Mobile interventionists in acute care networks – a
simulation study to improve functional patient
outcomes and onset until treatment times for stroke
patients.
Master Thesis
MSc Technology & Operations Management Faculty of Economics and Business
University of Groningen 25th of January 2021
Marc Bouma s2670445
1
ABSTRACT
Background and purpose
The mobile interventionist (MI) organizational model is explored to evaluate if this is a viable solution to improve the acute stroke pathway. The model proposes that MI travel to the primary stroke centres when intra-arterial thrombectomy is necessary. The time until treatment and functional patient outcomes are used to evaluate the feasibility of the organizational model. This study will look at the moment of notifying the MI, modes of transportation and the moment when the MI should be present at the primary stroke centre, to get insights in how the organizational model should be set up.
Methods
A simulation study is used that represents the acute care network of the North of the Netherlands. The simulation study is extended to fit the characteristics of the MI model. A variety of experiments have been included to test the effects on patient outcomes and time until treatment.
Results
The MI organizational model improves the time until treatment and patient outcomes compared to the drip-and-ship model (201.47 vs 241.73 minutes & 0.556 vs 0.508 likelihood of mRS 0-2). The MI should be notified as soon as possible (emergency medical services on scene), decreasing patient waiting times. Furthermore, the patient should be prepared for intra-arterial thrombectomy even when the MI is not present yet, this significantly decreases the time until treatment. The fastest mode of transportation for the MI should be chosen considering that the MI should not wait for the patient to be set. Furthermore, a combination of fast mode of transportation, early notification moments for the MI and that the MI only requires to be present right before the IAT, significantly improves patient outcomes and unset until treatment times.
Conclusions
The mobile interventionist organizational model is a viable substitution for the current dominant organizational models with respect to time until treatment and functional outcomes.
Keywords
2
ABBREVIATIONS
CSC Comprehensive stroke centre
CS Confidence interval
CT Computed tomography
CTA Computed tomography angiogram
DIDO Door-in door-out
DS Drip-and-ship
EVT Endovascular thrombectomy
EMS Emergency medical services
IAT Intra-arterial thrombectomy
IVT Intravenous thrombolysis
LVO Large vessel occlusion
MI Mobile interventionist
MS Mothership
mRS Modified Rankin scale
PSC Primary stroke centre
3
PREFACE
I would like to sincerely thank dr. ir. D.J. van der Zee for his patient advice and guidance throughout this thesis process. His expertise and dedication to a positive outcome helped me to keep going and to improve myself. Additionally, I would like to thank dr. M.M.H. Lahr for sharing his knowledge on the stroke care systems in the Netherlands which helped me enormously in understanding the challenges of organizing stroke care systems. Big appreciation goes out to W.J. Maas, who shared her simulation of the current organizational model and who gave various insights in how to extend this model to fit the research. Furthermore, I would like to give my thanks to dr. M. Uyttenboogaart for validating the simulation model and for sharing his expertise to improve this research.
4
1. INTRODUCTION
Recently there has been a shift to more specialization in healthcare (Kilgore & Langford, 2009). Specialization in hospitals may result in better cost control and improved quality (Tiwari & Heese, 2009). The shift to more specialized care is therefore a promising development. However, specialization in acute care can be problematic because acute care requires a timely response and specialized hospitals might be further away, extending the time until treatment. Organizing acute care to decrease the time until treatment, improving patient outcomes is a challenge.
This is not different for acute stroke, which is the second leading cause of death worldwide (Campbell & Khatri, 2020), furthermore treatment outcomes are extremely time sensitive. There are two types of acute stroke, ischemic stroke, which is 80% of the stroke cases and 20% are haemorrhagic stroke (Donnan et al., 2008). Ischemic stroke is where the blood flow to the brain is blocked due to a blood clot which can lead to tissue damage (Geoffrey A Donnan, Marc Fisher, Malcolm Macleod, 2009). There are two main ways to treat ischemic stroke, either by Intravenous thrombolysis (IVT) which needs to be administered within a time window of 4.5 hours or intra-arterial thrombectomy (IAT), with a time window of 6 hours (Campbell & Khatri, 2020). The IVT treatment can be given in both a primary stroke centre (PSC) or a comprehensive stroke centre (CSC) and IAT treatment can only be administered in the CSC, because it is an advanced treatment requiring expertise and involving relatively high-ranking costs.
The current stroke system set-up is a combination of two coexisting organizational models; the drip-and-ship model (DS model) and the mothership model (MS model) (Ciccone et al., 2019), where the transfer of the patient depends on the lowest estimated transportation time to the nearest hospital. With the DS model the acute care patient is transferred to the PSC to receive IVT treatment. When this patient is eligible for IAT treatment, he is transported to the nearest CSC. For the MS model the patient is transported directly to the nearest CSC to receive IVT treatment and when necessary IAT treatment. Intrahospital transportation between the PSC and CSC can lead to undesirable time delays and complications to the patient. The time from stroke onset until the IAT treatment is the time until treatment. Other organizational models might improve the time until treatment and therefore patient outcomes.
The organizational model that will be explored is called the mobile interventionist (MI) model (Ciccone et al., 2019). The MI model is closely related to the DS model. When a patient gets acute stroke, this patient will be transported to the nearest PSC to receive IVT treatment just like in the DS model. If an IAT treatment is necessary a specialized doctor(s), the MI(s), will be driven or flown from the CSC out to the PSC to conduct the IAT treatment, instead of transporting the patient to the CSC which would normally happen at the DS model. This organizational model potentially decreases the time until treatment.
5
stroke. The articles of Brekenveld et al (2018), Seker et al. (2018), Wei et al. ( 2017) and Uchiyama et al. (2016) all conclude that the MI model has great potential and is not inferior and maybe even superior to the current dominant organizational model. However, more extensive studies are necessary to confirm this (Seker et al., 2018). The articles do show a significant reduction in time until treatment and the article of Wei et al. (2017) gives insight that the risks of patient transport are lowered because no transport between the PSC and CSC is necessary. Nevertheless, they only show the potential of the model and do not give insights into organizational factors on the set-up of the MI model. These studies are clinical studies and do not considering the mode of transportation, when the MI should be notified, and at which stage the MI should be present.
There is an opportunity to improve the acute care pathway by using the MI model. When implementing this model into the acute care pathway there are still a lot of unknowns that have not been explored. Time until treatment is dependent on several factors; the base location of the MIs and where they are located, the number of specialists available, the transportation vehicle, when in the process the specialist is notified and the location of the CSC. These factors and variations in these factors due to regional characteristics can either positively or negatively influence the time until treatment and the functional patient outcomes. Examining the feasibility of the MI model taking in mind these factors is an important challenge.
This results in the following research objective:
Explore the viability of the mobile interventionist model, considering patient outcomes and time until treatment.
The viability of the model will be explored, by evaluating the time until treatment and functional patient outcomes, generated by experimental factors, compared to the current organizational model. Contributions to science are related to the implementation of the model considering amount and placement of CSC, optimal amount of specialist needed and what transportation vehicles are necessary. This is dependent on the characteristics of the system that the model is applied to. The research can help improve the operational side of acute healthcare in the future when the structure of hospitals is shifting more to specialized care.
6
modelling, experiments and verification, validation and evaluation. A Monte Carlo simulation will be made using Siemens Plant Simulation software (2012).
