A Hybrid Markov-SPC Approach to Assess Cost Differences in Urgent Care Utilization Using Patient-Reported Data in...

Cost is a key outcome in quality and value, but it is often difficult to estimate reliably and efficiently for use in real-time improvement efforts. We describe a method using patient-reported outcomes (PROs), Markov modeling, and statistical process control (SPC) analytics in a real-time cost-estimation prototype designed to assess cost differences between usual care and improvement conditions in a national multicenter improvement collaborative-the IBD Qorus Learning Health System (LHS).

The IBD Qorus Learning Health System (LHS) collects PRO data, including emergency department utilization and hospitalizations from patients prior to their clinical visits. This data is aggregated monthly at center and collaborative levels, visualized using Statistical Process Control (SPC) analytics, and used to inform improvement efforts. A Markov model was developed by Almario et al to estimate annualized per patient cost differences between usual care (baseline) and improvement (intervention) time periods and then replicated at monthly intervals. We then applied moving average SPC analyses to visualize monthly iterative cost estimations and assess the variation and statistical reliability of these estimates over time.

We have developed a real-time Markov-informed SPC visualization prototype which uses PRO data to analyze and monitor monthly annualized per patient cost savings estimations over time for the IBD Qorus LHS. Validation of this prototype using claims data is currently underway.

This new approach using PRO data and hybrid Markov-SPC analysis can analyze and visualize near real-time estimates of cost differences over time. Pending successful validation against a claims data standard, this approach could more comprehensively inform improvement, advocacy, and strategic planning efforts.