January 30th 2018
MSc Business Economics
Track: Competition Law and Economics
Master Thesis
The market for expensive medicines: a research about
the effectivity of purchasing groups in the Dutch health
care
by
Maura van Werkhoven
10243429
Supervisor
Barbara Baarsma
Statement of Originality
This document is written by student Maura Elisabeth van Werkhoven, who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
Healthcare costs continue to rise, putting the accessibility and affordability of medicines under pressure. The increasing demand of expensive medicines contribute to the increase in medicine’ costs within the total healthcare budget. One instrument to reduce healthcare costs, is to
stimulate joint purchasing groups for expensive medicines of medical specialist care (MSC). The idea of joint purchasing instead of purchasing expensive medicines individually, is the ability to exert countervailing power against powerful pharmaceutical companies. Purchasing groups that consist of hospitals only or in combination with a health insurer, are able to purchase medicines at pharmaceutical companies at large scales. This should lead to purchase discounts which are passed on to health insurers as in lower contract prices. In this study a qualitative and quantitative analysis have been performed to analyze the effectivity of 12 purchasing groups in the Dutch healthcare market for expensive medicines regarding medical specialist care. Specifically, 11 purchasing groups that consist of hospitals only and 1
purchasing group with the involvement of health insurer Achmea. Moreover, a 13th group,
control group, is formed with hospitals who purchase medicines individually. Besides 9 interviews with buyers of medical specialist care from hospitals and health insurers, a dataset provided by Vektis of contract prices for all expensive medicines for the period 2012 to 2016 is used. The difference between maximum prices set by the Dutch Health Authority (DHA) and the negotiated contract prices between health insurers and hospitals (defined as price deviation), allow me to assess the relation between purchasing groups and their ability to negotiate and pass on purchase discounts. In view of some assumptions, empirical evidence is found that
purchasing groups as well as hospitals who purchase individually, lead to positive price deviations. Indicating that negotiated purchase discounts at pharmaceutical companies, are passed on to health insurers as in lower contract prices. In addition, hospitals who are part of a purchasing group, show a larger positive price deviation (M=0.4266; SD=0.9372) than hospitals who purchase individually (M=0.3723; SD=0.8995). However, since purchase prices and discounts from pharmaceutical companies are not publicly available, observed price deviations may not be equal to the actual purchase discounts. Purchasing groups with significant market power may be able to negotiate larger discounts from pharmaceutical companies, but it does not indicate that all discounts are fully passed on to health insurers. Hospitals who purchase
individually may have less market power compared to health insurers, resulting in a larger share that is passed on to health insurers. Therefore the question remains to what extent the observed price deviations could have been larger and so, how much of the negotiated discount remains within hospitals. When comparing price deviation across 12 different purchasing groups, the purchasing group with health insurer Achmea shows a significantly larger price deviation than all other purchasing groups. This means that for a purchasing group that involves a health insurer, the health insurer is able to put hospitals under pressure so that on average more discount is passed on. However, only 1 purchasing group with a health insurer is analyzed, which weakens external validity. Besides determining the effect of group purchasing on price deviation, some other independent variables are tested as well. When comparing price deviation across 6 health insurer concerns, output results suggest that Menzis has a significantly larger price deviation than the other five health insurer concerns. This means that if all contract agreements between health insurers and hospitals are taken into account, Menzis was more able to obtain some of the purchasing discounts from hospitals, resulting in lower agreed contract prices, than the other 5 health insurer concerns. Moreover, price deviation significantly increases in time for the period 2013 to 2016. However, no evidence is found that price deviation is significantly higher for academic hospitals compared to top clinical hospitals and general hospitals. In addition, no evidence is found that the greater the size of the hospital (#beds) joining a purchasing group, the larger the price deviation. These results may be counterintuitive. Forming a purchasing group leads to larger price deviations as medicines are
bought jointly on a large scale, resulting in pharmaceutical companies who provide discounts. But, this is not reflected in the size of the hospital (#beds) within a purchasing group. Large academic hospitals are expected to purchase larger volumes, but may pass on less discount as they either have market power compared to health insurers or margins were necessary to finance researches. Future research with a different proxy of market power should prove whether this is the case. Results show that add-on medicines contribute to a larger significant price deviation compared to coagulation factors. And finally, evidence is found that price deviation is
significantly larger for medicines having a generic or biosimilar variant, compared to medicines not having a generic or biosimilar variant.
Keywords: purchasing groups, expensive medicines, medical specialist care, purchase discounts, bargaining power, Dutch healthcare market
Table of contents
1. Introduction……….. 6
2. Literature review……….. 8
2.1 Definition and price regulation………. 8
2.2 Purchasing groups……… 9
2.3 Market power and the healthcare market structure……….. 12
2.4 Empirical findings……… 13
2.5 Hypothesis……… 14
3. Methodology and data………. 15
3.1 Methodology………... 15
3.1.1 Data analysis……….. 15
3.1.2 Definition of the variables……….. 16
3.1.3 Assumptions………... 20
3.1.4 Statistical tools………... 21
3.2 Data………. 22
3.2.1 Qualitative data……….. 22
3.2.2 Quantitative data……… 24
4. Qualitative & Quantitative results………. 25
4.1 Qualitative results ……… 25
4.1.1 Results interviews Academic Hospitals ……… 25
4.1.2 Results interviews Top Clinical Hospitals………... 29
4.1.3 Results interviews Health Insurers………. 31
4.1.4 Similarities & differences hospitals and health insurers……… 35
4.2 Quantitative results ………. 37
5. Discussion & Conclusion………. 43
References………. 48
Appendix 1: 1A Hospitals and purchasing groups……….. 53
1B Interview guide Health Insurers……… 55
1C Interview guide Hospital pharmacists………... 57
Appendix 2: 2A Summary statistics………. 59
2B Statistics variable percentage……… 61
2C Statistics variable log percentage……….. 61
Appendix 3: Output results two sample t tests and one-way ANOVA……… 62
1 Introduction
According to the Central Bureau of Statistics (CBS), healthcare costs have been rising in the Netherlands the last few years and it is expected to increase until 2040 [45]. Besides that people are aging, the introduction and increasing demand of more expensive medicines, contribute to the increase in medicine’ costs within the total healthcare budget [43]. While the minister of Health, Welfare & Sports (hereinafter: HWS) decided that the expenditure of medical specialist care (hereinafter: MSC) is not allowed to increase by more than 1% per year, the costs of medicines within this segment are actually rising more. This is caused by expensive medicines, with an amount of half a million per patient per year. In 2014, costs of expensive medicines including coagulation factors were 1,7 billion, in 2015 1,8 billion and in 2016 these costs rose to an amount of 2 billion Euro [44]. On the other hand, the stake of these innovative expensive medicines led to patients living longer, having a better quality of life or more patients getting cured. Like for patients with lung cancer, cystic fibrosis and hepatitis-C [43].
