Close
Kaveri Harriss and Sarah Salway, London School of Hygiene & Tropical Medicine
(Project no. 30025)
This project aims to explore the economic and social consequences of limiting long-term illness (LLI) in adults of working age, in ethnic minority households. LLI may have serious implications for the well-being of a household, and for the perpetuation of poverty across generations. If illness affects a person's capacity to work, it can be a kind of economic shock, and families must make adaptations to cope with it (Andersen and Bury, 1988). Existing longitudinal research indicates that adult limiting illness is significantly associated with household poverty dynamics. Adults in certain ethnic minority groups in Britain are more heavily afflicted by LLI than whites, particularly Pakistanis and Bangladeshis (Nazroo, 1997). They are also more likely to be economically inactive due to permanent sickness (own calculations from 1991 SARS). Deprived ethnic minority households may have less access to the multiple resources that must be mobilised to cope with a limiting illness, such as financial and physical assets or social resources.
An 'ethnic penalty' in the economic impact of ill-health has been demonstrated elsewhere. In the USA, black people with LLI are more likely to be unemployed than white people with LLI (Bound et al, 2003). In Germany, foreign-born people with LLI are more likely to end up unemployed than healthy people, but in German-born people there is no difference in employment attachment according to health status (Arrow, 1996). This has not been examined in Britain. The position of different ethnic groups in the labour market varies hugely, with Bangladeshi, Pakistani and Afro-Carribean people most likely to be in low skilled manual work (Modood et al, 1997), which is the type of work most adversely affected by LLI (Lindholm et al, 2002). This study would examine whether there are ethnic differences in the risk of unemployment or downward social mobility conferred by LLI. The study would also examine whether any ethnic differences in the risk of adverse employment outcomes can be attributed to the different age structures and socio-economic positions of these ethnic groups, or whether there is any additional 'ethnic penalty' in the effects of LLI. Besides ethnic differences in socio-economic class, employer prejudices against disability may be compounded by racism and present additional barriers to remaining in employment (Molloy et al, 2003).
A second question is whether a high burden of adult illness is having an impact on the next generation. Qualitative research indicates that providing care for sick adults can pull children (particularly girls) out of school (Howard, 2001, Phillipson, Ahmed et al, 2003). A sample of children in 1991 would allow me to analyse whether parental illness influences educational achievement in children, whether there are ethnic differences in this effect, and whether these differences persist after adjusting for age and socio-economic position.
The data will be analysed using cross-tabulations and chi-squared tests, Mantel-Haenszel chi-squared tests and multiple logistic regression (where the sample size permits).
To examine the livelihoods consequences of LLI we need ideally to use cohort data, to follow people up over time. The LS is the only dataset with significant numbers from minority ethnic backgrounds that would allow us to be able to look at ethnic differences in the consequences of illness. A conservative estimate, based on half the 1991 SARS and assuming 30% linkage failure, gives 8183 'white', 159 'black', 159 'Indian', 121 'Pakistani/Bangladeshi' and 89 'other' people with LLI aged 25-49. The estimated number of children aged 8-16 in 1991 is 42,678 after assuming a conservative 30% linkage failure. Even using the LS, there may be problems of small sample sizes when making ethnic comparisons, but I would use a higher level of aggregation where necessary.
The LS would allow us to follow people up with known health status for ten years, a period over which significant associations between LLI and later adverse employment outcomes were found in a comparable study in Norway (Elstad and Krokstad, 2003). The 1991 census was the first to ask about LLI, and we are now in the position to be able to analyse what has happened to these people by 2001. The LS also has household demographic and socio-economic modules, which would allow us to examine the intersection between ill-health, ethnicity and socio-economic background, as well as examine what happens to other household members. I also hope to make use of the new 2001 questions on self-perceived illness and caring, to look at how stable LLI is over the ten year period and whether it seems to have an impact on care-giving by other household members.