A Tale of Two Thresholds: A Framework for Prioritization within the Cancer Drugs Fund

VALUE IN HEALTH 19 (2016) 567–576

Available online at www.sciencedirect.com

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

Health Policy Analysis

A Tale of Two Thresholds: A Framework for Prioritization within the Cancer Drugs Fund Simon Leigh, BSc, MSc1,2, Paul Granby, BSc, MSc1,3,* 1

Lifecodes Solutions, Liverpool, UK; 2Nexus Clinical Analytics, Euxton, UK; 3Certus Analytics, Formby, UK

AB STR A CT

Backgound: The Cancer Drugs Fund (CDF) has been the subject of controversy since its inception, with critics arguing that it creates a “backdoor” to the National Health Service (NHS), circumventing the National Institute for Health and Care Excellence and its health technology assessment program. Nonetheless, with its creation comes a new decision problem, how to best allocate resources among cancer drugs. Objectives: Our objective was to estimate CDF’s willingness and ability to pay for cancer drugs, providing guidance regarding where CDF funds are best spent, and determining the number of NHS quality-adjusted life-years (QALYs) displaced through the existence of the fund. Methods: Using CDF utilization figures, cost-per-QALY, and treatment episode costs from National Institute for Health and Care Excellence health technology assessment reports, the league-table approach was applied to determine appropriate cost-effectiveness thresholds to inform the CDF’s decision making. Results: The CDF exhibits a willingness-to-pay value of £223,627 per QALY, with 74% and 33% of expenditure for drugs with incremental

Introduction National Health Service (NHS) England's Cancer Drugs Fund (CDF) has been the subject of controversy since its inception, with many critics arguing that it creates a “backdoor” to the NHS, which circumvents National Institute for Health and Care Excellence (NICE) and its health technology assessment (HTA) program [1–4], imposing opportunity costs in terms of population health. Although the debate surrounding the existence of the fund has been lively, to date, very little attention has been paid to resource allocation within the fund. NICE applies a universal decision rule in which the incremental cost-effectiveness ratio (ICER) of an intervention must, at least in theory, be below a predefined cost-effectiveness threshold in order to be considered a good use of scarce NHS resources. The exact value this threshold should take has been the subject of much debate, with arguments for higher [5–7], lower [8,9], varying [10], and flexible [11] thresholds throughout the literature. Of particular relevance to the CDF is the move away from NICE’s default position

cost-effectiveness ratios of more than £50,000 and more than £90,000, respectively. During 2013-2014, CDF expenditure generated 4,677 QALYs, compared with a potential 13,485 if the same funds were used as part of routine NHS commissioning, displacing 8,808 QALYs. By ring fencing 10%, 25%, and 50% of the CDF budget for the provision of unevaluated drugs, cost-effectiveness thresholds of £149,000, £111,400, and £68,600 were calculated, respectively. Conclusions: Adopting the proposed framework for CDF prioritization would result in disinvestment from a number of highly cost-ineffective drugs applicable for CDF reimbursement. The present lack of a formal economic evaluation not only results in net health losses but also compromises a founding principle of the NHS, that of “equal access for equal need.” Keywords: Cancer Drugs Fund, health economics, prioritization. Copyright & 2016, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc.

that “a QALY is a QALY is a QALY,” the emergence of “value-based assessment,” and the growing support for reforming HTA to allow for the inclusion of “wider health benefits.” These include the continuing debate around the “super-QALY” [12–14], disease severity [15–18], and “end-of-life” weightings [19–21], all of which are of significant relevance to the CDF. However, perhaps the greatest barrier to such reform is not ideological or philosophical but technical. Although we are aware that funding one intervention necessitates imposing opportunity costs on others, we cannot be certain where these opportunity costs are borne. Comparing “known” and “unknown” means we cannot rationally attach an “equity weight” to an intervention under evaluation, unless we are able to identify and apply the correct weight to the unknown bearer of the opportunity cost and its corresponding patient group. Although the wider NHS decision problem encounters efficiency-limiting information constraints [22] and issues surrounding disinvestment [23,24], the CDF does not. Because the CDF is exclusively meant for the funding of drugs and exclusively for

* Address correspondence to: Paul Granby, Lifecode Solutions, 54 St. James Street, Liverpool, Merseyside, L1 0AB E-mail: [email protected]. 1098-3015$36.00 – see front matter Copyright & 2016, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. http://dx.doi.org/10.1016/j.jval.2016.02.016

568

VALUE IN HEALTH 19 (2016) 567–576

the therapeutic area of cancer, the technical problem of searching for an accurate cost-effectiveness threshold becomes much simpler. In contrast to the NHS budget, in which opportunity costs are borne somewhere within a vast array of interventions across numerous therapeutic areas, we need only consider a much narrower range of health technologies, 74 at the time of writing, making the adoption and disinvestment of cancer drugs and services an issue of mathematical programming. Moreover, we need not concern ourselves with whether a cancer quality-adjusted life-year (QALY) ought to be valued more highly than a non-cancer QALY because all QALYs are accrued to cancer patients. Despite the increased simplicity of cancer-specific threshold determination, it should be noted that the CDF’s purpose implies two potentially competing objectives, making this process more challenging. The first is to offer an extended willingness to pay for cancer drugs beyond that of the NICE cost-effectiveness threshold. The second, similar to the European Medicines Agency’s “adaptive pathways approach” [25], is to speed up and increase access to cancer drugs that are yet to be assessed by NICE, currently under evaluation, or have had their assessments suspended or terminated, henceforth referred to as “unevaluated” treatment indications. Because the relative importance of these two CDF objectives has not been made explicit, any threshold necessarily becomes a function of how resources are to be allocated between the two, and can only apply to the former case, in which cost-effectiveness is known. Given that we do not know the precise objective function of the CDF, but do know the likely arguments within, in an attempt to formalize this process we propose a series of funding splits for the provision of known (NICE evaluated) and unknown (unevaluated) cancer drugs. Using historical CDF utilization data and NICE HTA and evidence review group (ERG) reports, this article determines each drug’s impact on CDF’s ability to pay for drugs and the subsequent potential for QALY generation, calculating corresponding candidate thresholds for treatments of known cost-effectiveness for each of these funding splits.

