Shared physician scheduling policies

(1)

Shared physician scheduling policies

Balancing patient- & clinic-based performance

Master Thesis

MSc. Technology & Operations Management

University of Groningen, Faculty of Economics and Business Max Dannenburg

S2060337

First supervisor: University of Groningen dr. ir. D.J. van der Zee

Co-assessor: University of Groningen dr. X. Zhu

Company Supervisors: University Medical Center Groningen R.P. Borgers

(2)

Page | 2

Abstract

Purpose – Sharing is a common way of improving the utilisation of resources in hospitals.

However, the possible negative patient-based effects, such as an increase of throughput time are also recognised. Previous research has shown that scheduling policies can be used to balance patient- and clinic based performance. However, this research has been done regarding nurses and equipment, but not concerning physicians. This is relevant since physicians are known to be scarce resources. At the same time, by being key players in providing care services, they heavily impact the allocation of supportive staff, and/or fellow physicians, and the use of other resources like equipment or facilities. Therefore, this research strives to expand the current knowledge on shared physician scheduling policies.

Method – A design science research at the gastroenterology and hepatology department of the University Medical Centre Groningen (UMCG-GHD) is performed to attain this knowledge. Literature on physician scheduling policies is reviewed first in order to identify insights in important scheduling shared physicians elements such as their influence on performance and known best practices of dealing with them. Next, the UMCG-GHD physician scheduling system is described, analysed and redesign propositions are formulated and tested with the use of interviews and an qualitative cause and effect analysis.

Findings – The review of literature resulted in a four-phase staff scheduling terminology used to describe and analyse current UMCG-GHD practices; (1) forecasting demand, (2) determining staffing requirements, (3) shift scheduling and (4) rostering. This showed that the UMCG-GHD shared physician scheduling policies are designed based on historical figures and physician availability. These policy considerations result in a likely mismatch in patient-demand and physician supply. This is confirmed by an assessment on patient-based performance criteria that showed high, not norm meeting, admission times and admission time variability. An assessment on clinic-based performance showed high resource utilisation, high resource utilisation variability and a lot of resource restriction violations. An assessment on solution directions showed that more formal guidelines concerning forecasting and physician availability, the implementation of feedback loops on patient-based performance and the design scheduling policies based on patient demand can improve these performances. A comprehensive scheduling approach which incorporates all phases of staff scheduling showed to enhance the effects on patient- based performance even more.

Conclusions – A simultaneous view on clinic- and patient-based performance is needed in

shared physician scheduling policies. They should be designed following patient demand where formal and clear rules are applied considering physician availability and feedback loops on patient-based performance are implemented. A four-phase framework which entails forecasting demand, determining staff requirements, shift scheduling and rostering, can aid in this. A comprehensive scheduling approach, which incorporates all phases of staff scheduling, is advised for hospitals dealing with the challenge of scheduling shared physicians.

(3)

Page | 3

Preface

With this thesis I conclude my master Technology and Operations Management at the University of Groningen. The goal of this thesis was to expand the current knowledge on shared physician scheduling policies. Insights on shared physicians scheduling policies were obtained by doing a case study within the University Medical Centre Groningen gastroenterology and hepatology department.

Performing this, practical oriented, research was a pleasant experience. I therefore would like to thank the UMCG for the opportunity. Especially Ms. Nuus & Mr. Borges for their guidance, time and the providing of valuable insights in the problem faced. Furthermore, I also would like to thank Mr. Hoogstins for providing me with the needed data.

In addition, I would like to thank my supervisor Mr. Van der Zee for his experience and guidance for writing a thesis and setting up a research. I appreciated the structured process provided by him and the contribution of my fellow students, Evi & Simon, in this. Also, I would like to thank my co-assessor Mr. Zhu for his feedback on the research proposal.

Finally, I would like to thank my family for their unconditional support. Especially my partner Sophie, by helping me to put things into perspective.

(4)

Page | 4

Previous research ... 17 3.4.1 Performance criteria ... 18 3.4.2 3.5 Summary ... 18 4 System Description ... 20 4.1 System overview ... 20 Patient characteristics ... 20 4.1.1 Clinical processes ... 20 4.1.2 Patient routings ... 21 4.1.3 4.2 Physician scheduling ... 22 Forecasting Demand ... 22 4.2.1 Determining staffing requirements ... 23

4.2.2 Shift scheduling ... 25

(5)

Page | 5

4.3 Summary ... 26

5 System Analysis ... 27

5.1 Approach ... 27

Addressed patient population ... 27

5.1.1 Patient-based criteria ... 28 5.1.2 Clinic-based criteria ... 28 5.1.3 Data sources ... 29 5.1.4 Cause and effect analysis ... 29

5.1.5 5.2 Current patient-based performance ... 30

Admission time ... 30

5.2.1 Admission time variability ... 31

5.2.2 5.3 Current clinic-based performance ... 32

Resource utilisation ... 32

5.3.1 Resource utilisation variability ... 34

5.3.2 Resource restriction violations ... 34

5.3.3 5.4 Cause and effect analysis ... 35

Current performance ... 35

5.4.1 Current set-up ... 36

5.4.2 Causes of current UMCG-GHD performance ... 36

5.4.3 5.5 Summary ... 38

6 Solution Directions ... 39

6.1 Approach ... 39

Generation of solution directions ... 39

6.1.1 Testing of solution directions ... 39

6.1.2 Performance criteria ... 39

6.1.3 Addressed patient population ... 39

6.1.4 6.2 Forecasting demand ... 40

6.3 Determining staff requirements ... 40

6.4 Shift scheduling ... 40

6.5 Rostering ... 40

6.6 A comprehensive staff scheduling approach ... 41

6.7 Indication of effects ... 42

(6)

Page | 6

7 Discussion ... 45

7.1 Interpretation of results ... 45

7.2 Generalisation of the findings ... 46

7.3 Limitations ... 46

8 Conclusion ... 47

8.1 Summary of main findings ... 47

8.2 Reflection of research objectives ... 47

Sub-objective 1 ... 47 8.2.1 Sub-objective 2 ... 47 8.2.2 Sub-objective 3 ... 48 8.2.3 Main research objective. ... 48

8.2.4 8.3 Suggestions for further research ... 48

References ... 49

Appendices ... 51

Appendix A - Treatment times ... 51

(7)

Page | 7

1 Introduction

The environment of hospitals has changed in the last decades. Society is aging, causing an increasing demand for healthcare services. However, financial conditions for healthcare systems are not improving and even worsening (Poksinska, 2010). This pushes hospitals to improve on their operational performance. In order to enhance resource utilisation, hospitals often make use of shared resources; common-capacity sources in two or more networks (Hoekstra & Romme, 1992). However, when not managed properly, shared resources will lead to a decreased customer satisfaction because of an increase of throughput-times and throughput-time variability (Drupsteen, 2013; Vissers, 1994).

