The Anesthetic That Was Not the Problem

01

The pearl

The observational literature was consistent and it was believed: patients anesthetized for thrombectomy did worse than patients sedated. Three randomized trials — SIESTA, AnStroke and GOLIATH — then compared the two techniques directly under protocolized blood pressure and found no disadvantage to general anesthesia, with functional outcomes if anything favouring it. The cohorts were not wrong about the association. They were wrong about what produced it.

02

Why it is true

mechanism

Collateral perfusion to the ischemic penumbra is pressure-dependent. That single fact is the whole of the physiological argument that drove a decade of avoiding general anesthesia for thrombectomy, and it is correct.

When a large vessel occludes, the tissue that survives long enough to be worth reperfusing survives on leptomeningeal collaterals. Flow through those channels depends on the pressure gradient driving it, and autoregulation in that territory is impaired or abolished. Perfusion becomes pressure-passive. Any sustained fall in mean arterial pressure translates fairly directly into less flow to the tissue you are trying to save, and induction of general anesthesia produces exactly such a fall — sympatholysis, vasodilation, and reduced venous return arriving together in a patient who is often already dry.

Set against that is what general anesthesia buys. The patient does not move. Catheter navigation through tortuous intracranial vessels is easier and plausibly safer, and reperfusion success was higher in the general anesthesia arm of GOLIATH. Airway protection is secured rather than improvised in a patient who may deteriorate on the table.

The argument for avoidance treated the hemodynamic hazard as intrinsic to the technique. The trials treated it as a manageable feature of how the technique is delivered, and protocolized against it. That distinction is the whole day. What follows in slot 03 is not evidence that induction is hemodynamically benign — it plainly is not. It is evidence that when somebody decides in advance what pressure they will hold and has the means to hold it, the outcome penalty the registries predicted does not appear.

Figure 1. Why the cohorts and the trials disagree without either being fraudulent. In routine practice the decision to intubate tracks the thing that also determines outcome — a higher NIHSS, a worse ASPECTS, a patient who cannot protect an airway or hold still. The association between anesthetic and outcome is real; the causal arrow drawn from it is not. Randomization does not measure the association better. It removes the path that produced it.

03

The evidence

3 RCTs · n = 368

SIESTA

Schönenberger S, Uhlmann L, Hacke W, et al. Effect of conscious sedation vs general anesthesia on early neurological improvement among patients with ischemic stroke undergoing endovascular thrombectomy: a randomized clinical trial. JAMA. 2016;316(19):1986-1996. No association

Design: Single-centre randomized trial, Heidelberg
Population: 150 adults with anterior circulation large-vessel occlusion undergoing endovascular thrombectomy
Intervention: Conscious sedation
Control: General anesthesia with intubation
Primary: Change in NIHSS at 24 hours (early neurological improvement)
Result: No significant difference between groups in the primary endpoint. The prespecified 90-day secondary favoured general anesthesia, with functional independence (mRS 0–2) in 37.0% of the general anesthesia group versus 18.2% of the sedation group.

What it does not settle Single centre, with a dedicated neuroanaesthesia team and a strict blood pressure protocol. It shows that general anesthesia is not harmful when delivered this way; it cannot show that it is safe when induction is improvised by whoever is available at 2 a.m.

AnStroke

Löwhagen Hendén P, Rentzos A, Karlsson JE, et al. General anesthesia versus conscious sedation for endovascular treatment of acute ischemic stroke: the AnStroke trial (Anesthesia During Stroke). Stroke. 2017;48(6):1601-1607. No association

Design: Single-centre randomized trial, Gothenburg
Population: 90 adults with acute anterior circulation ischemic stroke undergoing thrombectomy
Intervention: General anesthesia
Control: Conscious sedation
Primary: Modified Rankin Scale at 3 months
Result: No significant difference in functional outcome at 3 months between the two techniques.

Why the blood pressure protocol matters This trial was explicit about maintaining systolic pressure during induction, and that protocol is plausibly the reason its general anesthesia arm did not suffer. It is evidence about a technique delivered under a rule, not about general anesthesia in the abstract.

GOLIATH

Simonsen CZ, Yoo AJ, Sørensen LH, et al. Effect of general anesthesia and conscious sedation during endovascular therapy on infarct growth and clinical outcomes in acute ischemic stroke: a randomized clinical trial. JAMA Neurol. 2018;75(4):470-477. No association

Design: Single-centre randomized trial, Aarhus
Population: 128 adults with acute ischemic stroke from anterior circulation large-vessel occlusion
Intervention: General anesthesia
Control: Conscious sedation
Primary: Infarct growth on MRI between baseline and 24 hours
Result: No significant difference in infarct growth. The 90-day modified Rankin distribution shifted in favour of general anesthesia, and reperfusion success was higher in that arm.

