Has the Medicare Sepsis Performance Measure (SEP-1) Catalyzed Better Outcomes for Patients With Sepsis?

Abstract

In October 2015, the Centers for Medicare & Medicaid Services implemented the Medicare Sepsis Performance Measure (SEP-1), requiringU.S. hospitals to report adherence to a sepsis management bundle that includes obtaining blood cultures, checking serum lactate levels, administering broad-spectrum antibiotics, and if patients are hypotensive or have elevated lactate levels, infusing 30 mL/kg of body weight of intravenous crystalloids within 3 hours of sepsis time zero, as defined by the measure. The bundle further requires rechecking serum lactate if the initial level was elevated, starting vasopressors for persistent hypotension, and documenting a repeated volume status examination within 6 hours. The SEP-1 program has proven to be controversial. Some believe it has brought welcome attention to sepsis and catalyzed hospitals to implement protocols that save lives. Others are concerned that the measure s rigid treatment requirements and aggressive timelines encourage excessive fluids and antibiotics, curtail clinicians latitude to calibrate their management to their certainty of infection and patients severity of illness, and unduly anchor providers diagnostic focus on sepsis to the exclusion of other serious conditions (1). Critics have also questioned the evidence supporting SEP-1. Supporters point to observational studies that report lower mortality rates after implementing sepsis bundles. However, these studies are difficult to interpret because bundle implementations are inevitably accompanied by awareness campaigns that increase the detection of milder cases of sepsis. This in and of itself lowers overall sepsis mortality rates, making it difficult to discern the effect of the bundle versus the effect of increased sepsis ascertainment. A new study in Annals provides welcome real-world data on the effect of SEP-1. Barbash and colleagues (2) used detailed electronic health record data to determine the effect of SEP-1 on processes of care and risk-adjusted outcomes from 2013 to 2017 among 11 academic and community hospitals affiliated with the University of Pittsburgh. They used an electronic clinical definition to define possible sepsis (orders for clinical cultures and concurrent organ dysfunction within 6 hours of arrival to the emergency department) to make the study less susceptible to ascertainment bias, an important strength. Using a sophisticated time series analysis, they reported that SEP1 was associated with a 50% increase in lactate checks (70% observed vs. 47% expected if pre–SEP-1 trends had continued), a 10% increase in broad-spectrum antibiotic administration (50% vs. 45%), and a 30% increase in infusions of 30mL/kg of intravenous fluid (13% vs. 10%) within 3 hours of culture orders. Despite these significant changes in processes, there were no significant changes in hospital mortality or the percentage of patients discharged to home. There are several possible explanations for SEP-1 s failure to affect outcomes in this study. First, the specific components of SEP-1 are controversial. Its primary effect in this study was on lactate testing, but elevated lactate levels are not specific for infection, and checking lactate in and of itself does not change a patient s prognosis. Randomized trial data suggest that following lactate levels is no better than careful perfusion assessments and may in fact lead to over-resuscitation, particularly for patients who are less sick (3). Overadministration of fluids is potentially harmful, and the data on time to antibiotic administration and outcomes are nuanced (4). Many observational studies report associations between delays and higher mortality rates, but these studies are at high risk for bias because the timing of antibiotics is not random. Patients with severe and obvious sepsis often receive antibiotics the soonest; this sometimes leads to early antibiotics being associated with higher mortality rates (5). Patients with ambiguous syndromes tend to receive antibiotics later; this has also been associated with higher mortality rates, but it is unclear if this is because of antibiotic delay or because patients with less obvious presentations also tend to have more comorbidities and serious noninfectious conditions that mimic sepsis (6). The 1 randomized trial that assessed administering antibiotics in the prehospital setting versus the emergency department found no difference in mortality (7). Some studies suggest a clearer relationship between time to antibiotic administration and mortality for septic shock, whereas others do not (5). A second possible reason for the limited effect of SEP-1 is because sepsis preferentially affects patients with severe comorbidities whose deaths are difficult to prevent once they become septic, even with optimal hospital-based care (8). Devastating deaths in young and healthy persons have helped catalyze sepsis measures and mandates, but patients hospitalized for sepsis are rarely perfectly healthy. A third possible reason for the limited effect of SEP-1 is misdiagnosis. Diagnosing infection and knowing whether organ dysfunction is due to infection or some other cause can be difficult. About a third of patients treated for serious infections in the emergency department or intensive care unit turn out to have noninfectious conditions (9, 10). Even among those who are infected, not all have bacterial infections. The most common cause of sepsis is pneumonia, but a third or more of cases of pneumonia are caused by viruses. The COVID-19 pandemic has dramatically brought home the fact that viruses can cause severe multisystem illness and that antibiotics and large volume fluid resuscitations are not appropriate for all patients with sepsis. Finally, the study was done in 1 health care organization that has long been a leader in emergency and critical care medicine. Results from this hospital system may not be generalizable to other organizations. Further studies