Local neighbourhood and mental health: Evidence from the UK

Abstract

This paper examines the association between neighbourhood and levels and changes in common mental disorders. Using data from a large scale nationally representative survey of individuals and households (the British Household Panel Survey), it locates individuals in their local neighbourhoods. These are defined as the nearest 500–800 persons centered around each individual in the survey. These ‘bespoke’ neighbourhoods are characterised according to five dimensions—disadvantage, mobility, age, ethnicity and urbanness—derived from factor analysis of the census characteristics of the residents of neighbourhoods in 1991. These dimensions measure characteristics of place that have been argued to be associated with mental ill health.

The paper estimates multilevel models of the level and 5-year changes of common mental disorders (measured by the twelve item version of the General Health Questionnaire). Three and two level models are estimated, all of which allow for individual and household characteristics that may act as confounders of any neighbourhood effect. The results show the extent of association between neighbourhood and both levels and changes in mental health is limited. Less than one percent of the variance across individuals is at the neighbourhood level. The neighbourhood characteristics are not generally statistically associated with levels or changes in mental ill health. There is some evidence of interaction between neighbourhood characteristics and gender and ethnicity, but while statistically significant these interactions are small in size compared to the main effects of individual and household characteristics. What appears to be important for levels of common mental disorders are the observed characteristics of individuals and their households, not of place.

Introduction

The interest in neighbourhoods and mental health has a long history. As early as 1939, Faris and Dunham (1939) examined the relationship between neighbourhood and patients admitted to hospital for psychiatric problems and concluded that there was a link between the disorganisation of the neighbourhood and mental disorders. Despite this early interest, evidence on the association between neighbourhood and mental health is somewhat limited, particularly compared to the much larger body of evidence of the effect of local area on other outcomes such as poverty, unemployment, and out of wedlock childbearing (e.g. Dietz, 2002).

Evidence is also limited by the fact that there has been little systematic exploration of whether those living in poor communities are sicker because they tend to be of lower socio-economic status, or because there is something unhealthy about living in such communities (Ellen, Mijanovich, & Dillman, 2001). As the location of individuals is not random, studies which omit individual characteristics are limited in the extent to which they can identify neighbourhood effects. It is also important in studies of area effects that the micro-individual outcome (conditioning on individual characteristics) and the macro-neighbourhood characteristics are modelled simultaneously.

Of the small, but growing, number of studies which meet these criteria (that they include data on both individual and neighbourhood characteristics in the analysis and use modelling techniques which allow for joint effects of individual and neighbourhood characteristics), the majority have examined the relationship between the economic context of the area in which the individual lives and their mental health. Measures of economic context include socio-economic disadvantage, poverty and income inequality. Several recent US studies (for example, Goldsmith, Holzer, & Manderscheid, 1998; Ross, 2000; Silver, Mulvey, & Swanson, 2002) find an association between the disadvantage of the geographic area, as measured by census tract, and poor mental health (depression or schziophrenia), as well as greater substance abuse (Goldsmith et al., 1998; Silver et al., 2002). Analyses from a recent voucher experiment in the US (Moving to Opportunity) also found decreased rates of depression amongst women (and children) after moving from high to lower poverty neighbourhoods (Katz, Kling, & Liebman, 2001). Recent UK studies have found an association between region (Cox et al., 1987; Dorling & Gunnell, 2003; Duncan, Jones, & Moon, 1995; Lewis & Booth, 1992) and neighbourhood, defined as electoral ward, (Weich, Holt, Twigg, Jones, & Lewis, 2003; McCulloch, 2001) and mental health, but all these studies have also found that after controlling for the characteristics of the individuals in these regions or neighbourhoods, this association was not statistically significant. There is less evidence that mental illness is associated with socio-economic inequality at the neighbourhood or census tract level (Muntaner, Eaton, Miech, & O’Campo, 2004), but Kahn, Wise, Kennedy, and Kawachi (2000) found that state level income inequality was associated with higher odds of depressive symptoms in women, net of individual income.

