2
Abstract
4
1. Introduction
As many other countries, England has been confronted with countrywide issues concerning overcrowded emergency departments (ED’s) over the last few decades (Lane, Monefeldt, & Rosenhead, 2000; Wiler et al., 2010). Rowe et al., (2011) define ED overcrowding as:‘A situation in which the demand for emergency services exceeds the ability to provide quality care within a reasonable time.’
This problem has risen extensively, with high utilisation of ED resources and soaring waiting times as a result. In an attempt to manage this prolonged crisis, The British Government introduced the four-hour rule in 2004, which implied that 98% of ED patients should be seen and either admitted, transferred or discharged within 4 hours. This was toned down in 2010 to 95% of admitted patients (Mason et al. 2012; Campbell & Mason 2017). According to Mason et al., (2012) the introduction of the four-hour rule has generally increased the proportion of patients leaving the ED within 4 hours. However, the National Health Service’s (NHS) latest figures still show that multiple hospitals only reach as little as 50 to 60% of total arrivals handled within the assumed four-hour maximum (Campbell & Mason, 2017). Much research on this problem has already been done. These studies have provided many improvement opportunities. Lane et al., (2000) for example, report that even though there is often a shortage of beds, simply increasing capacity does not always generate the desired results. Sayah et al. (2016) substantiate this claim. They demonstrated that the impact of process improvements and the implementation of rapid assessment within the front-end process was substantially greater than expanding capacity. Additionally, S. J. Traub et al., (2016) claimed that the front-end process is the place in which high-impact changes can easily be made. They argue this is possible because this process is almost completely under the emergency department’s control, with very few external stakeholders.
Various studies have already been conducted around design changes in the front-end process. Chan et al., (2005) implemented what they call ‘rapid entry and accelerated care at triage’ (REACT). REACT aims to parallelise processes, which after implementation resulted in significant improvements in length of stay (LoS). Another improvement opportunity explored in recent studies is the physician in triage (PiT) process (DeBehnke & Decker, 2002; Han et al., 2010; Traub et al., 2016). It allows for earlier contact between patient and physician and therefore an earlier advanced assessment and decision. This improved patient satisfaction and perceived waiting times. Even though these studies resulted in process improvements, the designs are all eminently inflexible. One might question if a static operating policy best suits the emergency department, since it deals with the non-stationary arrival of patients. Higginson, Whyatt, & Silvester, (2011) clearly show that the ED has to cope with a high level of variation in the number of patient arrivals. Additionally, Au-Yeung, Harder, McCoy, & Knottenbelt, (2009) state that the statistical properties of patient arrivals change over time, and are therefore non-stationary.
5
non-stationary. Their simulation study of job shops embodies control processes that change in case the system’s workload changes. Even though this study is put in another setting, the principles seem fit for other environments, like the ED. However, so far, no research has been done regarding the ED in this perspective. Even though, as explained by Au-Yeung et al. (2009), ED’s do cope with non-stationary customer arrival processes. Therefore, this research will focus on the influence of a non-stationary arrival pattern of patients on the front-end process design. This leads to the following research question:
“How does the non-stationary arrival pattern of patients influence the operational performance of the front-end process of the emergency department”
Much literature already exists on arrival patterns in the ED (Asplin et al. 2006, Jenkins et al. 2006, Au-Yeung et al. 2009,), as well as process improvements (DeBehnke & Decker 2002, Subash et al. 2004, Chan et al. 2005, Han et al. 2010, S. J. Traub et al. 2016). However, this research will contribute to literature by providing insight in how these non-stationary arrivals influence the operational performance of the ED. It will shed light on the impact of a surge in arrivals, as well as the need for a more flexible policy. The study will consist of a single case case-study in the ED department of a hospital in Hampshire, UK. It will provide a detailed overview on patient arrival patterns and surges. These will afterwards be linked to observations on the ED’s front-end process as well as performance measurements concerning patient throughput times.
The rest of this paper is structured as follows; the next chapter will discuss the theoretical background, after which the methodology section is included. This will be followed by a chapter covering the results and a chapter which contains a discussion and conclusion. Lastly a chapter with this study’s limitations and recommendations for further research.
6
2. Theoretical background
Within this section, first, an in-depth overview is provided concerning emergency department characteristics, after which the front-end process will be described in more detail. Lastly, more information concerning the arrival pattern of patients will be discussed.
2.1 Emergency department characteristics
Lane et al., (2000) explain that the ED is designed to provide access to hospitals for urgent cases, and is accessible 24 hours a day, 365 days a year. People enter the ED either by ambulance, or walking in themselves. The cases in which individuals walk in generally have a wide variation in severity. Therefore, the ED has an important role in sorting out patients and their severity level on arrival (“Urgent and emergency care services in England,” 2015). This is done during triage. According to Fitzgerald, Jelinek, Scott, & Gerdtz, (2010) triage is essential in modern day ED management, because it assigns patients to the resources it needs. But most importantly, a decision is made which patients need immediate care and which patients can wait. Most triage scales are from one to five, one being the most and five the least urgent.
Asplin et al., (2003) describe that the ED is mostly forced to cope with unscheduled visits by patients. Some caregivers have experimented with same day appointments for patients, however this does not seem to work. Patients’ conditions often worsen in the time they have to wait, or patients are simply not willing to wait the time until their appointment. Therefore, the great majority of ED visits is unscheduled. Additionally, it is mentioned that the ED is mostly the only open door for patients in need of acute unscheduled care, thus being an important safety net for the community. After arrival, patients are entered into the ED system. The process within the ED is usually split up in two parts. Wiler et al., (2010) state that the first is also known as the front-end process and consists of triage, rapid first assessment and bed placement. The second part of the process includes further diagnostic testing and treatment. According to Asplin et al., (2003) the latter will mostly cause the longest part of the patient’s total throughput time, because it needs extra and more explicit resources.
7
leadership. This low reliability on other stakeholders makes process improvements in the front-end process exceptionally manageable. In recent literature, many different designs are developed to improve the front-end process. Chan et al., (2005) described the (positive) results of the implementation of Rapid Entry and Accelerated Care Triage (REACT). Their aim was to introduce a new rapid entry process that would reduce the number of patients leaving the ED before being seen. The scope of their research only included the typical front-end process activities as mentioned before in this section. The developed design was focussed on changing serial processes into parallel processes and including various tests at the point of triage already. Even though their results were largely positive, they do acknowledge that the limited availability of staff can be a constraint. Also, since their policy was only used when no beds in the emergency were available, it could be seen as flexible. However, no explanation is given on why the policy is only used in this situation or when exactly it is activated. Therefore, one can only guess about the underlying reasons that drive this need.