One instrument to reduce health care costs, is to stimulate joint purchasing groups by health insurers and health providers [7]. The instruments that the House of Representatives of the Netherlands used in order to reduce costs and ensure accessibility of new medicines for all patients, are deficient now [46]. Because for instance, it involves quite large amounts for new medicines, leading to high health care costs. Besides, most of these new medicines does not yet have a cheaper and generic variant to exchange with. On the other hand, the Netherlands have a worldwide market share of 2% and so have little market power against the global
pharmaceutical companies and their product prices. Then to what extent are purchasing groups for the market of expensive medicines effective?
For the sustainability of a high quality and future-proof care, it is important to see in which way costs can be controlled without damaging the quality of care. Concerning purchases, potential savings can be achieved.
Very little research has been done on the topic of purchasing groups in the healthcare.
The Dutch Healthcare Authority (hereinafter: DHA) investigates twice a year at the request of HWS, the agreements made between health insurers and providers about expensive medicines (add-ons and coagulation factors) and what it means for the accessibility and affordability of care. They perform qualitative analyses based on opinions from providers and insurers [6]. Most studies have been performed in the United States. According to William O. et al., (1994), prior studies looked at hospitals’ supply items and discovered large price variances among hospitals for comparable or identical items [40]. But they did not agree on the effectiveness of group purchasing organizations due to inappropriate modeling of the explanatory factors. Cutler et al., (2000) [32] states that medical care organizations have bargaining leverage and therefore are significant in restraining medical care prices.
Moriya et al., (2010) analyze the relationship between insurer and hospital market concentration and hospital prices, while controlling for unobserved market effects [18]. They find that
increases in hospital concentration are not significantly related to hospital price increases, while increases in insurance market concentration are significantly leading to an decrease in hospital prices.
However, investigating the healthcare in Europe and especially in Germany, joint purchasing organizations exist. Although the German healthcare structure and the institutions are not completely comparable to the Dutch healthcare, it is estimated that approximately 80% of the health sector purchases jointly [42]. A research by Ecorys in 2012, confirms that savings in healthcare could be achieved by stepping up group purchasing [41].
This study investigates joint purchasing organizations in the Dutch healthcare and takes into account the actual agreed contract prices between hospitals and health insurers. In general, such data is often not publicly available and therefore studies are missing . Besides, the data is very recent as the time period is 2012 till 2016. As mentioned above, most research has been done in the United States while this study investigates the Dutch Healthcare market. Specifically, this study examines potential buyer power by different purchasing groups consisting of hospitals only and the purchasing combination of hospitals and an health insurer as well. In addition, a qualitative analysis as in conducting interviews, is done as well. The combination of a
quantitative and qualitative analysis provides an overall view about joint purchasing groups and the purchase of expensive medicines.
For the Authority of Consumers & Markets (hereinafter: ACM) it is of interest to investigate the group bargaining process. In 2016 they released a guideline [7] about the possibilities and boundaries for hospitals and insurers buying medicines for MSC jointly. However, a quantitative analysis of the effectivity of purchasing groups has not yet been performed. This research demonstrates whether purchasing groups were able to negotiate discounts by exerting buyer power against the monopolistic pharmacy and pass on to health insurers as in lower agreed contract prices. Because, if they are able to negotiate and pass on discounts, health premiums for insured patients might decrease and ACM could broaden their boundaries of purchasing jointly as it does not harm consumers.
To the Dutch Health Authority (hereinafter: DHA), this study seems relevant as well as it might contribute to their monitor ‘contracting and purchasing of medicines in the medical
specialization care’ (MSC). In which twice a year, agreements between insurers and health
providers as well as their implementation are monitored [8].
Furthermore, this research seems relevant for the Ministry of Health, Welfare and Sport (HWS), because if purchasing groups can exert buyer power and negotiate discounts leading to lower medicine prices, insured patients are benefited and purchasing groups could be encouraged at a higher (European) level. In this way the affordability of care can be guaranteed. Overall, this research may help the ACM and the DHA in monitoring the health care market.
By conducting this research, results can be used to understand cooperation’s and mergers more broadly in industries where prices are determined by negotiations between various sellers and buyers who act as intermediaries for the final consumer.
The aim of this research is to investigate purchasing groups in the Dutch healthcare market and especially for the market of expensive medicines. Having mentioned the contribution to the literature and the relevance of this study above, the accompanying research question is:
“Do purchasing groups in the Dutch Healthcare buying expensive medicines, lead to lower medicine prices for insured patients?”
In order to answer the research question, first of all interviews with health insurers and health providers are conducted to obtain more in-depth information about the purchasing process of joint purchasing groups. In addition, an empirical research is done for purchasing groups buying expensive medicines jointly and hospitals who buy individually and hence are not part of a purchasing group. A reason to buy individually is that interests of medical specialists may differ between hospitals which complicates forming a joint purchasing group. Purchasing groups consisting of hospitals only and the combination of insurer-hospitals are taken into account. Specifically, 76 out of 88 hospitals analyzed (86%) is part of a at least one purchasing group. The effect of a purchasing group on price is studied by computing for each different purchasing
group and for the group of hospitals that purchase individually, the price deviation for the period 2012-2016. Where price deviation is defined as the difference between the maximum price set by the DHA and the agreed contract price between hospitals and health insurers. Next to determining the effect of group purchasing on the selling price of medicines, some other independent variables are tested as well. These variables vary from medicine-specific variables as type of medicine segment, to hospital-specific variables as type of hospital.
This thesis is organized as follows. Chapter 2 provides a literature review and the hypothesis tested in this thesis. In chapter 3 a description of the methodology of this study is given as well as the data that will be used in this research is presented. In chapter 4 the results of the
interviews and the empirical analysis are presented. In the final chapter the implications of this research are discussed and a conclusion is given.
2 Literature review
More and more money is spent on expensive medicines, in particular medicines belonging to MSC. Because these medicines are difficult to make, e.g. cancer medicines, costs are increasing. 0.7 billion Euro was spent on expensive medicines in 2011, while in 2013 it increased to an amount of 1,5 billion [1]. To make sure that costs would not rise further and that care remains affordable to insured patients, health insurers and hospitals started purchasing expensive medicines jointly.
This section provides an overview of earlier research concerning purchasing groups for
expensive medicines in the Dutch healthcare. In the first paragraph, the definition as well as the price regulation concerning DBC-care products and expensive medicines is given. In the second paragraph, the current purchasing groups and their medicine segments of purchasing are
described. The third paragraph describes the healthcare market structure and the effect of market power. Finally, empirical findings about purchasing groups for expensive medicines and the effect on price is given.
2.1 Definition & price regulation
In the Netherlands, all health care services are generally subject to price regulation by the DHA. Since 2012, the funding of medicines for MSC is done per treatment with a performance system called the DTC-DOT-product structure (diagnosis treatment combination). This means that all medicines belonging to medical specialist care (MSC), are integral declared within a DTC-care product [2]. A DTC-care product exists of all activities and services provided by a hospital and medical specialist for a specific demand for care facility. So in principal, for the Dutch
healthcare market a one-tier funding system for all types of medicines holds [3]. However, because of the high costs of expensive medicines and the fact that costs vary greatly per insured patient, the DHA developed the add-on declaration system. Specifically, whenever medicines lead to inhomogeneity of costs for a specific DTC-care product, then the DHA determines that an additional add-on declaration holds for that specific medicine [3].
