Pharmaceutical Pricing in Germany: How Is Value Determined within the Scope of AMNOG?

VALUE IN HEALTH 20 (2017) 927–935

Available online at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/jval

Pharmaceutical Pricing in Germany: How Is Value Determined within the Scope of AMNOG? Victoria Desirée Lauenroth, MSc*, Tom Stargardt, PhD Hamburg Center for Health Economics, Hamburg, Germany

AB STR A CT

Objectives: To analyze how value is determined within the scope of the German Pharmaceutical Restructuring Act, which came into effect in 2011. Methods: Using data from all pharmaceuticals that had undergone assessment, appraisal, and price negotiations in Germany before June 30, 2016, we applied generalized linear model regression to analyze the impact of added benefit on the difference between negotiated prices and the prices of comparators. Data were extracted from the Federal Joint Committee's appraisals and price databases. We specified added benefit in various ways. In all models, we controlled for additional criteria such as size of patient population, European price levels, and whether the comparators were generic. Results: Our regression results confirmed the descriptive results, with price premiums reflecting the extent of added benefit as appraised by the Federal Joint Committee. On the substance level, an added benefit was associated with an increase in price premium of 227.2% (P o 0.001) compared with no added benefit. Moreover, we saw increases in price premium of 377.5% (P o 0.001),

90.0% (P o 0.001), and 336.8% (P o 0.001) for added benefits that were “considerable,” “minor,” and “not quantifiable,” respectively. Beneficial effects on mortality were associated with the greatest price premium (624.3%; P o 0.001), followed by such effects on morbidity (174.7%; P o 0.001) and adverse events (93.1%; P ¼ 0.019). Conclusions: Price premiums, or “value,” are driven by health gain, the share of patients benefiting from a pharmaceutical, European price levels, and whether comparators are generic. No statement can be made, however, about the appropriateness of the level of price premiums. Keywords: early benefit assessment, Germany, price negotiation, value-based pricing.

Introduction

an added therapeutic benefit over the current standard of care, defined as “appropriate comparative therapy” by the Federal Joint Committee (G-BA). (For the sake of simplicity, we use the term “comparator” in this article.) To do so, an initial advisory assessment is made by the Institute for Quality and Efficiency in Health Care, followed by a final appraisal of the G-BA. Second, pharmaceutical prices are negotiated between manufacturers and the National Association of Statutory Health Insurance Funds (GKV-SV). If the negotiations fail, prices are set by an arbitration board [9]. By law, substances that do not have an added benefit over their comparator should not lead to annual treatment costs that are higher than those of the comparator. For substances with an added benefit, however, the annual treatment costs may exceed those of the comparator by a premium that is in line with the extent and certainty of the added benefit specified in the appraisal of the G-BA. Although the determinants of coverage decision making have been studied well for countries that use formal economic evaluations (e.g., United Kingdom and Australia) [1,10–12], the determinants of decision making and pricing in two-stage administered systems [13], such as Germany, have not. Neither the determinants of decision making in Germany, which have been studied

Health systems vary in structure and in the needs and preferences of their populations. As a consequence, the assessment of health technologies and related decisions on the pricing and reimbursement of pharmaceuticals differ across countries as well [1–3]. Invariably, the concept of value plays an important role in health technology assessment. Yet looking at how different stakeholders define what attributes contribute to value [4] and how different countries have used the concept to help determine the prices of pharmaceuticals [5–7] reveals great heterogeneity. Paris and Belloni [6] even describe the application of value-based pricing (VBP) strategies as “more of an art than a science.” In Germany, a fourth-hurdle process, which leads to a change in launch prices after the pharmaceutical’s first year on the market, was introduced in 2011 with the German Pharmaceutical Restructuring Act (AMNOG). With this legislation, the German government aimed to ensure that pharmaceutical prices would be economically efficient while not inhibiting innovation [8]. The AMNOG process consists of two phases. First, new pharmaceuticals are assessed to determine whether they have

Copyright & 2017, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

* Address correspondence to: Victoria Desirée Lauenroth, Hamburg Center for Health Economics, Esplanade 36, Hamburg D-20354, Germany. E-mail: [email protected]. 1098-3015$36.00 – see front matter Copyright & 2017, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). http://dx.doi.org/10.1016/j.jval.2017.04.006

928

VALUE IN HEALTH 20 (2017) 927–935

only descriptively [14,15], nor the determinants of price negotiations’ outcomes [16–18] have been analyzed conclusively. In the present study, we analyzed whether price premiums negotiated in the second phase of the AMNOG process do indeed reflect 1) the G-BA’s previous appraisals on the pharmaceuticals’ added benefit; 2) the criteria, which are set out in a framework agreement between manufacturers and the GKV-SV [19], such as the size of the patient population; and 3) further aspects, such as the manufacturers’ experience in negotiating prices. Furthermore, we explored the magnitude of the price premiums granted for different extents of added benefit.

