Hospital Staff Capacity Planning

(1)

Hospital Staff Capacity Planning

An architecture for physician planning

Master thesis

MSc Technology & Operations Management

University of Groningen, Faculty of Economics and Business

Student: Petra de Jonge Student number: S2400901 E-mail: p.j.de.jonge@student.rug.nl

University supervisor: Dr.ir. D.J. van der Zee Second assessor: Dr. N.B. Szirbik Company supervisor: Ir. T.J. Hoogstins

(2)

Abstract

Nowadays hospitals are struggling with their capacity planning due to growing demand for care, higher expectations and constraints on the availability of resources. Problems are mostly solved when they are imminent, while tactically and strategically allocating and organizing available resources might have prevented these problems from happening. In current literature there is a lack of information on how staff capacity planning in healthcare should be organized. Alterna-tively, literature from other industries cannot directly be used within healthcare because of the unique nature of the healthcare industry. By means of a design science study this research de-velops and evaluates a Generic Process Planning Architecture (GPPA) for a physician shift planning system elaborating on decision support of the planning organization, procedures and techniques. The GPPA can be used in practice by setting up a capacity planning system for a new department or function as a reference architecture for existing departments that want to check the maturity of their existing system. This thesis contributes to literature by giving guid-ance on what kind of activities should be covered in a physician capacity planning and addresses the underdevelopment of planning and control operations in healthcare.

(3)

Preface

This thesis is my final assignment for the master Technology and Operations Management at the University of Groningen. During the process of writing several persons contributed to my thesis. In order to gather information and make decisions about the design, I had to interview employees of various departments of the University Medical Center Groningen. My thanks and appreciation to all those people who were willing and able to offer some of their valuable time to help me with my research. A special thanks to T.J Hoogstins and I.A van der Weide, for their feedback, made arrangements and given guidance. Without the great help of the people of the University Medical Center Groningen this research would not have been possible. Further, I gratefully acknowledge the invaluable feedback and suggestions from my supervisor dr. D.J. van der Zee and my co-assessor dr. N.B. Szirbik. Thank you both for your given time and energy. Finally, I would like to thank my family and friends for their confidence and support.

I hope reading my thesis will provide you with some valuable insight into hospital staff capacity planning.

Petra de Jonge

(4)

2.! Theoretical background _________________________________________________ 8! 2.1 ! Challenges faced in staff capacity planning _______________________________ 8! 2.2 ! Defining capacity planning – planning architectures ________________________ 9! 2.3 ! Common issues in staff capacity planning _______________________________ 11! 2.4! Capacity measurement and algorithmic support __________________________ 12! 2.5! Summary of findings ________________________________________________ 12!

3.! Methodology__________________________________________________________ 14! 3.1! Motivation of the research ___________________________________________ 14! 3.2 ! Research objective __________________________________________________ 14! 3.3! Conceptual model __________________________________________________ 15! 3.4! Research outline ___________________________________________________ 16!

Phase 1 – Problem investigation ___________________________________________ 17!

Phase 2 – System characterization and solution exploration _____________________ 17!

Phase 3 – Design of the artefact (GPPA) ____________________________________ 18!

Phase 4 – Illustration and evaluation of the artefact ____________________________ 18!

4.! System characterization ________________________________________________ 19! 4.1 ! Approach _________________________________________________________ 19!

4.1.1! Scope & detail ___________________________________________________ 19!

4.1.2! Process mapping - primer of BPMN use for describing physical architectures _ 19! 4.2! Department A _____________________________________________________ 21!

4.2.1! Resources and patient flow _________________________________________ 22!

4.2.2! Planning organization _____________________________________________ 22!

4.2.3! Time line _______________________________________________________ 22!

4.2.4! Tasks and triggers ________________________________________________ 22!

4.2.5! Information sources _______________________________________________ 24!

4.2.6! Decision support systems __________________________________________ 24! 4.3! Department B _____________________________________________________ 24!

4.3.1! Resources and patient flow _________________________________________ 24!

4.3.3! Time line _______________________________________________________ 26!

4.3.4! Tasks and triggers ________________________________________________ 26!

(5)

4.4.1! Resources and Patient flow _________________________________________ 27!

4.4.3! Time line _______________________________________________________ 29!

4.4.4! Tasks and Triggers _______________________________________________ 29!

4.4.6! Decision support systems __________________________________________ 30!

5.! Solution exploration: Analysis of the Physical Architectures __________________ 31! 5.1! Approach _________________________________________________________ 31! 5.2 ! Summary of elements ________________________________________________ 31! 5.3! Planning organization _______________________________________________ 34! 5.4! Timeline __________________________________________________________ 34! 5.5! Forecasting demand ________________________________________________ 34! 5.6! Determining staff requirements ________________________________________ 34! 5.7! Shift Scheduling ____________________________________________________ 35! 5.8! Rostering _________________________________________________________ 35! 5.9! Information sources _________________________________________________ 35! 5.10! Decision support systems ____________________________________________ 36! 5.11! Summary of findings ________________________________________________ 36!

6.! Design of the General Process Planning Architecture. _______________________ 37! 6.1! Key design steps ___________________________________________________ 37! 6.2! General Process Planning Architecture._________________________________ 38! 6.3! Planning organization _______________________________________________ 40! 6.4! Forecasting demand ________________________________________________ 41!

6.4.1! Collecting data ___________________________________________________ 41!

6.4.2! Analyze available data_____________________________________________ 41! 6.5! Determining staff requirements ________________________________________ 41!

6.5.1! Estimate expected required capacity __________________________________ 41! 6.6! Shift scheduling ____________________________________________________ 42!

6.6.1! Make initial year planning __________________________________________ 42! 6.7! Rostering _________________________________________________________ 43!

6.7.1! Make month planning _____________________________________________ 43!

6.7.2! Set physician planning open for patient planning ________________________ 44!

7.! Use of the General Process Planning Architecture __________________________ 45! 7.1! Approach _________________________________________________________ 45! 7.2! Department D _____________________________________________________ 46!

(6)

7.7.6! Decision Support Systems __________________________________________ 49!

7.7.7! Intended situation for department D __________________________________ 49! 7.3! Comparing the physical architecture of department D with GPPA ____________ 49!

