Fever

Sepsis

search

Sepsis, Septic Shock, Septicemia, Bacteremia in Adults, Septicemia in Adults, Surviving Sepsis Campaign, Sepsis Bundle, Early Goal Directed Therapy

  • See Also
  1. Sepsis in Children
  2. Bacteremia in Children
  3. Systemic Inflammatory Response Syndrome (SIRS)
  4. Sequential Organ Failure Assessment Score (SOFA Score)
  5. qSOFA Score
  6. Critical Care
  • Definitions
  1. Sepsis
    1. Sepsis 3 Definition (2016)
      1. Life threatening organ dysfunction resulting from dysregulated host response to infection
      2. Since organ dysfunction is part of the Sepsis 3 definition, Sepsis is now equivalent to severe Sepsis
    2. Older definition (Sepsis 1/2)
      1. Systemic Inflammatory Response Syndrome (SIRS) AND an infectious source
      2. Sepsis with organ dysfunction was considered severe Sepsis
  2. Severe Sepsis
    1. Sepsis AND organ dysfunction, hypoperfusion or Hypotension
    2. No longer distinct from Sepsis definition (2016 Sepsis 3 guidelines)
  3. Septic Shock
    1. Sepsis AND Hypotension refractory to fluid Resuscitation (typically 30 ml/kg Ideal Body Weight)
    2. Sepsis with circulatory, cellular and metabolic dysfunction that is associated with a higher risk of mortality
  4. Organ Dysfunction
    1. Organ function unable to sustain homeostasis
  5. Hypoperfusion
    1. Lactic Acidosis, Oliguria or acutely Altered Mental Status
  • Epidemiology
  • United States
  1. Incidence: 1.7 Million per year (2024)
  2. Prevalence: 3-5 cases per 1000 (>500 cases per 100,000)
  3. In-Hospital Mortality: 350,000/year
  4. Worldwide mortality rate 20% (rates vary widely 10 to 36%, some estimates as high as 52%)
  • Pathophysiology
  1. Inflammatory response (and Antagonizing anti-inflammatory response)
  2. Endothelial damage
  3. Increased vascular permeability
  4. Coagulation Pathway activation
  5. Impaired tissue oxygenation
  6. Decreased Vasopressin, Thyroxine, Cortisol and Growth Hormone
  • Risk Factors
  1. Age over 65 years (60 to 85% of all Sepsis cases)
    1. Sepsis risk >13 fold increased risk
    2. Sepsis mortality risk >2 fold increased risk
  2. Malnutrition
  3. Chronic comorbidity (Hazard Ratio >=2)
    1. Diabetes Mellitus
    2. Chronic lung disease (e.g. COPD)
    3. Peripheral Arterial Disease
    4. Chronic Kidney Disease
    5. Wang (2012) PLoS One 7(10):e48307 +PMID: 23118977 [PubMed]
  4. Immunosuppression
  5. Recent surgery
    1. Postoperative Infection is responsible for one third of Sepsis cases
  6. Recent hospitalization
  7. Indwelling catheter or other device
  8. History of prior Sepsis
  • Signs
  • Sepsis
  1. Constitutional changes
    1. Body Temperature abnormality
      1. Fever
        1. Most common presenting symptom
        2. However, absence of fever does not exclude Sepsis
      2. Hypothermia (<36 C): Poor prognostic sign
    2. Diaphoresis
    3. Rigors
    4. Myalgias
    5. Malaise
  2. Cardiovascular changes
    1. Occurs with myocardial depression and intravascular fluid redistribution
    2. Hypotension (seen on presentation in 40% of Sepsis cases)
      1. Systolic Blood Pressure <90 mmHg or
      2. Mean arterial pressure <65 mmHg or
      3. Systolic Blood Pressure drop >40 mmgHg from baseline
      4. Hypotension at presentation confers a 2 fold increased mortality risk
    3. Cold and clammy skin
    4. Mottling of skin
    5. Decreased Capillary Refill 3 seconds or greater
    6. Tachycardia
    7. Decreased Urine Output (<0.5 ml/kg)
  3. Respiratory Findings
    1. Hypoxia, Dyspnea or Tachypnea
    2. Pharyngitis, Dysphagia or Stridor
    3. Cough, Pleuritic Chest Pain
    4. Abnormal lung auscultation (consolidation)
  4. Gastrointestinal findings
    1. Abdominal Pain, distention and rigidity (peritoneal signs)
    2. Decreased Bowel Sounds
    3. Upper Gastrointestinal Bleeding (significant blood loss in Sepsis is rare)
    4. Vomiting
    5. Diarrhea
  5. Genitourinary Findings
    1. Dysuria, frequency, Hematuria, pyuria
    2. Costovertebral Angle Tenderness
    3. Pelvic Pain
    4. Vaginal Discharge or Vaginal Bleeding
  6. Neurologic changes
    1. Mental status changes
    2. Agitation or generalized weakness may be only presenting neurologic change in elderly
    3. Headache
  7. Dermatologic changes
    1. Direct Bacterial Infection (responsible for Sepsis)
      1. Abscess
      2. Cellulitis
      3. Necrotizing Fasciitis
    2. Vasculitis, microembolic, or Disseminated Intravascular Coagulation (DIC) induced lesions
      1. Ecchymosis
      2. Bullae
      3. Petechiae or Purpura
        1. Consider Neisseria Meningitidis
        2. Consider Rocky Mountain Spotted Fever
  • Diagnosis
  • Criteria (2015 Sepsis-3 definitions)
  1. Precautions
    1. Although SOFA has been preferred for Sepsis definition, it misses more cases of Sepsis than SIRS
      1. For Sepsis, Test Sensitivity is 82% for SIRS, while SOFA Score is 66% and qSOFA is 46%
      2. qSOFA is no longer recommended as a Sepsis screening tool due to low Test Sensitivity
    2. SIRS has poor Test Specificity for Sepsis
      1. For Sepsis, Test Specificity is 24% for SIRS, while SOFA Score is 75% and qSOFA is 82%
      2. SIRS has a particularly high False Positive Rate
    3. SOFA Score appears inadequate alone (in combination with infection) to diagnose Sepsis
      1. SOFA Score has been validated only to identify life threatening organ dysfunction
      2. SOFA Score is useful in prognosis
        1. SOFA Score worsening of >=2 at least doubles mortality, and increases overall mortality 10%
    4. References
      1. Wang (2022) PLoS One 17(4):e0266755 +PMID: 35427367 [PubMed]
  2. Sepsis
    1. Infection AND
