Josephine van Dongen

3 Updated cost-effectiveness and risk-benefit analysis 65 Sensitivity and scenario analyses Univariate sensitivity analysis was conducted to identify critical parameters driving our results. In shor t, parameter variations included 25% lower and 25% higher rotavirus hospitalization rates and hospitalizations costs; vaccine-induced IS rates of 1:20,000 and 1:100,000 (base case 1:50,000) 32-35 , and IS complication rates of 0% and 9.6% (base case 4.8%). We also included slightly higher QALY losses based on the sensitivity analysis of Marlow et al. (for hospitalizations 0.0039 vs 0.0030 and for GP visits 0.0030 vs 0.0022) 20 ) As caregiver work-loss estimates for rotavirus AGE are influenced by local employment conditions and parental leave plans, they can vary substantially by country. Our sensitivity analysis therefore also included 100% higher caregiver productivity losses. Subsequently, we tested the impact of old vs new parameter estimates including caregiver work loss for mild and moderate rotavirus cases 4 and QALY losses for hospitalized cases 11 .We applied various discount rates: 2% and 4% for both costs and effects (3% in the baseline), as well as the Dutch discount rates (1.5% for effects and 4% for costs 10 ). Extensive sensitivity analyses were conducted on vaccine costs to determine the thresholds at which the vaccination strategies would become cost-saving under base-case assumptions. Additionally, strategy-specific scenarios included the following: a lower vaccination coverage of 75% for “targeted vaccination” (baseline 86%); decreased or increased herd protection, or no herd protection at all in case of universal vaccination. Because a shift to a biennial rotavirus epidemic pattern could theoretically increase the average age of first infection as a result of the reduced force of infection, we assessed the impact of an “older” age when first infected.To this end, we simulated scenarios where 50% or 75% of the 0–1 years old patients with rotavirus from baseline were 1–2 years old instead, and consequently had lower probabilities of seeking medical care, both GP and hospitalization. Finally, an “alternative universal vaccination” scenario was also analyzed where we assumed that “universal vaccination” would be recommended, but not covered by the publicly funded national immunization program. Instead, vaccines would be individually purchased for each infant with or without par tial reimbursement from health insurance. For this scenario, we assumed a coverage of 60%, no herd protection due to the lower coverage, and the actual market price (i.e., €135.32/child). For more details see also Additional file 2: Tables S2 and S3 .