Reducing 30-day readmissions at a regional Alberta health authority
A 14-month study across four hospitals reduced 30-day readmissions by 17% relative to matched control sites.
Problem
The authority’s executive team suspected its post-discharge follow-up protocols were under-performing, but no rigorous evaluation existed. Funders required a defensible attribution of any reduction to the intervention rather than to secular trends in the patient population.
Approach
We designed a difference-in-differences study using two sets of hospitals: four intervention sites and four propensity-matched controls. Patient-level data from 2022–2024 anchored the baseline; the intervention rolled out site-by-site over a 6-month period.
Method
We modelled 30-day readmission probability as a function of intervention exposure, controlling for age, primary diagnosis, length of index admission, and social-deprivation index. Parallel-trends assumption verified across the 18-month pre-intervention window.
Finding
Intervention sites showed a 17% relative reduction in 30-day readmissions (95% CI: 11%–22%) compared to matched controls. The effect was concentrated in cardiac and respiratory diagnoses.
Impact
The authority extended the protocol to all 11 of its hospitals based on the evaluation. Estimated annual savings: $2.1M CAD in avoided readmissions.
Representative case study — methodology accurate, identifying details changed.