Cincinnati Prehospital Stroke Scale (CPSS)

The Cincinnati Prehospital Stroke Scale (CPSS) is a three-item bedside screen — facial droop, arm drift, and abnormal speech — designed to let a paramedic or EMT decide in under a minute whether a patient is likely having a stroke and should trigger a stroke alert. It is deliberately not a severity score, not a diagnosis, and not a substitute for the full neurological examination performed in hospital. Its entire design goal is a fast, reproducible yes/no gate at the front of the acute stroke pathway. This guide covers the three components and how each is administered, what the published accuracy figures actually say (and where they are routinely overstated), how the CPSS differs in purpose and scope from the NIH Stroke Scale (NIHSS), and how patient-safety and quality staff should treat CPSS documentation when auditing prehospital notification and door-to-needle performance.

What the CPSS is, and where it came from

The CPSS was developed at the University of Cincinnati and published by Kothari and colleagues in Annals of Emergency Medicine in April 1999 as "Cincinnati Prehospital Stroke Scale: Reproducibility and Validity." It was built as an explicit simplification of the NIHSS: the authors took the NIHSS items that were both quickest to perform and most reliably scored by non-neurologists, and reduced them to three.

That lineage matters, and it is a point of genuine confusion in stroke programmes. The CPSS and the NIHSS come from the same institution and the same research group — the NIHSS itself originated with Brott, Davis and colleagues at Cincinnati in the late 1980s for the NINDS rt-PA Stroke Study, and Brott is a co-author on the CPSS paper. They are relatives, not rivals, and they were never meant to do the same job. One is a field triage trigger; the other is a graded severity instrument for treatment decisions and registry reporting.

The three components

All three items are scored simply as normal or abnormal. There is no 0–4 gradation, no weighting, and no total score in the sense the NIHSS has one.

1. Facial droop

How it is elicited: ask the patient to smile or to show their teeth.

Normal — both sides of the face move equally.
Abnormal — one side of the face does not move as well as the other.

The comparison is symmetry, not absolute strength. A patient with a pre-existing facial asymmetry (an old Bell’s palsy, prior stroke, or facial surgery) will screen abnormal on this item regardless of whether anything acute is happening — one of several reasons the scale is a screen rather than a diagnosis, and a reason baseline history belongs in the handover.

2. Arm drift

How it is elicited: ask the patient to close their eyes and hold both arms straight out in front of them, palms up, for roughly ten seconds.

Normal — both arms move the same, or both arms do not move at all.
Abnormal — one arm does not move, or one arm drifts down relative to the other.

The eyes-closed instruction is not decorative: it removes visual correction, which is what makes a subtle pronator drift appear. Two symmetrically weak arms score normal on this item, because the item is testing for lateralised deficit, not for weakness in general. This is the single most commonly misapplied instruction in the scale.

3. Abnormal speech

How it is elicited: ask the patient to repeat a standard sentence — the phrase used in the original validation work is "You can’t teach an old dog new tricks."

Normal — the patient uses the correct words with no slurring.
Abnormal — the patient slurs words, uses the wrong words, or is unable to speak.

Note that this single item collapses two neurologically distinct problems: dysarthria (a motor articulation deficit) and aphasia (a language deficit). The NIHSS separates these into item 9 (Best Language) and item 10 (Dysarthria) precisely because the distinction carries localising information. The CPSS deliberately discards that distinction to keep the screen fast. If a prehospital record captures only "speech abnormal," the receiving team learns that something is wrong with speech and nothing about what.

What counts as a positive screen

The CPSS is interpreted on an any-one-of-three rule: if any single item is abnormal, the screen is positive and stroke should be suspected. It is not a scored threshold and there is no "two of three" requirement in the original scale. Some EMS systems record the number of abnormal items (0–3) as a crude severity proxy, which is a local practice layered on top of the instrument rather than part of it — if your protocol does this, say so explicitly in the protocol document rather than letting it look like validated scale behaviour.

