1004 — Using Spatial Analyses to Inform Implementation Activities
Lead/Presenter: Mary Bollinger,
COIN - North Little Rock
All Authors: Bollinger MJ (Center for Mental Health & Outcomes Research, CAVHS, NLR AR), Landes, Sara J, Center for Mental Health & Outcomes Research, CAVHS & Behavioral Health Query, VISN 16 South Central Mental Illness Research Education and Clinical Center (MIRECC),
When implementing evidence-based practices with limited resources at a broad level, such as a large multi-site healthcare system, it can be difficult to identify where to implement first. Community-level data can be used to identify initial implementation sites and to understand any variation in the use of evidence-based practice (EBP) once scaled up fully. Though not broadly used by implementation scientists, spatial modeling can provide context data to inform scale up of EBPs -- identifying where to target both interventions and evaluation activities in areas with uneven uptake. Using suicide as an exemplar, we show how sites could be selected for the implementation of suicide prevention activities (i.e., peer outreach).
Veteran self-directed violence (SDV) data was obtained from VA Center of Excellence for Suicide Prevention for enrollees with a primary address in VISN 16 or 17. To estimate the effect of the community context, rates of suicide by county and state were acquired from the CDC along with data from the 2017 County Health Ranking including opioid prescribing, rurality, the proportion of the population in deep poverty, the proportion of high school graduates, and the rate of firearm deaths. Geographic/map data was obtained from the US Census Bureau. County specific rates SDV rates were obtained using Bayesian disease mapping methods while controlling for community context variables.
Multiple clusters of counties with higher than average SDV rates were identified once we controlled for the geographic relationship between counties. Controlling for community context and geographic relationships reduced the number of clusters of SDV rates indicating that community context does influence SDV behavior.
Although we only looked at SDV as an outcome, it is strongly associated with suicidal behavior. Initial suicide prevention activities would be strengthened by targeting areas where SDV behavior was clustered to determine how best to strengthen community supports for Veterans to reduce suicidal behavior.
Implementation of EBPs could more efficiently be accomplished knowing where initial investments could be made to address Veteran needs. Spatial analyses are particularly helpful in identifying "hot spots" where services/programs are critically needed and can provide insight into the factors involved in diffusion and implementation.