An official website of the United States government
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Jessie King, MD, PhD, Lead, Carol E Chenoweth, MD, Peter C England, MD, Andrew Heiler, MBA, RN, Michael T Kenes, PharmD, FCCM, BCPS, BCCCP, Krishnan Raghavendran, MD, Winnie Wood, MSN, RN, CNS, and Shiwei Zhou, MD. Clinical Guidelines Oversight: Megan Mack, MD, Dave Wesorick, MD, and April Proudlock, BBA, RN.
Author Information and AffiliationsPatient Population: Sepsis is medical emergency, associated with high morbidity and mortality, and early recognition and standardized treatment of sepsis saves lives1. This guideline is intended for adult patients.
We recommend following the Surviving Sepsis Campaign’s International Guidelines for the Management of Sepsis and Septic Shock (2021)2, on which this guideline is based. We also endorse the SEP-1 treatment bundle (Figure 1) as standard of care for the initial management of a patient with sepsis.
Definitions: Sepsis is a syndrome characterized by life-threatening organ dysfunction caused by a dysregulated host response to infection.1 SEP3 clinically defined sepsis as acute organ dysfunction (e.g., ≥2 new SOFA points), plus evidence of infection2. However for billing, documentation, and treatment purposes, the older definitions of sepsis are still used.
Sepsis: two Systemic Inflammatory Response syndrome (SIRS) criteria plus evidence of infection.
Severe sepsis: Sepsis with evidence of organ dysfunction.
Septic shock: Sepsis with persistent hypotension, despite fluid resuscitation.
| Suspected site | Symptoms/signs | Initial microbiologic evaluation | Relevant Imaging Studies | Empiric Antibiotic Recommendations* |
|---|---|---|---|---|
| Upper respiratory tract | Pharyngeal inflammation plus exudate ± swelling and lymphadenopathy, sinus congestion/drainage | Throat swab for aerobic culture | None, unless concern for abscess or sinusitis, then CT | |
| Lower respiratory tract | Productive cough, pleuritic chest pain, consolidative auscultatory findings | Sputum of good quality, rapid influenza testing, urinary antigen testing (eg, pneumococcus, legionella; not recommended in children), quantitative culture of protected brush or bronchoalveolar lavage | Chest Xray, Chest CT if atypical infection is suspected | See MM guidelines |
| Urinary tract | Frequency, urgency, dysuria, loin or back pain | Urine culture and microscopy showing pyuria | None, unless concern for urinary obstruction (Renal U/S) or perinephric abscess (CT) | See MM guidelines |
| Vascular catheters: arterial, central venous | Redness or drainage at insertion site | Culture of blood (from the catheter and a peripheral site), culture catheter tip (if removed) | None | See MM Guidelines |
| Indwelling pleural catheter | Redness or drainage at insertion site | Culture of pleural fluid (through catheter) | None | See MM Guidelines |
| Peritoneal dialysis (PD) catheter | Cloudy PD fluid, abdominal pain | Cell count and culture of PD fluid | None | See MM Guidelines |
| Skin or soft tissue | Inflammation, edema, erythema, discharge of pus, lymphangitis | Gram stain and culture of draining pus, wound culture not reliable | None, unless concern for abscess, then CT | See MM Guidelines |
| Central nervous system | Signs of meningeal irritation | CSF cell count, protein, glucose, Gram stain, and culture | None, unless concern for abscess, then MRI brain/spine | See MM Guidelines |
| Gastro-intestinal and/or intra-abdominal | Abdominal pain, distension, diarrhea, and vomiting | GI: Stool culture for Salmonella, Shigella, or Campylobacter; detection of Clostridium difficile toxin; Intra-abdominal: cultures (aerobic and anaerobic) of percutaneously or surgically drained fluid collections | CT Abd/Pelvis | See MM Guidelines |
| Genital tract | Women: Low abdominal pain, vaginal discharge | Women: Endocervical and high vaginal swabs onto selective media | None, unless concern for abscess, then CT | See MM Guidelines |
| Men: Dysuria, frequency, urgency, urge incontinence, cloudy urine, prostatic tenderness | Men: Urine Gram stain and culture | None, unless concern for abscess, then CT | See MM guidelines | |
| Bone | Pain, warmth, swelling, decreased use | Blood cultures, bone cultures at surgery or by interventional radiology | Xray (subacute/chronic) or MRI (acute) | See MM Guidelines |
| Joint | Pain, warmth, swelling, decreased range of motion | Arthrocentesis with cell counts, Gram stain, and culture | None, unless concern for abscess then CT | See MM Guidelines |
This is not an exhaustive list of sources of sepsis. Rather, it intends to remind the user of common sources, and to demonstrate a general approach to the sepsis patient.
For patients with nosocomial infection, evidence of hypoperfusion or shock, or immunosuppression, an anti-pseudomonal antibiotic +/- MRSA coverage (e.g., cefepime/zosyn +/- vancomycin) should be started immediately. Stable sepsis patients should undergo a rapid evaluation to determine the likely source of infection to guide antibiotic selection (See Figure 2). Hemodynamically unstable patients, thought to be decompensating from unknown or resistant gram-negative bacterial infections, likely warrant double coverage of gram-negative organisms to ensure susceptibility of at least one antibiotic to presumed organism (i.e., addition of an aminoglycoside).
