Evaluate and manage neurological disorders in pregnancy

In one line

Two questions decide management: is this new neurology a pre-eclampsia complication or a primary CNS event?, and how do I keep a woman with epilepsy seizure-free on the least teratogenic regimen? Everything else (stroke, headache, myasthenia, IIH) is pattern recognition layered on those two reflexes.

This chapter assumes the groundwork in hypertension and pre-eclampsia basics; it covers the discrimination between mimics, the subtype-specific regimens and how they differ, the primary-literature appraisal, and the judgement calls.

Assessment

The seizure / new-deficit triage. Any first seizure, focal deficit or thunderclap headache after 20 weeks is eclampsia until proven otherwise — check BP, proteinuria and platelets before reaching for a CT. But eclampsia does not explain a lateralising deficit, a depressed conscious level that does not recover post-ictally, focal signs, or seizures before 20 weeks; those mandate neuroimaging. Do not let "it must be eclampsia" anchor you past a cerebral venous sinus thrombosis (CVST) or intracerebral haemorrhage — maternal stroke is now a leading direct cause of maternal death in high-income audits and is repeatedly missed as "atypical pre-eclampsia."

Five conditions sit on the same eclampsia look-alike spectrum, each with a different mechanism and a different treatment. The difficulty is not recognising textbook eclampsia; it is parsing the woman who almost fits.

• Eclampsia / PRES (posterior reversible encephalopathy syndrome). Mechanism: failure of cerebral autoregulation under acute hypertension → hyperperfusion, blood–brain-barrier breakdown and vasogenic oedema, classically parieto-occipital. The clinical consequence is cortical blindness, seizures and headache that recover as BP is controlled and the placenta is removed. PRES on MRI (bilateral posterior white-matter oedema, restricted-diffusion-negative) is the radiological signature of pre-eclampsia with severe features and resolves — if it does not resolve, question the diagnosis.

• Reversible cerebral vasoconstriction syndrome (RCVS) / postpartum angiopathy. This is the mimic that most often catches the clinician who reflexively gives magnesium. Mechanism: transient dysregulation of cerebral arterial tone → segmental, multifocal vasoconstriction ("string-of-beads" on angiography), typically in the first days to two weeks postpartum, often after an uncomplicated delivery. The distinguishing feature is recurrent thunderclap headache (maximal in under a minute) rather than the building headache of pre-eclampsia, and crucially the angiogenic-factor imbalance of eclampsia is not present. It can co-exist with PRES and can cause both ischaemic infarction and convexity subarachnoid haemorrhage. Vasoconstriction is the lesion, so calcium-channel blockade (nimodipine) is the directed treatment and serotonergic/vasoconstrictor drugs (ergometrine, triptans, SSRIs, sympathomimetics) must be avoided — the opposite reflex to "she's postpartum, give ergometrine."

• CVST. Mechanism: pregnancy/puerperium is a hypercoagulable, stasis-prone state (the puerperium especially) → dural-sinus thrombosis → venous hypertension → headache (often positional, worse lying flat), seizures, focal deficits and venous (often haemorrhagic) infarction that does not respect arterial territories. Because the problem is venous occlusion, the treatment is anticoagulation even in the presence of haemorrhagic venous infarction — counter-intuitive, and the most common error is withholding heparin because "there's blood on the scan."

• Arterial ischaemic / haemorrhagic stroke. Mechanism varies (paradoxical embolism, cardio-embolism, dissection, and in the pre-eclamptic the haemorrhagic stroke of uncontrolled severe hypertension). A lateralising deficit or a focal seizure is arterial/structural until imaged, and reperfusion is time-critical.

• Eclampsia in a woman who also has primary epilepsy. Do not assume a known epileptic's late-pregnancy seizure is "just her epilepsy" — superimposed pre-eclampsia is common, so check BP, platelets and urate in every peripartum seizure regardless of the past history.

• Imaging is not contraindicated. Non-contrast CT delivers a negligible fetal dose; MRI/MRV is first-line for suspected CVST and posterior circulation events; CT venography or contrast is justified when it changes management. Withholding imaging to "protect the fetus" is the classic medicolegal trap.

