The morning she talked herself out of it
A 66-year-old woman woke up one morning and her right arm did not feel right. Not paralyzed. Weak and heavy, like it was not quite hers anymore. She did what I suspect most of us would do. She decided she had slept on it funny, and she got on with her morning.
About 3 hours later, it got better. Almost completely better. And in her mind, that settled it, because you do not go to an emergency department for a limb you slept on badly, especially not one that has already recovered.
She came in later anyway. I am very glad she did, because her MRI showed a small stroke that had already happened, and her CT angiogram showed moderate atherosclerosis in the arteries supplying her brain.
She was, by any reasonable measure, doing more right than most people who come through my doors. She quit smoking in the nineties. She takes a blood pressure tablet every day. She plays pickleball a couple of blocks from her house. Her GP had flagged high cholesterol and told her she likely had prediabetes, and those threads were being followed. She still had an event.
She went home the same day on two antiplatelet drugs, a maximum dose statin, a heart monitor to wear at home, and a plan to work with her GP on aggressive risk factor management. That is a good outcome. It could have been much worse. But every one of those treatments is about the next stroke. We have nothing that undoes this one.
That is the whole reason this article exists.
FAST is good. BE FAST is better.
Most people know FAST. Face, Arms, Speech, Time. It is a genuinely good tool and it is far better than nothing. The problem is that it was built around symptoms that come from the front of the brain, and it misses strokes in the back.
There is a number on this. In a review of 736 consecutive patients admitted with acute ischemic stroke, 14.1 percent had no FAST symptoms at all when they presented. Of those, 42 percent had gait imbalance or leg weakness, 40 percent had visual symptoms, and 70 percent had one or the other. Adding those two categories cut the proportion of missed strokes from 14.1 percent down to 4.4 percent.
So FAST misses roughly 1 in 7 strokes. Adding two letters cuts that to about 1 in 23. That is what BE FAST is for, and it is the version the American Heart Association and American Stroke Association use in their own public materials.
B is for Balance
Sudden loss of balance, sudden clumsiness, or sudden vertigo arriving out of nowhere in someone who does not get vertigo. This is not the mild wobble of standing up too fast.
E is for Eyes
Sudden loss of vision in one eye or in part of the visual field, sudden double vision, or the one people struggle to describe, where half the world has simply gone missing. You can test peripheral vision quickly by asking them to look at your nose while you hold up 2 fingers out to one side and ask how many they can see.
F is for Face
Ask for a proper smile, teeth showing. You are looking for one side lagging behind the other.
A is for Arms
Both arms straight out in front, palms up, eyes closed, for 10 seconds. You are looking for one arm drifting down or rotating inward. This is the single most useful thing a family member can do and almost nobody does it.
S is for Speech
Slurred words, the wrong words coming out, or a blank look in response to a simple question. Ask them to repeat a specific short sentence rather than asking whether they feel okay, because "yeah, fine" is a two-word answer that hides a lot.
T is for Time
Time means two things. Go now, and note the time. The single most valuable piece of information you can hand a stroke team is the last moment this person was seen completely normal. Not when you noticed. When they were last definitely fine. If they woke up with it, the honest answer is bedtime, and it helps to know roughly when they fell asleep. The 2026 American Heart Association and American Stroke Association guideline uses the midpoint of the sleep period as a reference point for one of the treatment decisions, so even a rough estimate is useful.
You are not trying to be a doctor here. You are using a tool to decide whether to be seen. If you can only remember one letter, that is enough. Being unsure upgrades a green light symptom to amber and an amber light symptom to red.
The first five minutes
● Call 911. Do not drive.
● Note the last known well time, or the time they went to bed if they woke up with it, and say it out loud to the dispatcher.
● Do not give aspirin. This is different from chest pain. About 1 in 7 strokes is caused by bleeding rather than a clot, and aspirin can make that worse. Aspirin is a chest pain reflex, not a stroke reflex.
● Do not let them lie down and sleep it off.
● Grab the medication list, particularly any blood thinner. Bring the containers if you can, because some medications change what we can do in the first few hours.
● Unlock the front door, turn on the porch light, and stay on the phone.