DTCs and other care products are divided into two segments, A and B. In segment A,
(maximum) prices are set by the DHA while in segment B the fees for medical procedures are freely negotiable between healthcare providers and insurers [3]. Meaning, that prices of medicines in segment A are regulated and insurers and provides could bargain for the actual contract price only till a certain amount, the maximum price. Therefore for most of the
healthcare services in this segment, competition takes place on other market aspects than price. Moreover, hospitals face a budget ceiling in segment A and cannot spend beyond that ceiling.
Contrary, in segment B there is no budget ceiling and insurers and providers compete on DTC prices (DHA, 2009). As a consequence, competition should lead to lower negotiated contract prices and more choice for insured patients.
Segment A mainly consists of add-ons, coagulations factors and other care products for which maximum prices hold. Once a year the DHA determines maximum prices based on the pharmacy list price (AIP) [33]. The AIP cannot exceed the maximum price determined by the Law of medicine prices (WGP) as of the fact of external reference pricing, prices are decisive for other countries in Europe [34]. For some care products, maximum prices are allowed to increase by at most 10% [3]. Before the transfer of medicines to hospital care, an add-on declaration for a specific medicine could have been issued when average costs of the combination of name of substance, dosage form and indication exceeded €10.000 Euros per patient per year. Since 2015, this criteria changed to €1000 Euros per patient per year [3]. Moreover, insurers and providers are allowed to submit a request for including or removing medicines on the add-on medicines list of the DHA.
The add-on medicine list provided by the DHA, consists of mainly expensive medicines1.
However, the add-on medicines list contains some relative inexpensive medicines as well. These relative inexpensive medicines concern medicines who were previously expensive when patented, and therefore put on the add-on medicines list, but for which patents got expired now and for which generic or biosimilar variants entered the market. These relative cheaper variants should lead to savings [4]. Medicine segments for which an extra add-on declaration holds, are medicines transferred to hospital care e.g. TNF-alpha inhibitors, oncolytic, growth hormones and other oncolytic medicines. Besides add-on medicines, coagulation factors also have an additional declaration system because of the high costs. As the DHA indicates in their monitor ‘contract and purchase of medicines in MSC’ [8], the definition of expensive medicines
concerns add-on medicines and coagulations factors belonging to medical specialist care (MSC) [8].
2.2 Purchasing groups for medicines
When examining the effectivity of purchasing groups for expensive medicines, a clear explanation has to be given about what entails (group) purchasing of care and in which segments do purchasing groups operate. Besides, the types of purchasing groups that exist in healthcare and the possibilities for future purchasing cooperation’s are described.
Medicine segments
Following a research by Strategies in Regulated Markets (SiRM) about strengthening the purchase of medicines, four different expensive medicine segments are distinguished [5]. Better known as the Vuvuzela distribution of expensive medicine segments2. In addition, the add-on
medicine list maintained by the DHA, distinguishes the same different medicine segments [8]. Figure 1 shows the Vuvuzela distribution. Segment 1 contains patented monopoly medicines that do not have therapeutic alternatives, while segment 2 contains patented oligopoly medicines that do have some therapeutic alternatives like in the case of coagulation factors. Because there are no or few alternative medicines to exchange with, no full competition exist and medicines face high prices [5]. Contrary, segments 3 and 4 contain medicines in competition and
1
https://www.nza.nl/zorgonderwerpen/zorgonderwerpen/ziekenhuiszorg/Add-on-geneesmiddelen-en-stollingsfactoren/
2
source medicines respectively. These medicines are patented but have generic or biosimilar variants, like TNF-alpha inhibitors, and therefore face higher competition.
Figure 1: Vuvuzela distribution of medicine segments
source: Farma actueel1
For segments 3 and 4, substitution possibilities are available and therefore price negotiations occur, which in many cases results in purchase discounts and hence lower medicine prices[6]. For monopoly medicines, health providers do not have bargaining power since these expensive medicines are automatically added to the insurance package and volume shifts are not possible due to the lack of alternatives [6]. Therefore no price-effect is expected when purchasing groups buy medicines from this segment jointly instead of buying individually, and so the segment is left out in this research. Contrary, according to SiRM [5], for purchasing groups buying one of the other three segments jointly, negotiations are expected to be effective as in bargaining discounts, since larger volume shifts of these expensive medicines take place.
Types of purchasing groups
Different joint purchasing groups for the purchase of medicines belonging to the MSC are possible. For example, purchasing groups consisting of hospitals only, hospitals together with an health insurer or health insurers together [7]. Besides, care market participants as scientific societies can be part of joint purchasing as well. In this study, twelve different purchasing groups are analyzed. Eleven purchasing groups consisting of hospitals only and one purchasing group consisting of hospitals and an health insurer. Since 2012, a purchasing combination of Hospital Pharmacies of Academic Hospitals only (IZAAZ) exists who purchase expensive medicines jointly to achieve lower prices [11]. The purchasing combination of top-clinical hospitals called Santeon, buys biological medicines jointly while another purchasing group, consisting of health insurer Achmea and twelve hospitals, buys TNF-alpha inhibitors together since 2015. Table 1a of appendix 1, shows the hospitals and the different purchasing groups. It seems that 76 of 88 hospitals (86%) are part of at least one purchasing group. For twelve hospitals it was unobserved whether they join a purchasing group and therefore a thirteenth group was created with hospitals that purchase individually. Often, regional buying groups are formed like IGZ Friese, Haagse and Leiden/Rijnmond. Depending on the treatments that hospitals provide, purchasing groups can consist of academic hospitals only, like IZAAZ, top-clinical hospitals only, like Santeon, or a combination of for example, top-top-clinical and general hospitals like IZON. Moreover, hospitals could be part of different purchasing groups like MC Erasmus or Gelderse vallei. The DHA concludes in their two monitors of 2016 [6,8] ,that most purchasing groups of expensive medicines consist of health providers only (79% and 84% respectively) whereas 14 hospitals (21%) purchase jointly with insurers. One reason for this could be a lack of trust as insurers are not informed about the purchasing prices of hospitals. Or hospitals will not benefit if purchasing one segment jointly can be harmful for the purchase of
other segments [6]. Contrary, some hospitals (16% in 2016) do not join a purchasing group at all [6]. In this study, 14% of all hospitals analyzed purchase individually (see table 1a of appendix 1). A reason to purchase individually, is that medical specialists of different hospitals have different interests in prescribing the preferred medicine which could be a problem if the purchasing group becomes too big. Besides, it could be the case that hospitals are already able to negotiate large purchase discounts from the pharmacy themselves, whereas if they join a purchasing group, this discount has to divided among the involved parties.
