OBJECTIVES: Due to the increased number of high-cost orphan drug approvals and intensified economic pressure to drive down cost, payers are beginning to increase management of orphan drugs (US: step requirements, tier differentials; EU: price tenders, hospital formulary exclusions). This paper aims to identify key variables of an orphan product that drives the management level within disease categories.

METHODS: All products approved for orphan indications from 1983 to 2017 in US (n=669) were identified. We excluded products with multiple indications, as the level of management could not be solely attributed to the orphan indication, and products with no competition as this does not allow for preferential management of one (more cost effective) option versus another. We created a database of products/market attributes, including interchangeability of mechanism of action (MoA), route of administration (RoA), annual price, number of products available in indication, approval date, prevalence, and target pediatric population in the US/EU. We then classified management into separate levels. Separate random forest models were run in R to determine impact of each variable in US and EU3.

RESULTS: Model out-of-bag (OOB) error was 10%, 14%, and 21% for commercial, Medicare, and Medicaid books of business, respectively. In test data, the models correctly predict management at rates of 65%, 75%, and 80% for commercial, Medicare, and Medicaid, respectively. The strongest predictors of management were MoA, competition, and RoA in the commercial space and RoA, prevalence, and MoA in Medicare. RoA, launch year, and pediatric are the most predictive for Medicaid. In EU3, strongest predictors of management are MoA and number of competitors.

CONCLUSIONS: Variables included in the analysis are stronger predictors for management in commercial and Medicaid spaces than in the Medicare space. Predictors of management vary based on the channel. Within the EU3, the stronger predictors of management varied amongst countries.