• Headache red flags: thunderclap onset, postural component (worse lying down suggests raised ICP/CVST; worse upright after a dural puncture suggests post-dural-puncture headache), new aura, fever/neck stiffness, papilloedema, any focal sign, or seizure. A migraine diagnosis in pregnancy is one of exclusion — first-ever "migraine with aura" in the third trimester is a deficit until imaged. The differential of the peripartum thunderclap headache is RCVS, CVST, subarachnoid haemorrhage and pituitary apoplexy (Sheehan's in the haemorrhaged postpartum pituitary) — not migraine.

• Epilepsy booking review: seizure type and frequency, current antiseizure medication (ASM) and dose, last seizure, adherence, contraception/folate history, and whether the woman of childbearing potential is on valproate — which should already have been switched preconception, not in the booking clinic.

• Investigations: platelets, urate, LFTs and urine PCR to stage pre-eclampsia; ASM levels matter most for lamotrigine (clearance rises ≥65% by the first trimester and up to ~3-fold by late pregnancy → breakthrough seizures); lumbar puncture (opening pressure) for suspected idiopathic intracranial hypertension (IIH) or to exclude SAH/meningitis once mass effect is excluded.

Severity stratification and the judgement calls. Stratification turns less on a score than on three trajectory questions: (1) Is the conscious level recovering? A Glasgow Coma Scale that does not climb back after a post-ictal period is a structural lesion until proven otherwise. (2) Is the deficit focal or global? Global cerebral irritability fits eclampsia; a hemiparesis, a homonymous field defect or a III-nerve palsy does not. (3) Is the headache positional or thunderclap? — because that single discriminator separates CVST and RCVS from the pre-eclamptic headache. In epilepsy the stratification is different again: the woman whose seizures cluster, who has had a generalised tonic-clonic seizure in this pregnancy, or whose lamotrigine level has fallen below her preconception therapeutic level is the one at risk of status and SUDEP, and is the one to admit and re-titrate.

This builds on hypertension and pre-eclampsia basics; the discipline is not defaulting to that diagnosis.

Management

Immediate — the acute seizure / acute stroke.

The named magnesium regimens and exactly how they differ. Magnesium is the proven agent both for eclampsia treatment and pre-eclampsia seizure prophylaxis; there are three regimens, each making a different trade-off.

• Pritchard (IM, loading + IM maintenance). 4 g IV (as a 20% solution over 5–10 min) plus 10 g IM (5 g into each buttock), then 5 g IM every 4 hours, alternating buttocks, for 24 h after delivery or the last fit. Advantage: needs no infusion pump, so it is the SA district-hospital default where pumps and monitored beds are scarce. Disadvantage: painful injections, and a fixed dose that cannot be titrated to renal function — so reflexes, respiratory rate and urine output must be checked before each IM dose.
• Zuspan (IV loading + IV infusion). 4 g IV over 5–15 min, then a continuous 1 g/h IV infusion. Advantage: smoother serum levels, titratable, the regimen used in Magpie and most international practice. Disadvantage: requires a pump and closer monitoring — hence less suited to an unmonitored district setting.
• Sibai (higher IV loading + higher infusion). 6 g IV loading, then 2 g/h IV infusion. The higher-dose variant aims to reach therapeutic levels faster and in heavier women, at the cost of a narrower safety margin and a greater need for level/clinical monitoring.

The SA NDoH compromise is a hybrid: a 4 g IV load (where a line and a slow push are available) combined with the Pritchard IM maintenance where no pump exists, or a Zuspan-style 1 g/h infusion where one does. All three regimens target the same therapeutic window (~2–3.5 mmol/L); they differ only in route and rate, and the choice is a systems decision (do I have a pump and a monitored bed?) not a pharmacological one.

Magnesium monitoring and the antidote. Toxicity is monitored clinically, not by levels: loss of patellar reflexes is the first sign (~4–5 mmol/L), then respiratory depression (~5–6.5 mmol/L; hold the infusion if respiratory rate <12/min), then cardiac arrest at the extreme. Magnesium is renally cleared, so oliguria → accumulation: reduce or hold maintenance if urine output is <100 mL over 4 h, and check reflexes before every Pritchard IM dose. Keep 10% calcium gluconate 10 mL IV drawn up as the antidote.