The purchasing groups mainly use their countervailing buyer power with the purchase of add-on medicines like TNF-alfa inhibitors, coagulation factors or generic or biosimilar variants [9]. The DHA confirms this and shows that in 2016 for generic add-ons, medicines in competition and coagulation factors, 60% to 70% of the hospitals were able to negotiate lower purchasing prices than before [6]. Despite the fact that biosimilars cannot be substituted automatically by medical pharmacists and the price difference between the biological branded medicine and the biosimilar variant is lower compared generics, interviewees by SiRM expect a large negotiated discount as well [10].
It is important that insurers and hospitals make proper contract agreements about the costs of expensive medicines as since 2015 insurers face more risk. This increase of risk-bearing is a result of not being compensated anymore for unexpected high costs of add-on medicines. But since many insurers introduced a maximum budget for add-ons to hospitals, meaning that unexpected high costs must be financed by hospitals themselves, the risk is actually on
hospitals. Since hospitals are not able to predict future costs of expensive medicines, it could be the case that insured patients are asked to switch hospitals if the maximum budget for a specific hospital is already met. This could harm the quality of care. On the other hand, since unexpected costs are not compensated afterwards and the risk is on hospitals, hospitals should have an incentive to buy efficiently and negotiate discounts.
Instruments to encourage purchasing groups
To encourage the wide use of purchasing groups and to make collaborations between health providers and health insurers easier, some instruments are implemented or expected to happen in the near future. In June 2016, ACM released a guideline to inform providers and insurers about the opportunities within the framework of competition rules to purchase expensive medicines jointly [7]. To test whether purchase combinations do not harm competition, three rules apply. Costs of jointly purchased medicines cannot exceed 15% of a hospitals’ turnover while the limit of 5% holds for insurers. Besides, without objective and non-discriminatory criteria, entry barriers to potential incumbents of purchasing groups are not allowed. Finally, the purchasing combination must be sufficiently flexible for maintaining incentives to purchase efficiently [7].
Another recent instrument to optimize purchasing groups and the purchase of expensive medicines, is the platform of purchasing power for expensive medicines established by the minister of Health, Welfare and Sports (HWS) [35]. The idea is to address the problems and turtlenecks arising from current purchasing groups. Finally, to ensure sustainable access to innovative medicine at affordable cost for insured patients, the collaboration of BeNeLuxA (Belgium, The Netherlands, Luxembourg and Austria) has been launched recently. The
collaboration will be used for pricing and reimbursement, including joint price negotiations and horizon scanning of new expected expensive medicines. These medicines can be taken into account during purchase negotiations with pharmaceutical companies [37].
2.3 Theoretical model: The effect of market power – healthcare market structure – joint purchasing power
To investigate the effectivity of purchasing groups in healthcare, as in negotiating discounts and lowering contract prices of expensive medicines, first, market power in general and the
healthcare market structure is described. Market power
Market power arises when for a longer period of time, the buyer or seller has the ability to set prices above or below the perfect competition level. To investigate whether competition is restricted, hence market power arise, the market definition has to be determined [38]. This means defining the relevant product and geographic market in which the organizations are competing. However, problems arise in the market for medical specialist care (MSC). It assumes some price sensitivity of patients, assumes patients to make rational decisions ex-ante as to which hospital and which doctor provides the best care, and it assumes patients to make their own decisions [38].
The DHA defines purchasing power as the ability for a buyer to act to a certain extent independently from the seller [11]. Contrary, for supplier power the opposite holds. The existence of purchasing power in healthcare could decease consumer welfare if benefits of lower negotiated prices are not invested in healthcare [11]. Purchasing power from both the hospital and health insurer should be tolerated if the negotiated price discount at the pharmacy is, in some way, passed on to the insured patient [7]. This means passed on as in lower health premiums or better health plan terms.
The ability for hospitals to exercise bargaining power, and hence market power, means the ability to add incremental value to an insurance plans’ network [17]. Bargaining power is high (low) when there are less (many) substitutes available for a certain hospital.
Health insurers’ market power means the ability to direct insured patients to preferred providers [4]. Also known as the process of selective contracting. By selective contracting, a
concentration of care arises which should lead to better quality of care when the volume of treatments increases [10]. However, research shows that due to, among other things the lack of quality indicators of many treatments, incentives for insurers to channel patients to preferred providers are limited [39].
Joint purchasing power
Joint purchasing power can be achieved by forming purchasing groups [7]. As already mentioned in the paragraph above, these purchasing groups can consist of hospitals only or a combination of hospitals and an health insurer. The idea is that by forming purchasing groups, larger volumes of medicines can be bought at discount by exercising countervailing power against the pharmacy [12]. Hospitals receive this purchase discount and subsequently act as a seller of care when hospitals and health insurers negotiate contract prices for these medicines. The idea is that the negotiated purchase discounts at the pharmacy are to a certain amount passed on to health insurers as in lower agreed contract prices. However, the ACM noticed that hospitals and insurers are reluctant to form purchasing groups and buy medicines jointly, even if it is in the interest of patients and insured premium holders [12]. This could be due to the fact that in the past, health providers got fined for alleged cartel agreements [13] and nowadays are scared of violating competition rules. But ACM sees no harm when purchasing groups operate within certain boundaries. As in 2016, ACM introduced their guideline [7] to stimulate forming purchasing groups for medicines belonging to MSC. The European Commission (EC) already mentioned in their guidelines of horizontal cooperation, that purchasing groups with a market share of 15% and lower, will not restrict competition.
As Farma adds [9], purchasing groups would not pose significant market power as
pharmaceutical companies operate all over the world and the Netherlands only accounts for a small share (2%).
2.4 Empirical findings
There is substantial literature on the relationship between market concentration and market price. Within the healthcare market, previous studies focused on analyzing market power on only one side of the market; health insurers’ bargaining power or hospitals’ market power [14, 15]. However, since information about negotiated reimbursement rates between hospitals and insurers is sensitive, data on these rates is not publicly available and therefore empirical research of variation in rates across payers is rare.
Empirical research
Most research is done in the United States. For example, when looking at the relationship between health insurers’ buyer power and prices, most researchers find a negative relationship. Meaning that an increase in insurer size, hence more monopsony power, is associated with higher negotiated discounts [16,17]. Higher negotiated discounts indicate that insurers are more able to channel their patients to preferred hospitals. To analyze the effect of insurer and hospital market power on negotiated prices, [18] use a reduced-form structure-conduct-performance (SCP) method. Market concentration is measured by Herfindahl indices (HHI) and they find a negative significant relation as well between insurer concentration and hospital prices.
Contrary, a positive relationship is found between hospital concentration and market price [18,19, 20]. Capps and Dranove (2004) [21] support this result and by using multivariate regression analysis, they find in three of the four markets studied, a positive significant relation. They conclude that antitrust scrutiny of hospital concentration is warranted. Croes et al,. (2017) [22] adds that, in an environment of liberalized pricing, hospitals in concentrated markets have lower quality scores for two of the three diagnosis groups studied than hospitals in competitive markets. Meaning that less competition, hence hospitals having more market power, is
associated with lower quality scores.
