Article Text
PDF
Abstract
Objectives Regional accessibility and distribution of endovascular thrombectomy (EVT) capable facilities, that is, comprehensive stroke centres (CSCs), may significantly influence time to treatment. We analysed the impact of adding CSCs in the north of the Netherlands, a region with roughly 1.7 million inhabitants currently served by one CSC and eight primary stroke centres (PSCs).
Design Monte Carlo simulation modelling was used to establish new CSCs in our region by hypothetically upgrading existing PSCs to CSCs and ensuing adjustments in health services set-up.
Setting One CSC and eight PSCs in the north of the Netherlands.
Participants 165 patients with acute stroke treated with EVT and underwent interhospital transfer between PSC and CSC (drip and ship patients).
Primary and secondary outcomes Time from onset to groin (OTG) puncture and predicted probability of favourable outcome (modified Rankin Scale 0–2) after 90 days. Sensitivity analyses were performed to assess uncertainty in workflow efficiency of CSCs.
Results Adding one or two CSCs would reduce the OTG time up to approximately 17 min and increases the predicted probability of favourable outcome by approximately 2%. Sensitivity analyses revealed that ‘slow-acting’ CSCs may reduce OTG by 3–5 min compared with 24–32 min for ‘fast-acting’ CSCs.
Conclusions This study suggests that adding one or two CSCs in the north of the Netherlands would have modest impact. Improving workflow efficiencies seems to be more potent when aiming to improve existing acute stroke care systems.
- stroke
- epidemiology
- organisation of health services
Data availability statement
The participants of this study did not give written consent for their coded data to be shared publicly, so due to the sensitive nature of the research supporting data is not available. Nevertheless, used distributions of the collected data are available in the supplemental material.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
Strengths and limitations of this study
The results of this study adds important knowledge on the potential benefits of adding comprehensive stroke centres in underserved rural regions.
Input data for the developed models represent patient level data both from the hospital and prehospital setting thereby accurately representing clinical practice.
Simulation modelling is presented as a flexible and efficient methodological approach which has the potential for broader use by using input data from other regions.
Does not include the broader patients with stroke population of non-large vessel occlusion and haemorrhagic subtypes.
The simulation model includes time delay parameters on system performance that may have changed over time.
Introduction
Fast treatment of acute ischaemic stroke (AIS) due to large vessel occlusion (LVO) is pivotal to improve functional outcome.1 The effects of both intravenous thrombolysis (IVT) and endovascular thrombectomy (EVT) for AIS are highly time dependent. that is, every hour delay implies a 5%–6% decline in the chance of a favourable outcome.2 3
Two main organisational models for acute stroke care currently exist. In the mothership (MS) model, a patient with suspected stroke is directly transported to a comprehensive stroke centre (CSC), which can administer both IVT and EVT. In the drip and ship (DS) model, the patient is initially transported to the nearest IVT capable hospital, a primary stroke centre (PSC). In case the patient appears eligible for EVT, interhospital transfer by emergency medical services (EMS) is arranged between the PSC and CSC, which was shown to substantially increase the onset to groin (OTG) time and thus delaying the start of EVT.4–7 The dominance of a certain organisational model and associated delays to EVT, mainly depend on region specific spread of hospitals, the geographic location of stroke onset and protocols used by EMS. In some regions there are concerns regarding timely access to EVT, as patients require relatively long travel times towards a CSC.8 A possible solution is to upgrade other hospitals presently acting as PSCs to meet the standard of a CSC, thereby aiming to reduce interhospital travel times or avoid inter-hospital transfer delay altogether.
The aim of this paper is to estimate the potential effects of adding one or more CSC(s) in the northern region of the Netherlands on the OTG puncture time and 90 days functional outcome in patients treated with EVT, using simulation modelling.
Methods
Participants and setting
For the baseline model prospectively collected data of 183 patients from the MR CLEAN Registry was used.9 All patients were treated with EVT between July 2014 and November 2017 and routed according to the DS model in the northern region of the Netherlands. In this region, the University Medical Centre Groningen (UMCG) is the only CSC. Its catchment area serves 1.7 million inhabitants (209 per square kilometre), including eight PSCs at distances between 6 and 84 kilometers (figure 1).
