2017 HSR&D/QUERI National Conference
4037 — A Goal-Driven Approach to Aggregating Measures of PACT Performance
Lead/Presenter: Sylvia Hysong, COIN - Houston
All Authors: Che XX (Houston VA HSR&D Center of Innovation and Baylor College of Medicine, Houston, TX; Florida Institute of Technology, School of Psychology, Melbourne, FL)
Brown C (Houston COIN and Baylor College of Medicine)
Petersen LA (Houston COIN and Baylor College of Medicine)
Hughes AM (Houston COIN and Baylor College of Medicine)
Kuebeler M (Houston COIN and Baylor College of Medicine)
Hysong SJ (Houston COIN and Baylor College of Medicine)
The proliferation of clinical performance measures in healthcare has led to increasing reliance on composites to make clinical and operational decisions and to provide performance feedback. Guidance exists for different aggregation strategies, yet current practice lacks guidance for aggregating data when performance data is comparative. The current study proposes and demonstrates four strategies for aggregating multiple comparative performance indicators into a composite. As a case example, we employed care coordination improvement metrics for Patient Aligned Care Teams (PACTs) designed using the Productivity Measurement and Enhancement System (ProMES), an approach for monitoring and improving individual productivity, team effectiveness, and organizational performance.
In a VA HSR&D-supported care coordination intervention with PACTs, thirty-three participant teams provided monthly data on eight coordination indicators to generate and demonstrate four aggregation algorithms. Later, they were used to aggregate performance data according to the purpose of the assessment of the team performance 1: calculating absolute difference; 2: calculating interval percentage difference; 3: calculating the difference between max and min level percentage; 4: calculating Z-score difference.
Preliminary results showed that different aggregation strategies produced different ranking orders, in terms of teams' performance on coordination indicators. We found Algorithm 1 is best suited when PACTs need to identify the area where they have the most room to improve. Algorithm 2 should be adopted when PACTs aim to balance their performance in all areas. Algorithm 3 is ideal when PACTs wish to achieve acceptable performance in all areas while balancing performance for all indicators. Finally, Algorithm 4 is most appropriate when making cross-team comparisons.
Each data aggregation strategy exhibited both strengths and weaknesses. The goals of the performance feedback should guide which algorithm strategy to use in creating the composite.
This study evaluated four strategies for aggregating clinical performance indicator data for providing feedback in a coordination intervention for VA PACTs. Each algorithm was generated based on different feedback purposes. Using this approach, VA decision-makers can choose the approach best suited for their team's improvement needs, and thus significantly improve Veteran care though more evidence-based information sharing among the teams.