MASTER THESIS
Impact of lean in health
care
Assessing and reducing buffers in patient flows
Veerle Pellegrom - s2019043 Msc Supply Chain Management
22 June 2015
Abstract
Health care organisations are under increasing pressure to improve the quality of care and deliver care more cost effectively. Lean initiatives have been utilised to improve their performance. However, attention has mostly been paid to visible elements, and tools and techniques, while the essence of lean is minimising buffers for variability. This research utilises a multiple case study to analyse three different hospital patient flows where patient flow improvements are desired. The research indicates how buffers can be systematically assessed and reduced and it shows why improving patient flows requires different approaches due to variability and buffers. This research underlines the importance of variability and buffers in lean initiatives in health care and it aides health care organisations in analysing patient flow problems.
2
Table of content
1. Introduction p. 3
2. Theoretical framework p. 4
2.1 Lean and flow in health care p. 4
2.2 Variability p. 5 2.3 Buffers p. 6 2.4 Research focus p. 7 3. Methodology p. 7 3.1 Research setting p. 7 3.2 Case selection p. 8 3.3 Data collection p. 9 3.4 Data analysis p. 10
4. Analysis & results p. 11
4.1 Mammacare p. 11
4.2 Physiotherapy p. 14
4.3 Rheumatology p. 16
5. Discussion p. 19
5.1 Assessing and reducing buffers p. 19
5.2 Approaches p. 20 5.2.1 Mammacare p. 20 5.2.2 Physiotherapy p. 21 5.2.3 Rheumatology p. 21 5.3 Theoretical implications p. 22 6. Conclusions p. 24 6.1 Managerial implications p. 25
6.2 Limitations and future research p. 25
7. Literature p. 27
3
1. Introduction
Healthcare providers are expected to continuously improve their performance. There is increasing pressure from both governments, and insurers, to increase safety, and reduce costs. One way healthcare organisations attempt to improve their performance is through lean initiatives. These initiatives allow organisations to improve their processes without having to make high investments while utilising lean as the body of thought. Lean has proven to be successful in health care settings in terms of improved quality, access and efficiency (Mazzocato, Savage, Brommels, Aronsson, & Thor, 2010). However, more rigorous research is necessary to further understand the underlying factors influencing the impact and success of lean in health care (Poksinska, 2010). This research will contribute to this aspect.
Prior research on lean in health care has mostly focused on the different available tools and practices, and on 'quick wins' to improve performance which could obstruct optimal lean outcomes (Radnor & Osborne, 2013). This relates to Poksinska's (2010) findings that most articles on lean in health care focus on process improvement and continuous flow. Continuous flow often refers to patient flow, which is of importance as improving patient flow leads to increased hospital productivity and higher patient satisfaction (Litvak, 2009; Villa, Barbieri, & Lega, 2009). Hopp and Spearman (2004) state that the essence of lean is minimising the buffers for variability, which can often be applied in patient flow situations. These buffers can be divided into time, capacity or inventory. The related types of variability in health care settings are natural and artificial variability, which should be managed optimally or eliminated (Litvak & Long, 2000).
4
This research will therefore answer the following question: How can health care
organisations systematically assess and reduce buffers? The aim of this study is to analyse
the role of variability and buffers in patient flows and indicate why different approaches are required to improve patient flows. The chosen health care organisation in this study is a hospital as it encounters patient flows. A multiple case study, in which three patient flows will be analysed, will be adopted as it allows for an in-depth analysis by enabling a thorough exploration of the context of implementing lean, its content and processes (Burgess & Radnor, 2013).
This research is relevant for practice as health care organisations that wish to implement lean initiatives to improve patient flow can learn what a suitable approach could be for the analysis of patient flow problems to gain the most out of the initiatives. This could influence the delivered quality of care. Furthermore, health care organisations could become more aware of the role of variability and buffers that can cause patient flow problems.
This paper will continue as follows. Firstly, the theoretical framework will be presented in which the research focus will also be indicated. Afterwards, the methodology for this research will be discussed. This is followed by an analysis of the results. The subsequent part will discuss the results, and this paper will end with a conclusion where the managerial implications and limitations will be presented.
2. Theoretical Framework
This theoretical framework will start with an overview of lean and flow in health care, followed by a discussion on variability and buffers. This section will finish with the research focus.
2.1 Lean and flow in health care
5
mura (variability in a process) also causes waste, which is often overlooked (Hopp & Spearman, 2004). Hopp (2008) states that an organisation needs to reduce this variability and minimise buffers to become lean. Several types of variability and buffers have been identified in the health care environment, which will be discussed in the subsequent sections.
After the introduction of lean in the health care environment by Womack and Jones (1996), many authors have acknowledged that lean cannot be simply copied from the production to the health care environment (Hines, Holweg, & Rich, 2004). The main reason is that the inputs are people instead of products. Nevertheless, the transfer to the health care environment is still classified as relatively new (Burgess & Radnor, 2013). Several authors state that lean in health care is inconsistent as there are patients for which value needs to be considered, just as for highly educated professionals who need to collaborate in processes (Ballé & Régnier, 2007; Proudlove, Moxham, & Boaden, 2008; Young & McClean, 2008). Therefore, a more holistic view is necessary to avoid delivering only parts of best practices which might have an even wider negative impact on the whole system (Radnor et al., 2012; Towill & Christopher, 2005; Waldman & Schargel, 2006).
This holistic view can be obtained by improving processes that span multiple departments, of which patient flow is an interesting perspective as it requires department transcending process thinking. Patient flow can be defined as the speed at which patients transfer from one step in the care process to the next (Hopp & Spearman, 2001). Problems in patient flows contribute to longer waiting times and overcrowding of patients (Mazzocato et al., 2012). Therefore, improving patient flow is one of the main drivers in lean initiatives (Drupsteen, van der Vaart, & van Donk, 2013). According to Schmenner and Swink (1998), patient flow can be improved by overcoming three different barriers: bottlenecks, non-value-added activities (such as unnecessary waiting), and flow variability. Bottlenecks indicate the presence of few capacity buffers, non-value-adding activities can be classified as waste, and flow variability needs to be managed or buffered (Hopp & Spearman, 2004). Therefore, these aspects should be addressed in lean initiatives in health care.
