Accuracy Of Real-time Continuous Glucose Monitoring System Analysis Used In Patients With Septic Shock

Backgroud:Hyperglycemia is a common phenomenon in patients with septic shock, and they have large glucose variability which needs to measure glucose frequently. In present, blood glucose (BG) which is measured through bedside BG meter is used to evaluate patients in Intensive Care Unit (ICU). But it is not continuous blood glucose measurement, and may not accurately evaluate the trend of glucose variability. We may miss the incident of hyperglycemia and hypoglycemia. Continuous Glucose Mornitoring System (CGMS) measures the interstitial glucose continuously and it displays the trend of glucose variability all day which helps nurses monitor the glucose of patients better. Some of the researchers in China and abroad have indicated that CGMS is accurate used in critical patients in ICU. Other researchers in abroad think that although the CGMS could be clinically useful in critical patients, it is not sufficiently accurate and reliable at present to be used for clinical therapeutic decision. Septic shock is one of the special types of disease in ICU, it has special haemodynamic characteristics. There are less clinical researches to determine whether CGMS could be used in patients with septic shock.Objective:To examine the accuracy of Real-Time Continuous Glucose Mornitoring System (RTCGMS) used in patients with septic shock and compare the accuracy of RTCGMS used in patients with septic shock who have different tissue perfusion degrees.Method:From April to December in2013,61patients who were diagnosed septic shock were included in ICU of Peking Union Medical College Hospital. Patients were monitored interstitial glucose values (RTCGMS values) for at least48h using RTCGMS. Arterial blood glucose (ABG) samples were obtained following the protocol established in the ICU. The paired values between ABG and RTCGMS were calculated using numerical accuracy which includes median absolute difference (Median AD), median relative absolute difference (Median RAD), International Organization for Standardization (ISO) criteria, Clarke error grid analysis and Bland-Altman plot. The tissue perfusion indicators included lactate (Lac), central venous oxygen saturation (ScvO2) and perfusion index (PI). Using Bootstrap re-sampling technique, to compare whether the accuracy of RTCGMS used in septic shock patients with different tissue perfusion degrees had statistical difference. Results:In total,995(ABG and RTCGMS) pairs of glucose values were analyzed. In septic shock patients, the overall Median AD and Median RAD was respectively0.9mmol/L and10.9%, the ISO criteria were75.8%, and the correlation coefficient between ABG and RTCGMS values was0.776. In the Clarke error grid analysis,98.7%of the paired measurements were in the clinically acceptable zones A and B,0.8%and0.5%of the paired measurements were in zones C and D respectively. The Bland-Altman plot was shown that bias were-0.14mmol/L, upper and lower limits of agreement were3.44mmol/L and-3.73mmol/L Compared Lac>8mmol/L with2<Lac≤4mmol/L, the difference of ISO criteria in95%confidence intervals (Cl) were [0.026,38.710], patients with Lac>8mmol/L had better accuracy of RTCGMS than those with2<Lac≤4mmol/L (P<0.05). Compared60≤ScvO2<70%with ScvO2≤59%, the difference of Median RAD in95%CI were [0.007,0.127], patients with ScvO2≤59%had better accuracy of RTCGMS than those with60≤ScvO2<70%(P<0.05). Compared0.7<Pl≤1.4%with PI≤0.7%, the difference of Median RAD in95%CI were [0.002,0.076], patients with PI≤0.7%had better accuracy of RTCGMS than those with0.7<PI≤1.4%(P<0.05). Compared PI>3.0%with PI≤0.7%,0.7<PI≤1.4%and1.4<PI≤3.0%, the difference of ISO criteria in95%CI were respectively [3.322,28.302],[11.988,40.265] and [5.170,33.333], patients with PI>3.0%had better accuracy of RTCGMS than those with PI≤0.7%,0.7<PI≤1.4%and1.4<PI≤3.0%(P<0.05).Conclusion:RTCGMS used in patients with septic shock had certain accuracy, and RTCGMS was an effective supplement tool for regular blood glucose measurement. However, RTCGMS could not accurately estimate the hypoglycemia (ABG<4.4mmol/L). Moreover, when septic shock patients with low tissue perfusion (Lac, ScvO2and PI), the worse tissue perfusion was, the better accuracy of RTCGMS was. When septic shock patients with normal local tissue perfusion (PI), the better local tissue perfusion was, the better accuracy of RTCGMS was.