OBJECTIVE:

To determine the risk factors for requiring insulin treatment in women with gestational diabetes.

MATERIALS AND METHODS:

Observational, case-control, prospective, nested study in a cohort of patients with gestational diabetes. To determine the variables associated or not with the probability of requiring insulin treatment, a multivariate logistic regression analysis was performed. It was studied the discriminatory capacity of glucose and glycosylated hemoglobin basal concentrations, according to the need of treatment with insulin through the analysis of ROC curves; the area under the curve was estimated. With the Youden Index, the optimal cut-off point was calculated.

RESULTS:

113 patients diagnosed with gestational diabetes mellitus were studied, of which 27.4% (n = 31) required insulin treatment for glycemic control (case group); 72.6% (n = 82) achieved good metabolic control with diet and exercise (control group). The predictors of the need for insulin treatment were history of previous gestational diabetes, body mass index at the beginning of pregnancy, diagnosis of gestational diabetes before the third trimester, basal glycemia on the glucose tolerance curve and the value of glycosylated hemoglobin. The discriminating ability of basal glycaemia, by ROC curve analysis, had a C-statistic of 0.822 and for glycosylated hemoglobin of 0.718. Basal glycemia greater than 90 mg/dL predicted the need for insulin treatment with a sensitivity of 93.1% and specificity of 65.43%. Glycosylated hemoglobin value greater than 5% predicted the likelihood of insulin treatment with sensitivity of 93.3% and specificity of 41.4%.

CONCLUSIONS:

When diagnosing gestational diabetes, it is possible to select a group of pregnant women with a higher risk of needing medical treatment.