Other studies that analyze mergers, find that market concentration resulting from a merger of insurers is positively related to premiums [23, 24]. Indicating that cost-savings through lower negotiated prices after a merger, are not passed on to their enrollees.
Analyses relating the effectiveness of group purchasing defined as minimal purchase price, show that prices paid by hospitals who are part of a purchasing group are lower compared to hospitals not joining a purchasing group [25, 26, 27].
However, results should be taken with caution as these prior studies did not allow any factors for differences between the different purchasing groups. Besides, the healthcare system in the United States is different than in the Netherlands and it is questionable whether these studies are applicable to the Netherlands.
Recent research in the Netherlands
Since 2016, twice a year the DHA monitors the contracting process and the purchase of medicines for MSC [8]. They perform a qualitative analysis of which a survey has been set out and interviews are conducted. Besides, because the affordability and accessibility of new expensive medicines is under pressure, they monitor the development of medicine costs in MSC [28]. Overall, they conclude that the expenditure of expensive medicines is still rising, of which an increase from €1.73 to €1.76 billion between 2014 and 2015. Medicine segments as oncolytic and TNF-alpha inhibitors constitute the biggest part in 2015, €727.1 and €567.2 million
discounts through purchasing groups for medicines with competition. Partly caused by the fact that insurers are risk-bearing for add-ons since 2015 [6].
A recent study by ACM (2017), about price and volume effects of 12 hospital mergers, shows an increase in price for merged hospitals compared to non-merged hospitals. However, there are limited indications that volume of the merged hospitals developed differently than the volume of non-merged hospitals [47].
Contribution to the literature
The contribution of this study to existing literature is as follows. It is best thought of a combination of qualitative research and an empirical investigation of the idea that purchasing groups exert bargaining power and are able to negotiate and pass on discounts as in lower contract prices compared to non-purchasing groups.
First, interviews are conducted with both parties responsible for the contracting process of medicines. Subsequently, the effect of different purchasing groups consisting of hospitals only or a combination of hospitals and insurer on expensive medicine’ prices is analyzed. Whereas most aforementioned studies investigates the exercise of market power on only one side of the market. Second, actual contract prices are used. These prices are determined by the bargaining process between hospitals and insurers and normally are not publicly available. Third, most studies about the effects of market reforms are performed in the United States. The Dutch Healthcare market differs as providers and insurers are operating under more regulation [29]. Furthermore, as time passes and more data becomes available, the analysis can be extended over a longer period of time. Besides, it is expected that more purchasing groups will be formed or extended in the (near) future as the HWS recently launched a platform [30] to optimize purchasing groups and make expensive medicines affordable and accessible. In that case, this analysis provides a starting point for investigating the relation between purchasing groups buying expensive medicines and the price of medicines.
2.5 Hypothesis
The aim of this study is to analyze the effectivity of different purchasing groups consisting of hospitals only or the combination of hospitals and a health insurer. Where effectivity is defined as whether purchasing groups are able to exert bargaining power against the pharmacy and negotiate discounts in favor of the insured patient as in lower medicine contract prices. Besides, it is investigated if the negotiated discount for segments of medicines changed during the assessed period. The different purchasing groups analyzed are described in paragraph 2. The hypotheses are related to all different purchasing groups and created to examine the effect of purchasing groups on the sale price of expensive medicines (H2 and H3), against the effect of hospitals not joining a purchasing group but purchase individually (H1). In this way, the main research question can be answered. The expectation of the first hypothesis is that when hospitals negotiate and purchase medicines from the pharmacy individually, they do not have the benefits of a purchasing group making use of volume shifts and exerting great bargaining power.
Besides, it has only been a few years since hospitals are, to some extent financially responsible for the costs of medicines and therefore an incentive to purchase efficiently can be lacking [31].
H1- Hospitals not joining a purchasing group but who buy individually, are not able to negotiate discounts and therefore lower sales prices are not expected
For the second hypothesis, it holds that hospitals purchasing medicines jointly, are able to exert some bargaining power and volume shifts take place. Volume shifts can lead to purchase discounts from the pharmaceutical company a greater market is created for its medicine.
Nevertheless, hospitals joining a purchasing group with an health insurer will on average lead to the lowest medicine´ sales prices (H3). This is based on the following facts. From the literature, and mentioned in previous paragraph, it seems that the purchase price paid by hospitals who are part of a purchasing group is lower compared to hospitals who buy individually [25, 26, 27]. Cutler et al., 2000 [32] adds that the involvement of health insurers in a managed care organization with preferred health providers, leads to significant lower medical care prices because of bargaining leverage [32]. In the Dutch healthcare market, insurers have a duty of care towards their insured patients and compete with each other for insured patients by offering lower premiums or better contracts terms [11]. Besides, they are risk-bearing and therefore have an incentive to achieve purchasing benefits. Furthermore, since the reforms of the Dutch
healthcare sector in 2006, selective contracting of providers by health insurers takes place. By selective contracting, the idea is to stimulate competition between providers which should lead to strengthening their incentives to improve efficiency. This may improve their incentives to achieve purchasing benefits and hence negotiating discounts [2].
However, to what extent lower medicine prices are negotiated is not only determined by the fact that purchasing groups have the ability to buy larger volumes and exert countervailing power compared to hospitals not joining a purchasing group. It depends on the market power of pharmaceutical companies as well [8].
H2- Hospitals joining purchasing groups with other hospitals only, are able to negotiate discounts and therefore lower sales prices are achieved
H3- Hospitals joining purchasing groups with an health insurer will on average achieve the lowest sales prices
3 Methodology and data
In order to test the effectivity of different purchasing groups for the market of expensive medicines belonging to medical specialist care (MSC), a qualitative analysis combined with a quantitative analysis has been performed. Where MSC means, curative care and other specialist care that is provided by the hospital sector, like add-on medicines and coagulation factors [48]. Effectivity is measured as the ability of hospitals to negotiate discounts with pharmaceutical companies, and pass on discounts to health insurers as in lower sales prices. In the dataset, these lower sales prices are the agreed contract prices between health insurers and hospitals that are negotiated each year, for a specific DTC. By conducting semi-structured interviews and the consultment of a panel dataset provided by Vektisconsisting of contract prices, data is collected and analyzed respectively.
3.1 Methodology
3.1.1 Data analysis
To obtain the effects of different purchasing combinations on price, a dataset consisting of DTC contract prices between hospitals and health insurers for the period 2012 to 2016 is analyzed. Due to the lack of quality data, a price-effect analysis is performed only. Contract prices
regarding DTCs, provided by Vektis was the most important source. Neither purchase discounts that hospitals negotiate with pharmaceutical manufacturers, nor purchase volumes or costs are observed, since that kind of data is not publicly available.
