Ten patients (five women/five men), aged 51 ¡À 12 years, with type 1 diabetes for 24 ¡À 11 years treated by continuous subcutaneous insulin infusion (CSII) for more than 1 year (HbA1c: 7.5 ¡À 0.8 % ) performed a 30-min bout of exercise at a constant high-intensity load (15 % above their individual ventilatory threshold) on a cycle ergometer. All patients wore a subcutaneous ¡®needle-type¡¯ enzymatic glucose sensor linked to a portable monitor (Guardian? RT, Medtronic-MiniMed, Northridge, CA, USA) that had been inserted the previous evening. Sensor calibration was performed against capillary blood glucose immediately before the exercise. CGM values were recorded every 5 min from T-10 to T+30, then every 10 min during the recovery period from T+30 to T+90. These recorded values were compared with blood glucose assays performed on simultaneously collected venous samples.
Sensor functioning and tolerability raised no problems except for one sensor that could not be adequately calibrated. Data from this patient were excluded from the data analysis. An average blood glucose decrease of 63 ¡À 63 mg/dL (3.5 ¡À 3.5 mmol/L) (median decrease: 58 mg/dL [3.22 mmol/L]; range: -3 mg/dL [0.16 mmol/L] to 178 mg/dL [9.8 mmol/L]) occurred during exercise bouts, while CGM values decreased by 38 ¡À 49 mg/dL (2.11 ¡À 2.72 mmol/L) (median: 32 mg/dL [1.7 mmmol/L]; range: -15 mg/dL [0.83 mmol/L] to 58 mg/dL [3.22 mmol/L]). Cumulative paired glucose values (n = 135) could be analyzed. The correlation factor between CGM and blood glucose values was 0.957 with an intercept of 0.275. The mean difference between paired values according to Bland-Altman analysis was 10 ¡À 31 mg/dL (0.56 ¡À 1.72 mmol/L). Clarke error grid analysis showed 91 % of paired points in A and B zones, while 0 % , 9 % and 0 % of paired points were in the C, D and E zones, respectively.
Blood glucose changes during intensive physical-exercise bouts performed by CSII-treated type 1 diabetes patients can be estimated with acceptable clinical accuracy by online CGM.