Data, physician engagement key to decreasing mortality rate at UPHS

The University of Pennsylvania Health System is like a lot of major academic medical centers—high profile, prestigious and a magnet for many leading physicians in the Philadelphia metropolitan area. Yet seven years ago, the system's inpatient mortality rate was significantly higher than the national average.

With scrutiny on outcomes by payers rising, the chief medical officer at the three-hospital system launched a thorough review of why 2.5% of patients died while at UPHS, compared with a national average of 2.2%.

“We decided that we should get organized in a very deliberate way and study every death,” said UPHS Chief Medical Officer Dr. Patrick Brennan. He assembled a team of physicians, nurses and administrators to analyze outcomes and develop strategies to improve performance.

The committee's review of the system's electronic health records pointed to two glaring problems: excess mortality among patients with sepsis or delirium, the severe confusion that often accompanies hospitalization. Delirium was triggering numerous falls and breathing problems that led to unnecessary deaths.

UPHS is one of many hospitals and systems under increased pressure to decrease death rates since the CMS began requiring hospitals to publicly post outcomes data. Providers—even elite institutions such as UPHS—could lose patients when they start comparing mortality rates.

Many factors can drive up that rate, not only sepsis or delirium, and all can be prevented, said Dr. Don Goldmann, chief medical and scientific officer at the Institute for Healthcare Improvement. Hospitals must first identify the specific problems driving their own excessive mortality rate.

“Hot spots account for a very large population of the deaths and are areas for improvement in most hospitals,” Goldmann said. “Identifying these opportunities for quality improvement is vital.”

UPHS began its process by using data from the University HealthSystem Consortium to set benchmarks. UPHS then upgraded its data collection process by tapping a registered nurse to input deceased patient information into its EHR, which was reviewed by a physician.

A new mortality review committee recognized from the outset that physicians needed to shape the strategy for the quality-improvement initiatives if it was to be accepted by clinicians. Peer development—not something forced on them by the UPHS administration—was key.

Physicians began filling out forms that included their opinion on the cause of patient death and whether it was preventable. It was physicians who identified the major problems that led to poor outcomes.

The committee recognized that patients with sepsis needed faster treatment since their condition often worsened while waiting for a doctor. Staff needed training to recognize sepsis symptoms. They also identified inefficient processes that might trigger episodes of delirium, such as unsafe transports or prolonged waits when moving from the emergency room to the intensive-care unit.

The committee also found that antibiotics were underutilized. It developed an algorithm for administering drugs to combat sepsis. A pharmacist was added to the rapid-response team.

The moves worked. Hospital administrators estimate there have been 11,000 fewer deaths than expected since 2006, Brennan said. UPHS' mortality rate fell to 1.6%, lower than the national average of 2.2%.

“We had to lead by example,” said Dr. Rachel Kelz, an assistant professor of surgery at UPHS. “I don't think that people have a firm understanding of the difference each individual can make.”

Most hospitals with above-average mortality rates have the staff, the EHR and the analytical skills to significantly lower those rates. What's too often missing is the leadership and willingness to implement peer-generated solutions.

Follow Ashok Selvam on Twitter: @MH_aselvam