This research is motivated by a scheduling problem of the University Medical Centre Groningen gastroenterology and hepatology department (UMCG-GHD). The department consists of three sub-departments; (1) an outpatient clinic, (2) a clinical supervision ward and (3) an endoscopy centre. Gastroenterology and hepatology physicians are allocated to all three different sub-departments. Hence, they are shared resources. This, in combination with aspects such as patient-physician coupling, specific physician skills and a wide variety in patient diseases, complicates the overall scheduling of the department. The request of the department is twofold; gaining insight in the current shared physician scheduling policies and its effects on patient- and clinic-based performance and the providing of suggestions for improvement.

Vissers (1994) links the challenge of scheduling shared physicians to a trade-off, seeking to balance clinic- and patient-interests. This requires well-designed appointment policies which have the possibility to on the one hand decrease waiting times of patients and on the other hand increase the utilisation of expensive personnel and equipment-based medical resources. The effect of different staff scheduling policies, has, on a scarce notion, been assessed in the literature (Brusco & Showalter, 1993; Maenhout & Vanhoucke, 2013). However, these studies researched the allocation of shared nurses. Research on allocation policies of physicians and their effects on, clinic- and patient-based performance is not present. Physicians differ mostly in the fact that they are usually more specialised compared to nurses. Furthermore, physicians are the most important resources in hospitals regarding their interdependency with other resources (Vissers, 1994). Other resources, such as equipment and nurses, are namely often scheduled as a consequence of physician schedules. Therefore, the use of physicians has a major influence on the total performance of hospital departments. This generates a need for research on the field of physician allocation policies.

(8)

(9)

Page | 9

2 Problem Statement and Research Design

This chapter provides an overview of the content, scope and methodology of this research. Section 2.1. describes the background of the problem, section 2.2 states the research objectives that follow from this problem and section 2.3 includes the most important elements of section 2.1 and 2.2 in a conceptual model. Furthermore, section 2.4 discusses the research design and section 2.5 the data sources.

2.1 Problem Background

Shared resources, i.e., common-capacity sources used in two or more ways (Hoekstra & Romme, 1992), are often used in healthcare. (Drupsteen, 2013; Maenhout & Vanhoucke, 2013; Vissers, 1994). It is known that shared resources enhance clinic-based performance in terms of resource utilisation (Hoekstra & Romme, 1992), but it is also known that it can decrease patient-based performance (Drupsteen, Vaart, & Donk, 2013). This is acknowledged by Vissers (1994) who links the scheduling of shared resources to a trade-off, seeking to balance clinic and patient interests.

Scheduling policies can be used to facilitate this trade-off (Vissers, 1994). The process of scheduling staff may be composed of four phases (Defraeye & Van Nieuwenhuyse, 2016; Koole & Pot, 2006; Maenhout & Vanhoucke, 2013; Garry M Thompson, 1995): (1) Forecasting demand, (2) Determining staffing requirements, (3) Shift scheduling and (4) Rostering.

A scarce notion regarding shared staff allocation in hospitals is researched regarding the allocation of nurses (Brusco & Showalter, 1993; Maenhout & Vanhoucke, 2013). However, relating research regarding to physicians has not been performed before. Vissers (1994) differentiates leading from following resources whereas leading resources determine the possibility of the realisation of an appointment and following resources help facilitating the appointment. As depicted by Vissers (1994), physicians are often the leading resource in hospital scheduling and therefore have the most impact on overall resource utilisation and patient-based performance. An example of this is a consultation between physician and a patient. Without the physician this appointment cannot take place where there is no requirement for a specific blood-pressure measurer as there are plenty of these in hospitals. This interdependency of physicians causes a need for well-thought shared physician scheduling policies to facilitate that patients are served optimal and hospital resources are utilised efficiently.

(10)

Page | 10 due to sudden patient demand. Drupsteen (2013) has stated patient throughput-time and its throughput variability as the most important performance criteria regarding patients. Different definitions exist for the throughput time of patients in hospitals. Concerning the fact that this resource focuses on hospital scheduling where the time until an appointment can take place is relevant, admission times will be used. Klassen & Rohleder (2004) have defined admission times as the time between the awareness of a patient need for an appointment (for instance when he/she has made the call) and the date of the appointment. In hospitals, these admission times are often restricted by certain norms.

A scheduling shared physicians challenge is also faced by the University Medical Centre Groningen gastroenterology and hepatology department (UMCG-GHD). The department consists of three sub-departments; (1) an outpatient clinic, (2) a clinical supervision ward and (3) an endoscopy centre. The physicians are employed in all three sub-departments and therefore need to be scheduled over all three sub-departments. The objective of these schedules are the simultaneous enhancement of clinic-based performance, in terms of resource utilisation, resource utilisation variability and the amount and level of violations of resource restrictions and patient-based performance, in terms of the admission times and admission time variability.

2.2 Research Objectives

From the problem background, the following main research objective is determined:

‘To develop scheduling policies for shared physicians over hospital departments which improve physician utilisation, physician utilisation variability and physician restriction violations while meeting norms on patient admission time and admission time variability.’

The physician utilisation will be measured by the percentage of time spent to treating patients, the physician utilisation variability by the standard deviation and the coefficient of variation (CV) of the physician utilisation and the resource restriction violations by the amount of their occurrence. The admission times will be measured by determining the amount of days between the day of a request for an appointment until the actual appointment. The admission time variability will be measured by the standard deviation and the coefficient of variation (CV) of these admission times.

To realise the above objective, the following sub-objectives are stated:

1. To develop initial scheduling solutions based on empirical data and literature research. 2. To test different scheduling policies on the critical performance criteria in terms of

resource utilisation, resource utilisation variability, the amount and level of violations of resource restrictions and admission times and its corresponding variability.

3. To generalise the findings to establish widely applicable theory on shared physician scheduling polices.

(11)

Page | 11 research and to be able to investigate the effect of different shared physician allocation policies this research assumes the other resources to be sufficient when needed as a consequence of the use of physicians.

Besides the above mentioned performance criteria, the research objective will also be constrained by research constraints (such as budget and time).

2.3 Conceptual Model

As can be derived from the research objective, this research investigates the influence of certain shared physician allocation policies on patient- and clinic-based criteria. As stated in the problem background, sharing physicians has a positive influence on clinic-based performance and can have a negative influence on patient-based performance. The shared physician scheduling policies, which can influence these relationships, can furthermore be characterised by four different phases namely (1) forecasting demand, (2) determining staffing requirements, (3) shift scheduling and (4) rostering (Defraeye & Van Nieuwenhuyse, 2016; Koole & Pot, 2006; Maenhout & Vanhoucke, 2013; Thompson, 1995). Finally, the patient-based performance will be measured by the admission times and the admission time variability and the clinic-based performance will be measured by the resource utilisation, the resource utilisation variability and the resource restriction violations.

