Sepsis is an enormous public health problem with massive associated morbidity and a staggering cost to the US health care system estimated at $62 billion annually.1 Although the precise kinetics remain controversial, there is little disagreement that delayed initiation of appropriate therapy leads to measurable increases in mortality.2
Sepsis order sets are designed to streamline delivery of guideline supported interventions to improve the quality and consistency of sepsis care and adherence to metrics such as the SEP-1 bundle. Pressure to ensure comprehensive implementation of SEP-1 bundle components is intense due to public reporting of adherence data and connection to hospital reimbursement. These mandates have been the source of much controversy due to the challenges of early sepsis recognition and rapid implementation of resource-intensive interventions in crowded emergency departments, lack of high-quality data supporting many bundle elements, concern about incentivizing overtreatment, and perceived loss of autonomy in individualizing care.3,4 While some studies show that timely bundle implementation has a small but significant impact on mortality, others have shown no difference.5,6
Quantifying care value
Amid this background comes the study by Dale and colleagues published in the November issue of the journal CHEST® on the impact of sepsis order set use on care value among hospitalized patients.7 Care value refers to patient outcomes relative to their cost and, in this study, was operationalized as a composite of hospital mortality, direct costs, and health system reimbursement.
The authors hypothesized that sepsis order sets would improve all three of these component variables, with greater reduction in costs relative to reimbursement, thus increasing the hospital’s contribution margin, a measure of financial viability and profitability.
They conducted a large, multicenter, retrospective analysis including 97,249 adult patients receiving a sepsis diagnosis during admission across 51 hospitals between 2021 and 2022. Patients were divided into groups based on whether the institutional sepsis order set was (45.7%) or was not (54.3%) utilized in their care. The order set incorporated most elements of the SEP-1 bundle, as well as increased monitoring, decision support for antimicrobial drug selection, and a real-time view of completion status of various bundle components.
The main analysis conducted after propensity score matching showed that those treated via the order set received antibiotics around 35 minutes faster, despite a negligible (but statistically significant) two-minute shorter time from antibiotic order to administration. They also had shorter duration of hypotension, shorter length of stay, lower readmission rates, and, most importantly, lower mortality (11% vs 15%). Additionally, total cost of care was $1,487 lower, while reimbursement was reduced by only $465, resulting in a net increase in contribution margin of $1,022 per patient. Both the clinical and financial benefits observed were largely driven by the subset of patients with septic shock.
Presenting counterpoints
Better patient outcomes at a lower cost to the health system may begin to sound like a windfall, but are the findings too good to be true? The authors candidly point out that the mechanism underlying the observed benefits is not clear. The etiology of the cost savings is fairly intuitive—patients treated through the order set had better outcomes and shorter lengths of stay, particularly in the cost-intensive ICU setting, ergo the cost of their care was lower. But what about the mechanism that led to these improved outcomes? Is there face validity to the premise that after decades of negative trials on targeted interventions for sepsis, simply organizing standard-of-care orders into a streamlined package is the panacea we’ve been waiting for? It seems improbable that two medically equivalent patients treated by the same provider will have a different likelihood of survival if orders are entered through an order set vs being placed ad hoc.
The authors rightly emphasize that despite robust efforts to account for confounders, causation cannot be inferred from their study design. Unfortunately, significant sources of selection bias temper enthusiasm for the results. First, patients with clearcut manifestations of sepsis are more likely to trigger order set use, whereas those in whom the diagnosis is more obscure are often more complex and may have later initiation of therapy due to delayed recognition. This form of bias is inherent to observational time-to-intervention studies and has been a source of controversy around the dogma of hourly worsening of mortality with delayed antibiotics.8 Faster receipt of antibiotics despite minimal change in time from order to administration supports earlier recognition as the driver of the improved time to treatment, rather than order set use itself. Additionally, patients in the non-order set group were less likely to meet sepsis criteria at presentation (88% vs 96%), further supporting the likelihood of atypical presentations delaying recognition in the non-order set group. Some of these patients likely met sepsis criteria after arrival to the wards where order set use may be less common than in the emergency department, the target of many sepsis quality initiatives. Some of these later diagnoses also likely represented hospital-acquired sepsis, a condition with vastly different etiologies and higher mortality than community-acquired sepsis.9
Other confounders include overrepresentation of COVID-19 in the non-order set group, a population less likely to trigger use of a sepsis order set but with significant associated morbidity and mortality during the study period. Small but significant differences in baseline characteristics between groups magnified by the large statistical power may also have played a role, especially given that comorbidities were not assessed. While propensity score matching helps to counteract the prognostic imbalance imparted by the above confounders, it is likely that other unmeasured differences remain.
The study time period was also chosen due to increased emphasis at the health-system level on order set use as one component of a broader group of interventions aimed at improving sepsis mortality, including multidisciplinary huddles, focus on early antibiotics, and education on the morbidity of sepsis. As such, order set use may have been a marker for the uptake of a variety of interventions that could have influenced patient outcomes, or perhaps even the Hawthorne effect. It also may have been a marker for individuals who or centers that were more in tune with high-quality sepsis care or generally higher performing in other ways.
