Background
Incidental thyroid findings (ITFs) are increasingly detected on imaging performed for non-thyroid indications. Their growing prevalence and subsequent clinical implications remain poorly understood, raising concerns about overdiagnosis and unnecessary interventions. This study leverages artificial intelligence to systematically analyze radiology reports and assess the epidemiology and outcomes of ITFs.
Study Design
The retrospective cohort study included 115,683 adults without prior thyroid disease who underwent thyroid-capturing imaging at Mayo Clinic sites from July 1, 2017, to September 30, 2023. A transformer-based natural language processing (NLP) pipeline identified ITFs and extracted nodule characteristics from radiology reports across multiple imaging modalities.
Key Findings
Among the study population (mean age 56.8 years, 52.9% women), 7.8% had an ITF, with 92.9% being nodular. ITFs were more common in women, older adults, those with higher BMI, and in imaging ordered by non-Emergency Medicine specialties. Neck CT, PET, and nuclear medicine scans were more likely to detect ITFs compared to chest CT. Nodule characteristics were poorly documented, with size reported in only 44% of cases. Patients with ITFs had significantly higher odds of subsequent thyroid nodule diagnosis (OR 45), biopsy (OR 46.8), thyroidectomy (OR 55.8), and thyroid cancer diagnosis (OR 61.7). Most cancers were papillary thyroid carcinomas, larger when detected after ITFs (2 cm vs 1.3 cm).
Expert Commentary
This study provides compelling evidence that incidental thyroid findings frequently trigger diagnostic cascades leading to the detection of small, low-risk cancers. The findings underscore the potential for overdiagnosis and highlight the need for standardized reporting and more selective follow-up protocols for ITFs.
Conclusion
The AI-enabled analysis demonstrates that incidental thyroid findings are common and strongly associated with clinical cascades. These results call for a re-evaluation of current management approaches to minimize unnecessary interventions while ensuring appropriate diagnosis of clinically significant thyroid pathology.