Measuring hospital quality: can medicare data substitute for all-payer data?

Area of Science:

• Health Services Research
• Quality Improvement
• Healthcare Administration

Background:

• Administrative datasets are crucial for quality-of-care research.
• Assessing the validity of using Medicare patient data as a surrogate for all patients is essential for accurate quality measurement.

Purpose of the Study:

• To evaluate if adverse outcomes in Medicare patients can substitute for all-patient measures in quality-of-care research using administrative data.
• To compare the consistency of quality-of-care metrics derived from Medicare-only versus all-patient datasets.

Main Methods:

• Calculated rates for 10 adverse patient outcomes across three samples: all patients (11 states), Medicare patients (11 states), and national Medicare patients.
• Examined correlations between all-patient and Medicare adverse outcome rates.
• Conducted negative binomial regressions to compare results using all-patient and Medicare data, controlling for hospital characteristics.

Main Results:

• For medical patients, Medicare adverse outcome rates were higher than all-patient rates but highly correlated, yielding consistent regression results across samples.
• For surgical patients, Medicare rates were generally higher, but correlations were lower, and regression results were less consistent between Medicare and all-patient data.

Conclusions:

• Quality-of-care analyses for medical patients using Medicare data are likely comparable to those using all-patient data.
• Quality-of-care measures for surgical patients derived solely from Medicare data require cautious interpretation due to lower consistency and comparability.