Implementing INTERACT in Veterans Health Administration Community Living Centers: A pragmatic randomized trial

Hospital transfers from VA Community Living Centers (CLCs) are common. The objective of this study was to evaluate the effect of introducing the Intervention to Reduce Acute Care Transfers (INTERACT) program into VA CLCs.

Cluster randomized trial involving 16 pair-matched VA CLCs.

Intervention CLC nursing staff were trained in the use of INTERACT tools designed to identify early signs of a clinical change in condition and improve communication and documentation. One tool was embedded into the VA Electronic Medical Record. Intervention staff were supported by bi-weekly calls over 18 months to reinforce INTERACT tool use.

The primary outcome for intent-to-treat analyses was the rate of all-cause hospitalizations per 1000 person days. Secondary outcomes, assessed through structured record reviews and algorithms, were intervention CLCs change in 1) potentially inappropriate transfer decisions and 2) potentially preventable transfers. CLC staff implementation and engagement in INTERACT were documented.

Only five of the eight intervention CLCs substantially engaged in the intervention. Using a negative-binomial regression with random effects, adjusting for month, intervention, and the interaction of time and the intervention, we observed no statistically significant difference between intervention and control facilities in all-cause hospitalizations. This was confirmed with matched resident-level, as-treated, analyses among residents in the five engaged CLCs and their matched controls. Structured implicit review of intervention CLC's medical records revealed low rates of inappropriate transfer decisions both before and after the intervention.

Introducing INTERACT into volunteer VA CLCs did not reduce the rates of all-cause hospitalizations. In both the pre- and post-intervention periods, all-cause hospitalization rates were relatively higher, and inappropriate transfers lower in VA CLCs than commonly observed in community NHs. Low rates of potentially inappropriate transfers and higher nurse and physician staffing in CLCs may explain why INTERACT was not implemented as fully as planned.