7
2. THEORETICAL BACKGROUND
2.1. ACUTE CARE NETWORKS
The importance of regionalization of healthcare by directing acute patients to designated hospital with greater capabilities is growing (Horeczko, Marcin, Kahn et al., 2014, Kilgore & Langford, 2009). This can be explained by overcrowding and a shortage of specialists (Institute of Medicine, 2006). Specialization in hospitals may result in better cost control and improved quality (Tiwari & Heese, 2009). This requires a good cooperation between hospitals, EMS and other agencies and poses an organizational challenge, to take the patient to an optimal facility within their region, based on their condition and involved distances to the hospital (Institute of Medicine, 2006). Organizational models for acute care are consequently a necessity.
2.2. TYPES OF STROKE
Sacco et al. defines stroke as a neurological deficit attributed to an acute focal injury of the central nervous system by a vascular cause (2013). More generally speaking a stroke is a disruption of the supply of blood to the brain. There are two major types of stroke, namely the haemorrhagic stroke occurring 20% of the occasions and ischaemic stroke, which occurs 80% of the time (Donnan et al., 2008). The distinction between these subtypes is important because of the difference in management and treatment between them (Donnan et al., 2008). Haemorrhagic stroke is the rupture of a blood vessel, while ischaemic stroke reduces the blood flow, which is generally the result of an atrial occlusion (Campbell & Khatri, 2020). Both types of stroke results in a lack of oxygen in the brain, resulting in braincells that die, with the possibility of permanently damaging the brain (Churilov & Donnan, 2012). This research focusses on the ischaemic stroke treatment and organization. The time until treatment to reduce damage in the brain is important for acute stroke. A patient who suffers from acute ischaemic stroke loses around 1.9 million neurons each minute when stroke is untreated (Saver, 2006). This makes treating the patient as soon as possible an important challenge.
2.3. TYPES OF TREATMENT; INTRAVENOUS THROMBOLYSIS
There are two major treatment types for ischaemic stroke, namely intravenous thrombolysis (IVT) and intra-arterial thrombectomy (IAT). IVT is a treatment that aims to dissolve the thrombus causing the stroke. Treatment of IVT is only effective 4.5 hours after the stroke occurs (Campbell & Khatri, 2020), but this treatment is most effective when administered as soon as possible. The IVT treatment can be administered in most PSC’s, making this a general first treatment for patients.
2.4. TYPES OF TREATMENT; INTRA-ARTERIAL THROMBECTOMY
8
treatment. The importance of IAT has grown since around 2015 when the results of the treatment where proven to be significantly positive. Because of the selective locations of CSC’s available, this results in organizational challenges for patient routing (Appireddy et al., 2016). The IAT treats the patient by using a retrievable stent to achieve reperfusion (Rabinstein & MD, 2017). The IAT treatment should be administered within 6 hours of symptom onset (Appireddy et al., 2016).
2.5. ORGANIZATIONAL MODELS
The importance of patient routing for acute care resulted in a variety of organizational models; MS model, DS model, MI model and mobile stroke unit (Ciccone et al., 2019). The status quo for the patient routing, the stroke pathway, is currently a hybrid version between the MS and the DS model, where the patient is transferred to the nearest PSC or CSC (Holodinsky et al., 2018; Maas et al., 2020).
In the MS model the patients are transferred directly to the CSC that is closest to the current location of the patient. Patients will receive IVT treatment and if necessary IAT. The DS model transfers the patient to the nearest PSC, where he/she receives IVT treatment. When the patient is eligible for IAT treatment he/she will be transported to the nearest CSC.
The mobile stroke unit model consists of an ambulance including diagnostic devices and a telemedicine connection to the CSC. When ischemic stroke is detected the patient receives IVT treatment in the ambulance. The patient is then transported to the CSC to receive IAT treatment when eligible.
2.6. MOBILE INTERVENTIONISTS MODEL
This research focusses on the third model; the mobile interventionist model. The MI model explores the possibility of driving or flying the doctor to the PSC where IAT treatment is given, instead of transporting the patient to the CSC. This does however set requirements on the local infrastructure for PSCs. This relates to the IAT treatment given at the PSC, which should have the means available to provide the treatment, like an angio suite, equipment and support staff.
Figure 1 Graphical explanation of the organizational stroke systems.
A is a representation of the drip-and-ship model, B the mothership model and C the mobile interventionist model.
The zone of IV-tPA provision is the zone in where IVT can be given the zone of endovascular therapy is the zone
where the IAT treatment can be given.
9
2.6.1. Mobile interventionist clinical studies
Limited research on the effectiveness and feasibility of the MI model has been conducted. In the article of Brekenfeld et al. (2018) a study is conducted on the ‘drip-and-drive’ model, which is identical to the MI model, in comparison with the more commonly used DS model. Patients arrived at the nearest PSC, where they got a CTA scan to determine if the patient was eligible for the IAT treatment. If eligible, the specialist was called and driven (by taxi) to the PSC. While the specialist was transported, the patient was already prepared for IAT treatment. This includes transfer to a local angiography suite, anaesthesia and if possible, placing a catheter sheath and guiding catheter. This whole procedure takes place before the specialist arrives. This is a more time efficient approach, because both the preparation and the transportation of the specialist occurs parallel to each other, limiting waiting times. The results of the study show significant reduction of process times in comparison with the DS model.
The article of Seker et al. (2018) likewise compares the MI model with the DS model. The specialist needed to drive 100 km to the PSC, which is on average a 1-hour drive. The results show that the time until imaging for both models where approximately the same, but the time until groin puncture, IAT, was significantly shorter for the MI model. Seker et al. implies that it is faster to transport the doctor than to transfer the patient by ambulance or helicopter to the CSC (2018). However, it should be noted that the location of the PSC and CSC can change the outcomes of this study and that not every PSC has the means available to do an IAT treatment. The article of Wei et al. (2017) calls the MI model ‘trip-and-treat’. It looked at the initial door-to-puncture time compared to the DS model for 4 hospitals in Manhattan. This resulted in the MI model being on average 79 minutes faster. It is also mentioned that the risks of patient transport are eliminated when the doctor instead of the patient is transported. This is a legitimate point, because transport of patients carries inherent risks (Warren et al., 2004).
2.6.2. Organisation of the MI model
The articles (Brekenfeld et al., 2018; Seker et al., 2018; Uchiyama et al., 2016; Wei et al., 2017) give insights into the potential that the MI model has. It shows that this model is as good or better than the current dominant organizational models; DS and MS model. These articles however only show the potential of the model and do not give insights into organizational factors on the set-up of the MI model. The organizational factors of a general transport system can be based on control, process and modality. The base location of the MI is a process decision, while the moment of notification is a form of control. The actual transport of the patient and MI is covered by modality. The outcomes of each article are represented in table 1.
Base location MI
10
placement will affect the travel distance to the PSC. The MI base placement will be explored in this report.
Number of mobile interventionists needed.
The article of Brekenfeld et al. (2018) indicates that the travel time for the specialist to get to the PSC does not need to be shorter than the prepping time for IAT treatment. The number of MIs that need to be present at the base to deliver the IAT treatment is not mentioned in research, while this does affect the organization of an acute care organizational model. Therefore, more research on this subject should be conducted.
Mode of transportation
Likewise, the articles only used driving to transport the specialist to the PSC while the effect of flying has not been explored. An article by Diaz et al. (2005) investigated when to choose for ambulance or helicopter transport in relation with travel times. It is concluded that ground transport, with an ambulance, is more effective with distances closer than approximately 16 km. Knowing when to fly or when to drive the doctor is an important consideration when implementing the MI model and therefore taken into account in this report.
Moment of notification
Brekenveld et al. (2018), stated that the specialists were called when it was known that the patient was eligible for IAT treatment. The effects of when to notify the doctor might influence the time until treatment. Notifying the doctor before it is certain that the patient is eligible for IAT can shorten the time until treatment but can also lead to unnecessary trips for the doctor. The effects of the moment of notification are explored in this thesis. Also, no insight has been given into the operational costs of the model. It can be concluded that there is still an abundance of unknows regarding the MI model.
