Thyroid nodules are common, with an incidence 2–3 times higher in women than in men . In our sample the men to women ratio was 0.1 but the median age was similar, suggesting that other factors may have contributed to this marked predominance of women. In Portugal, women's use of health services may be greater than in men which may imply greater consumption of health resources. In a recent study including 1339 patients, where the quality of the US report was assessed, the proportion of women was also high (85%) .
According to our results maximum size was reported in all nodules. As expected, a positive correlation was found between maximum size and age . Previous studies also found a high reported maximum size feature [23, 27]. Maximum size is sometimes the only criterion used in daily practice to decide whether to perform FNAB or not. However, the size of the nodule alone has little value in assessing the risk of malignancy . Maximum size, although not necessary for calculating the malignancy risk of the nodule, is a very important feature for the decision on the recommendation for FNAB according to the EU-TIRADS and other risk stratification systems [1, 8,9,10]. The three-axis diameters of the nodules were only mentioned in 7.8% of the participants and only in one case was the nodular volume calculated. Nodule diameters were more frequently reported in NUTS2 Lisbon and Tagus Valley than in the other NUTS2, suggesting some regional variability. In the follow-up of thyroid nodules, US growth criteria according to ATA 2015 and ACR TI-RADS (20% increase in at least two nodule diameters with a minimal increase of 2 mm or more than a 50% change in volume) requires information about the three-axis diameters or alternatively about the volume of the nodule. Furthermore, if the nodule shape is not mentioned, it is necessary to know the nodule diameters to calculate it. Given that the non-oval shape is a malignancy risk feature, high risk nodules may go unnoticed if information about the nodule shape is not available. According to our data, vascularity was only reported in 5.8% of participants. Intra-nodular vascularity has low predictive value, sensitivity and specificity in the diagnosis of malignancy [29, 30]. The EU-TIRADS guidelines states that, although the routine use of Doppler US is not recommended for US malignancy risk stratification it can be used to differentiate solid tissue from thick colloid or to enhance the detection of the limits of a nodule in an isoechogenic parenchyma . In addition, the observation of a “spoke-and-wheel like” peripheral and central pattern of flow in a nodule can predict exceedingly bloody aspirates and non-diagnostic cytological findings .
The localization of the nodule is essential for its clinical integration and for a correct identification in future US exams and in preparation for FNAB or other diagnostic or therapeutic procedures. In the present study, the localization of the nodules was reported in the vast majority of cases. Other studies also confirm a high adherence by radiologists to the localization in the US report [23, 27].
Regarding the five main US features of the thyroid nodules, a very low report of shape, margins and echogenic foci was observed, with no significant differences between NUTS2. In contrast, the composition (76.5%) and echogenicity (53.2%) have often been reported, with significant differences between NUTS2. At NUTS2 Lisbon and Tagus Valley, the composition and the echogenicity features were reported more frequently than in the remaining NUTS2. In a previous study on the quality of US exams there was also a low report of the margins (8%) and echogenic foci (14%) and no reference to the shape . Composition and echogenicity were mentioned in 72% and 59% of the reports, respectively. In another study the composition (64.0%) and echogenicity (43.8%) features were also more reported than margins (11.0%), echogenic foci (35.1%) and shape (3.9%) features . A recent Canadian study also showed a higher report of composition (64.0%) and echogenicity (42.0%) than shape (2.0%), echogenic foci (33.0%) and margins (22.0%) features .
