Emergency Care Waiting Times; The Impact of
Dynamics and Room Utilization
By
VERA VAN MANEN
University of Groningen Faculty Economics and Business
MSc. Supply Chain Management February 2016
Supervisors: Dr. J.T. van der Vaart
TABLE OF CONTENT
1. INTRODUCTION ... 4
2. THEORETICAL BACKGROUND ... 6
2.1 Emergency department ... 6
2.2 Emergency department crowding ... 7
2.3 Emergency department capacity ... 7
2.4 Emergency department dynamics ... 8
2.5 Research framework ... 8 3. METHODOLOGY ... 10 3.1 Case description ... 10 3.2 Data collection ... 11 3.3 Data-analysis ... 11 4. RESULTS ... 14 4.1 General dataset-analysis ... 14 4.1.1 Input-information ... 14
4.1.2 Distribution of patients per day ... 15
4.1.3 Distribution of patients per hour ... 16
4.1.4 Throughput times of different process steps ... 17
4.2 Responsiveness analysis ... 18
4.2.1 Work shift-analysis ... 18
4.2.2 Throughput diagram analysis ... 19
4.2.3 Work-in-process analysis ... 21
4.2.4 Hourly input and output patterns ... 22
4.3 Room analysis ... 23
4.2.1 Room allocation policies ... 23
4.2.2 Room capacity ... 24
4.2.4 Prioritizing patients ... 25
4.2.5 Gantt chart analysis ... 27
5. CONCLUSION & DISCUSSION ... 28
5.1 The responsiveness of the emergency department ... 28
5.2 The interaction between room utilization and throughput times ... 28
6. RECOMMENDATIONS ... 30
6.1 Recommendations for the emergency department of the Martini hospital ... 30
ABSTRACT
This research paper focusses on a different approach in reducing waiting times at the emergency department. The studies up until now concetrated on the static factors influencing throughput time, and did not take into account that these processes are non-stationary. Therefore waiting times may result from earlier mismatches in demand and supply, and the daily dynamics of the emergency department may provide explaining starting points. In some preliminary research it is suggested that an inadequate reaction on an increase of input, due to a lack of responsiviness, is causing a delay in patient throughput time throughout the day. This research investigates if the same patterns apply within a different setting, and creates an in-depth understanding by considering the room capacity as an extra element. The input-output modelling techniques indicated the effect of these daily dynamics and revealed the interaction between room utilization and throughput times. This showed that in general an relative inadeqaute reaction on an increase of input is recognized, and additionally an inefficient assignment of room capacity is identified which causes concerns in care provided to more urgent patients. The awareness of creating output in the morning, and related changes in staff capacity profiles can potentially avoid waiting times and reduce throughput times. In addition, more consideration of triage codes in assigning treatment rooms to patients, and a focus on the sufficient use of room capacity can support in reducing these throughput times.
Key words: patient throughput time, waiting time, emergency department, emergency department dynamics, room capacity, room utilization, responsiveness.
1. INTRODUCTION
The emergency department (ED) of a hospital has a crucial role in saving lives. This crucial role ensures that is it important to increase quality and decrease process failures in providing emergency care. Due to an aging and growing population and at the same time financial pressures, ED’s have increasingly difficulties with matching highly fluctuating demand and available capacity (Jack & Powers, 2009). An important consequence of these aspects is ED crowding, which is one of the main causes of an increase in waiting time for patients (Derlet & Richards, 2000; Arkun, Briggs & Patel, 2010). Nowadays, waiting time at the ED is a common medical care issue and negatively affects patients’ safety, quality of care and patient and providers’ satisfaction (Kheirbek, Beygi, Zarhoush, Alemi, Smith, Fletcher, Seton & Hawkins, 2015). This research paper creates more understanding of the factors that are causing a delay in patient throughput times and consequently increases in patient waiting times.
In today’s literature several research papers indicate that crowding is a significant international problem of ED’s, which may affect the quality and access of healthcare (Hoot & Aronsky, 2008; Arkun et al., 2010; Derlet & Richards, 2000). The phenomenon of crowding describes the situation in which demand for service exceeds the ability to provide care within a reasonable time frame (Derlet & Richards, 2000). Several studies have found that crowding is associated with increased throughput times of patients at the ED (McCarthy, Zeger, Ding, Levin, Desmond, Lee & Aronsky, 2009; Asaro, Lewis, & Boxerman, 2008). The average patient throughput time at the ED in the United States can be determined at 5.5 hours (Arkun et al., 2010), whereas the average patient throughput time at the ED in the Netherlands is determined at approximately 2 hours (Van der Linden, Reijnen, Derler, Lindenboom, van der Linden, Lucas & Richards, 2013). Despite the relatively short throughput times of patients, crowding within the ED in the Netherlands occurs several times a week or even daily (Van der Linden et al., 2013). Next, several improvement ideas such as the ‘triage system’ and the ‘fast track system’ are introduced within the ED in order to decrease and control process flow times of patients (Van der Vaart, Vastag & Wijngaard, 2011). Furthermore, due to the fact that demand regularly outstrips capacity within the ED, capacity management is a critical component for maintaining and improving healthcare quality and patient safety (Trzeciak & Rivers, 2003).
these preliminary studies it is indicated that the ED has an inadequate response to patients’ arrival in the morning, due to a low level of responsiveness. The inadequate response to differences in arrival rates throughout the day has a significant negative impact on waiting times of patients at the ED. The focus of this paper is to find out if the same patterns and delaying factors are found in an ED working within a different setting and environment. Furthermore, analyzing the interaction between room utilization and patient throughput time creates an in-depth understanding of these dynamics. The preliminary studies focused on the total length of stay of patients, and did not take the room components of the ED into account. Therefore this paper will contribute to the in-depth knowledge of delaying factors at the ED, as the daily dynamics are affecting the available room capacity which can cause problems in providing on-time emergency care to incoming patients.
In order to create more understanding and insights into the effect of ED dynamics on patients’ throughput times, the pressure on room capacity and related opportunities to improve the use of capacity are investigated. By looking into other and multiple components of the dynamics, alternative causes of patient waiting times could potentially be identified. The additional insights in how and how long delaying factors are influencing throughput times, and the relation between the variations in these throughput times and interrelated room utilization are interesting and valuable. For instance, what will happen with the throughput times of patients when high utilization rates occur, and how is room capacity deployed when patients’ throughput times increases. The gap in today’s literature will therefore be limited by the in-depth understanding of the delaying factors influencing patient throughput time at the ED. The analysis of throughput times will be done by using the framework of Soepenberg, Land & Gaalman (2012). This framework originally diagnoses delivery reliability performance, and is designed to improve throughput times in make-to-order companies. The framework is applied in a healthcare setting, and the main approach that will be used is an input-output modelling technique based on historical data. The research paper aims to provide an answer on the following research question:
“How are the dynamics of the emergency department influencing patient throughput times?”