These elements and their (possible) relationships are visualised in the following conceptual model: Shared physicians Clinic-based criteria: · Resource utilisation · Resource utilisation variability · Resource restriction violations (?) (-) (+) (?) Patient-based criteria: · Admission time · Admission time variabilty Moderator Physician scheduling policies: (1)Forecasting demand (2)Determining staffing requirements (3)Shift scheduling (4)Rostering

(12)

Page | 12 As can be seen in figure 2.1, this research investigates the moderating effect of physician scheduling policies on the known possible negative and positive effect of using certain shared physicians on respectively patient- and clinic-based criteria. These, to be investigated relationships, are represented by question marks.

2.4 Research Design

To be able to achieve the research objectives and to discover the moderating effects of different shared physician scheduling policies, a design science approach is conducted. In this type of research, general applicable knowledge is developed by means of the creation of an artefact in practice (Holmstrom et al., 2009). This type of research is chosen to both satisfy the need of the company, the UMCG-GHD, and the overarching goal of developing knowledge on physician allocation policies.

The design science will start by means of a theoretical background to develop a solid base for the remainder of this research. In this chapter, systems are classified and typified, related research is summarised and research gaps are revealed. After this, by means of the regulative cycle of van Strien (1997), the problem faced by the UMCG is addressed. This regulative cycle consists of four steps namely (1) system description, (2) system analysis, (3) solution creation and (4) solution testing. The third (solution creation) and fourth step (solution testing) are in this research combined in one chapter; solution directions. In the discussion and conclusion, the findings will be generalised to generate widely applicable knowledge and limitations and suggestions for further research will be stated. The following table represents, based on the above mentioned research steps, the overview of this research:

Chapter Title 3 Theoretical Background 4 System Description 5 System Analysis 6 Solution Directions 7 Discussion 8 Conclusion

Table 2.1. Research design

The steps of the creation of the artefact, based on the regulative cycle of van Strien (1997), will be explained further in this section. As mentioned before this research is motivated by a problem of the UMCG-GHD.

System Description 2.4.1

(13)

Page | 13

System Analysis 2.4.2

Based on the system description and the performance criteria, the system will be analysed in this chapter. The aim of this chapter is to assess the current UMCG-GHD shared physician scheduling policies on the earlier mentioned performance criteria (section 2.2). Historical patient demand and physician scheduling data of the UMCG-GHD will be assessed by analytic deterministic techniques. To conclude this section, by means of an cause and effect analysis, elements of the scheduling system open for improvement are stated.

Solution Directions 2.4.3

This chapter will propose different directions for improving the scheduling of physicians of the UMCG-GHD and give indications of the effects of theses directions. Time-wise research constraints and data availability cause the scope of this research to be limited to organisational aspects instead of also to optimising the scheduling challenge faced. Since therefore no extensive assessment of proposed solutions is possible the third (solution creation) and fourth step (solution testing) of the regulative cycle of van Strien (1997) will be combined.

The identification of possible improvements will serve as starting points of the creation of possible solutions. Furthermore, findings of the department schedulers and supplementary literature findings will be used for the development of different options. Again, the different solutions will be classified by means of the four steps of staff scheduling.

The different possible solutions will be evaluated for their effect on the critical patient- and clinic-based performance criteria. The cause and effect analysis of chapter 5 will serve as base for the qualitative analysis of the possible effects of the proposed solutions.

2.5 Data Sources

(14)

Page | 14

3 Theoretical Background

The aim of this research is to specify the effect of shared physician scheduling policies on performance in hospital settings. This chapter strives to establish a solid theoretical base by reviewing previous research in this field and to consider insights obtained on the set-up of scheduling systems for allocating shared-staff. Section 3.1 discusses the trade-off between costs and service in hospitals. This sets the scene for the challenge that is faced in this research. Section 3.2 and 3.3 elaborate on the influences of the presence of shared resources and scheduling respectively. Section 3.4 discusses the challenges which occur when shared resources need to be scheduled. Finally, section 3.5 summarises the main findings.

3.1 Hospital organisation – improving costs and service

Hospitals are currently under a lot of pressure to improve on their clinic-based performance (costs) and patient-based performance (service). On the one hand, financial pressure and quality and safety demands are increasing due to changing governmental policies and on the other hand, patient demand is increasing due to the aging society and technological developments (Poksinska, 2010).

The enhancement of patient-based performance in terms of service and clinic-based performance in terms of resource utilisation can be seen as a trade-off (Vissers, 1994). Dellaert, Cayiroglu, & Jeunet (2015) have researched this trade-off and stated that when hospitals focus on the improvement of operational performance, this often has a negative influence on patient-based performance such as waiting times. Hospitals therefore should prioritise their performance goals and should include norms.

3.2 Shared resources in hospitals – aspects to take into consideration

(15)

Page | 15 combining appointments. This can result in a negative influence on patient-based performance.

Shared resources, as regular resources, can differ in the possibilities they have. Concerning physicians, this refers to the fact that they can have a different set of skills and therefore are capable for a different set of treatments. Two extreme configurations exist: specialisation (no flexibility) and full flexibility (pooling, total flexibility, full cross-training). However, many alternatives in between exist (Bokhorst & Gaalman, 2009). Logically, with full flexibility, more patients can be handled but research has shown that the same performance can be obtained with less than full flexibility (Campbell, 1999; Fry, Kher, & Malhotra, 1995). This however complicates the scheduling of shared physicians.

Vissers (1994) has distinguished shared resources to be 'leading' versus 'following' resources. A resource is 'leading' if it is the trigger for generating production on other resources, which 'follow'. Some resources are always leading or always following and for some resources this differs per situation. An example of a situation dependent resource is bed capacity, which is a following resource in an operating room, but a leading resource in an inpatient clinic. Vissers (1994) furthermore states that the physicians are the most important leading resource in hospitals. Physicians therefore enable production in hospital departments using the other resources. This makes physician-time, the time that a physician spends on patient treatment at certain (sub-) departments, a critical resource for the performance of other resources.

3.3 Scheduling in hospitals – opportunities, challenges & classifications

The complexity of hospital operations and the presence of shared resources require well designed schedules; blueprints that can be used to provide a specific time and date for patient consultation (Hulshof, Kortbeek, Boucherie, Hans, & Bakker, 2012). Scheduling policies are the set of rules with which the schedules are created (Maenhout & Vanhoucke, 2013). Several researchers have stated that scheduling policies play a central role in delivering care to patients (Cayirli & Veral, 2009; Maenhout & Vanhoucke, 2013). Well-designed scheduling policies can improve the matching between supply and demand to on the one hand decrease waiting times of patients and on the other hand increase the utilisation of expensive personnel and equipment-based medical resources (Gupta & Denton, 2008; Vissers, 1994).

