MSC. THESIS BA OPERATIONS & SUPPLY CHAINS
MAPPING FAILURE COSTS IN CARE PATHWAYS If you can measure it, you can manage it
PAUL SCHÜREN
University of Groningen, Faculty of Economics & Business MSc. Business Administration
Specialization: Operations & Supply Chains
Supervisor: prof. dr. ir. C.T.B. Ahaus Second supervisor: dr. J. Riezebos
Paul Schüren Gelkingestraat 34a 9711ND Groningen 06-52207148 s1613146
paulschuren@gmail.com nl.linkedin.com/in/paulschuren/
ABSTRACT
This thesis has the aim of validating failure costs and providing a suggestion on how to map them in a failure costs matrix. This is done by pinpointing and drawing up an inventory of the failure costs found in literature. The variables from literature are validated through a Delphi study in which 26 experts from the Netherlands participated (22 in the final round). Experts recognized 2 failure costs to a great/ very great extent and 10 failure costs were recognized to some extent. The recognized failure costs are presented in a failure costs matrix which is constructed from 1) a categorization of failure costs on overuse, underuse and misuse and 2) a categorization on clinical, service, team, process or financial domain. The final product of this thesis, the failure costs matrix, functions as a model for more guidance, consistency, coherence, feedback, commitment and consciousness with regards to failure costs in hospital care pathways. Future research should continue the search for failure costs occurring in one specific hospital or failure costs that occur throughout an entire care pathway, both intramural and extramural. Future research can learn from the Delphi design choices made in this thesis.
Keywords: healthcare, care pathway, failure costs, quality, failure costs matrix, explorative, Delphi study
PREFACE
In choosing the subject for my thesis I aimed at studying a new, surprising, meaningful and intriguing healthcare related topic. That topic became one focused on the increasing costs of healthcare; it was immediately clear that it would fulfil the aims which I set up front. Many interesting articles on cost related issues helped me in learning about a lot of factors involved in care provision. Answering the research question became an important quest while writing my thesis the past months and surprisingly even created attention from practitioners and enthusiasts throughout the Netherlands. I am pleased with my results in writing this thesis which can aid future research on using Delphi studies and hospitals in mapping and measuring failure costs occurrence within their work field.
I would like to thank the people who provided me with valuable information from practical applications. These experts offered me deeper insights and have each taught me interesting things with regard to organizing health care provision. Unfortunately, the scope of this thesis did not grant me the possibility to include all your valuable information in this research, but it was valuable none the less. Special thanks go out to my supervisor prof. dr. ir. Kees Ahaus for his cooperative feedback, open-mindedness towards the relatively new topic of mapping failure costs and the time he made available for me. I would also like to show appreciation towards my second supervisor dr. Jan Riezebos for his important role in my graduation. Finally, I would like to thank my friends, both here and oversees, for their interest in how my thesis was progressing, my brother and parents for their support and advice on key moments and my girlfriend Janine for her unlimited support, her help, motivation and full understanding of my lack of time for her. Without a single exception, all the people that I mentioned above have helped me gain insight on all sorts of things. These insights provided me a plethora of experiences and knowledge which I will definitely use in my current job as staff advisor at the oncology department at the University Medical Center Groningen.
“The pain of the transition state – the disruption of institutions, forms, habits, beliefs, and income streams in the status quo - is what denies us, so far, the enormous gains that integrated care could offer.”
Berwick, Nolan & Whittington, 2008:768
SUMMARY
Cooperation between care providers of different types of care is a requisite to handle the complexity of the demand for care. An approach towards this cooperation is known as care pathways. It is not so much the care pathway itself, but the economic impacts in the use of the clinical care pathways that requires attention. This economic aspect is known as failure costs, which is a specific category within the topic of costs of quality (CoQ). A well-defined and executed care pathway has two important consequences, namely decreased variable costs and increased overall quality. For this thesis it is assumed that the more failure costs are occurring within a care pathway, the smaller the effect on the decrease of variable costs is and the smaller the effect on the increase of overall quality is. At some tipping point, at a certain number of failure costs occurrences, variable costs may even increase and overall quality may decrease.
Therefore, the emphasis of care pathway improvement must lie on reducing the number of failure costs that occur within the care pathway. Our research question will be as follows: which failure costs occur in Dutch hospital care pathways and how can they be mapped? Our research objective is to provide a tool to map failure costs caused by non-quality in clinical care pathways. In supporting this objective, this thesis will construct a failure costs matrix in which failure costs can be filled mapped and an overview where errors occurs within the hospital is visualized.
Burton (2013) promotes the development of such an analytical system to amongst others
“identify, track, and measure the effectiveness of care for specific patient populations, types of care, and care units” (Burton, 2013:96). The research question is answered using an extensive search for literature; evidence that is provided in the found articles will be checked for its relevance and level of significance via a Delphi study. A Delphi aims at reaching consensus on important issues with the help of an expert panel (Beech, 1999, Clibbens, Walters and Baird, 2012) through a series of structured questionnaires (referred to as rounds; Hasson, Keeney &
McKenna, 2000). Finally suggestions are provided of how to map and operationalize the failure costs in healthcare in relation to clinical care pathways. This is done by providing a failure costs matrix as a product of this thesis. A failure costs matrix could in turn satisfy the need of an appropriate resource for desired improvement of the pathway process (Vanhaecht et al., 2007).
Of the 38 failure costs variables that we have started with and the 34 failure costs that are added by the experts during round 1 and 2, eventually after three Delphi rounds, only two failure costs are appreciated with a score of >79% by the Delphi group. In the end 10 failure costs attained a
score between a minimum of 50% and maximum of 79%. These 12 variables are displayed in the failure costs matrix on page 29 in figure 6. In this thesis a reflection is also presented on how to apply a Delphi study for such research. In addition a concise comparison to the RAND/UCLA method is discussed.
