ObjectiveThe ketogenic diet (KD) is a diet low in carbohydrates and rich in fats which has long been used to treat refractory epilepsy. The metabolic changes related to the KD may increase the risk of hypoglycemia, especially during the first days. The study focused on the impact of KD initiation on glycemia in non‐diabetic patients with refractory epilepsy.MethodsThe subjects were 10 pediatric patients (6 boys, mean age 6.1 ± 2.4 years), treated for intractable epilepsy. Blinded continuous glucose monitoring system (CGM) Dexcom G4 was used. Patients started on their regular diet in the first 36 hours of monitoring, followed by an increase in lipids intake and a gradual reduction of carbohydrates (relations 1:1; 2:1; 3:1; 3.5:1). We analyzed changes in glycemia during fat: nonfat ratio changes using a generalized linear model.ResultsThe mean monitored time per person was 6 days, 10 hours and 44 minutes. The mean ± SD glycemia for the regular diet was 4.84 ± 0.20 mmol/L, for the carbohydrates/fat ratio of 1:1 it was 4.03 ± 0.16, for the ratio of 2:1 it was 3.57 ± 0.10, for the ratio 3:1 it was 3.39 ± 0.13 and for the final ratio of 3.5:1 it was 2.79 ± 0.06 mmol/L (P < 0.001). The portions of time spent in glycemia ≤3.5 mmol/L (≤2.5 mmol/L respectively) were: on the normal diet 0.88% (0.31%) of the monitored period, during 1:1 KD ratio 1.92% (0.95%), during 2:1 ratio 3.18% (1.02%), and during 3:1 and 3.5:1 ratios 13.64% (2.36%) of the monitored time (P < 0.05).SignificanceContinuous glucose monitoring system shows the dynamic of glucose concentration in ketogenic diet treatment initiation. It may be a useful tool to control the effects of this diet on glucose metabolism, especially in hypoglycemia detection.