Background Despite considerable efforts to improve healthcare quality and patient safety, broad measures of patient outcomes show little improvement. Many factors, including limited programme evaluations and understanding of whether quality improvement (QI) efforts are sustained, potentially contribute to the lack of widespread improvements in quality. This study examines whether hospitals participating in a Veterans Health Affairs QI collaborative have made and then sustained improvements.

Methods Separate patient-level risk-adjusted time-series models for two primary outcomes (hospital length of stay (LOS) and rate of discharges before noon) as well as three secondary outcomes (30-day all-cause hospital readmission, in-hospital mortality and 30-day mortality). The models considered 2 years of pre-intervention data, 1 year of data to measure improvements and then 2 years of post-intervention data to see whether improvements were sustained.

Results Among 130 Veterans Affairs hospitals, 35% and 46% exhibited improvements beyond baseline trends on LOS and discharges before noon, respectively. 60% of improving LOS hospitals exhibited sustained improvements, but only 32% for discharges by noon. Additional subgroup analyses by hospital size and region found a similar performance across most groups.

Conclusions This quasi-experimental evaluation found lower rates of improvements than normally reported in studies of QI collaboratives. The most striking observation was that a majority of hospitals increased their rates of discharges before noon, but after completing the collaborative their performance declined. Future work needs to qualitatively and quantitatively assess what organisational features distinguish those hospitals that can improve and sustain quality.