Addressing the health and social services needs of homeless families: The application of geospatial methods towards identifying ideal shelter locations in Boston

INTRODUCTION

Homelessness can be a devastating experience for families, disrupting aspects of family life, physical and emotional health of family members, and interfering with children’s proper education and development. Recent evidence has confirmed that family homelessness is the fastest growing segment of the homeless population in every single state in the country, including Massachusetts. Through the state Emergency Assistance (EA) program, Massachusetts offers emergency housing to any family who applies for housing support. Since the Great Recession, the number of homeless families has risen by 78% in the state, burdening the existing shelter system and forcing the state to house families in motels and hotels. With an average spending of $3,000 per household for every month a family stays in a motel, there is a dire need for the state to ascertain a more cost-effective solution.

METHODS

This retrospective review concerned data for families experiencing homelessness from the 2012 Annual Point-in-Time Count of the Homeless in Boston, gathered by the Emergency Shelter Commission of the Boston Public Health Commission (BPHC). In addition, to conduct the spatial statistical analyses, demographic data was interlinked from the 2010 American Community Survey, 2010 Census, and various sources from the City of Boston. All the data collected were analyzed using ArcGIS version 10.2 and Stata version 13.  

RESULTS

Kernel density analysis showed that most cases of family homelessness are concentrated in the neighborhoods Roxbury and Mission Hill. Spatial global pattern analysis by nearest neighbor resulted in nearest neighbor ratio of 0.75, with Z-score of -5.59, p-value of <.01 and a z-score of -5.59. Spatial autocorrelation (Moran’s I) showed clustering significant with p<0.001,  z-score 3.14 and Moran’s Index of 0.007. Finally, a hotspot analysis was conducted to map clusters of spatial significance, relating a mismatch between salient needs (mental health and domestic violence services, shelters serving families, low-cost healthcare services) and causes (lack of affordable housing, concentrated poverty, and single-parenthood) of family homelessness, as described in the literature. Hot and cold spots were identified, with hotspots pinpointed on the southeast side of Boston, at the Roxbury-Dorchester interface and the Mattapan-Dorchester interface.

CONCLUSIONS

This study demonstrates significant spatial patterns for the ideal placement of a shelter for homeless families in Greater Boston. Knowledge about these spatial patterns can provide useful information to policymakers in the planning of housing, medical, and social services for homeless and underserved populations. Incorporating data from multiple sources into spatial models allows investigators to visualize and leverage information in innovative ways.