Reimbursement and value-based pricing: stratified cost-effectiveness analysis may not be the last word

HEALTH ECONOMICSHealth Econ. 20: 688–698 (2011)Published online 21 June 2010 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/hec.1625

REIMBURSEMENT AND VALUE-BASED PRICING: STRATIFIEDCOST-EFFECTIVENESS ANALYSIS MAY NOT BE

THE LAST WORD

NEIL HAWKINS� and DAVID A. SCOTT

Oxford Outcomes, Oxford, UK

SUMMARY

During recent discussions, it has been argued that stratified cost-effectiveness analysis has a key role inreimbursement decision-making and value-based pricing (VBP). It has previously been shown that whenmanufacturers are price-takers, reimbursement decisions made in reference to stratified cost-effectiveness analysislead to a more efficient allocation of resources than decisions based on whole-population cost-effectivenessanalysis. However, we demonstrate that when manufacturers are price setters, reimbursement or VBP based onstratified cost-effectiveness analysis may not be optimal. Using two examples – one considering the choice ofthrombolytic treatment for specific patient subgroups and the other considering the extension of coverage for acancer treatment to include an additional indication – we show that combinations of extended coverage andreduced price can be identified that are advantageous to both payers and manufacturers. The benefits of a givenextension in coverage and reduction in price depend both upon the average treatment benefit in the additionalpopulation and its size relative to the original population. Negotiation regarding trade-offs between price andcoverage may lead to improved outcomes both for health-care systems and manufacturers compared with processeswhere coverage is determined conditional simply on stratified cost-effectiveness at a given price. Copyright r 2010John Wiley & Sons, Ltd.

Received 29 August 2008; Revised 29 January 2010; Accepted 29 April 2010

KEY WORDS: value-based pricing; stratified cost-effectiveness analysis

1. INTRODUCTION

Recently the UK Office of Fair Trading (OFT) reviewed the existing system for drug pricing in the UKand recommended that a drug’s price should be set based on its potential health benefits. This is referredto in the literature as value-based pricing (VBP) (OFT, 2007). Claxton (2007) further argued that ‘Theanalysis of cost-effectiveness by subgroup provides the demand curve for the National Health Service(NHS) and the price structure that can be used in value-based pricing negotiations’ and maintained thatthis use of stratified cost-effectiveness analyses is essential if the NHS is to benefit from innovation in theshort term. In contrast, the 2009 Pharmaceutical Price Regulation Scheme (PPRS), a voluntaryagreement made between the Department of Health (DH) and the branded pharmaceutical industryrepresented by the Association of the British Pharmaceutical Industry (ABPI), includes a flexible pricingscheme that allows for potential price discrimination between indications (but not for subgroups withan indication) based on differences in value (ABPI/DH, 2008). This is equivalent to allowing pricingbased on average or whole-population cost-effectiveness.

*Correspondence to: Oxford Outcomes, Seacourt Tower, West Way, Oxford OX2 0JJ, UK. E-mail: [email protected]

Copyright r 2010 John Wiley & Sons, Ltd.

Reimbursement based on stratified, rather than whole-population, cost-effectiveness analysis hasindeed been shown to lead to a more efficient use of healthcare when prices are fixed (Coyle et al., 2002).However, this paper considers what happens when stratified cost-effectiveness analysis is used as thebasis of value-based pricing (VBP) or reimbursement decision-making when manufacturers have theability to set prices.

This paper is a not a review of the feasibility or desirability of reimbursement or VBP based on cost-effectiveness, which has been discussed elsewhere by a number of authors (Claxton et al., 2008; Towse,2007; Webb and Walker, 2007; Thornton, 2007; Claxton, 2007), rather it focuses on the consideration ofsubgroups should reimbursement or VBP based on cost-effectiveness be adopted.

