sepsis of unknown source
in review
5 min read
Updated 2026-09-29
sepsis of unknown source
Sepsis = life-threatening organ dysfunction from dysregulated host response to infection (Sepsis-3). Undifferentiated sepsis → broad empiric cover for the likely sources (lung, urine, abdomen, skin, line) while hunting for the source. Antibiotics within 1 hour in septic shock; within 3 hours for sepsis without shock. Narrow at 48–72 h. Gram-negative bacteraemia: 7 days if improving (BALANCE (2025)); S. aureus bacteraemia is its own pathway.
sepsis of unknown source — initial management
- blood cultures × 2 sets (peripheral + each CVC lumen) BEFORE antibiotics — do not delay antibiotics >45 min for cultures
- empiric antibiotics within 1 h (septic shock) or 3 h (sepsis without shock)
- first-line: piperacillin-tazobactam 4.5 g IV q6h
- penicillin allergy: ceftriaxone 2 g IV q24h + metronidazole 500 mg IV q12h, or meropenem — see penicillin allergy
- add vancomycin 25–30 mg/kg IV load if: suspected line infection, MRSA risk, purulent SSTI, or septic shock
- neutropaenic: see febrile neutropaenia
- IV balanced crystalloid 30 mL/kg within 3 h for hypotension or lactate ≥4 mmol/L — reassess after each bolus
- noradrenaline first-line — target MAP ≥65 mmHg
- source work-up: CXR, urinalysis + culture, lactate, examine skin/lines/joints/surgical sites; CT abdomen/pelvis if no source
- NARROW at 48–72 h once source and susceptibilities known
local antibiogram
Gram-negative empiric choice depends on local ESBL and fluoroquinolone resistance — prior culture results for the individual patient matter more than the unit antibiogram.
definitions (Sepsis-3)
| term | definition |
|---|---|
| sepsis | infection + acute organ dysfunction (SOFA ≥2 above baseline) |
| septic shock | sepsis + vasopressors to maintain MAP ≥65 + lactate >2 mmol/L despite adequate resuscitation |
qSOFA (prognostic prompt — not a screening or diagnostic tool)
- altered mental status (GCS <15)
- respiratory rate ≥22/min
- systolic BP ≤100 mmHg
≥2 → assess organ dysfunction (full SOFA, lactate). SSC 2021 recommends against qSOFA alone as a screening tool — low sensitivity.
source identification
| source | key investigations |
|---|---|
| respiratory | CXR, sputum culture, consider CT chest |
| urinary | urinalysis, urine culture; ultrasound/CT if obstruction possible |
| abdominal | CT abdomen/pelvis with contrast |
| skin/soft tissue | examination incl. perineum and pressure areas; wound/pus cultures |
| line-related | paired peripheral + line cultures (differential time to positivity) |
| endocarditis, meningitis, bone/joint, spine | guided by clinical findings |
Every patient: blood cultures × 2, CXR, urinalysis + culture, and a full examination (skin, lines, surgical sites, joints, spine). Beyond that, culture the suspected source — not reflexively everything.
empiric antibiotics — detail
community-acquired, immunocompetent
| clinical scenario | regimen |
|---|---|
| undifferentiated, no source | piperacillin-tazobactam 4.5 g IV q6h |
| likely respiratory | ceftriaxone 2 g IV + azithromycin 500 mg IV (community-acquired pneumonia) |
| likely urinary | ceftriaxone (urinary tract infections); ertapenem/meropenem if prior ESBL |
| likely abdominal | piperacillin-tazobactam (intra-abdominal infections) |
| likely skin/soft tissue | cefazolin 2 g IV q8h (non-purulent) or vancomycin (purulent/MRSA risk) |
| likely line-related | vancomycin + piperacillin-tazobactam; remove the line |
healthcare-associated / resistant organism risk
- piperacillin-tazobactam 4.5 g IV q6h + vancomycin
- prior ESBL or resistant Gram-negatives: meropenem 1 g IV q8h + vancomycin
- double Gram-negative cover (β-lactam + aminoglycoside) only in septic shock with high MDR risk (SSC 2021)
- prolonged/continuous β-lactam infusion suggested after the loading dose (SSC 2021) — BLING III (2024): no significant 90-day mortality difference alone, favourable in meta-analysis
penicillin allergy
- pip-tazo shares ~15–30% cross-reactivity with penicillin/amoxicillin allergy → avoid after a genuine immediate penicillin reaction
- ceftriaxone + metronidazole, or meropenem, usable after most penicillin reactions including anaphylaxis
- avoid all β-lactams only after SCAR, serum sickness, AIN, or haemolysis → ID consult; aztreonam IV is Special Access Programme only in Canada
- full framework: penicillin allergy
resuscitation
fluid
- 30 mL/kg crystalloid within 3 h for hypotension or lactate ≥4 (SSC 2021 weak recommendation) — reassess after each bolus
- restrictive vs liberal fluid after the initial bolus: no mortality difference (CLOVERS (2023), CLASSIC (2022)) → earlier vasopressors reasonable
- balanced crystalloid preferred over 0.9% saline (SMART (2018))
- albumin only when large crystalloid volumes are needed
- assess fluid responsiveness (passive leg raise, pulse pressure variation) — avoid fluid overload
vasopressors
- noradrenaline first-line — target MAP ≥65 mmHg; peripheral start acceptable while central access is arranged
- add vasopressin 0.03 U/min if noradrenaline ≥0.25 µg/kg/min
- dobutamine if low cardiac output despite adequate MAP and volume
- ProCESS (2014): protocolised EGDT not superior to usual care
corticosteroids
- hydrocortisone 50 mg IV q6h when noradrenaline ≥0.25 µg/kg/min for ≥4 h (SSC 2021); SCCM 2024 suggests for septic shock
- not for sepsis without shock
reassessment at 48–72 h
- review cultures → de-escalate
- no source and cultures negative → reconsider the diagnosis: pancreatitis, PE, adrenal crisis, drug reaction, vasculitis, malignancy, HLH
- not improving → repeat cultures, CT, echocardiography, ID consult
- trend lactate — normalising lactate tracks response
duration
| scenario | duration |
|---|---|
| Gram-negative bacteraemia, source controlled, improving | 7 days — BALANCE (2025): non-inferior to 14 days |
| S. aureus bacteraemia | see Staphylococcus aureus bacteraemia — ≥14 days from first negative culture |
| source-specific | CAP 5 days, UTI 7 days, IAI 4 days after source control |
| culture-negative sepsis, treated | reassess at 48–72 h; stop if alternative diagnosis; otherwise 5–7 days |
traps
- each hour of delay to effective antibiotics in septic shock increases mortality (Kumar (2006)) — this does not justify reflex antibiotics in stable patients with unclear diagnoses
- lactate is not specific to sepsis — seizures, mesenteric ischaemia, liver failure, adrenaline, salbutamol, metformin, thiamine deficiency
- qSOFA 0–1 does not exclude sepsis in a patient who looks unwell
- a single blood culture with CoNS is usually a contaminant; S. aureus in blood is never a contaminant → ID consult + echocardiography
- Candida in blood is never a contaminant → echinocandin, remove lines, ophthalmology assessment
- remove unnecessary lines and catheters as part of source control