The IMaCh programme (Interpolated Markov Chain) is used to
calculate period prevalences and their variance on the basis of
age-specific transition probabilities. A multistate model is
applied to cross-longitudinal surveys. Under GPL license, the
program was written by Nicolas Brouard and Agnès Lièvre.
Documentation is also available.
Health expectancy, that can be defined as life expectancy in good health, is subject to intense research in France and in other countries as well in order to measure and define it but also to make projections as it is done for the life expectancy.
The first so-called "cross-longitudinal" survey is the American LSOA survey (Longitudinal Study on Aging) which consists of persons aged 70 and over interviewed a first time in 1984 and interviewed again in 1986, 1988 and 1990. The cross-sectional prevalence of disability at age 90 and above is higher than the period prevalence at the same ages, showing the improvements realized since 30 or 40 years in the decrease of the incidence of disability and in the recovery of health. This very important result for the projection of disability has been recently published in 2003 (1) and should be confirmed by other surveys and studies.
Other cross-longitudinal surveys are now conducted in the world as in North America (United-States LSOA II, Canada), in Asia (China, Japan) or in Europe (Italy, Great Britain, The Netherlands) and also in France (the HID survey). There are numerous difficulties in analyzing such a surveys, whether they are statistical, demographical or computational. Most of these surveys are analyzed, at least for the estimation of the health expectancies, with the computational programme called IMaCh (Interpolation of Markov Chain) developed by Nicolas Brouard and Agnès Lièvre under GPL license. The documentation of the programme and the sources of the successive versions of the software are available in the box "more information".