For health-care technologies enjoying patent protection, the market typically approximates a bilateralmonopoly, with both monopoly (a single seller) and monopsony (a single buyer) existing in the samemarket (Bowley, 1928). Price setting in a bilateral monopoly can be described as a Nash bargaining gamewith the manufacturer and payer both demanding a portion of a good (in this case the value of theadditional health benefit or cost-saving associated with a new technology); neither party will benefit if thetwo proposals sum to more than the total good (Nash, 1950). Essentially there will be a range of pricesbetween the marginal cost of production and the maximum price at which a new technology is cost-effective for a given population where agreement between parties is possible. The allocation of the value ofadditional health benefit varies within this range of prices: at the minimum price, the health-care systemretains all of the value; at the maximum price, the manufacturer retains all of the value. Rather than beingconstrained to presenting a single endogenous demand curve, the payer is free – within political constraints– to determine the reimbursement strategy and hence the demand curve faced by the manufacturer.

If the payer determines reimbursement based on stratified cost-effectiveness analysis, the allocationof the value of the benefits of a technology between the manufacturer and the payer will depend on theexistence and identification of subgroups. Given that the payer has monopsony power, it seemsinefficient to rely on such accidents of nature and analysis to determine this allocation if it is deemeddesirable that the payer should gain from innovation in the short term.

Indeed, we show in this paper that reimbursement according to stratified cost-effectiveness definesone demand curve, but not necessarily the optimal demand curve. We show that where stratified cost-effectiveness analysis would lead a manufacturer to price a treatment such that it is cost-effective in onlycertain subgroups, it is possible to identify combinations of extended coverage and reduced price thatincrease both societal health benefit and manufacturer revenues.

To illustrate this point, we present two example analyses in this paper. In the first, we comparetreatment price, revenues, and societal health benefits when the use of tissue plasminogen activator(t-PA) for thrombolysis is considered under various reimbursement scenarios. This exampledemonstrates that combinations of coverage and price may exist that increase both total societalhealth benefit and manufacturer revenues compared with reimbursement based on stratified cost-effectiveness analysis. The second example is of an analysis of a proposed reduction in price andextension of reimbursement for a cancer treatment to include an additional second line indication. Thisillustrates the appropriate criteria for determining whether a given increase in coverage and reduction ofprice will increase both total societal health benefit and manufacturer revenues.

2. EXAMPLE ONE – COMPARISON OF DIFFERENT REIMBURSEMENT SCENARIOS

In this example, we consider the reimbursement of t-PA under three different scenarios:

1. Re-imbursement based on whole-population cost-effectiveness.2. Re-imbursement based on stratified cost-effectiveness.3. Re-imbursement based on negotiated price and coverage.

VBP AND STRATIFIED COST-EFFECTIVENESS ANALYSIS 689

Copyright r 2010 John Wiley & Sons, Ltd. Health Econ. 20: 688–698 (2011)

DOI: 10.1002/hec

The assessment of t-PA is based on a trial comparing streptokinase, the standard thrombolytictreatment at the time of the trial, to t-PA (Coyle et al., 2002; Mark et al., 1995). Subjects receiving t-PAlived longer on average than subjects receiving streptokinase; the incremental effectiveness of t-PAcompared with streptokinase is expressed as life-years saved (LYS) and is varied by patient age andlocation of the infarction (anterior or inferior). Patients under 40 years with inferior infarcts areestimated to receive the least benefit from t-PA (0.014 LYS, see the first two columns of Table I)whereas patients over 75 years with anterior infarcts receive the greatest (0.212 LYS). The analysis isconducted assuming that the price of streptokinase is fixed and that the manufacturer is free to set theprice of t-PA. Furthermore, for the purposes of this paper we have assumed that all subgroups are equalin size and that t-PA is more effective than streptokinase for each subgroup.

2.1. Estimation of cost-effectiveness

The incremental cost-effectiveness ratio (ICER) for t-PA is calculated by dividing its incremental costcompared with streptokinase by its incremental effect:

ICERt-PA ¼DCt-PA

DLYSt-PA

where DCt-PA is the incremental cost of t-PA compared with streptokinase and DLYSt-PA theincremental LYS compared with streptokinase.

When the estimated ICER is the basis of decision-making, a treatment is reimbursed if the ICER isbelow an acceptable threshold which, if the objective is to maximise health benefit gained within a fixedbudget, should represent the opportunity cost of investing in the new treatment (the benefit foregonedue to reduced expenditure on the existing treatments) (McCabe et al., 2008; Buxton, 2007; Culyeret al., 2007).