A pivotal exception: in myasthenia gravis, magnesium sulphate is contraindicated — it blocks presynaptic acetylcholine release and precipitates myasthenic crisis. If a myasthenic woman develops eclampsia, use levetiracetam/phenytoin for seizures and avoid magnesium; control BP with labetalol/methyldopa/hydralazine and involve neurology and anaesthesia early. The same caution applies to a woman with significant renal impairment, where any magnesium regimen must be reduced and levels followed.

Subtype-specific acute management — where the eclampsia reflex is wrong.

• CVST: start therapeutic-dose LMWH as soon as the diagnosis is made, including when there is haemorrhagic venous infarction (the venous occlusion, not the secondary blood, drives the disease). Decompressive surgery is reserved for impending herniation. Antiepileptics are used for seizures but are not given prophylactically without a seizure or a supratentorial lesion.
• RCVS / postpartum angiopathy: the directed treatment is nimodipine (a calcium-channel blocker) for the vasoconstriction, plus removal of any precipitant; avoid all vasoconstrictors — ergometrine, triptans, sympathomimetics. Magnesium and BP control are reasonable supportive measures, but the diagnosis must be made because the postpartum-haemorrhage reflex (give ergometrine) is actively harmful here.
• Acute arterial stroke: time-critical referral; pregnancy/puerperium is not an absolute contraindication to intravenous thrombolysis or to mechanical thrombectomy — the decision is by individualised risk/benefit and the same time windows as the non-pregnant patient. BP should be treated to non-pregnant thresholds, not the permissive ones obstetricians sometimes tolerate.
• Status epilepticus: the same ABC ladder as the non-pregnant — benzodiazepine first, then IV levetiracetam or phenytoin as second-line; protect the airway, nurse left-lateral, exclude and treat hypoglycaemia. Magnesium is not the drug for primary epileptic status (see "evidence" below for why).

Ongoing — the antenatal pathway.

• Epilepsy: the governing principle is seizure freedom on monotherapy at the lowest effective dose of the least teratogenic agent, ideally lamotrigine or levetiracetam. Do not stop ASMs because of pregnancy — a generalised tonic-clonic seizure carries real risk of trauma, hypoxia and SUDEP. Continue folic acid 5 mg daily (NDoH/RCOG); monitor lamotrigine levels and pre-empt the rising-clearance breakthrough by uptitrating against levels through pregnancy, then taper back to the preconception dose within ~3 weeks of delivery to avoid postpartum toxicity. Counsel on SUDEP and on never bathing alone.
• The lamotrigine pharmacokinetic trap in depth. Lamotrigine is glucuronidated, and oestrogen massively induces UGT1A4 — so clearance climbs early (≥65% above preconception by the first trimester for most women) and can rise up to ~3-fold by the third trimester, dropping the free drug level and precipitating breakthrough seizures in a previously stable woman. The directed approach is therapeutic-drug-monitoring-guided dosing: establish a preconception target level, check levels each trimester (or with any seizure), uptitrate to hold that level — then because clearance normalises within days of delivery, the dose must be reduced back toward preconception over the first two to three weeks postpartum or the woman becomes toxic (ataxia, diplopia, dizziness). Levetiracetam also has increased clearance in pregnancy but the relationship to seizures is less tight; phenytoin's protein-binding shift makes total levels misleading (measure free phenytoin if you must use it).
• Valproate carries the highest malformation rate and a dose-dependent IQ decrement; SA and international guidance is that it (and phenobarbitone) should never be initiated in a woman of childbearing potential and only continued if genuinely no alternative controls seizures, under specialist sign-off and pregnancy-prevention counselling. If a woman conceives on valproate, do not stop it abruptly in the first trimester — the neural-tube window has usually passed, organogenesis damage is already done or not, and an abrupt withdrawal seizure adds risk without removing the teratogenic exposure; refer urgently, optimise folate, arrange detailed anomaly scanning, and plan a postpartum transition.
• Migraine: paracetamol first-line; avoid ergots and routine triptans (sumatriptan has the most reassuring data if essential); low-dose aspirin/propranolol or amitriptyline for prophylaxis. Migraine with aura is itself a stroke/pre-eclampsia risk marker — consider aspirin prophylaxis.
• IIH: acetazolamide can be continued/used after counselling when vision is threatened; serial therapeutic LPs and weight-gain limitation are mainstays; protect vision with serial perimetry. IIH is not an indication for caesarean and not a contraindication to neuraxial analgesia. The threat is to vision, not to the pregnancy: management is driven by serial formal visual fields and fundoscopy, escalating from weight control and acetazolamide → serial therapeutic LPs → CSF diversion (shunt) or optic-nerve-sheath fenestration for fulminant, vision-threatening disease that fails medical therapy. The fetal calculus around acetazolamide is reassuring enough that sight-threatening papilloedema justifies its use.
• Myasthenia: continue pyridostigmine and, if needed, prednisolone/azathioprine; IVIg or plasma exchange for crisis. Avoid drugs that worsen transmission (aminoglycosides, magnesium). The course is unpredictable — exacerbation is commonest in the first trimester and the puerperium; the second stage exposes the weakest muscles (the striated voluntary muscles fatigue), which is why an assisted second stage is planned.

