Impact of Sepsis Mandates on Sepsis Care: Unintended Consequences

Abstract

The creation of dedicated sepsis guidelines and their broad dissemination over the past 2 decades have contributed to significant improvements in sepsis care. These successes have spurred the creation of bundled care mandates by major healthcare payers, such as the Center for Medicare and Medicaid Services. However, despite the likely benefits of guideline-directed sepsis bundles, mandated treatments in sepsis may lead to unintended consequences as the standard of care in sepsis improves. In particular, the heterogeneous spectrum of presentation and disease severity in sepsis, as well as the complexity surrounding the benefits of specific interventions in sepsis, argues for an individualized and titrated approach to interventions: an approach generally not afforded by care mandates. In this review, we review the risks and benefits of mandated care for sepsis, with particular emphasis on the potential adverse consequences of common bundle components such as early empiric antibiotics, weight-based fluid administration, and serum lactate monitoring. Unlike guideline-directed care, mandated care in sepsis precludes providers from tailoring treatments to heterogeneous clinical scenarios and may lead to unintended harms for individual patients.

sepsis, septic shock, mandate, bundle, SEP-1

The creation of clinical practice guidelines is an important endeavor, as it disseminates specialized research and expertise to a wide practice audience, provides benchmarks for quality, and sets standards to which individual providers and organizations can aspire. However, care must be taken to separate the recommendations of guideline-directed care from the requirements of mandated care, especially if the outcomes of such mandates drive financial incentives in the competitive marketplace of healthcare. Furthermore, informed providers must believe that guidelines and mandates are in the best interest of individual patients, if they are to feel comfortable using them in common clinical practice.

For the past 2 decades, the Surviving Sepsis Campaign (SSC) has been the predominant force in establishing and disseminating care guidelines in sepsis management. The SSC first released guidelines in 2004, which have been updated regularly at 4-year intervals. Guided in part by the trial of early goal-directed therapy by Rivers et al [1], the first sepsis care bundles released in 2005 recommended 6- and 24-hour time cutoffs for all listed interventions; in 2015, these were revised to 3- and 6-hour care bundles, and more recently an hour-1 bundle has been proposed by the SSC, although these recommendations were criticized by many major provider organizations [2, 3].

The benefits of SSC-recommended bundles were underscored by a wealth of observational data demonstrating improvements in patient outcomes with expedited, goal-directed and bundled care for sepsis [4]. In the largest study of the effect of the 6- and 24-hour sepsis bundles, among almost 30 000 subjects at >200 hospitals, there was a significant association between high bundle compliance at individual sites and hospital mortality rate, and 10% increases in the proportions of patients who met 6- and 24-hour bundle compliance were associated with overall 5% and 3% improvements in hospital mortality rates, respectively. Furthermore, at the individual patient level, complete compliance with the 6- and 24-hour bundles was associated with reductions in hospital mortality rates of 18% and 14%, respectively.

These results have been replicated in other studies [4, 5], as well as a meta-analysis of studies involving sepsis bundle compliance [6]. Studies performed after the transition to 3- and 6-hour bundles have similarly noted improved mortality rates with bundle compliance [7]. Based on these positive data, bundled care has been widely accepted for its overall benefits in furthering the goal of timely recognition and treatment of sepsis. However, critics have noted that the supporting literature was predominantly observational and retrospective in nature, which may have introduced biases in favor of bundled care. In addition, 3 large randomized controlled trials of protocolized care did not demonstrate benefit over usual care [8–10], although patients in the “usual care” groups likely benefited from the widespread recognition among providers of the importance of early detection and treatment in sepsis.

MANDATED SEPSIS CARE

Based on the strongly positive results described above and with the intention of further disseminating best practices in sepsis care, bundles derived from the SSC were subsequently incorporated into a core quality measure by the Center for Medicare and Medicaid Services (CMS SEP-1[11]; Boxes 1 and 2). Although the intention of these mandated bundled care measures, to create incentive for focusing providers on early recognition and treatment in sepsis, is admirable, there have been many voices of criticism. These criticisms of SEP-1 can be broadly distilled into 2 areas. First, providers are concerned that the mandate of expediting care within specific time periods will force them to make rapid therapeutic decisions without enough supporting data. Second, many worry that mandated bundled care will deprive first-line clinicians of the ability to make individualized treatment decisions, especially if the evidence for a given treatment is weak or lacking. Both of these concerns ultimately derive from the often-nonspecific initial presentation of sepsis and the broad spectrum of comorbid conditions among septic patients.