7.3.1! Commonalities and differences ______________________________________ 52! 7.4! Evaluation ________________________________________________________ 52!

8.! Discussion ____________________________________________________________ 54! 8.2! Research contributions ______________________________________________ 54! 8.1! Applicability of the General Process Planning Architecture. _________________ 55! 8.4! Limitations ________________________________________________________ 56!

9.! Conclusion ___________________________________________________________ 57!

(7)

1. Introduction

Staff are among the most important resources of healthcare systems. They account for the larg-est share of the healthcare budget in most countries (Vujicic and Zurn, 2006). Nowadays the availability of human resources is under pressure in terms of shortages of staff in hospitals and difficulties in hiring additional staff due to their unavailability at the labor market. According to the Dutch Institute for Employee Insurance (UWV) the shortage of available staff resources has led to an increased workload; “60% of the caregivers and nurses need to work extra shifts, the number of on-call workers is increasing and about 65% of the care givers and nurses indicate that they feel increased work pressure over the last couple of years” (2019). This pressure on human resources is mainly caused by decreasing healthcare budgets and ageing population re-sulting in an increasing demand and complexity of healthcare (Poksinska, 2013; Vissers, Ber-trand and De Vries, 2001).

Healthcare policy makers increasingly recognize the need for more integrated planning and control of human resources in healthcare, in particular making the management of human re-sources responsive to system needs and design, instead of vice versa (Bloor and Maynard, 2003). Other industries such as the manufacturing industry, provide examples and models on how to manage staff capacity. However, it should be noted that successful manufacturing plan-ning and control concepts cannot directly be copied, because of the unique nature of healthcare services (Hans, et al., 2012). Healthcare systems tend to be complex because processes are often highly coupled across, and along, organizational levels, while sharing scarce resources along clinical pathways and involve different time horizons (Nordlander, Van den Berghe, Schittekat, 2013).

(8)

however, is scarce. Furthermore, they do not provide guidance on how to organize and detail the various planning levels of the framework.

Another stream within research addresses algorithmic and mathematical support. Such as the models of Gar, McClean, Barton, Meenan and Fullerton (2012) and Costa, Ridley, Shahani, Harper, De Senna and Nielsen (2003). While algorithmic support is part of the puzzle when designing a capacity planning system, it does not directly clarify its design. An important limi-tation to algorithmic support is that optimization problems solved by these algorithms are mostly isolated, oversimplified problems, which do not consider shared resources across the organization (Nordlander et al., 2013).

In literature there is a lack of guidance on how to set up a staff capacity planning system at the various levels and - in doing so – to make (good) use of algorithmic support. Alternatively, the aim of this research is to contribute to existing literature by designing a generic process planning architecture (GPPA) for a physician shift planning system in hospitals. The focus on physicians is chosen in order to narrow the scope of this research and because physicians have a key role in hospitals. The generic architecture is meant to provide guidance to hospitals on how to or-ganize various planning tasks. Also the generic architecture should give a system view with information on how planning is organized amongst various planning levels and which support-ing information systems can, and should be used. To develop this generic architecture a design science study will be conducted in collaboration with 4 departments of a University Medical Center in the Netherlands. The study will be conducted according to the phases of the engineer-ing cycle of Wierengineer-inga (2014).

(9)

2. Theoretical background

This chapter strives to establish a solid theoretical base by reviewing previous research. Section 2.1 elaborates on the challenges faced in staff capacity planning. Thereafter capacity planning and its role in a planning hierarchy (Section 2.2) are defined. Followed by an elaboration on practical issues in capacity planning (Section 2.3) and a description of a capacity planning sys-tem set-up (Section 2.4). Further, an elaboration of capacity measurement (Section 2.5) and a summary of the main findings (Section 2.6) will be given.

2.1 Challenges faced in staff capacity planning

Hospitals are confronted with a growing demand for care, higher expectations for improved service delivery and have to cope with constraints on the availability of resources due to labor market issues (UWV, 2019; Vissers et al., 2001). The increase in demand for healthcare and a declining working population results in an increasing gap in supply and demand for healthcare, which in turn leads to an increased workload (Haczyński, Skrzypczak, and Winter, 2017). The increased workload puts the quality of care under pressure but also influences the health and private life of healthcare employees (UWV, 2019). More and more employees are unable to work due to a burnout (Prins, Hoekstra-Weebers, Van De Wiel, Gazendam-Donofrio, Sprangers, Jaspers and Van der Heijden, 2007). The unavailability of staff is increasing the workload, which leads to people leaving the system, and thus, to even more unavailability ( Rei-jenga, Zwinkels and Van Vuuren, 2004). This a vicious circle. Hence, pressure is put on capac-ity planning, seeking to adequately use staff that is available.

(10)

(5) Healthcare managers often feel that investing in better administration diverts funds from direct patient care.

2.2 Defining capacity planning – planning architectures

Hall (2012) defines hospital capacity planning as: “deciding on the amount of beds, staff, con-sulting rooms, equipment, etc. sufficient to enable an organization to meet demand for one or more packages of care while achieving specified service standards.” The level of resources required to have an adequate capacity depends on factors such as the productivity of the work-force, available medical technology, human resource education and training levels and a lot of other factors. (Vujicic and Zurn, 2006). The different facets of capacity planning are taken into account in multiple frameworks as proposed in literature, compare Vissers et al. (2001) and Hans et al. (2012).

(11)

Figure 1: Example application of the framework for healthcare planning and control to a general hospital (Hans et al., 2012).

Strategic planning can be seen as the bricks and mortar of an organization (Li, Benton, Keong, 2002). It addresses the structural decision making, has a long planning horizon and is based on highly aggregated information and forecasts (Hans, et al., 2012). Strategic planning on resource capacity sees to case mix planning, capacity dimensioning and workforce planning. Tactical planning addresses the “what, where, how, when and who” of operations (Hans, et al., 2012). The capacity planning on a tactical level is an important bridge between strategic goals and the operations. Operational planning (on- and offline), is aimed at short-term decision making, where offline planning concerns the in advance planning of operations and online planning involves monitoring the process and reacting to unforeseen or unanticipated events (Hans et al., 2012).

