LEAN IN HEALTHCARE - HOW HOSPITALS COPE WITH VARIABILITY ON A DEPARTMENTAL LEVEL

LEAN IN HEALTHCARE -

HOW HOSPITALS COPE

WITH VARIABILITY ON A

DEPARTMENTAL LEVEL

Master Thesis MSc. Technology and Operations management & MSc. Supply

Chain Management

26 januari 2015 Miriam Moes

s1911473

1

Abstract

2

Table of content

2.4 Current research lean in healthcare ... 8

2.5 Research model ... 11

3. Methodology ... 11

3.1 Introducing the hospital ... 12

3.2 Sample selection ... 12 3.3 Data collection ... 13 3.4 Data analysis ... 15 4. Analysis ... 17 4.1 Sources of variability ... 17 4.2 Buffers ... 20 4.2.1 Time buffers ... 21 4.2.2 Capacity buffers ... 21 4.2.3 Quality buffers ... 23

4.3 Current lean practices (and relation with variability) ... 23

4.3.1 Kaizen Board ... 23

4.3.2 Orange belt projects ... 24

5. Discussion ... 27

5.1 Presence of variability ... 27

5.2 Ways to buffer variability ... 28

5.3 How lean initiatives address variability ... 29

6. Conclusion ... 31

6.1 Theoretical implications ... 31

6.2 Managerial implications ... 32

6.3 Limitations and future research ... 32

7. References ... 34

3

1. Introduction

Nowadays, hospitals are under constant pressure to deliver high quality care and to reduce costs (Waring & Bishop 2010; Umble & Umble 2006), which implies that hospitals have to increase their performance. Lean has proven to be an effective management tool to improve operations performance (Hines, Holweg, & Rich, 2009; Shah & Ward, 2007). Although lean originated in the manufacturing industry at Toyota, it is applied in other sectors as well, such as the service industry (Hines et al. 2004). A more explicit example of the service sector, is the healthcare sector. A meta-analysis from Mazzocato, Savage, Brommels, Aronsson, & Thor (2010) of lean in healthcare reveals that lean may benefit the quality of care, the efficiency of care, and the reduction of mortality rates. Nevertheless, a thorough understanding of how lean principles influence healthcare seems lacking (Joosten et al. 2009; Mazzocato et al. 2010), especially on a system wide level (Radnor et al. 2012). This research contributes to the lean knowledge base, by addressing how lean initiatives in healthcare are related to established lean principles.

Lean manufacturing originated in Japan at Toyota and focusses on continuous improvement by reducing waste, such as overproduction and inventory (Womack & Jones 1996). Within this thesis, the focus of lean is on the reduction of variability. As suggested by Hopp and Spearman (2004, 2008), the reduction of variability will reduce the amount of buffers needed (time, inventory and capacity buffers) and as a result the process’ performance will be improved. The fact that the reduction of variability will increase the productivity of a process, is also explained by the ‘Theory of Swift, Even Flow’ (Schmenner 2009; Schmenner 2004; Schmenner & Swink 1998). Another motivation to study the reduction of variability in healthcare is given by Drupsteen, Van der Vaart, & Van Donk (2013), who state that future research should be aimed at the reduction of variability in hospital operations. On contrast, many studies about lean in healthcare tend to focus on sources of direct waste and omit variability reduction (Radnor et al. 2012; Mazzocato et al. 2010; Joosten et al. 2009; Poksinska 2010).

4

approach has proven to be effective in operations management (Hopp & Spearman 2004; Schmenner & Swink 1998; Schmenner 2004), it is worthwhile to investigate its merits in a healthcare setting. Lastly, Radnor and Osborne (2013) state that further research is needed to explore the end-user value as the key performance measure of public services, instead of internal efficiency. In this study the effect of lean initiatives is studied on the existence of variability and buffers.

This leads to the following research question: (How) can lean initiatives reduce variability in healthcare? The research question will be answered by comparing theory with practice via the following sub questions: (1) Which sources of variability exist? (2) How are these sources of variability buffered? (3) How are these addressed with the deployed lean initiatives? Considering the explorative nature of this research, a multiple case study is conducted among two different medical departments at a regional hospital. This will provide rich data that will help to create an objective view of the application of lean in healthcare. In addition to its theoretical contributions, this study also has a strong relevance for practice, as actual and frequent sources of variability in the studied environment will be identified.

The paper is structured as follows. In the next section the theoretical background will be presented, in which a research model is developed based on relevant literature. This is followed by an explanation of the methodology. The fourth section lists the results. Afterwards the results are discussed as well as the limitations of this research. Lastly the theoretical implications and managerial implications are presented.

2. Theoretical background

This section starts off with a short overview of lean, after which the roles of variability and buffers are discussed. Then the current research in lean is described, this section is concluded with the presentation of the research model.

2.1 Lean

5

most common approach to lean (Hines & Rich 1997). Although the reduction of the seven wastes is a very common approach to lean (Shah & Ward 2007; Shah & Ward 2003; Hines et al. 2004), this study will take a different perspective. In this study the roles of buffers and variability within lean are taken into consideration. According to Hopp & Spearman (2004), variability is an important source of waste that is often overlooked. Hines et al. (2004) support this by stating that a shortcoming of lean is its lack of ability to cope with variability, such as variability in demand. Furthermore, Shah & Ward (2007) confirm that the reduction of variability is at the core of eliminating waste. Not all waste in a system can be reduced unless firms address variability at the same time (Shah & Ward, 2007). Hence, the reduction of variability (mura) is a prerequisite for the reduction of waste (muda). In order to cope with variability buffers need to exist, which will be elaborated on later in this section.

According to Hopp and Spearman (2004), the continuous improvement of lean encompasses the following steps. First, all obvious waste in a system has to be removed. Second, all inventory buffers have to be replaced by capacity buffers, since inventory buffers hide problems in manufacturing. The next step is to reduce variability in a system. When variability is reduced, capacity buffers can be lowered. When variability is reduced, the continuous improvement ‘cycle’ starts over.

2.2 Variability

The effect of variability on process effectiveness is explained by the ‘Theory of Swift, Even Flow’. The ‘Theory of Swift, Even Flow’ states that the more swift and even the flow of materials and information through a process is, the more productive that process will be (Schmenner 2009; Schmenner 2004; Schmenner & Swink 1998). Flow is highly dependent on the variability of the process. The greater the variability is, the less goods flow through the systems which negatively affects a systems performance (Schmenner & Swink 1998). The variability approach to lean is interesting for healthcare, because a large part of lean healthcare literature focusses on continuous flow of patients (Poksinska 2010).

