Sepsis Guidelines - A Review

Sepsis is a dysregulated host response to infection that leads to life-threatening organ dysfunction. The pathophysiology is complex, involving a delicate balance between proinflammatory and anti-inflammatory responses that can drive both immune over-activation and immune suppression.

Key Mechanisms

• Dysregulated Immune Response: Sepsis is no longer just “blood poisoning” but rather a host-driven pathology. The immune system, recognizing pathogens via pattern-recognition receptors (e.g., Toll-like receptors), launches an inflammatory response that releases cytokines like TNF-α, IL-1, and IL-6. This response aims to clear infection but can lead to collateral tissue damage.
• Endothelial Dysfunction & Microvascular Thrombosis: The inflammatory cascade disrupts endothelial integrity, leading to capillary leak, hypotension, and tissue hypoxia. Additionally, sepsis triggers coagulation activation (via tissue factor and thrombin) while impairing anticoagulant pathways, often leading to disseminated intravascular coagulation (DIC).
• Mitochondrial Dysfunction & Organ Failure: Tissue hypoxia is compounded by impaired oxygen utilization at the cellular level due to mitochondrial dysfunction. Inflammatory mediators further induce cell death (apoptosis and necrosis), worsening organ dysfunction.
• Immunosuppression in Late Sepsis: While early sepsis is characterized by hyperinflammation, prolonged sepsis can lead to immune paralysis, with reduced HLA-DR expression on monocytes, increased apoptosis of lymphocytes, and a higher risk of secondary infections.

Source: Angus DC, van der Poll T. Severe Sepsis and Septic Shock. N Engl J Med. 2013;369(9):840-851. doi:10.1056/NEJMra1208623.

Sepsis remains a significant public health concern in the United States, with substantial morbidity and mortality. According to the most recent data, sepsis contributes to at least 1.7 million adult hospitalizations and 350,000 deaths annually in the U.S

• In Medicare patients sepsis hospitalizations surged from 981,027 (2012) to 1.7 million (2019) with associated mortalities below (Buchman et al, 2021)
• Septic shock: 59.9%
• Severe sepsis: 35.5%
• Unspecified sepsis: 26.5%
• Sepsis-related mortality remains disproportionately higher among Black, Native American, and Hispanic populations compared to White and Asian patients (Sathananthan et al, 2021)

The economic burden of sepsis in the United States is substantial.

• Total Sepsis Hospital Costs (2019): $57.47 billion across all payers (Buchman et al, 2021)
• Medicare Burden (Buchman et al, 2020)
• Sepsis hospital costs rose from $17.79 billion (2012) to $22.44 billion (2018).
• Skilled nursing facility (SNF) costs post-sepsis increased from $3.93 billion to $5.62 billion over the same period
• Cost Per Case (Reynolds et al, 2018)
• Sepsis without organ dysfunction: $16,324
• Severe sepsis: $24,638
• Septic shock: $38,298
• Readmission Costs: Sepsis is a top cause of 30-day readmissions, reportedly costing California hospitals $500 million annually (Chang et al, 2015)

Sepsis definitions have evolved through three major iterations:

• Sepsis-1 (1991): Introduced the classic definition that may look familiar from medical school.
• Sepsis-2 (2001): Added lactate and other lab markers to better capture organ dysfunction.
• Sepsis-3 (2016): Removed SIRS and severe sepsis, replacing them with SOFA scoring.

Sepsis-1 (1991)

Sepsis-2 (2001)

Sepsis-3 (2016)

The biggest issue? Sepsis definitions don’t actually diagnose sepsis well — See table below. Additionally, a retrospective study of 4,153 ED patients at a single academic center (Litell et al, 2021) found that 60-75% of patients meeting Sepsis-3 criteria were NOT diagnosed with sepsis at discharge.

What is AUC?

