Appraise intrapartum hypoxia, neonatal encephalopathy and cerebral palsy, including causation criteria and the medico-legal context

In one line

Neonatal encephalopathy is a clinical syndrome of disturbed neurological function in the term newborn with many possible causes, of which intrapartum hypoxia-ischaemia is only one; the consultant task is to grade it (Sarnat), cool the eligible baby within six hours, and — separately — apply the ACOG/AAP causation criteria honestly rather than assume that a depressed baby means a negligent labour.

Mechanism & pathophysiology

The single error that runs through both the clinical management and the medico-legal argument is conflating three distinct things: hypoxic-ischaemic encephalopathy (a pathophysiological process), neonatal encephalopathy (a clinical description), and cerebral palsy (a chronic motor outcome). They overlap but are not interchangeable. Neonatal encephalopathy (NE) is the umbrella clinical term — a term or near-term infant with depressed consciousness, abnormal tone, feeding or respiratory difficulty and often seizures — and intrapartum hypoxia is the cause in only a minority. Stroke, sepsis and meningitis, inborn errors of metabolism, genetic and channelopathy syndromes, and congenital brain malformations all produce an identical clinical picture, and a fixation on "birth asphyxia" misses the treatable mimics. Hypoxic-ischaemic encephalopathy (HIE) is NE that can be attributed, with reasonable confidence, to an acute peripartum or intrapartum hypoxic-ischaemic event.

The injury after an acute hypoxic-ischaemic insult is biphasic, and that biphasic shape is the entire rationale for therapeutic hypothermia. During the insult, failure of oxidative phosphorylation collapses cellular ATP, the Na⁺/K⁺-ATPase fails, cells depolarise, glutamate floods the synapse and excitotoxic Ca²⁺ entry begins — primary energy failure. If the infant is resuscitated, oxidative metabolism partially recovers over the next 30–60 minutes (the latent phase). Then, typically 6–48 hours later, comes secondary (delayed) energy failure: mitochondrial dysfunction, accumulation of oxygen free radicals, ongoing excitotoxicity, inflammatory cytokine release and apoptotic (programmed) cell death. The depth of this secondary deterioration correlates with the eventual neurodevelopmental outcome. The latent phase between the two is the therapeutic window — cooling started within it, before secondary energy failure is established, interrupts the cascade by lowering the cerebral metabolic rate, suppressing glutamate release and free-radical generation, and reducing apoptosis. Start cooling after secondary energy failure is under way and there is little left to rescue; this is why the six-hour rule is biological, not bureaucratic.

The topography of acute hypoxic injury in the term brain is also why the MRI pattern is diagnostically powerful. A profound, abrupt insult (a sentinel event such as cord prolapse, uterine rupture or abruption) preferentially damages the metabolically active deep grey nuclei — the basal ganglia and thalami — and the perirolandic cortex, producing the dyskinetic phenotype. A more prolonged, partial insult damages the parasagittal watershed zones between vascular territories, producing spastic quadriplegia with cognitive impairment. The timing of injury can be read from the imaging: established cystic change, ventriculomegaly or established atrophy in the first days of life points to an injury that predates labour, not an intrapartum event.

Cerebral palsy (CP) is the chronic endpoint — a permanent, non-progressive disorder of movement and posture from a lesion in the developing brain. Two facts dominate the consultant understanding. First, most cerebral palsy is not caused by intrapartum hypoxia. The large majority of CP in term infants arises from antenatal factors — prenatal stroke, congenital infection, thrombophilia, placental pathology, genetic and metabolic disorders — and in preterm infants from periventricular leukomalacia and intraventricular haemorrhage; the proportion of all CP attributable to an acute intrapartum hypoxic event is small (a widely cited figure is on the order of 10% or less). Second, the subtype carries causal information: the spastic forms (hemiplegic, diplegic, quadriplegic) predominate and most often reflect non-intrapartum pathology, whereas dyskinetic CP with the basal-ganglia–thalamic MRI signature is the phenotype most consistent with an acute term intrapartum insult. A child with spastic diplegia after a preterm birth almost never has an intrapartum-hypoxia aetiology, however dramatic the labour record looks.

