“How can the throughput time of patients in the Emergency Department be improved?”

(1)

“How can the throughput time of patients

in the Emergency Department be

improved?”

Tom Dijk

University of Groningen

Faculty of Economics and Business

(2)

2

4.2 Step 1: The selected days 14 4.3 Analyses of two comparable days 15 Monday 28-05-2012 versus Monday 16-07-2012 15 4.4 Analyses of the days 20 4.4.1 Step 2: Differences among order subsets 20 4.4.2 Step 3: Time dependency 21 4.4.3 Detail analyses 23 5. Conclusion 28 6. Recommendation 29 6.1 Suggestions 29 6.2 Limitations 29 6.3 Further research 29 References 30 Appendix 34

(3)

How can the throughput time of patients in the Emergency

Department be improved?

Tom Dijk

Master thesis, MSc BA, specialization Operations and Supply Chain Management University of Groningen, Faculty of Economics and Business

15 April 2013

Abstract

The focus of this research is to improve the throughput time at the Emergency Department. The diagnostic framework of Soepenberg et al. (2012) is used for structuring the analyses. Based on the analyses the responsiveness is the most important factor that determines if the throughput time will increase when there is a peak in arrival during one or more consecutive hours especially when urgent and more complex patients involved. Extra personnel available when needed should help to cope with the increase in demand during a certain time window and make the ED more responsive.

Keywords: throughput time, emergency care, hospital, throughput diagram, patient flow

1. Introduction

Over the past several years, changing social, economic, and public health forces have significantly changed the medical care. The increase in demand for medical care and the closure of facilities put pressure on the capacity of hospitals (Solberg et al., 2003). Therefore, hospitals have become

increasingly interested in minimizing patient throughput time to improve efficiency (Griffin et al., 2012; Rathlev et al., 2007). This thesis is about the patient throughput times and then in particular at the Emergency Department (ED). The investigation takes place in a large hospital that plays an important role in the regional care function in the northern part of the Netherlands. The ED is the most vital part of a hospital and therefore an important subject for investigation. Throughput time is defined as: “The time between the patient being enrolled at the ED and the time the patient leaves the department (Larsson et al., 2012).”

The ED has to be available to receive patients in need of treatment at any time, however the great deal of uncertainty regarding timing and volume of patients’ arrival and the scarce resources put pressure to patients for receiving timely emergency care (Asplin et al., 2003; Hwang et al., 2011; McNaughton et al., 2012; Van der Vaart et al., 2011). This results in crowding at the ED, when temporary demand for services outstrips available resources (Asplin et al., 2003; Hwang et al., 2011; Solberg et al., 2003). Overcrowded EDs have several negative consequences, prolonged waiting times, dissatisfied patients and ED personnel, diversion of ambulances (ambulances are send to other nearby hospitals), greater risks for poor outcomes, unnecessarily high costs, and several side effects, like more violence from patients, miscommunication and decreasing productivity of personnel (Cassidy-Smith et al., 2007; Derlet and Richards, 2000; Hoot and Aronsky, 2008; Solberg et al., 2003; Yoon et al., 2003).

(4)

4

measures for monitoring and managing ED crowding (Hwang et al., 2011; Solberg et al., 2003), improvements to reduce crowding (Joy et al., 2010; Olshaker and Rathlev, 2006), and also to

throughput capability as the key to overcome crowding (Bertoty et al., 2007; Byers, 2011; Scalise, 2006; White et al., 2012).

Soepenberg et al. (2012) developed a diagnostic framework that is based on general problem-solving literature and domain-specific knowledge. The framework structures the diagnosis process, enabling one to navigate from the achieved performance to the underlying causes related to production

planning and control (PPC). The framework is developed for make-to-order (MTO) companies, however previous research shows that methods from operation management for traditional manufacturing processes can also be useful in healthcare (Terwiesch et al., 2011). This framework will be used to structure the diagnosis.

The main aim of this research is to find the underlying causes of patients’ delays at the ED in a large hospital in the northern part of the Netherlands with the help of the framework of Soepenberg et al. (2012) to provide possibilities for improvement.

This thesis starts with a theoretical background which discusses the main variables that influence the patients’ delays in the ED. This is followed by the methodology part in which the research design will be explained. After this the analysis of the results will be discussed. The thesis will end with a conclusion of the results and the recommendation.

2. Theoretical background

The theoretical background will discuss the variables that influence the throughput time of patients in the ED. Bertoty et al. (2007), Byers (2011), Scalise (2006) and White et al. (2012) already saw that improving your throughput time is the key to reducing patients’ delays in the ED. Therefore, the variables here are all factors influencing the throughput time of patients. The variables will each be discussed profoundly. This chapter concludes with a conceptual model and research question.

2.1 Demand for emergency care

Crowding of the ED has become a major health care problem and a public issue throughout the past decade. One of the key expectations of EDs is the ability to provide immediate access and stabilization of patients who have an emergency medical condition. The timely access to an emergency provider is a critical dimension of quality for ED´s. However, overcrowding diminishes the capability of the ED to manage the emergencies effectively. Longer waiting times are a result of overcrowding which causes prolonged pain and suffering for patients and patients are facing a greater risk of poor outcomes (Derlet and Richards, 2000) which even results in a higher patient mortality (Begley et al., 2004; Richardson, 2006; Sprivulis et al., 2006).

(5)

Non-urgent visits and frequent-flyer patients (patients that visit the ED four or more times a year) are part of the demand that put pressure on the emergency care (Hoot and Aronsky, 2008). However, there isn´t consensus on the extent of these factors on crowding. The majority of the crowding resulting from the demand factor can be explained by the uncertainty in timing and volume of patients’ arrival (Asplin et al., 2003; Hwang et al., 2011; McNaughton et al., 2012; Van der Vaart et al., 2011). This is not only caused by the high variability in demand during a 24-hour period but also by non predictable

circumstances, for example multiple casualty incidents and influenza outbreaks, that result in a sudden increase of patients’ arrival at the ED.

Patient demand is not only uncertain in the timing and volume of patients’ arrival but also in the urgency and the seriousness of the complaint. Patients presenting in the ED show a wide range of problems, from life-threatening conditions to minor injury or illness. Because of the overcrowding, the severely ill patients compete with the less ill patients for the available resources. This increases the demand for a more systemic working in the ED. Therefore, the Manchester Triage Group developed the Manchester Triage System (MTS) to prioritize the patients in an ethic way (Mackway-Jones, 1997).

2.2 Triage process

Triage means: “a brief clinical assessment that determines the time and sequence in which patients should be seen in the ED (Van der Vaart et al, 2011).” The triage system is thus a method to assess patients’ severity of injury or illness within a short time after their arrival, assign priorities and transfer each patient to the appropriate place for treatment (Fernandes et al., 2005). Triage is a dynamic process as the patient’s condition may change rapidly (Robertson-Steel, 2006).

