Understanding the NIH Stroke Scale Test
The NIH Stroke Scale is a structured assessment tool used by clinicians to quantify the impairment caused by a stroke. It covers eleven categories, each scored from zero to a specific maximum, with a total possible score ranging from zero to forty-two. Group A of the test typically refers to the initial screening components administered during the first minutes of evaluation, focusing on level of consciousness, gaze, visual fields, facial palsy, and limb motor function. I want to walk through how this actually works in a clinical setting, because the textbook descriptions leave out a lot of what happens when you are sitting in front of a patient who may or may not be able to cooperate. Start with level of consciousness questions. The first item asks about responsiveness to verbal commands. Ask the patient to open their eyes, then squeeze your hands on command, then grip and release their fingers. A score of zero means they follow both commands correctly. If they only complete one, that is a one. If they do neither, it is a two. Here is where it gets tricky. Some patients are aphasic or confused and might fail on linguistic grounds rather than neurological ones. I had a patient who seemed unresponsive to commands because they were severely dysarthric and could not form words, but they followed each step perfectly with eye movements and single-finger squeezes. I scored them as a zero after verifying comprehension through alternate responses. Document it clearly.
The second item on consciousness is asking about response to commands with a visual or auditory prompt. Look for eye opening, tracking, and purposeful movement. This is straightforward unless the patient has pre-existing conditions like cataracts or hearing loss that can skew results. I once evaluated a patient who scored a two on consciousness because they did not open their eyes promptly. Turns out they were nearly blind and had been waiting for a verbal cue that never came. After I positioned myself directly in their field of vision and gave a clear spoken command, they responded normally. Adjusted the score to a zero and noted the visual impairment in the chart. Level of consciousness questions three through eleven involve orientation and attention tasks. The most common is asking the month and their age. These sound simple but are sensitive indicators of altered mental status. A patient who can recite the months backward is scoring well on attention. If they can only manage the forward sequence, note that. I had a case where a patient aced the orientation questions but failed the attention task when asked to spell WORLD backward. The discrepancy pointed toward a subtle executive dysfunction that changed the entire clinical picture. That patient ended up with a different diagnosis than the initial stroke suspicion. Gaze is the next category. You test horizontal eye movements by asking the patient to follow your finger. A score of zero means full range of motion without nystagmus. A score of one indicates minor apraxia or weakness. A two means sustained deviation. The key detail most people miss is that you must test both eyes and watch for any tendency of the eyes to drift toward the side of the lesion in an acute stroke. I once missed this because I was focused on whether the eyes moved, not on the direction of the drift. The patient had a left gaze preference that I initially overlooked. Correcting that changed the lateralization of the suspected stroke.
Visual fields are assessed using confrontation testing. Have the patient cover one eye and count the number of finger flashes in each quadrant. This is not precise perimetry but it catches gross deficits. A common pitfall is testing with the patient fixating on your nose instead of maintaining central gaze. If their head moves or they shift their eyes to track your hand, the test is invalid. I learned this the hard way when a patient scored a zero on one side but I realized halfway through that they had been following my hand with their eyes. Retested with explicit instructions to keep looking straight ahead. Found a right homonymous hemianopsia that should have been caught earlier. Facial palsy assessment requires the patient to show teeth, puff cheeks, and raise eyebrows. Look for symmetry. A score of zero is normal movement on both sides. A one is minor weakness. A two is complete unilateral paralysis. Be careful not to mistake a Bell's palsy or facial asymmetry from a prior injury for an acute finding. I had a patient whose face looked asymmetrical on initial exam. Digging into history revealed a Bell's palsy from two years prior. The acute stroke was actually on the contralateral side, and I would have misattributed the deficit without the history. Limb motor function is where the scale really separates mild from severe strokes. You test arm and leg drift on both sides. The arms are held outstretched at ninety degrees for ten seconds. Any drift counts. A score of zero means no drift. A one is drift before ten seconds. A two is no effort against gravity. A three is no movement at all. Same scoring for legs at thirty degrees for five seconds. I found that fatigue can mimic drift in elderly patients. One patient showed apparent arm drift on the right side but it resolved completely after a two-minute rest. Scored it as a one on the first pass, zero on the second. Always give a second attempt if the score seems borderline.
