Our analysis does not indicate what society should be prepared to pay for an orphan drug. Rather, in this paper we propose a method for establishing a reasonable price for an orphan drug in situations where a value-based price is deemed inappropriate. The method rests on the proposition that, although societal decision-makers may be willing to pay above their standard value-based price to make treatments for some orphan diseases available, they would still need a benchmark for use in price negotiations. One possible guide for setting prices would be to ensure that the manufacturers of orphan drugs do not make higher profits than manufacturers of drugs for non-orphan conditions. In order to establish a price based on this proposition, we have examined, for illustrative purposes, how the standard incremental cost-per-QALY cost-effectiveness threshold (CET) in the UK would need to be adjusted to reflect typical differences between orphan and non-orphan products in both (i) the costs of R&D and (ii) in the size of the expected treatment population. Whilst we recognize that there may be concerns about how well the QALY captures the value of the health gain for some rare conditions, we regard the (cost-per-QALY) CET as an effective tool for illustrating our reasonable price approach.
It is important to stress that our analysis does not indicate what society should be prepared to pay for an orphan drug, since this involves important societal judgments about whether some population health in total should be forgone in order to provide funding for treatments for rare conditions and, if so, how much. Rather, our approach could be viewed as one way of determining the maximum allowable price society should be willing to pay, based on allowing a reasonable rate of return. Thus, it might be used as a starting point for negotiations. Awarding a lower price would send a signal to manufacturers about the level of priority being assigned to the treatment of orphan conditions. We return to this point later.
While we believe our proposed method has some merit, it is evident that further research is required in order to improve the estimates produced. First, in estimating the costs of R&D, we only sampled novel drugs approved by the FDA in 2015. A larger sample, covering more years, might have generated different estimates of the relative research costs for orphans and non-orphans. Second, post-approval R&D costs (phase IV) should be included as they may differ as between ultra-orphan, orphan and non-orphan and would imply differences in the relative R&D cost estimates. However, our estimates, showing a lower R&D cost for orphan products, are consistent with those in earlier studies [9]. In addition, we show that the difference in the research cost between orphans and non-orphans is smaller for oncology products. This is consistent with our expectations, given that the research requirements for all oncology drugs are similar, with both orphan and non-orphan products being eligible for the FDA’s ‘fast track’ programmes for innovative drugs i.e. ‘Accelerated Approval’ and ‘Breakthrough Therapy’ [29]. These programmes often grant market approval based on less mature clinical data.
Second, we made assumptions about the differences in drug lifecycle costs other than R&D. Our simplest assumption was that only R&D costs, representing 34% of the total drug lifecycle costs, would be lower, with all other costs being the same for both orphan and non-orphan drugs. An alternative approach is to assume that the costs of manufacturing, marketing or distribution differed in the same proportion as R&D costs as orphans and non-orphans. This is the most conservative approach and produces slightly lower estimates of the adjusted CETs. Marketing and distributing an orphan drug to a small group of identified individuals, may be lower per patient than for a non-orphan drug. The truth is probably somewhere between these two extremes. However, since the assumption concerning the level of reduction (if any) in non-R&D costs makes a substantial difference to the adjusted CETs, these estimates should be verified by further research.
Third, we haven’t included the tax credit of the 50% of the phase III clinical testing costs for orphan drugs included in the US Orphan Drug Act of 1983Footnote 23 and which remained unchanged until 2018, when reduced to 27.5%.Footnote 24 Although in principle it could have an impact on our estimates of R&D cost for an orphan drug and consequently could reduce the adjusted CET for an orphan, the numerical impact is minor, since we are apportioning global R&D costs to the UK by the global market share of pharmaceutical sales. Furthermore, the tax credit will not be applicable for drug development based in Europe and Asia. Other push incentives given by governments, governmental agencies, product development partnerships and philanthropic donors are neither included in the estimation of the R&D cost of orphan medicines nor of non-orphans. These other push incentives are not orphan drug specific and therefore we assume they have a neutral effect on the relative R&D costs of orphans and non-orphans.
Fourth, the estimates of the costs of R&D are highly sensitive to the success rates in the development of new products. In our sample of oncology drugs, the numbers of patients in the different phases of clinical research were similar for orphans and non-orphans. However, applying different estimates of the success rate (2.4% for non-orphans and 28.5% for orphans) led to differences in overall R&D cost.Footnote 25 While there might be reasons to expect a higher success rate for orphans, given the more precise targeting of therapy, this issue requires further investigation.
