Прогностические модели для определения вероятности ранней неонатальной гипогликемии у детей, рожденных матерями с сахарным диабетом 1 типа

Abstract

The purpose — to identify the key predictors of neonatal hypoglycemia in children born to mothers with type 1 diabetes mellitus (DM1) and to develop predictive models to determine the likelihood of early hypoglycemia. Material and methods. We examined 130 children of mothers with DM1, who were born and treated in the «Mother and Child” Republican Scientific and Practical Center in Minsk. The main group consisted of 101 infants born to mothers with pregravid DM1. The patients were divided into group 1 (Gr1, n = 56) with diagnosed early hypoglycemia and group 2 (Gr2, n = 45) without hypoglycemia. Results. The most significant clinical and biochemical predictors associated with the risk of hypoglycemia in the first hours of life of infants are insufficient compensation for DM in women (the average level of HbA1c in the 2–3rd trimesters is more than 6.5%), levels of cholesterol, triglycerides and LDL-cholesterol in the mother’s serum on the eve of childbirth, the value of the z-score of the child’s body weight at birth, the level of cord blood leptin, the presence of signs of a perinatal infectious disease in the child. Mathematical models of the probability of early hypoglycemia in newborns of mothers with DM1 were developed. The control of the models’ correct operation was carried out on a sample of 29 children of women with DM. The high reproducibility of the developed prognostic models was established. Conclusion. The predictive accuracy of models for determining the likelihood of early hypoglycemia in newborns of mothers with DM1 was 83.1% (95% CI 74.1–92.0) and 83.2% (95% CI 74.1–92.3), which justifies the expediency of their use in healthcare organizations offering perinatal care of various levels.