Sepsis represents life-threatening organ dysfunction caused by a dysregulated response to infection that kills up to one in three affected people. Early identification and appropriate management can improve these outcomes. The Society of Critical Care Medicine has updated the guidelines of the Surviving Sepsis Campaign.

Identifying Sepsis

Performance improvement programs involving screening programs, education, sepsis bundles, and monitoring for process improvement reduce mortality in patients with sepsis and septic shock with an odds ratio of 0.66 (95% CI, 0.61 to 0.72). Mortality is lower when sepsis bundles are routinely used, at least in high resource countries. Machine learning algorithms are the most accurate way to identify sepsis, with an average 81% sensitivity and 72% specificity.

When using clinical criteria to diagnose sepsis, the quick Sequential Organ Failure Assessment should be avoided because of low sensitivity. The systemic inflammatory response syndrome criteria, National Early Warning Score, and Modified Early Warning Score are more accurate. An early lactate level is recommended if sepsis is suspected because an elevated level correlates with mortality. An elevated lactate level has a positive likelihood ratio of 5 for sepsis, whereas a normal lactate level has a negative likelihood ratio of 0.3.

Resuscitation

With sepsis-induced hypoperfusion or shock, resuscitation with 30 mL per kg of crystalloid fluids is recommended within the first three hours. Failing to meet this initial resuscitation guideline delays resolution of hypotension and increases intensive care duration and mortality.

Balanced crystalloids such as lactated ringers are recommended over normal saline because of weak evidence of decreased mortality. Albumin has no clear benefits over balanced crystalloids and is more expensive. After infusing large volumes of crystalloid fluids, albumin can be considered to improve blood pressure. Hydroxyethyl starch solutions should be avoided because of increased mortality, and synthetic colloid gelatin should not be used because of lack of evidence.

Subsequent resuscitation depends on intravascular volume status and organ perfusion. Static measures such as heart rate, central venous pressure, and systolic blood pressure are poor indicators of fluid status, and using dynamic measures reduces mortality with a risk ratio of 0.59 (95% CI, 0.42 to 0.83). Dynamic measures of fluid status include cardiac output measurement during passive leg raising, stroke volume measurement in response to fluid challenges or changes in intrathoracic pressure, and changes in pulse pressure or systolic blood pressure in response to fluid challenges.

Although lactate levels often will not decrease to normal, levels approaching normal suggest successful resuscitation. Normalization of capillary refill time can also demonstrate resuscitation.

If vasopressors are necessary after initial fluid resuscitation, the target mean arterial pressure is 65 mm Hg because higher targets do not improve survival. Vasopressors should be started peripherally rather than waiting for central venous access. Arterial blood pressure monitoring is essential and carries low risks of limb ischemia and bleeding. Intensive care admission should be completed within six hours because mortality increases when emergency department stays exceed that time frame.

Norepinephrine is the first-line vasopressor agent, with epinephrine or dopamine as acceptable alternatives although they increase lactate level and the risk of arrhythmia. Adding vasopressin at 0.03 units per minute should be considered if norepinephrine infusions reach 0.25 to 0.5 mcg per kg per minute without reaching blood pressure goals. Epinephrine can be considered if goals are not met with norepinephrine and vasopressin. For hypoperfusion with adequate volume status and arterial blood pressure, adding dobutamine to norepinephrine or using epinephrine alone are options.

Infection Control

Reducing time to administration of antimicrobials reduces mortality. For patients in septic shock, antimicrobials should be started within one hour because mortality increases with each hour of delay. In sepsis without shock, the evidence is less clear, and antimicrobials are recommended within three hours of recognition. Adding procalcitonin to clinical evaluation for antibiotic initiation does not improve mortality, length of intensive care, or hospitalization and is not recommended.

Because methicillin-resistant Staphylococcus aureus (MRSA) accounts for one in 20 infections in critically ill patients, empiric antimicrobial therapy should cover for MRSA in patients at high risk because delays can increase mortality. Although empiric combination antibiotics against gram-negative bacteria do not improve overall mortality, combination antibiotics are recommended for patients at high risk for multidrug-resistant organisms. Similarly, empiric antifungal agents do not improve short-term mortality in critically ill patients but are recommended for patients at high risk for invasive fungal infections.