The effects of performance funding in
Dutch hospitals
Master thesis, MSc BA Organizational & Management control University of Groningen, Faculty of Economics & Business
Thesis supervisor: Dr. B. Crom Co-assessor: B. Van der Kolk
Word count excluding references and appendices: 12 108 Word count including references and appendices: 19 055
Ad Boshoven Student number: 152997 b.a.boshoven@student.rug.nl
- 2 -
AbstractIn order for healthcare to remain affordable and quality of care to improve the Dutch government introduced performance funding for Dutch hospitals. Hospitals were no longer funded based upon a predetermined annual budget, but were funded based upon the care (performance) they actually delivered. Performance funding relates well to the principles of New Public Management (NPM), which promotes the idea of running government organizations more like private organizations. So far there is a lack of conclusive evidence regarding the relationship between performance funding and hospital efficiency and quality. This study aimed to provides such evidence by testing the relationship between performance funding on the one hand and efficiency and quality on the other hand. By comparing data prior to the introduction of performance funding and data after the introduction of performance funding two
conclusions were drawn. First the results showed no significant relation between performance funding and hospital efficiency. Second a positive relation was found between performance funding and hospital quality.
- 3 -
ContentsAbstract ... - 2 -
1.Introduction ... - 4 -
1.1 New Public Management... - 4 -
1.2 Problem statement ... - 5 -
1.3 Relevance ... - 5 -
2. Literature ... - 6 -
2.1 New Public Management... - 6 -
2.2 Existing literature on New Public Management in health care... - 8 -
2.3 Performance funding... - 9 -
2.4 Existing literature on performance funding ... - 10 -
2.5 Hypotheses... - 12 -
3.Methodology ... - 13 -
3.1 Sample ... - 13 -
3.2 Efficiency ... - 13 -
3.3Method of analysis for efficiency ... - 17 -
3.4 Quality ... - 18 -
3.5 Control variables ... - 18 -
4.Results ... - 22 -
4.1 Data Envelopment Analysis... - 22 -
4.3 Descriptive statistics... - 23 -
4.4 Regression analysis ... - 25 -
5.Discussion ... - 26 -
6.Conclusion... - 29 -
7.References ... - 31 -
Appendix A - Sample Dutch general hospitals for quality ... - 37 -
Appendix B – Efficiency scores calculated using Data Envelopment Analysis... - 47 -
Appendix C – Sample Dutch general hospitals for quality ... - 49 -
Appendix D – Results efficiency regression analysis performed in SPSS ... - 54 -
Appendix E – Results quality regression analysis performed in SPSS... - 55 -
Appendix F – Data Envelopment Analysis ... - 56 -
Appendix G – Algemeen Dagblad’s ‘ziekenhuis top 100’ Quality indicators... - 57 -
- 4 -
1. Introduction
Worldwide spending on healthcare has been rising steadily for several decades. These expenditures are putting pressure on public budgets, adding to that arising from other social spending programs and servicing of higher debt levels than in the past (De La Maisonneuve and Martins, 2013). Also in The Netherlands health care costs have risen sharply over the past 60 years and these costs are expected to increase even further over the next upcoming decades (Rijksoverheid, 2012). An average Dutch household is already paying nearly a quarter of their income on health care. If health care costs keep increasing at t his rate, this could add up to nearly half of their income according to the CPB1. Given the competing pressures from other social spending programs, these projected trends in public health and long-term care spending are likely to be a major source of concern for most governments (De La Maisonneuve & Martins, 2013). It is evident certain measures have to be taken in order to keep health care affordable both now and in the future.
The Dutch government is trying to contain healthcare costs by, among other things, introducing
performance funding in hospitals. Up to 2012 Dutch hospitals were funded based upon a predetermined annual budget. From 2012 hospitals no longer received an annual budget, but were funded based upon the care (performance) they actually delivered. The Dutch government expects this form of funding will
stimulate hospitals to perform treatments better, for healthcare to remain affordable and quality of care to improve1. This new form of funding was introduced gradually using a transitional arrangement for the first two years. In 2012 and 2013 potential differences in budget were reimbursed for respectively 95% and 70%. In 2014 Dutch hospitals were for the first time fully funded based on actual performance. 1.1 New Public Management
The introduction of performance funding for Dutch hospitals seems to be in line with a global trend, which started in the 1980s called New Public Management (NPM). New Public Management can be regarded as ‘a functionalist approach, in that one of the most important objectives of the changes it proposes is to
increase economic efficiency and effectiveness in public sector organizations’ (Ter Bogt, 2008, p. 210 ). It promotes the idea of running government organizations more like private organizations, with special attention for results-oriented performance, performance measurement, privatization and promoting competition and market forces (Korsten, 2011; Hood, 1991, 1995; De Vries en Van Dam, 1998).
The success of New Public Management however is disputable. The government organizations that initiated reforms probably consider some of the changes to be a success (ter Bogt, 2005, p. 62). In other cases, NPM reforms dismissed after a while, because of new developments or reforms were more difficult to
implement than was expected beforehand (Ter Bogt, 2008). Further Ter Bogt (2008) mentioned several researchers critized the effects of NPM measurements in the Netherlands, since in several cases ambtions were not realized (e.g. van Helden, 1998; Bordewijk and Klaassen, 2000; van Helden and Johnson, 2002; and ter Bogt, 2004). Groot (1999) states theoretically there are many reasons to believe that management instruments used in the private sector will fail in a public sector environment because outputs are difficult to measure, goals are ambivalent and most public sector organizations demand independency in order to perform their professional activities.
1
- 5 -
1.2 Problem statement
Performance funding for Dutch hospitals has been introduced in 2012 and was fully implemented in 2014. While New Public Management has been researched extensively, performance funding in health care is still a new phenomenon and existing research is limited. Just recently Dutch hospitals published their annual reports for 2014 and can potential benefits be identified.
At this point it is not clear if performance funding for Dutch hospitals actually lead to the potential benefits, expected by the Dutch government. Based upon NPM research it is, at the very least, doubtful that
expected potential benefits will be met. The goal of this study is to focus on the expected benefits, the realized benefits and the potential gap between those for performance funding in Dutch hospitals. This leads to the following research question:
To what extent has performance funding of hospitals led to increased efficiency and improvement of their quality of service?
1.3 Relevance
This study contributes to existing literature for several reasons. First of all this study tries to provide empirical evidence related to the use of NPM in health care. As mentioned by Alonso e.a (2014) there is a lack of conclusive empirical evidence regarding the relationship between the use of NPM related policies and efficiency in healthcare management. Second, to the best of my knowledge this is the first study to analyze the relation between performance funding, efficiency and quality of health care in hospitals. NPM has been studied extensively; however performance funding might actually differ from previously studied NPM implementations. Especially since NPM is such a broad term, existing studies on NPM don’t
automatically apply to performance funding in specific. Third this study focuses on Dutch hospitals in specific, while existing literature on the use and effects of NPM in hospitals was conducted in other countries. Effects of NPM might differ by country, as Hood (1995) argued: in spite of allegations of internationalization and the adoption of a new global paradigm in public management, there was considerable variation in the extent to which different countries adopted NPM.