TABLE OF CONTENTS
Abstract 2
Preface 3
Summary 4
Table of Contents 6
1. Introduction 7
2. Problem statement 8
2.1 The Problem of Rising Costs 9
2.2 The Current Situation of Failure costs 11
3. Conceptual Model 12
4. Methodology 13
4.1 Consensus Seeking 14
4.2 Selecting Experts 15
4.3 Duration of the Delphi 16
4.4 Delphi Rounds and Satisfactory Level of Concordance 16
4.5 Design Quality 17
5. Literature Review 18
5.1 Care Pathways 18
5.2 Quality 20
5.3 Failure Costs and its Classification 21
5.3.1 Failure costs from literature 23
5.4 The Leuven Compass 25
5.5 The failure costs matrix 26
6. Results 27
6.1 The Included Variables (>79%) 29
6.2 Variables with a High Degree of Recognition (50%-79%) 29
7. Discussion 30
7.1 Practical Implications 32
7.2 Remarks on Low Recognition by Delphi Experts 33
7.2.1 Delphi vs. RAND/UCLA 33
7.2.2 Other design choices explained 35
8. Conclusion 37
9. Limitations and Future Research 38
10. References 39
11. Appendix 49
11.1 Start List Failure Costs 49
11.2 Results Round 1 52
11.3 Results Round 2 53
11.4 Results Round 3 54
1. INTRODUCTION
In many European and Western countries the composition of the population is changing. An ageing society is a large challenge and can be explained by the persistently low fertility rates, increasing life expectancy, and a baby-boom generation that will soon start to reach retirement age (Eurostat, 2010). Because of this trend, the demand for healthcare becomes much more complex. Additionally, nowadays approximately one third of the population has a chronic disease (Eurostat, 2009). The increasing number of people with chronic diseases contributes to the growing complexity of health care delivery. Cooperation between care providers of different types of care is a requisite to handle the complexity of the demand for care. An approach towards this cooperation is known as care pathways. A care pathway is identified as a multifaceted intervention for the decision making and organization of care practices for a defined group of patients during a defined period (Vanhaecht, De Witte & Sermeus, 2007). The aim of a care pathway is to improve the quality of care “across the continuum (…) and [optimize] the use of resources” (Vanhaecht et al., 2007, p.138). In this way, patient care can be improved and the necessary inter-professional decision making can be standardized (Pettie, Dow, Sandilands, Thanacoody & Bateman, 2012). There is evidence that properly developed and implemented care pathways can both improve quality and safety of healthcare (Vanhaecht, Øvretveit, Elliott, Sermeus, Ellershaw, Panella, 2012). It is not so much the care pathway itself, but the economic impacts of the use of clinical care pathways that requires attention. The economic impact of care pathway projects is not really known so far according to Panella and Vanhaecht (2011) and Øvretveit and Tolf (2009) note that “the limitations of the evidence are a challenge to the research community to direct more attention to this area of high public concern” (Øvretveit &
Tolf, 2009:8). Therefore, this thesis will specifically focus on the economic aspect which plays a vital role in assessing the success of using care pathways. This economic aspect is known as failure costs, which is a specific category within the topic of costs of quality (CoQ). The occurrence of failure costs in care pathways is presented in this thesis by the development of a failure costs matrix (inspired by ‘the stakeholder costing matrix’ by Øvretveit (2009)) in which failure costs can be categorized and positioned under a hospital-specific domain. The aim is to fill a gap in literature; many articles pay attention to the wastes (failure costs are wastes) in healthcare (Berwick & Hackbarth, 2012; Berwick, Nolan & Whittington, 2008; Delaune &
Everett, 2008; Midwest Business Group on Health, 2003; Øvretveit, 2009; Øvretveit & Tolf,
2009; Smits et al, 2010), but only one explicitly tries to excavate on how to map them in order to make failure costs measurable and manageable; only Wallace and Savitz (2008) estimated waste in frontline health care worker activities. The aim of the failure costs matrix is thus to use it as an analysis tool of a pathway’s current situation regarding non-quality. Herewith, it fulfils part of the required system components for hospital waste reduction as proposed by Burton (2013). Burton (2013) described a framework consisting of system components that should drive out waste and enhance value in health care. He urges for the development of an analytical system to amongst others “identify, track, and measure the effectiveness of care for specific patient populations, types of care, and care units” (Burton, 2013:96). The failure costs matrix developed in this thesis gives professionals, practitioners and researchers the possibility to find out where (hidden factory) failure costs occur and to which hospital domain these costs are related. In practice, this matrix does not only make the costs of failure costs visible, and therefore measurable and manageable, but it also gives rise to more internal dialogue on failure costs among the employees of a hospital. As a consequence the intention is that the failure cost model stimulates employee’s assertiveness, situational awareness, creativity, innovation and teamwork, which aids in tackling the specific failure costs that occur in a care pathway. Finally, the model aids in making choices for correcting the failure costs in the near future and prioritizing them.
2. PROBLEM STATEMENT
We start by exploring the actual problem that has to be investigated. The problem exploration phase is performed by studying literature, which will be elaborated on in chapter 3. Before this literature study is discussed, a swift look is taken at an editorial that assists us in asserting the problem statement. This editorial is written by Panella and Vanhaecht (2011) and deals with clinical care pathways and the failure costs concept. According to Panella and Vanhaecht (2011) the failure costs is the expense of doing things wrong. Panella and Vanhaecht (2011) state that
“from an economic perspective, we could define the optimal level of quality at the point where the prevention-appraisal costs line and the internal-external failure costs line cross each other”
(Panella & Vanhaecht, 2011:2). This is called the Economic Conformance Level (ECL), displayed in figure 1.
FIGURE 1 - Model for Optimum Quality Costs - Emerging Processes (Juran and Gryna eds. 1988 4.19), adapted from:
Panella & Vanhaecht, 2011
Note that there are also other perspectives one can take towards attaining quality of products or services, for example perspectives by Crosby (1984) (the concept of Zero Defects). Despite this, fact is that decreasing the occurrence of poor quality in care pathways reduces failure costs, but may increase the use of resources required in prevention and appraisal activities. Now that the ECL is graphically explained, it seems advantageous for teams within a clinical pathway to search for their own optimum ECL. However, there is some criticism on this model. First of all, the definition of "conformance" can be questioned. Secondly, the model lacks precision, in that there are no metrics of the costs and conformance level included. These two shortcomings of the model makes that it will only be used for this thesis as a guidance for displaying the effects of failure costs on the efficient use of clinical care pathways. Panella & Vanhaecht (2011) argue that the care process should be re-designed through the usage of clinical care pathways, so that the focus will be put on the failure costs; which is the main topic of this thesis.