Fifth, we used estimates of target patient populations given in technology appraisals performed by the SMC and NICE in the UK. These may not reflect typical patient populations for the drugs studied for two reasons. First, in the case of orphan drugs, it is possible that some would also be indicated for larger, non-orphan populations, negating the argument for an adjusted threshold to compensate for a smaller treatment population [9]. In the case of the non-orphan drugs studied, the appraisal conducted by NICE or the SMC may have focused on a sub-set of the total population for the licensed indication for the drug concerned. For example, for oncology products in particular, it is common for a technology appraisal to focus on a given stage of disease, even if the product is licensed for other stages.
This issue was harder to investigate, but we did note that a small number (10) of the non-orphan oncology drugs in our sample had estimated patient populations in the appraisals that were lower than those for many orphan drugs in the sample.Footnote 26 Therefore, if patient population sizes were to be used as part of an argument to allow an adjusted threshold for orphan drugs, such a policy would require increased accuracy in the estimation of target patient populations and an understanding that the eligibility of an orphan drug for an adjusted threshold could be lost if the total patient population size were to increase beyond that typically designated ‘orphan’. Additionally, the policy should be designed to prevent potential perverse incentives to strategically narrow/stratify the scope of patients licenced in order to obtain higher prices. One approach to tackle any perverse incentives to expand patient populations by either approving medicines for multiple orphan and non-orphan indications and/or strategically narrowing/stratifying patient populations, would be to take the cumulative total patient population across all indications for the CET adjustment.Footnote 27
Sixth, we are aware that all drug prices, orphan and non-orphan, are subject to confidential discounts in many settings. Since the method described here establishes a reasonable price through an adjustment of the CET, the price that the policy maker of each country actually pays would be the one used to consider whether the drug was cost-effective, and not the published list price. However, we cannot observe confidential discounts. The existence of confidential discounts would only be problematic for our analysis if they were different as between the two types of drug. We have no reason to expect them to differ, so we make the assumption that they do not.
Seventh, we have used patient population sizes as a predictor of the likely revenue generated from the sales of the various products in the sample, at the price implied by the adjusted threshold in each case. However, it could be the case that the market exclusivity granted to orphan products means that revenue generation could be maintained for a given patient population for a longer period than that non-orphan drugs, since the latter would be more vulnerable to the entry of new, competitor products, including generics. Therefore, in the base case we assumed that a non-orphan product would only be used in 50% of the total patient population, due to the emergence of competitors.
The use of so many assumptions means that there is considerable uncertainty around the estimates of the adjusted CETs our method has produced. This uncertainty is explored in the sensitivity analysis. However, if our method were applied in practice, it may be feasible to obtain more accurate data for many of the parameters. We propose that industry wide data on R&D costs be used, together with bands of population size. Adjusting CETs by bands of patient population sizes avoids a ‘cost-plus’ pricing policy based on a company’s actual costs, which would give no incentives for efficiencies in research and development, be hard to verify, and need, somehow, to take account of failures.
Also, although we have suggested potential adjustments to the threshold ICER for orphan drugs, we have retained the standard cost per QALY rubric. Despite orphan status, it seems reasonable to expect products to show evidence of QALY gains, thereby maintaining the principle of assessing cost-effectiveness. However, some would argue that it is unreasonable to require orphan drugs to meet the same evidential standards as non-orphans, given their smaller patient populations and the likely lower level of understanding of the disease process [30]. This may be the case for many ultra-orphan products, but our research shows that oncology orphans often have trial population sizes equivalent to non-orphan cancer drugs. Furthermore, the trend for the FDA and EMA to offer various ‘accelerated approval’ programmes means that many oncology products, both orphan and non-orphan, will be licensed based on less extensive clinical data.
Finally, adopting this approach for establishing a reasonable price for an orphan drug does not tackle the broader issue of determining appropriate research priorities for the development of orphan drugs, or drugs in general. A value-based pricing policy not only ensures that the therapies adopted by the health care system are good value for money; it also encourages a shift in manufacturer research strategy towards delivering products that produce high added value in population health terms. Determining a reasonable price for the orphan drugs based on allowing a reasonable rate of return does not, of itself, appropriately drive the direction of future research. One could argue that allowing a higher CET for specialized services, as in the English NHS, but to limit this to £300,000 per QALY, is one way of sending such a signal to manufacturers. Namely, society may be willing to offer a reward to manufacturers for developing drugs for rare conditions, but this reward may not be as high as that to manufacturers developing drugs that have a major impact on population health.
This is a societal value judgment that we are not qualified to make. However, according to our data, the currently proposed CET for ultra-orphans in the UK—at its maximum of £300 k per QALY—would guarantee the average industry rate of return for most orphan drugs with patient populations as low as 3.2 per 50,000, but not for the target populations for drugs designated ‘ultra-orphan’ (i.e. 1 per 50,000 individuals). Therefore, there remains a case for having more discussion of priorities for research into rare diseases, given the large number of very rare diseases for which drugs could potentially be developed.