What it is:Sepsis is life-threatening organ dysfunction caused by a dysregulated host response to infection (Sepsis-3, 2016). Septic shock is the subset of sepsis where the circulation and cellular metabolism are deranged enough to markedly raise mortality: persistent hypotension that needs vasopressors to keep mean arterial pressure (MAP) at 65 mm Hg, plus a lactate above 2 mmol/L, despite adequate fluid resuscitation.
The core problem:the infection itself is often not what kills the patient, it's the body's own runaway response to it. Pro-inflammatory cytokines damage the vascular endothelium, protein C (which normally calms inflammation and supports fibrinolysis) gets consumed, and the vasculature dilates and leaks while microthrombi form in parallel. The net hemodynamic picture is a hyperdynamic state: high cardiac output, but pathologically low systemic vascular resistance, so end organs stop getting adequately perfused even though the heart is pumping hard.
What you do about it:you run against a clock. Cultures, broad-spectrum antibiotics, and a 30 mL/kg crystalloid bolus all start within the first hour, before you even know the organism. Everything after that is de-escalation, hemodynamic titration, and organ support.
Think of the first hours of sepsis care as four levers pulled simultaneously, not sequentially: (1) find and hit the bug, cultures plus broad-spectrum antibiotics plus source control, (2) refill the tank, fluids, (3) support the pressure, vasopressors once fluids alone aren't enough, and (4) protect the other organs, glucose control, coagulation, renal and GI protection. Nearly every guideline recommendation in this chapter is just a more detailed instruction for one of those four levers.
Sepsis-3 (2016) didn't just tweak the old criteria, it collapsed two tiers into one. "Severe sepsis" no longer exists as its own category, because the new definition of sepsis already requires organ dysfunction.
| Concept | Sepsis-2 (2012) | Sepsis-3 (2016) |
|---|---|---|
| Screening tool | SIRS: ≥2 of temp >38°C or <36°C, HR >90/min, RR >20/min, WBC >12,000 or <4,000/mm³ or >10% bands | SOFA score, acute change ≥2 points |
| Sepsis | SIRS + probable or documented infection | Life-threatening organ dysfunction from a dysregulated host response to infection |
| Severe sepsis | Sepsis + ≥1 organ dysfunction or hypoperfusion | Retired.Folded into "sepsis" since organ dysfunction is now baked into the definition |
| Septic shock | Sepsis + hypotension refractory to 30 mL/kg fluids, or lactate >1 mmol/L | Sepsis + vasopressor-requiring hypotension to MAP ≥65 mm Hg + lactate >2 mmol/L despite adequate fluids |
The lactate cutoff for septic shock moved from >1 mmol/L to >2 mmol/Lbetween the two guidelines, and it's now paired with a vasopressor requirement rather than an "or." SIRS criteria are sensitive but nonspecific (a patient with the flu can meet SIRS), which is exactly why Sepsis-3 shifted the screening tool to SOFA, which tracks actual organ dysfunction.
SOFA scores six organ systems, pulmonary, hepatic, cardiovascular, renal, coagulation, and neurological, each from 0 to 4. An acute rise in total SOFA of ≥2 pointsin a patient with presumed infection is the operational definition of sepsis, and it's associated with roughly a 10% increase in hospital mortality.
Risk climbs at the extremes of age and with preexisting conditions that either predispose to infection or blunt the ability to fight one off: heart failure, diabetes, COPD, cirrhosis, alcohol use disorder, end-stage renal disease, and other immunosuppressive states like malignancy or HIV.
| Source | Frequency |
|---|---|
| Lung | 40-42% |
| Intra-abdominal | 31-34% |
| Genitourinary | 11-15% |
On blood culture, gram-negative organisms show up in 44-59% of cases, gram-positives in 37-52%, anaerobes in 5%, and fungi in 4-10%. E. coli is by far the most common gram-negative isolate(55-60%), followed by Klebsiella, Proteus, Enterobacter, and Pseudomonas aeruginosa. S. aureusleads the gram-positives, followed by coagulase-negative staph, Enterococcus, and S. pneumoniae. Among fungi, Candida albicans is the dominant bloodstream pathogen.
Mortality climbs sharply with disease severity: 3.5% in sepsis, 9.9% in severe sepsis, 29% in septic shock.Pseudomonas raises mortality further within that curve. Candidemia is its own outlier: 30-day mortality for septic shock caused by candidemia is 54%, worse than most bacterial causes.