The understanding of CP aetiology has moved on, and the modern literature is part of a consultant's defence of a labour. Placental pathology — chronic villitis, fetal vascular malperfusion, severe chorioamnionitis with a fetal inflammatory response — is increasingly recognised as a substrate that both predates labour and lowers the fetal threshold for injury, so a baby may arrive in labour already compromised and tip into encephalopathy with an intrapartum stress that a healthy fetus would have tolerated. The inflammatory contribution is its own thread: chorioamnionitis and a fetal inflammatory response can produce encephalopathy and CP independent of frank hypoxia (the "neonatal encephalopathy of inflammatory-sensitised hypoxia" concept), which is one reason an isolated intrapartum gas does not settle causation. Most striking is the genetic signal: contemporary exome and genomic studies find a clinically meaningful proportion of children labelled as "cerebral palsy" carry pathogenic single-gene or copy-number variants — channelopathies, neurodevelopmental and movement-disorder genes — that were never of intrapartum origin at all. This is the current frontier and it cuts directly into the SA medico-legal picture: where the phenotype is atypical, the imaging does not show an acute term hypoxic pattern, or there is consanguinity or a family history, genetic testing can reclassify a "birth-asphyxia" diagnosis entirely, and increasingly should be offered.

Assessment

The assessment runs on two parallel tracks that must not be collapsed into one: grade the encephalopathy (to decide on cooling, urgently) and characterise the event (to test the causal hypothesis and exclude mimics, which can be done in parallel and over days).

Grading — the Sarnat staging. Sarnat and Sarnat described three clinical stages from a study of post-asphyxial term infants, and the grade drives both prognosis and the cooling decision.

Stage 1 generally recovers without sequelae and does not warrant cooling. Stage 2 (moderate) and Stage 3 (severe) are the cooling-eligible groups. In SA practice the Thompson score is widely used at the cot-side because it is a simple, reproducible numerical encephalopathy score that quantifies severity without an EEG; a Thompson score above 10 is a common cooling threshold where amplitude-integrated EEG (aEEG) is not immediately available. aEEG (cerebral function monitoring) adds prognostic and selection value — a persistently abnormal background (burst-suppression, continuous low voltage, flat trace) predicts poor outcome and was the selection tool in CoolCap.

Characterising the event — the building blocks of the causation argument. A cord arterial blood gas at delivery is the objective anchor: the markers of significant intrapartum metabolic acidosis are a cord arterial pH below 7.0 and/or a base deficit of 12 mmol/L or more (the interpretation of paired cord gases, the difference between metabolic and respiratory acidosis, and the sampling pitfalls that invalidate a result are developed in the companion chapter on cord-blood acid–base analysis). Beyond the gas, assemble: the Apgar scores (a low score persisting beyond five minutes is more meaningful than the one-minute figure); whether a recognised sentinel hypoxic event occurred (cord prolapse, uterine rupture, abruption, amniotic-fluid embolism, maternal collapse, ruptured vasa praevia); evidence of multisystem (multi-organ) involvement — acute kidney injury, hepatic transaminase rise, myocardial dysfunction, coagulopathy — because a single intrapartum hypoxic-ischaemic insult severe enough to injure the brain almost always injures other organs too, and an isolated encephalopathy with entirely normal kidneys, liver and heart should raise doubt about an intrapartum cause; and the neuroimaging pattern and its timing.

Investigations to exclude the mimics are mandatory, not optional. The infectious screen (cultures, lumbar puncture, consideration of congenital infection), metabolic work-up (lactate, ammonia, blood glucose, metabolic screen for inborn errors), cranial imaging for stroke or malformation, placental histology, and — increasingly — genetic testing where the picture is atypical. The ACOG/AAP framework is explicit that ascribing NE to intrapartum hypoxia requires exclusion of other identifiable causes, and skipping that work-up is both clinically negligent and, later, indefensible in court.

The specific red-flag features that point away from an acute intrapartum cause are worth naming because they are the ones a thorough assessment must actively look for: a normal cord gas with a deeply encephalopathic baby (the acidosis and the brain injury should match); microcephaly, contractures or established cystic change on early imaging (these take time to develop and mark an antenatal insult); dysmorphism or congenital anomalies; an isolated encephalopathy with no other organ injured; a family history of neonatal encephalopathy, consanguinity, or recurrence (which suggests a heritable metabolic or genetic disorder); and a clinical course that worsens after the first days or fails to follow the expected HIE trajectory. Placental histology is one of the most under-used investigations — it can reveal chronic villitis, fetal vascular malperfusion, or evidence of long-standing pathology that reframes the whole timeline, and the placenta should be sent for examination in every case of significant neonatal encephalopathy.

A sentinel hypoxic event is the strongest single piece of the causal jigsaw because it provides a plausible, timed mechanism for an acute global insult. The recognised sentinel events are a finite list to know: cord prolapse, uterine rupture (especially after a previous caesarean scar), placental abruption, ruptured vasa praevia, amniotic-fluid embolism, maternal cardiorespiratory collapse, and a sustained, profound bradycardia. Their value is twofold — they explain the physiology, and on the medico-legal track they shift the question from "was there an intrapartum cause?" to "was the event recognised and responded to within an acceptable interval?", which is where the standard-of-care argument actually lives.