The MTS is an algorithm consisting of 52 flowcharts, each chart representing a specific complaint which can be subdivided into five categories: illness, lesion, children, abnormal behavior and major incidents. The flowcharts describes six key discriminators – life threatening, haemorrhage, pain, conscious level, temperature and acuteness – as well as specific discriminators relevant to the presented complaint by which it is possible to assess the acuity of the patient’s problem. Five categories can be distinguished: red (immediate action required), orange (can wait 10 minutes), yellow (can wait 60 minutes), green (can wait 2 hours) and blue (can wait 4 hours) according to the Manchester Triage Group.

The reliability and validity of the MTS has been tested by several researchers and the overall conclusion is that the system possesses satisfactory to very good validity and reliability (Christ et al., 2010;

FitzGerald et al., 2010; Van Veen et al., 2008; Van der Wulp et al., 2008).

However, the problem with triage, as a prioritization tool, is that urgent patients will always override the less urgent ones even though there are time limits for all triage classes (Van der Vaart, 2011). Another aspect that influences the waiting time of patients at a particular moment in time are the available resources. Resources can be divided here in two parts: personnel and treatment rooms. This is the subject of the next paragraph.

2.3 ED Capacity

(6)

6

resources, physical space, equipment, personnel and organizational system of the ED to meet the patients demand (Bellow et al., 2012).

When the demand outstrips the available capacity there is mostly a shortage of physical space and/or personnel (Kellermann, 2006; Yen and Gorelick, 2007) and therefore the focus is on these two capacity resources. With the physical space researches mostly referred to the emergency rooms. Emergency rooms can be highly customizable by equipment. There are emergency rooms specialized with

expensive equipment (trauma room with X-ray, for example) or with specific supplies (casting, burns). On average the capacity outstrips the demand, however, due to the uncertainty in the timing and volume of patients’ arrivals the demand exceed the available capacity at particularly moments in time. Not only because the extreme variability in demand during each 24-hour period experienced by a typical ED but also because of accidents or multiple casualty incidents or an outbreak of influenza. Most analyses about capacity shortage are about staffing, the inadequacy of staffing (Lambe et al., 2003; Schneider et al, 2003; Schull et al., 2003), the impact of nurse staffing ratio legislation on hospitals and the quality of patient care (Chapman et al., 2009). Others research ties to develop methods to balance the personnel quantities over time (Ahmed and Alkhamis, 2009; Green et al., 2006; Vassilacopoulos, 1985).

Trying to match capacity levels to accommodate changing demand levels is a great challenge in EDs even in the case of constant patients demand levels over the day. Fluctuations in individual patient arrival times and the variability in the time needed by a nurse to treat patients can create long delays. Even when the overall average capacity is greater than average demand during the day which is mostly the case in EDs. In an environment like this, with time-varying demand, delays are likely to be even greater, if capacity is not carefully adjusted based on the actual fluctuations of the arrival rate over the day (Green et al., 2006).

2.4 Conceptual model and research question

The conceptual model is visible in Figure 1. The model is separated into two control decisions:

input and output. The input of the ED depends on the patient demand and the triage decision. The output of the ED depends on the available capacity of emergency rooms and personnel. Both the input and output influence the throughput time of patients.

Improving your throughput time as the key to overcoming crowding is well recognized in literature and is the underlying research question for this master thesis. This thesis is a problem solving research in

Input Output Patient demand • Timing • Volume Throughput time Availability of rooms Availability of personnel Triage decision • Urgency • Complexity

(7)

which the focus lies on one specific case; however, we might expect to derive some issues which will be of interest for many emergency departments applying MTS.

Research question

“How can the throughput time of patients in the Emergency Department be improved?”

3. Methodology

This research focuses on the ED of a large hospital in the northern part of the Netherlands. In

comparison with the ED here, a lot of other EDs use the MTS to prioritize patients as well and also have problems with the uncertainty in demand and therefore the results of this thesis may be applicable for other EDs. Paragraph 3.1 describes the specific ED that is used in this research. For diagnosing the underlying causes of patients’ delays the diagnostic framework of Soepenberg et al. (2012) is used. Paragraph 3.2 describes the framework. The last paragraph discusses how the data are gathered.

3.1 The ED

This research is conducted at the ED. The ED that is used here is a large, regional teaching hospital in the northern part of the Netherlands. The most vital part of the hospital is the ED on which the research focuses. The ED provides acute care to patients who arrive without prior appointment, either by own

(8)

8

means or by ambulance or helicopter, and patients who are referred to the ED by the general practitioner or specialist.

Figure 2 shows the layout of the ED. At the moment patients arrive at the ED, they are registered at the reception desk and their identifiers are entered in the computer system. Only extreme cases will be registered later because their condition is life-threatening. From the reception desk patients proceed to the waiting room where they wait for the triage nurse. The triage nurse is trained to assign the right urgency code to the patient, from red to blue, where red means direct help is required according to the MTS and blue means that the patient is non-urgent and can wait for maximum four hours. Depending on the patient’s condition, the code that is assigned and depending on the load in the ED, the patient is either sent back to the waiting room or transferred to the ED immediately for treatment. After the treatment the patient is dismissed from the ED and sent to another department in the hospital or sent home, depending on the condition of the patient.

Several patient data are obtained during the patient’s visit (i.e. reason for visit, the identifier of the patient, the potential cause of the visit, patient’s moment of arrival, triage code and other). The secretary that assists the nurses, residents and specialists creates the patient files in the registration system and the nurses, residents and specialists are responsible for the registration during the process. The data from the registration system will be used for this research, see paragraph 3.3.

3.2 Diagnostic framework

Terwiesch et al. (2011) shows that methods from operations management for traditional manufacturing processes can also be useful in healthcare. The diagnosing framework by Soepenberg et al. (2012) is developed for MTO companies and will be used here in a health care setting. However, there are some minor and major adjustments made to use the framework. Not only because the setting is different here but also because the limitations in the available data (see also paragraph 3.3). The adjustment made to the framework is mentioned in this paragraph per step of the framework.

The diagnosing framework consists of two layers, the upper layer enables us to gain a comprehensive overview and the lower layer carries out a more-detailed diagnosis (see appendix A, figure 11, for the upper layer of the framework). The first part of the framework is a breadth search strategy that enables one to gain a comprehensive overview of the performance in the ED and avoids getting too involved in the detail in an early phase of the diagnosis process (Wagner, 1993).

The first part of the framework consists of four steps. The first three steps question whether causes of bad performance must be sought in the direction of average lateness or variance of lateness. Based on this result you will continue in one of the two directions. The fourth step determines in which process the diagnosis must continue for the more detailed analysis.

