Autoregulation is impaired for hours after a craniotomy, and emergence supplies the hypertension. Coughing on the tube is the cheapest catastrophe in neuroanaesthesia — and the bleed it causes usually arrives long after everyone has stopped watching for it.
After craniotomy, cerebral autoregulation is often impaired for hours, so a hypertensive surge at emergence is transmitted more or less directly to a fresh surgical bed. Pre-haemorrhage hypertension in the first 12 postoperative hours carried an odds ratio of 4.6 for intracranial haemorrhage in a matched case-control study — and the haemorrhages themselves occurred at a median of 21 hours, interquartile range 4 to 52. The pressure you allow at extubation and the bleed it produces are separated by most of a day.
The last five minutes of a craniotomy can undo the previous five hours, and the mechanism is unglamorous: the patient coughs, the pressure goes up, and it arrives undamped at a vascular bed that was being cauterized a few minutes ago.
Two facts combine. Catecholamine release during emergence causes only a weak rise in oxygen consumption — so this is not primarily a metabolic problem — but it may raise blood pressure significantly. And cerebral autoregulation is often impaired in the hours following elective craniotomy.
Take those together and the buffer that normally stands between a systemic pressure surge and the cerebral circulation is not reliably there. The combination of systemic hypertension and altered autoregulation is what raises the risk of intracranial haemorrhage after craniotomy. Other risk factors are older age, larger tumour size, and underlying coagulopathy.
The evidence is a retrospective case-control study: 69 patients who developed postoperative intracranial haemorrhage, matched with 138 controls by age, date of surgery, pathologic diagnosis, surgical procedure and surgeon.
Pre-haemorrhage hypertension in the first 12 postoperative hours was present in 62% of cases against 25% of controls, with an odds ratio of 4.6.
The haemorrhages occurred at a median of 21 hours, interquartile range 4 to 52 hours.
Read the second figure carefully, because the commonly repeated version of it — that these bleeds typically happen within six hours — is not what the study found. Six hours is roughly the lower quartile. A quarter of these haemorrhages occurred later than 52 hours.
The caveat travels with the finding: this is observational and retrospective. A haemorrhage that has already begun can raise the blood pressure by itself, so the direction of causation is not established. The matching is careful and the association is strong, and it is still an association.
The listed measures are: adequate analgesia, prevention of hypothermia and shivering, and timely tracheal extubation without bucking and coughing. Those act on the cause. The drugs act on the consequence.
Remifentanil. Low-dose infusions reduce the haemodynamic changes and coughing associated with extubation, with minimal recovery delay. One thing to watch for: low plasma concentrations of 2.3 to 2.6 ng/mL can produce episodes of bradypnoea below 10 breaths per minute, typically settling within 20 minutes of arrival in recovery. Since airway reflexes and respiratory drive are commonly depressed after craniotomy anyway, these patients need watching for hypercapnia and its consequences for intracranial pressure.
Esmolol. A bolus of 1 mg/kg followed by 100 to 300 µg/kg/min for 15 to 30 minutes after extubation improves haemodynamic stability and reduces cerebral blood flow changes at emergence. Note the units again — the same thousand-fold trap as at induction.
Labetalol. Initial dose 5 to 20 mg intravenously, repeated to effect, followed by 1 to 3 mg/kg/h if needed. It reduces pressure through chronotropic and inotropic mechanisms rather than peripheral vasodilation, so it is safe with a raised intracranial pressure. Its half-life of 3 to 5 hours is the reason to think twice: if this patient is the one who bleeds, you have committed to a drug you cannot take back quickly.
Nicardipine. An intravenous loading dose of 15 µg/kg, then an infusion at 5 mg/h, increased by 5 mg/h no more often than every 5 minutes.
Emergence should be prompt and complete, and smooth. The endotracheal tube comes out when strict extubation criteria are met, and a train-of-four ratio above 0.9 should be confirmed first — neurosurgical patients are particularly susceptible to residual curarization, and visual or tactile assessment of blockade is unreliable.
Whether a patient must follow commands before extubation remains controversial. The consideration that resolves it in practice: failed extubation is associated with worse outcomes than delayed extubation, and it is commonly agreed that many patients who cannot follow commands but can clear pulmonary secretions are safely extubated.
