🧠 NIH Stroke Scale (NIHSS)（National Institutes of Health Stroke Scale）
Bedside scoring tool — not a hospital AI agent
- Standardize neurologic deficit severity in suspected AIS (0–42)
- Communicate serial change around reperfusion decisions
- Support severity framing with the local stroke pathway (not sole decision)
- Replace imaging to exclude LVO or intracranial hemorrhage
- Serve as the sole thrombolysis eligibility rule
- Score reliably without trained NIHSS technique
FormulaSum of 15 items (consciousness, gaze, fields, facial/limb motor, ataxia, sensory, language, dysarthria, extinction). 1–4 minor, 5–15 moderate, 16–20 moderate–severe, 21–42 severe.
Stratification0 no stroke symptoms; 1–4 minor; 5–15 moderate; 16–20 moderate–severe; 21–42 severe.
SourcesSources: AHA/ASA 2026 (doi 10.1161/STR.0000000000000513); China AIS 2023; Brott 1989
UseStandardised stroke severity assessment to guide treatment and track change.
Strata0 no stroke symptoms; 1–4 minor; 5–15 moderate; 16–20 moderate–severe; 21–42 severe.
LimitsRequires trained examiners and standardised technique. · Consciousness and cooperation affect reliability of several items. · Not a substitute for imaging or reperfusion eligibility checklists.
SourceBrott T, et al. Stroke 1989 (NIHSS)
This-shift next steps
After scoring: open the scene pack, peer scores, and the matching pathway. Local reference — no HIS write.
- Who it is for
- Adults with suspected acute stroke when quantifying deficit and deciding reperfusion, or TIA when judging short-term risk.
- Who it is not for
- Do not use ABCD² to decide thrombolysis. No new stroke scale.
- Formula and variables
- NIHSS is the neurologic deficit; items stay as on this page. FAST-ED is a prehospital large-vessel screen. ABCD² is age, blood pressure, clinical features, duration, and diabetes, for TIA.
- Result → next step
- This scene links only to the three tools below: deficit and time window open the thrombolysis page. A positive large-vessel screen opens thrombectomy. Drugs after the acute phase open secondary prevention. Thrombolysis shows no number until weight, onset time, pregnancy, and recent anticoagulation are known. Anticoagulant-related intracranial bleeding goes to reversal and ICH management.No numeric dose until age, weight, renal function, and pregnancy are all known. Pregnancy is contraindication or caution only.IV thrombolysis →
- Guideline
- China acute ischemic stroke guideline 2023
Paste the link in Slack, Teams, X, or LinkedIn — the preview image comes from this page’s Open Graph card.
Methodology · worked example, guidelines & full notes
Higher scores predict larger infarcts and worse outcome.
When to use
Standardised stroke severity assessment to guide treatment and track change.
How it works
Sum of 15 items (consciousness, gaze, fields, facial/limb motor, ataxia, sensory, language, dysarthria, extinction). 1–4 minor, 5–15 moderate, 16–20 moderate–severe, 21–42 severe.
Key points
- Higher scores predict larger infarcts and worse outcome.
- Used in reperfusion decisions and serial monitoring.
- Follow the official scoring rules for each item.
References
Worked calculation
The values below come from this tool's own example placeholders and are computed server-side with the formula shown on this page, so the arithmetic can be checked quickly. It demonstrates how to substitute values only — it is not clinical advice and not a real case.
| 1a. Level of consciousness | Drowsy (1) |
|---|---|
| 1b. LOC questions (month, age) | One correct (1) |
| 1c. LOC commands (open eyes, grip) | One (1) |
| 2. Best gaze | Partial palsy (1) |
| 3. Visual fields | Partial hemianopia (1) |
| 4. Facial palsy | Minor (1) |
| 5a. Left arm motor | Some effort vs gravity (2) |
| 5b. Right arm motor | Some effort vs gravity (2) |
| 6a. Left leg motor | Some effort vs gravity (2) |
| 6b. Right leg motor | Some effort vs gravity (2) |
| 7. Limb ataxia | One limb (1) |
| 8. Sensory | Mild–moderate loss (1) |
| 9. Best language | Mild–moderate aphasia (1) |
| 10. Dysarthria | Mild–moderate (1) |
| 11. Extinction/inattention | One modality (1) |
→NIHSS19/42
- Severity：Moderate–severe stroke
- Note：Higher scores predict larger infarcts and worse outcome; used to guide reperfusion decisions and track change.
| 1a. Level of consciousness | Alert (0) |
|---|---|
| 1b. LOC questions (month, age) | Both correct (0) |
| 1c. LOC commands (open eyes, grip) | Both (0) |
| 2. Best gaze | Normal (0) |
| 3. Visual fields | No loss (0) |
| 4. Facial palsy | Normal (0) |
| 5a. Left arm motor | No drift (0) |
| 5b. Right arm motor | No drift (0) |
| 6a. Left leg motor | No drift (0) |
| 6b. Right leg motor | No drift (0) |
| 7. Limb ataxia | Absent (0) |
| 8. Sensory | Normal (0) |
| 9. Best language | Normal (0) |
| 10. Dysarthria | Normal (0) |
| 11. Extinction/inattention | Normal (0) |
→NIHSS0/42
- Severity：No stroke symptoms
- Note：Higher scores predict larger infarcts and worse outcome; used to guide reperfusion decisions and track change.
Verification & sources · governance
FAQ
Frequently asked questions
- What NIHSS cut-off triggers thrombolysis or thrombectomy?
- Thresholds follow guidelines and the local stroke pathway; NIHSS alone does not decide. Combine imaging, time window, and contraindications. Computed locally in the browser; patient data are not uploaded. For licensed clinicians only; combine with history, exam, and local protocol — not a standalone decision.
- Can NIHSS replace head imaging to exclude hemorrhage or LVO?
- No. NIHSS does not replace imaging to exclude intracranial hemorrhage or large-vessel occlusion. Computed locally in the browser; patient data are not uploaded. For licensed clinicians only; combine with history, exam, and local protocol — not a standalone decision.
- Can untrained staff score NIHSS independently?
- No reliable independent scoring without standard NIHSS training. Computed locally in the browser; patient data are not uploaded. For licensed clinicians only; combine with history, exam, and local protocol — not a standalone decision.
- Are patient data uploaded, and where is the score calculated?
- Computed locally in the browser; patient data are not uploaded. For licensed clinicians only; combine with history, exam, and local protocol — not a standalone decision.