The prehospital tools most often used to flag a likely large vessel occlusion (LVO) are the Cincinnati Prehospital Stroke Severity Scale (CPSSS), the Los Angeles Motor Scale (LAMS), and the Rapid Arterial oCclusion Evaluation (RACE). All three are scored by paramedics in the field in under two minutes, and each uses a threshold rather than a single "positive" result. None of them diagnoses LVO; they sort patients toward the right destination, usually a thrombectomy-capable center.

The distinction matters because the destination decision is the whole point. A tool that over-calls sends stroke mimics and small-vessel strokes on long bypass journeys, delaying thrombolysis. A tool that under-calls leaves a treatable occlusion at a hospital without neurointervention. Every cutoff below is a trade-off between those two errors, and the trade-off shifts with the prevalence of LVO in the population your EMS system serves.

What CPSSS, LAMS, and RACE actually measure at the roadside

Each scale reduces a neurological exam to a handful of items a paramedic can score without a neurologist on the line. CPSSS, published by Katz and colleagues in 2015, scores three items: conjugate gaze deviation, arm weakness, and whether the patient can hold up both arms for ten seconds or is globally aphasic. The maximum is 4, and a score of 2 or more is the usual positive cutoff.

LAMS, described by Nazliel and colleagues in 2008, adds a grip item to the Los Angeles Prehospital Stroke Screen. It scores facial droop, arm drift, and grip strength, each 0 to 2, for a maximum of 5. A score of 4 or higher is the conventional LVO flag.

RACE, from Pérez de la Torre and colleagues in 2016, is the longest of the three at five items: facial palsy, arm motor function, leg motor function, gaze deviation, and either aphasia or neglect depending on which hemisphere is affected. It totals 9, and a score of 5 or more triggers the LVO alert in most published protocols.

The practical difference is exam burden versus discrimination. CPSSS takes the least time and misses more posterior and distal occlusions. RACE takes longer and picks up more, at the cost of more false positives. LAMS sits between them and is the easiest to teach to a rotating crew.

The cutoff number that decides whether a patient bypasses the nearest hospital

Cutoffs are where these tools earn or lose their keep, and the published values are specific. CPSSS at 2 or more, LAMS at 4 or more, RACE at 5 or more. Those are the numbers your protocol document should name, because a crew that remembers "LAMS is positive at 3" will over-triage by a wide margin.

Sensitivity and specificity vary by study population, and the honest statement is that no single scale dominates across all cohorts. A 2017 systematic review by Smith and colleagues in Stroke compared several prehospital scales and found broadly similar discrimination, with differences driven more by prevalence and EMS training than by scale design. Citing one scale as definitively superior overstates the evidence.

What determines the range is the mix of strokes in your region. A service with a high proportion of cardioembolic and proximal MCA occlusions will see better positive predictive value at any cutoff than a service where lacunar strokes dominate. That is arithmetic, not opinion: predictive value rises with prevalence.

Tool Items scored Maximum Usual LVO cutoff Field time Main weakness
CPSSS Gaze, arm, consciousness/aphasia 4 2 or more Under 1 minute Misses posterior circulation
LAMS Facial droop, arm drift, grip 5 4 or more About 1 minute Grip scoring varies between crews
RACE Face, arm, leg, gaze, cortical sign 9 5 or more 1–2 minutes More training needed for cortical items

Read the table as a menu, not a ranking. If your crews turnover fast and training hours are scarce, LAMS or CPSSS keeps inter-rater agreement higher. If your system already routes suspected LVOs to a comprehensive center and can absorb false positives, RACE's extra sensitivity is worth the exam time.

The Seattle failure case: when a negative screen hides a basilar occlusion

The clearest edge scenario is posterior circulation stroke, and it is the one most worth rehearsing. A basilar artery occlusion can present with vertigo, diplopia, gait instability, or fluctuating consciousness, and all three scales will score it low because none of them tests gait, eye movement beyond gaze deviation, or truncal ataxia. A patient with a catastrophic posterior occlusion can walk into the ambulance with a CPSSS of 1.

