Acute ischaemic stroke remains a time-critical medical emergency in which rapid diagnosis, efficient prehospital triage, and timely reperfusion therapy are central to reducing mortality and long-term disability. At the Hong Kong Stroke Society Annual Scientific Meeting 2025, Dr David Hargroves of the East Kent Hospitals University Foundation Trust, UK, outlined England’s 7-year, system-wide quality improvement programme in acute stroke care, delivered through the Getting It Right the First Time (GIRFT) initiative.
Integrated stroke networks and best practice models
Reducing unwarranted variation
GIRFT is a National Health Service (NHS) quality improvement programme designed to reduce unwarranted variation — differences in practice, outcomes and costs that cannot be explained by patient factors. Under the co-leadership of HarÂgroves and Professor Deb Lowe of UK’s Stroke Association, the GIRFT stroke programme analyzed 3 years of national data, encompassing >250,000 patient episodes across 122 hospitals.
Rather than benchmarking perforÂmance in isolation, the programme emÂphasized collaborative improvement. ClinÂical teams participated in regional quality improvement events that enabled transÂparent discussion of variation and shared learning. “This was not a test, but an open-book exercise to understand where we could do better,” said Hargroves.
The resulting national GIRFT stroke report outlined 29 evidence-based recÂommendations spanning the entire stroke pathway. Key priorities included strengthÂening clinical leadership through initiatives such as the GIRFT Stroke Leadership Academy, developing a capability-based multidisciplinary workforce, optimizing pathways from prehospital care through rehabilitation, expanding equitable acÂcess to thrombectomy, and improving the quality and use of stroke data to support service improvement. (Figure 1) [https://gmnisdn.org.uk/wp-content/ uploads/2022/04/Stroke-GiRFT-report- Apr-22.pdf]
Development of integrated stroke networks
Building on GIRFT findings, England established 20 Integrated Stroke DelivÂery Networks (ISDNs), aligned around 24 mechanical thrombectomy centres. These networks expanded earlier hub-and-spoke hyperacute stroke unit (HASU) models by coordinating the entire stroke pathway, from prevention and acute care to rehabilitation and long-term support. ISDNs bring together commissioners, acute and community providers, and third-sector partners to manage capacity, and coordinate quality improvement and workforce development. [Future Healthc J 2022;9:118-124; www.england.nhs. uk/wp-content/uploads/2021/05/stroke-service-model-may-2021.pdf]
Transparency across networks alÂlows services to respond dynamically, for example when a thrombectomy centre reaches temporary capacity. “Effective stroke care depends on collaboration and honest communication,” noted HarÂgroves. “No single centre can do everyÂthing perfectly.”
Linking acute care to rehabilitation
“Hyperacute diagnosis and intervention are critical, but they represent only a small portion of the stroke patient’s journey,” highÂlighted Hargroves. “Most recovery occurs over the following weeks and months. ReÂhabilitation services must therefore be availÂable when patients need them.”
England’s Integrated Community Stroke Service model provides strucÂtured, multidisciplinary follow-up after hospital discharge, supporting recovery, secondary prevention, and long-term parÂticipation. This approach reinforces stroke care as a continuous pathway rather than a single acute episode. [www.england. nhs.uk/wp-content/uploads/2022/02/ stroke-integrated-community-serÂvice-february-2022.pdf]
Optimized thrombolysis: Cornerstone of acute care
Addressing undertreatment
Intravenous (IV) thrombolysis is an esÂtablished therapy that improves outcomes in acute ischaemic stroke. Despite this, thrombolysis rates in England remained at approximately 11.7 percent across the country for more than a decade — well below the ambition of 20 percent that may be eligible for treatment. [J Clin Med 2024;13:5826; SSNAP Annual Report 2025; www.england.nhs.uk/wp-content/ uploads/2021/05/stroke-service-model-may-2021.pdf]
To address this gap, Thrombolysis in Acute Stroke Collaborative (TASC) was launched in 2023. The programme targeted statistically significantly low-thrombolysing stroke units in the country and focused on pathway redesign, workÂforce training, and process improvement. [www.nhselect.nhs.uk/uploads/files/1//2- NHS-TASC-Toolkit-V2.pdf]
“The first cohort of six hospitals achieved a 45 percent relative increase in thrombolysis rates and a 20-minute reduction in door-to-needle times within 1 year,” said Hargroves. “Following exÂpansion to 12 additional units, the proÂgramme delivered a 62.6 percent average relative increase in thrombolysis rates in the next year, contributing to a rise in the national thrombolysis rate to 15 percent.”