Management

Organise it as immediate → ongoing → long-term, and the immediate priority is the one with a time limit measured in hours.

Immediate. Resuscitate and stabilise to NRP/HBB principles, then make the cooling decision fast. Therapeutic hypothermia is the only neuroprotective intervention proven to improve outcome in moderate-to-severe HIE, and its eligibility is narrow and time-critical:

• Gestation ≥36 weeks (some protocols ≥35 weeks) and birthweight typically ≥1800–2000 g.
• Within 6 hours of birth — the therapeutic window; the earlier the better.
• Moderate or severe encephalopathy (Sarnat 2 or 3 / Thompson >10), supported where available by an abnormal aEEG.
• Evidence of a peripartum hypoxic-ischaemic event (acidosis on cord/early gas, low Apgar, need for prolonged resuscitation).

The target is a core (rectal/oesophageal) temperature of 33–34 °C for 72 hours, followed by slow controlled rewarming (≈0.5 °C per hour) — rapid rewarming can provoke rebound seizures and reverse the benefit. Servo-controlled cooling devices are preferred; where unavailable, passive cooling (switching off the radiant warmer with continuous temperature monitoring) is used to initiate and bridge cooling, especially during transfer, but it overshoots easily and demands close monitoring. Cooling the mild (Stage 1) baby is not indicated, and cooling the profoundly, irrecoverably injured baby (Stage 3 with absent brainstem function) requires honest discussion rather than reflexive intervention.

Ongoing — neuro-intensive supportive care, because cooling is delivered inside a tightly controlled physiology. Maintain normoglycaemia (both hypo- and hyperglycaemia worsen injury), normocapnia (hypocapnia from over-ventilation causes cerebral vasoconstriction and is associated with worse outcome; hypercapnia raises intracranial pressure), normotension and adequate cerebral perfusion, and normal electrolytes. Seizure management: clinical seizures are common in HIE and frequently subclinical (electrographic-only), so aEEG/EEG monitoring matters; phenobarbital remains first-line, with second-line agents (phenytoin, levetiracetam, midazolam) as needed. Anticipate and manage the multisystem sequelae — acute kidney injury (fluid balance, avoid nephrotoxins), hepatic dysfunction and coagulopathy, myocardial dysfunction and persistent pulmonary hypertension (which itself can preclude cooling), and the bradycardia and thrombocytopenia that cooling predictably causes. Avoid hyperthermia at all costs at every stage — pyrexia worsens hypoxic brain injury.

Long-term. Structured neurodevelopmental follow-up (the cooling-trial outcomes are reported at 18–24 months for a reason — that is when disability declares); early-intervention referral, physiotherapy and occupational therapy; hearing and vision surveillance; and honest, staged communication with the family. The MRI (best performed around days 4–10, when the pattern of acute injury is most clearly established) carries prognostic weight and anchors the conversation about likely outcome.

Prognostication brings the threads together, and it is part of management rather than an afterthought. The variables that predict outcome are the Sarnat grade and its trajectory (Stage 1 recovers; severe Stage 3 with absent brainstem function carries a grim prognosis; the moderate group is where cooling changes lives and where prediction is hardest), the aEEG background and how fast it recovers, the seizure burden, and — most powerfully — the MRI. Injury confined to the watershed cortex predicts cognitive and milder motor problems; bilateral deep grey-matter (basal ganglia–thalamus) injury predicts severe motor disability and dyskinetic cerebral palsy, and a posterior-limb-of-internal-capsule abnormality is an early adverse sign. This matters because families and clinicians make redirection-of-care decisions in the first days, and those conversations must be grounded in the evolving examination, the aEEG and (when available) the MRI rather than on the labour history alone. A normal MRI after moderate encephalopathy and adequate cooling is reassuring; a baby cooled appropriately and recovering is, importantly, a better prognosis than the same baby would have had before cooling existed — which is the clinical reason the six-hour window is worth fighting for.

Guidelines compared

The bodies agree on the core of cooling but diverge in emphasis and in how they frame causation.

The headline 2014 ACOG/AAP shift, important to state, was the move away from rigid single-criterion definitions (the older "essential criteria" approach) toward a comprehensive multidimensional assessment of the whole picture — pH and base deficit, Apgar, sentinel event, MRI pattern, multisystem involvement, and exclusion of other causes considered together. No single one of those, in isolation, proves intrapartum causation.

