Revisiting the threshold of statistical significance as a method for governing the increased expenditure for innovative drugs
In clinical trials, the threshold of statistical significance (or
statistical threshold, ST) is traditionally set at P<0.05. As a result,
approval of innovative treatments by regulatory bodies (such as FDA and EMA) is
strongly influenced by this level of ST, and also reimbursement decisions consequently
rely on this statistical cut-off.
At our institution, a retrospective study is ongoing in which all
positive reimbursement decisions made in 2009 and 2010 by the Italian regulatory
agency (AIFA) are revisited for purely speculative purposes to determine
whether the clinical material supporting individual favourable decisions met which of the following values of ST: STA, P<0.001; STB,
P<0.01; STC,
P<0.02; STC,
P<0.03; STD, P<0.04; STE,
P<0.05. The clinical material re-examined included only controlled clinical
trials or meta-analysis.
While the preliminary results of our study confirm that, as expected, nearly 100% of favourable reimbursement decisions
made by AIFA are supported by STE , the progressive cost-containment effect of using a more conservative
cut-off (e.g. STD , STC ,
STB, or STA) suggests that this approach could be a very simple method for governing the sustainability
issues raised by innovative drugs (particularly biotechnology treatments).
Of course, one limitation of this method is that cost-effectiveness issues and value-based pricing are not taken into
account; however, for reasons of marketing constraints and/or common sense and/or pharmacoeconomic
considerations the price values claimed by pharmaceutical industry are already
set, without excessive variations, around
the upper limits of current cost-effectiveness thresholds. Another limitation is
that this approach would introduce a discrepancy between clinical trial design
(that would of course continue to be based on STE) and regulatory
decisions (that could be based on a threshold other than STE).
Competing interests:
Although the author is member of the national price negotiation committee of AIFA, this Rapid Response reflects only his personal views.
21 January 2011
Andrea Messori
coordinator
Lab. of Pharmacoeconomics, c/o ESTAV Centro, SSR Regione Toscana, 50100 Firenze, ITALY
Rapid Response:
Revisiting the threshold of statistical significance as a method for governing the increased expenditure for innovative drugs
statistical threshold, ST) is traditionally set at P<0.05. As a result,
approval of innovative treatments by regulatory bodies (such as FDA and EMA) is
strongly influenced by this level of ST, and also reimbursement decisions consequently
rely on this statistical cut-off.
At our institution, a retrospective study is ongoing in which all
positive reimbursement decisions made in 2009 and 2010 by the Italian regulatory
agency (AIFA) are revisited for purely speculative purposes to determine
whether the clinical material supporting individual favourable decisions met which of the following values of ST: STA, P<0.001; STB,
P<0.01; STC,
P<0.02; STC,
P<0.03; STD, P<0.04; STE,
P<0.05. The clinical material re-examined included only controlled clinical
trials or meta-analysis.
While the preliminary results of our study confirm that, as expected, nearly 100% of favourable reimbursement decisions
made by AIFA are supported by STE , the progressive cost-containment effect of using a more conservative
cut-off (e.g. STD , STC ,
STB, or STA) suggests that this approach could be a very simple method for governing the sustainability
issues raised by innovative drugs (particularly biotechnology treatments).
Of course, one limitation of this method is that cost-effectiveness issues and value-based pricing are not taken into
account; however, for reasons of marketing constraints and/or common sense and/or pharmacoeconomic
considerations the price values claimed by pharmaceutical industry are already
set, without excessive variations, around
the upper limits of current cost-effectiveness thresholds. Another limitation is
that this approach would introduce a discrepancy between clinical trial design
(that would of course continue to be based on STE) and regulatory
decisions (that could be based on a threshold other than STE).
Competing interests: Although the author is member of the national price negotiation committee of AIFA, this Rapid Response reflects only his personal views.