Profiling hospitals based on emergency readmission: a multilevel transition modelling approach.

Area of Science:

• Health Services Research
• Biostatistics
• Health Informatics

Background:

• Emergency readmissions are a key UK government policy indicator for healthcare quality.
• Current performance profiling methods face statistical challenges, including hospital size variations and limited risk adjustment data.
• Patients and the public require reliable data on healthcare organization performance.

Purpose of the Study:

• To introduce a novel multilevel transition model framework for evaluating hospital performance.
• To develop a new performance index based on hospital-specific readmission propensities.
• To address statistical limitations in existing healthcare performance measurement.

Main Methods:

• A multilevel transition model was developed to analyze hospital performance.
• A dataset of over 5 million English patient readmissions (1997-2004) was utilized.
• To manage computational intensity, 1000 random samples were extracted, and a Grid implementation was employed for efficiency.

Main Results:

• Analysis of 1000 samples indicated that 4 out of the 5 worst-performing hospitals for cancer patients were located in London.
• The study identified specific hospital performance measures based on first, second, and further readmissions.
• The developed framework provides a new performance index for healthcare organizations.

Conclusions:

• The multilevel transition model offers a robust approach to hospital performance assessment, addressing prior statistical concerns.
• Findings highlight potential regional disparities in cancer patient readmission rates within England.
• The study underscores the need for refined performance metrics in healthcare, particularly for complex patient populations.