(16)

Page | 16 shift scheduling, and (4) rostering (Defraeye & Van Nieuwenhuyse, 2016; Koole & Pot, 2006; Maenhout & Vanhoucke, 2013; Thompson, 1995). With the help of this terminology, staff scheduling policies can be classified. The following sections elaborate on each of the phase of scheduling staff.

Forecasting demand 3.3.1

In this phase, the patient demand for a certain period is forecasted. Patient demand can be defined in different ways. Examples are the total number of patients served in a year, the patient mix or the total available time for a specific treatment. Forecasting can be defined as the art of predicting the occurrence of events before they actually take place (Archer, 1980). Archer (1980) has reviewed different forecasting methodologies and made a classification between numerical and qualitative methods. Different methodologies can be chosen for the realisation of a good fit to the to be forecasted patient population in order to improve overall shared physician scheduling policies.

Determining staffing requirements 3.3.2

This phase entails the selection of required staffing levels required over time, in order to meet a specific performance target at minimal cost. Arthur & James, (1994) have stated three different approaches for determining staffing requirements regarding nurses:

1. The consensus approach, which is rather subjective. By considering varying degrees of knowledge, training, or analytical skills the requirements of nurses for a certain period are determined.

2. The top-down approach; here staffing levels are determined by higher hierarchical stakeholders, either by nationally recognised recommendations, or statistical formulae. 3. The bottom-up approach; staffing levels are calculated from local, ward-level

information such as the dependency of patients.

Also, combinations of the three different approaches can be used where certain parts are for instance determined by a consensus approach and other by top-down regulations. Although the research of Arthur & James (1994) is performed on nurses, no nurse-specific characteristics are implied. These staff requirement methodologies can therefore be applied to physicians.

Shift scheduling 3.3.3

This phase determines how many shifts to attain for certain patient groups and how many workers to assign to each shift type, in order to cover the staffing requirements. Shift scheduling aims to satisfy the minimum coverage requirements; the minimum needed use of staff which results in meeting patient demands while meeting time-related rules and practices that define acceptable individual schedules for the staff and the hospital (Maenhout & Vanhoucke, 2013). This research has not found an overarching methodology to optimise staff shift scheduling.

Rostering 3.3.4

(17)

Page | 17 shown that this final step can, to a certain degree, reduce the effect of poor decisions made in the earlier tasks on customer service (Thompson, 1993). Uncertainties, such as the actual demand at a certain day, are namely clear at the time of rostering. This enable schedulers to adapt the final schedules to these possible changes and therefore to enhance the overall schedule quality.

3.4 Scheduling shared staff in hospitals – a trade-off

The scheduling of resources becomes more complex when they are shared. In this case, in each of the previous mentioned phases more complex, interdependent decisions need to be made. This section elaborates on previous research on scheduling shared staff in hospitals and furthermore states relevant performance criteria to assess different shared staff policies.

Previous research 3.4.1

Vissers (1994) has stated two ways of dealing with shared resources in hospitals: (1) set priority rules for patients who arrive at the same time (e.g. first come, first serve) or (2) establish allocate capacity policies to schedule staff to specific users. Drupsteen et al. (2013) have found that dedicated time slots are in practice preferred to allocate shared resource capacity.

Shared staff allocation in hospitals is researched regarding the allocation of nurses (Brusco & Showalter, 1993; Maenhout & Vanhoucke, 2013). The research of Maenhout & Vanhoucke (2013) assesses different scheduling policies on three kinds of decision variables namely (1) department dedication (the amount of staff flexibility), (2) work stretch (the amount of workdays that are sequentially worked) and (3) the amount and ratio of part-time & fulltime employees. The research of Maenhout & Vanhoucke (2013) has stated some important findings. Firstly, it is shown that if the level of sharing of nurses is increased, this enhances the resource utilisation but complicates scheduling and therefore possibly negatively influences patient service. This is in line with the in section 3.1 mentioned trade-off between clinic- and patient-based performance. Secondly, the research regarding the work stretch showed that lengthening the work stretch lowers the job satisfaction and tightens the effectiveness of care. Thirdly, the research showed that an increase of the percentage of part-time nurses enhanced the job satisfaction but decreased the overall quality of care. Unfortunately, these findings cannot be applied to physicians given their differing aspects before they are empirically tested. Physicians are usually far more specialised and therefore cannot be as easily be exchanged as nurses. Also, as mentioned in section 3.3, the patient-physician coupling hardens the flexibility of patient-physicians.

(18)

Page | 18 and rostering, have not been researched before. An inclusive overview of the scheduling process is however necessary according to (Hans, van Houdenhoven, & Hulshof, 2012). Furthermore, research regarding the influence of different allocation policies of shared physicians is also not done before. As stated before, physicians are the most important resources regarding patient-based performance and their interdependencies with other resources (Vissers, 1994). Therefore, the need for a research on the effect of a comprehensive shared physician scheduling policy approach which includes all the different phases of staff scheduling is worthwhile researching.

Performance criteria 3.4.2

Vissers (1994) has selected three clinic-based performance areas to focus on when researching the influence of allocation policies: (1) the level of resource need/use (which can e.g. be measured with the average utilisation), (2) the fluctuations in resource need/use (which e.g. can be measured with the deviation of the utilisation) and (3) the amount and level of violations of resource restrictions (which can e.g. with the amount of violated restrictions in a certain time period). Examples of restrictions are:

· The totalled sum of specialty resource allocation figures cannot exceed the total hospital capacity for each type of resource.

· The totalled sum of allocations for all types of resources to one specialty cannot exceed the capacity of the specialty, the same applies also to the individual specialist level.

· The allocations of specialists within a specialty to different activities for each day of the week should accord with the task structure agreed for the specialty, an example of this type of restriction would be that not more than two out of four surgeons should he allocated simultaneously to operating theatre sessions to allow for emergency attendance.

· Allocations should take into account opening hours of facilities or periods of restricted use of a department; an example of the latter type of restriction is the agreement between a nursing ward and specialists not to make a ward round to see patients during patients' lunch hours or relatives' visiting hours.

Regarding patient-based performance, Drupsteen (2013) has stated patient throughput-time and its corresponding variability as the most important performance criteria. Different definitions exist for the throughput time of patients in hospitals. In this research, admission times will be used. Klassen & Rohleder (2004) have defined admission times as the time between the awareness of a patient need for an appointment and the date of the appointment. The choice for admission times is that they can give a good indication of the used scheduling policies where better scheduling policies can result in shorter admission times. In hospitals, these admission times are often (top-down) restricted.

