Evaluation of accuracy and influence factors of bedside blood glucose monitoring in critically ill patients

Abstract

To evaluate the accuracy and influence factors of point-of-care testing (POCT) for glucose in critically ill patients. Two hundred and forty critically ill patients aged ≥18 years in department of critical care medicine were enrolled. According to blood glucose level (BGL) during glucose control, patients were divided into three groups: (1) hypoglycemia group, BGL<4.5 mmol/L, n=32; (2) euglycemia group, BGL 4.5-8.3 mmol/L, n=138; (3) hyperglycemia group, BGL>8.3 mmol/L, n=70. The blood samples from vein, artery and capillary of patients were collected synchronically and the blood glucose of POCT were determined with glucose oxidase (GOD) and glucose dehydrogenase (GDH) methods, respectively, compared with blood glucose reference values of laboratory [hexokinase method (HK method)]. The accuracy of POCT for glucose and influence factors were analyzed statistically by the logistic regression method. (1) The inaccurate rates of glucose values in blood samples from vein, artery and capillary in hypoglycemia group (GDH method: 25.00%, 40.62%, 40.62%; GOD method: 59.38%, 71.88%, 71.88%) were significantly higher than those in euglycemia group (GDH method: 2.90%, 9.42%, 7.97%; GOD method: 18.12%, 27.54%, 27.54%) and hyperglycemia group (GDH method: 1.43%, 8.57%, 4.28%; GOD method: 11.43%, 8.57%, 11.43%, all P<0.01). (2) The average levels of difference for the glucose reference value of laboratory and the glucose value measured by glucometer in hypoglycemia group were 0.41-0.69 mmol/L (GDH method) and 0.92-1.18 mmol/L (GOD method), in euglycemia 0.16-0.33 mmol/L and 0.77-0.90 mmol/L, in hyperglycemia group -0.06-0.18 mmol/L and 0.56-0.76 mmol/L, respectively. (3) The correlation coefficients between the laboratory and glucometer in hypoglycemia group were respectively 0.812-0.853 (GDH method) and 0.723-0.816 (GOD method). The correlation coefficients in euglycemia group were 0.862-0.890 and 0.768-0.857. They were elevated to 0.922-0.957 and 0.896-0.922 in hyperglycemia group (all P<0.01). (4)On the influence factors of accuracy in POCT for glucose, BGL, acute physiology and chronic health evaluation II (APACHEII) score and hematocrit level were into the logistic regression equation. The accuracy of POCT for glucose was markedly affected by the BGL, severity of patients and hematocrit level during glucose control. Under hypoglycemia, the agreement between laboratory and glucometer measurements was obviously fallen and it tended to overestimate the patient's real glucose value.