This study focuses on declaration codes (DTCs) that belong to expensive medicines for MSC, and the agreed contract prices for these DTCs between hospitals and health insurers. Following
the DHA, expensive medicines regarding MSC can be divided in add-on medicines and coagulations factors [6]. All other DTCs that belong to non-expensive medicines, are not taken into account. According to ACM (2016), medicines that belong to MSC and for which
purchasing groups are allowed to collaborate, are add-on medicines and other medicines for which reimbursements are based on the Health Insurance Act [7]. The database of Z-index3,
adds that both add-on medicines and coagulation factors fall into the category of MSC [49]. These are medicines for which maximum prices have been set by the DHA4. Therefore, add-on
medicines and coagulation factors that belong to MSC are analyzed in this study. As this study focuses on expensive medicines, DTCs other than add-on medicines and coagulation factors5
have been removed from the dataset, leading to 1023867 observations instead of 995500. In order to test whether purchasing groups passed on the negotiated discounts at pharmacy to health insurers as in lower sales prices, price differences with regard to maximum prices set by the DHA are calculated.
3.1.2. Definition of the variables
Dependent variable: price deviation
As mentioned above, the DHA sets maximum prices for DTCs belonging to add-on medicines and coagulation factors. In paragraph 2.1 it is explained that DTCs leading to inhomogeneity of costs, receive an additional acquirement, an add-on declaration, from the DHA [3]. Specifically, add-on medicines and coagulation factors are medicines that are charged separately from the treatment, in addition to its DTC product. These add-on medicines and coagulation factors are often expensive medicines [50].
For the different DTCs, health insurers negotiate contract prices with hospitals and these contract prices for the period 2012-2016 are part of the dataset. The negotiated contract price between health insurers and hospitals can be the DHA maximum price or lower. Most hospitals are part of a purchasing group and buy different medicines at pharmaceutical companies jointly, as for larger medicine volumes discounts can be negotiated. Subsequently, after hospitals negotiating purchase discounts at pharmaceutical companies, health insurers negotiate contract prices with hospitals. The idea is that when the negotiated contract price between health insurers and hospitals equals the DHA maximum price, no purchase discount could have been realized at the pharmacy and passed on to the health insurer. Because if hospitals were able to negotiate purchase discounts at the pharmacy, it is expected that to some extent benefits are passed on to health insurers as in contract prices lower than the DHA maximum price. Hence, a positive price difference indicates that a hospital was able to negotiate a purchase discount for a certain DTC and passed on the benefits to a particular health insurer.
The relative price deviation is then the DHA maximum price minus the negotiated contract price:
𝑃𝑟𝑖𝑐𝑒𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛𝑑ℎ𝑖𝑡= 𝑁𝑍𝑎 max 𝑝𝑟𝑖𝑐𝑒𝑑𝑡− 𝑐𝑜𝑛𝑡𝑟𝑎𝑐𝑡𝑝𝑟𝑖𝑐𝑒𝑑ℎ𝑖𝑡 (1)
Where d refers to a DTC, h to a hospital, i to a health insurer and t to a specific year. The price deviation variable contains many zeros, which means that many contract prices equal the maximum price set by the DHA. Therefore its distribution is skewed to the right and not normally distributed.
3 As an intermediary in customized healthcare information, Z-index offers products and services that are tailored to
the needs of the healthcare market
4 CI-15-48c Circulaire besluit regelgeving 2017 add-on geneesmiddelen en stollingsfactoren 5 For example first-line diagnostics as laboratory tests and paramedical services and researches
The percentage of price deviation with respect to the maximum price is calculated; 𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒𝑑ℎ𝑖𝑡=
𝑝𝑟𝑖𝑐𝑒 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛𝑑ℎ𝑖𝑡
𝑁𝑍𝑎 max 𝑝𝑟𝑖𝑐𝑒𝑑𝑡 ∗ 100 (2)
followed by a natural log-transformation in order to make the dependent variable less skewed. A log-transformation of the dependent variable is necessary as analyses with a non-normal distribution will yield biased results. The dependent variable used becomes:
𝐿𝑛𝑝𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒𝑑ℎ𝑖𝑡= 𝑙𝑛(1 + 𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒𝑑ℎ𝑖𝑡) (3)
Independent variable: purchasing group dummies
The main explanatory variable for hospitals to negotiate and pass on discounts as in lower contract sales prices, hence a positive expected price deviation, is operating in a purchasing group instead of purchasing expensive medicines individually. In this study, 12 different purchasing groups are examined; 11 purchasing groups that consist of hospitals only and 1 purchasing group that consists of hospitals and health insurer concern Achmea. For 12 of 88 hospitals (14%) (see table 1a, appendix 1), it was unobserved whether they join a purchasing group and therefore a 13th group, a control group, has been created. In this group it is assumed
that hospitals buy medicines individually. To correct for characteristics of the different
purchasing groups, purchasing group dummies are created. Besides, a dummy for the group of hospitals buying expensive medicines individually, is created.
The expected direction of the coefficient for the variable purchasing group, is positive for all years. As hospitals and health insurers can increase their bargaining strength by purchasing larger volumes of expensive medicines if they form a joint purchasing group, higher purchase discounts are expected and therefore lower sales prices compared to hospitals who purchase individually. Therefore, a positive price deviation is expected. With the contribution of a health insurer to a purchasing group consisting of different types of hospitals, e.g. the Achmea
purchasing group, a larger positive price deviation is expected. Health insurers compete with each other for subscribers and act as cautious buyers of care on behalf of their insured patients and so have a duty of care towards their insured. Therefore insurers have an incentive to negotiate contract prices as low as possible. To keep medicines affordable and accessible, Achmea has created a purchasing group with other hospitals with the same goal of saving expensive medicines’ costs by purchasing medicines jointly [9]. In addition, a clear agreement has been made on the distribution of purchase savings. 1/3 must be passed on to insured patients via premiums, 1/3 is spent on innovation and the remaining part goes to the participating
hospitals [9]. For purchasing groups without health insurers, it is unclear whether hospitals pursue such distributions of purchase savings as it cannot be monitored by health insurers. This leads to the following hypotheses that were described in paragraph 2.5 as well:
Hypothesis 1: Hospitals not joining a purchasing group but who buy individually, are not able to negotiate discounts and therefore lower sales prices are not expected
Hypothesis 2: Hospitals joining a purchasing group with hospitals only are able to negotiate discounts and therefore lower sales prices are achieved
Hypothesis 3: Hospitals joining a purchasing group with an health insurer concern will on average achieve the lowest sales prices
In addition, dummies for the different purchasing groups are included to correct for
characteristics for a specific purchasing group. Characteristics could be the purchasing policy for a certain purchasing group or the team that purchases medicines for MSC. By conducting
interviews, it became clear that not all purchasing groups purchase the same medicine segments jointly, hence they specialize and therefore different price deviations are expected.
Other important variables
There are several variables that may influence the contract price for a certain DTC as well. This includes hospital and DTC-specific variables as well.