On the driving question, the reason is not that you might crash. When you call 911, three things happen that cannot happen in your own car. The crew starts the assessment on the way, so the clock starts in your driveway rather than in our waiting room. They call ahead, so the team and the scanner are ready before you arrive. And they route you to a center that can actually treat you, with the stroke team mobilized before the doors open.
Why the clock is the whole story
A stroke is a plumbing problem. An artery gets blocked, usually by a clot, and the tissue downstream is starved of oxygen. The brain runs on what the blood delivers, second by second, so interrupting that supply is an immediate problem.
Where flow stops completely, tissue dies fast. Where some flow remains, the tissue is not dead yet. It has stopped working and it is waiting to see which way things go. Everything we do in the first few hours is a fight for that tissue.
The most quoted number in stroke medicine comes from a 2006 analysis by Saver. In a typical large-vessel supratentorial ischemic stroke, the brain loses roughly 1.9 million neurons every minute, which works out to about 3.6 years of accelerated brain aging for every untreated hour. That figure describes a large-vessel stroke specifically rather than every stroke, which is worth saying plainly. It is not a scare tactic. It is an argument for being seen when symptoms happen rather than after they settle.
What we can actually do, and how much it helps
Clot-busting drugs
There are two given intravenously, alteplase and tenecteplase, and the 2026 guideline treats them as equivalent first-line options within 4.5 hours of onset or last known well.
The evidence behind that window comes from a pooled analysis of individual patient data on 6,756 people across 9 randomized trials. Treated within 3 hours, 32.9 percent of patients had no significant disability at 3 to 6 months, compared with 23.1 percent of controls. Between 3 and 4.5 hours it was 35.3 percent versus 30.1 percent. Beyond 4.5 hours, the benefit was no longer statistically significant.
The window is also not as hard an edge as it used to be. The 2026 guideline says treatment may still be reasonable for people who wake up with symptoms within 9 hours of their sleep midpoint, or who are 4.5 to 9 hours from last known well, if perfusion imaging shows tissue worth saving. That is a softer recommendation and it needs imaging that not every hospital has. But it does mean the door does not slam shut at 4 hours and 31 minutes.
Pulling the clot out
For strokes caused by blockage of a large artery, the neurointervention team can go in and physically remove the clot. A pooled analysis of 1,287 patients across 5 randomized trials found a number needed to treat of 2.6 to reduce disability by at least one level on the modified Rankin Scale at 90 days. For roughly every 3 patients treated, 1 ends up meaningfully less disabled.
That is one of the strongest treatment effects in medicine. It is worth being precise about what it is not, though. That analysis did not show a reduction in 90-day mortality, and it applies to proximal anterior circulation large-vessel occlusion, not to every stroke. The 2026 guideline now supports thrombectomy in selected patients out to 24 hours from onset, which means the feeling that you have already left it too late is not a reason to stay home. Come anyway. We will work out what is still on the table when you get here.
Symptoms that resolve are not an all-clear
This is the part that got our patient, and it is the single most transferable idea in this whole piece.
Most odd sensations do pass, and most of the time the instinct to wait is correct. Our instincts are reasonably well calibrated for almost everything else in life. They are catastrophically miscalibrated for this one thing, because stroke is the rare condition where symptom improvement is not evidence that you are safe.
Her weakness lasted about 3 hours and then resolved completely, and that resolution is exactly what convinced her she was fine. If her MRI had come back clean, we would have called it a transient ischemic attack. It did not, so what she actually had was a stroke that happened to get better. From where she was standing that morning, those two possibilities looked identical. The only thing that separates them is a scanner.
The American Stroke Association calls a TIA a warning stroke in its own patient materials, and that is exactly the right name. The highest-risk window for a full stroke afterward is the days immediately following. It is the only warning shot most people get. Treat it as an appointment you were given, not a bullet you dodged.
You are not wasting our time
The most common reason people delay is not denial. It is embarrassment. "I don't want to waste their time."
So here is the most freeing number in this article. In a Dutch study of 3,317 consecutive stroke code activations across 4 ambulance regions, 1,077 patients, or 32.5 percent, turned out to be stroke mimics. Migraine, seizure, low blood sugar, fainting, and an old deficit flaring up with an infection can all present this way.
