OBJECTIVE: Expectations about the budget impact of new drug launches may affect payer behavior and ultimately consumer costs. Therefore, we evaluated the accuracy of pre-launch US budget impact estimates for a sample of new drugs. METHODS: We searched for publicly available budget impact estimates made pre-launch for drugs approved in the US from 1 September 2010 to 1 September 2015 and compared them to actual sales. Accuracy was calculated as the ratio of pre-launch estimate to actual sales. Quantitative analyses, including multivariate regressions, were used to identify factors associated with accuracy. RESULTS: We identified 25 budget impact estimates: 23 for one of 14 individual drugs and 2 for the category of PCSK9 inhibitors. The ratios of predicted to actual budget impact ranged from 0.2 (estimate was 20% of sales) for secukinumab to 37.5 (estimate was 37.5 × sales) for PCSK9 inhibitors. Mean ratio was 5.5. In multivariate analyses, larger eligible population, more recent estimate year (e.g. 2015 vs. 2012), and being first in class, were associated with statistically significant, greater overestimation of budget impact. CONCLUSIONS: For every $5.5 of predicted cost, there was $1 of actual cost to the healthcare system. This study, although based on a small, non-random sample, suggests possible cognitive bias on the part of the estimators. Overestimating budget impact may lead to early access restrictions, higher copays, and other changes that ultimately impact patients. Analysts and non-profits should be attuned to likely sources of error in order to improve their predictions.
OBJECTIVE: Expectations about the budget impact of new drug launches may affect payer behavior and ultimately consumer costs. Therefore, we evaluated the accuracy of pre-launch US budget impact estimates for a sample of new drugs. METHODS: We searched for publicly available budget impact estimates made pre-launch for drugs approved in the US from 1 September 2010 to 1 September 2015 and compared them to actual sales. Accuracy was calculated as the ratio of pre-launch estimate to actual sales. Quantitative analyses, including multivariate regressions, were used to identify factors associated with accuracy. RESULTS: We identified 25 budget impact estimates: 23 for one of 14 individual drugs and 2 for the category of PCSK9 inhibitors. The ratios of predicted to actual budget impact ranged from 0.2 (estimate was 20% of sales) for secukinumab to 37.5 (estimate was 37.5 × sales) for PCSK9 inhibitors. Mean ratio was 5.5. In multivariate analyses, larger eligible population, more recent estimate year (e.g. 2015 vs. 2012), and being first in class, were associated with statistically significant, greater overestimation of budget impact. CONCLUSIONS: For every $5.5 of predicted cost, there was $1 of actual cost to the healthcare system. This study, although based on a small, non-random sample, suggests possible cognitive bias on the part of the estimators. Overestimating budget impact may lead to early access restrictions, higher copays, and other changes that ultimately impact patients. Analysts and non-profits should be attuned to likely sources of error in order to improve their predictions.
Authors: Joost W Geenen; Svetlana V Belitser; Rick A Vreman; Martijn van Bloois; Olaf H Klungel; Cornelis Boersma; Anke M Hövels Journal: Eur J Health Econ Date: 2020-04-04