Development and validation of a new model for predicting hypocalcaemia after total thyroidectomy: the NuGra model.

Background: Hypocalcaemia is the most frequent complication after total thyroidectomy. Finding a method for its early detection has become a priority. A single-center prospective cohort study was conducted to identify risk factors for postoperative hypocalcaemia, develop an early detection model, and test its validity in a different group of patients.

Methods: The sample was composed of patients who underwent a total thyroidectomy between May 2012 and September 2015. Demographic, clinical, laboratory, and surgical data were collected. The incidence of hypocalcaemia and permanent hypoparathyroidism was calculated. Bivariate and multivariate analysis identified several independent predictors of hypocalcaemia, which were used to design a predictive model. The validity of the model was subsequently tested in a different cohort. Area under the ROC curve (AUROC) was calculated to determine its predictive power.

Results: The study and validation groups included 352 and 118 patients, respectively. Seventy-three patients developed laboratory-confirmed hypocalcaemia (20.7%), and symptomatic in 43 (12.2%). Multivariate analysis confirmed as independent predictors of hypocalcaemia the higher number of parathyroid glands identified [OR 1.41(0.98, 2.02); p = 0.063] and pre-to-postoperative gradient of parathormone decline [OR 1.06(1.04, 1.08); p < 0.001]. Based on these variables, the NuGra (Number of parathyroid glands identified-Gradient of decline) model was developed for predicting laboratory-confirmed hypocalcaemia. Its predictive power was high (AUROC 0.902, CI 0.857-0.947) for the study and the validation group (AUROC 0.956, CI 0.919-0.993).

Conclusions: A higher number of parathyroid glands identified and a higher gradient of parathormone decline are risk factors for post-thyroidectomy hypocalcaemia. The NuGra model is useful for early prediction of individual risk for hypocalcaemia.