| Agent | Concentration | Starting Rate and Maximum Rate | Receptor | Preload | Heart Rate | Stroke Volume | Systemic Vascular Resistance |
|---|---|---|---|---|---|---|---|
| Phenylephrine (Neo-Synephrine©) | 200 mcg/ml (50 mg/250 ml) | 50 to 300 mcg/min | α1 | ↑ | ↓(↔) | ↔ | ↑ |
| Norepinephrine (Levophed©) | 64 mcg/ml (16 mg/250 ml) | 0.1 to 1 mcg/kg/min | α1 > β1 | ↔(↑) | ↔(↑) | ↔(↑) | ↑ |
| Epinephrine | 20 mcg/ml (5 mg/250 ml) | 0.01 to 1 mcg/kg/min | Low dose: β1 > β2 | ↔ | ↑ | ↑ | ↔ |
| High dose: α1 =/> β1 | ↑(↔) | ↑ | ↑ | ↑ | |||
| Dopamine | 3.2 mg/ml (800 mg/250 ml) | 2-5 mcg/kg/min | DA > β1 | ↔ | ↔ | ↔ | ↔ |
| 5-10 mcg/kg/min | β1 > β2 | ↔(↓) | ↑ | ↑ | ↔(↓) | ||
| 10-20 mcg/kg/min | α1 > β1 | ↔(↑) | ↑ | ↑ | ↑ | ||
| Vasopressin (Pitressin©) | 1 unit/ml (50 units/50 ml) | 0.03 units/min | v1 | ↔ | ↔(↓) | ↔(↓) | ↑ |
| Dobutamine (Dobutrex©) | 4 mg/ml (1000 mg/250 ml | 1 to 20 mcg/kg/min | β1 > β2 | ↓ | ↑ | ↑ | ↓ |
| Milrinone (Primacor©) | 0.2 mg/ml (20 mg/100 ml) | 0.125 to 0.75 mcg/kg/min | PDE III Inhibitor | ↓ | ↑ | ↑ | ↓ |
Overgaard CB. Circulation 2008;118:1047-56.
| Variable | Human Plasma | Colloid | Crystalloids | ||
|---|---|---|---|---|---|
| 5% Albumin | 0.9% NaCl | Lactated Ringer’s | PlasmaLyte | ||
| Sodium | 135-145 | ~145 | 154 | 130 | 140 |
| Potassium | 4.5-5.0 | 4.0 | 5.0 | ||
| Calcium | 2.2-2.6 | 2.7 | |||
| Magnesium | 0.8-1.0 | 3.0 | |||
| Chloride | 94-111 | ~145 | 154 | 109 | 98 |
| Acetate | 0.02-0.2 | 27 | |||
| Gluconate | 23 | ||||
| Lactate | 1-2 | 28 | |||
| Osmolarity | 275-295 | 310 | 308 | 280 | 294 |
Myburgh JA, Mythen MG. N Engl J Med 2016; 369:1243-51.
Strength of Recommendation Classification
I = Generally should be performed; II = May be reasonable to perform; III = Generally should not be performed
Level of Evidence Classification
A = systematic reviews of randomized controlled trials with or without meta-analysis; B = randomized controlled trials; C = systematic review of non-randomized controlled trials or observational studies, non-randomized controlled trials, group observation studies (cohort, cross-sectional, case-control); D = individual observation studies (case study/case series); E = expert opinion regarding benefits and harm
Severe sepsis and septic shock are medical emergencies and delays in treatment lead to increased morbidity and mortality.3–5 Sepsis improvement programs highlight awareness of the disease and lead to improved recognition through education. The Electronic Health Record (EHR) can be utilized to help with screening and alerting providers about potential sepsis cases. However, these systems have traditionally performed sub-optimally6,7 and thus provider education and recognition is crucial to improved sepsis recognition and care.
Sepsis care should be standardized across the hospital since many studies have demonstrated improved clinical outcomes with bundled sepsis treatment.8–10 A large retrospective study evaluated compliance with a 2013 statewide mandate for bundled sepsis care in New York.5 This analysis showed that bundled, protocolized sepsis care was associated with improved clinical outcomes. A similar but larger study involving 1,012,410 patients among 506 hospitals located in five states demonstrated improved sepsis outcomes, including mortality, with protocolized sepsis care.11
Quick sequential organ failure assessment, or qSOFA, is a set of three variables developed in 2016 to predict mortality in patients with known or suspected sepsis.2 The three variables include a Glasgow Coma Score < 15, a respiratory rate ≥ 22 breaths/minute, and a systolic blood pressure ≤ 100 mmHg. If a patient has two or more variables present at a given time, then the patient is qSOFA positive. Studies have shown that qSOFA is a better predictor of mortality in patients with suspected or known sepsis when compared to other screening tools such as SIRS, NEWS or MEWS. However, the goal of a screening tool is high sensitivity, so that clinicians can identify patients with the disease process of interest, not prediction of mortality. 12
Here at Michigan Medicine, we are actively screening our patients for sepsis on presentation to the ED and on the inpatient floors. When patients screen positive with a concern for sepsis or who meet criteria for severe sepsis, a BPA and page are pushed to the First Contact. We recommend that the First Contact review the data in the Sepsis Navigator to understand why the BPA fired and to evaluate whether sepsis treatment should be initiated in the patient. Consideration of patient symptoms, laboratory values, and underlying risk of infection in determination of whether the patient should be treated for sepsis.
Lactate levels are important in the identification13–15, risk stratification16–18 and resuscitation19–22 of patients with sepsis, and an elevated lactate may be a sign of occult shock.16,18
Lactate has been studied as a method to identify patients with sepsis. In comparison to other lab markers, including white blood cell count, neutrophil count, and procalcitonin, lactate has been shown to be a better predictor of severe sepsis and septic shock.13–15
All patients with suspected or confirmed sepsis should have a lactate checked upon time of recognition.16 Lactate levels are important in the risk stratification5,6 and resuscitation7 of patients with sepsis, and elevated lactate may be a sign of occult shock.23 Elevated lactate levels have been associated with increased mortality for both intermediate lactate elevations (≥ 2) and high lactates (≥ 4).16,17 Thus, elevated lactate levels are concerning in patients with sepsis and should suggest to providers the need for fluid resuscitation, hemodynamic support, and additional resources such as rapid response intervention or a higher level of care.
