Nursing

J Nurs Manag. 2019;27:971–980. wileyonlinelibrary.com/journal/jonm  |  971© 2019 John Wiley & Sons Ltd

1   |   I N T R O D U C T I O N

Managing healthcare systems to achieve better quality of care, while reducing cost is a global concern (Bragadóttir, Kalisch, Smáradóttir, & Jónsdóttir, 2015). Contributing to fiscal pres‐ sures are an increased demand for healthcare services due to a

significant increase in an ageing population, the need for newer and costly healthcare technologies, and the increasing complexity of hospital care and treatment processes (Letiche, 2008). Amidst these challenges, hospital managers are faced with the competing priorities of improving cost efficiency in their operations while improving care quality, patient safety and maintaining a safe work

Received: 29 June 2018  | Revised: 14 January 2019  | Accepted: 5 February 2019 DOI: 10.1111/jonm.12757

O R I G I N A L A R T I C L E

Predicting the effect of nurse–patient ratio on nurse workload and care quality using discrete event simulation

Sadeem Munawar Qureshi MEng1  | Nancy Purdy RN, PhD2  | Asad Mohani BEng1 | W. Patrick Neumann PhD, LEL,EurErg1

1Human Factors Engineering Lab, Department of Mechanical and Industrial Engineering, Ryerson University, Toronto, Ontario, Canada 2Daphne Cockwell School of Nursing, Ryerson University, Toronto, Ontario, Canada

Correspondence Sadeem Munawar Qureshi, Human Factors Engineering Lab, Department of Mechanical and Industrial Engineering, Ryerson University, Toronto, ON, Canada. Email: s1qureshi@ryerson.ca

Funding information The Natural Sciences and Engineering Research Council of Canada (NSERC) for their generous Discovery grant(s)—Grant # 341664 and Grant # 2018‐05956.

Abstract Aim: A novel nurse‐focused discrete event simulation modelling approach was tested to predict nurse workload and care quality. Background: It can be challenging for hospital managers to quantify the impact of changing operational policy and technical design such as nurse–patient ratios on nurse workload and care quality. Planning tools are needed—discrete event simula‐ tion is a potential solution. Method: Using discrete event simulation, a demonstrator “Simulated Care Delivery Unit” model was created to predict the effects of varying nurse–patient ratios. Modelling inputs included the following: patient care data (GRASP systems data), in‐ patient unit floor plan and operating logic. Model outputs included the following: nurse workload in terms of task‐in‐queue, cumulative distance walked and Care quality in terms of task in queue time, missed care. Results: The model demonstrated that as NPR increases, care quality deteriorated (120% missed care; 20% task‐in‐queue time) and nursing workload increased (120% task‐in‐queue; 110% cumulative walking distance). Conclusions: DES has the potential to be used to inform operational policy and tech‐ nical design decisions, in terms of impacts on nurse workload and care quality. Implications for Nursing Management: This research offers the ability to quantify the impacts of proposed policy changes and technical design decisions, and provide a more cost‐effective and safe alternative to the current trial and error methodologies.

K E Y W O R D S

discrete event simulation, human factors, nurse management, nurse–patient ratio, quality of care

972  |     QURESHI Et al.

environment for the healthcare professionals delivering care. Since direct labour costs are the highest cost budget item, hos‐ pital managers are challenged to limit or reduce the direct labour costs despite potential adverse effects. Adverse effects arising from understaffing can lead to overtime and excessive workload giving rise to stress, fatigue, work‐related musculoskeletal dis‐ orders (WMSD), absenteeism and eventually burnout or injury (Silas, 2015). Therefore, hospital processes need to improve in ways that do not negatively impact the work environment or worker safety.

In 2016, nurses in Canada worked over 20.1 million hours of overtime, which is equivalent to 11,100 full‐time positions with an estimated cost of $968 million dollars (Canadian Federation of Nurses Unions, 2017b). In 2011, the rate of injuries and illness re‐ sulting in days away from work in the United States was highest for hospital workers among all industries, with an incident rate of 157.8/10,000 workers (Bureau of Labor Statistics, ). In the same year, the United Kingdom's National Health Service reported that 76% of the registered nurses worked extra hours; in 2012, this in‐ creased to 80% (International Council of Nurses, 2015). Nurses are absent from work at almost twice the rate of all other professions (Canadian Federation of Nurses Unions, 2017a). Nursing is a highly stressful job, where the incidence of burnout is high (Rizo‐Baeza et al., 2018). Furthermore, the healthcare sector was reported to have the highest number of lost time injuries including WMSD, workplace violence, exposures and falls; making nursing the highest risk job in 2014, over manufacturing and mining industries (CNFU, 2015). It is imperative that proposed changes to operational policies do not worsen working conditions for nurses. Hence, tools are needed to help understand the possible negative consequences of proposed changes.

