Efficiency and opportunity cost

Drummond believes NICE has done ‘more good than harm’ in terms of improving the efficiency of National Health Service (NHS) resource allocations. However, he provides no evidence to support his belief. He accepts that attempts to promote improvements in efficiency must be based on the explicit use of opportunity cost, although NICE decision-making has been free of explicit and systematic considerations of opportunity costs. Indeed Drummond identifies alternative options for incorporating opportunity costs into its future activities. Given the absence of opportunity cost considerations by NICE, how can Drummond’s belief of ‘more good than harm’ be justified? Evidence-based decision-making requires more than simply the beliefs of researchers (or clinicians or decision-makers).

GB, by ignoring NICE objectives, consider decision-making approaches as if they are ends in themselves. But, in this way, their arguments are essentially subjective with no approach, or proponent, being right (or wrong). They defend NICE’s performance through UK–US comparisons of levels and rates of increases in health care costs. This argument is irrelevant for at least two reasons. First, comparing oneself with a poorer performer is a prescription for mediocrity. It is like the England soccer team feeling good because they score more goals than the USA soccer team. Sadly for English soccer fans (and by implication the NHS) one is unlikely to win many trophies (or achieve many objectives) if your basis for measuring performance is comparison with only one of the poorest performing ‘competitors’. Second, the objective of NICE is not shared by US health care policy. So if the two systems aim for different things, what is the value in comparing one (UK) with the other (USA) on the basis of UK objectives? The analogy is judging the performance of the England soccer team, not in relation to the USA soccer team, but instead by comparison with the number of goals scored by the USA baseball team (for those not familiar with baseball there are no goals, the objective of baseball is to score more runs than the opposition).

The ‘need’ to spend more

BG claim that we focus attention on budget neutrality. This is wrong. Whether NHS expenditures should increase is a political decision and, as acknowledged by Drummond, health care spending in a publicly funded system such as the NHS is determined exogenously. Similarly, Drummond inaccurately attributes NICE as being described as the pharmaceutical industry’s ‘golden goose’ to us. Careful reading of our original paper shows that this is a quote from UK researchers who share some of our concerns (Cookson et al., Reference Cookson, McDaid and Maynard2001).

NICE’s stated objective is to maximize health gains from whatever resources are allocated to the NHS. However, NICE decision-making exists outside of the resources made available to the NHS. It excludes consideration of the opportunity cost of decisions and has been associated with substantial increases in NHS expenditures (£836 million in 2005 according to GB). If this represented a planned increase in resources and if NICE decisions had been based on maximizing health gains from this increased resource allocation, we would have no problems with NICE. Instead, NICE decisions have led to an unplanned increase in resources. In other words, no consideration has been given to the most efficient use of these additional NHS resources. On the contrary, NICE’s decisions based on invalid assumptions and arbitrary cost-effectiveness thresholds (Gafni and Birch Reference Gafni and Birch2006) define the procedures incorporated in NICE recommendations as the most efficient uses of these resources. In this way, the NICE ‘sponge’ simply ‘absorbs’ additional funds (and on a continual basis). Barrett et al. (Reference Barrett, Roques, Small and Smith2006) highlight the nightmare for NHS managers arising from the absence of the resource context and hence opportunity cost considerations in the development of NICE’s decisions. They conclude that by recommending more expensive treatments (Herceptin to treat breast cancer in their example) without suggesting what treatments to cut, NICE ‘creates more problems than solutions’ (p. 1119).

GB’s claim that the pressure for spending is the result of ‘the populace’s ever-growing desires for health and longevity’ (i.e., needs for care) is naïve both in its failure to consider opportunity costs (e.g., reductions in other publicly funded services) or the conceptual foundations of health care expenditures (the health care income-expenditure identity, Evans (Reference Evans1984)). For example, Shipman et al. (Reference Shipman, Lurie and Goodman2004) note that the number of general pediatricians in USA will expand by nearly 64% by the year 2020 compared with an increase of only 9% in the child population. They argue that for pediatricians to maintain workloads they may need to:

None of this relates to a populace’s ‘ever growing desires for health’. But let’s not be surprised if in coming years we note an increased demand for pediatric care on behalf of the populace.