Moreover, Physician in Triage (PiT) has been studied in different perspectives over the last two decades. DeBehnke & Decker, (2002) discovered that using a physician in triage positively influences patient satisfaction and perceived waiting times. However, the influence on actual waiting times is outside the scope of their research. Han et al., (2010) studied the effect of physician in triage on the length of stay. They concluded that the length of stay of non-admitted patients indeed decreased. The length of stay of admitted patients on the other hand was not affected. This difference can be explained as a result of delays in the back-end of the process, mostly caused a shortage of beds in hospital wards. Furthermore, the study of Subash, Dunn, McNicholl, & Marlow, (2004) focussed on the effect of introducing team triage on waiting times and length of stay in general. In the test periods, they saw a significant decrease in time to assessment, time to radiology and time to discharge. However, their intervention was only during the quieter moments. Therefore, it neither included the effectiveness in a highly-utilised environment, nor did it include any effects of potential arrival surges.
8
Saturdays until Mondays are usually the busiest days of the week. Also, they show that during summer more patients attend the ED.
This claim is substantiated by annual data from the NHS (Secondary care analysis, 2017). Their data shows that most patients indeed arrive on a Monday. Additionally, a daily peak in arrivals can be seen between around 4pm and 8pm. However, their data shows that there is a decrease in attendance during the July and August, typically explained by the summer holiday period.
Asplin, Flottemesch, & Gordon, (2006) define the facts as stated above as arrival patterns, in terms of volume, that can be predicted based on historic data. However, Au-Yeung et al. (2009) state that the arrival pattern of patients is non-stationary. This means that the statistical properties of the arrivals, such as the mean and the variance, change over time. For this reason, arrival patterns are difficult to predict. Moreover, Asplin et al. (2006) mention that the ED has to deal with unexpected surges. They define a daily surge as the unexpected and sudden increase in demand, using only the daily operating resources of a hospital. Additionally, they mention that in case either a disaster response plan is set in motion, or extra resources are added besides those used in day to day operations, the surge has exceeded the daily surge threshold. However, Jenkins, O’Connor, & Cone, (2006) state that because ED’s are in a state of overcrowding regularly, a real disaster is not needed to get the exceptional surge effect as described by Asplin et al., (2006). They claim that the effect of a sudden spike can already be seen on a daily basis in case of high resource utilisation in the ED. They also acknowledge that hospitals usually do not hold any resources they can add in times of higher demand. 2.4 Summary The ED is forced to deal with unscheduled patients 24 hours a day seven days a week. The process within the ED is split up in to two parts, the front-end process of which is most critical from both a clinical and an operational perspective. The front-end process usually includes activities from entrance to the ED until the first assessment. A lot of research has been done regarding front-end process improvements. However, the improvement designs are mostly static in nature, and lack an insight on the influence of arrival variation and surges. Arrival characteristics can be separated in the rather predictable daily trends as well as sudden unexpected spikes in arrivals. Therefore, this research will focus on the influence of the usual arrival pattern as well as surges in arrivals of patients on the operational performance of the emergency department.
9
3. Methodology
The aim of this research is to see how the non-stationary arrival pattern of patients influences the operational performance of the front-end process. As such, it is important to map the arrival pattern of patients and its variation. Afterwards, an analysis will be provided concerning waiting times and throughput numbers. Then, the influence of surges on the operational performance will be explored. Operational performance can be measured through throughput data, observations of the process and staff experiences.
3.1 Single case study
This case-study is explorative of nature because it aims to acquire familiarity, and provide new insights, into a subject not yet researched. For that reason, according to Voss, Tsikriktsis, & Frohlich (2002), a case study suits this research. It allows the effects (of arrival surges) to be directly observed in its own habitat, the emergency department. This is important since the process is under much variation, which is impossible to simulate. Additionally, because of the nature of the setting, it would be unsuitable for experimenting due to ethical reasons.
Furthermore, a case study will provide the ability to combine different types of data collection to increase validation. Hence, observations on the ED process can be compared to throughput times as well as physicians’ experiences. Additionally, a single case study will provide a more detailed understanding of an empirically rich and context specific phenomenon. Ideally, this study would be repeated in different cases to improve its generalisability. However, due to a limited time frame, this study consists of a single case case-study. Within the following sections the case description, data collection and data analysis will be further discussed.
3.2 Case description
approach to processing patients. All recommendations relate to the specifics of the front-
10
end process as mentioned in section 2.2. They include quick triage after patient arrival, registration, bed placement and a first assessment by a physician. More specifically, this process is, as the name already suggest, an example of a physician in triage (PiT) process. The new (static) design of the Pitstop process has been implemented approximately 9 months before this research was finished, and is only implemented on weekdays between 8am and 10pm. Even though physicians feel the process has improved since, the ED is still amongst the ten worst performing trusts in the country most days1. Since this ED operates
using a static operating policy, it is a very suitable setting to evaluate how arrival surges influence the front-end process and its need for flexibility.
Additionally, to clearly visualise the influence of these patient surges, the studied case should frequently have a high utilisation of capacity. In case of continuous spare capacity, a surge can more easily be absorbed. Particularly when an emergency department is (close to) overcrowded, surges are expected to cause significant troubles in the way the ED operates, with undesired higher throughput times as a result. For that reason, a hospital must have a combination of days with high and low bed utilisation to create a good overview. Since the studied hospital is unable to meet the NHS standards in LoS, as well as it regularly has a high bed utilisation, it is assumed that indeed this emergency department operates in an overcrowded state often. However, data shows that there are also many days with low to medium bed utilisation. Therefore this case is very appropriate for this research. (Anon, 2017; Anon, 2015; O’Leary, 2015; O’Leary & Bannister, 2015).
3.3 Data collection
Within this case study multiple types of data will be used. Firstly, data will be collected from the department’s information system, ‘Oceano’. The dataset will include: unique patient numbers, date of arrival, time of arrival, mode of arrival, time to triage, time to first assessment, time to full assessment (discharge of admittance), and where patients are referred to. This data will provide the possibility to analyse arrival patterns, its variation and unexpected surges in these arrivals. Additionally, a separate dataset will be acquired from the same system, which shows the bed utilisation within the different parts of the department. This will enable a separate analysis on the before mentioned factors in different states of crowdedness. Section 3.4 will further explain how this data will be analysed. Besides this quantitative dataset, qualitative methods will be used. Firstly, semi-structured interviews with ED staff will be used to substantiate the results from quantitative analyses. Lastly, observations will be used to analyse the actual operations. Moreover, observations can be used to back-up information gathered from interviews. Using these three types of data collection, called triangulation, will further increase construct validity (Voss et al. 2002).