Step 1: Analyse the distribution of lateness

Following the framework the diagnosis starts by determining the percentage of patients being late in a certain period. A distribution of lateness is constructed to discover if there is a high average and/or high variance of lateness.

(9)

only the moment until a specialist sees the patient. How to cope with the process after the contact with the specialist? Also the moment that the specialist interferes is not well listed and therefore hardly useable. So, the exact promised delivery time is missing.

Second, useful conclusions from the lateness distribution can only be drawn for relative homogeneous order (here: patient) populations. In the ED there is a great variety of patients, for example, there are 184 different codes for labeling the patient problems. These codes can be divided into 17 different groups, however, there are still major differences within groups, for example, in the group that includes the elbow/forearm problems there is a major difference between a radius fracture with reposition or without a reposition. Therefore, it is not possible to make a decent amount of homogeneous groups. Therefore, the distribution of lateness is not constructed here but with the help of the average

throughput time per day, the days with a longer throughput time than the overall average are depicted. Also the days with a shorter throughput time will be taken out to be able to compare both.

Step 2: Analyse differences among order subsets

In order to narrow down the later analysis one should check if a limited number of patients subsets are responsible for the high average and/or high variance of lateness. As mentioned in step 1, the days with on average a longer throughput time are interesting. Most of these days have a longer throughput time during a certain time window. However, we do not make a distinction here between a high average and/or high variance of lateness.

With the available data two difference subsets among patients can be distinguished. First, patients are divided into different codes based on the MTS. Second, patients are divided over the diagnostic codes they get, which refers to their problems.

With the codes from the MTS you get five groups (red, orange, yellow, green and blue) that also refer to the urgency of patients. With the codes referring to the problems of patients you can distract ‘fast’ track patients against ‘slow’ track patients (Table 1: ‘fast’ track versus ‘slow’ track patients” gives an overview). The average throughput time for all patients is 1:57:48 (hours:minutes:seconds). All the ‘fast’ track patients use on average 80% or less of the average overall time. Some categories have an

(10)

10

Step 3: Analyse differences over time

The third step is about the time dependency. The diagnosis continues by analyzing how lateness changes over time. In this research step 3 is used to see if certain moments in time over a day causes the longer throughput times. The throughput diagram is a useful tool in this step.

A throughput diagram is a graph that compares the cumulative input against the cumulative output in a certain time window. The horizontal axis in a throughput diagram shows the time during a day (here) and the vertical axis the amount of patients that arrive or depart during the day. The horizontal distance between the input and output curves relate to average throughput time and the vertical distance depicts the work-in-progress (WIP) on a particular moment in time (see figure 5).

Step 4: Analyse delivery time promising (DP) and realization process (RP)

The fourth step determines whether the diagnosis should continue in the process of delivery time promising and/or in the realization process. The time that is promised to patients is based on the triage process and is fixed, therefore the detail investigation of the causes of longer throughput time

continuous here in the realization process.

The detailed analyses

After the four steps the diagnosis continues to look into more detail. Because the available data (see also paragraph 3.3) do not include much detail information about the steps in the ED between arrival and departure the lower layer of the framework is not used. The detailed analysis continues here to look into aspects of individual patients and the personnel of the ED is ask to see if there were done things different on certain days that could explain the on average shorter or longer throughput times.

3.3 Data collection

For the diagnosis quantitative data are needed. The data are collected from the registration system at the ED. The secretary makes the patient files in the registration system for all the incoming patients at the ED. All the proceedings done by the nurses, residents and specialist are recorded in the same file by themselves. The data are gathered for a period of one year, from 1 November 2011 until the 31October 2012.

The data consist of several time moments: patient arrival, triage moment, the moment of treatment by a nurse, the first contact of the physician assistant, the time a patient is ready for departure and the actual departure time of the patient. Also the time a scan or photo is made for the patient is included. Other information that is recorded is: age, the transport mode to the ED, the reason for the visit, the triage code, the final diagnosis and the specialism for which the patients are eligible, the name of the practitioner (nurse and/or physician assistant), the destination after the ED and the emergency room where the patient is treated.

(11)

The arrival time and departure time are the most reliable and complete time data. With the other time moments, triage, treatment by a nurse, the first contact with the physician assistant and the time a patient is ready for departure, they are less reliable. For example, a patient that is more than a week at the ED according to the data or patients who depart before they have arrived isn’t possible. During busy hours at the ED the registration by the nurses and residents are done afterwards and based on their own estimations. It also happens that data is not registered at all. Therefore, these time moments are not used as a basis for the analyses.

The data are filtered based on incomplete records. However, records can’t be deleted based on one missing value. There is also the risk that a deleted record will disturb the other results on a certain moment. Therefore, the days of interest that have incomplete records will separately be judged, if the incomplete record can be deleted or have to be restored, also depending on the kind of analysis that will be made.

(12)

12

4. Results

In the last part of chapter three the datasets are discussed. This chapter will discuss the results that are found after analyzing the data. This chapter starts with some general comments about the data. The second paragraph pays attention to the selected days and the last paragraph discusses the selected days in more depth.

4.1 In general

As mentioned before the dataset consist of 366 days. The dataset represents one year from the 1 November 2011 till the 31October 2012. In total 23,624 patients have been treated in the Emergency Department. The average age of the patients is 50.1 years. Figure 3 shows the distribution.

1

The age of the patients varied between 0 and 100 years, 50 percent of the patients are between 29 and 70 years old. It is interesting to notice the double mode of the distribution. Main peaks occur both

around the age of 20 and the age of 65. The specialisms that see the most patients are surgery (38%), internal medicine (16%), cardiological (15%), lungs (7%), neurological, orthopedic and treatment by nurse (6%), see Table 3.

Looking to the arrival pattern during the week (Table 2), the Monday and Friday are the busiest days of the week and in the weekend it is a little bit calmer. However, the differences are not very large. The overall average of number of patients arriving per day is 64.54. More interesting is the arrival pattern during the day. Figure 4 shows the average arrival and departure of patients during an hour a day. Between 09.00-19.00 o’clock 69.7% of the patients visit the ED, before 09.00 o’clock this is 12.2% and after 19.00 o’clock this is 18.1%.

Figure 3: Distribution of the age

(13)

When comparing the input (arrival) versus the output (departure), the average amount of patients during an hour of the day that leaves the ED lies in the beginning above the arrival line because there

are some patients who arrive just before midnight and will depart after midnight and therefore more patients are leaving around 01.00 o’clock than arriving. During the day the departure line stays behind the arrival line, where you see a nod around 11.00-12.00 o’clock in the departure line. It seems that the output cannot follow the input. However, later in the afternoon the output curve reaches a higher peak than the arrival curve.