    2. Organ Dysfunction
      1. qSOFA Score or SOFA Score 2 or more is consistent with Sepsis
      2. SOFA Score (or qSOFA Score) are recommended to replace SIRS Criteria in Sepsis-3 Guidelines
        1. Previously: Systemic Inflammatory Response Syndrome (SIRS criteria) positive (2 of 4 present)
        2. However, SOFA Score has a much lower Test Sensitivity than SIRS for Sepsis
  3. Septic Shock
    1. Infection and SOFA Score or qSOFA Score >=2 (or SIRS) AND
    2. Full fluid Resuscitation trial (e.g. 40-60 ml/kg or 2 Liters in an adult patient or wedge pressure 12-20 mmHg) AND
    3. Lactic Acid >2 mmol/L AND
    4. Hypotension despite what should be adequate fluid Resuscitation
      1. Mean arterial pressure <65 mmHg (<80 mmHg if prior Hypertension) or
      2. Vasopressor (Dopamine, Norepinephrine, Epinephrine) required for pressure support
  • Diagnosis
  • Criteria for other definitions (not included in 2015 Sepsis definitions)
  1. Severe Sepsis
    1. Infection and SOFA Score or qSOFA Score >=2 (or SIRS) AND
    2. Markers of poor organ perfusion (tissue hypoperfusion precedes Hypotension)
      1. Increased serum lactate (>4 mmol/L cut-off)
      2. Capillary Refill 3 seconds or greater
      3. Mottled skin
      4. Urine Output <0.5 ml/kg for 1 hour or more (or renal replacement therapy)
      5. Abrupt onset of Altered Level of Consciousness or abnormal EEG
      6. Disseminated Intravascular Coagulation
      7. Acute Lung Injury or ARDS
      8. Platelet Count <100,000/ml
      9. Cardiac dysfunction (based on Echocardiogram)
  2. Refractory Septic Shock
    1. Septic Shock AND
    2. High Vasopressor dose required to maintain Mean arterial pressure <60 mmHg (<80 mmHg if prior Hypertension)
      1. Dopamine >15 mcg/kg/min OR
      2. Norepinephrine or Epinephrine >0.25 mcg/min
  3. Multiple Organ Dysfunction Syndrome (MODS)
    1. Most severe Sepsis with progressive organ dysfunction
    2. Parameters demonstrating progressive organ dysfunction
      1. Serum Creatinine
      2. Serum Bilirubin
      3. PO2 to FIO2 ratio
      4. GCS Score
      5. Platelet Count
  • Differential Diagnosis
  1. Hypovolemia
  2. Acute blood loss
  3. Acute Myocardial Infarction
  4. Acute Pancreatitis
  5. Transfusion Reaction
  6. Diabetic Ketoacidosis
  7. Adrenal Insufficiency
  8. Acute Pulmonary Embolism
    1. Consider evaluation early in course
  • Evaluation
  • Predictors of positive Blood Cultures (each doubles risk)
  1. Age over 30 years
  2. Heart Rate >90 bpm
  3. Temperature >37.8 C (>100 F)
  4. White Blood Cell Count >12,000
  5. Central venous catheter
  6. Hospital stay >10 days
  • Evaluation
  • Sources
  1. Most common organisms
    1. Bacteria are responsible for most Sepsis cases
    2. Gram-positive Bacteria (25 to 50% of all Sepsis cases)
      1. Staphylococcus aureus
      2. Streptococcus species
    3. Gram Negative Bacteria (30 to 60% of all Sepsis cases)
      1. Escherichia coli (most common overall)
      2. KlebsiellaPneumoniae
      3. Pseudomonas aeruginosa
  2. Most common sources (80% of cases)
    1. Respiratory infection
      1. Pneumonia is most common cause of Sepsis (43% of all Sepsis cases)
      2. Organism less likely to be obtained on culture
    2. Genitourinary infection
      1. Urinary Tract Infection or Pyelonephritis (16% of all Sepsis cases)
      2. Consider Septic Abortion in pregnancy or Chorioamnionitis in postpartum patients
      3. Organism most likely to be identified on culture
    3. Gastrointestinal infection
    4. Skin and Soft Tissue Infection
  3. Occult sources
    1. Meningitis (1% of cases)
      1. Consider Lumbar Puncture
    2. Bacterial Endocarditis (1% of cases)
      1. Consider Echocardiogram
    3. Acute Sinusitis
      1. Consider Sinus CT
    4. Cholecystitis
      1. Consider RUQ Ultrasound
  • Labs
  1. Cultures
    1. Obtain within 45 minutes of presentation
    2. Do not delay Antibiotics if cultures cannot be obtained within 45 minutes
    3. If surgical source culture is required, percutaneous, minimally invasive approach is preferred
      1. Consult with local surgery or Interventional Radiology
    4. Sources
      1. Blood Culture
        1. Rigors, high fever, and appetite loss are most predictive of positive Blood Cultures
        2. https://www.journalofhospitalmedicine.com/jhospmed/article/127086/prediction-rule-bacteremia
        3. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3841330/
      2. Urine Culture
      3. CSF Culture (if indicated)
      4. Central Line or PICC Line (if present, draw a culture through the line prior to removal)
  2. Complete Blood Count
    1. Leukocytosis (>12,000) or Leukopenia (<4000)
      1. Neutrophil predominance is typical
      2. Neutropenia is uncommon (except for chronic Alcoholism and the elderly)
    2. Thrombocytopenia
      1. Typically precedes DIC
      2. Platelet Count drop >30% is associated with increased ICU-mortality
    3. Hemolytic Anemia (microangiopathic)
      1. Present in DIC
  3. Coagulation Studies (INR, PTT)
    1. Coagulopathy in DIC
  4. Chemistry panel
    1. Hyperglycemia (Sepsis-induced Insulin Resistance)
    2. Acute Renal Failure
  5. Liver Function Tests
  6. Thyroid Stimulating Hormone
    1. Consider in refractory cases
  7. Serum Lactate (Lactic Acid)
    1. Marker of poor organ perfusion (see definition of severe Sepsis above)
    2. Obtain on all septic patients, when obtaining Blood Cultures and for those admitted with infection
    3. Serum Lactate >4 mmol/L is associated with increased mortality
    4. Higher serum lactates should be met with more aggressive management
    5. Measurement of serial Lactic Acid levels should be performed on the same machine