6

Variability is something that occurs in (almost) all operations, both manufacturing and service operations. In hospitals, examples of sources of variability are the arrival of patients (flow variability), the degree of patients’ illness (clinical variability) or the delivery of care by medical practitioners (professional variability) (Litvak & Long 2000). In essence, the presence of variability is a given, which makes the question how to cope with variability. Usually buffers are needed to cope with variability, which will be discussed next. The first step, after identifying variability, is to classify the variability. In their study concerning quality in healthcare, Litvak & Long (2000) distinguished two types of variability: natural and artificial variability. Natural variability occurs naturally and is a result of differences in patients and medical staff (Joosten et al. 2009; Litvak & Long 2000; Kolker 2008). Examples of natural variability are the above described clinical or professional variability. Artificial variability on the other hand, is related to controllable and non-random factors, such as scheduling (Litvak & Long 2000; Kolker 2008) and is the result of the way a system is designed (Joosten et al. 2009). The previously described study by McManus et al. (2003) reveals that scheduled surgeries have a greater impact on variability compared to emergency surgeries. Another effect of artificial variability can be found in the discharge of patients. It might be that the discharge of a patient is delayed due to poor planning for the care patients need after their discharge (Kim et al. 2006). This study emphasizes the need for thorough understanding of how to handle artificial variability in hospitals, since this is caused by own rules and choices and can therefore be influenced.

7 Sources of variability Explanation Found in Natural variability Arrival variability of patients / flow variability

Patients usually arrive in a random fashion with different means and standard deviations

(Litvak & Long 2000; Villa et al. 2008; Noon et al. 2003; Gupta & Denton 2008)

Clinical variability The degree of patients’ illness (Litvak & Long 2000) Professional

variability

It will take different medical practitioners different durations to provide the same care.

(Litvak & Long 2000; Mazzocato et al. 2010)

Bed availability The number of beds available for care

(McManus et al. 2003; Kolker 2008; Litvak & Long 2000) Artificial variability

Planning of patients

The planning of elective patients (Litvak & Long 2000)

New techniques New techniques (Litvak & Long 2000)

Table 1 - Sources of variability

2.3 Buffers

8

current pressure to reduce healthcare expenditures (Waring & Bishop 2010). By reducing variability, buffers van be minimized, and hence money can be saved.

Besides these three agreed upon buffers, a fourth buffer type is suggested as well by Hopp et al., (2007). The authors state that quality can be a buffer as well. According to the authors, most of the performance evaluation models in operations management assume that tasks possess standardized completion criteria. However, contrasting to manufacturing, in services one can decide how much time to allocate to one specific task (Hopp et al. 2007). By adjusting the amount of time spent on one job, workers can cope with variations in their work load. These adjustments are seen as a quality buffer.

Since research proves that the reduction of variability has a great impact on performance and variability has to be buffered, the definition of lean that will be used here is: Production of goods or services is lean if it is accomplished with minimal buffering costs (Hopp & Spearman, 2008; Hopp & Spearman, 2004). Coming back to the idea of continuous improvement (Hopp & Spearman, 2004), it becomes apparent that the second step (substitute inventory buffers for capacity) is not relevant in this study. Hence, time buffers should be substituted for capacity buffers, as a time buffer means patients have to wait for treatment.

2.4 Current research lean in healthcare

Before going in to depth about lean in an healthcare environment, is important to be aware of the differences between service industries, such as healthcare, and manufacturing. The main difference between service operations and manufacturing operations is that within the first production and consumption occur at the same time while in manufacturing this can be at different moments in time (Radnor & Osborne 2013). Another example of this difference is that in manufacturing reducing labour hours per manufactured product can increase efficiency, whereas the reduction of capacity (of employees), can lead to reduction of quality of the service offered (Radnor & Osborne 2013). Furthermore, healthcare is designed to be capacity led, and thus the ability to influence demand is limited (Radnor et al. 2012). The influence of (variability in) demand however, is seen as an integral part of lean in manufacturing (Hopp & Spearman 2004; Shah & Ward 2007; de Treville & Antonakis 2006).

9

from occurring. The downside of focussing on single processes is that the problem tends to shift to other processes (Joosten et al. 2009). Furthermore, lean at itself is an holistic approach, and thus concerns an entire organization. A limited application of lean, focussing on only one department or one tool, might hinder optimal lean outcomes (Radnor & Osborne 2013). Additionally, Curatolo et al. (2014) claim that the majority of the lean in healthcare studies lack methodological support. These statements reinforce the need for an objective view of lean in healthcare.

Although it is stated that the current research on lean in healthcare might lack methodological support and might be focussing on small aspects, many studies have made useful contributions regarding variability in healthcare. For example, Drupsteen et al. (2013) found that cross-departmental planning reduced variability, and subsequently increased patient flow performance (among other factors). The fact that the reduction of variability increases flow performance is supported by the findings of Chand et al. (2009). In their simulation study, the authors show that a change in the appointment system will reduce variability significantly, and as a result 37% more patients could flow through the process. Furthermore, McManus et al., (2003) found that the demand for intensive care services is extremely variable, particularly among patients that undergo scheduled surgery. This high variability results in high utilization of the process. As a result of that high utilization, some patients are placed off-service or denied access to the hospital. The effects of variability and elective surgery on bed availability are also demonstrated by Kolker (2008), in the specific study of an intensive care unit, a maximum of five elective surgeries should be planned per day. This constraint reduces the chance that no beds are available for incoming patients. These examples show that variability has a big impact on performance. Hence, variability is worthwhile to study, especially from a lean perspective.

10

As mentioned throughout this paper, lean is often used to increase performance. There is however no agreed upon definition in what constitutes lean performance. For example, Joosten et al. (2009), Fillingham (2007), and Radnor et al. (2012) state that the performance of lean in healthcare can be found in the value that lean tools create for the patient. Poksinska (2010) supports this statement by saying that customer/ patient value is at the core of improved processes in healthcare. Additionally, the majority of the articles about lean in healthcare revolve around ‘processes’, ‘value stream’, and ‘continuous flow’(Poksinska 2010). Table 2 shows the performance measures of selected articles, how often they occur, and what cases were used for the different studies. Strikingly, none of the studies had a holistic view, which is considered to be essential in lean healthcare studies (Radnor & Osborne 2013; Joosten et al. 2009; Poksinska 2010; Mazzocato et al. 2010). This holistic view implies that performance has to be optimized throughout the entire system, not only at different departments (Poksinska 2010). Furthermore, most of the positive results are found in out-patient clinics, especially in radiology and emergency departments.

