Island-Wide Stroke Bypass Protocol Cuts Treatment Times by Over an Hour on Oahu

When a major stroke strikes…

The Oahu experience offers a rare window into how a prehospital triage policy behaves when the usual confounders of real-world stroke systems are largely removed. In most metropolitan regions, patients with suspected large vessel occlusion can be routed to any of several thrombectomy-capable hospitals, each with its own workflows, staffing patterns, and activation criteria. That heterogeneity makes it difficult to attribute changes in treatment times to the bypass policy itself rather than to differences between receiving centers. On Oahu, by contrast, a single comprehensive stroke center receives all endovascular candidates, decisions about mechanical thrombectomy are made by one unified team of stroke neurologists and neurointerventionalists, and a single EMS agency handles prehospital transport for the entire island of roughly one million residents within 597 square miles. This closed-loop structure means that any shift in onset-to-door time observed after August 1, 2019—the date the bypass protocol took effect island-wide—can be interpreted with considerably more confidence than in open systems where patient migration between competing centers muddies the analysis.

The mechanics of the protocol deserve attention because they illustrate how prehospital stroke science is translated into dispatch-level practice. Under the Oahu scheme, EMS providers administer the Cincinnati Stroke Triage Assessment Tool, a brief field examination designed to discriminate cortical large vessel occlusions from smaller infarcts that do not benefit from thrombectomy. Patients scoring two or higher are classified as bypass-positive and transported directly past the nearest community hospital to the comprehensive stroke center, even when that nearest hospital would otherwise have been only minutes away. Before the policy change, standard EMS doctrine directed transport to the closest emergency department regardless of stroke severity, with exceptions reserved for major trauma or provider discretion. The consequence of the old doctrine was predictable: a patient with a disabling LVO would arrive at a community hospital, undergo initial evaluation and imaging, be recognized as needing endovascular therapy, and then require a second ambulance transfer to the comprehensive stroke center—consuming the very hours during which penumbral brain tissue remains salvageable.

The five-fold decline in interfacility transfers for endovascular therapy, from 55.9 percent of cases before implementation to 11.6 percent afterward, is arguably the most operationally meaningful finding in the study. Interfacility transfer is the single largest source of avoidable delay in stroke systems of care, because it compounds two ambulance trips, two emergency department evaluations, two imaging workflows, and two handoffs into a single treatment timeline. National quality registries have repeatedly shown that patients who require a drip-and-ship transfer to reach a thrombectomy-capable center lose substantial time compared with those who arrive directly. By collapsing that two-step journey into one, the Oahu protocol eliminated the dominant structural delay in the island’s treatment chain, and the 66-minute sustained reduction in onset-to-door time measured by the interrupted time series analysis is the direct arithmetic consequence of that collapse.

The interrupted time series design strengthens the causal interpretation in ways that a simple before-and-after comparison cannot. By grouping treatment times into quarterly averages and modeling both the change in level and any change in slope after implementation, the analysis distinguishes an abrupt policy effect from a gradual secular trend. The Durbin-Watson assessment for autocorrelation further guards against the possibility that the observed drop merely reflects a pre-existing downward drift in treatment times that happened to coincide with the protocol launch. The finding of a sustained level shift with no change in temporal trend indicates that the bypass produced a discrete, durable improvement rather than accelerating an ongoing improvement that would have occurred anyway.

Equally important is what did not get worse. A persistent concern with bypass protocols is that routing the sickest patients past community hospitals might delay thrombolysis for patients who turn out not to have an LVO, or who would have benefited from rapid intravenous thrombolytic administration at the nearest facility. The Oahu data showed no significant change in onset-to-thrombolytic time after implementation, which suggests that the field screening tool and the routing logic did not systematically penalize patients who still needed intravenous reperfusion therapy. This finding aligns with the broader literature on LVO scales, which generally accept a modest trade-off in thrombolysis timing in exchange for large gains in thrombectomy timing, but it is reassuring to see that trade-off was not required on Oahu.

The relationship between time and outcome that emerged from the multivariate modeling reinforces why these minutes matter. Every additional 30 minutes of delay from symptom onset to successful reperfusion increased the odds of a worse 90-day modified Rankin Scale score by six percent, an effect consistent in direction and magnitude with the time-to-treatment gradients reported in randomized thrombectomy trials and pooled patient-level meta-analyses. The biology underlying this gradient is well characterized: each minute of persistent arterial occlusion converts salvageable penumbral tissue into irreversibly infarcted core, so the functional benefit of reperfusion decays continuously with time. A 41-minute reduction in onset-to-reperfusion time, even one that fell short of conventional statistical significance in this cohort, would therefore be expected to translate into a modest but real shift in the distribution of disability—provided the cohort is large enough and the time savings reach the patients who stand to gain the most.

