# Educational gradients in the quality of mortality data: a nationwide, registry-based study on heart failure listed incorrectly as underlying cause of death in Norway

## TL;DR

Higher education levels in Norway are linked to fewer incorrect listings of heart failure as the underlying cause of death, suggesting better mortality data quality.

## Contribution

This study reveals educational gradients in the incorrect use of heart failure as the underlying cause of death in Norway over 30 years.

## Key findings

- Higher education levels are associated with lower odds of heart failure being incorrectly listed as the underlying cause of death.
- Educational gradients were more pronounced among younger individuals compared to older individuals.
- Similar patterns were observed in cardiovascular-specific mortality data.

## Abstract

In the context of mortality, heart failure (HF) represents an intermediate factor and should not be used to describe underlying cause of death (UCoD). We explored the potential educational gradients in use of HF to describe UCoD using national data spanning more than 30 years from Norway.

Using a cross-sectional design, we linked data from the Cause of Death Registry and the National Education Database. Logistic regression models were used to analyze the association between highest attained education and the odds of HF being listed as the UCoD: odds ratios (ORs) and corresponding 95% confidence intervals (CIs) are reported.

HF was listed as UCoD in 46,331 (3.7%) of 1,254,249 deaths analyzed. Compared to primary education, secondary and tertiary education were associated with 10% (OR = 0.90, 95% CI: 0.88–0.92) and 17% (OR = 0.83, 95% CI: 0.80-0.86) lower odds of HF incorrectly listed as UCoD, respectively. We observed no significant differences for the association between education and study outcomes between men and women and across place of death categories. However, educational gradients were greater among younger compared to older individuals (pinteraction, = 0.002). Similar educational gradients were observed in the analyses restricted to cardiovascular deaths (OR = 0.93; 95% CI: 0.91–0.94 for secondary vs. primary education, and OR = 0.91; 95% CI: 0.88–0.95 for tertiary vs. primary education).

Education was inversely associated with the use of HF to incorrectly describe UCoD. Addressing the observed educational gradients, would improve the quality of mortality data and allow for less biased descriptions of cause-specific mortality.