8.2 Factors that impact child development

We already know many of the factors that influence early child development. A higher level of education in the family promotes development. Infectious diseases like malaria slow down growth. Access to adequate nutrition, clean water and a stimulating, prosperous and safe environment is favourable for healthy development. And so on. Unfortunately, we do not have data on most factors, so we need to limit ourselves to a few background characteristics.

Table 8.1: Comparisons between on-track and off-track development
On-track
Off-track
n % n %
sex female 21136 0.9773873 489 0.0226127
male 20805 0.9721963 595 0.0278037
birth weight <2500gr 3388 0.9482228 185 0.0517772
>2500gr 36375 0.9779277 821 0.0220723
maternal education no education 1907 0.9665484 66 0.0334516
any primary 11764 0.9672751 398 0.0327249
any secondary 21576 0.9772182 503 0.0227818
higher secondary 6263 0.9841295 101 0.0158705
residence rural 1251 0.9889328 14 0.0110672
semi-urban 2236 0.9898185 23 0.0101815
urban 18740 0.9706827 566 0.0293173
metropolitan 11122 0.9793942 234 0.0206058
* Exludes children with missing DAZ or missing factor

 

Table 8.1 compares the frequency distributions of various factors for children on-track versus off-track. There are only tiny differences between boys and girls. Children with low birth weight (< 2500 gr) are more at risk for off-track development. This estimate does not correct for gestational age. We discussed techniques for such corrections elsewhere.

The influence of maternal education on off-track development follows the expected trend. Interestingly, it seems that a rural environment could prevent off-track development. We note that original measures of maternal education and residence were harmonised across studies. It would, therefore, also be interesting to study the impact per cohort using the actual factor coding.

Relative importance of the explanatory factors in this study

Figure 8.2: Relative importance of the explanatory factors in this study

We predicted DAZ by linear regressions with predictors country, sex, birth weight, maternal education, height for age and residential area. The percentage of explained variance was 11 percent. Figure 8.2 depicts the relative contributions of the individual factors to the prediction. Country differences explain over half the variances, followed by maternal education. Contributions of height-for-age (HAZ), low birth weight and residence are about equal in magnitude.

These analyses only scratch the surface. It is nowadays common to analyse the impact of interventions on height and HAZ by multivariate techniques and machine learning methods. The D-score and DAZ are drop-in replacements that allow similar procedures to study which factors contribute to healthy child development worldwide.