If new technologies, such as t-PA, are adopted only when they have ICERs below a threshold definedin this way, the total health benefit obtained for a given budget will inevitably increase andmanufacturers will be rewarded for innovation. The benefit gained through the adoption of newtechnologies will always be greater than the disbenefit arising from disinvestment in the existingtechnologies.

For the purpose of this paper, we have assumed that the acceptable threshold is set at £20 000 perLYS. The ICERs are presented for each subgroup in each scenario. In some of the scenarios, t-PA maybe adopted for individual subgroups where the ICER is above the cost-effective threshold.

2.2. Estimation of allocative efficiency

Stinnett and Mullahy (1998) demonstrated that a decision rule based on an ICER can also be expressedin terms of incremental net health benefit (NHB).1 In the current example, incremental NHB iscalculated by subtracting the incremental cost of t-PA divided by the acceptable cost-effectivenessthreshold (l) from the incremental effect.

DNHBt-PA ¼ DLYSt-PA � DCt-PA=l

The incremental NHB represents the health benefit offered by the new technology minus the healthbenefit that would be obtained from the equivalent investment in the existing technology. In this paper,the incremental NHB is used to compare the efficiency of different reimbursement strategies. If theestimated incremental NHB is greater than zero, the new technology offers a greater health benefit thanthe existing technology at the same cost and total health benefit will increase if it is adopted. The greater

1Following the suggestion of Peter Neumann to use language that makes cost-effectiveness more acceptable to clinicians, we haveused NHB in preference to net monetary benefit (Neumann, 2005).

N. HAWKINS AND D. A. SCOTT690


DOI: 10.1002/hec

the incremental NHB, the greater the total health benefit obtained for a fixed budget and the moreefficient the scenario. The incremental NHB thus provides a measure of the allocative efficiency of eachscenario from the perspective of the payer.

2.3. Manufacturer’s pricing setting behaviour

In each scenario, it is assumed that the manufacturer will set the price of t-PA to maximise revenue inanticipation of the payer’s reimbursement decisions. In these analyses, we make the simplifyingassumption that the marginal cost of production is (or at least is believed by the manufacturer to be)independent of volume and less than the revenue maximising price. The manufacturer thereforemaximises profit by maximising revenue.

2.4. Results

The incremental LYS, costs, NHB and the revenues when t-PA is considered under the three re-imbursement scenarios are summarised in Tables I–III. The values are presented for each subgroup andas a weighted average for the whole population.

Where reimbursement is based on whole-population cost-effectiveness, payers agree to reimburse atechnology if the ICER over the whole population is below the acceptable cost-effectiveness threshold atthe price set by the manufacturer; payers do not distinguish between subgroups.

The outcomes under this scenario are shown in Table I and the left-hand column of Figure 1.Demand will be equal to the whole population until the price increases to the point where the ICERexceeds the acceptable threshold, when the demand drops to zero (Figure 1, upper plot).A manufacturer maximises revenue by setting the price so that the ICER is just below the acceptablethreshold (middle plot). At this price, the incremental NHB approaches zero and the manufacturerretains the entire monopolist’s surplus (lower plot). Revenue is maximised when the incremental price oft-PA is £1898 leading to a whole-population ICER of £20 000. The population average incrementalNHB is 0, revenue is £1898, and treatment effect is 0.95 LYS (Table I). The same overall result would beobtained if reimbursement was based on stratified cost-effectiveness analysis and the manufacturer wasable to exercise perfect price discriminate between subgroups.

Where reimbursement is based on stratified cost-effectiveness analysis, payers agree to reimburse atechnology for specific subgroups where the ICER is below the acceptable cost-effectiveness ratio at theprice set by the manufacturer.

The outcomes under this reimbursement strategy are illustrated in Table II and the middle column ofFigure 1. In this second scenario, reimbursement based on stratified cost-effectiveness, a technology willbecome cost-effective for subgroups gaining progressively less benefit as the price decreases. The shape of

Table I. Reimbursement based on the whole-population cost-effectiveness ratio. In this scenario, the revenuemaximising price for t-PA is £1898

Position Age Cost LYs ICER Adopt t-PA? NHB

Inferior o40 1898 0.014 135 536 Yes �0.081Anterior o40 1898 0.023 82 500 Yes �0.072Inferior 41–60 1898 0.038 49 934 Yes �0.057Anterior 41–60 1898 0.057 33 289 Yes �0.038Inferior 61–75 1898 0.102 18 603 Yes 0.007Anterior 61–75 1898 0.138 13 750 Yes 0.043Inferior 475 1898 0.175 10 843 Yes 0.080Anterior 475 1898 0.212 8950 Yes 0.117

Weighted average� 1898 0.095 0.000

�based on the assumption that all subgroups are of equal size.