1.1 | Industrial engineering and human factors in healthcare

To improve healthcare processes, several healthcare organizations have implemented industrial engineering (IE) techniques such as

Lean interventions. But, long‐term negative effects have been re‐ ported such as an increased potential for making mistakes, injuries and missing less urgent care tasks which led to a drop‐in the quality of care (Moraros, Lemstra, & Nwankwo, 2016). Process improve‐ ment strategies to increase efficiency are sometimes accompanied by negative effects such as the degradation of the healthcare pro‐ fessional's (HCP) health, satisfaction and engagement, workload is‐ sues, availability of supplies, increased stress and reduced safety for patients (Carayon et al., 2014). Engineering tools applied in health‐ care processes have shortcomings when they do not consider the impact on the HCP. However, human factors engineering, and ergo‐ nomic principles and methodologies may help as most techniques are user‐centred (Carayon, 2016).

Human factors (HF) are the scientific discipline concerned with the understanding of interactions among humans and other ele‐ ments of a system to optimize system performance and human well‐ being (International Ergonomics Association, 2018). The conceptual model used in this study builds on the Systems Engineering Initiative for Patient Safety (SEIPS 2.0) model by Holden et al. (2013). SEIPS 2.0 is a framework for understanding the role of HF in healthcare systems. In this study, a more design‐oriented approach to the SEIPS 2.0 model was used with a broader goal to support efforts in im‐ proving performance in existing healthcare units. As illustrated in Figure 1, this design‐oriented approach addresses the needs of both the HCPs and patients in the improvement process. HCPs are central and factors affecting them affect the overall performance of these complex healthcare systems.

In the design and management of healthcare systems, ignoring HCP's workload means a lack of focus on the HCP's performance with a possible impact on efficiency, productivity, injury, burnout and increased costs (Alghamdi, 2016; Kalisch & Williams, 2009). Current approaches to testing design and management decisions such as the real‐life trial and error methods can be very expensive and hazardous (Gaba, 2007), as workers would be exposed to un‐ safe and untested environments that can not only affect their health but also negatively impact productivity and process efficiency with possible long‐lasting consequences. Hence, there is a need for a tool

F I G U R E 1   Illustrates how a design‐orientated approach can address the needs of the healthcare professionals (HCP) and patients, by focusing on the health and workload of HCP, and the quality of care that gets delivered to patients

     |  973QURESHI Et al.

that can “virtually” assess and predict the effects of design changes such as staffing on HCP and patients without the risk of real trials or at the expense of HCP's health. Simulation is a potential solution to this challenge.

1.2 | Discrete event simulation

Simulation imitates real‐world scenarios over time (Banks, Carson, Nelson, & Nicol, ). Discrete Event Simulation (DES) is an operational research technique used to assess, predict and op‐ timize the efficiency of a proposed or existing system (Clancy & Delaney, 2005; Jun, Jacobson, & Swisher, 1999). It is useful to study complex systems with emergent characteristics and com‐ plexity issues. DES uses mathematical formulas as a means of representing complex structures of a system/unit as a sequence of ordered events and stages, in which the variable(s) change at a discrete set of points (Banks, Carson, Nelson, & Nicol, ; Clancy & Delaney, 2005; Dode, Greig, Zolfaghari, & Neumann, 2016). When applied to health care, the interactions of interest could be between acute care patients and HCPs to examine patient or nurse outcomes.

DES has been widely used to model patient flow in hospi‐ tal units such as the discharge process in a paediatric hospital (Lambton, Roeder, Saltzman, Param, & Fernandes, 2017). DES has also been used to examine two family physician practice clinics by focusing on staffing and size of the facility (Swisher & Jacobson, 2002). DES was also used to model the operations of an emergency department to understand the system behaviour and examine the causes of excessive waiting times (Komashie & Mousavi, 2005), and injury levels of healthcare workers (Duguay & Chetouane, 2007). DES models have been used to predict the effect of different layouts of hospital units in term of nurses’ movements (Boucherie, Hans, & Hartmann, 2012; Choudhary, Bafna, Heo, Hendrich, & Chow, 2010). DES has also been used to monitor wait times in a Québec‐based haematology–oncology clinic using a patient‐flow simulation experiment (Baril, Gascon, Miller, & Bounhol, 2016). These DES studies have, however, gen‐ erally been limited to a focus on physicians and/or patients and not on nurses’ care delivery (Barnes, Golden, & Price, 2013), de‐ spite the fact that nurses are the largest group of healthcare pro‐ viders and nurses deliver over 75% of the hospital care (Nursing Task Force, 1999). While there has been a significant number of publications of DES research in healthcare settings, there has been a scarcity of research specifically focused on the healthcare professional and none that have modelled nursing workload, care quality and work environments.