      1. Most accurate methods are with Arterial Blood Gas machine (even for venous sample)
      2. Weingart and Orman in Majoewsky (2013) EM:Rap 13(10):5
    6. Recheck elevated Lactic Acid level in 2 hours after first draw, then every 4-6 hours until stabilized
  8. Urinalysis
    1. Urinary sediment (in association with Anuria or Oliguria)
      1. Acute Tubular Necrosis
  9. Procalcitonin (controversial)
    1. See Procalcitonin
    2. Emergency medicine literature highlights low efficacy and does not recommend
    3. Some hospitalist guidelines recommend Procalcitonin to guide Antibiotic management and duration
  • Imaging
  1. Chest XRay
  2. CT Abdomen and Pelvis
    1. Indicated for intraabdominal or pelvic findings or suspected GI Source
  3. Echocardiogram
    1. Indicated for suspected endocarditis with septic emboli (e.g. IVDA)
  4. Bedside Ultrasound
    1. Inferior Vena Cava Ultrasound for Volume Status
  • Management
  • Sepsis Decision Pathway
  1. Indications
    1. Modified SIRS Criteria 2 or more AND suspected infection
  2. Step 1: Emergency Department triage measures
    1. Serum Lactic Acid
    2. Mean arterial pressure
    3. Triage may also draw rainbow lab tubes and one Blood Culture bottle and set aside
      1. Triage Sepsis alert in the EMR significantly reduces time to fluids and Antibiotics
      2. Hayden (2016) Am J Emerg Med 34(1): 1-9 +PMID:26386734 [PubMed]
  3. Step 2: Mean Arterial Pressure <65 mmHg
    1. Immediate rooming
    2. Initial stabilization
      1. Initiate 30 cc/kg IV crystalloid bolus within 3 hours (with first liter within 30 minutes)
      2. Obtain 2 sets of Blood Cultures as well as specific source cultures
      3. Initiate broad spectrum Antibiotics
    3. Reassess mean arterial pressure and consider IVC Ultrasound
      1. Go to Steps 3a-d
  4. Step 3a: Mean Arterial Pressure <65 mmHg despite 2 Liters crystalloid in Step 2 (Septic Shock)
    1. Obtain Central Line access
    2. Empiric second round of fluid Resuscitation (or base on IVC Ultrasound)
      1. Age <75 AND no history of CHF: Give 4 L IV fluids within first 6 hours
      2. Age >75 OR history of CHF: Give 3 L IV fluids within first 6 hours
    3. Start Vasopressors (e.g. Norepinephrine)
      1. Target mean arterial pressure (MAP) >65 mmHg
    4. Disposition to Intensive Care Unit
  5. Step 3b: Mean Arterial Pressure >65 mmHg AND Serum Lactic Acid 4 mmol/L or more (Severe Sepsis)
    1. Obtain 2 peripheral IVs
      1. A single large gauge IV (e.g. 16 to 18g) may be sufficient
      2. Central Line access may be considered (optional)
    2. Initiate 2 Liter crystalloid bolus
      1. Empiric second round of fluid Resuscitation (or base on IVC Ultrasound)
      2. Age <75 AND no history of CHF: Give 4 L IV fluids within first 6 hours
      3. Age >75 OR history of CHF: Give 3 L IV fluids within first 6 hours
    3. Obtain 2 sets of Blood Cultures as well as specific source cultures
    4. Initiate broad spectrum Antibiotics
    5. Admit to Intensive Care unit
    6. Repeat serum Lactic Acid after fluid Resuscitation
  6. Step 3c: Mean Arterial Pressure >65 mmHg AND Serum Lactic Acid 2-4 mmol/L (Sepsis)
    1. Initiate 2 Liter crystalloid bolus
    2. Obtain specific source cultures
      1. Consider 2 sets of Blood Cultures (per hospital policy, discuss with admitting hospitalist)
    3. Initiate targeted Antibiotics for suspected source
    4. Repeat serum Lactic Acid after initial crystalloid bolus
      1. Serum lactate clearance >10% (percent drop in lactate from first value)
        1. Admit to regular medical ward
      2. Serum lactate clearance <10% (percent drop in lactate from first value)
        1. Repeat crystalloid bolus 1 Liter
        2. Admit to Intensive Care
  7. Step 3d: Mean Arterial Pressure >65 mmHg AND Serum Lactic Acid <2 mmol/L (Uncomplicated Sepsis)
    1. Antibiotics as indicated
    2. Fluid Resuscitation as indicated
  8. References
    1. Berg and Orman in Herbert (2015) EM:Rap 15(8): 8-11
  1. Appropriate Antibiotic choice and delivery
    1. Establish Emergency Department first dose protocols
      1. Based on local sensitivities and infectious disease recommendations
    2. Simultaneous Antibiotics without delay is ideal (obtain additional IV sites if needed)
    3. Vancomycin dosing should be calculated based on weight
      1. Dose: 15-20 mg/kg total body weight (up to 2 grams)
  2. Start Antibiotics as early as possible (within 1 hour of Septic Shock, within 3 hours of Sepsis without shock)
    1. Early Antibiotic delivery is most critical in severe Sepsis and Septic Shock
    2. Mortality increases with each hour of Antibiotic delay
    3. Kumar (2006) Crit Care Med 34(6): 1589-96 [PubMed]
    4. Gaieski (2010) Crit Care Med 38(4): 1045-53 [PubMed]
    5. Seymour (2017) N Engl J Med 376(23):2235-44 [PubMed]
  3. Consider source, but start broad spectrum Antibiotic
    1. Source not identified in 20-30% of cases
    2. Re-evaluate Antibiotic selection daily (esp. as cultures and sensitivity results are returned)
    3. Inappropriate Antibiotic selection is associated with a 34% increase in mortality
      1. Shani (2010) Antimicrob Agents Chemother 54(11):4851-63 [PubMed]
      2. Kumar (2009) Chest 136(5): 1237-48 [PubMed]
  4. Empiric therapy without obvious source (broad spectrum coverage)
    1. Vancomycin 15-20 mg/kg (up to 2 g) every 12 hours AND
      1. MRSA accounts for 1 in 20 causes of Sepsis in critically ill patients
      2. However, broad spectrum Antibiotic (see below) should be infused first, before the 2 hour Vancomycin infusion starts
    2. Choose one of the following broad spectrum Antibiotic