2.1 The Problem of Rising Costs
This thesis focuses on the healthcare sector in the Netherlands. For further understanding the current situation in the Netherlands, we will sketch the current state of the Dutch healthcare industry. The healthcare sector employs 1.3 million employees and has a turnover of almost 80 billion Euros, which makes it an important economical factor on the macroeconomic level (Wollersheim, 2011). An important purpose for writing this thesis has to do with the increasing costs of healthcare provision in the Netherlands. Expenditures in the healthcare industries have increased significantly over the past decades. In the period from 1972 to 2008 the average inflation corrected increases on healthcare spending were yearly 2.9%; from 6.5 billion euro per
expenditures have increased from 8.7% of total GDP in 1972 to 13.3% in 2008 (CBS, 2009;
Westert, Berg, Zwakhals, Heijink, Jong & Verklij, 2010). In addition yearly healthcare expenditures have risen 6% to 7% between 2007 and 2010, compared to relatively constant expenditures growth rates of 3% to 4% in the years before 2007 (Wollersheim, 2011; Westert et al., 2010). All in all, the CBS (2009) concludes that the price trend will hold and that the increases in healthcare expenditures are generally higher than the increases in GDP. Unhealthy behaviour currently is one of the reasons behind the increase in healthcare expenditures. For example, in the year 2003 over 2 billion Euros was expended on care related to the effects of smoking. These effects include cardiovascular disease, stroke, lung cancer and chronic respiratory conditions (COPD). In addition, figures from 2003 show that “healthcare costs related to the effects of overweight amounted to nearly 1.2 billion euros. Here too, cardiovascular disease is out in front, closely followed by diabetes and musculoskeletal conditions. In addition, no less than 2 billion euros of expenditure was related to high blood pressure and the associated diseases and conditions” (Van Baal, Heijink, Hoogenveen, Polder; 2006:11). When we look at smoking and overweight as a factor for increasing costs, we can see that smoking amounts for 3.7% of total care costs and effects from overweight accumulates to 2% of total healthcare expenditures (Van Baal et al., 2006). These amounts are in contrast “to the percentage costs of major disease groups such as psychological disorders (22%), cardiovascular disease (no less than 10%), and musculoskeletal disease (6.8%)” (Van Baal et al., 2006:11).
In addition to the unhealthy behaviour of patients, it seems evident that ageing takes a heavy toll on the healthcare system. However, Wollersheim (2011) found that if one corrects for this phenomenon, health care expenditures are still increasing with approximately 5% yearly (Wollersheim, 2011). Both unhealthy behaviour and the ageing society are important drivers in the increase of healthcare costs and triggered an increase in the attention of the economic impact of healthcare costs. In the Dutch hospital care a substantial increase of volume is noticeable; the number of clinical hospital admissions increased yearly with 3% on average and the number of day-care patients increased with an average of 10% (Westert et al., 2010).
The abovementioned cost distributions in the Netherlands are rather well documented and known. What is however unknown is how much of the total expenditures on healthcare in the Netherlands is related to costs spent as a result of failures. Therefore this research focuses on how we can control and reduce costs that are put on hospital healthcare by investigating the topic of
failure costs. Focusing on this aspect is more worthwhile than focusing on the increased autonomous growth of hospital (re)admissions since failure costs might in some cases be the cause of this increase in admission. Furthermore, “available evidence suggests that savings are possible through (…) skilful implementation using quality methods (…)” (Øvretveit, 2009:ixi).
2.2 The Current Situation of Failure costs
The emphasis of failure costs in this research lies on Dutch hospitals. Hospitals are a prime example of so called high reliability organizations (HRO’s) (Kemper & Boyle 2009; Lazear &
Gibbs, 2008), also known as high hazard organizations (Gaba, 2000). HRO’s are types of organizations in which the costs of mistakes are very high and in which a small mistake could lead to several nonlinear responses resulting in a possibly catastrophic event (Kemper & Boyle 2009; Lazear & Gibbs, 2008). The idea of high reliability has been picked up in the healthcare industry “because they want to know how to halt the alarming rate of errors and preventable complications and develop systems that are safer for patients” (Kemper and Boyle, 2009:14).
Smits et al. (2010) performed a research in 21 Dutch hospitals and found that 61% of all hospital failures (of various sorts) involved human causes, of which 61% was labelled as preventable. 14% of the hospital failures had organizational causes and technical factors contributed only 4%. Of the 14% organizational causes, 93% was labelled as preventable (Smits et al., 2010). The preventable failure costs are most interesting for this thesis research, although it must be noted that not all failure costs we discuss can necessarily be labelled as preventable.
Smits et al. (2010) denoted that the failures caused by for example organizational factors can be prevented by improving procedures. This thesis builds on this argument and aims at developing a failure costs matrix as a tool to improve procedures. We will use the matrix to sketch a measurement tool of failure costs and show that mapping failure costs can lead to improved management action aimed at reducing the numbers of failure costs. Furthermore, in the discussion section of this thesis we will provide the benefits of the failure costs matrix in further detail and emphasize on what managers can do with the information derived from the matrix.
This research objective is to provide a tool to map failure costs caused by non-quality in clinical care pathways. The corresponding research question will be as follows: which failure costs occur in Dutch hospital care pathways and how can they be mapped? The research question will be answered using a phased setting in which first an extensive search for published articles
on failure costs in hospitals will be performed. Secondly, the evidence that is provided in the found articles will be checked for its relevance and level of significance via a Delphi study performed among experts from various hospitals in the Netherlands. Thirdly, a suggestion is provided of how to map and operationalize the failure costs in healthcare in relation to clinical care pathways. This is done by providing a failure costs matrix as a product of this thesis.
3. CONCEPTUAL MODEL
Below, in figure 2 the conceptual model can be found. Ways of operationalizing the variables included in the conceptual model are explained.
FIGURE 2 – Conceptual model
It is assumed that the implementation of care pathways will lead to a reduction of variable costs and an increase in the overall quality of care provisions. Figure 2 shows the failure costs as a factor that intervenes in the outcomes of the implementation of clinical care pathways. In other words, because of the occurring failure costs, the positive results of the care pathways, namely decreased variable costs and increased overall quality, are not fully met any more. It is assumed that the more failure costs are occurring in a care pathway, the smaller the effect on the decrease of variable costs is and the smaller the effect on the increase of overall quality is. Therefore the emphasis must lie on reducing the number of failure costs that occur within the care pathway. In supporting this mission, this thesis will construct a failure costs matrix in which failure costs can be filled mapped and an overview where errors occurs within the hospital is visualized.
Implementation of clinical care pathways
Amount of variable costs Occurrence of failures
Degree of overall quality
4. METHODOLOGY
The entire research for this master thesis is divided into three distinct steps. First, current literature on the presence failure costs in healthcare is analysed. Secondly, the literature found is critically checked regarding the strength of the evidence provided. The third phase deals with an empirical research performed with the help of care professionals, practitioners and researchers who currently actively use, or focus on, care pathways. The empirical research is performed through a Delphi study. We will further emphasize on this approach later in this thesis. Finally, the last step deals with providing argumentation, derived from literature (step 1 and 2) and empirical research (step 3), to present an instrument (the failure costs matrix) to address and locate the occurrence of failure costs in clinical care pathways.
