Diagnose and manage puerperal sepsis and complications

In one line

Puerperal sepsis is the genital-tract face of maternal sepsis in the six weeks after birth — the historical "childbed fever", now read within the source-agnostic Sepsis-3 framework set out in maternal-sepsis, where sepsis is organ dysfunction from a dysregulated host response to infection rather than a site or a severity of fever. The consultant task is to recognise the deteriorating mother early — before the observations look frightening — name the source (the genital tract is only the commonest one), and combine prompt broad-spectrum antibiotics with the one intervention antibiotics cannot substitute for: source control.

Mechanism & pathophysiology

The puerperium is, in effect, a fresh wound. After delivery the placental bed is a raw, vascular surface the size of a palm, the cervix is patulous, and the lochia draining through it is an excellent culture medium; the involuting uterus, the breast establishing lactation, any caesarean or perineal wound, and a bladder emptied through a recently catheterised urethra are all portals. Normal involution is the background against which infection is read: the uterus descends roughly a fingerbreadth a day and is no longer palpable abdominally by about day 10–14, and lochia moves predictably from rubra (red, 3–4 days) to serosa (pink-brown) to alba (yellow-white) over a few weeks. A uterus that stays high and tender, lochia that turns heavy, offensive or reverts to red, and a fever that will not settle are involution going wrong. The Intermediate groundwork on the normal puerperium and its complications is assumed here; the focus is the consultant-level reasoning that turns a fever into a diagnosis and a plan.

Most puerperal sepsis is polymicrobial and ascends from the lower genital tract. Endometritis is the commonest source — colonisation of the decidua and myometrium by vaginal flora, dramatically more likely after caesarean section (the single biggest risk factor), and worse when products of conception are retained, because devitalised tissue is a protected nidus that antibiotics penetrate poorly. The organisms are the predictable mix of a gut-and-genital field: Escherichia coli and other Gram-negative aerobes, anaerobes (Bacteroides, peptostreptococci), enterococci, and staphylococci including community-acquired MRSA in wounds. Group B streptococcus and genital mycoplasmas contribute.

The organism that dominates the consultant's risk calculus is out of proportion to its frequency. Group A streptococcus (Streptococcus pyogenes) is the pathogen of "childbed fever" — the disease Semmelweis halted with hand-washing — and it behaves differently from the polymicrobial ascending infections. It is exotoxin-driven: streptococcal pyrogenic exotoxins act as superantigens, triggering massive non-specific T-cell activation and a cytokine storm that produces streptococcal toxic shock syndrome and tissue-destroying necrotising fasciitis. Invasive GAS is roughly twenty times more common in postpartum women than in the non-pregnant, can be acquired from the patient's own throat or from a household contact with a sore throat, and kills fast: once shock is established, mortality is of the order of 30–50%. The teaching point that falls out of the mechanism is that GAS can produce profound illness — pain out of proportion to signs, then hypotension — while the local findings are still unimpressive, because the damage is toxin-mediated and deep rather than a visible collection.

The other sources each have their own mechanism. Wound infection after caesarean or of a perineal repair follows the general surgical logic of contamination, haematoma and devitalised tissue; dehiscence is its structural consequence, and the dreaded version is fascial dehiscence with burst abdomen. Mastitis is milk stasis plus a skin organism (usually Staphylococcus aureus entering through a cracked nipple) — an inflammatory cellulitis of a lactating segment that becomes a breast abscess when a walled-off collection forms. Urinary tract infection exploits catheterisation and incomplete bladder emptying, and ascends to pyelonephritis in a ureteric system still dilated from pregnancy. Septic pelvic thrombophlebitis is the subtle one: bacterial seeding of the venous endothelium of the ovarian or deep pelvic veins, where thrombus and infection sustain each other and an antibiotic-resistant "enigmatic fever" persists despite an apparently treated source. Understanding that infection and clot are co-dependent in SPT is what justifies adding anticoagulation to antibiotics — the single counter-intuitive move in this whole topic.

