Ethnic counts on mortality and census data (mostly) agree for 2001-2004: New Zealand Census-Mortality Study update

Area of Science:

• Public Health
• Biostatistics
• Epidemiology

Background:

• Previous studies (1980s-1990s) indicated significant undercounting of Maori and Pacific deaths in mortality data compared to census data.
• The New Zealand Census-Mortality Study (NZCMS) aimed to assess if this numerator-denominator bias persists with more recent data.

Purpose of the Study:

• To evaluate the accuracy of ethnic group counts in mortality data against census data.
• To determine if undercounting of specific ethnic groups in mortality records has been resolved.

Main Methods:

• Anonymously and probabilistically linked 2001 census data with 3 years of subsequent mortality records (82,404 records).
• Compared ethnicity recording between census and mortality datasets using both 'total' and 'sole' ethnicity definitions.

Main Results:

• Using a 'total' ethnicity definition, census and mortality counts showed close agreement for Maori (0.98 ratio), Pacific (0.98), Asian (1.02), and other groups (1.01).
• A 'sole' Maori ethnicity definition revealed census counts were only 86% of mortality counts, indicating under-ascertainment of multi-ethnic individuals in census data, particularly in the South Island.

Conclusions:

• Little bias exists between census and mortality data when using a 'total' ethnicity definition.
• Mortality rate calculations by ethnicity using unlinked data and a 'total' definition are likely unbiased.
• Results support the consistent use of the census definition of ethnicity across all health datasets.