2.2 Variability
6
clinical, flow, and professional variability. The fact that patients can respond differently to the same treatments represents natural clinical variability. Natural flow variability implies that patients appear in a random fashion with varying arrival rates. An example of natural professional variability is that medical practitioners are not equal in their ability to provide care efficiently. Litvak and Long (2000) state that these types of variability should be managed optimally. However, artificial variability, which often results in flow or professional variability, should be actively reduced. For instance, a surgeon arriving late at the operating room and unfamiliarity with a new technology are types of artificial variability. There is no doubt that these types of variability should be eliminated. However, an organisation will first have to be able to identify these types of variability and understand their sources. This is supported by Andersen et al. (2014) who state that the characteristics of lean (e.g. managing and eliminating variability) and its local application should be given more attention.
2.3 Buffers
In a production system, all variability will be buffered in some way (Hopp & Spearman, 2004). Three types of buffers are identified: capacity, time, and inventory. These buffers can be translated to a health care setting. A capacity buffer in a hospital could consist of an excess of nurses, physicians, and equipment, which represents itself in idle capacity. A time buffer implies a patient needs to wait for treatment. However, as the objective in a hospital is that patient are seen as quickly as possible, a time buffer would not be the most preferable option. Conversely, capacity buffers for expensive resources are not preferred due to the high costs. Lastly, an inventory buffer means that extra medication or appliances are held. However, patients are the main input in hospitals and it is not possible to buffer them as inventory. In general, the inventory buffer is not applicable in this research and a choice needs to be made between capacity and time buffers as soon as variability has to be buffered.
7 2.4 Research focus
This research will answer the question how health care organisations can systematically assess and reduce these buffers, and indicate what kind of approaches are most suitable for analysing patient flow problems regarding variability and buffers. This is supported by Poksinska (2010) who states that to achieve continuous flow, it is necessary that the relevant persons gain an understanding about the different processes, are able to identify waste, and find out what the sources of the problems are. This can be related to artificial variability and the buffers that are utilised to cover for this variability. Especially in patient flow situations where problems can accumulate in longer waiting times, it is important to analyse the aforementioned barriers that hinder patient flow improvement, which will be incorporated in this research.
Lean initiatives should try to reduce artificial variability, as the presence of variability leads to time and capacity buffers, which has a negative effect on patient flow. Even though this is the essence of lean, it has not received much attention in literature. Therefore, this research will focus on these aspects. In order to improve patient flow, this research will uncover the sources of artificial variability to assess and minimise the related time and capacity buffers and provide a systematic way to assess these buffers.
3. Methodology
The main goal of this research was to generate knowledge on how health care organisations can systematically assess and reduce buffers. As this knowledge is currently unknown, new theory needed to be developed. Karlsson (2009) states that case research is very suitable for theory building, as well as gaining an understanding of the phenomenon in practice. Furthermore, case research is suitable when context knowledge is important, which is applicable in the health care environment (Voss, Tsikriktsis, & Frohlich, 2002). Therefore, a case study was performed in this research. A multiple case study was undertaken as this typically provides stronger support for theory building than single case studies (Yin, 2009). The way how buffers could be systematically assessed and reduced within the cases and the related approaches were compared by performing a cross-case analysis.
3.1 Research setting
8
for which lean initiatives had started, thus the importance of variability and buffers was expected. The hospital encountered several patient flow problems simultaneously, which was significant for the cross-case comparison. Within this hospital, three cases were selected while the unit of analysis was the patient flows as this allowed for identifying and assessing buffers and their variability sources, just as researching which approach was suitable to analyse the patient flow.
3.2 Case selection
Multiple cases provide the advantages of guarding for observer bias and augmenting the research validity (Karlsson, 2009). The case selection process of Miles and Huberman (1994) was followed by first analysing the boundary of the study, which was determined by the research question, the setting, and the unit of analysis. In the second step, a sample frame was generated by purposefully selecting cases. Three different cases were selected to analyse how the assessment and reduction of buffers could be accomplished in a systematic way. As propositions to analyse patient flow problems were also developed, it required exploring for similarities and differences in the patient flows. The most important variables to base the case selection on were: the current patient flow issues, the type of buffers that needed to be reduced (capacity or time), the stated goals of the initiative, and the type of department. This selection of cases ensured external validity and therefore should allow for generalisable results (Karlsson, 2009). This was vital for this research as the deliverables were a systematic way of assessing and reducing buffers and delivering propositions for analysing patient flow problems. Therefore, the cases were purposefully selected based on the above criteria. Table 1 provides an overview of the selection criteria and selected cases.
Table 1 - Case selection criteria and selected cases
Case 1 Case 2 Case 3
Selected department
Mammacare Physiotherapy Rheumatology Number of
professionals
10-15 10-15 5-10
Current issues Access times too long Expensive capacity utilised early on in process
Too much idle time for physiotherapist
Access times fluctuating and too long
Daily activities not efficient
Type of buffers Capacity and time Capacity Capacity and time Goal initiative Improve patient flow
Improve use of capacity
Improve patient flow Decrease waiting times physiotherapist
Improve patient flow Increase efficiency daily activities
Type of department
Outpatient clinic In- and outpatient clinic
9
The mammacare clinic wanted to introduce a big change in the step-wise diagnosis and treatment of patients with breast cancer. Access times were too long and expensive capacity was utilised early on in the process. The aspiration was to improve patient flow and the use of capacity. The second case revolved around the physiotherapy department where the main focus was on the capacity of the physiotherapist, which was classified as a shared resource as it was called upon by different departments. The main goal was to reduce waiting times for the physiotherapist during his inpatient treatments. For the third case, the focus was on shortening and stabilising the lengthy access times at the rheumatology department. Furthermore, the efficiency of the daily activities could be enhanced.
Theoretical replication logic was applied which entails that contrary results were produced, but for predictable reasons (Karlsson, 2009). Since the results of the selection criteria differed between the cases and different patient flow problems were encountered, contrary results were expected on how buffers could be assessed and reduced, as this was the main focus of the research.