The main research strategy focuses on a literature study on topics that assists us in answering the research question and fulfilling the research objective. Data gathering was constructed with the help of a search strategy as proposed by Fitch et al. (2001). The literature inquiry was started by setting up restrictions. The search strategy was only allowed articles written in Dutch or English and the abstract had to be available. At first, an in-depth look was taken on the work on care pathways in general, and its relevancy towards failure costs, which was published by the E-P-A (European Pathway Association). From extensive searches the reference list of the E-P-A-found articles, more interesting data was found. In addition, the databases JAMA & Archives, Business Source Premier, Cochrane, PubMed, ABI/Inform Databases, Web of Science and the British Medical Journal were used for searches on the following key words:
“care pathway[s]”, “clinical pathway[s]”, “costs of poor quality”, “failure costs”, “adverse events”, “AE”, “integrated care pathway”, “multidisciplinary care”, “consequence costs” and
“evidence-based care”. Often the results were sorted on “relevancy” to create a suitable list of relevant articles. In order to quickly get to know whether the articles found were really relevant for this paper the abstracts were read. These short texts provided enough information for deciding on whether the article was suitable or not. This literature inquiry revealed relevant literature on the basis of which more literature can be found and part of the fundament of the literature study can be build. For every article found on the key words described above, the reference list was analysed in order to find more useful literature.
For this thesis, a Delphi survey technique (hereafter referred to as Delphi) is undertaken.
This is because we want to extract actually occurring failure costs in Dutch hospitals from a list
of failure costs that were found in literature. The methodology of the Delphi study was constructed via the search terms “Delphi”, “Delphi study” and “Delphi healthcare”. The study by Minkman, Ahaus, Fabbricotti, Nabitz & Huijsman (2009) and Okoli and Pawlowski (2004) were most important for the design and construction of the Delphi study. A Delphi aims at reaching consensus on important issues with the help of an expert panel (Beech, 1999, Clibbens, Walters and Baird, 2012) through a series of structured questionnaires (referred to as rounds) (Hasson, Keeney & McKenna, 2000). Variations of Delphi studies exist and the type of method used for this thesis is slightly different from traditional Delphi studies in which the first phase is aimed at allowing experts to independently contributing to the research question (Thompson, 2009). This Delphi study has the aim of validating a list of failure costs derived from a literature review, researchers’ input and that of other professionals consulted during the Delphi. The key question to the panel experts was the following: to what extent do you recognize the following failure costs in your hospital? All variables were listed with a Likert scale ranging from the following items: 1. not/ hardly, 2. to some extent, 3. to a high extent, 4. to a very high extent. This 4-point Likert rating scale was used to avoid a tendency to score ‘in the middle’ (Minkman et al, 2009), as we are actively exploring the occurrence of failure costs and did not want to end up with failure costs that are neither occurring, nor not occurring. The 4 point Likert scale also forces the panellists to choose a direction.
4.1 Consensus Seeking
Delphi is a type of consensus seeking method. Pill (1971) and Rowe (1991) have constructed a number of features for consensus methods. They are described by Jones & Hunter (1995) and Yousuf (2007) as the following:
Anonymity: to avoid dominance; achieved by use of a questionnaire in Delphi and private ranking in nominal group
Iteration: processes occur in ‘rounds’, allowing individuals to change their opinions
Controlled feedback: showing the distribution of the group’s responses (indicating to each individual their own previous response in Delphi)
Statistical group response: expressing judgment using summary measures of the full group response giving more information than just a consensus statement.
In seeking consensus the Delphi method seeks to overcome some of the disadvantages in regular decision making groups, where coalitions or individuals may dominate the discussion with skewed outcomes as a result (Jones & Hunter, 1995). We have applied three of the four features of a Delphi as described by Jones and Hunter (1995) above. All our experts answered the answers stated in the study online and anonymously. They were only asked to write down their name in the beginning of the thesis for the researcher to ascertain the exact persons who joined the study and to extract the no-shows for the following rounds. Thus, anonymity is accounted for in our study. Secondly, iteration occurs since our study lasted for 3 separate rounds with an interval of approximately 3 to 4 weeks. This interval rate gave our respondents the time to reflect on their answers and change them if necessary. The iteration value was chosen beforehand and thus not influenced by the results of the individual Delphi rounds. Third, controlled feedback was also taken into account during our Delphi. After each round we placed the percentages of the response distribution behind each variable. This provided the experts with valuable information as would occur in normal face-to-face discussions. Fourth, the statistical group response comprises measures of central tendency. It provides the experts to decide on whether to change or to stick to his or her previous answer. These summary measures were provided as a percentage-based tendency based on previous rounds.
4.2 Selecting Experts
Okoli and Pawlowski (2004) note that “[a] Delphi study does not depend on a statistical sample that attempts to be representative of any population. It is a group decision mechanism requiring qualified experts who have deep understanding of the issues” (Okoli & Pawlowski, 2004:20).
Delbecq et al. (1975) suggest dividing all experts into several panels; we did not perform this step as the topic under investigation (failure costs in hospital care) situates at all hierarchical levels inside and outside the hospital (for example consultants). It is therefore chosen not to create panels based on job description of our panellists. Secondly, panels were not created based on geographical location or work place since many individuals from various hospitals and healthcare related organizations were involved. In addition, groups of panels were not created since Delbecq
et al. (1975) note that heterogeneous groups, characterized by panel members with widely varying personalities and substantially different perspectives on a problem, produce a higher proportion of high quality, highly acceptable solutions than homogeneous groups” (Powell, 2002:379). Therefore, we have formulated the following inclusion criteria and corresponding panels: the participant is active in either of the following four fields of expertise: 1) specialist, doctor, physician, physicians undergoing training to become a specialist (Dutch abbreviation:
AIOS) and physicians not undergoing training to become a specialist (Dutch abbreviation:
ANIOS), 2) physician assistant, nurse practitioner, senior nurse, nurse, nutritionist 3) (academic) researchers in the healthcare sector, 4) quality managers within a hospital, healthcare consultants, team managers within a hospital.
4.3 Duration of the Delphi
On forehand we have chosen to construct 3 iterations, since in most cases 3 iterations are enough to reach a level of saturation. This is supported by Okoli and Pawlowski (2004) and Minkman et al (2009). Researchers cannot send out the next questionnaire until all the results for a panel are in (Okoli & Pawlowski, 2004). Setting a deadline for participants to respond is therefore inevitable. Before closing each of the Delphi rounds, we have sent two reminder e-mails as an addition to the first invite e-mail that was sent each round. This prevented the rise of too many non-respondents, but slowed the data collection process down to some degree. Having non- respondents in one round continue to participate in further iterations is improbable (Hsu &
Sandford, 2007). Since our Delphi questionnaire was interrupted by the summary holidays here in the Netherlands, some rounds took little longer than other.
4.4 Delphi Rounds and Satisfactory Level of Concordance
Okoli and Pawlowski (2004) suggest that a Delphi study could be constructed as follows: 1) initial collection of factors, 2) validation of the list of factors, 3) choosing the most important factors (narrowing down) and 4) ranking the chosen factors.