Roughly 1 in 3 of the people we scramble the whole team for is not having a stroke. That is not the system failing. That same study found the prehospital scales paramedics use run high on sensitivity and low on specificity, with sensitivities from 66 to 84 percent against specificities of only 28 to 49 percent. That is the trade the system deliberately makes, because the cost of missing a stroke is so much higher than the cost of assessing someone who turns out to be fine.
We built the system on the assumption that you would call. If you turn out to be fine, that is not a wasted trip.
The half of the conversation nobody puts in a clip
You will find content telling you that clot-busting drugs cause brain bleeds and that hospitals push them anyway. The kernel of truth there is real, and you deserve both halves of it.
In that same analysis of 6,756 patients, thrombolysis raised symptomatic bleeding into the brain from 1.3 percent to 6.8 percent. Fatal intracranial hemorrhage within 7 days went from 0.4 percent to 2.7 percent. Nobody pretends those are small numbers.
Here is what gets left out. Ninety-day mortality was 17.9 percent in the treated group and 16.5 percent in controls, a difference that did not reach statistical significance. The authors' own summary is that despite an average absolute increase in early death from intracranial hemorrhage of about 2 percent, by 3 to 6 months that risk was offset by an average absolute increase in disability-free survival of about 10 percentage points for patients treated within 3 hours, and about 5 for those treated between 3 and 4.5 hours.
So it is a genuine tradeoff, and the size of the gain depends on how fast you get there. The exaggeration is not that the risk exists. It is presenting one half of a trade as the whole thing. And it is a decision made by a stroke physician with a scan in front of them, not by you at the kitchen table. Your job is to get into the room where that conversation can happen.
The boring part that actually prevents strokes
Everything above is about the worst 4 hours of somebody's life. But our patient did not arrive in my department because she had never heard of BE FAST. She arrived because of 20 years of physiology and some genetics.
You cannot do much about the genetics. You can do a great deal about the rest. High blood pressure is the single largest contributor to stroke risk at a population level, accounting for roughly half of strokes depending on how it is measured, so knowing your number and working with your primary care physician on it is the highest-yield thing on this list.
There are medications for that, and they work. There are also things you can start today. Move daily. Stop smoking. Reduce or stop alcohol. Build in some meditation or relaxation practice. Eat a plant-rich diet, with the DASH and Mediterranean patterns being the best studied: plant foods, seafood, low saturated fat, high fiber, low-fat dairy. Do all of that in a calorie-controlled way and you will likely lose weight, which drives blood pressure, lipids, and diabetes risk down together.
None of this is a Tier 2 problem. No supplement prevents a stroke. No wearable prevents a stroke. What prevents a stroke is a blood pressure cuff, a doctor who knows you, and 20 years of unglamorous decisions made regularly.
Come back to our patient. She went home on two antiplatelet drugs, a maximum dose statin, a heart monitor to look for the irregular rhythm that throws clots, and a referral back to her GP for her blood pressure, her cholesterol, and her prediabetes risk. Every item on that list could have been chased in a GP office 10 years earlier instead of in my department in an emergency. That is not a criticism of her, and the pickleball is not wasted. It is an argument for the fifth pillar: a primary care physician who knows you and can hold all of those threads at once.
Two numbers make that follow-up appointment the most important thing on her calendar. About 1 in 4 stroke survivors is at risk of another, and up to 80 percent of second clot-related strokes may be preventable.
How should this modify your practice?
If you are a patient or a family member
● Learn BE FAST rather than FAST. The two extra letters are the difference between missing about 1 in 7 strokes and about 1 in 23.
● Write down the last known well time the moment you suspect something, and say it out loud to the dispatcher. If they woke up with it, note when they went to bed.
● Call 911 rather than driving. The assessment, the pre-notification, and the routing decision all start earlier that way.
● Do not give aspirin for suspected stroke, and do not let anyone lie down to sleep it off.
● If symptoms resolve completely, go anyway. Resolution is the warning, not the all-clear.
● If you have already had a stroke or TIA, treat the follow-up appointment as the highest-value hour in your calendar.
● Send this to the person in your life most likely to say "I'm alright." You know exactly who that is.
