Can Adding Laboratory Values Improve Risk-Adjustment Mortality Models Using Clinical Percutaneous Cardiac...

Area of Science:

• Cardiovascular Interventions
• Health Services Research
• Medical Informatics

Background:

• Percutaneous coronary intervention (PCI) registries are used for risk-adjusting provider mortality rates.
• Current registries lack substantial numerical laboratory data for comprehensive risk adjustment.

Purpose of the Study:

• To evaluate the impact of appending laboratory data to existing PCI registry data.
• To compare the performance of risk-adjustment models with and without laboratory variables.

Main Methods:

• Statistical models were developed using New York PCI registry data (2008-2010) for 20 hospitals.
• Models predicted in-hospital/30-day mortality, comparing registry-only data versus registry plus laboratory data.
• Model discrimination, calibration, and hospital quality outlier status were assessed.

Main Results:

• Risk-adjustment model discrimination was similar (C-statistic 0.898 vs 0.908, P=.40).
• Laboratory data inclusion did not significantly alter hospital outlier status.
• Differences emerged in identifying hospitals with significantly higher risk-adjusted mortality.

Conclusions:

• Appending laboratory data did not substantially improve PCI mortality risk-adjustment model performance.
• The inclusion of laboratory variables did not dramatically alter hospital quality assessments.
• Further evaluation is needed on the benefits and drawbacks of incorporating laboratory data into PCI registries.