This research is also relevant in practice since for as well the government, civilians and hospitals it is important to understand the advantages and disadvantages that performance funding creates. Especially since performance funding is deployed to secure affordable healthcare both now and in the future . The structure of this paper is as follows. The next section elaborates on the concepts of New Public
- 6 -
2. Literature
2.1 New Public Management
In the 1980s the movement New Public Management (NPM) emerged, in which the market would serve as an example for control within the government apparatus. NPM can be described as all analyzes, ideas and best practices of public administration experts and other scientists about the renewal of the government in terms of running it more like a (business and) commercial organization. In this case with more attention to, among other things, a result-oriented culture, performance measurement, decentralized responsibilities (contract management) and an alternative construction by remote placement of government organizations, privatization, and promoting competition and market forces (Hood, 1991, 1995; De Vries en Van Dam, 1998).
Hood (1991) states New Public Management is a broad concept but is characterized by a several doctrines that prescribe how the public sector should be organized:
Table 1: Doctrines of new public management volgens Hood (1991), summarized by Korsten (2011)
No. Doctrine Meaning Justification
1 'hands-on professional management'
discretionary control, free to manage
accountability, clear responsability 2 standards and
performance measures
quantitave definition of goals, succes
accountability requires clear goals
3 output controls allocation linked to performance
stress results rather than procedures
4 disaggregation of units break-up monolyth into units with decentralized budgets
need for manageable units seperate provision and production
5 competition contracts and public tendering rivalry as the key to lower costs and better standards
6 private sector style management
away from ‘public sector ethic’ flexibility
use proven business management tools 7 discipline and parsimony
in resource use
cutting costs labour discipline do more with less
- 7 -
transferred to the public sector. For example most budgets in the private sector are short-term, while most public sector budgets must be planned long before the year of actual spending. (Simonet, 2015). Also most public service budgets are driven by incrementalism, in contrast to private sector rationality. Fotaki (2010) state NPM’s core values, like competition, freedom of choice and di fferentiation, frequently clash with core public values like socialy equity, universality of service and homogeneity (Fotaki, 2010).
Verbeeten e.a. (2015) studied NPM in Dutch municipalities and specifically examined two relations. First the relation between performance measurement systems, result-oriented culture and performance; and second the relation between internal deregulation, result-oriented culture and performance. Based upon their research Verbeeten e.a. (2015) concluded the results indicate the degree of result orientation is positively associated with organizational performance. However, no support was found for the claim that the use of performance measurement systems contributes to a results-oriented culture. Moreover, they found that the reliance on rules contributes positively to a results-oriented culture. For internal
deregulation a distinction between two types of deregulation was made. The first type of deregulation concerns ‘strategic decentralization’, which includes policy decision rights and investment decisions. The second type of deregulation conerns ‘operational decentralization’, which includes internal process design, personnel management and outsourcing decisions (Verbeeten e.a.,2015, p. 964). Decentralization of strategic decision rights was found to negatively affect the degree of results orientation, but it did show a positive direct effect on performance. However no significant relation between operational
decentralization and result orientation was found, while a negative direct effect of operational
decentralization on performance was found. Overall Verbeeten e.a. (2015) concluded NPM reforms may have had a negative affect on public sector performance. The assumed benefits of internal deregulation were illusory in their sample. They found no empirical support for one of NPM’s key tenets: that the use of performance information systems improves performance.
Hood who studied NPM extensively since 1991 concluded in 2015: ‘So what do we have to show for three decades or so of NPM reforms? The short answer seems to be: higher costs and more complaints’ . Hood (2015) looked at the last dree decades in the United Kingdom, a county embracing successive NPM-style reforms of its public services, and concluded that costs of public services increased substantially in spite of a reduction of roughly a third of civil service numbers. So even though several NPM-style reforms were introduced during those years, those reforms were not able to reduce costs of public services at all. Dan and Pollitt (2015) revealed a more balanced picture of NPM. They state NPM reforms have not always been successful however there is substantial evidence of NPM like measurements that have provided benefits for public sector organizations (Dan and Pollit, 2015). They state that NPM cannot be seen as the ultimate solution for all public sector issues, however empirical studies have showed in some cases NPM was able to provide some positive consequences. According to Dan and Pollit (2015) NPM reforms often lead to a variety of consequences, both positive and negative.
- 8 -
2.2 Existing literature on New Public Management in health care
Several studies examined the consequences of NPM implementations in healthcare. In 2015 Simonet analyzed healthcare reforms in the United Kingdom following the introduction of NPM theory -inspired reforms. Simonet studied the implementation of NPM to healthcare and the effects of NPM on several core elements of healthcare, such as costs, quality, and patient empowerment. Based on the ideas of NPM the UK government implemented several NPM reforms. First of all the British National Health Service (NHS) created an internal market:
‘An internal market or a quasi-market with a separation between care providers, the hospitals and funder was created, with the assumption that ownership and operations by competing organizations would lead to greater empowerment, efficiency, and equity (Le Grand, 2003).’ (Simonet, 2015, p. 808)
However Simonet (2015) concluded those reforms were only partial. The British government did little to to enforce competition and therefore market forces were limited.
Second the UK government instituted changes in funding:
‘Another prominent illustration of the quasi market was the fund-holding scheme in which a budget was transferred from the District Health Authorities to a group practice. The practice used that budget—a fixed sum (i.e., a capitated fee) that depended on the number of patients, between 1,500 and 1,800 people; and on patient category, for example, elderly or pregnant women—to purchase hospital care, prescriptions, and to cover physician practice operating expenses. Acting on behalf of his patients, the fund -holding physician negotiated the most suitable and effective care treatment available, such as, directly with hospitals a nd private clinics of his choice (e.g., specialized services, medical examinations, and pharmaceutical
prescriptions).’ (Simonet, 2015, p. 809)
These changes could not provide the expected benefits. The District Health Authority kept most of their budget in order to pay for the most expensive treatments, and if a practice did not use its whole budget, the District Health Authority would keep what was left of the budget. The reform was unable to contain costs and the complexity of this new way of funding was actually responsible for an increase in bureaucratic costs. Also group practices did not actually bear a financial risk, since ultimately the government bailed out those practices that had cost overruns (Simonet, 2015).
Third they instituted performance contracting:
‘Contracting out, or externalizations, to private or public providers via competitive bidding became common policy options under NPM, the latest being the sale of Britain’s blood bank.’ (Simonet, 2015, p. 810)
Simonet 2015 stated there is little evidence that performance contracting was able to improve outcomes. The goal of performance contracting was to increase competition, however purchasers stimulated cooperation between providers in order to create stability (Simonet, 2015). Therefore the realized
competition was only partial. Overall Simonet concluded that there was a large gap between NPM promises beforehand and the outcomes of those NPM reforms in practice.