A point of definition keeps the differential honest. The old puerperal pyrexia label — a temperature ≥38°C on any two of the first ten days after delivery (excluding the first 24 hours) — is a screening flag, not a diagnosis, and the modern habit is to treat any sustained postpartum fever as a sepsis screen pointing at a source rather than to wait for it to "declare itself". The timing of onset is a clue: GAS and other streptococcal disease tends to be early and fulminant (within the first day or two), classic ascending endometritis a few days in, wound infection towards the end of the first week, breast abscess as lactation establishes, and septic pelvic thrombophlebitis later still, as the fever that will not resolve.

Two host factors reshape this picture in South Africa. HIV — especially advanced disease with a low CD4 count — blunts the inflammatory response, broadens the organism list, and is associated with more severe and atypical puerperal infection; co-existing tuberculosis must be considered in the puerpera with persistent fever, weight loss or an atypical course, because genital and disseminated TB are part of the SA differential in a way they are not elsewhere. The general principle is that the immunosuppressed host both mounts a quieter signal and tolerates sepsis worse, so the threshold to investigate and to escalate falls. The severe end of any of these sources converges on the same final common pathway — the systemic inflammatory cascade, endothelial injury, capillary leak, and, in the worst cases, disseminated intravascular coagulation and multi-organ dysfunction — which is why the recognition and resuscitation are shared across sources even though the definitive treatment is source-specific.

Assessment

The first task is to recognise the deteriorating puerpera, and the trap is that a young, previously fit woman compensates well and then collapses. Numbers drift before they crash; tachycardia and tachypnoea precede hypotension, and a fall in urine output and a rising lactate are early. Tachypnoea is the single most under-weighted sign — a respiratory rate climbing above 20–24 is often the first marker of compensated metabolic acidosis and is exactly the observation a postnatal ward forgets to count. A structured MEOWS (Modified Early Obstetric Warning System) chart is the system safeguard — it is designed to catch the trending-but-not-yet-alarming observation set that a busy postnatal ward otherwise normalises, and it accounts for the fact that the normal ranges of pulse, blood pressure and respiratory rate shift in the puerperium so that a generic adult early-warning score misreads them. A single red MEOWS trigger, or two yellow triggers, in a febrile puerpera is a reason to escalate, not to recheck in an hour. Lactate carries the same weight here as in any sepsis: a level above 2 mmol/L signals hypoperfusion and above 4 mmol/L marks the patient who needs aggressive resuscitation and critical-care review.

• History: mode of delivery (caesarean and operative vaginal birth raise risk; ANODE-era practice now gives a single prophylactic dose after operative vaginal birth), prolonged rupture of membranes, retained products or manual removal of placenta, the character and odour of lochia, breast symptoms, urinary symptoms, wound pain or discharge, and a sore throat in the patient or a household child (the GAS clue). Pain that seems disproportionate to the wound, or that is rapidly worsening, is the necrotising-fasciitis history until disproven.
• Examination: temperature, the full observation set fed into MEOWS, and a deliberate search for the source — uterine height and tenderness, the caesarean and perineal wounds (erythema, induration, discharge, separation, and crucially crepitus, dusky skin or skin anaesthesia that signal necrotising infection), both breasts (a tender, red, wedge-shaped segment is mastitis; a fluctuant mass is abscess), the renal angles and suprapubic region, and the calves and chest because postpartum venous thromboembolism is the differential that masquerades as sepsis. Examine the throat.
• Investigations: blood cultures before antibiotics (but do not let drawing them delay the first dose), full blood count, CRP, U&E and creatinine, lactate and venous blood gas, and clotting if the picture is severe. A high vaginal and any wound swab, a midstream or catheter urine, and — when GAS or invasive infection is suspected — a throat swab. The bedside numbers that matter are the sepsis-screen physiology: lactate, urine output, and the trend.
• Imaging by question. Pelvic ultrasound to look for retained products (an echogenic endometrial mass with vascularity) or a pelvic or wound collection. CT is the workhorse for the deep and the occult: it shows the gas and fascial-plane fluid of necrotising fasciitis, the thrombosed enlarged ovarian or pelvic veins of septic pelvic thrombophlebitis, and a deep pelvic abscess the ultrasound misses. SPT is in large part a diagnosis reached when a fever persists despite adequate antibiotics and source control, with CT lending support rather than the other way round.