PSCs, CSC and NSCs in the north of the Netherlands. CSC, comprehensive stroke centre; NSC, non-stroke centre; PSC, primary stroke centre.
In order to acquire a complete overview of the acute stroke pathway, prehospital and interhospital transfer data was retrospectively collected at the regional EMS and subsequently linked. Inclusion criteria were prestroke modified Rankin Scale (mRS) score ≤2 and OTG time ≤390 min.
Baseline model
A Monte Carlo simulation model was developed based on DS time variables collected in the MR CLEAN Registry9 and served as baseline model. Input variables for the model included time of: symptom onset or last seen well, CT, start IVT, CT angiography (CTA), arrival at angiography suite and groin puncture. EMS variables included time of: 911 call, arrival at the stroke onset location, departure to PSC, PSC arrival, transfer notification (second 911 call), arrival at PSC, departure to CSC, and arrival at the CSC.
Prior to development of the simulation model, conceptual modelling was applied to capture the real-world acute stroke pathway for DS patients (figure 2). For each patient various logistical routes can be defined, starting from the moment of stroke onset up to groin puncture. For example, stroke onset may occur inside or outside the hospital, patients may be eligible for IVT or not, undergo CTA before IVT treatment or the other way around. Also, after interhospital transfer, patients may be assigned directly to the angiography suite, or first undergo additional diagnostics. The conceptual model represented all these variations and was validated using stroke experts participating in the nationwide COllaboration for New TReatments of Acute Stroke (CONTRAST) consortium, and based on findings from previous publications.10
Conceptual model of the acute stroke pathway (DS patients), baseline model and adapted time variables. Time variables and patient routing for the baseline model are represented by rectangles and diamonds. Time variables adapted when upgrading PSCs to CSCs are marked as coloured parts, referring to one or multiple time variables. Part 1: steps omitted, when patient is routed directly to the new CSC. Part 2: steps for which time variables are set equal to the distributions found for the existing regional CSC, when patient is routed directly to the new CSC. Part 3: time distribution for EMS interhospital transport is adjusted for patients still routed according to the DS model but transferred to the nearest new CSC. Part 4: time distribution adapted when prehospital routing is adapted to routing directly to the nearest CSC. CSC, comprehensive stroke centre; CT, computed tomography; CTA, CT angiography; EMS, emergency medical services; EVT, endovascular thrombectomy; IVT, intravenous thrombolysis; POC, point of care; PSC, primary stroke centre.
The Monte Carlo simulation model was coded using Plant Simulation software.11 Time intervals were quantified and presented as statistical distributions using ExpertFit,12 based on the original patient data (online supplemental table S1, online supplemental material).
Supplemental material
Adding CSC(s): Data and experiments
To evaluate the impact of adding one or more CSCs in our region, the baseline model was modified and patient routing adapted towards the nearest new CSC (online supplemental table S2, online supplemental material). Based on the original prehospital routing strategy, two categories of patients were distinguished: (1) patients routed directly towards the new CSC (modified MS model), and (2) patients routed according to the DS model, and subsequently routed towards the nearest new CSC (modified DS model). The time distributions underlying the baseline model were changed accordingly, see figure 2 (parts 1–3). For patients routed according to the modified MS model interhospital transfer time and time for additional diagnostics were set to zero (figure 2, part 1). In addition, for the new CSCs, we included time interval distributions from emergency department (ED) to angiography suite arrival and from angiography suite arrival to groin puncture (figure 2, part 2). For these distributions, we used data of MS patients (MR CLEAN Registry) treated with EVT in the UMCG.9
For patients still routed according to the DS model, but transferred to the nearest (new) CSC, the time distributions representing EMS interhospital transport were adapted (figure 2, part 3).