Lactate should be rechecked to evaluate the effectiveness of resuscitation if the initial lactate is greater than or equal to 2.24 Lactate clearance can be confounded by patient comorbidities, especially cirrhosis, or use of beta-agonists (i.e., albuterol or epinephrine) and should be interpreted within a broad clinical context and not as a single marker of resuscitation.25 Other markers of resuscitation such as central venous oxygen saturation26 and capillary perfusion27 have been shown to be non-inferior to lactate reduction and may aid in evaluation of resuscitation, along with lactate reduction (see Resuscitation & Reassessment).
A blood culture “set” includes 2 bottles, 1 aerobic and 1 anaerobic. A blood culture “site” refers to the location from which the culture is drawn. When infection is suspected, at least 2 sets of blood cultures should be drawn (aerobic & anaerobic). Blood cultures should be drawn from 2 separate peripheral sites in most patients. However, if the patient has an indwelling central catheter that is suspected to be infected, cultures should be performed through each lumen of the catheter, in addition to the peripheral site.
As antibiotic administration significantly decreases the sensitivity of blood cultures, blood cultures should be obtained prior to the initiation of antibiotics, unless the patient is unstable or critically ill, in which case antimicrobials should be initiated immediately, regardless of whether blood cultures have been obtained.28 How long to wait for cultures to be drawn in the setting of severe sepsis is at this discretion of the clinician, but it seems reasonable to start antibiotics if cultures are unable to be obtained within 45 minutes.
Antimicrobial therapy should be selected to include coverage for likely organisms at the presumed site of infection. Local data (https://www.med.umich.edu/asp/) should be utilized to tailor antimicrobial therapy. This includes local antibiograms (https://labportal.med.umich.edu/portal/apps/antibiogram/) and treatment guidelines according to disease state, which are held and maintained on the Antimicrobial Stewardship website (https://www.med.umich.edu/asp/adult.html). Patients with a reported Beta-lactam allergy should be considered for referral to the Beta-lactam Evaluation Team during their hospitalization (https://www.med.umich.edu/asp/pdf/adult_guidelines/Beta-lactam-Evaluation-and-Empiric.pdf).
Antimicrobial therapy should be prioritized for septic patients with presumed infection. Antibiotics should be targeted to cover presumed organisms based upon likely organisms at the presumed source of infection and tailored based upon local prevalence of, and risk factors for, resistance among those organisms. While risk factors for resistant organisms vary depending on geographic location of included study patients, specific type of infection, and number of variables collected, general risk factors include patients with a higher severity of illness, those with previous broad-spectrum antimicrobial exposure, those with exposure to high-risk healthcare settings (i.e. prolonged hospitalization, long-term care facilities, dialysis centers, etc.), immunosuppression, and previous recent resistant organisms.29–31 Double-coverage of gram-negative organisms, while exposing patients to additional antimicrobial agents and their adverse effects, is often used to maximize the chances that a suspected or proven bacteria is susceptible to one of the two antimicrobials until microbiologic susceptibilities result. The benefit of double-coverage likely is maximized in patients with highest risk of resistant organisms or in areas with a high local prevalence of resistant organisms. In patients at a high risk of immediate morbidity and mortality from their infection (i.e., septic shock with escalating vasopressor requirement), the benefit of double coverage likely outweighs the immediate risks, in contrast to patients at low risk of morbidity and mortality from infection, where judicious waiting for culture results to tailor antibiotics will not lead to additional harm.32 De-escalation strategies should be tailored based upon a patient’s clinical status, as well as laboratory or stewardship resources available to rapidly detect or rule-out bacterial organisms.
Antimicrobial delivery should be optimized to ensure rapid achievement of therapeutic targets and concentrations at the presumed site of infection. Strategies to optimize the pharmacokinetic-pharmacodynamic parameters should be utilized when able and include extended (or continuous)-infusion beta-lactams, and extended-interval aminoglycosides.33
MRSA accounts for a small percentage of culture positive infections in critically ill patients (i.e., 5%), yet a majority of critically ill septic patients receive anti-MRSA antibiotics.34 Individual risk factors for MRSA include recent IV antibiotics, history of recurrent skin infections or chronic wounds, presence of invasive devices, hemodialysis, recent hospital admission, severity of illness, and most importantly, prior history of MRSA infection or colonization. Failure to cover for MRSA in a patient with a true MRSA infection may be harmful, but unnecessary coverage for MRSA in a patient without MRSA may also be harmful.35 MRSA nasal swab testing is often used to avoid initiation of MRSA-targeted antibiotics, as it has a high negative predictive value for not only respiratory infections (96.5%), but also other systemic infections as well.36 Prior MRSA nasal swabs within the previous 60 days have also performed well when examining the durability of a previous negative result.37 Importantly, a positive MRSA nasal swab does not indicate an active MRSA infection or necessitate the use of anti-MRSA antibiotics. Rather, a positive swab indicates that the patient is colonized with MRSA, and that this organism should be covered if a S. aureus infection is presumed.
Early administration of effective antimicrobials has been shown to reduce mortality in patients with sepsis or septic shock in observational or retrospective studies.4,5,23,38 However, the importance of early antimicrobial treatment, must be tempered against the adverse effects of indiscriminate overuse of antibiotics, such as allergic or hypersensitivity reactions, renal dysfunction, Clostridioides difficile infections and antimicrobial resistance.39
In multiple large-scale studies of sepsis, the association of early antimicrobial administration and reduced mortality are strongest for patients with septic shock, and therefore support use of early administration of antimicrobials for patients with septic shock, in whom the risk of delayed, inadequate, or avoidance of treatment is the highest..5,40,41
In the above studies, the association of timing of antimicrobials in patients with sepsis without shock and mortality, is less consistent. Therefore, in patients with sepsis without shock, rapid assessment of the patient for infectious or non-infectious source for sepsis within 3 hours is recommended. The decision whether antimicrobials should be deferred or administered may be based on findings in the rapid assessment.