The area under the receiving operating curve (AUC) is a key metric for evaluating diagnostic test performance. It’s derived from the receiving operating curve (ROC), which plots sensitivity (true positives) vs. 1-specificity (false positives). AUC is the total area under the ROC which quantifies how well a test distinguishes between diseased and non-diseased patients (e.g. septic vs. non-septic).

AUC Value	Interpretation
< 0.5	Useless test (worse than chance)
0.5	Coin flip (equal to chance)
0.6-0.7	Poor
0.7-0.8	Fair
0.8-0.9	Good
0.9-1.0	Excellent

Example: If AUC = 0.85, the model correctly ranks a septic patient as higher risk than a non-septic patient 85% of the time

If sepsis definitions struggle to diagnose sepsis, what actually is the gold standard?

• Blood cultures? Too insensitive.
• Expert consensus? Subjective.
• Hospital discharge diagnosis? Potentially influenced by external pressures.

Given the limitations in diagnostic tests, heterogeneity, constantly evolving definitions, and reliance on clinical judgment, there is no gold standard for sepsis. The literature looking at diagnostic accuracy often rely on expert physician review of charts as the final word on whether a patient had sepsis. This can introduce bias—how do we know the experts are right?

Sepsis screening tools are useful but imperfect. Among them, SIRS and qSOFA are the weakest, while NEWS and MEWS perform better.

Clinical Gestalt

A prospective study (Knack et al, 2024) of 2,484 ED encounters found that emergency attendings’ early clinical judgment within 15 minutes of patient arrival outperformed standard sepsis screening tools.

Even a machine learning model (AUC 0.84) was less accurate than physician judgment. These findings highlight the superiority of experienced clinical assessment in early sepsis identification over algorithmic tools. Notably, this study did not assess EM residents.

Artificial Intelligence

While physician judgment outperforms traditional tools, it is impractical for ED attendings to assess every patient immediately. AI-driven models show promise in early sepsis detection.

• Machine learning models outperform traditional screening tools in sepsis prediction (Kijpaisalratana et al., 2022).
• AI models, including logistic regression and random forest, predict sepsis-related mortality better than clinical evaluations (Doorn et al., 2021).
• Deep learning applied to EHR event sequences achieves high accuracy in early sepsis detection (Lauritsen et al., 2020).
• A real-time AI-based early warning system (AISE) demonstrated improved sepsis detection with high accuracy (KH Goh et al., 2021). This deep learning model, trained on continuous vital signs and EHR data, detected sepsis hours before clinical recognition, highlighting its potential role in early intervention.

However, these studies often faced limitations such as reliance on single center and retrospective data, lack of external validation, use of EHR data that may be inaccurate and incomplete and challenges in integrating AI models into clinical workflows, which may affect their generalizability and real-world applicability.

• There is no definitive gold standard for diagnosing sepsis.
• Current sepsis definitions and screening tools fail to identify all cases.
• Even the best clinical judgment (gestalt) is imperfect and will miss some cases.
• AI-based screening shows promise but is not yet ready for widespread clinical use.

Prospective, randomized controlled trial (RCT) at an urban academic ED (1997–2000)

The study enrolled 263 patients with severe sepsis or septic shock, defined by:

• Two or more SIRS criteria, AND
• SBP ≤90 mmHg after fluid challenge or lactate ≥4 mmol/L.

Patients were randomized into two groups for 6 hours of treatment in the ED

Primary Outcome:

• In-hospital mortality

Secondary Outcomes:

• Resuscitation endpoints (ScvO₂, lactate, base deficit, pH)
• APACHE II scores (organ dysfunction severity)

One key difference between EGDT and standard therapy was continuous monitoring of ScvO₂ and CVP, allowing for real-time, goal-directed interventions.

Important Note: This study did not require hypotension for inclusion—some severe sepsis patients were normotensive but had elevated lactate.