For our Delphi research the initial collection of factors was performed through a literature study, since the aim of our Delphi was to check the list of literature-found factors on real-world occurrence. Our Delphi thus focused on step 2 and 3 as proposed by Okoli and Pawlowski (2004), namely choosing the most important factors. For selection the factors that will be
included in our model we have selected three threshold criteria. The first criterion follows Okoli and Pawlowski’s (2004) and Minkman’s et al. (2009) suggestion of dismissing all factors that are not selected by 50% or more of the experts. The second criterion aims at attaining approval of 80% of all the experts. The reason that this percentage is set relatively high is that the aim was to include variables which are supported by a substantive amount of practitioners (Minkman et al., 2009). 70% was judged by us as a too low percentage, hence 80% is chosen. Naturally, the third threshold consists of recognition percentages between 50% and 79%. These factors were included in the next round and were displayed with the controlled feedback as suggested by Jones and Hunter (1995). This controlled feedback showed the distribution of the group’s responses (indicating to each individual the previous group response in Delphi). This third threshold is chosen to be cautious about eliminating an element of more than 50% recognition so as not to miss a topic (Minkman et al., 2009). There remain those who question the value of Delphi studies since it can be referred to as more opinion-based than evidence-based (Powell, 2003). “However, it is noteworthy that in their seminal paper Sackett et al. (1996) describe the evidence-based movement as integrating best available external evidence from systematic research with individual clinical expertise. The latter is, in turn, described as the proficiency and judgment acquired through clinical experience and clinical practice. Such an interpretation is arguably more compatible with Delphi methodology than may at first appear, as expert opinion would presumably be evidence-based in precisely this way” (Powell, 2003:381). At the end of this thesis, we will reflect on the design choices we made in constructing our Delphi study;
limitations are provided, implementation of findings presented and suggestions for future research discussed.
4.5 Design Quality
In order to build construct validity we have included a relatively large Delphi group consisting of heterogeneous backgrounds, both with regards to job position and geographical location. These multiple sources of evidence are used to study same phenomenon, namely the occurrence of failure costs in Dutch hospitals. In addition the start list of failure costs is documented in the appendix. This provides the reader with further insights what the results are based on and increases construct validity (Yin, 2003). The findings derived from the Delphi study provide for a start towards theoretical fundaments with regards to mapping failure costs in healthcare.
However, external validity of the findings should be strengthened in the future when it is confirmed that future research is valued. The reliability of this research deals with the extent to which another researcher is able to produce the same results when performing similar interviews.
Although it is often heard that results obtained from Delphi (-like) studies may vary depending on the composition of the expert panel selected, our sample size of 26 experts (22 in the final round) reduces this critic on reliability to some extent. In order to increase reliability, the steps included in designing and constructing the Delphi study are documented in the methodology chapter. In addition, the outcomes of the Delphi rounds are documented in the appendix (12.2, 12.3 and 12.4). This makes sure that the reader can track where the results are derived from (Yin, 2003).
5. LITERATURE REVIEW
The literature review has the aim of guiding the reader through a set of concepts which are important, in that they are much related to the final product of this thesis, which is the development of a failure costs matrix within a care pathway. Hence, first of all the area of focus, namely care pathways, is discussed. After a description of its aim, features and criteria, we continue this literature review by focusing on theories of quality of products and services. Failure costs are a type of non-quality; therefore it is important to discuss the concept of quality of products and services in this literature review. The third topic that is investigated in this literature review is failure costs and its classification. Then, we will develop the failure costs matrix by discussing underuse, overuse, misuse, waste and inefficiency (Bailey, 2003; MBGH, 2003;
Orszag, 2008; Øvretveit & Tolf, 2009; Øvretveit, 2009) and the domains of the Leuven Compass (Vanhaecht and Sermeus, 2003). In the end of this chapter the failure costs matrix is presented.
5.1 Care Pathways
Kinsman, Rotter, James, Snow & Willis (2010) identified 84 different terms that refer to a care pathway (Kinsman et al., 2010). In this thesis the term care pathway will be maintained. In defining a care pathway it is needed to distinguish between the aim, criteria, features and prerequisites of a care pathway. All these elements of a care pathway provide for a complete explanation of what a care pathway is and will give us greater insight in exactly how to construct the failure costs matrix. A care pathway can be identified as a multifaceted intervention for the decision making and organization of care practices for a defined group of patients during a
defined period (Vanhaecht, De Witte & Sermeus, 2007). The aim of a care pathway is to improve the quality of care “across the continuum by improving risk-adjusted patient outcomes, promoting patient safety, increasing patient satisfaction, and optimizing the use of resources”
(Vanhaecht et al., 2007, p.138). The focus of the failure costs matrix is in line with the aim of the care pathways. Five criteria for a clinical care pathway were derived by Kinsman et al. (2010) from three articles (De Bleser et al., 2006; Campbell et al., 2006; Vanhaecht et al., 2006).
Describing these criteria can aid health care professionals, consultants and managers to apply measures against the occurrence of failure costs. In order to be defined as a care pathway an intervention must first of all be structured as a multidisciplinary plan of care, including both clinicians and managers. If they do not work together “all parties will continue to be driven by the distrust and related crises of confidence that pervade the field” (Vanhaecht, 2007, p.10).
Secondly, the intervention has to aim at a standardization of care for a specific clinical problem, procedure or episode of healthcare in a specific population. Standardized care helps to extract and map failure costs occurring in care pathways. Third, the intervention has to be used to channel the translation of guidelines or evidence into local structures. This criterion is related to the fact that care pathways bring standardized channels of care provision. These local structures are set up per type of care intervention which might increase the ease of finding failure costs. Fourth, the intervention must include detailed steps of the course of the care, pathway, algorithm, guideline, protocol or other 'inventory of actions'. These careful investigations are important to note because an action to scan for the occurrence of failure costs might very well be included in this criteria.
Finally, the intervention should include timeframes or criteria-based progression, which means that in order to meet designated criteria, pre-designed steps must have been taken. This aids the standardization of the local structures for each intervention (Kinsman et al, 2010). Vanhaecht et al. (2007) adds to Kinsman et al (2010), that the facilitation of the communication among the team members must be controlled for, the failure costs matrix presented in this thesis can stimulate communication about failure costs. Furthermore, variances and outcomes of interventions must be monitored, documented and evaluated. Failure costs form a sort of variances, it is therefore important to take them into account. A failure costs matrix could in turn satisfy the need of an appropriate resource for desired improvement of the pathway process (Vanhaecht et al., 2007).
5.2 Quality
The concept of quality has to be discussed for further understanding on failure cost recognition and mapping. Judging certain processes as failure costs is not reasonable without a proper framework on what quality actually is. It is decided to define quality using Reeves and Bednar (1994), since they are often cited (199 citations) and referred to (102 references) also in recent literature according to the Web of Science Citation Index (retrieved: 29-07-2013). In their article, they aimed at uncovering discrepancies in the usage of the term quality. Reeves and Bednar (1994) concluded that there was no single universal definition of quality. “The quality construct space is so broad and includes so many components that there would be little utility in any model that tried to encompass them all” (Reeves & Bednar, 1994:441). When discussing quality in relation to the topic of failure costs, it is logical to discuss the issue of cost of poor quality too.