When the decision to initiate antimicrobial agents is made, significant effort should be placed on ensuring rapid administration to the patient. This includes leveraging institutional strategies such as optimization of automated dispensing cabinets for medication availability closest to patient, ordering as “STAT” compared with “routine”, and prioritizing administration once medication is at the patient’s bedside.
Septic shock is classically designated as a distributive shock characterized by a profound immune response and release of vasodilatory substances resulting in inappropriate vascular dilation and tissue hypoperfusion. However, the over-reactive inflammatory response is commonly secondary to a separate disease process that frequently results in some period of poor volume intake, increased insensible loss (e.g. increased minute ventilation, hyperthermic losses secondary to fever) and potential GI losses, all contributing to intravascular depletion and hypovolemia. Lastly, the inflammatory response has been associated with myocardial dysfunction and decreased cardiac output. These separate but potentially additive causes for hypoperfusion makes septic shock difficult to protocolize. However, given the likelihood for intravascular depletion and hypovolemia it is suggested that initial volume resuscitation be attempted if evidence of tissue hypoperfusion.
The overarching goal of fluid therapy is to improve oxygen delivery to the tissue (DO2) by increasing stroke volume (SV) and cardiac output (CO). The goal of fluid resuscitation (or an endpoint for fluid therapy) is not to increase the mean arterial pressure (MAP).42 A patient is deemed “fluid responsive” when SV increases by 10-15% after a fluid challenge (i.e., 250-500 mL); such patients should receive judicious fluids until the appropriate increase in SV/CO subsides. Administration of fluids to patients deemed not fluid responsive only serves to overload the patient with unnecessary fluid which subsequently impairs organ function.
Fluid resuscitation has shown benefit in experimental studies by improving microvascular prefusion,43(p) improving cardiac output and oxygen delivery,44,45 and improved organ function.46 However, the data around impact of initial fluid administration on clinical outcomes is less robust. 30 mL/kg was utilized in the landmark paper, Early Goal Directed Therapy47 (EGDT), in which significant mortality benefit was demonstrated by protocolized sepsis care. The PROMISE,48 PROCESS49 and ARISE50 trials evaluated bundled sepsis care and did not focus on the timing and amount of fluid given and thus provide no specific insight into the benefit of early or large volume fluid resuscitation. The previously mentioned large clinical trials that were published post-EGDT showed no significant difference in “usual care” compared to protocolized EGDT, but the volumes administered in each of these trials did approach the similar 30 mL/kg given in EGDT. More recently the Classic trial51 and Clovers trial52 attempted to evaluate restrictive versus fluid resuscitation in septic patients. Unfortunately these papers do not provide additional guidance on initial fluid resuscitation as both studies do not account for fluid resuscitation received prior to randomization which was often 2-3 liters. It is thus unclear based on these large clinical trials whether fluid administration alone has benefit for patients with tissue hypoperfusion.
There is observational data53,54 that demonstrate clinical outcome benefits to bundled sepsis care and are frequently referenced in the discussion of early large volume fluid resuscitation, however none of these studies focuses specifically on early fluid administration alone.
There are current concerns in the medical community that over-administration of intravascular fluids to hospitalized patients is associated with worse clinical outcomes.55,56 However, these studies focus on large volume administration over longer durations of time and do not apply to the initial 30mL/kg fluid resuscitation. Ongoing volume administration, which is discussed below, is more likely an area with greater risk of harm to a patient. Thus, we continue to suggest the administration of 30 ml/kg of fluid resuscitation be administered in the first 3 hours of sepsis onset, especially given the potential for hypovolemia in our septic patients. Further research evaluating specific volume targets and time of administration is needed.
Given clinical concerns for over-resuscitation in our morbidly obese population, it is reasonable for the initial volume of resuscitation to be based on ideal or adjusted body weight, instead of actual weight. Literature57 currently is sparse on this subject.
The choice of fluids for resuscitation are typically separated into crystalloids and colloids. Crystalloids have the advantage of being more readily available and lower in cost when compared to colloids. Colloids have the theoretical benefit of increasing oncotic pressure intravascularly and thus preventing extravasation of the fluid into the extravascular space.
There have been several large trials that have attempted to address which resuscitative fluid, colloid (albumin) or crystalloid, is superior in critically ill patients. These studies show no clear benefit of one over the other in mortality at 28 days.58–60 Thus, given the cost difference and lack of clear benefit, we recommend crystalloids as first-line fluid for resuscitation in septic patients.
Several studies have shown benefit to balanced solutions (e.g. lactated ringers, Plasma-Lyte, or normosol) over normal saline, including reduction in major adverse kidney events within 30 days (MAKE-30),61 and reduced in-hospital mortality in patients with sepsis.62 Finally a meta-analysis63 of nearly 35,000 patients was recently published in the New England Journal that provides a signal for mortality reduction with the use of balanced crystalloids. The clinical impact or separation of effect is likely more pronounced in patients who require large volume resuscitation compared those with more conservative fluid needs. Regardless, choice of fluid therapy should be individualized for each patient. Balanced fluids likely produce less elevations in serum chloride levels, whereas 0.9% NaCl is likely advantageous for those at risk of cerebral edema or dysnatremia. Lacking robust, definitive evidence, and apart from a strong indication to choose one agent over another, we suggest administration of balanced crystalloids for patients with sepsis over 0.9% NaCl. Albumin may have a role in sepsis management, but not as a first line resuscitative fluid as discussed above. In patients who have received large volumes of crystalloids, the addition of albumin may have clinical benefit. The ALBIOS59 study, in which patients received approximately 3000 to 4000 mL of fluid over the seven day study period, demonstrated that patients receiving albumin, in addition to crystalloid, had a statistically higher mean arterial pressure and lower net fluid balance. A more recent retrospective analysis by Gomez et al.64 compared patients who received normal saline with 5% albumin versus normal saline alone in patients that received > 60 ml/kg of fluid resuscitation in 24 hours. This study demonstrated a lower adjusted hospital 30-day mortality in patients that received albumin. Additionally, those in the albumin arm had a lower incidence of renal injury. Thus, albumin may provide some benefit when providing large volume resuscitation. Further analysis of what constitutes large volume and timing of albumin administration is necessary for a stronger recommendation to be made.