      1. Piperacillin-Tazobactam (Zosyn, if low Prevalence ESBL) OR
      2. Imipenem 1 g IV every 8 hours (or Meropenem or Doripenem) OR
      3. Cefepime 2 g IV every 8 hours with Metronidazole
    3. Consider adding Aminoglycoside (e.g. Gentamicin)
  5. Remove obvious source within 6 hours (source control)
    1. Remove infected lines
    2. Abscess Incision and Drainage
    3. Necrotic tissue Debridement (e.g. Necrotizing Fasciitis)
    4. Culture the tip of infected device
  6. Consider risk factors for multi-drug resistance
    1. Recent Antibiotics
    2. Recent wound care
    3. Hospitalization in last 3 months
  7. Consider initial Antibiotics that may be given as IV bolus
    1. Some beta-lactams, Cephalosporins and Aminoglycosides may be administered IV bolus
    2. Larger molecule Antibiotics (e.g. Vancomycin, Quinolones) may not be given IV bolus
      1. Risk of adverse reactions
    3. References
      1. Lin and Spaulding in Herbert (2016) EM:Rap 16(3): 6-7
      2. Garrelts (1992) Pharmacoeconomics 1(2): 116-23 +PMID:10172048 [PubMed]
  8. Avoid delay in subsequent Antibiotic doses
    1. Antibiotic second dose delays are common in transition from the Emergency Department to admission
    2. Leisman (2017) Crit Care Med 45(6): 956-65 +PMID:28328652 [PubMed]
  9. Antibiotic course
    1. Typical total Antibiotic course: 7 to 10 days
    2. Longer Antibiotic duration indications
      1. Bacterial Endocarditis
      2. Osteomyelitis
      3. Endovascular device infections
      4. Orthopedic hardware infections
  • Management
  • Antibiotics - Empiric by Site of Infection and Risk Factors
  1. Unknown source
    1. See empiric therapy described above
  2. Lung Infection
    1. Evaluate Pleural Fluid
      1. Drain empyema if present
    2. Community Acquired Pneumonia without risk for multidrug resistance or Pseudomonas
      1. See Pneumonia Management
      2. Empiric Antibiotic Regimen
        1. Ceftriaxone AND Azithromycin OR
        2. Ceftriaxone AND Doxcycline OR
        3. Fluoroquinolone (Levofloxacin, Moxifloxacin)
    3. Pneumonia with risk for Multi-drug resistance or Pseudomonas
      1. Pseudomonas or Multi-Drug Resistance Indications
        1. Chronic Lung Disease (e.g. Cystic Fibrosis, COPD, Bronchiectasis)
        2. Frequent Antibiotics or Corticosteroids
        3. Gram Negative Rods on Sputum Gram Stain
        4. Hospital-Acquired Pneumonia
          1. See Nosocomial Pneumonia
      2. Empiric Antibiotic Regimen
        1. Fluoroquinolone (e.g. Ciprofloxacin) AND
        2. Piperacillin-Tazobactam 4.5 g IV q6-8h OR Cefepime OR Carbapenem (e.g. Meropenem) OR Aztreonam
        3. Add Vancomycin for MRSA if cavitary Pneumonia or empyema
  3. Urinary tract source
    1. See Acute Pyelonephritis
    2. Cover organisms associated with complicated Urinary Tract Infection
      1. Enterococcus
      2. Pseudomonas aeruginosa
      3. Staphylococcus aureus
    3. Empiric Antibiotics if no risk for multidrug resistance
      1. Ceftriaxone OR
      2. Fluoroquinolone (e.g. Ciprofloxacin, Levofloxacin)
        1. Avoid Fluoroquinolones if local Escherichia coli resistance rates >10%
    4. Empiric Antibiotics if risk of multidrug resistance (or Extended-Spectrum Beta-Lactamase, indwelling Foley Catheter)
      1. Cefepime OR
      2. Piperacillin/Tazobactam (Zosyn) 4.5 g every 6-8 hours OR
      3. Carbapenem (e.g. Meropenem, especially for ESBL organisms) AND Vancomycin OR
      4. Levofloxacin AND Gentamicin 7 mg/kg every 24 hours
        1. Other Antibiotics are preferred over fluouroquinolones in UTI with Sepsis due to increasing resistance
    5. Imaging indications
      1. Evaluate for obstruction (Nephrolithiasis)
  4. Intra-abdominal or pelvic source suspected
    1. Empiric Antibiotic coverage
      1. First-line single agents (choose one)
        1. Piperacillin-Tazobactam (Zosyn) 4.5 g every 6-8 hours OR
        2. Imipenem/Cilastin (Primaxin) 250-500 mg IV every 6-8 hours OR
        3. Meropenem 1 g IV every 8 hours OR
        4. Cefepime 2 g IV every 12 hours AND Metronidazole 500 mg IV every 6-8 hours
      2. Alternatives regimens (e.g. beta lactam Anaphylaxis)
        1. Vancomycin AND Aztreonam AND Metronidazole
      3. Prior guidelines added additional Gram Negative coverage (no longer recommended)
        1. Gentamicin 7 mg/kg every 24 hours OR
        2. Ciprofloxacin 400 mg IV every 12 hours
    2. Early surgical Consultation
      1. Consider percutaneous or open drainage of infection source
  5. Meningitis Suspected
    1. See Bacterial Meningitis Management
    2. Vancomycin 15-20 mg/kg (up to 2 g) every 12 hours AND
    3. Ceftriaxone 2 g every 12 hours (or Cefotaxime)
    4. Consider Ampicillin 2 g every 4 hours (for listeria if Immunocompromised or over age 50 years)
    5. Consider adding Dexamethasone 10 mg every 6 hours (for pneumococcal Meningitis)
    6. Consider adding Acyclovir 10 mg/kg every 8 hours (if herpes Encephalitis suspected, Delirium, focal deficits)
    7. Consider adding Rifampin
    8. Alternative regimen (e.g. Beta Lactam Anaphylaxis)
      1. Vancomycin AND Moxifloxacin AND Trimethoprim-Sulfamethoxazole
  6. Skin and Soft Tissue Infection or Necrotizing Fasciitis
    1. See Cellulitis
    2. See Necrotizing Fasciitis
    3. Empiric Antibiotics
      1. Vancomycin 15-20 mg/kg (up to 2 g) every 12 hours OR Linezolid AND
      2. Piperacillin/Tazobactam (Zosyn) 4.5 g every 6-8 hours OR Carbapenem OR Cefepime with Metronidazole
      3. Add Clindamycin 900 mg every 8 hours (esp. if toxin release, e.g. Necrotizing Fasciitis, gangrene)
    4. Surgical Consultation indications
      1. Necrotizing Fasciitis suspected
      2. Deep abscess