11
3.4 Data analysis
Firstly, the quantitative data received is analysed using Microsoft Excel. Data will be checked and any faulty or incomplete data will be removed from the sample. Additionally, case descriptives are provided in Appendix I and II. These include an overview of age and gender distributions, outflow destinations, and average arrivals and throughput times over 2016.
The first results section will be dedicated to revealing variation within arrival patterns. Line graphs will be plotted showing the actual arrivals as well as the average arrivals. These figures will be separated in to the total daily arrivals and a more detailed view at hourly arrivals during one week. Since the line graphs provide only limited information on the variation of arrivals, boxplots using percentiles will then be shown to disclose the dispersion in the arrival variations.
As mentioned in section 2.2, as well as by the ED’s physicians, problems with the outflow of admitted patients can cause long waiting times. For this reason, patients are divided in to two separate groups when analysing throughput times. The first group will consist of patients that have been admitted, the last group will consist of all other patients. This last groups thus consists of patients that are discharged without any further follow up, discharged with a follow up by a general practitioner (GP), and patients referred to other health care providers. Using these two groups will show if there indeed is a difference in throughput times for admitted patients. Additionally, it is interesting to see if there is any tendency of focussing on patients that do not need admittance in order to improve flow. The following analyses will be done using these patients’ groups. Firstly, a line graph will be provided showing the average times until seen and departure. Afterwards, a graph will be included presenting the average patients arriving, seen and departing on 15 minute intervals for the year 2016. Then, analyses including the expected time until seen and time until departure for both groups will be compared. This will be calculated using a cumulative in- and outflow analysis based on historical data over 2016.
Afterwards, throughput diagrams will be included to show the impact of arrival surges. To draw a good example, days will be chosen in which during one hour that day the number of arrivals is at least 170% of the average number of arrivals. Additionally, to increase the comparability of these days, only the surges during the day on weekdays will be analysed, as it is then that the pitstop process is used. Throughput diagrams are a powerful tool in showing the difference between input and output in to the ED. Therefore, it will also shed light on the consequences of a surge on the times until patients are triaged and seen. Since a surge is expected to have a different influence on a quiet day from a crowded one, three different examples will be included to compare the influence of a surge. The different diagrams show the influence of a surge hitting when the front-end process is 20%, 50% or 80% utilised. The crowdedness of the front-end department will be calculated based on bed utilisation.
12
bringing other factors to light that influence findings. Moreover, real-life observations give a good overview on the operational performance. They can also be used to back up statements drawn from interviews. Using these two types of data to back up the quantitative data and each other will also create a proper level of internal validity. Combining all these analyses will develop a detailed, all-round insight of the influence of patient arrivals on the operational performance of the ED.
13
4. Results
Within this chapter all data will be processed and elaborated on. Firstly, data concerning the variation in arrivals into the ED will be analysed. Secondly, arrival- and throughput data is split in to two groups as mentioned in section 3.4 and discussed. Afterwards, the effect of a surge will be illustrated using throughput diagrams. Lastly, qualitative data will be used to explain what happens to the process in case a surge hits. 4.1 Arrival variation
14
Figure 2: Arrival variation per hour
15
Figure 3: Arrival number distributions per weekday
Figure 3 shows that most patients attend the ED during the weekend and on Monday. Moreover, from Wednesday until Friday the arrivals are lowest. Even though Thursday and Wednesday both have the same average number of patient arrivals, the distribution is much higher on a Thursday than a Wednesday. This indicates more variation in arrivals on a Thursday. However, none of the interviewees could find an explanation for this difference in variation.
To get a more detailed look on the variation, Figure 4 shows the distribution of arrival numbers during the day. Figure 4: Arrival number distributions during the day The distribution of arrival numbers during the day, as can be seen in Figure 4, is separated in three hour blocks. Unlike Figure 3, this figure shows does not separate the different days of the week. It illustrates that the number of arrivals during the night is significantly lower than 80 90 100 110 120 130 140 150 160 170 180
Monday Tuesday Wednesday Thursday Friday Saturday Sunday
16
during the day. Also, even though arrivals remain relatively steady during the day, a slight peak in arrivals can be identified between 12:00 and 15:00. Moreover, least variation in arrival can be seen between 03:00 and 06:00. Between 21:00 and 24:00 on the other hand, the variation is largest. It shows that the lowest number of arrivals during 2016 was only 3, whereas the highest was 34.
This section has shown that the ED arrival number are subject much variation. Attendances can differ largely from day to day. Additionally, there is significant variation in arrivals at different times of the day. The next section will shed light on the arrivals and throughput times from moment of arrival, of both admitted and non-admitted patients.
4.2 Arrival and throughput analysis
Within this section, the attendances of 2016 are split in to ‘admitted’ and ‘other’ patients as explained in section 3.4. Firstly, average waiting times until seen and departure will be compared for both groups. Secondly, the average patient arrivals, seen and admitted will be shown and compared in 15-minute intervals for the year 2016. Lastly, figures showing the expected times until seen and departure, as well as the work in progress will be provided for both groups. Figure 5: Average time until seen and time until departure per 15 minutes over 2016 As Figure 5 shows, the average time until seen for discharged patients is always higher than the time until seen for admitted patients. The slight difference between the time until seen during the day is a logical result of treating patients according to triage levels. A physician explained:
“Generally speaking, patients that are being admitted are more poorly, as a group, than those going home. When patients arrive that are poorly, they are
17 obviously seen as a clinical priority. Therefore, on average, they are seen slightly sooner.” However, during the night the time until seen for both groups is further apart than during the day. This is a result of less physicians on shift during the night. Therefore, the physician that is available during the night usually treats the sickest patients, who are more likely to be admitted. As for the time until departure, there is a big difference between admitted and discharged patients. Discharged patients leave the departure earlier than admitted patients. This is caused problems moving admitted patients in to hospital wards. When the times until departure come closest together at 19:00, the time until departure for admitted patients is still an hour longer. To closer compare patient flow, the next section will focus on the average number of patients arriving, being seen and departing per 15 minutes, over 2016.