3.5 Summary

(19)

Page | 19 on patient-based performance criteria and a positive effect on clinic-based performance. Negative on patient-based performance, since it hardens scheduling and therefore can negatively influence admission times, and positive on clinic based performance since it enhances resource utilisation. When dealing with physicians, well-designed staff scheduling policies rules can be used to balance this. Staff scheduling policies can be classified in four different phases namely; (1) forecasting demand, (2) determining staffing requirements, (3) shift scheduling and (4) rostering.

The influence of staff scheduling policies on performance in hospital environments is, on a scarce notion, researched regarding nurses. Here, the determining staff requirements and shift scheduling stage are only addressed while dismissing the forecasting demand and rostering phases. It is found that hospitals should maintain an comprehensive overview of all phases when aiming to improve scheduling performance. Furthermore, the literature falls short on assessing staff allocation policies regarding physicians. Knowledge of this is relevant given their ‘leading’ nature and therefore possibly significant effects on patient-based performance. To conclude, the literature specifically falls short on:

· A comprehensive approach of scheduling policies which includes all phases of staff scheduling.

· The effect of different shared physician scheduling policies, classified on clinic-based performance criteria such as the resource utilisation, the resource utilisation variability and the amount of violations of resource restrictions and patient-based criteria such as admission times and its corresponding variability.

(20)

Page | 20

4 System Description

This chapter provides a system description for the department under study. In section 4.1 patient population and routings, care services as delivered in three sub-departments, and associated (staff) resources are characterized. Next, in section 4.2, physician scheduling is considered in more detail according to the four phase framework introduced in chapter 3.

4.1 System overview

This section will provide overview of the UMCG-GHD and will elaborate on the occurring processes.

Patient characteristics 4.1.1

Patients with gastroenterology- and hepatology-diseases are treated at the UMCG-GHD. These patients are classified into five, disease based, groups; (1) Inflammatory bowel disease (IBD), (2) hepatology, (3) transplantation, (4) oncology and (5) general gastroenterology- and hepatology-patients. An important characteristic of these diseases is that they mostly are chronic diseases which are seldom fully cured. Also taken the Dutch, by law arranged, free physician choice for patients, this means that a big share of the UMCG-GHD visits are done by known patients because generally, patients like to be served by the same physician.

Clinical processes 4.1.2

The UMCG-GHD consists of three sub-departments; the outpatient clinic, the endoscopy centre and the clinical supervision. The labour at the department, especially at the outpatient clinic and the endoscopy centre, is very physician labour-intensive.

In the outpatient clinic no equipment, besides some very basic equipment such as blood-pressures measuring systems, is used to perform treatment. Here, the advised treatment is therefore mostly diagnosed through consultation. The advised treatment can for instance take place at the clinical supervision, at another department of the UMCG or somewhere else, or patients can for instance be treated by a prescribed medicine. No nurses are needed to assist the physicians in their work at the outpatient clinic. If setting the diagnosis at the outpatient clinic is not possible, the patient is referred to the endoscopy centre. At the outpatient clinic, an important distinction that is made between patients is whether they are new or control patients. For new patients, a norm time of thirty minutes is set and for control patients, the norm is fifteen minutes.

(21)

Page | 21 The clinical supervision holds patients that need to recover. Patients can arrive from other UMCG-departments or from the endoscopy centre. If treatment is performed at the clinical supervision it only entails the ingestion of medicine. If complications arise, patients are redirected to another (sub-) department for further treatment.

Patient routings 4.1.3

Figure 4.1 below represents the possible patient routings at the UMCG-GHD. Here, the outpatient clinic is abbreviated with OC and the endoscopy centre with EC. This figure is explained on the next page.

(22)

Page | 22 Patients that are referred to the UMCG-GHD have to make an appointment at the department themselves. As can be seen in figure 4.2, a first important distinction is if patients are known to the UMCG-GHD or not. If not, it is important to mention that in principle, patients are first helped at their GP or other hospitals before they are referred to the UMCG-GHD. This is done because of the fact that the UMCG is a specialised hospital and therefore treatment there is more expensive than somewhere else. Furthermore, patients that have a disease that first needs treatment at another department of the UMCG but also need treatment to cure gastroenterology- and hepatology-symptoms are also referred to the UMCG-GHD. This for instance is the case with patients dealing with cancer in their intestines.

After the patients have translated their referral in an appointment they are treated at either the outpatient clinic or at the endoscopy centre, dependent of their symptoms. At the outpatient clinic, a physician can decide whether further treatment at the endoscopic centre or if the clinic is needed or not. For some endoscopic treatments extensive feedback of the endoscopy is needed at the outpatient clinic. For the new patients who follow this ‘outpatient clinic – endoscopy centre – outpatient clinic’ route, a short connection is required to be able to set the diagnosis as soon as possible. Furthermore, at the outpatient clinic, if it is discovered that the symptoms of a certain patient are really bad/dangerous they are immediately allocated to be cured at the clinical supervision for a while. This can differ from a couple of hours, to a couple of weeks. Patients from other UMCG-departments can also be moved to the clinical supervision of the UMCG-GHD directly. This is the case with patients with other type of diseases which also effect the gastroenterology- and/or hepatology-parts of the patient.

After patients are treated at an UMCG-GHD sub-department they can be referred to make an appointment at another UMCG-department (when the causes of the symptoms are not gastroenterology and hepatology based for instance) or somewhere else (at their GP or at another hospital for instance). Also, the patients can be dismissed to make a ‘control’ appointment to check the state of the disease of the patient later on. This control appointment needs to be made before a certain time span set by the physician. Lastly, after treatment, a patient can also be cured of their diseases. But as mentioned, this is hardly the case given the chronic nature of the gastroenterology- and hepatology-diseases. This means that patients usually stay in the system until they pass away.

4.2 Physician scheduling

This section further elaborates on the shared physician scheduling policies used by the UMCG-GHD. This description is structured by the four-phases of staff scheduling as presented in chapter 3.

Forecasting Demand 4.2.1

(23)

Page | 23 that strict in reality since the total amount of treated patients has really fluctuated over the last couple of years (about 5% growth from 2013-2014 and about 5% shrinkage from 2014-2015). Exceeding the limit simply means a certain non-funded amount of treated patients and not meeting patients targets results in a waste of budget. Physician capacity for both amounts of patients seemingly showed to be available. The forecasted demand is evaluated once per year, at the end of the year. Therefore, no feedback loops on patient-based performance are present. Finally, the forecasted is based on separate appointments. Therefore, routes that require a short connection (such as the ‘outpatient clinic-endoscopy centre-outpatient clinic- route’) are also not considered.