3.3 Data collection
Data was collected by utilising multiple methods and sources to achieve triangulation and a chain of evidence. On-site visits to the three different departments aided in gaining more insight in the patient flow. During the research period, the departments were visited several times to discuss research progress and intermediate findings. Next to case visits, a fellow researcher conducted observations at two departments the physiotherapists visited, to research their daily routines in order to identify buffers and their variability causes, and its influence on patient flow. Furthermore, the observations allowed for clarifications on why types of variability and buffers existed. Field notes were taken during these on-site visits and observations.
10
collaboration with fellow researchers with the project leaders, secretaries, a radiologist, and a surgeon. These interviews served in achieving more insight in the patient flow (problems) and how different disciplines approached the patient flow problems. Field notes were taken during these interviews, which were processed within 24 hours.
Furthermore, access was granted to secondary data on patient flows, internal reports and presentations reporting on the lean initiatives. These helped to determine what kind of approach was chosen in the initiatives, and if variability and buffers were considered. In addition, hospital databases provided insight in access times, process structure, waiting times, and planning schedules. Together, the different types of data collection methods and data sources facilitated construct validity which is essential for theory building (Karlsson, 2009).
3.4 Data analysis
Data from the observations was analysed by comparing field notes of the two departments. A search for patterns was executed to identify which buffers and variability sources were present and their effect on patient flow.
Interview data was analysed by following the steps suggested by Eisenhardt (1989). Firstly, an analysis was performed within a case, after which it was compared between cases to search for mutual aspects. Categorisation first occurred by determining what type of variability or buffer was discussed. Subsequently, through deductive reasoning, distinctions were made between relevant quotes and irrelevant data. All quotes were provided with comments to describe the presence and expression of a certain type of variability or buffer and its influence on patient flow. The within-case comments were compared between cases for a cross-case analysis. This cross-case analysis allowed for searching for similarities and differences between the types of buffers and variability sources, and where problems in the patient flow occurred. Furthermore, it was essential for the assessment and reduction of buffers and to generate propositions for the analysis of patient flow problems.
The quantitative data was analysed by utilising MS Excel to track the patient flow, available capacity, and appointment schedules. This allowed for indicating where variability and buffers arose in the patient flow.
11
4. Analysis & results
In this section of the paper, the analysis and results are provided. This section is structured as follows, first we introduce the specific department, subsequently identify the main issues in each patient flow, and finish with the identified sources of variability and buffers.
4.1 Mammacare
The mammacare department was troubled by high access times and long patient waiting times (especially between intake, final diagnosis, and the start of treatment). The main goals of the initiative were decreasing the access times to twenty-four hours, providing the final diagnosis within one day, and ensuring that ninety percent of the patients could start their treatment within four weeks. Figure 2 shows a simplified version of the patient flow from the patient perspective.
Figure 2 - Patient flow mammacare department
The mammacare department identifies three different patient inflows: all patients are referred by their general practitioner, either to visit the clinic, to directly undergo a mammography or echography, or after visiting a breast cancer screening facility. The largest inflow are patients that are referred to visit the clinic. Patients receive an urgency code which signals if a patient
Figure 1 – Overview methodology and deliverables research Information collected
on:
-Patient flow (problems) - Presence and types of
variability and buffers - Approaches taken by initiatives In 3 cases: Mammacare Physiotherapy Rheumatology Deliverables: -Answer how buffers can
be systematically assessed and reduced
- Propositions on approaches to analyse patient flows regarding
variability and buffers Information collected
via: -On-site visits and
12
could be asked to wait. After the intake, the patient undergoes a mammography or an echography, depending on the situation of the patient. These mammographies and echographies are performed by an echo analyst and assessed by a radiologist. Patients have to wait for the results, after which these are discussed with a surgeon. If the patient does not require treatment, the patient leaves the patient flow. However, if the results are not conclusive or if cancer is suspected, the patient has to undergo several other tests, where the same or more disciplines are involved. The clinic tries to tune this unpredictability by categorising patients with codes that determine the suspicion of cancer. The mammacare nurses tend to appointments where results are discussed. Apart from new patients, the department also has numerous returns for yearly check-ups after treatment. Both the surgeons and members of the radiology department are involved in these appointments. The number of appointment slots is limited per day, mostly at predetermined times, and are performed by five radiologists and five surgeons who work at the mammacare clinic one day a week, next to their work at other departments.
Main issue
The long access times, averaging six days, were partly caused by the limited capacity of the surgeons. Only a few new patients could be seen every day due to other appointments and the dependency on the capacity of the radiologists as they needed to assess the mammo- or echographies, and possibly conduct additional testing. Longer waiting times between the intake, final diagnosis and start of the treatment presented another issue. This was partly caused by the unpredictability of required tests for a patient and the conclusiveness of the results. For every test, several disciplines were involved who were dependent on each other. In other words, a patient sometimes waited because these disciplines were waiting for each other. For instance, if a patient was scheduled to meet a radiologist, it could take close to two hours before the patient could discuss the results with a surgeon. Another cause of waiting time could be found in the weekly meeting with all the disciplines where all treatment plans were ascertained. This implies that for some patients, it could take an extra week before they could continue with the next step in their process. Lastly, the clinic was closed during two afternoons per week which complicated appointment scheduling.
Variability and buffers
13
difficult to forecast. Natural clinical variability played a major role by causing unpredictability in the required appointments at the mammacare clinic. The first three appointments of new patients could be planned upfront, but if additional testing needed to occur, appointments had to be planned at that moment which caused waiting time. In addition, the dependency of disciplines for each appointment also caused variability as there were limited time periods in which the disciplines could work alongside each other and they were not simultaneously busy. Lastly, professional variability was eminent due to the large number of part-time employees at the clinic which complicated appointment scheduling and thereby caused waiting time. Overall, artificial variability had the most influence on performance by causing buffers. This occurred due to how the process was organised, which caused dependency between the different disciplines, the behaviour of the involved employees and their collaboration. An example of employee behaviour was purposefully letting patients wait for no medical reason. Furthermore, the unit manager stated: ''There is still a lot to do relating
to the collaboration between the different disciplines''. If this could be improved, waiting
times could decrease.
