Efetividade dos análogos da insulina comparados as insulinas humanas em gestantes diabéticas: revisão sistemática e metanálise

Abstract

Diabetes Mellitus is considered the main metabolic complication of pregnancy. A woman may have diabetes previously or develop it during pregnancy. This pathology has been related to unfavorable maternal-fetal outcomes, which are aggravated by inadequate glycemic control. Human insulins are the first drug choice in this period, due to the proven safety in its use. However, there are still questions about the use of insulin analogues during pregnancy. Thus, the objective of this research was to determine the effectiveness of insulin analogues compared to human insulin to treat pregnant women with diabetes through a systematic review with meta-analysis. A systematic review of original articles from randomized and observational clinical trials comparing the use of insulin analogs with human insulins in pregnant women with prior diabetes (type 1 or 2) or gestational diabetes was conducted at the time of the start of each database by July 2017 in MEDLINE, CINAHL, EMBASE, ISI Web of Science, LILACS, Scopus, SIGLE and Google Scholar. The variables analyzed were maternal glycemic control (fasting glycemia and glycated hemoglobin), congenital malformation, fetal death, maternal death, macrosomia, gestational age, abortion, neonatal weight, neonatal hypoglycemia and maternal hypoglycemic episodes. Through the search strategy, 29 original articles were selected, of which 11 were randomized clinical trials and 18 were observational studies. In total, 6382 participants were analyzed. The bias risk analysis showed that all original articles were classified as intermediate with high risk of bias, so it was not possible to make the sensitivity analysis. The variable in which the meta-analysis showed to be favorable to the use of insulin analogues was gestational age [mean difference of -0.26 (CI95% 0.03 to 0.49; P = 0.02), but with significant heterogeneity (I2 = 38%; x2 = 16.24, DF = 10, P = 0.09). This result, in clinical practice, does not compromise fetal well-being, since all babies were born to term. In the other variables studied, the results of the meta-analyzes showed no difference in relation to the use of human insulin or insulin analogues. It was not possible to perform a meta-analysis of the variable maternal death, since this variable was not present in the selected articles. There was publication bias in the variables gestational age and neonatal weight. To date, the evidence found has a high bias risk and does not allow the conclusion that insulin analogues are equivalent, more or less effective, compared to human insulin to treat diabetic pregnant women.