Development of an Insurance Framework for Funding ... - Health Policy

G Model HEAP-3773; No. of Pages 6

ARTICLE IN PRESS Health Policy xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Health Policy journal homepage: www.elsevier.com/locate/healthpol

Time for a change in how new antibiotics are reimbursed: Development of an insurance framework for funding new antibiotics based on a policy of risk mitigation Adrian Towse a,∗ , Christopher K. Hoyle b , Jonathan Goodall b , Mark Hirsch b , Jorge Mestre-Ferrandiz a,1 , John H. Rex c,2 a

Office of Health Economics, Southside, 7th Floor, 105 Victoria Street, London, SW1E 6QT, UK 1 Francis Crick Avenue, Cambridge Biomedical Campus, Cambridge, CB2 0AA, UK c AstraZeneca Pharmaceuticals, 35 Gatehouse Park, Waltham, MA, USA b

a r t i c l e

i n f o

Article history: Received 1 July 2016 Received in revised form 30 June 2017 Accepted 30 July 2017 Keywords: Healthcare-associated infection Antibiotics Healthcare system Multidrug resistance (MDR) bacteria Market dysfunction Premium price model Insurance model

a b s t r a c t Healthcare systems depend on the availability of new antibiotics. However, there is a lack of treatments for infections caused by multidrug resistant (MDR) pathogens and a weak development pipeline of new therapies. One core challenge to the development of new antibiotics targeting MDR pathogens is that expected revenues are insufficient to drive long-term investment. In the USA and Europe, financial incentives have focussed on supporting R&D, reducing regulatory burden, and extending market exclusivity. Using resistance data to estimate global revenues, we demonstrate that the combined effects of these incentives are unlikely to rekindle investment in antibiotics. We analyse two supplemental approaches: a commercial incentive (a premium price model) and a new business model (an insurance model). A premium price model is familiar and readily implemented but the required price and local budget impact is highly uncertain and sensitive to cross-sectional and longitudinal variation in prevalence of antibiotic resistance. An insurance model delivering risk mitigation for payers, providers and manufacturers would provide an incentive to drive investment in the development of new antibiotics while also facilitating antibiotic conservation. We suggest significant efforts should be made to test the insurance model as one route to stimulate investment in novel antibiotics. © 2017 Office of Health Economics. Published by Elsevier Ireland Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction One key challenge of developing new antibiotics is that expected returns and associated risk are not competitive with other therapeutic areas [1–8]. The conventional pharmaceutical business model with reward based on sales volume and price puts the twin objectives of (i) developing new antibiotics to tackle growing antimicrobial resistance, whilst (ii) restricting use of antibiotics to encourage appropriate stewardship, in opposition to each other. Initiatives to try to solve this problem are being implemented, but have focused primarily on providing funding for early stage research and development (R&D) [9,10] and on regulatory changes [11,12] with, in some cases, additional market exclusivity. Recent analyses [5,13,14] have identified other types of incentives required

∗ Corresponding author. E-mail address: [email protected] (A. Towse). 1 Permanent address: Rosario Pino 8, Madrid, Spain. 2 Current address: F2G Ltd, Lankro Way, Eccles, Manchester, M30 0LX, UK.

to make R&D for new antibiotics more attractive. In this paper we model three policy options. Firstly, we explore whether public support for R&D with regulatory change and market exclusivity is likely to be sufficient to stimulate new R&D. Secondly we model a premium price model in which the pricing mechanism is used to restrict use and to provide a return on R&D. Thirdly we look at the impact of introducing a new “insurance-type” commercial model, which involves the partial de-linkage of returns on R&D from the volume of sales. This is the only option which meets the twin objectives referred to above. We then discuss how this can be made to work in the context of other proposals for de-linkage. 2. Our economic model for evaluation of incentives Our approach was to combine a global estimate of the costs of developing a new antibacterial drug with global demand and revenue estimates scaled up from detailed modelling of demand and resistance estimates. We based our analysis on a hypothetical antibacterial drug targeted against specific, multi-drug resistant

http://dx.doi.org/10.1016/j.healthpol.2017.07.011 0168-8510/© 2017 Office of Health Economics. Published by Elsevier Ireland Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/).