Long-term / delivery & postpartum.

• Most neurological conditions allow vaginal birth; caesarean is for obstetric indications or maternal exhaustion (e.g. myasthenic respiratory fatigue in second stage — assist the delivery with forceps/vacuum rather than asking the woman to perform a prolonged Valsalva). Effective regional analgesia reduces the catecholamine surge that triggers seizures and myasthenic fatigue, and obtunds the hypertensive/intracranial-pressure response that matters in CVST and IIH.
• The puerperium is the highest-risk window for stroke and CVST — the hypercoagulable, fluid-shifting first two weeks. Maintain a low threshold for imaging postnatally, and remember RCVS/postpartum angiopathy presents into this window, often after an uncomplicated birth.
• Breastfeeding is encouraged on ASMs, including lamotrigine and levetiracetam; the small breast-milk exposure is outweighed by benefits (and is far less than the in-utero exposure the infant has already tolerated).
• Neonatal alerts: transient neonatal myasthenia (transplacental anti-AChR antibodies; observe babies of myasthenic mothers for feeding/respiratory weakness over the first days), and reinstating the mother's pre-pregnancy lamotrigine dose to prevent postpartum toxicity.
• Recurrence and contraception counselling. A woman who had a pregnancy-related CVST or stroke needs a documented thrombophilia and cause work-up and a future-pregnancy thromboprophylaxis plan; oestrogen-containing contraception is contraindicated after CVST/stroke (progestogen-only or non-hormonal methods instead). The epileptic woman needs interaction-aware contraception: enzyme-inducing ASMs (carbamazepine, phenytoin, phenobarbitone, topiramate) reduce the efficacy of combined and progestogen-only pills and the implant — the depot injection and the copper/levonorgestrel IUD are the reliable choices — and lamotrigine levels themselves fall when an oestrogen-containing pill is added (and rise in the pill-free week), a two-way interaction worth knowing.

The evidence & the controversy

Beyond stating that lamotrigine and levetiracetam are preferred, the substance is appraising which study answers which question and deriving the numbers.

The teratogenicity hierarchy is now well-grounded, not opinion. The EURAP prospective registry established a clean rank order of major congenital malformations — valproate ~10%, phenobarbital/phenytoin/carbamazepine intermediate, and lamotrigine, levetiracetam and oxcarbazepine ~2–3%, within the background range — and, critically, that risk is dose-dependent for valproate, carbamazepine, lamotrigine and phenobarbital. So "switch off valproate" is incomplete advice; the corollary is "use the lowest effective dose of whatever agent you choose." The dose effect is large: in EURAP the lowest-risk band was lamotrigine <300 mg/day at 2.0%, and valproate at high dose (above ~1450 mg/day) carried a malformation risk above 20% — roughly a ten-fold gradient created by drug-and-dose alone. The neurodevelopmental signal is separate and arguably more important: NEAD showed a valproate-specific, dose-dependent IQ deficit persisting to age 6, while the contemporary MONEAD cohort found no overall cognitive penalty for the newer ASMs (largely lamotrigine/levetiracetam) versus unexposed children. The malformation question and the cognition question are answered by different studies, and valproate fails both.