The elusive threshold

Drummond and GB explain the difficulties of identifying the ICER threshold (or different thresholds) and point to on-going research aimed at measuring society’s valuation of (or willingness to pay for) a QALY. GB chastise us for not relating to this research. Interestingly, this contrasts with the position of the US panel on Cost Effectiveness Analysis who noted that no absolute standard exists for deciding whether a specific ICER value represents an efficient use of resources (Gold et al. Reference Gold, Siegel, Russell and Weinstein1996).

Space does not permit us to show again how the notion of the ICER threshold is incompatible with an objective of maximizing health gain from available resources (see for example Birch and Gafni Reference Birch and Gafni2006a, Reference Birch and Gafni2006b and Gafni and Birch Reference Gafni and Birch2006). However according to the outline of this on-going research, the aim is to identify society’s willingness to pay for a QALY for use as an appropriate threshold for NICE decision-makers (http://pcpoh.bham.ac.uk/publichealth/nccrm/PDFs%20and%20documents/Ongoing/RM03JH12CD_QALY_Proj.pdf). Particular concerns we have with the application of the findings of this research to NICE decision-making include:

1. Resources cannot be invested in health care to ‘purchase’ individual QALYs. What is being purchased is an intervention that involves a ‘package’ of QALYs at a particular cost.

2. The methods described in the outline imply use of perfect divisibility and constant returns to scale assumptions to derive the WTP per QALY from the responses to surveys.

3. The application of a fixed WTP per QALY figure to the appraisal of different programmes producing different total QALYs assumes perfectly elastic societal demands for health improvements and other goods.

We fail to see, but are eager to learn how this can be used to improve the efficiency of resource use in the NHS.

Conflict of interest

Both commentaries support our stand on conflict of interest. However, GB again chastise us, this time for ‘innuendoes’ preventing them from discerning the prevalence of such conflicts of interest in the development and application of successive NICE guidelines. Our concern lies with establishing the principle of conflict of interest for others (e.g., NICE) to implement, not with the identification of individuals who might currently contravene such a principle. We provided a clear example to support our claim, without going so far as to name individuals. However, we are pleased to clear up any remaining ambiguities – some individuals listed as members of the Guidance Steering Committee of NICE 1 and the Methodology Working Party of NICE 2 had relationships with the pharmaceutical industry that are not acknowledged in the published guidelines documents.

Supporting NICE

GB believe that NICE is worthy of support from the academic community. We agree and have provided support by presenting methods for determining whether a health care intervention represents an unambiguous improvement in the efficiency of use of health care resources. Both Drummond and GB remain ‘wedded’ to the use of cost effectiveness ratios without showing how this achieves NHS objectives. Cost-effectiveness ratios provide no information about whether a particular intervention represents an improvement in efficiency of resource use (Birch and Gafni Reference Birch and Gafni1992, Reference Birch and Gafni2006b). Our concern is not with the existence of a body like NICE but with ensuring that the development and application of methods for use by NICE are consistent with improvements in efficiency of resource use. Drummond (1980) acknowledged that the only universal approach to choosing between programmes under a resource constraint with an objective of maximizing health gain is through the use of mathematical programming techniques. Although Drummond accuses us of being ‘heavy on diagnosis and light on treatment options’, while BG imply that the challenges we identify are presented in a non-constructive way, we have provided details of the application of such methods in several easily accessible publications (e.g., Birch and Donaldson Reference Birch and Donaldson1987, Birch and Gafni Reference Birch and Gafni1992, Sendi et al. Reference Sendi, Gafni and Birch2002).

The NICE experience of rapidly increasing expenditures emerging from the pursuit of improvements in efficiency without any evidence of increases in net health gains follows the outcomes of applying similar methods in other jurisdictions (Gafni and Birch Reference Gafni and Birch2003a, Reference Gafni and Birch2003b). Unlike Gold and Bryan, we don’t see this being something to cheer about.