1.3 | Aim

The aim of this study was to create and test a novel nurse‐focused DES modelling approach that could proactively assess care quality and the workload for nurses, by modelling the delivery of care for patients under different technical design and operational policies.

Specifically, the demonstrator model quantified the effect of chang‐ ing nurse–patient ratios on care quality and nurse workload.

2   |   M E T H O D S

“The Simulated Care Delivery Unit” (SCDU) is a computerized simu‐ lation model created using a commercial DES environment software (Rockwell ARENA). The SCDU is the representation of the HCP's work processes. The demonstration model was created in consulta‐ tion with a subject specialist—a registered nurse with extensive re‐ search and practical experience.

As illustrated in Figure 2, the inputs of the model consist of pa‐ tient care data, operating logic and virtual layout. The outputs con‐ sist of task in queue time and missed care, used here as care quality indicators and, task queue and cumulative distance walked as nurse workload indicators. These are further expanded in the next section.

2.1 | Patient care data

As illustrated in Figure 2, patient care data entail essential details of the daily patient care tasks that a nurse performs. These data were obtained from an inpatient unit of a large urban academic health centre in Canada for a period of one month. The data were part of a workload report generated from the hospital's GRASP software sys‐ tem (Grace Reynolds Application of the Study of PETO) (Farrington, Trundle, Redpath, & Anderson, 2000; Song et al., 2004). GRASP is a proprietary management information‐processing system used to collect data for analysis of nursing workload. Data contain informa‐ tion pertaining to the patient care tasks that were performed by nurses. The definitions of each task are specific to GRASP meth‐ odology; for example, assessment refers to the completion of the Braden Scale, Morse Fall Assessment, etc. and does not refer to the ongoing assessment that nurses conduct when delivering care. Data are manually entered by nurses at the end of their shift. For each sub‐task (such as: IV Maintenance), there is a pre‐set standardized time duration. Approximately 70% of the hospitals in Ontario use the GRASP system (Song et al., 2004). Patient care data are com‐ prised of task information, task frequency and task duration. (a) Task information includes basic task information such as task group, for in‐ stance nutrition; sub‐tasks within this category include feeding with minimal assistance; shift and date stamp. (b) Task frequency entails how frequently a certain task is completed along with the day and time stamps. Task frequency was calculated using an average of the task count for each task group across all patients per day for a period of one month. (c) Task duration is the amount of time required by the nurse to complete the task. Task duration for each of the task groups was calculated using a frequency‐weighted average of GRASP's standardized time duration for all sub‐tasks of in a Task group. Since the GRASP system uses a standardized time duration for each sub‐ task, a frequency‐weighted average was used in this study to reduce the volume of sub‐task programming in the model. Table 1 contains the cumulative time durations of the tasks for the SCDU model.

974  |     QURESHI Et al.

2.2 | Operating logic

As illustrated in Figure 3, the model's operating logic of the SCDU model consists of task priorities, nurse priorities, task schedules, task location and call tasks. Task priorities indicate which tasks have an increased priority for completion relative to other tasks in queue. As mentioned in Section 2.1, nursing care delivery task names, and associated task definitions, were taken from GRASP systems. For GRASP reporting purposes—the nurses are only required to enter the tasks delivered during that shift (i.e. task frequency), not their sequence or timing. The priority levels for these nursing care deliv‐ ery tasks were developed in consultation with a subject matter ex‐ pert (registered nurse with 25 years of experience) who considered the importance and urgency of each task based on their professional experience (Hendry & Walker, 2004). As illustrated in Table 1, the “Admission” task bears a priority level of 6 because it was believed that the nurse must attend to the immediate care needs of pa‐ tients already present in the unit (such as: delivering “Medication,” “Treatments”), before starting the admission process (recognizing that the acuity of the admission may vary between settings and model scenarios). It is anticipated that nurses may assign a different priority than those listed in Table 1 and this can be easily changed in the model to test alternatives. Nurse priorities can also be referred to as the “brain of the simulated nurse”—the logic rule identifies which task a nurse performs in retrospect of the task priorities. In this demonstrator model, the simulated nurse is programmed to do the highest priority task first. There may be occasions where more than one task bears the same priority. In this case, the task logic was built to direct the simulated nurse to perform the task for the closest

patient (at the least distance). Figure 3 represents the operating logic of this model.