If you are a clinician
● Consider auditing whether your local public education materials still say FAST. The 4.4 percent versus 14.1 percent miss rate is a straightforward argument for updating them.
● The 2026 guideline treats tenecteplase and alteplase as equivalent first-line options within 4.5 hours. If your department is still alteplase-by-default, that is worth a protocol conversation.
● Know your institution's perfusion imaging capability and your transfer pathway, because the extended windows out to 9 hours for thrombolysis and 24 hours for thrombectomy are only usable if the imaging exists or the transfer is fast.
● Sleep midpoint is a documentable data point for wake-up stroke. Ask for it and record it.
● A 32.5 percent stroke mimic rate among code activations is a feature of a high-sensitivity system, not a quality problem. Framing it that way to trainees and to EMS partners protects the low threshold that makes the system work.
● For minor noncardioembolic stroke of presumed large-artery atherosclerotic origin, the 2026 guideline supports dual antiplatelet therapy. Pair it with the secondary prevention conversation rather than deferring that entirely to primary care.
Author
Adrian Cois, MD is an Emergency Medicine physician and Assistant Professor. He writes and produces Overheard in the Emergency Room, an evidence-based health education series for patients, students, residents, and clinicians.
Disclosure: no financial relationships relevant to this content. No affiliate links appear on this site. Cronometer is referenced across this series as a diet-tracking tool with no financial relationship of any kind.
This article is for educational purposes only. It does not provide medical advice and does not establish a physician-patient relationship. If you have symptoms concerning for a medical condition, please seek care from a qualified clinician.
Related reading
● S2E1 (16), the "ER or Wait?" pillar episode
● S2E4 (19), Chest Pain: ER or Wait?
● S2E8 (23), What Your A1c Actually Means
● S2E9 (24), Statins: Should I Worry About My Meds?
● Season 1, Episode 10, Your Healthcare Team
● Overheard Journal Club Ep 7, Subarachnoid Hemorrhage and CT
References
1. Centers for Disease Control and Prevention. Stroke facts. Updated July 10, 2025. Accessed September 6, 2026. https://www.cdc.gov/stroke/data-research/facts-stats/index.html
2. Saver JL. Time is brain: quantified. Stroke. 2006;37(1):263-266. doi:10.1161/01.STR.0000196957.55928.ab
3. American Stroke Association. Five key facts about stroke. Accessed September 5, 2026. https://www.stroke.org/en/help-and-support/resource-library/five-key-facts-about-stroke
4. Prabhakaran S, Gonzalez NR, Zachrison KS, et al. 2026 guideline for the early management of patients with acute ischemic stroke: a guideline from the American Heart Association/American Stroke Association. Stroke. 2026;57(8):e316-e436. doi:10.1161/STR.0000000000000513
5. Correction to: 2026 guideline for the early management of patients with acute ischemic stroke: a guideline from the American Heart Association/American Stroke Association. Stroke. 2026;57(8):e461-e467. doi:10.1161/STR.0000000000000530
6. Emberson J, Lees KR, Lyden P, et al. Effect of treatment delay, age, and stroke severity on the effects of intravenous thrombolysis with alteplase for acute ischaemic stroke: a meta-analysis of individual patient data from randomised trials. Lancet. 2014;384(9958):1929-1935. doi:10.1016/S0140-6736(14)60584-5
7. Goyal M, Menon BK, van Zwam WH, et al. Endovascular thrombectomy after large-vessel ischaemic stroke: a meta-analysis of individual patient data from five randomised trials. Lancet. 2016;387(10029):1723-1731. doi:10.1016/S0140-6736(16)00163-X
8. Dekker L, Moudrous W, Daems JD, et al. Prehospital stroke detection scales: a head-to-head comparison of 7 scales in patients with suspected stroke. Int J Stroke. 2025;20(3):268-277. doi:10.1177/17474930241275123
9. Aroor S, Singh R, Goldstein LB. BE-FAST (Balance, Eyes, Face, Arm, Speech, Time): reducing the proportion of strokes missed using the FAST mnemonic. Stroke. 2017;48(2):479-481. doi:10.1161/STROKEAHA.116.015169