Performance measure # Articles Out-patient

Ward Pathway Other Not found

Decreased overall time that patients spent on care 8 4 - 1 1 1 Increase patient throughput 5 4 - - - 1 Reduced number of errors and incidents

7 3 1 - 3 1

Reduced waiting times 4 2 - - - 2

Increased patient satisfaction 5 3 1 - 1 - Increased employee satisfaction 5 3 1 - 2 - Reduction of overtime hours 2 2 - - 1 - Decreased inventory costs 2 1 - - - 1

Reduction in travel time (patient and staff)

3 2 - - 1 -

11

2.5 Research model

This study aims to find out the impact of a variability approach to lean in healthcare. As described throughout the theoretical section, the existence of variability in a process creates the need for buffers. These buffers then have an effect on performance, as shown in Figure 1. The first step is to find out which sources of variability exist, and how these sources of variability are buffered. The second step is to see how these buffers are reduced through lean initiatives. In other words, the aim is to see how lean initiatives can reduce variability, which reduces the amount of buffers needed which has a positive effect on performance, as shown in Figure 1.

Figure 1 - Conceptual model

Building on this, the need arises to find out how the presence of buffers affect performance. As previously stated, the implementation of lean practices in manufacturing is often related to operational performance measures, such as improvement in labour productivity and quality, reduction in customer lead time, cycle time and manufacturing costs (Shah & Ward 2003; Dal Pont et al. 2008). Within healthcare, most studies about lean are focussed around ‘processes’, ‘value stream’ and ‘continuous flow’(Poksinska 2010) as described previously.

3. Methodology

In this section the methodology applied in this research will be described. To answer the research question, a multiple case study was conducted. A case study is a methodology to ‘develop theory inductively by recognizing patterns of relationships among constructs within and across cases and their underlying logical argument’ (Eisenhardt & Graebner 2007). The aim of this research was to get insight in the use of lean initiatives in healthcare and how this relates to variability and the presence of buffers. Since the understanding of lean in healthcare is in its early stages, a case study is best suited (Karlsson 2009). Because of its explorative

12

nature, a case study will provide insights in contextual factors (Voss et al. 2002). In order to improve the generalizability of the conclusions, external validity, and to guard against observer bias, multiple cases were adopted (Voss et al. 2002; Karlsson 2009). Different data sources were used which allows data triangulation (Karlsson 2009).

3.1 Introducing the hospital

The hospital that was studied for the purpose of this research is a teaching hospital, with a total amount of beds of 643. At the hospital, lean is applied on different levels which can be seen in Figure 2. Different, but related departments, together form a Result Responsible Unit (RRU). Across these RRU’s, black belt projects are deployed. Within the RRU’s, there are green black projects. Both the black belt and the green belt projects, are driven by board of directors, data driven and are closely related to six sigma. Next to these projects, there are also orange belt projects. Orange belt projects are on departmental level, and are team based driven instead of driven by the board of the directors.

Figure 2 Structure of improvement projects within hospital

3.2 Sample selection

In order to collect the data for the study, two departments (cases) were selected; a dialysis clinic, and a pulmonology (lung) ward. Multiple cases were selected to ensure a mix of

RRU

RRU

RRU

Department

Department Department Department

Department

Green belt project Black belt project

Driven by board of directors

Team based driven

Department

Orange belt project

Department Department

13

current lean practices. Both an outpatient clinic and a hospital ward have been selected, since these serve different types of patients and are organized differently. Furthermore, these departments were selected because they have the most experience with lean. Other departments have less experience with lean, and therefore it might be that not all lean initiatives have been deployed fully yet. Another motivation for studying both an out-patient clinic and a ward is that these are organized differently, which may result in different outcomes. On the outpatient clinic, the processes are to a large extent controllable, whereas the controlling processes on the ward will be difficult. This is in line with the suggestions of Eisenhardt (1989), who states that it is best to choose polar types as cases for case study research. Because the departments are different and different outcomes may be expected, theoretical replication is applied (Karlsson 2009).

The (outpatient) dialyses clinic has a capacity of 18 beds, and works together closely with other regional dialyses clinics. Because the clinic under study handles the most complex patients of the region, emergencies from the entire region have to be treated at the outpatient clinic. Patients are dialyzed during three shifts, one starting in the morning, one in the afternoon and one starts in the evening. The procedure for dialysis is characterized by standardized processes. The dialysis process takes four hours on average in the first two shifts, people who start in the evening are dialyzed throughout the night because their treatment is longer than the regular four hours. The high levels of standardization should make the identification of variability in this outpatient clinic (fairly) easy.

The pulmonary ward has a capacity of 28 beds, and treats patients with a wide variety of lung related problems. However, once a year capacity is increased to 32 beds because of higher demand during winter. The ward serves a wide variety of patients, and therefore identifying variability will be difficult.

3.3 Data collection

The data is collected at the two selected departments. In order to increase the reliability of the data, and content validity, multiple data sources were used which enhances data triangulation (Karlsson 2009). Semi-structured interviews with unit managers1 and care coordinators2 were conducted, to find out which sources of variability exist, how these are buffered and, how these are addressed by the lean initiatives. An overview of the interviews can be found in

1 _{The unit manager is responsible for the entire department, management of employees on the long run, budgets}

etc.

14

Table 3. During the interviews an interview protocol (Appendix A) was used to increase reliability and validity (Voss et al. 2002; Karlsson 2009). The interviews were recorded (audio), and notes were made during the interviews. Following the 24-hour rule, the recordings of the interviews were transcribed within 24 hours after the interview took place (Eisenhardt 1989). Furthermore, open discussions were held with orange belt trained nurses to obtain more in depth knowledge about the orange belt projects that they conducted.

Department Interviewee function Experience with lean

Dialysis (P1) Unit manager Green belt trained

Dialysis (P1) Care coordinator Orange belt trained

Lung ward (W1) Unit manager Black belt trained

Lung ward (W1) Care coordinator No training in lean

Table 3 - Interview details

Besides interviews, data was also collected through observations, and on-site visits. During the observations, a nurse was followed during the first part of her shift to get insights in the nurses work routine. According to Eisenhardt (1989), empirical observations make the theory-building process more powerful. Furthermore, two orange belt projects were studied within each department. The reason for studying orange belt projects, is that they are on departmental level only, and thus help to get insights in how variability is addressed on departmental level. A short overview of these projects can be found in Table 4. The orange belt trained nurses provided data about their problem definition, how they approached the problem, how they made measurements, and (preliminary) results.

