What this is

The ZODIAC trial randomized 92 large-vessel-occlusion stroke patients awaiting thrombectomy to 0° vs 30° head-of-bed. Early neurological deterioration (≥2 NIHSS points) occurred in 1/45 patients at 0° vs 26/47 at 30° (HR 34.40, 95% CI 4.65–254.37, P < .001), and 90-day mortality was 4.4% vs 21.7% (P = .03). A free, reversible positioning maneuver appears to halve the rate of pre-thrombectomy deterioration — if the result holds up at scale.

Shadow Audit

What the article is NOT saying. What claims are presented without evidence. What population is excluded. What comparison is missing.

Inversion Engine

What would have to be true for the opposite conclusion to be the right one. What mechanism would need to be wrong. What assumption breaks first.

What would have to be true for the opposite conclusion to be the right one. What mechanism would need to be wrong. What assumption breaks first.

If head position is irrelevant to LVO physiology, the only explanation for 1 vs 26 events is some unmeasured confounder. The 97-minute consent-time gap is the obvious candidate — the 0° arm may have been sicker or different in ways not captured at baseline. The assumption that breaks first is the integrity of randomization given such large time imbalance.If 30° head elevation is a protective maneuver for a subgroup, then flat positioning could be net-harmful in patients with impaired autoregulation, raised ICP, or posterior circulation occlusions. Posterior circulation strokes (3.26% of cohort, n=3) are too few to refute this. The trial explicitly excluded “conditions challenging tolerance of 0° head positioning” — so the published finding is conditional on a clinical filter the reader must replicate.If the 30° group’s deterioration is iatrogenic from positioning itself (e.g., orthostatic hypoperfusion unmasked by the protocol), then 0° “works” by avoiding harm, not by augmenting penumbral flow. The mechanistic claim in the introduction (“positional cerebral ischemia… coined by Caplan and Sergay in 1976”) is much stronger than what the data support. The data only show that flat beats elevated on a short-term NIHSS endpoint.If the 90-day functional and mortality signals are chance (note: P = .03 mortality, P = .22 functional independence, 90-day dichotomized mRS not significant), then the entire practice change rests on a 10–12 event mortality outcome. Replication in a larger cohort is required to know if the long-term benefit is real.The early-stopping DSMB decision is also a fragility flag. The trial was halted at the second interim with n=92 of 182 planned. The pre-specified stopping rule (z > 2.963 efficacy) was met, but with such a wide CI on the primary (4.65–254.37) and tiny event count, a single patient reclassified would change the conclusion. The fact that 1/45 deteriorated in 0° vs 26/47 in 30° is a very clean split — too clean, perhaps.## Second-Order Catalyst The downstream consequence if the result is real. Who changes practice first. Who changes it last. What does the field do in 12 months.

The downstream consequence if the result is real. Who changes practice first. Who changes it last. What does the field do in 12 months.

First movers (3–6 months): Mobile stroke units and comprehensive stroke centers with long door-to-puncture times (rural transfer, drip-and-ship). These are the exact patients with the longest “head position” exposure window where a 50%+ reduction in END is the largest absolute gain. EMS protocols would update first — prehospital positioning of suspected LVO is currently 30° by AHA default (“head of bed elevated for suspected stroke, except when hypotension suspected”).Guidelines (6–12 months): AHA/ASA would likely issue a Class IIa, Level of Evidence B-R (single RCT) focused recommendation for 0° positioning in confirmed LVO patients awaiting thrombectomy, conditional on airway/pulmonary safety. The 2018 HeadPoST Class IIb recommendation of 0–30° would be narrowed. ESO would mirror.Last movers (12–24 months): Individual neurointerventionalists, particularly those who already do procedures at 0° for anatomical reasons. Nursing workflow changes (head-of-bed alarms, aspiration precautions) lag guideline changes by 12–18 months in most US hospitals.**Second-order consequences:**Aspiration pneumonia rates will be tracked prospectively as the safety counter-signal. If even a 1–2% excess pneumonia rate emerges with widespread 0° implementation, the practice will reverse. The trial’s zero-pneumonia finding in n=92 is a fragility point — the upper 95% CI for pneumonia in the 0° arm is around 8%.Pre-hospital triage algorithms (RACE, LAMS, VAN) might integrate “head position tolerance” as a transfer-decision modifier.Trial-design spillover: Future LVO trials will need to control for head position as a confounder — every ongoing thrombectomy adjunctive trial (e.g., tenecteplase, nerinetide, cooling) has a new covariate to manage.The 90-day mortality signal, if confirmed, would feed into cost-effectiveness models and quality metrics for thrombectomy-capable centers (door-to-puncture plus head-position fidelity).Asymmetry: if real, the practice is essentially free (no drug, no device, no capital) — making adoption low-regret even with weak evidence. If not real, the only cost is rare aspiration events in a small subset.## Asymmetric Leverage Where is the asymmetric payoff? What is the largest claim that, if true, gives the most leverage? What is the smallest claim that costs the most if wrong?