However, for the sake of clarity it is chosen to elaborate on cost of poor quality when discussing the topic of failure costs in sub-chapter 6.3. The definition of quality in the field of care interventions is focused on two different definitions of quality, namely meeting expectations and conformance to specifications. One might expect the principle of quality if value here too, however we argue in line with Stahl and Bounds (1991) the meeting expectations definition of quality already incorporates the value of considerations in customers’ expectations. That is, value is a subcomponent of quality. The actual price of value is not included in sticking to the definitions of meeting expectations and conformance to specifications, but for this thesis this factor is not desired immediately as the focus lies on the occurrence of failure costs and not the precise price of the occurrence of failure costs.
The definition ‘quality is meeting expectations’ stems from the services marketing literature (Reeves & Bednar, 1994; Normann, 1984; Shostack, 1977; Zeithaml, 1981). According to these researchers the much older conformance-to-specifications definition of quality did not manage to address the distinctive characteristics of the quality aspects within the service sector.
The strength of this definition is that it holds a strong customer focus. It is argued that only
“customers can articulate how well a product and/or service meets their expectations (…)”
(Reeves & Bednar, 1994:432). The definition of meeting customer expectations is an externally oriented definition of quality. Reasons behind choosing externally focused definitions can be fed through expected changes in the marketplace. These changes inhibit amongst others the costs of healthcare in the Netherlands, which are rising quickly, and the increase in pressure that is being
put on health care provision due to an ageing society with a higher life expectancy (CBS, 2009, Westert et al., 2010, Wollersheim, 2011). With the customer focus of quality managers can react to these changes in the marketplace and enhance its competitive advantage (Reeves & Bednar, 1994). One must take into account that it is found by several researchers that in many cases customers simply do not know what their actual expectations are. This is particularly the case when services are used infrequently, as is the case for most patients visiting hospitals (Cameron
& Whetten, 1983; Lawrence & Reeves, 1993). Additionally, researchers using the meeting expectations-definition of quality must be cautious towards the fact that a service may on the short term be regarded as high in quality and in the long term as low in quality and vice-versa (Curry, 1985).
Quality can also be defined from the perspective of ‘conformance to specifications’. This definition is internally focused and has the advantage of being relatively easy to understand and apply. A hospital can set a number of specifications in advance and subsequently monitor progress in achieving the pre-set objectives on a continuous basis. The specifications must, according to the advocates of a conformance-to-specifications definition (Crosby, Deming, Feigenbaum and Juran), be determined strongly based on the customer’s wants (Reeves &
Bednar, 1994). An important advantage is that objective measurement items can be assessed on different levels within the hospital, across hospitals (benchmarking) and over a sustained period of time (Reeves & Bednar, 1994). Furthermore, “defining quality as conformance to specifications should lead to increased efficiency on the part of the organization” Reeves &
Bednar, 1994, p. 430). These two perspectives on quality create possibilities for recognizing and registering failure costs. Research on failure costs should therefore always include a predefined perspective on quality principles.
5.3 Failure Costs and its Classification
Failure costs can be referred to as costs create through adverse events (Øvretveit, 2009), which does not necessarily mean that failure costs lead to damage. Westert et al. (2010) refers to the solution of combating the rising healthcare expenditures, namely through reducing the number of failure costs, also known as adverse events. This would reduce failure costs, which in turn reduces overall healthcare expenditures. Many recent articles referring to adverse events (AEs) are doing so from the perspective of patient safety (Smits et al 2010; Sharek et al. 2011; Zhu et al
2011; González-Formoso, 2011) and not from the perspective of reducing costs. In addition, research on AEs is mostly about failure to use effective treatments on patients, rather than about quality problems with rather indirect clinical outcomes, “such as poor patient experiences, long waiting times, or wasted personnel time and materials” (Øvretveit, 2009:15). This thesis aims at taking the cost of quality perspective when referring to AEs. In order to make a clear distinction between the patient’s perspective and the cost of quality perspective, the term ‘failure costs’ is used in order to refer to expenditures made as a result of failures that could have been avoided or can be avoided in the future with better measurements and management.
The variable failure costs can be operationalized and categorized using the report written by the Midwest Business Group on Health (MBGH, 2003) and the article by Bailey (2003) that is derived from the MBGH’s report on reducing the costs of poor-quality health care. MBGH (2003) made estimates on the occurrence of failure costs and noted that “even if these figures are off by 50%, poor-quality health care exacts a several-hundred-billion-dollar toll on [the US] each year (MBGH, 2003:ii). Already in 2003, they found that the vast amount of broad statistics on increasing health care costs offered “no insights into the nature of a dilemma that pervades the health care system. To better understand this dilemma, it is useful to classify the different types of quality problems into four categories: overuse, underuse of evidence-based care [and] misuse (…)” (MBGH, 2003:ii). Based on the findings by Bailey (2003), MBGH (2003) and Orszag (2008) each of the failure costs categories is discussed. Although opinions on how to define the categories might differ, for the sake of clarity the following definitions are upheld for this thesis.
The first type of poor quality consists of overuse of resources. This occurs when a service is provided, but when it is not wanted on medical grounds. In such cases, the risk of harm and burden exceeds its likely benefit (Orszag, 2008). This might expose patients to risks of complication and sometimes even death (MBGH, 2003). Examples are unnecessary surgeries, tests, medication, treatments and other procedures (Bailey, 2003).
Secondly, opposite to overuse, in the case of underuse patients do not get provided diagnostic and therapeutic services, medications and procedures even though they would have been medically beneficial (Orszag, 2008; MBGH, 2003). Orszag (2008) found evidence in studies that patients typically only receive half of the advised services. Underuse consists for example of a lack of preventive screenings and services (Bailey, 2003). This leads to “illness, relapses, complications, and other conditions that could have been avoided” (Bailey, 2003:24).
MBGH (2003) also mentions underused services such as “the administration of influenza and pneumococcal vaccines; screening tests for depression, breast cancer, and chlamydia; and follow- up after discharge from behavioral health care” (MBGH, 2003:iii). Underuse, according to MBGH (2003), clearly causes premature death and diminishes quality of life.
Third, misuse occurs when medical errors add additional costs to the health care process.
It includes incorrect diagnoses, medical errors (such as never events) and other sources of avoidable complications (such as infections during hospital stay) (Orszag, 2008). Of all misuse,
“medical errors represent the most common form of misuse within the health care system, with drug misuse representing the most frequent form of error” (MBGH, 2003:iii).