Earlier in 2013 Simonet also studied NPM in French public hospitals, where NPM reforms were
- 9 -
with high uncertainty (e.g., medical outcomes are difficult to predict), and in bureaucracies that are closed and self- regulated (e.g., hospitals) while failing to fulfill its promises (e.g., transparency and accountability) Alonso e.a. (2014) studied NPM in Madrid Hospitals, and deployed a data development analysis to compare efficiency scores in traditionally managed hospitals and those operating with new (NPM) management formulas. The study could not reveal any statistical significant differences in efficiency between traditionally managed hospitals and those using NPM. Alonso e.a. (2014) suggested that what really matters might be the individual hospital management, not the management model itself.
In conclusion existing research shows little to no evidence of NPM implementations that were able to provide the desired outcomes. Outcomes were either below expectations or the desired outcomes were not achieved at all. Based upon existing research it is doubtful performance funding for Dutch hospitals will achieve the desired benefits.
2.3 Performance funding
In the nineties of the last century many governments have fallen under the spell of placing organizations that deliver public duties on a distance. After privatization these task organizations had to be controlled based on the performance they delivered. This introduction of performance funding should lead to a business-like, less informal relation between the funding government and the concerned organization. The intention is that, after the implementation of performance funding, the government does not have to pay attention to the financial healthcare of the task organization. The funding government bears still limited responsibility in case the concerned task organization would go bankrupt (Herweijer, 2010). Except to funding based on results, funding governments also aimed to boost market forces under the new
conditions, meaning the executive organizations are competing for the favor of citizens or respectively of the funding authority (Herweijer, 2010).
Performance funding fits within the doctrines of New Public Management as mentioned in Table 1. The focus is on 'output controls’ (pay per performance), 'competition' (market forces with the objective of lower costs and higher quality) and ‘discipline and parsimony in resource use’ (work efficiently). Proponents of New Public Management say that performance funding leads to more effective policy implementation but not in itself result in a substantive change of policy. Their claim is: NPM leads to cost savings
(efficiency), but not a change in policy (Herweijer, 2010). Performance funding in Dutch Healthcare
Before performance funding was fully implemented in 2014 Dutch hospitals were funded mostly (for around 65%) based upon a fixed yearly budget, set by the Dutch Government. If these budgets proved not to be sufficient, the budget could be extended. The other 35% included the B-segment. For this part a free market applied where hospitals and health insurance organizations make arrangements about the volume and prices of treatments that fall under this B-segment (Inview, 2012).
In 2012 performance funding was introduced in the Dutch healthcare system, meaning health care providers were now only funded based upon the care they actually deliver. So the income of a hospital is equal to the price per performance (healthcare) and the number of times (volume) they deliver these performances2. Both segment A and B are funded based upon performance funding, but only for segment B
2
- 10 -
prices are freely negotiable. For segment A there is a maximum rate set by the government. There is also a fixed segment, which contains some exceptions like for example donor removal teams and helicopter facilities for trauma care. Table 2 shows the situation for Dutch hospitals before performance funding and after.
Table 2: Funding of Dutch hospitals
Funding/Financing Dutch hospitals 2011 and before
A-segment (approx. 66%) B-segment (approx. 34%)
Budget / fixed rates Performance funding / freely negotiable rates Funding/Financing Dutch hospitals 2012 and after
A-segment (approx. 30%) B-segment (approx. 70%)
Performance funding / set maximum rates Performance funding / freely negotiable rates
In theory the use of market forces and performance funding should lead to lower costs and higher quality. Since hospitals compete with one another, have to negotiate with health insurance organizations, and only get paid for their actual delivered care, hospitals have to deliver high quality care at a competitive price. This lead the Dutch government to believe that through performance funding health care remains affordable and improves the quality of care3.
Unlike other NPM implementations performance funding creates a sense of urgency. If hospitals fail to offer quality care at a competitive price, health insurance organizations might no longer offer healthcare at that specific hospital. This could lead to a drop in patients and financial trouble (since hospitals get only paid for actual performance) for the concerned hospital, or ultimately even bankruptcy. Especially since the Dutch government does not automatically help hospitals in financial need, as it did in the past decades. For the first time hospitals will be fully responsible for their own financial well-being and will no longer be able to rely on the Dutch government. Some Dutch hospitals have already gone bankrupt, while another 20% of the hospitals are in the danger zone financially (BDO, 2015). Here the Dutch situation differs from the British, as mentioned before, where Simonet (2015) had to conclude the British government fell short in ensuring competition and bailed out those who had cost overruns eliminating the financial risk. In the Netherlands competition seems to be ensured and hospitals do actually carry genuine financial risks. 2.4 Existing literature on performance funding
Marc Robinson (2002) studied the feasibility of performance funding for the funding of tax -financed public services. In this study Robinson (2002) uses the term ‘output-purchase funding’ instead of performance funding, which also relates to funding based upon the quantity of outputs a government agency actually delivers. Robinson (2002) stated output-purchase funding in the public sector is often implemented in order to increase competition and therefore efficiency. This is in line with one of the reasons the Dutch
3
- 11 -
government introduced performance funding in the Netherlands; to increase efficiency . Robinson
concluded funding based upon actually delivered output is unworkable or too costly administratively when outputs are: ‘highly heterogeneous, moderately heterogeneous but produced in low volume, or contingent capacity services’ (Robinson, 2002, p.33). A selective use of output-purchase funding is possible when outputs are more homogeneous, however output-purchase funding should not be implemented as a basis for overall government funding (Robinson, 2002). Since the outputs of a hospital are highly heterogeneous one would expect that performance funding is not suitable for Dutch hospitals.
Performance funding in the Netherlands
Total performance funding in Dutch health care is relatively new, however performance funding has already been introduced in other public sectors in The Netherlands. Herweijer (2010) examined the effects of performance funding in the province of Drenthe and in the Dutch administration of justice.
Province of Drenthe
The first case occurred in Drenthe, which is one of the twelve Dutch provinces. Every year around 20 public organizations - varying from the regional public broadcast to the provincial library, to the Youth Welfare Office – receive subsidy from the provincial government. Since 2004 these institutions are funded based on performance. The introduction of performance funding in Drenthe lead to an increase in bureaucracy costs since public organizations had to reform their internal administration. After implementation of
performance funding public organizations spend more time drawing up budgets, conducting hour
administrations and documenting the actual performance. An important assumption behind performance funding is that the funding authority should be able to devote less attention to the financial health of the task organization. Nevertheless Herweijer (2010) found the focus on the financial aspects of task
organizations increased rather than decreased.
Performance funding resulted in a limited increase in competition, but was less than expected before implementation. This was partially caused by several executing task organizations merging, as a response to increased risks; decreasing the number of competitors in the market. Funding based on actual
performance, happened less pressing than announced. In practice pledged budgets were often paid, despite demonstrably inadequate performances.