DOI: 10.1002/hec

the demand curve therefore depends on the distribution of effects across subgroups (upper plot). Themanufacturer maximises revenue by reducing the price of a technology until the additional revenue gainedby including an additional subgroup is less than the loss of revenue caused by the price reduction across thepreviously reimbursed subgroups (middle plot). This is the point where demand becomes ‘elastic’. If morethan one subgroup is included at this price, the incremental NHB will be greater than zero (lower plot).

In this scenario, revenue is maximised when the incremental price of t-PA is £2760. At this price, t-PAis reimbursed for all patients over 75 years and for patients over 60 years with anterior infarcts. If themanufacturer were to reduce the price to £2040, t-PA would also be cost-effective for patients over 60years of age with inferior infarcts. However, the increase in coverage by a factor of 1.33 would be offsetby a reduction in price by a factor of 0.74. Similarly, if the manufacturer were to increase the price to£3500, t-PA would not be cost-effective for patients over 60 years of age with anterior infarcts. However,the increase in price by a factor of 1.27 would be offset by a decrease in coverage by a factor of 0.67.

Compared with reimbursement based on whole-population cost-effectiveness, re-imbursement basedon stratified cost-effectiveness leads to the population average incremental NHB increasing from 0 to0.014 LYS, revenue decreasing from £1898 to £1035, and treatment effect decreasing from 0.95 to 0.066LYS (Table II).

Where reimbursement is based on negotiated price and coverage, payers agree to reimburse a technologyfor a given price and coverage although the ICER may be above the acceptable threshold for some of thereimbursed subgroups. In this scenario, coverage over the whole population is offered at a price of £1328 –this is below the optimal price indicated by whole-population cost-effectiveness (£1898). The outcomesunder this reimbursement strategy are illustrated in Table III and the right-hand column of Figure 1.

Compared with reimbursement based on stratified cost-effectiveness analysis, reimbursement for thewhole population at £1328 leads to the population average incremental NHB increasing from 0.014 to

Table II. Reimbursement based on the stratified cost-effectiveness ratio. In this scenario, the revenue maximisingprice for t-PA is £2760


Inferior o40 0 0 197 143 No 0Anterior o40 0 0 120 000 No 0Inferior 41–60 0 0 72 632 No 0Anterior 41–60 0 0 48 421 No 0Inferior 61–75 0 0 27 059 No 0Anterior 61–75 2760 0.138 20 000 Yes 0.000Inferior 475 2760 0.175 15 771 Yes 0.037Anterior 475 2760 0.212 13 019 Yes 0.074

Weighted average 1035 0.066 0.014

Table III. Reimbursement based on extended coverage and reduced price. In this scenario, the negotiated price is£1328 for coverage of all subgroups


Inferior o40 1328 0.014 94 857 Yes �0.052Anterior o40 1328 0.023 57 739 Yes �0.043Inferior 41–60 1328 0.038 34 947 Yes �0.028Anterior 41–60 1328 0.057 23 298 Yes �0.009Inferior 61–75 1328 0.102 13 020 Yes 0.036Anterior 61–75 1328 0.138 9623 Yes 0.072Inferior 475 1328 0.175 7589 Yes 0.109Anterior 475 1328 0.212 6264 Yes 0.146

Weighted average 1328 0.095 0.028



DOI: 10.1002/hec

0.028 LYS, revenue increasing from £1035 to £1328, and treatment effect increasing from 0.066 to 0.095LYS (Table III).

2.5. Summary

We have illustrated that in some circumstances it may be possible to identify a combination of price andcoverage that leads to increases in both incremental NHB and manufacturers’ revenues, an outcomethat should be attractive to both parties. Figure 1 is annotated to illustrate the range of prices wherereimbursement for the total population would lead to a better outcome than stratified cost-effectivenessanalysis. The grey shaded region on the lower plot shows the range of prices where total health benefit isincreased compared with stratified cost-effectiveness analysis; in the middle plot it shows the range ofprices where revenue is increased; and in the upper plot it shows the range of prices where bothincremental NHB and revenues are increased.

