Blog

To quantify the cost-effectiveness of rescue medications for pediatric status epilepticus: rectal diazepam, nasal midazolam, buccal midazolam, intramuscular midazolam, and nasal lorazepam.

Decision analysis model populated with effectiveness data from the literature and cost data from publicly available market prices. The primary outcome was cost per seizure stopped ($/SS). One-way sensitivity analyses and second-order Monte Carlo simulations evaluated the robustness of the results across wide variations of the input parameters.

The most cost-effective rescue medication was buccal midazolam (incremental cost-effectiveness ratio ([ICER]: $13.16/SS) followed by nasal midazolam (ICER: $38.19/SS). Nasal lorazepam (ICER: −$3.8/SS), intramuscular midazolam (ICER: −$64/SS), and rectal diazepam (ICER: −$2,246.21/SS) are never more cost-effective than the other options at any willingness to pay. One-way sensitivity analysis showed the following: (1) at its current effectiveness, rectal diazepam would become the most cost-effective option only if its cost was $6 or less, and (2) at its current cost, rectal diazepam would become the most cost-effective option only if effectiveness was higher than 0.89 (and only with very high willingness to pay of $2,859/SS to $31,447/SS). Second-order Monte Carlo simulations showed the following: (1) nasal midazolam and intramuscular midazolam were the more effective options; (2) the more cost-effective option was buccal midazolam for a willingness to pay from $14/SS to $41/SS and nasal midazolam for a willingness to pay above $41/SS; (3) cost-effectiveness overlapped for buccal midazolam, nasal lorazepam, intramuscular midazolam, and nasal midazolam; and (4) rectal diazepam was not cost-effective at any willingness to pay, and this conclusion remained extremely robust to wide variations of the input parameters.

For pediatric status epilepticus, buccal midazolam and nasal midazolam are the most cost-effective nonintravenous rescue medications in the United States. Rectal diazepam is not a cost-effective alternative, and this conclusion remains extremely robust to wide variations of the input parameters.