Exploring the limitations of age-based models for health care planning

Health care decision makers are required to make planning decisions over a medium to long term planning horizon. Whilst population ageing is an important consideration for planners, age-stratified demographic models may produce misleading estimates of future resource requirements if the actual relationship between age and health is not fixed. We present a methodology which tests whether the assumption of a fixed age–health relationship is valid and estimate the magnitude of planning errors using a long time-series of measures of chronic health and service utilisation (N = 2419) taken from the Great British General Household Survey (1980–2008). We find that age-only models contain significant omitted variable bias, and that the relationship between age and health varies significantly across birth cohorts. Chronic sickness has fallen across birth cohorts born between 1890 and 2008, particularly before birth year 1930. Generational health improvements have mitigated the effects of population ageing, meaning that the population rate of sickness fell between 1980 and 2008. Planning based only on age leads to overestimation of the population level of health care need if successive cohorts are becoming healthier. Many alternative approaches exist which allow planners to relax the assumption of a fixed relationship between age and health.