BACKGROUND: Assessment for thyroid nodules includes ultrasound (US) and cytology according to the Bethesda classification. There is no firm consensus regarding clinical management for nodules classified as Bethesda III. Our aim was investigate the value of US to predict malignancy in these nodules. METHODS: Patients with Bethesda III nodules who underwent thyroid surgery from July 2011 to July 2013 were included. Inclusion criteria mandated that US were available for review by two observers blinded to each other's results and histological outcome. The nodules were scrutinized with six US criteria: hypoechoic attenuation (HA), irregular margins (IM), taller than wide, microcalcifications (MC), loss of halo, and increased central vascularity. Disagreements between observers were solved by consensus. RESULTS: There were 141 patients (121 women) with a mean age of 55 years. Mean nodule size was 25 mm. The malignancy rate was 13 %. Interobserver ratios were moderate to very strong for all six predictors (kappa = 0.60-0.94). However, only HA, IM, and MC were predictors of malignancy by univariate analysis (all p < 0.002). Logistic regression revealed an odds ratio of malignancy versus no malignancy for HA 4.8, IM 3.3, and MC 4.0 (all p < 0.05). The positive and negative predictive value for malignancy when having one or more of these three criteria was 22 % and 98 %, respectively. CONCLUSION: HA, IM, and MC were predictors of malignancy in Bethesda III nodules. In addition, the negative predictive value for any of these three criteria was high; a nodule that lacks all of these three criteria is thus unlikely to be malignant.
BACKGROUND: Assessment for thyroid nodules includes ultrasound (US) and cytology according to the Bethesda classification. There is no firm consensus regarding clinical management for nodules classified as Bethesda III. Our aim was investigate the value of US to predict malignancy in these nodules. METHODS:Patients with Bethesda III nodules who underwent thyroid surgery from July 2011 to July 2013 were included. Inclusion criteria mandated that US were available for review by two observers blinded to each other's results and histological outcome. The nodules were scrutinized with six US criteria: hypoechoic attenuation (HA), irregular margins (IM), taller than wide, microcalcifications (MC), loss of halo, and increased central vascularity. Disagreements between observers were solved by consensus. RESULTS: There were 141 patients (121 women) with a mean age of 55 years. Mean nodule size was 25 mm. The malignancy rate was 13 %. Interobserver ratios were moderate to very strong for all six predictors (kappa = 0.60-0.94). However, only HA, IM, and MC were predictors of malignancy by univariate analysis (all p < 0.002). Logistic regression revealed an odds ratio of malignancy versus no malignancy for HA 4.8, IM 3.3, and MC 4.0 (all p < 0.05). The positive and negative predictive value for malignancy when having one or more of these three criteria was 22 % and 98 %, respectively. CONCLUSION: HA, IM, and MC were predictors of malignancy in Bethesda III nodules. In addition, the negative predictive value for any of these three criteria was high; a nodule that lacks all of these three criteria is thus unlikely to be malignant.