Department Topic

Dialysis (P1) Preparing materials for patients

Dialysis (P1) Why does treatment for some patients start late? Lung ward (W1) Discharging patients

Lung ward (W1) Treatment with Xolair (medication)

Table 4 - Overview analysed Orange belt projects

15

later meetings. The researcher was present during these meetings for a period of two months, and made notes and took pictures. This provided insights in the day to day problems employees (mainly nurses) encounter.

3.4 Data analysis

In order to answer the question how lean practices are used in healthcare, three sub-questions have to be answered which are: Which sources of variability and buffers exist in the studied departments?; How are these sources of variability buffered?; How do lean initiatives (Kaizen board and orange belt projects) relate to the reduction of variability and buffers?

In order to answer the questions, the data was analysed following the steps of Miles & Huberman (1984). The first step is data reduction. The data reduction was done by transcribing the data in to useful quotes, which were labelled in related categories and sub-categories. An overview of the coded categories and sub categories can be found in Table 5. The definitions from the variability sources and buffer types were derived from the theoretical section. The codes regarding the lean initiatives on the other hand, were derived from the interview transcripts.

Categories and subcategories Variability Examples from interviews

 Arrival variability “There is definitely variability in when patients arrive. On this

department patients have more complaints during winter than during summer”.

 Clinical variability “Patients suffer from chronical illness, but sometimes they arrive

more ill than other moments and thus need more care”.

 Professional variability “I think the moment the doctors do their rounds is an important

variation, they come on different days on different times”.

 Bed availability “Sometimes we do not have enough beds. Currently seven or

eight of our patients are treated outside our department”.

 Planning of new patients “The planning of new patients is something that a separate

department plans, we cannot influence this”.

 Experience employees “The experience of employees varies. There are 60 employees, all

with different experiences and problems”.

Buffers

16

of half an hour before treatment can start”

 Capacity buffer “We have an additional nurse who can help the acute patients”.  Quality buffer “If it comes to quality, people do not always act up on the agreed

levels of quality”.

 Flexibility “We use the traffic light model. When a nurse has a green magnet

next to her name, she can help somewhere else”.

Lean initiatives

(derived from interviews)

 Efficiency “The goal of the Kaizen board and orange belt projects is to

lower the inefficiencies in a process”.

 Ownership “The biggest effect of the Kaizen board is that nurses have the

feeling that the problem, and a possible solution, is within their own influence”.

 Clarity “One of the projects created clarity about who is responsible for

the preparation of materials for treatment”.

 Team based “The power of lean is that the solutions you come up with, fits

with your team. Our solution do not always fit another team as well”.

Table 5 - Category definitions

17

4. Analysis

Within this section, the sources of variability will be presented first. Afterwards, the types of buffers that are used to cope with that variability are discussed. Lastly the lean initiatives and their relationship with variability are elaborated on.

4.1 Sources of variability

There is a clear distinction in literature between the different sources of variability, which can be either natural or artificial. However, the interviews showed that one type of variability can have both an artificial, and a natural component. For example, the studied departments experience variability in the arrival of patients. This arrival variability results from the moment when patients are in need of treatment (natural), but also from when they are relocated from another department. The unit manager of the pulmonary ward commented: “The variability in the arrival of patients has to do with when a patient enters the hospital […]. Often the emergency department “batches” patients, which gives peaks in patients arrival at the department”. This statement shows that due to the actions of another department, many patients can arrive at the ward at the same time. These peaks in arrivals can cause increased levels of work pressure and subsequently result in lower quality of care, as mentioned by the care coordinator: “Nurses are, especially during the afternoon, very busy. They have to a lot of work within (the last) two hours, so I think having more patients will definitely decrease the quality of work.” The arrival of patients on the outpatient clinic is more stable, as patients arrive according to a planned schedule, and it is known beforehand which patients arrive. However, there is variability in their arrival for their planned appointments. Most patients arrive with taxis, and these taxis are allowed (by the insurance companies) to arrive 30 minutes before or after the scheduled appointment time. Furthermore, because this rule is set by insurance companies, this source of arrival variability cannot be controlled by the department.

18

patient, while another patient only keeps you busy for one and a half hour”. As a result of the level of care the patient needs, it can take a nurse longer than 20 minutes to start treatment for a patient which may leave a second patient waiting. The fact that there is many clinical variability on the outpatient clinic can also be found in one of the orange belt projects. In 35% of the measurements, the treatment of a patient started too late because nurses had difficulties with finding the vein to which the dialysis machine has to be connected. Even though the ward has the same kind of clinical variability, here variability is seen a characteristic of the nursing job, and the interviewees indicated that they do not have additional capacity to cope with this variability. According to theory however, this cannot be true since all variability in a system has to be buffered.

19

The availability of beds also varies between the two studied departments. At the dialysis clinic, the group of patients that are treated remains relatively constant. Moreover, all patients need treatment every two days, and the time they arrive is mostly set. As a result there is very little variability in the availability of beds. On the ward on the other hand, this is highly variable. Throughout the year, more patients come in during winter than during summer. Furthermore, on a day to day basis, the availability has to do with the number of discharges and admissions. During the day that observations were done, seven patients were discharged, while at other days only one are two patients were discharged. This is mainly natural variability and hence cannot be controlled. However, the previously described professional variability can also affect the availability of beds.

The planning of new patients is different between the two studied departments. Within the dialysis clinic, all patients are planned. As stated by both the care coordinator and the unit manager, patients (both acute and planned) that need to be treated on different departments have a big impact on the work of nurses and subsequently on flow. “Sometimes patients that are too ill, need to be dialyzed on for example the ICU. They can have the same pattern of treatments as the other patients have. For dialysis outside our department, you have to be one on one, so a nurse has to leave the clinic which is a disruption in the logistics of our department”. On the ward, 20 percent of the patients is planned. These planned patients mostly come in before they undergo surgery. “A large part of the patients that undergo surgery, need to go to the ICU after their surgery. But if there are no ICU beds available, it is not possible to plan these patients (and they cannot undergo surgery)”. This statement of the care coordinator shows that the planning of patients is not only based on their own capacity, but on the capacity of other departments as well. It can be that patients are already admitted on the ward, when it becomes clear that there are no ICU beds available.

