26.1 Postoperative Seizures, Hemorrhage, Ischemia, and Hyperperfusion (04.A, C, D, H)

A patient who was following commands in the recovery area and now is not is a clock, not a narrative. Independent OpenExamPrep teaching for ABIM Neurocritical Care outline items 04.A, 04.C, 04.D, and 04.H is the four-way split you run before you reach for a sedative: postoperative seizure, hematoma, ischemic stroke (arterial or venous), and cerebral hyperperfusion. Those entities share a declining examination and they do not share a blood-pressure target. This chapter is perioperative complication recognition. Disease-state subarachnoid hemorrhage, AVM natural history, and carotid stenosis decision-making live in earlier chapters; here the question is what the operation just did.

This is not an American Board of Internal Medicine, American Board of Psychiatry and Neurology, American Heart Association, or neurosurgical-society product. Name the source when a number comes from a trial or a guideline, and do not imply official endorsement of this guide.

Early versus late postoperative seizures, and when to load

The International League Against Epilepsy treats a seizure within 7 days of an acute brain insult as an acute symptomatic (early) seizure. After a craniotomy, that insult is cortical irritation, blood, edema, electrolyte shift, or a hematoma you have not imaged yet. Late postoperative seizures begin after day 7. Early seizures do not, by themselves, diagnose epilepsy. Late seizures have a much higher risk of unprovoked recurrence and are treated as new epilepsy until you find a reversible trigger (abscess, residual tumor, hyponatremia, missed hematoma).

Load an antiseizure drug (ASD) when a postoperative seizure has already occurred, when the examination is too poor to detect convulsions and continuous EEG shows electrographic seizures, or when the patient already had epilepsy and missed doses. A first early seizure after craniotomy is not “just a twitch”: image first if the patient does not wake cleanly, then load. Levetiracetam is the usual intensive-care load (for example 20–30 mg/kg intravenous, then a maintenance schedule) because it does not require level-chasing and does not interact with nimodipine the way phenytoin does. Phenytoin still appears in older stems; it is a worse perioperative choice because of hypotension with rapid infusion, rash, and enzyme induction.

Prophylaxis without a prior seizure is a shorter, more political question:

Worked contrast: a seizure-naive convexity meningioma wakes, has a 60-second focal-to-bilateral tonic-clonic event at hour 4, CT is clean — load levetiracetam and keep a short course. A Hunt-Hess 2 anterior-communicating aneurysm that never seized does not need phenytoin “for the clip.” A late seizure on postoperative day 18 after glioblastoma resection is a new epilepsy workup and a longer ASD plan, not another 7-day “prophylaxis” sticker.

If the patient is still unresponsive after a reported seizure, treat status epilepticus as in chapter 24 and obtain a noncontrast CT before you blame residual anesthetic. Nonconvulsive seizures after craniotomy are common enough that a poor examination plus a dirty EEG is not “ICU psychosis.”

Hematoma at the operative bed versus remote hemorrhage

Operative-bed hematoma is bleeding into the cavity you just left: residual tumor bed, evacuated hematoma cavity, clip site, or the epidural space under a bone flap. Most clinically important bed hematomas declare in the first 6 hours; almost all announce themselves within 24 hours (Taylor and colleagues, 1995, remain the series boards still paraphrase). The story is wake-then-decline, a blown pupil on the operative side, or a sudden hypertension that is the Cushing response, not “pain.” Risk factors you can name: incomplete hemostasis, residual vascular nidus, hypertension in the first hours, coagulopathy, antiplatelet agents that were not reversed, and thrombocytopenia. The move is STAT CT, reverse what is reversible, and call the operating room — not another 4 mg of morphine.

Remote hemorrhage is blood away from the surgical field. The pattern you must not miss is remote cerebellar hemorrhage (RCH) after supratentorial craniotomy, aneurysm clipping, or spinal surgery with large CSF loss. CSF hypovolemia lets the cerebellum sag, stretching bridging veins; the bleed is venous. CT shows streaky blood in the cerebellar sulci — the zebra sign — sometimes with hydrocephalus from fourth-ventricle compression. This is not a “missed posterior-circulation aneurysm” until you have thought about the drain and the CSF leak. Clamp or raise any suction subgaleal or lumbar drain, stop overdrainage, reverse coagulopathy, and decompress if the patient is herniating. Remote supratentorial hematomas also occur with hypertension and coagulopathy; they are arterial or venous depending on the scan, but they are still not the clip site.

Do not call every new blood on the first postoperative CT a “surgical disaster.” A thin cavity film after a glioblastoma resection is expected. Expanding hematoma plus decline is the emergency.

