[Deprivation and mortality: a deprivation index suitable for geographical analysis of inequalities]

FOREGROUND: A national deprivation index suitable for geographical analysis of inequalities in health is not yet available in Italy, although the link between deprivation and health has been clearly demonstrated in our country in a number of studies.

1) To describe a deprivation index in Italy at municipal level, based on the percentage of selected census variables (simple components) 2) To analyse general mortality according to deprivation categories at municipal level in Italy 3) To measure the size of ecological bias using the Turin Longitudinal Study.

Italy (1-2), Turin (3).

1) 1991 census data base at municipal level 2) General mortality in Italy by municipality and age bands in 1990-92. 3) Turin Longitudinal study 1991-1995.

1) Percentages of selected indicators of inequalities for all the 8.100 Italian municipalities present at 1991 census were calculated. Factorial analysis were performed in order to help in selecting the most valuable ones. The final choice led to five indicators (low education, unemployment, rented occupier housing, no indoor bathroom, lone parent with childhood). A simple additive index was computed using the method of the sum of normal standard deviates of each component 2) Correlation of indexes with mortality was performed; SMR for each deprivation category were computed; 3) Computation of the index at different levels: individual (one million individuals), census ward (3657 wards), statistic zone (92 statistic zones), two neighborough levels (23 and 10 neighboroughs) was conducted; differential analysis was performed for each level and index category, allowing for comparison among results, adjusting for age.

Increasing deprivation was significantly associated with mortality from all causes: there was up to a 20 percent increase in mortality between the highest and lowest quintile. The relationship was linear with no apparent threshold. The slope of the relationship between deprivation and mortality varied among regions. Differentials in inequalities measured by individual and census tract level resulted to be very small; ecological bias arises when larger areas are taken into account.

An area based measure of deprivation has proved a valuable tool in examining differentials in death and is likely to prove of continuing value to health authorities in planning the delivery of health care.