• In-hospital mortality was significantly lower in the EGDT group (30.5%) vs. Standard Care (46.5%) (P=0.009) —> NNT ~ 6
• Physiologic improvements (7–72 hrs post-randomization) in EGDT patients:
• Higher ScvO₂: 70.4% vs. 65.3% (P<0.02)
• Lower lactate: 3.0 vs. 3.9 mmol/L (P<0.02)
• Higher pH: 7.40 vs. 7.36 (P<0.02)
• Lower APACHE II scores: 13.0 vs. 15.9 (P<0.001)

• EGDT significantly reduced mortality in severe sepsis and septic shock.
• Early optimization of hemodynamics and oxygen delivery (via fluids, vasopressors, transfusion, and dobutamine) improves outcomes.
• The study established early sepsis resuscitation as a critical intervention and influenced future sepsis guidelines.

To be completed within 6 hours of severe sepsis or septic shock identification:

1. IV Antibiotics 
2. Blood cultures before antibiotics
3. Serum lactate
4. 30 cc/kg bolus for hypotension or lactate ≥4 
5. Vasopressors if hypotensive despite fluids
6. CVP ≥ 8-12 mm HgScvO2 ≥ 70% or mixed SvO2 ≥ 65% (EGDT)

EGDT was fully embedded in this first sepsis bundle. At the time, Sepsis-2 definitions classified severe sepsis as lactate ≥2, which meant a large number of patients were potentially being aggressively resuscitated. I can only imagine this era produced an entire generation of EM doctors exceptionally skilled in central line placement.

Between 2014 and 2016, three major randomized controlled trials—ProCESS, ARISE, and ProMISe—challenged the findings of the original EGDT trial. All three studies found no mortality benefit of EGDT compared to usual care.

A 2017 meta-analysis pooled data from 3723 patients at 138 hospitals across seven countries and confirmed:

• EGDT did not reduce all-cause mortality at 90 days or 1-year survival
• EGDT was associated with increased ICU days and hospital costs

With no clear benefit and increased resource utilization, EGDT was removed from the 2016 Surviving Sepsis Guidelines.

With EGDT removed, the 2016 Surviving Sepsis Guidelines introduced a revised 3-hour and 6-hour bundle, focusing on early identification and standardized interventions for sepsis and septic shock. Of note, severe sepsis had been phased out of the sepsis definitions.

Look familiar? These bundle recommendations were also adopted by CMS as the SEP-1 bundle, which became the foundation of most hospital sepsis protocols. While the Surviving Sepsis Campaign (SSC) and CMS are entirely separate entities, their guidelines have become closely linked in clinical practice. We’ll break down CMS’s SEP-1 measures later in this post—so keep reading.

With growing evidence showing increased mortality for every hour of delayed antibiotics, the Surviving Sepsis Campaign (SSC) introduced the 1-hour bundle in 2018 to emphasize rapid, protocolized intervention for sepsis and septic shock.

The implementation of the 1-hour bundle faced significant pushback from the medical community with concerns were raised about the potential for antibiotic overuse, rigid fluid resuscitation and lack of high quality evidence. In response to these concerns, the Surviving Sepsis Campaign revised its guidelines in 2021 to allow a more flexible, patient-centered approach, especially in patients without shock. The quality of evidence behind each intervention in the 2021 guidelines is outlined above. We will break down the evidence behind these recommendations in the next section.

1. Lactate alone is not diagnostic of sepsis. While lactate is commonly used in sepsis assessment, it lacks specificity. The differential diagnosis for hyperlactatemia is broad, encompassing both Type A (tissue hypoxia) and Type B (non-hypoxic) causes (see table below). With a sensitivity and specificity of <70%, lactate cannot definitively rule in or rule out sepsis.

