Health perspectives: Exploring differential reporting across sex and generations

Abstract

In health metrics, health surveys are an important source of information, particularly for the estimation of health risk exposures and outcomes. This dissertation studies how the same survey questions are answered differently depending on whether the respondent is a mother of a child in the 90s, the child as a young adult, or whether the respondent is female or male. Topics explored were chosen as those in which sex and social norms around sex and across generations can influence how we experience health, the risk factors we are exposed to, and our relationship with morbidity. In the first chapter, Differences in reporting of child abuse by mothers and young adults, we used a longitudinal study to compare mothers’ prospective accounts of their child experiencing different forms of violence against children (VAC), and young adults’ retrospective self-reports of experiencing VAC. We then studied the socioeconomic factors associated with mothers reporting abuse of their child among children that were classified as having experienced abuse. For this end, we used the Avon Longitudinal Study of Parents and Children (ALSPAC), a 30-year prospective birth cohort study in England. This chapter addresses a well discussed issue in the field, the underreporting of VAC depending on the survey respondent. We used longitudinal data, unlike previous work where mostly cross-sectional data was used. We found that when questions are asked in the same way, there was no evidence of mothers underreporting physical or psychological abuse in comparison to children, even though there was little reliability across respondents. Among the pairs of mothers and children in which at least one of them reported abuse, we found that the sex of the child and other mother characteristics are associated with mothers’ reporting of physical or psychological abuse. Finally, the first chapter reflects on the social norms around discipline, as both mothers and the young adults described physical cruelty to be related to acts of severe physical violence, in which case acts such as pushing, smacking or kicking, would not be classified as abuse if mothers were the only respondents. Chapters two and three focus on a fundamental topic in the measurement of the burden of disease, the measurement of morbidity through disability weights. In chapter two, Differential health loss valuation by sex of the respondent in the Global Burden of Disease (GBD) study we analyzed differences by sex of the respondents in the disability weights used for the GBD study. Similar to literature focused on paired comparison questions from Martens de Noordout et al. in 2018, Liu et al. in 2020, and Haagsma et al. in 2024, we found high correlation of health preferences by sex. This translated into anorexia nervosa being the only health state for which there a was a significant difference between disability weights estimated with female only and male only paired comparison data. In contrast, the sex stratification of the population health equivalence questions resulted in significantly different disability weights for females and males in almost all health states measured. In other words, we found that in the disability weights used in the GBD study, preferences for health states do not differ by sex, but females are less willing to accept disability as health program evaluations in comparison to males. Finally, in chapter three, Differential health loss valuation by sex on population health equivalence questions (PHE), we further explored the sex differences found in chapter two and analyzed willingness to accept disability using the disability weights data from the GBD study. Through a marginal logistic model using generalized estimating equation, we found that even when we take age and education into account, females are more likely to choose programs that avoid deteriorating health over preventing death for the relative few. Before this study there was no empirical evidence on the differences in PHE valuation by sex of the respondent, mainly due to the use of PHE data as a methodological step. These questions are not used to rank health states but to anchor the preferences revealed through paired comparison questions in values that are useful for the estimation of Years Lived with Disability (0 to 1 ranges). It is in these questions, that we found females are more likely to choose the program that averts lifelong consequences of disease over programs that avert death as creating the greater population health benefit. Consequently, we estimated that if all disability weights input data were stratified by sex, female disability weights would be larger for every heath state. Notably, because 70 percent of the respondents of population health equivalence questions are female, the current set of disability weights in the GBD study reflect more the preferences of disability weights of females than males.