Taking the Right Decision at the Right Time: A Case Study to Investigate Timely Disposition Impeding Factors at the Emergency Department

(1)

Taking the Right Decision at the Right Time:

A Case Study to Investigate Timely Disposition Impeding Factors

at the Emergency Department

Master Thesis Dual Degree Operations Management

9 December 2019

Evi Dubbink

Groningen supervisor: M. J. Land Newcastle supervisor: A. De

S2973669 B80573439

(2)

ABSTRACT

Crowding in emergency departments (EDs) is a major barrier for providing timely patient care. However, timely determination of a patient requiring discharge or admission, i.e. the disposition decision, might contribute to a shorter patient length of stay (LOS), improve departmental flow and reduce the probability of crowding to occur. The objective of this study is to identify process- and contextual-related factors that result in a prolonged duration of the disposition process by dividing the disposition process into two consecutive stages, namely recognition of patient’s readiness for disposition and disposition decision. A case study is conducted at the University Medical Center Groningen (UMCG) including 28 discharged (=43%) and 37 admitted (=57%) patients. Results show that diverse process-related factors might influence the total duration of the disposition process. However, waiting is most strongly determined by consults from another specialist required (p<0.05), contact with supervisor after all diagnostic input (p<0.05) and waiting for supplementary results to be available (p<0.05). The relationship between operational processes and timely disposition decision-making is moderated by different contextual factors, e.g. the experience of a physician, different medical specialties involved in patient care and the level of crowding at the ED. This study contributes to a greater understanding of the disposition process, allowing to support the ability for timely disposition decision-making to reduce crowding in the ED.

Keywords: Crowding, emergency department, disposition process, timely disposition

(3)

1. INTRODUCTION

Emergency departments (EDs) are a critical component of the healthcare safety net, available 24 hours per day for everyone who demands for care. However, crowding in the ED has become an increasingly significant and international health care challenge over the last decade (Di Somma et al., 2015). According to Van der Linden et al. (2013), 68% of the hospitals in the Netherlands experience crowding several times a week or even daily. The consequences of crowding are mainly negative and assumed to reach even beyond the particular ED (Eckstein & Chan, 2004). Solutions are not universal but can be found in diminishing inflow, shortening throughput times or improving outflow (Morley et al., 2018). Concerning outflow, Di Somma et al. (2015) argue that one of the main contributors to crowding occurs within the hospitalization process: an inability to move admitted patients to inpatient beds promptly. Although several strategies are proposed to improve the hospitalization process, crowding can already originate in the throughput stage concerning the decision to either discharge or admit a patient, i.e. the disposition decision. However, before efficiency of disposition decision-making can be improved, factors should be identified that currently cause a prolonged duration of the disposition process.

Delays in patient flow result in bottlenecks in hospital operations and impact admissions from the ED, the general admitting unit and the operating room (El-Eid et al., 2015). Previous research identified that timely disposition contributes to preventing delays (Pourmand et al., 2013), allowing for a shorter patient LOS (White et al., 2012). However, timely disposition is challenging since it requires coordination of multiple different groups including physicians, nurses, patients and service staff (Watts & Gardner, 2005). According to Ebrahimian et al. (2014), decision-making of these groups related to patient logistics is not solely based on the status of the emergency care patient, but also on cultural and socioeconomic status, characteristics of the staff, and conditions of the mission. However, it is urgent to assess the disposition decision early in the process to plan patient movements ahead. Although research identified what factors influence decision-making of staff, it is unknown what process-related factors support the ability for timely disposition decision-making and how this relationship is influenced by different contextual-related factors.

(5)

level of crowding) and timely disposition decision-making. This paper contributes to the acceleration of the disposition decision-making process by analysing the current causes of delay in two stages of the disposition process, namely at patient’s readiness for disposition and the disposition decision. In the remainder of this thesis, recognition by a physician whether a patient should be discharged or admitted will be acknowledged as the readiness for disposition, whereas communication of the disposition with coordinating staff will be referred to as the disposition decision. To identify what factors are currently contributing to a prolonged duration of the disposition process, this paper addresses the following research question:

“What are the main factors contributing to a delay in the disposition decision-making process?”

To impede negative impacts on hospital operations, one must understand the underlying causes of delay since this allows to positively impact revenue, save time, and increase patient and staff satisfaction (Institute for Healthcare Optimization, 2018). Therefore, identification of factors that cause delay in disposition decision-making might support health care professionals and hospital management in the process to identify and mitigate ED crowding (Ebrahimian et al., 2018). A case study at a Dutch ED of the University Medical Centre Groningen is conducted, since delays in reaching a disposition decision is one of the main contributors to crowding in the Netherlands (Van der Linden et al., 2013). In terms of this researchs practical contribution, the findings of this study improve understanding of the disposition decision-making conditions within Dutch EDs, allowing to identify bottlenecks affecting patient flow.

(6)

2. THEORETICAL BACKGROUND

The basis of this thesis is grounded in literature, providing a framework of previous research while addressing the importance of this study. Concepts and characteristics relevant for timely disposition are defined, to create a structured framework for the methodology of this thesis.

2.1 Crowding in emergency departments

The American College of Emergency Physicians (2006) defined crowding as the situation in which the identified need for emergency services exceeds available resources for patient care in the emergency department, hospital, or both. Crowding is a major barrier to provide timely emergency care and comes along with various negative consequences. Studies reported that frequent consequences are an increased length of stay (LOS) (Higginson, 2012), higher patient mortality and complication rates (Schull et al., 2004), reduced staff morale and a lower level of patient satisfaction (Derlet & Richards, 2000).

Asplin et al. (2003) established a conceptual model applying operational management principles in order to understand causes of crowding at the ED. This model partitions ED crowding into three different components: input, throughput and output. First, waiting in the input component refers to the total demand for ED services. Reducing input can be realized by e.g. enhanced identification of patients who can be treated in a non-urgent care setting, improving collaboration with general practitioners or a temporary ambulance ban. Second, the throughput component refers to the total LOS of patients. Within this component, one needs to look at internal ED care processes and modify them if necessary to enhance efficiency and effectiveness. Shorter ED throughput times result from minimizing the time needed for diagnostic tests or by increasing personnel capacity (Jarvis, 2016). Lastly, the output component refers to the inefficient discharge of ED patients (Derlet, Richards & Kravitz, 2009). Inefficient discharge entails the inability to move admitted patients from the ED to an inpatient bed and can be improved by e.g. creating observation wards (Jarvis, 2016).