Methods Study Setting We included data on all pharmaceuticals that had completed the second phase of the AMNOG process between January 1, 2011, and June 30, 2016. In cases in which the price of a pharmaceutical had been renegotiated (e.g., because of a re-assessment or an extension to a new indication), we included results from only the first negotiations to eliminate potential bias resulting from strategic behavior of manufacturers. We excluded assessments of so-called orphan drugs because these do not have to demonstrate an added benefit over a comparator and thus do not include data on comparator costs. We also excluded pharmaceuticals for which manufacturers had decided not to start or complete negotiations and had therefore opted out of the German market. Last, we excluded benefit assessments with comparators defined as “best supportive care” because the G-BA does not publish comparator costs in these cases. We extracted data from publicly available G-BA appraisals as well as from the German price database Lauer-Taxes and various international, publicly accessible price databases [20–31].

Outcome Variable We computed the relation between the annual treatment costs of a pharmaceutical and those of its comparator from the statutory health insurer’s perspective (i.e., using pharmacy retail prices, including value-added tax minus manufacturer and pharmacy rebates as regulated by law): Relation of costs ¼

Annual treatment costs of pharmaceutical  100 Annual treatment costs of comparator

If the pharmaceutical and its comparator have equivalent costs, the relation amounts to 100%. The relation of costs can be transformed into a proportional price premium by subtracting 100 for descriptive analysis or by calculating marginal effects when used in regression models. To calculate the annual treatment costs of a pharmaceutical, we used the expected treatment duration and dosage given in G-BA’s appraisals in combination with postnegotiation prices from the German price database Lauer-Taxe. We also extracted annual treatment costs for the comparators directly from G-BA’s appraisals. When applicable, we additionally extracted costs for concomitant medication and for procedures associated with the use of the pharmaceutical or its comparator from G-BA’s appraisals. Treatment costs were collected at a patient subgroup level. In cases in which several interchangeable comparators were eligible for the same patient subgroup, we calculated an average of their costs. When the comparator varied between patient subgroups, the population-weighted mean was calculated to determine a pharmaceutical’s comparator’s cost. We obtained patient population sizes from G-BA’s appraisals.

Variable of Interest When determining the level of a pharmaceutical’s added benefit for each patient subgroup, the G-BA considers the following end points: mortality, morbidity, adverse events, and quality of life. We therefore specified the G-BA’s appraisal of a pharmaceutical’s added benefit in our six regression models in the following ways. In model A.1 we considered whether an added benefit (“not quantifiable,” “minor,” “considerable,” or “major”) had been assigned for at least one patient subgroup or not (“less benefit” or “no added benefit”). In cases in which no added benefit has been assigned, the annual treatment costs of the new substance should not exceed those of its comparator. In model A.2 we took into account differences in added benefit across patient subgroups, specifying the share of the patient population that had been appraised as experiencing an added benefit. For models B.1 and B.2 we differentiated between “no added benefit” and the four categories of added benefit. We distinguished between the greatest extent of added benefit that had been assigned at the substance level (model B.1) and the particular shares of the patient population that had been assigned a major, considerable, minor, or not quantifiable added benefit (model B.2). Models C.1 and C.2 reflect whether the G-BA decided that the pharmaceutical showed beneficial effects in particular end point categories (mortality, morbidity, adverse events, and quality of life) (model C.1) as well as the corresponding shares of the patient population (model C.2).

Control Variables The selection of control variables is based 1) on the framework agreement [19] between manufacturers and the German statutory health insurance and 2) on previous literature on coverage decision making [1]. First, we included the total number of patients eligible for a new pharmaceutical as specified in its marketing authorization [11,19,32,33]. This was obtained from G-BA’s documentation. Second, as required in the framework agreement, the annual treatment costs of so-called comparable medication are to be taken into account [19]. According to the framework agreement, a comparable medication is authorized within the same medical indication as the new pharmaceutical and its usage is appropriate according to international standards of evidence-based medicine. Yet, it is unequal to the G-BA– defined “comparator” (standard of care). Because no information on the comparable medication used is disclosed after price negotiations are completed, we used an approximation by specifying a binary variable that captures whether a comparable medication is available at the fourth level of the Anatomical Therapeutic Chemical Classification System. In addition, the comparable medication had to be 1) listed in the German LauerTaxe price database at the time of completed price negotiation and 2) authorized for the same medical indication as the new pharmaceutical [19]. The presence of alternative treatments or the “innovativeness” of a new pharmaceutical has also been used as a control variable in the literature [10–12,33–35]. Last, price negotiations must take account of the new pharmaceutical’s actual sales prices in other European countries [19]. We therefore attempted to extract the pharmaceutical’s exfactory prices for 14 European countries from publicly available European databases, and were able to do so for the Czech Republic [21] and France [22]. Because ex-factory prices were not always available, we surveyed the pharmaceutical’s price to pharmacy for Ireland [23] and Sweden [24], and the pharmacy retail prices for Belgium [25], Denmark [26], Finland [27], Italy [28], the Netherlands [29], Portugal [30], and Slovakia [31]. To calculate the ex-factory prices, we used average wholesaler and pharmacy margins estimated by Kanavos et al. [36]. The framework

VALUE IN HEALTH 20 (2017) 927–935

agreement specifies that price data from Austria, Spain, and the United Kingdom should also be taken into account in negotiations, but these were not publicly available. Contrary to the framework agreement, we did not include price data from Greece because the GKV-SV and manufacturers have agreed to temporarily exclude Greece from the European price basket [9]. We calculated an average ex-factory price per defined daily dose for each substance per country and weighted prices across countries according to purchasing power parities and population sizes from the Eurostat database [37] to obtain a substance's European price level. In addition, we controlled for whether a pharmaceutical’s comparator or all substances constituting a comparator were solely generic, because the price level of generics is substantially lower than that of substances under patent protection [38]. Also, because negotiation performance can be improved by experience [39], we included the number of price negotiations a manufacturer had previously undertaken with the GKV-SV. Furthermore, we included indication fixed effects, distinguishing between oncological, infectious, and all other diseases [32,33,35,40–42] (Table: "Overview of variables used within the analysis").