As illustrated in Table 1, Task schedule refers to tasks that fol‐ low either an established schedule or those that occur randomly throughout the shift, or both. For example, hygiene is scheduled for once a day. However, the hygiene task can happen at any time (randomly) as well as the need arises. In this model, nutrition, hy‐ giene, admission and discharge are identified as both, scheduled and random tasks. Within the simulated environment, there are also “call” tasks that are called directly by the patient. For exam‐ ple, within the task group of Vascular Access, a patient's IV may become blocked. Therefore, the nurse performs IV maintenance, a task that was not scheduled or a random task but in fact this was a task that was called directly by the patient. The Task location was determined for each task, that is occurring at the nurses’ station or patient bedside. Task priority level and task scheduling for the SCDU model are listed in Table 1.

2.3 | Virtual layout of the simulated care delivery unit

The virtual layout was developed using Microsoft Visio software to define the overall floor plan details of an inpatient unit such as the nurse station location, total beds and the distances reflecting the simulated unit layout in the DES model. The virtual layout is also used for visual verification while running the simulation. It allows the software to display the nurse's movement on the layout diagram that helps to visually verify the simulated‐nurse's movement pat‐ terns during simulation trials.

F I G U R E 2   Inputs are patient care data, operating logic and virtual layout, and outputs to the model are care quality (task in queue time, missed care) and nurse workload (task queue and cumulative distance walked)

     |  975QURESHI Et al.

2.4 | Outputs

In this demonstrator simulation model, nurse workload is assessed by task queue, a mental workload indicator representing the num‐ ber of pending tasks which has been associated with medical er‐ rors (Potter et al., 2009). Tasks are generated stochastically by the model according to the frequency and schedule of the unit's histori‐ cal GRASP data (per 2.1). These tasks are recorded in a sequence/ queue as a “stack” for the simulated nurse to perform according to the task priority rules, this stack is called the task queue. Cumulative distance walked by nurse, the total distance walked by the nurse during a shift in metres. Care quality is assessed by calculating the amount of missed care, the number of pending tasks that were not started by the nurse before the end of the shift, and task in queue time, the average amount of time a task has been in queue waiting to be completed.

2.5 | Demonstrator model testing

NPR is defined as the number of patients assigned to a nurse. The SCDU model was simulated on different NPR conditions: Low (1 nurse: 2 patients), Medium (1:4) and High (1:6), each for a period of 252 shifts which is approximately the total working days in a year. Each shift consists of 12 hr which is the standard shift length in nursing for North America. Data for 10 replications were recorded for each operating condition to calculate warm‐up period for the model and to analyse 10 years of nursing data for each operating condition. Warm‐up times are used in simulation for the model to reach an optimal operating state. For this model, a warm‐up period of 41 days was established using Welch's method (Hoad, Robinson, & Davies, 2008); averages across shifts were taken for missed care, task in queue time, task queue and cumulative distance walked.

Task group Priority level (rank)

Task schedule type

Task delivery location

Time duration (min)

Medication 1 Random intervals Bedside 6.51

Vital signs 2 Random intervals Bedside 5.26

Assessment and planning

3 Random intervals Bedside and nurse station

6.93

Vascular access 4 Random intervals Bedside 31.50

Treatments 5 Random intervals Bedside 9.50

Activity 6 Random intervals Bedside 26.10

Consultation 6 Random intervals Bedside 6.00

Hygiene 6 Random intervals and Scheduled interval (8:00 a.m.)

Bedside 13.32

Nutrition 6 Random intervals and Scheduled intervals (8 a.m., 12 p.m., 5 p.m.)

Bedside 17.05

Other direct nursing care

6 Random intervals Bedside 25.65

Admission 6 Scheduled interval (7:30 a.m.)

Bedside 32.10

Discharge 6 Scheduled interval (7:30 a.m.)

Bedside 21.40

Evaluation 6 Random intervals Bedside and nurse station

3.00

Non‐patient care 6 Random intervals Bedside and nurse station

13.79

Elimination 7 Random intervals Bedside 19.91

Teaching and emotional support

8 Random intervals Bedside 19.68

T A B L E 1   List of tasks programmed in the SCDU model. The list contains the task name along with their respective priority levels, task schedule type and time duration where 1 = highest task priority. Time duration for each task group is calculated using a frequency‐weighted average of the sub‐tasks for each group, as reported by GRASP systems

976  |     QURESHI Et al.

3   |   R E S U LT S

A nurse‐focused DES modelling approach was developed, the SCDU, that demonstrated the ability to assess the impact of changing nurse–patient ratios on care quality and nurse work‐ load. The demonstrator model exhibited that as the NPR in‐ creased (Low, Medium, High), nursing workload increased (tasks in queue: 2, 15, 33 tasks, respectively; cumulative walking distance: 279, 269, 595 metres, respectively) and care quality deteriorated (missed care: 17, 24, 53 tasks, respectively; task in queue time: 0.3, 1.0, 1.2 hr, respectively). A summary of these results is presented in Table 2.