(7)

discharge or admit a patient, i.e. readiness for disposition. A patient is considered ready for disposition when it is authorized by ED staff as safe to depart in accordance with the hospital policy principles (NSW Health, 2014). Generally, patients are discharged in case no further treatment at the hospital is required, whereas patients are admitted in case further treatment is necessary. Once readiness for disposition is recognized, the disposition decision should be communicated with coordinating staff, allowing the patient to depart the ED.

2.2 Importance of timely disposition decision-making

The disposition decision is a logistic decision made by the physician. It determines a patient’s pathway throughout the hospital and allows the physician to have control over the system (Kingsman et al., 2010). Early assessment of a logistic situation creates a responsive manner to cope with discrepancies, improves flexibility to reduce working pressure by managing scarce time and space, and allows to create a more robust system (Nugus et al., 2011). However, early assessment may be challenging since the disposition decision is a complex, multiphase phenomenon (Bobay et al., 2010).

(8)

2.3 Characteristics influencing disposition decision-making

The disposition of a patient is influenced by hospital- (e.g. medical resources or existing care protocols), patient- (e.g. medical condition, diagnosis or preference) and physician-related factors (e.g. knowledge or self-defined risk thresholds) (Calder et al., 2012). For patients in critical conditions, the decision to discharge or admit a patient might be straightforward. However, in some situations decisions are not evident and factors must be carefully weighed to make a decision based on clinical gestalt. Decision-making based on clinical gestalt is the theory that physicians actively organize clinical perceptions into coherent construct wholes (Cook, C. 2009). Hereby, different processes may be involved to allow for timely decision-making.

How the multi-disciplinary team within the ED is organized and functions amongst varying circumstances is fundamental for timely disposition (Epstein, 2014). Therefore, it is important to increase understanding about how patient pathways are affected within different contextual ED settings. Three contextual factors are included, namely physician experience, medical specialty and level of crowding. All factors are related to total patient LOS and might affect the disposition process to some extent. The following sections discuss literature of each contextual factor on total patient LOS and its potential effect on disposition decision-making.

2.3.1 Physician experience

Experienced physicians have accumulated skills and knowledge during years of practice, allowing to improve departmental flow (Li et al., 2016). However, experienced physicians are more conservative in their decisions. They encounter lower discharge rates at all triage levels and are more likely to retain ED patients compared to their junior colleagues. Therefore, although experienced physicians are able to offer the best quality of care, this goes accompanied with a slightly longer LOS because of a prolonged duration of the discharge process.

(9)

2.3.2 Medical specialty

Previous studies have mentioned alterations in managing the disposition decision between emergency physicians and other specialists (Colucci, Kordick & Chan, 2004; Hack, O’Brien & Benson, 2005). According to Vegting et al. (2011) there is a dependency between medical specialty and total time to completion, including examination, treatment and disposition. Furthermore, in case multiple specialists are involved in the care for a patient, patients are more likely to experience a longer LOS. An underlying reason for this is that different specialties tend to work individually, and not in a team. Apparently, a coordination of care is lacking when multiple specialties are involved (Vegting et al., 2011).

2.3.3 Level of crowding

(10)

2.4 Summary

This literature review highlights the importance of timely disposition decision-making. However, what operational processes support timely disposition decision-making and how this relationship is moderated by different contextual factors, is not mentioned. However, this will be further investigated in this research using the research model as visualized in figure 1. According to literature, the identified contextual factors influence total LOS, but it is unknown how they influence the disposition process specifically. Therefore, these factors are deliberately appropriate to investigate.

(11)

3. METHODOLOGY

This chapter provides a discussion on the research methodology of the study, explaining the method used to provide an answer to the stated research question. The studied case is discussed with an explanation of the sampling method. Furthermore, the method of data collection is described followed by an analysis of the data, including examination of the moderating variables.

3.1 Embedded multiple case study

An embedded multiple-case study design offers the opportunity to explore the nature of the complex disposition decision-making phenomenon through an inductive lens, while also applying quantitative measures to objectively assess physician experience, medical specialties and crowding level factors common to the successful timely disposition decision-making context. Yin (2009) indicates that a case study is appropriate when investigators aim to define broad research topics, cover complex conditions and rely on multiple sources of evidence. Therefore, a case study allows to create an improved understanding of how physicians interpret, construct, and attribute meaning to their experiences (Merriam, 1998).

Merriam (1998) suggests that including multiple cases in a study makes the findings and interpretations more compelling. Studying multiple cases of the same phenomenon might corroborate, qualify, or extent the findings that might occur from studying only one case. Within this study, a single case stands for the pathway of one specific patient during the disposition process, i.e. the unit of analysis of this study is a patient’s pathway. To identify the moderating effects, multiple patients are studied in different contextual factors, namely at different levels of physician experience, medical specialties and levels of crowding within the ED.

This case study can be qualified as exploratory. The exploratory nature allows for an extensive examination of the disposition process in a rich content, enabling collection of a wide variety of data from different sources. This varied and rich data enables to discuss and interpret the findings in the context of the literature, allowing for a more holistic and meaningful understanding of the complex disposition phenomenon while contributing to validity of the research (Yin, 2009).

(12)

complex disposition process than if only one type of data were to have been utilized. Qualitative data was gathered from the interviews, where after quantitative statistical relationships are constructed based on these qualitative interview findings and time stamps from the hospital’s electronic system Epic.

In an embedded design, different levels of data are collected and subunits reside within the main unit (Yin, 2009). As such, the main unit in this study are patients in the ED. Hereby, a distinction is made between discharged and admitted patients, since it appears that differences exist in the duration of the disposition decision-making process between discharged and admitted patients (Van der Linden et al., 2016). The subunits within the main units are patients treated by different 1) physician experience levels, 2) medical specialists and 3) levels of crowding.