  7. Post-splenectomy
    1. Ceftriaxone 2 g IV every 24 hours (every 12 hours if Meningitis) OR
    2. Piperacillin-Tazobactam (Zosyn) 4.5 g IV every 6-8 hours OR
    3. Carbapenem (e.g. Meropenem) OR
    4. Clindamycin OR
    5. Fluoroquinolone (e.g. Levofloxacin OR Moxifloxacin)
      1. Do not use in Dog Bite (risk of Capnocytophagia, with Fluoroquinolone resistance)
  8. Febrile Neutropenia
    1. See Febrile Neutropenia
    2. Cefepime OR
    3. Piperacillin-Tazobactam (Zosyn) 4.5 g IV q6-8 hours OR
    4. Carbapenem OR
    5. Ceftazidime (Fortaz) OR
    6. Aztreonam AND Vancomycin OR
    7. Ciprofloxacin AND Clindamycin
    8. Indications to ADD Vancomycin
      1. Septic Shock
      2. Pneumonia
      3. Catheter Related Bloodstream Infection
      4. Skin or Soft Tissue Infection
      5. Severe Mucositis
  9. Vascular Device Infection
    1. Vancomycin 15-20 mg/kg (up to 2 g) IV every 12 hours AND
    2. Piperacillin-Tazobactam (Zosyn) 4.5 g IV q6-8 hours (or Aztreonam 2 g IV q8 hours) AND
  10. Illicit Drug use
    1. Include Vancomycin in regimen (to cover MRSA)
  11. Petechial rash
    1. Include Ceftriaxone 2 g IV q12 hours
    2. Cover Neisseria Meningitidis and Rocky Mountain Spotted Fever (also consider DIC)
  12. Acute Diarrheal Syndrome
    1. Metronidazole 500 mg IV every 6 hours AND
    2. Vancomycin 250 mg orally every 6 hours (OR enema 1 g/500 ml rectally every 8 hours)
  • Monitoring
  • Serum Lactate
  1. Serial serum lactate levels can help guide Resuscitation and response to management
    1. Consider obtaining at presentation and again at 3 and 6 hours
    2. Lactate clearance can drive goal directed therapy
    3. Jones (2010) JAMA 303(8):739-46 [PubMed]
  2. Indications for more aggressive Sepsis management and monitoring with serial serum lactate levels
    1. Serum lactate >4 mmol/L
    2. Hypotension despite 2 Liters of IV fluids
    3. Hypotension responsive to IV fluids (may be a harbinger of later more severe episodes)
  • Management
  • Stabilization - General Measures
  1. ABC Management
  2. Surviving Sepsis Guidelines in 2018
    1. Criticism that prior 3 and 6 hour bundles have been compressed into 1 hour
    2. Weak evidence to support the aggressive changes and increased risk of adverse effects
    3. Swaminathan, Spiegel, Farkas in Herbert (2018) EM:Rap 18(10): 1-2
  3. Precautions: ProCESS Trial results (2014)
    1. Pragmatic aggressive care (fluids, Antibiotics, pressors)
      1. As effective as Early Goal Directed Therapy (with Svo2 monitoring)
    2. Emphasis
      1. Aggressive fluid hydration starting with 30 cc/kg bolus (4.4 Liters on average)
      2. Early Antibiotics
      3. Vasopressors (44%)
      4. Central Lines (>50% of cases)
    3. However, specialized Sepsis catheters with Svo2 monitoring are not required
      1. Did not improve outcomes beyond otherwise aggressive care
    4. References
      1. (2014) N Engl J Med 370(18): 1683-93 [PubMed]
  4. Oxygenation
    1. Maintain Oxygen Saturation >90%
    2. Maintain superior vena cava Oxygen Saturation (Scvo2) >70%
    3. Maintain mixed venous Oxygen Saturation (Svo2) >65%
    4. Advance to High Flow Oxygen as needed to maintain adequate oxygenation
  5. Ventilation (BIPAP or Mechanical Ventilation)
    1. Indicated for septic patients with Oxygen Saturation <90% or significant Tachypnea despite High Flow Oxygen
    2. Use a low threshold for intubation of the elderly patient with severe Sepsis
    3. Be alert for transient Hypotension with intubation in Sepsis (consider Push Dose Pressor)
    4. Lung Protective Ventilator Strategy (using ARDSPEEP and FIO2 Tables)
      1. Tidal Volumes 6 cc/kg of Ideal Body Weight (up to 8 cc/kg/IBW)
      2. Maintain median inspiratory plateau pressure (IPP) <30 cm H2O in ventilated patients
  6. Central venous access indications (e.g. internal jugular venous catheterization)
    1. Vasopressor delivery (Norepinephrine, Epinephrine)
    2. Unreliable Intravenous Access
    3. Monitoring (controversial - see above)
      1. Monitoring may be done instead non-invasively (e.g. follow IVC Ultrasound)
      2. Sepsis catheters (e.g. Vigileo) can monitor Central Venous Oxygen Saturation (ScvO2)
        1. ProCESS Trial showed no added benefit to otherwise aggressive Sepsis management (see above)
  1. Crystalloid selection
    1. Lactated Ringers is preferred instead of NS
      1. May also consider Plasma-Lyte after several liters of NS have been given
      2. However, several Antibiotics (Ceftriaxone and Zosyn) cannot be run in same line as LR
        1. Consider using Normal Saline until dual Intravenous Access, or initial Antibiotics completed
    2. Normal Saline (NS) increases acidosis in contrast to Lactated Ringers (LR)
      1. NS in the large quantities used for Sepsis, may have adverse renal and inflammatory effects
  2. Crystalloid volume
    1. Start with 1 liter of isotonic crystalloid (NS or LR) in first 30 minutes
      1. Improved survival when crystalloid bolus started within first 30 minutes
      2. Liesman (2016) Ann Emerg Med 68(3): 298-311 +PMID:27085369 [PubMed]
    2. Typical total initial crystalloid bolus of 30cc/kg (2-3L in 70-100 kg adult) within first 3 hours
    3. Typical total crystalloid fluid over first 24-48 hours may approach 5-7 Liters
    4. Overall Crystalloid Total: 2-4 Liters initially of NS or LR (up to 10 L have been used in some cases)
    5. Hemodynamic effects of fluid boluses lasts only one hour
  3. Goals
    1. Inferior Vena Cava Ultrasound with <50% collapse on inspiration
    2. Passive Leg Raise Maneuver (fluid responsiveness)
    3. Increased Pulse Pressure (difference between systolic Blood Pressure and diastolic Blood Pressure)