Article Study set-up Regional setting Proposed interventions Effects on lead time Effects on patient outcomes Brekenfeld et al. (2018)
Clinical trial 2 PSCs and 1 CSC Catchment area 180.000 to 200.000 citizens DS model vs MI model MI transported by taxi. Time until treatment: DS 433 min MI 244 min Not provided
Seker et al. (2018) Clinical trial Distance PSC and CSC is about 100 km DS model vs MI model Time until treatment: DS 189 min MI 302 min Patient outcomes where similar for both DS and MI model
Wei et al. (2017) Clinical trial 1 CSC and 3 PSCs for MI, 8 PSCs for DS in Manhattan. DS model vs MI model Time until treatment: DS 267 min MI 199 min Patient outcomes where slightly better for MI model (P=0.0704) Uchiyama et al. (2016) Clinical trial 14 PSCs in Ishikawa. Distances between 20 and 80 km
DS model Time until
treatment: DS 297 min
mRS 0-2 at 90 days is 50%
Table 1 Outcomes and characteristics of the articles where the MI organizational model was
11
3. CASE DESCRIPTION
3.1. CHOICE OF REGION
The research will be based on a representative acute stroke care casus of the North of the Netherlands. The setting for which the simulation study will be designed concerns tree provinces in the North of the Netherlands; Groningen, Friesland and Drenthe. The population density lays around 209 citizens per squared kilometres with a total of 1.7 million inhabitants. Furthermore, there is an adequate road-network in the North of the Netherlands and every PSC and the CSC have a landing area for helicopters.
3.2. PATIENTS AND CHARACTERISTICS
Included in this study are patients who underwent acute ischaemic stroke and got an IAT treatment, based in the setting of the North of the Netherlands. The patients followed the MS model when the nearest hospital was a CSC. When a PSC was is closest proximity the patient was routed using the DS model to the PSC and if necessary, transported to the CSC for IVT treatment. The patients routed to the CSC are not included in this research.
3.3. REGIONAL HEALTH INFRASTRUCTURE
The research region contains one CSC, located in the city of Groningen and 9 community hospitals distributed over the three provinces. The University Medical Centre Groningen (UMCG) acts as the CSC, where currently the IAT treatment is given, furthermore it has the means to provide IVT treatment. The CSC has the ability to give stroke care whenever needed, including CTA scans and an angiosuite for IVT treatment. There are 9 community hospitals, that act as PSCs. The distances between the hospitals differ between 5.7 and 83.7 kilometres. The hospitals are located at densely populated areas, mostly bigger cities, resulting in an uneven distributed network of hospitals. Patients must travel dissimilar distances from the PSC to the CSC if IAT treatment is deemed necessary. Higher travel times can increase risks of endangerment to the patient.
3.4. CURRENT PATIENT ROUTING
12
patients outside the 4.5-hour window (Campbell & Khatri, 2020) or patients for whom the IVT treatment works counteractive. Currently a shift is taking place where more hospitals prefer the second route. The CTA scan shows if patients are eligible for IAT treatment.
If the patients are eligible for IAT treatment the patient will either be transferred to a CSC (DS model) or when already present (MS model), transferred to an angiography suite, where the IAT treatment will take place. For the DS model intra-hospital activities occur; the patient needs to be transferred by the EMS, who need to be notified, respond, get on the scene (PSC) and transport the patient to the CSC. When these patients arrive at the CSC, they will either be transported to the angiography suite immediately or additional examination (CT/CTA) will take place to revaluate the decision of EVT treatment. At the angiography suite the IAT treatment will take place.
3.5. IMPLEMENTING THE MOBILE INTERVENTIONIST MODEL
The MI model has close resemblance to the DS model. The patient is transferred to the PSC where diagnostics and IVT treatment takes place. Instead of transporting the patient to the CSC if IAT treatment is necessary, the patient stays at the PSC. MIs are located at the CSC and are notified by the PSC about the stroke patient. The MI will investigate the patient’s data to judge if a LVO has occurred and that the IAT treatment is necessary. After this the MI goes to the transportation vehicle and travels to the PSC. The MI will then either redo the diagnostic CTA scan or move to the angiosuite to do the IAT treatment directly. All this time the patient has stayed in the PSC and only have been moved to the diagnostic or treatment location inside the PSC. The patient could also still receive IVT treatment while the MI was moving to the PSC, reducing time until treatment. A flow diagram is included in figure 2, showing all steps of the MI organizational model.
There are essential choices that can influence the total time until treatment for the patient. The moment of notification of the MI influences the time it takes before the MI present at the PSC. Depending on the moment of notification the patient can receive IVT treatment or diagnostics while the MI is transported to the PSC, reducing the waiting time, because of simultaneous processes.
Furthermore, the mode of transportation has an influence on the transportation times of the MI. A choice can be made to transport the MI using an ambulance, car or helicopter. Each mode of transportation has their advantages and disadvantages. A suitable choice is necessary to reduce the waiting time for the patient, but it is not necessary for the MI to be present too early, as the MI then must wait for the patient to be ready.
Additionally, it is important to determine at which stage the MI should be present. Preparations of the angiosuite can be made before the MI is available, so that the MI can start with the IVT treatment right on arrival. This could reduce waiting times but is dependent on the mutual agreements between the PSCs and CSCs.
13
any location depending on transportation times. This can be beneficial when other hospitals might lay more central in the region.
14
4. DESIGN OF SIMULATION STUDY
4.1. SIMULATION STUDY
For this research, a simulation study is chosen as an appropriate methodology. A Monte Carlo simulation model will be built for the MI organizational model. The importance of simulation studies in operations research for healthcare has been addressed in literature. Simulation can be an important tool when systems have variability, interconnectedness and complexity. The advantages of simulation studies are that they are cost effective compared to experimentations in real life systems, also it is less time consuming and there is high control of the experimental conditions (Robinson, 2014). These advantages are applicable on the MI model, which has a high degree of variability and complexity. Conducting real life experiments are therefore not feasible. Simulation will enable realistic modelling of stroke care to enable capturing the complexity of regional organizational stroke models in detail (Maas et al., 2020). Research has also shown that simulation studies for accurate representation of the acute stroke pathway can be developed in different settings (Lahr et al., 2020).
4.2. RESEARCH QUESTIONS
The research is investigating the possible application and relevance of the MI model. Relevance is measured by determining the stroke onset to treatment times and the patient outcomes. If the time until treatment is low and the patient outcomes are the same or better, in comparison to the DS and MS model, this organizational model might be a viable solution in acute stroke care. The main research question is therefore:
Is the mobile interventionist model a viable way of improving the acute stroke pathway, considering functional patient outcomes and time until treatment?
It is expected that the MI organizational model is an equal or even better substitution to the current dominant organizational models and that the possible potential should be further explored to be implemented in the near future.
To get further insights into the different characteristics of the MI model a variety of sub questions have been determined. The moment of notification for the MI can has an influence on the time that the patient must wait on the MI to receive IVT treatment. It is expected that the time until treatment is shortest when the MI is notified as early as possible. However, this might not be the optimal moment of notification, because of the possibility that the MI must wait until the patient is done with the initial IVT treatment or diagnostics and because the MI might be transported to the PSC while no LVO has occurred. To determine the actual influence and best moment of notification the following sub question needs to be answered;
1. What is the influence of the moment of notification on the time until treatment and what is the optimal moment of notification?
15
the MI more travel time. To measure the effect of the moment that the MI should be present the following sub question should be answered;
2. How do the time until treatment and patient outcomes differ when the MI is present just before the IAT treatment or when the MI is present at the preparation and possible extra diagnostic stages?
