The general framework of the surveys used to derive prevalence ratios also need to be the same to allow comparisons. The sample of the unhealthy population is very sensitive to the way to proceed interviews (face to face interview, telephone interview...). Some surveys exclude institutionalized persons. In that case, a bias is implied (notably) for older population.

To take into account the fact that the french survey (and many other national surveys) excluded the Institutions, and as most of institutionalized persons are disabled, we should adapt our calculation of the age-specific prevalence ratios.

So, let :

- = the institutionalization ratio in the whole population
for age group
*i*. - = the prevalence ratio of disability derived from the
survey (also for age group
*i*).

The new prevalence ratio is a weighted mean of the prevalence in Institutions (which is 1) and outside Institutions (which is given by the survey, .

The new formula 2 is then:

Then, the obtained indicator would be interpreted as an approximation of future times spent non institutionalized and in good health.

Fri Apr 25 22:40:35 DFT 1997