Development and validation of a predictive scoring system for hypoglycaemic agents for optimal control of blood glucose during glucocorticoid therapy.

Abstract

Glucocorticoid (GC) treatments are often used. There is limited information on the prediction of hyperglycaemia after GC administration. This study aimed to identify the risk factors for hyperglycaemia after glucocorticoid (GC) administration and the need for hypoglycaemic agents to correct it and to develop and validate a novel scoring system for predicting GC-induced hyperglycaemia. In a development set, 508 adults receiving prednisolone (PSL) for the first time were divided into two groups based on treatment with or without hypoglycaemic agents. Clinical and laboratory parameters were compared, and risk factors were identified using logistic regression analysis after performing univariate analyses between the two groups. A point-addition scoring system with several categories and coefficients for each risk factor was constructed to predict the need for hypoglycaemic agents. The scoring system was then applied and validated on two validation sets: A and B. Older age, higher glycated haemoglobin percentage, body mass index and initial PSL dosage were identified as risk factors. The sensitivity, specificity and accuracy of the scoring system were 70.6%, 81.9% and 77.1% in the development set; 75.8%, 78.4% and 77.4% in validation set A; and 79.4%, 73.9% and 75.3% in validation set B respectively. By fitting the total score in the development set and the probability of hyperglycaemia to a logistic curve, a figure was created to show the probability of GC-induced hyperglycaemia in patients scheduled to receive GC. This scoring system is a novel, valid and reliable tool for predicting GC-induced hyperglycaemia and the need for hypoglycaemic agents to correct it.