Year dummies
In this study, purchasing groups are analyzed over a 5 year period. Dummies are created for years 2012 to 2016. For all years, positive coefficients are expected with an increase in time. Since 2014, expensive medicines have been brought to the attention in the health sector. In 2014 and 2015, KWF released reports6,7 about the accessibility of expensive cancer medicines and the
bottlenecks in the near future. They show that costs of expensive medicines will continue to rise due to the increase of new medicines and the increase in the number of patients. In addition, pharmaceutical companies keep asking exorbitant medicine prices and therefore they argue for a different funding system3. Moreover, in 2016, ACM issued a guideline on the possibilities of
joint purchasing of medicines for MSC. ACM sometimes noticed some reluctance to cooperate in the procurement of medicines for MSC [7]. Due to the reports of KWF and the guideline of ACM, social interest of keeping medicines affordable and accessible has increased and therefore a positive price difference is expected with an increase in time.
Hypothesis 4: Over the period 2012-2016, a positive trend in price deviation is expected
Health insurer concern dummies
In the Dutch health insurance market, 24 health insurers labels are active which are divided over 9 different health insurer concerns 8. 4 small health insurer concerns (ASR, ONVZ, ENO and
Zorg&Zekerheid) are part of Multizorg and the 5 other concerns are Achmea, VGZ, Menzis, CZ and DSW. Health insurers belonging to these different health insurer concerns, negotiate
contract prices for each DTC for each year with hospitals. As mentioned in the literature, hospitals compete with each other for contracts with health insurers and therefore differences in hospital-health insurer combination are expected. Therefore, at the level of hospital, health insurer and for a certain DTC, 6 health insurer concern dummies are created. A positive price deviation for all health insurer concerns is expected. As explained above, health insurers compete for subscribers and have an incentive to negotiate lower contract prices – i.e. prices that are further away from the DHA maximum price. However, health insurers differ in market share, hence in bargaining power and therefore the magnitude of the price deviation may vary between health insurer concerns.
Hospital-specific variables
Two hospital-specific variables that do not change over time but who characterize hospitals, and hence purchasing groups, are included, since they may affect price deviations for a certain DTCs. Annual reports of the different hospitals were consulted to obtain the following two hospital-specific variables.
6 KWF Kankerbestrijding (2015) Effectieve nieuwe middelen tegen kanker, maar het financieringssysteem kraakt -
Belemmeringen en oplossingen bij de inzet van dure geneesmiddelen tegen kanker.
7 https://www.kwf.nl/over-kwf/Pages/SCK-rapport-dure-geneesmiddelen.aspx 8 https://www.zorgwijzer.nl/zorgverzekering-2016/infographic-zorgverzekeraars
Type of hospital dummies
The Dutch hospital market consists of about 90 hospitals which can be divided in three types of hospitals; academic, top clinical and general hospitals [51]. The sample size of this study contains 8 academic hospitals, 24 top clinical hospitals, 55 general hospitals and 1 foundation called Cello which belongs to hospital Jeroen Bosch. Due to hospital mergers and acquisitions, the amount of hospitals vary. The different types of hospitals all have their own purchase policy and specialization in type of care they provide, hence differ in the purchase of medicines. Differences in the effects of the coefficients of the dummies on price deviation are expected. In general, academic hospitals carry out complex care and these large hospitals receive patients from a wide region compared to smaller, general hospitals [51]. On the other hand, top clinical hospitals specialize in one or more healthcare areas and general hospitals treat patients with more common diseases. Based on several interviews, purchasing groups differ in the types of medicine segments and the amount of medicines that are purchased. Besides, purchasing groups differ in how long they have been cooperating. Purchasing groups Santeon and Izaaz, consisting of top-clinical hospitals and academic hospitals respectively, already purchase medicines jointly for around 30 years.
Hypothesis 5: Academic hospitals joining a purchasing group will on average achieve the lowest sales prices compared to top clinical and general hospitals
Size hospital
The size of the hospital is measured by the number of beds. Specifically, the total bed capacity of a hospital. This hospital-specific characteristic corrects for the size of the hospital and the association with the possibility of economies of scale. Hospitals with a large bed capacity, might purchase larger volumes and therefore higher discounts can be negotiated with pharmaceutical companies.
Hypothesis 6: The greater the size of the hospital joining a purchasing group, the greater the negotiated discounts and therefore on average the lower the sales prices
DTC-specific variables;
Several DTC (medicine) –specific variables that do not change over time but who characterize medicines and may affect prices, are included as well.
Type of MSC dummy
DTCs for MSC are divided in add-on medicines and coagulation factors. As mentioned in the literature, add-on medicines are registered differently than DBC medicines because of the high costs [6]. Higher costs for a certain DTC means higher contract prices and therefore a lower price deviation for this particular DTC. On the other hand, based on several interviews,
hospitals were able to negotiate high purchase discounts for coagulations factors because of the introduction of alternatives, biosimilars. Where biosimilar is defined as a biological medicine, highly similar to the original biological reference medicine of which the patent has expired [6]. In 2016, these high negotiated purchase discounts at the pharmacy resulted in lower agreed contract prices between health insurers and hospitals [6]. The following hypothesis holds.
Hypothesis 7: Higher discounts can be negotiated for coagulation factors and therefore on average lower sales prices are achieved compared to add-on medicines
Type of segment dummies
Based on DHA documents, DTCs for add-ons and coagulation factors could be subdivided into the following segments; TNF-alpha inhibitors, fertility hormones, growth hormones,
immunoglobin, oncolytic, other oncolytic and coagulation factors. The more complex the medicine, like oncolytic, the fewer alternative medicines exist and therefore higher costs and higher contract prices are expected. According to several interviews and the monitor from the DHA of 2016, purchasing groups are able to negotiate purchase discounts for DTCs having alternative medicines like TNF-alpha inhibitors, coagulation factors and/or biosimilars/generics. The existence of alternative medicines allow hospitals to negotiate with pharmaceutical
companies as hospitals can choose to switch medicines (substitution possibilities). For unique medicines without alternative medicines, like oncolytic, pharmaceutical companies have market power in setting a price and therefore purchase discounts are not expected. Therefore the following hypothesis hold:
Hypothesis 8: For TNF-alpha inhibitors and coagulation factors, highest discounts can be negotiated and therefore on average lowest sales prices are achieved
Generic variant dummy
A dummy for DTCs having a generic variant is created to make a distinction between DTCs having a generic variant and DTCs not having a generic variant. Where generic medicine is defined as the therapeutic equivalent of an original branded medicine of which the patent has expired [6].Basically, a generic medicine is equivalent to the original branded medicine as both medicines have the same active ingredients [6]. For DTCs whose patents have expired and generic medicines have entered the market, substitution is possible between the original branded DTC and the generic variant. This results in hospitals having bargaining power against
pharmaceutical manufacturers in terms of the price of medicines. DTCs with a generic variant will therefore lead to lower contract prices resulting in positive price deviations compared to DTCs not having a generic variant.
Hypothesis 9: DTCs having a generic variant will lead to higher discounts and therefore on average lower sales prices compared to DTCs without a generic variant
Biosimilar variant dummy
Having explained the definition of biosimilars under the type of MSC, a same positive price deviation as for DTCs having a generic variant is expected. For DTCs whose patents have expired and biosimilar variants have entered the market, substitution is possible between the original branded DTC and the biosimilar variant. Therefore the following hypothesis holds.