Statistical Analysis After analyzing price premiums descriptively, we estimated six regression models. The dependent variable (i.e., the relation between a pharmaceutical’s costs and those of its comparator) appeared to be gamma-distributed (many values at or close to 0), which we confirmed by applying the Modified Park Test [43]. Therefore, we could not use a simple ordinary least squares (OLS) regression. To avoid retransformation problems we preferred a generalized linear model (GLM) over a log OLS model [43]. Applying the Pregibon Goodness of Link Test [44] we confirmed the usage of a log-link. Thus, we used a GLM with a log-link function to analyze the impact of added benefit on price premiums. To allow for meaningful interpretation, we calculated marginal effects for each variable, with all other variables set to their means. We conducted several sensitivity analyses. First, because single observations may have a large impact on results based on such a small sample, we excluded substances with a Cook’s D greater than the conventional cutoff point of 4/n [45]. Second, we controlled for whether pharmaceuticals had been assessed during the first 7 months after the AMNOG legislation came into effect. During this transitional period, manufacturers were advised on the completeness of their dossiers by the G-BA and, if required, were granted an additional 3 months to complete them [46]. Third, we included a variable that controlled for whether the final price had been set by the arbitration board. Last, instead of using average comparator costs when several interchangeable comparators were eligible for the same patient subgroup, we reran our models using the least and the most costly comparators. All analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, NC).

Results Between January 1, 2011, and June 30, 2016, 148 pharmaceuticals completed the second phase of the AMNOG process. Of the appraisals related to these, 26 involved orphan drugs, 28 were renegotiations, 13 had comparators defined as “best supportive care,” and 20 involved manufacturers that had opted out of the German market, making it impossible to collect data on prices. Our final sample therefore contained 61 appraisals eligible for analysis (Fig. 1).

929

Fig. 1 – Selection of study sample.

Of the 61 pharmaceuticals covered in these appraisals, 31 (50.8%) had not been assigned an added benefit by the G-BA, 16 (26.2%) had been appraised as having a considerable added benefit, 12 (19.7%) as having a minor added benefit, and 2 (3.3%) as having an added benefit that was not quantifiable (Table 1). The 61 appraisals included in our study sample covered 153 patient subgroups. The comparator consisted of only one substance in 17 subgroups, of multiple substances in each of 29 subgroups, of a choice of substances or substance combinations in each of 103 subgroups, and of a nonmedicinal treatment in each of 4 subgroups.

Results of the Descriptive Analysis For pharmaceuticals with no added benefit (N ¼ 31), the average price premium over the annual treatment cost of comparators was 9.1% (median 0%). For pharmaceuticals with an added benefit (N ¼ 30), however, this amounted to 286.4% (median 110.1%). Taking the extent of added benefit into account, price premiums were the highest for pharmaceuticals with a considerable added benefit (N ¼ 16) (arithmetic mean 444.7%; median 151.2%), followed by those for pharmaceuticals whose added benefit was not quantifiable (N ¼ 2) (arithmetic mean 110.1%; median 110.1%) or minor (N ¼ 12) (arithmetic mean 104.8%; median 35.0%) (Fig. 2). Price premiums varied substantially across therapeutic areas (Table 2). For example, the average price premium for a considerable added benefit was 127.7% for pharmaceuticals used in infectious diseases (N ¼ 8) and 1900.3% for a pharmaceutical used in diseases of the eye (N ¼ 1).

Results of the Regression Analyses In model A.1 an appraisal of added benefit was associated with an increase in price premium of 227.2% (P o 0.001) compared with no added benefit. Correspondingly, model A.2 showed that any 1% increase in the share of patient population experiencing an added benefit was associated with an increase of 2.4% (P o 0.001) in price premium. Looking at price premiums according to the extent of added benefit, model B.1 showed an increase of 377.5% (P o 0.001) for pharmaceuticals with a considerable added benefit, of 90.0% (P o 0.001) for a minor added benefit, and of 336.8% (P o 0.001) for a not quantifiable added benefit compared with no added benefit. In model B.2, a 1% increase in the share of patient population experiencing an added benefit was associated with an increase in price premium of 3.8% (P o 0.001) for a considerable added benefit, 1.5% (P ¼ 0.012) for a not quantifiable added benefit, and 0.6% (P ¼ 0.07) for a minor added benefit. Model C.1 showed that an appraisal indicating a beneficial effect on mortality was associated with an increase in price premium of 624.3% (P o 0.001) compared with an appraisal that did not acknowledge such an effect. An appraisal of a beneficial effect on morbidity or adverse events was also associated with an