Where is the asymmetric payoff? What is the largest claim that, if true, gives the most leverage? What is the smallest claim that costs the most if wrong?

Largest claim, if true: 0° positioning before thrombectomy reduces 90-day all-cause mortality by ~17 absolute percentage points (21.7% → 4.4%, NNT ≈ 6). This would make head positioning one of the highest-leverage interventions in the entire LVO pathway, comparable to thrombectomy itself in absolute terms — and free.Smallest claim that costs the most if wrong: “30° head elevation is actively harmful in LVO.” If adopted at scale and 30° is later shown to be neutral (per HeadPoST), then 0° implementation across hundreds of thousands of LVO patients per year will have caused rare but real aspiration/pulmonary events in patients who would have tolerated 30° fine. The ZODIAC exclusion criteria (vomiting, pneumonia, pulmonary edema) define a population that the field cannot perfectly replicate in real-world practice.**Leverage asymmetry:**Benefit side: 1 line in every LVO protocol, no cost, immediate effect, applies to every confirmed LVO patient awaiting thrombectomy — leverage on the order of “give aspirin for TIA” or “time is brain.”Harm side: Aspiration in patients with borderline airway protection, unrecognized posterior circulation stroke with raised ICP, operator delay if a flat patient takes longer to position on the angio table. The paper notes “Head position protocol termination for catheterization table movement was fastest in patients with a head position at 30°… At 50 minutes, 7 patients in 0° group had not moved to catheterization” — workflow friction is real.Net leverage: asymmetrically positive. The intervention is reversible (just elevate the head), cheap, and the ceiling of benefit (mortality reduction) is enormous. The downside is bounded by patient selection.## Paradigm Destroyer What does this paper kill? What prior belief or practice is now untenable? Be specific.

What does this paper kill? What prior belief or practice is now untenable? Be specific.

Killed: The 2018 HeadPoST-derived belief that “head position is irrelevant to acute stroke outcome.” While HeadPoST was a heterogeneous population (all ischemic stroke, not specifically LVO awaiting thrombectomy), the field has used it to defer head-position protocols. ZODIAC makes that deferral untenable in the LVO-to-thrombectomy interval.Killed: The default EMS and ED practice of elevating the head of bed to 30° in all suspected strokes. At minimum, this is no longer appropriate in confirmed LVO awaiting thrombectomy. The “30° is safer for aspiration” assumption, dominant in nursing education, must be re-examined in the LVO subgroup.Did not kill: The belief that the dominant lever in LVO is time-to-reperfusion. The 0° arm consented 97 minutes later than 30° and still won on END. If anything, the trial says “if you cannot accelerate reperfusion, at least don’t make the patient deteriorate while you wait.”Did not kill: The generalizability of “flat is better” to non-LVO strokes. Lacunar, posterior circulation, and stroke mimics are not addressed.Did not kill: The thrombectomy effect size or the imaging-based patient selection. The trial’s intervention is a positioning adjunct, not a substitute.## MVP — Minimum Viable Proof The cheapest test (single trial, single biomarker, single dataset) that would change the paper’s evidentiary weight if granted.

The cheapest test (single trial, single biomarker, single dataset) that would change the paper’s evidentiary weight if granted.