The four levels of control of the framework of Hans et al. (2012) can be linked to a four-step process as discussed in various literature (Defraeye & Van Nieuwenhuyse, 2016; Koole Pot, 2006; Maenhout & Vanhoucke, 2013). The four-step process consists out of the following steps:

1.! Forecasting demand (based on empirical data)

2.! Determining staff requirements to meet specific performances 3.! Shift scheduling to cover staffing requirements

(12)

examination of historical information (Cote and Tucker, 2001). Different methodologies can be chosen for the realization of a good fit to the to be forecasted population (Archer, 1980). How-ever, the methods should at least satisfy the following concerns: data should be readily availa-ble, existing staff members equipped with readily available tools should be able to perform the forecasting in-house and the forecasting method and its results should be understandable for either financial management staff and decision makers. Determining staff requirements sees to selecting the required staffing levels over time, in order to meet a specific performance target at a minimal cost (Defraeye & Van Nieuwenhuyse, 2016). The shift scheduling phase deter-mines how many workers needs to be assigned to each shift type, in order to cover the staffing requirements (Defraeye & Van Nieuwenhuyse, 2016). In order to do this correctly different types of shift should be determined based on different kind of patient groups. Rostering is the final step in which employees are assigned to shifts.

The four levels of control and the four-step process cannot be linked one to one in a sense that all levels correspond with the different steps. However, step one and two (forecasting demand and determining staff requirements) are linked to the first level (strategic planning), step three (shift scheduling) is linked to the second level (tactical planning) and step four (rostering) is linked to the third level (offline operational). An extra step to link to the last level could be re-rostering because monitoring and emergency coordination would most likely lead to re-roster-ing of the schedule.

2.3 Common issues in staff capacity planning

(13)

resources may be more effective and cheaper than solving problems at the last moment (Hans et al., 2012).

2.4 Capacity measurement and algorithmic support

In order to accurately forecast demand and determine the amount of shifts and staff required for those shifts, accurate measurement of theoretical and available capacity is of vital importance (Bamford and Chatziaslan, 2009). Because focus on capacity on a micro (departmental level) can potentially make a direct contribution to the ability of both the department and the organi-zation to achieve and maintain its quality and cost objectives (Bamford and Chatziaslan, 2009). Furthermore, it offers possibilities for safeguarding tolerable staff workload levels in order to prevent staff from being overworked. Yet, measuring capacity in healthcare is a challenging task due to different characteristics in terms of resource requirements, length of stay etc. (Adan and Vissers, 2002). Measurements and reports on the use of capacity are often not straightfor-ward, since the measurements are open to bias and misinterpretation (Bamford and Chatziaslan, 2009). This is mainly caused by conflicting stakeholder priorities (Simons, Russell and Kraus, 2004), interaction of capacity decisions with each other and shared resources (Bamford and Chatziaslan, 2009).

When searching for more in depth measurement methods in literature, one finds mostly algo-rithmic and mathematical support such as the models of Gar, McClean, Barton, Meenan and Fullerton (2012) and Costa, Ridley, Shahani, harper, De Senna and Nielsen (2003). While al-gorithmic support is an important part of the puzzle when it comes to capacity planning (it helps to make calculations and decisions based on data), it does not directly clarify capacity planning design. The current models mostly focus on the operational levels of various planning systems and still rely on rather theoretical and strict assumptions, lack a real-life implementation and do not consider shared resources (Defraeye & Van Nieuwenhuyse, 2016).

2.5 Summary of findings

(14)

information on algorithmic and mathematical support, which is mainly focused on the opera-tional levels of staff capacity planning. Algorithmic and mathematical support helps to make calculations and decisions based on data, but it does not directly clarify capacity planning de-sign. Thus, it is not clear how those two streams should lead to a capacity planning system that can be used in practice.

(15)

3.!Methodology

This chapter elaborates on the design of the research by explaining the motivation of the re-search (Section 3.1), stating the rere-search objective (Section 3.2) and providing a conceptual model (Section 3.3). Thereafter the choice of research methodology will be motivated and the research outline will be highlighted (Section 3.4).

3.1 Motivation of the research

This research is motivated by a capacity planning problem of a department of the University Medical Center Groningen. The department consists of three sub-departments: an outpatient clinic, a nursing ward and an endoscopy center. There is an increasing demand for care in terms of more, and more complex patients. This puts increasing pressure on available staff and re-sources, which are not always easy to come by. The complexity of care in this hospital is higher than in other hospitals because of a regional agreement: only the complex patients are forwarded to the University Medical Center Groningen. Whereas, more standardized services are provided at other hospitals in the region. The request of the department is to research how capacity plan-ning should be organized in order to be sustainable for the future. When searching in literature on how to organize a capacity planning system, a general structure of production planning and control of hospitals can be found. Zooming in a bit further into tactical capacity resource plan-ning it is found that a capacity resource planplan-ning should consist out of 5 stages: forecasting demand, determining staff capacity, shift scheduling, rostering and monitoring. However, when one wants to zoom in a little more and know what kind of activities should be covered in a physician capacity planning, the answer cannot be found in literature.

3.2 Research objective

(16)

Therefore, the following main research objective is determined:

To develop a generic architecture enhancing the efficient and effective development of a physi-cian planning system in hospitals aiming to contribute to patient service and cost minimization.

The primary objective of the generic architecture is to structure the planning process of a de-partment. The generic architecture should help departments by developing a physician capacity planning in an effective and efficient way. It should improve the efficiency of the developing process by giving insight in which functions should be developed and help make better choices. Further, it should decrease the lead time of the development process by making it easier to divide resource uses. The generic architecture should also increase the effectiveness of the de-veloped planning process in the sense that it should improve the performance of the department in terms of patient service and by minimizing costs.

3.3 Conceptual model

In order to design and evaluate the GPPA a conceptual model is established to define the struc-ture of the artefact and its context. The model in figure 2 shows the context of the conceptual architecture.

(17)

Figure 2: Conceptual model

The five levels of planning as described in figure 2 influence the performance of the department. The performance of the department can be measured in terms of patient service and costs.

