (GSH) laboratory in the Western Cape province. During this time, an increase in low-level viraemia (LLV; 50 copies/mL-1000 copies/mL) was observed at the GSH laboratory, raising concerns about possible pre-analytical and analytical issues, including compromised specimen quality and possible contamination. Objective: To determine whether the observed increased LLV was due to analytical errors (e.g. contamination), pre-analytical factors such as prolonged turnaround times (TAT), or underlying epidemiological differences. Methods: HIV VL data from 2023-2024 for Limpopo and the Western Cape were analysed using Python. Viral load results were grouped into predefined categories and compared. Turnaround times were plotted and Thembisa model estimates were used to assess HIV prevalence and antiretroviral therapy (ART) coverage. Longitudinal patient-level analysis evaluated VL trends as a proxy for adherence. In addition, quality control data were evaluated at testing sites. Results: Limpopo specimens showed higher LLV (20%) and lower viral suppression (< 50 copies/mL) at 70%, compared to Western Cape (13% LLV, 81% suppression), where follow-up and suppression outcomes were also higher. No clear evidence indicated that extended TAT or potential instrument contamination affected VL results significantly. Conclusion: The increase in LLV at GSH was linked primarily to processing specimens from Limpopo, highlighting regional differences in HIV VL result distributions. These differences probably reflect variations in ART access and adherence, rather than laboratory-related issues such as delayed TAT, sample quality, or contamination.
Keywords:
laboratory testing
HIV
viral load
data science
low-level viraemia
pre-analytical factors
ART programme
contamination
Journal
A
IF:
1.2
Papers:
30
Citations:
0
