Recognise and manage maternal sepsis

In one line

Maternal sepsis is life-threatening organ dysfunction caused by a dysregulated host response to infection in pregnancy, labour, the puerperium or after abortion; the mother who dies is almost always the one in whom the diagnosis was made late, because the physiology of pregnancy hides the early signs — and survival turns on two things done within the first hour, broad-spectrum antibiotics and a relentless hunt for the source to control.

Mechanism & pathophysiology

Sepsis is not severe infection and it is not bacteraemia. The Sepsis-3 consensus reframed it as life-threatening organ dysfunction due to a dysregulated host response to infection — the lethal element is the host response, not the organism. An invading pathogen, recognised through pattern-recognition receptors, triggers a coordinated release of pro-inflammatory cytokines (TNF-α, IL-1, IL-6) that is appropriate when local and catastrophic when systemic. The endothelium is the organ that fails first and worst: glycocalyx shedding, loss of barrier integrity and widespread microvascular leak produce the clinical triad of vasoplegia, maldistribution of flow and capillary leak. Nitric-oxide-driven vasodilatation drops systemic vascular resistance; arterial and venous tone collapse; fluid escapes the intravascular space. The result is a paradox — a patient who is volume-deplete intravascularly yet oedematous, with tissue hypoperfusion despite a normal or high cardiac output. Mitochondrial dysfunction means oxygen that does reach the cell cannot be used; lactate rises as a marker of that cellular distress, not simply of anaerobic debt. Coagulation is activated in parallel, consuming platelets and clotting factors and producing the microthrombi of disseminated intravascular coagulation that occlude the very capillaries already starved by maldistribution.

The immune derangement is not a simple cytokine storm but a two-phase dysregulation, and the distinction has bedside consequences. The early hyperinflammatory phase is followed — sometimes overlapping — by a compensatory anti-inflammatory and immunoparalytic state, in which monocyte HLA-DR expression falls, lymphocytes are lost to apoptosis, and the patient becomes susceptible to secondary and nosocomial infection. This is why a septic mother who appears to turn the corner can deteriorate again days later from a new organism, and why immunosuppression — including HIV — compounds an already maladaptive host response. The complement and contact pathways are activated alongside cytokines, generating the anaphylatoxins that amplify vasodilatation and capillary leak. Tissue factor exposed on damaged endothelium initiates coagulation, while the natural anticoagulant systems (protein C, antithrombin, tissue-factor-pathway inhibitor) are consumed and downregulated; the net pro-thrombotic shift, uncoupled from effective anticoagulation, is the molecular basis of the DIC that clinically presents as bleeding from puncture sites alongside microvascular thrombosis. Understanding that coagulation and inflammation are the same dysregulated process is what makes the failure of activated protein C as a therapy intelligible — restoring one node of a system-wide collapse was never going to rescue the whole.

Septic shock is the subset in which these circulatory, cellular and metabolic derangements are profound enough to carry a substantially higher mortality — operationally, a patient who needs a vasopressor to hold a mean arterial pressure of 65 mmHg and has a lactate above 2 mmol/L despite adequate fluid resuscitation. That two-part definition matters: a low blood pressure that corrects with fluid is not septic shock, and a persistently raised lactate after resuscitation signals shock even when the pressure has been propped up. The lactate deserves a mechanistic caveat the consultant should be able to defend: in sepsis it is only partly a marker of anaerobic, hypoperfusion-driven metabolism — adrenergic stimulation of aerobic glycolysis and impaired hepatic clearance both contribute — so a raised lactate is a marker of severity and a target for clearance, not a literal oxygen-debt gauge to be chased with fluid alone.