In addition, the strategy of prehospital routing to the nearest CSC was added. Prehospital routing was adapted when transportation times to the nearest CSC were shorter than 30–45 min,13 both for added and existing CSCs. Transport times from stroke onset location to the added or existent CSC were based on times collected by a web-based route planner (figure 2, part 4),14 and adjusted for presumed higher ambulance speeds by reducing transport times by 23% (calculated by comparing route planner car times and EMS time variables that were collected for our region). Furthermore, similar adaptions were made as for the modified MS model (figure 2, part 1 and 2).
Model scenarios
Three scenarios were tested to assess the impact of adding new CSC(s). The first two scenarios included selection of PSCs to be hypothetically upgraded to a CSC, based on their distance to the existing CSC, expected treatment volumes and available resources. Based on a distance of approximately 60 km from the UMCG and an expected treatment volume of 50 patients per year (based on collected patients with EVT of the last year in its catchment area),9 scenario 1 adds a CSC in the western section of the region (figure 3A). Scenario 2 adds an additional CSC in the south-east section at approximately 60 km from the UMCG (figure 3B). This centre would have a treatment volume of approximately 15 patients per year,9 but an increase in treatment volume is considered likely because this hospital is on a provincial border and thereby might attract patients from adjacent regions. In scenario 3, all PSCs are upgraded to CSCs.
CSCs added in the north of the Netherlands. (A) One CSC is added in the western part of northern Netherlands (scenario 1). (B) Two CSCs are added in the western part and southern part of northern Netherlands respectively (scenario 2). CSC, comprehensive stroke centre; NSC, non-stroke centre; PSC, primary stroke centre.
In addition, a subscenario was to scenario 1 (1A) and 2 (2A) for adapted prehospital routing, that is, directly routed to the nearest CSC, when transportation time was shorter than30-45 minutes.
On adding CSC(s), we assumed that workflow efficiencies (time from door/last examination at the ED to groin) within the new CSCs would be comparable to the existing CSC in the region (the UMCG). However, as not every CSC will perform equal, we performed a sensitivity analyses representing practice variation observed in the Netherlands. Based on MR CLEAN Registry data of hospitals providing EVT in the Netherlands, we studied the impact of a 25% increase and decrease in workflow efficiency.
Outcome measures
For each scenario, we calculated the clinical benefits in terms of reduction in OTG, and favourable functional outcome defined as mRS score of 0–2.
Statistical analysis
Missing values were excluded from analyses, as statistical imputation techniques were not necessary to obtain intact model distributions. Time distributions of the baseline model were numerically validated by comparing model output (mean, median, SD, minimum and maximum) with real-world data of patients.
Ordinal regression was used to estimate the likelihood of each of the seven outcomes according to the mRS score. Known prognostic variables were: OTG, age, National Institutes of Health Stroke Scale score and CTA collateral grading score in four categories. The predicted probability of favourable outcome (PPFO) was predicted using the formula obtained by ordinal regression, that is, still discerning all possible mRS and subsequently dichotomising.
Model outcomes for scenarios were compared with the baseline model. Testing for significance was deemed redundant since the aim was to assess the potential gains that may be expected based on a hypothetical cohort of 100.000 individuals.
Simulation model access
The simulation model will be available on reasonable request for other researchers.
Public and patient involvement
Patients and the public were involved in the conception of the topics to be addressed in the CONTRAST consortium. Study results will be disseminated through newsletters, poster presentations and publications in newspapers, lay journals and publication in peer-reviewed journals.
Results
Baseline characteristics
Out of the 179 patients, we included 165 patients. Fourteen patients were excluded because of an unknown or >2 prestroke mRS, and four because of an OTG >390 min. Baseline patient characteristics are presented in table 1.
Characteristics, diagnostics and time delays of the baseline model
Input data, adding CSC(s)
The median (IQR) time intervals from hospital arrival to angiography suite arrival and from angiography suite arrival to groin puncture were 58 (44–82) and 28 (25–35) minutes, respectively.
Simulation results, adding CSC(s)
Results for all simulated scenarios are presented in table 2. Adding a new CSC (scenario 1), would reduce OTG by 14 min and increase PPFO by 1.6%. For patients routed specifically according to the modified MS model, OTG would be reduced by 43 min and PPFO would increase by 5.2%. For all patients routed according to the DS model, OTG is reduced by 7 min and PPFO would increase by 0.8%. The strategy of direct prehospital routing to the nearest CSC would result in a 35 min reduction in OTG, and an increase by 4.1% in the PPFO.