The practical contribution of this research paper is to develop recommendations to improve the overall throughput times of patients at the ED, and consequently decrease patient waiting times. The research question will be studied by assessing waiting time components, using a case study at the Martini Hospital in Groningen, The Netherlands.
The remaining part of this paper is structured into five parts. In the second section the theoretical background will be discussed. Next, the methodology will be addressed, which explains how this research is conducted. Furthermore, the results will be presented followed by the conclusion and discussion of the conducted research. Finally, the recommendations and suggestions for further research are presented.
2. THEORETICAL BACKGROUND
This section will describe the theoretical foundation of this study. The first paragraph is focused on the ED of a hospital, were the distinctive characteristics of this type of department are elaborated. In the second paragraph ED crowding is discussed, as this has an important impact on waiting times. Next, the capacity of an ED is elaborated in order to obtain an better understanding of the challenges ED’s are facing. Furthermore, the ED dynamics are reviewed. Finally, the research framework is provided, which summarizes the fundaments of this study. 2.1 Emergency department
The health expenditures within the Netherlands account for 14.5% of the GDP in 2014, in addition the overall spending on health and welfare in the Netherlands continues to grow (CBS, 2014). Nowadays, in a time of many governmental budget cuts, the smaller hospitals within the Netherlands are having significant financial problems (NU, 2015). The emergency care is the most visible component of the healthcare system, where process failures lead to significant public scrutiny and debate (Vaart et al., 2011). The ED is specialized in providing emergency medicine, while coping with the unplanned nature of patient’s attendance and a broad spectrum of illnesses and injuries. Due to this complex environment, the ED’s are facing important challenges according to excessive waiting times (Asplin et al., 2003). These excessive waiting times in the ED are causing patient dissatisfaction and an increase in the probability of patients leaving without treatment, which has an important impact on the quality of care (Kaushal, Zhao, Peng, Strome, Zhang & Chochinov, 2015). In addition, nowadays the ED’s are facing the challenge of providing high quality patient care while being cost-effective at the same time (El-Rafai, Garaix, Augusto & Xie, 2015).
The stochastic nature of emergency care ensured that rules were developed for situations where demand exceeds available resources (van der Vaart et al., 2011). Therefore the triage system is an essential element of modern medical care as it is necessary to assign relatively scarce resources to unlimited medical needs (FitzGerald, Jelinek, Scott & Gerdtz, 2010). The system should identify patients who need immediate attention and patients who can safely wait for a longer time or who might not need emergency care at all (Van Veen, Steyerberg, Ruige, van Meurs, Roukema, van der Lei & Moll, 2008). The most common used triage system within the Netherlands is the Manchester triage system, whereas the distinction between five different levels or urgency is made.
2.2 Emergency department crowding
According to van der Linden et al. 2013 crowding within the ED occurs several times a week or even daily, and is becoming a nationwide problem within the Netherlands. Due to an increase in the number of patients and a decrease in the number of resources, (over)crowding can be described as one of the important aspects in the ED. Subsequently, ED crowding occurs when the identified need for emergency services exceeds available resources for patient care in the in the ED (American College of Emergency Physicians, 2006). The phenomenon of crowding is therefore the result of a weak relationship between the interaction of supply and demand. Therefore waiting time arises when the demand for work goes beyond the facility’s service capacity (El-Rafai et al., 2015). In addition, overcrowding within the ED is difficult to determine specifically because definitions based on precise waiting times, or quantitative delays in actual ED care are lacking.
Derlet & Richards (2000) identified fourteen causes of overcrowding in the ED, of which the most important causes within the scope of this research are: (1) delays in service provided by radiology, laboratory, and ancillary services, (2) a shortage of nursing staff, (3) a shortage of administrative/clerical support staff, (4) a shortage of on-call specialty consultants or lack of availability, and (5) a shortage of physical plant space within the ED. In addition, the most important consequences of overcrowding at the ED are: (1) public safety at risk, (2) prolonged pain and suffering, (3) long waits and dissatisfaction of patients, and (4) decreased physician productivity (Derlet & Richards, 2000).
2.3 Emergency department capacity
influencing the ED (Jack & Powels, 2009): (1) the static or declining number of hospital beds, (2) increasing financial pressures on hospitals, (3) increasing number of hospital closures and capacity reduction, and (4) growing labor shortages for skilled health-care providers and caregivers. In addition, approximately 70% of hospitals do not maximize the use of existing resources, and hospitals often “feel full” when they are not (McCaughey, Erwin, & DelliFraine, 2015).
Because of the relatively unpredictable nature of the ED, the capacity needs to be flexible in order to cope with variability and eventually optimize the overall throughput times of the patients. The patient demand at the ED could to some extent be forecasted by analyzing
historical data (Asplin, Flottemesch, & Gordon, 2006; Ekelund et al., 2011). In the
Netherlands the highest peak in demand is indicated on Monday and Friday, whereas the indication with the lowest peak in demand is on Sunday (Verbree, 2013). Furthermore, the variability in patient demand within the days causes a required focus on the processes of patient flow. According to Terwiesh et al. (2011) waiting time can be the consequence of weak (capacity) responsiveness, but also other causes may apply.
2.4 Emergency department dynamics
The great sense of urgency of overcrowding at the ED ensured extensive research into the causes of long waiting times at the ED. The main causes that are identified in literature can be summarized and distinguishes into static external and internal factors. The external factors include for example the number of patients and patients’ urgency (Ekelund et al, 2011; Arkun et al., 2010; Van der Vaart et al., 2011), whereas the internal factors are mainly related to capacity shortages and delays of external services (Arkun et al, 2010; Ekelund, 2011; Van der Vaart et al., 2011; Kheirbek et al., 2015; Derlet & Richards, 2000; Hoot & Aronsky, 2008). The external factors are hard to control, as the ED is not able to plan incoming patients and cannot deny admission to patients in need for care. In addition, the internal factors influencing patients waiting times are also hard to control as in today’s environment increasing capacity levels (i.e. bed and care givers) is generally not a viable option due to its costs (Land, van der Vaart & Fredendall, 2015). Therefore the dynamics of the ED could provide starting points for decreasing long waiting times. The previous mentioned studies used surveys, observations and regression analyses in order to identify the delaying causes at the ED. These methods did not recognize that patients arrival are non-stationary processes, and long waiting times at late hours may result from earlies mismatches between demand and supply (Land et al., 2015). The understanding of the daily dynamics at the ED could provide more insights in the delaying factors influencing throughput time, by considering the influence of an increase in input on patient throughput time during the day.