BOX 1.

SEP-1 RECOMMENDATIONS FOR SEPSIS^a

Within 3 hours of presentation:

Measure serum lactate

Obtain blood cultures before administering antibiotics

Administer antibiotics

Within 6 hours of presentation:

Repeat serum lactate measurement if initial lactate level is >2 mmol/L

^aAdapted from the American College of Emergency Physicians [11].

BOX 2.

SEP-1 RECOMMENDATIONS FOR SEPTIC SHOCK^a

Within 3 hours of presentation:

Measure serum lactate

Obtain blood cultures before administering antibiotics

Administer antibiotics

Resuscitation with crystalloid fluids (30 mL/kg)

Within 6 hours of presentation:

Repeat volume status and tissue perfusion assessment

Vasopressor administration (if hypotension persists after fluid resuscitation)

^aAdapted from the American College of Emergency Physicians [11].

Given the recent implementation (2015) of SEP-1, the literature on compliance and outcomes is relatively limited. In a systemic review of the evidence underpinning the SEP-1 guidelines, it was noted that no moderate- or high-quality studies suggested improvements in patient outcomes with SEP-1 compliance, although benefit was seen in a few low-quality trials, per the CMS framework for evaluating evidence [12]. In one study involving 851 sepsis cases, there was a 33% pass rate for SEP-1 compliance, and crude mortality rates were higher in noncompliant than in compliant cases; however, this association was not seen after adjustment for clinical characteristics and severity of illness [13]. There is also evidence of significant variability in SEP-1 compliance among different healthcare institutions; smaller, nonacademic hospitals seem better able to comply with SEP-1 measures than academic and safety-net hospitals, probably owing at least in part to patient mix [14]. There are concerns that financial repercussions due to lower SEP-1 compliance rates could have a detrimental effect on safety net hospitals in particular [15].

Providers also struggle with the changing definitions of sepsis and septic shock in the context of these mandates, with the greatest change occurring between the prior SSC guideline of 2012 (on which the SEP-1 mandate was based) [16] and the “Sepsis-3” guidelines of 2016 [17]; there are significant differences in these populations in terms of illness severity and mortality rates [18]. Moreover, the strength of evidence for the efficacy of various mandated interventions varies across the severity spectrum, with the strongest support in septic shock; the Infectious Diseases Society of America Sepsis Task Force highlighted these issues in a 2018 article [19]. Delivering optimal individualized care within the rigid structure of core quality mandates such as SEP-1, even when major provider guidelines do not fully concur on the various definitions of sepsis and septic shock, poses significant clinical as well as regulatory challenges.

The remainder of the current article focuses on the potential unintended consequences of mandated sepsis care, with specific appraisal of 3 fundamental diagnostic or therapeutic recommendations in the SEP-1 mandate: empiric antibiotic therapy, fluid administration, and lactate measurement. As with much of sepsis care, what may be helpful in one subset of patients can be potentially harmful in another; we will discuss the appropriateness of each treatment strategy in the context of the spectrum of sepsis severity. Note that many aspects of sepsis care, including timely source control, choice and route of vasopressor therapy, and the administration of other supportive care (eg, mechanical ventilation and renal replacement therapy) are not covered by currently mandated care and will not be discussed herein.

EMPIRIC ANTIBIOTIC THERAPY

Clinicians and guidelines uniformly agree that antibiotic therapy remains the cornerstone of sepsis management. Because sepsis is a life-threatening condition deserving of timely intervention, and because many sepsis cases are caused by bacterial organisms for which effective antibiosis can limit disease severity, early initiation of empiric antibiotics is the foundation of sepsis management. As with any empiric therapy, the challenge lies in differentiating the patients who will benefit from those patients with mimics of bacterial sepsis who will not respond and may experience adverse effects.

In keeping with prior SSC guidelines, SEP-1 mandates initiation of antibiotics within 3 hours of presentation for all patients with sepsis, regardless of initial severity [11]. The most commonly cited data supporting rapid initiation of antibiotics comes from a study of >2000 intensive care unit (ICU) patients with septic shock by Kumar et al [20], in which every hour of delay in antibiotics was associated with a 7.6% absolute decrease in survival. Other studies have supported this result, although noting smaller effects [21, 22]; improved mortality rates with rapid antibiotic administration for septic shock may potentially even extend to the <1-hour time cutoff [23]. However, this effect was not universally seen in all studies; a prospective evaluation of delays in antibiotic delivery in 3 emergency departments found that hourly delays <6 hours were not associated with increased mortality rates in septic shock (though antibiosis before shock onset was associated with improved mortality rates) [24]. In addition, a systematic review and meta-analysis of hourly delays in antibiotics among patients with severe sepsis or septic shock [25] did not find a significant association between hourly delays and mortality rate, despite including the strong results of Kumar et al [20].

