Abstract 17464: Calibration of Heart Failure as the Cause of Death: The ARIC Study

Abstract

Introduction: Underlying causes of death (UCD) are often (mis)coded as ill-defined causes of death such as heart failure (HF), thus compromising accurate mortality estimates. Statistical redistribution of UCD has been used to examine the effect of misclassification of UCD attributed to HF, but sex- and race-specific redistribution of deaths and cause-specific mortality in the U.S. has not been examined to our knowledge. Hypothesis: We hypothesized that deaths attributed to HF would redistribute to plausible UCDs differently by sex and race in the Atherosclerosis Risk in Communities (ARIC) Study communities.

Methods: We examined North Carolina vital records from 2005 to 2010 for decedents ≥55 years old (N=312,594). A coarsened exact matching method was used to match records where HF was the UCD to records in which HF was a multiple cause of death (MCD). We matched on 5 year age groups, sex, race, education and year of death. Each death in which HF was the UCD was redistributed to plausible UCDs proportional to the frequency among the matched records. Here we include results of redistribution proportions overall and by sex and race from 1 of the 4 ARIC communities.

Results: A total of 9,005 deaths with HF as the UCD were matched to 31,357 deaths with HF as a MCD. After redistribution, the proportion of deaths significantly increased for ischemic heart disease (IHD) (1.1%), chronic obstructive pulmonary disease (COPD) (0.3%), and diabetes, hypertensive heart disease and cardiomyopathy (all 1%) compared to pre-redistribution, p<0.05. The overall pattern reflected the larger sample of whites as IHD and other cardiovascular diseases only significantly increased in blacks. Mortality attributed to IHD, cardiomyopathy and COPD increased in males and females, while diabetes and hypertensive heart disease only significantly increased in females.

Conclusion: Our data suggest that reclassification of HF deaths modify cause-specific mortality estimates, and may significantly impact mortality patterns for subgroups or regions prone to miscoding of UCD. Calibration by means of reclassification of ill-defined causes of death would improve the accuracy of mortality statistics needed by those setting public health policies and monitoring mortality trends.