Please cite this article in press as: Towse A, et al. Time for a change in how new antibiotics are reimbursed: Development of an insurance framework for funding new antibiotics based on a policy of risk mitigation. Health Policy (2017), http://dx.doi.org/10.1016/j.healthpol.2017.07.011

G Model

ARTICLE IN PRESS

HEAP-3773; No. of Pages 6

A. Towse et al. / Health Policy xxx (2017) xxx–xxx

2

Table 1 Global cost of R&D for a new antibiotic and effect of incentives on R&D and eNPV. PCRD Global cost (million)a Progression rateb Duration Total programsc Global capitalised cost (million)d eNPV ($m) Global capitalised cost (million) eNPV ($m) Global cost (millions) Global capitalised cost ($m) eNPV ($m) Global capitalised cost (million) eNPV ($m)

Phase I

Base case without incentives $19 $16 0·35 0.67 5 years 0 months 1 year4 months 15.22 5.33 $708 $153

Phase II

Phase III

Registration

Post-launch study Total (excl. post-launch)

$54 0·46 2 years2 months 3.57 $292

$196 0.70 2 years5 months 1.64 $393

$29 0·87 0 years11 months 1.15 $35

$40 N/A 6 years $40

With matched (50%) co-funding as part of Public-Private Partnerships for R&D $354 $76 $146 $197 $17

$20

Implementation of the Tier Be framework for registration $19 $16 $54 $98 $708 $153 $292 $197

$40 $40

$29 $35

With matched funding as part of PPP for R&D and implementation of the Tier B framework for registration $354 $76 $146 $99 $17 $20

11.8 years 15.22 $1581 -$1510 $791 -$701

$1385 -$1313 $692 -$603

a

Out-of-pocket costs to support a single program. Progression rates are defined as the probability of progression to the next stage of R&D or successful registration. Two progression rates for pre-clinical R&D (PCRD) have been published; 0·35 [35] and 0·10 [36]. We used the more optimistic estimate of 0·35 which results in lower R&D costs. Progression rates for the subsequent clinical development phases were estimated from our analysis of clinical studies for antimicrobials conducted between 2005 and 2011 (see supplementary material). Note that for Phase III, we use the (more optimistic) average of 0.70 reported by Paul et al. [35] which results in lower R&D costs. c This is the total number of programs that are required to be initiated at the pre-clinical R&D stage to produce a single licensed product. d A cost of capital of 10% was used to calculate the capitalised cost of R&D. It represents the expected return required from an alternative investment portfolio with a similar level of risk [17].The capitalised cost is risk-adjusted to account of failures i.e. programs that do not progress to next phase. e Under Tier B pathway [11], there is one phase III trial rather than two (standard) phase III trial for each of two indications. Thus, we assume Phase III costs are a quarter relative to our base case, which is a conservative estimate of the cost of late-stage development. b

(MDR) pathogens with expected use exclusively within acute care. Prevalence of the MDR pathogens would be very low at the point of registration of a new antibiotic, which would therefore, initially, be rarely used. Resistance prevalence and subsequent growth was based on a combination of high and low European country-specific prevalence data for healthcare-associated infections (HAIs) [15] and the most recently available data for rates of resistance of E. coli to third generation cephalosporins [16] (see the supplementary material). To model lifetime revenues we estimated an average price per course and, using increased prevalence of antibiotic resistance as the key driver, estimated volumes. We are not aware of any other modelling using this approach. Economic viability was assumed to require a global estimated pre-tax NPV (eNPV) of $100 million [5]. Programs were assumed to target global registration. The model covered R&D from drug target to registration, and revenue, manufacturing and distribution costs for twenty years post-registration. To model the effect of recent efforts to streamline antibacterial development, our base case included estimated Phase 3 costs of four large non-inferiority studies (two for each of two different indications to obtain registration in the US and EU). We also include a single post-launch study costing $40 million to cover the costs of additional indications.

3. Cost of developing a new antibiotic and resulting estimated net present value (eNPV) − base case Our assumptions for the base case model (i.e., with no incentives) R&D costs and eNPV are shown in Table 1. The estimated global cost of developing a new, targeted antibiotic, excluding the post launch study, was calculated at $1581 million (at 2011 US prices), which is similar to that of Mestre-Ferrandiz et al. [17]. Pre-clinical R&D contributes the greatest share of the capitalised development cost as (i) a large number of projects need to be initiated at this stage due to low probability of progression and (ii) from the long time remaining to drug licensing impacts the cost of capital. For the base case, the market for a new antibiotic was based on an assumed 3% annual growth in the number of HAIs and a preva-