Methods An analysis was undertaken to identify the cost-effectiveness and affordability of all treatments currently subject to reimbursement under the CDF, enabling the estimation of a range of suitable cost-effectiveness thresholds that may be used to inform CDF decision making, dependent on a range of potential funding allocations between evaluated (known) and unevaluated (unknown) treatment indications.

Data Sources and Extraction We conducted a search of the NHS UK Cancer Drugs Fund Web site [26] to reveal all drugs subject to reimbursement under the most recent edition of the “CDF-approved list” at the time of writing. Because the treatment duration, clinical effectiveness, and cost-effectiveness of these drugs differ with respect to the form of cancer they are indicated to treat, every therapeutic indication for each of these drugs was identified. Following the identification of all relevant treatment options reimbursed by the CDF, a literature search was conducted with the sole purpose of identifying evidence concerning cost-effectiveness and drug acquisition costs. For the purpose of data collection and synthesis, no systematic search criteria or date restrictions were applied. The selective literature search was initially confined to the NICE database, to limit our results to those solely from an NHS perspective. For each intervention, where available, full copies were obtained of the NICE technology appraisal resulting in the initial

rejection or “optimized” recommendation for funding under routine NHS care. Data were extracted concerning total drug acquisition costs per indication and the estimated ICERs versus the next best standard of care, expressed in terms of the incremental cost per QALY. Because NICE HTA reports often contain multiple estimates of ICERs, contingent on numerous assumptions and conditions being met, the ICERs used were those identified as “most plausible” by the ERGs. Furthermore, because the remit of the CDF only permits the reimbursement of “drugs,” and not the associated nursing and chemotherapeutic costs, every effort was made to ensure that the costs reported were in fact those solely associated with drug acquisition. The expected utilization of each CDF intervention was obtained from historical CDF audit reports [26]. For indications for which NICE technology assessments were not available, this information was obtained using a range of resources including the NICE Web site, ERG reports, notices of HTA suspension and/or termination, and final appraisal determinations. If data were still unavailable, either due to HTA suspension and/or termination or simply having not yet undergone NICE HTA, these treatment indications were labeled as unevaluated due to being in use but having ultimately unknown cost-effectiveness or budget impact. In cases in which evidence obtained from NICE was incomplete, that is, information were available concerning the estimated cost per QALY but not the estimated cost per treatment episode, these were estimated using recommended dosing regimens contained within the HTA. If this information were also not available, treatment costs were sourced from the literature, converted to pounds sterling, and inflated to represent their net present value as appropriate.

Threshold Search: League-Table Approach This model makes use of the league-table search approach to threshold determination and applies it to the CDF. First proposed by Gafni and Birch [27,28], and the subject of much debate within the health economics literature [9,29], the league-table approach is a well-validated method for cost-effectiveness threshold elicitation, both in theory and in practice. Although excessive informational requirements render this approach infeasible for informing NHS-level questions of resource allocation [23,24,29], this approach has been successfully applied to more modest NHS decision problems, including the estimation of “local” costeffectiveness thresholds at the primary care trust level [24]. As such, given the highly bounded nature of our hypothesis, a resource allocation problem considering just 74 treatment options within a single therapeutic area, we deemed this method to be the most accurate, flexible, and practical means of determining a CDF-specific threshold to inform future resource allocation. Under this approach, interventions are ranked in order of decreasing cost-effectiveness (increasing ICERs), expressed in terms of the cost per QALY gained, and adopted sequentially until the budget is exhausted. As presented in Table 1, adapted from Appleby et al. [24], the relevant threshold theoretically lies between indications X and Y, the point between the ICER of the last (least cost-effective) or marginal intervention funded (CDF indication X) and that of most cost-effective service not currently funded (CDF indication Y). To adopt interventions beyond this threshold (e.g., CDF intervention N) necessitates disinvestment from others of greater cost-effectiveness that produce greater health gains for every pound spent, resulting in an unambiguous loss in population health. In the event that the CDF budget exceeded the costs of satisfying demand for every treatment indication, the benchmark interventions approach [30] was applied to estimate the threshold on the basis of the maximum willingness to pay demonstrated when funding previous CDF interventions.

569

VALUE IN HEALTH 19 (2016) 567–576

Table 1 – Theoretical implications of the league-table approach to threshold determination. Treatment indication CDF CDF CDF CDF CDF CDF

indication indication indication indication Indication Indication

ICER (cost-per QALY)

Cost per year

Cumulative budget

Remaining budget

£ICER1 £ICER2 £ICER3 £ICERX £ICERY £ICERN

£C1 £C2 £C3 £CX £CY £CN

£C1 P £C1 þ £C2 ¼ £ C2 P P £ C2 þ £C3 ¼ £ C3 P P £ C(X1) þ £CX ¼ £ CX P P £ CX þ £CY ¼ £ CY P £ CN

£CDF  £C1 4 0 P £CDF  £ C2 4 0 P £CDF  £ C3 4 0 P £CDF  £ CX 4 0 P £CDF  £ CY o 0 P £CDF  £ CN o 0

1 2 3 X Y N

CDF, Cancer Drugs Fund; ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life-year.

Scenario Analyses: Range of Candidate Thresholds Because there are no existing data regarding the budget impact of unevaluated drugs, it is unclear how the purchase of such drugs will affect the existing budget for drugs of known cost-effectiveness and budget impact, and ultimately the affordability of these evaluated drugs. As such, we propose a total of four candidate scenarios that cover a broad range of potential levels of investment in unevaluated treatment indications, equal to 0%, 10%, 25%, and 50% of budget spending.