Given the limited availability of the blood product albumin, and its associated cost, efforts have been taken to produce synthetic colloids (i.e., hydroxyethyl starches, gelatins, dextrans). Unfortunately, multiple studies have shown the harm of using synthetic colloids in a variety of clinical settings for resuscitation in comparison to crystalloids.65–69 These are resuscitation fluids which contain large molecules aimed at increasing intravascular oncotic pressure with the physiologic goal of keeping resuscitative fluid from extravasating into extravascular tissue. However, these studies all found that patients receiving synthetic colloids have higher rates of AKI and need for renal replacement therapy, and failed to find a mortality benefit in favor of colloids.
Pre-existing conditions that make patients more vulnerable to fluid overload (heart failure, cirrhosis, ESRD, etc) are generally not a contraindication to initial fluid resuscitation (30 mL/kg) and reduced volumes do not lead to lower rates of intubation, or increased ventilator-free days as demonstrated in one retrospective propensity-score matched analysis.70 There is a mortality benefit to initial volume resuscitation in these patients,71,72 however, and if lower volumes are being considered, appropriate documentation of their current volume status and contraindication to fluid is recommended.
Since the original Goal Directed Therapy article was published,47 there have been numerous studies trying to elucidate the best method by which to assess a patient’s volume status in the setting of sepsis. Many methods, including Swan-Ganz catheter,73 measurement of central venous pressure (CVP),74 and mixed venous oxygenation75 are invasive and are correlated with mortality outcomes, but have not been definitively shown to improve survival. Newer non-invasive technologies for measuring cardiac output have been developed, however these require additional training and are not widely available outside of the intensive care units.
After the initial 30 mL/kg bolus, we recommend reassessment of the clinical status of the patient. Lactate should be rechecked to evaluate the effectiveness of resuscitation if the initial lactate is greater than or equal to 2.24 If there is recurrent or refractory hypotension, we recommend performing a physical examination that includes whether the patient remains responsive to additional fluid resuscitation which may be accomplished through an additional small volume bolus (250-500 mL given rapidly, over 10 minutes), or by performing a passive leg raise (see Figure 3), and then followed by objective measures of SV/CO (non-invasive cardiac output systems, ECHO, esophageal doppler, etc.). When measurement of CO or SV is not possible, a >15% increase in pulse pressure could indicate that the patient is fluid responsive utilizing a passive leg-raise test for 60–90 seconds.
Other markers of resuscitation such as central venous oxygen saturation26 and capillary perfusion27 have been shown to be non-inferior to lactate reduction and may aid in evaluation of resuscitation. Physical exam findings such as skin mottling, and poor capillary refill indicate ongoing poor perfusion.
There is recent evidence that after initial fluid resuscitation, there is no difference in outcomes when additional fluids are given or vasopressors are initiated. 76
Norepinephrine is recommended as a first-line agent in the treatment of septic shock and we suggest against dopamine use. Both observational studies77–79 and a multicenter RCT80 suggested norepinephrine to be superior to dopamine in terms of 28 day mortality, likely driven by an increase in arrhythmias due to dopamine. This finding was confirmed in a subsequent meta-analysis in 2012.81
Studies have evaluated additional agents including vasopressin, epinephrine, and phenylephrine in comparison to norepinephrine as a first-line vasopressor in septic shock. There was no significant difference between vasopressin and norepinephrine82,83 in terms of 28-day mortality, and when vasopressin is used in conjunction with norepinephrine there may be a survival benefit over high dose norepinephrine monotherapy.82 Additionally, in meta-analysis,84 vasopressin was found to cause less tachyarrhythmias than norepinephrine. A RCT85 comparing norepinephrine and dobutamine to epinephrine alone in the treatment of septic shock found no difference in 28-day mortality between groups or in the incidence of side effects. These results were subsequently confirmed in a large prospective observational study.86
We suggest starting peripheral vasopressors to achieve MAP goals instead of delaying initiation until central access may be achieved. However, central access should be obtained if ongoing vasopressor support is needed. Traditionally, vasopressor administration mandated immediate placement of a central venous catheter in either the internal jugular, femoral or subclavian veins (to avoid tissue injury, should extravasation occur). However, this frequently led to delays in vasopressor administration which placed patients at risk for ongoing hypotension and continued organ injury. A number of case reports and case series65–67,87–89 had documented the risk of complications from peripherally administered vasopressors, however the majority of these studies were very old. More recently, an observational study68, two systematic reviews69,90 and one meta-analysis91 have demonstrated low incidence of serious adverse injury secondary to vasopressor administration through a peripheral IV. The incidence of extravasation seems to be greater when IVs are placed distal to the antecubital fossa. Thus, we suggest starting peripheral vasopressors to achieve MAP goals instead of delaying initiation until central access may be achieved. However, central access should be obtained if ongoing vasopressor support is needed.
For adults with septic shock requiring norepinephrine for at least 4 hours at doses of ≥ 0.25 μg/kg/min, we suggest using stress dose steroids (i.e. IV hydrocortisone 50mg q6hr plus/minus oral fludrocortisone 50 mcg q24hr).
All of the major trials on stress dose steroids in septic shock have consistently shown faster resolution of shock with the administration of steroids, with several demonstrating a mortality benefit. Sepsis is defined by a dysregulated host response to infection involving unregulated inflammation, organ dysfunction and hemodynamic instability. In addition, there is evidence that the hypothalamic-pituitary-adrenal (HPA) axis becomes dysregulated in sepsis.92 Steroids have thus been an attractive treatment option to counteract the widespread inflammatory cascade and re-store the dysregulated HPA access.