2.5 Research framework
patients are stacking up. Furthermore, following research suggest evidence for a positive effect of increased responsiveness on patient waiting times at the ED (Van Achteren, 2014). The increased responsiveness to an increase in patient demand early in the day, will logically lead to less waiting time for patients in the treatment room to be seen by a physician. This will consequently lead to less overcrowding at the ED. It could be concluded that the dynamics within the day can explain large parts of the waiting times, and the waiting time later on the day is a result of a queue building up earlier in the day. The adequate responses to arrival rate increases are a key to reducing emergency care waiting times (Land et al., 2015). However, these preliminary studies are conducted within only one hospital in the Netherlands. Therefore it is important to conduct research within more hospitals in the Netherlands, in order to analyze whether the same patterns can be identified in another environment and setting. In addition, it is an interesting aspect to reconsider and reevaluate these understandings of the causes of increased throughput times at the ED. In the preliminary studies particular attention is paid to the moments of arrival, triage and departure of patients. The moments when the patient enters and leaves the treatment rooms are not included. These moments could provide additional insights, as the reaction on increases of input has a significant impact on room utilization. For example, the moments when the throughput times of patients are increasing and output cannot keep up with the increase of patients’ arrival might cause conflicts among the available room capacity. The rates of room utilization can provide a delay in start of treatment, which causes an increase in waiting time for new incoming patients and potentially exceeds a reasonable and safe timeframe of handling patients at the ED. Furthermore, the efficient and effective use of room capacity ensures that these delays in start of treatment are avoided or reduced. Therefore the interrelated relationship between room utilization and patient throughput time provides other or additional insights into the dynamics of the ED.
3. METHODOLOGY
The aim of this research is to gain more insight and understanding into waiting times of the emergency care in the Netherlands. In order to answer the research question a single case study is conducted. A case study resulted in an in-depth analysis of a specific phenomenon by observing actual practice in its natural setting (Voss, Tsikriktsis & Frohlich, 2002; Yin, 1994). The advantages of this approach are the strong base for theory building and explanation, and the rich data-analyses, which is due to the different sources of evidence. In this section the case description, data collection and data-analysis will be discussed.
3.1 Case description
3.2 Data collection
The data is collected by reviewing the information received from the information system, observing techniques and semi-structured interviews. The combination between quantitative and qualitative research provided an optimal aggregation of research in order to answer the research question. The focus of the data-analyses is on the quantitative data, whereas the qualitative data provided additional information to test interpretations of the results.
The Martini hospital uses an information system to keep record of patients, within this system the data relating to patient treatments are recorded. The dataset received from this information system includes data from 1 January until 31th of October 2015, whereas a total of 20.560 patients visited the ED. The period of 1 until the 12th of June is excluded from the dataset, due to a pilot performed during this period that influences the data. This dataset includes information about patients’ age, assigned specialism, triage codes, patients’ origin, mode of transport, patients’ destination, moment of arrival the ED, moment of triage, moment of start treatment, moment of departure the ED, moment arrival observatory and moment of departure the observatory. The data was almost complete, but missed moments of triage of 174 patients. These missing values were replaced by the time of the prior process, in this case the time of arrival of the patients. The data of arrival, departure and start of treatment were found to be complete and reliable after correction. In some data the information was entered falsely, as there were sometimes treatment times of more than 24 hours. These treatment times are verified and corrected by checking a related dataset, were every process step of the patient was briefly described. The patients that did not have treatment at the ED are deleted from the dataset. The dataset is analyzed in order to ‘take a look in time’ and review/evaluate the daily operations during the period of time in the ED.
In addition to this quantitative analysis, qualitative methods are used in order to obtain understanding of issues in which the ‘how’ and ‘why’ questions are relevant. Firstly, observation on the work floor are used in order to identify and map the work process of the ED. Secondly, semi-structured interviews are used in order to provide an improved understanding of the work process and to test the interpretations of the results. By using three different sources of evidence, the (construct) validity is ensured within this research paper, which can also referred to as triangulation (Voss et al., 2002). The case at the Martini hospital provided enough data to satisfactorily address the research question.
increase of input is recognized. Next, the overall throughput time of patients is determined, and a distinction is made between average throughput times per weekday. The relationship between patient throughput time and amount of patients visited is tested by means of a regression analysis. In addition, the time until triage after patients’ arrival and the time until start of treatment are determined to indicate patterns in patient processes.
The next step involves the identification of the responsiveness at the ED. The work-shifts are established in order to determine when and how nursing and physician capacity is deployed. These work-shifts are compared with the average input-output patterns in order to diagnose how capacity is related to the dynamics. Next, four days are selected with a high average work-in-process in order to make throughput diagrams analyzing the daily dynamics of the ED. These dynamics are analyzed by taking a closer look to input and output of patients throughout the day, by identifying how these patterns are influencing total throughput time and by determining the recurring patterns. The throughput diagram shows the cumulative input and output of individual process steps, where the horizontal axis shows the cumulative time and the vertical axis show the general work unit (Soepenberg et al., 2012). In this analysis the time is represented in hours of the day, and the general work unit is expressed in the number of patients. The vertical distance between two process lines depicts the work-in-process, whereas the horizontal distance relates to the average throughput times. The throughput diagrams show the interaction between input and an increase in patient throughput time, whereas the moments of arrival, triage, start treatment and departure of patients are included. Furthermore, the work-in-process is analyzed by counting at each moment in time how many patients are present at the ED, in order to establish the effect of the analyzed responsiveness and related dynamics on crowding. Thereafter, the input and output rates for the selected days per hour are determined in order to assess the responsiveness in more detail. The last step involves the diagnoses of how room capacity has an impact on the previous identified dynamic assumptions, and related patient throughput time. The current room allocation process is described and mapped to indicate the main decisions affecting room capacity. Furthermore, the room utilization rates per hour for the whole dataset and the selected days are determined in order to analyze how the input and output during the day influences room utilization. Next, prioritization of patients is analyzed by considering the assignment of patients to treatment rooms at different moments in time. These moments are determined by the work-in-process of rooms, where each moment in time indicates how many patients occupy the (treatment) rooms. The prioritization of triage codes when taking patients into treatment is diagnosed, and the effects on on-time treatment of patients is determined. Finally, the role of specific room utilization in ED dynamics is analyzed by means of Gantt charts, whereas the inflow and outflow of patients into treatment rooms is mapped during the four selected days.
In addition to this comprehensive quantitative analysis, the identified causes are verified and validated using qualitative data. There are multiple meetings with the unit coordinator, meetings with the nursing staff, small interviews with nurses and physicians, and days were the daily work is observed, used to improve the understandings of the daily dynamics of the ED.