Type A (tissue hypoxia)		Type B
Cardiogenic shock	Respiratory Failure	Albuterol	Ketoacidosis	Physical Exertion
Hemorrhage / anemia	Bowel Ischemia	Alcohol Intox	Metformin	Toxins
Trauma	Other shock	Alcohol Withdrawal	Malabsorption	Seizure
		Anti-retroviral Meds	Increased Metabolic demand	Thiamine deficiency
		Cancer	Metabolic disorders	Delayed lab processing
		Hyperglycemia

2. Lactate ≥ 4 is associated with a significant increase in sepsis mortality.
• Multiple studies have shown that lactate ≥4 mmol/L is strongly associated with increased mortality in sepsis. Casserly et al. found 44.5% mortality in patients with lactate >4 and hypotension (vs. 29% without), while Gotmaker et al. reported a 1.7x higher 90-day mortality in patients with isolated hyperlactatemia compared to those with isolated refractory hypotension.
• Thomas-Rueddel et al. and Mikkelsen et al. both identified hyperlactatemia (lactate > 4) as an independent predictor of mortality, even when controlling for shock or organ dysfunction—highlighting its prognostic value across the sepsis spectrum.
3. Trending intermediate lactates (2-4 mmol/L) may not impact patient outcomes.
• A 2021 retrospective cohort study (n=12,517) examined the necessity of repeating lactate measurements in ward patients with suspected sepsis and initial lactate 2.0–3.9 mmol/L.
• Failure to reduce lactate by >10% within 6 hours was NOT associated with increased risk of ICU transfer or in-hospital mortality.
• These findings suggest that routine lactate rechecks in intermediate cases may not be clinically necessary—a contrast to prior guidelines recommending lactate clearance as a therapeutic target.

4. Serial cap refill times may be an alternative to trending lactates in sepsis.
• The ANDROMEDA-SHOCK trial (2019) investigated capillary refill time (CRT) as a non-invasive alternative to lactate clearance in septic shock resuscitation. This multicenter RCT across 28 hospitals in South and Central America enrolled 424 ICU patients with early septic shock (<4 hrs).
• CRT Group – Cap refill assessed every 30 min until normalization (<3 sec)
• Lactate Group – Serum lactate checked every 2 hrs until normalization or ≥20% reduction

• Key Findings
• Primary Outcome: 28-day mortality was lower in the CRT group, but not statistically significant (p=0.06).
• Secondary Outcomes: CRT led to a lower 72-hour SOFA score, but there was no difference in mechanical ventilation, RRT, vasopressor use, ICU or hospital length of stay.
• These findings suggest while CRT did not significantly reduce mortality, it may be a viable, non-invasive alternative to lactate for guiding sepsis resuscitation.

1. Do not delay antibiotic administration to draw blood cultures if this leads to significant delay. This is supported by the 2021 SSC guidelines, which uses 45 minutes as an example.

2. Blood cultures have low yield but may identify a subset of high risk patients.
• Blood cultures have a low overall yield in sepsis, with positivity rates of 14% in community-onset cases (Ohnuma et al. 2023) and 54% in ICU-admitted sepsis patients (Mellhammar et al. 2021).
• However, culture-positive sepsis is consistently associated with higher mortality. A study by Mellhammar et al. found that 90-day mortality was 47% in bacteremic sepsis patients vs. 36% in non-bacteremic patients, even after adjusting for confounders. Similarly, a PHANTASi trial sub-analysis reported significantly higher 28-day and 90-day mortality in culture-positive sepsis patients, with relative risks of 1.43 and 1.41, respectively. Additionally, Verway et al. demonstrated that bloodstream infections (BSIs) were linked to a 17% 30-day mortality rate, with an adjusted odds ratio of 1.47 for death in culture-positive cases.
3. The routine use of blood cultures in all septic patients is not always justified due to high false-positive rates, cost implications, and limited impact on treatment decisions.
• Low Yield in Certain Conditions – In cellulitis, simple pyelonephritis, and community-acquired pneumonia, blood cultures have a low likelihood of yielding a true positive result, often outweighed by the risk of false positives (Long et al)
• High False Positive Rates – Blood culture contamination is a common issue. One study found that 30.6% of positive cultures were contaminants, with overall contamination rates ranging from 4.2% to 9.6% in routine practice (Aiesh et al, Vebroom et al)
• Increased Costs and Length of Stay – False positives lead to unnecessary antibiotic use, additional testing, and prolonged hospital stays, driving up healthcare costs and exposing patients to unnecessary treatments. (Geisler et al)
• Limited Impact on Clinical Decision-Making – In some settings, such as critically ill surgical patients, studies suggest that blood cultures rarely alter management and may not be cost-effective. (Henke et al)

