Diagnostic Accuracy of Midline Pediatric Neck Masses.

OBJECTIVE: To assess clinical evaluation, ultrasound, and previously published predictive score at preoperatively diagnosing midline neck masses and demographic or clinical associations that aid in differentiation of thyroglossal duct and dermoid cysts.

STUDY DESIGN: Retrospective chart review.

SETTING: Tertiary care children's hospital.

SUBJECTS: Patients <18 years undergoing primary midline neck mass surgery with histopathologic diagnosis of thyroglossal duct or dermoid cyst who had preoperative ultrasound performed were included.

METHODS: An electronic medical record query generated 142 patients whose histopathologic diagnosis was thyroglossal duct cysts (TGDCs) or dermoid cysts (DCs). Charts were reviewed for demographic and clinical features. A radiologist blindly reviewed patients' ultrasounds for SIST (septae + irregular walls + solid components = thyroglossal) score components. Each patient received 3 preoperative diagnoses: clinical, ultrasound, and SIST. Statistical analyses were conducted to determine association of demographic, clinical, or radiographic variables with diagnoses. Specificity, sensitivity, and predictive values were evaluated for each candidate diagnosis.

RESULTS: There were 83 TGDCs and 59 DCs. Tenderness, infection history, depth relative to strap muscles, and SIST components were more common among TGDCs. Sensitivity and positive and negative predictive values surpassed 63% for each diagnostic modality. SIST score outperformed other diagnostic modalities with sensitivity, positive predictive value, and negative predictive value of 84%, 91%, and 81%, respectively. Clinical and ultrasound assessments were largely inconclusive for dermoid cysts, but SIST correctly identified 89% of DCs.

CONCLUSION: SIST score was the most accurate predictor of pediatric midline neck masses. Clinical and radiographic findings may help guide preoperative diagnosis, although further evaluation is required to develop more efficacious diagnostic tools.