A systematic review of relations between neighborhoods and mental health.

Abstract

BACKGROUND: The relationship between neighborhood characteristics and resident mental health has been widely investigated in individual studies in recent years, but this literature is not adequately reviewed. AIMS OF THE STUDY: To systematically review relevant individual research of the relation between neighborhoods and adult mental health by identifying and synthesizing all relevant studies in this literature. METHODS: We conducted an electronic search with PubMed and PsycINFO, and manual reference-checking, resulting in 8,562 screened studies of which 29 were selected. Studies were included in the main synthesis if they (i) were published in English in peer reviewed journals; (ii) had relevant definitions and measures of neighborhood characteristics; (iii) utilized standardized measures of adult mental health; (iv) controlled for individual characteristics; (v) reported quantitative results; and, (vi) studied a population in a developed country. We focused on two key areas within this literature: the methodologies utilized to study neighborhood effects and quantitative results. With regard to the former, we examined five major issues: (i) definitions and measures of neighborhoods; (ii) definitions and measures of mental health; (iii) controls for individual level characteristics; (iv) conceptual models; and (v) analytical models. As for quantitative results, the relation was reviewed by types of neighborhood characteristics. We summarized general quantitative findings and drew common conclusions across groups of studies. RESULTS: 27/29 studies found statistically significant association between mental health and at least one measure of neighborhood characteristics, after adjusting for individual factors. This association was evident for all types of neighborhood features, varying from sociodemographic characteristics to physical environment, and from objective to subjective measures. Neighborhood effects were weakened when adding individual-level characteristics into the regression models, and were generally modest relative to individual effects. DISCUSSION AND LIMITATIONS: Although the evidence is highly consistent across studies, detailed evaluation of each individual study reveals a complex reality. The included studies used various study designs and measures of mental health and neighborhoods, making generalization of their relationships less meaningful. It is not possible to conclude that studies with weaker controls for individual characteristics found stronger association between neighborhoods and mental health and vice versa. As only two studies used randomized and controlled trials, causal effects must be interpreted with caution. Formal meta-analysis techniques cannot be used due to large heterogeneity across the included studies. Efficient methods for quantitative analysis remain a great challenge. IMPLICATIONS FOR HEALTH POLICY: The current evidence suggests that efforts to improve mental health may be limited if only individual-level interventions are implemented. The calculation of the costs and benefits of neighborhood-level interventions deserves more attention. Moreover, policy makers may want to incorporate mental health as a measure for evaluating neighborhood improvement programs. IMPLICATIONS FOR FUTURE RESEARCH: There are not enough replicated or comparable studies in this literature to make more precise quantitative conclusions of this relation. Key aspects of study design and analyses could be improved to better understand the true nature of causal relationships. The data resolution of neighborhood characteristics needs to better match with the scale of neighborhood definition that is hypothesized to affect the residents' mental health. As experimental designs are rare in this area, thoughtful use of panel data, instrumental variable (IV) techniques, and other non-experimental approaches deserves further exploration.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 212 citations exist upstream.