930

VALUE IN HEALTH 20 (2017) 927–935

Table 1 – Characteristics of the study sample. Characteristic

Frequency

Percentage

30

49.20

Added benefit Population share with added benefit Extent of added benefit Considerable Minor Not quantifiable Population share with added benefit Considerable Minor Not quantifiable Beneficial effect Mortality Morbidity Adverse events Quality of life Population share with beneficial effect Mortality Morbidity Adverse events Quality of life Generic Population (in 1000s) European price level Comparable pharmaceutical other than the comparator(s) Disease category Infectious diseases Oncological diseases Metabolic diseases Respiratory diseases Diseases of the nervous system Eye diseases Cardiovascular diseases Skin diseases Diseases of the blood and blood-forming organs Diseases of the musculoskeletal system Psychological diseases Diseases of the digestive system Diseases of the genitourinary system Others Experience

Average

Median

33.80% 16 12 2

26.20 19.70 3.30 13.00% 17.50% 3.30%

6 19 16 4

9.80 31.10 26.20 6.60 6.70% 20.40% 14.20% 3.80%

21

34.40

44

72.10

14 9 6 6 5 4 3 3 2 2 2 1 1 3

23.00 14.80 9.80 9.80 8.20 6.60 4.90 4.90 3.30 3.30 3.30 1.60 1.60 4.90

649.6 €111.8/DDD

57.4 €15.8/DDD

2.8

2

DDD, defined daily dose.

increase in a pharmaceutical’s price premium of 174.7% (P o 0.001) or 93.1% (P ¼ 0.019), respectively. An appraisal of a beneficial effect on quality of life, however, was associated with a decrease in price premium of 118.7% (P ¼ 0.007). In model C.2, a 1% increase in the share of patient population benefiting in the end point categories mortality, morbidity, and adverse events was associated with increases in price premium of 3.7% (P o 0.001), 1.9% (P o 0.001), and 0.8% (P ¼ 0.211), respectively. Again, a 1% increase in the share of patient population benefiting in the end point category quality of life was associated with a decrease in price premium of 2.6% (P ¼ 0.027). Five of the six models (A.2, B.1, B.2, C.1, and C.2) showed significant increases in price premium associated with the comparators being solely generic. Marginal effects ranged from 69.8% (P ¼ 0.002; model B.1) to 105.9% (P o 0.001; model B.2). The availability of a comparable medicine other than the comparator at Anatomical Therapeutic Chemical level 4 was associated with significant decreases in price premium in models A.1, B.1, and C.1, ranging from 73.6% (P ¼ 0.004) in model B.1 to 119.0% (P ¼ 0.002) in model A.1. An increase in the European price level of €1

per defined daily dose was associated with significant increases in price premium throughout all models, ranging from 0.06% (P ¼ 0.05; model B.2) to 0.13% (P ¼ 0.001; model C.2). No other variables were significant throughout all models, except for indication fixed effects (model B.1) and the manufacturer’s experience in negotiating prices (model B.2) (Table 3).

Results of the Sensitivity Analyses Removing all substances from the study sample with a Cook’s D greater than the conventional cutoff point of 4/n resulted in a reduction in sample size from 61 to 57. Subsequent regression results revealed substantially smaller but still highly significant estimates for the variables that described added benefit. The variable that indicated whether a pharmaceutical was assessed during the 7-month transitional period after the AMNOG legislation had come into effect was not significant throughout all models. In four of the six models (A.2, B.1, B.2, and C.2), the inclusion of a variable that indicated whether a pharmaceutical’s price had

931

VALUE IN HEALTH 20 (2017) 927–935

Fig. 2 – Price premiums on the cost of comparator by extent of added benefit. been set by the arbitration board was associated with a significant increase in the price premium for pharmaceuticals compared with no arbitration. Effects ranged from 121.0% (P ¼ 0.015; model B.1) to 167.0% (P ¼ 0.007; model C.2). At the same time, there was no change in the significance of any other variable in any of the six models. To account for potential uncertainties when it comes to the price of the comparator, we reran our analyses using the least costly and the most costly comparator instead of the average comparator costs. Doing so did not change regression results in the six models, with one exception. When using the least costly comparator to calculate average comparator costs, the variable for the comparators being solely generic became nonsignificant in all models.

Discussion In this study, we analyzed whether the price premiums negotiated for new pharmaceuticals in Germany between January 1, 2011, and June 30, 2016, reflect previous appraisals made by the G-BA about the pharmaceuticals’ added benefit, as was intended by legislation. We also examined whether the premiums reflect further criteria agreed upon in the framework agreement between manufacturers and the GKV-SV. Furthermore, we have

