Improving stroke pathway efficiency: outcomes of a quality improvement collaborative across a national stroke network

Abstract

Background

Timely endovascular thrombectomy (EVT) is crucial for improving outcomes in acute ischemic stroke (AIS). This study evaluated the effectiveness of a national quality improvement collaborative (QIC) in reducing process times for potential EVT candidates across a national stroke network.

Methods

A pre-post intervention design using a modified Breakthrough Series approach was implemented across 24 hospitals. Multidisciplinary teams participated in monthly learning sessions and action periods focused on reducing ‘Door to Decision’ (time from hospital arrival to EVT decision) to under 30 min. Mixed-effects linear models and mixed-effects ANOVA were used to analyse the impact of the QI program on Door to Decision and Door to CT times, comparing intervention and control cohorts.

Results

The QI program significantly reduced Door to Decision time in the intervention cohort by 15.9% (p<0.001) from a mean of 92.8 min to 78.9 min. Door to CT time also decreased by 15.6% (p<0.001). No significant changes were observed in the control cohort. Mixed-ANOVA revealed a significant interaction effect for both Door to Decision (p<0.004) and Door to CT (p<0.04), indicating that the QI program impacted these times as compared with the control group. The QIC effectively improved the efficiency of stroke care pathways across a national stroke network. This effect was sustained across the network and over time. This success was facilitated by a bottom-up approach, fostering collaboration and shared learning within and across hospitals.

Conclusions

This study demonstrates the effectiveness of a collaborative, network-wide QI program in reducing critical process times for AIS patients. Continued efforts to sustain these improvements and optimize stroke care pathways are warranted.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Rapid treatment is essential for optimal stroke outcomes. While individual hospitals have implemented quality improvement (QI) initiatives to reduce in-house delays, achieving network-wide improvements in stroke care efficiency remains challenging.

WHAT THIS STUDY ADDS

• This study demonstrates that a national QI project can significantly reduce key process times, such as Door to Decision and Door to CT, for potential endovascular thrombectomy candidates across a network of hospitals.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• This study supports the adoption of collaborative QI approaches to improve stroke care efficiency at a national level. It highlights the value of empowering local teams and fostering shared learning across hospitals to achieve meaningful and sustainable improvements in stroke care delivery.

Introduction

Endovascular thrombectomy (EVT) is the standard of care for all eligible patients presenting with acute ischemic stroke (AIS) due to a large vessel occlusion. The speed at which the patient receives revascularization remains a critical factor in improving stroke patient outcomes,^{1 2} with each minute of delay being associated with a decrease in the likelihood of a good functional outcome.

The acute treatment of stroke across a system of care presents considerable logistical and organizational challenges, particularly within the ‘hub-and-spoke’ thrombectomy center model. This model involves multiple steps: patient identification and transport to a local stroke center by emergency medical services, local assessment, rapid imaging, consultation with a comprehensive stroke center for thrombectomy, patient transfer, the thrombectomy procedure itself, and repatriation following the procedure.³ While previous work has focused on improving in-house process times for stroke patients,⁴,⁶ these have tended to focus on single-center improvements. This is perhaps reflective of the difficulties in engaging and sustaining change across a broad network of care, given the disparate stakeholder and elements involved.

Recognizing these issues in Ireland, the National Thrombectomy Service (NTS), along with the Royal College of Physicians of Ireland (RCPI) and the Royal College of Surgeons of Ireland (RCSI), embarked on a quality improvement collaborative (QIC) project entitled ‘Door to Decision in Under 30!’. This initiative aimed to reduce the process times in patients with acute stroke. We focused primarily on the time from patient arrival at the primary hospital to clinical and radiological diagnosis and decision-making for EVT (Door to Decision), with a target of making this decision within 30 min. This was obtained through development and implementation of a comprehensive QI program. The primary objective of this study was to evaluate the effectiveness of a national QIC in reducing process times for acute ischemic stroke patients across the Irish stroke network.

Specifically, we aimed to:

• Measure the impact of the QI program on 'Door to Decision' time for potential thrombectomy candidates as a primary outcome measure, with a target of achieving this decision within 30 min of hospital arrival.