Dutch administration of justice
In 2002 performance funding was introduced in this sector. Herweijer (2010) found that productivity increased by 8% between 2002 and 2005. This corresponds to an annual increase in productivity of 2% a year, which is a remarkable development for government work. However many judges felt their workload as high to very high. In 2008 59% of the judges experienced their workload as high to very high, compared to only 18% in 2002. Partly as a result of this high workload a worrying development emerged, namely that judges and councilors now indicate they have no more time for professional development. Over 52% mention they are regularly not able to participate in jurisprudential- or professional consultation. Another 40% indicates they are often not able to participate in professional courses during working hours and a whopping 71,9% of the judges state it is not possible to commit self -study during working hours.
- 12 -
judges experienced an increase in workload. Also judges spend less time on professional development, which does not benefit the quality of the administration of justice.
2.5 Hypotheses
Existing literature shows little to no evidence of NPM based implementations that actually lead to the desired results. So based upon this literature one would expect that the introduction of performance funding will not lead to an improvement in efficiency or quality of service for hospitals.
So the hypotheses of this thesis are formulates as follows:
H1 The introduction of performance funding for Dutch hospitals does not lead to increased efficiency H2 The introduction of performance funding for Dutch hospitals does not lead to increased quality of
- 13 -
3. Methodology 3.1 Sample
To empirically test the hypotheses, data were collected from all general Dutch Hospitals that were functional both before and after the introduction of performance funding. Since performance funding in the Netherlands was introduced in 2012 and fully implemented in 2014 we compare data from 2011 and 2014. Data from 2012 and 2013 is not included, since there was a transitional arrangement and so
therefore performance funding was not yet fully implemented. A total of 86 Dutch general hospitals were operational in 2011 and/or 2014. However some hospitals had to be excluded from the sample since either necessary data was not available, the hospital was part of a merger or the hospital went bankrupt between 2011 and 2014. Academic hospitals were also excluded from the sample, since they provide some
additional features relative to general hospitals. A general hospital is a concentration of facilities for
research, treatment and care that in addition educates future doctors and nurses. Academic hospital have a number of functions corresponding to the general hospitals, namely the regular patient care, clinical care and job training for medical specialists. However academic hospitals also have a top referral function, a research function (scientific research and education for the medical faculty) and a development function (development of new medical technologies and treatments)4. In order to provide these additional functions academic hospitals receive subsidies from the government. Because both the function and funding of academic hospitals differ from general hospitals, academic hospitals were excluded from the sample. Finally a total of 60 Dutch general hospitals were included in the sample to test the efficiency hypothesis (see Appendix A) and 72 Dutch general hospitals were included in the sample to test quality (see Appendix C). Table 3 provides an overview of the structure of the samples used.
Table 3: Samples structures
Efficiency sample Quality sample Number of Dutch General hospitals functional in 2011 and/or 2014 86 86 Number of Dutch general hospitals excluded in sample because of:
Missing data -10 -12
Merger -14
Bankruptcy -2 -2
Number of Dutch general hospitals in sample 60 72
Number of years 2 (x) 2 (x)
Total sample 120 144
In the following sections I will discuss how efficiency and quality are measured and how the necessary data was collected.
3.2 Efficiency
In the theory of production and costs of organization, three aspects of efficiency are mostly used (Ludwig, 2008): technical efficiency, economic efficiency and allocative efficiency. Technical efficiency relates
quantities of inputs to the quantity of output, while economic efficiency relates the currency value of inputs
4
- 14 -
to the currency value of output. An organization is technically efficient when it produces a certain level of output with the least amount of input. Economic efficiency would be achieved when a certain level of output is produced with the lowest cost of inputs. Allocative efficiency occurs when there is an optimal distribution of goods and services, taking into account consumer’s preferences; i n particular, every good or service is produced up to the point where the last unit provides a marginal benefit to consumers equal to the marginal cost of producing.
From a measurement perspective, efficiency can be divided in technical and economic efficiency. Technical efficiency refers to the ability to minimize input in the production of a given output vector, or the ability to obtain maximum output from a given input vector. Technical efficiency imposes no behavioral objective on the organization since the output (or the input) is considered given (Blank, 1998a). Economic efficiency imposes a behavioral objective on the organization: cost minimizing behavior, profit maximizing behavior or revenue maximizing (Ludwig, 2008). Almost all international studies on hospital efficiency assumed cost minimizing behavior for hospitals. Cost minimizing behavior implies that hospitals minimize costs where the production is given. The main reason that most models build on this assumption is that many health care systems hospitals have no incentives to pursue profit maximization (Ludwig, 2008). In this study hospital efficiency is defined as economic efficiency; producing a certain level of output with the lowest cost of inputs (cost minimizing behavior).
To study the efficiency of hospital three issues deserve further discussion (Ludwig, 2008): • What are inputs, and how do we measure them?
• What are outputs, and how do we measure them?
• What efficiency benchmark group will be used to compare the output divided by input measure? What are inputs, and how do we measure them?
In order to provide healthcare a hospital uses several inputs, which can be divided into three general types of input: labor, capital and materials. The costs of these three general inputs together make up for the total costs of a hospital. Most studies measuring efficiency in hospitals use these total costs as input in order to measure efficiency (e.g. Zuckerman, 1994; Chirikos, 1998; Chirikos, 2000; Rosko, 1999, Li, 2001; Sari, 2003, Street, 2003; Rosko, 2004). Following these studies in this thesis input is measured as the total costs of a hospital, divided into three general types of expenses: labor, capital and material. The costs are measured based upon the actual costs made by hospitals in 2011 and 2014. These actual costs derive from the financial statements provided by each hospital. Table 4 shows an overview of the three general types of inputs for a hospital, how these inputs are measured and what they consist of.
Table 4: Inputs of an average Dutch hospital Inputs Measured as costs derived
from income statement
Consists of:
Labor Labor costs -Wages and salaries
- 15 -
Material Other operating expenses -Food and hotel costs -General costs -Patient-related costs
-Maintenance and energy costs -Rental and leasing
-Additions and release provisions
Labor costs include all personnel related costs. Capital costs consist of the deprecation costs of both tangible assets (i.e. land, equipment, machinery) and intangible assets (i.e. patents, medical records, service contracts, licensing agreements). Costs for materials include other operating expenses such as: rent, overhead, maintenance and provisions for possible future costs.
What are outputs, and how do we measure them?
In general the output of a hospital should be health gain. However health gains are difficult to measure. Even if health gain could be measured, it is difficult to measure which part of health gain can be attributed to the hospital (Ludwig, 2008). Empirical studies on hospital efficiency therefore use several other ways to measure output. First of all several studies use Diagnosis-related groups (DRGs), which is a system of classification to identify the ‘products’ that a hospital provides (e.g. Chirikos, 2000; Rosko, 2001; Brown, 2003; Linna, 1998). Second some studies use the number of patients and/or admissions. Thirdly there is a small number of studies using other output measurements such as morbidity, indicator variable per doctor (Bradford, 2001), patient and hospital characteristics (Sari, 2003), alive and dead discharges (Dismuke, 1999), number of free beds, waiting time, and emergency admissions (Jacobs, 2003).