Pregnancy is the reason this presents late and progresses fast. The adaptive physiology of a normal pregnancy is, point for point, a near-perfect camouflage for early sepsis. Resting heart rate is already raised by 10–20 beats per minute, so a compensatory tachycardia looks like pregnancy. Systemic vascular resistance and blood pressure are physiologically lower, so the relative hypotension that should alarm reads as normal for a gravid woman. The white cell count is elevated in healthy pregnancy and rises further in labour, so leucocytosis loses its discriminatory value. Tidal volume and minute ventilation are already high, so a rising respiratory rate — the single most sensitive early sign of deterioration — is easily dismissed. The gravid uterus splints the diaphragm and shifts abdominal landmarks, blurring peritoneal signs. Onto this baseline the mother brings a relatively immunotolerant state that permits the fetus and, in doing so, blunts containment of infection. The net effect is a wider, faster trajectory from infection to organ failure on a background that disguises every early warning. The young, previously fit obstetric patient also compensates impressively — until she does not, decompensating precipitously from what looked like a stable picture an hour earlier.

The sources cluster by timing and anatomy. Chorioamnionitis in the antenatal and intrapartum period, ascending from the genital tract, often with prolonged rupture of membranes. Endometritis is the dominant puerperal source — retained products and an open, devascularised placental bed make the postpartum uterus a culture medium, and it is the commonest infective focus after both vaginal and caesarean delivery. Septic abortion, whether after unsafe termination or incomplete miscarriage, drives some of the most fulminant disease, with retained tissue as an uncontrolled source. Beyond the genital tract sit the non-obstetric sources that kill exactly because they are not looked for: pyelonephritis (pregnancy's ureteric dilatation and stasis make ascending urinary infection common and severe), pneumonia and influenza (to which pregnant women are unusually susceptible and from which they deteriorate quickly), wound and episiotomy infection including necrotising soft-tissue infection, and mastitis progressing to breast abscess. The organism that disproportionately kills is group A streptococcus (Streptococcus pyogenes): it produces superantigen-driven toxic shock, necrotising endomyometritis and fasciitis, and a deceptively well-looking woman can be peri-arrest within hours; it is the single most frequent organism in fatal maternal sepsis, often community-acquired from a child with a sore throat. Escherichia coli and other Gram-negatives dominate urinary and genital sources; polymicrobial pictures are common in retained-tissue and abscess disease.

South Africa adds its own drivers, and they reshape the differential. HIV is the dominant modifier — both a direct cause of overwhelming opportunistic infection and a state that worsens the course and completion of treatment of any sepsis. Disseminated tuberculosis is a specifically South African trap: it presents as an indolent, then suddenly fulminant, septic picture, is frequently extrapulmonary in HIV co-infection, and is missed when the team is anchored on a genital-tract source. Influenza and bacterial pneumonia carry disproportionate mortality. The implication is concrete: in this population a septic mother without an obvious genital source is TB or HIV-related sepsis until proven otherwise, and the HIV status is part of the assessment, not an afterthought.

The Saving Mothers confidential enquiries make the epidemiology unambiguous for SA practice: non-pregnancy-related infection — overwhelmingly AIDS-related — has been the leading cause of maternal death since 1999, and pregnancy-related sepsis sits among the top direct causes alongside hypertension and haemorrhage. The maternal mortality ratio in HIV-infected women runs roughly an order of magnitude above that in uninfected women, and the great majority of NPRI deaths occur in HIV-positive women. The advanced-HIV mother does not mount the febrile, leucocytotic response that flags sepsis in the immunocompetent; she may present with a normal temperature and a normal white count while disseminated TB, cryptococcal disease, pneumocystis or bacterial pneumonia smoulders. A blunted inflammatory response in an unwell, low-CD4 mother is therefore a reason for more concern, not reassurance — the absence of the classic signature reflects an immune system too depleted to produce it.

Source by source — the reasoning the differential demands

Naming the focus changes both the empirical antibiotic and the source-control move, so the differential is structured, not a memorised list:

• Endometritis — the dominant puerperal source. Offensive lochia, a tender bulky subinvoluted uterus, fever and a raised white count days after delivery; the question is always whether there are retained products, because antibiotics will not clear an infected retained focus and ultrasound that shows retained tissue mandates evacuation. Post-caesarean endometritis is more often polymicrobial and may overlie a wound or pelvic collection.
• Chorioamnionitis — maternal fever with uterine tenderness, fetal and maternal tachycardia, offensive liquor, often after prolonged membrane rupture. Here the uterus is the source while the fetus is in it, so resolution requires delivery.
• Septic abortion — fever, offensive discharge, lower abdominal pain and a history of recent (often unsafe) termination or incomplete miscarriage; the most fulminant maternal sepsis arises here, and uterine evacuation is the resuscitation.
• Urinary (pyelonephritis) — flank pain, dysuria, vomiting and rigors; pregnancy's ureteric dilatation makes ascending infection common and severe, and Gram-negative bacteraemia is the danger. The "source control" is occasionally drainage of an obstructed, infected system.
• Necrotising soft-tissue infection — pain wildly out of proportion to a wound or episiotomy, rapidly advancing erythema, crepitus, dusky skin and systemic toxicity. This is the surgical emergency hidden among ordinary wound infections; it needs immediate radical debridement, not a course of antibiotics.
• Respiratory — pneumonia and influenza, to which the pregnant woman is unusually vulnerable, with rapid progression to hypoxia and ARDS; in SA, TB pneumonia and disseminated TB sit squarely in this differential.
• Mastitis / breast abscess — a tender, erythematous, wedge-shaped breast segment progressing, if untreated or if an abscess forms, to systemic sepsis; the abscess needs drainage.

Assessment

Recognition is the whole game, and it depends on suspicion rather than on any single number, because every individual vital sign is unreliable in pregnancy. The discipline is to treat any unwell pregnant or recently delivered woman with a plausible infective focus as septic until proven otherwise and to act before the diagnosis is certain.

• A structured early-warning trigger, not clinician gestalt alone. The Modified Early Obstetric Warning Score (MEOWS) is the standard SA tool — a colour-banded chart tracking respiratory rate, oxygen saturation, temperature, blood pressure, heart rate, conscious level and urine output, designed precisely because isolated obstetric observations mislead. A single red or two amber triggers mandate senior review. Its value is that it forces serial, charted observation of trends, catching the slow drift the eye misses.
• The single most useful early sign is the respiratory rate. A rising respiratory rate (≥20–24/min) precedes hypotension and is the most sensitive marker of deterioration; it is also the observation most often unrecorded. Tachypnoea in a febrile postpartum woman is a reason to escalate, not to reassure.
• The limits of qSOFA — and why MEOWS/NEWS2 are the screen. The Sepsis-3 bedside screen, qSOFA (respiratory rate ≥22, altered mentation, systolic BP ≤100), is specific but insensitive — in its derivation only about a quarter of infected patients who went on to do badly scored ≥2, so it misses early sepsis. The Surviving Sepsis Campaign therefore makes a strong recommendation against qSOFA as a single screening tool, favouring SIRS, NEWS/NEWS2 or MEWS-type scores together with lactate. In obstetrics the case is stronger still: qSOFA and the full SOFA score were derived in non-pregnant adults, and the lower baseline blood pressure and higher respiratory rate of pregnancy degrade their thresholds further. Screen with an obstetric early-warning chart (MEOWS) — or NEWS2 where MEOWS is unavailable — plus lactate, not qSOFA; and never let a normal qSOFA override clinical concern.
• Lactate is the key biochemical severity marker. A venous or arterial lactate quantifies tissue hypoperfusion when the blood pressure still looks acceptable; a level above 2 mmol/L flags hypoperfusion and above 4 mmol/L signals high mortality and mandates aggressive resuscitation. Serial lactate (clearance) tracks response.
• Cultures before antibiotics — but never delay antibiotics to obtain them. Take blood cultures (ideally two sets) and site-directed cultures — high vaginal and endocervical swabs, urine, sputum, wound, throat (group A strep often colonises before it invades), and placental swabs — before the first antibiotic dose only if they can be obtained within minutes. If sampling would delay antibiotics, give the antibiotic and culture around it. Add a full blood count, U&E and creatinine, liver function, CRP, coagulation profile and an HIV test in every patient.
• Find the source with imaging. Pelvic ultrasound for retained products or a tubo-ovarian/pelvic collection; chest radiograph for pneumonia; CT for an abscess or to define necrotising soft-tissue infection. The diagnostic question is not only "is she septic" but "what is the source and can I control it", because antibiotics alone never cure an undrained collection or retained tissue.
• Stratify severity. Organ dysfunction — oliguria, a rising creatinine, deranged liver enzymes, thrombocytopenia and coagulopathy, confusion, hypoxia or a lactate above 4 — defines the woman who needs critical-care-level support, not ward management. The pregnant patient who has decompensated has little reserve left.