Gram-negative organisms trigger this cascade through lipopolysaccharide (endotoxin), and specifically its Lipid A component, which is intensely immunoreactive and drives most of the downstream toxicity. Gram-positive organisms do something analogous through exotoxin and peptidoglycanon the cell wall surface. Either trigger sets off a flood of pro-inflammatory cytokines (TNF-α, IL-1, IL-6, IL-12) that's supposed to be balanced by anti-inflammatory mediators (IL-1 receptor antagonist, IL-4, IL-10), but in sepsis that balance fails. Early TNF-α also stimulates cyclooxygenase-derived products (thromboxane A2, prostaglandins) that directly injure vascular endothelium.
That endothelial injuryis the central lesion: damaged endothelium lets granulocytes and plasma constituents leak into tissue that shouldn't be inflamed, and that's what produces organ damage. At the same time, protein C, which normally enhances fibrinolysis and dampens inflammation, gets consumed and depleted, tipping the whole system further toward clotting and inflammation simultaneously. That combination (clotting and leaking, at once) is what produces DIC.
The hemodynamic hallmark of sepsis is low systemic vascular resistance with a high (or preserved) cardiac output, the vessels are pathologically dilated, not the pump that's failing. That's why the first-line drug is a vasoconstrictor(norepinephrine), not an inotrope, and it's why dopamine, which leans harder on beta-1/chronotropic effects at moderate doses, is a worse fit than a drug that's almost purely alpha-1.
Presentation depends heavily on the site of infection, the patient's baseline health, and how far organ dysfunction has progressed. It ranges from vague and easy to miss, to obviously critical.
| Stage | Findings |
|---|---|
| Early / nonspecific | Malaise, myalgia, fever orhypothermia, chills, tachycardia, tachypnea, change in mental status |
| Organ dysfunction (progressing sepsis) | Oliguria, hemodynamic instability or frank shock, lactic acidosis, hyperglycemia or hypoglycemia, leukopenia, DIC, thrombocytopenia, ARDS, GI hemorrhage, coma |
| Hemodynamic signature (septic shock) | Warm, flushed skin early (low SVR), tachycardia, widened then narrowing pulse pressure as shock progresses, refractory hypotension |
Don't anchor on fever. Hypothermia is just as much a sepsis finding as fever, especially at the extremes of age, and a normal or low WBC (leukopenia) is just as consistent with sepsis as leukocytosis. A patient who "doesn't look infected" on a quick vital sign scan can still be septic.
There's no single confirmatory test. Diagnosis is the combination of a documented or probable infection plus objective evidence of organ dysfunction, operationalized as an acute SOFA rise of ≥2 points.
| Test / Marker | Threshold | What it tells you |
|---|---|---|
| Lactate | ≥2 mmol/L (≥18 mg/dL) | Marker of tissue hypoperfusion; repeat within 2 hours if elevated |
| Lactate | ≥4 mmol/L (≥36 mg/dL) | Bundle trigger for immediate 30 mL/kg fluid resuscitation |
| SOFA score | Acute rise ≥2 points | Operational definition of sepsis; associated with ~10% higher mortality |
| MAP | <65 mm Hg after fluids | Trigger to start vasopressors |
| Cultures | Blood, urine, sputum, or other suspected site | Obtain before antibiotics if it doesn't delay treatment; never let it push antibiotics past the 1-hour mark |
A single lactate value tells you how sick the patient is right now. The trendtells you whether resuscitation is working. Normalizing an elevated lactate is itself a resuscitation target, not just a lab you note and move on from.
The goals of treatment are timely diagnosis and pathogen identification, prompt hemodynamic support, source control (medical and/or surgical), early broad-spectrum antimicrobials, and avoiding progression to organ failure or shock. The Sepsis-3 performance bundle packages the first hour of care into four things that all start at once, not in sequence.
Measure immediately.
Blood, urine, sputum, etc.
Broad-spectrum IV, against likely pathogens.
Crystalloid bolus, then vasopressors if still hypotensive.
Compliance with this bundle is tracked as a quality metric, and the outcomes that matter downstream are ICU/hospital length of stay, rate of organ dysfunction, and mortality. The whole point of packing everything into 60 minutes is that sepsis can progress even with good fluids and antibiotics if source control lags, so speed on every front matters simultaneously.
