Understanding Hospital Performance Metrics

In this discussion, I’ll share a narrative about well-meaning initiatives that have gone awry within hospitals across the United States. The primary challenge? Sepsis.

Sepsis stands as the foremost cause of death for hospitalized individuals, claiming approximately 300,000 lives annually in the U.S. Timeliness in diagnosis and treatment is crucial; the sooner we identify and treat a patient suffering from severe sepsis, the greater their chances of survival. Our critical time frame for saving lives in such situations is hours, not days.

In recognition of the urgent need for swift action, various states, particularly New York, have introduced mandatory sepsis protocols and public outcome reporting. The Center for Medicare Services (CMS) has also acknowledged this issue, emphasizing sepsis care as a fundamental quality measure. They publish data on how well hospitals adhere to sepsis protocols and the outcomes achieved. For instance, here’s the data from my own hospital:

At first glance, this seems promising. After all, transparency can lead to improved care. However, there’s a significant complication.

The principal metric these organizations use to evaluate hospital performance concerning sepsis is inpatient mortality. This outcome measure appears logical until you delve into the underlying data.

Recent research published in JAMA Network Open highlights the complexities of measuring hospital effectiveness.

The study analyzed Medicare data covering 2.5 million older adults admitted for sepsis or septic shock from 2011 to 2019. Researchers examined patient outcomes based on whether they were treated at safety-net hospitals or not.

Patients entering safety-net hospitals, even those with sepsis, generally present with more severe health issues compared to those at other facilities. They often have additional comorbidities and less effective pre-hospital care. Despite these adjustments, the findings reveal a troubling pattern: inpatient mortality rates for sepsis are significantly higher in safety-net hospitals. Specifically, 28.2% of sepsis patients in safety-net facilities succumb during their hospital stay, compared to 26.4% in other hospitals.

This data raises concerns about the efficacy of our critical safety-net hospitals, suggesting a systemic failure. However, this perceived disparity does not stem from inferior care; rather, it reflects the differing approaches of non-safety-net hospitals regarding end-of-life care.

Any physician working extensively in an ICU can recount the various types of cases they encounter. Some patients show rapid improvement, while others linger in a precarious state between life and death for days or weeks, referred to as “meta-stable.” For these patients, the trajectory can be grim, marked by escalating treatments and a gradual decline. What options do we have for such patients?

In non-safety-net hospitals, the protocol typically involves palliative care consultations, discussions with families, and transitioning patients to hospice care. Notably, patients who pass away in hospice—regardless of whether the care occurs within an acute care setting—do not contribute to the inpatient mortality statistics reported by CMS. Conversely, safety-net hospitals, often lacking sufficient palliative resources, record these patients as inpatient fatalities.

This disparity largely explains the differences in sepsis outcomes between safety-net and other hospitals. The study indicates that if we used 30-day mortality rates as a measure instead of inpatient mortality, the results would be nearly identical.

The following breakdown illustrates the 30-day death rates, categorized by hospital type, highlighting the tendency of non-safety-net hospitals to refer dying patients to hospice care.

I am not suggesting that the increased hospice transfers are a strategy to manipulate statistics. On the contrary, I believe that appropriate hospice care is a valuable outcome. However, publicly criticizing safety-net hospitals for their lack of hospice services does little to resolve the issue.

Why not utilize 30-day mortality as a more appropriate metric? It would be beneficial. Critics may argue that tracking this measure is more challenging—hospitals excel at recording deaths occurring during care, but it can be more difficult to ascertain death dates once patients return to the community. Nonetheless, this study demonstrates that with proper integration of national and state databases, it is feasible.

Let’s take a moment to recognize the efforts of our safety-net hospitals. Despite operating with fewer resources, less staffing, and limited access to supportive services, they are achieving outcomes comparable to wealthier institutions when treating sepsis. Given these circumstances, they deserve significantly more respect than they currently receive.

A version of this commentary was first published on Medscape.com.

Chapter 2: The Future of Hospital Care

The first video discusses why inpatient mortality is not a reliable measure for assessing hospital performance, highlighting biases against safety-net institutions.

The second video explores the potential for "Hospital at Home" models, which may become a prevalent care option in the near future.