Regarding the buffers, the time buffer for the patient was prevalent. Patients had to wait before they could visit the clinic and they encountered in-process waiting time by waiting on results, occupied capacity, or capacity that waited for other capacity. A capacity buffer was present, which expressed itself in two ways. There was an unwanted form of capacity buffer use since personnel worked during their breaks or performed overtime. There was not a clear pure capacity buffer since the access times usually increased when patient inflow enlarged. However, for the radiologists, there was a capacity buffer as not all of their appointment slots were always booked, and they ensured a build-up of work.
Figure 3 shows the main aspects with regards to variability, patient flow problems, and the buffers of the mammacare department.
Figure 3 - Overview main aspects mammacare department Mammacare department Waiting times in complex patient flow: unpredictability possible appointments and dependency disciplines Artificial professional variability: organisation process, collaboration and behaviour disciplines
14 4.2 Physiotherapy
For the physiotherapy department, the problems mentioned were that patients could not be treated when the physiotherapists wanted to due to external disruptions, patients could not be treated at all during (a part of) a day due to a lack of time, or material was not present. This caused the physiotherapist to sometimes experience idle time as he waited on the patient. External disruptions were caused by patients' unavailability because of other medical disciplines, or patients were not prepared to receive treatment for instance. Therefore, the capacity use, and the waiting times for the physiotherapist needed to be analysed at two designated departments where most of the problems occurred. The physiotherapists treat several types of patients at different departments and face time pressure. A group of patients, called the Rapid Recovery patients, need to be mobilised within four hours after surgery. Furthermore, the majority of patients only stays for around two days and requires treatment (twice) daily due to care pathway rules. Only if patients stay longer due to complications, the physiotherapist has more time to treat the patient and can more easily provide the treatment on another day if necessary. At the two departments, there are differences in collaboration, ranging from extensive discussion with nurses about the patient's conditions to hardly any contact. Furthermore, the other department plays different roles in the preparation of patients for treatment which could influence the available time for the physiotherapist.
Main issue
An interesting aspect of this patient flow was that the physiotherapists were deemed an important shared resource as various departments called upon them to provide treatments. However, some of these departments have attempted to improve their own patient flow which implied that there was less time for the physiotherapists to provide their treatments and their capacity was made subordinate to that of the other department. Next to the limited time of patient presence, this allowed for less time buffers, even if the patient did not experience the buffer, since the patient was present nevertheless. Although physiotherapy was important, patients did not have to wait for physiotherapy treatment if they were medically approved to be discharged.
15
for the physiotherapists occurred during the morning as the patients had to be seen by the physiotherapists, but also by nurses and interns.
Furthermore, during the observations it became apparent that direct waste still had a major impact on the activities of the physiotherapists. For instance, we often observed unnecessary motion, as the physiotherapists consulted the computer for patient information, only to find out later the patient was not available. Furthermore, treatment materials often had to be located, and the physiotherapists were unaware of which physiotherapist was going to treat which patient. If these forms of direct waste could be eliminated, the physiotherapists would have more available treatment time, and face less pressure on their buffers.
Variability and buffers
Even though all types of variability could be distinguished, these were not the main causes of the issues. Flow variability was barely present as patient arrival showed a predictable, steady pattern. Clinical variability could only facilitate the physiotherapists as they could make more use of a time buffer when patients were more ill and stayed longer. Professional variability played a more significant role, but mainly the artificial component. Every physiotherapist was specialised in treatments at a specific department. However, flexible capacity was utilised as physiotherapists would assist each other if the available capacity was insufficient. Because of this specialisation, physiotherapists required more time at another department due to unfamiliarity with the patients, treatments, and administration. Another type of artificial variability could be found in the collaboration with other departments as this caused patients to be unprepared or unavailable, and thereby led to idle time for the physiotherapist.
The buffers however, were of greater interest. There was no time buffer for the patients in the sense that patients did not experience it as waiting time. However, the physiotherapists utilised the mechanism of a time buffer since they transferred patients within a time period. One patient essentially waits when the physiotherapist is treating another patient, while the physiotherapist can visit another patient if one patient is unavailable. The capacity buffer expressed itself in two ways. Occasionally, the physiotherapists worked during their breaks or performed overtime, therefore the unwanted form of capacity buffer use was present. However, a pure capacity buffer was also applicable as the physiotherapists still often managed to treat all patients because of this patient transfer within a time period. Conversely, the unit manager stated: ''If you compare the amount of patients to the available time, it
16
very busy times, the physiotherapists would work harder. On the other hand, the observations showed that the physiotherapists take more time for other activities if given the opportunity, and the unit manager stated that ''If there are no patients going home, and no new patients
arriving, you have a lot of time to do your job in a relaxed way''. This led to the observation
of the quality buffer as described by Hopp, Iravani, and Yuen (2007). This indicates that processing time is adjusted which influences job quality, as a response to system congestion. The unit manager stated that the quality of care could sometimes diminish because specialised care could not be provided due to a lack of time.
Figure 4 indicates the main aspects regarding the variability, patient flow problems, and the buffers of the physiotherapy department.
Figure 4 - Overview main aspects physiotherapy department
4.3 Rheumatology
The rheumatology department was troubled by long and unstable access times. Furthermore, the wish was expressed to free up production capacity. It was determined that the access times could fluctuate between three and eighteen weeks. As stabilising the access times was required, an analysis needed to be performed on the reasons behind the access times fluctuations. Patient demand was relatively stable over the year. Figure 5 shows a simplified version of the patient flow from the patient perspective.
Physiotherapy department
Capacity under pressure due to Rapid Recovery patients and limited time patient presence
Idle time physiotherapist Special trade off: waiting time and
capacity Direct waste more
influential than variability Artificial professional variability: specialisation
Time buffer: use of buffer, but patient does not experience
waiting time Capacity buffer: unwanted form and
pure form Quality buffer
17
After the referral and triage of the rheumatologist, the first appointment is often with one of the rheumatologists where tests are performed and medication prescribed as patients cannot be cured from their condition. The majority of patients returns as a follow-up patient over time. The team at the rheumatology clinic is small with only two rheumatologists, one specialist nurse, one counsellor and several secretaries. Patients with specific conditions can be seen by the specialist nurse. The counsellor only performs supporting activities for the patients and no medical procedures. The capacity of the rheumatologists for new patients is limited due to other appointments and one rheumatologist having commitments at another department.