Among nonshock subgroups, estimates of risk with hourly delays in antibiotic administration are decidedly mixed [21, 22, 25]. In a large retrospective study of 35 000 patients randomly selected from a large hospital network and admitted for sepsis, hourly delays in antibiosis were found to be weakly correlated with hospital mortality rate among patients without shock, with an increase in hospital mortality rates of 0.3%–0.4% per hour (in contrast to 1.8% per hour in septic shock within the same study) [22]. Seymour et al [21] evaluated almost 50 000 patients presenting to the emergency department with sepsis, over a wide range of severity; although hourly delays in antibiotics were associated with a significant odds ratio of 1.07 for death among the 160 000 patients who required vasopressor therapy, there was no significant association between hourly antibiotic delays and death among the almost 33 000 patients who did not require vasopressors.

Critics of hourly targets for antibiotic administration argue that all data supportive of an hourly increase in risk before antibiosis are retrospective and are complicated by concerns about the difficulty of accurately and reproducibly determining “time zero” in sepsis [26]. In the randomized PHANTASi trial of prehospital antibiotics administered in the ambulance, there was no difference in 28-day mortality rates among septic patients, despite a >1.5-hour difference in time to antibiotic administration between the groups [27]. This represents the only published trial randomizing patients to earlier versus later antibiotic administration; however, the study was significantly underpowered to find a difference in the primary outcome of all-cause mortality rates, and there have been concerns regarding misallocation of patients between control and antibiotic arms.

When guideline-directed or mandated care is dictated within time constraints, a common problem is overdiagnosis and overuse of standard treatment. A prescient historical example is the former CMS core measure mandating administration of antibiotics for community-acquired pneumonia (CAP) within 4 hours of presentation. Although based predominantly on favorable retrospective data [28], later studies failed to reproduce the favorable effect of early antibiotics among a similar patient population; a large cross-sectional analysis of almost 100 000 emergency department admissions for pneumonia noted no correlation between time to first antibiotic dose and inpatient mortality rates from pneumonia [29]. Furthermore, there were clear signs of overdiagnosis, likely due to time constraints imposed by the core measure.

In a retrospective review performed after the implementation of the pneumonia CMS measure, the time to first antibiotic was associated with a reduced accuracy in the diagnosis of pneumonia [30]. In a survey of 121 emergency physicians with awareness of the CAP core measure, 55% reported prescribing antibiotics to patients they did not suspect of having pneumonia in order to comply with the measure; almost half of these recalled that they did so >3 times per month [31]. Eventually, based on concerns raised by these studies, the CAP core measure was eliminated.

Similar overuse could occur with mandating empiric antibiotic coverage for all potential sepsis cases within the first 3 hours of presentation, or even within 1 hour, as recommended most recently by the SSC. Indeed, the problem would likely be much greater in scope, because the diagnosis of sepsis is often more fraught with difficulty than pneumonia. In a prospectively collected cohort of 2579 patients admitted to the ICU with clinically suspected sepsis, 43% were later categorized as “unlikely” to have had an infection [32]. Beyond the concerns of inappropriate resource use, there are also substantial risks to individual patients from antibiotic overuse or misuse. In another trial of 681 patients with pneumonia admitted to a general medicine ward, almost two-thirds received a duration of antibiotic therapy greater than recommended by underlying diagnosis, and each excess day of antibiotics was associated with a 5% increased risk of either a patient-reported or provider-reported adverse effect (among which were included allergic reactions, acute kidney injury, and Clostridium difficile infections) [33].

It has been shown that as many as 20% of patients receiving antibiotics during a hospitalization will experience an adverse effect [34]. Even among critically ill populations, delaying initial antibiotics to allow for appropriate diagnostic workup may not be as harmful as is often thought. In a prospective before-after study of a strategy of either conservative or aggressive antibiotics, in which a conservative strategy often involved culture of the causative organism before initiation of therapy, the conservative group received more appropriate initial therapy, had a shorter duration of therapy, and had a lower mortality rate (13% vs 27%) [35].