Management

The plan organises immediate → ongoing → long-term, and the immediate hour is where mortality is decided.

Immediate (the first hour). The actionable bundle in SA practice is the Sepsis Six (UK Sepsis Trust), a deliberately simple set of six tasks any front-line clinician can deliver within an hour, drawn from the Surviving Sepsis Campaign: high-flow oxygen, blood and site cultures, intravenous broad-spectrum antibiotics, intravenous fluids, serum lactate, and urine-output monitoring (a catheter). Three of these carry the survival benefit and bear emphasis:

• Antibiotics within one hour. Broad-spectrum empirical cover, started within 60 minutes of recognising sepsis, is the intervention most strongly tied to survival. The SA empirical choice covers genital-tract and Gram-negative organisms and must cover group A streptococcus and anaerobes — a regimen along the lines of a broad β-lactam (e.g. amoxicillin–clavulanate or a cephalosporin) plus metronidazole, with the addition of an aminoglycoside (gentamicin) for severe or Gram-negative-predominant disease, escalated per local antibiogram and NDoH/EML availability; clindamycin is added in suspected streptococcal toxic shock or necrotising infection for its anti-toxin (anti-exotoxin) effect. The principle to defend is broad early, then narrow on culture and sensitivity.
• Balanced-crystalloid resuscitation, guided by response. For hypoperfusion or shock, an initial fluid challenge using a balanced crystalloid (Ringer's lactate / Plasma-Lyte rather than large volumes of 0.9% saline) — the Surviving Sepsis Campaign suggests up to 30 mL/kg in the first three hours — then reassess dynamically. Fluid is titrated to response (blood pressure, lactate clearance, urine output), not poured in by protocol: the pregnant patient with capillary leak and a splinted diaphragm tolerates over-resuscitation badly, and the modern evidence is that fluids are not innocuous.
• Lactate and its clearance anchor the resuscitation target — a falling lactate is the signal that perfusion is recovering.

Source control is the curative step and the one most often delayed. Antibiotics buy time; they do not cure an uncontrolled source. The septic mother needs the focus removed or drained without delay once she is being resuscitated:

• Evacuation of retained products in septic abortion or postpartum endometritis with retained tissue — the single most important intervention in those patients, and one not to defer for "stabilisation" that resuscitation alone will never achieve while the source remains.
• Delivery where the uterus is the source antenatally (chorioamnionitis with a deteriorating mother): expediting delivery is source control, and a septic intrauterine focus is a reason to deliver, not to temporise — though the timing and mode are individualised to maternal stability and fetal condition.
• Drainage of a pelvic or tubo-ovarian abscess, debridement of a wound or, urgently, of necrotising soft-tissue infection (which is a surgical emergency — radical early debridement saves life and antibiotics alone do not).
• Hysterectomy as a last resort for an unsalvageable, gangrenous or clostridial uterine source in a woman who is dying of it.

Ongoing (the deteriorating mother). When fluid resuscitation does not restore perfusion, start noradrenaline (norepinephrine) as the first-line vasopressor to hold a mean arterial pressure of ~65 mmHg, ideally through central access but not delayed for it in extremis. This is the threshold at which care belongs in a high-care/HDU or ICU, with senior obstetric, anaesthetic/critical-care and microbiology input — the multidisciplinary escalation is itself part of the standard. Support failing organs (oxygen and, where needed, ventilatory support for ARDS; renal replacement for severe acute kidney injury), correct coagulopathy in DIC, and re-image to confirm the source is truly controlled if the patient is not improving — failure to respond usually means inadequate source control or the wrong antibiotic, not insufficient fluid. Manage the HIV-positive or TB-co-infected mother with the explicit recognition that she carries less reserve, may have an unexpected source (disseminated TB), and needs her antiretroviral and TB therapy optimised alongside the sepsis care.