How the CPSS differs from the NIH Stroke Scale

These two instruments are frequently treated as interchangeable in documentation audits, which produces misleading findings in both directions. The differences that actually matter operationally:

Question answered. CPSS: "is this plausibly a stroke, yes or no?" NIHSS: "how severe is the deficit, and where?"
Items. CPSS: 3 items, binary. NIHSS: 11 numbered items split into 15 individually scored components, each with defined behavioural anchors.
Output. CPSS: positive or negative. NIHSS: a 0–42 total, plus the item-level pattern, which carries more clinical meaning than the total does.
Time to administer. CPSS: well under a minute, performable at the roadside or in a moving ambulance. NIHSS: several minutes, requiring a cooperative patient and controlled conditions.
Who performs it. CPSS: EMTs and paramedics, no formal certification requirement. NIHSS: carries a formal training-and-certification expectation that stroke programmes are expected to track for individual examiners.
Where it lands in the record. CPSS: prehospital patient care report and the stroke-alert activation trigger. NIHSS: the treatment-eligibility record, the thrombolysis/thrombectomy documentation, and the certification and registry data set.
What a change in it means. CPSS: nothing — it is not designed for serial monitoring. NIHSS: serial scores are the standard way deterioration or response to treatment is documented.

The practical rule for a quality reviewer: a CPSS is not a "lightweight NIHSS" and a positive CPSS does not substitute for an NIHSS anywhere in the pathway. If your abstraction process is pulling a CPSS result into a field that expects a severity score, that is a data-definition defect, not a clinician documentation failure. The NIHSS guide covers the severity side of that pathway in detail.

CPSS, FAST, BE-FAST — and the scale that is not the CPSS

FAST (Face, Arm, Speech, Time) is the public-facing mnemonic built on the same three examination components, with "Time" added as a call-to-action prompt rather than a fourth exam item. Functionally, a FAST assessment and a CPSS assessment test the same three things; FAST is the lay/awareness framing, CPSS is the clinical instrument with a validation literature attached to it. BE-FAST prefixes Balance and Eyes, specifically to catch the posterior-circulation presentations the three-item screen misses.

Separately, and much more consequentially for anyone writing protocol documents: the Cincinnati Prehospital Stroke Severity Scale (CPSSS, also written C-STAT) is a different instrument from the CPSS. CPSSS is a large-vessel-occlusion prediction scale, intended to help decide whether a patient should bypass a nearer hospital for a thrombectomy-capable centre. The CPSS answers "is this a stroke;" the CPSSS answers "is this a large vessel occlusion." One extra S changes the clinical question, the destination decision, and the downstream quality metric. Spell both out in full on first use in every protocol and audit tool — the abbreviation collision is a real source of routing error, and it is not caught by spellcheck.

Reported accuracy — the figures, with their caveats

In the original 1999 validation, physician observation of an abnormality in any one of the three items had a sensitivity of 66% and specificity of 87% for identifying a stroke patient. Restricted to anterior circulation strokes, sensitivity was 88%. Reproducibility between physicians and prehospital providers was strong: intraclass correlation of 0.92 (95% CI 0.89–0.93) for total score, and 0.91, 0.87 and 0.78 for the individual items.

A later systematic review and meta-analysis of the CPSS in the emergency department setting (Open Access Emergency Medicine, 2019; 11 studies, databases searched to December 2018) reported a pooled sensitivity of 82.46% (95% CI 74.83–88.09%) and specificity of 56.95% (95% CI 41.78–70.92%), with a diagnostic odds ratio of 6.22, positive likelihood ratio 1.92 and negative likelihood ratio 0.31.

Three things follow from those numbers that are worth stating plainly, because staff education materials routinely quote a single figure without them:

There is no one CPSS sensitivity. The figure moves substantially with the population and the setting — a screen applied to an undifferentiated 911 call performs differently from the same screen applied to patients who have already reached an emergency department. Quoting "88%" without saying "anterior circulation" is the most common misstatement of this instrument.
The confidence intervals are wide, particularly on specificity. A pooled specificity with a 95% CI spanning roughly 42% to 71% is not a precise operating characteristic; it is a range consistent with a lot of false positives.
False positives are the intended trade-off. A screen at the front of a time-critical pathway is tuned to over-trigger. The cost of a false positive is an unnecessary stroke alert; the cost of a false negative is a missed treatment window. If your programme’s stroke-alert positive predictive value looks low, that may be the instrument working as designed rather than a problem to be optimised away — but it should be measured and known, not assumed.