There also remain concerns with overly-broad empiric antibiotic coverage, which, though common, is not strongly correlated with improved outcomes among patients with severe infections, especially those not in shock. The SSC guidelines recommend initial broad-spectrum empiric antibiotic coverage, guided by evidence of improved outcomes with combination therapy to cover multidrug-resistant organisms [36]; however, the mortality benefit for this approach may only be confined to sicker subsets of patients with sepsis, especially septic shock [37]. In addition, guideline-directed care may not offer benefit over the decisions of individual physicians armed with knowledge of local antimicrobial resistance patterns.

In an observational multicenter cohort study, patients admitted to ICUs with multidrug-resistant pneumonia were followed up via chart review, and the antibiotic regimen used in each case was compared with American Thoracic Society—Infectious Disease Society of America guidelines on pneumonia treatment to evaluate compliance. Among 303 patients studied, 28-day survival in the compliant group was surprisingly lower than in the noncompliant group (65% vs 79%), even when adjusted for illness severity; regimens deemed noncompliant with these guidelines were narrower in antibiotic coverage than compliant regimens [38]. In cases where the causative agent was eventually identified, there was no significant difference between the 2 groups in terms of activity of the empiric treatments, suggesting that providers were providing adequate coverage even when not compliant with treatment guidelines. Again, the history of treatment guidelines for severe infections shows that it can be difficult to identify high-risk immunocompetent patients who benefit from broader-spectrum empiric antibiotics; for example, the most recent CAP guidelines have dispensed with the previously recognized predictors of multidrug resistance in “healthcare-associated” pneumonia, owing to concerns about overuse [39].

Overall, although data are predominantly retrospective and conflicting, there seems to be strong evidence in favor of rapid, empiric broad-spectrum coverage for patients who are more ill, and especially for those in septic shock, with de-escalation based on clinical response and further diagnostic workup. Among normotensive patients with sepsis, the available literature does not strongly demonstrate a significant mortality benefit from rapid time-directed antibiotic administration, and there are inherent risks with antibiotics for patients who do not derive benefit; therefore, a mandate to provide rapid antibiosis within a specific time frame for those patients who are not in shock at presentation likely risks overuse and could lead to harm related to antibiotic adverse effects. It may be reasonable to withhold initial antibiotics in stable patients presenting with possible sepsis until further diagnostic workup reveals a clear infectious source or a sepsis mimic.

FLUID RESUSCITATION

Intravenous fluid resuscitation has always been considered one of the pillars of initial sepsis management, and guidelines have enshrined this common practice. Septic patients are often volume depleted from lack of intake and increased insensible losses; as well, intravascular volume resuscitation will quite often improve hypotension due to vasodilation and third spacing of fluids from capillary leakage, at least transiently. However, transient benefits due to improvement in blood pressure and organ perfusion with additional fluid resuscitation may be offset later by increased end-organ edema as well as potential injury to the endothelial glycocalyx, which can exacerbate microcirculatory defects and capillary leakage [40]. Furthermore, detailed assessments of septic shock patients often reveal only a minority who are truly “fluid responsive,” in that they respond to fluid resuscitation with an adequate increase in cardiac output [41, 42]; delays in vasopressor use in patients who are overly resuscitated with additional intravascular volume may cause harm.

The physiologic rationale for fluid resuscitation in septic shock is strong, barring the case of septic shock patients in volume-overloaded states. However, in normotensive septic patients, even in the presence of a high lactate level, the benefit for fluid administration rests on an assumption that the underlying organ dysfunction is due to hypoperfusion that will respond to increased cardiac preload, an assumption heavily dependent on the type of organ dysfunction and the underlying comorbid conditions and volume status of the individual patient. Therefore, the optimum strategy for initial and repeated intravenous fluid administration should be highly individualized and is often one of the main challenges confronting the bedside physician.

SEP-1 defines septic shock as hypotension (systolic blood pressure <90 mm Hg) despite adequate fluid resuscitation or an initial lactate measurement >4 mmol/L [11]. For these patients, the “adequate” fluid resuscitation mandated by SEP-1 is an intravenous crystalloid bolus of 30 mL/kg of body weight for these patients; although it is difficult to trace this dosing to any particular evidence, it seems to have emerged from data on pediatric septic shock. Weight-based dosing in adults obviously can lead to overadministration of fluids in obese patients or in patients already experiencing intravascular volume overload.