3. EXAMPLE TWO – CRITERIA FOR EVALUATING CHANGES IN PRICE AND COVERAGE

In this example, we outline the criteria for determining whether a given extension of coverage andreduction in price is desirable. A change of coverage with an associated reduction in price will lead to an

Whole Pop. CE

Dem

and

(%)

0

Rev

enue

(£,

000s

)

0 1 2 3 4

Price (£,000s)

INH

B (

LYs)

0.0

0.1

Stratified CE

0 1 2 3 4

Negotiation

0 1 2 3 4

Region where revenue and health benefitincreased

Region where revenue increased compared to stratified CEanalysis

Region where heath benefit increasedcompared to stratified CEanalysis

1

0

1

2

Figure 1. The relationship between price, demand, and revenue. The three columns of plots representing the threescenarios, the upper row shows the relationship between demand and price, the middle row shows the relationshipbetween revenue and price, and the bottom row shows the relationship between price and revenue. The black dotsindicate the revenue maximising price for each scenario. The grey shaded areas in the third column show the rangeof prices that results in both increased revenue and health benefit (top plot), increased revenue (middle plot), and

increased health benefit (bottom plot)



DOI: 10.1002/hec

increase in NHB and will be advantageous to payers if the new price (cnew) is below a given threshold:

cnew � ðccurr1r � enew � lÞ=ð11rÞ ð1Þ

where ccurr is the current price, r is the ratio of the size of the additional population to the currentpopulation, enew is the mean effect over the additional population, and l is the maximum acceptablecost-effectiveness threshold.

Adopting the alternative perspective, increased coverage with an associated reduction in price will beadvantageous to manufacturers if the new price is above a given threshold:

cnew � ccurr=ð11rÞ ð2Þ

For a given increase in coverage, at the minimum identified price revenue will be the same as understratified cost-effectiveness analysis, at the maximum identified price the total health benefit will be thesame as under stratified cost-effectiveness analysis. At a negotiated price between the minimum andmaximum, manufacturer revenues, total health benefit, and the covered population are all increased.

If a new price and coverage are proposed, and the current price (ccurr) and mean effect (enew) over theadditional population are known, we can describe these thresholds in terms of the ratio of the additionalpopulation to the current population (r). The ratio that would lead to an increase in NHB is:

r� ðcnew � ccurrÞ=ðenew � l� cnewÞ ð3Þ

And the ratio that would lead to an increase in revenue for manufacturers is:

r� ðccurr=cnewÞ � 1 ð4Þ

The acceptable cost-effectiveness threshold may vary between different patient subgroups to allow forsocietal preferences, for example end-of-life treatments (NICE, 2009; Waugh and Scott, 1998) and mayalso vary between investment and disinvestment decisions (Dowie, 2004). In these cases, fomulae 1 and3 would be applied individually to subgroups to account for the variation in acceptable thresholds.

3.1. Example application

In this section, we demonstrate an application of these criteria. Consider a cancer treatment that hasapproval for both first- and second-line use. The mean incremental benefit of treatment is 1 QALY inthe first-line indication and 0.5 QALYs in the second-line indication (referred to as enew in the previousequations). The manufacturers have set the price of a course of treatment at £20 000 (ccurr). Payersconsider the maximum acceptable cost-effectiveness ratio to be £20 000 (l) and with reimbursementdetermined according to stratified cost-effectiveness analysis, the payer currently only funds treatmentfor the first-line indication.

It is proposed that reimbursement is extended to include second-line treatment in return for areduction in price. Figure 2 shows both the maximum price (from Equation (1)) that would beacceptable to the payer and the minimum price (from Equation (2)) that would be acceptable to themanufacturer as a function of the ratio of the size of the second-line treatment population comparedwith the first-line population.

For example, if the second-line population were half the size of the first-line population, themaximum price that would be acceptable to the payer is £16 667 and the minimum price that wouldacceptable to the manufacturer is £13 333. Whereas, if the second-line population were of the same sizeas the first-line population, the maximum price that would be acceptable to payers is £15 000 and theminimum price that would be acceptable to manufacturers would be £10 000.



