In acute ischemic stroke, time is the variable that most reliably predicts outcome. The phrase "time is brain" — coined by Camilo Gomez in 1993 and quantified by Jeffrey Saver in 2006 — captures the central premise of acute stroke care: every minute of cerebral ischemia destroys tissue that no subsequent intervention can recover.1 The trials of the last three decades, from NINDS in 1995 through HERMES, DAWN, and SELECT2, have refined which patients benefit and how — but they have not weakened, and have in many ways sharpened, the dependence of outcome on speed.

How much brain is lost per minute in ischemic stroke?

Saver's 2006 analysis, published in Stroke, used average values from neuropathology and infarct kinetics to estimate the per-minute cost of an untreated supratentorial large-vessel occlusion: approximately 1.9 million neurons, 14 billion synaptic connections, and 12 kilometers of myelinated axonal fibers per minute.1 Each hour of ischemia therefore accelerates the brain through roughly 3.6 years of normal aging.

These are population averages. The actual rate of infarct growth in any given patient depends on collateral circulation, the location of the occlusion, baseline tissue health, and physiological factors such as blood pressure and glucose. But the order of magnitude has been replicated in serial imaging studies and is consistent with the time-dependent outcomes seen in every major reperfusion trial.

What does the evidence show about time-to-treatment?

Intravenous thrombolysis. The original NINDS trial in 1995 demonstrated that recombinant tissue plasminogen activator improved 90-day functional outcomes when given within three hours of symptom onset.2 The Lees pooled analysis of eight randomized trials (n = 6,756) in The Lancet in 2010 demonstrated a clear dose-response with time: the odds ratio for a favorable outcome (mRS 0–1 at 90 days) was 2.55 in the 0–90-minute stratum, 1.64 in the 91–180-minute stratum, and 1.34 in the 181–270-minute stratum, with no significant benefit beyond 4.5 hours.3

Saver and colleagues, working with the AHA Get With The Guidelines registry (n = 58,353 patients), translated these effect sizes into clinical terms: each 15-minute reduction in door-to-needle time was associated with approximately one additional month of disability-free life.4 Meretoja's "Save a minute, save a day" analysis showed that, on average, every minute saved from onset to thrombolysis added 1.8 days of healthy life.5

Endovascular thrombectomy. The HERMES collaboration — a meta-analysis of individual patient data from MR CLEAN, ESCAPE, REVASCAT, SWIFT PRIME, and EXTEND-IA — established that mechanical thrombectomy was superior to medical management for anterior-circulation large-vessel occlusion, and that the benefit was strongly time-dependent.6 Saver's 2016 JAMA analysis using the HERMES dataset quantified the relationship: for each 9-minute delay in symptom-onset-to-reperfusion, one fewer patient out of 100 achieved a functionally independent outcome at 90 days.7

Time is not a soft variable in stroke. It is the lever that determines whether a patient walks out of rehabilitation or does not.

What are the AHA door-to-needle and door-to-puncture targets?

The 2019 AHA/ASA guidelines for the early management of acute ischemic stroke set process benchmarks that have since been tightened under the Target: Stroke Phase III initiative.8 The current targets are:

  • Door-to-needle ≤60 minutes in ≥85% of eligible patients (baseline expectation).
  • Door-to-needle ≤45 minutes in ≥75%.
  • Door-to-needle ≤30 minutes in ≥50% (Target: Stroke Phase III Honor Roll Elite Plus).
  • Door-to-puncture ≤90 minutes for patients presenting directly to a thrombectomy-capable center.
  • Imaging-arrival-to-puncture ≤60 minutes for transferred patients.

Achieving the 30-minute door-to-needle benchmark is not a function of effort alone — it requires re-engineering the stroke alert workflow so that parallel processing replaces sequential handoffs.

Practice point

The single highest-yield intervention for compressing door-to-needle time is parallel processing: imaging, point-of-care labs, NIHSS, history, and pharmacy preparation should run on different staff in different rooms simultaneously. Centers that linearize these steps rarely achieve sub-45-minute door-to-needle times. Centers that parallelize them routinely break 30.

Where do the minutes hide?