3.1 | Nurse workload indicators

As illustrated in Figure 4, the demonstrator model showed an in‐ crease in the number of tasks in queue by 120% when the NPR is increased from medium to high and decreased by 86% when NPR levels changed from medium to low. However, the cumulative dis‐ tance walked increased in both cases, that is when the NPR is in‐ creased from medium to high and medium to low by 110% and 3%, respectively. With the increase in NPR (Low, Medium, High), nurs‐ ing workload increased in terms of task queue by 2, 15, 33 tasks, respectively, and cumulative distance walked by 279, 269 and 595 m, respectively.

F I G U R E 3  Flow chart representing the operating logic of the simulated care delivery unit model

     |  977QURESHI Et al.

3.2 | Care quality indicators

When the NPR is increased from medium to high, the demonstrator model showed an increase in missed care by 120%, and a decrease in missed care by 86% when NPR levels changed from medium to low. Furthermore, task in queue time increased by 20% when the NPR is increased from medium to high and decreased by 70% when NPR levels changed from medium to low. With the increase in NPR (Low, Medium, High), care quality deteriorated–task queue time increased by 0.3, 1.0, 1.2 hr and missed care increased by 17, 24, 53 tasks, re‐ spectively. This effect is illustrated in Figure 5.

4   |   D I S C U S S I O N

In this study, a nurse‐focused DES modelling approach was devel‐ oped, to evaluate the impact of healthcare system design policy

choices on nurse workload and care quality. This is a novel approach in DES as previous simulation studies have only focused on model‐ ling patient flow.

The number of missed care tasks generated from the demon‐ strator model was found to be consistent with the RN4CAST study conducted in medical and/or surgical units of 488 hospitals across 12 European countries where missed care was estimated before the end of the shift (Ausserhofer et al., 2014). The top three missed care tasks reported in this international study were comfort/talking, care planning and patient education which is consistent with the most prevalent areas of missed care identified by the simulated model (“teaching and emotional support” and “assessment and planning”). Therefore, the simulation model was able to demonstrate similar results regarding the types of missed care adding to the validity of this test of DES. Since “teaching and emotional support” were as‐ signed the lowest task priority level, it was expected to have a higher occurrence within missed care tasks. However, “assessment and

Nurse–patient ratio (NPR)

Care quality indicators Nurse workload indicators

Missed care (no. of task)

Task in queue time (hr)

Task in queue (no. of task)

Cumulative walking distance (m)

Low (1:2) 17 0.3 2 279

Medium (1:4) 24 1.0 15 269

High (1:6) 53 1.2 33 595

T A B L E 2  The results for care quality (missed care, task in queue time) and nurse workload indicators (task in queue time, cumulative walking distance)

F I G U R E 4  The nurse workload indicators: mean and St. Deviation of “no. of task queue” (left) and “distance walked by nurse” (right)

F I G U R E 5  The care quality indicators: mean and St. Deviation of “missed care” (left) and “task in queue time” (right)

978  |     QURESHI Et al.

planning” with a priority level of 3 also constituted a greater propor‐ tion of missed care tasks—an unanticipated finding. This is due to the higher task frequency or a number of occurrences for “assessment and planning” in comparison with others. High‐frequency unsched‐ uled tasks that occur towards the end of shift are less likely to be completed in time if any higher priority tasks remain to be done. The quantity of missed care in the simulated model is much larger (17–64 missed care tasks) than reported in the RN4CAST study (range of 1.5–7.5 and mean of 3.6 missed care tasks). One possible expla‐ nation is that the simulated model measured actual missed care, whereas the RN4CAST study measured nurse perceptions of missed care. Another possible reason for the large volume of missed care in the model could be caused by real nurses rushing to keep up with a heavy workload (with possible quality implications) while the simu‐ lated‐nurse only ever operates at the designated GRASP time. The effects of varying task priority levels and the sources of difference between modelled and reported missed care estimates need further research.

Dabney and Kalisch (2015) reported that increased nurse–pa‐ tient ratios were associated with a greater incidence of missed care. A similar relation was observed with the demonstrator modelling results of missed care as high NPR had greater missed care in compar‐ ison with lower NPR. Chapman, Rahman, Courtney, and Chalmers (2016) reported that increased missed care led to increased overtime which can lead to increased workload for nurses (Alghamdi, 2016; McGillis Hall et al., ; Silas, 2015). As illustrated in Figure 5, a small fraction of “missed care” can also be observed for low NPR. Even though a NPR of 1:2 may be lower than is realistic in such wards, it still shows that there are still missed tasks. This was caused by the arrival of tasks at the end of shift that the simulated nurse was un‐ able to complete before shift‐end.

