368 chapter 6 Modelled lifelong costs and utilities To estimate lifelong costs and utilities, a Markov model with 12 health states and a one-year cycle length was constructed using Stata (version 14.2. StataCorp, College Station, TX, USA). Model structure The model represented health states that may occur from the second year onwards for either an [18F]FDG-PET/CT-driven workup or diagnostic surgery (Figure 2). These health states included active surveillance (i.e., follow-up of the thyroid nodule with yearly ultrasound), end of followup (i.e., patients discharged from active surveillance without thyroid surgery), observation after thyroid surgery (i.e., hemithyroidectomy [HT], total thyroidectomy [TT], completion TT [cTT], and/or radioactive iodine [RAI] ablation), medication-dependent hypothyroidism following HT, permanent complications due to HT or (c)TT, recurrent (including persistent) malignant disease after HT or (c) TT and/or RAI, or death. Health states following HT or TT may apply to patients with either benign or malignant disease. The “cTT + RAI” procedure and recurrent disease states (grey-shaded shapes in Figure 2) only apply to patients with malignant disease. Model parameters Values for the (time-dependent) probabilities in the Markov model were collected from a comprehensive Medline literature search, from Statistics Netherlands, and/or from the EfFECTS trial and adhered to the Dutch national guidelines (Table 2) [17, 467, 501, 527]. Parameters for which no information was found or that varied highly among literature were estimated by a local expert panel, including an endocrinologist, a nuclear medicine physician, and a health economist. For patients undergoing active surveillance for an [18F]FDG-negative nodule, a mean follow-up of three years was assumed. The negative predictive value (NPV) of [18F]FDG-PET/CT was 95.1% in the EfFECTS trial [501]. To prevent overestimation of the accuracy of an [18F]FDG-PET/CT-driven workup, we used this NPV to assume a 0.049 (=1-0.951) probability of missed malignancies in unoperated patients (i.e., a false-negative [18F] FDG-PET/CT), even though none were reported in the EfFECTS trial and its extended follow-up [501]. We assumed that any missed malignancies would be detected within the first five years of follow-up, and could occur among patients residing in the “active surveillance” or “end of follow-up” state (Table 2). Cost parameters Costs for thyroid-related procedures and costs for each cycle in a particular health state were derived from reference prices, 2019 reimbursement rates, and previous cost-utility studies, where appropriate and available, and adhered to the national guidelines (Table 3) [29, 176, 178, 476, 477, 495, 496, 558]. For the “active surveillance” state, we assumed one yearly visit to the endocrinologist and an ultrasound of the neck every 12-24 months. Productivity losses for thyroid-related procedures were inferred from the reported iPCQ data
RkJQdWJsaXNoZXIy MTk4NDMw