DOI: 10.1002/hec

3.2. Summary

This example shows that the acceptability of proposed extension in coverage and reduction in price topayers and manufacturers will depend on their respective beliefs as to the relative size of the additionalpopulation and the benefit gained from the treatment in this population.

4. DISCUSSION

Where manufacturers are able to set the prices, they will do this to maximise profits in anticipation ofthe demand curve they face. The demand curve is determined primarily by payers’ reimbursementdecisions. In some jurisdictions, such as France, manufacturers are explicitly notified of the maximumprice at which a technology will be reimbursed. In other jurisdictions, such as the UK, manufacturerscurrently have to anticipate future reimbursement decisions when setting prices (Garattini et al., 2007).In the absence of a formal system of VBP, the methods used to determine reimbursement may influencepricing decisions.

We have shown, using the two examples presented in this paper, that where manufacturers of health-care technologies have price setting power, reimbursement or VBP based on stratified cost-effectivenessanalysis may lead to a sub-optimal combination of price and coverage. Specifically, we show that wherethe application of stratified cost-effectiveness analysis would lead a manufacturer to set a price resultingin reimbursement of selected subgroups, it is possible to negotiate combinations of increased coverageand reduced price where both manufacturer revenues and total societal health benefit are increased.

This negotiated reduction in price and extension in coverage both increases equity, as a wider rangepatients receive the new technology, and increases allocative efficiency. However, this observation isonly of practical relevance if (1) manufacturers are able to set prices and (2) there is a viable mechanismfor the negotiation of combinations of extended coverage and reduced price.

In general, manufacturers are free to set the price for a technology in markets where there are fewclose substitutes. This commonly occurs when manufactures enjoy patent protection, although theexistence of a patent does not preclude the development of novel substitute technologies. Where close

0.0 0.2 0.4 0.6 0.8 1.0

10000

12000

14000

16000

18000

20000

Ratio of Additional Population to Existing Propulation

Pric

e

Maximum Price

Minimum Price

Figure 2. The relationship between maximum and minimum price and the ratio of the additional populationto the existing population



DOI: 10.1002/hec

substitutes exist, the price of a technology will be driven down to the marginal cost of production. Theobservation that for a number of technologies, such as statins, prices dropped markedly when genericsentered the market (OFT, 2007; Hoyle, 2008) suggests that, at least for some technologies,manufacturers do enjoy some price setting ability.

In principle, manufacturers maximise global profits by maximising profits in each local market. Theywill do this by setting prices independently in each market. However, parallel trade and reference pricingarrangements may restrict manufacturers’ ability to price discriminate in this way (Kyle et al., 2008;Schneeweiss, 2007; Danzon and Ketcham, 2004; Danzon and Towse, 2003; Danzon, 1998). In response,manufacturers may attempt to counter these arrangements by offering discounts from list prices, the useof risk share schemes, and other forms of rebate (Lansley, 2008; Wind, 2008; Pollack, 2007).

If manufacturers are unable to counter these arrangements, they will set prices to maximise profits inthe larger markets at the expense of profits in the smaller markets. Countries with lower acceptable cost-effectiveness thresholds will also want to encourage price discrimination, whereas those with higheracceptable cost-effectiveness thresholds will tend to favour reference pricing and parallel trade (OFTreport). We might therefore expect some payers to respond by aiding manufacturers’ efforts to pricediscriminate. For example in the UK, the report on VBP produced by the OFT explicitly mentioned theuse of rebates to relax global trading constraints (OFT, 2007).

Overall, it is likely that at least for some products in some markets manufacturers will have pricesetting power. Regarding the viability of negotiation, one option is that the manufacturer initiallyproposes a price, the decision-maker then determines which (if any) subgroups it is prepared to cover atthat price. If the decision-maker is not prepared to cover all subgroups at that price, it can then offer themanufacturer the option of extended coverage at a reduced price. The manufacturer would then acceptthis option if it would lead to increase profits. This process is not predicated on the decision-makerhaving any knowledge of the manufacturer’s costs of production or price setting ability.

