Sepsis and heart failure (HF) are among the most mentioned hospital conditions when discussing unneeded treatment variation with health system CMOs, according to Trever Burgon, PhD. Data suggest these two conditions are also the most frequent principal diagnoses among hospitalized Medicaid patients. Additionally, research indicates that sepsis is present in one-third to one-half of all hospital deaths, with HF listed as a contributing facture in more than 10%. Dr. Burgon and colleagues sought to determine if a patient simulation-based engagement program could significantly impact physicians’ actual care decisions and reduce variation in sepsis and HF care at the bedside.

Virtual Patients

For a study published in the Journal of Hospital Medicine, Dr. Burgon and colleagues designed a library of online interactive clinical performance and value (CPV) sepsis and HF virtual patients designed to represent typical presentations and disease courses. Real-world quality and cost metrics from hospitalists at hospitals enrolled in the program were compared with non-participating hospitals that were used as control group to measure the impact of the program. Program goals were:

  1. Measure how sepsis and HF patients are cared for using the CPV simulations.
  2. Facilitate a forum for hospitalists to review their care decisions together.
  3. Reduce unneeded variation to improve quality and reduce costs.

“Physicians need engaging, effective, and efficient ways to stay current on the latest guidelines,” explains Dr. Burgon. “They want to deliver the best care, but are busy, trained at different places and at different times, and often don’t have visibility into what their colleagues are doing.”

The study team developed 12 CPV simulated cases (6 sepsis and 6 HF), with case-specific and guideline-supported scoring criteria. Possible scores ranged from 0% to 100%, with higher score reflecting greater alignment with best practice recommendations. Every 3 months over a course of six rounds, participants treated two virtual patients (20-30 minutes), received custom feedback on where they could improve, and then participated in group discussions to review the data together. All participants treated the same patients, so the study team could understand where the group differed or aligned in treatment.

Big Improvements

To analyze the impact of the program, the team compared participant performance during the first two rounds with that of the last two rounds (Table 1). “With measurement and feedback, clinicians became better at making care decisions that were aligned with the latest evidence-based guidelines,” adds Dr. Burgon. The scores showed an overall 7.8% relative increase, with improvements seen in all care domains. Improvement was significant in all domains except the workup, with the greatest improvement in diagnostic accuracy (+19.1%).

When they looked at the real-world data at the end of the first year, the team found that sepsis and HF patients cared for by participants enrolled in the program spent 892 fewer days than expected in the hospital. By the end of the 2-year program, participants in the program spent $6.2 million less than expected to care for their sepsis and HF patients, driven by more efficient, more evidence-based care. An analysis by the team showed that these improvements, above and beyond the control group, came primarily from reduced variation between the providers.

“With the CPV simulations, we could pinpoint gaps in care and how these change with feedback,” notes Dr. Burgon. “For example, we found high rates of unnecessary sputum cultures and urinary antigen testing for pneumonia patients who did not have risk factors that would indicate such testing (Table 2). Through the serial feedback, the group became less likely to order these when they weren’t indicated.”

“We have found that combining short, validated patient simulations with individual and group feedback increases guideline-based care,” says Dr. Burgon. “Over time, we saw that the clinicians were much more likely to make evidence-based care decisions in the simulations. At the end of the project, we then looked at the real-world, patient-level data and found that participating in this interactive measurement and feedback process led to significant reductions in length of stay and cost per case for HF and sepsis patients that were above and beyond improvements seen across the health system.”