In this model, each room consists of two patient beds; the oper‐ ational logic is programmed in a way that the simulated‐nurse can walk to the nurse station only when all patient bedside priority tasks are completed. For medium NPR level, the simulated‐nurse had to walk between two rooms and a nurse station. Since the two rooms are arranged closely to each other, the simulated‐nurse walked less. However, for low NPR level since there is just one room and a nurse station, the simulated‐nurse walked relatively more (i.e. 4% more). The virtual layout programmed consists of a hypothetical floor layout with scaled drawings of patient rooms and a nurse station. Further research is needed to estimate the impact of floor layout and bed assignment on workload and care quality.

In this study, task(s) in queue is treated as a mental workload in‐ dicator (Potter et al., 2009), but it also related to care quality. The number of tasks in queue has a direct impact on care quality indica‐ tors. If the number of tasks in queue is substantial, then task in queue time and missed care will also be greater, as observed in high NPR.

4.1 | Implications to nursing management

The ability to create a computerized model to simulate nursing care, staffing conditions and related outcomes offers a promising strategy

to test the impact of various administrative decisions on a range of nurse and patient outcomes. For instance, the implementation of en‐ gineering techniques such as Lean may lead to an increased poten‐ tial for making mistakes, injuries and missing less urgent care tasks which lead to a drop‐in the quality of care (Moraros et al., 2016). This novel nurse‐focused approach to DES modelling can provide insight to the impact of this new design policy proactively. This framework for nurse‐focused DES modelling can be adapted to proactively quantify the impacts of proposed policy changes and technical de‐ sign decisions. This could be useful for hospital managers, healthcare practitioners, researchers, architects, engineers and policymakers, and provide a more cost‐effective and safe alternative to the current trial and error methodologies.

4.2 | Methodological issues

Like all computer models, the current model will suffer from the “gar‐ bage‐in, garbage‐out” (GIGO) phenomenon. The current modelling approach needs to be further developed to test and adjust for possi‐ ble in‐data errors. The current demonstrator model was built on ex‐ isting 1‐month data (GRASP) from a metropolitan area hospital and from a single inpatient unit. This data set (GRASP) consisted of only standardized task durations, lacked variability in terms of nurse skill level (novice/expert), and did not reflect patient acuity. If the GRASP data set failed to capture other nurse activities, then the workload in the model would be an underestimate and the care quality would likely decline. Further study is needed on the extent to which the GRASP system captures all relevant nurse activities. Other limita‐ tions included the use of a single subject matter expert to construct the nurse operating logic in the model, and the use of scaled draw‐ ings rather than actual floor plans. Further field validation studies are needed to address these issues. The modelling method itself al‐ lows for testing of the potential impacts of different task prioritiza‐ tion strategies or changes in task mix and this issue required further research to as well. The authors are currently working with a nursing team to explore and refine the approach to establishing and testing different task priority logics.

Future work includes exploring additional indicators for work‐ load and care quality, such as fatigue, biomechanical loading and error rates, testing other unit layouts and design factors such as patient acuity. Using up to 1 year of historical care delivery data (GRASP). A field validation study incorporating nurse experience/ competency levels (novice, expert) and using acuity sensitive time duration inputs would be a needed next step in the development of this DES tool. The model needs to be extended, validated and tested for utility to support real‐world management and decision‐making.

5   |   C O N C L U S I O N

This study demonstrated the capability of a novel nurse‐focused simulation approach that simulated the nurse's process of care deliv‐ ery to help hospital administrators understand, quantify and predict

     |  979QURESHI Et al.

the impact of changing NPRs in terms of nurse workload and care quality. In this simulation, as the NPR increased (from Low, Medium, High), nursing workload increased (120% increase in task in queue; 110% increase in walking distance) and care quality deteriorated (120% increase in missed care; 20% increase in task in queue time). A field validation study is needed to support further development of the SCDU model.

E T H I C A L A P P R O VA L

This research has been reviewed by Ryerson's Research Ethics Board (REB).

O R C I D

Sadeem Munawar Qureshi https://orcid. org/0000‐0003‐4309‐6674

Nancy Purdy https://orcid.org/0000‐0002‐5294‐0729

W. Patrick Neumann https://orcid.org/0000‐0002‐1560‐3870

R E F E R E N C E S

Alghamdi, M. G. (2016). Nursing workload: A concept analysis. Journal of Nursing Management, 24(4), 449–457. https://doi.org/10.1111/ jonm.12354

Ausserhofer, D., Zander, B., Busse, R., Schubert, M., Geest, S. D., Rafferty, A. M., … Sjetne, I. S. (2014). Prevalence, patterns and pre‐ dictors of nursing care left undone in European hospitals: Results from the multicountry cross‐sectional RN4CAST study. BMJ Quality & Safety, 23, 126–135. https://doi.org/10.1136/bmjqs‐2013‐ 002318

Banks, J., Carson, J. S. I., Nelson, N. L., & Nicol, D. M. (2005). Discrete event system simulation, 4th ed.. Upper Saddle River, NJ: Prentice Hall International Series in Industrial and Systems Engineering.