Modelling results of adding CSC(s) and its effect on OTG and PPFO
Adding another CSC in the region (scenario 2) would lead to an overall reduction in OTG of 19 min and would improve PPFO by 2.2%. Modified MS patients would be treated 43 min faster and PPFO would increase by 5.2%. All patients routed according to the DS model would be treated 9 min faster and PPFO would increase by 1.2%. Using the adapted prehospital routing directly to the nearest CSC reduces OTG by 38 min and increases PPFO by 4.6%.
Upgrading all PSCs to CSCs effectively routing all patients according to the modified MS model (scenario 3), OTG would be reduced by 43 min and the PPFO increased by 5.5%. Increasing the number of CSCs showed a shift towards lower predicted mRS scores (online supplemental figure S1).
Sensitivity analysis
Results of the sensitivity analysis are presented in online supplemental table S3. When considering adding a new CSC in which workflow processes were 25% slower compared with the original CSC, OTG would be reduced by 3 min compared with the baseline model and PPFO would increase by 0.4%. In contrast, when implementing a CSC that is 25% faster, OTG would be reduced by 24 min and PPFO would increase by 3.0%. Adding two CSCs to the region, with 25% slower or faster workflow would reduce OTG by 5 and 32 min, respectively. PPFO would increase by 0.7% and 3.9%.
By upgrading all PSCs to CSCs, and assuming all the new CSCs would be 25% slower or faster compared with the baseline model, OTG would reduce by 19 and 67 min, respectively. PPFO would increase by 2.4% and 8.2%. In addition, if the original single CSC would achieve a 25% faster workflow, this would reduce OTG by 15 min and PPFO would increase by 1.8%.
Discussion
This modelling study demonstrated that, adding one or two CSC(s) in our region with comparable workflow efficiency as the current CSC, would reduce OTG between 15 and 20 min and improve PPFO at 90 days by 1%–2% (absolute benefit). A likely explanation for this modest effect may be the relatively short travel distances, the well-developed road network and the well-organised EMS within our region. Upgrading all PSCs to CSCs would have a much larger impact, reducing the OTG time by more than 40 min and improving PPFO by more than 5% (absolute benefit). This latter scenario is considered unfeasible, for reasons of low treatment volumes, which is related to the quality of EVT,15 availability of staff and equipment and higher costs.
When prehospital routing was adapted to direct transfer to the nearest CSC, the OTG may be reduced by 30–40 min and PPFO at 90 days improves by 4%–4.5%. Although this option appears to be beneficial, we did not estimate onset to IVT times for non-LVO patients. In addition, the expected increased workload for CSCs when routing all patients with stroke directly towards CSCs was not taken into account. A more comprehensive simulation model would be needed to study this option further. Patients routed directly to the new CSCs (MS model) were observed to be treated much faster compared with patients routed according to the DS model. This is in line with previous research indicating that interhospital transfer significantly contributes to longer OTG times.4–7 Analysing our data clarifies how the time interval from completing CTA to call for transfer explains more than one third of the difference among patient groups. Rapid LVO detection after CTA,16 early EMS notification or even EMS waiting for release at the PSC17 are therefore clear recommendations to further reduce the interhospital transfer time.
Importantly, sensitivity analyses revealed that workflow performance would have a major impact on performance of additional CSCs. Our sensitivity analysis revealed a difference of approximately 25 min in OTG times between ‘slow’ versus ‘fast’ acting CSC(s) compared with the baseline model. Given the practice variation between hospitals, the estimated impact might even be greater, and may be further enhanced by workflow improvements.18 In addition, our sensitivity analysis suggests that improving workflow might be even more efficient than adding a second or third CSC. However, prior to offering clear-cut recommendations we suggest performing cost and cost-effectiveness studies.