The fetus follows the mother. Resuscitate the mother first; uteroplacental perfusion and fetal oxygenation improve as maternal physiology is restored. Continuous fetal monitoring is appropriate in a viable pregnancy, but a non-reassuring trace in an unstable septic mother is usually a reflection of maternal compromise to be corrected by maternal resuscitation, not an automatic call to deliver — and delivery into ongoing maternal instability is hazardous unless the uterus is the source. The exception is the uterus-as-source scenario, where delivery is the source control that saves both.

Long-term. Survivors of severe sepsis carry a real burden — post-intensive-care syndrome, the sequelae of any organ injury, and the psychological aftermath of a near-death event around the birth. Debrief the woman and family, document the episode and its likely cause for future pregnancies (a group A streptococcal puerperal sepsis or a septic abortion carries counselling implications), and ensure follow-up of any residual organ dysfunction. In SA, link to district-level care so the tertiary unit is not the only safety net.

Guidelines compared

The evidence & the controversy

The modern management of sepsis is in some ways a story of de-escalation, and the obstetric clinician has to hold both the things that genuinely save lives and the things that, on testing, did not. What survived rigorous trial is unglamorous and time-critical: early antibiotics and source control. Everything elaborate built on top of it — early goal-directed protocolised resuscitation, activated protein C, fixed large-volume fluid loading — has either failed or been substantially qualified, and a candidate who recites the discredited bundle as current is making the error the trials corrected.

The antibiotic-timing evidence is the foundation. Kumar's retrospective cohort in septic shock put a number on delay — each hour without effective antimicrobial therapy over the first six hours cost a mean 7.6% in survival — and although its absolute figures reflect an older, slower era, the gradient is real and reproducible. Seymour's analysis of mandated emergency sepsis care confirmed it prospectively and, crucially, dissected the components: faster antibiotics and faster completion of the three-hour bundle lowered mortality, but faster delivery of a fluid bolus did not. That dissociation is the heart of the modern argument — speed to antibiotics and source control is what cures; speed and volume of fluid is not a virtue in itself.

The fluid story is where current practice diverges from the protocol many were taught. Early goal-directed therapy, once dogma, was overturned by the pooled patient-level meta-analysis of ProCESS, ARISE and ProMISe (the PRISM collaboration), which found no mortality benefit over usual care and higher cost — and no benefit even in the sickest. The composition question moved towards balanced crystalloid over saline (the SMART trial reduced major adverse kidney events; the larger BaSICS and PLUS trials were neutral on mortality but consistent with avoiding chloride-rich saline), and the volume question was reopened by CLOVERS, which found no difference between a restrictive (earlier vasopressor) and a liberal fluid strategy. The defensible synthesis for the septic mother — already prone to capillary leak and pulmonary oedema — is balanced crystalloid, titrated to response and lactate clearance rather than to a protocol number, with noradrenaline introduced early rather than chasing perfusion with ever-larger fluid volumes.

A genuinely current controversy worth holding is antimicrobial stewardship versus the one-hour mandate. The Surviving Sepsis Campaign's insistence on antibiotics within an hour of suspected sepsis is challenged by those who argue it drives over-treatment of patients who turn out not to be septic and fuels resistance — a live concern in SA, where antimicrobial resistance is a national-priority problem. The reconciliation in obstetrics is that the maternal-sepsis population is high-stakes and the camouflaging physiology makes under-recognition the larger danger, so early broad cover is justified provided it is paired with disciplined de-escalation on culture results within 48–72 hours. The skill being examined is treating fast and stopping/narrowing thoughtfully, not blanket broad-spectrum therapy.

Landmark trials & key evidence

A worked figure makes the antibiotic point concrete: if each hour of delay costs on the order of 7.6% in survival, a mother whose effective antibiotic is delayed three hours from recognition has lost roughly 20–23 percentage points of survival to the clock alone — which is why the one-hour target, in a population whose physiology already hides the diagnosis, is treated as a hard standard rather than an aspiration.