systematically reported in this article on the magnitude of price premiums granted for different dimensions of added benefit. Our results suggest that the intentions of the AMNOG legislation are generally being met. For pharmaceuticals with no added benefit, the prices negotiated during our study period yielded annual treatment costs that were no higher than those of their comparator. At the same time, price premiums were negotiated for pharmaceuticals that had been appraised as having an added benefit. Contrary to what is stated within the framework agreement, the size of the eligible patient population did not have a significant effect on the magnitude of price premiums. This could be explained by the assumption that substances eligible for large patient populations (e.g., diabetes) are being assigned no added benefit more frequently. Descriptive analyses indeed showed a greater average population for substances without an added benefit (926,315; median 95,000) than for substances with an added benefit (363,751; median 46,000). Nevertheless, conducting regression analysis for the subgroup of beneficial substances only did not show significant effects for the size of patient population. This, however, might also be the result of small sample size. The share of patients benefiting from a pharmaceutical, however, was valued in negotiations, as the results of our models A.2, B.2, and C.2 suggest. Two of our models (C.1 and C.2) indicate that a pharmaceutical’s effects on mortality were valued the highest, followed by its impact on morbidity and adverse events. Unexpectedly, the effects of a pharmaceutical on a patient’s quality of life appeared to play a contradictory role. It should be noted, however, that only 4 of the 61 pharmaceuticals in our sample had been appraised as having a beneficial effect on quality of life. In all four cases, beneficial effects were seen not only for quality of life but also for other categories, which might suggest that beneficial effects in several end point categories are not valued additively. Also, the negative impact we observed on the price premium could merely be the result of small sample size. Effects resulting from changes in the European price level were rather small but significant throughout all models. Thus, in accordance with the observations of Paris and Belloni [6] for member countries of the Organisation for Economic Co-operation and Development, international price referencing also plays a (minor) role in the pricing of new pharmaceuticals in Germany. One could argue that the European price level is considered only after an added benefit has been assigned, as is stipulated by law. Nevertheless, when we used this as an interaction effect, our analyses suggested that the impact of the European price level on price premiums does not apply to substances with added benefit only (P ¼ 0.227).

Table 2 – Mean price premiums on the cost of comparator by therapeutic area. Therapeutic area

N

None (%)

Not quantifiable (%)

Minor (%)

Considerable (%)

Infectious diseases Oncological diseases Respiratory diseases Metabolic diseases Diseases of the nervous system Eye diseases Cardiovascular diseases Skin diseases Diseases of the blood and blood-forming organs Psychological diseases Diseases of the musculoskeletal system Diseases of the digestive system Diseases of the genitourinary system Others

14 9 6 6 5 4 3 3 2 2 2 1 1 3

0.00 0.40 10.00 88.80 7.47 7.80 0.00 1.04 4.60 3.80 36.40 2.20 151.40 43.60

110.10

1.40 147.20 28.30 262.80 9.70

127.70 532.30 33.00

26.60

1900.30 382.40

1648.40

932

Table 3 – Regression results. Explanatory variables

Model A.1 (added benefit yes/no)

Estimate

Intercept No added benefit

* P o 0.001. † P o 0.01. ‡ P o 0.05.

Estimate

Marginal effect

4.670* Reference category 227.224 *

1.265

Model B.1 (extent of added benefit)

Estimate

Marginal effect

Model B.2 (share of population with different extents of added benefit) Estimate

Marginal effect

4.236* Reference category

Reference category

4.485*

1.590

377.459*

2.179

3.810*

0.658

89.991*

0.351

0.608

1.501

336.781*

0.886

1.540†

Model C.1 (effects of end point categories)

Estimate

Marginal effect

4.165*

Model C.2 (share of population with effects of end point categories)

Estimate

Marginal effect

4.381*

2.365 *

0.306 0.000

58.550 0.005

0.405 0.000

80.628† 0.010

0.460 0.000

69.758‡ 0.005

0.555 0.000

105.941* 0.006

1.606 0.798 0.453 0.937 0.390 0.000

624.285* 174.743* 93.057† 118.686‡ 76.469† 0.006

2.026 1.034 0.422 1.438 0.448 0.000

3.710* 1.884* 0.767 2.589† 87.765‡ 0.009

0.569

118.940‡

0.299

59.624

0.469

73.625‡

0.139

24.839

0.519

108.147†

0.282

54.619

0.001

0.104‡

0.001

0.119‡

0.000

0.056†

0.000

0.065†

0.001

0.124‡

0.001

0.130‡

Reference category 0.401

Reference category 62.440

Reference category 0.423

Reference category 78.320

Reference category 0.480

Reference category 53.791†

Reference category 0.151

Reference category 25.819

Reference category 0.056

Reference category 8.354

Reference category 0.086

Reference category 14.710

0.085 0.014

16.805 2.508

0.168 0.025

35.202 4.661

0.164 0.011

25.131 1.604

0.048 0.041

8.673 7.151†

0.348 0.008

60.714 1.505

0.126 0.031

22.032 5.763

VALUE IN HEALTH 20 (2017) 927–935

Share of population with an added benefit Share of population with a … Considerable added benefit Minor added benefit Not quantifiable added benefit Share of population with a beneficial effect on … Mortality Morbidity Adverse events Quality of life Generic Population (in 1000s) Unavailability of a comparable pharmaceutical other than the comparator(s) European price level Disease category Oncological diseases Infectious diseases All other diseases Experience

4.322* Reference category 1.169

Marginal effect

Model A.2 (share of population with added benefit)