Secondary Analyses Secondary analyses demonstrate the net health impact of CDF spending, in terms of QALYs gained, versus the marginal productivity of alternative services provided through the NHS. In this instance, the marginal productivity of NHS services is represented by the lower limit of the NICE cost-effectiveness threshold, equal to £20,000 per QALY gained. Additional analyses highlight the potential for QALY generation by treatment indication, to determine which services provide the greatest health gains in terms of mortality and morbidity among cancer patients, in addition to estimating overall investment and ability for health improvement in the following forms of cancer:

















Melanoma Leukemia Myelodysplastic syndromes Gastrointestinal cancer Renal cell carcinoma Myeloma Hepatocellular carcinoma Prostate cancer Colorectal cancer Non–small cell lung cancer Breast cancer Myelofibrosis Ovarian/fallopian/epithelial cancer Head and neck cancer

As such, an “average cost per QALY” by cancer type was estimated for each of the 14 different forms of cancer for which treatments were available through the CDF. In doing so, the total expenditure on all drugs within a specific etiology was divided by the total number of QALYs generated as a result. This provided an estimate of the current average cost per QALY within each of the 14 forms of cancer reimbursed under the CDF, and providing an indication as to where these funds are presently best spent.

Results Health Technology Assessment: Evaluated versus Unevaluated In total, 74 treatment indications were provided through the CDF during the study period of April 2013 to March 2014. Of these, 31 were yet to issue HTA reports estimating their clinical/costeffectiveness and budget impact, 6 resulting from terminated appraisals, 2 with guidance currently in development, and 23 with no guidance at all. In total, these treatment indications collectively accounted for more than one-third (38.3%) of CDF activity, corresponding to 7,386 of 19,282 funded treatment episodes during 2013-2014. Abiraterone for the treatment of metastatic castration-resistant prostate cancer accounted for more than 40% of this “unevaluated” utilization, with a total of 3023 treatment episodes approved, equal to 15.7% of total CDF activity and making this unevaluated treatment indication the most commonly approved within the CDF.

CDF: Willingness to Pay Table 2 lists all evaluated CDF treatment indications, their indications approved during the study period, and costs per treatment episode and cost-effectiveness (ICER), in terms of cost per QALY gained, while Figure 1 demonstrates the respective utilization of CDF-funded treatments grouped by ICER. Those with ICERs of less than £50,000, £50,000 to £90,000, and more than £90,000 accounted for 26.1%, 41%, and 32.9% of CDF expenditure, respectively, reaching a maximum of £223,627 per QALY in the case of cetuximab for the second- or third-line treatment of metastatic colorectal cancer.

CDF: Ability to Pay Applying the league-table approach to our “base-case” scenario in which only evaluated treatment indications are subject to reimbursement, demand for all treatment indications, from the least to the most cost-effective, could be met with a surplus budget of £10.3 million remaining. This suggests that in the base case it is not possible to determine an appropriate threshold using the league-table approach because ability to pay is at least as great as willingness to pay. As such, the benchmark interventions approach suggests that the appropriate threshold should be equal to the ICER of the least cost-effective treatment indication subject to reimbursement at that time, corresponding to £223,627. Under scenarios 1, 2, and 3 with 10% (£28 million), 25% (£70 million), and 50% (£140 million) of the £280 million annual CDF budget formally ring fenced for the provision of unevaluated drugs, the league-table approach estimates appropriate costeffectiveness thresholds, and corresponding CDF ability to pay up to maximum ICERs of £149,000, £111,400, and £68,600 per

570

VALUE IN HEALTH 19 (2016) 567–576

Table 2 – Cost per QALY and treatment cost of “evaluated” CDF-approved treatment indications. Treatment

Indication

Cost per QALY (£)

Indications approved (2013-2014)

Treatment cost (£)

Evidence

Bendamustine Dabrafenib

Chronic lymphocytic leukemia Unresectable or metastatic BRAF V600 mutation-positive melanoma Refractory chronic-phase chronic myeloid leukemia Myelodysplastic syndromes associated with a deletion 5q cytogenetic abnormality Adjuvant treatment of gastrointestinal stromal tumor (high-risk) Chronic-phase chronic myeloid leukemia First-line metastatic colorectal cancer Accelerated-phase chronic myeloid leukemia Bortezomib relapsed multiple myloma Advanced breast cancer Relapsed multiple myeloma Advanced renal cell carcinoma Chronic-phase chronic myeloid leukemia where there is intolerance of treatment(s) Castrate-resistant metastatic prostate cancer In combination with irinotecan-based chemotherapy for second-line treatment of metastatic colorectal cancer Third-/fourth-line treatment of metastatic colorectal cancer as a single agent First-line advanced hepatocellular carcinoma Refractory accelerated- phase chronic myeloid leukemia Blast crisis chronic myeloid leukemia ALK þve advanced or metastatic non– small cell lung cancer Adjuvant treatment of gastrointestinal stromal tumor (significant-risk) Refractory blast crisis chronic myeloid leukemia Second-/third-line treatment of advanced colorectal cancer Accelerated-phase chronic myeloid leukemia where there is intolerance of treatments Second-line treatment of multiple myeloma Advanced breast cancer Advanced breast cancer First-line treatment of advanced colorectal cancer with combination chemotherapy Metastatic renal cell carcinoma Chronic lymphocytic leukemia Maintenance treatment of advanced nonsquamous non–small cell lung cancer

9,400 11,000

337 6

7,673 78,396

NICE TA216 [31] NICE TA321 [32]

20,972

31

130,752

NICE TA299 [33]

25,300

55

54,243

NICE TA322 [34,35]

27,600

33

15,017

NICE TA196 [36]

28,000

61

314,413

NICE TA241 [37]

33,300 34,500

58 4

22,796 135,570

NICE TA176 [38] NICE TA241 [37]

35,750 37,336 38,000 40,967 43,000

106 341 294 496 31

25,000 25,207 3,000 21,069 44,799

NICE TA129 [39] NICE FAD [40] NICE TA129 [39] NICE TA333 [41,42] NICE TA299 [33]

48,000

656

25,269

51,000

778

8,816

NICE press release [43] NICE press release [44]

52,054

42

11,789

NICE TA118 [45]

52,600

427

11,665

NICE TA189 [46,47]

53,789

2

126,237

NICE TA299 [33]

54,093 55,157

14 78

88,181 44,546

NICE TA241 [37] NICE TA296 [48]

56,350

33

14,795

NICE TA196 [36]

59,191

0

49,936

NICE TA299 [33]

59,993

397

16,824

NICE TA118 [45]

60,000

2

44,799

NICE TA300 [49]

63,000

147

51,800

NICE TA171 [50]