3. ATS/IDSA do not recommend blood cultures in non-severe CAP creating a point of conflict with current sepsis guidelines.
• The guidelines recommend against routine blood cultures in non-severe CAP due to the low diagnostic yield, high false-positive rates, unnecessary antibiotic use, and increased length of stay.
• In non-severe CAP, blood cultures have a low yield (2-9%) and rarely alter empiric therapy. Exceptions include severe CAP and patients at risk for MRSA or Pseudomonas.
• Notably, not all septic CAP cases meet criteria for severe CAP.

1. Observational studies suggest increased mortality with each hour of antibiotic delay in sepsis.
• Large retrospective studies (Liu et al., 2017; 35,000 patients) and (Seymour et al., 2017; 49,000 patients) found each hour of delay was associated with higher mortality.
• A systematic review/meta-analysis (Johnston et al., 2017) across 11 observational studies reported a 33% reduction in mortality when antibiotics were given within 1 hour of sepsis identification.
2. Adjusting for confounders shows that mortality association is primarily in septic shock but not sepsis without shock.
• Pak et al. (2023) retrospectively analyzed 104,248 patients across 5 hospitals over 7 years, adjusting for the following factors:
• Severity stratification ensured that comparisons were made within similar illness groups (sepsis vs. septic shock), preventing sicker patients (who get earlier antibiotics) from biasing results.
• Adjusting for covariates (e.g., comorbidities, organ dysfunction, source of infection) helped isolate the true effect of antibiotic timing on mortality rather than confounding factors driving outcomes.
• Capping time-to-antibiotics at 6 hours removed extreme delays, which in previous studies may have exaggerated the impact of timing by including outliers who had worse outcomes for other reasons.
• They found an hourly mortality association in septic shock, but not in sepsis without shock for antibiotic timing ≤6 hours. However, mortality increased in sepsis without shock when antibiotics were delayed ≥6 hours.

3. Subsequent studies show no benefit between immediate vs. early antibiotics.
• Puskarich MA et al. (2011) – Randomized controlled trial showed no increased mortality with each hour delay of antibiotics ≤6 hours from triage in septic shock.
• Ryoo et al. (2015) – Prospective study found no increase in 28-day mortality with hourly delays ≤5 hours in septic shock.
• Rothrock SG et al. (2020) – Systematic review found no difference in mortality between immediate (0–1 hr) and early (1–3 hrs) antibiotic administration.
• Seok H et al. (2020) – Prospective study showed no impact on 7-, 14-, or 28-day mortality when antibiotics were given within 6 hours.
• Alam N et al. (2018) – Multicenter randomized trial found that prehospital antibiotics did not reduce 28-day mortality in sepsis patients.

1. The 30cc/kg initial fluid bolus recommendation originates from observational studies and was popularized by the EGDT trial—yet no prospective studies have compared different initial resuscitation volumes in sepsis or septic shock.

Wait... wasn’t EGDT disproven? Yes. The ProCESS, ARISE, and ProMISe trials ultimately showed no mortality benefit with EGDT. But interestingly, the "usual care" arms in those studies all received 30cc/kg fluids, meaning this bolus was already standard practice—even before EGDT was debunked.

