Ethnic mortality estimates for the UK – how reliable are they?

Pia N. Wohland, Hull York Medical School
Stefanie Doebler, Queen's University Belfast

The ethnic composition of the UK is changing and diversity is increasing. By 2011, about 20% of the population in England and Wales was not White British. Still, information on mortality for ethnic groups, an important population health indicator, is not routinely collected. This even though numerous UK health studies found varying health outcomes by ethnic group and immigrant mortality also varies significantly between groups. In the course of developing population projection for UK ethnic groups, we estimated ethnic mortality by combining information on the spatial distribution of ethnic groups and overall mortality with our Geographical Distribution Method (GDM). We found significant variations in mortality between groups, with the majority of ethnic minorities experiencing higher mortality compared to the White British majority. But how reliable are these GDM estimates? Even though we unearthed other studies using similar methods to estimate ethnic mortality, to our knowledge no validation of results has ever been undertaken. Northern Ireland is the only UK home country that has a 100% linkage of census population to the death register, with data held in the Northern Ireland Mortality study (NIMS). Traditionally only a small number of ethnic minorities lived in Northern Ireland compared to England and Wales, but the proportion of ethnic minorities doubled since 2001 and reached about 32400 by 2011. Nationally large ethnic groups in 2011 were Chinese, Indian, Mixed, Other Asian and slightly smaller Irish Travellers, with big enough numbers to robustly estimate mortality in these populations using NIMS data. This presentation sets out to compare mortality estimates for ethnic groups using GDM with ethnic mortality estimates derived from actual data. Demonstrating the effectiveness of GDM is not only immensely useful for estimating ethnic mortality, but could encourage applying this method in any setting were demographic intensities are not actually measured.