The posterior-circulation blind spot

The gap between the ~88% anterior-circulation sensitivity and the ~66% all-stroke figure in the original study is largely the posterior circulation. Strokes in the vertebrobasilar territory commonly present with vertigo, ataxia, diplopia or visual field loss, nausea and vomiting, or isolated gait instability — none of which the three CPSS items test. A patient with a cerebellar stroke can score entirely normal on facial symmetry, arm drift and speech repetition.

This is a known and documented limitation, not a defect in how the scale is used, and it is the reason BE-FAST exists and the reason many EMS protocols carry an explicit override: a normal CPSS does not close the door on stroke if the presentation is otherwise suggestive. Any protocol that treats a negative CPSS as a rule-out has misconfigured the instrument. In audit terms, this is worth checking directly — sample the cases where a stroke diagnosis was made in hospital and look at what the prehospital screen said.

Role in EMS stroke-alert activation

Operationally, the CPSS sits at a specific junction: it converts a clinical suspicion into a system action. In most EMS stroke protocols a positive CPSS, combined with a documented last-known-well time, is what authorises prehospital notification — the call that starts the receiving hospital’s stroke pathway before the patient physically arrives.

The elements that typically travel with that notification, and that a receiving-hospital audit should expect to find:

The three CPSS item results individually, not just "CPSS positive" — the receiving team plans differently for an aphasic patient than for one with isolated arm drift.
Last known well time, as a clock time rather than a duration. This is the single most consequential prehospital data element for thrombolysis eligibility, and the one most often recorded as "approximately 2 hours ago" and rendered useless by transport delay.
Blood glucose — hypoglycaemia is the classic stroke mimic and is correctable in the field.
Anticoagulant and antiplatelet use, and a contact who can give a collateral history.
Baseline functional status and any pre-existing deficit, which is what allows the receiving team to tell a new deficit from an old one.

AHA/ASA acute ischemic stroke guidance recommends that EMS personnel use a validated prehospital stroke screening tool, and prehospital notification based on such a screen is a well-established contributor to shorter in-hospital treatment times. The CPSS is one of several validated options; the recommendation is to use one consistently, not to use this one specifically.

Where CPSS documentation shows up in quality metrics

The CPSS itself is not usually a reported measure. Its influence on quality reporting is indirect but substantial, because it gates the events that are measured.

The anchor metric is door-to-needle time. Under the AHA/ASA Target: Stroke Phase III goals introduced in 2019, participating hospitals aim for door-to-needle within 60 minutes in at least 85% of eligible patients treated with IV thrombolytics, within 45 minutes in at least 75%, and within 30 minutes in at least 50%. Prehospital notification — which the CPSS triggers — is one of the recognised strategies for compressing that interval, because it moves CT availability, laboratory turnaround, and stroke-team assembly to before arrival rather than after it.

Practical consequences for quality and patient-safety staff:

A missed CPSS is a missed notification, and a missed notification is a longer door-to-needle time. When door-to-needle performance degrades, the prehospital screen is a legitimate place to look before assuming an in-hospital process failure.
Screen-to-alert concordance is auditable. How often does a positive CPSS actually produce a stroke alert? A gap here is a protocol-adherence finding with a clear owner.
Stroke-alert positive predictive value is worth tracking as a paired metric, so that a push to increase alert sensitivity does not quietly overload the pathway without anyone noticing.
Last-known-well capture rate is frequently a bigger driver of treatment eligibility than the screen result itself, and is easier to fix.
Stroke-centre certification programmes expect a defined, documented prehospital-to-ED handoff process — see disease-specific care certification for how that expectation is structured, and hospital core measures for how stroke measures sit alongside other reported measure sets.

Common documentation and audit failure modes

Recording a composite result only. "CPSS positive" loses the item-level detail that the receiving team and the abstractor both need.
Confusing CPSS with CPSSS/C-STAT in protocol documents, EHR field labels, or audit tools — different instruments, different clinical question, different destination decision.
Treating a negative CPSS as a stroke rule-out, against the documented posterior-circulation limitation.
Scoring arm drift with eyes open, which suppresses subtle drift and produces a false negative on the most reproducible item in the scale.
Not documenting pre-existing deficit, so an old facial droop or hemiparesis reads as an acute finding downstream.
Last known well recorded as a duration ("about an hour ago") rather than a clock time, which decays into uselessness with every minute of transport.
Quoting the 88% figure without the anterior-circulation qualifier in staff education, which builds false confidence in a negative screen.