In the early 2000’s two large RCTs90,91 demonstrated a faster resolution of shock (i.e. shorter duration of vasopressor therapy). One study90 which enrolled sicker patients (63% 28-day mortality in placebo group) demonstrated a mortality benefit from steroid therapy.
Two more recent trials92,93 have since been published which have reinforced the administering of stress dose steroids in patients with septic shock. Both of these trials demonstrated a shorter time to shock reversal with steroids, and again a reduction in mortality for the trial93 that enrolled sicker patients.
For adults with sepsis with hypotension, non-invasive blood pressure cuff measurements can be inaccurate compared to invasive monitoring11–15, particularly in patients with elevated BMI. Arterial catheters for arterial blood pressure measurement allow continuous, real-time measurements so that therapeutic decisions can be made with accurate and immediate information. While there are associated risks to invasive arterial monitoring16, these are small and using ultrasound guidance may increase placement success rates and decrease complication rates.17,18
Septic shock is characterized by inappropriate arterial vasodilation secondary to an exaggerated inflammatory response in the patient. This inappropriate arterial vasodilation causes a reduction in the MAP and can lead to decreased organ perfusion if MAP falls below the autoregulatory zone of each organ. Decreased organ perfusion leads to ischemia and organ failure if not corrected. Fluid administration aims to improve cardiac output and maintain MAP but for many patients with septic shock volume administration will not be sufficient and vasopressor support will be required to raise MAP and organ perfusion pressure.
The goal MAP recommended by surviving sepsis guidelines24 is equal to or greater than 65. This MAP goal was first based on small clinical studies94,95 in which biomarkers (lactate, creatinine), cardiac index and urine output were observed following changes in NE infusions in patients with septic shock. These studies showed no significant benefit in lactate reduction, creatinine or urine output when MAP was raised from 65 to 85 with vasopressors.
In 2014 the SEPSISPAM96 study was published which compared 28-day mortality between MAP goals of 65 to 70 mmHg against those with a goal of 80-85 mmHg. The study found no significant difference in 28-day mortality and patients with the higher MAP target experienced longer duration on vasopressors and an increased rate of atrial fibrillation. Interestingly, upon subgroup analysis, patients with chronic hypertension demonstrated less adverse renal outcomes when randomized to a higher MAP target. Two large retrospective studies97,98 published following SEPSISPAM demonstrated significant increases in hospital mortality and the incidence of AKI when MAP was < 65 in patients with septic shock.
Additionally an individual patient-data meta-analysis99 performed on previous studies, of which SEPSISPAM was the majority of the patient data included, showed a non-significant trend for increased mortality in patients > 65 when higher MAP targets were chosen. This led to the more recent 65 trial100 which was a RCT comparing permissive hypotension (MAP 60-65 mmHg) to usual MAP goals in patients older than 65 years of age requiring vasopressor support in the ICU for vasodilatory shock. The study found no significant difference between groups in terms of 90-day mortality, AKI or vasopressor duration. The mean difference in blood pressure between groups was approximately six, thus it is not surprising that no significant difference was found.
Thus, given the current available data, an initial MAP goal of ≥ 65 mmHg for patients in septic shock seems the best initial goal for all patients with septic shock. The SEPSISPAM and 65 trial remind us that individualized care is important and that adjusting the MAP goal based on individual patient data may be reasonable.
Definition: Source control is defined as the identification and possible removal of the anatomical area responsible for sepsis and septic shock. Excluding areas that are not responsible for the initiation of sepsis is additionally an important element of source control. Source control is identified as a key element of the surviving sepsis guidelines established and now widely implemented across the globe.24
Clinical Decision making in source control: Adequate source control is associated with increased survival.101,102 There are a multitude of factors that determine the clinical approach to performance of a procedure in patients with sepsis.103 The timing of the procedure is often dependent on the presence of hemodynamic stability. In patients with profound septic shock and vasoplegia, source control procedures should be undertaken as soon as the patient is resuscitated. While there is limited data to suggest the appropriate timing of the performance of procedures for source control, it is our recommendation based on a multitude of smaller studies that it be performed within 6-12 hours from admission.104,105 It is also clear from clinical observations that some patients may not achieve hemodynamic stability despite maximal resuscitative efforts without adequate source control.106
Treatment alternatives: Source control in a critically ill patient could include many different treatment paradigms.
The selection of the type of procedure depends on multiple factors. A risk-benefit analysis dictates the timing and the location of performance of the procedure. In general, the least invasive procedure is the first line of therapy. Additional factors include clinical expertise, availability of interventions and other logistic institutional factors, patient preference and the nature of clinical status.107 The nature of intra-operative procedures depends on hemodynamic and respiratory stability, presence of severe metabolic acidosis, and hypothermia with diffuse coagulopathy. In some patients, anatomic resection of the infected area is not possible. A diversion or drainage of the infected bed is the suggested alternative. Additionally, a surgical procedure such as a laparotomy may be warranted to exclude specific causes of septic shock where further radiologic studies may not be possible. Most procedures are either performed in the operating room or in interventional radiology suites. However, in patients with extreme hemodynamic instability and respiratory distress despite maximal ventilatory support, performance of a bedside procedure in the ICU, is recommended.
