Public Health.

Area of Science:

• Epidemiology
• Gerontology
• Public Health

Background:

• Ascertaining dementia diagnoses in longitudinal studies is resource-intensive.
• Mortality registries may offer a more efficient alternative for dementia diagnosis.
• The accuracy of cause-of-death coding for dementia is not fully established.

Purpose of the Study:

• To evaluate the diagnostic accuracy of death certificates for dementia.
• To assess factors influencing the accuracy of dementia diagnosis from mortality data.

Main Methods:

• Utilized data from 6280 participants in the population-based Rotterdam Study.
• Compared in-life dementia diagnoses with certified causes of death (ICD-9 and ICD-10) from 1989-2018.
• Analyzed concordance based on coding, place of death, age, dementia duration, and comorbidity.

Main Results:

• Death certificates demonstrated high specificity (>96%) for dementia.
• Sensitivity increased from <10% (ICD-9) to ~70% (ICD-10) over time.
• Accuracy was influenced by ICD version, age at death, dementia duration, and comorbidity, particularly stroke and heart failure.

Conclusions:

• Death certificates are specific but not highly sensitive for dementia diagnosis.
• Accuracy is lower in older individuals, those with recent diagnoses, or cardiovascular comorbidities.
• Findings inform the use of mortality data versus in-person screening in longitudinal dementia studies.