4. RESULTS
This section summarizes the results of this research paper. Firstly, a general analysis is done in order to describe the general process characteristics and identify input information of the ED. Next, a more detailed analysis of the dynamics and the responsiveness to deal with these dynamics is executed to determine the delaying throughput factors at the ED. Finally, the room capacity is investigated more in-depth to address the interrelated relationship with patient throughput times.
4.1 General dataset-analysis
The dataset provided by the ED covers a ten-month period. This period is from January 1, 2015 until October 31, 2015, excluding 1 until 12 June because a pilot that took place during this period which could affect the outcomes of the analyses. During this ten-month period there are 294 days included in which a total of 20.560 patients visited the ED.
The original dataset contained 21.648 patients, which included patients without any treatment at the ED. The requirement of being treated at the ED is taking as a starting point, and therefore the patients without any treatment are removed from the dataset.
4.1.1 Input-information
The input of the ED is highly variable in terms of timing, mix and volume. Therefore the factors contributing to this variation are analyzed in order to evaluate and discuss the input uncertainty of the processes at the ED.
The age of the 20.560 patients that had treatment at the ED is varying between 0 and 104, and the average age is determined at 50.3 years (appendix A1). There are eight different routes in which a patient can arrive at the ED. The majority of the patients are referred by the general practitioner, and only a small amount of patients arrive on their own initiative (appendix A2). This can be attributed to the Dutch policy which is focused on the prevention of patient self-referring to the ED. Next, there are several ways patients can arrive at the ED, most of the patients arrive by own transport (63%) or by ambulance (35%).
The triage process of patients takes place after patients’ arrival, which determines the urgency of patients complaints. The ED uses the Manchester triage system, which distinguishes five different levels of urgency, namely: red (immediate evaluation/treatment), orange (evaluation within 10 min), yellow (evaluation within 60 min), green (evaluation within 120 min), and blue (evaluation within 240 min). Table 4.1 shows that most of the patients are classified as yellow (43.1%), and thereafter green (34.8%). Only 1.1% of the patients’ needs immediate treatment, and only 0.5% of the patients can wait for four hours and does actually not belong at the ED.
Triage code Triage color Meaning # Patients Percentage
1 Red Acute 221 1.1%
2 Orange Emergency 3960 19.3%
3 Yellow Urgent 8858 43.1%
4 Green Standard 7153 34.8%
5 Blue Not urgent 97 0.5%
Unknown 271 1.3%
Table 4.1- Number of patient with triage codes
Furthermore, the patients that arrive at the ED are assigned to a certain specialism. The highest number of patients is assigned to the surgical specialism (41%). In appendix A3 is shown that the other most frequently visited specialisms are cardiology (15%), internal medicine (14%), pediatrics (7%) and lung (7%). Lastly, the patients that completed treatment are send to home (52%), are hospitalized (47%) or have other destinations as for example external facilities (1%).
4.1.2 Distribution of patients per day
The ED has to cope with high variation in the arrival patterns of patients. This high variety is caused by the unannounced arrival of patients at the ED. The ED does not make use of a reservation system, and is therefore not capable to plan the arrival of patients. This is reflected in figure 4.1 where there is a range in amount of arriving patients between 30 and 102, in the 294 days considered.
Figure 4.1 – Boxplot amount of patients per day
0,0 1,0 2,0 3,0 4,0 5,0 6,0 7,0 Ra te (p a( en ts p er h ou r) Time average hourly input per day average hourly output per day
4.1.3 Distribution of patients per hour
Due to the unannounced nature of arrival at the ED, patients can come in at any time of the day. The ED therefore has to cope with the uncertainty of patients input per day. In order to identify the structural input and output patterns at the ED, the hourly input and output rates of the ED are shown in figure 4.2. A more detailed analysis of the rates per hour can be found in appendix B. The data of all 20.560 patients is used in order to analyze these patterns, as there were no missing values in the arrival and departure information of the dataset. Therefore the rates per hour of the day are averaged across the 294 days. The vertical axis represent the number of patients, and the horizontal axis represent the timeframe. The patients who arrived half an hour before and half an hour after the full hour are grouped together in the graph. For example, the patients that arrived between the intervals of 09:30-10:30 hours are grouped together and plotted at 10:00 hours.
Figure 4.2- Average hourly arrival and departure rates
0:00:00 0:30:00 1:00:00 1:30:00 2:00:00 2:30:00 3:00:00 3:30:00 0 20 40 60 80 100 120 Av er ag e th ro ug hp ut ( m e
Amount of pa(ents
patients. The same patterns are found in the hourly input and output rates of particularly Monday and Friday, as these days are the busiest days at the ED (appendix C).
4.1.4 Throughput times of different process steps
The average throughput time of patients at the ED is 02:23:16 hours. However, the average is quite sensitive to outliners and therefore medians are used in order to indicate throughput times of patients. The overall throughput time median is determined at 02:10:00 hours, whereas the throughput times are measured as the time between arrival and departure of the patients. Figure 4.3 shows the relationship between the average throughput time of patients and the amount of patients arriving at the ED, whereas each marker represents a particular day. The regression line implicates that there is a weak relation between these variables, indicating that on average a small effect is recognized. The individual markers show a number of fluctuating values, suggesting a limited explanation for the increase in throughput time.
Fig 4.3- Scatter diagram of average throughput times
Time until triage after patients arrival Cumulative percentage 0:00:00 0.6% 0:05:00 8.2% 0:10:00 28.1% 0:15:00 48.3% 0:20:00 64.3%
Table 4.2- Time until triage after patients arrival
Furthermore, the time between arrival and start of treatment of all the patients is analyzed, and showed an overall median time of five minutes. The start of treatment indicates the moment when patients are placed into a treatment room. In appendix D2 the medians per weekday are shown, were a range between 00:03:00 and 00:06:00 in the longest and shortest times can be recognized. In the 75, 90 and 95% percentile is shown that this distribution is extremely tailed to the right. This implicates that most patients hardly have to wait before start of treatment, and for the remaining part of the patients waiting time before start of treatment in increasing rapidly. The right-tailed distribution seems a logical consequence of moments when capacity is fully utilized at the ED. The observations and interviews indicated that patients, with for example cardiac complaints or complaints after chemotherapy, have priority over other patients and are directly taken into treatment.
4.2 Responsiveness analysis
In order to establish the factors that influence patient throughput times at the ED, an analysis is made comparing and considering the dynamics and the responsiveness to deal with these dynamics.
