Servier Laboratories Ltd, R (on the application of) v National Institute for Health & Clinical Excellence & Ors

Fairness normally requires NICE to disclose the fully executable economic model and underlying data used in a technology appraisal. Exceptionally, NICE may accept confidential material where its importance to the robustness of the appraisal is sufficiently great, disclosure has been properly pursued, and confidentiality is narrowly confined.

Even then, NICE has a positive duty to take all reasonable steps to obtain permission for disclosure. The court must allow NICE a margin of appreciation, but may intervene where reasonable steps have not been taken. NICE is entitled to evaluate scientific evidence independently and is not bound by the assessment of another expert body. Its reasons need only be intelligible and adequate to a technically informed reader.

Servier challenged revised Final Appraisal Determinations issued by the National Institute for Health and Clinical Excellence concerning treatments for osteoporotic fractures. Three grounds remained: failure to disclose an economic model and its underlying data; failure properly to consider a post hoc analysis of the TROPOS trial; and unlawful disability discrimination under the Disability Discrimination Act 1995.

The earlier appeal process had rejected or dismissed the relevant challenges. The central issues were whether confidentiality justified non-disclosure, whether NICE’s treatment of the scientific evidence was irrational or inadequately reasoned, and whether the treatment recommendations discriminated against disabled patients.

1. The claim for judicial review was allowed on ground (a), concerning non-disclosure of the economic model and underlying data. Grounds (c) and (e) were rejected.

2. Fairness and transparency normally required disclosure of the fully executable model and the data on which it was based. However, Eisai v NICE did not establish an absolute rule that confidentiality could never justify non-disclosure. The present case differed because the algorithms and coefficients were subject to express confidentiality undertakings given by their owner.

3. Even where confidentiality has been justifiably accepted, NICE remains under a positive duty to take all reasonable steps to obtain permission to disclose, particularly where the economic model and its data are central to understanding the appraisal. The court must give due weight to NICE’s judgment, but may intervene where the steps taken are unreasonable.

4. NICE had failed to discharge that duty. It should have promoted a meeting with the data owner and explored practical arrangements, including disclosure to named representatives subject to confidentiality undertakings.

5. NICE was not bound by the conclusions of EMEA or the Scottish Medicines Consortium concerning the post hoc TROPOS analysis. It had to consider the evidence, and did so. The appraisal committee was entitled to attach less weight to a post hoc subgroup analysis than to the trial’s overall results, and its reasons were intelligible and adequate to a technically informed reader.

6. The disability discrimination grounds failed. The relevant comparator was a patient unable to take alendronate for a reason unrelated to disability, rather than a patient able to take it. NICE applied the same treatment thresholds to both groups. Its approach also satisfied the general duty under section 49A of the Disability Discrimination Act 1995, and was justified by the high cost of alternative treatment and the need to protect the fair allocation of NHS resources.

7. NICE was required to negotiate disclosure of the model and data to all consultees or identified representatives on suitable confidentiality terms, allow further representations, and reconsider the FADs. The same recommendations could lawfully be retained after proper disclosure and reconsideration.