The moment of departure of patients from the ED depends on a lot of factors such as the moment of arrival, the amount of patients that are already waiting in the ED but it also depends on the urgency and the complexity of the patients. In Tabl the urgency of the patients is shown by the triage code they get. Red means immediate action is required where blue means the patient can wait four hours. The codes in the middle, yellow and green, represent 78% of the patients. The urgent codes, red and orange, represent 18% of the patients. There are not so many patients with the code blue and patients that get the code blue are mostly send home or to the general practitioner. More than 50% of the patients with code blue is only treated by a nurse and see no physician or specialist.

On the left side in Tabl the categories of specialisms are mentioned. There are 22 categories and if the patients do not fit in one of the categories there is an overall (remaining) category. There are 11 patients without a code and in the remaining category there are only 24 patients.

Most of the patients get code yellow or green, within the categories this is also the case. However, there is one notable exception and that is in the category cardiological. In this category 60% of the patients get an urgent code (red or orange). When looking overall, 50% of the urgent patients fall into the category cardiological.

(14)

14

;

4.2 Step 1: The selected days

For selecting the days (step 1, see section 3.2), the average throughput time from arrival until the moment of departure was crucial. These start and end points were chosen because the data about the moment of arrival and the moment the patient leaves the ED were the most reliable and complete time data, as mentioned in chapter 3.

The average throughput time from arrival until departure for the 366 days is 1:57:48

(hours:minutes:seconds). The day with the longest throughput time is 11 July 2012, 2:39:29. The day with the shortest throughput time is 28 October 2012, 1:21:46. For the detailed analyses 10% of the days was used, 5% with a relative short throughput time and 5% with a relative long throughput time. Table 4 shows the chosen days with the throughput times, the amount of patients and the average age. The overall average age is 50.1 years, the days with a long throughput time have almost the same average age with 51 years and the days with a short throughput time have an average just below the overall average. This suggest that the age of treated patients doesn’t provide a main explanation for longer or shorter average throughput times on specific days. The average amount of patients arriving on the days with a long throughput time is 71.8 patients, a little bit higher than the overall average of 64.5 patients. 11 February 2012 was an extremely busy day. The average for the days with a short

(15)

throughput time is 58.8 patients. Though differences occur, the difference in the average number of arrivals between days with long and short throughput time is not very large. On the other hand

differences in arrivals across the days within each group seem relatively high. This suggests that also the number of arrivals is not the main explaining factor for the longer throughput times. The difference between the average throughput time is more than 50 minutes and that is relatively long. The next paragraph will go further into the difference between the days with long and short throughput times.

4.3 Analyses of two comparable days

In this part a day with a short throughput time will be compared with a day with a long throughput time. The days are compared based on the arrival pattern, urgency, ‘fast’ track versus ‘slow’ track and also on a detailed level. In the next section an explicit example of how the graphs and diagrams are used is explained, the other graphs and diagrams are put in appendix B and C. In paragraph 4.4 the other days are discussed.

Monday 28-05-2012 versus Monday 16-07-2012

As just mentioned in the title of the paragraph two Mondays will be compared. The first one, 28-05-2012 has an average throughput time of 1:36:14 (hours:minutes:seconds) with 66 patients overall (see also Table 4). The other day, 16-07-2012 has an average throughput time of 2:36:58 with 64 patients overall (see Table 4). Both days have almost the same amount of patients arriving, however, the throughput time differs more than one hour.

(16)

16

First the arrival and departure pattern of both days are compared. Figure 5 shows Monday 28-05-2012 and Figure 6 shows Monday 16-07-2012. The horizontal distance between the curves is the

Figure 5: Throughput diagram Monday 28-05-2012

(17)

throughput time and the vertical distance is the work-in-progress (WIP) on a particular moment in time. In Figure 5 the departure line (red) lies much closer to the arrival line (blue) both in horizontal distance as in vertical distance than in Figure 6. This means that on average the WIP is lower and the throughput time is shorter on the 28 May than on the 16 July. Looking at the arrival pattern both days show hours with more arrival (see Table 5). As mentioned the WIP is lower for the 28th than for the 16th, Table 5 shows that for the 16th the WIP is constant above ten patients from 11.00 to 22.00 o’clock. More WIP causes longer throughput times because patients have to wait for nurses, residents and available rooms. However, the 28th shows also hours with more arrival of patients but the WIP does not increase,

only shortly after 13.00 o’clock. Therefore, the question arises why does the WIP not increase on the 28th_{so much?}

Figure 7 and Figure 8 will help finding the answer. The figures show the WIP divided into the different codes. On the 28th there are less urgent (red and orange) patients than on the 16th. On the 28th_{the urgent patients are around}

noon and in the beginning of the evening. On the 16th the urgent patients are arriving the whole day at the ED. On the 16th there are two red patients that disturb the process extra, both arriving around 16.00 o’clock (15.55 and 16.04 o’clock). Those two patients are

WIP=10 Throughput time=1:15:00

Table 5: The arrival and WIP per hour

(18)

18

acute patients and goes directly to the acute (trauma) room of the hospital (room one and two, see Figure 2). When a patient goes to the acute room two nurses must be available directly to stabilize the patient. Here, two acute patients arriving almost in the same time and therefore, four nurses are needed directly and that causes longer waiting time for the other patients that arrive or that are already waiting around 16.00 o’clock.

On the 28th there are relatively more patients with the code blue and green during the peak moments

Figure 8: Work-in-progress – the urgency - Monday 16-07-2012

(19)

than on the 16th. This doesn’t mean that the throughput time will be shorter if there are more non-urgent patients because this also depends on the problems of the patients, however, non-non-urgent patients don’t disturb the process such like the urgent patients.

Part of the problem is the urgency of the patients. Another factor that influences the throughput time is the problem of the patient, as mentioned before. For example a patient with a wound that needs to be attached has a shorter throughput time than a patient with vague complaints, like pain in the stomach. The patient with the wound can be helped directly and afterwards the patient can go home, however to diagnose the patient with the vague complaints correctly will take more time, the treatment will take

longer and sometimes the patients have to stay in hospital and have to wait before the patient can be transferred to another department. Figure 9 and 10 make a distinction between the ‘fast’ track and ‘slow’ track patients (also mentioned as the ‘long’ respectively the ‘short’).

Blue are the fast track (short throughput time) patients and red are the slow track (long throughput time) patients. During the busy hours on the 28th there are much more ‘fast’ track patients than during the peaks of arrival on the 16th. In total 42% of the patients is fast track on the 28th and for the 16th it is only 19%.

The 16th has longer throughput times than the 28th because there are hours with more arrival of patients in combination with urgent patients that disturb the process. There are more ‘slow’ track patients and therefore the WIP increases and the throughput time is rising and vice versa. This also means for the 16th that patients have to wait longer for triage, especially when acute patients arrive, which results in a crowded and cluttered ED. The busy hours on the 28th correspond with more ‘fast’ track patients that results in a more stable throughput time during the day.