Early, effective fluid resuscitation is what prevents progression from sepsis-induced hypoperfusion to full septic shock. Give at least 30 mL/kg of IV crystalloid within the first 3 hours(strong recommendation, low-quality evidence), then reassess.
Hydroxyethyl starch products increase the risk of renal failure, need for renal replacement therapy, and deathacross multiple studies. There's no scenario in septic shock where hetastarch is the right choice.
| Phase | Timeframe | Fluid goal |
|---|---|---|
| Resuscitation | Minutes | Positive fluid balance is expected, especially right after the 30 mL/kg bolus |
| Optimization | Hours | Neutral balance, intake roughly matching output |
| Stabilization | Days | Neutral to net negative; maintenance-only fluids (~30 mL/kg/day) |
| Evacuation | Days to weeks | Net negative balance, actively removing accumulated fluid |
The practical takeaway: the same patient who needed aggressive fluids on day 1 can be fluid-overloaded by day 3 if nobody moves them through these phases. Fluid therapy isn't a one-time decision, it's a moving target across the admission.
Empiric IV broad-spectrum antibiotics start within 1 hourof recognizing sepsis or septic shock (strong recommendation, moderate evidence), aimed at the most likely bacterial or fungal pathogens given the suspected source.
Cultures should be drawn beforeantibiotics when possible, but that sequencing must never delay antibiotic administration past the 1-hour window. If cultures can't be obtained fast enough, treat first.
Give empiric antifungal therapy to patients at high risk of fungal infection (weak recommendation, low-quality evidence). For suspected invasive candidiasis in nonneutropenic ICU patients, the preferred empiric choice is an echinocandin(anidulafungin, micafungin, or caspofungin). Triazoles(fluconazole, voriconazole) are an option in hemodynamically stable patients who haven't had recent triazole exposure and aren't known to be colonized with an azole-resistant Candida species.
Duration:the average course for bacterial sepsis is 7-10 dayswithout source control complications; fungal infections typically need 10-14 days. The Surgical Infection Society recommends no more than 4 full daysof antimicrobials when source control was adequate, and no more than 5-7 dayswhen definitive source control wasn't achieved. Clinical improvement plus a falling procalcitonin can support shortening a course further (weak recommendation, low-quality evidence).
Coverage splits along two axes: where the infection is, and whether it's community- or hospital-acquired (hospital-acquired regimens lean broader and add antipseudomonal or MRSA coverage).
| Source | Community-Acquired | Hospital-Acquired |
|---|---|---|
| Urinary tract | ||
| Ceftriaxone or ciprofloxacin/levofloxacin | Ceftriaxone/ceftazidime or ciprofloxacin/levofloxacin | |
| Respiratory tract | ||
| Levofloxacin 750 mg daily or moxifloxacin, or ceftriaxone + clarithromycin/azithromycin | Piperacillin/tazobactam or ceftazidime or cefepime or a carbapenem, + levofloxacin/ciprofloxacin or an aminoglycoside, ± vancomycin or linezolid | |
| Intra-abdominal | ||
| Ertapenem or ciprofloxacin/levofloxacin + metronidazole, or ceftriaxone + metronidazole | A carbapenem or piperacillin/tazobactam or ceftazidime/cefepime, + metronidazole | |
| Skin / soft tissue | ||
| Vancomycin or linezolid or daptomycin | Vancomycin + piperacillin/tazobactam | |
| Catheter-related | ||
| Vancomycin | ||
| Unknown source | ||
| Piperacillin/tazobactam or a carbapenem | A carbapenem | |
Carbapenem here means imipenem, meropenem, or doripenem. Levofloxacin for respiratory sepsis is dosed higher than routine outpatient dosing (750 mg once daily).
Once fluids alone aren't holding MAP ≥65 mm Hg, start a vasopressor and titrate carefully to an endpoint of adequate organ perfusion, not just a number on the monitor.