Bailey (2003), MBGH (2003) and Orszag (2008) focus on clinical failure costs in applying the categories underuse, overuse and misuse. There are however also numerous failure costs in the field of uncoordinated care, delays, waiting times and redundant services (MBGH, 2003). Bailey (2003) and MBGH (2003) place these factors under a fourth category, namely waste and inefficiency. This fourth category is however rather disputable since they can also be grouped under one of the three other categories. Uncoordinated care for example can be seen as misuse, as it might lead to medical errors. Delays can result from underuse, as it is a consequence of not providing the patient with service that would have been beneficial. An example of this is a test result that is not used because there is a delay in getting them to the ordering clinician (Øvretveit, 2009). Redundant services can be grouped as overuse, as it is a consequence of services that are not necessarily needed by protocol. Since the indicator waste and inefficiency is rather ambiguous and overlapping with the other types of failure costs, it is decided not to use waste and inefficiency as an operationalization of failure costs in this thesis. The three classification factors overuse, underuse and misuse will therefore form the vertical indicators of the failure costs matrix as sketched in figure 5 on page 25. In applying failure costs to a certain category, for this thesis the definitions by Bailey (2003), MBGH (2003) and Orszag (2008), as outlined above, are maintained. To strengthen this, similar definitions are also maintained by Øvretveit (2009) and Øvretveit & Tolf (2009).
5.3.1 Failure costs from literature. Numerous failure costs are collected by a literature study in various research papers (Dowsey, Kilgour & Santamaria, 1999; Gray and Goldman, 2004; Husbands, 1999; Murff et al, 2003; Panella, 2003; Sloan, Chatterjee, Sloan, Holland, Waters, Ewins & Laundy, 2008; Wachtendonk, 2010; Zegers et al, 2008; Zhu et al, 2011) and
derived from inspiration by Julian (2007), Øvretveit (2009) and Schönsleben (2007). In addition, failure cost variables were derived from various informal discussions with experts and practitioners in the healthcare industry. The failure costs that are appearing in these papers are said to exist in hospitals. Before we apply them directly into our failure costs matrix, the failure costs from literature are checked for appearance in Dutch hospitals via a Delphi study. It is up to the Delphi panel of experts to decide whether the failure costs are recognized sufficiently in order to be included in the failure costs matrix. The entire list of failure costs variables can be found in the appendix. Some examples can be found below. It must be noted that the classification under the various domains are not stringent. This means that some failure costs can fit into more than one domain and there is no one best way for some failure costs as they tend to overlap.
Clinical domain, costs as a consequence of:
- Number of hospital-acquired infections (nosocomial infections) (Dowsey, 1999; Panella, 2003; Zhu, 2011; Zegers, 2008)
- Wrong drug prescription/ nursery (Gray and Goldman, 2004)
- Too early dismissal of patients (requires readmissions) (Zegers, 2008) Service domain, costs as a consequence of:
- Extra time spent on slow amount of time to ambulation by (for specific patients treatments) (Dowsey, 1999)
- Degree of patient dissatisfaction (Øvretveit & Tolf, 2009)
- Patients failing to continue their treatment (e.g. because of their dissatisfaction with the quality of the service) (Øvretveit & Tolf, 2009)
Team domain, costs as a consequence of:
- Communication and coordination failures (Smits, 2010; Wachtendonk, 2010) - Lack of relational coordination (Smits, 2010, Wachtendonk, 2010)
Process domain, costs as a consequence of:
- Too long time between two interventions (interoperation time)
- Wrong ranking of medical interventions (from diagnoses to dismissal of the patient) (Øvretveit & Tolf, 2009)
Financial domain, costs as a consequence of:
- Excessive Length of Stay (LoS) (Sloan et al, 2008; Panella et al, 2003; Husbands, 1999) - High number of recidivism (Dowsey, 1999; Panella et al, 2003; Hanna et al, 1999) - Patients leaving without treatment (lost customers)
5.4 The Leuven Compass
The Leuven Compass (Vanhaecht & Sermeus, 2003) is discussed because it is an important part of our failure costs matrix as it forms the vertical indicators. The Compass consists of five domains, which are the clinical, service, team, process and financial domain. These distinct perspectives help interdisciplinary care teams to set up goals, indicators and performance measures to track the quality and efficiency of their work (Vanhaecht & Sermeus, 2003) and thus track failure costs that occur within the care pathway more precisely.
The first domain deals with clinical and functional indicators. Clinical indicators are directly related to a specific disease and functional indicators relate to the impact of the disease and its treatment on health outcomes. The second domain relates to service indicators. Indicators under this domain measure the degree of service of the medical team for a specific patient population of a care pathway. Service indicators that fall under this domain are amongst others:
patient satisfaction, patient experiences and patient attitudes. The third domain is called the team domain and is more difficult to assess due to a number of reasons. First of all, the team working at a distinct care pathway is in most cases not a well-defined team, but is rather fluid and may be part of a virtual team as well (Vanhaecht & Sermeus, 2003). This makes that team members are not always aware of the fact that they are team members. Additionally, there is no visible team from the patient’s perspective. Despite these difficulties, the team domain is accounted for as a valid domain to use in this thesis. Indicators such as relational coordination, team satisfaction (Team Satisfaction Survey & Spath, 1994) and team effectiveness (Leuven Team Effectiveness Scale, Haspelagh, Vanhaecht, De Witte & Sermeus, Van de Waeter & Serra, 2002) fall under this domain. The fourth domain consists of process indicators. Examples of important process indicators are brought together under the term ‘lead time’. Important aspects here are; the administration time, operation time, interoperation time and transportation time (of medicine, samples and patients). All these types of lead time must be accounted for in every care intervention, because they can turn out to be failure costs if the lead time is unwanted, unnecessary or unknown due to a lack of measurement. In addition, there is an important aspect in sequencing all the interventions appropriately (Schönsleben, 2007). Within the indicators discussed above, variances can occur through the patient condition, the health care worker condition, the hospital condition and the society condition (Vanhaecht & Sermeus, 2003).
Variance tracking is a way of measuring the processes, and aids in reducing the variance to a
minimum. The final domain is aimed at the financial indicators and deals with the monetary costs of healthcare interventions. The main challenge here is to balance between lowering the costs of healthcare, while maintaining or increasing the levels of quality and service. According to Vanhaecht and Sermeus (2003), the development of a Bill-of-Services (BOS) for every clinical pathway separately makes that “every element is given a standard costs based on the principles of activity-based costing or another cost accounting method. The comparison of the BOS before and after the implementation of the pathway will show the efficiency of the pathway on cost reduction” (Vanhaecht and Sermeus, 2003:5).