Figure 6: Average number of admitted patients arrived, seen and departed per 15 minutes over 2016
Figure 6 shows that the number of patients arriving to the ED that need admittance is increasing rapidly during the morning. This corresponds with the arrival pattern of all patients, as shown in figure B of Appendix II. However, that figure shows that arrivals are high all day and even reach a peak at 21:00. Figure 6 on the other hand, shows a decrease in arrivals in the evening. 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 0: 00 1:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00 10:00 11:00 0012: 13:00 14:00 15:00 16:00 17:00 0018: 19:00 20:00 21:00 22:00 23:00 0:00 Av er ag e nu m be r o f p at ie nt s Time of the day
18
The number of patients seen follows the pattern of arrivals to some extent. More patients are seen during the day than during the night, and the number of patients seen is highest in the afternoon. However, the line illustrating the patients seen shows some high peaks at 8:00, 12:00, 16:00 and 22:00. As stated by a physician, this can easily be explained by new staff coming on the floor. At these points the new staff quickly picks a patient. Besides these peaks, serious drops in productivity can be seen just before 8:00 and 22:00. A physician explained that this does not have to do with handover time, as one might expect.
19 Figure 7: Average number of other patients arrived, seen and departed per 15 minutes over 2016
Contrarily to admitted patients, the number of other patients arriving to the ED is more evenly distributed during the day. The number of arrivals during the day is only a little higher than during the night. The arrivals rise slightly after 8:00, but it is only in the evening, between 20:00 and 23:00, that the attendances are highest. This was to be expected, as the total number of arrivals (as shown in figure B, appendix II) stays high until late in the evening, but the arrivals of admitted patients in Figure 6 showed a small decrease.
Looking at the patients seen, similar peaks as for the admitted patients can be identified. However, contrarily to Figure 6, the difference in patients seen between day and night is smaller. Yet this is to be expected, as the arrivals follow the same pattern.
Lastly, the number of departures is also more evenly distributed over the day. Contrarily to Figure 6, Figure 7 does not show a strong delay for patients departing compared to patients arriving. Thus, one would expect there are less patients in the department waiting to leave. In order to create more understanding concerning this pattern, the next section will look further in to the expected time until seen and the expected total length of stay for both
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 0: 00 1:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00 10:00 11:00 0012: 13:00 14:00 15:00 16:00 17:00 0018: 19:00 20:00 21:00 22:00 23:00 0:00 Av er ag e nu m be r o f p at ie nt s Time of the day
21 The expected length of stay for admitted patients is much higher than for other patients all through the day. Observations and conversations with employees in the operations centre as well as the ED brought to light that this is mainly caused by problems moving the patients from the ED in to the hospital wards. Since the hospital wards are highly utilised, there is little space for patients coming in through the ED, with lengthy waits as a result. Figure 9: Expected time until seen and work in progress in the ED over 2016 When looking at the expected time until seen in Figure 9, it is clear that admitted patients are seen sooner than patients that are not. During the day, both times never differ more than 30 minutes. During the night, the time until seen increases largely for all patients. However, non-admitted patients have to wait even longer to be seen, with a peak waiting time of approximately 4 hours around 04:00. A physician explained this as follows:
22
“When the attendances are high, and they remain high until 9 o'clock in the evening, you have a decent number of staffing causing the time until seen to be better than during the night.” Moreover, after 22:00 the time until seen rises quickly, as staff numbers drop strongly at this hour. “At 10 o'clock that number drops down. Then you’re down to one consultant in the entire department. “…” It is relatively easy for a relatively small number of attendances to get behind the curve of seeing them.”
Thus, the expected length of stay is largely dependent on staffing. During the night, the waiting times are longer as less staff is on. During the day, more staff is on shift resulting in lower times until seen as well as times until departure. However, during the day there is some influence of arrival patterns on the front-end process visible. As the number of attendances reach a relatively steady state after 12:00, the time until seen is rather stable as well. Moreover, waiting times are increasing after 17:00 as more patients arrive. Where the figures above all show the expected daily fluctuations, the next section will shed more light on the influence of unexpected surges. 4.4 Surges
This section will provide more insight into the influence of a surge of arrivals on the ED. Firstly, the effect of a sudden surge will be analysed in a situation of low bed utilisation. Additionally, the effect of a sudden surge of arrivals will be reviewed in a case of both medium and high bed utilisation at the moment the surge starts. Lastly, the influence of a surge on the process itself and how it is run will be described. 4.4.1 Surge at moment of low bed utilisation Figure 10 shows the in- and output flow for December 23rd 2016. On this day, between 12:00 and 13:00, 13 patients arrived at the ED, where on average only 7 patients arrive during this hour. With almost the double amount of average arrivals, this clearly represents an unexpected surge. When the surge started just after 12:00, only 2 bays were occupied in the front-end process, which has room for 10. Therefore, bed utilisation of the front-end process was only 20%.
23
Figure 10: In- and output flow Dec. 23rd 2016
In general, Figure 10 once more shows that during the day the number of arrivals is higher than during the night. Additionally, the number of patients triaged follows the arrival numbers closely, until a few patients arrive in a short period of time. The number of patients triaged then distances from the number of patients arrived, identifying a short delay. Concerning the number of patients seen, there is a high increase right after 8:00 and 16:00. Also, the number of patients seen is kept well aligned with the arrivals during the afternoon after the catch up around 16:00. However, after 22:00 the number of patients seen clearly cannot follow the arrivals anymore. According to an ED physician, this is a direct result of staffing.
