You're on scene. Patient's smiling, but the left side of their face doesn't quite move. Arm drift? Negative. Speech? Day to day, clear as day. Just the face.
Now what?
If you've run EMS calls for more than a week, you've had this exact moment. One positive finding on the Cincinnati Prehospital Stroke Scale. In real terms, not two. That's why not three. In real terms, just one. And you're wondering — is that enough to call a stroke alert? Transport lights and sirens? Wake up the neurologist at 2 a.m.?
Quick note before moving on.
Short answer: yes. But the why matters. And most people — even seasoned medics — don't fully understand what that single abnormal finding actually tells you.
What Is the Cincinnati Prehospital Stroke Scale
The Cincinnati Prehospital Stroke Scale — CPSS for short — is a three-item neurological screening tool designed for speed. It was derived from the NIH Stroke Scale back in the late 1990s by researchers at the University of Cincinnati who needed something paramedics could perform in under a minute, on a stretcher, in a moving ambulance, with a patient who might be confused, combative, or just plain uncooperative Not complicated — just consistent..
Three tests. That's it.
Facial droop
Ask the patient to smile or show their teeth. Look for asymmetry. One side doesn't move? That's abnormal.
Arm drift
Have them close their eyes and hold both arms out straight, palms up, for 10 seconds. If one arm drifts down, pronates, or doesn't move at all — abnormal.
Speech
Ask them to repeat a simple phrase. "You can't teach an old dog new tricks" is the classic. Slurred words, wrong words, or no words at all — abnormal Easy to understand, harder to ignore. No workaround needed..
Each component gets scored as normal or abnormal. Day to day, that's the whole scale. But no points. Think about it: no grading. Just yes or no, three times Simple, but easy to overlook..
The original validation study, published in Annals of Emergency Medicine in 1999, showed that any single abnormal finding carried a positive predictive value of around 72% for stroke. Two or three abnormal findings pushed that higher. But here's the kicker — even one finding was enough to justify stroke team activation.
That hasn't changed.
Why It Matters / Why People Care
Stroke is time-sensitive in a way few other emergencies are. 5 hours from last known well. On top of that, mechanical thrombectomy? 9 million neurons. The IV tPA window — 3 to 4.Every minute of large-vessel occlusion costs the brain roughly 1.Up to 24 hours in select patients, but only if you identify them fast Nothing fancy..
The CPSS exists because we needed a tool that worked in the real world. Also, m. Even so, a dark bedroom at 3 a. A nursing home hallway. Not a neurology clinic. Not a research setting. The back of a moving rig.
And it works. Meta-analyses consistently show sensitivity in the high 80s to low 90s. Now, specificity sits lower — around 70% — but that's by design. We'd rather over-triage a few stroke mimics than miss a real one Most people skip this — try not to..
The problem? People forget that one abnormal finding is a positive screen. " Not "keep an eye on it.On top of that, not "maybe. " Positive. Full stop.
I've seen medics document "CPSS 1/3" and then write "low suspicion for stroke" in their narrative. That's not how this works. The scale doesn't have a "low suspicion" setting. Worth adding: it's binary. Positive or negative. And positive means stroke alert until proven otherwise Not complicated — just consistent..
How It Works (or How to Do It Right)
You think you know this. You've done it a hundred times. But I've watched plenty of providers rush through it and miss the subtle stuff. Let's break down each component the way it's supposed to be done — not the way it looks when you're running late for shift change.
Facial droop — look closer than you think
"Smile for me." Simple, right?
Watch the nasolabial folds. The forehead — though forehead sparing suggests upper motor neuron lesion, which is what we're hunting for, but don't get distracted by that distinction in the field. The corners of the mouth. Just look for asymmetry.
Pro tip: Ask them to show teeth and puff cheeks. Two different cranial nerve pathways. Facial nerve (CN VII) handles both, but upper face gets bilateral innervation. A central lesion spares the forehead. A peripheral lesion (Bell's palsy) doesn't. In the field, you're not diagnosing Bell's vs. stroke. You're flagging asymmetry. But if the forehead moves symmetrically and only the lower face droops, that's more specific for stroke, not less.
And please — don't just glance. Good lighting. Sit at eye level. Take three seconds. It feels like forever. Do it anyway.
Arm drift — the 10-second rule exists for a reason
"Close your eyes. Hold your arms out straight, palms up. Don't let them move.