Published case series of basilar occlusion repeatedly describe delayed recognition, and the prehospital scales are part of why. If your protocol treats a negative screen as "no LVO, go to the nearest hospital," that patient loses hours. The fix is not a better score; it is a protocol escape hatch for any patient with sudden severe vertigo, new diplopia, or unexplained coma, routed to a center with CT angiography regardless of score.

The second edge case is the patient who cannot be scored at all. Intubated, sedated, or aphasic in a way that prevents arm commands, and the gaze and grip items become uninterpretable. Document the unscored items explicitly rather than defaulting them to zero. A defaulted zero silently converts a possibly positive patient into a negative one, and that documentation gap is the kind of thing that surfaces in a case review months later.

How prevalence changes the meaning of a positive LAMS

Take a service where 10% of suspected strokes turn out to have LVO. A scale with 80% sensitivity and 80% specificity, run on 1,000 suspected strokes, flags 80 of the 100 true LVOs and 180 of the 900 non-LVOs. That is 260 alerts for 80 real occlusions, a positive predictive value of about 31%. Roughly two of every three activations are false.

Now raise prevalence to 25%, which happens in systems that pre-screen by EMS dispatch criteria before applying the scale. The same sensitivity and specificity on 1,000 patients flags 200 true positives and 150 false positives, for a positive predictive value near 57%. Same tool, same cutoff, very different operational burden.

This is why copying another region's cutoff without checking your own case mix is a mistake. The number on the protocol is only meaningful alongside the prevalence it was tuned for, and that figure comes from your own stroke registry, not from the journal article. Ask your stroke coordinator for the local LVO rate among EMS-transported suspected strokes before you defend a cutoff at a committee meeting.

Where the three scales diverge on the same patient

Consider a patient with dense left arm weakness, mild facial droop, and no gaze deviation, fully alert, grip weak on the left. LAMS scores 1 for face, 2 for arm, 1 for grip, totaling 4 and crossing the threshold. CPSSS scores 1 for arm and 0 for gaze and 0 for consciousness, totaling 1, which is negative. RACE scores facial palsy 1, arm 2, leg 1, gaze 0, and aphasia or neglect 0, totaling 4, just under its cutoff of 5.

Three tools, three answers, one patient. Which is right depends on whether that arm weakness reflects a proximal MCA occlusion or a lacunar infarct, and the field exam cannot tell you. This is the honest limit of prehospital severity scoring, and it is why protocols pair the score with clinical judgment and a destination policy rather than treating the number as a verdict.

If you want to run this comparison at your own service, pull twenty recent LVO-confirmed charts and score them retrospectively from the PCR narrative. You will find the divergence quickly, and you will learn which cutoff your crews actually apply in practice, which is often not the one in the protocol binder.

Building the destination protocol around the score

A working protocol names four things: the scale, the cutoff, the destination, and the override. The scale and cutoff are above. The destination is a thrombectomy-capable center, and the transport time penalty is the variable that decides whether bypass is justified. Most systems tolerate a bypass of roughly 15 to 30 minutes beyond the nearest hospital, though that figure is local policy, not a universal rule.

Compute the penalty before you adopt it. If the nearest hospital is 8 minutes away and the comprehensive center is 35, the bypass adds 27 minutes outbound. If the patient instead goes to the nearest hospital and is transferred, the door-in-door-out interval at a non-thrombectomy facility typically runs well over an hour in published series. That comparison, not the scale's sensitivity, is what justifies the bypass.

Put the override in writing. Any patient with sudden severe vertigo, new diplopia, unexplained coma, or a clinical picture the crew believes is stroke despite a negative score should be discussed with medical direction or routed by the posterior-circulation pathway. A protocol without that clause is betting that every LVO looks like an MCA occlusion, and some do not.