This again is down to staffing. ‘…’ When the attendances are high, and they remain high until 9 o'clock in the evening, you have a decent number of staffing causing the time until seeing patients to be better than during the night. ‘…’ At 10 o'clock that number drops down. ‘…’ It is relatively easy for a relatively small number of attendances to get behind the curve of seeing them. 0 20 40 60 80 100 120 140 160 180 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 Cu m ul at iv e nu m be r o f p at ie nt s Time
24 When looking at the surge between 12:00 and 13:00, the number of people triaged and seen does not keep up with the arrivals. However, it does not seem to have a strong influence on the rest of the day. For closer analysis, Figure 11 will show a more detailed overview of the surge between 12:00 and 13:00. Figure 11: Surge analysis in case of low bed utilisation
When looking at the number of patients being triaged in Figure 11, one can see that immediately after a couple of patients arrive right after each other, the number of patients triaged cannot keep up. However, a response can be seen after 25 minutes, after which the number of patients triaged rises quickly again. For this reason, the number of patients triaged catches up even before the end of the surge. This is also shown in the time until triage. When the first patient enters arrives after 12:00, the time until triage is 12 minutes. Shortly after, the time until triage is down to 15 minutes. However, the time until triage for the last patient arriving before 13:00 is only 6 minutes. In this analysis, time until triage is based on a first come first serve basis and, might therefore differ slightly in reality. As with the number of patients triaged, the number of people seen cannot keep up with the arrivals during the surge either. However, the number of patients seen needs more time to recover. It is not until 14:20 that the number of patients seen compared to the number of patients arrived is at approximately the same level as before the surge. Figure 11 shows that first patient arriving after 12:00 is seen within 42 minutes. However, the time until seen of the last patient arriving before 13:00 is 1 hour and 16 minutes. An ED physician confirmed the above-mentioned findings. It was stated that in case of low bed utilisation the ED is capable of managing a sudden surge of patients rather well. If there are spaces available, there is good order in the ED and the nurses as well as physician can do their work in a structured manner. Also, there is good overview on which patients still need to be triaged or seen and their location. 60 80 100 120 11:00 12:00 13:00 14:00 15:00 Cu m ul at iv e nu m be r o f p at ie nt s Time
25 4.4.2 Surge at moment of medium bed utilisation Figure 12 shows the in- and output flow on November 28th 2016. Between 13:00 and 14:00 there was a surge of 12 patients, where on average only 7 patients come in at this hour. The bed utilisation at the moment the surge started was 50%. Also, during the day the total number of arrivals is approximately the same as on December 23rd. This makes both cases adequately comparable. Figure 12: In- and output flow on Nov. 28th 2016 Figure 12 visualises that during the morning there is an effort to catch up with the backlog developed overnight. At 12:00 the number of patients arrived, triaged and seen is approximately the same. Thus, at the moment the surge happens an hour later, there is only little backlog, just like on December 23rd. However, at this moment there is a bed utilisation
of 50% instead of 20%. When looking at the entire day, Figure 12 shows that the effect of the surge lasts longer than on December 23rd. The number of patients seen only catches up around 18:00. However, the situation was not as it was like before the surge hit until
0 20 40 60 80 100 120 140 160 180 200 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 Cu m ul at iv e nu m be r o f p at ie nt s Time
26 midnight. The time until triage seems to be affected less severely, but to judge so properly, a more detailed view of the surge will be analysed. Figure 13: Arrival surge in case of medium bed utilisation
Figure 13 clearly shows that the time until seen is indeed affected largely, and does not catch up the with number of patients arriving as quickly as was seen in Figure 11. On the other hand, the number of patients triaged is less affected. The first patient after coming in at 13:00 already had to wait 27 minutes before being triaged. However, the first patient arriving after 14:00 only had to wait 7 minutes until being triaged. During the surge the time until triage goes down gradually.
Important to note is that 27 minutes is already a long time for a patient to wait triage. The department aims to have every patient triaged within 15 minutes. Also, observations showed that there is a live system showing when ambulances are expected to arrive exactly. Therefore, nurses might have made sure they were extra prepared, as they saw the surge was about to happen soon. Additionally, extra resources might have been pulled from elsewhere in the department, as will be explained more in section 4.4.4. Unfortunately, there is no evidence proving this. Hence it can only be concluded that the number of patients seen lags behind for the rest of the day after the surge hits, whereas the number of patients triaged catches up quickly.
4.4.3 Surge of moment of high bed utilisation
Figure 14 shows the in- and output flow in the front-end process of the ED at the 17th of
November 2016. Between 13:00 and 14:00, there was a sudden unexpected surge in patients. During this hour 13 patients arrived to the ED, where the average number of arrivals is only 7. However, unlike the before, the bed utilisation at the beginning of this surge was 80% instead of 20% or 50%. 80 100 120 140 12:00 13:00 14:00 15:00 16:00 Cu m ul at iv e nu m be r o f p at ie nt s Time
27 Figure 14: In- and output flow Nov. 17th 2016 When looking at Figure 14 it clearly shows that from the moment the surge starts at 13:00, the number of patients being triaged stays behind. Where before this number caught up quickly at the end of the hour, in this case the triage seems to lag behind for the rest of the day until approximately 18:00. Additionally, contrarily to what was shown in Figure 10 and Figure 12, at the moment the surge starts there is already a significant backlog in the number of patients seen. In this case, the number of patients seen does not catch up to the number of arrivals for the rest of the day. This is a more extreme effect than seen before. With a 20% utilisation, the number of patients seen caught up only two hours after the surge. With a 50% utilisation, it was much later before the number caught up, yet it was still the same day. However, it is important to realise that the 17th of November the total number of arrivals is
higher than before, which might also affect the throughput. Unfortunately, since bed 0 20 40 60 80 100 120 140 160 180 200 220 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 Cu m ul at iv e nu m be r o f p at ie nt s Time
28 utilisation is mostly a result of many arrivals there is little possibility to evade this. Figure 15 will now, like before, shed a more detailed light on the direct influence of the surge. Figure 15: Arrival surge in case of high bed utilisation Unlike the situation with low bed utilisation as shown in Figure 11, Figure 15 shows that in case of high bed utilisation the number of patients triaged does not catch up within the hour. Even though some effort seems to be made around 13:30, the number of patients being triaged is not sufficient to meet the number of arrivals. This is confirmed by the time until triage. Looking at the hour of the surge, the time until triage was 8 minutes for the first patient arriving after 13:00. The last patient before 14:00 had to wait 15 minutes until being
triaged.
Moreover, the number of patients seen at the moment the surge starts is already more behind on the number of arrivals than in Figure 11 and Figure 13. The first patient arriving after 13:00 has to wait 1 hour and 7 minutes before being seen. The last patient arriving before 14:00 has to wait 1 hour and 53 minutes before being seen. When looking at the hour of the surge only, the time until seen is increased more than before. Additionally, the time until seen does not seem to catch up within the same time frame as before. An ED physician mentioned that the operational performance is much more affected by a surge in times of high bed utilisation. The biggest issue is said to be space. In case there are no beds available, patients are being placed on the corridor. If there is no place in the corridor, the patients wait in the ambulances. This causes much trouble for triage as well as for the first assessment. When the front-end area is full, a physician has to move patients that currently occupy a bay but are already seen towards the corridor, and place patients that are still to be seen from the corridor in a bay. This is the aim, since seeing patients in the corridor comes with difficulties around privacy. These movements take much time of the
80 100 120 140 160 12:00 13:00 14:00 15:00 16:00 Cu m ul at iv e nu m be r o f p at ie nt s Time
29
physician’s time and negatively influences the number of patients they can see in a certain amount of time.
