A blood-draw timing- and season-adjusted PTH-calcium model for predicting post-endoscopic/robotic thyroidectomy hypocalcemia.

Abstract

Background: Post-thyroidectomy hypocalcemia is a common complication after endoscopic/robotic surgery and may delay discharge. Early postoperative parathyroid hormone (PTH) and calcium (Ca) are useful predictors, but their circadian and seasonal variation is rarely considered. We assessed whether incorporating blood-draw timing and season improves prediction of hypocalcemia within 48 h.

Methods: In this single-center cohort, we developed the model retrospectively (January 2019-December 2023) and validated it prospectively (January 2024-June 2025) in adults undergoing endoscopic or robotic total/completion thyroidectomy. The primary outcome was biochemical hypocalcemia within 48 h after the earliest postoperative blood draw, defined as corrected Ca < 2.10 mmol/L or ionized Ca < 1.10 mmol/L. We compared Model A (PTH only), Model B (PTH + total Ca), and Model C (Model B plus hours since surgery, diurnal and seasonal terms, clinical covariates, and a prespecified PTH × hours interaction).

Results: Among 1,052 patients (derivation n = 784; validation n = 268), biochemical hypocalcemia occurred in 24.9%. In derivation, Model C achieved an AUROC of 0.87 (95% CI: 0.84-0.89; optimism-corrected: 0.85), outperforming Model B (0.81) and Model A (0.78; all P < 0.001). Calibration was excellent (intercept: -0.01, slope: 0.99, Brier: 0.13). In temporal validation, Model C maintained strong performance (AUROC: 0.84, 95% CI: 0.80-0.88; intercept: 0.03, slope: 0.97, Brier: 0.14). Decision curve analysis showed a greater net benefit; at a 20% threshold, Model C reduced unnecessary prophylactic supplementation by about 11 per 100 patients versus Model A.

Conclusion: Modeling blood-draw timing and season improves early prediction of post-thyroidectomy hypocalcemia beyond standard PTH/Ca measures and may support more efficient postoperative management.