Sensitivity and specificity of dementia coding in two Swedish disease registries

Area of Science:

• Gerontology
• Epidemiology
• Medical Informatics

Background:

• Accurate dementia identification in disease registries is crucial for epidemiological research and healthcare planning.
• Swedish disease registries are valuable resources, but their accuracy for specific conditions like dementia requires validation.
• Population-based studies provide a robust gold standard for evaluating registry data.

Purpose of the Study:

• To assess the sensitivity and specificity of dementia detection in two Swedish disease registries: the Inpatient Discharge Registry and the Cause of Death Registry.
• To compare the performance of these registries against clinical diagnoses from population-based studies.
• To inform the use of these registries for dementia research and surveillance.

Main Methods:

• Utilized clinical diagnoses from two Swedish population-based studies as the gold standard for dementia.
• Calculated sensitivity (dementia detected) and specificity (non-dementia correctly identified) for the Inpatient Discharge Registry and the Cause of Death Registry.
• Analyzed data for both prevalent (existing) and incident (new) dementia cases.

Main Results:

• The Inpatient Discharge Registry demonstrated a sensitivity of 55% for prevalent dementia cases and 31% for incident cases.
• The Cause of Death Registry showed lower sensitivity compared to the Inpatient Discharge Registry.
• Specificity was high for both registries: 98% for the Inpatient Discharge Registry and 100% for the Cause of Death Registry.

Conclusions:

• The Inpatient Discharge Registry offers moderate sensitivity for identifying prevalent dementia cases in Sweden.
• The Cause of Death Registry exhibits excellent specificity but limited sensitivity for dementia detection.
• Findings highlight the need for careful consideration of registry limitations in dementia research and the potential for data linkage to improve accuracy.