Irrelevance of explicit cost–effectiveness thresholds when coverage decisions can be reversed

Cost–effectiveness analysis (CEA) compares a new treatment to a standard, already covered treatment by estimating an incremental cost–effectiveness ratio (ICER) – the estimated incremental costs divided by the estimated incremental benefits. When the ICER is less than a socially acceptable threshold then the new treatment is considered to be cost effective and, therefore, suitable for insurance coverage.

There are no clear norms as to what a socially acceptable threshold should be [1]. The National Health Service in the UK has used a threshold range of GBP£20,000–30,000 over several years [2], albeit with little empirical evidence. In nationalized healthcare systems, such as that in the UK or Australia, where a fixed healthcare budget (in real currency) is allocated every year, a normative threshold is the ICER of the ‘marginal technology’, that is, the technology with the highest ICER that was covered by health insurance just before the budget cap was reached. If any new technology is additionally covered, then the budget impact of this coverage should lead to forgo the benefits of this marginal technology in order to maintain a fixed budget [2]. In fact, this conceptualization of the threshold as the ‘shadow price of the budget constraint’ is consistent with the theoretical foundations of CEA [3–5]. However, this also implies that the threshold will change over time, even when the budget is held fixed.

Recent rigorous empirical analyses estimate the UK’s central policy threshold for cost–effectiveness to be GBP£18,317 per quality-adjusted life year (QALY) [101], although these are deemed to be ‘positive’ analyses of the effects of changes of overall budget spending on health outcomes and, hence, the estimate will most likely deviate from the normative threshold. An added concern about the threshold approach lies in its explicit public announcement, which can set up perverse incentives for manufacturers of future technologies who would tend to price products such that their ICERs are just below this threshold, even when the marginal costs of producing these technologies are much lower.

In this editorial, I argue that a dynamic decision-making framework using CEA league tables that does not attempt to define a threshold may be more suitable to achieve dynamic efficiency in this setting, whereby efficient allocation of resources are achieved in the long run.

A dynamic decision-making framework

Consider that a healthcare system is assigned a fixed amount of budget every year; at present, a set of medical treatments are already covered by this healthcare system (grandfathered in), which implies that a portion of the annual fixed budget has already been committed over the next few years. The goal is to allocate the fixed amount of a residual budget every year so as to maximize health.

Suppose four new technologies are considered for coverage in the first year; Table 1 shows the ICER for each of these technologies and, if covered, the impact they will sequentially have on the residual budget. Such presentation of ICERs in league tables is common practice in CEA [6]. The highest ICER among these new technologies is US$75,000 per QALY. A decision-making framework that uses comparison with a threshold, say US$50,000/QALY, might deem this not to be cost effective and, hence, not extend coverage in the first year, even though the residual budget is able to accommodate it. However, if this threshold is publicly known a priori, then there is no reason to believe that the other three technologies that have ICERs of US$10,000, 25,000 and 40,000 per QALY respectively, would not raise their prices such that their ICERs are just below $50,000 per QALY [7]. As long as they can be accommodated in the residual budget, they will all be covered, in essence transferring a substantial portion of the consumer surplus to the manufacturers. If, together, their budget impacts exceed the residual budget, the policy maker will have difficulty deciding which technology not to cover since they all have the similar ICERs. Therefore, even though an advertised threshold may help with proper evaluation of a marginal technology, it is also likely to induce inflated ICERs from non-marginal technologies, thereby sacrificing dynamic efficiency.

Now let’s consider a dynamic decision-making framework where the coverage of the US$75,000 per QALY technology in the first year is not relevant as long as it can be revoked in later years, which would set forth the incentives to achieve efficiency in the long run. This is shown in Table 1, where it is assumed that all four technologies were covered in the first year and the residual budget in the first year was able to accommodate all. Now assume that three more new technologies applied for coverage in year 2. These, along with the four technologies from year 1, sorted by their ICERs, are reported in Table 1. The budget impacts in year 2 show the US$75,000 per QALY technology can no longer be fully covered. Either a high copayment must be introduced to accommodate it within the budget or coverage must be altogether stopped. Alternatively, the manufacturer may decide to adjust the price in order to retain coverage (this may be unlikely, especially if other countries use this price for referencing).

Such a framework possesses four unique features that help in achieving dynamic efficiency in a transparent manner:

• • Reversal of coverage decisions weeds out high-ICER technologies from the budget to make place for low-ICER technologies. This framework makes it explicit what new technologies will be accommodated in order to reverse the coverage on the US$75,000 per QALY technology. Multiple decision-making criteria other than ICERs can also be incorporated at ease in this framework with public transparency [8];

• • Manufacturers are unable to game against a fixed threshold, and instead they face pressures to compete with both the currently covered technologies (with known ICERs) and the future technologies whose ICERs are unknown. This can be viewed as a sealed-bid auction process, where the auctioneer is the healthcare authority trying to allocate a budget and the bidders are the manufacturers bidding to claim a residual fraction of that budget [9]. New technologies must come at ICERs that are competitive with covered technologies, cautious about their budget impact and also mindful of how easily future technologies may be able to supplant coverage, even if they are approved today, thereby giving them the incentives to reveal their true reservation prices;

• • Manufacturers are incentivized to invest in outcomes research and also to promote individualization of technology use in the population, thereby improving overall population health [10,11]. By doing so, they can lower the ICER estimate and the budget impact of their product for that indication as compared with those at the introduction of the technology and, in this process, help secure the place of the technology in the healthcare budget;

• • Over time, the saturation of budgets with increasingly efficient (low ICERs) technologies would reach a point where many low-ICER (seemingly efficient) technologies cannot be funded. This would naturally lead to the proposition of increasing the health budget, which can be strongly supported by evidence of what seemingly efficient technologies are being forgone.

In summary, fixating over a CEA threshold may not lead to dynamically efficient allocation of budgets. Instead, a decision-making framework that can set up healthcare technologies to compete with each other for a place on the healthcare budget could generate better outcomes over time. It is not necessary to have full ICER information on all technologies covered today to implement this process. Grandfathered technologies will be weeded out over time, reaching a point where full ICER information of all covered technologies becomes reality.

Table 1. Technologies that arrive in year 2.

Year 1				Year 2
Technology	ICER US$/QALY	Budget impact (%)	Cumulative budget impact (%)	Technology	ICER US$/QALY	Budget impact (%)	Cumulative budget impact (%)
1	10,000	5	5	1	10,000	5	5
2	25,000	10	15	2	25,000	10	15
3	40,000	25	40	5^†	35,000	15	30
4	75,000	50	90	3	40,000	25	55
				6^†	45,000	15	70
				7^†	50,000	20	90
				4	75,000	50	Potential reversal on coverage

^†Technologies that arrived in year 2.

ICER: Incremental cost–effectiveness ratio; QALY: Quality-adjusted life year.

Acknowledgements

The author is grateful for discussions with Sean Sullivan and Karl Claxton on the material presented.

Disclaimer

The opinions expressed in this article are solely the author’s and do not reflect those of his professional affiliations.

Financial and competing interests disclosure

The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.