The MI can either be transported by an ambulance, car or helicopter. All these transportation vehicles influence the transport times of the MI. It is expected that the helicopter is fastest, but that other modes of transportation are also sufficient. Furthermore, it could also be a possibility to combine different transportation modes, depending on the distance between the CSC and PSC. This results in the following sub question;
3. What is the influence of either an ambulance, car or helicopter as a transportation vehicle for the MI on the patient outcomes and time until treatment?
Additionally, the base location of the MI can be changed to another hospital. It is explored if the hospital in Heerenveen which is located more central in the region might improve the time to treatment and functional outcomes of the patients. It is assumed to be a better location, because of its central location and close access to multiple freeways.
4. How does changing the base location of the MI to Heerenveen influence the time until treatment and functional outcomes of the patients?
Lastly the influence of combining experimental factors together to see the influence on the time until treatment and patient outcomes is explored. It is expected that more then one improvement gives a better outcome, however it might also be possible that combining improvements has a limited effect. Therefore, the last sub question is:
5. What effect does a combination of experimental factors has on the time until treatment and functional outcomes of the patients?
4.3. SIMULATION MODEL
16
simulation is validated using expert opinion, solution validation and step by step validation (Robinson 2014). Technical details and coded model descriptions, involving the simulation model can be found in appendix A. The MI model follows the distributions shown in appendix A2. A flow diagram of the MI organization model is shown in figure 3.
17
4.4. EXPERIMENTAL FACTORS
To determine the best moment to notify the MI, three notify moments have been included in the MI organizational model. The notification moments happen when the ambulance arrives on the scene, when the patient arrives at the hospital or when the patient has gone through the CTA scan. At the moment of notification, the MI enters the system.
There are two moments in the system when the MI should be present. This is either after the patient leaves the CTA scan or when the actual IAT treatment should commence. If the MI is not present the patient has to wait until the MI arrives to move further in the system. The waiting time for the MI to arrive can differ depending on the time characteristics of the patient. There are three modes of transportation for the MI; the ambulance, car or helicopter. Furthermore, a combination of both ambulance and helicopter is explored. When the distance between the CSC and PSC is higher than 40 km a decision is made to use the helicopter instead of the ambulance, based on the judgement of an expert. The distribution for the ambulance transportation times is already known because these are the same as in the DS model where the patient is transported to the CSC. The distribution of the car has been determined using Google Maps data. The helicopter distributions have been determined using a constant start-up time and the linear distances multiplied by the average helicopter speed of 3.5 km/min (Regenhardt et al. 2018). All distributions can be found in appendix A2.
Based on the judgement of an export, Heerenveen has been picked as a suitable base location for the Mis. Looking at Heerenveen to be a new base location for the MI in the North of the Netherlands the transportation distributions had to be adjusted to fit the travel times. For both the car and the helicopter transportation distributions have been determined similar as done for the UMCG. The ambulance distributions have been determined by using the original distribution for the UMCG simultaneous with a scale factor of 0.6823. The scale factor is derived by dividing the mean transport time for the ambulance by the mean transport time by car, both in the situation where the UMCG is the base location for the MI.
Lastly the influence of a combination of the experimental factors is explored to see if this results in better patient outcomes and a lower time until treatment. For all experiments where a combination is explored, it is assumed that the MI should be present right before the IAT treatment.
18
Experiment Experimental factor
1a 1b 1c
Moment of notification of the MI. When ambulance is on scene When patient enters the hospital After CTA scan
2a 2b
Moment that the MI should be present After the CTA scan Just before IVT treatment 3a
3b 3c 3d
Mode of transportation for the MI Ambulance Car Helicopter
Combination (Ambulance + Helicopter) 4a1 4a2 4b1 4b2 4c1 4c2 4d1 4d2
The base location for the MI UMCG (current base location) + ambulance Heerenveen + ambulance
UMCG (current base location) + car Heerenveen + car
UMCG (current base location) + helicopter Heerenveen + helicopter
UMCG (current base location) + combination Heerenveen + combination 5a1 5a2 5a4 5b1 5b2 5b3 5c1 5c2 5c3 5d1 5d2 5d3
Combination of experiments Ambulance + On Scene Ambulance + Arrival at PSC Ambulance + After CTA Car + On Scene
Car + Arrival at PSC Car + After CTA Helicopter + On Scene Helicopter + Arrival at PSC Helicopter + After CTA Combination + On Scene Combination + Arrival at PSC Combination + After CTA
Table 2 Overview of all experiments conducted in the simulation model.
4.5. MODEL OUTPUT
The primary model outputs are the onset to groin times, which is the total delay from the stroke onset until treatment. Furthermore, the likelihood of a good functional outcome at 90 days is calculated in percentages. A good functional outcome is defined with a modified ranking (mRS) score between zero and two.
Additional outcomes and results that give better insights into the performance of the system have been included as secondary outputs.
4.6. EXPERIMENTAL DESIGN
19
4.7. MODEL CODING
20
5. RESULTS
5.1. BASE SETTING
The results of the experiments will be compared with the DS model of the North of the Netherlands, where the patients are transported to the nearest PSC when after receiving IVT treatment they will get transported to the CSC by ambulance to receive IAT treatment there. The delay from stroke onset until the IAT treatment is 241.73 minutes and the likelihood of a functional outcome after 90 days are 0.508 on a scale of 0 to 2. Which means that patients have a probability of 50.8% that the treatment will help them recover to a good functional health.
Time until treatment (min) Functional outcome (0-2) Base setting 241,73
[241.37; 242.10]
0,508 [0.507, 0.509] Table 3 The outcome measures of the DS model (base setting)
5.2. NOTIFICATION MOMENT
Three notification moments have been identified to notify the MI, namely; 1) On scene (1a), 2) Arrival at PSC (1b) 3) After CTA scan (1c). The results indicate that all times until treatment times are lower than the base case. The DIDO times differs for the base case significantly because the patient leaves the PSC to go to the CSC after the CTA scan, meaning no insights can be generated by comparting the DIDO times of the base case with the experiments. The outcomes of the study show that the total waiting time for the MI to arrive is the lowest when the MI is notified as early as possible (1a). Also, the functional outcome improves when the notify moment is earlier. This is the result of the MI being transported while the patient is still going through the steps in the system. Earlier notification means more time that the patient and MI move parallel to each other, reducing delays in the total waiting time until MI arrives. The total waiting time until the MI arrives can be zero, depending on how fast the patient moves through the system, meaning that the MI must wait for the patient. The time that the MI must wait does not affect the time until treatment or the patient outcomes. These results assume that the MI is transported by ambulance transport.
Notify moment
Time until treatment (min)
[min CI ; max CI]
DIDO Times (min)
[min CI ; max CI]
Total waiting time until MI arrives (min)
[min CI ; max CI]
21
5.3. MI PRESENT LOCATION
The MI can be present only just before the IAT treatment commences (2b) or after the CTA and IVT treatment, for preparations or additional diagnostics (2b). Results indicate that the functional outcome improves greatly when the MI only needs to be present when the treatment should be performed (2b). Otherwise, the patient should wait on average half an hour longer until IAT treatment is received. These results assume that the MI is transported by ambulance transport.
