Estimating the Prevalence of Dementia in New Zealand Using Capture-Recapture Analysis on Routinely Collected Health Data

Abstract

Objectives

Community based dementia prevalence studies are expensive and resource intensive. Aotearoa New Zealand (NZ) has never had a community based dementia prevalence study representing all major ethnic groups. In recent years, dementia prevalence estimates have been derived from routinely collected health data but issues of underdiagnosis and undercoding limit their utility. Capture-recapture techniques can estimate the number of dementia cases missing from health datasets by modelling the ascertained overlaps between linked data sources.

Methods

Three routinely collected national health data sets—interRAI, Public hospital discharges, and Pharmaceuticals—were linked and all prevalent cases of dementia in NZ for the year 1 January 2021–31 December 2021 were identified. Capture-recapture analysis fitted eight loglinear models to the data, with the best fitting model used to estimate the number of prevalent cases missing from all three datasets.

Results

We estimated that almost half (47.8%) of dementia cases are not present in any of the three datasets. Dementia prevalence increased from 3.7% to 7.1% (95% CI 6.9%–7.4%) in the NZ 60+ population and from 4.9% to 9.2% (95% CI 8.9%–9.6%) in the NZ 65+ population when missing cases were included. Estimates of missing cases were significantly higher (p < 0.001) in Māori (49.2%), Pacific peoples (50.6%) and Asian (59.6%) compared to Europeans (46.4%).

Conclusions

This study provides updated estimates of dementia prevalence in NZ and the proportion of undiagnosed dementia in NZ, highlighting the need for better access to dementia assessment and diagnosis.