The UMCG-GHD, as most UMCG-departments, is known for its high level of expertise and its wide variety of treatment capabilities. This, supported by the free physician choice, enhances the (local) patient demand for the UMCG-GHD instead of other hospitals. However, for non-complex treatments, the treatment can also take place in other (less specialised) hospitals and the treatment is more expensive at the UMCG-GHD. Therefore, patients are if possible advised to go to another hospital for treatment. Because of this it can be said that the actual total demand exceeds the restricted amount. Since the UMCG-GHD does not collect these actual figures it unfortunately cannot be stated how much the current demand is in total.

Determining staffing requirements 4.2.2

The UMCG-GHD staff requirements are not determined based on patient demand but have historically grown. Past results have shown that the needed to be treated patients can be treated by this amount of physicians. The general policy that the department cannot grow furthermore also holds for the staff capacity. In practice, this means that new physicians are only hired when others leave.

(24)

Page | 24 Physician Outpatient Clinic Endoscopy centre Clinical

supervision 1 Fully skilled except

hepatology

- Stomach &

intestines

2 IBD and

GHD-general

Diagnostic and therapeutic endoscopy, expendable stents, endo-echography & ERCP.

Stomach & intestines

3 GHD-general Diagnostic and therapeutic

endoscopy, expendable stents, double balloon and ERCP.

Stomach & intestines 4 Hepatology & transplantation - Liver 5 Hepatology, & oncology

Diagnostic and therapeutic endoscopy, expendable stents, endo-echography & ERCP.

Stomach & intestines and liver 6 Hepatology & transplantation Diagnostic, therapeutic endoscopy and expendable stents. Liver 7 IBD & transplantation (only small intestine) Diagnostic, therapeutic endoscopy and expendable stents.

8 Hepatology and

transplantation

Diagnostic and therapeutic endoscopy, expendable stents and ERCP.

Liver

9 IBD and

GHD-general

Diagnostic and therapeutic endoscopy, double balloon, expendable stents, ERCP and function research. Stomach & intestines 10 Hepatology and transplantation - Stomach & intestines and liver 11 IBD and GHD-general

Diagnostic and therapeutic endoscopy, expendable stents, double balloon and ERCP

12 IBD Diagnostic and therapeutic

endoscopy, expendable stents, endo-echography & ERCP.

Stomach & intestines 13 Hepatology, transplantation and oncology Diagnostic, therapeutic endoscopy and expendable stents..

Liver

(25)

Page | 25 Furthermore, the department also employs junior physicians. These have completed their education but still need some practical experience before they can work as standalone physicians. Junior physicians are capable of performing all skills of table 4.1 but they need supervision in the clinical supervision and in the endoscopy centre. In the endoscopy centre, a maximum of three junior physicians can be supervised by one senior physician. In the clinical supervision, the junior physicians (three) are always supervised, but require only a certain physicians capacity, namely two hours every day and one full day a week by two senior physicians. Junior physicians are completely free in choosing their educational preferences. This makes the availability of junior physicians very unstable and hard to predict.

The department finally also holds interns. These are medicine students who help at the department to attain some field experience. However, these are not authorised to set diagnoses or perform treatments alone. Since they always need supervision when performing treatments, which requires capacity of the physicians, they are disregarded in this research.

Shift scheduling 4.2.3

This stage holds the determination of the amount and length of shifts in such a way that the forecasted demand can be satisfied. At the UMCG-GHD, the length and amount of shifts are however not based on the to be treated patients but on the availability of the physicians. To facilitate this, a base-roster is made where the availability of physicians is translated into half-day shifts. This determines how many shifts of the outpatient clinic and endoscopy centre can be realised. This base-roster is made once per month and therefore does not have a rolling planning horizon. The formal procedure is that physicians communicate their expected availability two months in advance of the realisation of the base-roster, so that it can be realised one month in advance of the start of the roster. However, the base roster is usually not established before this monthly deadline but usually one or two weeks in advance of the start of the base-roster. According to the UMCG-GHD head scheduler, this is due to the uncertainty of junior physician availability.

The amount of physicians allocated to the clinical supervision ward is always the same. There are always three junior physicians needed and for two hours a day also two senior physicians. Except for Wednesday where, because of extensive patient visits, the senior physicians are needed the entire day.

Rostering 4.2.4

This step normally entails assigning physicians and subsequently patients to shifts. Furthermore final adjustments due to changes in patient or physician availability, can be made. In the UMCG-GHD, the physicians are already assigned to shifts in the shift scheduling step due to the fact that the shifts are designed according to their availability. Therefore, the remaining work entails coupling the patients to the shifts and adjusting the base-roster to changes in the, in the base-roster used, availability of physicians.

(26)

Page | 26 physicians are the leading resource. In the clinical supervision the availability of beds is leading. In order for the scheduling department to be able to make appointments, the different sub-departments therefore have to communicate their availability of physicians and beds respectively. This is represented in figure 4.3 by the dotted lines. Although the UMCG-GHD does not schedule patients in the clinical supervisions directly they however need to know its availability because of the scheduling of endoscopies which need treatment at the clinical supervision (see figure 4.2).

The availability of physicians at the UMCG-GHD is hard to predict. Besides the fact that they also perform educational and research activities, they can also take days off for conferences or vacation which is not restricted. Therefore, it often happens that physicians plan vacations or conferences or sorts when they are already assigned to shifts in the base-roster. If this happens the scheduling needs to be adjusted. To stress this fact, this freedom in availability for physicians even causes the head scheduler to be uncertain of the total yearly availability of physicians. Furthermore, as mentioned, the junior physicians have total freedom in their educational preferences. This also decreases the total physician availability.

4.3 Summary

In this chapter, the UMCG-GHD is characterized on the presented four phases of staff scheduling. This has led to the following findings:

1. Forecasting demand. The UMCG-GHD does not forecast it’s expected demand but aims to meet the same amount of treated patients as the respective previous year. These are, as a consequence of the influence of the insurers, caused by budget wise considerations. However, reality deviates from these seemingly formal rules. The total amount of treated patients namely has fluctuated substantially the last couple of years. Furthermore, no feedback loops on patient-based performance are present and routes that require short admission times are not regarded as a chain but as separate appointments.

2. Determining staffing requirements. The staff turned out not to be based on forecasted

patient demand, but on historical results and showed to be sufficient to meet patient targets in the past. Furthermore, it was found that the UMCG-GHD has no flexibility in adapting the available staff with respects to the amount of physicians and their skills.

3. Shift scheduling. The shifts at the outpatient clinic and the endoscopy centre are not

designed based on patient demand but on physician availability. This is done via a base roster which is monthly renewed. The shifts at the clinical supervision are constant.