Phase when MI must be present
Time until treatment (min) [min CI ; max CI]
DIDO Times (min) [min CI ; max CI]
Total waiting time until MI arrives (MI) [min CI ; max CI]
% of mRS score 0-2
[min CI ; max CI] Base case 241,73 [241.37; 242.10] 93,92 [93.72; 94.12] - 0,508 [0.507, 0.509] After CTA and IVT treatment (2a) 233,04 [ 232,68 ; 233,41] 170,92 [ 170,66 ; 171,18] 68,80 [ 68,63 ; 68,98] 0,518 [ 0,517 ; 0,520] Just before IAT begins (2b) 201,47 [ 201,13 ; 201,80] 139,31 [ 139,09 ; 139,52] 37,23 [ 37,00 ; 37,45] 0,556 [ 0,554 ; 0,557] Table 5 Results of experiment 2a and 2b, where the moment that the MI should be present at
the PSC is an experimental factor.
5.4. VEHICLE TYPE
Outcomes of the study show that both the helicopter and a combination of the helicopter and ambulance transport (3d, 3e) reduces the MI vehicle transport time drastically, affecting both onset time until treatment and patient outcomes. It should be noted that the distinction between flying or driving was made at a maximum mark of 40 km for driving after which the MI is flown to the PSC. This mark of 40 km can be altered if deemed necessary, decreasing transport times with less than 40 km and an increase for more than 40 km. The vehicle transport time affects the total MI time from call until present, consisting of the call for MI, MI response, MI transport and MI on scene. The distributions for call for MI, MI response, and MI on scene are identical for each experiment making the MI transport the only variable, altering the total MI time. It should be noted that the influence of the transport times for the MI decreases with earlier notification or the MI needing to present just before the IAT treatment commences, making the MI transportation by car (3c) a suitable mode of transportation as well.
22
Vehicle type Time until treatment (min) [min CI ; max CI] DIDO Times (min) [min CI ; max CI] vehicle transport MI time (min) [min CI ; max CI] Total MI time from call until present (min) [min CI ; max CI] Total waiting time until MI arrives (min) [min CI ; max CI] % of mRS score 0-2 [min CI ; max CI] Base case (3a) 241,73 [241.37; 242.10] 93,92 [93.72; 94.12] - - - 0,508 [0.507, 0.509] Ambulance (3b) 233,04 [ 232,68 ; 233,41] 170,92 [ 170,66 ; 171,18] 24,32 [ 24,26 ; 24,39] 73,94 [ 73,70 ; 74,18] 68,80 [ 68,63 ; 68,98] 0,518 [ 0,517 ; 0,520] Car (3c) 244,19 [ 243,82 ; 244,56] 182,01 [ 181,75 ; 182,27] 35,64 [ 35,55 ; 35,72] 85,26 [ 85,00 ; 85,51] 80,09 [ 79,91 ; 80,27] 0,505 [ 0,504 ; 0,507] Helicopter (3d) 222,32 [ 221,96 ; 222,68] 160,12 [ 159,87 ; 160,38] 13,67 [ 13,63 ; 13,70] 63,29 [ 63,08 ; 63,49] 58,21 [ 58,05 ; 58,37] 0,531 [ 0,530 ; 0,532] Combination (3e) 225,63 [ 225,27 ; 225,99] 163,43 [ 163,18 ; 163,68] 17,01 [ 16,98 ; 17,04] 66,63 [ 66,43 ; 66,83] 61,52 [ 61,36 ; 61,68] 0,527 [ 0,526 ; 0,529] Table 6 Results of experiments 3a, 3b, 3c and 3d, where the transportation mode of the MI is
an experimental factor.
5.5. BASE LOCATION FOR MI
The new base location for the MI has been explored using the PSC in Heerenveen. Results indicate that both the difference in time until treatment and patient outcomes are small and slightly worse for Heerenveen, but nevertheless a suitable option.
Base location for the MI Vehicle type Time until treatment (min) [min CI ; max CI] DIDO Times (min) [min CI ; max CI] vehicle transport MI time (min) [min CI ; max CI] Total MI time from call until present (min) [min CI ; max CI] Total waiting time until MI arrives (MI) [min CI ; max CI] % of mRS score 0-2 [min CI ; max CI] UMCG (4a1) Ambulance 233,04 [ 232,68 ; 233,41] 170,92 [ 170,66 ; 171,18] 24,32 [ 24,26 ; 24,39] 73,94 [ 73,70 ; 74,18] 68,80 [ 68,63 ; 68,98] 0,518 [ 0,517 ; 0,520] Heerenveen (4a2) Ambulance 235,36 [ 235,00 ; 235,72] 173,16 [ 172,91 ; 173,42] 26,79 [ 26,73 ; 26,85] 76,41 [ 76,17 ; 76,64] 71,26 [ 71,08 ; 71,43] 0,516 [ 0,514 ; 0,517] UMCG (4b1) Car 244,19 [ 243,82 ; 244,56] 182,01 [ 181,75 ; 182,27] 35,64 [ 35,55 ; 35,72] 85,26 [ 85,00 ; 85,51] 80,09 [ 79,91 ; 80,27] 0,505 [ 0,504 ; 0,507] Heerenveen (4b2) Car 247,80 [ 247,43 ; 248,17] 185,60 [ 185,34 ; 185,86] 39,26 [ 39,17 ; 39,35] 88,88 [ 88,62 ; 89,14] 83,69 [ 83,51 ; 83,88] 0,501 [ 0,499 ; 0,502] UMCG (4c1) Helicopter 222,32 [ 221,96 ; 222,68] 160,12 [ 159,87 ; 160,38] 13,67 [ 13,63 ; 13,70] 63,29 [ 63,08 ; 63,49] 58,21 [ 58,05 ; 58,37] 0,531 [ 0,530 ; 0,532] Heerenveen (4c2) Helicopter 223,32 [ 222,96 ; 223,68] 161,12 [ 160,87 ; 161,38] 14,68 [ 14,64 ; 14,71] 64,30 [ 64,09 ; 64,50] 59,22 [ 59,05 ; 59,38] 0,530 [ 0,529 ; 0,531] UMCG (4d1) Combination 225,63 [ 225,27 ; 225,99] 163,43 [ 163,18 ; 163,68] 17,01 [ 16,98 ; 17,04] 66,63 [ 66,43 ; 66,83] 61,52 [ 61,36 ; 61,68] 0,527 [ 0,526 ; 0,529] Heerenveen (4d2) Combination 227,56 [ 227,20 ; 227,92] 165,37 [ 165,12 ; 165,62] 18,96 [ 18,93 ; 18,99] 68,58 [ 68,38 ; 68,78] 63,46 [ 63,29 ; 63,62] 0,525 [ 0,523 ; 0,526] Table 7 Results of experiments 4a, 4b, 4c and 4d, where a new base location for the MI is
23
5.6. COMBINATION OF EXPERIMENTS
A variety of combinations of experimental factors have been explored. All experiments have the location of when the MI should be present set to right before the IAT treatment. All possible combinations of the notification moment together with the transportation mode has been explored. The results indicate that all the times until treatment are significantly lower then without combining experimental factors. The total waiting times until the MI arrives in minutes go from 48.49 minutes (5b3) to a minimum of 3.15 minutes (5c1). It should be noted that it is likely that with a low waiting time until the MI arrives that it is more likely that the MI must wait for the patient. Furthermore, the likelihood of a good functional outcome goes up.
