408 chapter 7 It was previously hypothesized that patients undergoing active surveillance may experience some degree of long-term, cyclic psychological distress concentrated around their periodic follow-up, caused by the suspense and uncertainty of repeated ultrasound and/or FNAC procedures and the outcome thereof [495]. No indications of such an effect were observed during the one-year followup of the current study, which included one protocolled follow-up visit. We assessed the MIC to estimate which HRQoL differences could be perceived as important by the patient, but only for the ThyPRO. Following a systematic literature search, we identified no suitable country- and/or disease-specific MIC values for the EQ-5D-5L and RAND-36 questionnaires for our study population. As cultural differences play an important role in the perception of HRQoL and the EQ-5D-5L uses country-specific algorithms to calculate utilities, we did not use the available generic EQ-5D-5L MIC values for other countries [603-605]. For the ThyPRO questionnaire, MIC values were recently established in 435 Danish patients with benign thyroid disease, including 135 (31%) nontoxic goitre patients [599]. The ThyPRO does not apply country-specific algorithms to calculate its scale scores [597]. Finally, because the ThyPRO was considered the most relevant for the current study and because of limitations of the small group sizes, we chose not to apply Norman’s rule of thumb to assess the MIC for the EQ-5D-5L and RAND-36 [606]. Our previous intention-to-treat analysis of the EQ-5D-5L results demonstrated non-inferiority of an [18F]FDG-PET/CT-driven workup with regard to general HRQoL based on QALYs [28, 501]. For the current comprehensive HRQoL study, a per protocol analysis with exclusion of crossovers was considered most appropriate, despite its known drawbacks with regard to bias [607]. As we aimed to compare the perceived HRQoL of active surveillance after a negative [18F]FDG-PET/CT (i.e., reassurance by a negative test result that the risk of malignancy was low, followed by the effects of active surveillance in absence of a definitive histopathological diagnosis) to the HRQoL of diagnostic surgery (i.e., a definitive diagnosis at the cost of effects caused by recovery from surgery and potential surgical complications), patients who crossed over between strategies fitted neither group in light of the counselling and treatment they received. A main limitation of the current study is the limited sample size of the malignancy and active surveillance groups. Designed as a secondary, explorative analysis of HRQoL data acquired during a randomised controlled trial, the study was not a priori powered to distinguish HRQoL effects [501]. Post hoc sensitivity analysis, however, showed that in the smallest cohort (i.e., active surveillance group, n=23) we have 80% power to detect an effect size of approximately 8.9 points on the ThyPRO scales (i.e., 0.539 standard deviations of the observed mean difference between two dependent groups), a number smaller than considered clinically relevant, predefined by MIC data [599]. Yet, the limited group sizes may have caused underreporting of the statistical significance of any smaller effects, especially as compared to similar-sized effects in the larger benign histopathology group. Also, it limited the multivariate analysis by lacking the statistical power to construct a mixed linear
RkJQdWJsaXNoZXIy MTk4NDMw