For example, in the second example presented in this paper, the manufacturer might initially suggesta price of £20 000 for the cancer treatment, anticipating reimbursement for first-line use only. Thedecision-maker could then respond with the option of coverage for both first- and second-line use at areduced price of £12 500. The manufacturer would be free to accept or reject this offer based on theirbelief in the relative size of the first- and second-line markets and the incremental cost of production.

This type of negotiation may require a change in constitution for some pricing and reimbursementauthorities to allow for explicit negotiation on price and coverage. For example, despite the fact thatanticipation of reimbursement decisions may influence pricing, the Chairman of NICE, Sir MichaelRawlins, does not view NICE as a pricing agency: ‘What we look at is the price that is being proposedby a company, and that’s the beginning and end of our remit in this area’ (Barham and Mansell, 2009).However, recent negotiation with Roche on their oncology drug erlotinib suggests some flexibility maybe creeping into this role (NICE, 2008) and the latest UK PPRS negotiation describes ‘flexible pricing’arrangements that formalise the process of such VBP negotiations (ABPI/DH, 2008). And as we haveshown in this paper, restricting the ability of reimbursement agencies to negotiate potential changes inprice and coverage may lead to allocative inefficiency.

There is, however, the possibility that the option of negotiation as described in this paper might leadto gaming. For example, manufacturers might pre-empt the option of negotiation by initially offering ahigher price then they would otherwise have done. If, in the absence of negotiation, manufacturerswould have set a price that would have led to all subgroups being covered, this gaming will lead to areduction in total societal health benefit. In the previous example, if negotiation was not an option theymight have set a price of £10 000, anticipating reimbursement for both first- and second-line use.Decision-makers must balance this possibility of gaming against the potential increases in allocativeefficiency if negotiation is allowed.

The benefits of a potential extension of coverage and reduction in price depend on the size of theadditional population relative to the current population. In practice, this proportion may be difficult to



DOI: 10.1002/hec

estimate, especially in advance. This uncertainty might be addressed by designing a price volumeagreement that ensures that the price is within an acceptable range whatever be the relative size of theadditional population.

A wider concern that a move to reimbursement or VBP based on cost-effectiveness may increase theopportunity cost associated with the adoption of new technologies increasing as manufacturers’ price tothe ‘upper limits of cost [-effectiveness]’ has also been raised (Webb and Walker, 2007). However, in theabsence of reimbursement or VBP decision-making based on cost-effectiveness manufacturers may pricebeyond the upper limits of cost-effectiveness. In this case, the adoption of such technologies will lead toa reduction in total population health.

The cost-effectiveness threshold is a function of available health-care budget, prevailing technologies,and prices, and hence may vary over time (McCabe et al., 2008; Buxton, 2007). Consequently, previousreimbursement and pricing decisions should be periodically re-evaluated and existing technologiessubject to evaluation and potential disinvestment. This may require consideration of any ‘kink’ in theacceptable cost-effectiveness threshold between investment and disinvestment decisions reflectingdifferences between consumer thresholds for giving up existing QALYs compared with gaining newQALYs (Buxton, 2007; Dowie, 2004; O’Brien et al., 2002).

In summary, reimbursement or VBP processes that allow for negotiation regarding trade-offsbetween price and coverage may lead to improved outcomes, both for health-care systems andmanufacturers, compared with processes where coverage is determined based on a stratified cost-effectiveness at a given price.

ACKNOWLEDGEMENTS

Funding: The authors did not receive any specific financial support for this work.

Conflict of interest: The authors have no conflicts of interest to declare with respect to this work.

REFERENCES

The Association of the British Pharmaceutical Industry and the Department of Health. 2008. The PharmaceuticalPrice Regulation Scheme 2009. Department of Health: London.

Barham L, Mansell P. 2009. Assessing NICE’s expanding role. PharmaTimes, 27–29 April.Bowley A. 1928. Bilateral monopoly. The Economic Journal 38: 651–659.Buxton M. 2007. Looking for willingness-to-pay (WTP) threshold for a QALY – does it make sense? a practical

view. ISPOR Connections, 15 August 9–11.Claxton K. 2007. OFT, VBP: QED? Health Economics 16(6): 545–558.Claxton K, Briggs A, Buxton MJ, Culyer AJ, McCabe C, Walker S, Sculpher MJ. 2008. Value based pricing for

NHS drugs: an opportunity not to be missed? British Medical Journal 336(7638): 251–254.Coyle D, Buxton M, O’Brien B. 2002. Stratified cost-effectiveness analysis: a framework for establishing efficient

limited use criteria. Health Economics 12(5): 421–427.Culyer A, McCabe C, Briggs A, Claxton K, Buxton M, Akehurst R, Sculpher M, Brazier J. 2007. Searching for a

threshold, not setting one: the role of the National Institute for Health and Clinical Excellence. Journal of HealthServices Research and Policy 12(1): 56–58.