Combination of transportation mode and notification moment
Time until treatment (min) [min CI ; max CI]
Total waiting time until MI arrives (MI) [min CI ; max CI]
% of mRS score 0-2 [min CI ; max CI]
Ambulance + On Scene (5a1) 169,94 [ 169,63 ; 170,26] 5,70 [ 5,62 ; 5,79] 0,593 [ 0,592 ; 0,594] Ambulance + Arrival at PSC (5a2) 189,54 [ 189,23 ; 189,86] 25,34 [ 25,17 ; 25,50] 0,570 [ 0,569 ; 0,571] Ambulance + After CTA (5a3) 201,47 [ 201,13 ; 201,80] 37,23 [ 37,00 ; 37,45] 0,556 [ 0,554 ; 0,557] Car + On Scene (5b1) 173,61 [ 173,30 ; 173,92] 9,51 [ 9,40 ; 9,62] 0,589 [ 0,587 ; 0,590] Car + Arrival at PSC (5b2) 198,30 [ 197,98 ; 198,62] 34,19 [ 34,01 ; 34,38] 0,560 [ 0,558 ; 0,561] Car + After CTA
(5b3) 212,60 [ 212,26 ; 212,94] 48,49 [ 48,27 ; 48,72] 0,543 [ 0,541 ; 0,544] Helicopter + On Scene (5c1) 167,25 [ 166,94 ; 167,57] 3,15 [ 3,09 ; 3,21] 0,596 [ 0,595 ; 0,598] Helicopter + Arrival at PSC (5c2) 181,87 [ 181,56 ; 182,19] 17,77 [ 17,63 ; 17,91] 0,579 [ 0,578 ; 0,580] Helicopter + After CTA (5c3) 190,99 [ 190,66 ; 191,32] 26,88 [ 26,67 ; 27,10] 0,568 [ 0,567 ; 0,570] Combination + On Scene (5d1) 167,79 [ 167,47 ; 168,11] 3,68 [ 3,62 ; 3,75] 0,596 [ 0,594 ; 0,597] Combination + Arrival at PSC (5d2) 183,92 [ 183,60 ; 184,23] 19,81 [ 19,67 ; 19,96] 0,577 [ 0,575 ; 0,578] Combination + After CTA (5d3) 194,21 [ 193,88 ; 194,54] 30,11 [ 29,89 ; 30,32] 0,564 [ 0,563 ; 0,566]
Table 8 Results of experiment 5a, 5b, 5c and 5d, where a combination of experimental factors
24
6. DISCUSSION
6.1. MOMENT OF NOTIFICATION
The MI organizational model shows to have better results, concerning both time until treatment and patient outcomes. The MI model uses three moments of MI notification; 1) On scene, 2) Arrival at PSC and 3) After CTA scan. With early notification the time that the patient must wait for the MI to arrives decreases drastically, decreasing the time until treatment. The difference in DIDO times between notification after the CTA scan and when the EMS are on scene is approximately 53 minutes. The process of MI transportation proceeds parallel with the patient following the remaining steps in the system. The notify moment should commence as early as possible, maximizing parallel times while the MI is transported to the PSC. However, there are drawbacks of early notification. The need for IAT treatment can only be determined after the CTA scan, showing an LVO for approximately 30% of the patients (Venema et al. 2019). When the CTA scan deems the patient eligible for IAT treatment it is necessary to transport the MI. When the MI however is notified of a possible LVO earlier in the process there is the chance that the MI drives unnecessarily to the PSC. The simulation study only takes patients with an LVO into account and so the implication of time lost on unnecessary MI travel with early notification is not known. The development of prehospital stroke scales can be used to identify patients that are likely to have an LVO (Venema et al. 2019). The use of these diagnostical tools can be used to decrease the probability that the MI must travel unnecessary to the PSC, hence reducing resources and allowing for early notification. Furthermore, if the notification moment for the MI is set after the CTA treatment it is recommended to follow the route where the patient receives the CTA scan before IVT treatment, to reduce waiting time until the MI arrival.
6.2. MOMENT THAT MI SHOULD BE PRESENT
25
the aspiration that each PSC is accustomed to preparation phase of the IAT treatment, with the intention of MI arrival just before the IAT treatment should commence.
6.3. MODE OF TRANSPORTATION
The transportation of the MI from the base location to the PSC can either be by ambulance, car, helicopter or a combination of both the ambulance and the helicopter. Compared to the DS model the time until treatment is reduced when using an ambulance, helicopter or combination (reduction: 8.69 min, 19.41 min, 16.1 min). When the car is used the time until treatment is slightly longer (increase: 2.46 min). The use of the ambulance, helicopter or a combination is therefore a recommended outcome as a mode of transportation. However, using the car collectively with an earlier notification moment or with the MI present just before the IAT treatment begins, will likewise generate better outcomes then the DS model. This means that the recommended mode of transportation depends on the full set up of the organizational model. The aforementioned also corresponds to the article of Brekenveld et al. (2018) where the MI was transported by taxi; the time until treatment was significantly better compared to the DS model.
The choice of mode of transportation does however depends on regional characteristics. When the distances between the MI base location and the PSCs are far apart, it might be better to re-evaluate and try to make helicopter transport available as much as possible. This is however dependent on weather conditions, amount of helicopters available and landing areas. Moreover, the helicopter start-up times might differ. Additionally, when a car is chosen, there might be an abundance of traffic jams, increasing transportation times, making an ambulance a better option, since an ambulance is allowed to bypass traffic jams. Additionally, the choice can be made depending on the resources available. If ambulance and helicopter transport is scarce it might be beneficial to make the decision to use a car. Additionally helicopter transportation costs are exorbitant compared to transportation by car or ambulance.
6.4. BASE LOCATION OF MI
With the MI organizational the MI base location is not necessarily dependent on the current CSC location, because the MI is transported to the PSC. However, it is beneficial to locate the base location somewhere central in the region and furthermore the MI regularly has additional responsibilities in a hospital. The more central base location in Heerenveen has been explored as an alternative to the UMCG as base location, in consultation with a field expert. While the location is more central, the region has a few PSCs that lay further away, therefore extending the average transportation times in comparison with the UMCG. The simulation study does nevertheless show the possibility of determining an optimal base location for the MI organizational model.
6.5. COMBINATION OF EXPERIMENTS
26
experiments the waiting time until the MI arrives was close to zero. This results in a reduction of the total time until treatment and improved patient outcomes. However, the chance is more then likely that the MIs must wait on the patient before the IAT treatment can commence. If both time and transportation resources are available helicopter transportation will be a good choice, otherwise it is recommended to choose another mode of transportation, with a slightly higher waiting time for the MI to arrive. Due to a combination of experimental factor this will have a limited effect on the time until treatment Furthermore, the effect on the decline of the time until treatment gets smaller when the waiting time until the MI arrives reaches zero for more patients.
6.6. LIMITATIONS
This simulation study explored the MI model for patient with an LVO. This means that the MI should always be transported to the PSC to do the IAT treatment. In real live stroke systems this is not always the case. When early notification is used, the MI is transported unnecessarily to a PSC on a regular basis, meaning a waste of resources. Early notification in combination with screening scales (Venema et al. 2019) will limit the unnecessary MI transport.
Additionally, the time that de MI needs to drive back to the base location is not included in this simulation. Therefore, there might be no MI available when more than one stroke occurs at the same time or at close proximity to eachother.
Lastly this simulation study has been based on the specific case of the North of the Netherlands. Regions with dissimilar regional and organizational characteristics might not receive similar outcomes. All PSCs included in the simulation have the means to give IAT treatment to the patient. This might differ for other regions.
6.7. FUTURE RESEARCH
27
7. CONCLUSION
28
BIBLIOGRAPHY
Appireddy, R., Zerna, C., Menon, B. K., & Goyal, M. (2016). Endovascular Interventions in Acute Ischemic Stroke: Recent Evidence, Current Challenges, and Future Prospects.
