Estimates of Health Adjusted Life Expectancy from the World Health Surveys using an enhanced methodology

Presenter: Stefan Baeten, Erasmus University Rotterdam

Abstract

Introduction
Promoting health is the defining goal of any health system. Health attainment in a country is an important indicator of the performance of a health system. A high level of health enables the attainment of higher productivity both for society and the individual.
However valid international comparisons of health attainment still demand much methodological work. This study aims to enhance estimation methods and provide evidence of health differences between countries.

Data
We use data from the World Health Organization’s (WHO) World Health Surveys (WHS). This dataset contains, among other data, responses from individuals from 68 countries about their health and attitudes towards health.

Methods
To obtain healthy life expectancy we combine WHO life tables with the World Health Survey 2002. The latter is used to obtain health utility for each age-sex group. Responses of self assessed health may be prone to response heterogeneity. This hampers international comparison. We correct for response heterogeneity using a Hierarchical Ordered Probit regression. We first calculate individual performance on different health domains. We aggregate these domain scores into a single measure of health utility. Next, we average health utilities by country, age and gender. Finally, these estimates are combined with life table survival data to measure health adjusted life expectancy.
This approach resembles the existing approaches. However we enhance this existing approach through (1) using more recent, improved, more representative and richer data, (2) using better defined health domains, (3) using new natural and comparable domain scales based on TTO (4) using a newly developed valuation function to aggregate health domains to obtain a health utility measure.

Results
On average health utility was higher among men than women and was highest in Irish males (0.969) and lowest in Turkish women (0.585). Health adjusted life expectancy was also lowest in Turkish women (42.0), but highest in Norwegian men (79.2).

Conclusions
We introduce adjustments to an existing methodology for the measurement of population health. These adjustments lead to different results than the currently available estimates of health adjusted life expectancy. Decomposition of the results allows a better understanding of the main drivers for country differences. Our results are comparable across different countries and give new insights about the distribution of health around the world.