3.2 Hospital case description

The study is performed at the University Medical Centre Groningen (UMCG) which is a Dutch academic hospital with approximately 34.000 ED visits each year. UMCG functions as a fundamental base for this research since it enables complex care treatment: patients requiring complex investigations and/or treatments, or people suffering a rare disease or other disorders. Because of this complex care treatment, the interaction level between physicians is more complicated compared to a general hospital. Nonetheless, the high level of interaction complexity allows to create a wide analysis of factors promoting timely disposition decision-making processes.

The Dutch setting of EDs is characterized by the presence of both emergency physicians and other specialists. In total, six different medical specialities are operating at the ED of UMCG:

 Emergency medicine: Illnesses or injuries requiring immediate medical attention for unscheduled and undifferentiated patients of all ages

 Internal medicine: Prevention, diagnosis, and treatment of adult diseases

 Surgery: Operative manual and instrumental techniques used on a patient to investigate or treat a pathological condition such as disease or injury, to help improve bodily function or appearance or to repair unwanted ruptured areas

 Cardiology: Diseases and abnormalities of the heart  Neurology: Disorders of the nervous system

(13)

Walk-in patients are initially checked by the triage nurse. The triage nurse decides on the urgency of need for care, which indicates which patients have to be treated first and what the total waiting time will be. When a bed is available and the physician is ready for treatment, the physician will request the patient to enter the ED. In case a patient arrives by ambulance, the ambulance team will place the patient under coordination of the nurse within one of the treatment rooms. Treatment of a patient may require the work of one or multiple specialists involved. After treatment, physicians determine whether a patient can be discharged home or requires admission to a hospital ward.

In case of discharge, physicians will either discharge the patient themselves or will instruct the nurse to discharge the patient if still additional care of the patient is necessary, e.g. bandage. In case of admission, after treatment the physician has to find an available bed at a ward. Physicians may either search for an available bed themselves, or may request the coordinating nurse to locate one.

However, before leaving the ED, patients have to be recognized as being ready for disposition which may require conclusions from multiple parties. Furthermore, before the disposition decision can take place, additional time may expire. In UMCG there are no established guidelines regarding when can be recognized that a patient is ready for disposition, which results in a grey area between a patient’s readiness for disposition and the time the disposition decision is made.

3.3 Data collection

(14)

To reach a high degree of credibility, criteria according to Strauss & Corbin (2014) for good interviews will be followed. According to Strauss & Corbin (2014), credible findings are “trustworthy and believable in that they reflect participants, researchers, and readers’ experiences with a phenomenon”. To achieve credibility, data will be collected from different viewpoints, i.e. different physicians. To increase validity of interview data, all physicians will be asked the same questions and answers are recorded by note-taking and digitally using a mobile phone.

Based on the interviews, primary data about the disposition is collected on the last patient treated by the interviewed physician. Factors that might contribute to a prolonged duration of the disposition are identified and categorized in one of the following two intervals: 1) last diagnostic input available until patient ready for disposition or 2) patient ready for disposition – disposition decision. Clarification of the highlighted concepts can be found in table 3.1 below.

Table 3.1 - Clarification of key concepts

Key event Explanation

Last diagnostic input available

The last diagnostic input that is required to form a diagnosis is available. Input at this stage can consist of treatments and procedures, diagnostic testing, monitoring and reassessment or consults. It may differ per patient what the last diagnostic input required is before diagnosis can be formed. At this stage, the physician still has to integrate and interpret the information obtained

Readiness for disposition Moment in time diagnosis is formed (including medical

decision-making) based on last diagnostic input available

Disposition decision

(15)

To ensure convergence of views, a printed version of the timeline of key events being studied as visualized in figure 2 is presented to the interviewee.

Figure 2 Timeline key events being studied

3.3.1 Sampling method

In case research, a sample of cases is selected according to different criteria (Eisenhardt, 1989). Purposive sampling is used to identify and select information-rich cases for the most effective use of the limited resources (Patton, 2002). To be eligible to participate in this study, a participant is either 1) a doctor in training for one of the medical specialties, 2) a doctor not in training for one of the medical specialties or 3) a specialist. Physicians working in the ED for other medical specialties than mentioned in paragraph 3.2 are excluded from the study. Physicians are requested to provide in-depth and detailed information about the patient stream under investigation, allowing to increase generalizability by offering the interviewed physicians the opportunity to provide a detailed indication into their overall thoughts and outcomes. To promote reliability, interviews are conducted until responses become more consistent across larger number of samples. Moreover, interviews are conducted until no significant new themes emerge from the data.

3.4 Data analysis

(16)

labels for the condensed meaning units are developed. Codes allow to describe the condensed meaning unit and may help this study to reflect on data in new ways (Erlingsson & Brysiewicz, 2017). Codes are divided into two primary codes based on the two different studied intervals. Secondary codes are grouped elements causing a delay in either one of the intervals. Hereby, the goal is to become familiar with each case as a stand-alone entity, and to allow unique patterns of each case to emerge before it can be generalized across cases (Eisenhardt, 1989).

(17)

3.4.1 Analysis of moderating effects

To identify the moderating effects of the three contextual factors, a regression analysis is conducted indicating the interacting effect of operational processes and moderators on the ability of timely disposition decision-making. Hereby, it is examined whether the moderator strengthens (or not) the relationship between an identified operational process and a prolonged duration of the disposition process.

Physician experience

In total, this study included 65 interviews with physician’s seniority ranging from 1 to 23 years with an average experience level of 4,5 years. Physicians are categorized into three groups according to their levels of experience:

 Junior: ≤2 years (n=27)

 Intermediate: 2-5 years (n=18)  Senior: >5 years (n=20)

To identify in which group the interviewed physician belongs, the first question of the interview is “How much years of working experience as a physician do you have?”.