4.2.1 Work shift-analysis
Fig 4.4- Schedule of nurses compared with hourly input-output patterns
The A-shift includes one nurse who is solely focusing on the coordination. Next, the B-shift includes one nurse who is solely focusing on the triage process, and one nurse focusing on pediatrics. Lastly, the D-shift includes one nurse who takes over the coordination of the other shift, and one extra nurse focusing on pediatrics. Figure 4.4 shows that from 09:30 hours there are six nurses working, and from 12:00 hours even eight nurses are working at the ED. In the ED of the MCL less capacity is allocated in the morning, which causes an even more inadequate reaction on an increase of input in the morning. Furthermore, figure 2.1 shows that between 16:00 and 18:00 hours the highest amount of nurses (10) are working at the ED, including B, C and D shifts. The average input and output patterns showed that the average arrival rate between those hours is determined at approximately 5.5 patients. An even higher arrival rate of approximately six patients is reached around 11:00 hours, however this average arrival rate is than handled with only six nurses. In the end of the afternoon handling over shifts takes place, which make it quite logical that somewhat more staff is assigned. However, maximum capacity is reached too late which facilitates an inadequate reaction on an increase of input in the morning. The interviews and observation revealed that the length of the queue was an indicator to increase capacity. Therefore ten nurses are assigned between 16:00-18:00 hours as this timeframe felt as the busiest moment at the ED by the nurses. This is comparable with the findings at the MCL were pressure to be more productive was only perceived in the afternoon. In the interviews the nurses indicated that in the morning there is some time to drink a cup of coffee, or do some administrative work such as reading protocols. An increase in the awareness of the importance of creating output in the morning can potentially improve the output rate even more, which prevents work-in-process from building up and eventually the queue at later moments of the day will be avoided or decreased.
4.2.2 Throughput diagram analysis
0 10 20 30 40 50 60 70 80 90 100 110 Cu mu la (ve n umb er o f p a(en ts Time arrival triage start treatment departure
average process to diagnose how these days could build up the highest work-in-process of the total dataset and to indicate the effect on the pressure of treatment rooms. Therefore the days for the analysis are selected based on the principles of Little’s law. These principles imply that the average work-in-process is equal to the average output per hour times the average throughput time in hours. The combination of these numbers is used in order to determine the days were on average most patients were at the same time in process at the ED. The days that were selected are shown in table 4.3, these days were selected on the above-mentioned requirement and the completeness of the data. In addition, the weak relation between throughput times and amount of patients’ arrivals ensures that the days with on average highest work-in-process are also the days with the highest amount of arrivals.
Table 4.3 - Days selected for throughput diagram analysis
Figure 4.5 shows the throughput diagram of 29-09-2015, the other three throughput diagrams can be found in appendix D. The throughput diagrams show the cumulative input and cumulative output of certain steps in the process. The horizontal axis shows the cumulative time in hours, whereas the vertical axis shows the number of patients in the process. The horizontal distance between two curves depicts the average throughput times, and the vertical distance relates to the work-in-process (Soepenberg et al., 2012).
Figure 4.5- Throughput diagram of September 29, 2015.
Date Average throughput
The four lines represent the first time the patients comes into one of the specified processes, whereas the start of treatment is the first time a patient is placed into a treatment room. Figure 4.5 shows that around 09:00 hours the arrival of patients starts to increase. There is a considerable variation in the arrival rate per hour, between 10:00-11:00 hours ten patients’ arrive, and between 14:00-15:00 only four patients arrive at the ED. The throughput diagram shows that the ED does not appear to directly have a response on the increase in patients arrival. In detail, there are respectively nine patients arriving between 09:00-10:00 hours, and ten patients arriving between 10:00-11:00 hours. The highest numbers of output occur between 14:00-15:00 hours were nine patients leave, and between 17:00-18:00 were ten patients leave the ED. The extreme high amount of patients’ arrival in the morning ensures that the ED is not able to respond adequate, more specific the output cannot keep up with the increase of input in the morning. This ensures an increase in throughput time (horizontal distance) and work-in-process (vertical distance) right after 09:00 hours. However, a relatively consequent output rate is realized between 11:00-18:30 hours indicating that the ED is working at maximum capacity and is therefore not able to handle these extreme peaks in patients’ arrival. In addition, as the throughput times and respectively work-in-process are catching up in the end of the afternoon, there is almost zero output for an hour between 19:00 and 20:00 hours. This is causing an increase in throughput time and respectively work-in-process again in the evening. An adequate reaction on an increase of input will reduce throughput times, and prevents work-in-process from building up. The high amount of patients’ arrival raises the question if the inadequate reaction of the output could be avoided, as a relatively stable and maximum output is generated during a considerable part of the day. The comparison with the throughput diagrams of the other selected days show the same high amount of patients’ arrival during the morning, and subsequently an increase in work-in-process and throughput times (appendix E).
4.2.3 Work-in-process analysis
0 2 4 6 8 10 12 14 16 18 20 22 24 26 0:00 1:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00 0:00 num be r of pa( ents Time WIP total WIP in treatment nr rooms
nr treatment rooms
Figure 4.6- Work-in-process of September 29, 2015
The red line in figure 4.6 represent the maximum capacity of the total amount of rooms (i.e. including specific rooms such as trauma rooms), whereas the green line represents the maximum capacity of the treatment rooms. In addition, the blue line represents the total work-in-process at the ED, and the purple line represents the work-work-in-process of the patients in treatment at the ED. The blue line in figure 4.2 shows that after 10:00 hours more people are present at the ED than the total number of available rooms. This implicates that the ED is getting overcrowded, and incoming patients have to wait in the waiting room. The total work-in-process reaches its peak between 12:00 and 13:00 hours, in this timeframe there are 24 patients present at the ED. At the end of the day the total work-in-process is decreasing, but due to the lack of output between 19:00 and 20:00 hours the work-in-process is increasing again in the evening. Furthermore, figure 4.6 shows that the number of people for which treatment has started (the purple line) exceeds the total number of rooms between 12:00-13:00 and 16:00-17:00 hours. This implicates that there are more patients in treatment than rooms are available, meaning that there are patients waiting in the waiting area while being in treatment. The extreme high input rates in the morning are causing an increase in waiting time for patients arriving in the end of the morning, which continues until the end of the afternoon. These patterns are also recognized in the other three selected days, which are shown in appendix F. The increase in total work-in-process ensures that the queue is building up, and eventually more patients are taken into treatment than room capacity is available.