Concept of damage control: Evidence of damage control or performance of an abbreviated operation was first initiated in trauma patients with severe non-compressible hemorrhage in the chest or abdomen. Over the last several decades mortality benefits have been observed and now extended to diverse types of traumatic insults.108,109 These principles have been adapted for patients with septic shock where source control has been undertaken. The basic principles of the initial procedure involve arresting areas of hemorrhage or contamination with luminal bowel contents in the operating room. Such maneuvers in emergency general surgery include resection or stapling of small and large bowel. The next steps involve resuscitation in the ICU where facets of coagulopathy, lactic acidosis and hypothermia are addressed through resuscitative measures and goal directed therapy. The final steps include return to the operating room where definitive surgical procedures along with closure of the abdomen is achieved, once the patient is appropriately resuscitated. The definitive procedure may be performed at any time but typically between 48-72 hours after resuscitation in the ICU. There are a number of observational and retrospective case studies evaluating damage control laparotomy versus conventional single stage surgery. A recent meta-analysis of 5 non-randomized and 16 observational studies suggested no difference in crude mortality rates between a damage control versus conventional single stage surgical intervention.110 However, the standardized mortality ratio displayed clear benefit for damage control surgery. In presence of severe septic shock, metabolic acidosis and coagulopathy, we strongly suggest taking a damage control approach to surgery and resuscitation.
Adequacy of source control: There is little evidence in literature that suggest how to conclude whether source control has been successful or further interventions are needed. For patients where the operative source control has been definitively performed (examples sigmoid colectomy and appendectomy) it is expected that patients will respond with clinical improvement within 48 hours of the intervention. If the patient does not respond to what was thought to be a definitive procedure, further imaging or surgery may be warranted. However, for clinical scenarios where definitive source control is not surgically attainable (for example, uncontrolled enteric fistula with ongoing leak into the abdomen that has been subjected to multiple laparotomies), a combination of non-surgical and long term antibiotics may be necessary. Removal of potentially infected lines and tubes should be performed after initial resuscitation as part of an overall source control plan19. For intravascular lines, these should be removed when alternative intravascular access has been established. There may be circumstances in which it is not practical to remove an intravascular line, and the patient should be treated through the line for an extended period of time. We recommend Infectious Disease consultation in this circumstance.
Patients who recover from septic shock often have a positive fluid balance.111,112 After initial resuscitation and stabilization, de-escalation of fluid management and optimization of volume status is required. The goals of de-escalation include restoration of organ function by attaining negative fluid balanced (i.e. progressing toward net even for the hospitalization) through goal-directed fluid removal. This is achieved through use of diuretics or renal replacement therapy and the limitation of fluid intake, including medication carriers and diluents.
While data continues to emerge, it is clear that prolonged positive fluid balance is associated with worsened organ function and clinical outcomes, and that de-resuscitation (i.e. achieving a negative fluid balance after stabilization) is associated with improved clinical outcomes.113,114 Fluid balance therefore is a practice-dependent and modifiable risk factor for clinical outcomes, including mortality. Patients should be assessed frequently and early after initial stabilization utilizing physiologic data (i.e. pre-admission volume status, dynamic and static hemodynamic parameters, assessment of organ function, etc.) when available to guide aggressive volume removal.115
While early antimicrobial treatment is important for improving mortality in patients with septic shock, limiting the continued overuse of broad spectrum antimicrobials is also important to decrease potential for development of antimicrobial resistant microorganisms.116 Once a microbial pathogen and susceptibilities had been identified, or a non-infectious cause for acute illness is identified, stopping unnecessary antimicrobials and/or narrowing the spectrum of antimicrobials to target only the identified organisms is recommended. Michigan Medicine Antimicrobial Stewardship guidelines provide appropriate target based treatment guidelines: University of Michigan Health System (umich.edu) If infection has been ruled out, discontinuing all antimicrobials is recommended. De-escalation is generally safe, may decrease cost associated with unnecessary antibiotic use, and can decrease selection pressure for antimicrobial resistance, C.difficile infection and antibiotic adverse events. A recent meta-analysis reviewed 13 studies and found improved short-term mortality and decreased hospital length of stay in patients who had antimicrobial de-escalation.117
An anterior nares screen for methicillin-resistant Staphylococcus aureus (MRSA) can be a useful tool for discontinuation of anti-MRSA antibiotics, such as vancomycin.36 A negative MRSA nasal screen may be used to guide the safe discontinuation of anti-MRSA antimicrobials.
Resolution of sepsis physiology may lag behind control of the infection, and procalcitonin can aid in the de-escalation and discontinuation of antimicrobials in a safe and timely manner. Measurement of serum procalcitonin may assist with de-escalation of antimicrobials during treatment of sepsis. A meta-analysis of 14 RCTs assessing the use of procalcitonin compared to controls to guide antimicrobial therapy duration in patients with sepsis suggested improved mortality in those patients managed with use of procalcitonin. In these studies there was no difference in lCU or hospital length of stay. Antibiotic exposure was consistently lower in patients managed using procalcitonin, with minimal adverse effects.24 Based on benefit, with minimal undesirable effects, procalcitonin is suggested for use, along with clinical evaluation, to guide duration of antimicrobial treatment in patients with sepsis or septic shock and adequate source control.
Post-intensive care syndrome (PICS), or post-hospital syndrome, is a collection of problems that patients can experience after surviving a life-threatening illness, like sepsis or septic shock.118,119 The three domains include alterations in physical function, mental health, and cognitive impairment. More than half of all survivors of critical illness and sepsis will suffer from at least one of these problems.120,121
These problems can be identified through normal patient (or caregiver)-healthcare provider interactions, asking specific questions, or formal testing in each of the three domains. Patients with pre-existing health problems in a specific domain are more likely to develop problems in that area post-hospitalization.122 Patients with more severe illness are also more likely to develop PICS.
Treatment involves identification of specific symptoms, providing therapy, as able, and referral to specialized providers (i.e. physical therapy, psychiatric services, occupational therapy, etc.).123 Even with identification and treatment, PICS symptoms are often present at six to twelve months post-hospitalization. For example, in one randomized controlled study, cognitive impairment at 12 months post discharge was seen in 34% and 24% of patients, comparable to moderate traumatic brain injury and mild Alzheimer’s disease, respectively.124
PICS not only affects patients, but also their caregivers. It is important to counsel caregivers and suggest referral to their own healthcare providers for support.125 Structure for follow up for sepsis patients after discharge is evolving, and we recommend discussing follow up options with your Care Management team.