VALUE IN HEALTH 20 (2017) 927–935

The estimates associated with a solely generic comparator were positive and significantly different from 0 throughout all models. The pharmaceutical industry has frequently argued that generic prices are inappropriate anchors when negotiating prices for pharmaceuticals with patent protection, because generic prices do not include costs for research and development or marketing [47]. Our results indicate that this objection may be taken into account during negotiations. Whether this price premium is large enough to address the industry’s concern is another matter. Last, because orphan drugs are assigned an added benefit per se by law, we might conclude that the rarity of an indication is also valued within the pricing of pharmaceuticals in Germany, even though we could not analyze the value of this directly. Our results are contradictory to what Schwarz and Freiberg [16], Radic et al. [17], and Theidel and von der Schulenburg [18] found in their studies on the connection between added benefit and negotiated prices. From a methodological perspective, however, all three studies are not comparable with ours. Schwarz and Freiberg [16] as well as Theidel and von der Schulenburg [18] used the difference between initial launch prices and negotiated prices as dependent variable, which is, thus, influenced by a manufacturer’s pricing strategy. Furthermore, the specification of independent variables in their OLS differs from our specification in the GLM. This is also true for the analyses of Radic et al. [17] who, in addition, looked at pharmaceuticals that were appraised an added benefit only [16–18]. Overall, our results are comparable with those of Antoñanzas et al. [4] and Paris and Belloni [6] in their analyses of how different countries refer to value in the context of pharmaceutical pricing. Compared with UK’s understanding of VBP, however, the AMNOG process, including the second stage of price negotiations, is not nearly as elaborate (yet). Claxton et al. [48,49] look at VBP through the lens of current productivity levels in the National Health Service (NHS). According to their definition, a value-based price should ensure that the health gain associated with the use of a new technology is at least as great as the health expected to be forgone because of the new technology’s additional costs. They posit that this should be assessed using empirically determined cost-effectiveness thresholds that reflect current NHS productivity levels in different therapeutic areas [48,49]. Sussex et al. [50] regard the essence of VBP as being to ensure “that the maximum price the NHS will pay for a medicine is set where the incremental value of using it relative to the comparator treatment (…) just balances the incremental costs.” As components of value they explicitly include criteria that go beyond a pharmaceutical-associated health gain, such as social value judgments used by the National Institute for Health and Care Excellence in current health technology assessment or criteria grounded in bioethical arguments. Nevertheless, determining which criteria to consider; how to measure, value, and aggregate these; and how ultimately to link these to a maximum price are not only methodologically challenging but also depend on a range of normative value judgments [50,51]. Unlike the United Kingdom, Germany does not explicitly define a limited health care budget. There is therefore no need to identify and measure health forgone. Instead, new pharmaceuticals directly replace the current standard of care (i.e., the comparators defined by the G-BA). Intuitively, if additional money is not paid in cases in which there is no additional benefit, productivity levels will stay constant. At the same time, paying additional money for additional benefit is plausible. In the German setting, however, no statements can be made about the extent to which negotiated price premiums change current productivity levels. Indeed, considering that neither productivity levels nor incremental benefits are measured as part of the assessment methodology in Germany, and given the confidential

933

nature and lack of transparency in price negotiations within the German system, no statements can be made about the appropriateness of price premiums. This confirms what Gerber et al. [7] pointed out shortly after AMNOG’s introduction in 2011. The introduction of health economic evaluation could not only be of help to enable the assessment of the appropriateness of price premiums, but could also increase the transparency of the process [52]. Until now, pharmaceutical pricing in Germany can be considered only to be “value-based” in the very broadest sense —namely, that the negotiated price premiums reflect the extent of added benefit as appraised by the G-BA.

Study Limitations Our study has several important limitations. First, although we included information from every negotiation currently available, our sample size was small, potentially rendering influential independent variables nonsignificant and restricting the number of variables we could add to the model. Our descriptive and regression results must therefore be interpreted with caution. Second, although our sensitivity analyses suggest that our results are quite stable for the variables of interest, outliers can have an impact in such a small sample. This can be seen in the sensitivity analysis based on Cook’s D, in which the four observations with the highest price premiums were excluded and, consequently, effect sizes changed considerably. Third, when collecting our data, we depended on information that was publicly available. We could neither observe nor account for things “between the lines” or compromises made in real-life negotiations. For example, there may have been compromises between manufacturers and GKV-SV concerning the interpretation of comparators, especially in cases in which the comparators consist of combinations and/or multiple alternatives. Here, it is conceivable that recent data on physician prescribing patterns could be introduced into the negotiations, revealing that the use of one comparator alternative is much more common than that of another. Similarly, it is plausible that negotiations may make use of more recent data on disease prevalence, leading to changes in the size of patient groups thought to benefit from a pharmaceutical. Although this would create only minor distortions in our calculations, it could explain why the mean price premium for no added benefit in our descriptive analysis was 9.1% (median 0%) although it should be 0% by law.

Conclusions When it comes to the pricing of pharmaceuticals in the context of AMNOG, our results suggest that the intent of the legislation is generally being met. When there is no evidence of an added benefit, the negotiated prices yield annual treatment costs that are no higher than those for the comparators. At the same time, higher price premiums are generally negotiated for more beneficial pharmaceuticals. The price premiums themselves appear to be driven by health gain, the share of patients benefiting from a pharmaceutical, and the rarity of diseases. Furthermore, the generic status of comparators and European price levels do play a role in pricing. Nevertheless, this all being said, we are unable to make any statements about the appropriateness of the level of price premiums. Although in practice policy decides on “appropriate” price premiums, this rather is (or should be) the answer to normative questions related to a population’s preferences and its associated willingness to pay [53,54]. Questions like this will most likely be addressed to some extent through frameworks for VBP such as those currently being discussed in the United Kingdom [49,50]. Pricing within the scope of AMNOG may

934

VALUE IN HEALTH 20 (2017) 927–935

therefore be considered only to be value-based in the very broadest sense, namely, in that negotiated price premiums reflect the extent of added benefit as appraised by the G-BA. Nevertheless, compared with the concept and operationalization of VBP, the AMNOG process, including the second stage of price negotiations, is not nearly as elaborate (yet). Source of financial support: For the development of the Hamburg Center for Health Economics AMNOG appraisal database, part of which was used in this analysis, funding was received by the German Research Foundation (grant nos. FI 1950/2-1 and STA 1311/2-1).