Arterial versus venous infarct after aneurysm or AVM surgery

Arterial infarct after clipping or AVM resection follows the arteries you operated on. Parent-vessel stenosis or occlusion from a clip that kinks the M1, A2, or a perforator; intraoperative thrombosis; embolus from a temporary clip; and, later, delayed cerebral ischemia from vasospasm after aneurysmal SAH (chapter 21) all produce arterial maps. Named perforator disasters: recurrent artery of Heubner after anterior-communicating work (caudate and anterior limb of internal capsule), lenticulostriates after MCA aneurysm clipping, and brainstem perforators after basilar-apex surgery. The deficit appears as the patient fails to wake or drops a dense, territorial examination. CTA, catheter angiography, or a return trip to the microscope answers “is the vessel open?” Diffusion MRI confirms the infarct once the patient is stable enough to travel.

Venous infarct follows sacrifice or thrombosis of a vein that still mattered. Classic setups: vein of Labbé or a dominant temporal bridging vein during a subtemporal approach; superior sagittal sinus or cortical veins during a parasagittal meningioma; and AVM draining-vein thrombosis when the nidus is only partly gone and a residual feeder still pressurizes a compromised vein. Venous infarcts are often hemorrhagic, more edema than a clean MCA wedge, and they do not respect a single arterial tree. Treatment is blood-pressure sanity, seizure control, and — when the vein is thrombosed and the nidus is gone — sometimes anticoagulation after a hard conversation about the fresh cavity. Do not anticoagulate a residual AVM because the scan “looks venous.”

Normal perfusion pressure breakthrough (NPPB) after AVM resection is the AVM-specific cousin of hyperperfusion (next heading): chronically steal-conditioned brain suddenly sees full arterial pressure. Surrounding edema and hemorrhage are not always a residual nidus, but residual nidus must still be excluded with angiography.

Cerebral hyperperfusion after CEA, CAS, and AVM resection

Cerebral hyperperfusion syndrome (CHS) follows carotid endarterectomy (CEA) or carotid artery stenting (CAS) when a chronically hypoperfused hemisphere loses autoregulation and then receives a sudden jump in flow. The clinical triad boards expect is ipsilateral throbbing headache, seizure, and intracerebral hemorrhage (or focal edema that looks like a stroke until you check the pressure and the TCD). Onset is hours to about 2 weeks, with many cases in the first 7 days. A Vascular Quality Initiative analysis (published 2025) found CHS in about 0.25% of carotid revascularizations, more often after transfemoral CAS than after CEA or transcarotid revascularization; uncontrolled hypertension roughly doubled the odds. That rarity is why the syndrome is missed: the nurse calls it a migraine and the intern raises the pressure “for the new stroke.”

Transcranial Doppler that shows a marked jump in ipsilateral middle-cerebral velocity (classically a doubling of the preoperative baseline, or a large absolute increase) supports the diagnosis when you have a window. Perfusion imaging shows ipsilateral hyperperfusion, not a mismatch core. Treatment is blood-pressure reduction, not induced hypertension. Many units keep postoperative pressure below 140/90 mm Hg, and drop further (sometimes toward 120/80 mm Hg) once CHS is suspected. Prefer agents that do not raise cerebral blood flow: labetalol or clonidine. Pure vasodilators (nitroprusside, high-dose dihydropyridine infusions) can worsen hyperemia. Treat the seizure with an ASD. Do not give intra-arterial vasodilators as if this were vasospasm.

The same physiology follows AVM resection (NPPB) and some extracranial-intracranial bypasses. Staged embolization exists to shrink that risk; your ICU job is still a tight pressure plan and a low threshold to re-image hemorrhage.

Timeline table you can carry into a stem

Exam traps

Treating a 4-hour decline after craniotomy as residual propofol. Starting phenytoin for every clipped aneurysm. Calling zebra-sign cerebellar blood a “posterior-circulation embolus from the heart.” Raising blood pressure for ipsilateral headache and seizure the day after CEA. Anticoagulating a residual AVM because the hemorrhage “looks venous.” Keeping levetiracetam for 6 months after an uneventful, seizure-naive meningioma because “the brain was touched.”

In practice

Hour 3 after a temporal glioblastoma: the patient seizes, then the pupil enlarges — CT first, not a second levetiracetam push as the only move; bed hematoma beats “early seizure” when the eye is blown. Hour 8 after sitting-adjacent CSF-lossy aneurysm clipping: delayed wake, zebra-sign cerebellum — clamp the suction drain. Day 2 after CEA for a 90% stenosis: ipsilateral pounding headache and a first seizure at systolic 190 mm Hg — this is CHS, lower the pressure with labetalol, do not start a vasospasm hypertension protocol. Day 12 after AVM resection: a late seizure with a clean cavity — that is a late seizure, not a reason to invent a 7-day prophylaxis rule after the fact.