OBJECTIVES : ICER estimates a “potential annual budget impact threshold” (currently $991M for 2018-2019) to determine the maximum price at which a new technology may exceed affordability. This threshold is determined using US economic indicators and is, by definition, an estimate and not a fixed value; how sensitive this estimate is to variation in economic conditions is nevertheless unclear.

METHODS : Sensitivity of ICER’s threshold to variation in the underlying parameters was evaluated using a budget impact calculator developed (Microsoft Excel®) specifically for this purpose. A case study was derived from ICER’s recent assessment of Nonmetastatic Castration-Resistant Prostate Cancer (nmCRPC) adopting the US payer perspective and obtaining inputs from ICER’s published report. Using ICER’s $991M threshold as the reference case, underlying parameters were altered to examine the influence on maximum annual drug price. Using the most recent US gross domestic product (GDP) growth rate and number of new molecular entities approved in 2018, an alternative threshold of $1,228M was calculated. An additional scenario analysis averaged values for 2016-2018 and derived a threshold of $740M.

RESULTS : Our calculator produced similar results to ICER’s budget impact analysis in an initial validation step: 11% market share uptake for nmCRPC therapies resulted in $858M-$890M estimated budget impact among 59,000 treated patients across therapeutic options. Using the new “2018” threshold, maximum potential drug prices increased by 21.6%-23.9% for APA and 24.0%-30.0% for ENZ. Using the average 2016-2018 threshold resulted in drug price reductions ranging from 22.9%-25.3% for APA and from 25.4%-31.8% for ENZ.

CONCLUSIONS : This analysis indicates that assumptions and estimates underlying calculation of ICER’s budget impact threshold may have a notable impact on estimated drug prices. ICER’s threshold may be subject to variation depending on the source and perspective. Future ICER analyses should include scenario analysis to address uncertainty around the estimated threshold value.