4.2.4 Hourly input and output patterns
Fig 4.7– Hourly arrival and departure rates four selected days
Figure 4.7 shows an extreme peak in arrival rate (blue line) on the four selected days compared with the average hourly input rate (orange line). In all four hourly input-output figures is shown that the input rate is approximately twice as much as the average hourly input rate in the morning. The output rate on these four selected days is comparable or higher than the average, and due to these peaks in arrival rate logically a more fluctuating output rate is recognized. The high and early peaks in arrival rate and the relatively low compared output rate ensures that the work-in-process keeps building up. For example, on 29-09-2015 there are 34 patients arriving between 09:00-13:00 hours, and in the same time frame only 16 patients are leaving the ED. This obviously creates a difficult situation to handle for the ED. As a result these days are developing on average high work-in-process and are perceived as ‘problem days’ at the ED.
4.3 Room analysis
In order to identify how and when room capacity is influencing throughput times at the ED, an analysis is made considering the different aspects of room utilization contributing to the dynamics of the ED.
4.3.1 Room allocation policies
implies that these patients have priority over patients that are waiting in the waiting room. The precise and specific arrival time of patients via ambulance is unknown, therefore the care coordinator anticipates on the notification of patients’ arrival by reserving treatment rooms when the ED is crowded.
The ED uses an observatory with 16 beds, which serves various diverse functions. Firstly, patients who need hospitalization for less than 24 hours can access the observatory. Secondly, patients involving waiting time of already two hours at the ED without any clarity could be referred into the observatory. Lastly, the observatory is used for through flow of the ED as this is an alternative opportunity to treat patients when the ED is crowded.
4.3.2 Room capacity
The average room utilization per hour across the whole dataset in analyzed, in order to indicate how the input-output patterns are influencing the room utilization rates. Next, an analysis of the four selected days is conducted in order to indicate the consequences of on average high work-in-process on the room utilization at the ED. The utilization of each room is determined, and then averaged to come up with the utilization rate of the total set of rooms (appendix G). An analysis of the utilization of all individual rooms at the emergency department showed that the utilization of the rooms with a specific function (i.e. plaster room, reanimation room and trauma rooms) do not pass a utilization rate of 40%. Therefore the analysis of the utilization will solely focus on the treatment rooms. Figure 4.8 shows the hourly input-output pattern across the 294 days and the corresponding utilization rate of the treatment rooms.
Fig 4.8- Average hourly input-output pattern and utilization rate
In figure 4.8 is shown that the average utilization rate does not pass a rate of 70% throughout the day. The hours with the highest utilization rates correspond with the most busiest hours of the day at the ED. An analysis of the capacity at individual days showed the utilization rate of treatment rooms never reaches 100% in an hour, and a maximum of 23 days reached an utilization rate above 90% in an hour (appendix G). These relatively low amounts of average utilization rates indicate that the inadequate reaction on an increase of input in the morning do
average utilization rates corresponds to the times were output cannot keep up with the increase of input which implies that a more sufficient use of room capacity could be reached by a more adequate reaction on an increase of input in the morning.
Furthermore, the utilization rates of the four selected days are analyzed in order to indicate the consequences of average high work-in-process and the related high input rates in the morning on the utilization of treatment rooms. Therefore the hourly input-output rates related to the utilization rates of treatment rooms are shown in figure 4.9.
Fig 4.9- Hourly arrival and departure rates vs utilization rate four selected days
Figure 4.9 shows an increase in utilization rates short after the peaks in arrival rate during the morning. The utilization rates on all four selected days reach at least 90%, and on 04-05-2015 is reaches even 97% between 14:00 and 15:00 hours. On three of the four days the utilization rate is higher than 70% between 11:00 and 20:00 hours, except 29-09-2015 were between 14:00 and 16:00 hours a lower utilization rate is reached. This implicates that days with on average high work-in-process, were the arrival rate in the morning is extremely high, are consequently encountering problems and conflicts among the capacity of rooms.
4.3.4 Prioritizing patients
Fig 4.10- Work-in-process of (treatment) rooms 29-09-2015
The blue line in figure 4.10 shows the work-in-process of all the rooms (16), whereas the green line shows the process of the treatment rooms (12) at the ED. This work-in-process analyses is excluding patient in the waiting room, allowing to see the total utilization of the room capacity throughout the day. Figure 4.6 shows that between 00:00-10:00 hours there is still enough capacity left to treat incoming patients, between 10:00-11:00 hours the maximum capacity is almost reached, and between 11:00-14:00 hours the ED is working around the maximum capacity. The same analysis is done at the other three selected days, were these different moments of capacity utilization are determined (appendix H).
ensures that it is not possible or harder to focus on patients’ arrival with more urgent attention needed.
4.3.5 Gantt chart analysis
The high utilization rates of (treatment) rooms at the four selected days are analyzed in more detail in order to indicate the effect of the relatively high utilization of treatment rooms with non-urgent patients on the room allocation process at the ED. In figure 4.11 a Gantt chart of the rooms is shown, whereas the x-ray and the waiting room are included as patients have to go there during treatment. An enlarged, and therefore better quality chart, of 29-09-2015 can be found in appendix I, were the other three charts are also shown.
Fig 4.11- Gantt chart room utilization 29-09-2015
5. CONCLUSION & DISCUSSION
This research paper focused on the impact of the ED dynamics on patients’ throughput times. The responsiveness of the ED was analyzed in order to measure the relationship with overcrowding. Next, the room capacity and utilization of the ED was analyzed in order to investigate the interaction and opportunities to improve patient throughput. The main findings of this research are presented and discussed in this section.
5.1 The responsiveness of the emergency department
The overall input-output analysis showed that the ED appears to have an inadequate reaction on an increase of input in the morning. The output cannot keep up with the increase of input in the morning, which causes increases in patient’s length of stay at the ED. The comparison with previous research conducted in the MCL, indicated that the ED of the Martini hospital has been able to reach an output rate equal to the input rate measured at an earlier moment in time. This difference is mainly attributed to the higher amount of staff capacity assigned in the morning. In addition, a higher amount of patients’ arrival is reached in the morning, which ensures the requirement to achieve a higher output rate at an earlier moment in time. In detailed analyses of days with on average high work-in-process is showed that these days are mainly characterized by high inflow rates in the morning, and relatively consequent and high output rates during a considerable part of the day. The high peaks in inflow rates ensures that the capacity is insufficient, which causes an increase in the average work-in-process during these days. This indicates that these days are not specifically influenced by an inadequate reaction on an increase of input during the morning, due to the unfeasibility of coping with these situations. One of the reasons that the ED is not able to have an adequate response on an increase of input in general is that staff is not aware of the importance of creating output in the morning. Furthermore, the capacity of the staff at the ED is planned in such a way that the highest amount of staff is working in the end of the afternoon. In the morning the same arrival rate as in the end of the afternoon is already reached, while a significant less amount of staff is assigned. A more adequate responsiveness to an increase of input in the morning will consequently lead to less queueing of patients during the day. The patients who enter the ED in the morning will have to wait less time to be seen by a (assistant) physician. Consequently, the patients will leave the ED earlier, allowing incoming patients to be treated at an earlier moment in time. This will prevent patients from stacking up at the ED, and ensures that crowding occurs at a later moment in time or is completely avoided. In this way the throughput times of patients could potentially be improved and additionally the ED can cope with higher amounts of input.