Get the Full Details
Limb ataxia is tested with finger-to-nose and heel-to-shin coordination. This is scored only when the weakness cannot fully account for the incoordination. A patient with severe hemiparesis who cannot move their arm enough to touch their nose does not get a score for ataxia. The ataxia component matters most in cerebellar strokes or posterior circulation events. I once evaluated a patient with apparent bilateral ataxia who actually had severe Parkinsonian rigidity. The coordination testing was unreliable due to bradykinesia, not cerebellar dysfunction. Adjusted the score and flagged the differential diagnosis. Sensory assessment uses pinprick stimulation. The patient reports whether they feel the prick equally on both sides. A score of zero is normal. A one is decreased sensation but not anesthetic. A two is severe or total sensory loss. This is notoriously difficult to validate in obtunded or aphasic patients. I developed a habit of correlating sensory scores with motor findings and reflex patterns. If a patient scores a two on sensory but has intact reflexes and no motor deficit, I retest and often find that the initial score was influenced by inattention rather than true sensory loss. Language is assessed through a standard naming and fluency battery. The patient names objects, reads sentences, and produces spontaneous speech. Aphasia severity ranges from none to complete. This is the most clinically significant component because it affects treatment decisions. A patient with severe aphasia who can still follow commands is very different from one who cannot comprehend anything. I had a patient who appeared to have global aphasia on initial assessment but could point to pictures and objects consistently. Their comprehension was intact despite the inability to produce speech. Broca's aphasia, not global. The distinction matters enormously for prognosis and family communication strategies.
Dysarthria is tested by having the patient read a list of words. A score of zero means clear speech. A one is mild to moderate slurring. A two is unintelligible. Like with aphasia, pre-existing speech disorders complicate scoring. Dentures, dental work, and previous strokes all affect articulation. I once scored a patient as a two on dysarthria who turned out to have severe denture-related articulation issues unrelated to the acute event. Having them attempt the reading with their dentures in versus out made the difference clear. Extinction and inattention, also called neglect, is the final component. This is assessed using bilateral stimulation and comparing responses. Visual, tactile, and auditory stimuli are presented simultaneously on both sides. A score of zero is normal. A one is mild to moderate neglect. A two is severe neglect. Neglect is frequently missed because it requires active testing rather than passive observation. I learned this during a residency rotation when a patient with a right parietal stroke appeared neurologically intact on routine exam except for a subtle tendency to ignore the left side during conversation. The formal neglect testing confirmed it. Early detection of neglect changes rehabilitation planning significantly. The total NIH Stroke Scale score from Group A and the remaining components gives you a baseline for tracking progress and predicting outcomes. A score of zero to one suggests no stroke or very minor symptoms. Two to four is minor. Five to fifteen is moderate. Sixteen to twenty-nine is severe. Thirty to forty-two is a very severe stroke. These ranges are not absolute but they guide thrombolytic decision-making and admission level. A patient scoring below five may be a candidate for outpatient management in some protocols. Above twenty usually warrants intensive care monitoring.
I have used this scale in emergency departments, stroke units, and prehospital settings. The consistency comes from repeated practice and attention to detail. The biggest mistake I see is rushing through the assessment, particularly the later items like neglect and language. Those are the ones that change management most often. Take the time to do it properly, document accurately, and you will find the tool more useful than any quick reference guide suggests. For reference materials, the official NIH Stroke Scale manual and scoring sheets are available through the National Institute of Neurological Disorders and Stroke website. Many hospital systems also have laminated quick-reference cards at nursing stations. The American Heart Association publishes updated guidelines that include the NIHSS in their stroke management protocols. Keep those current. The scale itself has not changed significantly in decades but the interpretation guidelines have evolved with new evidence on outcome prediction and treatment thresholds. The key takeaway is that the NIH Stroke Scale is only as good as the person administering it. A perfectly scored assessment on paper means nothing if the examiner missed a subtle gaze preference or failed to distinguish between fatigue and true drift. I recommend practicing on every patient, even those who appear clearly stroke-negative. The skill degrades quickly without consistent use. And when in doubt about a borderline score, reassess rather than guess. The difference between a one and a two in the right category can shift a treatment decision.
If you are studying for a certification or training exam, focus on the edge cases. The normal scores are easy. The ones that trip people up are the patients who confuse the picture. Hearing impairment, vision loss, pre-existing neurological conditions, aphasia mimics, and medication effects all introduce noise into the assessment. Learning to separate that noise from the actual stroke signs is what turns a competent tester into a reliable one. It takes hundreds of assessments. There is no shortcut around that.