68,000 68,600 69,300

633 791 2198

27,086 6,896 16,826

NICE TA295 [51] NICE TA250 [52] NICE TA212 [53]

70,450 71,000 74,500

284 105 520

59,752 13,565 11,520

NICE TA219 [54,55] NICE TA202 [56] NICE TA309 [57]

Bosutinib Lenalidomide*

Imatinib Dasatinib Cetuximab Dasatinib Bortezomib Lapatinib Bortezomib Axitinib* Bosutinib

Enzalutamide Aflibercept

Cetuximab

Sorafenib* Bosutinib Dasatinib Crizotinib Imatinib Bosutinib Bevacizumab Bosutinib

Lenalidomide Everolimus Eribulin Bevacizumab

Everolimus* Ofatumumab Pemetrexed

continued on next page

571

VALUE IN HEALTH 19 (2016) 567–576

Table 2 – continued Treatment

Pemetrexed

Bosutinib

Cabazitaxel Dasatinib

Temsirolimus* Ruxolitinib

Bevacizumab Bevacizumab

Bevacizumab

Cetuximab Trastuzumab emtansine (Kadcyla) Cetuximab

Indication

Cost per QALY (£)

Indications approved (2013-2014)

Treatment cost (£)

Evidence

Second-line treatment of advanced nonsquamous non–small cell lung cancer Blast crisis chronic myeloid leukemia where there is intolerance of treatments Castrate-resistant metastatic prostate cancer Philadelphia chromosome positive (Phþ) acute lymphoblastic leukemia and lymphoid blast crisis chronic myeloid leukemia Advanced renal cell carcinoma Symptomatic splenomegaly in primary myelofibrosis, after polycythemia vera myelofibrosis or after essential thrombocythemia myelofibrosis Advanced breast cancer First-line advanced epithelial ovarian, fallopian tube, or primary peritoneal cancer Second-line advanced epithelial ovarian, fallopian tube, or primary peritoneal cancer First-line treatment of advanced head and neck cancer HER2-positive locally advanced/ unresectable or metastatic (stage IV) breast cancer Second-/third-treatment of metastatic colorectal cancer with combination chemotherapy

74,500

520

11,520

NICE TA190 [58]

85,000

0

44,799

NICE TA300 [49]

87,518

465

22,936

NICE TA255 [59]

96,000

28

30,477

NICE TA251 [60]

97,480 111,500†

29 427

28,867 43,200

NICE TA178 [61,62] NICE TA289 [63]

139,806 144,500

190 488

38,924 36,078

NICE TA263 [64] NICE TA284 [65]

149,000

403

44,428

NICE TA285 [66]

166,307

205

13,241

NICE TA172 [67]

166,400

129

90,831

NICE TAG350 [68]

223,627

52

11,739

NICE TA118 [45]

ALK, anaplastic lymphoma kinase; CDF, Cancer Drugs Fund; ERG, evidence review group; FAD, final appraisal determination; HTA, health technology assessment; ICER, incremental cost-effectiveness ratio; NICE, National Institute for Health and Care Excellence; QALY, qualityadjusted life-year; TA, technology appraisal. * Data contained within NICE HTAs incomplete. Treatment cost data sourced from the literature, converted to pounds sterling, and inflated to represent their net present value as appropriate. † NICE Guidance unclear as to the most plausible estimate of ICER, suggesting that it lies somewhere between the manufacturers’ base case of £74,500 and the ERG case of £149,000. £111,500 represents the midpoint between the two.

QALY, respectively. Under these scenarios, satisfying demand for treatment indications with ICERs beyond these points would necessitate overspending of £17.6 million, £73.7 million, and £129.7 million, respectively, alternatively introducing horizontal inequity where those with equal clinical need cannot receive equal access to treatment funding.

Comparison of CDF and NHS QALY Generation In the base-case scenario, whereby clinical demand for all evaluated treatment indications is met, these treatments would generate approximately 4677 QALYs, at a cost of £269.7 million. If instead this funding were used for day-to-day NHS activity, applying £20,000 per QALY as an estimate of NHS marginal productivity, the estimated number of QALYs gained would equal 13,485, resulting in the displacement of approximately 8,808 QALYs over the study period, with a corresponding net present value of £176.2 million. Table 3 highlights total QALY generation by treatment indication, with 8.3 and 4.9 potential NHS QALYs forgone for every 1 CDF QALY generated through the use of

trastuzumab emtansine and cetuximab (all indication average), respectively.

QALY Generation and “Average Cost per QALY” by Therapeutic Area Figure 2 highlights the CDF’s ability to generate QALYs, and an estimate of “average cost per QALY gained,” in each of the 14 therapeutic areas covered by the CDF during the study period. Those treated for leukemia, colorectal, and breast cancer accrued a total of 1259, 831, and 685 QALYs, respectively, whereas those with melanoma (43 QALYs), head and neck (38.2 QALYs), and gastrointestinal (26.7 QALYs) cancers generated considerably less QALYs. Furthermore, despite total annual costs for the treatment of ovarian cancers and leukemia varying by just 10% (£35.5 million vs. £31.6 million), using CDF funds for the latter generated some 1,017 additional QALYs per year, resulting in average ICERs of £146,738 and £25,094 per QALY, respectively, and highlighting the significant differences in marginal benefits and marginal costs between therapeutic areas.

572

VALUE IN HEALTH 19 (2016) 567–576

Fig. 1 – Utilization of CDF treatment indications by ICER. CDF, Cancer Drugs Fund; ICER, incremental cost-effectiveness ratio