Rapid diagnosis and treatment
Rapid admission to a stroke unit reÂmains a defining element of organized stroke care. Earlier thrombolytic treatÂment, ideally within 3 hours of symptom onset, is strongly associated with imÂproved functional outcomes. (Figure 2) [Lancet 2004;363:768-774]
Tissue-based decision making
While speed remains essential, UK stroke practice has increasingly shifted from rigid time windows to tissue-based decision making, supported by advanced imaging. “Current guidelines permit thrombolysis up to 9 hours from last-known-well, and recommend perfusion imaging when onset time is unknown,” explained Hargroves. CT perfusion (CTP) allows clinicians to identify salvageable brain tissue and exclude patients unÂlikely to benefit, enabling safe treatment beyond traditional time limits in selected cases. [www.strokeguideline.org]
The National Stroke Service ModÂel recommends 24/7 access to mulÂtimodal brain and vessel imaging, inÂcluding CT, CT angiography (CTA), CTP and MRI , supported by rapid image interpretation to guide timely clinical decisions. (Figure 3) [www.england.nhs. uk/wp-content/uploads/2021/05/stroke-service-model-may-2021.pdf]
Prehospital video triage
Diagnostic uncertainty remains a major obstacle in acute stroke care. PreÂhospital workers (eg, paramedics and ambulance technicians) must rapidly identify stroke patients, exclude mimics, and transport patients to the most apÂpropriate facility. One UK audit reported that over half of suspected stroke adÂmissions were ultimately diagnosed as suffering a stroke mimic. [www.strokeauÂdit.org/Documents/National/AcuteÂOrg/2021/2021-MimicReport.aspx]
To address this, prehospital video triage was introduced in pilot regions in England, enabling stroke specialists to assess patients remotely and supÂport conveyance decisions. EvaluaÂtions demonstrated reductions in hosÂpital conveyance of 13–19.5 percent, with fewer unnecessary admissions to HASUs. [Health Soc Care Deliv Res 2022;10.26]
Another study showed clinically meaningful time savings with prehospital video triage: door-to-CT was 40 minutes faster, door-to-thrombolysis 30 minutes faster, and door-to-groin puncture 50 minutes faster, highlighting the potenÂtial of early specialist input to streamÂline hyperacute pathways. [Eur Stroke J 2025;10:3-789]
Transforming hyperacute stroke pathways with AI
As imaging becomes more complex and time pressures increase in hyperÂacute stroke care, decision-support techÂnologies are emerging as practical tools to facilitate consistent, high-quality care.
AI in an overburdened workforce
“Clinicians and therapists spend a disproportionate amount of time on docÂumentation and administrative tasks,” noted Hargroves. “Artificial intelligence [AI] and ambient technologies have the poÂtential to reallocate cognitive effort back to direct patient care.”
AI-driven decision support
AI imaging software, including BrainoÂmix or RapidAI, is increasingly used to assist with interpretation of acute stroke imaging and support patient selection for interventions such as endovascular thrombectomy. (Figure 3)
A large prospective observational study in England evaluated the impact of AI implementation across 107 hospitals, encompassing >450,000 stroke admisÂsions. At 26 evaluation sites (n=71,017) where the AI software was adopted, thrombectomy rates increased from 2.3 to 4.6 percent post-implementation, repÂresenting a 100 percent relative increase. Among 747 patients at these sites with available transfer data, the median door-in door-out time was 128 minutes with AI vs 192 minutes without AI (Mann WhitÂney test, p<0.0001). [Lancet Digit Health 2026:100927]
At patient level, AI use was associated with higher likelihood of receiving thrombecÂtomy (odds ratio [OR], 1.57; 95 percent conÂfidence interval [CI], 1.33–1.86; p<0.0001) and IV thrombolysis (OR, 1.99; 95 percent CI, 1.79–2.22; p<0.0001) vs no AI use.
“[It is important to remember that] AI is not the panacea,” emphasized HarÂgroves. “It is an assistive tool that helps clinicians rapidly identify where to focus their attention.”