2. Studies comparing restrictive vs. liberal IV fluid strategies only looked at fluid administration once the patient was admitted to the ICU.
• CLASSIC Trial (2022) - This international randomized trial found that restricting intravenous fluids in ICU patients with septic shock did not reduce 90-day mortality compared to standard fluid therapy.
• CLOVER Trial (2023) - In this multicenter randomized trial, early restrictive fluid management did not significantly affect 90-day mortality before discharge home compared to a liberal fluid strategy in patients with sepsis-induced hypotension.

1. The CENSER Trial (2019) showed that early vasopressors during initial fluid resuscitation may improve hemodynamics faster without affecting mortality.
• The CENSER Trial (2019) was a Phase II, single-center, double-blinded RCT in Thailand that evaluated the use of early norepinephrine (0.05 µg/kg/min) versus standard care in adult sepsis patients with MAP <65 mmHg.
• If MAP remained <65 mmHg after a 30cc/kg fluid bolus, open-label vasopressors were initiated in both groups.
• The primary outcome was shock control within 6 hours, defined as achieving MAP ≥65 mmHg with urine output ≥0.5mL/kg/hr or a ≥10% reduction in lactate. Early norepinephrine significantly improved shock control, with 76.1% of patients meeting this outcome compared to 48.4% in the standard care group.
• Among secondary outcomes, early norepinephrine use was associated with a lower incidence of pulmonary edema and new-onset arrhythmias. However, there was no significant difference in 28-day mortality, total IV fluid administration, or ICU admissions between the groups.
2. Subsequent studies had similar results.
• Elbouhy et al (2019): Small RCT comparing early (<25 min) vs. late (<120 min) norepinephrine found early use led to faster MAP goal achievement, better lactate clearance, and improved in-hospital survival.
• Xu et al (2022): Propensity score analysis of 2862 patients showed norepinephrine within 3 hours was linked to lower 28-day mortality, shorter ICU stays, and reduced hospital length of stay.
• Ahn et al (2024): Meta-analysis of 7181 patients (4 RCTs, 8 observational studies) found no overall mortality difference, but early norepinephrine reduced pulmonary edema and improved mortality when fluid restriction was not used.

The table below presents a side-by-side comparison of the Surviving Sepsis Campaign recommendations and the current evidence discussed in this section. Notably, most recommendations are based on studies showing benefit to septic shock patients, with limited evidence supporting their application in sepsis without shock.

• The Centers for Medicare & Medicaid Services (CMS) is a federal agency within the U.S. Department of Health and Human Services that administers the nation's major healthcare programs, including Medicare, Medicaid, and the Children's Health Insurance Program (CHIP). 
• Severe Sepsis and Septic Shock Management Bundle (SEP-1) was introduced by CMS in 2015 as a quality measure requiring hospitals to report compliance with a sepsis treatment bundle. This bundle includes several interventions such as timely administration of broad-spectrum antibiotics and intravenous fluids, serial lactate measurements, and reassessment of volume status and tissue perfusion. The goal of SEP-1 is to standardize and improve the early management of sepsis and septic shock, which are critical conditions often first identified and treated in the ED

1. CMS uses a modified version of Sepsis-2 definition of sepsis which includes severe sepsis. As per CMS, SEP-1 Bundle only applies to severe sepsis or septic shock.

Sepsis	Severe Sepsis	Septic Shock
2 SIRS+ AND Confirmed infection OR Sepsis or infection on differential diagnosis	Sepsis AND One of the following AMS Hypotension (SBP <90, MAP <65, SBP ↓40) New NIV or Mechanical Ventilation	Severe sepsis + hypotension despite fluids OR Lactate ≥ 4
OR Documented suspicion for sepsis, severe sepsis, or septic shock	OR one of the following Lactate > 2, Cr > 2, Total bili > 2 Platelets <100k, INR > 1.5, aPTT > 60

2. SEP-1 Bundle is identical to the Surviving Sepsis 3 and 6 hour bundle. Time Zero is when SEP-1 criteria for severe sepsis or septic shock is met.