4. Rostering. Because of the fact that the shifts are designed according to physician

(27)

Page | 27

5 System Analysis

This chapter assesses the UMCG-GHD shared physician scheduling policies on the performance criteria as presented in conceptual model of section 2.3. Section 5.1 elaborates on the taken approach of the performed analysis. Section 5.2 assesses the results of patient-based performance and section 5.3 of clinic-patient-based performance. A cause and effect analysis presented in section 5.4, that seeks to relate current set-up of the physician scheduling system to observations made on patient-based and clinic based performance. Section 5.5. summarises the main findings.

5.1 Approach

Scheduling policies are in this research assessed for their effects on clinic- and patient-based performance criteria. These are again presented in figure 5.1:

Clinic-based criteria: · Resource utilisation · Resource utilisation variability · Resource restriction violations Patient-based criteria: · Admission time · Admission time variabilty

Figure 5.1 Patient- and clinic-based performance criteria

The method of assessment of the current UMCG-GHD physician scheduling policy is explained in the upcoming sections.

Addressed patient population 5.1.1

The performance criteria will be tested on patients according to the patient group/treatment classification of table 4.1. Although a wider variety of patients are treated in the outpatient clinic and a wider variety of treatments is provided in the endoscopy centre, these variations will be represented by their belonging ‘main category’. This is done to enhance the overview of this analysis.

Patients visiting the clinical supervision are not as deeply analysed as the other two sub-departments since the leading resource of this department are not physicians but beds. Therefore, the physician scheduling policies cannot as much be analysed by clinical supervision performance levels as by the performance levels of the other two sub-departments.

(28)

Page | 28

Patient-based criteria 5.1.2

The admission time can be measured by the time between the awareness of a patient need for an appointment and the date of the appointment (Klassen & Rohleder, 2004). The admission times will be presented by the 2015 average. Furthermore, the percentage of times where the norm times are met is also presented. The admission time variability is represented by the standard deviation and the relative variation; the coefficient of variation (𝐶𝑉 = 𝜎/𝜇 ), which provides more insight. The CV has the standard deviation and the mean as parameters.

The UMCG-GHD admission times of new patients will be compared by governmental determined norms. These are the same for both the outpatient clinic as the endoscopy centre; 28 days. Furthermore, the UMCG-GHD has a target of 95% of the times in which these 28 days limit should be met.

Besides departmental focused assessments were patient admission time performance is analysed on separate sub-departments, the interdepartmental performance is also analysed. The in section 4.1 described ‘outpatient clinic – endoscopy centre – outpatient clinic’ route for new patients is analysed. This route is chosen because is concerns patients for which the admission times are critical and therefore can provide valuable insights.

Clinic-based criteria 5.1.3

The resource utilisation represents the share of time that physicians spend to the treatment of patients. Therefore, this is analysed in the outpatient clinic and the endoscopy centre by a combination availability figures and the percentage of time spent on treating patients when available. Bed occupancy data of the clinical supervision is analysed to be able to say something about the amount of physicians allocated there. To provide more insight in the skills determination at the UMCG-GHD, an analysis on skills utilisation is also performed. This is done by dividing the amount of time for demanded treatments by the times it is employed. The higher this number, the higher the utilisation i.e. the more critical a certain skill is. Also, a rank on the criticalness is given. Here the highest utilisation gets the highest rank and so on. Here junior physician skills are neglected because they can be assigned to all skills.

The resource utilisation variability is represented by the standard deviation and the coefficient of variation (𝐶𝑉 = 𝜎/𝜇) of the resource utilisation.

As the name suggests , resource restriction violations are unwanted. Therefore, they give a good indication of the quality of the scheduling policies at the UMCG-GHD. The resource restriction violations will be presented by the amount of their occurrence. Based on experiences by the UMCG-GHD head scheduler and the resource restriction violation examples as presented in the theoretical background the following resource restriction violations were determined:

(29)

Page | 29

· Rescheduled appointments: these are appointments that are rescheduled because of wrongly bookings or changes in physician availability.

· Overtime: This occurs when physicians are not able to deal with the amount of patients in pre-planned slots. Therefore, they have to work longer than expected. Considerations for the determination of the above violations were their occurrence in the UMCG-GHD and the availability of data.

Data sources 5.1.4

The patient-based criteria are analysed with the help of patient-data which is tracked digitally by the UMCG logistics department with the help of scheduling software named X-care. The UMCG-GHD has provided data on patient appointment-data which hold data about the date of a patients request for an appointment, the date of the appointment and the patients diagnosis (outpatient clinic) or type of treatment (endoscopy centre). Unfortunately, the data showed imperfections. For the endoscopy data, X-care does not make a distinction between new and control patients. To be certain of the fact that data about new patients is analysed therefore a sample of the data is used. Because of the fact that new patients have to visit the outpatient clinic first before they can visit the endoscopy centre, this sample is determined by focussing on endoscopy patients who were new patients in the outpatient clinic and after that had to visit the endoscopy centre. Furthermore for the ‘outpatient clinic – endoscopy centre – outpatient clinic’ route a sample of the data is also used to be certain that is concerns data about new patients. This sample is determined in a similar way as the endoscopy sample, but this time also adding the constraint that they have to have visited the outpatient clinic afterwards.

The resource utilisation and resource utilisation variability are analysed with the help of provided physician-availability data. This data is provided by the head of the UMCG-GHD. Furthermore, this data is expanded by interviewing local experts (UMCG-GHD schedulers and the head scheduler) on their observations on system performance. Furthermore, the assessment on the physician skills is done based on combining the data of appendices A & B and the information of table 4.1. Of the resource restriction violations, the amount of rescheduled appointments is provided by X-care and the amount of double bookings and the amount of overtime is based on the UMCG-GHD head scheduler’s estimations.

Cause and effect analysis 5.1.5

After the performance criteria of the UMCG-GHD are determined, the possible causes of the determined performances are determined. This is done by relating the set-up of the scheduling system (chapter 4) by the findings of this chapter, the theoretical background and by input from interviews with UMCG-GHD staff.

(30)

Page | 30 5.2 Current patient-based performance

The patient-based performance is analysed on the admission time and admission time variability.

Admission time 5.2.1

Table 5.1 represents the average admission times and the percentages of patients that are treated within the norm of the outpatient clinic and the endoscopy centre for 2015:

Patient group

Outpatient clinic Endoscopy centre

Average

admission time

Share of patients treated within the norm Average admission time Share of patients treated within the norm

New 37 days 51% 20 days 72%

Control 90 days - 71 days -

Table 5.1 2015 admission time performance at the outpatient clinic and the endoscopy centre

As can be seen in table 5.1, the average admission time criteria for the outpatient are (by far) not met. However, for the endoscopy centre, the threshold of 28 days is met on average but the share of patients treated within the norm is unwanted low.