5.2 The interaction between room utilization and throughput times
morning due to patient with non-urgent triage codes. These high utilization rates are causing a delay in start of treatment of the most urgent patients arriving at the ED. The set timeframes of the triage codes are subsequently exceeded, which implies that the room capacity serves as a bottleneck when too many rooms are occupied and therefore causes a lack of focus on patients with most urgent triage codes. The room capacity is functioning as a bottleneck because it is preferred to generate other output, while this is not possible since all the rooms are occupied. In addition, these findings suggest that deliberately not all the (treatment) rooms should be taken into use in order to ensure on-time treatment when more urgent patients arrive at the ED. Furthermore, the detailed analyses of room utilization throughout the selected days implicated an inefficient use of (treatment) rooms at the ED. The ED is focused on freeing up treatment rooms, however these rooms are not always used in the meantime while patients are in the waiting room or x-ray while being treated. A more efficient use of available room capacity at the emergency department ensures a decrease in patients waiting time before start of treatment. This will avoid the work-in-process from building up, and eventually decreases patient throughput time. Additionally, the ED creates more available and accessible capacity, which could potentially support the on-time treatment of urgent patients.
In conclusion, the main causes of increased throughput times of patient at the ED are the inadequate reaction on an increase of input, and the relatively inefficient assignment of available room capacity.
6. RECOMMENDATIONS
This section will describe various recommendations to the ED of the Martini hospital in order to decrease patient throughput times. Next, there will be various recommendations for further research.
6.1 Recommendations for the emergency department of the Martini hospital
The findings suggest that an adequate response to arrival rate increases could potentially reduce patient throughput times and therefore dynamics within the day can provide starting points of understanding and considering patients waiting times. Furthermore, it is suggested that assigning patients to (treatment) rooms in a more controlled way could potentially improve the quality and speed of care for more urgent patients and the efficiency of the available room capacity.
An increased responsiveness of the ED could be created by assigning more staff capacity in the morning and creating awareness of the importance to create output in the morning. In the current situation, there are six nurses working between 09:30 and 12:00 hours, including one nurse solely focusing on the triage process, one nurse solely focusing on the coordination and one nurse solely focusing on pediatrics. This implies that only three general nurses have to deal with the treatment of a patients arrival rate of almost six per hour during those moments. It is recommended that the two nurses, who are beginning their shifts at 12:00 hours, are starting at 09:00 hours in order to create a more adequate reaction in the morning. This shift is then working until 17:30, and still five nurses are working after this time. Next, flexible working shifts could optionally be used in order to deal better with a sudden increase of input overtime. This allows the ED to increase additional capacity when necessary, which avoids the work-in-process from building up. The staff at the ED additionally has to realize how important it is to create output in the morning. There is time spend at additional activities during the morning because no pressure of creating output is experienced. The awareness of creating output in the morning will ensure an increased outflow rate in the morning, and subsequently prevents patients from stacking up at the ED.
Furthermore, it is recommended that the care coordinator uses the triage codes more as a guideline in assigning (treatment) rooms to patients. The triage codes of patients are currently not taken into consideration when there is still sufficient capacity to treat patients. This ensures that the room capacity is acting as a bottleneck when high utilization rates with less urgent patients are causing a delay in treatment of incoming urgent patients. Therefore it is also recommended that not all the treatment rooms are taken into use when capacity is reaching its maximum. In addition, a certain limit of room utilization must be taken into account in order to ensure that no capacity is lost. The remaining room capacity could be used for new incoming patients with more urgent triage codes, which ensures that these patients are assigned to a treatment room more rapidly. In this way the ED is able to focus on treatment of patients who need the most urgent and immediate care.
several tasks besides this such as answering the phone and assisting in treatments. The additional tasks ensure that at moments of crowding there is no clear overview of the room utilization. This results in the inefficient use of treatment rooms during moments of high work-in-process. The assumption is made that when the coordinator is only focusing on the assignment of (treatment) rooms it is noticed more quickly when room capacity is available. As a result, patients have to wait less time before start of treatment due to a more efficient and effective use of capacity.
6.2 Recommendations for further research
Further research could potentially investigate the consequences of deliberately not taking all the treatment rooms into use when maximum capacity is almost reached. The interesting aspect is related to the effects on patients with non-urgent triage codes, and the performance relating to the care of most urgent patients. In addition, further research can focus on which limit must be considered in assigning treatment rooms, as a significant loss of capacity should be avoided.
Furthermore, the data regarding the moments in time were the patients were seen by a physician and/or nurses were missing in the dataset of the ED. It would be interesting to investigate if and how long patients are waiting inside the treatment rooms in order to estimate the efficiency of the room utilization rates even further.
Finally, the days that were selected for detailed analyses are based on the average work-in-process. In this way it was not possible to filter any days with a single short and intense peak in work-in-process. It would be interesting to see what happens with the room utilization during these days and to indicate how this further affects the responsiveness to deal with the dynamics of the ED.
REFERENCES
American College of Emergency Physicians. (2006). Crowding. Annals of Emergency
Medicine, 47:585.
Asplin, B.R., Flottemesch, T.J., & Gordon, B.D. (2006). Developing models for patient flow and daily surge capacity research. Academic emergency medicine: official journal of the
Society for Academic Emergency Medicine, 13(11): 1109-1113.
Asplin, B.R., Magid, D.J., Rhodes, K.V., Solberg, L.I., Lurie, N., & Camargo, C.A. (2003). A Conceptual Model of Emergency Department Crowding. Annals of Emergency Medicine, 42(2): 173-180.
Arkun, A., Briggs, W., & Patel, S. (2010). Emergency department crowding: factors influencing flow. Western Journal of Emergency Medicine, 11(1): 10–15.
Asaro, P.V., Lewis, L.M., & Boxerman, S.B. (2008). The Impact of Input and Output Factors on Emergency Department Throughput. Academic Emergency Medicine, 14: 235-242.
CBS Site Redirect — Centraal Bureau voor de Statistieken. 2015. [ONLINE] Available at: http://www.cbs.nl/enGB/menu/themas/gezondheidwelzijn/publicaties/artikelen/archief/2015/z orguitgaven-stijgen-met-18-procent-in-2014.htm. [Accessed 17 September 2015].