A stroke alert, broken into discrete steps, exposes the points at which time accumulates without anyone defending it. The most common sources of delay include:

  • Pre-hospital recognition. Patients and bystanders frequently underrecognize stroke symptoms; median symptom-onset-to-911 time remains over 60 minutes in most US registries.
  • EMS pre-notification. Hospitals that receive advance EMS notification of an incoming stroke alert reduce door-to-needle time by approximately 25–35% compared with those that do not.
  • Direct-to-CT pathways. Routing the patient directly from EMS to the scanner — bypassing the ED bay — saves 10–20 minutes.
  • Transfer delays. Patients with large-vessel occlusion who present first to a non-thrombectomy-capable center experience a median additional delay of 90–120 minutes before puncture.
  • Imaging interpretation. Centers without automated software for ASPECTS scoring and LVO detection accumulate time during human interpretation, particularly overnight.
  • Pharmacy turnaround. Alteplase requires reconstitution; tenecteplase, given as a single bolus, removes this step.

Each of these has been successfully shortened, somewhere, by someone. The work of running a stroke system is largely the work of borrowing those solutions and refusing to accept any individual minute as unrecoverable.

Do late-window trials change the urgency?

DAWN (Nogueira et al., NEJM 2018) and DEFUSE-3 (Albers et al., NEJM 2018) demonstrated that mechanical thrombectomy improved outcomes out to 24 hours from last-known-well in carefully selected patients with small ischemic cores and large salvageable penumbras on perfusion imaging.910 These trials moved the field from a clock-based to a tissue-based selection paradigm.

They did not, however, change the underlying biology. The conditional probability of mRS 0–1 at 90 days is still highest when reperfusion is achieved early; selection by perfusion imaging identifies patients in whom the penumbra remains viable longer, but does not slow its rate of conversion to infarct. A patient who can be treated at 3 hours will, on average, do better than the same patient treated at 9 hours, even when both are eligible.

How do mobile stroke units fit in?

Mobile stroke units (MSUs) — ambulances equipped with a CT scanner, point-of-care laboratory, and a trained team — push the door-to-needle clock upstream of the hospital door. The BEST-MSU trial (Grotta et al., NEJM 2021) randomized patients in seven US cities and demonstrated that MSU dispatch was associated with a higher proportion of patients achieving a 90-day mRS of 0–1 compared with conventional EMS transport (55.0% vs 44.4%; adjusted odds ratio 2.43).11

The European B_PROUD study (Ebinger et al., JAMA 2021) reached similar conclusions: MSU care was associated with lower 90-day disability scores in patients with thrombolysis-eligible stroke.12 The remaining questions are not whether MSUs work, but where their cost is justified — a question of population density, EMS volume, and existing door-to-needle performance.

What practical steps reduce treatment delays?

Published quality improvement work and the AHA Target: Stroke initiative converge on a small set of interventions that consistently move door-to-needle times below 45 minutes:

  1. EMS pre-notification with structured pre-arrival data (last-known-well, LVO screening tool, blood pressure, glucose).
  2. Single-call activation of the stroke team to the CT suite.
  3. Direct-to-CT from the EMS stretcher, bypassing the ED bay.
  4. Parallel processing of imaging, point-of-care labs, neurological exam, and consent.
  5. Tenecteplase at bedside, drawn into a syringe and ready to administer, removing pharmacy reconstitution from the critical path.
  6. Standardized imaging protocols with automated post-processing (ASPECTS scoring, LVO detection, perfusion).
  7. Drip-and-ship pathways for thrombolysis at the spoke followed by transfer for thrombectomy, with image transfer occurring in parallel with patient movement.
  8. Continuous QI feedback with case-level review of every stroke alert exceeding internal time benchmarks.

The bottom line

Acute ischemic stroke is one of the few neurological emergencies in which speed of intervention is itself the treatment. The trials that built modern stroke care — NINDS, the alteplase pooled analyses, the five 2015 thrombectomy trials, HERMES, DAWN, DEFUSE-3, BEST-MSU — converge on a single clinical reality: every minute compresses the menu of what can be offered, and every minute lost is tissue that no subsequent intervention can replace.

The system that delivers reperfusion is built out of minutes that no one defends by default. The work of acute stroke care is to defend them — one stroke alert at a time.