Main issue
Regarding the access times, the capacity needed to be analysed, since the demand of patients did not fluctuate greatly. For instance, the capacity decreased for several weeks due to absence of one of the rheumatologists, or the capacity increased by letting interns or a resident treat new patients as well. If interns and a resident were present at the clinic, the amount of new patients that could be seen in a time period increased greatly, which decreased the access times. However, this caused another effect as well. More patients were referred as general practitioners noticed that the access times at this hospital were lower compared to other hospitals. This would lead to an increase in patient demand, which subsequently caused an increase in the access times because the capacity remained equal. Furthermore, as more new patients were being treated at the clinic, this led to an increase in the amount of follow-up patients which the clinic could not handle over a longer time period. Therefore, the focus was on freeing up production capacity, especially for the rheumatologists. An attempt has been made to free up this capacity by letting the counsellor also see a specific group of new patients, and by shifting some activities to the secretary.
Referred via general practitioner Rheumatologist performs triage Appointment often with rheumatologist Undergoes tests or medication is prescribed
18 Variability and buffers
After identifying the main issues, several variability causes emerged. Flow variability played an enormous role, however this was mostly caused by an artificial component as the capacity fluctuated. This influenced performance by affecting patient demand and access times. If the capacity would remain stable, the same result would be achieved for the access times. The clinical variability was of less importance as half of the patients were diagnosed with the same condition. The professional variability was clearly present as every team member only performed specific tasks and appointments. However, there was a large artificial component as it was not always medically necessary that only the rheumatologists would treat the patients. If some of their activities or appointments would be performed by other team members, the access times could decrease.
In terms of buffers, there was a clear time buffer for patients as they had to wait before they could visit the clinic. There was regular waiting time within the process if test results needed to be obtained. The time frame of scheduling follow-up appointments was almost always met. Therefore, the time buffer was minimal within the process after the patient has visited the clinic, and it expressed itself by patients waiting on occupied capacity and information flows. The unwanted form of capacity buffer use was present as the personnel worked during their breaks or worked overtime. The pure form of the capacity buffer was also found, but only for the specialist nurse as she did not have appointments for around a quarter of her time per year. The former unit manager explained this as follows: ''She is dependent on the production of
new patients, because she does not play a role for the new patients, but she does have a role when the new patients become follow-up patients, so then she needs more capacity''.
Figure 6 indicates the main aspects regarding the variability, patient flow problems, and the buffers of the rheumatology department.
Figure 6 - Overview main aspects rheumatology department Rheumatology
department
Unstable access times due to fluctuations in capacity that lead to
reaction of demand Interaction with follow-up patients Artificial flow variability: fluctuations in capacity
Time buffer: access time Capacity buffer: unwanted form and pure form specialist
19
5. Discussion
The review of the patient flow problems, and the presence of variability and buffers, produced results which will now be combined in answering the research question, and generating propositions for analysing patient flow problems in lean initiatives.
5.1 Assessing and reducing buffers
The question posed in this research was how health care organisations can systematically assess and reduce buffers. By comparing the three different patient flows, the analyses showed there is no clear cut way. However, there are several aspects which can be systematically analysed that need to be considered which facilitate in the assessment. The two types of buffers (time and capacity) can arise in various parts of the patient flow, the types and level of variability that cause the buffers can differ greatly, different people can be involved, and the buffers can express themselves in several ways. For example, regarding time buffers, it has to be established who is waiting, on what or whom is that person waiting and when. An interesting aspect is that patients can choose to implement a time buffer sometimes, for example at the mammacare department. A patient can decide she needs more time to reflect on her options and chooses to come back later than possible to the clinic. The capacity buffer can differ in how it expresses itself; the unwanted form (performing overtime) and the pure form (excess capacity), and who is involved in the capacity buffer.
20
The reduction of buffers is possible to a certain extent. Some waiting time will always occur as patients need to wait on test results. A capacity buffer might be difficult to reduce because of strict hospital rules as to the use of capacity. The professionals have to execute a specific amount of tasks per year to hold on to their license for instance. There needs to be another way of filling up freed up capacity as well. Concluding, there is no clear cut way to assess and reduce buffers, but systematically covering several aspects can facilitate in the assessment and reduction.
5.2 Approaches
While analysing the patient flow problems, different approaches emerged. Next to the assessment of buffers, this led to several propositions on approaches to analyse patient flow problems with regards to the interaction of variability and buffers. Per department, a different proposition has been generated. Comparisons are made with the approaches taken by the hospital.
5.2.1 Mammacare
For the mammacare department, it was important to take both a macro and micro view due to the complexity of the patient flow. The macro view included the complete patient (in)flow and overall processes where the role of the unpredictability of appointments was included. The micro view was important to look at phases and the order of appointments where waiting times occurred and include the dependency, behaviour, and collaboration of the different disciplines. Naturally, the limited capacity of the specialists and radiologists was of importance in determining the waiting times as well since they are dependent on each other. This situation and approach leads to the following macro-micro proposition.
If there is a high level of complexity in the patient flow due to the unpredictability of appointments and dependency of disciplines, a complex dependency approach needs to be taken which consists of a macro and micro view to analyse buffers throughout the patient flow.
21
patients. The dependency between the different disciplines, their collaboration issues and behaviour did not became clear. The unpredictability of appointments might also have been underestimated as no figures were given on the different paths within the patient flow, even though this could play a big role in future process improvements.
5.2.2 Physiotherapy
It is characteristic of the physiotherapy department that treatments are provided in patient flows of other departments and that patients quickly leave the patient flow. The approach that needed to be taken for this patient flow was for a department classified as a shared resource, that was experiencing pressure on their capacity buffers, while it could easily utilise time buffers, and where direct waste influenced performance more than variability. The physiotherapists faced a trade-off between waiting time and capacity since less time buffers were allowed due to Rapid Recovery patients, which might caused the need for more capacity buffers. Furthermore, the synchronisation with other departments needed to be considered as this caused idle time for the physiotherapist. This approach could be typified as a capacity-collaboration approach which is explained in the following proposition.
If a shared resource faces less allowance for time buffers and increased pressure on its capacity buffers, an approach needs to be taken that includes this trade-off between waiting time and capacity, and the collaboration with other departments. This could be typified as a capacity-collaboration approach.