Danzon PM. 1998. The economics of parallel trade. Pharmacoeconomics 13: 293–304.Danzon PM, Ketcham JD. 2004. Reference pricing of pharmaceuticals: evidence from Germany, the Netherlands

and New Zealand. Frontiers in Health Policy Research 7: 1–54.Danzon PM, Towse A. 2003. Differential pricing for pharmaceuticals: reconciling access, R&D and patents.

International Journal of Health Care Finance and Economics 3: 183–205.Dowie J. 2004. Why cost-effectiveness should trump (clinical) effectiveness: the ethical economics of the South West

quadrant. Health Economics 13: 453–459.



DOI: 10.1002/hec

Garattini L, Cornago D, Compadri PD. 2007. Pricing and reimbursement of in-patent drugs in seven Europeancountries: a comparative analysis. Health Policy 82(3): 330–339.

Hoyle M. 2008. Future drug prices and cost-effectiveness analysis. Pharmacoeconomics 26(7): 589–602.Kyle MK, Allsbrook JS, Schulman KA. 2008. Does reimportation reduce price differences for prescription drugs?

Lessons from the European Union. Health Services Research 43: 1308–1324.Lansley A. 2008. Nice doesn’t have to be nasty. Daily Telegraph, 10 September.Mark D, Hlatky M, Califf R, Naylor C, Lee K, Armstrong P, Barbash G, White H, Simoons ML, Nelson CL. 1995.

Cost effectiveness of thrombolytic therapy with tissue plasminogen activator as compared with streptokinase foracute myocardial infarction. New England Journal of Medicine 332: 1418–1424.

McCabe C, Claxton K, Culyer AJ. 2008. The NICE cost-effectiveness threshold: what it is and what it means.Pharmacoeconomics 26(9): 733–744.

Nash J. 1950. The bargaining problem. Econometrica 18: 155–162.National Institute for Health and Clinical Excellence. 2008. Final Appraisal Determination – Erlotinib for the

Treatment of Non-small-cell Lung Cancer. NICE: London, September.National Institute for Health and Clinical Excellence. 2009. Appraising Life-extending, End of Life Treatments.

NICE: London, January.Neumann P. 2005. Using Cost-effectiveness Analysis to Improve Health Care: Opportunities and Barriers. Oxford

University Press: Oxford.O’Brien BJ, Gertsen K, Willan AR, Faulkner LA. 2002. Is there a kink in consumers’ threshold value for cost-

effectiveness in health care? Health Economics 11: 175–180.Office of Fair Trading. 2007. The pharmaceutical price regulation scheme. An OFT market study. Technical Report,

Office of Fair Trading, London, UK.Pollack A. 2007. Pricing pills by the results. New York Times, 14 July.Schneeweiss S. 2007. Reference drug programs: effectiveness and policy implications. Health Policy 81: 17–28.Stinnett A, Mullahy J. 1998. Net health benefits: a new framework for the analysis of uncertainty in cost-

effectiveness analyses. Medical Decision Making 18: S68–S80.Thornton S. 2007. Drug price reform in the UK: debunking the myths. Health Economics 16(10): 981–992.Towse A. 2007. If it ain’t broke, don’t price fix it: the OFT and the PPRS. Health Economics 16(7): 653–665.Waugh N, Scott DA. 1998. How should different life expectancies be valued? British Medical Journal 316: 1316.Webb DJ, Walker A. 2007. Value-based pricing of drugs in the UK. Lancet 369: 1415–1416.Wind K. 2008. Risk sharing schemes – improving patient access to new drugs. Hospital Pharmacist 15: 114.



DOI: 10.1002/hec

Documents

Reimbursement and value-based pricing: stratified cost-effectiveness analysis may not be the last word