Derlet, R. W., & Richards, J.R. (2000). Overcrowding in the nation’s emergency departments: complex causes and disturbing effects. Annals of Emergency Medicine, 35(1): 63-68.
Ekelund, U., Kurland, L., Eklund, F., Torkki, P., Letterstal, A., Lindmarker, P., & Castrén, M. (2011). Patient throughput times and inflow patterns in Swedish emergency departments. A basis for ANSWER, A National Swedish Emergency Registry. Scandinavian journal of
trauma, resuscitation and emergency medicine, 19(1): 37.
El-rafi, O., Garaix, T., Augusto, V., & Xie, X. (2015). A stochastic optimization model for shift scheduling in emergency departments. Health Care Management Science, 18(3): 289-302.
FitzGerald, G., Jelinek, G.A., Scott, D., & Gerdtz, M.F. (2010). Emergency department triage revisited. Emergency Medicine Journal, 27: 86-92.
Hoot, N. R., & Aronsky, D. (2008). Systematic Review of Emergency Department Crowding: Causes, Effects, and Solutions. Annals of Emergency Medicine, 52(2): 126-136.
Jack, E.P., & Powers, T.L. (2009). A review and synthesis of demand management, capacity management and performance in health-care services. International Journal of Management
Reviews, 11(2): 149-174.
Kaushal, A., Zhao, Y., Peng, Q., Strome, T., Zhang, M., & Chochinov, A. (2015). Evaluation of fast track strategies using agent-based simulation modeling to reduce waiting time in a hospital emergency department. Socio-Economic Planning Sciences, 50:18-31.
Kheirbek, R.E., Beygi, S. Zarhoush, M., Alemi, F., Smith, A.W., Fletcher, R.D., Seton, P.N., & Hawkins, B.A. (2015). Causal Analysis of Emergency Department Delays. Quality
Management in Health Care, 24(3): 162-166.
McCarthy, M.L., Zeger L.Z., Ding, R., Levin, S.R., Desmond, J.S., Lee, J., & Aronsky, D. (2009). Crowding Delays Treatment and Lengthens Emergency Department Length of Stay, Even Among High-Acuity Patients.
McCaughey, D., Erwin, C.O., & DelliFraine, J.L. (2015). Improving Capacity Management in the Emergency Department: A Review of the Literature, 2000-2012. Journal of Healthcare
Management, 60(1): 63-75.
NU Site Redirect — 2015. [ONLINE] Available at: http://www.nu.nl/binnenland/3509496/kle ine-ziekenhuizen-in-financiele-problemen.html. [Accessed 10 October 2015].
Richardson, D.B. (2006). Increase in patient mortality at 10 days associated with emergency department overcrowding. Medical Journal of Australia, 184(5): 213-216.
Ridge, J. C. (1998). Capacity planning for intensive care units. European Journal of
Operational Research, 105: 346-356.
Soepenberg, G.D., Land. M.J., & Gaalman, G.J.C. (2012). A framework for diagnosing the delivery reliability performance of make-to-order companies. International Journal of
Production Research, 50(19): 5941-5507.
Solberg, L. I., Asplin, B. R., Weinick, R. M., & Magid, D. J. (2003). Emergency department crowding: consensus development of potential measures. Annals of emergency medicine, 42(6): 824–834.
Terwiesch, C., Diwas, K.C., Kahn, J.M. Working with capacity limitations: operations management in critical care. Annals of Emergency Medicine, 15:308.
Trzeciak, S., & Rivers, E.P. (2003). Emergency department overcrowding in the United States: an emerging threat to patient safety and public health. Emergency Medicine Journal, 20: 402-405
Van Achteren, G. (2014). Care for waiting: the effect of emergency department responsiveness on patient waiting times. Master Thesis. University of Groningen, NL
Van der Linden, C., Reijnen, R., Derler, R.W., Lindenboom, R., van der Linden, N., Lucas, C., & Richards, J.R. (2013). Emergency department crowding in The Netherlands: managers’ experiences. International Journal of Emergency Medicine, 6: 41.
Van der Vaart, T., Vastag, G., & Wijngaard, J. (2011). Facets of operational performance in an emergency room (ER). International Journal of Production Economics, 133(1): 201-211. Van Veen, M., Steyerberg, E.W., Ruige, M., van Meurs, A.H.J., Roukema, J., van der Lei, J., & Moll, H.A. (2008). Manchester triage system in paediatric emergency care: prospective observational study. British Medical Journal, 337.
Verbree, J.M. (2013). Diagnosing and improving the patient throughput time in emergency departments. Master Thesis. University of Groningen, NL.
Voss, C. Tsikriktsis, N., & Frohlich, M. (2002). Case research: case research in operations management. International Journal of Operations and Production Management, 22(2): 195-219.
APPENDIX A2 – Origin of patients
Origin of patients # Patients Percentage
General practitioner 9892 48,1%
112 3771 18,3%
General practitioner post 2817 13,7%
Other 1667 8,1% Rontgen 1182 5,7% Own initiative 1131 5,5% UMCG 60 0,3% External hospital 40 0,2% APPENDIX A3 – Specialisms
Specialism # Patients Percentage
Surgery 8334 41% Cardiology 2994 15% Internal 2851 14% Pediatrics 1473 7% Lung 1401 7% Orthopedics 1031 5% Neurology 1012 5% Other 1464 7%
APPENDIX A4 – Amount of patients per day
Days # Patients Median Percentile 0,75 Percentile 0,9 Percentile 0,95
APPENDIX C1 – Average input-output Friday
APPENDIX D1 – Patients throughput times
Days Median Percentile 0,75 Percentile 0,9 Percentile 0,95
Monday 02:10:00 03:01:00 04:04:00 04:43:00 Tuesday 02:10:00 03:00:00 04:05:00 04:42:00 Wednesday 02:16:00 03:08:00 04:06:00 04:52:00 Thursday 02:15:00 03:05:00 04:06:00 04:42:00 Friday 02:15:00 03:11:00 04:05:00 04:45:00 Saturday 02:06:00 03:00:00 03:52:24 04:24:00 Sunday 02:01:00 02:45:00 03:39:00 04:18:00
APPENDIX D2 – Time until start treatment after patients arrival
Days Median Percentile 0,75 Percentile 0,9 Percentile 0,95
0 10 20 30 40 50 60 70 80 90 100 110 Cu mu la (ve n umb er o f p a(en ts Time arrival triage start treatment departure
APPENDIX E2 – Throughput diagram 25-09-2015
APPENDIX F3 – Work-in-process 30-10-2015
APPENDIX H1 – Work-in-process rooms 04-05-2015
APPENDIX I4 – Gantt chart 30-10-2015