Baril, C., Gascon, V., Miller, J., & Bounhol, C. (2016). Studying nurse workload and patient waiting time in a hematology‐oncology clinic with discrete event simulation. IIE Transactions on Healthcare Systems Engineering, 6(4), 223–234. https://doi.org/10.1080/19488300.2016 .1226212

Barnes, S., Golden, B., & Price, S. (2013). Applications of agent‐based modelling and simulation to healthcare operations management. International Series in Operations Research & Management Science, 184, 45–74.

Boucherie, R. J., Hans, E. W., & Hartmann, T. (2012). Health care lo‐ gistics and space: Accounting for the physical build environment. Proceedings – Winter Simulation Conference, (January). https://doi. org/10.1109/WSC.2012.6465222

Bragadóttir, H., Kalisch, B. J., Smáradóttir, S. B., & Jónsdóttir, H. H. (2015). Translation and psychometric testing of the Icelandic version of the MISSCARE Survey. Scandinavian Journal of Caring Sciences, 29(3), 563–572. https://doi.org/10.1111/scs.12150

Bureau of Labor Statistics (2011). U.S. Department of labor, occupational outlook handbook: A review of 50 years of change. Monthly Labor Review. USA: Bureau of Labor Statistics.

Canadian Federation of Nurses Unions (2015). Overtime and Absenteeism Factsheet.

Canadian Federation of Nurses Unions (2017a). Enough is Enough: Putting a Stop to Violence in the Health Care Sector.

Canadian Federation of Nurses Unions (2017b). Quick Facts 2017, 1–6.

Carayon, P. (2016). Handbook of human factors and ergonomics in health care and patient safety, 2nd ed.. Boca Raton, FL: CRC Press Taylor & Francis Group.

Carayon, P., Wetterneck, T. B., Rivera‐Rodriguez, A. J., Hundt, A. S., Hoonakker, P., Holden, R., & Gurses, A. P. (2014). Human fac‐ tors systems approach to healthcare quality and patient safety. Applied Ergonomics, 45(1), 14–25. https://doi.org/10.1016/j. apergo.2013.04.023

Chapman, R., Rahman, A., Courtney, M., & Chalmers, C. (2016). Impact of teamwork on missed care in four Australian hospitals. Journal of Clinical Nursing, 26, 170–181. https://doi.org/10.1111/jocn.13433

Choudhary, R., Bafna, S., Heo, Y., Hendrich, A., & Chow, M. (2010). A pre‐ dictive model for computing the influence of space layouts on nurses’ movement in hospital units. Journal of Building Performance Simulation, 3(3), 171–184. https://doi.org/10.1080/19401490903174280

Clancy, T. R., & Delaney, C. W. (2005). Complex nursing systems. Journal of Nursing Management, 13(3), 192–201. https://doi. org/10.1111/j.1365‐2834.2004.00518.x

Dabney, B. W., & Kalisch, B. J. (2015). Nurse staffing levels and patient‐ reported missed nursing care. Journal of Nursing Care Quality, 30(4), 306–312. https://doi.org/10.1097/NCQ.0000000000000123

Dode, P., Greig, M., Zolfaghari, S., & Neumann, W. P. (2016). Integrating human factors into discrete event simulation: A proactive approach to simultaneously design for system performance and employees’ well being. International Journal of Production Research, 54(10), 3105. https://doi.org/10.1080/00207543.2016.1166287

Duguay, C., & Chetouane, F. (2007). Modelling and improving emergency department systems using discrete event simulation. Computer Science and Software Engineering, 1(63), 311–320.

