Background Increasing numbers of elderly persons reside and die in institutions, yet there are few studies that analyse the effect of this on mortality in small areas and its ensuing effect on the association between material deprivation and mortality.

Methods A cross-sectional, ecological study in the region of Madrid covering 3906 census tracts (median 1000 inhabitants), using mortality data for 1996–2003 and socioeconomic deprivation from the 2001 census. Standardised mortality ratios (SMR) were calculated for each census tract. Using the Besag–York–Mollié model, RR of dying and their 95% CI according to the deprivation index considered (with the fourth quartile, Q, being the most unfavourable situation) were calculated for deaths among: the total population and the population excluding residents who died in institutions.

Results 6% of the deceased had been residing in institutions, which affected 16.5% of census sections (644) and accounted for 17% of the variability in SMR among men and 10% among women, p<0.001. Mortality increased with socioeconomic deprivation, whereas the RR for the total population in Q4 with respect to Q1 was 1.46 among men (95% CI 1.41 to 1.50) and 1.12 among women (95% CI 1.08 to 1.17), these figures rose to 1.48 (95% CI 1.43 to 1.53) and 1.14 (95% CI 1.10 to 1.18), respectively, for the population excluding residents who died in institutions.

Conclusions Deaths of residents in institutions affect the variation in small-area mortality, and confound the relationship between mortality and socioeconomic deprivation. This variable should be recorded in mortality statistics so that its effect can be controlled for in subsequent analyses.