Using order sets, with discretion
Despite these limitations, any intervention that proposes to save lives from a devastating disease deserves careful consideration. Use of protocolized care is well-established for other time-sensitive diseases like stroke and myocardial infarction. In sepsis, the benefit of bundled care can be traced back to early goal-directed therapy. The authors of the current study cite numerous others that have demonstrated similar cost savings and outcome benefits associated with components of their intervention, including timely implementation of care bundles, provider education, early detection systems, and decision support for antimicrobial selection. While most of these studies are retrospective and subject to many of the same limitations as the current one, the trend in the literature is clear.
Increased emphasis on order set utilization should be expected after the recent publication of “Hospital Sepsis Program Core Elements” from the Centers for Disease Control and Prevention.10 This document provides guidance on best practices for creating and maintaining effective systems for delivering quality sepsis care and emphasizes the importance of creating order sets to improve bundle compliance and optimize patient outcomes. Anecdotally, at our own institution, we have observed significant improvement in bundle compliance when the order set is utilized compared with ad hoc order entry.
As we are unlikely to see randomized controlled trials on this topic, we must applaud the authors for undertaking this ambitious and enlightening study. Ultimately, it is hard to argue that sepsis order sets are not a good thing, even if the magnitude of purported benefit may be uncertain. Order sets make it easier to ensure that the right things get done and harder to miss something important due to our human limitations.
Automating the mundane components of care delivery allows us to focus our intellect on the nuances of individualized evaluation and personalized care, where it performs best, rather than making sure we remember to order the repeat lactate or calculate the fluid bolus correctly.
While the downsides of order set use are likely negligible, several potential pitfalls should be noted. First, caution is warranted against incorporation of mandated order set use into reportable metrics such as SEP-1 in the absence of strong evidence to support them. Second, the onus remains on providers to ensure that order set components are selected thoughtfully rather than automatically out of fear of reprisal for nonadherence to mandated bundles. While it is easy to click through boxes, we must remember that each click represents an important clinical decision with downstream effects. Perhaps the fluid bolus should be decreased in the floridly hypervolemic dialysis patient and broad-spectrum antibiotics should be withheld for the tachycardic, febrile college student with the flu.
The future of sepsis care is personalized precision medicine targeting distinct phenotypes rather than a blanket, one-size-fits-all approach. As we adapt to the increasing integration of human and artificial intelligence, we must embrace the areas where technology enhances and extends our capabilities, while clinging to those where our personal touch will always make the practice of medicine a human endeavor.
References
1. Buchman TG, Simpson SQ, Sciarretta KL, et al. Sepsis among Medicare beneficiaries: 1. The burdens of sepsis, 2012-2018. Crit Care Med. 2020;48(3):276-288. doi:10.1097/CCM.0000000000004224
2. Liu VX, Fielding-Singh V, Greene JD, et al. The timing of early antibiotics and hospital mortality in sepsis. Am J Respir Crit Care Med. 2017;196(7):856-863. doi:10.1164/RCCM.201609-1848OC
3. Chen AX, Simpson SQ, Pallin DJ. Sepsis guidelines. N Engl J Med. 2019;380(14):1369-1371. doi:10.1056/NEJMCLDE1815472
4. Wang J, Strich JR, Applefeld WN, et al. Driving blind: instituting SEP-1 without high quality outcomes data. J Thorac Dis. 2020;12(Suppl 1):S22. doi:10.21037/JTD.2019.12.100
5. Seymour CW, Gesten F, Prescott HC, et al. Time to treatment and mortality during mandated emergency care for sepsis. N Engl J Med. 2017;376(23):2235-2244. doi:10.1056/NEJMOA1703058
6. Rhee C, Yu T, Wang R, et al. Association between implementation of the severe sepsis and septic shock early management bundle performance measure and outcomes in patients with suspected sepsis in US hospitals. JAMA Netw Open. 2021;4(12):e2138596-e2138596. doi:10.1001/JAMANETWORKOPEN.2021.38596
7. Dale CR, Chiu ST, Schoepflin Sanders S, et al. Sepsis order set use associated with increased care value. Chest. 2024;166(5):1046-1055. doi:10.1016/j.chest.2024.05.032
8. Weinberger J, Rhee C, Klompas M. A critical analysis of the literature on time-to-antibiotics in suspected sepsis. J Infect Dis. 2020;222(Suppl 2):S110-S118. doi:10.1093/INFDIS/JIAA146
9. Ginestra JC, Coz Yataco AO, Dugar SP, Dettmer MR. Hospital-onset sepsis warrants expanded investigation and consideration as a unique clinical entity. Chest. 2024;165(6):1421-1430. doi:10.1016/J.CHEST.2024.01.028
10. Centers for Disease Control and Prevention (CDC). Hospital Sepsis Program Core Elements. 2024.