Assessing Symptomatic Hypocalcemia Risk After Total Thyroidectomy: A Prospective Study

Abstract

Introduction The most common postoperative complication of total thyroidectomy is hypocalcemia, usually monitored using serum parathyroid hormone and calcium values. Objective To identify the most accurate predictors of hypocalcemia, construct a risk assesment algorithm and analyze the impact of using several calcium correction formulas in practice. Methods A prospective, single-center, non-randomized longitudinal cohort study on 205 patients undergoing total thyroidectomy. Parathyroid hormone, serum, and ionized calcium were sampled post-surgery, with the presence of symptomatic or laboratory-verified asymptomatic hypocalcemia designated as primary outcome measures. Results Parathyroid hormone sampled on the first postoperative day was the most sensitive predictor of symptomatic hypocalcemia development (sensitivity 80.22%, cut-off value ≤2.03 pmol/L). A combination of serum calcium and parathyroid concentration sampled on the first postoperative day predicted the development of hypocalcemia during recovery with the highest sensitivity and specificity (94% sensitivity, cut-off ≤2.1 mmol/L, and 89% specificity, cut-off ≤1.55 pmol/L, respectively). The use of algorithms and correction formulas did not improve the accuracy of predicting symptomatic or asymptomatic hypocalcemia. Conclusions The most sensitive predictor of symptomatic hypocalcemia present on the fifth postoperative day was PTH sampled on the first postoperative day. The need for algorithms and correction formulas is limited.

Keywords: calcium; hypocalcemia; parathyroid hormone; risk assessment; total thyroidectomy.

Fig. 1

CONSORT flow diagram.

Fig. 2

ROC curves of correlations between laboratory parameters and occurrence of symptomatic hypocalcemia. ( A ) PTH concentration determined one hour after surgery predicts the onset of hypocalcemia symptoms with a sensitivity of 69.89% and a specificity of 84.91% at a cut-off value of ≤1.51 pmol/L (ROC analysis; AUC 0.832; 95% CI 0.773–0.881; P value <0.0001). ( B ) Serum calcium concentration determined one hour after surgery predicts the occurrence of symptomatic hypocalcemia with a sensitivity of 72.63% and a specificity of 88.07% at a cut-off value of ≤2.01 mmol/L (ROC analysis; AUC 0.807; 95% CI 0.746- 0.859; P value <0.0001). ( C ) PTH concentration determined on the first postoperative day predicts the occurrence of symptomatic hypocalcemia with a sensitivity of 80.22% and a specificity of 78.22% at a cut-off value of ≤2.03 pmol/L (ROC analysis; AUC 0.835; 95% CI 0.775- 0.884; P value <0.0001). ( D ) Serum calcium concentration determined on the first postoperative day predicts the occurrence of symptomatic hypocalcemia with a sensitivity of 79.07% and a specificity of 85.88% at a cut-off value of ≤2.02 mmol/L (ROC analysis; AUC 0.891; 95% CI 0.835–0.934; P value <0.0001).

Fig. 3

ROC curves of correlations between calcium correction formulas and symptomatic hypocalcemia. ( A ) The Clase correction formula predicts the development of symptomatic hypocalcemia with a sensitivity of 65.26% and specificity of 89.91% at the cut-off value of ≤1,801 mmol/L (ROC analysis; AUC 0.751; 95% CI 0.686–0.809; P value <0.0001). ( B ) The Jain correction formula predicts the development of symptomatic hypocalcemia with a sensitivity of 65.26% and a specificity of 89.91% at a cut-off value of ≤1.847 mmol/L (ROC analysis; AUC 0.756; 95% CI 0.691–0.814; P value <0.0001). ( C ) The Payne correction formula predicts the development of symptomatic hypocalcemia with a sensitivity of 63.16% and specificity of 88.99% at a cut-off value of ≤1.8525 mmol/L (ROC analysis; AUC 0.746; 95% CI 0.680–0.804; P value <0.0001). ( D ) The Ridefelt correction formula predicts the development of symptomatic hypocalcemia with a sensitivity of 66.32% and a specificity of 88.99% at a cut-off value of ≤1.8426 mmol/L (ROC analysis; AUC 0.753; 95% CI 0.688–0.811; P value <0.0001).

Fig. 4

ROC curves of the correlation of laboratory parameters and the occurrence of laboratory verified hypocalcemia on the fifth postoperative day. ( A ) PTH concentration determined one hour after surgery predicts the occurrence of hypocalcemia on the fifth postoperative day with a sensitivity of 55.86% and a specificity of 88.64% at a cut-off value of ≤ 1.27 pmol/L (ROC analysis; AUC 0.760; 95% CI 0.695–0.818; p  < 0.0001). ( B ) Serum calcium concentration determined one hour after surgery predicts the occurrence of hypocalcemia on the fifth postoperative day with a sensitivity of 58.41% and a specificity of 87.91% at the cut-off value of ≤ 1.99 mmol/L (ROC analysis; AUC 0.757; 95% CI 0.692- 0.814; p  < 0.0001). ( C ) PTH concentration determined on the first postoperative day predicts the occurrence of hypocalcemia on the fifth postoperative day with a sensitivity of 57.94% and a specificity of 82.56% at a cut-off value of ≤ 1.55 pmol/L (ROC analysis; AUC 0.747; 95% CI 0.679–0.806; p  < 0.0001). ( D ) Serum calcium concentration determined on the first postoperative day predicts the occurrence of hypocalcemia on the fifth postoperative day with a sensitivity of 94% and a specificity of 66.67% at a cut-off value of ≤ 2.1 mmol / L (ROC analysis; AUC 0.860; 95% CI 0.799–0.908; P-value <0.0001).

Fig. 5

ROC curves of calcium correction formulas predicting the occurrence of laboratory verified hypocalcemia on the fifth postoperative day. ( A ) The Clase correction formula predicts the development of hypocalcemia on the fifth postoperative day with a sensitivity of 70.80% and a specificity of 69.23% at the cut-off value of ≤1.8791 mmol/L (ROC analysis; AUC 0.707; 95% CI 0.639–0.768; P value <0.0001). ( B ) The Ridefelt correction formula predicts the development of hypocalcemia on the fifth postoperative day with a sensitivity of 64.60% and a specificity of 76.92% at the cut-off value of ≤1.8867 mmol/L (ROC analysis; AUC 0.715; 95% CI 0.647–0.776; p  < 0.0001). ( C ) The Jain correction formula predicts the development of hypocalcemia on the fifth postoperative day with a sensitivity of 51.33% and a specificity of 91.21% at the cut-off value of ≤1,803 mmol/L (ROC analysis; AUC 0.719; 95% CI 0.652–0.780; P value <0.0001).

Fig. 6

Algorithm for assessing the risk of hypocalcemia on the fifth postoperative day using variables identified in the logistic regression model (age in years, PTH in pmol/L, Ca in mmol/L, iCa in mmol/L).

Fig. 7

ROC curve of the algorithm assessment of hypocalcemia risk on the fifth postoperative day (ROC analysis; AUC 0.820; cut-off value ≥2; sensitivity 69.91%; specificity 83.52%; 95% CI 0.760–0.870; P-value <.0001).