Exploring differential response to an emergency department-based care transition intervention

Area of Science:

• Health Services Research
• Patient Outcomes
• Clinical Trial Analysis

Background:

• Care transition interventions aim to improve patient outcomes after emergency department (ED) visits.
• Identifying patient subgroups with differential responses is crucial for optimizing intervention effectiveness.
• Previous analyses often rely on pre-specified subgroups, potentially missing novel insights.

Purpose of the Study:

• To identify multivariable patient subgroups with varied responses to a nurse-delivered care transition intervention post-ED visit.
• To employ an emerging data-driven method, model-based recursive partitioning (MoB), for subgroup discovery.
• To compare data-driven findings with traditional pre-specified subgroup analyses.

Main Methods:

• Secondary analysis of a randomized controlled trial (RCT) involving 512 participants.
• Applied model-based recursive partitioning (MoB) using 37 baseline variables to identify treatment effect heterogeneity.
• Compared MoB results with pre-specified subgroups (ED super-user status, sex, age) using logistic regression interaction terms.

Main Results:

• MoB analysis identified marital status as a key factor in differential treatment response.
• Unmarried patients receiving the intervention showed reduced repeat ED visits (22% vs. 34% usual care).
• Married patients receiving the intervention had increased repeat ED visits (27% vs. 12% usual care). No significant interactions were found for pre-specified subgroups.

Conclusions:

• Patient social factors, like marital status, may significantly influence care transition intervention effectiveness post-ED.
• Data-driven approaches, such as MoB, can uncover unexpected subgroup effects missed by traditional analyses.
• Findings highlight the need for personalized care transition strategies to improve ED patient outcomes.