(20)

20

4.4 Analyses of the days

In paragraph 3.2 the steps of the framework of Soepenberg et al. (2012) are explained and how they will be used in this research with some adjustments. Step 1, selecting the days, was already used in paragraph 4.2. In this paragraph the other steps are used. At the end of this chapter there is a summary of all the 36 days that are selected with detail information about the reasons for longer or shorter throughput times (see table 7,8,9 and 10).

4.4.1 Step 2: Differences among patients subsets

In step 2 differences among patients subsets will be analyzed. There are two subsets among patients that are distinguished here, first, patients get different codes based on the MTS and second, patients are divided over the diagnostic codes they get, see also Table 1.

I start with the first one, the codes patients get, in appendix B and C the work-in-progress figures about the urgency are used to see how the urgency is spread over the day in the ED. You can see when and how much patients are in the ED with a particular urgency code. For example, in figure 13, that is about Sunday the 20th of November 2011, there are four patients with code green, two patients with code yellow and one patients with code orange at 12.00 o’clock.

From the data it become clear that patients with a more urgent code will take more time, see Table 6. Patients with the code yellow will take on average the most time, however, the less urgent codes (green and blue) have the shortest time. Especially code blue is very short but that is because a part of this category goes directly to a general practitioner or go home and have a throughput time of only a few minutes.

There are 18 days selected with on average long throughput times, you should aspect that on these days there are more urgent (red and orange) and semi-urgent (yellow) patients than on the 18 days with short throughput times. This is only partly true, for 10 out of 18 days (28-11-2011, 12-1-2011, 03-08-2012, 11-05-2012, 24-08-2012, 03-09-2012, 17-09-03-09-2012, 22-09-03-09-2012, 23-09-2012 and 02-10-2012) with long throughput times this is the case. These dates have during the day or during a part of the day a combination of longer throughput times and more urgent and semi-urgent patients. In the other category (the days with short throughput times) there are four days with more urgent and/or semi-urgent patients, 27-12-2011, 03-03-2012, 28-05-2012 and 28-08-2012. However, on these four days there are more urgent and/or semi-urgent patients but the throughput time will not rise much.

The code patients get influence the patients individually but influence the other patients only when there are more urgent and/or semi-urgent patients in the same time arriving and/or waiting. The acute patients (mostly code red) can disrupt the process more and have a direct negative influence on patients that are waiting or arriving shortly after the acute patient was arrived. Because when an acute patient arrive then the patients will directly go to room one or two (the acute rooms of the ED, see Figure 2) and two nurses must directly go to the patient and other patients have to wait. On the 16th of July, two acute patients arrive around 16.00 o’ clock and thus four nurses were needed and that causes also longer throughput times for the other patients waiting and/or arriving around 16.00. In the evening

(21)

on the 15th of April there was also a acute patients that shortly disrupt the process. The acute patients have a direct negative influence, however, this is only shortly the case. The MTS code patients get influence the patient more individually than that it has an impact on others.

The other subset that I distinguish is the diagnostic code patients get. There is a different between ‘fast’ track (or short) and ‘slow’ track (or long) patients. The diagnostic codes are divided in Table 1. ‘Fast’ track patients have on average a much shorter throughput time than the ‘slow’ track patients. The Figures in appendix B and C, about ‘work-in-progress – ‘short’ versus ‘long’, are used here. With the help of the figures you can see how the patients with their diagnostic code are divided over the day. For example in Figure 71 in appendix C, you can see that just before 20.00 o’clock there are only ‘fast’ (or short) track patients at the ED. This has a positive effect on the throughput time, Figure 69 in appendix C shows around 20.00 o ‘clock and afterwards a relative short throughput time and an extreme short throughput time in comparison to the afternoon of the 11th of February (the particular day that the figure described).

Three of the four days that have more urgent and/or semi-urgent patients have instead of longer throughput times on average much shorter throughput times (03-03-2012, 28-05-2012 and 28-08-2012). The explanation for this is that these days have a lot of ‘fast’ track patients, respectively 39%, 42% and 46%. On the days with long throughput times there are relative less ‘fast’ track patients where 20% or less ‘fast’ track and 80% or more ‘slow’ track is not uncommon.

13 out of 18 days that have on average short throughput times have relative more ‘fast’ track patients (27-11-2011, 31-12-2011, 01-01-2012, 28-01-2012, 12-02-2012, 03-03-2012, 08-04-2012, 15-04-2012, 29-04-2012, 28-05-2012, 09-06-2012, 28-08-2012 and 28-10-2012). On the other side, there are 5 out of 18 days which have longer throughput times, that also have relative more ‘fast’ track patients (11-02-2012, 11-05-(11-02-2012, 29-06-(11-02-2012, 11-07-2012 and 22-09-2012). With the remark that four of these days have 31% or 32% ‘fast’ track patients where percentages of 40% or more ‘fast’ track patients is not uncommon for the days with on average short throughput times. Only the 11th of February have a lot of ‘fast’ track patients, with 54% but still relative long throughput times. The 11th of February is the busiest day in history. There were 134 patients and that is an all time record for the ED, here. Between 09.00-18.00 o’clock six or more patients arrive every hour, with 19 patients between 15.00-16.00 o’clock. Also between 19.00-21.00 o’clock seven patients arrive every hour. Surprisingly there are no patients arriving between 17.43-18.51. This is notable because it is extremely busy all day long, also short before and just after this moment it is extremely busy. However during 68 minutes time there are no patients arriving. There is no explanation for this finding. With 134 patients the ED is overloaded and you should expect extreme long throughput times the whole day but with all the ‘fast’ track patients the average throughput time will not rise above four hours, however still too long.

Relative much ‘fast’ track patients (30% or more) have a positive influence on the average throughput time during the day. The code from the MTS patients get have less influence on the average throughput time over the day.

4.4.2 Step 3: Time dependency

(22)

22

the two discussed in the section before. However, more hours in a row with more arrival will increase the WIP over a longer period and will probably cause longer throughput times around the arrival peaks. 18 out of the 18 days with on average longer throughput times have during the day hours with more arrival. On 10 days there are consecutive busy hours (28-11-2011, 12-01-2012, 11-02-2012, 11-05-2012, 29-06-2012, 03-08-2012, 24-08-2012, 03-09-2012, 17-09-2012 and 04-10-2012), here consecutive busy hours: is a day with three or more hours in a row with an arrival of five or more patients, with one exception: between 16.00-18.00 o’clock there must be six or more patients because then there are also six nurses available. Between 23.30-09.30 there are three nurses at the ED, between 09.30-13.00 there are four, between 13.00-15-30 there are five, between 15.30-18.00 there are six, between 18.00-21.00 there are five again and between 21.00-23.30 there are four.