That last caveat is where the Oahu results become most instructive about the limits of systems interventions. Despite the substantial and sustained reduction in treatment times, the post-bypass era was not associated with a statistically significant improvement in good 90-day outcomes, defined as modified Rankin Scale scores of zero to two, and the adjusted odds ratios for both good outcomes and the ordinal shift in disability hovered near unity. There was, however, a suggestive trend toward more excellent outcomes—scores of zero or one—which rose from 23.5 percent to 34.5 percent in unadjusted comparison and retained a positive but non-significant adjusted association. The most plausible reading is that the study was underpowered to detect the modest outcome benefit that a 41-minute time saving would be expected to produce. With 102 patients in the pre-bypass era and 249 in the post-bypass era, and with functional outcome influenced by many factors beyond timing—age, baseline stroke severity, clot location, reperfusion success, and symptomatic hemorrhage among them—the signal from a time saving of this size may simply be diluted below the threshold of detectability.

Several other factors may also have blunted the observable outcome benefit. The institutional criteria for mechanical thrombectomy during the study period required a baseline modified Rankin Scale of zero to two, an NIH Stroke Scale score of at least six, and an ASPECTS score between six and ten, which restricts the treated population to patients with moderate to severe anterior circulation strokes and reasonable pretreatment function. Patients at the milder end of the spectrum, who might have gained the most from faster reperfusion in absolute terms of preserved independence, were not eligible for the procedure at all. Additionally, the bypass protocol changed not only where patients went but who arrived at the comprehensive stroke center; if the post-bypass cohort included more severe or more bypass-positive patients by design, case-mix shifts could offset some of the timing gains in the adjusted models. The investigators addressed this by adjusting for age, baseline NIHSS, clot location, reperfusion status, and symptomatic hemorrhage, but residual confounding in a retrospective design is inevitable.

The comparison with RACECAT and TRIAGE-STROKE, the two randomized trials of prehospital bypass strategies, adds an important layer of interpretation. Both trials confirmed that direct transport to thrombectomy-capable centers shortens onset-to-puncture times, yet neither demonstrated improved functional outcomes, mirroring the Oahu pattern of time savings without proven outcome benefit. The convergence of these findings across a randomized European trial, a randomized Catalan trial, and a natural experiment in a closed island system suggests that the time-outcome relationship, while real at the individual patient level, is difficult to demonstrate at the system level when the time savings are measured in tens of minutes rather than hours. It also raises the possibility that the benefit of bypass is heterogeneous: patients with the longest prehospital delays, the largest salvageable penumbras, or the most accessible occlusions may benefit substantially, while the cohort average dilutes that benefit.

For stroke systems planners, the practical lessons are nonetheless concrete. First, a simple, validated field tool administered by a single EMS agency can reliably redirect LVO candidates and produce large, durable reductions in door time without harming thrombolysis delivery. Second, the magnitude of benefit from bypass depends heavily on the baseline system: in regions where interfacility transfer rates are as high as the 55.9 percent seen on Oahu before implementation, the headroom for improvement is enormous, whereas systems that already route most patients directly may see smaller gains. Third, geography matters. Oahu’s compact, single-center catchment meant that bypass added relatively little transport time for most patients; in sprawling regions where the nearest thrombectomy-capable center is far away, the calculus of bypass versus nearest-hospital-first becomes more complex and may favor drip-and-ship for some patients. Finally, the Oahu experience underscores that process metrics, while necessary, are not sufficient—demonstrating that a systems intervention improves the lives of stroke patients ultimately requires either larger cohorts, longer follow-up, or composite endpoints sensitive enough to capture the modest population-level shifts that faster reperfusion can deliver.

Subject of Research: Prehospital large vessel occlusion bypass protocol and its effect on endovascular stroke treatment times and outcomes

Article Title: Reducing onset-to-door times through the implementation of a prehospital large vessel occlusion bypass protocol: the Oahu experience

Article References: Sun, C.-H., Chun, I., Nakagawa, K., Brown, S., Tsappidi, S., Hui, F., Zhang, Y., Salem, R., Zhang, G., Galanis, D., & Koenig, M. (2026). Reducing onset-to-door times through the implementation of a prehospital large vessel occlusion bypass protocol: the Oahu experience. Frontiers in Stroke, 5, Article 1903852. https://doi.org/10.3389/fstro.2026.1903852

Image Credits: AI Generated

DOI: 10.3389/fstro.2026.1903852

Keywords: stroke, large vessel occlusion, endovascular thrombectomy, EMS bypass protocol, prehospital triage, onset-to-door time, reperfusion, comprehensive stroke center, Oahu, interrupted time series, modified Rankin Scale, mobile stroke units