lence growth of MDR infections requiring treatment with a new antibiotic of half the growth in prevalence of infections caused by E. coli resistant to third generation cephalosporins. Data came from the European Centre for Disease Control Antimicrobial Resistance Interactive Database (EARS-Net) [16]. Assuming a price per day of $120, roughly comparable to recently launched branded antibiotics for certain Gram positive infections, and treatment duration of 14 days for a complicated Gram-negative infection [18], results in a treatment course cost of $1680. Operating cost assumptions are set out in the supplementary material. The global eNPV for the base case, including R&D and registration cost, and post-registration revenue and costs, was negative (- $1510 million). We estimated the numbers of MDR infections that would be needed in our model to make the current price/volume arrangements viable for new drugs, i.e. generating the target eNPV of $100m. If the cost of treatment for each case of infection caused by a MDR pathogen remains at $1680, the total number of MDR cases for Europe and North America at the time of launch required to produce a positive eNPV would need to be more than 375,000 with cases of HAI needing to almost double from the 12.56 million cases in 2012–23.67 million with 1.58% caused by a MDR pathogen requiring a new antibiotic. If 80% of these MDR patients were treated with the new antibiotic, equating to an annual revenue of $630 million, the required eNPV would be achieved. This compares to our base case estimate of 200 cases in the first year post-registration based on data from EARS-NET for E. coli resistant to third generation cephalosporins. The analysis is set out in the supplementary material. Such a high prevalence of MDR infections would likely produce a large increase in attributable excess mortality [19], morbidity [20,21] and a decrease in the number of certain medical procedures (e.g. surgical implantation) that patients would be willing to risk and which hospitals would be willing to perform [22,23]. In short, relying on growth in MDR infections to make the existing commercial model for drug development work would require such a high prevalence of MDR infections that the sustainability of developed country health systems would be threatened. Patients would be reluctant to undertake routine treatment because of the


G Model

ARTICLE IN PRESS


A. Towse et al. / Health Policy xxx (2017) xxx–xxx Table 2 The eNPV under different scenarios. Scenario

eNPV ($m)a

Five-year exclusivity

eNPV ($m)a

Base case PPP Tier Bb PPP + Tier B

-$1510 -$701 -$1313 -$603

Base case + 5 year exclusivity PPP + 5 year exclusivity Tier B + 5 year exclusivity PPP + Tier B + 5 year exclusivity

-$1436 -$627 -$1239 -$529

PPP: Public Private Partnership; eNPV: estimated Net Present Value. a eNPV represents the global estimate extrapolated from the eNPV of the EU5. b Tier B regulatory pathway described by Rex et al. [11].

high risk of getting an MDR infection. The impact on GDP would also be considerable [24]. 4. Current incentives for antibiotics We next modelled the impact on the base case eNPV of incentives previously proposed [25–31]. We examined: (1) publicprivate partnerships (PPPs), assuming public funding covered 50% of total R&D (pre- and post-launch) costs; (2) implementation of the Tier B regulatory pathway as set out by Rex et al. [11] with one phase III trial rather than two (standard) phase III trial for each of two indications; (3) use of both a PPP and Tier B; (4) 5-year extended market exclusivity, used in combination with the base case and each of (1) − (3). Note that the Tier B regulatory framework addresses the shortcomings of the historical approach of regulatory bodies to the approval of antibiotics. It relies on the totality of the data, both clinical and pre-clinical, rather than just large randomised trials, providing for registration of an indication based on a single phase III trial plus small pathogen focused studies. Table 1 shows the impact of these initiatives on total capitalised R&D cost and eNPV. Footnote (e) explains how we model Tier B. Table 2 shows the impact of extended market exclusivity (in isolation and in combination with other initiatives) on eNPV. The results show that none of these initiatives, even in combination, are able to bring the eNPV to the target level. The combination (3) “PPP funding plus Tier B” yields the highest eNPV with or without any additional market exclusivity. Table 2 highlights the limited impact of a five-year patent extension on eNPV. If annual revenues are low, extending them for five years has little effect. Thus, neither the current business model, nor the use of public partnership funding of R&D together with regulatory reform and a five-year patent extension will increase the pipeline of new antibiotics. Additional action is required. 5. Evaluation of commercial incentives for new antibiotics: ‘premium price’ and ‘insurance’ model We evaluated the impact of two business models on eNPV using the base case and incentive combinations (1)-(4): a ‘premium price model’ based on a single enhanced price per unit; and an ‘insurance model’ based on a global flat annual fee, apportioned to, and paid by, each healthcare system, coupled with a price paid by healthcare providers to the manufacturer for each unit of drug used. We suggest this price be set higher than the cost of a generic antibiotic, which would promote stewardship, whilst limiting the financial liability of the provider should localised demand rise due to a large increase in prevalence of infection within a single hospital. The structure of the insurance model provides insurance in the following ways: 1) The company is insured against the commercial risk of both a low mean price and highly variable use, both of which risk large negative eNPVs. This means that companies have an incentive to develop new treatments;

3

2) New antibiotics will be developed. Health care systems are insured against a lack of availability of antibiotics to treat patients; 3) Health care systems financial obligations can be capped, so insuring against the financial risks of a resistant infection outbreak under a premium price model.