| Drug | Receptor emphasis | HR | SVR | CO | Role in septic shock |
|---|---|---|---|---|---|
| Norepinephrine | α1 dominant, mild β1 | ↔/↑ | ↑↑ | ↑ | First-line.Raises pressure without heavy chronotropic cost |
| Vasopressin | V1 | ↔/↓ | ↑↑ | ↔/↓ | Add-on to norepinephrine instead of escalating the norepinephrine dose further |
| Epinephrine | α1 = β1 = β2, all strong | ↑ | ↑↑ | ↑↑ | Second-line/refractory shock; more arrhythmogenic |
| Phenylephrine | α1 only | ↔/↓ | ↑↑ | ↔ | Pure vasoconstrictor, no inotropy; reflex bradycardia can drop CO |
| Dopamine, low (0.5-2 mcg/kg/min) | Dopaminergic | ↔/↑ | ↔/↑ | ↑ | "Renal-dose" dopamine; no longer recommended, no proven renal protection |
| Dopamine, mid (5-10 mcg/kg/min) | β1, some α1/β2 | ↑ | ↔/↑ | ↑↑ | Inotropic/chronotropic range |
| Dopamine, high (10-20 mcg/kg/min) | α1 dominant | ↑ | ↑↑ | ↑ | Vasopressor range; more arrhythmogenic than norepinephrine |
| Angiotensin II | AT1 | ↑ | ↑ | ↑ | Adjunct in refractory shock |
Compared to dopamine, norepinephrine is less arrhythmogenic and is associated with lower mortalityin septic shock. Dopamine's heavier beta-1 chronotropic push and higher arrhythmia risk are exactly the trade-offs you're trying to avoid in a patient who's already hemodynamically unstable. "Renal-dose" dopamine is an outdated concept: it never showed kidney protection and just adds arrhythmia risk.
IV hydrocortisone is a weak recommendation, low-quality evidenceaddition for septic shock patients with an ongoing vasopressor requirement, it's an adjunct for refractory shock, not routine first-line therapy.
Trigger and dose:add low-dose hydrocortisone 200 mg/daywhen the patient needs an ongoing norepinephrine dose >0.25 mcg/kg/min for at least 4 hoursto maintain the target MAP.
Taper hydrocortisone once vasopressors are no longer needed. Abrupt cessation has been linked to hemodynamic and immunologic rebound, essentially un-doing the stability you just built.
Start an insulin dosing protocol once two consecutive blood glucose readings exceed 180 mg/dL (10 mmol/L)(strong recommendation, high-quality evidence). Target an upper blood glucose below 180 mg/dL, with a practical target range of 144-180 mg/dL (8-10 mmol/L). Tight, near-normal glycemic control isn't the goal here, avoiding both marked hyperglycemia and iatrogenic hypoglycemia is.
VTE prophylaxis:daily subcutaneous low-molecular-weight heparinis preferred over unfractionated heparin (strong recommendation, moderate evidence) for essentially all ICU patients admitted with sepsis or septic shock. If pharmacologic prophylaxis is contraindicated (active bleeding, severe thrombocytopenia), use mechanical prophylaxis instead.
Stress ulcer prophylaxis:give it to patients who have bleeding risk factors (strong recommendation, low-quality evidence). Either a proton pump inhibitor or an H2-receptor blocker is acceptable (weak recommendation, low-quality evidence), this isn't a drug-specific mandate, it's a class-level one.
Uncontrolled sepsis, or sepsis without source control, doesn't just plateau, it produces a specific and predictable set of downstream organ injuries.
Sepsis can keep progressing even with rapid, appropriate fluids and antibiotics if the underlying source, an abscess, an infected line, necrotic tissue, isn't controlled medically or surgically. Antibiotics buy time; they don't substitute for draining or removing the source.
| Parameter | When | Watching for |
|---|---|---|
| Lactate | Baseline, repeat within 2 hours if initial ≥2 mmol/L | Trend toward normal = resuscitation is working; persistently high = ongoing hypoperfusion |
| MAP | Continuously once on vasopressors | Keep ≥65 mm Hg; guides pressor titration |
| Urine output, mental status, skin perfusion | Continuously | End-organ perfusion; don't chase the MAP number while ignoring the patient |
| Blood glucose | Frequently while on an insulin protocol | Target 144-180 mg/dL, avoid hypoglycemia |
| SOFA / organ function labs | Serial, at least daily | Worsening trend = escalate support; improving = consider de-escalation |
| Culture results and clinical status | Daily | De-escalate or narrow antibiotics as soon as it's safe; also guides total duration |
| Platelets, PT/INR, fibrinogen | If DIC is suspected | Evolving consumptive coagulopathy |
| Vasopressor dose trend | Continuously | Norepinephrine >0.25 mcg/kg/min for ≥4h triggers hydrocortisone; a falling requirement triggers the steroid taper |
Septic patients are frequently altered, sedated, or intubated, so most of this "counseling" happens with family at bedside, plus the patient at the point of discharge.