Based upon the financial statements of Dutch general hospitals we are able to identify three general types of output for Dutch hospitals: healthcare, training and other outputs. Table 5 provides an overview of these outputs and what they consist of.
Table 5 Outputs of an average Dutch hospital
Outputs Stated as income on financial statement Consisting of:
Healthcare Budgeted care benefits -Specialized care of burn wounds -Education fund
Non-budgeted care benefits -Benefits on behalf of other institutions -Other non-budgeted care
Turnover DBC’s B-segment -Income resulting from providing healthcare belonging to the B-segment Turnover DBC’s A-segment -Income resulting from providing
healthcare belonging to the A-segment
Training Subsidies -Dutch hospitals receive subsidies for
training of medical personnel
Other Other operating income -This includes other income created by a
- 16 -
Combination’. The DBC system groups healthcare into a DBC healthcare product, which is a total hospital treatment. So the entire treatment process, from diagnosis of a specialist, possible hospital treatment, and appropriate follow-up(s). In total there are around 4.400 different DBC healthcare products. DBC
healthcare products belong to either the A or B segment. For DBC healthcare products in segment B free prices apply. Hospitals and insurance organizations can negotiate prices for treatments in this segment. Segment A includes all treatments that are not defined as B-segment. Prices of treatments in the A-segment are set by the Dutch Governments5. The DBC system has been introduced in the Netherlands in order to gain more insight into the care hospitals provide. In short the DBC system is a system that allows healthcare facilities to register the performed care, in order to eventually declare this to the patient or health insurance organization6. There are also some exceptions of healthcare that are not included in either the A or B segment. Budgeted care benefits include the few services a hospital deliver that are not included in either the A or B segment but are covered in the fixed segment. These include for example: donor removal teams, emergency hospitals and helicopter facilities for trauma care.
In this study a mix of several variables is used to measure the outputs of hospitals. The choice of these variables represents a compromise between what is ideal and what is feasible. Ideally all mentioned variables are used to measure output, however this study contains a large sample and given time is limited. So only publicly available data will be used to measure output. Data will be collected from the annual reports, financial statements and DigiMV files that Dutch hospitals are legally bounded to publish every year. A DigiMV file contains, among other things, key data of the production of a hospital. The production of hospitals in terms of healthcare will be measured based upon these DigiMV files. These files provide the following key data regarding healthcare production: number of opened DBC’s, number of surgical
operations, number of admissions, number of patients, number of clinical nursing days, and the number of visits to the policlinic. All of these numbers were used to measure healthcare output, except for the number of surgical operations and the number of patients since too many hospitals in the sample did not publish these numbers. Training and other outputs were measured based upon the annual reports and financial statements. Table 6 provides an overview of the three general types of output for a Dutch hospital and what variables will be used to measure these outputs.
Table 6 Outputs of a Dutch hospital and how they are measured
Output Measured based upon
Healthcare Number of opened DBC’s in A and B segment
Number of admissions Total clinical nursing days
Total number of visits to the policlinic
Training Measured as subsidies on the financial
statement of a hospital.
Other Measured as other operating income on the
financial statement of a hospital
5
https://www.nza.nl/zorgonderwerpen/zorgonder werpen/ziekenhuiszorg/veelgesteldevragen/dbc -systematiek/ 6
- 17 -
What efficiency benchmark group will be used to compare the output divided by input measure?
The benchmark group in this study includes general Dutch hospitals that were operational in both 2011 and 2014.
3.3 Method of analysis for efficiency
There are several ways to measure efficiency of hospitals. The simplest efficiency measure would be to calculate the input/output ratio. This ratio however might be too restrictive because:
-multiple outputs are produced; -multiple inputs are used;
-scale and scope effects are neglected;
-no distinction between chance and (in)efficiency can be made (Ludwig, 2008).
In existing literature studies on efficiency in hospitals two methods to measure efficiency are used most often; the Stochastic Frontier Analysis (SFA) and the Data Envelopment Analysis (DEA). SFA is a parametric approach that assumes all firms are not efficient and accounts for noise. SFA requires input and output quantities for empirical estimation of production functions (Ozcan, 2008). According to Jacobs (2011) SFA: ‘constructs a smooth parametric frontier which accounts for stochastic error but requires s pecific
assumptions about the technology and the inefficiency term which may be inappropriate or very restrictive (such as half-normal or constant inefficiency over time)’ (p.104). DEA is a popular data-oriented approach applied to the evaluation of relative work efficiency in the public agencies that use multiple inputs to produce multiple outputs (Cooper e.a. 2011). DEA was created specifically with the work of non-profit organizations and public agencies in mind (Charnes e.a., 1978). No assumptions are made about the shape of the (production) function when using DEA. The method can handle multiple numbers of inputs and outputs simultaneously.
Both DEA and SFA have their advantages and disadvantages. DEA has the advantage of requiring no
assumptions about the production frontier, but have the disadvantage of assuming no statistical noise. SFA has the advantage of allowing for statistical noise, but require strong assumptions about the production frontier (Jacobs, 2011). DEA is a non- parametric method, which by means of linear programming comes to an efficiency score of organizations. DEA does not work with standard errors (as is the case with SFA ), but blames deviations from the optimal frontier to inefficient processes (Ozcan, 2007). Disadvantage of this is that due to the non-statistical nature of this method, the robustness of the model is difficult to assess, and this method is more sensitive to errors in the data (Jacobs, 2000). However, Jacobs (2000) indicates that DEA has the advantage that it can work with more complex organizations, which often use multiple input and output variables, such as hospitals.
Given these advantages a Data Envelopment Analysis was used in this study to measure efficiency of Dutch hospitals. DEA is a mehod that evaluates the efficiency of so called Decision Making Units (DMUs). These DMUs can be business units, government agencies, police departments, hospitals, educational institutions, and even people. All DMUs under comparison are assumed to operate homogeneously, they receive the same inputs and produce the same outputs (differing in quantities) and these inputs and outputs are representative of the whole population7. From this set of inputs and outputs, and a minimum of
assumptions a technological feasible maximum is derived. This is called the ‘frontier’. A hospital which is on
7
- 18 -
the ‘frontier’ position is ‘best practice’, based upon the data it is not possible to conclude this hospital could do any better. The frontier is derived from actual performance (inputs and outputs) of existing institutions. For the other hospitals it is now possible to calculate the distance to the ‘frontier’. This distance we can translate to a cost-effectiveness score or performance score. Each hospital is compared with a combination of existing hospitals under more or less the same conditions (Nza, 2012). In appendix F the DEA method is explained in some more detail. The DEA analysis was performed using MaxDEA software. The by MaxDEA created efficiency scores will be imported in SPSS, which allows the comparison of efficiency scores from 2011 and 2014.