    4. Serial serum lactate decrease (see above)
    5. Improved Capillary Refill
    6. Central Venous Pressure >8 (>12 if on mechanical Ventilator)
    7. Central venous oxygen level >70 mmHg
    8. Urine Output >0.5 ml/kg/h
    9. Systolic Blood Pressure >90 mmHg or mean arterial pressure >65-70 mmHg
      1. Blood Pressure is an unreliable marker of Sepsis severity and response to therapy
      2. Blood Pressure can be normal or high immediately before decompensated shock
      3. Patient positioning is also an unreliable marker to predict fluid Resuscitation response
        1. Technique involves raising legs and observing for increase in Blood Pressure
  4. Precautions
    1. Aggressive fluid hydration is key to Sepsis management, even in those at risk of Fluid Overload (CHF, CKD)
      1. CHF and CKD patients have decreased mortality with aggressive fluid hydration in Sepsis
      2. Aggressive fluid hydration offers significant benefit even if Lactic Acid 2-4 (intermediate range)
      3. Levy (2016) Am J Respir Crit Care Med 193(11):1195-6 +PMID:27248586 [PubMed]
      4. Liu (2016) Am J Respir Crit Care Med 193(11): 1264-70 +PMID:26695114 [PubMed]
    2. However, as of 2020, nuanced fluid approach is advocated (not part of Sepsis protocol)
      1. Early Sepsis trials used 20 cc/kg (not 30 cc/kg)
      2. Intravenous Fluids have a short duration in the intravascular space
        1. Third spacing of fluids
        2. Ongoing insensible losses
      3. Deliver fluids in 500 cc increments with reassessment in those at risk of fluid overloaf (CHF, CKD)
        1. See Fluid Responsiveness Markers
        2. Reevaluate with Inferior Vena Cava Ultrasound
      4. Consider early initiation of Vasopressors (e.g. Norepinephrine) via reliable peripheral IV
        1. Counters Sepsis-related vasodilation
        2. Allows for maintenance of mean arterial pressure (MAP) until fluid Resuscitation is adequate
      5. References
        1. Weingart and Swaminathan in Herbert (2020) EM:Rap 20(8): 3
        2. Weingart and Swaminathan in Swadron (2021) EM:Rap 21(11): 4-6
    3. Limit fluids in later stages of Sepsis management (day 2 to 3)
      1. Net fluid balance at 72 hours should approach Zero
      2. Excessive, persistent overhydration is associated with Acute Respiratory Distress Syndrome (ARDS)
      3. Each liter of Positive Fluid Balance at 72 hours increases mortality
        1. Sakr (2017) Crit Care Med 45(3): 386-94 +PMID:27922878 [PubMed]
    4. Avoid Albumin 5% (500 ml bolus) or similar colloid (lack of efficacy)
      1. Previously indicated for high volume crystalloid Resuscitation (e.g. >4-6 Liters)
      2. Was postulated to help prevent fluid third spacing
      3. No benefit to albumin over Isotonic Saline in Sepsis (as of 2014)
        1. Caironi (2014) N Engl J Med 370(15): 1412-21 [PubMed]
    5. Avoid hyroxyethyl starch
      1. Initially promoted to correct Metabolic Acidosis from high volume crystalloid Resuscitation
      2. Hydroxyethyl Starch is associated with increased risk of bleeding and Renal Failure
    6. Avoid Synthetic colloid gelatin
      1. Insufficient evidence as of 2022
  5. Transfusion: pRBC Indications
    1. Indications are controversial (follow a restrictive transfusion policy)
      1. Most current guidelines suggest transfusion for Hemoglobin <7 g/dl (Hematocrit <21)
      2. Other studies suggest transfusion for Hemoglobin <10 g/dl (Hematocrit <30)
        1. Goal Hematocrit >30% if Central Venous Oxygen Saturation <70% after restoring mean arterial pressure
    2. More important after the initial stabilization
    3. Mortality increases for transfusion for mild Anemia
    4. Herbert (1999) N Engl J Med 340:409-17 [PubMed]
    5. Rivers (2001) N Engl J Med 345(19): 1368-77 [PubMed]
  6. Transfusion: Platelet Indications
    1. Platelet Count <10,000/ul OR
    2. Platelet Count <20,000/ul and significant bleeding risk OR
    3. Platelet Count <50,000/ul if active bleeding, surgery or invasive procedure required
  1. Precaution
    1. Intravenous Fluids should be maximized prior to starting first pressor (>2 L) and second pressor (4 L)
    2. Vasopressor delay after refractory Hypotension (MAP <65) increases mortality 5% per hour of delay
      1. Bai (2014) Crit Care 18(5): 532 +PMID:25277635 [PubMed]
    3. Vasopressors ultimately require central venous access
      1. Vasopressors may be initiated peripherally during stabilization
        1. Temporize until Central Line access is obtained (typically <1-2 hours)
          1. Norepinephrine has been used via reliable large bore peripheral IV up to 24 hours
        2. Complication rates are higher with peripheral Vasopressor (but not life threatening)
        3. Risk outweighed by the benefit of initiating Vasopressors without delay
          1. Indicated after failed response to aggressive hydration
        4. Ricard (2013) Crit Care Med 41(9): 2108-15 [PubMed]
      2. Peripheral extravasation of Vasopressors may result in severe local tissue necrosis
        1. Potential with risk of Compartment Syndrome
        2. Vasopressors in peripheral lines must be observed very closely
          1. Frequently recheck for extravasation (e.g. every 10-15 min)
        3. Plan for immediate response to extravasation including available antidote (e.g. Phentolamine)
  2. Target perfusion
    1. Central Venous Pressure (CVP) 8-12 mmHg
    2. Mean arterial pressure (MAP) >65 mmHg
    3. Urine Output >0.5 ml/kg/h
    4. See venous oxygenation goals (Scvo2 or Svo2) above
  3. First agent
    1. Norepinephrine (preferred first line)
      1. Start at 2 to 5 mcg per minute (or 0.2 mcg/kg/min)