Variability in the experience of employees is found on both departments as well, and is something that occurs natural and cannot be controlled. However, there are ways to minimize the effect of this variability. For example, on the dialysis clinic less experienced nurses treat less difficult patients than nurses who are more experienced.

20

variability, buffers need to exist. Table 6 shows for each type of variability, how they affect performance and how they are buffered. The next section provides more detailed information about the buffers being used. In the last section of the results, the lean initiatives will be discussed as these should be aimed at the reduction of artificial variability.

Variability Effect on performance

Natural variability Artificial variability Throughput time Quality Employee satisfaction

W OP W OP W OP W OP W OP Arrival variability Yes *, **, ***, **** Acute patients *, **** Patients from ED**, ***, **** Taxis are allowed to shop up 0.5 hour early/ late **, ****

- Negative Negative Negative Negative Negative

Clinical variability Yes Yes*,**,* **,**** - - - Can be negative - Can be negative - Neutral Profes-sional variability

Yes**** Yes**** Rounds of doctors** ** Rounds of doctors**** Can be negative Not directly Can be negative Negative Can be negative Negative Bed availability Yes*, ****

Very little - - Not

directly - Not directly - Not directly - Planning of new patients

- - Yes** Yes Not

directly - Not directly - Not directly - Experience employees

Yes**** Yes**** - Not

directly Not directly Not directly Not directly Not directly Not directly

* Capacity buffer ***Quality buffer W = Ward

** Time buffer **** Being flexible OP = Outpatient clinic Table 6 - Overview variability and buffers

4.2 Buffers

21

the three buffer types. As could be expected, at both departments buffers are present to cope with variability.

4.2.1 Time buffers

At first, all interviewees stated that patients do not have to wait for their treatment, indicating an absence of a time buffer. However, throughout the interviews it became clear the dialysis clinic uses a small time buffer. As stated by the care coordinator of the dialysis clinic “Sometimes a nurse is not ready for a new patient yet, and the patient has to wait for fifteen minutes up to a maximum of half an hour before treatment can start”. With this time buffer, different types of variations are buffered. As mentioned previously, clinical variability may result to a nurse being behind schedule. A time buffer can be used to cope with this variability, which results in longer throughput times of patients. Furthermore the fact that patients are allowed to show up 30 minutes later than there appointment time can also result in the fact that a nurse is behind schedule, and thus a time buffer might be used. However, due to the nature of the dialysis clinic, the treatment of a patient cannot be postponed over a longer period, as patients need to be treated every two days. What is also interestingly to note, is that the dialysis process itself creates a small time buffer. The duration of one treatment is typically around four hours, and the highest workload for the nurses is when patients have to start their or have to finish their treatment. In between these two phases, the workload is lower and thus nurses have time to do additional tasks such as the distribution of medicines. Currently the distribution of medicines is done at the end of the process, but the unit manager stated that they are planning to this earlier to spread the workload.

On the lung ward time buffers are barely used. When patients need to be admitted to the ward, they have to be admitted, so it is not possible to buffer the arrival variability. A small exception is mentioned by the unit manager of the ward: “Sometimes we use the ED as a buffer. By letting a patient wait on the ED for half an hour, we can prepare for the patients arrival”.

4.2.2 Capacity buffers

22

described clinical variability, by starting treatment for a next patient to prevent a time buffer. Furthermore, this nurse can absorb the variability of acute patients that need to be dialyzed at other departments.

The capacity buffers at the ward are organized differently. The variability in the arrival of patients is partly known beforehand. Since more patients come during winter than during summer, the ward has more beds and nurses available during winter. “During winter periods we have more beds than during summer, so you have less employees during summer”. Furthermore, capacity buffers are also being used when variability affects the process of care delivery on a day to day base. Every morning (during working days) the unit managers of all wards in the hospital gather to discuss how busy they are. So if one department has excess capacity in nurses, they can help on other departments where the workload is high. “This whole week we have had help from cardiology and paediatrics, they help with washing because the workload is currently very high”. Furthermore, if all beds are occupied there is an opportunity to admit patients to other departments. If possible, a dedicated unit is opened on another department where one specialized nurse can go to. It should be noted however that this is not additional capacity on departmental level, but rather on organizational level. It might also be argued that the treatment of patients with lung problems outside the pulmonary ward can be seen as a time buffer. So, reasoning from a department level only, the treatment of a patient on another department is actually a time buffer until a patient can go to the pulmonary ward.

23

The dialysis clinic ‘buffers’ the variability in experience similar : “On the heavier units only certified nurses will work, and no students”. Furthermore, on both departments nurses are flexible in the way they plan their work. Each nurse works on projects teams to improve or clarify certain subjects, for which time is planned to work on. However, when workload is high, nurses do not work on their projects but work with patients instead. This can also be seen as flexible capacity.

4.2.3 Quality buffers

More interestingly, is the fact that in both cases the unit managers admitted that the quality of care is used to buffer high workload. One could argue that it might be that organizations use additional capacity to ensure quality, but the unit manager of the dialysis clinic denied this. She stated: “Sometimes we do not work according to our own quality standards. It is hard to say if quality really diminishes, but it is a different quality than we agreed up on”. A quality buffer does not directly imply that work is done below quality standards, but rather that quality is used to absorb variability. A good example is given by the care coordinator of the dialysis clinic: “When the workload is high, everyone delivers the basic levels of care. Sometimes you have to give in on additional things for patients”. Another example is given by the care coordinator of the ward: “Sometimes it happens that antibiotics are not checked twice because it is busy, while medications should be checked twice”. This example shows that rules are sometimes undermined in order to get work done. Similar things happen on the dialysis clinic. It happens that patients help nurses with their work, which is not according to the rules. However, this can save up to ten minutes, so a patient can leave earlier.

4.3 Current lean practices (and relation with variability)

4.3.1 Kaizen Board

24

Of the 12 issues written down on the Kaizen board at the outpatient clinic, six were related to ambiguities in the work of nurses, four were related to the quality of care, and two were related to variability. However not directly to the variability as described in the theoretical section. The first issue written down related to variability, is that nurses have to open doors in the evening for taxi driver who come to pick up patients. As this is during the time patients have to finish their treatment, this creates variability in the process of care delivery as well. Up to the point the meetings were attended, no solution was found for this problem. The second source of variability was that tanks, containing fluids that need to be used during dialysis, cause disruptions. Occasionally the tanks have to be replaced with new ones, which means that the treatment is delayed, which results in variability in the process. The solution was that tanks that had fluids levels below a certain threshold had to be replaced before the treatment starts. This can be regarded as preventive maintenance, which is also used at Toyota to reduce variability.

