AN ANALYSIS OF NICE ‘RECOMMENDED IN LINE WITH CLINICAL PRACTICE’ HEALTH TECHNOLOGY

ASSESSMENT DECISIONS Phill O’Neill1, Amanda Chapman1, Nancy Devlin1, Aurea Duran2

1. Office of Health Economics 2. Pfizer Ltd.

For further information, contact poneill@ohe.org

1. BACKGROUND The outcome of technology appraisal (TA) decisions by the National Institute of Health and Care Excellence (NICE) in the UK are of 4 ‘types’: Recommended (either in line with clinical practice or in line with marketing authorisation), Optimised, Only in research and Not Recommended [1]. Between January 2007 and September 2014 NICE made 77 “recommended in line with clinical practice” (RiLwCP) decisions (64 of which were for medicines). These are counted as “recommended” in NICE’s summary statistics of its decisions. The implications for patient access of this categorisation are not clear. In order to capture common features of these decisions we develop a framework for the criteria used to reach decisions, as reported in guidance documentation. To assess patient access we use a previously developed method [2] - the ‘M’ score - to calculate the access associated with each RiLwCP decision.

Acknowledgements This research was funded and initiated by Pfizer Ltd.

References [1] NICE (2014) Technology appraisal decisions. NICE Statistics. Available at: http://www.nice.org.uk/News/NICE-statistics [Accessed 5 October 2014) [2] O’Neill P, Devlin N (2010) “An analysis of NICE’s restricted (or ‘optimised’) decisions” Pharmacoeconomics. 28(11), 987-993

2. AIMS • To develop a framework to assess the criteria underlying

RiLwCP decisions.

• To assess the level of patient access associated with NICE TA decisions for medicines that are categorised as RiLwCP.

3. METHODS

• By considering the final recommendations, summaries of evidence and committee’s considerations, we identified six emerging themes related to choice of recommendation.

• We calculate a measure of patient access (M: 0=no access, 100=full access) for each of the 64 RiLwCP decisions for medicines made between January 2007 and September 2014. M=(p/P)X100, where P is the set of potentially eligible patients (given the TA scope and license), and p is the number of patients for whom NICE did recommend the treatment.

4. RESULTS For many of the RiLwCP decisions assessed we were able to attribute more than one criterion to the decision, and found there to be some subjectivity in interpreting guidance documentation. The table below provides an example of each criterion.

5. DISCUSSION 6. CONCLUSIONS

• Our analysis provides some context for NICE reported outcomes of TA decisions, specifically the nature of recommendations ‘in line with clinical practice’.

• Whilst each appraisal and committee deliberation is different, the terminology used to describe NICE decision outcomes is important, as this acts as a reference point for many users of NICE guidance.

• The use of RiLwCP as a decision outcome is not as transparent as other classes of NICE decision. In particular, there is ambiguity in some cases between the criteria for defining an outcome as RiLwCP and ‘optimised’.

• It is often appropriate for HTA bodies to make recommendations for products which do not extend to the full marketing authorisation, for example due to evidence of clinical or cost-effectiveness in different populations. These decisions involve optimising treatment.

• According to NICE, a recommendation in line with clinical practice reflects how the new treatment would fit with current clinical practice. Whereas criteria 1 – 4 outlined above reflect clinical practice (to varying degrees of formality), in many cases we could identify only clinical/cost-effectiveness as the motivation for limiting the patient population (with respect to marketing authorisation). Along with the fact that many decisions were associated with M scores below 50 (i.e. recommended for less than half the patients for whom it is licensed), this makes it difficult to distinguish between RiLwCP and ‘optimised’ recommendations.

Criterion Example

1. Reference to previous NICE TA

Explicitly refers to a previous NICE TA of the same or a related product

TA294 Afilbercept for wet age-related macular degeneration. Afilbercet recommended for use in patients according to same criteria as previous NICE TA for competitor product, until both can be appraised in a multiple technology appraisal. This was despite suggestion by clinical experts that Afilbercept may provide benefits to a wider patient population.

2. Reference to relevant clinical guideline

Recommendations are matched to existing clinical guidelines including

NICE guidelines

TA188 Somatropin for the treatment of growth failure in children. BTS/SIGN guideline is referred

to describe the step-wise approach to treatment that Somatropin use should conform with.

3. Fits within an established pathway of care

Recommendation aligned with a pre-existing patient management

pathway (with no particular ‘guideline’ for reference)

TA223 Treatment of intermittent claudication in people with peripheral arterial disease, where

Naftidrofuryl “represents one part of a wider programme of management”.

4. "Clinical opinion“

Guidance is explicitly influenced by input from clinical experts

TA315 Canagliflozin in combination therapy for treating type 2 diabetes. Only recommended

where sulfonylurea was not appropriate given extensive experience of clinicians with the current

treatment.

5. Clinical/Cost-effectiveness evidence matching

Guidance recommendation is shaped by the clinical and/or cost-

effectiveness evidence for the technology

TA181 Pemetrexed for the first-line treatment of non-small-cell lung cancer. Recommended for

specific histologically-determined subgroup based on clinical effectiveness evidence. Requires a

change in clinical practice to implement.

6. Non-pharmaceutical

All non-medicinal products in our sample that were recommended were done so ‘in line with

clinical practice’.

We were able to calculate the M score for 32 of the 64 RiLwCP decisions; the average M score for those recommendations was 35 (range: 3 – 100); this means that the medicine was recommended for on average 35% of potentially eligible patients (based on license and TA scope). If we conservatively assume that M was 100 for all cases where M could not be calculated, then the average M would be 67 - still only two-thirds of potentially eligible patients.