Training and inter-rater reliability are the real bottleneck

A scale is only as good as the agreement between the paramedic who scores it tonight and the one who scores it next Tuesday. Grip strength in LAMS and cortical signs in RACE are the items most prone to drift, and drift shows up as inconsistent destination decisions rather than as a formal error. Annual competency checks with video cases catch it better than a slide deck.

Build the training around disagreement rather than around the scale's history. Have crews score the same recorded exam and compare, then discuss the divergent items. Ten minutes per quarter with real cases does more for consistency than a two-hour initial lecture, and it surfaces the crews who are defaulting unscored items to zero.

Track your own numbers. Alert volume, confirmed LVO rate among alerts, and the interval from EMS arrival to groin puncture for bypassed patients are the three metrics that tell you whether the protocol is working. If confirmed LVO among alerts falls below roughly a third, your cutoff or your prevalence has drifted and the protocol needs review.

What the scales are not: a substitute for imaging or judgment

None of these tools confirms an occlusion. CT angiography at the receiving hospital does that, and the scales exist only to get the right patient to the right scanner first. Treating a positive score as a diagnosis leads to inappropriate thrombectomy alerts and, in some cases, to patients being told they have a clot before anyone has imaged them.

The scales also do not account for time from onset, baseline disability, or anticoagulant use, all of which change the treatment decision at the receiving center. A positive RACE in a patient who woke with symptoms eight hours ago means something different from the same score in a patient whose symptoms started forty minutes ago during a witnessed event.

Document score and item-level findings, not just the total. The receiving neurologist can interpret a gaze deviation plus arm weakness far better than a bare number, and the item detail is what lets your service audit its own accuracy later. A total alone is a data point; the items are the story.

FAQ

No single scale carries a universal mandate. The American Heart Association's guidance on EMS destination emphasizes routing suspected LVOs to thrombectomy-capable centers and leaves scale selection to regional protocols. Check your state EMS office and local stroke system's written policy for the mandated tool.

Can a patient with a negative LAMS still have a large vessel occlusion?

Yes, and posterior circulation strokes are the classic miss. A basilar occlusion can score 1 or 2 on LAMS because the scale does not test gait, diplopia, or truncal ataxia. Any patient with sudden severe vertigo, new double vision, or unexplained coma needs a posterior-circulation pathway regardless of the score.

How long does it take to score RACE compared with LAMS in the field?

RACE adds roughly 30 to 60 seconds over LAMS in published comparisons, mainly because of the leg motor and cortical sign items. That extra minute matters in a time-critical call, which is why some high-volume services accept LAMS's lower sensitivity to keep the scene interval short.

Does a positive score mean the patient needs thrombectomy?

It means the patient needs imaging at a center that can perform thrombectomy, which is a different statement. CT angiography, onset time, baseline function, and infarct size all feed the actual decision. Roughly two of every three positive alerts in a 10%-prevalence system turn out not to have an LVO.

Which tool is easiest for new paramedics to learn?

CPSSS has three items and the shortest training curve, and LAMS is close behind with three items and simple scoring. RACE's five items, particularly the aphasia-versus-neglect branch, take more practice to score consistently across a rotating crew.

What happens if a crew cannot score the gaze item because the patient is intubated?

Document the item as not assessable rather than scoring it zero. A defaulted zero can flip a positive screen to negative. Discuss the patient with medical direction and consider routing to a center with immediate CT angiography on clinical grounds.

Does a higher maximum score mean the scale is more accurate?

No. RACE's 9-point maximum reflects more items, not better discrimination. Head-to-head comparisons have found broadly similar accuracy across CPSSS, LAMS, and RACE, with differences driven by local LVO prevalence and crew training more than by scale design.

This article is general educational information about prehospital stroke screening and is not a diagnosis, treatment plan, or EMS protocol. Scale selection, cutoffs, and destination policies vary by jurisdiction and must be set by your local EMS medical director and stroke system; consult those authorities and a qualified clinician before changing practice. For a patient with active stroke symptoms, call local emergency services immediately rather than relying on any score described here.