The quality of social networks and social cohesion, as distinct from socio-economic status, may be particularly important to health (Ellen et al., 2001). Yet there have been relatively few studies of the effect of these aspects of area or neighbourhood that also control for individual characteristics. Some studies have focused on youth, and the results of these studies are mixed and, in some cases, causality difficult to establish, because the quality of networks is based on respondent assessment (Ellen et al., 2001). Silver et al. (2002) examined the relationship between neighbourhood mobility and poor mental health and found that neighbourhood residential mobility was associated with higher rates of schizophrenia, major depression and substance abuse disorder. They tested for, but did not find, any evidence that the effects of individual characteristics on mental disorders varied significantly across neighbourhood characteristics. Ross, Reynolds, and Geis (2000) found an association between neighbourhood mobility and depression/anxiety, but also found that it was sensitive to the level of poverty in the neighbourhood. A recent study of youth in one city in Columbia found that there was only one aspect (trust) of social capital which was important for mental ill health once measures of poverty (poor education and employment) were controlled for, and once measures of violence were included, even this association become insignificant (Harpham, Grant, & Rodriguez, 2004).

However, despite this growing research interest, several methodological issues exist which make unambiguous interpretation of the results of studies of the impact of neighbourhood on difficult. Ellen et al. (2001) draw attention to the methodological challenges of measuring relevant neighbourhood conditions and capturing non-linear effects. With respect to neighbourhood conditions, there are two related issues. First, few studies have gone beyond conventional census measures of deprivation. Second, because it is difficult to find data sets which include both information on an individual's health status and the characteristics of neighbourhood, many studies define neighbourhood or community as large areas (Ellen et al., 2001). In most US studies that also control for individual characteristics, the smallest spatial scale used is the census tract, containing around 4000 individuals. Of the UK studies which have examined mental health and neighbourhood, the smallest spatial scale is the electoral ward, containing around 2400 persons (McCulloch, 2001; Weich et al., 2003). True neighbourhoods may consist of only a few streets round an individual's home: wards and census tracts are considerably larger than this.

Non-linear effects may be present both in terms of the responses by different individuals to neighbourhoods, and in the fact that the effect of neighbourhood on outcomes may not be linear. Research on mental health has identified the importance of different responses to neighbourhood conditions by women and individuals from different ethnic backgrounds (e.g. Chen, Subraminian, Acevedo-Garcia, & Kawachi, 2005; Weich et al., 2003). Ross et al. (2000) has highlighted the importance of interactions between deprivation and other aspects of neighbourhood.

In this paper, we explicitly address the issues of neighbourhood size and measures of neighbourhood characteristics. We examine the impact of neighbourhood using a very small definition of neighbourhood and measure the attributes of these neighbourhoods on five dimensions that have been found to be associated with poor social outcomes and, in some cases, poorer mental health. Using a panel data set of around 9000 individuals, which contains detailed measures of the sample individuals’ mental health, their demographic and socio-economics characteristics and those of their household (Taylor, Brice, Buck, & Prentice-Lane, 2003), we construct ‘bespoke’ neighbourhoods (Buck, 2001; Johnston & Pattie, 2004) that contain the nearest 500–800 people to each individual in the sample. Using factor analysis of characteristics from the 1991 UK census we measure neighbourhood type on five dimensions—socio-economic disadvantage, population mobility, demographic structure, ethnicity and urbanness—based on the characteristics of the population in the neighbourhood.

This measure of neighbourhood size is considerably smaller than that used in recent US and UK studies. A neighbourhood of 500 persons is approximately one tenth of the size of the census tract measure used in recent US studies (e.g. Ross, 2000; Silver et al., 2002), and at a fifth of the size of the ward definition used in recent UK studies (McCulloch, 2001; Weich et al., 2003). In addition, this study measures more dimensions of neighbourhood characteristics than have been used in large-scale studies to examine the relationship between health and place in both the UK and the US to date. The closest UK study (McCulloch, 2001) examines only one of the five dimensions analysed here: that of deprivation.

Using these measures, we investigate the relationship between neighbourhood and the levels and changes in mental ill health, controlling for individual and household characteristics that may be associated with neighbourhood type. Changes in mental ill health have received little attention in the UK: using large scale data they have only been examined by Hauck and Rice (2004) who did not examine neighbourhood effects at all.¹ We estimate models that allow for within household correlation of errors and for the clustering of individuals within neighbourhoods.

We also address the second of the methodological issues raised by Ellen et al. (2001) by allowing for non-linearities, both in the impact of neighbourhood, and in the response to neighbourhood of different individuals. We allow for neighbourhood to have a non-linear impact on outcomes and, given the findings of differential responses to neighbourhood conditions by gender, ethnicity, and education, we allow for differential response by women, non-whites and the less well educated.