As mentioned in section 4.1, a patient route that requires a short connection is the ‘outpatient clinic – endoscopy centre – outpatient clinic’ route for new patients. Figure 5.2 presents the total average admission times for this route, divided in sub routes (where date of appointment call --> OC1 represents the time between the call and the first outpatient clinic visit, OC1-->EC the time between the first outpatient clinic visit and the endoscopy centre visit and EC-->OC2 the time between the endoscopy centre visit and the second outpatient clinic visit):

Figure 5.2 2015 average UMCG-GHD new patients admission times for the ‘outpatient clinic - endoscopy centre - outpatient clinic’ route

(31)

Page | 31 For the ‘outpatient clinic-endoscopy centre-outpatient clinic’ route, the total norm time is 84 days. As can be seen in the above picture, the total average admission time is 124 and therefore the norm time is likely not to be met for most patients. The average admission time for the sub routes is quite evenly divided and is 47 for the admission time until the first outpatient clinic visit, 42 for the admission time between the first outpatient clinic visit and the endoscopy centre and 35 between the endoscopy centre visit and the second outpatient clinic visit. Concluding it can be stated that the total admission time threshold of 84 days and the admission time threshold of 28 days at any of the sub routes are not met.

Admission time variability 5.2.2

Table 5.2 represents the variation (measured with the standard deviation) and coefficient of variation (relative variation) of the admission times of all patients at the different sub-departments:

Patient group Outpatient clinic Endoscopy centre

Standard deviation Coefficient of variation Standard deviation Coefficient of variation

New 35 days 1.0 16 days 0.8

Control 100 days 1.1 111 days 1.6

Table 5.2 2015 admission time variability at the outpatient clinic and the endoscopy centre

The standard deviation and the coefficient of variation are relative high. This causes the earlier presented data, on the share of patients treated within the admission time norms, to not be met.

Table 5.3 presents the 2015 average admission times and admission times variabilities per patient group/treatment for new patients:

Department Patient group/treatment Admission times Standard deviation Coefficient of variation Outpatient clinic GHD-general 60 38.4 0.6 Hepatology 19 18.8 1.0 IBD 58 56.9 1.0 Oncology 13 16.1 1.2 Transplantation 34 30.4 0.9 Endoscopy centre Diagnostic endoscopy 33 25.3 0.8 Therapeutic endoscopy 26 28.1 1.1 Endo-echography 32 21.1 0.7 ERCP 21 18.0 0.9 Double-balloon 10 1.4 0.1 Function research 13 13.2 1.0 Expandable stents 5 4.9 1.0

(32)

Page | 32 Table 5.3 shows that at the outpatient clinic, the GHD-general and IBD patients cause the highest admission times and that the IBD and oncology scored the highest coefficient of variation. The fact that oncology patients relatively score the lowest average admission times is that because of the serious nature their disease, they are often prioritised. At the endoscopy centre, the diagnostic- & therapeutic-endoscopy and the endo-echography skills cause the highest admission times and admission time variability.

5.3 Current clinic-based performance

The clinic-based performance is measured on three criteria namely resource utilisation, resource utilisation variability and resource restriction violations.

Resource utilisation 5.3.1

The resource utilisation is determined by the percentage of time that physicians can spend to treating patients when available and the percentage of time that physicians are available. An interview with the UMCG-GHD head scheduler has shown that about 20% of the time of physicians cannot be spent to treating patients because of administration tasks. Furthermore, an analysis of the patient data showed that about 1% of the appointments is cancelled last minute. Because of the current physician availability focused design of the scheduling policies this however means that the remaining time can fully be spent to treating patients. Therefore, it can be stated that the time that physicians are available is the bottleneck in the determination of the resource utilisation.

Ideally physicians are 42 weeks a year available for the treatment of patients. This is 210 days a year. Based on this and the availability of physicians in 2015, the following availability percentages are derived:

Physician # days available Availability percentage

1 206 98% 2 142 68% 3 190 90% 4 209 100% 5 104 50% 6 211 100% 7 182 87% 8 201 96% 9 200 95% 10 173 82% 11 167 80% 12 103 49% 13 163 78%

(33)

Page | 33 0 5 10 15 20 25 30 1 8 16 23 30 38 45 52 O c c upi e d be ds Weeknumber Maximum daily occupation As can be seen in table 5.4, the physicians in general have a quite high availability. The low resource of physician 2 can be explained because he spends a lot of time in educating junior physicians and interns. Physician 5 & 12 where ill for a long time in 2015 and this therefore explains their relative low level of availability.

For the junior-physicians, the data was however not so sufficient as the physician availability data. Therefore, assumptions had to be made. Given the fact that junior-physicians have to meet certain hour-targets to be able to finish their education. Therefore, an availability of 80% is assumed. The amount of junior-physicians was quite stable in 2015. There worked nine junior physicians in Q1 in the UMCG-GHD and eleven in Q2, Q3 and Q4.

To the clinical supervision, as stated before, a fixed amount of physicians is always assigned. If the assumption that more patients leads to a higher resource utilisation is applied, the following 2015 physician resource utilisation can be obtained:

Figure 5.3 Maximum weekly occupation at the clinical supervision in 2015

(34)

Page | 34 The table below represents the utilisation of physicians skills.

Department Skill # Employed Total needed time (days) Time demanded per skill (days) Rank of level of criticalness Outpatient clinic IBD 6 195 33 3 Hepatology 6 188 31 4 Transplantation 7 124 18 8 Oncology 3 36 12 10 GHD-general 4 134 34 2 Endoscopy centre Diagnostic endoscopy 13 312 24 6 Therapeutic endoscopy 13 477 37 1 Endo-echography 3 82 27 5 ERCP 7 106 15 9 Double-balloon 3 16 5 11 Function research 1 22 22 7 Expandable stents 13 16 1 12

Table 5.5 2015 utilisation of physician skills

As can been seen in table 5.5, some skills are utilised a lot. At the outpatient clinic, the most critical skills (by time) are GHD-general, IBD & hepatology. For GHD-general and IBD, this is in line with the analysis about the admission times per disease but with hepatology patients this is not the case. The most critical skills at the endoscopy centre is therapeutic endoscopy although they are employed by all physicians. This is in line with wat is found in table 5.3.

Resource utilisation variability 5.3.2

With the help of the availability-data on physicians, the resource utilisation variability is determined. This is done by determining the total availability of physicians for each week in 2015 and afterwards the variability and the coefficient of variation are calculated. The standard deviation of the availability in 2015 turned out to be 8.95 days and the coefficient of variation turned out to be 0.21. This relative small variability in availability of physicians however can still have an impact on the variability of admission times.

Resource restriction violations 5.3.3

The following results were gathered concerning the resource restriction violations:

Shared physician scheduling policies