We assume that, at a certain period of time (for example, a given year), every living individual of a population on which we want to calculate the Sullivan indicator can be unambiguously described as being in exactly one of a fixed number of states of health.
Consider a collection of individuals classified in increasing W+1 age groups (numbered from i=0 to i=W) and a particular health state which has been located on some of them. Suppose you want to have a health indicator on the whole population which doesn't depend on its age structure. Using a common notation , let:
The health state indicator proposed by Sullivan can be calculated as:
Values of can be estimated from a cross-sectional health survey on the whole population. and are provided by an ordinary current life table.