3.4 Research outline

(18)

This thesis is a design science study in the sense that the design is bringing together different parts of the physician capacity planning process in a architecture, based on empirical knowledge obtained from a case study. The set-up of the research is based on the engineering cycle of Wieringa (2004) consisting out of 4 phases:

1.! Problem investigation

2.! System characterization and solution exploration 3.! Design of the artefact

4.! Illustration of the artefact and evaluation

There are four departments of the UMCG involved. The departments are chosen because col-lectively they are a good reflection of the variety of departments in this, and most likely other hospitals. Due to privacy and business confidentiality the UMCG requested that the depart-ments remain anonymous and will therefore be marked as department A till D. Department A till C are used in order to obtain empirical knowledge in order to construct the design. Depart-ment D is used to illustrate the relevance of the design.

Phase 1 – Problem investigation

Phase 1 consists out of a literature study to establish a solid theoretical base by reviewing pre-vious research and an unstructured interview with the planner and manager of department D, giving a first indication on how the capacity planning of the hospital is organized now.

Phase 2 – System characterization and solution exploration

(19)

Phase 3 – Design of the artefact (GPPA)

In phase 3, the analysis of phase 2 is used as input for the design. The relevant activities, tech-niques and decisions are integrated and used to design a generic process planning architecture of the planning organization, procedures and relevant techniques. This integration will lead to the aimed GPPA.

Phase 4 – Illustration and evaluation of the artefact

(20)

4.!System characterization

This chapter describes the capacity planning process of three departments. Section 4.1 elabo-rates on the approach taken and explains the notation adopted to detail physical architectures. Next, sections 4.2-4.4 describe planning processes for each department.

4.1 Approach

4.1.1 Scope & detail

The physical architectures describing each department entail four levels of planning, i.e., fore-casting demand, determining staff requirements, shift scheduling and rostering. Each process is described by detailing its physical architectures in terms of activities undertaken, their inter-linkage, planning staff involved, information sources and decision support systems as witnessed in practice.

4.1.2 Process mapping - primer of BPMN use for describing physical architectures

The physical architectures are developed using the process diagramming language BPMN. A legend of the various symbols that are used in the physical architectures can be found in figure 3 and the symbols will be discussed from here further.

Figure 3: Legend of used BPMN symbols.

(21)

in the process which can vary from user tasks to sending a task to another person, for example if one would request feedback or would need someone else to do something. A user task results into an intermediate event, which is the event (or state) that leads to a new task, or to an end

event, which means that the process has ended. A task can be informed by various data sources

such as a database or a data object. All types of data (digital and non-digital) that can influence the task can be marked as data objects. There are no limitations on what can be marked as data. A sending task results in a message intermediate catch event, meaning that someone else re-ceives a document or a task, leading to a task for the receiving participant. When there are multiple participants involved in the process, the process is placed in a pool in which every participant has his own lane, showing clearly for which part of the process the participant has a responsibility.

Sometimes tasks may happen at the same time without influencing each other. When this is the case a parallel gateway is shown in the process functioning as a fork gateway. Meaning that from that moment on, the process splits into multiple process paths that all must be followed, but that the sequence of the paths does not matter. If thereafter a second parallel gateway is shown, this functions as a join gateway. All paths coming in to the parallel (join) gateway have to be completed before advancing the process. Alternatively, it can happen that the flow of the process depends on a situation or a choice. If this is the case an exclusive gateway is shown, representing a decision to take exactly one path in the flow. The paths are mutually exclusive.

(22)

4.2 Department A

(23)

4.2.1 Resources and patient flow

The process of department A is described in the physical architecture in figure 4. This depart-ment is an outpatient clinic where 4 main sub specialisms dominate. Basically all patients can be assigned to one of the 4 sub specialisms. Every sub specialism is executed on a different day of the week. So if a patient is diagnosed with a disease in sub specialism A for example and this group is planned on Mondays, then appointments for that patient will basically always be on a Monday. The department has specific sub specialisms on every weekday except Wednesday. On Wednesday the department is open for general hours; for patients who are not (yet) assigned to one of the main sub specialisms.

4.2.2 Planning organization

On all levels of the process only the planner of the department is involved. However, the format for the clustering per sub specialism, in the sense of determining which illnesses belong to which sub specialism, is determined by the physicians. This format is established once and will most likely stay the same for the coming years.

4.2.3 Time line

Forecasting demand, determining staff requirements and shift scheduling are executed once every year before the start of the new year. Rostering happens every month when the new monthly schedule needs to be made.

4.2.4 Tasks and triggers

Forecasting demand

(24)

Determining staff requirements

Once the demand for next year is forecasted, the staff requirements can be determined by cal-culating the expected needed capacity for next year leading to a capacity forecast for that year. This calculation is made based on the expected demand in terms of appointments and the time it takes per appointment type. Within this calculation a distinction between new and returning patients is made.

Shift Scheduling

The capacity forecast for next year is the input for making an initial year planning. The initial year planning of the department is based on the period of a week, meaning that every week of the year should basically look the same, except for the physicians that have taken time off.

Rostering

After creating the initial year planning, the month planning can be made by determining which physicians will work which shifts. The month planning is made three months in advance and has a rolling horizon, in order to leave enough time for patient planning. Thereafter the planning can be inserted in the patient planning system. This part of the process has a rolling horizon. Thus, for the first month planning the trigger is the creation of the year planning giving input for the month planning, for the other months of the year this task is triggered time-based every month.

The physical architecture shows the process of making a monthly physician planning as a sub-process, as is indicated as part B of figure 4. When making the monthly planning, the planner analyzes appointment data for the last month. He then decides based on the analysis whether the year planning needs to be adjusted or that the planning for last month is on track and waiting lists are not increasing. Alongside the analysis the time-off requests of physicians are handled. Eventually leading to the scheduling of the month planning which can be used as in put for patient planning.

(25)

physi-4.2.5 Information sources

The information sources used in this phase are the EPD and a related production database which consists out of all kinds of data regarding production, a time-off request document, and the year planning that is made during the process. In the EPD and the production database various data can be found such as appointment types, durations, characteristics of appointments, number of scheduled and executed production time, reasons of cancelations, time frames of cancelations, all with the possibility of dividing the numbers per sub specialism.

4.2.6 Decision support systems

As decision support systems the department uses spreadsheets, some pivot tables within the spreadsheets and the EPD. The planner uses the spreadsheets in order to calculate realized pro-duction, estimate demand and capacity and make the year and month planning. Within the spreadsheets a formula is placed leading to a suggestion of how many hours per sub specialism should be scheduled in order to meet demand. It should be noted that the demand calculation also is determined within the same spreadsheet. The spreadsheets do not give suggestions on how to make the planning in terms of which physician should do what. It can just be used as a tool to work the numbers and create an overview of production and capacity.