Data from prior trials in septic shock suggest that volume overload, even in the early stages of hospitalization, is common and detrimental. In the VASST trial of vasopressin, involving 778 patients with septic shock, a more positive fluid balance in the first 12 hours was associated with increased mortality rate [43]. In a retrospective study of patients treated with early goal-directed therapy, two-thirds of patients were perceived to be volume overloaded on hospital day 1, and this persisted in close to half of patients by day 3. The presence of clinically apparent fluid overload was associated with an increased rate of diuretic use, thoracentesis, and hospital mortality rate [44]. In a study of 151 patients with septic shock randomized after initial fluid resuscitation to either a restrictive strategy (median, 500 mL) or standard care (median, 1250 mL), there was no significant difference in hemodynamics, suggesting a lack of benefit to further fluid boluses, even among patients deemed to be fluid responsive.[45, 46]

Among normotensive septic patients, the benefits of routine weight-based initial fluid resuscitation have been disputed. In a trial of 3-hour sepsis care bundles, despite showing a hospital mortality benefit for full bundle completion, the individual component of time-based intravenous fluid resuscitation was not associated with improved in-hospital mortality rate [21]. A trial of protocol-mandated care for sepsis in Zambia, which compared initial weight-based fluid resuscitation with usual care, was stopped early owing to a 100% mortality rate among patients who had initially presented in hypoxemic respiratory failure [47].

In a follow-up trial, despite attempts to exclude patients with respiratory failure, patients who received protocol-mandated care were more likely to receive larger fluid boluses initially and had higher respiratory failure and in-hospital mortality rates [48]. However, the patients in these studies differed from those in most other studies of fluid administration, given high rates of human immunodeficiency virus, tuberculosis, and malnourishment not seen in resource-rich areas. In addition, it is difficult to compare the risks and benefits of weight-based fluid administration in a resource-poor setting with those of similar resuscitation in areas with high rates of critical care use, especially mechanical ventilation, as seen in the large trials of protocol-driven sepsis care in resource-rich areas (PROCESS, ARISE, and PROMISE trials) [8–10].

Nevertheless, these trials suggest that without high quality critical care, weight-based initial fluid administration can lead to harm in a subset of patients. Although initial weight-based fluid administration is not mandated by SEP-1 for normotensive patients with a lactate level <4.0 mmol/L, it is advocated by SSC guidelines for any septic patient with sepsis-induced hypoperfusion [49]; in attempting to follow expert-adjudicated “best practices,” busy practicing clinicians may not realize these distinctions between mandated and guideline-directed care and therefore could routinely deliver inappropriate fluid volumes to certain septic patients.

In interpreting the current literature on initial fluid resuscitation in sepsis, it is far from clear that protocolized fluid administration offers a significant benefit to patients compared with clinician-directed fluid resuscitation, even among patients with septic shock. On the other hand, multiple studies suggest that overly aggressive fluid resuscitation may lead to harm among certain subgroups of patients who are not “fluid tolerant.” An individualized, physiologic approach to fluid administration in sepsis may be the most appropriate; indeed the results of the FENICE study suggest that most intensivists in common practice resuscitate with fixed boluses of 500–1000 mL at a time, guided only in the minority of cases by markers of fluid responsiveness [50].

The ongoing CLOVERS trial (NCT0343028) may aid in deciding the critical question of how to optimize the initial approach to fluid resuscitation in septic shock. For patients not in shock, the available literature and an understanding of cardiovascular physiology do not support the mandating of specific volumes of crystalloid for all patients; a better approach would be the observance of guideline-directed fluid resuscitation among most patients (especially those with septic shock) and documentation of specific contraindications to such resuscitation among individual patients.

LACTATE MEASUREMENT

Serum lactate levels have been interpreted as a marker of organ hypoperfusion in sepsis since the initial guidelines on sepsis definitions were released [51], and lactate has remained strongly linked with organ ischemia throughout the subsequent iterations of the SSC guidelines. The CMS SEP-1 mandate requires an initial lactate measurement and serial measurements if the initial value is >2.0 mmol/L [11]. However, while the serum lactate level seems to be correlated with sepsis severity and disease mortality rate [52], and its “clearance” (often defined by a decrease of ≥10%) can be an important prognostic indicator [53], it is not clear that lactate levels in the absence of other supportive clinical data can provide meaningful therapeutic guidance.

