Interpret intrapartum fetal surveillance and CTG, and manage suspected fetal compromise in labour

In one line

The cardiotocograph is a screening test for fetal hypoxia with high sensitivity and dismal specificity; the consultant skill is not pattern-spotting but distinguishing the fetus that is compensating from the one that is decompensating, acting on reversible causes first, and confirming or refuting suspected acidaemia before an irreversible decision — because continuous CTG halves neonatal seizures but does nothing for cerebral palsy or mortality while sharply raising the caesarean rate.

Mechanism & pathophysiology

The fetal heart rate is a window onto the autonomic nervous system, and the autonomic nervous system reports on cerebral oxygenation. Everything the trace shows — the baseline, the variability, the shape and timing of decelerations — is a readout of how the fetal brainstem is being perfused, which is why a CTG can warn of hypoxia before the tissue is injured and why it cannot tell you that injury has already occurred.

Oxygen reaches the fetus by a delivery chain that can fail at any link: maternal arterial oxygen content, uteroplacental blood flow through the spiral arteries, transfer across the placental villous membrane, umbilical venous return, and fetal cardiac output. Labour stresses every link. Each contraction transiently occludes the spiral arteries and interrupts intervillous perfusion, so the fetus must withstand repeated 60–90-second ischaemic insults and recover in the gaps between them. A healthy term fetus has the reserve to do this many hundreds of times; a growth-restricted, post-dates, or acidotic fetus does not.

When oxygen delivery falls, the response unfolds as a sequence, and reading where a fetus sits on that sequence is the whole of intrapartum interpretation:

• Hypoxaemia — reduced oxygen in the blood, no tissue effect yet. The fetus mounts a chemoreceptor- and catecholamine-mediated response: peripheral vasoconstriction, redistribution of cardiac output to brain, heart and adrenals, and a rise in baseline rate. Decelerations may appear but variability is preserved. This is compensation, and it can continue for hours without harm.
• Hypoxia — oxygen lack now reaches the tissues. Anaerobic glycolysis in peripheral beds generates lactate; a respiratory (CO₂-driven) acidosis is buffered, but as redistribution is sustained the picture shifts.
• Metabolic acidaemia — once anaerobic metabolism in the central organs cannot be offset, fixed acid accumulates, the base deficit climbs, and myocardial function and central nervous control begin to fail. Variability is lost, the baseline may fall, and decelerations deepen and stop recovering. This is decompensation, and it is the state that precedes hypoxic-ischaemic injury.

The individual features map onto this physiology:

• Baseline (normal 110–160 bpm) is set by the balance of sympathetic and parasympathetic tone. A rising baseline is an early catecholamine-driven sign; tachycardia (>160 for >10 min) accompanies maternal pyrexia (the commonest cause), early hypoxaemia, beta-agonists and fetal tachyarrhythmia. Bradycardia (<110 for >10 min) is most ominous as a terminal sign of profound hypoxia but also occurs with maternal hypothermia, beta-blockade and heart block.
• Variability (normal 5–25 bpm) is the most informative single feature because it requires an intact, oxygenated brainstem oscillating sympathetic against parasympathetic output beat to beat. Reduced variability (<5 bpm for >50 min, or >3 min within a deceleration) reflects central depression — from acidosis, but equally from fetal sleep, opioids and magnesium sulphate, which is why isolated reduced variability is interpreted cautiously and in context. Variability that is preserved despite decelerations is the single most reassuring finding on a worrying trace: a fetus that can still oscillate its rate is still compensating.
• Accelerations denote a neurologically responsive, non-acidotic fetus; their absence in labour is of uncertain significance and is not, on its own, pathological.
• Decelerations are where mechanism and management converge:
  • Early decelerations are shallow, symmetrical, and mirror the contraction; they are vagally mediated by head compression and carry no hypoxic meaning.
  • Variable decelerations are abrupt (onset to nadir <30 s), vary in size and shape, and reflect a baroreceptor response to cord compression. The majority of intrapartum decelerations are variable, and most are benign. They acquire significance when they evolve "atypical" features — slow recovery, loss of variability within the deceleration, a combined U-shaped component, or duration beyond 3 minutes — signalling an emerging chemoreceptor (hypoxic) element.
  • Late decelerations are smooth, U-shaped, begin after the contraction peak and return to baseline after the contraction ends; they are chemoreceptor-mediated and indicate that contraction-associated hypoxaemia is reaching the chemoreceptors. Repetitive late decelerations with reduced variability are the classic decompensation pattern.
  • Prolonged decelerations (>3 min) demand the operator at the bedside: below 80 bpm with reduced variability they carry a chemoreceptor component and frequently accompany acute events.