Interpretation matters as much as the panel. The puerperium itself produces a physiological leucocytosis and a raised CRP for days after a normal delivery, so a modestly elevated white count and CRP are not in themselves proof of infection; it is the trend, the magnitude, and the clinical picture that count, and a normal white count never reassures in suspected GAS, where the count can be low or even neutropenic in fulminant disease. A markedly raised CRP with a soft-tissue picture, a creatine kinase that is rising, hyponatraemia and a falling platelet count together point at a necrotising soft-tissue infection. Falling platelets, a prolonged clotting profile and a rising D-dimer signal evolving DIC and a patient who has moved from sepsis towards septic shock.

The diagnostic discipline is to name the source. "Puerperal sepsis" is a syndrome, not an endpoint; the management is entirely different for endometritis with retained products (evacuate), a breast abscess (drain), a necrotising soft-tissue infection (operate now) and septic pelvic thrombophlebitis (anticoagulate). The overlap with the wider syndrome of maternal sepsis is real — the recognition physiology, the Sepsis Six and the resuscitation are shared — but the puerperal sources are specific and each carries its own definitive step. Where no source is obvious, the search widens to the non-genital causes a postnatal fever can hide: chest (aspiration or hospital-acquired pneumonia after anaesthesia), the cannula or epidural site, Clostridioides difficile after broad antibiotics, and, in the SA setting, malaria, HIV-related opportunistic infection and tuberculosis.

Management

Organise it immediate → ongoing → long-term, and let the source dictate the definitive move. The immediate phase is identical regardless of source: it is sepsis resuscitation. The Sepsis Six within the first hour — high-flow oxygen, blood cultures, intravenous broad-spectrum antibiotics, intravenous fluid, serum lactate, and hourly urine output (the "take three, give three") — is the floor, escalated to the wider sepsis bundle and to critical care if organ dysfunction appears. Fluid resuscitation is a balanced crystalloid bolus (around 30 mL/kg in the hypoperfused or hypotensive patient) with reassessment after each bolus rather than open-ended loading, because the leaky septic vasculature and the post-partum heart tolerate over-transfusion poorly and pulmonary oedema is a real iatrogenic risk; the patient who stays hypotensive after appropriate fluid needs vasopressors and a critical-care environment, not more litres. The first antibiotic dose should be in within the first hour — every hour of delay in established septic shock costs survival, and this is the time-critical step the SA Saving Mothers analyses repeatedly find was missed.

Empirical antibiotics are broad and started before the organism is known; the SA regional default is a combination covering Gram-negatives, anaerobes and streptococci, typically ampicillin (or amoxicillin) plus gentamicin plus metronidazole, or co-amoxiclav plus gentamicin, with the threshold for adding clindamycin (for its anti-exotoxin effect) low whenever invasive GAS or toxic shock is in the differential. A single severe penicillin allergy redirects to a regimen such as clindamycin plus gentamicin, which conveniently is also the most evidence-supported endometritis combination. Gentamicin dosing is weight-based and renal-function-aware, and in the woman with sepsis-related acute kidney injury its level needs watching. The choices are tuned to the source below; the principle is early, broad, and intravenous, then narrowed on culture — and de-escalated, not simply continued, once the organism and sensitivities are back.

