Sex-stratified assessment of the concordance between self-reported diagnoses and the administrative health databases across eight cardiovascular and neurological traits

Abstract While sex-based differences are well-documented in health-related research, it remains unclear whether sex influences agreement between self-reported and administrative health data (AHD) diagnoses. Neurological and cardiovascular diseases pose substantial burdens, and present sex differences in terms of prevalence, progression and treatment response. We assessed whether AHD and self-reported data concordance differs by sex. Self-reported data was obtained from CARTaGENE, a population-based cohort in Québec, Canada, and linked with health insurance records. Agreement between self-reported and administrative data was assessed for eight neurological or cardiovascular diseases using Cohen’s Kappa. Additionally, through sex-stratified analyses, we evaluated the correlation between disease prevalence and published polygenic risk scores. Kappa estimates had non-overlapping 95% confidence intervals between the sex-stratified analyses for myocardial infarction (0.67 [0.63–0.70] in males vs. 0.48 [0.39–0.57] in females) and stroke (0.53 [0.47–0.59] in males vs. 0.38 [0.29–0.46] in females), indicating higher agreement among males. Our analysis of comparing published polygenic risk scores to this cohort in a sex-stratified manner based on either administrative health data diagnoses or responses to self-report survey questions suggested differences between female and male prevalence by score decile for myocardial infarction, hypertension, depression and multiple sclerosis. Knowledge on agreement between self-reported and health administrative data by sex helps clarify, for a given disease and sex, whether diagnoses from different sources of data can be reliably compared or integrated.