Nice Restrictiveness Compared To the Market Authorization.

Abstract

To determine how often NICE recommendations are more restrictive than the market authorizations. 161 NICE Technology Appraisal decisions from 2007-2013 were evaluated. These reviews included 80 unique drugs from 37 disease conditions. For each generic drug included in a review, the corresponding market authorization was retrieved from the EMA or MHRA. NICE positive decisions were compared to the market authorizations. Any decision that included language that restricted the population eligible for reimbursement for a given therapy was categorized as “recommend with restrictions. ” NICE positive decisions that were not more restrictive than the market authorization were categorized as “recommend. ” Negative decisions were categorized as “do not recommend. ” Restrictions were also quantified and categorized. NICE issued “do not recommend” decisions in 32% of the reviews from 2007-2013. The overall rate at which NICE issued “do not recommend” decisions increased after 2010, but this did not pass traditional levels of statistical significance (p=. 21). NICE issued positive decisions in 68% of reviews, but the decision was more restrictive than the market authorization in 52% of the positive decisions. NICE’s restrictiveness has decreased since 2007, with the exception of 2013 where 60% of NICE’s positive decisions were “recommend with restrictions. ” For the “recommend with restrictions” reviews, there are 1.7 restrictions on average (range 1-4, s. d.. 11) added to the market authorization. The most prevalent type of restrictions were for contraindicated or intolerance. ” In 2007-2013, NICE issued “recommend with restrictions” decisions in 36% of reviews and issued both “recommend” and “do not recommend” decisions in 32% of reviews. NICE was more restrictive than the market authorization in 52% of the positive decisions, though NICE’s restrictiveness seems to be declining over time. An independent analysis of NICE decisions in 2007-2013 found a statistically significant different distribution of decisions than reported in the NICE website (p=. 01).