Introduction

Treatment with tissue plasminogen activator (tPA) or thrombolysis is the most effective therapy for acute ischemic stroke patients within the first 4.5 hours following the onset of stroke symptoms [1], [2]. However, thrombolysis is substantially underused. Of all stroke patients, currently between 1–8% [3]–[5] are treated with tPA worldwide and around 11% (ranging from 4–26%) within the Netherlands [6], while 24–31% may be achieved in optimized settings [7], [8]. Reasons for this undertreatment are multi-factorial. Patients may not be familiar with the symptoms of a stroke or may not know how to act. Also, access to stroke care and expertise may vary by location [9], [10]. Other factors influencing the use of thrombolysis ensue from the organization of acute stroke care. Recently we demonstrated a 50% greater likelihood and up to 22% overall rate of treatment with thrombolysis achieved by a stroke center in a centralized organizational model versus nine community hospitals united in a decentralized organizational system of acute stroke care [11].

So far, it remains largely unclear how to improve thrombolysis delivery in decentralized stroke care; i.e. community hospitals. A method to study improvement of thrombolysis delivery is using Randomized Controlled Trials (RCTs). Although the benefits of RCTs have been clearly established in case of single interventions, their efficacy may be limited in case of complex patient pathways such as thrombolysis. For example, two recently published RCTs showed nonsignificant increases in thrombolysis rate following an intensive multicomponent intervention [12], [13]. Taken together, these studies focused on singular aspects of thrombolysis delivery (i.e. the in-hospital phase) and seem to yield very little gains- in terms of an increase in thrombolysis use- for the time and money invested. This warrants the question whether alternative research strategies are needed to study delivery of thrombolysis.

Various studies have shown how simulation models may be used as an efficient alternative or precursor to clinical trials, for example by predicting the prognosis after aortic heart valve replacement before implementation of the therapy [14], and studying the clinical benefit of reducing in-hospital delays to maximize the population benefit of thrombolysis in acute ischemic stroke [15]. As opposed to RCTs, simulation models allow for studying the effect of multiple factors along the entire stroke pathway at high pace and low costs.

The aim of this simulation study was to assess the impact of previously identified success factors in a central model on thrombolysis rates and patient outcome when implemented for a decentral model.

Methods

Results

Observational study – performance for organizational models

In the observational study, the difference in thrombolysis rate between the decentralized and centralized model was 7.8% (14.1% vs. 21.9%, respectively). Table 1 describes the performance for each factor investigated in both models, and Table 2 shows the results of the nine scenarios selected for this study on primary and secondary outcome measures. Details on the distributions underlying activity durations and diagnostic accuracy of both the decentralized- and centralized organizational model are presented in Table S1 and Table S2, respectively. Key activities of the acute stroke pathway and treatment decision are presented in Figure S1 and Figure S2, respectively.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Organizational model performance for the pre- and in-hospital acute stroke pathway.

https://doi.org/10.1371/journal.pone.0079049.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Re-configuration decentralized model: results simulation experiments.

https://doi.org/10.1371/journal.pone.0079049.t002

Secondary outcomes

Improving all factors led to an 11.4% increase in patients with good functional outcome, of which 8.7% ascribed to pre-hospital factors and 2.7% to in-hospital factors. The OTT decreased 17 minutes; of which 7 minutes ascribed to pre-hospital factors and 10 minutes to the in-hospital factors. A shift was observed in the proportion of patients thrombolyzed within 1.5 hours; increasing by 14.1%, of which 5.6% ascribed to pre-hospital factors, and 8.5% to in-hospital factors. Treatment between 1.5–3.0 and 3.0–4.5 hours decreased by 9.3% and 4.8%, respectively; of which 4.3% and 2.2% attributed to pre-hospital factors and 5.0% and 2.6% to in-hospital factors, respectively. Figures 1 and 2 show the proportion of patients treated with thrombolysis according the OTT intervals and the proportion of patients with good functional outcome, respectively.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Number of patients treated with thrombolysis according to onset-treatment-time (OTT) intervals.

https://doi.org/10.1371/journal.pone.0079049.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Patient outcome at 90 days for onset-treatment-time (OTT) intervals.

https://doi.org/10.1371/journal.pone.0079049.g002

Discussion

This simulation study quantifies the potential benefit of implementing success factors from a centralized organizational model in a decentralized model. Combining all factors raised thrombolysis rates by 7.9% thereby closing the gap observed between both models. Furthermore, our model demonstrates that the proportion of patients with a good functional outcome (mRS 0–1) may increase by 11.4% and those thrombolyzed within 1.5 hours by 14.1%.