3.4 Quality
The quality of hospitals is measured based upon the results of Algemeen Dagblad’s ‘ziekenhuis top 100’, which is a yearly ranking of Dutch hospitals based upon quality indicators. Algemeen Dagblad uses a total of 36 quality criteria and assigns a score to each Dutch hospital. These criteria relate to medical treatments at the hospital, such as the number of reoperations, the number of risky operations, but also to organizational processes in the hospital and the quality of nursing. Most of these criteria (32 out of 36) are derived from the Dutch Health Inspectorate, which uses these criteria to keep track of the medical quality of hospitals. Algemeen Dagblad selected these criteria for several reasons: a mixture was created of common, chronic and very specialized conditions; but also criteria which are relevant for each patient are included, such as the quality of the nursing staff. Starting-point for criteria to be considered is that the number reflects the quality of care and whether clear medical standards exist to which a hospital would have to comply. The other 4 criteria are partly focused on patient friendliness, which include four patient quality labels awarded by patient associations. Such a quality label is assigned based upon patient desires and in consultation with medical specialists8. A full list of indicators used by Algemeen Dagblad to rate hospital quality can be found in appendix G.
Algemeen Dagblad translates the scores of all these indicators to a total score, which is 64 points if a hospital meets all measured quality criteria. The percentage of points obtained of the maximum achievable number of points determines the final score and the ranking. So the maximum score is 100% (64 points) and the lowest possible score would be 0% (0 points). These percentages will be used in order to create a dataset for both 2011 and 2014. These datasets will be compared and potential differences in quality scores will be measured for significance using SPSS.
Elsevier also provides a yearly list of hospitals quality, which is somewhat similar to Algemeen Dagblad’s ‘ziekenhuis top 100’. Since Elsevier uses a total of 395, instead of 36, quality indicators this was the first choice of measurement for hospital quality. However the scores Elsevier appoints to each hospitals do not allow for comparison between years.
3.5 Control variables
Besides the introduction of performance funding other variables might have influenced either efficiency or quality of service for hospitals between 2011 and 2014. Some control variables have been identified based upon existing literature, examination of government measures related to Dutch hospitals between 2011 and 2014, as well as examination of publications by Dutch healthcare institutions and authorities between 2011 and 2014. These control variables are further discussed below.
8
- 19 -
Increase in demand for hospital care
The demand for hospital care has increased over the last decades as a result of an aging population and also the possibilities of (medical) technology9. This increase in demand led to an increase in production of hospitals over the years. This could potentially affect efficiency since efficiency is basically an input/output ratio; for example if output increases and input remains the same efficiency will thereby increase. To see if the increase in demand (and therefore production) might affect efficiency one has to look at the increase in total costs as well. In the 5 years prior to the introduction of performance the total costs of Dutch hospitals increased by 5,6% on average, while the total production increased by 5,7%10. So both production and costs increase almost equally because of the trend of an increasing demand for hospital care. Based upon these numbers the trend of increasing demand for hospital care does not influence efficiency of Dutch hospitals and will therefore not be a control variable in this study.
Scale up and concentration
A trend has emerged in The Netherlands where hospitals scale up and concentrate in order to save costs and improve quality11, which occurred mostly through mergers between existing hospitals. Also between 2011 and 2015 several hospitals. Since these mergers might influence efficiency or quality those hospitals were not included in the sample of this study.
Introduction of quality- and transparency standards
According to the Dutch Association Hospitals (NVZ) and the Dutch Healthcare Inspectorate (IGZ) the quality of hospital care has increased in the years prior to the introduction of performance funding. Between 2008 and 2010 almost all quality indicators for hospitals treatments have improved12. In addition the turnaround time per treatments improved for both the A- and B segment2 and mortality numbers decreased
throughout the years1314. The NVZ mentions the development of quality- and transparency standards as a reason for this increase in quality. In 2007 the Dutch government started a program ‘zichtbare zorg’ (which translates to ‘visible care’) in order to both improve quality of hospital care and create more transparency about quality of hospital care. Goal of this program was that hospitals would collect and provide data of a total of 498 quality indicators. Between 2007 and 2014 several measures were taken to enforce hospitals to collect and provide this data. The NVZ concludes there is no conclusive scientific evidence that providing public information about quality indicators improves the quality of care, yet in practice it seemed some public indicators caused a positive effect on quality of healthcare. Based upon this trend of improving quality prior to the introduction of performance funding one would expect this trend to continue and for quality of hospital care to improve between 2011 and 2014, regardless of the influences of performance funding.
Hospital size
Existing research indicates a positive relationship between firm size and efficiency (Schiersch, 2013) , mainly because small firms cannot exploit economies of scale (Audretsch, 1999). Also research specific on hospitals
- 20 -
found that hospital size was consistently and positively related to efficiency due to economies of scale (Ozcan & Luke, 1993; Coyne, 2009). Furthermore existing research indicates bigger organizations perform better than small organizations in terms of quality (Fareed, 2012). Contributors to these increase quality varies from increased availability of resources, better economies of scale to more opportunities for helping advance medical knowledge for a given problem (Fareed, 2012). These claims are supported by several empirical research, such as Keeler (1992) and Fareed (2012) who both found that large hospitals have better quality in general than small hospitals.
Since hospital size is able to influence both efficiency and quality, hospital size will be a control variable in this study. In imitation of previous research (Coyne, 2009; Fareed, 2012; Keeler, 1992) hospital size is measured by the number of available beds.
Age of the hospital
Previous research also indicated the age of a hospital might influence financial performance of a hospital (Abor, 2015; Marlin e.a., 2013). Marlin e.a. (2013) found that older hospitals positively influenced financial performance, while Abor (2015) found older hospitals perform better on efficiency. The fact that older hospitals perform better attributes to the fact that these hospitals have more experence and are well -resourced. Based on the learning curve, such a hospital might therefore be able to operate at lower cost or at a higher efficiency level than their younger counterparts (Abor, 2015). Therefore hospital age will be a control variable in this study and will be defined as the number of years the hospital has been in e xistence. There is no evidence found hospital age possibly influences quality, however in order to be complete hospital age will also be controlled for the quality analysis.
Location/population density
Several studies found that hospital location influences financial and non-financial performance of hospitals (Abor, 2015; Collum, 2014; Keeler, 1992). Abor (2015) found that national hospitals located in the national capital showed higher occupancy- and discharge rates and therefore influence efficiency. Abor (2015) relates these high occupance rates to the high number of population in national capitals, which therefore leads to a high occupancy of available beds for hospitals located in such a capital. Furthermore Keeler (1992) found that more urban hospitals have better quality in general than rural hospitals. According to Keeler (1992) rural hospitals may have difficulty in attracting skilled physicians, and these physicians may not have enough patients or contact with other physicians to maintain their ski lls. Since the population density in the region where a hospital is located possibly influences efficiency and/or quality this will be a control variable in this study. The population density is measured according to the number of inhabitants per square kilometer in the Dutch province where the hospital is located.
Inflation
Inflation influences the general level of prices for goods and services. So in order to compare cost and benefits of a hospital between 2011 and 2014, these numbers have to be correcte d for inflation in between these years. The inputs capital and material of 2011 were therefore corrected for inflation, by increasing them by 2.53% (for 2012), 2.47% (for 2013) and 0,98% (for 2014). Labor costs will be discussed hereafter. For the outputs only the ‘other outputs’ were corrected for inflation, since healthcare output is not measured in economic quantities and subsidies for training were not subject to inflation.