Supplemental Materials Supplemental material accompanying this article can be found in the online version as a hyperlink at http://dx.doi.org/10.1016/j. jval.2017.04.006 or, if a hard copy of article, at www.valueinhealth journal.com/issues (select volume, issue, and article).

R EF E R EN CE S

[1] Fischer KE. A systematic review of coverage decision-making on health technologies—evidence from the real world. Health Policy 2012;107:218–30. [2] Stafinski T, Menon D, Davis C, McCabe C. Role of centralized review processes for making reimbursement decisions on new health technologies in Europe. Clinicoecon Outcomes Res 2011;3:117–86. [3] Eddy D. Health technology assessment and evidence-based medicine: What are we talking about? Value Health 2009;12(Suppl 2):S6–7. [4] Antoñanzas F, Terkola R, Postma M. The value of medicines: a crucial but vague concept. Pharmacoeconomics 2016;34:1227–39. [5] Drummond M, de Pouvourville G, Jones E, et al. A comparative analysis of two contrasting European approaches for rewarding the value added by drugs for cancer: England versus France. Pharmacoeconomics 2014;32:509–20. [6] Paris V, Belloni A. Value in pharmaceutical pricing. OECD Health Working Paper, Organisation for Economic Cooperation and Development, Paris, France, 2013. [7] Gerber A, Stock S, Dintsios C-M. Reflections on the changing face of German pharmaceutical policy. Pharmacoeconomics 2011;29:549–53. [8] German Bundestag. Gesetzesentwurf: Drucksache 17/2413, July 6, 2010. [9] Ludwig S, Dintsios C-M. Arbitration board setting reimbursement amounts for pharmaceutical innovations in Germany when price negations between payers and manufacturers fail: an empirical analysis of 5 years’ experience. Value Health 2016;19:1016–25. [10] Dakin HA, Devlin NJ, Odeyemi IA. “Yes”, “No” or “Yes, but”? Multinomial modelling of NICE decision-making. Health Policy 2006;77:352–67. [11] Devlin N, Parkin D. Does NICE have a cost-effectiveness threshold and what other factors influence its decisions? A binary choice analysis. Health Econ 2004;13:437–52. [12] Harris AH, Hill SR, Chin G, et al. The role of value for money in public insurance coverage decisions for drugs in Australia: a retrospective analysis 1994–2004. Med Decis Making 2008;28:713–22. [13] Drummond M, Jönsson B, Rutten F. The role of economic evaluation in the pricing and reimbursement of medicines. Health Policy 1997;40:199–215. [14] Ruof J, Schwartz FW, Schulenburg J-M, Dintsios C-M. Early benefit assessment (EBA) in Germany: analysing decisions 18 months after introducing the new AMNOG legislation. Eur J Health Econ 2014;15:577–89. [15] Fischer KE, Stargardt T. Early benefit assessment of pharmaceuticals in Germany: manufacturers’ expectations versus the Federal Joint Committee’s decisions. Med Decis Making 2014;34:1030–47. [16] Schwarz R, Freiberg M. Black Box AMNOG Rabatte. Mark Access Health Policy 2015;1:27–8. [17] Radic D, Haugk S, Radic M. Benefit assessment and price negotiation under AMNOG: Calculable process or unfair poker game? Gesundheitswesen [published online ahead of print September 16, 2016]. http://dx.doi.org/10.1055/s-0042-113600. [18] Theidel U, von der Schulenburg J-MG. Benefit assessment in Germany: implications for price discounts. Health Econ Rev 2015;6:33. [19] Social Security Code V (SGB V). Framework agreement according to §130b (9), March 19, 2012.