Discussion Implications of Study Findings The results of this study suggest that the existence of the CDF results in considerable opportunity costs within NHS England, which can be measured in terms of forgone population health. Making reference to the NICE cost-effectiveness threshold of £20,000 per QALY, we estimate a net loss of 8,808 QALYs per annum, with the CDF generating 4,677 QALYs compared with a potential 13,485 if the same funds were used as part of routine NHS commissioning. With respect to the evaluative requirements of CDF drugs, of the 74 treatment indications reimbursed through the CDF during the study period, just 58% (43) were subject to evaluation by NICE, with the remaining 42% of largely unknown clinical and cost-effectiveness and budget impact. These treatment indications accounted for more than a third (38.3%) of CDF activity in 2013-2014, corresponding to 7,386 of 19,282 treatment episodes. Our base-case analysis, without the formal ring fencing of funds for the provision of unevaluated drugs, demonstrated a CDF willingness-to-pay value of £223,627 per QALY gained, suggesting a willingness-to-pay value for access to cancer treatments some 11 times greater than that applied by NICE as part of routine NHS commissioning. Furthermore, our results indicate that 26%, 41%, and 33% of the CDF budget was dedicated to treatments with ICERs of less than £50,000, £50,000 to £90,000, and more than £90,000 per QALY, respectively. As such, with the average CDF treatment exhibiting an ICER of £75,086 per QALY, the efficiency of the CDF, in terms of ability to generate QALYs, appears considerably lower than that of the wider NHS, seemingly willing to accept a much lower level of clinical benefit for every pound spent, despite having a much lower budget to begin with. Our study findings also suggest that the CDF is currently funding interventions with ICERs considerably above our imputed thresholds, based on 10%, 25%, and 50% of the annual CDF budget being ring fenced to provide drugs of largely unknown cost-effectiveness. This is despite the fact that funding these interventions will necessitate that others with lower associated ICERs and greater gains in survival and quality of life per pound spent are subsequently not funded. This not only

represents poor resource allocation in terms of maximizing population health subject to a budget constraint but also raises concerns regarding horizontal equity, widely considered a fundamental founding principle of the NHS. In fact, by formalizing the dual function of the CDF and ring fencing 10%, 25%, or 50% of its budget for the reimbursement of unevaluated drugs, the league-table approach derived an absolute maximum ability-topay value of £149,000, £111,400, and £68,600 per QALY, respectively, while still preserving equal access for equal need among all approved interventions. As such, in the absence of any clear rationing framework, drugs with ICERs beyond any given threshold can currently be funded only for a subset of patients with equal need before the budget is exhausted, compounding the potential inequities created by the fund’s mode of operation, which requires individual applications rather than blanket funding for a given clinical indication or patient subpopulation.

Dealing with Unevaluated Treatment Indications The considerable reduction in ability to pay for cancer QALYs as increasingly more of the CDF budget is diverted to funding unevaluated treatment indications is a clear signal of the implicit trade-off between offering an extended willingness to pay for cancer drugs and providing a means of reducing waiting times for those accessing “novel” cancer drugs. Our findings suggest that the CDF budget, at its current level, is sufficiently large to service demand for all treatment indications subject to prior health technology assessment, signifying that only through the funding of unevaluated treatments of largely unknown budget impact does the CDF then risk exhausting its budget and the need to seek additional funding. This finding may be informative to considerations of how much of the fund ought to be allocated to reimbursing drugs of unknown cost-effectiveness, and how evaluated and unevaluated drugs may be traded off. As alluded to within this analysis, one potential solution is to decide in advance how much to dedicate to funding these unevaluated treatments, then only funding treatments within this group for which the entirety of patient demand can be met, that is, which treatments can be provided equitously. In doing so, and through the prospective collection of real-world evidence during treatment, it is possible that any treatments with ICERs below the chosen threshold for

573

VALUE IN HEALTH 19 (2016) 567–576

Table 3 – Overall QALY generation and comparison with NHS marginal productivity by treatment indication. Treatment Bevacizumab Bevacizumab Bevacizumab Ruxolitinib Everolimus Trastuzumab emtansine (Kadcyla) Enzalutamide Cabazitaxel Bevacizumab Cetuximab Dasatinib Axitinib Lenalidomide Bevacizumab Pemetrexed Pemetrexed Aflibercept Lapatinib Eribulin Sorafenib Crizotinib Bortezomib Ofatumumab Dasatinib Bosutinib Dasatinib Temsirolimus Lenalidomide Cetuximab Cetuximab Bortezomib Imatinib Cetuximab Dasatinib Bosutinib Bosutinib Imatinib Bosutinib Bosutinib Everolimus Bosutinib Bendamustine Dabrafenib Total

Indication

ICER (£)

Budget impact (£)

QALYs generated

vs. NHS marginal productivity

Advanced colorectal cancer Advanced epithelial ovarian cancer Advanced epithelial ovarian cancer Symptomatic splenomegaly in primary myelofibrosis Advanced breast cancer Advanced unresectable metastatic breast cancer Metastatic prostate cancer Metastatic prostate cancer Advanced breast cancer Advanced head and neck cancer Chronic-phase CML Advanced renal cell carcinoma Multiple myeloma Advanced colorectal cancer Advanced nonsquamous NSCLC Advanced nonsquamous NSCLC Metastatic colorectal cancer Advanced breast cancer Advanced breast cancer Advanced hepatocellular carcinoma Metastatic NSCLC Relapsed multiple myeloma Chronic lymphocytic leukemia Blast crisis CML Chronic-phase CML Advanced renal cell carcinoma Advanced renal cell carcinoma Myelodysplastic syndromes Metastatic colorectal cancer Metastatic colorectal cancer Relapsed multiple myeloma Gastrointestinal stromal tumor (significant-risk) Metastatic colorectal cancer Accelerated- phase CML Chronic-phase CML Refractory accelerated- phase CML Gastrointestinal stromal tumor (high-risk) Accelerated- phase CML Refractory blast crisis CML Metastatic renal cell carcinoma Blast crisis CML Chronic lymphocytic leukemia Metastatic melanoma

69,300 149,000 144,500 111,500

36,983,548 17,904,484 17,606,064 18,446,400

533.7 120.2 121.8 165.6

1315.5 775 758.5 756.7

68,000 166,400

17,145,438 11,717,199

252.1 70.4

605.2 515.5

48,000 87,518 139,806 166,307 28,000 40,967 63,000 59,993 74,500 74,500 51,000 37,336 68,600 52,600 55,157 35,750 71,000 54,093 43,000 96,000 97,480 25,300 223,627 33,300 38,000 56,350

16,576,464 10,665,240 7,395,560 6,288,990 19,179,193 10,450,224 7,614,600 6,679,128 5,990,400 5,990,400 6,858,848 8,595,587 5,454,736 4,980,955 3,474,588 2,650,000 1,424,325 1,234,534 1,388,769 853,356 837,143 2,983,365 610,428 1,322,168 882,000 488,235