- 21 -
- 22 -
4. ResultsThe first hypothesis of this study was tested using a Data Envelopment Analysis and then a regression analysis. The second hypothesis was measured using a regression analysis. The results of these analyzes are presented below.
4.1 Data Envelopment Analysis
To measure efficiency in Dutch general hospitals first a Data Envelopment Analysis was performed to calculate the efficiency scores for each hospital in 2011 and 2014. Table 7 provides a summary of the performed DEA. The calculated efficiency scores for each hospital can be found in appendix B. Table 7 Summary Data Envelopment Analysis
Model Type Envelopment model
Number of DMUs (Decision Making Units)
120 Number of Inputs 3 Number of Outputs 6 Assumptions Distance Non-radial Orientation Input-oriented
Returns to Scale Variable
As table 7 shows the DEA was performed with a sample of 120 (60 hospitals for 2 years), 3 inputs and 6 outputs (see appendix A). In all variations of the DEA models, the DMU’s with the best inherent efficiency in converting inputs into outputs is identified, and then all other DMU’s are ranked relative to that most efficient DMU (Schaar e.a., 2008). However there are some variations possible in the assumptions made when performing the DEA. Table 7 shows the assumptions that were made when performing the DEA of this study.
The first assumption relates to the choice between a radial or non-radial measurement of efficiency. Radial measurement assumes proportionate reduction of input resources in order to become efficient. So in other words if the organization unit under study has two inputs, this model aims at obtaining the maximum rate of reduction while maintaining the same proportion between input 1 and 2. In contrast, the non-radial models put aside the assumption of proportionate contracti on in inputs and aim at obtaining maximum rates of reduction in inputs that may discard varying proportions of original input resources (Avkiran e.a., 2008). Most studies using DEA follow the concept of radial efficiency measures. However, using radial efficiency measures often leads to the case where a lot of DMUs have the same efficiency score of 1 and hence difficulty in ranking the environmental performance of these DMUs only based on their efficiency scores. Since non-radial efficiency measures have a higher discriminating power in evaluating the
efficiencies of DMUs, non-radial DEA-based models seem to be more effective in measuring environmental performance (Zhou e.a., 2007). So in this study the non-radial measurement of efficiency was chosen.
- 23 -
indicates how much a firm can increase its output for a given level of input (Huguenin, 2012). A hospital cannot simply increase its output, because it relies on a given demand for hospital care. So in this study an input-oriented DEA was performed, following most international studies on hospital efficiency that also assume cost minimizing (input minimizing) behavior for a given output (Ludwig, 2008).
Third there are two basic models used in DEA, leading to the identification of two different frontiers. The first model assumes constant returns to scale technology (CRS model). This is appropriate when all firms are operating at an optimal scale. However, this is quite an ambitious assumption. To operate at an optimal scale, firms should evolve in a perfectly competitive environment, which is seldom the case. The second model assumes variable returns to scale technology (VRS model). This is appropriate when firms are not operating at an optimal scale. This is usually the case when firms face imperfect competition, government regulations, etc. (Huguenin, 2012). In this study a variable returns to scale model is performed since no perfectly competitive environment can be assumed. The calculated efficiency scores were exported to SPSS in order to perform a regression analysis.
4.3 Descriptive statistics
Before analyzing the data, a winsorizing test was completed in order to identify possible outliers. This is necessary to control for impact of outliers on a linear regression. Only a few outliers were identified, and brought back to within the range of 3 times the standard deviation plus/minus of the mean. Table 8 presents the descriptive statistics for all variables used in either the DEA or regression analyzes related. Table 8 Descriptive Statistics
Efficiency sample (H1)
N Minimum Maximum Mean Std. Deviation
Hospital size 120 60 1075 443,900 209,725 Hospital age 120 ,00 171 55,417 45,322 Population density 120 183 1227 689,767 368,6399 Performance funding 120 ,00 1,00 ,500 ,502 Labor 120 29933759 250807311 99209077 50145856,240 Capital 120 602500 37606955 13678442 7684235,761 Material 120 16359499 163421329 59445374 33911333,399 Training 120 0 101760000 5965627 10642783,192 Other outputs 120 0 38202363 10625900 7477002,893 Opened DBC’s total 120 56269 461952 187025 90443,539 Number of admissions 120 2637 47423 20223 9400,075
Clinical nursing days 120 5811 235874 99766 51239,648
Policlinic vists total 120 73160 3141952 342596 294238,719
- 24 -
The descriptive statistics provide the minimum, maximum, mean and standard deviation of the data found in the sample. As the table shows there is a wide variation in hospitals, for example the hospital with the lowest labor costs only spends around 29 million on labor while the maximum found in this sample was over 250 million. The maximum score found for efficiency is 1, which means a hospital is on the ‘frontier’ position. A score of 1 implicates that based upon the data it is not possible to conclude this hospital could do any better. The worst score of efficiency was found to be 0.334 which correlates with 33.34%, which means this specific hospital should improve by over 66% in order to reach the frontier by either decreasing their inputs or increasing their output. Table 8 also presents the descriptive statistics for all variables related to hypothesis 2. The values for the variables might differ from the variables of the efficiency sample, since the sample used to test hypothesis 2 differs from the sample used to test hypothesis 1.As table 8 shows the lowest quality score found in the sample was 48.05 and the highest 89.54.
Table 8 shows a minimum value of 0 and maximum value of 1 for performance funding. In order to
compare the results of hospitals between 2011 and 2014 a grouping variable for performance funding was created in SPSS. This grouping variable has only two possible values; either 0 or 1. The value of 0
corresponds to a hospital with no performance funding, so a hospital in 2011. The value of 1 corresponds to a hospital with performance funding, so a hospital in 2014.
A multicollinearity test was conducted to make sure the dependent and independent variables used for the regression analysis do not influence each other. Table 9 presents these correlations for both the efficiency and quality sample.
Table 9 Correlations efficiency sample Efficiency sample
Hospital age Performance funding Population density Size Efficiency score
Hospital age 1 Performance funding .033 1 Population density -.075 .000 1 Hospital size .098 -.095 -.098 1 Efficiency score .024 -.145 .197* .162 1 Quality sample
Hospital age Performance funding Population density Size Quality score
Hospital age 1
Performance funding .030 1
Population density -.095 .000 1
Hospital size .069 -.109 -.116 1
Quality score -.098 .284** .112 .100 1
*. Correlation is significant at the 0.05 level (2-tailed).
*
*. Correlation is significant at the 0.01 level (2-tailed).
- 25 -
4.4 Regression analysis
Regression analysis is a statistical tool for the investigation of relationships between variables. The main results of the regression analysis for efficiency are showed below in table 10. The full results of this regression analysis can be found in appendix D and appendix E.
Table 10 Regression analysis
H1 H2
Variable Efficiency Score Sig. Quality Score Sig.