• Assess changes in 'Door to CT' time as a secondary outcome measure.

• Compare these metrics between hospitals participating in the QI program and those not participating, to determine the specific impact of the intervention.

• Evaluate the effectiveness of a collaborative, network-wide approach to improving stroke care efficiency across multiple hospital sites.

Methods

Study design

The study comprised a pre-post intervention design, implementing a QI program based on a modified Breakthrough Series approach, adapted from the Institute for Healthcare Improvement Breakthrough Series model.^{7 8} This model involved multidisciplinary teams (MDTs) from hospitals across the stroke network participating in monthly learning sessions, interspersed with action periods during which teams implemented the learned strategies in their respective hospitals. Detailed descriptions of the methodology can be found in the online supplemental appendix.

QI program implementation

Timeline and structure

The QI program was coordinated by the NTS at the mothership hospital, commencing in January 2018 and running through October 2020. The program featured formal learning sessions led by expert QI faculty, interspersed with action periods during which the QI leads provided ongoing support. Although the formal learning sessions concluded in 2019, QI leads continued to support participating hospitals, with data collection and service monitoring extending into 2023. Hospital visits by the QI leads also persisted through 2023, ensuring sustained engagement and continuous improvement.

Participating hospitals and team composition

Twenty-four hospitals across the Irish stroke network participated in the QI program, with an additional four hospitals providing data to the NTS during the study period. Participating hospitals were divided into two cohorts: QI 1 and QI 2. Hospitals were given the opportunity to participate in the QI program at its inception. The decision to participate was voluntary and based on each hospital’s assessment of their local resources and interest in engaging in the initiative. Hospitals not actively participating in the QI program, referred to as the non-QI group, served as a natural control group. The control hospitals, while providing valuable data, did not actively participate in the learning sessions or implement the QI interventions during the study period. This highlights the bottom-up, collaborative nature of the program, which prioritized local ownership and voluntary engagement over a top-down, mandated approach.

Each participating hospital identified a subset of their broader stroke MDT to attend the monthly learning sessions. These MDTs included key stakeholders involved in stroke management, such as emergency care clinicians, stroke physicians, clinical nurse specialists, radiologists, radiographers, and hospital administrators. Institutional support was critical for these teams to attend training sessions, adjust rosters, and carry out the testing and implementation phases of their QI initiatives. To facilitate this, each hospital established a local steering group with key leaders responsible for governing and guiding the QI process.

Teaching strategy and methods

The monthly learning sessions, guided by expert QI faculty, focused on specific themes rooted in the Institute for Healthcare Improvement’s Model for Improvement. These themes included healthcare system understanding, change management, goal setting, measurement for improvement, understanding variation, and testing changes through Plan-Do-Study-Act (PDSA) cycles. The QI program utilized a hybrid format, combining lectures, group exercises, and problem-based learning. Interactive activities, such as simulation exercises, were designed to effectively merge theoretical knowledge with practical application. This instructional approach was delivered both online and in-person, optimizing engagement and enhancing learning outcomes.

During the action periods between learning sessions, each local hospital team employed QI tools to analyze their stroke care pathways. Supported by QI leads both remotely and on-site, teams focused on identifying specific friction points and bottlenecks that impeded patient flow. Root cause analysis was performed to pinpoint areas for improvement, and PDSA cycles were used to test and refine interventions. This approach allowed for the development of customized solutions tailored to the unique challenges of each site, avoiding a ‘one-size-fits-all’ methodology.

Interventions

A literature review was conducted to identify interventions in the AIS pathway and benchmark against best international practices. Interventions were mapped to specific steps in the patient journey from pre-hospital through to the decision regarding suitability for thrombectomy and onwards to definitive intervention, as needed.