This simulation study provides an example of an alternative research method to study delivery of thrombolysis in acute ischemic stroke. Contrary to traditional approaches that evaluate improvement strategies (i.e. RCTs), simulation models may be better suited to investigate multiple factors of stroke system organization simultaneously and allow for quicker answers at lower costs [15]. Instead of time-consuming and expensive trials such as double-blinded randomized trials, we advocate action research. This means implementing critical success factors determined from the simulation study in the existing acute stroke pathway and study the effects of interventions with baseline measurements. Simulation experiments obviously precede implementation to check whether comparable results can actually be achieved in clinical practice. By using the results obtained from simulation, implementation studies may be tailored towards improving factors impacting most on thrombolysis rates and patient outcome, while excluding others that do not.

Based on the identified potential for improving thrombolysis rates and patient outcomes pre-hospital factors seem primary targets for change. The importance of the pre-hospital phase in advancing the acute stroke care pathway is supported by a recent review article emphasizing that every link in the care pathway matters and should be studied for potential improvements [18]. In addition, we observed an interaction effect between improving symptom onset to call for help and EMS transport. This interaction effect may be explained by a volume effect – meaning more patients arriving at the hospital in time for thrombolysis, and by shortening the time to treatment for those already thrombolyzed. As treatment rates with thrombolysis are relatively high in our study compared to international standards, we would expect even larger effects when implemented in regions with low treatment rates.

Our study has limitations. Firstly, we were limited by the documentation of symptom onset times by ambulance and hospital personnel, i.e. possibly some of the ‘unknown’ onset times were actually known but never documented. However, the proportion of patients with unknown or estimated onset times was comparable to or even better than in previous studies [19], [20]. Secondly, the interaction between hospital size, i.e. stroke center and community hospitals, may be difficult to interpret because requirements of stroke centers are such that community hospitals are unlikely to meet them.

Conclusions

This study demonstrates that simulation may be employed as research tool to target opportunities for improving the acute stroke care pathway and therefore thrombolysis rates. Pre-hospital factors proved to be the most promising targets for improving acute stroke care in an implementation study.

Supporting Information

Acknowledgments

Members of the study group

We wish to express our gratitude to all participating hospitals, emergency medical services, and general practitioner out of hours co-op for providing data for this study: S.Schade van Westrum, MD, Martini Hospital Groningen, Netherlands, K.M.Horvath, MD, PhD, Refaja Hospital Stadskanaal, Netherlands, A.van As, MD, Ommelander Hospital Group, Location Delfzicht Delfzijl, Netherlands, C.Felisiak, MD, Ommelander Hospital Group, Location Lucas Winschoten, Netherlands, W.J.Schuiling, MD, PhD, Medical Center Leeuwarden, Netherlands, R.F.Duyff, MD, Hospital The Tjongerschans Heerenveen, Netherlands, K.Beintema, MD, Hospital Nij Smellinghe Drachten, Netherlands, A.M.H.P. Boreas, MD, PhD, Antonius Hospital Sneek, Netherlands, G.Sulter, MD, PhD, Hospital Sionsberg Dokkum, Netherlands, J.N.Wessel, MD, Wilhelmina Hospital Assen, Netherlands, Netherlands, L.Koops, MD, and J.Trip, MD, Diaconessen Hospital Meppel, Netherlands, C.de Langen, MD, en P.Oomes, MD, Bethesda Hospital Hoogeveen, Netherlands, R.de Vos, MD, Emergency Medical Services Groningen, Netherlands, T.Sandjer, MD, Emergency Medical Services Drenthe and Friesland, Netherlands, R.Kijlstra, Emergency Medical Services Friesland, Netherlands, J.Post, MD, PhD, General Practitioner Out Of Hours Co-op Groningen, Netherlands.