The distinction between respiratory and metabolic acidosis is not academic, because it separates a fetus that will recover within minutes of birth from one that has sustained a tissue insult. A respiratory acidosis is pure CO₂ retention from a brief interruption of gas exchange — a low pH with a high pCO₂ and a near-normal base deficit — and it corrects rapidly once the cord is cut and the lungs aerate. A metabolic acidosis reflects accumulated fixed acid from anaerobic metabolism — a low pH with a raised base deficit (the threshold of concern is a base deficit ≥12 mmol/L) — and it is the type associated with hypoxic-ischaemic injury. The CTG tracks the trajectory between these states: a fetus with preserved variability and recovering decelerations is buffering a respiratory load; a fetus with absent variability and non-recovering late decelerations is building the metabolic deficit that confirmation at delivery (paired cord gases) will later quantify. This is why a worrying trace is interrogated during labour with a test of acid–base status rather than left to declare itself, and why the cord gases drawn at birth are the retrospective check on whether the intrapartum read was right.

The corollary that governs everything downstream: the CTG detects the physiological state, not the outcome. A fetus can sit in stable compensation indefinitely and a normal trace is genuinely reassuring; but suspicious and pathological traces have a very high false-positive rate for actual acidaemia, because the same pattern is produced by a fetus that will be perfectly well and by one that is decompensating. That asymmetry — excellent at confirming health, poor at confirming harm — is the reason every intervention that follows is aimed at reversing a cause and confirming the suspicion, not at delivering on the trace alone.

Assessment

Interpretation is a structured read of five features against a stated risk context, then a category, then a search for cause — never a category in isolation.

• Establish the risk context first. Continuous CTG is indicated where the risk of hypoxia is raised: intrapartum risk factors include meconium-stained liquor, antepartum or intrapartum bleeding, maternal pyrexia or sepsis, oxytocin augmentation, suspected growth restriction or oligohydramnios, prematurity, post-dates, hypertensive disease, diabetes, previous caesarean, and a multiple pregnancy. The same trace means different things in a low-risk multipara and a growth-restricted post-dates primigravida; the denominator changes the index of suspicion.
• Read the four basic features, then classify. Baseline, variability, accelerations and decelerations are each judged over a 10-minute window against the contraction pattern, and the trace is then categorised. Re-evaluate at least every 30 minutes, and continuously when abnormal — the signal changes through labour.
• Tachysystole and the contractions. Excessive uterine activity (>5 contractions in 10 minutes over two successive 10-minute periods) is the commonest reversible driver of an abnormal trace, because it shortens the recovery interval the fetus needs between perfusion insults. Always read the toco channel alongside the rate.
• Exclude the maternal heart rate masquerading as the fetal. Doppler external monitoring can double-count or lock onto the maternal pulse, especially in the second stage; an apparent "acceleration with every push" is the giveaway. Where any doubt exists, palpate the maternal pulse against the trace, or move to a fetal scalp electrode — a recurring and preventable medico-legal trap is a reassuring trace that was, in fact, recording a well mother over a dying fetus.
• Sinusoidal versus pseudosinusoidal. A true sinusoidal pattern — regular, smooth, undulating, 5–15 bpm amplitude, 3–5 cycles/min, lasting >30 min with absent accelerations — is pathological and points to fetal anaemia (rhesus alloimmunisation, fetomaternal haemorrhage, twin-to-twin transfusion, ruptured vasa praevia) or severe acute hypoxia, and triggers urgent action. The pseudosinusoidal pattern is jagged, transient (typically <30 min), often follows maternal opioids or fetal mouthing, and is benign; the discriminator is the smoothness and the duration.

Categorisation — what the classification actually means

The international systems share one three-tier architecture — normal / suspicious / pathological — built from the same four features. The point of the category is not to label but to set the tempo of response: a normal trace needs nothing, a suspicious trace needs a search for and correction of reversible causes plus closer observation, and a pathological trace needs immediate action on reversible causes and a decision about expediting delivery if it does not recover.

The FIGO 2015 criteria, which the SA setting most closely mirrors:

(Decelerations are "repetitive" when present with >50% of contractions.) The NICE NG229 and RCOG schemes carry the same tiers but classify the individual features slightly differently and frame the decision around the whole clinical picture rather than the trace alone — addressed below.

The acid test of any category is whether you can name the reversible cause and the state of fetal reserve, not whether you can recite the box it falls in.

Management

The plan is built immediate → ongoing → definitive, and the governing principle is that a suspicious or pathological trace is a prompt to correct what is reversible and confirm whether the fetus is truly acidotic, not an instruction to deliver. Most abnormal patterns recover when the cause is addressed.

