Estimating the impact of HIV self-testing on HIV testing services, diagnoses, and treatment initiation at the population-level with routine data: the example of the ATLAS program in Côte d'Ivoire


BACKGROUND: HIV self-testing (HIVST) is a critical testing approach particularly for reaching those at HIV risk who are hesitant or unable to access existing services. While the discreet and flexible nature of HIVST is appealing to users, these features can limit the ability for programmes to monitor and estimate the population-level impacts of HIVST implementation. This study triangulates publicly available routine programme data from Côte d'Ivoire in order estimate the effects of HIVST distribution on access to testing, conventional testing (self-testing excluded), HIV diagnoses, and antiretroviral treatment (ART) initiations.
METHODS: We used quarterly programmatic data (Q3-2019 to Q1-2021) from ATLAS, a project that aims to promote and implement network-based HIVST distribution in West Africa, in addition to routine HIV testing services program data obtained from the PEPFAR dashboard. We performed ecological time series regression using linear mixed-models.
RESULTS: Between Q3-2019 and Q1-2021, 99,353 HIVST kits were distributed by ATLAS in 78 health districts included in the analysis. The results (Table 1) show a negative but non-significant effect of the number of ATLAS HIVST on the volume of conventional tests (-190), suggesting the possibility of a slight substitution effect. Despite this, the the beneficial effect on access to testing is significant: for each 1000 HIVST distributed via ATLAS, 390 to 590 additional HIV tests were performed if 60% to 80% of HIVST are used . The effect of HIVST on HIV diagnosis was significant and positive, with 8 additional diagnoses per 1,000 HIVST distributed. No effect of HIVST was observed on ART initiations.

Table 1: Linear effect of number of HIVST distributed via ATLAS on access to HIV testing, on 'conventional' testing, HIV diagnosis and ART initiation, in 78 health districts monitored by PEPFAR in Côte d'Ivoire (Q3-2019 to Q1-2021), adjusted by quarter year and region.
Outcome variablesEffect for 1000 HIVST distributed by ATLAS95% CIp-value
HIV testing (utilisation rate of HIVST of 80%)+590356 to 821<0·001
HIV testing (utilisation rate of HIVST of 60%)+390160 to 625<0·001
Conventional testing-190-427 to 380·10
HIV diagnosis+80 to 150·044
ART initiations-2-8 to 50·66
1CI = Confidence Interval. ART = antiretroviral treatment.

CONCLUSIONS: Our study provides a standard methodology for estimating the population-level impact of HIVST that can be used across countries. It shows that HIVST distribution was associated with increased access to HIV testing and diagnosis in Côte d'Ivoire. Wide-scale adoption of this method will improve HIVST data quality and inform evidence-based programming.

Download the e-Poster (Movie)

Download the e-Poster (PDF)