Let cost effectiveness models be open to scrutiny

In light of the recent ruling over the National Institute for Health and Clinical Excellence's (NICE) decision on donezepil,^{1 2} we wish to comment on the adversarial system of drug evaluation and the inadequacy of NICE providing read-only versions of cost effectiveness models for the purposes of reviewing their decisions.

The assertion that “[NICE] is not in a position to deal with the reality of restrictions being placed on [the models] by those who supply them”² is not defensible. NICE could specify terms and conditions to technology assessment groups (TAGs) to allow full disclosure of what, after all, is publicly funded research. Indeed, they should be disclosed under the Freedom of Information Act, with non-disclosure acceptable only when in the public interest.

Recently, the Sheffield TAG published a correction to a cost effectiveness model for multiple sclerosis treatment caused by a coding error.³ This model underpinned a high profile and novel shared risk policy.⁴ Because even the simplest deterministic models need thousands of calculations, usually coded by one person, errors are hardly surprising. More sophisticated simulation models use complex computational routines, the detail of which is rarely scrutinised. Under current TAG contracts NICE cannot quality assure these models. Furthermore, a recent objective comparison of validated health economic models for diabetes found wide variability.⁵

The current adversarial system of economic evaluation is unacceptable to all stakeholders. Cost effectiveness models could be produced by consensus under the joint direction of NICE and industry. This would reduce costs and hasten access to health technologies that all agree are good value for money.