Dominant nodules classified as EU-TIRADS 5 (with at least one suspicious feature) were reported in 8.7% of participants. In a previous study, which included 1029 nodules 7.4% of the nodules were categorized as EU-TIRADS 5 . In addition the same study showed the presence of marked hypoechogenicity, microcalcifications, irregular margins and non-oval shape in 1.5%, 3.3%, 1.4% and 2.8% of nodules. In another study in which the quality of US reports was analyzed, non-oval shape and marked hypoechogenicity were never reported . In addition, microcalcifications and irregular margin were only reported in 3% and 1% of nodules, respectively. Compared to our results, microcalcifications and irregular margins were less reported in those studies. Our data confirms the low non-oval shape US report. According to our results, malignancy risk features were reported more often in men than in women. Radiologists may be more careful in men due to their increased risk of differentiated thyroid carcinoma (DTC). In fact, men with nodular thyroid disease undergoing FNAB have a higher risk of DTC than women, although the prevalence of the disease and the absolute number of cancers is higher in women . In addition, men with DTC may have a more advanced disease at the time of diagnosis, although male sex is not an independent prognostic factor in itself .
According to our results, the utility score of 6 was observed in only 1.7% of the nodules. In addition, only 21.8% of the nodules had a score ≥ 4. Our data is slightly better compared to the results of a previous study in which the score of 6 and the score ≥ 4 were identified in only 0.4% and 13.7%, respectively .
Most of the reports (92.2%) did not refer a risk category. The risk category was reported more often in men than in women. As mentioned above, this finding may be due to greater attention from radiologists in men due to its increased risk. The reference to the EU-TIRADS classification system was made in only one participant. These results document the low adherence in Portugal, namely in NUTS2 Center and Lisbon and Tagus Valley to these classification systems in clinical practice. In addition, and although the European-based EU-TIRADS classification, developed by ETA, was published in 2017, its use by Portuguese radiologists is residual. Other studies have shown greater adherence by radiologists to the US-based risk stratification systems. A Canadian study showed that 29.9% of reports cited a risk stratification system .
The description of non-dominant nodules with criteria for FNAB was made in only 8.4% of the participants. These results suggest that, in most cases, the report was focused on just one nodule considered to be "dominant".
According to our results, only 7.0% of reports did not mention the presence or absence of cervical lymph nodes. These results are better than those found in another study that included 136 participants and showed that 74% of the US examinations did not mention lymph node status at all .
According to several scientific societies, the use of a thyroid imaging and reporting data system is very useful for physicians in their clinical practice [1, 10]. The use of US pattern categories, which include several individual sonographic features, when compared to individual US characteristics, increase the interobserver report reproducibility. In addition, these systems offer a simple communication of results, facilitating decision-making regarding the indication for diagnostic FNAB. Taking into account our results, most reports did not allow the establishment of the nodules’ malignancy risk and US features of aggressive thyroid cancers, thus preventing an informed decision on the recommendation of diagnostic FNAB. It is therefore urgent to make radiologists aware of the need to adhere to these US-based risk stratification systems. Studies like this can contribute to the improvement of clinical practice and to a reduction of unnecessary diagnostic interventions. Finally, there are some strengths and limitations of this study that should be mentioned. It is a multicenter one involving participants from the three largest NUTS2 in mainland Portugal. The thyroid US exams were performed by radiologists in their routine clinical practice as part of their activity in private radiology centers and not in specialized centers. In this way our results show the real life clinical practice. In this study, a non-probabilistic sampling method was used, with convenience selection of participating endocrinologists. For this reason, biases with interference in the representativeness of the results cannot be excluded. However, unselected participants were consecutively included. Furthermore, most of the thyroid US exams had been ordered by a general practitioner and endocrinologists involved in the study did not choose the radiologists who performed the exams. It should also be noted that, despite three endocrinologists from each NUTS2 region being selected, there was no further sampling stratification, namely with regard to the volume of exams performed by each radiologist, as it was not possible to access this information. The retrospective nature of this study did not prevent access to the full content of the reports. The final sample size was determined by the number of participants included within the approved study time. NUTS2 Center did not reach the expected number of participants. In contrast, NUTS2 Lisbon and Tagus Valley included more participants than expected. It was assumed that the non-reference in the report to a nodule feature did not mean the absence of that finding. In some cases, this may have been the radiologist's intention, but the failure to include that information in the report prevented the risk assessment of the nodule.