QI support

The program was funded internally by the ‘mothership’ hospital, which provided one full-time equivalent (FTE) post which was dedicated to running the QI program for the duration of the program. The program was directed by one of the coauthors (JT). Regular hospital site visits by the NTS QI leads were instrumental in facilitating exercises to understand local patient journeys and pinpoint improvement areas. They played a pivotal role in coordinating local teams, challenging existing processes, and translating QI methodology from theory into practical teaching to ensure the sustainability of improvements, as well as ongoing data collection. Crucially, a focus on sharing learnings from individual sites across the broader stroke network was a key component of the program, thus facilitating the propagation of solutions and success stories. This enabled other sites to adapt these proven solutions to meet their specific needs, thereby enhancing the overall effectiveness of stroke care across the broader stroke network. Importantly, the QI leads focused on supporting rather than directing, encouraging local teams and initiatives, and involving ‘local champions’ to drive processes forward. This was a deliberate ‘bottom-up’ supportive approach, contrasting with a ‘top-down’ directive from the national strategy or the central mothership thrombectomy site.

Data collection and analysis

Data collection

Inclusion criteria were any AIS patients who underwent thrombectomy and who had been previously flagged/identified as FAST+ (Face, Arm, Speech, Time positive) during the study periods. The collected timepoints included the time of arrival at hospital, time of CT scan (as measured by the initial localizer sequence), and the recorded time of decision regarding suitability for thrombectomy (Door to Decision).

Baseline data for QI 1 were gathered between January 2016 and October 2018, while baseline data for QI 2 were collected between January 2016 and June 2019. Following the interventions, post-intervention data for QI 1 were recorded from November 2018 to October 2020. Similarly, post-intervention data for QI 2 were collected from July 2019 to June 2021, allowing for a comprehensive analysis of the impact of the learning sessions on the stroke care pathway (see online supplemental appendix for more details).

All data were collated and analyzed centrally by the QI leads on an ongoing basis for both annual reporting and to assess the impact of the various initiatives in real time.

Data analysis

For analysis, two cohorts were created: an intervention cohort and a control cohort. The intervention cohort’s baseline period included all Door to Decision and Door to CT time points for both QI 1 and QI 2 up until the end of their respective baseline periods. The post-intervention period included all Door to Decision and Door to CT time points for both QI 1 and QI 2 hospitals for the 2 years following the completion of their respective interventions. The control cohort included all Door to Decision and Door to CT time points from hospitals not participating in the QI intervention. The baseline and post-intervention periods for the control cohort followed the same dates as the QI2 group (see online supplemental appendix).

Entries with missing values were excluded from both cohorts. Door to CT and Door to Decision times were positively skewed; therefore, a natural log transformation was applied to normalize the data prior to analysis, ensuring the validity of the parametric tests used. Baseline and post-intervention periods were then compared to identify significant changes in these key time metrics.

The primary outcome measure analyzed was the ‘Door to Decision’ time, with the ‘Door to CT’ time as the secondary outcome measure. All data analysis was performed using Python 3 in Google Colaboratory. To assess the impact of the QI intervention, mixed-effects linear models were employed. In these models, the cohort (intervention vs control) was treated as a fixed effect, while the hospital was treated as a random effect to account for differences across the various hospital sites. This approach allowed for the consideration of both the overall effect of the intervention and the variations between individual hospitals.

Mixed-ANOVA (analysis of variance) analyses were used to further compare the differences between groups.

Results

Overview

The QI program was implemented in 24 hospitals across the Irish stroke network, with a further four hospitals also providing data to the NTS during the course of the QI program. The analyses included 28 hospitals with 909 observations for Door to Decision time and 1126 observations for Door to CT time.

Primary outcome: Door to Decision time

The mean baseline Door to Decision time was 92.75 min (SD 92.31) in the intervention cohort, compared with 81.33 min (SD 48.81) in the control group (table 1). Following the intervention, the mean Door to Decision time in the intervention cohort decreased to 78.94 min (SD 61.64), representing a 15.97% reduction (SE 4.29%, p<0.001) (see online supplemental appendix). In contrast, the control cohort showed no significant change in Door to Decision times during the post-intervention period (p=0.622).

Table 1. Impact of quality improvement program on Door to Decision times before and after the intervention.

Primary endpoint: Door to Decision times
Cohort	Timepoint	n	Median (min)	IQR	SE	95% CI	P value
Intervention	Before	400	68.54	57.60	5.78	64.17 to 74.86	<0.001
	After	409	62.22	49.58	3.82	58.33 to 67.08
Control	Before	64	67.57	51.77	7.64	54.44 to 82.64	0.622
	After	36	69.03	40.59	28.70	58.33 to 79.72

CI, confidence interval; IQR, interquartile range; SE, standard error.