The evidence & the controversy

Three threads define the modern argument, and each is a place where a candidate either reasons like a consultant or recites a slogan.

Cooling works — but read the trials precisely. The pivotal trials are not uniformly "positive" on their primary endpoint, and knowing which showed what is the difference between citing the evidence and gesturing at it. The NICHD whole-body trial (Shankaran 2005) reduced death-or-disability with a relative risk of 0.72. TOBY (Azzopardi 2009) did not significantly reduce its composite primary endpoint of death-or-severe-disability (RR 0.86, not significant), but it significantly increased survival free of neurological abnormality and reduced cerebral palsy among survivors — a result that is positive but more nuanced than "cooling halves disability." CoolCap (Gluckman 2005) missed significance overall but benefited the less-severe-aEEG subgroup. The case for cooling is therefore made most cleanly by the Cochrane meta-analysis (Jacobs 2013), which pooled 11 RCTs and 1505 infants and found a clear reduction in death or major neurodevelopmental disability (typical RR 0.75) with a number needed to treat of about 7 — a genuinely large effect for a neurological intervention, and the number worth carrying into a viva.

Cooling in resource-limited settings is unresolved, and it matters most to SA. The intuitive extrapolation — if cooling works in Detroit and London it must help everywhere — was directly tested and failed. HELIX (Thayyil 2021), a randomised trial in tertiary NICUs in India, Sri Lanka and Bangladesh, found no reduction in death-or-disability and a significant increase in death in the cooled group, leading the authors to conclude that hypothermia should not be offered for NE in low- and middle-income settings even where tertiary intensive care exists. The probable explanation is that the injury profile differs (more antenatal and infective contribution, a higher antenatal-onset fraction) and that without the full envelope of intensive care the cooling itself adds risk. Yet SA tertiary units do cool, and report benefit: a Chris Hani Baragwanath cohort (Nakwa et al, 2023) cooled only around two-thirds of eligible neonates — a proportion that fell over the study years, with equipment unavailability and late presentation among the documented barriers. The defensible SA position is not "cool everyone" or "never cool", but cool the right baby, in a unit with the supportive care to do it safely, while acknowledging that the LMIC evidence is genuinely cautionary — and that the access gap (a cooling device that is broken, or a baby who reaches the unit after six hours) is itself a quality-of-care failure with measurable consequences.

Continuous CTG does not prevent cerebral palsy — and pretending otherwise is the root of much avoidable litigation. Continuous electronic fetal monitoring was introduced on the assumption that catching hypoxia would prevent brain injury; it has not delivered that. The Nelson 1996 case-control analysis is the landmark: specific CTG abnormalities (multiple late decelerations, decreased beat-to-beat variability) were associated with an increased risk of CP, but the false-positive rate was 99.8% — the overwhelming majority of abnormal traces are not followed by cerebral palsy, so acting reflexively on them generates caesareans without benefit. Continuous CTG, compared with intermittent auscultation, reduces neonatal seizures but has not been shown to reduce cerebral palsy or perinatal death, at the cost of more operative delivery. The candidate-level synthesis is that CTG is a screening test for fetal acidaemia with poor specificity for long-term neurological outcome, not a CP-prevention tool — a point developed in detail under intrapartum-fetal-surveillance-ctg. The intervention that genuinely prevents a form of cerebral palsy is a different one entirely: antenatal magnesium sulfate given before anticipated very preterm birth, which reduces cerebral palsy in survivors (BEAM, 2008) and is the basis for SA and international neuroprotection protocols at gestations below roughly 32 weeks.

A live and unsettled question sits at the moment of birth itself: what to do with the cord when the baby is born flat. Deferred (delayed) cord clamping improves haematological and some outcomes in vigorous infants, but the depressed, non-breathing baby is exactly the one who classically needs to reach the resuscitaire — and also, in theory, the one who might benefit most from continued placental transfusion and an intact circulation while breathing is established. Intact-cord (bedside) resuscitation, keeping the cord unclamped while initial resuscitation proceeds, is being actively studied for this group, and the honest position is that the evidence does not yet support a firm recommendation either way for the compromised neonate; in a resource-limited delivery room the practical constraint (equipment at the mother's side, a second pair of hands) often decides it. It is a genuine current controversy rather than a settled protocol, and a defensible answer states the uncertainty rather than overcommitting.