345.3 121.9 52.9 38.2 685 255.1 120.9 111.3 80.4 80.4 134.5 230.2 79.5 94.7 63 74.1 20.1 22.8 32.3 8.9 8.6 117.9 2.7 39.7 23.2 8.7

483.5 411.4 316.9 276.2 274 267.4 259.8 222.7 219.1 219.1 208.4 199.6 193.2 154.3 110.7 58.4 51.1 38.9 37.1 33.8 33.3 31.3 27.8 26.4 20.9 15.7

52,054 34,500 20,972 53,789 27,600

495,138 542,280 4,053,312 252,474 495,561

9.5 15.7 193.3 4.7 18

15.3 11.4 9.4 7.9 6.8

60,000 59,191 70,450 85,000 9,400 11,000

89,598 0 0 0 2,585,801 470,376

1.5 0 0 0 275.1 42.8 4676.7

3 0 0 0 (þ) 145.8 (þ) 19.3 8806.1

CML, chronic myeloid leukemia; ICER, incremental cost-effectiveness ratio; NHS, National Health Service; NSCLC, non–small cell lung cancer; QALY, quality-adjusted life-year.

evaluated treatments may then be transferred to the “evaluated” group, thereby displacing the least cost-effective (marginal) interventions within this group, gradually improving the efficiency of the CDF. We believe this point to be of great importance, with our results demonstrating that the possible QALY gains achievable through disinvesting from evaluated but highly cost-ineffective drugs could be substantial. Considering treatments individually, it was demonstrated that for every 1 CDF QALY generated

through the use of Kadcyla, cetuximab (all indication average), and bevacizumab (all indication average), the NHS will forgo 8.3, 4.9, and 4.7 QALYs, respectively, clearly demonstrating the extent of the trade-off in terms of population health. Furthermore, if considering the impact of such extensive differences in costeffectiveness solely with respect to CDF efficiency, our findings suggested that the reimbursement of Kadcyla generates some 60% fewer QALYs per pound than the “average” CDF treatment indication, further highlighting the benefits that could be

574

VALUE IN HEALTH 19 (2016) 567–576

Fig. 2 – QALYs generated by form of cancer and average cost per QALY achieved. QALY, quality-adjusted life-year achieved through the adoption of a more formally driven approach to disinvestment within the CDF. These findings may be interpreted as a quantification of the implicit premium for cancer QALYs that goes beyond that of the willingness to pay for noncancer QALYs by NICE. Expressed another way, the considerable variability exhibited in the ICERs attached to CDF-funded treatments may simply reflect an increased focus on the perceived wider benefits of extending treatment options within cancer. These may include addressing any unmet needs or perceived lack of innovation, aspects of disease severity or end-of-life considerations, or simply a shift in emphasis to improving access to health care interventions rather than a focus on health maximization itself, as not all health technologies possess the ability to generate improvements in health at the same price.” This was observed with respect to Figure 2, whereby the average cost per QALY by the form of cancer being treated was shown to vary significantly, ranging from £10,990 and £25,094 in the cases of melanoma and leukemia, respectively, to £146,738 and £164,633 in the case of drugs for ovarian and head and neck cancers, respectively. As such, our results suggested that although spending on the treatment of ovarian cancers and leukemia was almost identical during the study period (£35.5 million vs. £31.6 million), the latter generated some 1017 additional QALYs per year, highlighting the extent of the variability in marginal costs and benefits exhibited between CDF treatments. We believe this to be an important consideration when deciding on the future provision of CDF treatment indications, particularly if there exists any unsatisfied demand for the most cost-effective treatments currently made available through the CDF.

Care Costs versus Drug Costs Because the CDF funds only cancer drugs and not the accompanying costs of complementary health care in order to provide these drugs, including chemotherapy, nursing time, and complementary services, it is unclear what effect this has on program budgeting and commissioning within the NHS. Under the current operating model of the CDF, it is unlikely that these activities can be easily anticipated by administering health care providers. Hence, it is possible that formalizing a clear resource allocation protocol, as proposed in this study, may provide positive externalities by facilitating better planning of resource use outside of the fund, including any care costs relating to CDF treatments that

are borne elsewhere in the NHS. It is unclear whether future economic evaluations undertaken by the fund should be restricted solely to cancer drug acquisition costs, or include NHS-wide costs including the costs of administering these treatments. Future research may wish to determine the relative contribution of drug acquisition costs to the costs of overall treatment episodes, to increase the internal validity of any cost-effectiveness ranking.

Strengths and Limitations Strengths of the study include the quality of data sources used and the novel nature of the results obtained. The use of historical CDF utilization data leaves little room for error in terms of the real-world demand for specific cancer drugs, whereas NICE HTA and ERG reports provide rigorously considered estimates of the ICERs attached to these drugs. Moreover, no previous study has provided a means of improving resource allocation within the CDF and thereby minimizing the opportunity costs that result from the ring fencing of these funds in the first place. An acknowledged limitation of this research is the existence of any patient access scheme pricing, risk sharing, or other discounts received by the CDF to the prices reported in health technology appraisals, on which no data were publicly available. Unlike NICE the CDF has no obligation to disclose the prices paid for drugs to the public, and as such, the ranking of these drugs by costeffectiveness, their budget impact, and the respective placements of these drugs in the league table may differ from those reported in this analysis. However, given that this information is available within the CDF, this limitation would not apply were the fund to adopt such a framework. Second, again because of paucity of data, the analyses used point estimates of parameters including ICERs and treatment costs with no explicit consideration of uncertainty around these parameter estimates using sensitivity analyses. As such, it is unclear whether the results obtained are robust to changes in model inputs, and whether recommendations for specific treatment indications can be put into action without seeking further information not available in the public domain.