The concerns with overinterpretation of lactate as a therapeutic biomarker of adequacy of resuscitation relate to the complex pathophysiology of hyperlactatemia in critical illness. The assumption that the presence of hyperlactatemia equals organ ischemia, especially among normotensive patients, can be fraught with peril. In fact, many lactate elevations are independent of tissue oxygenation and are caused by catecholaminergic stimulation of skeletal muscle [54]. For example, it is commonly observed that lactate levels tend to rise with epinephrine administration and inhaled β-agonists in normotensive and even in hypertensive patients. Furthermore, even among patients with septic shock, elevated lactate levels are more often associated with decreased oxygen extraction by tissues than by decreased oxygen transport, when simultaneous central venous oxygenation is measured [55].

Unfortunately, the complex pathophysiology of hyperlactatemia in sepsis is often oversimplified by sepsis guidelines. Physicians attempting to follow mandated sepsis care will therefore often respond to an initial elevated lactate level with more aggressive resuscitation, which can include broadening of empiric antibiotic therapy and weight-based fluid administration even in normotensive patients; in fact, SEP-1 mandates such resuscitation in normotensive patients with an initial lactate level >4 mmol/L.

Repeated measurements of a nonclearing lactate level often engender a resuscitation strategy of repeated fluid boluses, in the absence of other more robust clinical data supporting ischemic organ dysfunction. This was in evidence in the ANDROMEDA-SHOCK trial comparing lactate-guided versus capillary refill time–guided therapy, in which the lactate-guided group received significantly more fluid, without any significant improvement in mortality rate (in fact, with a trend toward lower mortality rate as well as significantly less organ dysfunction in the capillary refill time–guided group) [56].

Even when lactate monitoring has shown benefit in sepsis, the direct physiologic mechanisms remain unclear. For example, in a trial of 348 ICU patients randomized to repeated monitoring of lactate levels within the first 8 hours after ICU admission versus no serial lactate monitoring, the lactate group had shorter ICU stays and a slightly lower risk-adjusted in-hospital mortality rate [57]. However, despite the lactate group’s receipt of more fluid resuscitation and, controversially, vasodilators to treat microperfusion defects, lactate levels did not decrease faster in the lactate monitoring group than in the control group. This result argues that there may be a beneficial effect to serial patient monitoring, of which lactate measurements may be part, even in the absence of a physiologic rationale behind lactate-based resuscitation.

In all, though there are conclusive data that hyperlactatemia in sepsis is a poor prognostic indicator and there may be a mortality benefit to lactate monitoring, there is an absence of strong data supporting lactate measurements as a guide to the adequacy of resuscitation. A monitoring strategy is unfortunately only as good as the therapeutic decisions that it engenders; as discussed above, our therapeutic approach to sepsis beyond timely antibiotics and supportive care remains limited. With this caveat, it seems reasonable to allow providers to use lactate measurement for close monitoring of septic patients and for prognostic purposes, without mandating its use, in order to avoid impulsive treatment decisions that may harm individual patients.

CONCLUSIONS

Over the past 20 years, guideline-directed care for sepsis and septic shock has, without question, improved the overall quality of care for these patients. However, the wide spectrum of sepsis severity and significant variability in patient comorbid conditions and health status argue against a mandated approach to each and every septic patient; international guidelines appropriately stress these individual differences and allow physicians to tailor diagnostic and therapeutic strategies to the individual patient. Unlike guidelines, mandates such as SEP-1 provide no such flexibility to the bedside provider, thereby opening the door to avoidable harms among patient subgroups. Risks to individual patients can be compounded by mandated bundled care, wherein the decision to forego a single component leads to a failure to complete the entire bundle; this is concerning when each component has significant limitations and can worsen outcome if inappropriately applied to the wrong patient subgroup.

We favor a focus on mandating care for those patients who are most likely to benefit and least likely to be harmed by early broad-spectrum antibiotics, weight-based fluid resuscitation, and serial lactate monitoring, such as patients presenting in septic shock. If bundled care is mandated, physicians and hospitals should not be penalized if there is clear documentation regarding contraindications to specific components of the bundle, especially fluid administration. If sepsis mandates do not allow for variation in response to individual patient characteristics, the unintended consequences of mandated care may become one of the greatest obstacles toward progress in surviving sepsis.

Note

Supplement sponsorship. This supplement is sponsored by bioMérieux, the Gordon and Betty Moore Foundation and Beckman Coulter.

Potential conflicts of interest. Both authors: No reported conflicts. Both authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.