This is a form of de-linkage model in that most of the revenues received by the manufacturer are not linked to the volumes of the drug used. For illustration, we assumed that the annual constant real terms fee (i.e. adjusted for inflation) would be paid for over ten years. Shorter terms would need higher annual fees to generate the same level of revenue. In the absence of any incentives, the price per treatment (14 days) required to increase the global eNPV from minus $1510 million to our target of a global eNPV of $100 million in the premium price model was calculated to be $14,596 per course of treatment (Table 3), which is equivalent to a daily price of $1043. If duration of treatment decreased, the daily price would need to increase proportionally. Combined with matched PPP funding for R&D and the Tier B registration pathway, price decreased to $7319. A five-year extension of market exclusivity in addition to PPP and Tier B, reduced the price to $4493. However, the daily cost of $321 per day would still produce a significant barrier to use. Under the insurance model, with no additional incentives and with a fixed price per day of $120 ($1680 per 14 day treatment course), the fee for global markets combined would be $262 million per annum for ten years (a total of $2.6 billion) to attain an eNPV of $100m. When the same incentives for R&D were applied to the insurance model, the annual premium reduces to $114 million (with PPP funding and a Tier B regulatory pathway). There is a large cross-sectional variation in antibiotic resistance. We show in Table 4 resistance prevalence rates for three drug classes to three pathogens in 7 European countries [19]. Accordingly, we conducted two-way sensitivity analysis with variation in annual growth in HAIs of 0% to 5% and a four-fold variation in the growth in prevalence of an MDR pathogen that requires treatment with the new antibiotic. The results are shown in Table 3 in brackets under each point estimate. Assuming there is matched PPP funding, Tier B, and no additional exclusivity, the lowest price per treatment that achieves the target eNPV ($3581) results when there is a 5% growth in HAIs combined with high growth in resistance equivalent to the growth of E. coli resistance to third generation cephalosporins (Table 3); however, with no growth in HAIs and low growth in resistance (equivalent to a quarter of the growth of E. coli resistance to third generation cephalosporins), the price increased to $17,611. There is therefore a five-fold difference in the drug revenues for manufacturers and drug expenditures for health care systems under plausible differences in HAI and MDR growth rates. Whilst the value of the new antibiotic will increase over time as resistance to existing antibiotics develops, the greatest volume of use is likely to come after patent expiry. Successful antibiotic stewardship would minimise growth in HAIs and spread of antibiotic resistance, making the higher price more likely to be necessary. This makes the premium price model highly unstable as a source of expected revenue for a return on investment to manufacturers and highly unpredictable to health care systems as to budget impact. It creates a tension between the need for conservation and the need for the manufacturer to get a return on R&D investment during the patent period. In the case of the insurance model, two-way sensitivity analysis of the base case produced variation in the annual fee of less than 15% (Table 3). No growth in HAIs and low growth in MDR prevalence required an annual fee of $272 million. High growth in both required an annual fee of $240m.


G Model

ARTICLE IN PRESS


A. Towse et al. / Health Policy xxx (2017) xxx–xxx

4

Table 3 Modelling ‘premium price’ and ‘insurance’ models under different scenarios (target global eNPV = $100m. Base Case

R&D incentives modelled

Business model Premium Price (per patient)a with 5 year extension of IPR Annual Insurance Premium (local price = $1680 per treatment)a with 5 year extension of IPR

$14,596 ($6632–$36,518) $8553 ($4008–$21,806) $262 m ($240 m–$272m) $250 m ($209 m–$268m)

Matched PPP Funding

Matched PPP funding & Tier B registration pathway

$8108 ($3912–$19,661) $4933 ($2539–$11,916) $130 m ($108 m–$140m) $118 m ($77 m–$136m))

