Background Social Fragmentation is the idea that isolation and disorganisation within an area influences individual health. Some, but not all, studies have shown it to be related to suicide and parasuicide risk, higher GHQ12 scores and higher admission rates for psychoses. The aim of this study is to determine if fragmentation per se affects mental health or if the association is due to other factors relating to fragmented areas.

Methods A measure of social fragmentation was constructed from four census variables (as per Condgon, 1996) for each of the 890 super-output areas in Northern Ireland (avg pop. 1900). These were divided into quintiles and added to the 2005 Health and Social Wellbeing Survey (HSWB) as a contextual variable. Respondent characteristics known to be associated with mental health were included such as age and sex, marital status, living alone, perceived social support, socio-economic status (based on car availability and housing tenure) and health status (based on limiting long-standing illness (LLTI)). A GHQ-12 score of 4 or more was taken as indicative of significant psychological ill health. Logistic regression analysis was restricted to 3306 individuals aged 25–74 years.

Results As expected, people in the most fragmented quintile were more likely to be unmarried and living in single person households, much more likely to be deprived, and were more likely to have a significant psychological disorder (OR 1.70, 95% CI 1.30 to 2.24), after adjusting for age and sex. Although level of perceived social support was strongly associated with GHQ12 score, adjustment for this did not significantly explain the likelihood of poor mental health across fragmentation quintiles (OR 1.44, 95% CI 1.08 to 1.91). However, adjustment for SES and LLTI completely eliminated the association between social fragmentation and psychological ill health.

Conclusions Social Fragmentation is associated with poor mental health, but only because these areas tend to be more deprived. After adjustments are made for SES, social fragmentation has no association with the likelihood of psychological disorder. It's who you are not where you live that determines mental health. However, before we completely sound the death knoll for social fragmentation we should take into consideration the recognised imperfections of the construct and modify it. Until then, policies to improve mental health should focus on reducing individual poverty and material disadvantage rather than changing the character of areas.