14 chapter 1 aspiration cytology (FNAC) alone, as cytology has limited insight in the tissue structure, including the assessment of capsular and/or vascular invasion that distinguish follicular adenoma (FA) from follicular thyroid carcinoma (FTC) [25]. Indeterminate cytology makes up approximately a quarter of all FNAC results [16, 18, 26]. To subsequently obtain a definitive diagnosis, diagnostic thyroid lobectomy would be required, resulting in a histopathological diagnosis of thyroid carcinoma in approximately 25% of indeterminate nodules [16, 18, 26, 27]. In other words, approximately 75% of these patients would undergo diagnostic thyroid surgery for a benign nodule: futile surgery from an oncological perspective, with associated costs, morbidity, and unwarranted risks of surgical complications [28, 29]. Improving the diagnostic workup Additional diagnostics should be considered to aid the malignancy risk stratification and hopefully obtain a more definitive diagnosis before proceeding to diagnostic surgery. That way, unbeneficial diagnostic surgery for benign nodules can be avoided when malignancy can accurately be ruled out. When malignancy is confirmed or highly suspicious, depending on other clinical and pathological characteristics, a (sub-)total thyroidectomy can be considered at once instead of two-step surgery that starts with a diagnostic hemithyroidectomy [14]. The 2015 American Thyroid Association guidelines proposed that an ideal rule-out diagnostic for thyroid carcinoma should have a negative predictive value similar to a benign cytological diagnosis (~96.3%) and the ideal rule-in test a positive predictive value that is at least similar to a malignant cytological diagnosis (~98.6%) [17]. Diagnostic accuracy versus clinical utility In times where shared-decision making and cost-effectiveness are increasingly important in daily clinical practice, both physicians and patients desire more from a diagnostic test than merely well-validated diagnostic accuracy and high rule-in and/or rule-out capacity. They are additionally interested in actual changes in outcomes that matter to patients. Therefore, instead of simply focusing on the highest sensitivity and/or specificity, a diagnostic test is better appreciated by end points such as desired minimal rates of accurately prevented unbeneficial surgeries or accurately managed carcinomas. The extent to which the use of a diagnostic test improves health outcomes relative to the current best alternative, is defined as clinical utility [30]. Similar to the four well-known phases in clinical drug research, several comparable hierarchical systems have previously been proposed to evaluate diagnostic tests. The most well-know is the six-step Fryback-Thornbury hierarchy, which was originally presented in 1991 and designed for the evaluation of imaging techniques (Figure 1) [31, 32]. The evaluation of diagnostic accuracy occupies a central position in this and other systems, as the accurate identification of patients with the index
RkJQdWJsaXNoZXIy MTk4NDMw