$7319 ($3581–$17,611) $4493 ($2361–$10,714) $114 m ($92 m–$124m) $102 m ($61 m–$120m)

IPR: Intellectual Property Rights; eNPV: estimated Net Present Value. The point estimates are based on a base case of annual growth in the prevalence of HAIs of 3% and growth in the prevalence of resistance over 20 years equal to half of that of E. coli to third generation cephalosporins. The intervals (shown in brackets under each point estimate) are based on (i) a best case for the health system of no growth in HAIs and an MDR growth rate of a quarter of the growth in prevalence of in resistance, and (ii) a worst case of 5% annual growth in HAIs and growth in prevalence of MDR infections requiring treatment with new antibiotic equivalent to growth in resistance to that of E. coli to third generation cephalosporins. The proportion of HAIs caused by E. coli was 15.9% and reflects the mean for the EU5 countries. a Price per course of treatment. Table 4 Prevalence of antibiotic resistance in seven EU countries (2012). Country Pathogen

Class

FRA

DEU

ITA

ESP

GBR

SWE

GRC

E. colia

Cephalosporins Carbapenems fluoroquinolones cephalosporins carbapenems fluoroquinolones Cephalosporinsb carbapenems fluoroquinolones

10.0% 0.0% 17.8% 22.6% 0.5% 24.4% 18.0% 14.1% 22.2%

8.8% 0.0% 21.1% 13.0% 0.0% 13.7% 10.7% 9.6% 19.6%

26.3% 0.2% 42.0% 47.7% 28.8% 49.6% 25.1% 25.5% 31.3%

13.5% 0.1% 33.9% 16.7% 0.8% 16.5% 16.4% 8.9% 21.0%

13.1% 0.2% 16.6% 11.8% 0.5% 7.4% 6.3% 3.9% 4.8%

4.4% 0.0% 11.2% 2.8% 0.0% 3.7% 5.3% 6.2% 6.7%

16.2% 1.4% 29.1% 70.9% 60.5% 69.7% 47.7% 31.0% 44.3%

K. pneumoniae

P. aeruginosa

Source: Antimicrobial Resistance Interactive Database (EARS-NET), ECDC [16]. Data extracted November 2013. FRA: France; DEU: Germany; ITA: Italy; ESP: Spain; GBR: United Kingdom; SWE: Sweden; GRC: Greece. Percentage of pathogens resistant to selected classes of antibiotics: third-generation cephalosporins (ceph); carbapenems (carb); and fluoroquinolones (fluor). Countries are France, Germany, Italy, Spain, United Kingdom, Sweden and Greece. a Prevalence for E.coli resistance to cephalosporins for prior years are shown in Table 5 of the supplementary material. The rates shown here are for year 14 in Table 5 of the supplementary material. b For P. aeruginosa, rates are shown for ceftazidime only.

A reimbursement mechanism based on use, such as the premium price model, compounds the problem of what price to set and creates significant risk for both the health care system and the innovator. Outbreaks of resistant pathogens can be highly localised with high variation over time creating potential financial problems where funding is borne at the local level.

6. Discussion Our model suggests current incentives are not sufficient alone to drive investment in antibiotics targeted at specific resistant mechanisms. Similar conclusions have been reached by others [5;8;13;14;30]. We compare our modelling assumptions with three of these approaches in the supplementary material. All highlight the need for powerful incentives that could revitalize antibiotic innovation to protect public health, while promoting long-term sustainability. We have explored two alternative commercial models − ‘premium price’ and ‘insurance’. Although a premium price is intuitively the simplest solution, there are two important disadvantages to such a model. Firstly, it poses a large financial risk for both health care systems and manufacturers. High growth in pathogen prevalence and hence drug use would produce high costs for hospitals in areas where outbreaks occurred which required the use of high cost antibiotic treatments, exposing local health care systems to considerable financial risk. This risk of significant financial burden could act as a barrier to use of a new antibiotic during an outbreak. Implementation of the premium price model would be difficult given current reimbursement systems. The cost could be moved