Immediate — conservative resuscitation and the search for a cause. The mnemonic is less important than doing all of it at once while someone senior reviews the trace:

• Reposition to left or right lateral to relieve aortocaval compression and improve uteroplacental flow; this alone reverses a large share of variable and prolonged decelerations.
• Stop or reduce oxytocin and consider acute tocolysis with a beta-agonist (salbutamol or terbutaline) or other uterine relaxant where tachysystole is driving the trace — relieving excessive activity restores the recovery interval between contractions.
• Correct maternal hypotension, classically after epidural or spinal, with an intravenous fluid bolus and a vasopressor (ephedrine or phenylephrine).
• Treat maternal pyrexia/sepsis (antipyretics, fluids, cultures, antibiotics), since pyrexia both raises the baseline and worsens any hypoxic insult.
• Perform a vaginal examination to exclude cord prolapse and to assess progress and station, which determine the route of delivery if escalation is needed.
• Ask the mother to stop pushing in the second stage if the trace deteriorates with active pushing, allowing recovery before resuming.

Routine maternal facial oxygen and routine intravenous fluids in a normovolaemic woman are not supported for the purpose of improving the trace and are no longer recommended for that indication; treat them as interventions for a maternal need, not for the CTG.

Confirming or refuting acidaemia — second-line tests. When a pathological trace persists despite conservative measures and the clinical picture does not mandate immediate delivery, a test of fetal acid–base status can convert a low-specificity screen into a decision:

• Fetal scalp stimulation is the simplest and cheapest: digital stimulation of the scalp that provokes an acceleration makes significant acidaemia very unlikely, and a positive response can spare an operative delivery. It requires no equipment and is the only second-line test available in most SA labour wards.
• Fetal blood sampling (FBS) quantifies the acid–base state from a scalp capillary sample. Conventional pH thresholds: ≥7.25 normal (repeat if the trace persists), 7.21–7.24 borderline (repeat within ~30 min), and ≤7.20 abnormal (expedite delivery). Lactate thresholds (analyser-dependent, commonly): <4.2 mmol/L normal, 4.2–4.8 borderline, >4.8 mmol/L abnormal. Lactate needs a far smaller sample and fails far less often than pH, which is its practical advantage. FBS requires adequate cervical dilatation, ruptured membranes, an accessible vertex, and is contraindicated where maternal blood-borne infection (HIV, hepatitis, active herpes) risks vertical transmission via the scalp wound or where a coagulopathy or prematurity <34 weeks applies — a real limitation in the SA HIV setting, where scalp stimulation often becomes the de facto second-line test.
• STAN / fetal-ECG ST-analysis combines the CTG with automated analysis of the ST segment of the fetal ECG (via a scalp electrode), flagging the rise in T/QRS ratio that signals myocardial anaerobic metabolism. It is used only as an adjunct to a CTG that already shows an abnormal pattern, never as a primary screen, and its evidence is contested (below).

Escalation and expediting delivery. Where the trace is pathological and does not recover, or an acute event (cord prolapse, abruption, scar rupture, sustained bradycardia) is present, the decision moves to delivery. The route follows the cervix: full dilatation with a suitable station permits operative vaginal delivery; otherwise caesarean. For acute, irreversible compromise the standard is delivery within ~30 minutes of decision (the "category-1" caesarean), but the clock is a target, not a justification for delay — in an acute bradycardia the relevant interval is "as fast as safely possible." Sustained terminal bradycardia, a true sinusoidal pattern with anaemia, or a confirmed scalp pH ≤7.20 each mandate immediate delivery without further testing.

The South African context shapes every step. CTG machines, scalp-blood analysers and reliable cardiotocograph paper are not uniformly available across district and regional facilities. The NDoH Guidelines for Maternity Care in South Africa therefore reserve continuous CTG for high-risk labour and where intermittent auscultation is abnormal, and use intermittent auscultation as the default for low-risk women: a Pinard stethoscope or hand-held Doppler, listening immediately after a contraction for at least one minute, at least every 15–30 minutes in the first stage and after every contraction (or at least every 5 minutes) in the second stage. An abnormal auscultatory finding (a baseline outside 110–160, or decelerations heard) escalates to continuous CTG where available, or to expedited referral where it is not. FBS is rarely available below tertiary level, so the realistic SA second-line test is scalp stimulation, and the realistic safety net for a worrying trace in a district unit is transfer or delivery, not further monitoring. ART optimisation and the PVT framework apply when an HIV-positive woman is in labour, and the increased vertical-transmission risk of scalp electrodes and FBS in the setting of unsuppressed viraemia is a real reason to prefer non-invasive surveillance.