Endometritis. The most evidence-supported regimen is clindamycin plus gentamicin, which the Cochrane review shows produces fewer treatment failures than penicillin- or cephalosporin-based regimens — the difference being good activity against penicillin-resistant anaerobes. Once-daily gentamicin is at least as effective as thrice-daily dosing. The other practical Cochrane finding is that for uncomplicated endometritis that has responded to intravenous therapy, continuing oral antibiotics after the patient is afebrile adds no benefit — stop the IV at around 24–48 hours of apyrexia and do not reflexively prescribe a take-home course. The definitive step when products are retained is evacuation: antibiotics alone will not clear an infected nidus, and a gentle ultrasound-guided evacuation of an infected uterus carries a real perforation risk in soft postpartum tissue, so it is timed once antibiotics have begun and performed by an experienced operator.

Mastitis and breast abscess. The counter-intuitive instruction that wins this question is keep the milk moving. Effective and frequent emptying of the affected breast — by the baby or by expressing — is treatment, not a contraindication; stopping breastfeeding worsens stasis and promotes abscess formation. Antibiotics (flucloxacillin or cloxacillin against S. aureus) are added when systemic features or no improvement over 12–24 hours suggest infection rather than simple engorgement. A breast abscess needs drainage. The SA-authored Cochrane review compared needle aspiration (ideally ultrasound-guided) with incision and drainage: aspiration is less invasive and can be faster to resolve, but carries a substantially higher treatment-failure rate, so I&D remains the more reliably definitive option, with aspiration reasonable for smaller, accessible collections. Breastfeeding continues from both breasts throughout.

Wound infection and dehiscence. An infected caesarean or perineal wound is managed on surgical principles: open it, drain the collection, debride devitalised tissue, and swab. Antibiotic cover must include staphylococci, with MRSA cover where local prevalence demands it. A wound left to heal by secondary intention, negative-pressure wound therapy (VAC) to accelerate granulation, and delayed secondary closure are the reconstructive ladder. The decision that must not be missed is distinguishing a contained wound infection from a spreading soft-tissue infection — the moment the cellulitis is advancing by the hour, the pain is disproportionate, or the skin is dusky or anaesthetic, this is no longer a wound infection.

Necrotising fasciitis is the emergency of this topic, and the management is surgical and immediate. Antibiotics, fluid and critical care are supportive; they do not treat the disease. The treatment is urgent, radical surgical debridement of all non-viable tissue, repeated as often as the infection demands ("second look" within 24 hours is routine), because survival is determined by how early and how complete the first debridement is. Antibiotics are broad-spectrum plus high-dose clindamycin for its ribosomal suppression of toxin production (the Eagle effect — beta-lactams are less effective against the high-inoculum, slowly-dividing organisms), and intravenous immunoglobulin is added in streptococcal toxic shock to neutralise circulating exotoxin. Delay to theatre is the commonest avoidable cause of death; a suspicious wound goes to theatre for exploration even when imaging is equivocal, because waiting for a confirmatory scan costs tissue and life.

Septic pelvic thrombophlebitis is the diagnosis to reach when a puerpera has persistent swinging fevers despite adequate antibiotics and no other source, sometimes with a palpable tender cord in the adnexa. The management is to continue the broad-spectrum antibiotics and add therapeutic anticoagulation with heparin or low-molecular-weight heparin; the classic teaching is a prompt defervescence on heparin, which is itself near-diagnostic. Warfarin is avoided in the acute septic phase, and the optimal duration of anticoagulation is genuinely unsettled — a short course continued for several days of apyrexia is usual.

Escalation and the sepsis pathway. Throughout, escalation follows organ dysfunction, not a single number: rising lactate, falling urine output, hypotension unresponsive to a fluid bolus, confusion, or a deteriorating MEOWS score mandate senior obstetric, anaesthetic and critical-care involvement and consideration of a high-care or ICU bed. In the SA district setting this is also a referral decision — broad antibiotics, resuscitation and a phone call to the regional or tertiary unit are started simultaneously, and a patient with necrotising fasciitis or refractory shock is stabilised and transferred without waiting for definitive imaging. The realities that the Saving Mothers analyses keep naming are practical: a district hospital may have no ICU bed, no surgeon comfortable with extensive debridement, and a transfer that takes hours — so the consultant plan is to start everything that can be started locally (oxygen, fluids, the first antibiotic dose, source control where feasible) while the referral is in motion, rather than treating transfer as the first step.

