A scoring system to predict recurrence in patients with differentiated thyroid cancer

Abstract

Objectives: Though predictive models have been constructed to determine the risk of recurrence in differentiated thyroid carcinoma, various aspects of these models are inadequate. Therefore, we aimed to construct, internally validate and implement on a mobile application a scoring system to determine this risk within 10 years. Design: A retrospective cohort study in 1984-2016. Setting: A Spanish region. Participants: We enrolled 200 patients with differentiated thyroid carcinoma without distant metastasis at diagnosis. Main outcome measures: Time-to-recurrence. A risk table was constructed based on the sum of points to estimate the likelihood of recurrence. The model was internally validated and implemented as a mobile application for Android. Results: Predictive factors were follicular histology, T, N and multifocality. This risk table had a C-statistic of 0.723. The calibration was satisfactory. Conclusions: This study provides an instrument able to predict rapidly and very simply which patients with differentiated thyroid carcinoma have a greater risk of recurrence.