Effects of an increase in schooling on HIV infection and behavioural risk: a multicountry instrumental variable analysis of 23 sub-Saharan African countries
Justin Parkhurst, Olivia Klaassen, Pepita Barlow, Ranjeeta ThomasObjectives
To estimate the causal effect of increased schooling arising from national policy reforms on HIV/AIDS infection and testing, along with associated indicators of risk across a wide range of settings in sub-Saharan Africa.
Design
Multicountry quasi-experimental analysis using two-stage least squares (instrumental variable) analysis with national schooling reforms as the instrument for years of schooling.
Setting
A pooled sample from 23 sub-Saharan African countries with historical peak HIV-prevalence rates ranging from under 1% to over 20%, surveyed from 2003 to 2019.
Participants
A sample of 191 128 cases with linked HIV test results was obtained from nationally representative Demographic and Health Surveys and AIDS Indicator Surveys (and a related survey in Botswana), restricted to adults aged 18–49 years in birth cohorts spanning 1955–1998.
Primary and secondary outcome measures
Primary outcomes were HIV test positivity and ever having been tested for HIV. Secondary outcomes were comprehensive HIV knowledge, age at first intercourse, condom use at last intercourse, number of sexual partners in the preceding 12 months and lifetime number of sexual partners.
Results
Exposure to an education reform increased mean schooling by 0.45 years (95% CI 0.28 to 0.63). Pooled instrumental variable analysis found no significant effect of an additional year of schooling on HIV test positivity (−0.5 percentage points, coefficient −0.005, 95% CI −0.018 to 0.008). One additional year of schooling was associated with a 4.6 percentage point increase in ever having been tested for HIV, significant at the 10% level (coefficient 0.046, 95% CI −0.001 to 0.094). Effects on secondary behavioural outcomes were limited to specific subgroups: women showed a small delay in age at first intercourse while men and urban residents showed increased condom use. No significant effect on comprehensive HIV knowledge was identified in the pooled sample.
Conclusions
Across a wide range of African settings, policies expanding access to schooling appear to increase HIV testing but do not directly reduce HIV infection. Policymakers should look beyond education alone as an HIV prevention strategy, with realistic expectations of schooling expansion as a strategy on its own.