      2. Titrate to 35-90 mcg/min to adequate perfusion parameters including MAP >65 mmHg
        1. Dose range 0.1 to 1 mcg/kg/min
      3. May be initiated via peripheral access (while awaiting central access)
        1. Observe closely for extravasation and move to Central Line within 1-2 hours
        2. With large bore, reliable peripheral IVs, Norepinephrine has been used peripherally for 24 hours
      4. Add Vasopressin if refractory Hypotension
        1. Consider for Hypotension despite Norepinephrine 0.3 mcg/kg/min
  4. Second agent (added to first)
    1. Indicated for Hypotension despite fluid bolus and other additional measures listed below
    2. Vasopressin 0.03 units/minute (previously 0.04 units/min was recommended)
      1. Indicated for additional Vasopressor support if BP not at goal despite norephinephrine 0.25 to 0.5 mcg/kg/min
      2. Indicated when bedside Echocardiogram demonstrates good cardiac contractility with adequate inotropy
    3. Epinephrine
      1. Dose 20 to 50 mcg/min
      2. Indicated for combined Vasopressor and inotropic support
      3. Indicated when bedside Echocardiogram demonstrates poor cardiac contractility (Cardiomyopathy)
      4. Consider adding to Norepinephrine and Vasopressin when they are insufficient to maintain target pressures
  5. Agents to generally avoid in most Sepsis cases
    1. Avoid Dopamine in Sepsis
      1. Dosing range: 2-20 mcg/kg/min
      2. Do not use "renal dose" Dopamine - misnomer
      3. No longer recommended in Sepsis due to less effective than Norepinephrine and arrhythmogenic
      4. De Backer (2010) N Engl J Med 362(9): 779-89 [PubMed]
    2. Avoid Phenylephrine in Sepsis
      1. Lacks inotropic activity and overall efficacy in Sepsis shock when compared with Norepinephrine
      2. Has been used to bridge Vasopressor use until securing central venous access
        1. However Norepinephrine can be safely used for hours until central access has been obtained
      3. Has also been used to avoid Norepinephrine in Tachycardia
        1. However, Norepinephrine may still be used despite Tachycardia
      4. References
        1. Orman and Weingart in Herbert (2016) EM:Rap 16(8):7-8
  • Management
  • Stabilization - Additional measures when poor response to Resuscitation efforts
  1. Early Intensive Care Unit Admission
    1. ICU admission within 6 hours is preferred (ED boarding >6 hours is associated with increased Sepsis mortality)
  2. Corticosteroids
    1. Hydrocortisone at physiologic dose
      1. Hydrocortisone 50 mg IV every 6 hours (some give 100 mg as initial dose) for a total of 200 mg/day
    2. Indicated for severe Sepsis refractory to aggressive fluid Resuscitation and Vasopressor therapy
      1. Consider if patient is requiring 2 pressors for support (e.g. Norepinephrine and Epinephrine or Vasopressin)
      2. Benefit may be limited to early administration
    3. Efficacy (variable evidence)
      1. Surviving Sepsis Campaign downgraded Corticosteroid recommendation to weak evidence in 2012
      2. Marked mortality benefit from Corticosteroids in severe Sepsis
        1. Annane (2002) JAMA 288(7):862-71 [PubMed]
      3. Has short-term benefit in duration and severity
        1. Annand (2009) JAMA 301:2362 [PubMed]
      4. CORTICUS trial found no benefit to Corticosteroids overall
        1. However of those who did respond to shock reversal, those on Corticosteroids improved faster
        2. Sprung (2008) NEJM 358(2): 111-24 [PubMed]
      5. ADRENAL Study found no significant benefit
        1. However some argue higher doses given earlier in course are beneficial
        2. Swaminathan and Weingart in Herbert (2018) EM:Rap 18(8): 9
        3. Venkatesh (2018) N Engl J Med 378(9): 797-808 [PubMed]
  3. Consider Hypothyroidism
  4. Consider additional Intravenous Fluids (if suspect still volume down)
    1. Give additional 1-2 Liters on top of already administered 2 liters
  5. Occult Hemorrhage (e.g. Gastrointestinal Bleeding)
    1. Stop bleeding and Consider pRBC transfusion if actively bleeding or Hemoglobin <7.0 mg/dl
  6. Inotropes (e.g. Dobutamine)
    1. Consider for inotropic support when inadequate tissue perfusion/oxygenation
      1. Especially where myocardial dysfunction is suspected
    2. Consider when perfusion markers fail to improve despite MAP >65 mmHg and Oxygen Saturation >90%
      1. SvcO2 <70% or
      2. Lactic Acid fails to improve or
      3. Serum Creatinine fails to improve
    3. Approach
      1. Dobutamine 2.5 mcg/kg/min
      2. Obtain beside Echocardiogram if available
  7. Other agents (avoid in general)
    1. Activated Protein C
    2. Drotecogin Alfa (Xigris)
      1. No survival benefit
      2. Removed from the U.S. market in october 2011 (as well as surviving Sepsis guidelines)
    3. Sodium Bicarbonate
      1. Not generally recommended
      2. Acidosis improves with improved perfusion
    4. Vitamin C
      1. No benefit in Sepsis or Septic Shock
  • Management
  • Supportive Measures
  1. See Critical Care
  2. Nutrition
    1. Maintain adequate nutrition
      1. Initiate Enteral Nutrition via oral Gastric Tube or nasal Gastric Tube within first 24-48 hours
        1. Enteral Nutrition is preferred over Total Parenteral Nutrition
        2. May consider Parenteral nutrition at 7 days if inadequate Enteral Nutrition
          1. Dextrose infusions may be used in first 7 days
    2. Blood Glucose
      1. Conventional therapy (non-intensive Blood Sugar management: 144-180) is safer in critically ill
      2. Tight Glucose control is associated with Hypoglycemia and increased mortality
      3. Hyperglycemia is associated with apoptosis, ischemia and delayed healing
