How do disparities in health change with economic development?

Abstract

Background: Socio-economic disparities in health are persistent, pervasive and difficult to modify. Disparities, particularly for non-communicable chronic diseases, appear to change with economic development and may be specific to the stage of economic development and/or epidemiological transition. Most evidence concerning health disparities comes from Europe and North America, where contextual or cultural effects may have built up over centuries of economic development. Evidence from other socio-historical settings, with a much more recent history of economic development, can be particularly valuable. Uniquely, the Hong Kong population has already experienced transition from pre- to post- industrial living conditions over one lifetime, and as such serves as a unique setting to examine health disparities during economic development. In this context, I assessed the association of small area-based (tertiary planning unit (TPU)) income with potential years of life lost (PYLL), life expectancy and mortality over 30 years of rapid economic development, and the association of the level of TPU-based income inequality with mortality. I also assessed the contribution of major causes of death to any changes. Finally, to provide additional insight concerning the observed changes I took a novel approach and examined whether any changes overall and by cause of death related to population wide exposures (period effects), generational exposures (cohort effects) or differences with aging.

Methods: I used concentration indices to assess neighbourhood (TPU-based) income disparities in PYLL. I used abridged life tables to assess changes in life expectancy by income group. I used negative binomial regression to assess the association of TPU-based income and the level of TPU-based income inequality with mortality rates during the period. Finally, I used an age, period, cohort model, stratified by income group, to identify the contribution of age, period and cohort to changes in mortality disparities from 1976 to 2005 in Hong Kong.

Results: The concentration index for all-cause TPU-based PYLL was negligible in 1976, but increased over the period. PYLL attributable to injury and poisoning was consistently associated with lower TPU-based income, but PYLL attributable to cardiovascular diseases and cancer reversed from an association with higher TPU-based income in 1976 to an association with lower TPU-based income in 2006. Life expectancy at 0-4 years increased by 8.1 years during that period, but gains were greater for people with higher TPU-based income (10.1 years) than lower TPU-based income (6.2 years). A positive association of neighbourhood (TPU-based) Gini with all-cause mortality became more evident during the later period. Disease-specific cohort effects contributed to the widening TPU-based income disparities in Hong Kong from 1976 to 2006; the lower TPU-based income group tended to be more strongly affected by population wide exposures, such as economic downturns.

Conclusion: Health disparities widened in Hong Kong during a period of rapid economic development from 1976 to 2006. Social patterning of major chronic diseases, such as cardiovascular disease, emerged, partially as a disease-specific cohort effect. Disparities in health in a developed setting may partially be the result of long-term disease-specific processes that have accumulated over generations, with corresponding implications for contemporaneous interventions.