The reperfusion confounder, running the other way Better reperfusion in the general anesthesia arm makes the functional signal hard to attribute to the anesthetic itself — a still patient is easier to recanalize. That is a real mechanism, but it means the trial supports "general anesthesia does not harm" more strongly than it supports "general anesthesia helps."

Synthesis. Three randomized trials totalling 368 patients compared general anesthesia with conscious sedation for thrombectomy. None found the harm the observational literature had predicted, and two showed 90-day functional signals favouring general anesthesia. The trials are small and each is single-centre, so this is not the same weight of evidence as a multi-centre programme — but it is the only evidence in which the anesthetic was assigned rather than chosen.

The physiology that motivated avoidance was never wrong. Induction can drop perfusion pressure, and collateral flow to the penumbra is pressure-dependent in exactly the way the old argument claimed. What the trials establish is that this is a manageable hazard rather than an intrinsic property of the technique, and that the managing is done with a blood pressure target and a vasopressor drawn up before induction, not by avoiding the airway.

04

What you actually do

in the room

Treat blood pressure, not technique, as the variable you are actually controlling. Every trial that found general anesthesia safe protocolized pressure. That is the active ingredient, and it is the part that transfers to your institution.
Have the vasopressor drawn up and running before you induce. The dangerous interval is induction to recanalization, and it is short enough that reacting to a pressure drop is already too late.
Set an explicit pre-induction floor and say it out loud to the room. A target nobody has stated is a target nobody will defend when the interventionalist asks you to stop moving the table.
Do not let the anesthetic delay reperfusion. Time to reperfusion outweighs anything in this day. An elegant induction that costs fifteen minutes has given back more than the technique choice could ever win.
Choose general anesthesia when the patient cannot hold still or protect the airway. This is the situation the observational data was actually describing, and the trials remove the reason to resist it.
Avoid the emergent mid-procedure conversion if you can foresee it. Intubating an agitated patient with a catheter in the intracranial circulation combines the hemodynamic hit with the worst possible timing. Deciding early is better than deciding correctly.
Plan emergence around the neurological examination. The exam after reperfusion is the outcome measure everyone downstream will use. An emergence that obscures it for two hours has a real cost.

05

The pitfall

Inheriting an association

The characteristic error is reading a consistent observational association as a causal claim, and then carrying the resulting practice forward long after it has been tested directly.

It is worth being precise about why the cohorts came out the way they did, because the mechanism is not fraud or sloppiness. Anesthesiologists intubate the patients who cannot lie still, cannot protect an airway, or arrive with a devastating deficit. Those same features predict a poor 90-day outcome through the stroke itself. The anesthetic sits downstream of severity and upstream of nothing, and it collects the credit for an outcome that severity produced. No amount of statistical adjustment fixes this reliably, because the clinician's judgement that drove the assignment is not fully captured in any variable the registry recorded.

Watch for the same shape elsewhere in neuroanesthesia: it is the same error as reasoning from correct physiology to an untested clinical action, only arriving from the opposite direction. There the mechanism was real and the outcome was never tested. Here the association was real and the mechanism was assumed.

06

Sources

primary first

01 Schönenberger S, Uhlmann L, Hacke W, et al. Effect of conscious sedation vs general anesthesia on early neurological improvement among patients with ischemic stroke undergoing endovascular thrombectomy: a randomized clinical trial. JAMA. 2016;316(19):1986-1996. doi 10.1001/jama.2016.16623 UTMCK proxy
02 Löwhagen Hendén P, Rentzos A, Karlsson JE, et al. General anesthesia versus conscious sedation for endovascular treatment of acute ischemic stroke: the AnStroke trial (Anesthesia During Stroke). Stroke. 2017;48(6):1601-1607. doi 10.1161/STROKEAHA.117.016554 UTMCK proxy
03 Simonsen CZ, Yoo AJ, Sørensen LH, et al. Effect of general anesthesia and conscious sedation during endovascular therapy on infarct growth and clinical outcomes in acute ischemic stroke: a randomized clinical trial. JAMA Neurol. 2018;75(4):470-477. doi 10.1001/jamaneurol.2017.4474 UTMCK proxy

Can't get the full text? The doi link always reaches the publisher's record. For one-click access through Preston holdings, install LibKey Nomad once — it follows you between devices instead of depending on which link you clicked.

07

Check yourself

3 items · one attempt each

Item 1 Analysis Neuro 2 · Day 5

Multiple large registries found that thrombectomy patients receiving general anesthesia had worse 90-day outcomes than those receiving conscious sedation, yet three randomized trials found no such disadvantage. What best explains the discrepancy?

Item 2 Application Neuro 2 · Day 5

Your institution cites SIESTA, AnStroke and GOLIATH to justify a new protocol allowing general anesthesia for thrombectomy. Which element of those trials most needs to be reproduced for that conclusion to hold?

Item 3 Recall Neuro 2 · Day 5

During thrombectomy under general anesthesia, which period carries the greatest hemodynamic threat to the ischemic penumbra, and what follows for how you prepare?