Medical specialty

Six different medical specialties are active at the ED of UMCG: 1. Emergency medicine 2. Internal medicine 3. Surgery 4. Cardiology 5. Neurology 6. Pulmonology

(18)

Level of crowding

The occupancy rate1_{as a measure of ED crowding makes real-time assessment of crowding}

feasible for Eds (McCarthy et al., 2008). The average occupancy rate, including standard deviation, is set as a baseline to indicate at what level this study considers the ED to be crowded. The average amount of patients in the ED during the period of interviews is measured using Epic. Furthermore, 23 treatments beds are available at the ED of UMCG. On average, the occupancy rate was measured at 0,45 (=10 beds) with ± standard deviation of 0,30 (=7 beds). Therefore, in case more than 17 beds were occupied, the ED is considered to be crowded. This measurement is validated by asking physician about their opinions regarding at what level they consider the ED to be crowded.

(19)

4. FINDINGS AND DISCUSSION

This chapter reports the results of this study. First, an indication is created about what operational processes are relevant in the disposition context. Second, insight is provided about factors that might potentially cause a prolonged duration of the disposition process, distinguishing between factors occurring in the first (last diagnostic input available until readiness for disposition) and second (readiness for disposition until disposition decision) interval. To identify whether interrelations exist between factors, a correlation analysis is conducted. Lastly, the moderating effect of physician experience, medical specialty and level of crowding on the relationship between operational processes and timely disposition decision-making will be analyzed, to identify how disposition is affected in different contexts.

4.1 Disposition operational processes

In total 65 interviews are conducted, analysing pathways of 28 discharged (=43%) and 37 (=57%) admitted patients. Five operational processes are identified emerging relevant for the disposition decision. These operational processes are identified as visualized in figure 3 below.

Figure 3 Cycle of operational processes relevant for disposition

(20)

Operational processes as mentioned in figure 3 can be classified into one of the studied intervals. An oversight and explanation of each process can be found in table 4.1 below.

Table 4.1 - Operational processes classified based on relevant interval Operational

process Explanation

Last diagnostic input – readiness for disposition

Interpretation of diagnostic

results

Analysation of last diagnostic results available, e.g. physical examination,

laboratory or radiology testing Communication

supervisor/co-assistant

Confirmation from supervisor/involvement of co-assistant in recognition of disposition

Forming a diagnosis

Indication of the disposition readiness (discharge versus admission) based on all

information collected Readiness for disposition –

disposition decision Arrangement

additional supplements

Additional supplements required before patient can depart the ED, e.g. medication

or available bed. These additional supplements do not determine whether patient requires discharge or admission Communication

of diagnosis

Communicating the decision to patient or (coordinating) nurse

According to the interviews, dissimilarities exist between operational processes depending on whether patient is discharged or admitted. Therefore, this study differentiates analysis between discharged and admitted patient’s pathways. Relevant dissimilarities emerge within three out of five operational processes, namely in interpretation of diagnostic results, arrangement of additional supplements and communication of diagnosis.

(21)

Figure 4 Last diagnostic result required to form diagnosis in discharge versus admitted cases

For discharged patients, in most cases a diagnosis was formed based on laboratory or radiology results (16 out of 28 cases). For admitted patients, diagnosis was mostly formed based on status of the patient at arrival of the ED (13 out of 37 cases).

Arranging additional supplements for discharged patients only refers to medication, if necessary. However, for admitted patients, besides medication, an available bed at a ward also needs to be arranged.

Lastly, communication of the diagnosis for discharged patients is only centred on direct communication to the patient. After contact, the patient can leave if no additional care from the nurse is required. In case of admission, this decision has to be communicated with the (coordinating) nurse. Hereby, the nurse can prepare the patient for admission by e.g. arranging an available bed.

(22)

4.2 Factors related to the first interval

Based on outcomes of the interviews, four factors are identified that might influence the total duration between last diagnostic input available until patient ready for disposition. For all factors, identification of the frequency a factor occurred in discharge versus admitted cases is provided.

i. Contact with supervisor

Because of the academic background of the UMCG, physicians are educated during their shifts before becoming a fully trained professional. In case a physician is either a ‘doctor in training’ or a ‘doctor not in training’ for one of the medical specialties, contact with the supervisor is obligatory. However, since no established guidelines exist, it varies per physician when contact with the supervisor happens. While some physicians contact the supervisor before all diagnostic input is available, others contact the supervisor after all diagnostic input is available. Communication at an early stage, i.e. before all diagnostic input available, contributes to an efficient disposition process since minimum time is wasted.

Discharge

 21 out of 28 (=75%) cases contact with the supervisor occurred

o 13 out of 28 (=46%) cases contact with the supervisor occurred after all diagnostic input available

Admission

 33 out of 37 cases (=89%) cases contact with the supervisor occurred

o 12 out of 37 (=32%) cases contact with the supervisor occurred after all diagnostic input available

(23)

ii. Involvement of co-assistant

Physicians regularly involve co-assistants2 in the disposition process. Although physicians might be able to indicate a patient’s readiness for disposition individually, co-assistants are occasionally involved because of the academic background of the hospital. The physician asks the co-assistant challenging questions, contributing to the co-assistants learning process.

Discharge

 3 out of 28 (=11%) cases co-assistant involved

Admission  None

Although in none of the admitted cases a co-assistant is involved, it cannot be assumed that co-assistants are never involved in admission of a patient. For both discharge and admission, communication with co-assistants might occur, but it does not happen frequently.

iii. Consult from other specialist

Physician can request consults from specialists active within or outside the ED. In some cases, this consult cannot be provided immediately, e.g. in case specialist requires additional information from another professional. However, UMCG has created a target goal stating consults should be provided within one hour after request.

Discharge

 4 out of 28 (=14%) cases additional consult required

Admission

 9 out of 37 (=24%) cases additional consult required

For admission, consult from another specialist is required more often. An underlying reason for this could be the increased complexity of care required for admitted patients compared to discharged patients.

(24)

iv. Laboratory/radiology results availability

In case readiness for disposition is recognized by the physician based upon laboratory or radiology results, the probability exists that results are not seen at time of availability. Figure 4 in section 4.1 indicates when patient’s readiness for disposition was recognized based upon laboratory or radiology results.

Discharge

 16 out of 28 (=57%) cases based upon laboratory or radiology results o 6 out of 16 (=38%) cases not seen at time of availability

Admission

 8 out of 37 (=21%) cases based upon laboratory or radiology results o 3 out of 8 (=38%) cases not seen at time of availability

Readiness for discharge is more frequently recognized upon laboratory or radiology results compared to admission. The reason behind this could be the fact that laboratory or radiology results might already provide sufficient information to recognize a patient is ready for discharge. Although, one could not argue the chance of not seeing the results on time of availability is higher for either discharged or admitted patients.