4.3! Department B

Department B, as described in figure 5 on the next page, is somewhat different from other de-partments in the sense that it has an outpatient clinic and a surgery section that work closely together. Physicians can be assigned to either hours in the outpatient clinic or surgery hours within the same day or week.

(26)

(27)

4.3.3 Time line

Shift scheduling is executed once every year before the start of the new year. Rostering happens over a rolling horizon every month, for three months in advance.

4.3.4 Tasks and triggers

The Shift Scheduling phase consists of making an initial standard year planning which is made once a year before the start of the next year and is divided over even and odd weeks. For every even week the schedule is the same and for every odd week it is also the same. However, the even and odd week might differ from each other since not every sub specialism has enough patients to fill the hours every week.

The year planning is influenced by the duties of each physician. The various duties are influ-enced by the sub specialism and the schedule is established based on experience. The planner stated that the department is mainly doing the same over the last couple of years. However, with the general shift of the hospital towards treating more complex patients and sending the less severe patient to non-academic hospitals, change is needed. The ratio between office- and sur-gery hours is shifting because more sursur-gery patients from other hospitals are sent to the univer-sity medical center. This leads to deviations from their forecast (which is not updated) in this phase. Hence, shift scheduling (and patient demand) act as a bottom up trigger to invest in forecasting.

Rostering

The planner makes a month planning every month and thereafter sends the planning to the CDC for feedback. If the CDC is satisfied with the planning the planning is ready for patient planning. If not, the month planning is altered by the planner until the CDC is satisfied. The physician planning gets manually inserted into the Electronic Patient Dossier (EPD), the program used for patient scheduling, by opening and closing calendars and adding room numbers into the calendar.

(28)

sub-then reviews the requests and checks if all the mandatory hours of the department are covered. If so, the planning is ready. Mandatory hours depend on the importance of the duties per sub specialism that need to be executed. This is not recorded on paper but something the planner knows by heart.

4.3.5 Information sources

The information sources used in the process are documents with planning requirements from management, staff changes, initial planning of other staffing groups, time of requests and the year planning made within the process.

4.3.6 Decision support systems

The department uses spreadsheets and the Electronic Patient Dossier as decision support sys-tems.

4.4! Department C

4.4.1 Resources and Patient flow

Department C, as described in figure 6 on the next page, consists out of an outpatient clinic, a nursing ward and a surgery team. The department recently got a new manager who wants to reorganize the planning process on a more detailed level. Currently it is not clear yet what kind of changes this will entail. However, the manager would like to shift towards a more data driven capacity planning.

(29)

(30)

4.4.3 Time line

Shift scheduling is executed once every year before the start of the new year. Rostering happens every month when the new monthly schedule needs to be made.

4.4.4 Tasks and Triggers

The Shift Scheduling phase consists of making an initial standard year planning which is di-vided over even and odd weeks and has been mostly the same for multiple years. When making this planning the planner uses inputs as physician preferences, surgery planning and fixed pro-portions of appointment types. Within the various weeks, sub specialisms are distributed over the days and all hours are linked to a physician. This planning is made once a year, before the start of the new year.

Rostering

After creating the initial year planning, the month planning can be made by taking the year planning as a starting point, removing all unavailable and inserted in the patient planning sys-tem. This part of the process has a rolling horizon making every month a planning for three months in advance. So, for the first month planning the trigger is the creation of the year plan-ning giving input for the month planplan-ning, for the other months of the year this task is triggered time-based every month.

Part B of figure 6 shows the process of making a monthly physician planning as a sub-process. When making the monthly planning, the planner reviews the time off requests and edits the year planning by removing all unavailable physicians. All physicians need to hand in their time-off requests at least three months in advance. The planner then reviews the requests and checks if all the mandatory hours of the department are covered. If so, the planning is ready. Mandatory hours are the hours that there is a surgery room made available for the department. It is im-portant for the department to always use the given hours in the surgery rooms, because vacancy of surgery hours is sanctioned by reducing the available surgery hours for the department.

(31)

surgery rooms at their scheduled hours. Due to required special equipment and availability of surgery rooms these hours cannot be swapped with each other to reduce the waiting list. There-fore the department should get time in surgery rooms that is equipped for day treatments, from other departments. Yet, at times the department has trouble filling up surgery rooms it is re-sistant to giving the hours to other departments themselves, afraid of reduction of their surgery time when they would need it at a later time.

The department uses various documents with physician preferences, surgery planning, fixed proportions of appointment types, time off requests and the year planning that is made within the process as information sources for their planning process.

4.4.6! Decision support systems

(32)

5.!Solution exploration: Analysis of the Physical Architectures

This chapter gives an analysis of the physical architectures. The approach taken (Section 5.1) suggest to first summarize elements of the PAs (Section 5.2). Next, an analysis of the PAs will be given (Sections 5.3-5.10), followed by a summary of findings (Section 5.11).

5.1 Approach

Based on the interviews with planners and managers of various departments three physical ar-chitectures were described in chapter 4. These arar-chitectures describe the physician capacity planning process of each department and give an indication on how the physician planning is created. The various planning process are analyzed by comparing the various elements of the physical architectures. Distinguishing commonalities and differences between the physical ar-chitectures of the departments provides insight in possible methods and functioning of elements which can contribute to an initial architecture. In order analyze the physical architectures, first a summary of the elements of the physical architectures will be given. Thereafter, a separate analysis on the planning organization and the timeline will be given, because this is not directly dependent on the various levels of the architectures. This will be followed by an analysis on how the four different levels of the conceptual model are captured in the PA’s.

5.2 Summary of elements

(33)

Level Element Department A Department B Department C

Forecasting demand

Who is involved? Planner X X

Timeline Once a year, every year X X

Triggers New year X X

Activities -Extract historical production data

-Cluster historical production data per sub specialism - Calculate realized production per sub specialism - Calculate expected needed capacity for next year

X X

Information Sources - Historic production data - Workforce X X Decision Support systems Spreadsheets X X Determining staff capacity

Who is involved? Planner X X

Timeline Once a year, every year X X

Triggers Document with numbers of expected demand per sub specialism available

X X

Activities Calculate expected need capacity for the next year X X

Information Sources - Demand analysis - Workforce X X Decision Support systems - Spreadsheets X X Shift Scheduling

Who is involved? Planner Planner Planner

Timeline Once a year, every year Once a year, every year Once a year, every year

(34)

Level Element Department A Department B Department C

Shift Scheduling

Activities Make initial year planning (per week) Make initial year planning (per even and odd week)

Copy planning of last year and edit preference in order to make initial year planning

Information Sources

- Capacity forecast - Work force

- Planning requirements management - Staff changes

- Initial planning other staffing groups

- Physician preferences - Surgery planning

- Fixed proportion of appointment types Decision Support

systems

Spreadsheets Spreadsheets Spreadsheets

Rostering

Who is involved? Planner Planner and CDC Planner

Timeline Once, every month Once, every month Once, every month

Triggers First of the year: finishing year planning Rest of the year: time-based every month

First of the year: finishing year planning Rest of the year: time-based every month

First of the year: finishing year planning Rest of the year: time-based every month Activities - Make monthly physician planning

- Insert physician planning in EPD

- Make monthly physician planning - Send planning to CDC

- Check planning

- Either suggest change and process change or accept planning

- Insert physician planning in EPD

- Make initial month planning - Insert Physician planning in EPD

Information Sources

- Initial year planning - Time off requests - Production data

- Initial year planning - Time off requests

- Overview mandatory hours

- Initial year planning - Time off requests

- Overview mandatory hours Decision Support

systems

- Spreadsheets

-Electronic Patient Dossier

- Spreadsheets

- Electronic Patient Dossier

- Spreadsheets

(35)

5.3 Planning organization

The first thing that can be seen from the PA’s is that responsibility for, and control over the planning lays mostly with the planner. At almost all departments the planner is mainly making their own decisions or following routine procedures. However, when too many physicians want to take time off at the same time, acceptance of time off requests is mostly informally discussed with the physicians before making a decision. There is only one department where the initial planning is supervised by the Chef de Clinique and only one department stated that there was some kind of supervision on utilization and employable hours at all.

5.4 Timeline

Every department makes a new year planning every year in the Shift Scheduling phase. There-after, in the rostering phase, this year planning is the basis for their monthly planning. This planning is made three months ahead, because the physician planning is the input for patient planning, which opens two months in advance. Therefore, time off requests of the physicians need to be handed in at least three months in advance. Only one department executes tasks in order to forecast demand and determine staff requirements, this happens once a year before making the year planning.

5.5 Forecasting demand

Most departments do not have a focus on forecasting demand. Instead, they focus mainly on physician availability and take a basic planning from a previous year, or look at physicians preferences, as input for their shift planning phase. This will be explained further under the heading Shift Scheduling. There was only one department that actually had a focus on forecast-ing demand. The department uses historic production data in order to forecast demand. Even though there was only one department actually working with a demand forecasting, all depart-ments indicated that they would be open to start forecasting demand and change towards a more demand driven planning, but that for now they do not know exactly how to change it.

5.6 Determining staff requirements

(36)

not available full time because of education obligations and are not always allowed to operate on their own. Residents mostly need supervision of a physician. However, the level of supervi-sion can variate between residents. This should be taken into account when determining the capacity of a department.

5.7 Shift Scheduling

Every department makes (or has) a basic year planning, but for every department the year plan-ning looks different. Some are more detailed than others. Some are based on data while others are more or less based on experience or personal feeling of the planner. While the plans that are based on historic data mostly change every year after analyzing the data of the previous year, the plans that are sort of grown over time are used years in a row. Another distinction that can be made is that for some departments the basic planning is the same every week, while for others there is a distinction made between even and odd weeks. It might happen that for exam-ple, some office hours for a certain specialism only need to be held once per two weeks in order to have enough production for this type of specialism. In this case, there is no reason to schedule the production hours every week. Thus, the distinction between even and odd weeks is made.

5.8 Rostering

The final rostering of the schedule happens every month for a period that lays three months in the future. For most departments the rostering phase consist of altering the basic planning by crossing out the physicians that are unavailable. Physicians need to hand in their time-off re-quests at least three months in advance. The planner then decides based on simultaneity require-ments whether the time-off requests are accepted or that the physicians should discuss who needs to withdraw his or her time-off request. Meaning that physicians that execute the same specialties cannot be unavailable at the same time. For now that is the only restriction on a time off request. However, in the future, when focusing more on actual capacity, the impact of una-vailability of the physician on the capacity should be considered.

5.9 Information sources

(37)

5.10 Decision support systems

Even though all departments use spreadsheets, the exact content of these spreadsheets differ per department and vary from a simple block schedule towards more in depth calculations. All departments use the same, hospital wide, electronic patient planning system. None of the departments has a planning system that can be linked to the EPD. Thus, the planning needs to be manually inserted into the EPD by opening and closing calendars and assigning rooms.

5.11 Summary of findings

(38)

6.!Design of the General Process Planning Architecture.

In this chapter the design of the GPPA is described. First, the key design steps are explained (Section 6.1), elaborating on various decisions in making the design. Thereafter, the design will be introduced (Section 6.2) and the various steps of the design, following the various levels of the conceptual model, will be discussed (Sections 6.3-6.7).

6.1 Key design steps

The analysis of chapter five is the first step in designing the GPPA. The analysis showed that despite the departments having own system and procedures when taking a close look, when taking a more distant look as shown in the architectures, a few general steps can be discovered. Such as the year planning and the month planning. These commonalities are the basis for the GPPA. The analysis also showed differences. For example the fact that there was only one department executing tasks that were related to forecasting demand and determining staff re-quirements. In case of differences the best practices were taken as input for the GPPA.

The next step in designing the GPPA is validation by and requesting input of domain experts. Multiple domain experts were asked to provide input on how the process could be improved. Mainly all domain experts acknowledged that in practice most of the departments make their planning based on the availability of the physicians and are therefore focusing on supply instead of demand. The domain experts indicated their wish to turn this around towards a more demand driven capacity planning in order to prevent capacity problems and bottlenecks, increase pro-duction and reduce cancelation and rescheduling of patient appointments.