The approach that was used by the hospital mostly looked at external influences that complicated the work of the physiotherapist in terms of waiting time. This can be found in the synchronisation with other departments, but also the activities that a patient has to undertake which makes a patient unavailable for the physiotherapist.
5.2.3 Rheumatology
22
interaction between new and follow-up patients. The following proposition summarises these aspects into a dynamic process approach.
If there are artificial fluctuations in the capacity of a department, a dynamic process approach needs to be taken where the interactions between capacity and patient demand over time are considered which influences the time buffer for a patient.
The approach that was taken by the hospital was a more static approach. It was determined that the access times were not stable, but no research was executed regarding the reasons for this occurrence. Furthermore, there were often discussions on the role and task division of the team at the clinic to reduce buffers, but it was difficult to achieve real progress in that area. We can conclude that every patient flow needed a completely different approach to analyse where and what kind of patient flow problems occurred and what the role of variability and buffers was, which is summarised in Figure 7. To know which approach is the most suitable could be very beneficial as Poksinska (2010) states that to achieve continuous flow, it is necessary that the relevant persons understand the different processes, are able to identify waste, and find out what the sources of the problems are.
Figure 7 - Overview approaches analysis patient flow problems
5.3 Theoretical implications
This research has several theoretical implications. In current theory, Litvak and Long (2000) distinguish between natural and artificial variability and divide those in flow, clinical, and professional variability. These labels have been useful by indicating the main areas where variability exists within a process. Especially for artificial variability, as this needs to be buffered (Hopp, 2008). However, for the three analysed patient flows, sources of variability were also found in many other aspects as expected by literature (Kolker, 2008; Noon,
• Complex dependency approach • Macro-micro view
Mammacare
• Capacity-collaboration approach • Trade-off waiting time - capacity
Physiotherapy
• Dynamic process approach • Interaction demand and capacity
over time
23
Hankins, & Côté, 2003; Villa et al., 2009). For instance, at the rheumatology department, a variability source was found in the capacity, for the mammacare clinic in the dependency of disciplines. These sources do not correspond with one of the distinguished variability types. Therefore, it is useful to think broader than these labels and scan for variability in several aspects. Even though variability was found in many sources, it was not always recognised by the hospital, as could be seen by the missing analysis of fluctuations in capacity at the rheumatology department. In other initiatives, variability and buffers were recognised, but they were not always acknowledged. An explanation could be that variability is not always considered as a negative element, or the presence is known and the hospital does not believe it could be managed. Thereby this study supported the findings of Tucker and Edmondson (2003) when health care staff experiences an interruption or problem, they do not pay attention to it and continue working around the problem. Their focus is on finishing the job and helping the patient, but the cause of the problem remains uninvestigated. This was noticeable at the physiotherapy department for instance, as direct waste was observed, while the daily routines were not critically researched. This elimination of direct waste is crucial since it is the first source of excess buffering (Hopp & Spearman, 2004).
Regarding buffers, three types of buffers (time, capacity, and inventory) are identified in theory. However, this research indicated that a quality buffer could also arise in an healthcare environment. Hopp et al. (2007) indicate this buffer for the production environment where the processing time is adjusted as a result of congestion in the system, which could influence job quality. This buffer was recognised at the physiotherapy department. In addition, we found that buffers could express themselves in multiple ways, for instance, an unwanted form of capacity buffer use, or the time buffer could differ by determining what the patient was waiting on, and where the buffer occurred in the patient flow. The reduction of buffers could therefore not be completed in a clear cut way. Furthermore, an effect relating to buffers was found, the service bullwhip effect, which occurs when the variation of demand is progressively amplified up the supply chain (Akkermans & Voss, 2013). This effect was found at the rheumatology department when more new patients were treated, which led to more follow-up patients further down the service supply chain, which the clinic could not handle over a longer time.
24
bottlenecks in the patient flow as almost every patient had to be seen by them and their capacity was limited. By freeing up their capacity, the patient flow could benefit, to which optimally managing or minimising variability could contribute.
Furthermore, the analysis showed that eliminating waste in the form of unnecessary use of capacity could also be very beneficial. For example, at the mammacare department, eighty percent of a group of patients where a mammography had to be taken in order to determine whether cancer was suspected, did not require treatment. But all of these patients were seen by the specialist before the mammography, which was a form of unnecessary capacity use. They have tried to free up this capacity by letting a regular doctor perform intakes as well. In addition, the outcomes of this study indicate that standard lean tools might not be as useful to improve performance as patient flows require different approaches due to the different interactions between variability and buffers. While these standard tools have been the focus in literature on lean (Radnor & Osborne, 2013), this research indicated that specific approaches are required. Thereby, this research has extended the findings of Poksinka (2010) by proving that the recognition of variability and buffers are underlying factors that could influence the impact of lean initiatives in health care. Lastly, this research has furthered lean theory in health care by proving that variability and buffers certainly need to be considered when attempting to improve patient flow.
6. Conclusions
25
the waiting time-capacity trade-off for the physiotherapy department, and the dynamic process approach with the demand-capacity interaction over time for the rheumatology department. These propositions indicate that a systematic assessment requires that these main elements need to be considered. Obviously, these propositions will not be exhaustive, and do not cover all possible patient flow situations. Still, they should provide a direction as to which aspects need to be analysed when attempting to improve patient flow.
6.1 Managerial implications
This research has several managerial implications. In all of the patient flows, artificial variability was found. However, these aspects were not regarded as the main aspects that needed to be covered by the hospital in order to improve the patient flow. Artificial variability was often found in the behaviour of involved professionals or collaboration with other disciplines and departments. It is therefore important for future initiatives that a critical view is taken to the way of working and how collaboration with other professionals can be improved in order to reduce buffers. In addition, lean initiatives should incorporate a micro view that considers variability, their causes, and buffers in all steps of the patient flow. Lastly, project leaders should acknowledge that different approaches are required per patient flow improvement due to different sources of variability and buffers.
