[The effect of blood glucose fluctuations on prognosis of critically ill patients in intensive care unit].

OBJECTIVE:To determine the association between glucose fluctuations and hospital mortality in intensive care unit (ICU) patients. METHODS:A retrospective study involving 90 critically ill patients in ICU according to the patients' outcome were divided into survivors (49 cases) and nonsurvivors (41 cases), in whom the blood glucose level was monitored in the first 72 hours, and the initial blood glucose (GluAdm), the average blood glucose (GluAve), glucose standard deviation (GluSD), coefficient of variation glucose (GluCV) were determined, then GluAdm, GluAve, GluSD, and GluCV were compared between survivors and nonsurvivors, and the receiver operating characteristic curve (ROC curve) was applied to evaluate the association between blood glucose fluctuation and prognosis. According to the values of GluSD, GluCV, the critically patients were divided into four subgroups, and mortality in ICU and hospital was compared. RESULTS:The levels of GluAdm, GluAve, GluSD, GluCV of nonsurvivors were higher than those of survivors [GluAdm: (11.47+/-3.91) mmol/L vs. (9.23+/-2.96) mmol/L, GluAve: (9.22+/-1.31) mmol/L vs. (8.28+/-1.15) mmol/L, GluSD: (2.62+/-0.97) mmol/L vs. (1.66+/-0.64) mmol/L, GluCV: 0.28+/-0.10 vs. 0.20+/-0.07, all P<0.05]. When ROC was applied, the area under the curve (AUC) of GluSD, GluCV were 0.782+/-0.049 and 0.757+/-0.053, they were higher than that of the GluAdm and GluAve (0.669+/-0.058 and 0.690+/-0.056, both P<0.05). When GluSD was 4.35-5.66 mmol/L, the ICU mortality was 95.7%, hospital mortality was 98.6%; when GluCV was 0.378-0.500, the ICU mortality was 83.3%, hospital mortality was 100.0%. CONCLUSION:The increase in GluSD and GluCV in critically ill patients is significantly correlated with ICU mortality and hospital mortality, and they are more valuable in predicting ICU mortality than GluAdm, GluAve. Diminution in fluctuation of blood glucose might be an important aspect of glucose management.