Health Expectancy

Health expectancy is a general term referring to the entire class of indicators that extend the classic life expectancy by dividing the total number of life years – which is the area under the survival curve – into two quality dimensions, i.e., life years spent in good health and those spent in poor health:

Health expectancy indicators have become important instruments for public health. To properly use and interpret these indicators, one should be aware of how they are computed and what are the underlying health measures. Indeed, there are as many variants of health expectancy as health measures, and each of them corresponds to particular health issues and needs. The most common health expectancy indicators are disability-free life expectancy, chronic disease-free life expectancy and life expectancy in good perceived health.

Examples and explanations to better understand health expectancies and their use can be found here:
EHEMU Team (2007). Interpreting health expectancies. EHEMU Technical Report 2007_1.

Sullivan Method

The method developed by Sullivan (1971) is a very simple and frequently used method to compute health expectancies. It combines data on the number of life years from a regular period life table and data on the prevalence of a given health state from a cross-sectional survey, e.g., disability. The age-specific prevalence is directly applied to the age-specific person-years of the life table, providing the total number of years spent with disability, the total number of years lived without disability, and summing both, the total number of years lived.

Once the table is modified, the period life expectancy is calculated in the traditional manner, according to various states of disability. In this way, one can obtain a series of total life expectancy, disability-free life expectancy and life expectancy with disability.

The main advantage of the Sullivan method is the separate collection of mortality and disability data and the wide availability of these data. Basic cross-sectional surveys are sufficient to collect the observed prevalence of disability within the population. However, the problem of this method is the approximation of the period prevalence by the observed prevalence of disability.

Original publication:

For further information and guidance in applying the Sullivan method, see:

Multistate Method

The multistate life table method has been proposed by Rogers and colleagues (1989) in order to take into account the recovery of lost functions and the return to a state of good health. Data comes from longitudinal surveys which allows the calculation of the following probabilities for a period of several years:

  • onset of disability or other health problems
  • recovery from disability or other health problems
  • probability of dying for people who had initially disability, probability of dying for people who were initially disabled and probability of dying for people who were initially healthy.

Based on this set of probabilities, multistate life tables can be computed by simulating from age to age the risks of entering disability, recovering and dying and deriving the person-years with and without disability.

The advantage of this method – based on transitions between states of health – is that it gives a period indicator that takes into account the reversibility of disability. The specific drawback of the multistate life table method is the scarceness of adequate data. Moreover, biases arise when the gaps between successive waves of longitudinal surveys are too long, thus failing to capture a part of the flows between health states between the survey waves.

Original publication:

For methods and programs to compute multistate life tables see: