Timely and efficient planning of treatments through intelligent scheduling

When people experience health problems, this often leads to a period of concern and stress. Both for patients’ physical and mental well-being, it is essential that healthcare processes are organized in the best possible way. This thesis aims to support hospitals in realizing excellent quality of patient service, whilst utilizing resources efficiently. By developing methodologies based on techniques from operations research, also known as mathematical decision theory, the research presented provides decision support for the planning and scheduling of healthcare processes. Three specific challenges are addressed in the three parts this thesis consists of. Part I focuses on online appointment scheduling, providing patients with prompt responses to their appointment requests. Part II presents an alternative to appointment scheduling: enabling patients to walk in for diagnostic examinations without an appointment. Part III focuses on organizing inpatient care services in such a way that quality of service, quality of care, and efficiency are guaranteed. Online appointment scheduling When a patient submits an appointment request, a hospital can either reject the patient or choose between three response delays: (i) responding immediately by scheduling an appointment, (ii) postponing the response temporarily, meanwhile collecting appointment requests from other patients, and performing the scheduling activity periodically (for example, at the end of each day), or (iii) collecting all appointment requests for a particular resource in a particular period before scheduling the appointments. In this thesis, we refer to these three different scheduling horizons as online, near-online, and offline appointment scheduling, respectively. While offline appointment scheduling has received considerable attention in the literature and has been the focus of several literature reviews, the literature on online and near-online appointment scheduling, receiving increased attention from researchers over the past decades, has not been reviewed yet. In Chapter 2, we review this literature according to a taxonomy consisting of: the number of appointments each customer requires, the number of resource types at the service facility, and the horizon at which the scheduling decisions are made. We provide an overview of the scheduling decisions, the objectives, and the operations research methods applied in different application areas. We identify similarities and differences between application areas and categories of our taxonomy, and highlight gaps in the literature that represent opportunities for future research. By reviewing the literature across various application areas, the chapter aims to stimulate mutual interchange of research results in the field of online appointment scheduling. Timely and efficient planning of treatments through intelligent scheduling. Outpatient rehabilitation treatment planning is one of the areas in healthcare where online appointment scheduling is frequently applied. Rehabilitation outpatients, especially those treated in an academic medical center, often require repeated treatment by therapists from various disciplines. Thus, a patient’s request consists of multiple appointments for which various resource types are required. A short access time (the time between the appointment request and the first treatment), coordination of the appointments over the disciplines involved in a patient’s treatment, and continuity of the treatment process are essential for realizing high-quality rehabilitation care. Moreover, combining multiple appointments on one day is an important service metric from an outpatient’s perspective, while utilizing the available therapist capacity efficiently is crucial from a rehabilitation clinic’s point of view. In Chapter 3, we develop an integral treatment planning methodology, based on an integer linear programming (ILP) formulation, that takes all these performance indicators into account. We evaluate its performance by means of computer simulation, for a case study at the rehabilitation outpatient clinic of the Academic Medical Center (AMC) in Amsterdam. From the results we conclude that by implementing the developed methodology, more patients can be treated with the same therapist capacity, while patients benefit from both a higher quality of care and a higher quality of service. Walk-in as an alternative Enabling patients to walk in for certain types of diagnostics or treatment has considerable potential. When the patient walks in right after the consultation in which the necessity of an examination or treatment has been revealed, access time is eliminated completely and the patient saves one hospital visit. Also, the time until treatment, during which deterioration of the patient’s health condition may be a serious risk, is shortened. Moreover, the patient is enabled to autonomously choose his preferred date and time for service. In terms of system efficiency, compared to an appointment system, walk-in eliminates both scheduling efforts and no-shows (a patient not showing up for his appointment without prior notification to the hospital). Despite its potential, a disadvantage of a walk-in system is the high possible variability in patient arrivals, which results in a highly variable system utilization and high patient waiting times during busy periods. In hospital outpatient clinics that serve patients on a walk-in basis, there are various organizational and medical reasons to give a patient an appointment. Therefore, most of these clinics also offer a limited number of appointment slots, and thus employ a combined walk-in and appointment system. Combining walk-in and appointments facilitates having the best of both worlds, because appointments can be scheduled in periods with low walk-in demand such that the arrivals are evenly distributed over each day. The objective in a combined system is twofold: providing appointment patients with timely access, and providing walk-in patients with acceptable waiting times. In Chapter 4, we develop a methodology for generating a cyclic appointment schedule for a facility employing a combined walk-in and appointment system. Our methodology prescribes how many slots to reserve for appointments on each day of the cycle (e.g., week), and at which moments during the day to schedule these slots. To this end, our approach consists of decomposing the appointment scheduling process and the service process during the day. We use two analytical evaluation models, a discrete-time cyclic queueing model and a Markov reward model, which we link by an iterative procedure. In a case study for the CT-scan facility at the radiology department of the AMC, we find that our methodology generates schedules that effectively balance the workload over time. Both appointment patients and walk-in patients experience high-quality service, also when the number of patients increases. While enabling patients to walk-in for their diagnostic examination has considerable potential, the impact of implementing a combined walk-in and appointment system has not been quantitatively investigated before. In