5.5 The Failure Costs Matrix
The domains from the Leuven Clinical Pathway Compass by Vanhaecht and Sermeus (2003) are each used our failure costs matrix as sketched in figure 5 below. The failure costs matrix consists of a matrix-structure where all domains from the Leuven Compass are situated vertically and the type of failure costs are displayed horizontally. Figure 3 displays the failure costs model. It is constructed from research by Bailey (2003), MBGH (2003), Orszag (2008), Øvretveit (2009) and Øvretveit & Tolf (2009) on underuse, overuse and misuse and the Leuven Clinical Pathway Compass by Vanhaecht & Sermeus (2003).
FIGURE 5 - Failure costs matrix
1. Clinical domain 2. Service domain 3. Team domain 4. Process domain 5. Financial domain
1. Clinical domain 2. Service domain 3. Team domain 4. Process domain 5. Financial domain
1. Clinical domain 2. Service domain 3. Team domain 4. Process domain 5. Financial domain
Underuse Overuse Misuse
Failure costs
6. RESULTS
As earlier discussed, the Delphi study has the aim of validating a list of failure costs derived from a literature review, input from professional’s experiences and that of other professionals who were part of the Delphi study. The key question to the panel experts was the following: to what extent do you recognize the following failure costs in your hospital? All variables were listed with a Likert scale ranging from the following items: 1. not/ hardly, 2. to some extent, 3. to a high extent, 4. to a very high extent. The initial list of failure costs that the Delphi study was started which can be found in the appendix. This chapter describes the results of the Delphi study per round.
TABLE 1 – Origin of Delphi experts (anonymized)
Origin (all from the Netherlands) Number of experts
Hospital A 16
Hospital B 1
Hospital C 1
Hospital D 1
Hospital E 1
Healthcare consultancy (3 companies) 3 Healthcare education/ training company 1 Healthcare research (2 institutes) 2
Total 26
TABLE 2 – Delphi panel results
Response (n = 26) Round 1 Round 2 Round 3
100% 92.3% (n = 24) 84.6% (n = 22)
Elements (numbers) 38 27 17
Included (>79%) 2.6% (1) 0% (0) 5.9% (1)
Unchanged (50%-79%) 21.1% (8) 29.6% (8) 58.8% (10)
Excluded (<50%) 76.3% (29) 70.4% (19) 35.3% (6)
New elements 50% (19) 29.6% (8) 0% (0)
Recognition (percentages)
Not/hardly 18.71% 10.49% 1.65%
To some extent 46.15% 47.53% 35.54%
To a high extent 28.06% 33.02% 62.81% (combined top
scores “to a high extent” and “to a very high extent”)
To a very high extent 7.09% 8.95% lost data*
* Due to closure of the Qualtrics data-account, the data of round 3 cannot be retrieved for the ‘to a very high extent’ recognition measurement. Only the combined top scores were still obtainable.
The first round of the Delphi study was conducted among 26 panel members, of which all experts completely filled in the questionnaire. The total number of participants that rated each variable as recognized to a high or very high extent were summed up, the totals were divided by the sum of all participants; 26. Derived from this calculation, out of the total of 38 initial variables, only one failure cost is recognized sufficiently to be included in our failure costs matrix and 8 failure cost variables were presented in the next Delphi round. All other 29 variables are excluded by the Delphi experts from further inclusion in the failure cost model, since they were not deemed relevant enough. However, 19 failure cost factors were added by the participants individually.
Round 2 consisted of 24 participants, 2 participants failed to fill in the questionnaire before the deadline despite two reminders. The total amount of failure costs with which this round started was 27. Not a single failure costs attained a score of >79%, meaning that no variable was directly added in the included list of failure costs. 8 failure costs were presented in the next Delphi round and the remaining 19 were eliminated. However, 15 variables were added by the experts. Out of these 15, 8 were regarded as sufficiently relevant by the researcher to be included in round three.
Round 3 was continued with 17 failure costs (8 newly introduced, 8 included from round 2). This final round was finalized by 22 experts. Of the total amount of 16 variables, 1 failure costs gained a score of 81.81%, which means that it is included for the end product of this research. 9 variables gained a score of >50% or <80%. The remaining 6 variables were excluded.
6.1 The Included Variables (>79%)
As could be read above, at the end of the study two failure costs are included, as they received a score of >79%. All included variables are ranked. These variables are:
1. Costs as a result of time spent on different forms containing the same information (81.81%, based on 22 experts).
2. Costs as a result of much time spent documenting patient records (80.76%, based on 26 experts).
6.2 Variables with a High Degree of Recognition (50%-79%)
A total of 10 variables did not meet the inclusion criteria of attaining a score of >79%, but were still significantly important for a majority of the experts (50%-79%). These variables are also ranked in order of recognition:
3. Costs as a result of time spent waiting for colleagues at meetings, surgery’s or other procedures (72.7%, only 1 expert did not recognize it at all, mean 2.95 out of 4).
4. Costs as a result of general communication and coordination errors; lack of relational coordination (72.7%, 0 experts did not recognize it at all, mean 2.86 out of 4).
5. Costs as a result of time spent on unnecessary registrations; of whatever kind (63.6%, 0 experts did not recognize it at all, mean 2.82 out of 4).
6. Costs as a result of bad logistics (59.1%, 0 experts did not recognize it at all, mean 2.77 out of 4).
7. Costs as a result of inconsiderate projects (63.6%, 0 experts did not recognize it at all, mean 2.73 out of 4).
8. Costs as a result of insufficient knowledge and overview of the employee in the entire patient-process (50%, 0 experts did not recognize it at all, mean 2.59 out of 4).
9. Forgone earnings as a result of a long time between interventions; inefficient execution of the pathway (63.6%, only 1 expert did not recognize it at all, mean 2.59 out of 4).
10. Foregone earnings as a result of long waiting times; inefficient execution of the pathway (54.5%, 0 experts did not recognize it at all, mean 2.59 out of 4).
11. Costs as a result of inefficient (bad) teamwork (50%, 2 experts did not recognize it at all, mean 2.58 out of 4).
12. Costs as a result of extra time spent to patients due to a lack of good cooperation by and with the patient (50%, 0 experts did not recognize it at all, mean 2.56 out of 4).
Of the 38 failure cost variables that we have started with and the 34 failure costs that are added by the experts during round 1 and 2, eventually after three Delphi rounds, only two failure costs are appreciated with a score of >79% by the Delphi group. In the end 10 failure costs attained a score between a minimum of 50% and maximum of 79%. These 10 variables are not relevant enough to be included according to our prior inclusion criteria. However, they will still be discussed in the discussion part of this chapter, since they cannot be ignored.
7. DISCUSSION
In this chapter, we will elaborate on two things. The failure costs matrix and the reasons behind the reported lack of relevancy according to the Delphi experts of the failure costs. First of all, the failure costs matrix is presented with the failure costs that were found relevant by the Delphi experts included. The two most relevant variables which attained the pre-set threshold of inclusion are presented in bold. The 10 variables which did not meet the >70% threshold, but did grab the attention of 50% or more of the experts are displayed in italics.