During the two months, 22 issues were written down on the Kaizen board at the ward. Of these issues, 12 were related to quality, eight to ambiguities in working procedures, and two were (indirectly) related to variability. The first variability related issue was that nurses had difficulties in answering questions from patients because the transfer of information between different nurses was incomplete. As a result, nurses had to ask other nurses or doctors to answer the patients’ problem, which creates variability in process of care delivery. The solution was to write something in the weekly newsletter to communicate to the nurses that this transfer has to be complete. However, there is no proof that this issue did improve after an item in the newsletter. A second issue related to variability was that a certain injection that has to be mixed from two ingredients was available in a pre-mix. In the old situation, two nurses had to prepare the medicine together since medication has to be double checked, which took eight minutes on average. The pre-mix thus reduces variability since it takes two nurses eight minutes to prepare the injection. Although this type of variability has no direct link with the main process, it makes the available time for patients less variable, and subsequently has an indirect effect.

4.3.2 Orange belt projects

25

insights in problems you cannot solve immediately”. So a more structured analysis of the problem is needed for an orange belt project.

Within the dialysis clinic, a few orange belt projects have been deployed. One of these projects was about preparing materials for treatment. Before a patient’s arrival, boxes are prepared with the right equipment (artificial kidney, needles etc.). The motivation for this project was that it frequently happened that equipment that needed to be in a patients’ box, was not collected correctly, and this was only discovered when the patient was present. As a result, nurses had to check which materials were wrong and which were missing, and get the correct equipment. This was a disturbance for the process because less time was available for the patient, and as a result there was variability in the process. Due to the project, there now are clear agreements who is responsible for what which part of preparation, so mistakes are less likely. Although the measurements done by the orange belt trained nurse did not show improvements (before 32% of the boxes had errors, afterwards this was 41%), the unit manager stated: “It is changed, there are less VIM-reports3_{, and there are less complaints that}

boxes are prepared incorrectly”. This example shows how variations in a sub-process (preparation of materials) can have effect on the variations in the main process (dialysing patients). However, the fact that the measurements performed by the orange belt trained nurse showed an increase in the amount of mistakes, shows that it takes more than an improvement plan to achieve improvements. An explanation can be that it takes a while before new procedures or ways of working are embedded in the work routine of nurses. Another project was related to variability as well. This project was concerned with why it sometimes takes too long (longer than the available 15-20 minutes) to start treatment of a patient. The analysis revealed several reasons why treatments started too late and the fact that in 2.6% (13) of all the measurements (506) an echo was not available, resulted in the purchase of an new echo. This is an example of how direct waste, waiting for an echo, can cause variability in the process. Moreover, this orange belt project provided a solid argumentation for buying a new echo, as patients’ treatment frequently was delayed due to the fact that there were not enough echo’s available.

Furthermore, the unit manager stated that projects are planned on how to deal with clinical variability. With this project the aim is to see which levels of care (clinical variability) can be identified and how these can be planned. Additionally, there also projects that are not related

26

to variability, but rather to quality and the clarity of information. One project is aimed at haemodialysis, which is a type of dialysis where patients are dialyzed at home. When these patients have questions they can call the dialysis clinic for answers. It turned out that many nurses did not know how to answer all the questions, so the aim of this project was to train all nurses in providing the right information to patients. Another project that is not related to variability is the double check of medication. This double check is based on rules set by the hospital and the project seeks out how to make sure all medicines are double checked. Lastly, the unit manager of the dialysis clinic stated that she wants to do a new orange belt project (or green belt project) on the division of time during the treatment of one patient. Currently, medicines are distributed at the end of a treatment. As described previously, there are many sources of variability than can result in delay, especially at the start and end of a treatment. When medicines are distributed earlier, the workload will be more spread throughout the treatment and subsequently the unit manager expects that the impact of variability will be less.

27

A second orange belt project on the ward was concerning patient discharge. The goal was mainly to provide the right information to patients and their relatives before, and after their discharge. For example, how prescribed medication should be used. In order to find out if patients were satisfied with the information that was provided, patients are called one day after their discharge. This project does however only focus on quality, and does not relate to variability. In previous sections, professional variability has shown to have an impact on the discharge of patients as well. This professional variability was however not covered in this project.

5. Discussion

This study used a multiple case study to examine lean initiatives at departmental level in a hospital. Adapting multiple cases improves the generalizability of results (Voss et al. 2002; Karlsson 2009). Although the studied cases are different in their nature, interesting findings were derived from this study.

Important to note is that not all nurses had complete understanding of lean. For example, the care coordinator of the pulmonary ward could not indicate what the overall goal of lean on her department was and during the observations, one nurse asked if lean was a new computer programme. As stated by Kim et al. (2006), everyone in an organization should understand what lean is and believe why it will benefit the organization. Therefore it can be that the benefits of lean are suboptimal, since not all employees have that required understanding of lean.

5.1 Presence of variability

28

pooling. Lastly, there are sources of variability that should be labelled as artificial according to theory, but cannot be controlled. For example the fact that patients are allowed to show up half an hour after their appointment time is due to regulations of insurance companies. Regulations are seen as artificial, but the hospital has no control over this source of artificial variability.

5.2 Ways to buffer variability

The second research question was aimed to find out how the different sources of variability are buffered, since all variability in a system has to be buffered (Hopp & Spearman 2004; de Treville & Antonakis 2006). The studied departments use different types of buffers. First, both departments use a capacity buffer. At the outpatient clinic this capacity buffer is found in the presence of an additional nurse. At the ward however, the capacity of other wards is used when there are no beds available on the pulmonary ward. One could argue, reasoning from a departmental perspective, that the use of the capacity of other departments can be seen as a time buffer, as patients have to wait for available beds. Moreover, the results show that the capacity often is flexible. A large part of the capacity of the studied department, is in the form of nurses. There is however not a fixed amount of patients they can help within a certain period of time. Rather, nurses that have spare to time help other nurses who experience a high workload.

Furthermore, it seemed that both departments barely used a time buffer when new patients have to be treated. Both departments stated that the time that patients have to wait never exceeds one hour. It might even be that quality buffers are used more often that time buffers, with a note however that a basic care level is always provided.