      4. Ideally mean Glucose of 150 mg/dl is preferred (range 144-180 mg/dl or 8-10 mmol/L)
      5. Consider initiating Insulin for Serum Glucose >180 mg/dl (10 mmol/L) targeting above range
        1. Monitor Serum Glucose every 1-2 hours if initiating Insulin, then every 4 hours
      6. (2009) N Engl J Med 360:1283-97 [PubMed]
  3. Manage Hypocalcemia (based on Ionized Calcium or Corrected Serum Calcium for albumin)
    1. Replace with Calcium Gluconate or Calcium Chloride if hypocalcemic
  4. Stress Ulcer or Peptic Ulcer prophylaxis
    1. Indications
      1. Thrombocytopenia
      2. Multiorgan failure
      3. Mechanical Ventilation
    2. Agents (either are equivalent)
      1. H2 Blocker (e.g. Ranitidine) or
      2. Proton Pump Inhibitor (e.g. Protonix)
        1. Reduces gastrointestinal Hemorrhage risk in high risk septic patients
        2. Does not appear to increase risk of C. Diff or Pneumonia
  5. DVT Prophylaxis
    1. DVT occurs in up to 14% of critically ill patients not on prophylaxis
    2. DVT Prophylaxis options
      1. Low molecular-weight Heparin (preferred) or
      2. Low dose Unfractionated Heparin or
      3. Mechanical compression devices if Heparin contraindicated
  1. Precautions
    1. Septic Shock in ESRD is of the highest risk
    2. ESRD patients have decreased mortality with aggressive fluid hydration in Sepsis (see above)
  2. Fluid management
    1. All septic patients with Hypotension or Lactic Acidosis will typically need upwards of 30 cc/kg replacement
      1. Early intubation may be preferred
      2. If monitoring (e.g. Ultrasound, CVP) is unreliable or unavailable
        1. Aim for 3-4 Liter IV fluid Resuscitation (in 500 cc increments)
    2. Emergency Dialysis may be required to manage Fluid Overload
      1. Aggressive fluid Resuscitation needed in Sepsis
    3. Use Inferior Vena Cava Ultrasound to drive fluid Resuscitation
      1. See Inferior Vena Cava Ultrasound for Volume Status
      2. Hypotension or Lactic Acidosis
        1. Vena cava collapses with inspiration: Give 500 cc IV bolus of fluid
        2. Vena cava does not collapse: Start Vasopressors (e.g. Norepinephrine)
      3. Repeat cycle of fluid bolus or pressors followed by Ultrasound until Hypotension and Lactic Acidosis resolve
        1. Once IVC does not collapse with inspiration, move to Vasopressors
        2. Consider starting Vasopressors earlier to help increase Preload
        3. Consider monitoring Central Venous Pressure (CVP) to confirm that it remains low following fluid bolus
  3. Airway management
    1. Early intubation is preferred over crash airway management
    2. BIPAP or intubation may be needed with fluid Resuscitation related to Fluid Shifts and Pulmonary Edema (expected)
  4. Labs
    1. Serum lactate is an accurate reflection of Sepsis status in ESRD (lactate undergoes Hepatic Clearance)
  5. References
    1. Weingart and Orman in Majoewsky (2013) EM:Rap 13(10): 6
  • Management
  • Pregnancy and Sepsis
  1. Sepsis Criteria in pregnancy
    1. SIRS Criteria and qSOFA Criteria do NOT apply to pregnant patients
    2. No obstetric Sepsis scoring system (MEWS, MOEWS, SOS) has been found accurate enough for clinical use
      1. Edwards (2015) Am J Obstet Gynecol 212(4):536.e1-8 +PMID:25446705 [PubMed]
      2. Albright (2017) Obstet Gynecol 120(4):747-55 +PMID:28885400 [PubMed]
    3. Blood Pressure, Heart Rate, Respiratory Rate are unreliable markers of serious infection or instability in pregnancy
    4. Lactic Acid cutoffs do not apply well to pregnancy, but higher Lactic Acids are associated with worse outcomes
      1. Albright (2015) Am J Perinatol 32(5):481-6 [PubMed]
  2. General measures
    1. Left lateral decubitus position
    2. Aggressive fluid Resuscitation (as with other Sepsis cases)
    3. Blood Products as needed
    4. Early antbiotics
    5. No clear evidence for one Vasopressor over another in pregnancy (Norepinephrine is reasonable)
  3. References
    1. Swadron, Schmitz, Bridwell, Carius in Herbert (2019) EM:Rap 19(3): 12-4
  • Prognosis
  1. Positive Blood Culture
    1. Confers 150% increase in mortality risk
  2. Mortality
    1. Severe Sepsis: 25-30%
    2. Septic Shock: 40-70%
  3. Longterm outcomes
    1. Increased risk of readmission and death in the subsequent year after Sepsis discharge
    2. Significantly decreased physical and cognitive function after Sepsis discharge
    3. Yende (2016) Crit Care Med 44(8):1461-7 [PubMed]
  • Resources
  1. Surviving Sepsis Protocol
    1. http://www.annalsofintensivecare.com/content/pdf/2110-5820-1-17.pdf
  • References
  1. (2016) CALS, 14th ed, 1:46
  2. Goldberg (2015) Crit Dec Emerg Med 29(3): 9-19
  3. Guirgis and Khadpe (2017) Crit Dec Emerg Med 31(12): 3-8
  4. Palizzolo, Dunne and Gaieski (2024) Crit Dec Emerg Med 38(6): 4-13
  5. Swadron and Goldberg in Majoewsky (2013) EM:RAP 13(6): 2-4
  6. Weingart and Orman in Majoewsky (2014) EM:RAP 14(8): 3-5
  7. Orman and Weingart in Majoewsky (2012) EM:RAP 12(10): 4-7
  8. Khoujah (2013) Crit Dec Emerg Med 27(4):12-21
  9. Marik (2011) Annals of Intensive Care 1:17
  10. Angus (2013) N Engl J Med 369(9): 840-51 +PMID:23984731 [PubMed]
  11. Annane (2005) Lancet 365(9453): 63-78 [PubMed]
  12. Cunha (2008) Crit Care Clin 24(2): 313-34 [PubMed]
  13. Dellinger (2013) Crit Care Med 41(2):580-637 [PubMed]
  14. Evans (2021) Crit Care Med 49(11):e1063-143 +PMID: 34605781 [PubMed]
  15. Gauer (2013) Am Fam Physician 88(1): 44-53 [PubMed]
  16. Gauer (2020) Am Fam Physician 101(7): 409-18 [PubMed]
  17. Jaimes (2004) Clin Infect Dis 38:357-62 [PubMed]
  18. Lever (2007) BMJ 335(7625): 879-83 [PubMed]
  19. Singer (2016) JAMA 315(8): 801-10 [PubMed]