6.2 Limitations and future research
An overall limitation of the research was the available time to research the three patient flows. Therefore, it was not possible to interview all the involved professionals to incorporate multiple views, which might have led to some biased information. Furthermore, two of the three initiatives were led by a former employee of the department, which might have clouded the initiative view. Another important aspect is that the initiatives were influenced by the Six Sigma process improvement method which might have undermined lean as the body of thought and thereby limit the applicability of this research.
26
Although this research has provided interesting insights, further research could try to discover more aspects that cause differences in buffers and complicate the reduction of buffers in an health care environment. Moreover, further research could test whether the delivered propositions of this research hold in other hospital departments. Furthermore, additional research could be executed at different departments to find more interactions of variability and buffers to facilitate in analysing patient flow problems.
27
7. Literature
Akkermans, H. & Voss, C. 2013. The service bullwhip effect. International Journal of
Operations & Production Management, 33(6): 765-788.
Andersen, H., Røvik, K. A., & Ingebrigtsen, T. 2014. Lean thinking in hospitals: Is there a cure for the absence of evidence? A systematic review of reviews. BMJ Open, 4(1): 1-8. Ballé, M., & Régnier, A. 2007. Lean as a learning system in a hospital ward. Leadership in
Health Services, 20(1): 33-41.
Burgess, N., & Radnor, Z. 2013. Evaluating lean in healthcare. International Journal of
Health Care Quality Assurance, 26(3): 220-235.
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): 912-933.
Eisenhardt, K. M. 1989. Building theory from case study research. Academy of Management
Review, 14(3): 532-550.
Fredendall, L. D., Craig, J. B., Fowler, P. J., & Damali, U. 2009. Barriers to swift even flow in the internal supply chain of perioperative surgical services department: A case study.
Decision Sciences, 40(2): 327-349.
Hines, P., Found, P., & Harrison, R. 2008. Staying lean: Thriving, not just surviving (1st ed.). Cardiff: Lean Enterprise Research Centre, Cardiff University.
Hines, P., Holweg, M., & Rich, N. 2004. Learning to evolve: A review of contemporary lean thinking. International Journal of Operations & Production Management, 24(10): 994-1011.
Hopp, W. J. 2008. Supply chain science. Long Grove, IL: Waveland Press.
Hopp, W. J., Iravani, S. M. R., & Yuen, G. Y. 2007. Operations systems with discretionary task completion. Management Science, 53(1): 61-77.
28
Hopp, W. J., & Spearman, M. L. 2004. To pull or not to pull: What is the question?
Manufacturing & Service Operations Management, 6(2): 133-148.
Jayaram, J., Vickery, S., & Droge, C. 2008. Relationship building, lean strategy and firm performance: An exploratory study in the automotive supplier industry. International
Journal of Production Research, 46(20): 5633-5649.
Karlsson, C. (Ed.) 2009. Researching operations management. New York: Routledge.
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): 27-40.
Litvak, E. (Ed.) 2009. Managing patient flow in hospitals: Strategies and solutions (2nd ed.). Oakbrook Terrace, IL: Joint Commission Resources.
Litvak, E., & Long, M. C. 2000. Cost and quality under managed care: Irreconcilable differences? American Journal of Managed Care, 6(3): 305-312.
Mazzocato, P., Holden, R. J., Brommels, M., Aronsson, H., Bäckman, U., Elg, M., & Thor, J. 2012. How does lean work in emergency care? A case study of a lean-inspired intervention at the Astrid Lindgren Children's Hospital, Stockholm, Sweden. BMC Health Services
Research, 12(1): 28-40.
Mazzocato, P., Savage, C., Brommels, M., Aronsson, H., & Thor, J. 2010. Lean thinking in health care: a realist review of the literature. Quality & Safety in Health Care, 19(5): 376-382.
Miles, M. B., & Huberman, M. A. 1994. Qualitative data analysis. London: Sage Publications.
Noon, C. E., Hankins, C. T. & Côté, J. 2003. Understanding the impact of variation in the delivery of healthcare services. Journal of Healthcare Management, 48(2): 82-97.
Poksinska, B. 2010. The current state of lean implementation in healthcare: Literature review.
Quality Management in Health Care, 19(4): 319-329.
29
Radnor, Z. 2010. Literature review of business process improvement methodologies. London: Advanced Institute of Management Research.
Radnor, Z. J., Holweg, M., & Waring, J. 2012. Lean in healthcare: The unfilled promise?
Social Science & Medicine, 74(3): 364-371.
Radnor, Z., & Osborne, S. P. 2013. Lean: A failed theory for public services? Public
Management Review, 15(2): 265-287.
Schmenner, R. W., & Swink, M. L. 1998. On theory in operations management. Journal of
Operations Management, 17(1): 97-113.
Shah, R., & Ward, P. T. Defining and developing measures of lean production. Journal of
Operations Management, 25(4): 785-805.
Towill, D. R., & Christopher, M. 2005. An evolutionary approach to the architecture of effective healthcare delivery systems. Journal of Health Organisation Management, 19(2): 130-147.
Tucker A. L., & Edmondson, A. C. 2003. Why hospitals don’t learn from failures: organizational and psychological dynamics that inhibit system change. California
Management Review, 45(2): 55-72.
Villa, S., Barbieri, M., & Lega, F. 2009. Restructuring patient flow logistics around patient care needs: Implications and practicalities from three critical cases. Health Care
Management Science, 12(2): 155-165.
Voss, C., Tsikriktsis, N., & Frohlich, M. 2002. Case research in operations management.
International Journal of Operations & Production Management, 22(2): 195-219.
Waldman, J. D., & Schargel, F. P. 2006. Twins in trouble II: Systems thinking in healthcare and education. Total Quality Management & Business Excellence, 17(1): 117-130.
Womack, J., & Jones, D. 1996. Lean thinking: Banish waste and create wealth in your
corporation. New York: Simon & Schuster.
30
Yin, R. K. 2009. Case study research: Design and methods. Thousand Oaks, CA: Sage Publications.
Young, T. P., & McClean, S. I. 2008. A critical look at lean thinking in healthcare. Quality
31
8. Appendix I
Interview guide in Dutch 1. Inleiding
Allereerst, welkom bij dit interview en hartelijk dank voor uw deelname. In dit interview zullen wij vragen stellen over het proces waar uw afdeling haar werkzaamheden verricht. We zullen zo eerst beginnen met een aantal algemene vragen over uw eigen werkzaamheden, uw afdeling en uw benadering ten opzichte van lean aangezien het project met een lean
benadering is opgezet.