(25)

4.2.1 Factors contributing to prolonged duration of the first interval

Table 4.2 below summarizes the frequencies a factor appeared in a patient’s pathway, as discussed in the previous section.

Table 4.2 – Factors occurring within the first interval

Discharge (n=28) Admission (n=37)

Contact supervisor 21 33

Involvement co-assistant 3 -

Consult other specialism 4 9

Laboratory/radiology results availability 16 8

Although table 4.2 indicates the frequencies factors appeared within a specific case, it is important to note that these factors do not necessarily result into a prolonged duration of the disposition process. To provide insights in the duration of the first interval for discharged and admitted patients, an overview of the mean, standard deviation and median duration in minutes is reviewed in table 4.3 below.

Table 4.3 - Mean, standard deviation and median duration in minutes of first interval

Mean duration 38,1 76,4

Standard deviation 29,4 48,9

Median 34 73

Table 4.3 implies the mean duration of the first interval for admitted patients is approximately twofold compared to discharged patients. Furthermore, the coefficient of variation3 indicates discharged patients (CV=0,77) experience relative higher levels of variation compared to admitted patients (CV=0,64). Therefore, admitted patients experience slightly more consistent durations of the first interval compared to discharged patients. Assuming unequal variances, the difference between the two means is highly significant (p<0.001).

(26)

Figure 5 below shows a distribution curve indicating data is close to normally distributed around the mean. An histogram dividing all values into series of intervals and the frequency of how many values fall in each interval can be found in Appendix C.

Figure 5 Nominal distribution curve discharge and admission

Table 4.4 below indicates whether factors occurred more often in the right-hand tail, i.e. highest 50% of the distribution according to figure 5 by using the median as a cut-off point. If the duration of the first interval for a discharged case is >34 minutes and for an admitted case >73 minutes, this study assumes this case experiences a prolonged duration of the disposition process. Subsequently, the probability for the disposition process to result in a delay is higher.

0 0,002 0,004 0,006 0,008 0,01 0,012 0,014 0,016 0 50 100 150 200 250 Pro b a b il ity d en siity Duration (minutes)

Distribution curve: Interval 1

(27)

Table 4.4 - Significance table factors causing delay using Chi-Square test

Comparing percentages of cases where duration of the interval is above the median compared to cases with duration below the median, one can assume contact with supervisor after all diagnostic input, consult other specialist required and results not seen when available are all more prospective to result in a longer duration of the disposition process for both discharged and admitted patients.

In three discharged cases, consult with co-assistant is mentioned as a factor that might contribute to a prolonged duration of the disposition process. However, according to table 4.4, contact with co-assistant is in most cases not expected to result in a longer duration of the disposition process, i.e. 33,3% of cases only resulted in above median duration.

Factor Prolonged duration? p-value

No Yes

Discharge Admission Discharge Admission Discharge Admission

(28)

However, when testing for relationships between prolonged cases in which a factor occurred versus cases were factor did not occur, only two significant differences exist. These significant differences suggest that consult from another specialist (for discharge) and contact with supervisor after all diagnostic input (for admission) are expected to significantly contribute to higher probability of delay within the first interval. Therefore, occurrence of delay is most strongly determined by waiting for consult from another specialist and contact supervisor compared to other factors.

For discharged patients, longer durations were observed in cases where consult from another specialist was required (100%), compared to cases were no consult from another specialist was required (41,7%). These result might seem rather obvious, since extra complication results in higher probabilities of delay. However, for admitted patients, consult from another specialist does not significantly contribute to a higher probability of delay. For admitted patients, longer durations were observed in cases where contact with the supervisor occurred after all diagnostic input available (75%), compared to cases where contact with the supervisor occurred before all diagnostic input available (36%). An underlying reason for delay to be more probable to occur in case contact is sought after all information is available, has to do with the fact that waiting for diagnostic results might be time-consuming. Therefore, in case contact with the supervisor occurs already before all information available, waiting time can be reduced since based on the supervisor’s advice a patient’s readiness for disposition might already be recognized. This allows to recognize a patient’s readiness for disposition at an earlier stage.

(29)

4.3 Factors related to the second interval

Based on outcomes of the interviews, five factors are identified that might influence the total duration between patient ready for disposition until disposition decision.

i. Supervisor requires additional patient check

The supervisor may require an additional patient check to obtain extra information. This information is required before the disposition decision can be made as it might result in e.g. patient requiring prescriptions.

Discharge

 5 out of 28 (=18%) cases supervisor required additional patient check

Admission

 6 out of 37 (=16%) cases supervisor required additional patient check

One cannot assume that for either discharge or admission the supervisor more likely requires an additional patient check, since the frequencies this factor appeared are relatively identical.

ii. Waiting for supplementary results

Physicians might need to wait for supplementary results to be available to obtain full certainty about the disposition decision. Often, physicians wait for all supplementary results to be available before communication with (coordinating) nurse or patient occurs.

Discharge

 7 out of 28 (=25%) cases waiting for supplementary results

Admission

 20 out of 37 (=54%) cases waiting for supplementary results

(30)

iii. Patient requires prescriptions

Physicians must determine if a patient requires medication for use at home or at a ward. Occasionally, a pharmacist or microbiologist has to be involved to recognize which medication is suitable.

Discharge

 9 out of 28 (=32%) cases patient required prescription

Admission

 5 out of 37 (=14%) cases patient required prescription

In both discharged and admitted cases prescriptions were frequently required. Although, in case of discharge, prescription is more likely to be necessary. An underlying reason for this could be the fact that a patient is not further treated at the hospital and therefore requires prescriptions to continue for safe care at home.

iv. Search for an available bed

Before a bed can be searched, a suitable department to admit the patient to should be determined. Physicians might either search a bed/ward individually, or request the nurse coordinator to do so. According to information from UMCG, beds for Internal Medicine patients are always arranged by the coordinator.