3. Determining time zero in SEP-1 compliance is a major source of confusion—even for the reviewers.

A study found that abstractors agreed on time zero in only 36% of cases, highlighting the deep inconsistency and subjectivity in SEP-1 compliance reviews. That’s because CMS defines time zero not as when sepsis is suspected, but rather:

"The earliest time at which all three clinical criteria for severe sepsis are simultaneously present in the medical record — even if sepsis is not yet recognized by the treating team." (CMS Quality Net)

In other words, time zero can be assigned retrospectively — before any diagnostic test confirms infection or before the clinician suspects sepsis. According to CMS, time zero is triggered when all three of the following are present:

SIRS criteria (≥2)

A clinical or lab marker of organ dysfunction (e.g., lactate ≥2, hypotension, new oxygen requirement)

Confirmed or suspected infection

Even if infection isn’t suspected initially, CMS allows the infection component to be applied in hindsight — once a diagnosis like pneumonia is confirmed later, time zero can be backdated. There’s a lot of confusion about alternative definitions floating around, but here’s the key takeaway:

The clinician’s documentation of "suspected sepsis" does not set time zero.

Retrospective abstraction does.

Example:

11:00: Patient with asthma presents with cough and wheezing. Vitals: HR 110, RR 24. Afebrile. Clinically appears to be an asthma exacerbation.

12:00: Lactate returns at 2.8 (”organ dysfunction”).

15:00: CXR returns showing right lower lobe infiltrate. Diagnosis revised to pneumonia → patient now meets criteria for severe sepsis (SIRS + elevated lactate + confirmed infection).

15:15: Antibiotics are started.

According to CMS, once pneumonia is diagnosed, they can retroactively assign time zero to 12:00, when SIRS and elevated lactate were already present. This would trigger a SEP-1 failure because antibiotics were not given within 3 hours of that backdated time zero — even though the treating clinician reasonably didn’t suspect infection until hours later. This rigid, retrospective definition is a major reason why SEP-1 is so problematic — it penalizes appropriate clinical judgment and fuels inconsistency across reviewers.

A recent systematic review (Ford et Al) assessed the impact of SEP-1 compliance and implementation on sepsis mortality by analyzing 17 studies.

1. Is SEP-1 compliance associated with lower mortality?
• Of the 12 studies evaluating SEP-1 compliance, 5 showed a mortality benefit but had significant limitations:
• Methodological issues: One study lacked adjustment for confounders, another found benefit only in septic shock patients, and one showed improvement only in Medicare beneficiaries.
• Strongest study: Townsend et al. (2022) found 5.7% lower mortality in Medicare patients but no benefit for septic shock.
• 7 studies found no mortality benefit, with no consistent pattern of improvement across different settings or patient groups.

2. Is SEP-1 implementation at the system level associated with lower mortality?
• Five before–after studies assessed mortality trends before and after SEP-1 implementation.
• Only one study found a mortality benefit, but it did not account for pre-existing mortality trends before implementation.
• Four studies, including high-quality interrupted time-series (ITS) analyses, found no mortality benefit.
3. Confounding factors was a common theme among the studies analyzed.

Confounder	Description
Baseline Severity of Illness Bias	Sicker patients (e.g., those in septic shock or with multiple comorbidities) were less likely to receive full SEP-1 bundle compliance, making it appear as though compliance improved outcomes when, in reality, healthier patients were just more likely to complete the bundle.
Hospital-Level Differences	Hospitals with higher SEP-1 compliance may have had better overall sepsis care (e.g., stronger ICU staffing, faster recognition, better resource availability), making it difficult to isolate whether SEP-1 itself reduced mortality or if these hospitals were just performing better overall.
Timing of Interventions vs. Clinical Deterioration	Patients who improved rapidly (e.g., responded quickly to initial resuscitation) may have met SEP-1 criteria naturally, whereas deteriorating patients may have been diverted to ICU or required individualized care, leading to selection bias.
Changes in Sepsis Recognition and Documentation	Some hospitals implemented sepsis screening alerts and documentation changes alongside SEP-1, which may have artificially improved sepsis mortality statistics rather than reflecting an actual improvement in patient outcomes.
Lack of Adjustment for Pre-Existing Trends	Before-and-after studies evaluating SEP-1 did not always account for pre-existing improvements in sepsis care (e.g., declining mortality trends before SEP-1), which could falsely attribute improvements to SEP-1 rather than to broader advancements in sepsis management.