The eclampsia-treatment evidence is a distinct literature from the eclampsia-prophylaxis literature. Magpie answered prophylaxis (does magnesium stop a pre-eclamptic woman fitting?); the Collaborative Eclampsia Trial (1995) answered treatment (which anticonvulsant for a woman who has already fitted?). It is the trial that made magnesium first-line over the older agents:

• vs diazepam: recurrent convulsions 13.2% (60/453) with magnesium vs 27.9% (126/452) with diazepam — a 52% relative reduction (95% CI 37–64%). The absolute risk reduction is 27.9% − 13.2% = 14.7%, so the NNT ≈ 1 / 0.147 ≈ 7 women treated with magnesium rather than diazepam to prevent one recurrent fit.
• vs phenytoin: recurrent convulsions 5.7% (22/388) with magnesium vs 17.1% (66/387) with phenytoin — a 67% relative reduction (95% CI 47–79%). ARR = 17.1% − 5.7% = 11.4%, so NNT ≈ 1 / 0.114 ≈ 9. Women given magnesium rather than phenytoin were also less likely to be ventilated, to develop pneumonia, or to need intensive care.

Magnesium is superior for eclamptic seizures (a cerebral-vasospasm/endothelial mechanism), but it is an eclampsia drug, not a general anticonvulsant — it has no role as monotherapy for primary epileptic status, and it is frankly harmful in myasthenia.

The live controversy is maternal stroke, where the 2026 AHA scientific statement reframes practice. It stresses that pregnancy and the puerperium are independent stroke risk states, that hypertensive thresholds for treatment in pregnancy should not be more permissive than outside it (the old habit of tolerating systolic 160s "because she's pregnant" is dangerous), and that acute reperfusion therapy should not be withheld solely because a woman is pregnant or postpartum. The friction in SA practice is access — thrombectomy and rapid MRV are concentrated at tertiary centres, so the district/regional task is recognition, BP control and time-critical referral rather than definitive treatment. The same access reality shapes the Saving Mothers signal: in SA, non-pregnancy-related sepsis and pre-existing medical conditions (including epilepsy and the consequences of poorly controlled seizures) sit among the indirect causes of maternal death, and the avoidable factors are recurrently delayed recognition and delayed transfer — the cheap, system-level steps, not the heroic ones.

CVST evidence and the counter-intuitive treatment. The principle that anticoagulation is given even with haemorrhagic venous infarction rests on the trial and registry data showing better outcomes with early heparin/LMWH despite intracranial blood — pregnancy-associated CVST treated with anticoagulation reaches recovery rates comparable to non-pregnancy CVST. Withholding heparin "because of the haemorrhage" inverts the evidence.

RCVS / postpartum angiopathy is the differential that most changes management: it is not eclampsia, the angiogenic-factor imbalance is absent, the treatment is nimodipine and avoidance of vasoconstrictors, and giving ergometrine for postpartum haemorrhage in such a woman can precipitate or worsen the vasoconstriction. Note the honest uncertainty: nimodipine relieves the thunderclap headache in a majority of patients but the largest case series did not show an outcome benefit over symptomatic care — so the directed treatment is reasonable and the avoidance of vasoconstrictors is the firmer recommendation.

Landmark trials & key evidence

Worked viva — how to structure the answer

A common stem is "a 31-year-old, day-3 postpartum after an uncomplicated normal delivery, presents with a thunderclap headache and a brief seizure; BP 148/92, platelets normal, no proteinuria." A high-scoring answer runs:

1. Frame the discrimination, don't anchor. "My first reflex is eclampsia, but two things don't fit: the headache is thunderclap and there is no proteinuria or thrombocytopenia. After an uncomplicated delivery into the high-risk puerperal window, I must actively exclude CVST, RCVS/postpartum angiopathy and subarachnoid haemorrhage, not just treat as eclampsia."
2. Resuscitate and make safe — ABC, left-lateral, control any active seizure with a benzodiazepine, control BP to non-pregnant targets, and image early: non-contrast CT then MRI/MRV. State explicitly that imaging is not withheld for a breastfeeding mother.
3. Let the imaging set the treatment — "If MRV shows sinus thrombosis I start therapeutic LMWH even if there is venous haemorrhage; if angiography shows segmental vasoconstriction I treat as RCVS with nimodipine and avoid all vasoconstrictors — importantly I would not give ergometrine for any uterine atony in that woman; if it is genuinely eclampsia I give magnesium by the regimen my unit can monitor."
4. Name the regimen and why — "If I use magnesium I'd use a Zuspan IV infusion where I have a pump and a monitored bed, or the Pritchard IM regimen at a district hospital with no pump, monitoring reflexes, respiratory rate and urine output, with calcium gluconate drawn up."
5. Justify from evidence — Collaborative Eclampsia Trial for magnesium over phenytoin/diazepam in treatment; the CVST anticoagulation principle; the AHA position that reperfusion is not withheld for pregnancy.
6. Close the loop — cause work-up, avoid oestrogen contraception after CVST/stroke, document a future-pregnancy thromboprophylaxis plan, and the SA reality of stabilise-and-transfer to a centre with MRV/neurology.

