Challenges in comparing the quality of life of older people between ethnic groups, and the implications for national well-being indicators: a secondary analysis of two cross-sectional surveys

Policy makers world-wide are increasingly interested in promoting and measuring societal well-being, which is a dynamic, multi-faceted concept with social and psychological dimensions, overlapping with measures of quality of life (QOL) [1‐3]. A great deal of the well-being of any nation will be determined by physical and mental health and existing QOL scales are being examined as potential well-being indicators. The aim of incorporating well-being measures into governmental policy is hopefully to stimulate improvements over time and to promote equity, as defined most recently by ongoing work in the United Kingdom and Europe [4, 5]. The well-being of minority and marginalised groups, whose members are typically disadvantaged, will be key to success, but creating a set of indicators that is valid for such groups is a considerable challenge. Official statistics struggle to include accurate information on "hard to reach" groups, often relying on assumptions and extrapolation. Bajekal and colleagues reported differences between ethnic groups in seven facets of QOL among older people in a national survey in England and Wales [6]. It is unknown whether such variations are attributable to differences in priorities or true underlying QOL. Most measures have been developed largely by experts with some lay input, although more recently developed QOL scales have used a grounded approach incorporating participants' priorities [7].

We have seen that our data do not provide simple solutions to the problem of making fair comparisons between ethnic groups. Regional comparisons will similarly be affected because of differing population characteristics. There are two main limitations. Firstly, we cannot fully adjust for socio-economic status, which is known to be difficult to measure in retired people [10]. We have used housing tenure as a proxy because of its availability in the data but we recognise that it is a coarse indicator. Secondly, we cannot rule out differences between ethnic groups arising from differences between the two surveys. There were only 16 people in Omnibus who did not regard themselves as White British, not enough to allow us to test this by comparing them with Ethnibus. The Ethnibus participants were younger on average than the Omnibus, but we have adjusted for age in our analyses.

The differences seen between ethnic groups could as plausibly arise from cultural norms, expectations or semantics, as from community structures and economic disadvantage, but regardless of the reason, the fact that inter-group differences can be distorted is enough to raise concern. The WHOQOL-OLD scale was developed by modifying a generic QOL scale for adults and tested by a series of convenience samples internationally, which may explain its detection of fewer differences between ethnic groups compared to OPQOL. Before any QOL scale can be used as a well-being measure, it would need to be developed or tested in a variety of socio-cultural groups in the population, not simply analysed by groups to assess differences. The potential exists for policy informed by well-being statistics inadvertently to increase inequity or assign resources inefficiently because of over-simplification. As we have seen problems in comparing ethnic minorities' well-being, we can also expect problems in other minority and marginalised groups such as older people, children, recent immigrants, mental health service users, those whose first language is not English, those living in institutional care, homeless people and disabled people including communication difficulties.