Consecutive busy hours results in a higher WIP this you can see in the ‘work-in-progress’ Figures in appendix B and C. For example the 17th of September has nine consecutive busy hours. Figure 100 and 101 in appendix C shows a WIP above ten from 10.00 o’clock until 20.00 o’clock which results in a throughput time between 2.5-3 hours from noon until mid evening (see Figure 99).

5 out of 18 days (28-01-2012, 12-02-2012, 28-05-2012, 14-07-2012 and 28-08-2012) with on average shorter throughput times have also consecutive busy hours. However, the WIP and throughput time only shortly rise but it soon goes back to values below normal (the WIP below 10 and the throughput time around the two hours are on average the normal values). There are different reasons why these five days will not have extreme long throughput times and/or long periods with a WIP above ten in contrast with the 10 days that have on average long throughput times during the day and also have consecutive busy hours. The 28th of January have busy hours followed by very calm hours with only one or two patients. On the days that have on average short throughput times, there are 10 out of 18 days (20-11-2011, 27-11-2011, 4-12-2011, 31-12-2011, 28-01-2012, 08-04-2012, 15-04-2012, 29-04-2012, 09-06-2012 and 26-08-2012) that have busy hours followed by calm periods with zero, one or two patients in a hour and that helps to avoid longer throughput times during the day. For the days with on average long throughput times this only happens sporadic.

Going further with explaining why some days with consecutive busy hours not have longer throughput times during the day: the 12th of February is the day after the most busiest day and have consecutive busy hours but on average not a longer throughout time because extra capacity was assigned which results in a shorter throughput time (see also the paragraph about the detail analyses). The 28th of May and August has relative much ‘fast’ track patients this helps during busy hours and last, the 14th of July has less urgent and/or semi-urgent patients, however this is more a factor that influence the patient individually than that it affects other patients.

(23)

4.4.3 Detail analyses

As mentioned in paragraph 3.2, step 4 was straightforward and push the research in the direction of the realization process. In this part the detail analysis, were necessary, will be described. Only the days were on average the longer or shorter throughput time can’t be explained by the differences in patients subsets or time dependency will be discussed here.

There are 5 out of 36 days (27-12-2011, 31-12-2011, 01-01-2012, 11-02-2012, 12-02-2012) that needs a more detail focus. The detail analyses focus on other aspect than the reasoning mentioned before. Other aspects for example: are there more patients waiting on the same physician at a certain moment that causes a longer throughput time for some patients, is there a group of patients in the same time waiting for radiology and/or are there other factors which are not directly visible from the data. In this section only the five days will be discussed but in the summary after this section (see table 7,8,9 and 10) more detail information about the other days are mentioned. Some of the information here is gathered by interviews with the personnel at the ED.

Starting with the 11th of February, with 134 patients it was the busiest day in history. This was because it was possible to skate on natural ice and that caused a lot of accidents. As mentioned earlier, you would suspect astronomical throughput times, however, the throughput time did not rise above the four hours because of all the ‘fast’ track patients (54%). But there was another reason, they respond very well during the day. A lot of patients, 56, had elbow/forearm/finger complaints and after the diagnosis it became clear that they had a ‘good’ fracture. So these patients were bundled together and they were plaster right after each other. This you can see in Figure 69 (appendix C) were the departure line suddenly increases and smoothens again around 15.00, 16.00 and 17.00 o’clock. Not only the bundling was a good response but they also had two separate lines in the ED, one for the ‘normal’ every day patients and one line for the patients with skating accidents. With this separation you had two patients streams that were served by two different nurses and physicians pools and that helped to serve the patients quicker.

With the knowledge of the 11th of February there was extra personnel assigned for the 12th of February. Three plaster masters were specially assigned to the ED and every patient that arrives with a fracture was directly referred to the plaster masters. There were again a lot of patients with fractures and they were served directly. When you take all the patients with a fracture that day out of the patient pool that day than you had a relatively calm day. That was actually what happened with sending the patients with fractures directly to the plaster masters and therefore the average throughput time that day was relatively short.

The 31th of December and the 1st of January had on average shorter throughput times not only because the calm periods after the busy hours on the 31th or because the more ‘fast’ track patients on the 1st but also because of an extra nurse that was assigned.

(24)

(25)

(26)

26

(27)

(28)

5. Conclusion

With the help of the diagnostic framework of Soepenberg et al. (2012) this research was done. In paragraph 3.2 the steps are explained and also the adjustments that are made for this research. In chapter four the results are discussed. Starting with the first step, 36 days were chosen, the 5% days with on average a short throughput time during the day and the 5% days with on average a long throughput time during the day. The differences between the two groups is more than 50 minutes and that is really long.

In step 2, two difference subsets among patients are distinguished. The first one, is based on the urgency code patients get which is based on the MTS. After analyzing the 36 days that were depicted it seems that the urgency level of patients do not explain the on average longer or shorter throughput times during a day. The urgency patients get influence the patient more individually, only the acute patients that needs directly assistant of two nurses can have a direct negative influence on other patients, however, this is only shortly the case.

The second subset that can be distinguished, the diagnostic code patients get that determines if the patient could be in the ‘fast’- or ‘slow’ track group, have more influence on the average throughput time during the day. Days with on average longer throughput times have relative more ‘slow’ track patients, were 80% or more is not uncommon, than the days that have on average shorter throughput times, were 60% or less ‘slow’ track is not uncommon.

In step 3 the time dependency was the focus. Special attention was drawn to the arrival pattern during the day. Consecutive busy hours seems to be the biggest upset. More hours in a row with extra patients arriving at the ED put pressure on the existing capacity resources. Sometimes, calm periods after busy hours can help, however, when the calm periods not follows after the busy hours the ED can get overloaded, the WIP increases and the throughput time for the patients also increase. From the ‘work-in-progress’ Figures in appendix B and C it seems that two hours or more with a WIP around or above ten have a negative influence on the throughput time.

(29)

6. Recommendation

The last chapter pays attention to the recommendations for the specific ED to improve the throughput time for the patients, which is discussed in the first paragraph. The second paragraph mentions the limitations of this research and this chapter ends with suggestions for further research.

6.1 Suggestions

It is impossible to influence the arrival pattern and the complaints patients have, however, this influence the throughput time much and therefore it is important to get responsiveness to overcome overcrowding of the ED and to improve the throughput time of the patients. To get responsiveness you need a clear overview on all moments about the patients that are in the system. In one glance you should know how many patients are in the system, what the codes of the patients are and what the complaints of the patients are, how long they have already been at the ED and whether they are

waiting or helped in a room. This is important to know to assign extra capacity when needed or to make two patients streams, for example, which happened on the 11th of February. Important for correct monitoring of the ED is a good registration. Nurses, specialist, residents and others have to register on time and correctly, otherwise you can´t react adequate.