TCD-mechanistic confirmatory study in 50 LVO patients awaiting thrombectomy: measure middle cerebral artery mean flow velocity (MFV) and pulsatility index (PI) on the affected side at 0° vs 30° crossover, 5 minutes per position, before catheterization. If the 0° arm shows a ≥15% MFV increase and PI drop, the mechanistic claim is mechanistically vindicated and the END reduction has a physiological substrate. If MFV is unchanged, the END signal becomes more suspect (unmeasured confounder).A larger pragmatic trial (n≈400) at 20 centers, with 90-day mRS as the primary endpoint and pre-specified aspiration/pneumonia safety, would resolve the 90-day functional non-significance. Cost: ~$4–6M. Time: 3 years.A registry analysis of Get With The Guidelines-Stroke or similar, comparing pre- vs post-protocol implementation at hospitals that adopt 0° positioning, would give a real-world effectiveness signal in months, not years, at near-zero marginal cost.## Best Combination The single move with the highest asymmetric payoff based on the 5 frameworks.

The single move with the highest asymmetric payoff based on the 5 frameworks.

Adopt 0° head-of-bed positioning for confirmed LVO patients awaiting thrombectomy, with explicit safety exclusion criteria (vomiting, active pulmonary disease, unprotected airway, posterior circulation with raised ICP suspicion) and a documented aspiration-precaution nursing bundle (head-of-bed alarm, suction at bedside, NPO except medications, hourly lung auscultation). Combined with a 12-month institutional audit of aspiration pneumonia incidence and 90-day mRS.Cost: zero, no new equipment.Reversibility: high (just elevate).Asymmetry: massive — if HeadPoST was right and ZODIAC was a 92-patient fluke, the harm is a small increase in monitored aspiration events; if ZODIAC is right, you’ve added a 17-point mortality reduction to a previously unmodifiable moment in the LVO pathway.The combination matters: a naked 0° recommendation would invite the very aspiration events the exclusion criteria protect against. The win is the bundle, not the angle.## Overvalue Warning Where the analysis is most likely to overstate the result. Bias to watch for. Sample size, novelty, recency, mechanism plausibility.

Where the analysis is most likely to overstate the result. Bias to watch for. Sample size, novelty, recency, mechanism plausibility.

Small-sample fragility. n=92 with a binary primary outcome. The 95% CI on the hazard ratio (4.65–254.37) is so wide that the lower bound is consistent with a clinically trivial effect, and the upper bound is essentially “impossibly large.” Replication in a larger cohort is essential before guideline-level practice change.Early-stopping inflation. The DSMB halted the trial at the second interim for efficacy. Stopped-early trials systematically overstate treatment effects (the “winner effect bias”). The 0°/30° split of 1/26 is too clean — a single reclassification would have changed the HR by an order of magnitude.Novelty / recency bias. Head-of-bed positioning is a cheap, intuitive, visually dramatic intervention. The “lay-flat” idea has been floating in TCD/transcranial Doppler circles for a decade. There is a halo effect around positive, mechanism-plausible findings from respected groups (the senior author of ZODIAC is also the senior author on the prior TCD pilot work — investigator allegiance is real and documented in the field).Mechanism plausibility trap. Positional cerebral ischemia sounds physiologically real, but the trial did not measure flow. Readers will assume the mechanism without evidence.Time-imbalance confounder not addressed. The 97-minute consent-time gap favoring 30° should bias against 0° on END, but the 30° arm is the one that deteriorated. The discussion does not wrestle with this directly — a healthy sign of analytic caution is missing.Mortality signal is an exploratory outcome. The paper’s primary endpoint is END. The 90-day mortality difference (P = .03) is from a small event count and a wide CI (1.20–28.22). Treating mortality as a “headline” finding is a Type-I error risk in a trial already stopped early.Single geographic region (US), 12 hospitals, mostly comprehensive stroke centers. The intervention may behave differently in drip-and-ship networks, mobile stroke units, or low-resource settings where patient monitoring during the pre-thrombectomy interval is less intensive.## Source Alexandrov AW, Shearin AJ, Mandava P, et al. (for the ZODIAC Investigators). Optimal Head-of-Bed Positioning Before Thrombectomy in Large Vessel Occlusion Stroke: A Randomized Clinical Trial. JAMA Neurology. 2025.

DOI: https://doi.org/10.1001/jamaneurol.2025.2253PMID: 40465238PMC full text: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12138796/