outside of mechanisms such as the Diagnosis-Related Group (DRG) reimbursement system used in the US and Europe, and reimbursed separately, but the financial risk would remain high and unpredictable. Meanwhile low growth in pathogen prevalence and low drug use would produce insufficient lifetime manufacturer revenue to drive investment in future antibiotics. In contrast, the insurance model represents a risk reduction strategy for both the manufacturer and health care system. A central element of an insurance-like system is risk mitigation at the individual (provider) level by risk pooling at the aggregate level. As demonstrated in the two-way sensitivity analysis described above, the insurance model provides a mechanism by which the health care system (at local and national level) can know with relative certainty what the annual cost of providing a new antibiotic to the healthcare system will be and the manufacturer knows with relative certainty the lifetime revenue, with a net positive return to drive investment in a pipeline of new antibiotics. It can do this because it is a form of de-linkage model, with most of the revenue being separated from use of the product. At least five issues around the insurance model merit further research. First, should countries all contribute equally per capita to the annual premium? Differential contribution rates as outlined in the differential pricing literature [20] could involve an annual premium based on income per capita. There would be a need to avoid free riding by health care systems who are not contributing to the annual payment but still seek the low price in the insurance model at a local level. Second, there is a need to address the number of new drugs required and funded under the insurance model. This combines


G Model HEAP-3773; No. of Pages 6

ARTICLE IN PRESS A. Towse et al. / Health Policy xxx (2017) xxx–xxx

the question of how many drug-resistant pathogens are likely to emerge over the next decade and how the model should treat competition (e.g., second and subsequent entrants which potentially address similar unmet needs). The latter issue was considered when setting up the Advanced Market Commitment (AMC) for pneumococcal vaccine, operated by GAVI [32]. Third, institutional arrangements will be needed for a global solution to tackle the lack of new antibiotics. The insurance model would appear to offer a framework by which national health care systems can initiate pilots and coordinate implementation of complementary intra-regional schemes. For instance, European legislation for “Joint Procurement of Medical Countermeasures” includes “antimicrobial resistance and healthcare-associated infections” [33] and could facilitate implementation of a European wide insurance scheme. The evidence of the AMC suggests that using existing institutions speeds up implementation of new global mechanisms [34]. The O’Neill Report [14] proposed “a system of market entry rewards” for new antibiotics operated at a supranational level, via a political agreement at the G20. This is supported by a report commissioned by the German government, which chairs the G20 [37]. Proposals for the market entry reward to be delivered via a transferable market exclusivity voucher to overcome issues of credible commitment have also been proposed, both for the US [38] and for Europe [39]. A report from the Trans-Atlantic Task Force on Antimicrobial Resistance [40] supports, on balance, a partial delinkage option of the sort that we propose. Funding arrangements and choice of de-linkage model may need to differ by country or region as the system would need to fit with existing health system supply arrangements. Fourth, a key variable in our analyses is the target eNPV i.e. the level of returns that would be deemed as reasonable to encourage further R&D in this area. There is no specific target eNPV that can be set; Sharma and Towse [5] use $200 million (or D 147 million) while Sertkaya et al. [30] use $100 million as a benchmark agreed by the authors and the US Department of Health and Human Services. We also use a global target eNPV of $100 million as a conservative assessment. For completeness, we modelled incentives required with a (global) target eNPV of $200 million. Under the premium price model, with no additional incentives, the premium price per treatment was $15,398 (vs. $14,596 with a global eNPV target of $100 million). Under the insurance model, the annual fee increases to $278 million (vs. $262 million with a global target of $100 million). Given the size of R&D and cumulative manufacturing costs, only relatively small changes in revenue are needed to double the eNPV. Fifth, the cost of use at the local level would need to be agreed. Ideally this would be set to promote effective stewardship; sufficiently high to discourage widespread, uncontrolled use and sufficiently low to contribute minimally to the lifetime revenue for the manufacturer and discourage marketing to maximise sales volume. Under circumstances where resistance rates increase dramatically, and thus the new antibiotic needs to be used widely, there could be an option to cap volume sales with any excess revenue returned by the manufacturer to the health care system. As with the AMC agreements, the contractual arrangements between the manufacturer and the health system would cover both premium revenue and supply price together with any rebate “cap” arrangements.

7. Conclusions The current set of incentives of public support for R&D, with regulatory change, and market exclusivity are not sufficient to stimulate new investment in antibiotic R&D. Appropriate reimbursement mechanisms for future antibiotics that are founded