Certain groups of patients are more vulnerable to sepsis and more likely to die from sepsis. These include patients with history of heart failure, cirrhosis, or neutropenia. Patients with heart failure and cirrhosis are less likely to get standard-of-care treatment, especially with regards to fluid resuscitation in the setting of hypotension or elevated lactate. Patients with neutropenia are vulnerable to opportunistic infections as well as community acquired ones, and are profoundly immunosuppressed, limited in their ability to fight these infections at all. Early, and broad spectrum, antibiotics are critical to their survival. Specific considerations for these patient populations are addressed below.
Sepsis accounts for almost 25% of the deaths in patients with chronic heart failure20,21 and patients with CHF have higher mortality rates than patients without CHF who are admitted for sepsis.22,23
While chronic heart failure is not an exception to the SEP1 Bundle mandate by CMS, septic heart failure patients are less likely to receive adequate fluid resuscitation than non-heart failure patients72 The 3 hour bundle has been shown to reduce mortality in heart failure patients24. A large prospective observational study demonstrated that the subgroup of patients with heart failure had reduced mortality when compliant with the 3 hour bundle24, and this included the 30 mL/kg fluid bolus, when applicable.
Studies around treatment of sepsis in patients with decompensated heart failure are lacking.
Immunocompromised states are associated with lower sepsis survival, with neutropenia conveying a particularly high risk of mortality in critically ill sepsis patients.25 Neutropenia is most frequently associated with malignancy and/or chemotherapy and these patients frequently have additional risk factors such as recent surgeries, central venous catheters, and pre-treatment with antibiotics. National guidelines around management of sepsis in neutropenic cancer patients were published in 2018.26 These guidelines suggest we follow standard of care (SEP1 bundle) for the initial treatment of sepsis in the neutropenic population. At Michigan Medicine, hematological malignancy patients have increased risk of antibiotic resistance and therefore antibiotic choice should be adjusted per the Neutropenic Fever Guidelines, with anti-pseudomonal antibiotics recommended. There may be a role for antifungal treatment in the setting of prolonged neutropenia for patients not on fungal prophylaxis, given their increased susceptibility to invasive fungal infections.
Source control in the setting of neutropenia or pancytopenia should be no different than in non-neutropenic patients, though it is recommended that the least invasive procedure is used. Surgical interventions, however, should not be delayed due to pancytopenia, but rather blood products should be transfused to the surgeon’s specifications to mitigate associated risk.26 There is no evidence that granulocyte transfusion is beneficial.
Routine use of G-CSF or GM-CSF is not recommended in patients with neutropenic sepsis, as rapid neutropenia recovery carries a risk of respiratory deterioration due to ARDS.27,28
Patients with cirrhosis and septic shock experience higher mortality than those without cirrhosis.29 This may be due, in part, to altered hemodynamics at baseline in cirrhosis patients, as it is known they have increased renal and neurological organ dysfunction in the setting of sepsis. Unfortunately, there have not been many studies looking at the best practices in management of sepsis in this vulnerable population. Cirrhosis does not exclude patients from standard treatment bundles (SEP1), however there has been debate amongst physicians as to the best fluid for resuscitation in this unique population. A recent single-center randomized controlled trial of albumin infusion (250 mL of 5% solution) versus 30 mL/kg of normal saline demonstrated improved blood pressure (MAP > 65) and lactate clearance at 3 hours as well as improved mortality at 6 days with albumin resuscitation as compared to normal saline.30 However, the average MELD-Na score in both groups was over 33, and 30-day and 60-day mortality rates were not reported. Although albumin can be used for volume expansion in patients with cirrhosis, the evidence supporting this approach is limited. Therefore, in cirrhosis patients with sepsis, we recommend initial fluid resuscitation with 30 mL/kg of crystalloids, as recommended for patients without cirrhosis. It may be reasonable to add albumin in some circumstances as well.
Within the Medline (Ovid) database, the following search strategy was used.
The Main search retrieved 13,772 references. When the search hedges for Guidelines, Clinical Trials, and Cohort Studies were added, the base results are as follow:
Within the Cochrane Database of Systematic Reviews, 55 reviews were found using the strategy in the search strategies document.
The search was conducted in components each keyed to a specific causal link in a formal problem structure (available upon request). The search was supplemented with very recent clinical trials known to expert members of the panel. Negative trials were specifically sought. The search was a single cycle.
Level of evidence supporting a diagnostic method or an intervention:
Drafts of this guideline were reviewed in clinical conferences and by distribution for comment within departments and divisions of the University of Michigan Health System to which the content is most relevant: Emergency Medicine, Family Medicine, General Medicine, Infectious Disease, Gastroenterology, and Radiology. Medication recommendations were reviewed by the Pharmacy and Therapeutics Committee. The Executive Committee for Clinical Affairs of the University of Michigan Hospitals and Health Centers endorsed the final version.
These guidelines should not be construed as including all proper methods of care or excluding other acceptable methods of care reasonably directed to obtaining the same results. The ultimate judgment regarding any specific clinical procedure or treatment must be made by the physician in light of the circumstances presented by the patient.
These links to Internal UMHS Guidelines contain proprietary information so are only accessible to appropriate Michigan Medicine staff. For more information, contact the authors or publisher.
Supplementary material can be found at http://www.uofmhealth.org/provider/clinical-care-guidelines
Internal UMHS Guidelines contain proprietary information so are only accessible to appropriate Michigan Medicine staff. For more information, contact the authors or publisher.
Created: November 2023.
Except where otherwise noted, this work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. To view a copy of this license, visit https://creativecommons.org/licenses/by-nc-sa/4.0/
Your browsing activity is empty.
Activity recording is turned off.
See more...