During meetings with the manager of the ED it became clear that there is no extra personnel available for the ED for moments when it is busy. This makes the system rigid, problems will arise sooner and you can´t react responsiveness. Also the personnel that works at the ED have to work harder during busy hours and that can give stress and can influence the performance negatively. Therefore, extra available personnel is needed for busy periods. This can be done in different ways: personnel that works at other places in the hospital can come over immediately when needed, but calling standby personnel at home would be another option.

6.2 Limitations

The biggest limitation during the research was the dataset. Especially, the incomplete data of the processes in the ED itself formed a problem. The registration of arrival, the way of arrival, the triage code a patient gets, the final diagnoses, the destination after the ED and the time of arrival were listed correctly. But the times and things that were done between arrival and departure, were not registered well. Therefore, it was not possible to see if the capacity resources were a problem and maybe the bottleneck of the system. This limited the research as it was not possible to look to the output factors mentioned in the conceptual model.

6.3 Further research

(30)

30

References

Ahmed, M.A., and Alkhamis, T.M., 2009. Simulation optimization for an Emergency Department healthcare unit in Kuwait. European Journal of Operational Research, 198 (3), 936-942.

Andersson, G., and Karlberg, I., 2001. Lack of integration, and seasonal variations in demand explained performance problems and waiting times for patients at Emergency Departments: A 3 years evaluation of the shift of responsibility between primary and secondary care by closure of two acute hospitals. Health Policy, 55 (3), 187-207.

Arkun, A., Briggs, W.M., Patel, S., Datillo, P.A., Bove, J., and Birkhahn, R.H., 2010. Emergency

Department crowding: Factors influencing flow. Western Journal of Emergency Medicine, 6 (1), 10-15. Asaro, P.V., Lewis, L.M., and Boxerman, S.B., 2007. The impact of input and output factors on

Emergency Department throughput. Academic Emergency Medicine, 14 (3), 235-242.

Asplin, B.R., Magid, D.J., Rhodes, K.V., Solberg, L.I., Lurie, N., and Camargo, C.A, 2003. A conceptual model of Emergency Department crowding. Annals of Emergency Medicine, 42 (2), 173-180.

Begley, C.E., Chang, Y., Wood, R.C., and Weltge, A., 2004. Emergency Department diversion and trauma mortality: Evidence from Houston, Texas. The Journal of Trauma and Acute Care Surgery, 57 (6), 1260-1265.

Bellow, A.A., Flottemesch, T.J., and Gillespie, G.L., 2012. Application of the Emergency Department census model. Advanced Emergency Nursing Journal, 34 (1), 55-64.

Bertoty, D.A., Kuszajewski, M.L., and Marsh, E.E., 2007. Direct-to-Room: One Department’s approach to improving ED throughput. Journal of Emergency Nursing, 33 (1), 26-30.

Byers, J., 2011. Emergency Department IT: Throughput is key. CMIO, 1-2.

Cassidy-Smith, T.N., Baumann, B.M., and Boudreaux, E.D., 2007. The disconfirmation paradigm:

Throughput times and Emergency Department patient satisfaction. The Journal of Emergency Medicine, 32 (1), 7-13.

Chapman, S.A., Spetz, J., Seago, J.A., Kaiser, J., Dower, C., and Herrera, C., 2009. How have mandated nurse staffing ratios affected hospitals? Perspectives from California hospital leaders. Journal of Healthcare Management, 54 (5), 321-335.

Christ, M., Grossmann, F., Winter, D., Bingisser, R., and Platz, E., 2010. Modern Triage in the Emergency Department. Deutsches Ärzteblatt International, 107 (50), 892-898.

Derlet, R.W., and Richards, J.R., 2000. Overcrowding in the Nation’s Emergency Department: Complex causes and disturbing effects. Annals of Emergency Medicine, 35 (1), 63-68.

(31)

Green, L.V., Soares, J., Giglio, J.F., and Green, R.A., 2006. Using queueing theory to increase the effectiveness of Emergency Department provider staffing. Academic Emergency Medicine, 13 (1), 61-68.

Griffin, J., Xia, S., Peng., and Keskinocak, P., 2012. Improving patient flow in an obstetric unit. Health Care Management Science, 15 (1), 1-14.

Hoot, N.R., and Aronsky, D., 2008. Systematic review of Emergency Department crowding: Causes, effects, and solutions. Annals of Emergency Medicine, 52 (2), 126-136.

Hwang, U., McCarthy, M.L., Aronsky, D., Asplin, B.R., Crane, P.W., Craven, C.K., Epstein, S.K., Fee, C., Handel, D.A., Pines, J.M., Rathlev, N.K., Schafermeyer, R.W., Zwemer, F.L., and Bernstein, S.L., 2011. Measures of crowding in the Emergency Department: A systematic review. Academic Emergency Medicine, 18 (5), 527-538.

Joy, T., Blue, L., and Fecher, A., 2010. Attacking the problem of hospital diversion. The Journal of Nursing Administration, 40 (4), 177-181.

Kellermann, A.L., 2006. Crisis in the Emergency Department. New England Journal of Medicine, 355 (13), 1300-1303.

Lambe, S., Washington, D.L., Fink, A., Laouri, M., Liu, H., Scura Fosse, J., Brook, R.H., and Asch, S.M., 2003. Waiting times in California’s Emergency Department. Annals of Emergency Medicine, 41 (1), 35-44.

Larsson, A., Johansson, M., Bååthe, F., and Neselius, S., 2012. Reducing throughput time in a service organization by introducing cross-functional teams. Production Planning & Control, 23 (7), 571-580. Mackway-Jones, K., 1997. Emergency triage: Manchester Triage Group. London: BMJ Publishing Group. McCarthy, M.L., Zeger, S.L., Ding, R., Levin, S.R., Desmond, J.S., Lee, J., and Aronsky, D., 2009. Crowding delays treatment and lengthens Emergency Department length of stay, even among high-acuity patients. Annals of Emergency Medicin, 54 (4), 492-503.

McNaughton, C., Self, W.H., Jones, I.D., Arbogast, P.G., Chen, N., Dittus, R.S., and Russ, S., 2012. ED crowding and the use of nontraditional beds. The American Journal of Emergency Medicine, 30 (8), 1474-1480.

Olshaker, J.S., and Rathlev, N.K., 2006. Emergency Department overcrowding and ambulance diversion: The impact and potential solutions of extended boarding of admitted patients in the Emergency Department. The Journal of Emergency Medicine, 30 (3), 351-356.

Rathlev, N.K., Chessare, J., Olshaker, J.S., Obendorfer, D., Mehta, S.D., et al., 2007. Time series analysis of variables associated with daily mean Emergency Department length of stay. Annals of Emergency Medicine, 49 (3), 265-271.