[20] Lauer Taxe. Kundencenter—LAUER-FISCHER GmbH, Fürth. Available from: https://www.lauer-fischer.de/LF/Seiten/Verwaltung/ Kundencenter/1.aspx. [Accessed January 23, 2016]. [21] State Institute for Drug Control. Medicinal products database. Available from: http://www.sukl.eu/modules/medication/. [Accessed January 23, 2016]. [22] L’assurance maladie en ligne. Base des Médicaments et Informations Tarifaires. Available from: http://www.codage.ext.cnamts.fr/codif/ bdm_it/index.php?p_site=AMELI. [Accessed January 23, 2016]. [23] Health Service Executive. Primary care reimbursement service. Available from: https://www.sspcrs.ie/druglist/pub. [Accessed January 23, 2016]. [24] Dental and Pharmaceutical Benefits Agency. Databas läkemedel. Available from: http://www.tlv.se. [Accessed January 23, 2016]. [25] Belgisch Centrum voor Farmacotherapeutische Informatie. Belgisch Centrum voor Farmacotherapeutische Informatie. Available from: http://www.bcfi.be. [Accessed January 23, 2016]. [26] Danish Medicines Agency. Lægemidler styrelsen. Available from: http:// www.medicinpriser.dk. [Accessed January 23, 2016]. [27] Kela. Medicinal products database. Available from: https://easiointi. kela.fi/laakekys_app/LaakekysApplication?kieli=en. [Accessed January 23, 2016]. [28] Federazione nazionale unitaria titolari di farmacia. Cerca un farmaco. Available from: http://www.federfarma.it/Farmaci-e-farmacie/ Cerca-un-farmaco.aspx. [Accessed January 23, 2016]. [29] Zorginstituut Nederland. Medicijnkosten. Available from: http://www. medicijnkosten.nl. [Accessed January 23, 2016]. [30] Autoridade nacional do medicamento e produtos saude. Pesquisa de Medicamentos. Available from: http://www.infarmed.pt/infomed/ pesquisa.php. [Accessed January 23, 2016]. [31] Ministerstvo zdravotníctva Slovenskej republiky. Kategorizácia a ceny. Available from: http://www.health.gov.sk/?kategorizacia-a-ceny. [Accessed January 23, 2016]. [32] Chim L, Kelly PJ, Salkeld G, Stockler MR. Are cancer drugs less likely to be recommended for listing by the Pharmaceutical Benefits Advisory Committee in Australia? Pharmacoeconomics 2010;28:463–75. [33] Cerri KH, Knapp M, Fernandez J-L. Decision making by NICE: examining the influences of evidence, process and context. Health Econ Policy Law 2014;9:119–41. [34] Linley WG, Hughes DA. Reimbursement decisions of the All Wales Medicines Strategy Group. Pharmacoeconomics 2012;30:779–94. [35] Dakin H, Devlin N, Feng Y, et al. The influence of cost-effectiveness and other factors on NICE decisions. Health Econ 2015;24:1256–71. [36] Kanavos P, Schurer W, Vogler S. The Pharmaceutical Distribution Chain in the European Union: Structure and Impact on Pharmaceutical Prices. Brussels, Belgium: European Commission, 2011. [37] Eurostat. Available from: http://ec.europa.eu/eurostat/de. [Accessed January 28, 2016]. [38] Sloan FA, Hsieh C-R. Health Economics. Cambridge, UK: MIT Press, 2012. [39] Thompson L. The influence of experience on negotiation performance. J Exp Soc Psychol 1990;26:528–44. [40] Anis AH, Guh D, Wang X. A dog’s breakfast: prescription drug coverage varies widely across Canada. Med Care 2001;39:315–26. [41] Segal L, Dalziel K, Mortimer D. Fixing the game: Are between-silo differences in funding arrangements handicapping some interventions and giving others a head-start? Health Econ 2010;19:449–65. [42] Mauskopf J, Chirila C, Masaquel C, et al. Relationship between financial impact and coverage of drugs in Australia. Int J Technol Assess Health Care 2013;29:92–100. [43] Manning WG, Mullahy J. Estimating log models: To transform or not to transform? J Health Econ 2001;20:461–94. [44] Pregibon D. Goodness of link tests for generalized linear models. Appl Stat 1980;29:15–24. [45] Bollen KA, Jackman RW. Regression diagnostics: an expository treatment of outliers and influential cases. In: Fox J, Long JS, eds. Modern Methods of Data Analysis (2nd ed.). NewburyPark, CA: Sage, 1990. [46] Regulation on the Benefit Assessment of Pharmaceuticals (AM-NutzenV). Available from: http://www.gesetze-im-internet.de/ bundesrecht/am-nutzenv/gesamt.pdf. [Accessed May 3, 2017]. [47] Cassel D, Ulrich V. AMNOG auf dem ökonomischen Prüfstand— Funktionsweise, Ergebnisse und Reformbedarf der Preisregulierung für neue Arzneimittel in Deutschland. Baden-Baden, Germany: Nomos Verlagsgesellschaft, 2015. [48] Claxton K, Sculpher M, Carroll S. Value-based pricing for pharmaceuticals: its role, specification and prospects in a newly devolved NHS. CHE Research Paper, Centre for Health Economics, York, UK, 2011. [49] Claxton K, Martin S, Soares M, et al. Methods for the estimation of the National Institute for Health and Care Excellence cost-effectiveness threshold. Health Technol Assess 2015;19:1–503, v–vi. [50] Sussex J, Towse A, Devlin N. Operationalizing value-based pricing of medicines. Pharmacoeconomics 2013;31:1–10.

VALUE IN HEALTH 20 (2017) 927–935

[51] Walton SM, Basu A, Mullahy J, et al. Measuring the value of pharmaceuticals in the US health system. Pharmacoeconomics 2017;35:1–4. [52] Franken M, Heintz E, Gerber-Grote A, Raftery J. Health economics as rhetoric: the limited impact of health economics on funding decisions in four European countries. Value Health 2016;19:951–6.

935

[53] Daniels N, Sabin JE. Setting Limits Fairly—Can We Learn to Share Medical Resources? Oxford, UK: Oxford University Press, 2002. [54] Linley WG, Hughes DA. Societal views on NICE, cancer drugs fund and value-based pricing criteria for prioritising medicines: a crosssectional survey of 4118 adults in Great Britain. Health Econ 2013;22: 948–64.