Intercept .560 .000 67.515 .000 Performance funding -.130 .148 .304 .000** Hospital size .167 .066 .155 .057 Hospital age .028 .754 -.106 .185 Population density .215 .018* .120 .138 R-squared .089 .126 Adjusted R-squared .057 .100 F value 2.812 4.989
*. Correlation is significant at the 0.05 level (2-tailed).
*
*. Correlation is significant at the 0.01 level (2-tailed).
Regarding hypothesis 1 no significant relation is found between performance funding and efficiency (p=0.148>0.05). The results found in the regression analysis are in line with hypothesis 1; performance does not significantly influence efficiency and therefore performance funding did not increase efficiency for Dutch hospitals. Population density is found to have a significant (p=0.018<0.05) and positive influence on hospital efficiency. These results are in line with previous research of Abor (2015), who stated hospitals located in national capitals show higher occupancy- and discharge rates and therefore higher efficiency. Abor related these high occupancy to the fact that usually national capitals have a high number of population. Since this study focused on the Netherlands only, in this study a distinction was made based upon population density. This result suggest hospitals located in high population density areas tend to have higher efficiency scores.
- 26 -
5. Discussion
The aim of this study was to examine the influence of performance funding in the Dutch hospital se ctor on efficiency and quality of those hospitals. The results of this study are mixed and will be further discussed below.
H1 The introduction of performance funding for Dutch hospitals does not lead to increased efficiency No significant relation was found between performance funding and hospital efficiency. These results are in line with the majority of existing literature. Most previous literature agrees the ideas of NPM are often not suitable for the public sector, because the government has its own character, ethos and complexities (Steen et.al., 2009; Van den Berg & Van der Meer, 2011; Talbot, 2009; Simonet, 2013) . Also Robinson (2002) concluded performance funding does not fit companies with highly heterogeneous outputs, such as a hospital. It is quite possible performance funding was not able to increase efficiency, simply because the principles of NPM are not suited for the public sector.
However the results conflict with the expectations of the Dutch government, which expected performance funding would increase both hospital efficiency and quality. The idea behind these expectations were based upon two principles. First of all performance funding creates a financial need to deliver healthcare,
because a hospitals income is dependent upon the amount of healthcare a hospital delivers. Thereby also a financial risk is created, since failure in delivering sufficient healthcare will lead to financial distress. Second hospitals have to compete with one another to negotiate a contract with Dutch health care insurance companies. Such a contract enables hospital to deliver healthcare to customers of the concerned insurance company. The combination of a financial need to deliver healthcare and competition between hospitals should theoretically stimulate hospitals to offer high quality healthcare at competitive prices17. Also Simonet (2015) highlighted the importance of a financial risk and competition when implementing NPM measurements. Simonet (2015) in fact stated one of the reasons NPM implementations in th e United Kingdom were below expectations, was that the British government fell short in ensuring competition and eliminated the financial risk by bailing out those who had cost overruns.
The financial risk created by the Dutch government is what distinguished performance funding from other NPM implementations studied before. Dutch hospitals do carry a genuine financial risk in the sense that a hospital could potentially go bankrupt. However to what extent does this financial risk influence financially healthy hospitals? Does performance funding stimulate a financially healthy hospital to increase its
efficiency? The financial need to deliver healthcare arguably only affects hospitals in financial need. Possibly the majority of Dutch hospitals still lack the necessity to increase its efficiency. Both organizations and people often display satisficing behavior, meaning they tend to search for satisfactory solutions rather than optimal solutions (Peters and Zelewsk, 2006). The Dutch government assumes hospitals desire to achieve maxium efficiency. However in reality the majority of Dutch hospitals might operate at a satisficing level of efficiency, and do not have the ambition to increase efficiency unless the are forced to do so financially. Such satisficing behavior might explain the lack of increased efficiency.
Furthermore the competition between Dutch hospitals is far from perfect. Hospitals do theoretically compete with one another in order to negotiate a contract with insurance companies. However in practice this competition is rather limited. First of all hospital competition is limited to only a few hospitals in each region. Insurance companies have to negotiate a contract with at least one hospital for each regions, since its customers expect to be able to visit a hospital nearby. Especially when a hospital has no other hospitals located nearby competition is quite small. Also the number of health care insurance companies purchasing healthcare is limited. The insurance market is dominated by four insurance companies which hold 90% of the market. Furthermore most insurance companies state beforehand they have the ambition to conclude
17
- 27 -
a contract with every Dutch hospital, thereby undermining the basic principles of competition. Lastly the Dutch government encourages hospitals to compete on the one hand on price and product, but on the other hand requires more cooperation between the institutions18. This form of market is hardly comparable to a genuine economic-based market, so it is not likely it will provide an increase in efficiency as expected by the Dutch government.
Finally it must be noted that performance funding might be able to increase efficiency over time. As
mentioned Dutch hospitals carry a financial risk and could potentially go bankrupt, since the introduction of performance funding. Therefore it is possible that inefficient hospitals will cease to exist in the future, since inefficiency will eventually lead to financial distress which could ultimately result in either bankruptcy or a forced merger with another hospital. Since the introduction of performance funding two hospitals already went bankrupt, while other financially troubled hospitals already intervened by entering into mergers with financially healthier parties19. Such a development would eliminate inefficiency by eliminating hospitals, leaving only those hospitals efficient enough to survive financially. There is also some evidence hospital mergers do actually increase efficiency (Harris e.a., 2000; Kristensen e.a., 2010). However one may wonder whether such consequences of performance funding and competition were intended by the Dutch
government and are desirable for Dutch citizens.
H2 The introduction of performance funding for Dutch hospitals does not lead to increased quality of service
The performed regression analysis showed performance funding does positively influence hospital quality. This result does contradict the hypothesis and the majority of existing literature. The results do support the Dutch governments’ expectations of performance funding, which are based upon the same idea as for efficiency. Hospitals are dependent upon delivering healthcare and have to do with market forces; so besides offering competitive prices the government expects hospitals to try and distinguis h themselves from other hospitals by providing high quality care20.The results regarding hypothesis 2 are somewhat surprising. First of all most existing literature argues NPM is mostly not able to deliver the expected benefits. Second these results are in contrast with the results for efficiency. There may be a variety of reasons why performance was able to increase hospital quality, some of which are further discussed below. First of all it is possible performance funding did stimulate hospitals to compete with one another. However this competition would be based purely on quality and not price, since efficiency did not increase. For example a development that became visible in recent years is the fact that hospitals waste much money on unnecessary end expensive technology, in order to distinguish themselves from other hospitals in the battle for patients21. Such a development supports the idea of competition in terms of quality. Hospitals increase quality in order to become competitive, however as a result they have to increase costs and therefore efficiency decreases.
Second Dutch insurance companies pay increasing attention to the quality of care when negotiating a contract with hospitals (Zorginstutuut Nederland, 2015). Therefore hospitals are forced to i ncrease their quality of care. Especially since several quality- and transparency standards were introduced in the Dutch