Hypothesis 10: DTCs having a biosimilar variant will lead to higher discounts and therefore on average lower sales prices compared to DTCs without a biosimilar variant
3.1.3 Assumptions
Due to the lack of the following information, some assumptions are made in order to analyze the data.
First of all, in general specific information, e.g. purchasing volumes and the types of medicines purchased, about the different purchasing groups is not publicly available. In this study, 1 purchasing group with health insurer Achmea and 12 purchasing groups consisting of hospitals only are formed and analyzed. However, the first weakness of the study is that based on several
interviews, it appears that almost every hospital participates in a purchasing group for at least one medicine segment. Moreover, it may happen that hospitals join several purchasing groups for different medicine segments which could differ over the period 2012-2016. How many purchasing groups actually exists and which medicine segments they purchase, is unknown. Based on information provided by ACM, in this study it is assumed that 12 purchasing groups are active. The remaining 12 hospitals of table 1a in appendix 1, are assumed to purchase individually and are called the non-purchasing group. However, it may happen that a number of hospitals that are assigned to the non-purchasing group, actually are part of one or more
purchasing groups. In addition, the assumption holds that all purchasing groups buy the different medicine segments as defined above, while the purchase of medicine segments might differ between purchasing groups. For example, purchasing group Izaaz consists of academic hospitals only while purchasing group Midden-NL consists mainly of general hospitals. The hospitals differ in the kind of the treatments they provide and therefore the type of medicines they need and purchase as a purchasing group.
In the ideal situation, information about which DTCs belong to which medicine segments that each purchasing group buys jointly, should be taken into account. In addition, all purchasing groups must be included in the analysis.
Furthermore, dummies for having a generic or biosimilar variant are included. The assumption of a DTC having a generic or biosimilar variant holds for the entire period, 2012-2016.
However, an exact date of when the generic/biosimilar variant entered the market is more reliable. Besides, information about DTCs having more than one generic or biosimilar variant is missing as well as the substance names of these particular variants.
Finally, contracts are based on price and volume agreements but information about purchase volumes of the different hospitals within a purchasing group is not observable. In the ideal situation where purchase volumes are observed, the contribution of the different hospitals within a purchasing group to price deviations can be calculated. Besides, purchase volumes can be used as a proxy of market power concerning the different purchasing groups for a certain medicine segment. Moreover, by including purchase volumes, volume effects of lower negotiated contract prices can be examined.
3.1.4 Statistical tools
In this study, different types of statistical tools are used to compare price deviations across different purchasing groups and the performance of the variables mentioned above, on price deviation. Specifically, one-way ANOVA and two-sample t tests are used in order to answer the hypotheses. This section briefly explains the tools used in this study.
The one-way analysis of variance (ANOVA) and t tests are widely used to compare the means of population groups. One-way ANOVA determines whether the mean of the dependent variable is the same in two or more independent unrelated groups [52]. Yet, one-way ANOVA is typically only used when there are three or more independent, unrelated groups and
independent-samples t tests are commonly used for comparing two population means. For example, the comparison of the means of salary of five different types of degree types. Where degree types are defined as e.g. psychology, biological sciences, business studies, history and law. While the statistical methodology is the same, ANOVA uses F statistic and two-sample t statistics use its t statistic and p-value to test if all groups have the same mean [53]. If
statistically significant differences between groups exist, post hoc tests (e.g. Bonferroni & Tukey) are used to examine which groups were significantly different from each other. Some assumptions must be met in order to run one-way ANOVA. If the assumptions are not met, one-way ANOVA provide invalid results which cannot be interpreted. One of the
of the independent variable [54]. In this study, meeting this requirement could be problematic as the dependent variable price deviation is right-skewed due to many contract prices that equal the maximum price set by the DHA. In order to obtain an approximate normal distribution, the dependent variable price deviation is transformed into a natural logarithm percentage as explained above. To investigate whether the transformed dependent variable has a more normal distribution and is an appropriate dependent variable, summary statistics of skewness and kurtosis are provided in table 2b and 2c in appendix 2. Summary statistics of both variables
percentage and ln percentage, show skewness and kurtosis in its distribution. Indicating that
both variables are not approximately normally distributed but right-skewed. However, by transforming the dependent variable percentage into a natural logarithm, both skewness and kurtosis decrease dramatically. The natural logarithm transformation of the dependent variable
price deviation, leads to a more approximate normal distribution and therefore the variable lnpercentage is a more appropriate dependent variable and will be used in the analyses. Another
reason why one-way ANOVA can still provide valid results when the dependent variable is not normally distributed, is a large sample size. This dataset contain 1,023 million observations for the period 2012 to 2016 of which 96 hospitals are divided over 13 groups; 12 purchasing groups and 1 individually-purchasing group. According to the central limit theorem, normal probability calculations can still be used when its sample size is sufficiently large, even when the
population distribution is not normal [52]. Moreover, ANOVA is quite robust to normal
distribution violations, meaning that ANOVA still provide valid results when the requirement is a little violated [54]. Therefore, given the above reasons, ANOVA tests can be performed. Another assumption of the one-way ANOVA test is that there needs to be homogeneity of population variances as a pooled standard deviation is used. If the equal variance requirement is violated, individual sample and Satterthwaite’s or Welch’s approximation of the degrees of freedom have to be used [53]. Nevertheless, the hypothesis remains the same.
The following structure of answering the hypotheses is maintained during this study.
For the first hypothesis, a t test has been performed to determine whether there is a statistically significant difference in price deviation between the group of hospitals who purchase
individually and the hospitals who are part of a purchasing group. One-way ANOVA is then used to determine whether statistically significant differences in price deviation exist between the 12 independent purchasing groups to answer hypotheses 2 and 3. Moreover, to test the performance of the variable dummies year, type of hospital, health insurer concern, type for
MSC, type of medicine segment and having a generic/biosimilar variant on price deviation, and
hence answering hypotheses 4 to 10, one-way ANOVA and t tests including all purchasing groups are performed. In addition, to answer hypothesis 6, one-way ANOVA tests are performed for two specific purchasing groups, to determine whether there exists statistically significant differences in price deviation between the hospitals within a purchasing group. The specific purchasing groups are IZAAZ and Santeon. These purchasing groups are chosen as
IZAAZ and Santeon are homogeneous groups of hospitals (academic and top clinical hospitals
respectively) for which it is known with certainty that the purchasing groups were active in the 5 year period analyzed, and nowadays still exist in this formation of hospitals.
3.2 Data
3.2.1 Qualitative data
In this study, first a qualitative analysis is performed to obtain more in-depth information [55]. By conducting 5 interviews with health insurers and 4 interviews with hospital pharmacists, information is gathered about the purchasing process of expensive medicines and the use of purchasing groups. The underlying motivations of the market parties are revealed and