This is where the day earns its place. The pressure control you provided ends when you hand the patient over. The haemorrhage risk does not — its median is 21 hours away, and its upper quartile is beyond two days.
So the handover has to carry two things that are easy to leave out. What the blood pressure did during emergence and in recovery, because that is the exposure the study identified. And what the patient’s neurological baseline was at the moment they left your care, because a delayed emergence or a new deficit is diagnosed by comparison with it — and the causes of delayed emergence are categorized as physiologic, pharmacologic, neurosurgical and neurologic, which is a list you can only work through if you know where the patient started.
No randomized trial addresses this.
The physiology is established. Catecholamine release at emergence causes only a weak rise in oxygen consumption but can raise blood pressure significantly. Cerebral autoregulation is often impaired in the hours following elective craniotomy. The combination of systemic hypertension and altered autoregulation increases the risk of intracranial haemorrhage — the pressure is transmitted to a vascular bed that has lost its ability to buffer it, and there is a fresh surgical field waiting.
The timing and the magnitude come from a case-control study. Sixty-nine patients who developed postoperative intracranial haemorrhage were matched with 138 controls by age, date of surgery, pathologic diagnosis, surgical procedure and surgeon. Pre-haemorrhage hypertension in the first 12 postoperative hours was present in 62% of cases against 25% of controls, with an odds ratio of 4.6. Haemorrhages occurred at a median of 21 hours, interquartile range 4 to 52 hours.
That is an association from a retrospective design. It is the best available evidence, the matching is careful, and it cannot establish that the hypertension caused the bleeding rather than accompanying it — a haemorrhage that has begun can itself raise blood pressure.
The drug choices are consensus. Esmolol has been shown to blunt the rise in cerebral blood flow velocity during emergence, and the specific regimens for esmolol, labetalol, nicardipine and remifentanil are accumulated practice rather than comparative trial results.
It is widely taught that intracranial haemorrhage after craniotomy typically occurs within six hours, and the source text repeats it. Go to the study the claim rests on and the numbers are different: median 21 hours, interquartile range 4 to 52 hours.
Both statements come from the same case-control series, and only one of them is in it. A quarter of these haemorrhages occurred after 52 hours. Under half had happened by the end of the first day.
The reason this matters is not academic. "Within six hours" sets a watching period, and it is the shorter half of the real distribution. A patient handed over at hour seven as through the danger period is, on these data, more likely than not still ahead of it.
What is concentrated early is the exposure rather than the event. Pre-haemorrhage hypertension was recorded in the first 12 postoperative hours — 62% of cases against 25% of controls, odds ratio 4.6. So the pressure you allow at emergence and in the first half day is plausibly doing its damage long before the bleed declares itself.
Two windows, and collapsing them into one produces exactly the wrong handover: watch the pressure for six hours, then stop watching the patient. The correct version is closer to the reverse — the pressure matters most early, and the patient needs watching much longer than that.
One honest caveat belongs with all of it. This is a retrospective case-control study, carefully matched but observational, and a haemorrhage that has already started can raise the blood pressure by itself. The association is strong and the direction of causation is not established.
Full text · Preston goes through LibKey to whatever Preston holds — the PDF where there is one — and works on a phone. The doi link always reaches the publisher's record, whether or not we hold a subscription. If you are reading on a computer, LibKey Nomad adds the same one-click PDF button to PubMed and publisher pages everywhere else; it is a desktop browser extension, so it does nothing on a phone or tablet.
In the case-control study relating perioperative hypertension to intracranial haemorrhage after craniotomy, what was the timing of the haemorrhages?
A patient is being extubated after resection of a frontal meningioma. As the tube comes out the blood pressure rises to 190/100 with coughing. Which agent is preferable?
Why is a hypertensive surge at emergence more dangerous after craniotomy than after comparable non-neurosurgical operations?
The internal vertebral venous system drains into the external vertebral system and then the caval veins, and its veins have no valves. That is precisely why it engorges in pregnancy and obesity, and it is why a free abdomen is the first safety goal of prone positioning. Freeing the abdomen is the fastest correction available.
The 2019 ASA Practice Advisory on perioperative visual loss warns against the routine use of intraoperative deliberate hypotension, and the physiology supports the position — epidural venous plexus pressure and intraosseous pressure are the important determinants of blood loss in spine surgery and are largely independent of arterial pressure. The higher-yield conversation is about abdominal support and table tilt.