Identifying Communities Which May Benefit from Targeted Resources for Child Maltreatment Prevention: A Spatial and Temporal Analysis

Background: Child maltreatment (CM) remains a pervasive public health problem which can have a lifelong impact on survivors, resulting in increased risk for chronic conditions, psychological distress, and early death. Many studies have shown that the overall rate of CM has been decreasing over the past 20 years, however less is known about changes in CM rates at local levels. State agencies charged with child protection are increasingly confronting limited resources for funding maltreatment prevention and service provisions for families. Determining how to distribute these limited funds in order to provide targeted support to local areas who are in need still proves to be a difficult task for state and local agencies.

In order to address these concerns and add to knowledge about these issues, this study will explore spatial and temporal variation in CM rates in Virginia and identify clusters of high and low CM rates among localities. Methods demonstrated in the study could be used by any local or state agency to help locate communities which may benefit from more targeted funding for prevention efforts.

Method: We used administrative data of CM rates aggregated by county, collected by the Virginia Department of Social Services. Data includes between 5,000 to 9,000 annual incidents of founded CM reported between 2000-2013. The data set reports the total number of incidents of CM by county, the calculated rate of founded maltreatment for each county, as well as the population of children in each county for each reporting year. First, using QGIS 2.1, exploratory maps were created to visually explore patterns in CM rates. Next, GeoDa software was used to examine data for spatial autocorrelation (Moran’s I, Queen contiguity weights) in addition to completing local indicators of spatial association tests for each year of available data. Results were compared in order to examine local trends in CM rates over time as well as to identify clustering of high/low rates of CM.

Results: Results from Moran’s I tests demonstrated that CM rates remained significantly spatially correlated during the study period and generally strengthened overtime. This indicates that the clustering of high/low CM rates may be a result of spatial effects and are not due to chance. For example, data from 2000 resulted in Moran’s I = 0.118, z = 1.99, p = 0.03, data from 2008 resulted in Moran’s I = 0.402, z = 6.70, p = 0.001 and data from 2012 resulted in Moran’s I = 0.280, z = 4.49, p = 0.002. In addition, over the course of 13 years, significant clusters of high CM rates were maintained in the Western part of the state while low CM clusters were found throughout the Northern and Central regions.

Conclusion: CM rates remained significantly clustered in particular areas of the state throughout the study period, indicating a potential need for targeted prevention efforts. Our study demonstrates how local agencies can harness spatial analysis tools in determining need for targeted prevention programs or service delivery. Implications for research, policy and practice will be addressed.