Guidelines compared

The major bodies agree on the three-tier architecture and on conservative-measures-first; they diverge on the fineprint of feature classification and on how mechanically the category drives the decision.

Two recent shifts are worth naming. First, admission CTG for low-risk women has been actively de-adopted — it raises intervention without improving outcomes and NICE advises against it. Second, the ACOG/NICHD category II problem (a third or more of all traces fall here, and the category alone does not tell you what to do) is exactly why the Clark standardised management algorithm exists; the European schemes avoid an "indeterminate" middle category by forcing every feature into normal/suspicious/pathological, which trades a smaller grey zone for slightly more aggressive escalation.

The evidence & the controversy

The uncomfortable foundation of this entire field is that continuous CTG, the standard of care across high-resource labour wards for fifty years, has never been shown to prevent the outcome it was introduced to prevent. The Cochrane review of continuous CTG versus intermittent auscultation — 13 trials and over 37,000 women — found that continuous monitoring halves neonatal seizures (RR 0.50, 95% CI 0.31–0.80) but produces no reduction in cerebral palsy (RR 1.75, not significant) and no reduction in perinatal mortality (RR 0.86, not significant), while raising caesarean section (RR 1.63) and operative vaginal delivery (RR 1.15). A test that cuts seizures but not cerebral palsy or death, at the cost of a 60% relative rise in caesarean, is a test with a real but narrow benefit and a large iatrogenic footprint — and the candidate who can hold both halves of that sentence at once is reasoning correctly.

The reason the trace does not translate into prevented cerebral palsy is causal, not technological. Most cerebral palsy does not arise from intrapartum hypoxia at all — it originates antenatally, from infection, genetic and developmental causes, and events before labour — so a perfect intrapartum monitor could only ever prevent the minority of cases that are intrapartum in origin. Worse, the trace is a poor predictor even of that minority: in the case–control work, late decelerations and reduced variability carried a 99.8% false-positive rate for predicting cerebral palsy. The clinical translation is that an abnormal CTG identifies a population at slightly higher risk while being wrong about the individual nearly every time, which is precisely the recipe for over-intervention.

Attempts to rescue the test's specificity by adding technology have mostly failed. Computerised interpretation, the obvious fix for inter-observer disagreement (itself a documented weakness of CTG), was tested at scale in the INFANT trial — over 47,000 women, decision-support software added to the trace — and made no difference to poor neonatal outcomes (adjusted RR 1.01) or to development at two years. Fetal-ECG ST-analysis is genuinely contested: the original Swedish Amer-Wåhlin trial showed reductions in metabolic acidosis and operative delivery for non-reassuring fetal status, but the large American MFMU trial (Belfort) found no improvement in any perinatal outcome. The honest position is that STAN is at best an adjunct of marginal and inconsistent benefit, not a solution to CTG's specificity problem, and certainly not a priority for a resource-limited service.

What has improved outcomes is unglamorous: protecting variability as the key reassuring feature, acting on reversible causes, confirming suspicion with a cheap test before committing to delivery, and — most of all — the human factors of structured interpretation, fresh-eyes review and timely escalation. The recurring theme of the confidential national audits and of the litigation literature alike is that babies are harmed not by the wrong CTG criteria but by a correctly abnormal trace that was misread, not escalated, or recorded the maternal heart rate. The current direction of travel, embodied in NICE NG229, is therefore away from ever-finer pattern taxonomy and towards the whole clinical picture, risk-stratification and human-factors training — an acknowledgement that the limiting factor is the interpreter, not the algorithm.

A second strand of current debate is the move to interrogate the physiological pattern of deterioration rather than to count features against a static table. Approaches that read the trace as a story — a progressive loss of accelerations, then deepening and broadening decelerations, then a rising baseline, then loss of variability and a terminal fall — aim to catch the fetus on the downslope of compensation rather than waiting for a feature count to cross a threshold. The attraction is physiological coherence; the caution is that no such scheme has yet shown an outcome benefit over the standard categories in a randomised trial, and adopting an unvalidated interpretive framework as if it were established is its own error. The defensible stance is to use the recognised categorisation while reasoning explicitly about where on the compensation–decompensation curve a given fetus sits, which is what the categories are a proxy for.

This is also the field most shaped by medico-legal pressure. The litigation cost of a hypoxic-ischaemic encephalopathy claim is enormous, and the defensive instinct it creates — to deliver on any worrying trace — is itself a driver of the rising caesarean rate that the Cochrane data warn about. The defensible position is not "monitor more" but "interpret well, confirm before acting, and escalate early," which both serves the fetus and withstands later scrutiny.