A mixed-design ANOVA was performed in order to evaluate the effects of cohort (intervention vs control) and time point (before vs after the intervention period) on the Door to Decision time.

• Main Effect of Cohort: The analysis revealed no significant difference in Door to Decision times between the intervention and control cohorts when considered as a whole (p=0.128). This indicates that the average times were not significantly different between groups overall.

• Main Effect of Timepoint: Similarly, there was no significant change in Door to Decision times when comparing the pre- and post-intervention periods across all participants, irrespective of cohort (p=0.265). This suggests that general improvements in hospitals over time did not account for any significant differences in Door to Decision times in the various cohorts.

• Interaction Effect: A significant interaction effect was, however, observed between cohort and timepoint (p=0.004), indicating that the change in Door to Decision time before and after the intervention period differed significantly between the intervention and control cohorts. This interaction suggests that the QI program had a significant impact on reducing Door to Decision time, but this effect was primarily evident when comparing the temporal changes within each cohort (online supplemental figure 1: Interaction plot of Door to Decision Time by timepoint and cohort).

In summary, this mixed-effects analysis shows that the interaction between the various cohorts and the time of the intervention had a significant impact on the Door to Decision time compared with the control group. In other words, the temporal relationship of the intervention to the outcome was significant. (These results are summarized in the online supplemental appendix.) In addition, significant differences in the Door to Decision time were observed before and after the intervention periods for both the QI1 (p=0.002) and QI2 (p=0.004) groups separately (see online supplemental appendix).

Secondary outcome: Door to CT time

The mean baseline Door to CT time was 35.45 min (SD 36.82) in the intervention cohort, which was not significantly different from the control cohort (p>0.999) (table 2). After implementing the QI program, the mean Door to CT time in the intervention cohort decreased to 32.16 min, representing a 15.63% reduction (p<0.001). In contrast, the control cohort showed no significant change in Door to CT times during the post-intervention period (p=0.914).

Table 2. Impact of quality improvement program on Door to CT times before and after the intervention.

Secondary endpoint: Door to CT times
Cohort	Timepoint	n	Median (min)	IQR	SE	95% CI	P value
Intervention	Before	530	27.22	25.28	2.00	24.31 to 29.17	<0.001
	After	489	22.36	21.39	2.02	21.39 to 24.31
Control	Before	71	29.17	23.33	4.93	24.31 to 32.08	0.914
	After	36	26.74	25.52	20.18	20.42 to 33.54

CI, confidence interval; IQR, interquartile range; SE, standard error.

A mixed-design ANOVA was again conducted to evaluate the effects of cohort (intervention vs control) and time point (before vs after the intervention period) on the Door to CT time.

• Main Effect of Cohort: There was no statistically significant difference between the intervention and control cohorts (p=0.402).

• Main Effect of Timepoint: There was no significant change in Door to CT time from before to after the intervention period (p=0.397).

• Interaction Effect: There was a significant interaction effect between cohort and timepoint (p=0.032), demonstrating that the change in Door to CT time from before to after the intervention period differed significantly between the intervention and control cohorts (online supplemental figure 2: Interaction plot of Door to CT Time by timepoint and cohort).

Significant improvements in Door to CT times were independently noted in both QI1 (p=0.002) and QI2 (p=0.019) after the intervention (see online supplemental appendix) demonstrating the effectiveness of the strategies. This suggests that the interventions were successful across different settings within the study.

Discussion

Our findings demonstrate that a nationwide QIC can significantly and sustainably reduce critical process times for AIS patients across a network of hospitals. Specifically, we observed a 16% reduction in both Door to Decision and Door to CT times in the intervention cohort, an effect not seen in the control group. The significant decrease in the standard deviation also highlights that following the implementation of the QI program, variability in patient process times through the system were also reduced. This interaction effect highlights the direct impact of the QI program on streamlining these time-sensitive stages of the stroke care pathway. Notably, this improvement was not a general trend across hospitals over time, as evidenced by the lack of a similar effect in both Door to Decision and Door to CT analyses in the control group. This underscores the specific impact of the QIC in driving these positive changes.

