Gender differences in the associations between health and neighbourhood environment.

There was no difference in the overall age-adjusted prevalence of less than good self-rated health between men and women. This is consistent with findings from recent studies that have found either no gender difference or that the difference between men and women was not consistent across the life course.

Between-neighbourhood differences in self-rated health were larger for women than for men. Additionally, aspects of the socio-political environment, amenities and the physical environment, and economic characteristics were more consistently associated with women's self-rated health. Both these findings suggest that the residential environment may be more important for women's health. On the other hand, individual economic activity was more strongly related to self-rated health for men. Other studies of gender differences in the effects of work and domestic conditions suggest that occupational factors are more important for men's health whereas the home environment is important for women's health. Factors such as control at home and effort–reward imbalance outside work predict depression in both men and women, with larger effects for women. In a prospective study of civil servants, Chandola, Kuper, Singh-Manoux, Bartley, & Marmot (2004) found that low control at home predicted coronary heart disease in women but not in men. However, we note that the gender differences found in these studies may be cohort- or context-specific. A study in a white-collar organisation found similar levels of work-home conflict for men and women, and similar relationships between work-home conflict and health for both sexes. The relative effects of the work and home context are not the focus of this study, but our findings add to this literature by highlighting the influence of the residential environment on women's health especially. Further work is needed to determine how features of the neighbourhood might influence social roles; the Health Survey for England and Scottish Health Survey data available within this project did not allow us to investigate this.

A strength of this study is that neighbourhood characteristics have been measured externally to the health data. These findings add to the existing literature documenting an association between perceptions of problems in the neighbourhood and perceived health because they avoid the problem of affect bias. A tendency for women to report negatively about all aspects of their life, including their neighbourhood and their health, cannot be the sole explanation for larger associations between neighbourhood environment and health in women. An alternative explanation for the gender differences seen is that the exposure to various aspects of the local environment differs between women and men. When the analyses were limited to participants in current employment, gender differences in the association between neighbourhood environment and health were less clear, with the exception of neighbourhood unemployment rate. This suggests that some of the stronger association between neighbourhood characteristics and health seen amongst women may be due to the fact that women spend more of their time in the neighbourhood (assuming that employed women and men spend the same amount of time in their neighbourhood). Data on time spent in the neighbourhood were not available, however a study of housing and health in the West of Scotland suggests that women do not spend a considerably greater amount of time in the home than men. In that study, the average time spent at home was 16.6 hours per day for women and 15.2 for men on weekdays, and 17.2 for women and 16.2 for men at weekends. Although both sex differences were statistically significant, they were smaller than expected and do not suggest a substantial exposure difference. Considering employed participants only, residence in a neighbourhood with high unemployment was associated with poorer health for women but not for men. Greater vulnerability to features of the residential environment may play an additional part in explaining the larger associations seen amongst women.

Differences between neighbourhoods in the prevalence of less than good health were fully explained by the variables captured here for both sexes. For men, between-neighbourhood differences were explained by the spatial separation of people from higher and lower socioeconomic backgrounds. For women, some differences between neighbourhoods remained after accounting for individual socioeconomic factors and the socio-political, service, and economic characteristics of neighbourhoods measured here fully explained those remaining differences. This suggests that we have not omitted important features of the neighbourhood from this study.

On the whole, correlations between socio-political aspects of the neighbourhood and the physical environment and amenities were low indicating that “social” and “physical” domains can, to some extent, be considered separately. In contrast with previous work which suggested that social and physical domains are associated with men and women's health differently, we found no clear gender division in the characteristics of neighbourhood which were most salient for health. The lack of association between social cohesion and poor quality physical environment was unexpected, given previous work in the US. It is possible that Wilson and Kelling's theory does not apply in the UK context. Alternatively, the lack of association may reflect the fact that the items used (number of missed waste collections, public sector housing vacancy rate, and vacant and derelict land) reflect institutional rather than informal lack of care over the physical environment.

Some limitations with this study should be acknowledged. As with the majority of quantitative studies of neighbourhoods and health, neighbourhoods are defined by administrative boundaries rather than by a theoretical approach. However, qualitative work suggests that many people do express some identification with the electoral ward they live in, suggesting a degree of overlap between geographically and psychologically defined communities. Some of the data capturing amenities and the physical environment used here were obtained from routine data sources and it is possible that the quality varies across neighbourhoods. We cannot rule out the possibility that measurement error has masked some true effects of the neighbourhood characteristics investigated here. Regression modelling is limited in its suitability for investigating the complex relationships between the different aspects of neighbourhood and health. Future work should use an approach which tests a priori causal pathways and considers how aspects of neighbourhoods can feedback or reinforce each other over time.

This observational study is based on cross-sectional data and so selection of less healthy participants into neighbourhoods with a poorer quality residential environment cannot be ruled out as a possible explanation for the associations between neighbourhood characteristics and health. However, we think it is unlikely that selection explains all of the associations presented here. Analysis of the 1991 Sample of Anonymous Records showed that migration did not account for the association between economic deprivation and health. Also note that socioeconomic position is likely to be an important determinant of where a person lives. This aspect of selection has been accounted for in the modelling since social class and economic activity were included in all models.

In the UK, where responsibility for looking after the home and family tends to fall on women and where social and economic resources and amenities are not equitably distributed across areas, we find that women's self-rated health is more strongly associated with features of the neighbourhood than is men's health. These findings need to be tested in other settings.

Previous studies of neighbourhoods and health have largely used a summary index based on routine data which describes concentrated deprivation. The current study has looked at different facets of the socio-political, physical and economic environment as well as local amenities as a step towards identifying points for intervention. There is evidence that each of these is associated with self-rated health over and above individual demographic and socioeconomic factors and that the associations are larger in magnitude for women. The health impact of area regeneration policies may be different for men and women. Our findings suggest that improving features of the local residential environment could be beneficial for women's health, but could have less impact on men's health.