Current Atherosclerosis Reports, 18(7). https://doi.org/10.1007/s11883-016-0588-z
Brekenfeld, C., Goebell, E., Schmidt, H., Henningsen, H., Kraemer, C., Tebben, J., Flottmann, F., Thomalla, G., & Fiehler, J. (2018). “Drip-and-drive”: Shipping the neurointerventionalist to provide mechanical thrombectomy in primary stroke centers. Journal of
NeuroInterventional Surgery, 10(10), 932–936.
https://doi.org/10.1136/neurintsurg-2017-013634
Campbell, B. C. V., & Khatri, P. (2020). Stroke. In The Lancet. https://doi.org/10.1016/S0140-6736(20)31179-X
Chalos, V., LeCouffe, N. E., Uyttenboogaart, M., Lingsma, H. F., Mulder, M. J. H. L., Venema, E., Treurniet, K. M., Eshghi, O., van der Worp, H. B., van der Lugt, A., Roos, Y. B. W. E. M., Majoie, C. B. L. M., Dippel, D. W. J., Roozenbeek, B., Coutinho, J. M., van Oostenbrugge, R. J., van Zwam, W. H., Boiten, J., Vos, J. A., … Alves, H. (2019). Endovascular Treatment With or Without Prior Intravenous Alteplase for Acute Ischemic
Stroke. Journal of the American Heart Association, 8(11), 1–15.
https://doi.org/10.1161/JAHA.118.011592
Churilov, L., & Donnan, G. A. (2012). Operations Research for stroke care systems: An opportunity for The Science of Better to do much better. Operations Research for Health
Care, 1(1), 6–15. https://doi.org/10.1016/j.orhc.2011.12.001
Ciccone, A., Berge, E., & Fischer, U. (2019). Systematic review of organizational models for intra-arterial treatment of acute ischemic stroke. International Journal of Stroke, 14(1), 12–22. https://doi.org/10.1177/1747493018806157
Diaz, M. A., Hendey, G. W., & Bivins, H. G. (2005). When is the helicopter faster? A comparison of helicopter and ground ambulance transport times. Journal of Trauma - Injury, Infection
and Critical Care, 58(1), 148–153. https://doi.org/10.1097/01.TA.0000124264.43941.41
Donnan, G. A., Fisher, M., Macleod, M., & Davis, S. M. (2008). Stroke. In The Lancet. https://doi.org/10.1016/S0140-6736(08)60694-7
Erasmus MC University Medical Center. (n.d.). MR CLEAN-R REGISTRY. Retrieved October 24, 2020, from https://www.mrclean-trial.org/
Fit Data in DataFit. (n.d.). Retrieved October 25, 2020, from
https://docs.plm.automation.siemens.com/content/plant_sim_help/15/plant_sim_all_in_o ne_html/en_US/tecnomatix_plant_simulation_help/step_by_step_help/modeling_the_m aterial_flow/preparing_data_for_the_simulation_with_datafit/step_3_decide_which_distri bution_to_use/fit_data_in_datafit.html
Geoffrey A Donnan, Marc Fisher, Malcolm Macleod, S. M. D. (2009). Emergency and
comprehensive care for stroke needed. The Lancet, 373(9674), 1496.
https://doi.org/10.1016/S0140-6736(09)60833-3
Holodinsky, J. K., Williamson, T. S., Demchuk, A. M., Zhao, H., Zhu, L., Francis, M. J., Goyal, M., Hill, M. D., & Kamal, N. (2018). Modeling Stroke Patient Transport for All Patients with
Suspected Large-Vessel Occlusion. JAMA Neurology, 75(12), 1477–1486.
29
Horeczko, T., Marcin, J. P., Kahn, J. M., & Sapien, R. E. (2014). Urban and Rural Patterns in Emergent Pediatric Transfer : A Call for Regionalization. The Journal of Rural Health,
30, 252–258. https://doi.org/10.1111/jrh.12051
Jauch, E. C., Saver, J. L., Adams, H. P., Bruno, A., Connors, J. J. B., Demaerschalk, B. M., Khatri, P., McMullan, P. W., Qureshi, A. I., Rosenfield, K., Scott, P. A., Summers, D. R., Wang, D. Z., Wintermark, M., & Yonas, H. (2013). Guidelines for the early management of patients with acute ischemic stroke: A guideline for healthcare professionals from the
American Heart Association/American Stroke Association. Stroke.
https://doi.org/10.1161/STR.0b013e318284056a
Kilgore, R. V., & Langford, R. W. (2009). Reducing the failure risk of interdisciplinary healthcare
teams. In Critical Care Nursing Quarterly.
https://doi.org/10.1097/CNQ.0b013e3181a27af2
Lahr, M. M. H., Maas, W. J., Van Der Zee, D. J., Uyttenboogaart, M., & Buskens, E. (2020). Rationale and design for studying organisation of care for intra-arterial thrombectomy in
the Netherlands: Simulation modelling study. BMJ Open, 10(1), 1–7.
https://doi.org/10.1136/bmjopen-2019-032754
Maas, W. J., Lahr, M. M. H., Buskens, E., Zee, D.-J. van der, & Uyttenboogaart, M. (2020). Pathway Design for Acute Stroke Care in the Era of Endovascular Thrombectomy. Stroke,
November, 3452–3460. https://doi.org/10.1161/STROKEAHA.120.030392
Medical, A. E. (2006). The Future of Emergency Care in the United States Health System.
Annals of Emergency Medicine, 48(2), 115–120.
https://doi.org/10.1016/j.annemergmed.2006.06.015
Morey, J. R., Oxley, T. J., Wei, D., Kellner, C. P., Dangayach, N. S., Stein, L., Hom, D., Wheelwright, D., Rubenstein, L., Skliut, M., Shoirah, H., De Leacy, R. A., Singh, I. P., Zhang, X., Persaud, S., Tuhrim, S., Dhamoon, M., Bederson, J., Mocco, J., … Weinberger, J. (2020). Mobile Interventional Stroke Team Model Improves Early Outcomes in Large
Vessel Occlusion Stroke. Stroke, December, 3495–3503.
https://doi.org/10.1161/strokeaha.120.030248
PLM, P. S. (2012). Available at: http://www.plm.
automation.siemens.com/en_us/products/tecnomatix/plant_design/plant_ simulation.shtml.
Rabinstein, A. A., & MD, F. (2017). Treatment of Acute Ischemic Stroke. Continuum, February, 62–81.
Regenhardt, R. W., Mecca, A. P., Flavin, S. A., Boulouis, G., Lauer, A., Zachrison, K. S., Boomhower, J., Patel, A. B., Hirsch, J. A., Schwamm, L. H., & Leslie-Mazwi, T. M. (2018). Delays in the air or ground transfer of patients for endovascular thrombectomy.
Stroke, 49(6), 1419–1425. https://doi.org/10.1161/STROKEAHA.118.020618
30
Sacco, R. L., Kasner, S. E., Broderick, J. P., Caplan, L. R., Connors, J. J., Culebras, A., Elkind, M. S. V., George, M. G., Hamdan, A. D., Higashida, R. T., Hoh, B. L., Janis, L. S., Kase, C. S., Kleindorfer, D. O., Lee, J. M., Moseley, M. E., Peterson, E. D., Turan, T. N., Valderrama, A. L., & Vinters, H. V. (2013). An updated definition of stroke for the 21st century: A statement for healthcare professionals from the American heart
association/American stroke association. Stroke, 44(7), 2064–2089.
https://doi.org/10.1161/STR.0b013e318296aeca
Saver, J. L. (2006). Time is brain - Quantified. Stroke, 37(1), 263–266.