This evidence sits inside an acute South African controversy: an obstetric medico-legal crisis in which cerebral-palsy claims dominate state liability. Contingent liabilities for medical negligence have run into the tens of billions of rand (claims lodged against the state were reported at roughly R68 billion as of March 2023, with birth-related brain-injury payouts a large share and individual CP claims averaging around R20 million), and the great majority of obstetric claim value is for cerebral palsy. The clinical and the legal collide here: the causation criteria exist precisely to distinguish the labour that genuinely caused the injury from the far more common situation in which a baby with antenatal or genetic pathology happens to have had a difficult birth. Misapplying "abnormal CTG → asphyxia → CP" — in either direction — is both bad medicine and the engine of unaffordable, often unmeritorious, litigation. The professional response is meticulous intrapartum documentation, honest cord-gas and multi-organ assessment, MRI to read the timing of injury, and a willingness to state, where the evidence supports it, that the cerebral palsy was not of intrapartum origin.

Landmark trials & key evidence

An arithmetic worth holding: the Cochrane pooled NNT of about 7 means cooling roughly 7 eligible babies prevents one death or major disability — a large effect — yet HELIX showed that the same intervention, transplanted into NICUs without the full supportive-care envelope, produced no benefit and excess death. Same therapy, opposite sign, because the denominator (the case-mix and the surrounding care) changed. That is the appraisal point: an NNT is not portable across settings.

Exam traps & red flags

• Equating neonatal encephalopathy with birth asphyxia. NE has many causes; the work-up to exclude stroke, infection, metabolic and genetic disease is mandatory before attributing it to intrapartum hypoxia.
• Claiming most cerebral palsy is intrapartum in origin. It is not — the attributable fraction is small, most CP is antenatal (or, in preterm infants, from PVL/IVH), and spastic CP after preterm birth is almost never an intrapartum-hypoxia story.
• Missing the six-hour cooling window. Cooling started after secondary energy failure is established loses its rationale; the window is biological. A baby who arrives at the cooling unit at seven hours has been failed by the system, not by physiology.
• Cooling the wrong baby. Mild (Sarnat 1) encephalopathy is not an indication; cooling a preterm or low-birthweight infant outside protocol, or a profoundly injured Stage 3 baby without honest discussion, is an error in the opposite direction.
• Rapid rewarming. Rewarm slowly (≈0.5 °C/h); fast rewarming provokes rebound seizures and can undo the benefit.
• Over-ventilating the cooled baby. Hypocapnia causes cerebral vasoconstriction and is associated with worse outcome; aim for normocapnia, normoglycaemia, normotension, and never let the baby become hyperthermic.
• Reading the CTG as a cerebral-palsy test. An abnormal trace has a ~99.8% false-positive rate for CP; it is a poor-specificity screen for acidaemia, not a CP-prevention tool, and treating it as proof of intrapartum causation drives both unnecessary caesareans and unmeritorious litigation.
• Extrapolating cooling trials uncritically to a district setting. HELIX is the counter-example; cooling belongs in a unit that can deliver the surrounding neuro-intensive care.
• Forgetting antenatal magnesium for neuroprotection. The intervention that demonstrably prevents a form of CP is MgSO₄ before very preterm birth, not intrapartum CTG.
• Asserting intrapartum causation without multi-organ evidence. An isolated encephalopathy with entirely normal kidneys, liver, heart and clotting is hard to attribute to a single global hypoxic-ischaemic insult — that mismatch should prompt a search for another cause.

Evidence anchors

• NICHD whole-body hypothermia — Shankaran et al, N Engl J Med 2005
• CoolCap selective head cooling — Gluckman et al, Lancet 2005
• TOBY — Azzopardi et al, N Engl J Med 2009
• TOBY study protocol — Azzopardi et al, BMC Pediatr 2008
• Cooling for newborns with HIE (Cochrane) — Jacobs et al, 2013
• HELIX — Thayyil et al, Lancet Glob Health 2021
• Uncertain value of EFM in predicting cerebral palsy — Nelson et al, N Engl J Med 1996
• BEAM — MgSO₄ for prevention of cerebral palsy — Rouse et al, N Engl J Med 2008
• Sarnat & Sarnat staging — Arch Neurol 1976
• Neonatal Encephalopathy and Neurologic Outcome, 2nd ed (executive summary) — ACOG/AAP Task Force, Obstet Gynecol 2014
• Template for defining a causal relation between intrapartum events and CP — MacLennan, BMJ 1999
• Therapeutic hypothermia for intrapartum asphyxia at a SA public tertiary hospital — Nakwa et al, BMC Pediatrics 2023
• South African NCCEMD Saving Babies / Perinatal Problem Identification Programme (PPIP) — intrapartum asphyxia and birth trauma a leading cause of perinatal death, with provider-associated avoidable factors in a substantial proportion of cases.