Chapter 5, we develop a reusable (i.e., also applicable to other systems of similar type) discrete event simulation model that can be used to investigate the consequences of implementing such a system for any type of diagnostic examination, and apply it to evaluate the impact of enabling patients to walk in for CT-scans. Our model is component-based and can therefore be applied to various types of diagnostic facilities with various configurations of process steps. For simulating a prospective combined walk-in and appointment system, an appointment schedule is required, which we develop by employing the methodology from Chapter 4. We apply our approach to the same case study as that of Chapter 4. Based on our results, the CT-facility at the AMC will shift from a complete appointment system to a combined walk-in and appointment system on October 25, 2015. Timely and efficient inpatient care Outpatient consultations and diagnostics may lead to the conclusion that a patient needs to be hospitalized, as may emergencies. For non-emergency patients, also called elective patients, admission to clinical wards is often driven by their scheduled treatments. For surgical nursing wards, the number of patients present at any given point in time, so-called bed census, is determined by the outflow from the operating rooms, arrivals of emergency patients, and patients’ length of stay. Moreover, a nursing ward may receive patients from an other ward that is fully occupied (so-called overflow). The bed census, which thus results from a complex interaction of several hospital departments, determines the number of nurses required during a given shift. Because nurse staffing and rostering is typically performed several weeks in advance, ward managers would benefit from reliable bed census predictions. With their fluctuating patient populations, they would further benefit from the ability to only completely fix the number of nurses working on a given ward just before the start of the actual shift, to be able to dynamically respond to the bed census observed. Once this number is fixed, the available nurses have to be assigned to the patients present, a process which affects quality and safety of care, and both nurse and patient satisfaction. As a first step in supporting ward managers in this process, Chapter 6 presents a stochastic analytical model that predicts the bed census on nursing wards by hour, as a function of the operating room schedule and arrival patterns of emergency patients. Patient overflow between nursing wards is also incorporated. Based on overflow and rejection (a patient not being admitted due to bed unavailability) thresholds set by hospital management, our model can be used to determine the number of beds required on each ward. For a case study of four surgical wards in the AMC, we demonstrate the decision support abilities of the model. It can be used to evaluate changes in care unit partitioning (which care units are created and which patient groups are assigned to these units), care unit size, the operating room schedule, admission and discharge policies, and patient overflow policies. For our case study we show that such changes potentially yield a 10-25% increase in bed productivity (the number of patients treated per bed per year). In a second step, we use the hourly bed census predictions from Chapter 6 as input for a methodology that supports nurse staffing decisions, presented in Chapter 7. Our model, which is based on stochastic programming techniques, determines the number of nurses to staff on each ward for each shift. To indeed enable head nurses to dynamically respond to the bed census observed just before the start of a shift, we investigate the concept of flexible nurse staffing. In this concept, multiple wards share a flex pool consisting of cross-trained nurses. These nurses are told several weeks in advance that they have to work during a given shift, but only at the start of the shift is it decided at which ward they will work. For the same case study as in Chapter 6, we show that changing the currently applied staffing policy to staffing based on bed census predictions increases nurse productivity (the number of patients treated per employed FTE per year) by 20-30%, while guaranteeing more consistent quality and safety of care. Based on the outcomes of both studies, the bed census prediction model and the subsequent flexible staffing method are embraced by the AMC as valuable instruments to support the resource capacity planning of its inpatient care services. The nurse-to-patient assignment process, that recurs daily at the start of each working shift on nursing wards, is time-consuming and complex due to the many considerations involved. Creating well-balanced, high-quality assignments is crucial to ensuring patient safety, quality of care, and job satisfaction for nurses. In Chapter 8, we develop a computerized decision support system (CDSS) for nurse-to-patient assignment based on an ILP. We design the CDSS in close cooperation with two nursing wards in the AMC and subsequently test and evaluate it in a before-and-after study, for which we include an additional nursing ward. The CDSS promotes several aspects relating to quality and safety of care such as spreading workload equally over nurses, striving for continuity of care (a patient being cared for by the same nurse on consecutive days, whenever possible), and minimizing nurses’ walking distances, such that nurses have more time for patient care. The measurements after implementation reveal a 30% decrease of the time required for the nurse-to-patient assignment process, while nurses experience a lower workload. We conclude that the developed CDSS increases both the quality and safety of care as well as the nurses’ job satisfaction. Concluding, all studies in this thesis exemplify how healthcare operations research enables simultaneous improvements in both quality of care and logistical efficiency. Thereby, it enables and supports the future of healthcare processes. A future, in which the planning and scheduling of healthcare processes is such that patients receive their consultations, examinations, and treatment at the medically desired instants, to facilitate high-quality care, resulting in the best possible health outcomes. Also, by being offered excellent quality of service, patients do not experience organizational burdens on top of their illness and receive healthcare appointments aligning with their activities in everyday life. Healthcare professionals are enabled to focus on what they have been trained for: patient care. And healthcare organizations are financially healthy, because the limited resources they have available are utilized efficiently. Although the healthcare sector will always experience certain inefficiencies due to the uncertainty inherent in care processes, operations research methodologies support healthcare organizations in keeping these to a minimum. The implementation of operations research methodologies – such as the ones presented in this thesis – in decision support systems will enable healthcare organizations to organize their processes in a way that supports excellent quality of both care and service.