[Skip to Content]
[Skip to Content Landing]
Views 109
Citations 0
Comment & Response
June 10, 2019

Benefits of Stock Epinephrine Likely Misestimated—Reply

Author Affiliations
  • 1Dartmouth-Hitchcock Medical Center, Section of Allergy and Immunology, Lebanon, New Hampshire
  • 2Dartmouth Geisel School of Medicine, Hanover, New Hampshire
  • 3Children’s Hospital Colorado, University of Colorado School of Medicine, Section of Allergy and Immunology, Aurora
JAMA Pediatr. Published online June 10, 2019. doi:10.1001/jamapediatrics.2019.1533

In Reply We appreciate the opportunity to comment on Gerald’s cost-effectiveness analysis of our model of school-based stock epinephrine programs,1 which confirms our conclusion that epinephrine autoinjectors are not cost-effective if purchased at $715 per unit. On our review, we note Gerald’s abridged analysis omits considerations of natural tolerance to peanut during school-age, all-cause age-adjusted mortality, and stochastic variation in rare food fatality outcomes afforded by Markov microsimulation and instead uses a simple deterministic model. However, our findings are aligned.1 Consistent with our original analysis, his model confirms the rarity of fatal food allergy as a measurable outcome and the high societal burden associated with steep epinephrine costs. Given unique modeling assumptions, it is not surprising that results of these decision analyses are not identical. We believe the difference in degree of cost-ineffectiveness between the models ($270 000 per QALY in our Markov model and $3.6 million per QALY in his approach) may result to a large degree from different applications of a 10-fold risk reduction applied in the base-case stock epinephrine analysis. Because the actual risk reduction afforded from epinephrine as a shared resource (either as a supplemental or universal approach) is unknown, our model applied a global 10-fold reduction to patients evaluated under both supplemental and universal stock epinephrine strategies. In an attempt to define a ceiling for value-based costing, we did not adjust this risk reduction by applying hours of school attendance. Because there is variation in hours of attendance between both children and adults, the 10-fold risk reduction benefit was applied globally. In contrast, it appears that Gerald’s model assumed additional discounts in the fatality risk reduction by only applying the 10-fold risk reduction for 7.25 hours per day and 176 days per year.