Farrington, M., Trundle, C., Redpath, C., & Anderson, L. (2000). Effects on nursing workload of different methicillin‐resistant Staphylococcus aureus (MRSA) control strategies. Journal of Hospital Infection, 46(2), 118–122. https://doi.org/10.1053/jhin.2000.0808

Gaba, D. M. (2007). The future vision of simulation in healthcare. Simulation in Healthcare: the Journal of the Society for Simulation in Healthcare, 2(2), 126–135. https://doi.org/10.1097/01. SIH.0000258411.38212.32

Hendry, C., & Walker, A. (2004). Priority setting in clinical nursing prac‐ tice: Literature review. Journal of Advanced Nursing, 47(4), 427–436. https://doi.org/10.1111/j.1365‐2648.2004.03120.x

Hoad, K., Robinson, S., & Davies, R. (2008). Automating warm‐up length estimation. Proceedings – Winter Simulation Conference, 532–540. https://doi.org/10.1109/WSC.2008.4736110

Holden, R. J., Carayon, P., Gurses, A. P., Hoonakker, P., Hundt, A. S., Ozok, A. A., & Rivera‐Rodriguez, A. J. (2013). SEIPS 2.0: A human factors framework for studying and improving the work of healthcare pro‐ fessionals and patients. Ergonomics, 56(11), 1–30. https://doi.org/10. 1080/00140139.2013.838643

International Council of Nurses (2015). International Classification for Nursing Practice (ICNP®).

International Ergonomics Association (2018). Definition and Domains of Ergonomics. Retrieved from http://www.iea.cc/whats/index.html.

Jun, J. B., Jacobson, S. H., & Swisher, J. R. (1999). Application of dis‐ crete‐event simulation in health care clinics: A survey. Journal of the Operational Research Society, 50(2), 109–123. https://doi. org/10.1057/palgrave.jors.2600669

Kalisch, B. J., & Williams, R. A. (2009). Development and psychomet‐ ric testing of a tool to measure missed nursing care. Journal of Nursing Administration, 39(5), 211–219. https://doi.org/10.1097/ NNA.0b013e3181a23cf5

Komashie, A., & Mousavi, A. (2005). Modeling emergency departments using discrete event simulation techniques. In Proceedings of the 37th Conference on Winter Simulation (pp. 2681–2685). Winter Simulation Conference. Retrieved from http://dl.acm.org/citation. cfm?xml:id=1162708.1163203.

980  |     QURESHI Et al.

Lambton, J., Roeder, T., Saltzman, R., Param, L., & Fernandes, R. (2017). Using simulation to model improvements in Pediatric bed placement in an acute care hospital. Journal of Nursing Administration, 47(2), 88– 93. https://doi.org/10.1097/NNA.0000000000000445

Letiche, H. (2008). Making healthcare care: Managing via simple guiding principles. USA: IAP.

McGillis Hall, L., Doran, D., Tregunno, D., McCutcheon, A., O’Brien‐ Pallas, L., Tranmer, J., … Thomson, D. (2005). Quality work environ‐ mentsfor nurse and patient safety (L. McGillis Hall, Ed.). Sudbury, MA: Jones and Bartlett Publishers Inc.

Moraros, J., Lemstra, M., & Nwankwo, C. (2016). Lean interventions in healthcare: Do they actually work? A systematic literature review. International Journal for Quality in Health Care, 28(2), 150–165. https://doi.org/10.1093/intqhc/mzv123

Nursing Task Force (1999). GoodNursing, GoodHealth: An investment for the 21st century. Ontario, Canada: Ministry of Health and Long‐Term Care.

Potter, P., Wolf, L., Boxerman, S., Grayson, D., Sledge, J., Dunagan, C., & Evanoff, B. (2009). An analysis of nurses’ cognitive work: A new perspective for understanding medical errors. International Journal of Healthcare Information Systems and Informatics, 4(3), 39–52.

Rizo‐Baeza, M., Mendiola‐Infante, S. V., Sepehri, A., Palazón‐Bru, A., Gil‐Guillén, V. F., & Cortés‐Castell, E. (2018). Burnout syndrome in nurses working in palliative care units: An analysis of associated

factors. Journal of Nursing Management, 26(1), 19–25. https://doi. org/10.1111/jonm.12506

Silas, L. (2015). Creating safe cultures and work environments for nurses. In Quality and Safety Summit: Leveraging Nursing Leadership. November 23 & 24, 2015. Toronto.

Song, D., Chung, F., Ronayne, M., Ward, B., Yogendran, S., & Sibbick, C. (2004). Fast‐tracking (bypassing the PACU) does not reduce nursing workload after ambulatory surgery. British Journal of Anaesthesia, 93(6), 768–774. https://doi.org/10.1093/bja/aeh265

Swisher, J. R., & Jacobson, S. H. (2002). Evaluating the design of a fam‐ ily practice healthcare clinic using discrete‐event simulation. Health Care Management Science, 5(2), 75–88. https://doi.org/10.1023/A: 1014464529565

How to cite this article: Qureshi SM, Purdy N, Mohani A, Neumann WP. Predicting the effect of nurse–patient ratio on nurse workload and care quality using discrete event simulation. J Nurs Manag. 2019;27:971–980. https://doi. org/10.1111/jonm.12757