The observed reduction in Door to Decision time is particularly interesting, as this metric reflects the complex interplay of multiple stakeholders within the initial assessment and diagnostic phase. This improvement signifies enhanced communication, coordination, and resource utilization across emergency care clinicians, stroke physicians, radiologists, radiographers, and hospital administrators, all working in concert towards the shared goal of timely EVT decision-making. Moreover, the reduction in Door to CT times reflects a successful optimization of the very first (and often overlooked) stage of in-hospital stroke care, emphasizing the importance of streamlined processes and efficient resource allocation on patient arrival.

The 'bottom-up' approach adopted by this QI initiative is a key driver of its success. By empowering local teams, fostering a collaborative learning environment, and sharing best practices across the network, the program fostered a sense of ownership and engagement that is crucial for sustainable change. This contrasts with top-down mandates which are often met with resistance and limited long-term impact. The dedicated QI leads provided vital support, facilitating knowledge transfer, process mapping, problem-solving, and promoting cross-hospital learning, thus creating a network of shared expertise.

While our study focused on process time reduction, it is important to acknowledge that the ultimate goal of any stroke care intervention is to improve patient outcomes. Although this study was not powered to detect changes in functional outcomes, existing literature demonstrates a strong correlation between faster treatment times and better outcomes for AIS patients. Future research is needed to investigate the effect of these process time improvements on patient outcomes, and a focus on process times should be a component of future EVT trials.

Our results are consistent with prior QI initiatives in stroke care published in this Journal. Busby et al demonstrated significant reductions in door-to-needle times using the 'CODE FAST' protocol, emphasizing streamlined processes and team coordination.⁴ Similarly, Rai et al employed a 'pit-crew' model, achieving faster ER-to-CT, CT-to-angio lab, and lab-to-puncture times for EVT through a Six Sigma approach.⁹ Asif et al reviewed prehospital and in-hospital delays to mechanical thrombectomy, identifying key areas for improvement,¹⁰ while Aghaebrahim et al showed that a standardized, multidisciplinary approach could significantly reduce door-to-puncture times.¹¹ While these studies primarily focused on single-center improvements, our study extends these findings by demonstrating the effectiveness of a collaborative, national QI program, highlighting the potential for widespread improvements in stroke care through shared learning and local adaptation within a diverse hospital network. However, while encouraging, our results also highlight that there is still room for improvement. Achieving Door to Decision times consistently under 30 min remains a challenge for many hospitals, emphasizing the need for continued efforts to refine and optimize stroke care pathways, particularly during off-hours. Future research should explore the impact of various transport paradigms on EVT outcomes, especially in the context of mobile stroke units and drip-and-ship versus mothership models.

It is important to acknowledge the limitations inherent in this study’s design. As a pre-post intervention study without randomization, we cannot definitively rule out the influence of confounding factors unrelated to the QI program that may have contributed to the observed improvements. Additionally, the study relied on retrospective data collection, which may be subject to inaccuracies and missing data points. Future prospective studies with larger sample sizes and robust outcome measures are needed to further validate the impact of this QIC on patient outcomes and to explore the sustainability of improvements over the long term. Furthermore, the generalizability of our findings to other healthcare systems may be limited due to the unique context of the Irish stroke network.

Our findings underscore the need for ongoing national-level efforts to promote and support collaborative QI initiatives within stroke networks. By investing in training, fostering inter-hospital collaboration, and prioritizing data-driven process improvement, healthcare systems can effectively translate the growing body of evidence for time-sensitive stroke interventions into tangible improvements in patient care and outcomes. This study serves as a model for other nations seeking to enhance stroke care efficiency through a collaborative, network-wide QI approach.

Conclusions

This study demonstrates that a collaborative, network-wide QI program can significantly reduce key process times, such as Door to Decision and Door to CT, for potential EVT candidates across a network of hospitals. By empowering local teams and fostering shared learning, this bottom-up approach resulted in sustained improvements in stroke care efficiency, underscoring the potential of such initiatives to enhance stroke care delivery at a national level.

Data availability statement

No data are available.