522 chapter 13 based literature are available, the consulted experts should preferably be independent and have no interest in the outcomes of the research project, to prevent optimistic overestimation. When parameters are estimated using indirect evidence from literature on related topics – with or without intervention of an expert panel – because no direct evidence is available on the research topic in question, it should be reasonably justifiable why the related literature is considered applicable. For example, Heinzel et al. inappropriately applied favourable diagnostic accuracy parameters of MIBI scintigraphy in unselected thyroid nodules to their index population of indeterminate thyroid nodules even though (less favourable) performance parameters in indeterminate thyroid nodules were already available [177, 296, 300]. In addition, this study probably also underestimated the modelled costs for MIBI and thyroid surgery [29, 177]. As such, even though their conclusion that MIBI scintigraphy is likely cost-effective as compared to the Afirma® GEC may be correct, the inappropriately idealized parameters regrettably weaken their study’s evidential value and validity. Second, while adhering to the applicable national guidelines for economic evaluations, cost-utility studies should attempt to construct a model that on the one hand is methodologically manageable and clear to interpretation by researchers as well as readers, and on the other hand also reflects the real-world clinical practice as best as possible. Although a cost-utility model will always remain a simplification, I believe that it should be extensive enough to not ignore (clinically) relevant details, such as the consequences of incidental [18F]FDG-PET/CT findings or nondiagnostic MD results. It should also include both the protocolled patient management as well as the possibility of treatment noncompliance, such as delayed surgery for a benign nodule in a patient initially undergoing active surveillance. The Dutch guidelines recommend to separately distinguish all cost units that contribute to the total and incremental costs of the disease and its treatment [477, 777]. In our cost-utility study of the EfFECTS trial, we adhered to the 2015 Dutch guidelines, resulting in a comprehensive cost-utility model based on the 1-year trial results [28, 477, 777]. Most previous costutility studies of additional diagnostics in indeterminate thyroid nodules, however, constructed a more simplified cost-effectiveness model using a theoretical base case [29, 176-178, 495, 496, 558, 579]. A limited 1- to 5-year time horizon was often applied, ignoring any long-term HRQoL and costs [29, 176, 178, 495, 579]. In addition, possible nondiagnostic test results, treatment noncompliance, and crossover between management strategies were regularly disregarded, even though these realistic scenarios generally decrease the efficacy of the index test and result in additional health care expenses [158, 159, 176, 178, 495, 496, 579]. Although different national guidelines may have been adhered, I feel that some of these studies may have cut some corners and insufficiently reflect real-world clinical practice. This may have resulted in overestimation of the impact of the index test and may compromise the validity of the conclusions of these studies. Future high-quality, more extensive cost-utility studies are required for other additional diagnostics than [18F]FDG-PET/CT in indeterminate thyroid nodules, preferable based on RCT data and with stricter adherence to the relevant practice guidelines for economical evaluations.
RkJQdWJsaXNoZXIy MTk4NDMw