For the epilepsy stem ("a woman with epilepsy on lamotrigine, now 28 weeks, has had a breakthrough seizure"), the structured answer is: confirm adherence and check a lamotrigine level against her preconception target (clearance has risen), uptitrate to restore that level rather than switching drugs, continue 5 mg folate, counsel SUDEP and bathing safety, plan a postpartum dose reduction within ~3 weeks, and screen for superimposed pre-eclampsia before assuming it is "just her epilepsy."

Exam traps & red flags

• Anchoring on eclampsia. A focal deficit, pre-20-week seizure, thunderclap (rather than building) headache, or non-resolving consciousness is not eclampsia — image it. Missed CVST, RCVS and haemorrhagic stroke kill.
• Withholding imaging or thrombolysis "for the fetus." CT/MRI doses are negligible; reperfusion is not contraindicated by pregnancy.
• Withholding anticoagulation in CVST because there is haemorrhagic venous infarction — the venous occlusion is the disease; treat it with LMWH.
• Giving ergometrine (or a triptan) to a woman with postpartum angiopathy / RCVS — vasoconstrictors worsen the lesion; this is the postpartum mirror of the magnesium-in-myasthenia error.
• Magnesium in myasthenia gravis — precipitates crisis; the single most dangerous reflex error in this topic. Magnesium also has no role as monotherapy for primary epileptic status.
• Stopping ASMs in pregnancy to "protect the baby" — risks SUDEP and traumatic seizures; the malformation risk is managed by drug choice and dose, not cessation. And do not abruptly stop valproate in the first trimester of an established pregnancy — refer and plan a postpartum switch.
• Forgetting lamotrigine clearance rises (≥65% by the first trimester) → third-trimester breakthrough seizures, then postpartum toxicity if the dose is not reduced back within ~3 weeks.
• Initiating valproate in any woman of childbearing potential, or failing to document pregnancy-prevention counselling if it is unavoidable.
• Tolerating systolic 160 mmHg in a pregnant woman with headache — that is a stroke risk, not a benign pregnancy reading.
• Labelling first-ever third-trimester "aura" as migraine without excluding a deficit.
• Prescribing oestrogen-containing contraception after a pregnancy-related CVST or stroke, or forgetting that enzyme-inducing ASMs undermine the pill/implant — choose depot or an IUD.

Evidence anchors

• Tomson T et al. Comparative risk of major congenital malformations with eight ASMs (EURAP). Lancet Neurol 2018;17:530–538.
• Tomson T et al. Dose-dependent risk of malformations with ASMs (EURAP). Lancet Neurol 2011;10:609–617.
• The Eclampsia Trial Collaborative Group. Which anticonvulsant for women with eclampsia? Lancet 1995;345:1455–1463.
• Meador KJ et al. Fetal ASM exposure and cognition at age 6 (NEAD). Lancet Neurol 2013;12:244–252.
• Meador KJ et al. Cognitive outcomes at age 3 (MONEAD). Lancet Neurol 2023.
• Miller EC et al. Prevention and Treatment of Maternal Stroke. AHA Scientific Statement. Stroke 2026.
• The Magpie Trial: magnesium sulphate vs placebo for pre-eclampsia. Lancet 2002;359:1877–1890.
• RCOG Green-top Guideline No. 68: Epilepsy in Pregnancy (2016).
• SA National Department of Health Maternity Care Guidelines (epilepsy: 5 mg folic acid, avoid initiating valproate/phenobarbitone; eclampsia: magnesium per Pritchard/Zuspan-hybrid regimen); SAMF for ASM, magnesium and antihypertensive dosing.