The last type of buffer that is used it the quality buffer as described by Hopp et al., (2007). The results show that a quality buffer is used to cope with variability. Since the name of a quality buffer can imply that quality diminishes or work is below quality standards, a better name would be a ‘processing time buffer’. This name is more suitable, since it is the processing time that is used to cope with variability. A result of this can be either higher or lower quality, but adjusting the time spent on a patient does not necessarily affect quality of care.

29

care varies as well. This implies that due to the flexible capacity, a quality buffer might be used as well.

5.3 How lean initiatives address variability

Lastly, the last research question was aimed to find out how lean practices address the previously described sources of (artificial) variability. The lean initiatives deployed at the studied departments, did however not cover the aspects of variability. The issues that were found to be related to variability, are not the types of variability that were described in the theoretical section. Rather they influence the available time for patients and subsequently create variability in the core process. Furthermore, most of the issues discussed on the Kaizen board were about uncertainties or quality related issues. This issues can be seen as quick wins as described by (Radnor et al. 2012), or direct waste.

Shah & Ward (2007) stated that the reduction of variability, and the reduction of waste are intertwined. This study showed results that support this statement. One of the orange belt projects was aimed to reduce mistakes in the preparation of materials. Because mistakes were made, all materials had to be checked again which can be considered as defects and thus as a source of waste. This source of waste caused variation in the process of care delivery to patients. So with the removal of a type of waste (wrongly prepared materials), variability can be removed as well. A second example on the dialysis department was that nurses had to wait for an available echo, this source of waste created variability in the core process which resulted in longer treatment times of patients.

30

Mazzocato et al. 2010). However, the improvements made based on the lean initiatives are not communicated between different departments, which hinders this holistic view. According to the unit manager of the ward, this is not shared since the issues within the lean initiatives are focused on specific departments only. What can be a problem for one department does not have to be a problem for another department. However, it might be wise to share the outcomes of relevant orange belt projects or improvement based on the Kaizen board. For example, at the outpatient an orange belt project is done about the double check of medication and how to make sure that medication is double checked. Since the double check of medication has to be done hospital wide, it might be wise to share the outcomes of this project with other departments.

Furthermore, the results showed that performing an orange belt project does not directly lead to improvements. For example, the results of the orange belt project about the preparation of materials on the outpatient clinic, showed an increase in the amount of mistakes. The unit manager stated that it is necessary to remind nurses to work according to plan made based on the project. This indicates that a cultural change is needed as well, which is also stated by Radnor and Osborne (2013).

Lastly, the two studied departments used the Kaizen board differently. On the pulmonary ward, issues were discussed plenary and the improvement approach often was to communicate standard procedures through a newsletter. On the dialysis clinic on the contrast, issues were analysed in more detail and sometimes it took weeks or months before an issue was resolved. Because of the differences in how the departments work with the Kaizen board, there might be differences in the outcomes as well. Moreover, during the time the Kaizen board meeting were attended, there were 10 more issues written on the Kaizen board on the ward than on the outpatient clinic.

31

Figure 3 - Research model

6. Conclusion

The aim of this study was to see how lean initiatives, at a departmental level, can help to reduce sources of variability. At the studied departments, many sources of variability were found. However, these types of variability were not addressed through the applied lean initiatives. The lean initiatives rather focussed on the quality of care, and disruptions in the work process. These disruptions can be variations in sub-processes, but are not directly related to the sources of variability that were discussed within this thesis. Linking these findings to the idea of lean from Hopp and spearman, it can be stated that lean initiatives on departmental level primarily focus on the elimination of direct waste (which is the first step). The reduction of variability is something that occurs on higher levels, for example in this case study through the green and black belt projects.

One of the surprising findings was about the use of buffers. In theory three types of buffers are described: inventory, capacity and time. Besides these three, a fourth form is described as well, namely a quality buffer. The results show that a quality buffer is used to cope with variability. Since the name of a quality buffer can imply that quality diminishes or work is below quality standards, a better name would be a ‘processing time buffer’.

6.1 Theoretical implications

The results of this research provided some useful insights for theory. In current theory, the most frequent discussed buffer types are inventory, capacity and time buffers. Within this

32

research, many sources of variability were discussed as well as how the cases cope with those sources of variability. Counter to expectations based on literature, time buffers are barely used. Furthermore, the capacity buffers that are used are partly flexible. This means that nurses are not always fully dedicated to a patient group, but nurses can help other nurses as well. Moreover, variability is buffered is in a ‘processing time’ buffer as described by Hopp et al. (2007).

Furthermore, this thesis showed that it is hard to control variability on departmental level, as variability often involves multiple departments. For example the flow of patients between two departments, and the variations of that flow, cannot be controlled at departmental level. This means that controlling and (or) reducing variability, is only possible at higher levels in an organization.

6.2 Managerial implications

One of the main findings was that is hard to control variability on departmental level, and that variability should be addressed at higher levels. This is however already the structure of lean practices at the studied hospital. One topic that has strong managerial relevance is the professional variability induced by the doctors. On both departments the interviewees indicated that the way doctors work affects the work of nurses on the departments. On the ward, the discharge of patients is not always communicated correctly. As a result, it happens that patients stay longer than necessary which increases the workload for the nurses. On the outpatient clinic, there is no set moment when doctors come to see their patients. This can counteract the work of nurses as doctors may show up when workload is already high. Because nurses have to be present when doctors visit their patients, the even higher workload might affect patient flow. Therefore it might be useful for management to set up protocols or guidelines for both doctors and nurses, so that their work is more aligned and hence does not affect patients.

6.3 Limitations and future research

33

statement of Poksinksa (2010). He states that for an holistic approach, the impact of actions from one department on other departments should be considered as well.

A second limitation lies in the methodology. First, only two cases have been studied. Although two cases have more generalizable results than a single study, it is recommended to have between four and ten cases. With fewer than four cases it is often difficult to generate theory with much complexity (Eisenhardt 1989). However, with the available time it would not be possible to study four cases in depth. Furthermore, no analytical data was used to prove variability in processes which would have strengthened the results.

Furthermore, the studied departments work with lean for just over a year. The finished projects were part of the orange belt education for nurses. The fact that they do not have much experience yet, can explain why the projects do not consider flow. Lastly, only the nurses take part in the lean initiatives. Since the results show that doctors also have a big impact on the work of nurses, it might be better if doctors were involved as well, to achieve most positive results.

34

7. References

Chand, S. et al., 2009. Improving patient flow at an outpatient clinic: study of sources of variability and improvement factors. Health Care Management Science, 12(3), pp.325– 340.