5

on a policy of reducing market uncertainty are needed to drive re-investment in new antibiotics to address the challenge of antimicrobial resistance. Our analysis shows that an insurance model may prove to be part of the optimal solution. It is a partial delinkage model that is consistent with recent global initiatives [14; 37; 40]. It offers (i) a reasonable return to the manufacturer; (ii) risk mitigation for payers, providers and manufacturers, (iii) stewardship by delinking revenues from volume of use, and (iv) conservation of antibiotic effectiveness by ensuring the continued availability of new antibiotics in the long-term. By contrast, a premium price model would lead to high uncertainty in lifetime revenue/expenditures due to cross-sectional and longitudinal variation in need, and would not be aligned with the development of good stewardship. Countries with the highest prevalence of resistance, and local areas suffering outbreaks would face the highest financial burden. We suggest significant efforts should be made by innovators, policy makers, physicians and payers to test the insurance model as one route to stimulate investment in novel antibiotics which meet the needs of key stakeholders. Funding This project was funded by AstraZeneca Pharmaceuticals. Disclosures AT and JMF are, respectively, an employee of, and a Visiting Fellow at, the Office of Health Economics, which received funding from AstraZeneca for the analysis and manuscript development. CKH and JJG, are employees of and shareholders in AstraZeneca Pharmaceuticals; MWH and JRH are former employees of AstraZeneca. JHR is a non-executive Board Member, Senior Advisor, and shareholder in F2G Pharmaceuticals; a consultant to and investor in Advent Life Sciences; a non-executive Board Member of Adenium Biotech ApS; and an occasional scientific advisor to other pharmaceutical companies. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.healthpol.2017. 07.011. References [1] DiMasi JA, Grabowski HG, Vernon J. R&D costs and returns by therapeutic category. Drug Information Journal 2004;38(3):211–23. [2] Falagas ME, Fragoulis KN, Karydis I. A comparative study on the cost of new antibiotics and drugs of other therapeutic categories. PLoS One 2006;1:e11. [3] Grabowski H, Vernon J, DiMasi JA. Returns on research and development for 1990 new drug introductions. Pharmacoeconomics 2002;20(Suppl. 3):11–29. [4] Projan SJ. Why is big Pharma getting out of antibacterial drug discovery? Current Opinion in Microbiology 2003;6(October (5)):427–30. [5] Sharma P, Towse A. New Drugs to Tackle Antimicrobial Resistance: Analysis of EU Policy Options. London, UK: Office of Health Economics; 2011. [6] Tillotson GS. Where does novel antibiotics R&D stand among other pharmaceutical products: an industrial perspective? Expert Review of Anti-infective Therapy 2008 Oct;6(5):551–2. [7] Williams KJ, Bax RP. Challenges in developing new antibacterial drugs. Current Opinion in Investigational Drugs 2009;10(February (2)):157–63. [8] Outterson K, Powers JH, Daniel GW, McClellan MB. Repairing the broken market for antibiotic innovation. Health Affairs (Millwood) 2015;34(February (2)):277–85. [9] Rex JH. ND4BB: addressing the antimicrobial resistance crisis. Nature Reviews Microbiology 2014;10(March (12)):231–2. [10] Sinha G. BARDA to pick and choose next-generation antibiotics. Nature Biotechnology 2013;31(August (8)):665. [11] Rex JH, Eisenstein BI, Alder J, Goldberger M, Meyer R, Dane A, et al. A comprehensive regulatory framework to address the unmet need for new antibacterial treatments. The Lancet Infectious Diseases 2013;13(March (3)):269–75.