Our results may be generalisable to other underserved rural regions, which indeed reflects 31% of the European population, and a higher proportion of the older population.8 Using simulation modelling to obtain early estimates on the potential impact of implementation policies is in line with stroke guidelines indicating that organisation of stroke care should be analysed and/or adjusted per region.13 19 Moreover, simulation could be applied to other regions using the same approach by repopulating the model with region-specific distributions and assumptions. For example, similar analyses might be performed using data from urban regions possibly overserved currently by an abundance of CSCs. Thus, the effect of reducing the number of CSCs may be assessed.
Limitations
Our study has limitations. Model input is only described for patients eligible for EVT, although in reality organising acute stroke care requires a comprehensive approach beyond this specific group. that is, by including a wider group of patients with stroke including non-LVO patients and haemorrhages. Also, our simulation study has a main focus on logistic gains implied by adding CSCs, leaving quality issues as reflected in outcome models out of scope.
Conclusions
Our study suggests that the impact of adding CSCs on treatment times and clinical outcome in the north of the Netherlands will be modest. Workflow efficiency and prehospital routing (using a prehospital triage scale) seems important when considering to add CSCs to existing infrastructures. Both the performance of the existing CSC(s) as well as additional CSC(s) are important in this consideration.
Data availability statement
The participants of this study did not give written consent for their coded data to be shared publicly, so due to the sensitive nature of the research supporting data is not available. Nevertheless, used distributions of the collected data are available in the supplemental material.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was approved by central medical ethics committee and research board of Erasmus University Medical Centre (MEC-2014-235) Participants gave informed consent to participate in the study before taking part.
Acknowledgments
The CONTRAST consortium acknowledges the support from the Netherlands Cardiovascular Research Initiative, an initiative of the Dutch Heart Foundation (CVON2015-01: CONTRAST), and from the Brain Foundation Netherlands (HA2015.01.06). The collaboration project is additionally financed by the Ministry of Economic Affairs by means of the PPP Allowance made available by the Top Sector Life Sciences & Health to stimulate public-private partnerships (LSHM17016). This work was funded in part through unrestricted funding by Stryker, Medtronic and Cerenovus. The funding sources were not involved in study design, monitoring, data collection, statistical analyses, interpretation of results, or manuscript writing. Furthermore, we acknowledge the UMCG Emergency Medical Services, Kijlstra Emergency Medical Services and Emergency Medical Services Groningen.
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Collaborators CONTRAST investigators Diederik Dippel, Charles Majoie, Heleen van Beusekom, Hugo ten Cate, Ruben Dammers, Rick Dijkhuizen, Jaap Kappelle, Karin Klijn, Peter Koudstaal, Hester Lingsma, Aad van der Lugt, Moniek de Maat, Paul Nederkoorn, Robert van Oostenbrugge, Yvo Roos, Denis Vivian, Wim van Zwam.
Contributors WJM, D-JvdZ, EB, MU and MMHL designed the study with EB and MU as principal investigators. EB and MU applied for, received and organised study funding. WJM, D-JvdZ, MU and MMHL analysed the data. WJM drafted the manuscript, and D-JvdZ, EB, MU and MMHL revised the manuscript for intellectual content and approved the final version of the manuscript for publication. MMHL is responsible for the overall content as guarantor.
Funding The CONTRAST consortium is supported by Netherlands Cardiovascular Research Initiative, an initiative of the Dutch Heart Foundation (CVON2015-01: CONTRAST), by the Brain Foundation Netherlands and powered by Health~Holland, Top Sector Life Sciences and receives unrestricted funding from Medtronic and Cerenovus. The collaboration project is additionally financed by the Ministry of Economic Affairs by means of the PPP Allowance made available by the Top Sector Life Sciences & Health to stimulate public-private partnerships. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Map disclaimer The inclusion of any map (including the depiction of any boundaries therein), or of any geographic or locational reference, does not imply the expression of any opinion whatsoever on the part of BMJ concerning the legal status of any country, territory, jurisdiction or area or of its authorities. Any such expression remains solely that of the relevant source and is not endorsed by BMJ. Maps are provided without any warranty of any kind, either express or implied.
Competing interests None declared.
Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting or dissemination plans of this research. Refer to the Methods section for further details.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.