Landmark trials & key evidence

A worked figure makes the trade-off concrete. The Cochrane caesarean risk ratio of 1.63 means that for every 100 caesareans that would occur under intermittent auscultation, roughly 163 occur under continuous CTG — 63 extra caesareans per 100 baseline — to achieve a halving of neonatal seizures — an outcome that itself does not translate into less cerebral palsy or death. The arithmetic is the whole argument for reserving continuous CTG for genuinely high-risk labour and not defaulting every low-risk woman to it.

Exam traps & red flags

• Treating the category as the decision. "Pathological CTG, therefore caesarean" is the wrong reflex. The correct move is to correct reversible causes (reposition, stop oxytocin, treat hypotension and pyrexia, exclude cord prolapse) and, where it persists and the picture allows, confirm acidaemia before an irreversible decision. Most abnormal traces recover.
• Missing the maternal heart rate masquerading as fetal. A "reassuring" Doppler trace that accelerates with every push may be recording the mother. Palpate the maternal pulse against the trace; move to a scalp electrode if in doubt. This is a recurring cause of unrecognised intrapartum death.
• Ignoring variability. Variability is the most informative feature: preserved variability through decelerations is the strongest sign of ongoing compensation, and reduced variability is the warning of decompensation — but only after sleep, opioids and magnesium have been excluded.
• Mistaking pseudosinusoidal for sinusoidal. A true sinusoidal pattern (smooth, >30 min, absent accelerations) points to fetal anaemia or severe hypoxia and demands urgent action; the benign, jagged, transient pseudosinusoidal pattern after opioids does not. Discriminate on smoothness and duration.
• Defaulting low-risk women to continuous CTG. Admission CTG and routine continuous monitoring in low-risk labour raise intervention without improving outcomes; intermittent auscultation is the evidence-based and NDoH-endorsed default.
• Forgetting the reversible cause. Tachysystole from oxytocin is the commonest correctable driver of an abnormal trace; reading the toco channel and stopping the oxytocin is first-line, not an afterthought.
• Over-claiming STAN or computerised CTG. Presenting ST-analysis as established or computerised interpretation as a fix for CTG's specificity ignores the negative MFMU and INFANT trials; both are contested or null adjuncts, not solutions.
• Confusing screening accuracy with outcome. CTG is sensitive but very non-specific for acidaemia, and abnormalities are very poor predictors of cerebral palsy (99.8% false-positive); an abnormal trace raises population risk while usually being wrong about the individual.
• Using FBS where it is unsafe or unavailable. Scalp sampling is contraindicated with maternal blood-borne infection (the SA HIV reality), coagulopathy and prematurity; where it is unavailable, scalp stimulation is the realistic second-line test and delivery/transfer the safety net — not more monitoring.
• Letting the 30-minute caesarean clock excuse delay in an acute bradycardia. A sustained terminal bradycardia, abruption, scar rupture or cord prolapse is delivered as fast as safely possible; the 30-minute figure is an audit standard, not a permitted waiting time.

Evidence anchors

• Continuous CTG (EFM) for fetal assessment during labour — Alfirevic et al., Cochrane Database Syst Rev 2017
• The Dublin randomized controlled trial of intrapartum fetal heart rate monitoring — MacDonald et al., Am J Obstet Gynecol 1985
• FIGO consensus guidelines on intrapartum fetal monitoring: Cardiotocography — Ayres-de-Campos et al., Int J Gynaecol Obstet 2015
• Computerised interpretation of fetal heart rate during labour (INFANT) — INFANT Collaborative Group, Lancet 2017
• CTG plus ST analysis of the fetal ECG, Swedish RCT — Amer-Wåhlin et al., Lancet 2001
• Fetal ECG ST-segment analysis (MFMU) — Belfort et al., N Engl J Med 2015
• pH or lactate in fetal scalp blood, RCT — Wiberg-Itzel et al., BMJ 2008
• Fetal scalp lactate sampling — East et al., Cochrane Database Syst Rev 2015
• Uncertain value of EFM in predicting cerebral palsy — Nelson et al., N Engl J Med 1996
• Neonatal Encephalopathy and Neurologic Outcome (2nd ed), Executive Summary — ACOG/AAP, Obstet Gynecol 2014
• Intrapartum management of category II FHR tracings — Clark et al., Am J Obstet Gynecol 2013
• NICE NG229 — Fetal monitoring in labour (2022), nice.org.uk/guidance/ng229
• South African National Department of Health — Guidelines for Maternity Care in South Africa (the "Maternal Book"): intermittent auscultation as the default for low-risk labour; CTG for high-risk and for abnormal auscultation.