G Model HEAP-3773; No. of Pages 6 6

ARTICLE IN PRESS A. Towse et al. / Health Policy xxx (2017) xxx–xxx

[12] Tillotson GS. GAIN Act legislation: is it enough? The Lancet Infectious Diseases 2012;12(November (11)):823–4. [13] O’Neill J. Securing new drugs for future generations: the pipeline for antibiotics; 2015. [14] O’Neill J. Review on Antimirobial Resistance. Takling drug-resistant infections globally: Final report and recommendations; 2016. [15] European Centre for Disease Prevention and Contro. Point prevalence survey of healthcare-associated infections and antimicrobial use in European acute care hospitals. Stockholm: ECDC; 2013. [16] European Centre for Disease Prevention and Control. Antimicrobial resistance interactive database; 2014. [17] Mestre-Ferrandiz J, Sussex J, Towse A. The R&D cost of a new medicine. Office of Health Economics. London: Office of Health Economics; 2012. [18] Rex JH, 2016. Personal Communication. [19] Schwaber MJ, Navon-Venezia S, Kaye KS, Ben-Ami R, Schwartz D, Carmeli Y. vClinical and economic impact of bacteremia with extended- spectrumbeta-lactamase-producing Enterobacteriaceae. Antimicrobial Agents and Chemotherapy 2006;50(April (4)):1257–62. [20] Cosgrove SE, Kaye KS, Eliopoulous GM, Carmeli Y. Health and economic outcomes of the emergence of third-generation cephalosporin resistance in Enterobacter species. Archives of Internal Medicine 2002;162(January (2)):185–90. [21] Cosgrove SE. The relationship between antimicrobial resistance and patient outcomes: mortality, length of hospital stay, and health care costs. Clinical Infectious Diseases 2006;42(January (Suppl. 2)):S82–9. [22] Laxminarayan R, Duse A, Wattal C, Zaidi AK, Wertheim HF, Sumpradit N, et al. Antibiotic resistance-the need for global solutions. The Lancet Infectious Diseases 2013;13(December (12)):1057–98. [23] Smith R, Coast J. The true cost of antimicrobial resistance. BMJ 2013;346:f1493. [24] O’Neill J. Review on Antimicrobial Resistance. Antimicrobial Resistance: Tackling a Crisis for the Health and Wealth of Nations; 2014. [25] Kettler H, Towse A. Public-private partnerships for research and development: Medicines and vaccines for diseases of poverty. London: Office of Health Economics; 2002. [26] Morel CM, Mossialos E. Stoking the antibiotic pipeline. BMJ 2010;340:c2115. [27] Mossialos E, Morel C, Edwards S, Berenson J, Gemnmill-Toyama M, Brogan D. Policies and incentives for promoting innovation in antibiotic research. London School of Economics and Political Science; 2009. [28] Nathan C, Goldberg FM. Outlook: the profit problem in antibiotic R&D. Nature Reviews Drug Discovery 2005;4(November (11)):887–91.

[29] Outterson K, Samora JB, Keller-Cuda K. Will longer antimicrobial patents improve global public health? The Lancet Infectious Diseases 2007;7(August (8)):559–66. [30] Sertkaya A, Eyraud JT, Birkenbach A, Franz C, Ackerley N, Overton V, et al., Analytical framework for examining the value of antibacterial products: Submitted to U.S. Department of Health and Human Services (HHS), Washington D.C. Task order no. HHSP23337004T. 2014. [31] Towse A, Sharma P. Incentives for R&D for new antimicrobial drugs. International Journal of the Economics of Business 2011;18(2):331–50. [32] Levine R, Kremer M, Albright A. Making Markets for Vaccines. Washington DC: Center for Global Development; 2015. [33] Decision No 1082/2013/EU of The European Parliament and of The Council of 22 October 2013 on serious cross-border threats to health and repealing Decision No 2119/98/EC. 2013., European Commission, (2013). [34] Process and Design Evaluation for the Pilot Advance Market Commitment (AMC) for Pneumococcal Vaccines. Dalberg Global Development Advisors. 2013. [35] Paul SM, Mytelka DS, Dunwiddie CT, Persinger CC, Munos BH, Lindborg SR, et al. How to improve R&D productivity: the pharmaceutical industry’s grand challenge. Nature Reviews Drug Discovery 2010;9(March (3)):203–14. [36] Payne DJ, Gwynn MN, Holmes DJ, Pompliano DL. Drugs for bad bugs: confronting the challenges of antibacterial discovery. Nature Reviews Drug Discovery 2007;6(January (1)):29–40. [37] Stern S, Chorzelski S, Franken L, Völler S, Rentmeister H, Grosch B. Breaking through the Wall; follow-up report for the German GUARD Initiative. The Boston Consulting Group & The Federal Ministry of Health; 2017. [38] Seabury, S. Sood, N. Toward a New Model for Promoting the Development of Antimicrobial Drugs. Health Affairs Blog May 18, 2017. Available at: http://healthaffairs.org/blog/2017/05/18/toward-a-new-modelfor-promoting-the-development-of-antimicrobial-drugs/. (Accessed 21 June 2017). [39] Ferraro, J., Towse, A., Mestre-Ferrandiz J. Incentives for New Drugs to Tackle Anti-Microbial Resistance Research Briefing, Office of Health Economics, May 2017. Available at: https://www.ohe.org/publications/incentives-new-drugstackle-anti-microbial-resistance. (Accessed 21 June 2017). [40] Ardal C., Rottingen J-A., Opalska A., Van Hengel A., Larsen J. 2017, Pull Incentives for Antibacterial Drug Development: An Analysis by the Trans-Atlantic Task Force on Antimicrobial Resistance.Clinical Infectious Diseases, published on line 9 June 2017 https://doi.org/10.1093/cid/cix526.