Daarna volgen meer specifiekere vragen. Het kan zo zijn dat bepaalde zaken al eerder aan het licht zijn gekomen tijdens onze informele gesprekken, maar wij willen dit dan nu graag formeel vastleggen.
Dit interview zal ongeveer een uur duren. Als u zelf tussendoor vragen heeft of iets wil toevoegen, voel u vooral vrij om dat te doen. Als u op een vraag geen antwoord wil geven, kunt u dat aangeven. Uw antwoorden zullen vertrouwelijk en anoniem verwerkt worden. Vindt u het goed als dit interview opgenomen wordt? Dit wordt uitsluitend voor onze verwerking gebruikt en daarna vernietigd.
Heeft u op dit moment vragen? 2. Algemene vragen
Wat is uw functie binnen het ziekenhuis en hoe lang voert u deze functie al uit? Wat zijn uw verantwoordelijkheden?
Wat is lean volgens u?
Wanneer ziet u een verbetering als een ‘lean’ verbetering? Hoe lang werkt u met lean op uw afdeling?
Hoeveel mensen op uw afdeling hebben een lean training gehad? Heeft u zelf ook een lean training gevolgd?
Is al uw personeel op de hoogte van de ideeën en doelen van lean?
32
Met dit in uw achterhoofd, zouden we u graag de volgende vragen willen stellen. Voor elke soort variabiliteit hebben we dezelfde vragen opgesteld.
3. Specifieke vragen
Zou in percentages uit kunnen drukken tot op welke hoogte geplande en ongeplande aankomstvariabiliteit een rol speelt? Gepland is dat u van tevoren weet hoeveel patiënten er die dag een afspraak hebben. Ongepland is dat er tijdens een dag nog spoedaanvragen binnenkomen die u niet van tevoren heeft kunnen plannen.
o Hoe beïnvloedt dit de kwaliteit van zorg die gegeven wordt aan een patiënt? o Hoe beïnvloedt dit de tijd die besteed kan worden aan een patiënt?
o Hoe beïnvloedt dit de doorstroom van de patiënt?
o Tot op welke hoogte is aankomstvariabiliteit voorspelbaar? Verschilt dit misschien per dag of per seizoen?
o Hoe gaat u met aankomstvariabiliteit om, hoe probeert u deze op te vangen? o Zijn er nog dingen die u wilt toevoegen over aankomstvariabiliteit en de
effecten hiervan op het proces op uw afdeling?
Tot op welke hoogte speelt variabiliteit in de mate van ziekheid op de klinieken? Kunt u dit uitdrukken op een schaal van 1 tot 5 waarbij 1 laag is en 5 hoog?
o Hoe beïnvloedt dit de kwaliteit van zorg die gegeven wordt aan een patiënt? o Hoe beïnvloedt dit de tijd die besteed kan worden aan een patiënt?
o Hoe beïnvloedt dit de doorstroom van de patiënt?
o Tot op welke hoogte is variabiliteit in mate van ziekheid voorspelbaar? Verschilt dit misschien per dag of per seizoen?
o Hoe gaat u met deze variabiliteit in de mate van ziekheid om, hoe probeert u deze op te vangen?
33
Tot op welke hoogte ervaart u variabiliteit door specialisme op de klinieken? Kunt u dit uitdrukken op een schaal van 1 tot 5 waarbij 1 laag is en 5 hoog?
o Hoe beïnvloedt dit de kwaliteit van zorg die gegeven wordt aan een patiënt? o Hoe beïnvloedt dit de tijd die besteed kan worden aan een patiënt?
o Hoe beïnvloedt dit de doorstroom van de patiënt?
o Hoe wordt de kwaliteit van de fysiotherapeut bepaald en gecontroleerd? o Tot op welke hoogte is variabiliteit door specialisme voorspelbaar?
o Hoe gaat u met variabiliteit door specialisme om, hoe probeert u deze op te vangen?
o Zijn er nog dingen die u wilt toevoegen over variabiliteit door specialisme en de effecten hiervan op het proces op uw afdeling?
Kunstmatige variabiliteit ontstaat door de manier van werken, het gedrag en de manier van organiseren van het proces. Kunt u voorbeelden geven van deze soort variabiliteit?
o Hoe beïnvloedt dit de kwaliteit van zorg die gegeven wordt aan een patiënt? o Hoe beïnvloedt dit de tijd die besteed kan worden aan een patiënt?
o Hoe beïnvloedt dit de doorstroom van de patiënt?
o Tot op welke hoogte is kunstmatige variabiliteit voorspelbaar? Verschilt dit misschien per dag(deel) of per persoon?
o Hoe gaat u met kunstmatige variabiliteit om, hoe probeert u deze op te vangen? o Zijn er nog dingen die u wilt toevoegen over kunstmatige variabiliteit en de
effecten hiervan op het proces op uw afdeling?
34
Hoe (vaak) denkt u dat de tijdsbuffer, dus wachttijd voor de patiënt, voorkomt uw afdeling? Is dit dagelijks, hoe groot is het deel van de patiënten die wachttijd ervaart?
o Waardoor ontstaat deze wachttijd voor de patiënt normaal gesproken? o Waar wacht de patiënt vooral op?
o Hoe beïnvloedt dit de kwaliteit van zorg die gegeven wordt aan een patiënt? o Hoe beïnvloedt dit de doorstroom van de patiënt?
o Hoe denkt u dat deze wachttijd voor de patiënt verminderd kan worden? Wordt er gebruik gemaakt van een capaciteitsbuffer, dus af en toe onnodig beslag
leggen op capaciteit? Of uit de capaciteitsbuffer zich op een andere manier, bijvoorbeeld doorwerken in pauzes?
o Waardoor ontstaat de behoefte voor een capaciteitsbuffer?
o Hoe beïnvloedt dit de kwaliteit van zorg die gegeven wordt aan een patiënt? o Hoe beïnvloedt dit de doorstroom van de patiënt?
o Hoe denkt u dat dit onnodig beslag leggen op capaciteit verminderd kan worden?
3. Einde