Conclusions Rationing, or prioritization, is clearly inevitable in a health care system with a fixed budget. The combination of scarce resources

VALUE IN HEALTH 19 (2016) 567–576

and near infinite demand implies that unmet medical need will persist. We have shown that the existence of the CDF imposes considerable opportunity costs in terms of forgone QALYs, and appears to be at odds with the principle of horizontal equity. Yet whether justifiable, rational, or otherwise, cancer is and is likely to continue to be afforded special consideration ahead of other therapeutic areas. We have demonstrated that the opportunity costs of ring fencing these funds can be reduced through the adoption of a formal QALY maximization mechanism. We therefore recommend that moving forward, the CDF adopts such a mechanism to make best use of this ring-fenced budget, promote horizontal equity, and maximize the potential health benefits to patients in need of high-quality cancer treatments. R EF E R EN CE S

[1] Graham J, Cassidy J, Hughes D, et al. Ring-fencing a budget for cancer drugs: is it fair? J R Coll Physicians Edinb 2011;41:224–8. [2] New 50 million pound cancer fund already intellectually bankrupt. Lancet 2010;376:389. [3] Duerden M. Cancer drugs fund: a NICE mess. Prescriber 2010;21:8–9. [4] Maynard A, Bloor K. The economics of the NHS cancer drugs fund. Appl Health Econ Health Policy 2011;9:137–8. [5] Towse A. Should NICE’s threshold range for cost per QALY be raised? Yes. BMJ 2009;338:b181. [6] Speight J, Reaney M. Wouldn’t it be NICE to consider patients’ views when rationing health care? BMJ 2009;338:297. [7] Mason AR, Drummond MF. Public funding of new cancer drugs: is NICE getting nastier? Eur J Cancer 2009;45:1188–92. [8] Williams A. What Could be Nicer than NICE? Office of Health Economics 2004. [9] Claxton K, Martin S, Soares M. Methods for the estimation of the NICE cost effectiveness threshold (CHE Research Paper 81). 2013. Available from: 〈https://www.york.ac.uk/media/che/documents/papers/ researchpapers/CHERP81_methods_estimation_NICE_ costeffectiveness_threshold_(Nov2013).pdf〉. [Accessed April 11, 2015]. [10] 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. [11] McCabe C, Claxton K, Culyer AJ. The NICE cost-effectiveness threshold – what it is and what that means. Pharmacoeconomics 2008;26:733–44. [12] Dolan P, Tsuchiya A. The elicitation of distributional judgements in the context of economic evaluation. In: Jones A, ed. Companion to Health Economics. Cheltenham: Edward Elgar, 2006. [13] Nord E, Pinto JL, Richardson J, et al. Incorporating societal concerns for fairness in numerical valuations of health programmes. Health Econ 1999;8:25–39. [14] Wailoo A, Tsuchiya A, McCabe C. Weighting must wait: incorporating equity concerns into cost-effectiveness analysis may take longer than expected. Pharmacoeconomics 2009;27:983–9. [15] Nord E. The relevance of health state after treatment in prioritizing between different patients. J Med Ethics 1993;19:37–42. [16] Nord E. The trade-off between severity of illness and treatment effect in cost–value analysis of healthcare. Health Policy 1993;24:227–38. [17] Gyrd-Hansen D. Investigating the social value of health changes. J Health Econ 2004;23:1101–16. [18] Nord E. Concerns for the worse off: fair innings versus severity. Soc Sci Med 2005;60:257–63. [19] Round J. Is a QALY still a QALY at the end of life? J Health Econ 2012;31:521–7. [20] Pinto-Prades JL, Sánchez-Martínez FI, Corbacho B, et al. Valuing QALYs at the end of life. Soc Sci Med 2014;113:5–14. [21] Shah KK, Tsuchiya A, Wailoo AJ. Valuing health at the end of life: an empirical study of public preferences. Eur J Health Econ 2014;15:389–99. [22] Rooshenas L, Owen-Smith A, Hollingworth W, et al. “I won’t call it rationing…”: an ethnographic study of healthcare disinvestment in theory and practice. Soc Sci Med 2015;128:273–81. [23] Karnon J, Carlton J, Czoski-Murray C, et al. Informing disinvestment through cost-effectiveness modelling: is lack of data a surmountable barrier? Appl Health Econ Health Policy 2009;7:1–9. [24] Appleby J, Devlin N, Parkin D, et al. Searching for cost effectiveness thresholds in the NHS. Health Policy 2009;91:239–45. [25] European Medicines Agency. Adaptive pathways. Available from: 〈http://www.ema.europa.eu/ema/index.jsp?curl=pages/regulation/ general/general_content_000601.jsp〉. [Accessed May 13, 2015]. [26] NHS England. The Cancer Drugs Fund. Available from: 〈http://www. england.nhs.uk/ourwork/pe/cdf/〉. [Accessed March 26, 2015].

