A scoping review of methods for assessment of sex differentials in early childhood mortality

Abstract

Abstract Background While assessment of sex differentials in child mortality is straightforward, their interpretation must consider that, in the absence of gender bias, boys are more likely to die than girls. The expected differences are also influenced by levels and causes of death. However, there is no standard approach for determining expected sex differences. Methods We performed a scoping review of studies on sex differentials in under-five mortality, using PubMed, Web of Science, and Scopus databases. Publication characteristics were described, and studies were grouped according to their methodology. Results From the 17,693 references initially retrieved we included 154 studies published since 1929. Indian, Bangladeshi, and Chinese populations were the focus of 44% of the works, and most studies addressed infant mortality. Fourteen publications were classified as reference studies, as these aimed to estimate expected sex differentials based upon the demographic experience of selected populations, either considered as gender-neutral or not. These studies used a variety of methods – from simple averages to sophisticated modeling – to define values against which observed estimates could be compared. The 21 comparative studies mostly used life tables from European populations as standard for expected values, but also relied on groups without assuming those values as expected, otherwise, just as comparison parameters. The remaining 119 studies were categorized as narrative and did not use reference values, being limited to reporting observed sex-specific estimates or used a variety of statistical models, and in general, did not account for mortality levels. Conclusion Studies aimed at identifying sex differentials in child mortality should consider overall mortality levels, and report on more than one age group. The comparison of results with one or more reference values, and the use of statistical testing, are strongly recommended. Time trends analyses will help understand changes in population characteristics and interpret findings from a historical perspective.