Validation of death certificate diagnosis for coronary heart disease: the Atherosclerosis Risk in Communities (ARIC) Study

• Published in: Journal of clinical epidemiology 2001, Vol.54 (1), p.40-50
• Main Authors: Coady, Sean A, Sorlie, Paul D, Cooper, Lawton S, Folsom, Aaron R, Rosamond, Wayne D, Conwill, David E
• Format: Article
• Language: English
• Subjects: Abstracting and Indexing as Topic - standards; Adult; Age Distribution; Aged; Bias; Biological and medical sciences; Cardiology. Vascular system; Cause of Death; Community surveillance; Coronary Disease - classification; Coronary Disease - diagnosis; Coronary Disease - mortality; Coronary heart disease; Death certificate; Death Certificates; Female; Heart; Hospital Mortality; Humans; Male; Maryland - epidemiology; Medical Records - standards; Medical sciences; Middle Aged; Minnesota - epidemiology; Mississippi - epidemiology; Mortality; North Carolina - epidemiology; Population Surveillance - methods; Residence Characteristics; Sensitivity and Specificity; Sex Distribution; Sudden death; Surveys and Questionnaires; Validity
• ISSN: 0895-4356; 1878-5921
• DOI: 10.1016/S0895-4356(00)00272-9

The validity of the death certificate in identifying coronary heart disease deaths was evaluated using data from the community surveillance component of the Atherosclerosis Risk in Communities Study (ARIC). Deaths in the four ARIC communities of Forsyth Co., NC; Jackson, MS; Minneapolis, MN; and Washington Co., MD were selected based on underlying cause of death codes as determined by the rules of the ninth revision of the International Classification of Diseases (ICD-9). Information about the deaths was gathered through informant interviews, physician or coroner questionnaires, and medical record abstraction, and was used to validate the cause of death. Sensitivity, specificity, and positive predictive value of the death certificate classification of CHD death (ICD-9 codes 410–414 and 429.2) were estimated by comparison with the validated cause of death based on physician review of all available information. Results from 9 years of surveillance included a positive predictive value 0.67 (95% CI 0.66–0.68), sensitivity of 0.81 (95% CI 0.79–0.83), and a false-positive rate (1-specificity) of 0.28 (95% CI 0.26–0.30). Comparing CHD deaths as defined by the death certificate with validated CHD deaths indicated that the death certificate overestimated CHD mortality by approximately 20% in the ARIC communities. Within subgroups, death certificate overestimation was reduced with advancing age (up to age 74), was consistent over time, was not dependent on gender, and exhibited considerable variation among communities.