575

[27] Gafni A, Birch S. Guidelines for the adoption of new technologies—a prescription for uncontrolled growth in expenditures and how to avoid the problem. Can Med Assoc J 1993;148:913–7. [28] Gafni A, Birch S. Incremental cost-effectiveness ratios (ICERs): the silence of the lambda. Soc Sci Med 2006;62:2091–100. [29] Drummond M, Torrance G, Mason J. Cost-effectiveness league tables: more harm than good? Soc Sci Med 1993;37:33–40. [30] Weinstein M, Zeckhauser R. Critical ratios and efficient allocation. J Pub Econ 1973;2:147–57. [31] Bendamustine for the first-line treatment of chronic lymphocytic leukaemia. NICE technology appraisal guidance [TA216]. 2011. [32] Dabrafenib for treating unresectable or metastatic BRAF V600 mutation positive melanoma. NICE technology appraisal guidance [TA321] 2014. [33] Bosutinib for previously treated chronic myeloid leukaemia. NICE technology appraisal guidance [TA299]. 2013. [34] Lenalidomide for treating myelodysplastic syndromes associated with an isolated deletion 5q cytogenetic abnormality. NICE technology appraisal guidance [TA322]. 2014. [35] Goss TF, Szende A, Schaefer C, et al. Cost effectiveness of lenalidomide in the treatment of transfusion-dependent myelodysplastic syndromes in the United States. Cancer Control 2006;13(Suppl):17–25. [36] Imatinib for the adjuvant treatment of gastrointestinal stromal tumours. NICE technology appraisal guidance [TA196]. 2010. [37] Dasatinib, high-dose imatinib and nilotinib for the treatment of imatinib-resistant chronic myeloid leukaemia (CML) (part review of NICE technology appraisal guidance 70), and dasatinib and nilotinib for people with CML for whom treatment with imatinib has failed because of intolerance. NICE technology appraisal guidance [TA241]. 2012. [38] Cetuximab for the first-line treatment of metastatic colorectal cancer. NICE technology appraisal guidance [TA176]. 2009. [39] Bortezomib monotherapy for relapsed multiple myeloma. NICE technology appraisal guidance [TA129]. 2007. [40] Lapatinib for the treatment of women with previously treated advanced or metastatic breast cancer. NICE technology appraisal guidance [TA387]. 2009. [41] Axitinib for treating advanced renal cell carcinoma after failure of prior systemic treatment. NICE technology appraisal guidance [TA333]. 2015. [42] International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Economic evaluation of axitinib for second line treatment in adult patients with advanced renal cell carcinoma – the Portuguese case. 2014. Available from: 〈http://www.ispor.org/research_pdfs/48/ pdffiles/PCN146.pdf〉. [Accessed August 23, 2015]. [43] National Institute for Health and Care Excellence. NICE issues final draft guidance recommending enzalutamide for prostate cancer. NICE press release. 2014. Available from: 〈https://www.nice.org.uk/news/ press-and-media/ nice-issues-final-draft-guidance-recommending-enzalutamide-forprostate-cancer〉. [Accessed May 24, 2015]. [44] National Institute for Health and Care Excellence. NICE issues final guidance aflibercept for colorectal cancer. NICE press release. 2014. Available from: 〈https://www.nice.org.uk/news/press-and-media/ nice-issues-final-guidance-aflibercept-for-colorectal-cancer〉. [Accessed May 15, 2015]. [45] Bevacizumab and cetuximab for the treatment of metastatic colorectal cancer. NICE technology appraisal guidance [TA118]. 2007. [46] Sorafenib for the treatment of advanced hepatocellular carcinoma. NICE technology appraisal guidance [TA189]. 2010. [47] Zhang P, Yang Y, Wen F, et al. Cost-effectiveness of sorafenib as a firstline treatment for advanced hepatocellular carcinoma. Eur J Gastroenterol Hepatol 2015;27:853–9. [48] Crizotinib for previously treated non-small-cell lung cancer associated with an anaplastic lymphoma kinase fusion gene. NICE technology appraisal guidance [TA296]. 2011. [49] Peginterferon alfa and ribavirin for treating chronic hepatitis C in children and young people. NICE technology appraisal guidance [TA300]. 2013. [50] Lenalidomide for the treatment of multiple myeloma in people who have received at least one prior therapy. NICE technology appraisal guidance [TA171]. 2009. [51] Everolimus in combination with exemestane for treating advanced HER2-negative hormone-receptor-positive breast cancer after endocrine therapy. NICE technology appraisal guidance [TA295]. 2013. [52] Eribulin for the treatment of locally advanced or metastatic breast cancer. NICE technology appraisal guidance [TA250]. 2012. [53] Bevacizumab in combination with oxaliplatin and either fluorouracil plus folinic acid or capecitabine for the treatment of metastatic colorectal cancer. NICE technology appraisal guidance [TA212]. 2010. [54] Everolimus for the second-line treatment of advanced renal cell carcinoma. NICE technology appraisal guidance [TA219]. 2011. [55] Casciano R, Chulikavit M, Di Lorenzo G, et al. Economic evaluation of everolimus versus sorafenib for the treatment of metastatic renal cell carcinoma after failure of first-line sunitinib. Value Health 2011;14:846–51.

576

VALUE IN HEALTH 19 (2016) 567–576

[56] Ofatumumab for the treatment of chronic lymphocytic leukaemia refractory to fludarabine and alemtuzumab. NICE technology appraisal guidance [TA202] 2010. [57] Pemetrexed maintenance treatment following induction therapy with pemetrexed and cisplatin for non-squamous non-small-cell lung cancer. NICE technology appraisal guidance [TA309]. 2014. [58] Pemetrexed for the maintenance treatment of non-small-cell lung cancer. NICE technology appraisal guidance [TA190]. 2010. [59] Cabazitaxel for hormone-refractory metastatic prostate cancer previously treated with a docetaxel-containing regimen. NICE technology appraisal guidance [TA255]. 2012. [60] Dasatinib, nilotinib and standard-dose imatinib for the first-line treatment of chronic myeloid leukaemia (part review of technology appraisal guidance 70). NICE technology appraisal guidance [TA251]. 2012. [61] Bevacizumab (first-line), sorafenib (first- and second-line), sunitinib (second-line) and temsirolimus (first-line) for the treatment of advanced and/or metastatic renal cell carcinoma. NICE technology appraisal guidance [TA178]. 2009. [62] Hoyle M, Green C, Thompson-Coon J, et al. Cost-effectiveness of temsirolimus for first line treatment of advanced renal cell carcinoma. Value Health 2010;13:61–8.

[63] Ruxolitinib for disease-related splenomegaly or symptoms in adults with myelofibrosis. NICE technology appraisal guidance [TA289]. 2013. [64] Bevacizumab in combination with capecitabine for the first-line treatment of metastatic breast cancer. NICE technology appraisal guidance [TA263]. 2012. [65] Bevacizumab in combination with paclitaxel and carboplatin for firstline treatment of advanced ovarian cancer. NICE technology appraisal guidance [TA284]. 2013. [66] Bevacizumab in combination with gemcitabine and carboplatin for treating the first recurrence of platinum-sensitive advanced ovarian cancer. NICE technology appraisal guidance [TA285]. 2013. [67] Cetuximab for the treatment of recurrent and/or metastatic squamous cell cancer of the head and neck. NICE technology appraisal guidance [TA172] 2009. [68] Trastuzumab emtansine for treating HER2-positive, unresectable locally advanced or metastatic breast cancer after treatment with trastuzumab and a taxane. NICE Final Appraisal Determination. 2014. Available from: 〈https://www.nice.org.uk/guidance/gid-tag350/ resources/breast-cancer-her2-positive-unresectable-trastuzumabemtansine-after-trastuzumab-taxane-final-appraisal-determinationdocument2〉. [Accessed June 28, 2015].