Discharge  None

Admission

 12 out of 37 (=32%) cases search for available bed/ward

(31)

v. Finishing patient administration before communication

In UMCG, no established guidelines exist about when patient administration in Epic should be completed and when communication with the (coordinating) nurse or patient is expected. In case of admission, it is most efficient for the physician to timely contact the coordinating nurse to start the search for an available bed. However, if a physician arranges a bed individually, this factor is irrelevant for causing a potential delay since communication then does not contribute to a patient being able to depart the ED earlier.

For discharged cases, this study considers whether communication with the patient (instead of with the coordinating nurse) occurred before or after administration. In case administration is completed before the patient is noted, this factor might result in a delay of the disposition decision since the patient is unnecessarily waiting at an ED room.

Discharge

 12 out of 28 (=43%) cases physician finished administration before communicating with the patient

Admission

 12 out of 37 (=32%) cases physicians finished administration before communicating with the coordinating nurse

(32)

4.3.1 Factors contributing to prolonged duration of the second interval

Table 4.5 below summarizes the frequencies factors occurred within a patient’s pathway as is discussed in the previous section.

Table 4.5 - Factors occurring within second interval

Supervisor additional patient check 5 6

Supplementary results 7 20

Patient requires prescriptions 9 5

Bed/ward availability - 12

Administration 12 12

To identify if a factor contributed to a prolonged duration of the disposition process, table 4.6 first provides information about the mean, standard deviation and median duration in minutes of the second interval.

Table 4.6 - Mean, standard deviation and median duration in minutes of second interval

Mean duration 23,0 53,7

Standard deviation 27,5 50,6

Median 16 41

(33)

Figure 6 below shows a distribution curve indicating data is close to normally distributed around the mean. An histogram dividing all values into series of intervals and the frequency of how many values fall in each interval can be found in Appendix C.

Figure 6 Nominal distribution curve interval 2

According to table 4.6, if the duration of the second interval for a discharged case is >16 minutes and for an admitted case >41 minutes, this case experiences a prolonged duration of the second interval. Table 4.7 below examines the cases in which a factor resulted in a prolonged duration of the disposition process.

0 0,002 0,004 0,006 0,008 0,01 0,012 0,014 0,016 0 50 100 150 200 250 Pro b a b il ity d en sity Duration (minutes)

Distribution curve: Interval 2

(34)

Table 4.7 - Significance table factors causing delay using Chi-Square test

Comparing percentages of cases where duration of the second interval is above the median to cases with duration below the median, it can be identified additional patient check, waiting for supplementary results and patients requiring prescriptions are all probable to contribute to a longer duration of the disposition process for both discharged and admitted patients.

Factor Prolonged duration? p-value

No Yes

Discharge Admission Discharge Admission Discharge Admission

(35)

Search for an available bed/ward only resulted in 50% of the cases to a prolonged duration of the disposition process. Therefore, the search for an available bed does not necessarily cause a prolonged duration of the second interval. The same is measured for finishing administration before communication in admitted cases, where this also only resulted in a prolonged duration in 50% of the cases. However, finishing administration before communication resulted in a prolonged duration in 66,7% of the discharged cases, indicating higher probabilities for a delay to occur in case a patient is being discharged.

Waiting for supplementary results to be available is most probable to result in a delay of the second interval for admitted patients (p<0.05). Other factors did not appear to have a significant effect on the mean duration of the disposition process. Therefore, reducing the duration required for laboratory or radiology to be available is probably the most effective method of reducing the total duration of the second interval for admitted patients, which is approximately twice as long as discharged cases.

4.4 Correlation analysis

To identify whether some factors are interrelated, a correlation matrix is developed examining the interaction between all factors previously mentioned. Hereby, correlation is measured for the probability that two factors occur together in a single case. Therefore, two codes are assigned:

 0 codes: if factor did not exist for the delayed patient  1 codes: if factor did exist for the delayed patient

(36)

4.4.1 Discharge

Table 4.8 below recognizes the probability of two factors occurring together for discharged cases.

Table 4.8 - Correlation matrix: Discharge factors

C ontact supe rv isor Co -assi stant C onsul t spe cial ist Re sult s av ail able Addit ional pati ent che ck Supp. Re sult s Pre sc ript ion Adm ini str ati on Contact supervisor 1 Co-assistant -0,322 1 Consult specialist 0,234 -0,141 1 Results available 0,037 0,382* -0,213 1 Additional patient check 0,127 -0,162 -0,190 -0,243 1 Supp. Results -0,207 -0,200 0,236 -0,101 -0,269 1 Prescription 0,279 -0,238 -0,063 0,199 -0,121 -0,221 1 Administration 0,207 -0,300 -0,147 0,075 0,349 -0,1667 0,640*** 1

Significance of p-value: *p<0.05, **p<0.01 and ***p<0.001

According to table 4.8, two positive significant correlations exist:

1. A positive correlation exists between results not seen when available and involvement of co-assistant (r=0.382)

(37)

4.4.2 Admission

Table 4.9 below recognizes the probability of two factors occurring together for admitted cases.

Table 4.9 - Correlation matrix: Admission factors

C ontact supe rv isor C onsul t spe cial ist Re sult s av ail able Addit ional pati ent che ck Su p p . re sult s Pre sc ript ion Be d/w ard av ail abil it y Adm ini str ati on Contact supervisor 1 Consult specialist 0,011 1 Results available 0,429** -0,168 1 Additional patient check 0,008 -0,078 0,138 1 Supp. Results -0,288 -0,615*** -0,322* -0,036 1 Prescription 0,064 -0,039 0,172 -0,174 0,047 1 Bed/ward availability 0,013 0,280 0,005 0,322* -0,404* -0,274 1 Administration -0,233 -0,124 0,005 0,165 0,175 0,233 0,013 1

Significance of p-value: *p<0.05, **p<0.01 and ***p<0.001

According to table 4.9, two positive significant correlations exist:

1. A positive correlation exists between results not seen when available and contact supervisor after all diagnostic results (r=0.429)

(38)

Additionally, three negative significant correlations exist:

1. A negative correlation exists between waiting for supplementary results and consult from another specialist required (r=-0.615)

2. A negative correlation exists between waiting for supplementary results and result not seen when available (r=-0.322)

3. A negative correlation exists between search for an available bed/ward and waiting for supplementary results (r=-0.404)

4.4.3 Analysis correlation coefficients discharge versus admission

According to the results, which factors are interrelated depends on whether a patient is being discharged or admitted. According to the absolute value of r as identified by Evans (1996), most relationships are relatively weak or moderate and therefore considered to be too unreliable for valid interpretation. However, two strong linear relationships exist.