The invasive-GAS case is also a public-health event. A confirmed invasive group A streptococcal puerperal infection should prompt infection-control involvement, isolation of the patient, and consideration of screening and chemoprophylaxis for close contacts and, in clustered cases, of healthcare-worker carriage — because the organism transmits and outbreaks on maternity units, though rare, are devastating. Treating the index case without closing the transmission loop misses half the problem.

Long-term. Most recover fully. The longer arc is counselling on future pregnancy risk (a woman with caesarean wound problems or retained products has implications for subsequent delivery), support to continue or re-establish breastfeeding after mastitis or abscess, debriefing after a frightening critical illness, and — for the small number left with deep wound or fascial complications — staged reconstruction.

Prevention

The cheapest cure is not catching it. Antibiotic prophylaxis before skin incision at caesarean section is one of the best-evidenced interventions in obstetrics: a single dose reduces wound infection, endometritis and serious infectious morbidity by roughly 60–70%, given before the incision rather than after cord clamping. Antibiotic prophylaxis after operative vaginal birth is now recommended on the strength of ANODE — a single dose of co-amoxiclav roughly halves maternal infection. Hand hygiene is the oldest and still the most powerful infection-control measure on a postnatal ward, and the GAS chain of transmission means a staff member or family contact with a streptococcal sore throat is a genuine hazard to a recently delivered woman. Perineal care, aseptic technique at delivery and with catheters, and early recognition systems (MEOWS) close the loop.

Guidelines compared

The major bodies agree on the architecture — recognise early, resuscitate with the Sepsis Six, give broad-spectrum antibiotics fast, and achieve source control — and diverge mainly in emphasis and in the resources they assume.

The substantive contrast is between the prevention emphasis of WHO (which suits a high-volume, resource-limited service) and the recognition-and-resuscitation emphasis of the sepsis bundles, with the SA Saving Mothers analyses repeatedly identifying the failure mode as delay — the fever that is rechecked rather than escalated, the transfer that is started too late. The most recent change worth stating is the RCOG consolidation into a single GTG 64 (2025) and the post-ANODE shift to prophylaxis after operative vaginal birth.

The evidence & the controversy

The defensible modern position rests on a few well-grounded points and a couple of genuinely open ones. The settled ground is source control over antibiotic escalation: retained products are evacuated, abscesses are drained, and necrotising infection is debrided, because no antibiotic clears a protected nidus or dead tissue. The endometritis regimen is settled too — clindamycin plus gentamicin is the most evidence-supported combination, once-daily gentamicin is fine, and the reflex of sending every woman home on a week of oral antibiotics after she has responded to intravenous therapy is not supported by the evidence and should be dropped.

The genuinely contested area is necrotising fasciitis adjuncts. The role of intravenous immunoglobulin in streptococcal toxic shock is biologically rational (it neutralises superantigen exotoxin) and supported by observational and some trial data, but the evidence base is not definitive, and hyperbaric oxygen is offered in some centres on weak evidence; what is not contested is that none of these adjuncts substitutes for early, repeated surgical debridement, and presenting any of them as a reason to delay theatre is the error. Septic pelvic thrombophlebitis sits on old, largely observational evidence: the "heparin defervescence" is classic teaching but the trials are small or absent, the optimal anticoagulation duration is unknown, and improving CT has shifted it from a pure diagnosis-of-exclusion towards an imaging-supported one — so a confident, dogmatic protocol overstates the data.

A current thread worth holding is antimicrobial stewardship against the reflex to broaden. Puerperal sepsis is one of the settings where the right answer is paradoxically both "start broad immediately" and "de-escalate fast on culture and stop the unnecessary oral tail" — the Cochrane endometritis data are, read correctly, a stewardship argument as much as an efficacy one. Set against rising community MRSA and resistant Gram-negatives in SA, getting cultures before the first dose and narrowing promptly is not a nicety; it is how the regimens stay effective.