Observations confirmed these statements. The ED consultant was constantly channelling between his computer to identify the next patient, and the floor to find these patients. The ED consultant at that moment explained that it is a big challenge finding patients when it is this busy as they can be in so many different areas at the front-end process. This also explains the fact that there are less patients seen than in moments of low bed utilisation, as consultants’ time is wasted on locating patients. According to physicians, an effort is made to deploy extra physicians when the department gets overcrowded. However, this is usually very hard since these physicians have their own patients for which they are responsible as well. As far as bringing staff to the front door, it is difficult, because they're looking after their own patients ‘…’ Medical wise if we have a lot of patients arriving all at once we endeavour to put 2 people at the front door but this has to be balanced with the rest on the department and the needs in the other areas. “…” The clinical space is more of an issue, we don't have the physical space so actually we struggle do an effective Pitstop at that moment.
Observations on a busy Monday afternoon confirmed this statement. The front-end area at that moment was fully clogged, and all beds in the corridors were occupied as well. Additionally, there were on average 10 ambulances with patients waiting to be seen outside. At this point an extra physician was requested to join the floor, even though officially he had no clinical duty at that time. This consultant was based outside in the parking area, to assess patients in the back of the ambulances. As explained, his job was to pick the sickest patients and make sure they get the tests urgently needed. At that point, he was solely firefighting, and making sure the situation remains safe for patients.
4.4.4 Surge influence on processes
Interviews and observations showed that sometimes a policy is activated called ‘pitstop surge policy’. Even though this sounds as a real structured policy, it only implicates that staff is pulled from the back-end process and moved towards the front-end process. Where possible, up to two or three nurses or health care practitioners are pulled from the back-end process to assist during the surge in arrivals. Also, if possible, an extra consultant is deployed to the pitstop area. Yet, as explained in section 4.4.3, there is only limited flexibility in moving them.
30
amongst nurses. This ensures they are prepared to the best of their ability. However, the abovementioned actions can only be taken when bed utilisation allows these movements. If all bays on the corridor are already occupied, actions are limited.
In a situation where there is flow out of the pitstop area and beds utilisation is around 50%, observations showed that a surge can relatively easily been processed. However, in the morning the system can get clogged up quickly, because of limited admitted patients moving out of the ED as shown in Figure 6. This is confirmed by all interviewees, who mention that problems in outflow have a significant impact on the pitstop process. Therefore, when a surge hits at a moment where there are problems moving patients out of the pitstop area, the consequences are more significant. Since the outflow blockage happens regularly, consultants do take this in to account when prioritising patients. A physician stated:
“I quickly look at the triage notes and ambulance notes. I will then cherry pick patients that we can dealt with quickly. “…” I will pick low clinical priority patients who will otherwise wait around for a long time to be seen. If I know that I can do 10 minutes of work and get them out the queue.”
Obviously, patients needing immediate attention are seen first. As there are issues with outflow, it might be expected that physicians where possible pick the patients that will eventually be discharged. However, as mentioned by a physician, this is difficult to see up front. Therefore, the focus tends to be on patients that can move to departments where space is available. This is confirmed when analysing three days at which a surge hit between 13:00 and 14:00, as shown in Figure 16. This figure shows that there is no clear prioritisation in seeing patients that are later admitted or not. Figure 16: Time until seen for admitted and discharged patients arriving during a surge Again, since there is no set policy to deal with these incoming surges, they are dealt with in different ways. Different consultants have different ways of working, and thus will prioritise patients how they see fit. A physician explained: 00:00:00 00:30:00 01:00:00 01:30:00 02:00:00 02:30:00 03:00:00 03:30:00 04:00:00 04:30:00 05:00:00 13:00:00 13:15:00 13:30:00 13:45:00 14:00:00 14:15:00 14:30:00 14:45:00 15:00:00 Nov. 17th admitted patients Nov. 28th admitted patients Dec. 27th admitted patients
31 “It's such a fluent thing. It's basically down to clinical experience. “..” However, when there is a sudden surge in ambulances, and I know there is 3 or 4 beds in a specialty ward, I will have a look if I can find a patient I can quickly send through. There are so many variables, and different doctors do things differently.” Additionally, different nurses will have different ways of leading the department. One of the nurses mentioned that this can cause difficulties, especially during shift changes, as you have to adapt to the way of working of the person in charge. In case the department is already overcrowded, very little actions can be taken as there is no flexibility to move patients around. In this case, ambulances will stay in queue after arriving until space becomes available in the department. During observations on the 19th of June, patients were waiting in the back of ambulances for one to two hours. In this situation, an extra consultant was deployed in the pitstop area from administrative duty, to ensure patient safety. It was then mentioned that he was mostly there to make sure the sickest patients in the back of ambulances got the necessary treatment. There was no procedure followed to ensure how the addition of an extra consultant could be used most efficiently. Additionally, the physician and nurses based in the pitstop area lose a lot of time looking for patients, as mentioned in section 4.4.3. This is a result of the many areas patients are placed in busy times. Patients can be located in the pitstop area, on the corridor, in resus, on in the spare resus room called ‘the star suite’. The nurse in charge usually writes down all patients’ locations on a board, to make sure all other staff can find their patient. However, this board only covers the six bays and 4 chairs in the pitstop area. Thus, when the department is crowded, the patients located outside this area are hard to trace.
Besides issues in locating patients, movement itself also becomes more complicated. One of the reason is the fact that the corridors are filled with patients and family is usually standing close by the bed too. Also, medical equipment might be needed for these patients as well, which may block movement. During one of the observations a physician said that during these busy periods, moving patients around was like solving a sliding puzzle, as there is so little empty space in the department.
32
5. Discussion and conclusion
Within this section, the before mentioned findings will be discussed and conclusions drawn. This chapter is divided into four sections. Firstly, the main findings will be discussed. Secondly, this study’s limitations are elaborated on. Thirdly, recommendations for future research will be mentioned. Afterwards, managerial recommendations will be discussed. Within the last section, conclusions will be drawn.
5.1 Main Findings
This section will discuss the main findings of this study. The aim of this research was to discover how the non-stationary arrival pattern of patients influences the front-end process.