First, a positive significant relationship exists for discharged cases between finishing administration before communications and patient requiring prescriptions. An underlying reason for this could be that physicians finish administration first because the patient requires medication.

Second, a negative significant relationship exists for admitted cases between waiting for supplementary results and consult from another specialist required. Therefore, it is probable in case physician is waiting for supplementary results to be available, no additional consult from another specialist is required, and vice versa. This positively contributes to timely disposition decision-making, since it prevents multiple factors to occur within a patient’s pathway.

4.5 Moderating effects

Lastly, to identify how the relationship between operational processes and timely disposition decision-making is influenced by varying contextual factors, the moderating effect of physician experience, medical specialty and level of crowding is identified.

4.5.1 Physician experience

(39)

to table 4.9 in section 4.4.2, the search for an available bed is correlated with supervisor requiring an additional patient check. Although this relationship is considered weak (r=0.322), it is significant and therefore one can assume there is some evidence to believe that this relationship exists. Therefore, an underlying reason for junior physicians to more likely experience a delay due to the search for an available bed could rely on the fact that additional supervisor check is required before the search for a bed can be started. Intermediate and senior physicians do not encounter this required supervisor check, and are therefore less likely to encounter delay. Furthermore, a negative significant relationship (r=-0.404) exists between search for an available bed and waiting for supplementary results, indicating factors are not interrelated. This may indicate junior physicians more often wait for supplementary results to be available before a bed is searched. Contrary, intermediate and senior physicians more often start the search for a bed already before all supplementary results are available.

4.5.2 Medical specialty

According to the interviews, in 13 out of 65 cases an additional consult from another specialist was required for disposition. As can be identified from figure 7 below, primarily patients qualified with specialty Emergency Medicine required additional consults before readiness for disposition was recognized (9 out of 13).

Figure 7 Additional consult required between different medical specialties

(40)

consult required and specialty Emergency Medicine on timely disposition decision-making (p<0.05).

The average duration of the search for an available bed variates between different medical specialties. According to table 4.7 in section 4.3.1, in 60% of all cases the search for an available bed contributed to a prolonged duration of the disposition process, the patient was qualified with specialty Emergency Medicine. Therefore, one can assume the search for a bed results more often in a prolonged duration for Emergency Medicine patients compared to other specialties. However, an underlying reason for this could be the fact that only specialism Emergency Medicine does not possess its own ward. Therefore, it is more complicated to search for a bed, probably resulting in the physician needing additional time for the search of a bed, compared to other specialties.

4.5.3 Level of crowding

The search for an available bed mostly resulted in a prolonged disposition duration in case it was crowded at the ED, compared to cases when not crowded. The level of crowding showed a significant effect (p<0.01) on the search for an available bed and timely disposition making. As such, if crowded, there is a higher probability disposition decision-making cannot occur in a timely fashion

However, when crowded, physicians experience a higher urgency to communicate the diagnosis with the patient or coordinating nurse timely. 80% of the cases physicians communicate first with the (coordinating) nurse or patient before finishing administration it was crowded at the ED. Therefore, besides a negative influence on timely disposition, crowded EDs also positively influence the ability for timely disposition decision-making.

4.5.4 Analysis of moderating effects

(41)

4.6 Limitations

A relatively small sample size is studied with information restricted applying only to academic hospitals located in the Netherlands. Therefore, findings of this study cannot be generalized across larger cases. However, the reason for studying a relatively small group of patients is because it is time consuming to record all steps in the disposition process of a patient, due to the lack of detailed insights of the hospital’s electronic tracking system.

Furthermore, the period used to collect data from interviews covered only four weeks and only occurred during day-shifts. However, seasonal influences and day of the week may influence patient population in the ED. Besides, this study is partly based on time-series data collected by the hospital’s electronic information system Epic. Therefore, accurateness of time series registration is depending on ED personnel and may not represent actual data.

(42)

5. CONCLUSION

In this multiple embedded case-study, the disposition decision-making process in UMCG is analysed. Before efficiency of the disposition process can be improved, factors that influence the ability for timely disposition decision-making should be identified. These factors may be process- or context-related and contribute to a prolonged duration of the disposition process. Therefore, the question to be answered in this case study is as follows:

“What are the main factors contributing to a delay in the disposition decision-making process?”

The disposition process is divided into two consecutive decision-making stages, namely recognition of patient’s readiness for disposition and the disposition decision. According to interviews with physicians, different factors emerged that cause a prolonged duration of the disposition process resulting in a higher potential of delay to occur in decision-making. Since difference of the mean duration of the disposition process between discharged and admitted patients is significant (p<0.05), this study analyses both patient’s pathway separately.

In the first stage, for both discharged and admitted patients, three factors might contribute to late recognition by a physician regarding patient’s readiness for disposition. These factors are contact with supervisor after all diagnostic input, consult from another specialist required and diagnostic results seen later than available.

In the second stage, for both discharged and admitted patients, another three factors might contribute to late establishment of the disposition decision. These factors are supervisor requiring additional patient check, waiting for supplementary results and patient requiring prescriptions. Additionally, for discharged patients, finishing administration before communication might also result in late establishment of the disposition decision.

(43)

available. For discharged patients, the focus should be on reducing the need or total duration of the consults that are required for disposition.

Correlation analysis indicated interrelation between factors depends on whether a patient is being discharged or admitted. Therefore it is important for physicians to recognize differences exists in the pathways of discharged and admitted patients. This does not only allow to predict a patient’s pathway throughout the disposition process, but also allows to identify what factors might result in a prolonged duration of the disposition process. Recognizing these factors beforehand may increase awareness of the physicians, allowing to prevent the factor to result in a delay while supporting timely disposition-making.

