374 chapter 6 heteroscedastic data [479, 501, 577]. Minor imbalances in baseline characteristics and malignancy rates were observed across the allocated groups despite stratified randomisation (Table 1). To avoid an impact of these imbalances on costs and utilities over the lifelong period, we also adjusted for these covariates: the local benign/borderline or malignant histopathological diagnosis, EQ-5D-5L utility score at baseline, medical history (binary, represented by the periodic use of non-thyroid medication), and productivity at baseline (represented by the patient-reported contractual work hours per week. Unadjusted results are presented in Supplementary tables 3-5 and Supplementary figure 1. Results are presented as means and their 95% confidence intervals (CI), mean difference and 95% CI, and p values, where appropriate. All analyses adhered to the intention-to-treat principle. A p value ≤ 0.05 is considered statistically significant. Data analysis was performed using SPSS Statistics version 26 (IBM Corp., Armonk, NY, USA). Cost-utility analysis Cost-effectiveness acceptability curves (CEACs) were used to graph the probability that an [18F]FDGPET/CT-driven workup is cost effective compared to diagnostic surgery, as a function of willingness to pay (WTP) for a QALY. In the Netherlands, a willingness-to-pay threshold of €50,000 per QALY is recommended by the Dutch Council for Public Health and Health Care for conditions with an intermediate disease burden [578]. The probability of cost-effectiveness was calculated as the onesided p value for the difference in net benefit (net benefit = WTP × QALYs − costs). The statistical analysis of the net benefit was identical to the analysis for costs and QALYs separately. To explore the impact of individual parameters in the Markov model, univariate sensitivity analyses were performed and presented in a tornado diagram. Individual parameters were set at extreme values (Table 4), while keeping the other parameters at their base-case value and for each trial patient simulating 10,000 extrapolated patient histories beyond one year. Results Between July 2015 and October 2018, 132 adult patients with a Bethesda III or IV thyroid nodule were enrolled in the EfFECTS trial (Table 1). All patients completed all study-related procedures and one-year follow-up. Diagnostic surgery was avoided for 25 of 91 (27%) patients in the [18F]FDG-PET/ CT-driven group, as compared to 1 of 41 (2%) in the diagnostic surgery group (p=0.001) [501]. The unoperated index nodules remained unchanged in size and unsuspicious on ultrasound surveillance and were considered benign after one year. During study follow-up, 106 (80%) patients underwent diagnostic surgery: 29 (22%) nodules were malignant, 8 (6%) were borderline tumours, and 69 (52%) were benign. For the [18F]FDG-PET/CT-driven group, this resulted in avoided futile diagnostic surgery for 25 of 62 (40%) benign nodules.
RkJQdWJsaXNoZXIy MTk4NDMw