Firstly, analysis of the arrival numbers over 2016 showed that the actual arrivals per day predominantly deviate largely from the averages. This is different from what was claimed by Asplin, Flottemesch, & Gordon, (2006), who stated that the arrival numbers can largely be predicted from historic data. However, when looking at individual days, a pattern more closely related to the average arrivals can be identified. Yet, there are still many peaks and troughs at certain hours deviating from the hourly averages. As the arrival variation is a crucial part of the research aim, the above-mentioned findings make the selected case very suitable for this study. Secondly, when looking at the time until seen by a physician, compared to the daily arrival variation, the data shows that expected throughput times are longest during the night. This is explained by lower staffing between 22:00 and 8:00. However, during the day, the effect of arrival variation can be noticed in the front-end process, as the expected time to be seen by a physician is higher when the number of arrivals increase. Furthermore, the results when comparing the impact of unexpected surges in arrivals in a situation with low, medium and high utilisation are in line with the expectations. The data shows that in case of low bed utilisation the effect of the surge is relatively small and easily caught up with. In case of medium bed utilisation, the effect of triage is hard to identify, as it goes down strongly during the period of the surge. On the other hand, the number of patients seen does not catch up with the number of patients arrived until the end of the day. Since the number of patients arriving during the day is approximately the same as in the analysis with low bed utilisation, the effect is most certainly caused by the surge.
Moreover, a surge arriving when the bed utilisation is high is proven to cause most problems. It regularly results in patients in the corridor, and a queue of ambulances outside the hospital. The time until triage is almost doubled, and the time until seen does not catch up for the rest of the day. However, this might also be influenced by a higher total number of patients coming in that day than during the days of the two earlier surge analyses. Unfortunately, this is hard to go around, as high bed utilisation is mostly the logical result of many arrivals.
33
need, and necessary test are taken as soon as possible. These actions are in line with theory stated by Wiler et al., (2010), who argued that clinical and operational criticality lays in the front-end process. Additionally, physicians put in some prioritising effort. However, it is not focussed on non-admitted patients, as was expected, because it usually is not clear if patients need to be admitted or not at first sight. It was discovered that in crowded situations physicians look at what patients can be seen quickly and then referred to wards that have space available.
Lastly, the physician in triage process as studied by DeBehnke & Decker (2002) and Han et al. (2010) generally seems to improve operational performance in this emergency department as well. However, the process experiences significant problems in moments of high utilisation, as stated by Affleck et al. (2012). There is no different protocol on how to operate in case of a surge, nor during regular really busy times. However, physicians and nurses in charge usually do have their own way of adjusting to these situations. Since this differs per member of staff, it costs time for other staff to adapt. This confirms the statement made by Ward et al. (2015), explaining that flexibility is usually only informally integrated in to the ED. Additionally, there is no knowledge if a certain way of working in these situations is most efficient. 5.2 Limitations
This study has three main limitations, two of which concerning the generalizability of this study. Firstly, the studied case is one of the worst performing trusts in the United Kingdom. Even though this created a good setting for the study, generalising results might be difficult when looking at better performing hospitals. Secondly, this ED gets clogged up regularly, obstructing day-to-day operations. This is mostly caused by a lack of patients flowing out of the department. The problem concerning flow impacts all results, going through all the way towards the front-end process. For that reason, results might be different for hospitals not dealing with these issues. To back-up the results from this study it would be useful to repeat the study in a better performing ED with less outflow issues to see if the results are comparable. Moreover, this study does not shed light on the severity of patients arriving. However, this might have an influence on the operations in the ED. Some effort has been made by splitting admitted and non-admitted patients, however no detailed information on triage was used. If a surge arrives including many severely injured patients, this will be harder to deal with for the department than in case only mildly ill patients arrive to the hospital. 5.3 Recommendations for future research
This study is a first step in exploring the effect of arrival variation and more specifically surges in arrivals on the front-end process of the emergency department. It is important to acknowledge that it is just a starting point. More in-depth research is needed to further look into these problems, and how to overcome the issues explored in this study.
34
implementation of a specific protocol in times of a surge would positively influence operational performance.
Moreover, it was found that if a surge hits in times of high utilisation, physicians try and prioritise patients that take little time and can easily be referred into hospital wards that have beds available. However, their decisions are made based on clinical experience only. More detailed research in how the prioritisation of specific patients may influence the flow in the department in periods of high utilisation could bring to light interesting improvement opportunities. Additionally, in the process of this study, it was found that outflow issues cause problems all the way through to the front door. Therefore, research mapping the impact of the outflow blockage, and its causes would be highly beneficial in improving operational performance as well. 5.4 Managerial implications
This study also exposed some managerial opportunities. Firstly, a policy adapting to the system’s state would seem beneficial. The current policy only deploys more staff to the front-end process. As there is no protocol to follow in overcrowded periods or surges, staff at this moment organise the ED as they see fit. Unfortunately, there is only one consultant permanently in the front-end process. Thus, it is difficult to create a policy to including flexibility to employ resources, as at this point there isn’t any. Therefore, it is essential to expand the consultants staffing at the front-end first, and afterwards try to design such a protocol. Moreover, during observations and in interviews with staff, it seemed that the main issue in the performance of this emergency departments was caused by the outflow block instead of the arrival surges. This outflow blockage causes significant operational problems, and long waiting times. Therefore, managerial actions to improve the outflow would be advised. 5.5 Conclusion
39
Appendix II: Throughput times
Table A: Average throughput times 2015 & 2016 2015 2016 Average Time until Triage (Min:Sec) 21:16 14:01 Average Time until seen (Hour:Min) 01:30 01:45 Average Time until Departure (Hour:Min) 04:58 05:44 Table B: Average throughput times per weekday Weekday Average of Time until Triage (Min:Sec) Average of Time until seen (Hour:Min) Average of Time untildeparture (Hour:Min) Average no. of arrivals Monday 15:50 01:52 06:19 142 Tuesday 14:24 01:37 06:10 137 Wednesday 13:17 01:29 05:46 135 Thursday 12:41 01:28 05:22 135 Friday 12:19 01:29 05:18 136 Saturday 13:59 02:01 05:24 143 Sunday 15:29 02:17 05:50 148 Total 14:01 01:45 05:44 139 Figure A: Average throughput times vs average number of arrivals 100 110 120 130 140 150 160 00:00:00 01:30:00 03:00:00 04:30:00 06:00:00 07:30:00 09:00:00
Monday Tuesday Wednesday Thursday Friday Saturday Sunday
41
Appendix III: Interview set-up