Limitations to state explicitly in local protocol

It is a screen, not a diagnosis, and not a severity instrument.
It does not reliably detect posterior circulation stroke.
It does not distinguish ischaemic from haemorrhagic stroke — that requires imaging, and nothing in the field screen substitutes for it.
It does not identify large vessel occlusion, and so cannot by itself support a thrombectomy-centre bypass decision.
It does not exclude stroke mimics — hypoglycaemia, seizure with Todd’s paralysis, migraine with aura, and Bell’s palsy all screen positive on one or more items.
Pre-existing deficits degrade specificity in exactly the population most likely to be having another stroke.

Frequently asked questions

What are the three components of the Cincinnati Prehospital Stroke Scale?

Facial droop (ask the patient to smile or show teeth), arm drift (both arms held out for about ten seconds with eyes closed), and abnormal speech (repeat a standard sentence). Each is scored normal or abnormal.

What counts as a positive CPSS?

Any one of the three items being abnormal. There is no numeric threshold and no requirement for two or three abnormal items.

How accurate is the CPSS?

The original 1999 validation reported 66% sensitivity and 87% specificity for stroke overall, rising to 88% sensitivity for anterior circulation stroke. A 2019 meta-analysis of 11 emergency-department studies reported pooled sensitivity of 82.46% and specificity of 56.95%, with wide confidence intervals. Cite the setting and circulation territory alongside any figure.

How is the CPSS different from the NIHSS?

The CPSS is a three-item binary field screen that answers whether stroke should be suspected. The NIHSS is a 15-component, 0–42 severity instrument used in hospital for treatment decisions, serial monitoring and registry reporting, and it carries a formal examiner certification expectation the CPSS does not.

Can the CPSS detect a posterior circulation stroke?

Not reliably. None of the three items tests balance, gait, eye movements or visual fields, so a cerebellar or brainstem stroke can screen completely normal. This is the principal reason BE-FAST adds Balance and Eyes.

Is the CPSS the same as FAST?

They test the same three examination components. FAST is the public awareness mnemonic and adds Time as a call-to-action prompt rather than a fourth exam item; the CPSS is the clinical instrument with the validation literature behind it.

What is the difference between the CPSS and the CPSSS?

They are different instruments despite nearly identical abbreviations. The Cincinnati Prehospital Stroke Scale (CPSS) screens for stroke. The Cincinnati Prehospital Stroke Severity Scale (CPSSS, or C-STAT) predicts large vessel occlusion to support thrombectomy-centre routing decisions. Always write both out in full in protocol and audit documents.

Does a CPSS need to be repeated during transport?

The CPSS was not designed or validated for serial monitoring, and a change in it should not be documented as if it were a change in severity. Serial neurological assessment during transport is appropriate clinical practice; the instrument for quantifying change once in hospital is the NIHSS.

Related assessment instruments on this site

NIH Stroke Scale (NIHSS) — the in-hospital severity instrument the CPSS was simplified from.
Glasgow Coma Scale — level-of-consciousness assessment, frequently recorded alongside a stroke screen.
Modified Early Warning Score — physiological deterioration detection, a comparable trigger-to-action instrument.
Barthel Index and Functional Independence Measure (FIM) — functional outcome measurement after stroke.
Montreal Cognitive Assessment (MoCA) — post-stroke cognitive screening.
Patient safety — the wider cluster this guide belongs to.

Sources

Kothari RU, Pancioli A, Liu T, Brott T, Broderick J. Cincinnati Prehospital Stroke Scale: Reproducibility and Validity. Annals of Emergency Medicine, April 1999;33(4):373–378. PubMed 10092713.
The role of the Cincinnati Prehospital Stroke Scale in the emergency department: evidence from a systematic review and meta-analysis. Open Access Emergency Medicine, 2019. PubMed 31410071.
American Heart Association, Target: Stroke clinical tools and resources — Phase III door-to-needle goals.

This guide is written for EMS, prehospital and hospital patient-safety, quality and risk staff as a documentation and quality-reporting reference. It is not clinical guidance and does not replace your local stroke protocol, medical direction, or the current AHA/ASA acute ischemic stroke guidelines.