ABSTRACT
Background: Young to middle-aged U.S. adults are burdened by the obesity, opioid, and COVID-19 epidemics. However, the ability to detect cardiovascular disease (CVD) manifestations of population-wide changes in risk factors within contemporary cohorts in young to middle adulthood is unknown. Objective(s): To assess inter-rater reliability of death certificate (DC), obituary, coroner/medical examiner (CME) autopsy report, and hospital record ion in young to middle adulthood. Setting(s): The National Longitudinal Study of Adolescent to Adult Health (Add Health), a nationally representative school-based sample of 20,745 U.S. adolescents in grades 7-12 (1994-1995) followed for the last 26 years and aged 37-45 years in 2020. Method(s): We traced all participants, identified decedents, then collected DCs, obituaries, CME autopsy reports, and hospital records = 1 month before death dates. Among a random sample of 28 decedents and an oversample of 28 hospitalized decedents enriched 3:1 for CVD, two trained and certified staff used an electronic data entry system to data needed for outcome classification from the four information sources following standardized, item-by-item instructions. We measured item-specific reliability of categorical data ion as the agreement between ors (%) and prevalence- and bias-adjusted kappa coefficient (PABAK). We measured reliability of interval-scale data (e.g. creatinine;troponin;creatinine kinase;CK-MB;pro-BNP concentrations) as an intra-class correlation coefficient (ICC). Result(s): We identified 578 (2.8%) participants who were deceased through December 2020. Of those, 577 (99.8%) had high scoring National Death Index matches uniquely identifying decedents in 44 U.S. states. We collected and ed 531 (92%), 445 (77%), 178 (66%), and 95 (39%) of their DCs, obituaries, CME autopsy reports, and hospital records. CVD was the underlying cause of death in 10% (95% confidence interval [CI]: 8%-13%) of decedents. Mean, item-specific agreement (95% CI) was 0.86 (0.84-0.89), 0.90 (0.87-0.93), 0.93 (0.92-0.95), and 0.94 (0.92-0.95) for each source. The corresponding mean, item-specific PABAK (95% CI) was 0.83 (0.80-0.86), 0.86 (0.83-0.90), 0.92 (0.90-0.94) and 0.91 (0.89-0.93). The mean, biomarker-specific ICC (95% CI) was 0.96 (0.95-0.98). Conclusion(s): Overall, CVD was a major cause of mortality and reliability of ion was excellent across a range of measures. Ongoing investigation of deaths, and as needed, targeted staff retraining and improvement of ion protocol will enable high quality studies of CVD emergence within this large, nationally representative U.S. cohort. Such studies will provide generalizable insight into the biological mechanisms underlying cardiovascular manifestations and thereby inform understanding of changing CVD burden in the U.S. population.