The following set-up was followed (loosely) during the different interviews with both ED physicians as operational staff. • Short description on job specifics and responsibilities • During the week and day different numbers of people arrive at the ED. Do you adapt the process to these arrival flows and if so how? • When looking at the figure above, the time until triage is indeed higher on a Sunday and a Monday, when more patients come in. What other factors might influence this? • The time until seen seems quite directly linked to the number or arrivals, with higher times during Saturday to Monday, are there other factors causing these times to increase than just the arrival numbers? • Remarkably, time until departure is much lower than on a Saturday compared to Sunday and Monday when many patients arrive as well. Could you explain this? • The time until departure is higher when arriving on a Tuesday, could you explain why the time until departure is lower when arriving on a Saturday than for example on a Tuesday? 125 130 135 140 145 150 00:00:00 01:30:00 03:00:00 04:30:00 06:00:00 07:30:00 09:00:00Monday Tuesday Wednesday Thursday Friday Saturday Sunday
42 • When looking at the number of patients arriving per hour and the different throughput times, you can see that the triage times increase when the number of arrivals increases, however the time until seen and departure is a lot lower during the day. Could you explain that? • The figure below shows that even though Wednesday, Thursday and Friday have approximately the same average of patients coming in. The variation of the actual numbers is much higher on a Thursday, what do you think could explain this? 0 2 4 6 8 10 00:00:00 01:12:00 02:24:00 03:36:00 04:48:00 06:00:00 07:12:00 08:24:00 00: 00 01: 00 02: 00 03: 00 04: 00 05: 00 06: 00 07: 00 08: 00 09: 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 Nu m be r of p at ie nt s ar riv in g Ti m e in h ou rs Time of the day in hours
Average number of patients arriving per hour
Average of Time until Triage Average of Time until seen Average of Time until departure Average number of patients arriving per hour 80 90 100 110 120 130 140 150 160 170 180Monday Tuesday Wednesday Thursday Friday Saturday Sunday
43 • Does it happen often that besides this variation in arrivals, there is a sudden surge or arrivals? o How does this influence the operations of the ED. Are there any changes being made to how patients are seen? o How does this influence the work in the Operations Centre? How does a surge change the way of working? • When looking at the in- and output flow in the ED figure, how would you explain the graph? Note: these are total output numbers during the day, at a certain point of time. • Do you see a difference is how surges are handled on both busy and quiet days? 0 20 40 60 80 100 120 140 160 180 200 220 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 Cu m ul at iv e nu m be r o f p at ie nt s Time
In- and output flow on a surge day
44
Appendix IV: interviews
Interview 1Job and responsibilities: General manager Operations Centre. Managing the teams that manage the flow in the hospital <Introduction> During the week as well as during the day there is always a variety of patients arrive. How do you cope with this, how do you respond?
45
Is it true that during the weekend there is less people discharged, which causes the blockage? There are less patients discharged on Saturdays and Sunday. We choose the targets for the number of discharges. One: we don't have the same number of staff in discharging. Also, we don't have the same support services in the hospital, as well as out in the community. This means that the system gets clogged up during the weekend which strongly effects flow in the entire ED right after. On a Saturday, the number of people arriving is higher, yet the times different throughput times are lower than on Tuesday. Historically we have more people going home on a Friday, so extra space opens up in the Emergency Department as well at the Friday evening. This gives you a better start on a Saturday morning. And, the services that are open in the community and in the hospital to support, are most likely to be still open on the Saturday morning.
This is the same graph, but on an hourly basis. You can see the amount of people coming in at night is relatively low, compared to the day. You see that the triage seems to be linked to the number arrivals, being higher as more people getting in. However, the time until seen and the time until departure do not follow the pattern. Could you elaborate on that? I think there are a couple of things. We still have Pitstop during the day, so the time until seen is really practiced by Pitstop, people come in and go straight to Pitstop. The blockage doesn't tend to be Pitstop, even if the department gets congested and there is no flow out of Pitstop, most of the consultants will use a couple of the bays in Pitstop to use practically. So, you will be still be seen in time, however this might take a little longer because people are to be moved around. Is there a consultant making decisions during the night? The number of senior decision makers, registrars and consultants, starts to fall after 10 in the evening and falls again after 2am. The senior decision maker will pre-dominantly looking after the sickest patients, so will be in resus. The number of patients seen by a decision makers thus goes down, and goes to a hold to some extent.
The following graph shows the variation of people coming in during the week. You see the average of arrival on the Wednesday and Thursday is the same, and the Friday is close. However, for some reason the variation on a Thursday is much bigger. Can you explain that?
No I can't really think of any reason. This is quite odd really. It might have to do with the closing of GP surgeries. You might want to find out what the target days are. Target days are when all the primary care, the CCG's, the GP surgeries have training. A lot of the GP surgeries close for the afternoon at those moments. If that is on a Thursday it could give you some peaks. However, I think these are usually on Wednesday. It might be that Thursday are before bank holidays, but it would all be guessing.
Does it happen often that besides the regular variation there is a sudden surge of arrivals?
46
come to A&E in the morning, and then afterwards ring they can't get in to work. We certainly see more attendances when the GP surgeries close, but also there is a bit of a handover period for the Ambulance service. They have a handover in the morning and afternoon. Therefore, in a period during the day you have more ambulances available to convey people in.
However, those are relatively stable every day, the football cups will create more of a sudden surge, you didn't prepare for on a daily basis. That is true, that also counts for the surgeries off course. Does it happen often that all of a sudden there is a big surge of patients coming in than you are used to? Sometimes there is no sense to it indeed. Some people say if the sun is shining more people go out and might come in to A&E because they're drinking all day. If there are activities in the community this might cause it as well. How does this influence the operations of the ED. Is there a way of adapting policies to cope with these surges better? If Tesco's have more than 2 people queuing they open another till. We should to, but we don't have the same response here because I would suggest that we are fully powered the full time. However, even if we do require it, we don't have a robust way of handling it. Some hospitals have a way to escalate up, and have a much more robust way of responding. Is there any way you change the way of working here in the ‘Operations Centre’? It would be nice. We do escalate, we do let people know that they're coming. If we got lots of patients to be transferred we'll put in some extra effort and make some transfer teams. But we don't have a robust in being to respond to that really. This is the in- and output flow of November the 21st of November. I show this because there the biggest surge compared to average between 10 and 11. It is a Monday. (explanation of the figure). Sudden after the surge it seems that triage cannot keep up with the number of arrivals anymore. This effect is even bigger for the time until seen. Can you explain this?