Count out loud. Ten full seconds. Not "one Mississippi, two Mississippi" while you're grabbing the BP cuff. Actual seconds.
What you're watching for:
- Pronation (palm turns inward)
- Downward drift
- Complete inability to maintain position
- Asymmetric effort — one arm shaking, the other rock steady
Common cheat: Patient keeps eyes open, watches their own arms, corrects subconsciously. That's why eyes closed matters. If they refuse to close them, document it. "Patient unable/unwilling to close eyes — drift assessed with eyes open."
And if they can't follow commands? Here's the thing — try the "arm drop" method — lift both arms yourself, let go simultaneously, watch what happens. Not perfect, but better than nothing.
Speech — use the phrase, not your own words
"You can't teach an old dog new tricks."
Why that phrase? Day to day, it's standardized. On top of that, it's been validated. Plus, it tests articulation, language content, and fluency in one go. Your made-up sentence — "The sky is blue today" — hasn't And that's really what it comes down to..
Listen for:
- Dysarthria (slurred, effortful, mechanical — motor problem)
- Aphasia (wrong words, empty speech, can't repeat — language problem)
- Mutism (no output at all)
Here's what most people miss: A patient with expressive aphasia might understand you perfectly but can't get the words out. They'll look frustrated. They might gesture. Don't mistake that for confusion or non-compliance. That's a positive speech finding.
And if they're intubated? Non-verbal baseline? Document "unable to assess speech — intubated" or "baseline non-verbal per family." Don't leave it blank. Blank looks like you forgot Small thing, real impact..
What One Abnormal Finding Actually Means
This is the section that changes how you practice.
The math behind the "any one" rule
The original Koth
The original Kothari validation study gave us the numbers: any single abnormal finding — face, arm, or speech — yields roughly 87% sensitivity for anterior circulation stroke. That means if your patient has one positive sign, there's a high probability they're having a stroke. But 13% of strokes will have a completely normal CPSS. On the flip side, specificity sits around 70%. Zero findings does not rule it out And it works..
Worth pausing on this one.
Posterior circulation strokes — cerebellar, brainstem, vertebrobasilar — often present with none of the three. That's not a flaw in the tool. The CPSS misses them by design. It's a boundary. Now, vertigo, ataxia, diplopia, dysphagia, crossed findings. Know the boundary.
The "negative" exam that isn't
You've seen this patient. Or they're vomiting with hiccups. Because of that, normal speech. Think about it: normal face. Worth adding: normal arms. But they're "just not right." Family says they're confused. Or they can't walk but strength is 5/5 Practical, not theoretical..
Document the negative exam explicitly. "CPSS negative: symmetric smile, no arm drift at 10 seconds, repeats phrase without error." Then document why you're still concerned. "Persistent vertigo, direction-changing nystagmus, unable to tandem walk — high suspicion for posterior circulation stroke despite negative CPSS."
That documentation protects the patient. On top of that, it protects you. And it forces the receiving team to think beyond the checklist.
Handoff: the exam lives or dies in the transfer
You did the work. Even so, you sat at eye level. But you counted ten real seconds. That said, you used the validated phrase. You documented the forehead sparing. You noted the eyes-open cheat.
Now hand it off like it matters Small thing, real impact..
"CPSS positive for right facial droop (lower face only, forehead spared), right arm pronation at 6 seconds, mild dysarthria — repeats phrase with slurring but correct words. Last known well 06:45. On apixaban. Glucose 112. GCS 14 (confused to time)."
That's a handoff. Not "stroke alert, positive CPSS.Still, " The details change triage. Consider this: they change imaging priority. They change whether the neurologist meets the patient at the door or finishes their coffee first.
The Bottom Line
The three-item stroke screen isn't a diagnostic test. Worth adding: it's a trigger. A structured, validated, reproducible trigger that forces you to look at the right things, in the right way, every single time — even at 0300, even on the "frequent flyer," even when the radio is crackling and the family is screaming and your partner is already pulling the cot Nothing fancy..
Asymmetry is the signal. Which means standardization is the filter. Documentation is the receipt That's the part that actually makes a difference..
Do the exam. Do it right. Write it down It's one of those things that adds up..
The patient can't afford for you to be casual about the only tool that catches the treatable ones in time.
