Background: Deaths from suicide, as recorded within the Northern Ireland Coroner's Office for the years 2005-2011 inclusive, were analysed in terms of standardised mortality ratios (SMRs), within Wards and Local Government Districts (LGDs). The aim of the study is to examine factors relating to the ecological context of the area within which the person resided at time of death. Area deprivation, religious composition and age structure are examined in terms of SMRs, while controlling for the number of individuals living within a designated area.
Methods: Random-intercept Poisson regression models were used in conjunction with empirical Bayes prediction to examine area effects.
Results: Considerable variation occurs between the numbers of recorded deaths within each area. A strong association is shown between deprivation and the number of deaths by suicide within an area. There was considerable variation at the LGD level in terms of the number of deaths, but once the nested nature of Wards was taken into account and adjusted for level of deprivation, the variation between LGD was no longer statistically significant. When adjusted for the number of individuals within each age group, the number of deaths in the younger and middle-aged groups did not show a statistical difference (0.05 level), nor did the religious composition of the area in terms of the number of recorded deaths.
Conclusions: Based on SMRs, using empirical Bayes prediction, area effects were shown to be substantial, especially in urban locations where there are high rates of deprivation.
Keywords: poisson; Suicide; empirical bayes; northern ireland; standardised mortality ratios.