We thank Priya Vart for the critical reading of our study1 of the association of parental
education and global mortality in children younger than 5 years. Our models accounted
for study-level controls for the confounding effects of wealth or income, partner’s
years of schooling, and sex of the child. The key challenge of systematic reviews
is that they are limited by the characteristics of the data from published studies,
and this constrained our choice of covariates. This constraint brings us to a larger
point—namely, in characterising the average effect of published studies, we had to
choose an ideal study to effectively normalise all non-ideal studies to.
Lire l’article complet sur : www.thelancet.com