Landmark trials & key evidence

Reading the prophylaxis numbers together makes the SA argument concrete. Caesarean prophylaxis cuts infectious morbidity by roughly two-thirds for the cost of a single antibiotic dose, and ANODE's operative-vaginal-birth result is of the same order: 19% of infections fell to 11%, an absolute risk reduction of about 8 percentage points, so the number needed to treat is around 1/0.08 ≈ 12 women given a single dose of co-amoxiclav to prevent one maternal infection — a high-value, low-cost intervention that is entirely deliverable in a district setting.

Exam traps & red flags

• Pain out of proportion is necrotising fasciitis until proven otherwise. Severe, rapidly worsening pain, then dusky or anaesthetic skin and crepitus, in a sick puerpera, goes to theatre — waiting for a confirmatory CT costs tissue and life. The treatment is debridement; antibiotics are adjuncts.
• Stopping breastfeeding for mastitis or a breast abscess. Continued effective milk drainage is part of the treatment; stopping worsens stasis and abscess formation. Breastfeeding continues from both breasts.
• Treating endometritis with retained products by antibiotics alone. The infected nidus must be evacuated; antibiotics will not clear it.
• The reflex oral antibiotic tail. For uncomplicated endometritis that has responded to IV therapy, continuing oral antibiotics after apyrexia adds no benefit — a stewardship and an evidence point at once.
• Missing group A streptococcus / toxic shock. Profound illness with unimpressive local signs, a household sore-throat contact, and rapid deterioration is invasive GAS; add clindamycin for anti-toxin effect and consider IVIG, and isolate the patient.
• Rechecking instead of escalating. A MEOWS trigger or a fever that will not settle in a postpartum woman is a reason to escalate now; SA Saving Mothers data repeatedly attribute sepsis deaths to delay.
• Forgetting the fever that defies adequate antibiotics. Persistent swinging fever with no demonstrable source and a good antibiotic regimen is septic pelvic thrombophlebitis — add anticoagulation; defervescence on heparin supports the diagnosis.
• Calling VTE "sepsis" (and vice versa). Postpartum pulmonary embolism and pelvic vein thrombosis mimic infection — examine the calves and chest and keep VTE in the differential of the breathless, tachycardic puerpera.
• Cardiac prophylaxis timing. Giving caesarean prophylaxis after cord clamping instead of before skin incision throws away part of the benefit.
• Under-treating because she is "young and fit". Compensation hides deterioration; the crash, when it comes, is abrupt.

Evidence anchors

• Smaill FM, Grivell RM — Antibiotic prophylaxis versus no prophylaxis for preventing infection after cesarean section, Cochrane Database Syst Rev 2014
• Knight M et al. — ANODE: prophylactic antibiotics after operative vaginal delivery, Lancet 2019
• Mackeen AD et al. — Antibiotic regimens for postpartum endometritis, Cochrane Database Syst Rev 2015
• Irusen H et al. — Treatments for breast abscesses in breastfeeding women, Cochrane Database Syst Rev 2015
• RCOG Green-top Guideline No. 64 — Identification and Management of Maternal Sepsis during and following Pregnancy, BJOG 2025
• Surviving Sepsis Campaign: International Guidelines for Management of Sepsis and Septic Shock 2026 — Prescott/Evans et al., Crit Care Med 2026
• WHO recommendations for prevention and treatment of maternal peripartum infections, 2015
• South Africa NDoH Maternity Care Guidelines and the Saving Mothers reports of the National Committee for Confidential Enquiries into Maternal Deaths (NCCEMD) — pregnancy-related sepsis a leading-but-declining cause of maternal death, with recurring avoidable factors of delayed recognition and transfer.
• South African Medicines Formulary (SAMF) / NDoH Essential Medicines List — empirical puerperal-sepsis antibiotic availability and dosing.