(39)

Further, this study recites that creating an adequate physician capacity planning starts with ex-tensive use and understanding of data because of the garbage in, garbage out principle; the quality of the output is determined by the quality of the input. Thus, if the data used in deter-mining demand and supply is incomplete or incorrect, the capacity planning will most likely be inefficient as well.

6.2 General Process Planning Architecture.

(40)

(41)

Level Element GPPA

Who is involved? Planner and preferably Healthcare logistics experts

Timeline Once a year, every year

Triggers New year

Activities - Collect data

- Analyze available data Information Sources - Historic (patient)data

- Management requirements - Insurance requirements

- Trends in diseases and treatments - Demographics

- Other available data Decision Support systems - Spreadsheets

- Analytic tools

Determining staff capacity

Who is involved? Planner

Triggers Insight in (past) capacity, needs and available resources available Activities Calculate expected required capacity

Information Sources - Demand analysis - Workforce

- Obligations besides patient care Decision Support systems - Calculation tools

- Digital decision support systems Shift

Scheduling

Triggers Requirements for initial year planning ready

Activities Make initial year planning

Information Sources Planning of other staffing groups Decision Support systems Schedule application

Rostering

Timeline Once, every month

Triggers First of the year: finishing year planning. Rest of the year: time-based every month (rolling horizon)

Activities - Make initial month planning

- Set physician planning open for patient planning Information Sources - Initial year planning

- Time off requests

- Overview mandatory hours Decision Support systems - Schedule application

- Patient planning system Table 2: Summary of the Generic Process Planning Architecture.

6.3 Planning organization

Following the analysis, the GPPA, as displayed in figure 7, lays responsibility for, and control over the planning mainly with dedicated staff, i.e., the planner(s). If the planner has the right information and tools available, he or she should be able to make the planning by him/herself. However, it is advised that in case of uncertainties, the planner seeks guidance by other plan-ners, management or healthcare logistic experts.

(42)

appointments with patients within a span of two weeks and therefore, in practice it is seen that patient planning opens two months in advance.

6.4 Forecasting demand

6.4.1 Collecting data

In order to have an adequate capacity planning it is important to gain insight in what the demand will be for the next year. Demand includes both known and unknown claims and has a variation in volume, resource-mix requirements, urgency and patient throughput (Larsson and Fredriks-son, 2018). In order to predict demand, various data sources can be used. Historic (patient)data shows what services and treatments the hospital has provided in the past. Management require-ments influence the direction that a department is going, which might influence the physician planning. Insurance requirements are of interest, because in the end the insurance companies determine the hospitals budget and therefore influence the production of the hospital. Known trends in diseases and treatments and demographics can be useful since the various trends and the characteristics of a population can give insight in what kind of illnesses are expected to need treatment in the coming year. There might be other department specific data sources that could be relevant and should therefore be used. Since they might vary per department these are qual-ified as ‘Other available data’.

6.4.2 Analyze available data

In order to make an estimation of how much capacity is needed, the collected data should be analyzed in order to show trends in procedures and treatments, give information on no show of patients or the number of empty appointment slots. No show of patients and empty appointment slots are lost capacity. Lost capacity can be seen as capacity that was reserved for patient treat-ment, but that was not used because of unavailability of patients. An adequate analysis of the available data is important as the output of the analysis is the input of the capacity estimation that follows after.

6.5 Determining staff requirements

6.5.1 Estimate expected required capacity

(43)

make sure that he or she has information on the workforce, in the sense of which physicians are available, for how many hours a week, what other kind of obligations the physician has and to which degree the physician can work independently or needs supervision. The planner also needs to know which spaces and equipment are available at what times, because this directly influences how much production the hospital can run. It is important to calculate whether the estimated available capacity is sufficient to fulfil the expected demand. If not, reorganizing physicians tasks and obligations or hiring new staff might be needed.

Not all physicians work full time and almost none of the physicians are full time available for patient care. Some physicians need to do research, some physicians need to provide supervision and some physicians need to do both. Also, residents can sometimes work independently, but most of the time they need some kind of supervision. Besides that, residents need training time. Thus, it is important to know what the time distribution is for all physicians and residents in order to make an adequate capacity estimation.

6.6 Shift scheduling

6.6.1 Make initial year planning

After determining how much capacity is needed based on the calculations and the other relevant information extracted from the available data, an initial year planning can be made. The year planning is mainly a standardized format that shows for example on what day, which office hours will be held. The office hours can be directly linked to a physician, but this can also be determined later in order to leave more flexibility and the possibility to indent more easily on changes in demand. The year planning can be made either with a view per month, per week, per even and odd weeks or any other format, depending on whatever fits best in the given cir-cumstances. In practice it is seen that most departments have the same weekly schedule with sometimes a distinction between even and odd weeks.

(44)

week planning where a certain specialism needs 208 hours for the coming year. Then, consid-ering the whole year (52 weeks) there should be 4 hours per week scheduled for this specialism. However, the physician will most likely go on holiday or have congresses that he or she needs to attend to. Thus, the physician will not be available for all of the 52 weeks. Considering that the physician would for example be available for 40 weeks next year, there should be 5,2 hours (208 divided by 40) per week for this specialism instead of 4. By scheduling the hours over the weeks a physician will be available, all hours are covered, even though it is not determined at which exact weeks these hours will be worked.

For residents this calculation is even a bit more complex than described above. When for ex-ample there are two residents at a department, the first thought would most likely be that these two residents cannot take time off at the same time, in order to equally spread the available capacity. However, if the residents are supervised by the same physician, it would be more efficient to give them time off at the same time. Then, during times the residents are available, they both work and while they are unavailable, the physician can see patients himself. If only one resident is unavailable at a time, the other resident still needs supervision all the other times and the physician does not get any extra time to see patients himself. In practice it is seen that by making the year planning there is a lot of discussion on how many weeks a physician can and should be available. Clear agreements with physicians about their availability are needed in order to prevent problems during the year.

6.7 Rostering

6.7.1 Make month planning

(45)

Currently it is seen in practice that most of these schedules are made in spreadsheets where all decisions are made by the planner. However, it would probably be more effective to invest in more advanced tooling, such as developed planning programs or algorithms that can help the planner by giving suggestions on how to make the planning. The maturity of the needed deci-sion support systems might be influenced by the capabilities of the planner and can vary per department.