Richardson, D.B., 2006. Increase in patient mortality at 10 days associated with Emergency Department crowding. The Medical Journal of Australia, 184 (5), 213-216.

(32)

32

Schneider, S.M., Gallery, M.E., Schafermeyer, R., and Zwemer, F.L., 2003. Emergency Department crowding: A point in time. Annals of Emergency Medicine, 41 (2), 167-172.

Schull, M.J., Lazier, K., Vermeulen, M., Mawhinney, S., and Morrison, L.J., 2003. Emergency Department contributors to ambulance diversion: A quantitative analysis. Annals of Emergency Medicine, 41 (4), 467-476.

Soepenberg, G.D., Land, M.J., and Gaalman, G.J.C., 2012. A framework for diagnosing the delivery reliability performance of make-to-order companies. International Journal of Production Research, 1-17.

Solberg, L.I., Asplin, B.R., Weinick, R.M., and Magid, D.J., 2003. Emergency Department crowding: Consensus development of potential measures. Annals of Emergency Medicine, 42 (6), 824-834. Sprivulis, P.C., Da Silva, J.A., Jacobs, I.G., Jelinek, G.A., and Frazer, A.R.L., 2006. The association between hospital overcrowding and mortality among patients admitted via Western Australian Emergency Departments. The Medical Journal of Australia, 184 (5), 208-212.

Terwiesch, C., Diwas, K.C., and Kahn, J.M., 2011. Working with capacity limitations: Operations management in critical care. Critical care, 15 (1), 308-313.

Yen, K., and Gorelick, M.H., 2007. Strategies to improve flow in the pediatric Emergency Department. Pediatric Emergency Care, 23 (10), 745-752.

Yoon, P., Steiner, I., and Reinhardt, G., 2003. Analysis of factors influencing length of stay in the Emergency Department. Canadian Journal of Emergency Medicine, 5 (3), 155-161.

Van Veen, M., Steyerberg, E.W., Ruige, M., Van Meurs, A.H.J., Roukema, J., Van der Lei, J., and Moll, H.A., 2008. Manchester Triage System in paediatric Emergency care: Prospective observational study. British Medical Journal, 337, 1501-1507.

Van der Vaart, T., Vastag, G., and Wijngaard, J., 2011. Facets of operational performance in an emergency room (ER). International Journal of Production Economics, 133 (1), 201-211.

Van der Wulp, I., Van Baar, M.E., and Schrijvers, A.J.P., 2008. Reliability and validity of the Manchester Triage System in a general Emergency Department patient population in the Netherlands: Results of a simulation study. Emergency Medicine Journal, 25 (7), 431-434.

Vassilacopoulos, G., 1985. Allocating doctors to shifts in an Accident and Emergency Department. Journal of the Operational Research Society, 36 (6), 517-523.

Wagner, C., 1993. Problem solving and diagnosis. International Journal of Management Science, 21 (6), 645-656.

Waldrop, R.D., 2009. Don’t be put out by throughput in the Emergency Department. Physician Executive Journal , 35 (3), 38-41.

(33)

Wiler, J.L., Gentle, C., Halfpenny, J.M., Heins, A., Mehrotra, A., Mikhail, M.G., and Fite, D., 2010. Optimizing Emergency Department front-end operations. Annals of Emergency Medicine, 55 (2), 142-160.

(34)

34

Appendix A: Diagnostic framework

(35)

Appendix B: Short throughput time

In appendix B the three figures show per day the arrival pattern, urgency of patients and the ‘fast’ track versus the ‘slow’ track are included. Only the 28 May 2012 is not included because the figures of this day are already included in chapter 4.

(36)

36

Figure 14: Work-in-progress – ‘short’ versus ‘long’ – Sunday 20-11-2011

(37)

Figure 16: Work-in-progress – the urgency – Sunday 27-11-2011

(38)

38

Figure 18: Throughput diagram Sunday 04-12-2011

(39)

Figure 20: Work-in-progress – ‘short’ versus ‘long’ – Sunday 04-12-2011

(40)

40

Figure 22: Work-in-progress – the urgency – Tuesday 27-12-2011

(41)

(42)

42

(43)

(44)

44

Figure 30: Throughput diagram Saturday 28-01-2012

(45)

Figure 32: Work-in-progress – ‘short’ versus ‘long’ – Saturday 28-01-2012

(46)

46

(47)

(48)

48

(49)

(50)

50

(51)

(52)

52

(53)

.

(54)

54

(55)

Figure 52: Work-in-progress – the urgency – Saturday 14-07-2012

(56)

56

(57)

(58)

58

(59)

(60)

60

(61)

Appendix C: Long throughput time

In appendix C the three figures show per day the arrival pattern, urgency of patients and the ‘fast’ track versus the ‘slow’ track are included. Only the 16 July 2012 is not included because the figures of this day are already included in chapter 4

(62)

62

Figure 66: Throughput diagram Thursday 12-01-2012

(63)

Figure 67: Work-in-progress – the urgency – Thursday 12-01-2012

(64)

64

Figure 6: XX

(65)

(66)

66

Figure 73: Work-in-progress – the urgency – Friday 17-02-2012

(67)

(68)

68

Figure 78: Throughput diagram Friday 11-05-2012

(69)

(70)

70

Figure 81: Throughput diagram Friday 29-06-2012

(71)

Figure 84: Throughput diagram Wednesday 11-07-2012

(72)

72

Figure 85: Work-in-progress – the urgency – Wednesday 11-07-2012

(73)

Figure 87: Throughput diagram Tuesday 17-07-2012

(74)

74

Figure 89: Work-in-progress – ‘short’ vs ‘long’ – Tuesday 17-07-2012

(75)

(76)

76

Figure 1: XX

(77)

Figure 2: XX

Figure 95: Work-in-progress – ‘short’ vs ‘long’ – Friday 24-08-2012

(78)

78

(79)

Figure 99: Throughput diagram Monday 17-09-2012

(80)

80

(81)

Figure 103: Work-in-progress – the urgency – Saturday 22-09-2012

(82)

82

Figure 1: Throughput diagram Tuesday 02-10-2012

(83)

(84)

84

Figure 109: Work-in-progress – the urgency – Tuesday 02-10-2012

(85)

(86)

86

“How can the throughput time of patients in the Emergency Department be improved?”

“How can the throughput time of patients

in the Emergency Department be

improved?”

Tom Dijk

University of Groningen

Faculty of Economics and Business

Table of contents

How can the throughput time of patients in the Emergency

Department be improved?

Abstract

1. Introduction

2. Theoretical background

3. Methodology

4. Results

5. Conclusion

6. Recommendation

References

Appendix A: Diagnostic framework

Appendix B: Short throughput time

Appendix C: Long throughput time