Objectives To offer reasons for the gender difference in care home admission risk.

Design A prospective longitudinal study collecting information from 2001 census returns, death registration, and health card registration information for a 28% sample of the Northern Ireland population, along with information from the local care home inspectorate. Analyses were performed using Cox regression models with hazard of care home admission as the outcome.

Participants 55 440 people aged 65 years or older and not living in care homes at the time of the Census.

Main outcome measures Permanent admission to a care home for older people, identified by change of address (from health card registration information) to a registered nursing or residential home (from inspectorate information).

Results In unadjusted models, women were 80% more likely to be admitted to a care home than men (HR 1.80 95% CI 1.65 to 1.96), while in fully adjusted models, the risk averaging across all living arrangements was 10% higher (HR 1.10 95% CI 1.00 to 1.20). There was however variation in the risk by living arrangements. After controlling for age, there was no raised admission risk for females among people living alone (HR 1.05 95% CI 0.93 to 1.19), or with siblings (HR 1.04 95% CI 0.64 to 1.68), however there was a higher risk when looking at the 20 972 cohort members living with a partner (HR 1.34 CI 1.14 to 1.59). There was no evidence of variation with health status of coresidents.

Conclusions Apart from age, the single biggest contribution to the raised admission risk is living arrangements. There are no apparent gender differences among people living alone or with siblings, whereas the risk is higher for women living with a partner. This suggests that the support provided within the home is different, and that women receive less support from their husbands than men receive from their wives. Further research should investigate the effect of coresident gender on living arrangements among people living with children.