6.7.2 Set physician planning open for patient planning

After making a monthly physician planning the department knows which physicians will be available for what hours, for that month. Since a physician needs patients to actually do his job, the physician planning should be linked to the patient planning. Once the physician planning is linked to the patient planning, the medical administration office can enter the patient appoint-ments into the schedule.

(46)

7.!Use of the General Process Planning Architecture

This chapter will give an illustration and evaluation of the GPPA by using it as an aid in setting up the physical architecture of department D. The approach taken (Section 7.1) suggests to first describe the planning process of department D by making a physical architecture (Section 7.2). Next, an overview of the commonalities and differences between the GPPA and department D will be given (Section 7.3), followed by an evaluation of the working of the GPPA for depart-ment D (Section 7.4).

7.1 Approach

The primary objective of the GPPA is to structure the planning process of a department. The GPPA can be used either for setting up a new planning system for a department or as a reference architecture for existing departments.

When setting up a new planning system the GPPA can give an indication of the functions that should be generated, which can be used as building blocks for developing a planning system. The GPPA provides an organized structure which can help determining necessary tasks. It could also help with appointing and directing project teams. If one would want to set up a planning system, one could for example set up a project team for each level of GPPA. These project teams can then focus on setting up their part of the planning process without having to figure out where to start and determine what to focus on.

When using the GPPA as a reference architecture for existing departments, it helps mapping the process of a department in a fast and efficient way by indicating the various levels of a planning system and describing possible tasks. After mapping the process of the department, the GPPA can be used as a reference to verify the maturity of the planning system. Comparing the two architectures can help discovering underdeveloped parts of the process or determining which tasks need to be added or improved.

(47)

Once the physical architecture of department D has been described, the next step is comparing the physical architecture with the GPPA. Comparing the two architectures and identifying dif-ferences leads to a better understanding of the current planning process and gives the oppor-tunity to check the completeness of either the process of the department even as the complete-ness of the GPPA.

The department is currently working on an improvement of the planning process themselves, this will also be considered while comparing the department with the GPPA.

7.2 Department D

Department D, as described in figure 8 on the next page, consists out of three sub departments: an outpatient clinic, a nursing ward and an endoscopy center. The department recently got a new healthcare logistics planner who is actively working on changing the process in order to improve their capacity planning. The schedule of the sub departments is now really focused on the availability of the physician. The planner wants to turn this around towards a more demand driven capacity planning. In order to prevent capacity problems and bottlenecks, increase pro-duction and reduce cancelation and rescheduling of patient appointments.

Forecasting demand and determining staff requirements are not really happening in the process. For this part of the process, the department relies on historical information and is mainly fo-cused on physician preferences. On the levels that do exist only the planner of the department is involved.

7.2.3 Time line

(48)

(49)

7.2.4 Tasks and Triggers

Shift scheduling

The shift scheduling phase consists of making an initial standard year planning per week, per physician. When making this planning the planner uses inputs as physician preferences, plan-ning of other staffing groups and fixed proportions of appointment types.

Rostering

After creating the initial year planning, the month planning can be made by taking the year planning as a starting point, removing all unavailable physicians and inserted in the patient planning system. This part of the process is repeated every month. Thus, for the first month planning the trigger is the creation of the year planning giving input for the month planning, for the other months of the year this task is triggered time based every month.

Part B of figure 8 shows the process of making a monthly physician planning as a sub-process. When making the monthly planning, the planner reviews the time off requests and edits the year planning by removing all unavailable physicians. The time off requests are accepted or rejected based on simultaneity requirements. Meaning that there should always be enough phy-sicians available for every specialty. So, not all phyphy-sicians that execute the same specialties can be unavailable at the same time. The planner then reviews the planning and checks if all the mandatory hours of the department are covered. If so, the planning is ready. If not, the planning is rescheduled in informal consultation with the physicians.

The planner stated that nowadays physicians have a really high workload due to the increasing number of tasks that physicians take on. A lot of physicians learn new skills and procedures or join new research while they already have a busy schedule. Most of the time it is not possible to reduce the already scheduled time in order to make time for the new things. Also the planner needs to take into account that some physicians work part time or need to take days of for research. Thus, not all physicians are available full time.

(50)

7.7.6 Decision Support Systems

11 months ago the department shifted from the use of spreadsheets to a cloud schedule appli-cation specifically for medical specialists.

7.7.7 Intended situation for department D

For now the planning process is as described in the physical architecture and is almost similar as the process of department C. However, the healthcare logistics planner of department D is, in collaboration with two healthcare logistics experts, actively working on changing the plan-ning process in order to improve capacity planplan-ning. The idea is that based on various data sources the demand and supply is forecasted and based on that a standard planning is made. The standard planning should give an overview of the amount hours per appointment type that should be made available for patients given a certain number of working weeks per physician. Thereafter the monthly planning is made based on availability of the physicians. The acceptance of time of requests for the physicians might be limited if unavailability of the physician will lead to a loss of capacity at one sub-department and lead to overcapacity at another sub-depart-ment. This might happen if, for example at the end of the year, a physician has not made enough hours at the outpatient clinic and wants to take a day off at which he should be available for the outpatient clinic.

7.3 Comparing the physical architecture of department D with GPPA

(51)

Level Element GPPA Department D (current situa-tion)

Department D (intended situation)

X Planner and Healthcare logistics expert

Timeline Once a year, every year X Once a year, every year

Triggers New year X New year

Activities - Collect data

- Analyze available data

X - Collect data

- Analyze available data Information

Sources

- Historic (patient)data - Management requirements - Insurance requirements

- Trends in diseases and treatments - Demographics

- Other available data

X - Historic (patient) data

-Expected growth/shrinkage per patient group - Workforce Decision Support systems - Analytic tools - Spreadsheets X Spreadsheets

Determining staff ca-pacity

X Planner and Healthcare logistics expert

Timeline Once a year, every year X Once a year, every year

Triggers Insight in (past) capacity, needs and available resources available

X Insight in (past) capacity, needs and available resources available

Activities Calculate expected required capacity X Calculate expected required capacity

Information Sources

- Demand analysis - Workforce

- Obligations besides patient care

X Undefined

Hospital Staff Capacity Planning