Curatolo, N. et al., 2014. A critical analysis of Lean approach structuring in hospitals. Business Process Management Journal, 20(3), pp.433 – 454.

Dal Pont, G., Furlan, A. & Vinelli, A., 2008. Interrelationships among lean bundles and their effects on operational performance. Operations Management Research, 1(2), pp.150– 158.

Drupsteen, J., Van der Vaart, T. & Van Donk, D.P., 2013. Integrative practices in hospitals and their impact on patient flow. International Journal of Operations & Production Management, 33(7), pp.912–933.

Eisenhardt, K.M. & Graebner, M.E., 2007. Theory Building From Cases: Opportunities and Challenges. Academy of Management Journal, 50(1), pp.25–32.

Eisenhardt, M., 1989. Building Theories from Case Study Research. The Academy of Management Review, 14(4), pp.532–550.

Fillingham, D., 2007. Can lean save lives? Leadership in Health Services, 20(4), pp.231–241.

Gupta, D. & Denton, B., 2008. Appointment scheduling in health care: Challenges and opportunities. IIE Transactions, 40(9), pp.800–819.

Hines, P., Holweg, M. & Rich, N., 2004. Learning to evolve: A review of contemporary lean thinking. International Journal of Operations & Production Management, 24(10), pp.994–1011.

Hines, P. & Rich, N., 1997. The seven value stream mapping tools. International Journal of Operations & Production Management, 17(1), pp.46–64.

Hopp, W.J., 2008. Supply Chain Science, Longe Grove: Waveland Press.

Hopp, W.J., Iravani, S.M.R. & Yuen, G.Y., 2007. Operations Systems with Discretionary Task Completion. Management Science, 53(1), pp.61–77.

Hopp, W.J. & Spearman, M.L., 2008. Factory Physics, Longe Grove: Waveland Press.

Hopp, W.J. & Spearman, M.L., 2004. To Pull or Not to Pull: What Is the Question? Manufacturing & Service Operations Management, 6(2), pp.133–148.

35

Karlsson, C., 2009. Researching operations management, New York: Routledge.

Kim, C.S. et al., 2006. Lean health care: what can hospitals learn from a world-class automaker? Journal of hospital medicine, 1(3), pp.191–9.

Kolker, A., 2008. Process Modeling of ICU Patient Flow: Effect of Daily Load Leveling of Elective Surgeries on ICU Diversion. Journal of Medical Systems, 33(1), pp.27–40. Litvak, E. & Long, M.C., 2000. Cost and Quality Under Managed Care : Irreconcilable

Differences? The American Journal of Managed Care, 6(3), pp.305–312.

Mazzocato, P. et al., 2010. Lean thinking in healthcare: a realist review of the literature. Quality & safety in health care, 19(5), pp.376–82.

McManus, M.L. et al., 2003. Variability in surgical caseload and access to intensive care services. American Society of Anesthesiologists, 98(6), pp.1491–1496.

Miles, M.B. & Huberman, A.M., 1984. Meaning from Drawing Valid Data Qualitative Toward: Towared a Shared Craft. , 13(5), pp.20–30.

Noon, C.E., Hankins, C.T. & Côté, J., 2003. Understanding the impact of variation in the delivery of healthcare services. Journal of Healthcare Managment, 48(2), pp.82–97.

Poksinska, B., 2010. The Current State of Lean Implementation in Health Care: Literature Review. Quality Management in Health Care, 19(4), pp.319–329.

Radnor, Z. & Osborne, S.P., 2013. Lean: A failed theory for public services? Public Management Review, 15(2), pp.265–287.

Radnor, Z.J., Holweg, M. & Waring, J., 2012. Lean in healthcare: the unfilled promise? Social science & medicine, 74(3), pp.364–71.

Schmenner, R.W., 2009. Looking Ahead By Looking Back: Swift, Even Flow in the History of Manufacturing. Production and Operations Management, 10(1), pp.87–96.

Schmenner, R.W., 2004. Service Businesses and Productivity*. Decision Sciences, 35(3), pp.333–347.

Schmenner, R.W. & Swink, M.L., 1998. On theory in operations management. Journal of Operations Management, 17(1), pp.97–113.

Shah, R. & Ward, P.T., 2007. Defining and developing measures of lean production. Journal of Operations Management, 25(4), pp.785–805.

36

Shannon, R.P. et al., 2006. Using real-time problem solving to eliminate central line infections. Joint Commission journal on quality and patient safety / Joint Commission Resources, 32(9), pp.479–87.

De Treville, S. & Antonakis, J., 2006. Could lean production job design be intrinsically motivating? Contextual, configurational, and levels-of-analysis issues. Journal of Operations Management, 24(2), pp.99–123.

Umble, M. & Umble, E.J., 2006. Utilizing buffer management to improve performance in a healthcare environment. European Journal of Operational Research, 174(2), pp.1060– 1075.

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

Villa, S., Barbieri, M. & Lega, F., 2008. Restructuring patient flow logistics around patient care needs: implications and practicalities from three critical cases. Health Care Management Science, 12(2), pp.155–165.

Voss, C., Tsikriktsis, N. & Frohlich, M., 2002. Case research in operations management. International Journal of Operations & Production Management, 22(2), pp.195–219.

Waring, J.J. & Bishop, S., 2010. Lean healthcare: rhetoric, ritual and resistance. Social science & medicine, 71(7), pp.1332–40.

37

Appendix A: Interview protocol

 What is your function within the hospital?

 Can you describe the work activities performed on your department?

 What is Lean according to you?

 When is an improvement Lean?

 What measures do you use for this?

 How long do you work with Lean on your department?

 How many people have had a Lean training?

 Are all the employees aware of the ideas and goals behind Lean?

2. Variability and buffers

Day to day probably you experience a lot of variations in the work process.

 To what extent do you experience the following sources of variability?

 How do these sources of variability affect each of the following aspects?

 How do you cope with these sources of variability?

 Are there sources of variability that you experience which we have not discussed yet?

Type of variability Brief explanation Influences (for all sources of variability)

Arrival variability Variability in when and how many patients arrive.

 Throughput time of patients  Quality of care  Satisfaction of patients  Employee satisfaction  Amount of mistakes made  Costs  Travel time  Over time

 Are there any other effects?

Clinical variability Variation in the illness of patients.

Professional variability Variation in the working procedures of nurses and doctors, for example how long it takes them to perform a certain procedure.

Bed availability The availability of beds. Patient service time The amount of time needed

to treat one patient, can be dependent of age.