https://doi.org/10.1161/01.STR.0000196957.55928.ab
Seker, F., Möhlenbruch, M. A., Nagel, S., Ulfert, C., Schönenberger, S., Pfaff, J., Ringleb, P. A., Steiner, T., Bendszus, M., & Herweh, C. (2018). Clinical results of a new concept of neurothrombectomy coverage at a remote hospital—“drive the doctor.” International
Journal of Stroke, 13(7), 696–699. https://doi.org/10.1177/1747493018765267
Siemens. (n.d.). Technomatix Plant Simulation. Retrieved October 25, 2020, from https://www.dex.siemens.com/plm/tecnomatix/plant-simulation
Tiwari, V., & Heese, H. S. (2009). Specialization and competition in healthcare delivery networks. Health Care Management Science. https://doi.org/10.1007/s10729-008-9096-1
Uchiyama, N., Misaki, K., Mohri, M., Kamide, T., Hirota, Y., Higashi, R., Minamide, H., Kohda, Y., Asahi, T., Shoin, K., Iwato, M., Kita, D., Hamada, Y., Yoshida, Y., & Nakada, M. (2016). Treatment result in the initial stage of Kanazawa mobile embolectomy team for acute
ischemic stroke. Neurologia Medico-Chirurgica, 56(12), 737–744.
https://doi.org/10.2176/nmc.oa.2016-0101
Uyttenboogaart, M. (2017). REGIONAAL PROTOCOL INRICHTING CVA
( BEROERTEZORG ) Noord Nederland. Acute Zorgnetwerk Noord Nederland, september. https://www.acutezorgnetwerk.nl/inhoud/uploads/Regionaal-Protocol-Inrichting-CVA-beroertezorg-Noord-NL-AZNN-Definitief-25-september-2017.docx.pdf
Venema, E., Duvekot, M. H. C., Lingsma, H. F., Rozeman, A. D., Moudrous, W., Vermeij, F. H., Biekart, M., Lugt, A. Van Der, Kerkhoff, H., Dippel, D. W. J., & Roozenbeek, B. (2019). Prehospital triage of patients with suspected stroke symptoms ( PRESTO ): protocol of a prospective observational study. BMJ Open, 1–6. https://doi.org/10.1136/bmjopen-2018-028810
Warren, J., Fromm, R. E., Orr, R. A., Rotello, L. C., & Mathilda Horst, H. (2004). Guidelines for the inter- and intrahospital transport of critically ill patients. In Critical Care Medicine. https://doi.org/10.1097/01.CCM.0000104917.39204.0A
Wei, D., Oxley, T. J., Nistal, D. A., Mascitelli, J. R., Wilson, N., Stein, L., Liang, J., Turkheimer, L. M., Morey, J. R., Schwegel, C., Awad, A. J., Shoirah, H., Kellner, C. P., De Leacy, R. A., Mayer, S. A., Tuhrim, S., Paramasivam, S., Mocco, J., & Fifi, J. T. (2017). Mobile interventional stroke teams lead to faster treatment times for thrombectomy in large
vessel occlusion. Stroke, 48(12), 3295–3300.
31
APPENDICES
A. SIMULATION MODEL
1. Mobile interventionist simulation model
Figure 4 The mobile interventionist simulation model including names of each step.
Figure 5 The methods of the mobile interventionist simulation model including names of each
method.
32
additional diagnostics are needed. However, if the MI should join the patient at MI_present2, the patient goes to wait for the patient at MI_present2.
There are three notification moments set where a MI is notified; 1) OnScene, 2) Arrival_to_CT or 3) Ct_to_CTA2, IVT_to_CTA1 or CT_toCTA3, depending on the patient’s route.
When the MI is notified the MI enters the system at the Source. Henceforth the MI reviews the know patient data and diagnostics to determine if the IAT treatment is necessary (CallforMI). In the simulation model only stroke patients who are eligible for IAT treatment are included. The MI then goes to the transportation vehicle (ResponseMI) after which the MI is transported (TransportMI) in either a car, ambulance or helicopter. When the vehicle arrives, the MI goes to the needed location (OnSceneMI). Depending on the agreed location in the system the MI moves to either MIPresent or MIPresent2a and MIPresent2b. Only when both the patient as the MI is present at the MIPresent or MIPresent2b location the patient can continue his or her path through the system.
2. Distributions
Activity duration Distribution Parameters
Hospitalized vs. patients
outside hospital
Discrete empirical Value Frequency Hospitalized 15 Outside hospital 152
Time from stroke onset in hospital to CT (hospitalized patients)
Continuous empirical Lower Bound Upper Bound Frequency
0 20 4
20 50 7
50 230 2
Time from stroke onset to 911 call
(patients outside hospital)
Continuous empirical Lower Bound Upper Bound Frequency
0 1 26 1 5 22 5 10 18 10 15 11 15 20 10 20 30 11 30 40 8 40 50 7 50 75 10 75 100 6 100 150 6 150 200 3
EMS Response Beta Lower endpoint = 2.29; Upper endpoint = 30.53; α1 = 2.59; α2 = 7.21
EMS on Scene Gamma Location = 1.30; α = 5.57; β = 2.71
EMS Transport Weibull Location = 0.00 α = 2.10; β = 13.05
Time from hospital arrival to CT
Continuous empirical Lower Bound Upper Bound Frequency
0 5 8 5 10 18 10 15 40 15 20 28 20 25 14 25 35 12 35 55 3
ED routing (3Catergories) Discrete empirical Value Frequency
Route 1: CT to IVT to CTA 57 Route 2: CT to CTA to IVT 62 Route 3: CT to CTA 30
Time from CT to IVT (route 1) Erlang μ = 13.58; σ = 16.96
Time from IVT to CTA (route 1) Erlang μ = 15.31; σ = 14.12
Time from CT to CTA (route 2)
Gamma Location = 0.00; α = 2.63; β = 3.53
Time from CTA to IVT (route 2) Erlang μ = 12.57; σ = 16.96
Time from CT to CTA (route 3) Lognormal μ = 23.06; σ = 21.72
33
Time of call with mobile interventionist (route 1)
Gamma Location = 0.00; α = 2.57; β = 14.18
Time of call with mobile interventionist (route 2)
Continuous empirical Lower Bound Upper Bound Frequency
0 5 12 5 15 7 15 25 14 25 35 13 35 60 9 60 90 3
Time of call with mobile interventionist (route 3)
Continuous empirical Lower Bound Upper Bound Frequency
0 15 6
15 30 5
30 45 5
45 60 9
60 95 3
Time for mobile interventionist to arrival at vehicle
Continuous empirical
(distribution values are multiplied by 0.667)
Lower Bound Upper Bound Frequency
0 5 5 5 10 29 10 15 60 15 20 31 20 30 11 30 40 2
Transportation of the mobile
interventionist to PSC -
Ambulance (MI base = UMCG)
Beta Lower endpoint = 0.00; Upper endpoint = 50.06; α1 = 2.16; α2 = 2.29
Transportation of the mobile interventionist to PSC - Car (MI base = UMCG)
Discrete empirical distribution Value Frequency 9 1 26 1 27 1 31 1 40 1 42 1 43 1 46 1 57 1
Transportation of the mobile
interventionist to PSC -
Helicopter (MI base = UMCG)
Discrete empirical distribution (for every value the startup time = 2 minutes is added)
Value Frequency 1.0543 1 7.3771 1 7.5457 1 9.6314 1 15.8086 1 14.9657 1 14.7029 1 15.2486 1 18.8086 1
Transportation of the mobile
interventionist to PSC –
Combination ambulance and helicopter
(MI base = UMCG)
Discrete empirical distribution Value Frequency
6.1414 1 17.7419 1 18.4242 1 21.1538 1 17.8086 1 16.9657 1 16.7029 1 17.2486 1 20.8086 1
Transportation of the mobile
interventionist to PSC -
Ambulance
(MI base = Heerenveen)
Discrete empirical distribution Value Frequency