4. Key Takeaway: No Moderate or High-Level Evidence Support SEP-1 Compliance or Implementation

This systematic review found no strong evidence that adherence to or implementation of SEP-1 reduces sepsis mortality. While some studies suggested a potential benefit, findings were inconsistent and prone to bias. All 17 studies were observational, none had a low risk of bias, and confounding issues were common. High methodological variability across studies prevented a meta-analysis from being performed.

💡

You’re probably thinking: Wait a second. SEP-1 is built on outdated bundles and terminology… and it might not even improve patient outcomes? So why should I follow it at all? Let’s move onto the next section.

1. SEP-1 compliance has been “pay for reporting” to CMS since 2015.

Since 2015, hospitals have been required to report their SEP-1 compliance data to CMS, but reimbursement has not been directly tied to performance. Instead, hospitals receive financial incentives simply for reporting, regardless of the quality of care provided. Hospitals that participate and submit data are eligible for their full annual payment update, which accounts for about 2% of their total CMS reimbursements.

2. Effective 2026 Fiscal Year, SEP-1 compliance will move to the Hospital Value-Based Purchasing (VBP) Program.

Under VBP, CMS withholds 2% of total payments annually, and hospitals must earn back that money through good performance on quality measures, including SEP-1. High-performing hospitals can recover their withheld funds and even earn additional bonuses, while low-performing hospitals risk financial losses.

3. 2% may not seem like much, but for hospitals heavily reliant on CMS payments, the financial impact is significant.

Take, for example, a large urban safety-net public health system with projected Medicaid and Medicare revenue of nearly $6 billion. Losing 2% of $6 billion equates to $120 million—a substantial financial hit.



For hospitals operating on thin margins, this level of financial risk makes SEP-1 compliance a priority—whether or not it improves patient outcomes.

1. More Resources, More Compliance

A study by Barbash et al. found that SEP-1 reporting was more common in larger, nonprofit hospitals, which typically have greater resources. Hospitals that performed well on SEP-1 measures were smaller, for-profit, non-teaching institutions with intermediate-sized ICUs. Safety-net hospitals, which serve vulnerable populations and operate with fewer resources, consistently performed worse on SEP-1 measures compared to non-safety-net hospitals (Barbash & Kahn). This disparity was more pronounced in hospitals unaffiliated with larger health systems.

This should not be surprising—hospitals with resources can afford to hire teams of chart reviewers and coders to ensure SEP-1 compliance. These hospitals can retroactively adjust documentation and refine workflows to meet compliance targets, while under-resourced hospitals struggle to keep up.

2. Problems with a Penalty System

The penalty-based system of SEP-1 compliance disproportionately harms lower-resource hospitals, worsening sepsis mortality for their patient populations and creating a vicious cycle.

• Hospitals with fewer resources struggle to meet SEP-1 compliance.
• Poor SEP-1 compliance leads to reduced CMS reimbursements
• Lower CMS funding limits access to staffing, training, and quality improvement initiatives
• Limited resources contribute to higher sepsis mortality and further decompensation of chronic conditions (e.g., diabetes, heart disease, lung disease, etc.) which also contribute to higher sepsis mortality
• Exacerbation of chronic illness and worsening sepsis mortality put strain on the hospital which already is under-resourced which leads to poorer SEP-1 scores, restarting the cycle