Research question
We investigated the research question: Does childhood vaccination improve school grade attainment?  It is plausible that childhood vaccinations improve school grade attainment. Vaccinations prevent diseases that can lead to school absenteeism and decreased school performance. Vaccine-preventable diseases can lead to physical weakness (preventing children from attending school), fatigue (reducing children’s ability to concentrate in school), and impair cognitive development (limiting children’s capacity to acquire new skills). However, based on our current knowledge of the diseases prevented by specific childhood vaccinations, we would expect some vaccinations to have a much greater impact on school attainment than others.

Fit of the research question within the Africa Centre research strategy
The research question fits in broadly with the research agenda at the Africa Centre, investigating the determinants of child development and the effects of health interventions on cognitive and socio-economic outcomes.

The study was co-funded by The Pneumococcal Vaccines Accelerated Development and Introduction Plan (PneumoADIP) at the Johns Hopkins Bloomberg School of Public Health.

Data sources and methods
We used data from the Africa Centre demographic surveillance. To observe educational attainment up to the highest ages possible, we chose the earliest vaccination data available in the surveillance (from 1995 through 2000). In this observation period, vaccination data was available for five routine childhood vaccinations, namely measles vaccination (measles), diphtheria-tetanus-pertussis vaccination (DTP), hepatitis B vaccination (HepB), poliomyelitis vaccination (polio), and the Bacille Calmette-Guérin vaccination (BCG).

The direct comparison of school grade attainment in vaccinated versus unvaccinated children will lead to biased estimates of vaccination effect if the children in the two groups differ systematically in characteristics that influence their school attainment. This bias can be reduced if information on observed variables is taken into account in the analyses. Here, we achieve such a reduction in bias by using propensity score matching methods to balance the distribution of observed variables between vaccinated and unvaccinated children in the comparison. The propensity score is the conditional probability of receiving a particular vaccination given a vector of observed variables. 

We included variables in the propensity score regressions which are likely to affect vaccination, or to affect both vaccination and school grade attainment, namely, a child’s sex and year of birth, wealth and self-reported financial status of a child’s household, distance to the nearest fixed primary health care clinic and distance to the nearest mobile health clinic, distance to the nearest road, urban vs. rural residence, mother’s age, mother’s education attainment, and birth order. If vaccination and school grade attainment are conditionally independent given these observed variables, matching vaccinated to unvaccinated children on the propensity score will tend to produce unbiased estimates of the effect of vaccination on school grade attainment.  We used a range of different propensity score estimators (nearest neighbour with different calipers, radius with different calipers, and kernel with different bandwidth).

As a second estimation strategy, we used household fixed effects regressions, exploiting variation in vaccination status across children growing up in the same household in order to identify the effect of vaccination on school grade attainment.

Findings
The measles vaccination effect was significant at the 5 percent level across all matching estimators. The size of the estimates of measles vaccination effect on school grade attainment ranged from 0.15 to 0.17 grades across the different propensity score methods. None of the other vaccination effects on school grade attainment was significant. Before matching, school grade attainment in the group that received measles vaccination was 0.38 grades lower than in the control group (p < 0.001).  The household fixed-effects regressions confirmed the results of the propensity score estimation.

Policy implications 
We find that measles vaccination coverage significantly increases school grade attainment in a rural community in South Africa. The size of the effect of measles vaccination on school grade attainment is considerable. The effect size estimate based on propensity score matching equals about one child attaining one additional school grade for every six children. 

Our finding is biologically plausible. Measles infections are likely to lead to complications (otitis media, undernutrition, and neurological damage), which can reduce a child’s capacity for learning. Further, as our data show, measles vaccination coverage did not reach the level required to ensure herd immunity (i.e. 90%) in the period from 1995 through 2000, so that measles outbreaks were still possible in the community. A related recent study at the Africa Centre found that measles coverage levels in the same community did not reach the herd immunity threshold in 2006. Moreover, data from South Africa show significant rates of measles throughout the study period.

Achieving universal coverage with vaccinations of proven efficacy is desirable for many reasons, foremost to save lives and avoid disease. Because worldwide child mortality could be substantially reduced if all children received the measles vaccination, the proportion of one-year old children immunized against measles is one of the targets used to monitor the progress towards the fourth Millennium Development Goal (MDG) of reducing child mortality.

Our findings suggest that measles vaccination could also be instrumental in attaining the second MDG of achieving universal primary education. Finally, as more educated children will be more productive and inventive as adults, increases in measles coverage could accelerate a country’s economic development.