SOCIOECONOMIC CIRCUMSTANCES AND THE RISK OF BOWEL-CANCER IN NORTHERN-IRELAND

Objective - To describe the variation in the incidence of colorectal c ancer across Northern Ireland and relate it-to factors associated with community deprivation. Design - This was a cross sectional descriptiv e study. Setting - Incidence data were obtained fi-om a population bas ed register for the period 1990-91. Small areas were characterised by their ''affluence'', or lack of it, by deriving a Townsend deprivation score for each electoral ward, using information from the 1991 census . Participants, main outcome measures, and statistical methods - The a ge standardised incidence was calculated for all colorectal cancer cas es diagnosed histologically in 1990-91. Electoral wards were grouped i nto quintiles of the population after ranking of their Townsend scores and the association with incidence was studied using Poisson regressi on. Results - The age standardised colorectal cancer incidence ranged from 22.5 (for quintile 1) to 29.9/100 000 (quintile 5) for men but th e trend for women was less regular and rates were 18.4, 23.8, 27.3, 26 .5, and 23.9/100 000 for quintiles 1-5 respectively (that is, from the most ''affluent'' to the most ''deprived'' fifths of the population). After adjusting for age and sex in Poisson regression, there was a si gnificant association between the total colorectal cancer incidence an d levels of community deprivation. The rate ratio for the most deprive d quintile of the population (compared with the least) was 1.28 (95% C I 1.06,1.53). The effect was stronger for rectal cancer than for colon ic cancer. There was no association between community deprivation and the cancer stage at-diagnosis. Conclusions - In this population, the c olorectal cancer incidence is associated with the level of material de privation. The disease stages at the time of diagnosis in patients fro m more deprived areas seem to be comparable with those of patients fro m affluent areas. As others have shown, associations such as these are not explicable entirely on the basis of the distribution of known ris k factors. Further research is needed to determine plausible mechanism s for the association.