Lastly, contextual factors, i.e. physician experience, medical specialty and level of crowding, significantly moderate the relationship between the search for a bed and timely disposition decision-making.

In case junior physicians were involved in the search for a bed for the admitted patient, it was more likely the studied case resulted in a prolonged duration of the disposition process. An underlying reason for this could be that junior doctors might have to wait for additional consult from the supervisor or because they are too uncertain about their disposition decision due to a lack of experience. Therefore, to allow for timely disposition decision-making, UMCG might focus on educating junior physicians the most efficient methods of conducting the search for a bed in a timely manner. Experience positively influences the search for a bed and therefore allows for more timely disposition decision-making.

Mainly patients qualified for specialty Emergency Medicine encounter longer durations in the search for an available bed compared to other medical specialties. However, specialty Emergency Medicine does not have its own ward, resulting in the fact that the search for a bed is more complicated. Therefore, UMCG might focus on increasing the ease of searching a bed for Emergency Medicine qualified patients. Furthermore, additional consults are more often required for Emergency Medicine patients, contributing to a higher potential duration of the disposition process. Therefore, physicians should take into account the importance of requesting additional consults in a timely manner, since this may support to prevent the disposition process resulting in a delay.

(44)

duration of the disposition process is observed. However, when it is crowded, physicians also feel higher urgency to communicate with the coordinating nurse before finishing patient administration. Crowding might therefore also positively affect timely disposition decision-making since early communication allows the coordinating nurse to search a bed early in the process. Although this indicates physicians currently feel “less pressure” to communicate with the coordinating nurse when it is not crowded. Therefore, to improve efficiency of the disposition process, physicians may be stimulated to communicate in a timely manner despite the ED’s circumstances, i.e. also encourage timely communication when it is not crowded.

The results of this case study can be used by the University Medical Centre Groningen as a source of inspiration to improve understanding of factors impeding timely disposition decision-making. It provides understanding of the disposition process for both discharged and admitted patients. Greater understanding of the disposition process supports the hospital in optimization of the disposition process by increasing efficiency of the processes currently causing high potential for delay in decision-making to occur. An efficient disposition process decreases crowding in the ED, while contributing to a shorter patient LOS.

5.1 Future research

Future research could test for causality between factors that contribute to a prolonged duration of the disposition process. This enables physicians to timely recognize all processes that are relevant within a patient’s pathway, allowing to predict a patient’s pathway early in the process. Furthermore, the results found within this study could be extended by studying hospitals with other sizes. It would be interesting to learn if other factors influence the ability for timely disposition decision-making in different contexts. Finally, findings of this study cannot be generalized across larger populations due to the relative small sample size. Therefore, future studies could study to what extent one of the identified factors contributes towards crowding across a larger sample case.

(45)

REFERENCES

American College of Emergency Physicians. (2006). Crowding, Annals of Emergency Medicine, 47(6), 585

Asplin, B. R., Magid, D. J., Rhodes, K. V., Solberg, L. I., Lurie, N. & Camargo, C. J. A. (2003). A conceptual model of emergency department crowding, Annals of Emergency Medicine, 42(2), 173-180

Bobay, K. L, Jerofke, T. A., Weiss, M. E. & Yakusheva, O. (2010). Age-related differences in perception of quality of discharge teaching and readiness for hospital discharge, Geriatric Nursing, 31(3), p. 178-187

Bovier, P. A., Perneger, T. V. 2007. Stress from uncertainty from graduation to retirement: a population-based study of Swiss physicians, Journal of General Internal Medicine, 22(5), 632-638

Calder, L. A., Forster, A. J., Stiel, J. G., Carr, L., K., Perry, J. J., Vaillancourt, C. & Brehaut, J. (2012). Mapping Out the Emergency Department Disposition Decision for High-Acuity Patients, Annals of Emergency Medicine, 60(5), 567-576

Cameron, P. A., Jospeh, A. P. & McCarthy, S. M. (2009). Access block can be managed, Medical Journal of Australia, 190(7), 364-368

Campell, I. (2007). Chi-squared and Fisher-Irwin tests of two-by-two tables witih small sample recommendations, Statistics in Medicine, 26(19), 3661-3675

Cassell, C. & Symon, G. (2004), Essential guide to qualitative methods in organizational research (pp. 11-22). London: Sage.

(46)

Cook, C. (2009). Is Clinical Gestalt Good Enough?, Journal of Manual and Manipulative Therapy, 17(1), 6-7

Derlet, R. W. & Richards, J. R. (2000). Overcrowding in the nation’s emergency departments: complex causes and disturbing effects, Annals of emergency medicine, 35(1), 63-68

Derlet, R., Richards, J. & Kravitz, R. (2009). Frequent overcrowding in US emergency departments, Academic Emergency Medicine, 8(1), 151-155

Di Somma, S., Paladino, L., Vaughan, L., Lalle, I., Magrini, L. & Magnanti, M. (2015). Overcrowding in emergency department: an international issue, International Journal of Emergency Medicine, 10(2), 171-175

Ebrahimian, A., Hashemi-Amrei, S. H. & Monesan, M. (2018). Exploring Factors Affecting the Emergency Specialists’ Decision-Making in Case of Emergencies in Patients, Critical Care Research and Practice, 1-7

Ebrahimian, A., Seyedin, H., Jamshidi-Orak, R. & Masoumi, G. (2014). Exploring Factors Affecting Emergency Medical Services Staff’s Decision about Transporting Medical Patients to Medical Facilities, Emergency Medicine International, 1-6

Eckstein, M. & Chan, L. S. (2004). The effect of emergency department crowding on paramedic ambulance availability, Annals of emergency medicine, 43(1), 100-105

Eisenhardt, K. M. (1989). Building theory from case study research, Academy of Management Review, 14(4), 535-556

El-Eid, G. R., Kaddoum, R., Tamin, H. & Hitti, E. A. (2015). Improving Hospital Discharge Time: A successful Implementation of Six Sigma Methodology, Medicine, 94(12), 1-8

Taking the Right Decision at the Right Time: A Case Study to Investigate Timely Disposition Impeding Factors at the Emergency Department