Exam traps & red flags

• Anchoring on "normal for pregnancy". A tachycardic, tachypnoeic, borderline-hypotensive febrile woman is septic until proven otherwise; reading her vitals as physiological pregnancy is the commonest fatal error. The respiratory rate is the sign that betrays her — record it and act on it.
• Treating qSOFA/SOFA as the obstetric screen. They were validated in non-pregnant adults; a normal qSOFA does not exclude maternal sepsis. Use MEOWS and clinical suspicion.
• Antibiotics alone, source ignored. The patient who is not improving despite correct antibiotics almost always has an uncontrolled source — retained products, an undrained abscess, necrotising infection — or the wrong drug. Re-image and control the source; do not simply give more fluid.
• Delaying antibiotics to complete the cultures or the work-up. Cultures before antibiotics only if they take minutes; otherwise give the antibiotic and culture around it. Every hour costs survival.
• Missing group A streptococcus and its toxic shock. A deceptively well-looking woman with rapidly progressive disease, often after a child's sore throat, can be peri-arrest within hours; add clindamycin for anti-toxin effect and consider necrotising fasciitis early.
• Missing the necrotising soft-tissue infection. Pain out of proportion, rapidly spreading erythema, crepitus or systemic toxicity around a wound or episiotomy is a surgical emergency needing immediate radical debridement, not a course of antibiotics and review.
• Over-resuscitating with fluid. Fixed large-volume crystalloid into a leaking, splinted obstetric circulation causes pulmonary oedema; titrate to response and reach for noradrenaline early in fluid-refractory shock.
• Forgetting HIV and TB in the SA mother. A septic mother without an obvious genital source may have disseminated TB or an HIV-related opportunistic infection; test HIV in every patient and keep TB in the differential.
• Deferring source control to "stabilise" first. In septic abortion or retained-products endometritis, evacuation is the resuscitation; stabilisation rarely succeeds while the source remains.
• Quoting discredited care as current. Presenting early goal-directed therapy or activated protein C as standard, or the old GTG 64a/64b as the current guideline, reads as out of date.

Evidence anchors

• Sepsis-3 — Singer et al., Third International Consensus Definitions for Sepsis and Septic Shock, JAMA 2016;315(8):801–810
• Surviving Sepsis Campaign: International Guidelines 2026 — Prescott/Evans et al., Crit Care Med 2026 (the current guideline; updates the 2021 edition)
• RCOG Green-top Guideline No. 64 — Lissauer et al., Identification and Management of Maternal Sepsis during and following Pregnancy, BJOG 2025;132(4):e61–e85
• Kumar et al. — duration of hypotension before effective antimicrobial therapy, Crit Care Med 2006;34(6):1589–1596
• Seymour et al. — Time to Treatment and Mortality during Mandated Emergency Care for Sepsis, N Engl J Med 2017;376(23):2235–2244
• PRISM (Rowan et al.) — Early, Goal-Directed Therapy for Septic Shock: A Patient-Level Meta-Analysis, N Engl J Med 2017;376(23):2223–2234
• SMART (Semler et al.) — Balanced Crystalloids versus Saline in Critically Ill Adults, N Engl J Med 2018;378(9):829–839
• CLOVERS — Early Restrictive or Liberal Fluid Management for Sepsis-Induced Hypotension, N Engl J Med 2023;388(6):499–510
• PROWESS-SHOCK (Ranieri et al.) — Drotrecogin Alfa (Activated) in Adults with Septic Shock, N Engl J Med 2012;366(22):2055–2064
• Global Maternal Sepsis Study (GLOSS) — Bonet et al., Lancet Glob Health 2020;8:e661–e671; the WHO Statement on Maternal Sepsis defines it as life-threatening organ dysfunction from infection in pregnancy, childbirth, post-abortion or the postpartum period.
• Sepsis Six — Daniels et al., Emerg Med J 2011;28(6):507–512 (UK Sepsis Trust bundle)
• South Africa NDoH / NCCEMD Saving Mothers — non-pregnancy-related infection (mainly AIDS) the leading cause of maternal death since 1999; pregnancy-related sepsis a top-five direct cause; HIV testing mandatory.