[Risk factors for death and their predictive value on diabetic kidney disease patients in intensive care unit based on MIMIC-III database].

Objective: To analyze the influencing factors of prognosis of patients with diabetic kidney disease (DKD) in intensive care unit (ICU), and analyze their predictive value.

Methods: Based on the inpatient information of more than 50 000 patients from June 2001 to October 2012 in the latest version of American Intensive Care Medical Information Database (MIMIC-III v1.4), the data of DKD patients were screened out, including gender, age, body weight, comorbidities [hypertension, coronary heart disease, chronic obstructive pulmonary disease (COPD), chronic kidney disease (CKD)], sequential organ failure assessment (SOFA) score, the length of ICU stay, the incidence of mechanical ventilation, vasoactive drugs and renal replacement therapy during the ICU hospitalization, complications of other diseases [ventilator-associated pneumonia (VAP), urinary tract infection (UTI), diabetic ketoacidosis (DKA), acute myocardial infarction (AKI)] and prognosis of ICU. At the same time, the blood routine and biochemical data of the first 24 hours in ICU and the extremum values during the ICU hospitalization were collected. Multivariate Logistic regression analysis was used to screen the prognostic factors of DKD patients in ICU, and receiver operating characteristic (ROC) curve was plotted to analyze the predictive value of death risk factors.

Results: 416 DKD patients were screened out, 20 patients were excluded due to data missing, and finally 396 patients were enrolled, including 220 survival patients and 176 dead patients. Compared with the survival group, the patients in the death group were older (years old: 57.13±13.04 vs. 52.61±14.15), with lower rates of hypertension and CKD (11.4% vs. 23.6%, 26.7% vs. 41.4%), higher SOFA scores and baseline values of blood urea nitrogen (BUN), serum creatinine (SCr) and blood K [SOFA score: 5.86±2.79 vs. 4.49±2.56, BUN (mmol/L): 18.4±10.0 vs. 14.8±9.0, SCr (μmol/L): 387.2±382.8 vs. 284.6±244.9, K (mmol/L): 4.64±0.99 vs. 4.33±0.86], and longer ICU stay [days: 2.65 (1.48, 5.21) vs. 2.00 (1.00, 4.00)], and the differences were statistically significant (all P < 0.01). Further analysis of laboratory tests extremum values during ICU hospitalization showed that the maximum (max) and minimum (min) values of white blood cell (WBC), BUN and SCr, and K in the death group were significantly higher than those in the survival group [WBC (×10/L): 17.3±10.3 vs. 14.5±7.3, WBC (×10/L): 7.9±4.1 vs. 6.7±2.7, BUN (mmol/L): 23.8±10.4 vs. 18.8±10.2, BUN (mmol/L): 11.0±6.6 vs. 9.3±6.6, SCr (μmol/L): 459.7±392.5 vs. 350.1±294.4, SCr (μmol/L): 246.6±180.3 vs. 206.9±195.4, K (mmol/L): 5.35±0.93 vs. 5.09±0.99], and the minimum values of hemoglobin (Hb) and glucose (Glu) were significantly lower than those in the survival group [Hb (g/L): 87.4±14.5 vs. 90.6±16.5, Glu (mmol/L): 4.0±1.7 vs. 4.6±2.0], and the differences were statistically significant (all P < 0.05). The incidences of mechanical ventilation and vasoactive drugs during ICU hospitalization in the death group were significantly higher than those in the survival group (37.5% vs. 24.1%, 32.4% vs. 20.0%, both P < 0.01), and the incidences of UTI and AMI in the death group were significantly higher than those in the survival group (29.5% vs. 19.1%, 8.5% vs. 3.6%, both P < 0.05). Multivariate Logistic regression analysis showed that age [odds ratio (OR) = 1.019, 95% confidence interval (95%CI) was 1.003-1.036, P = 0.023], SOFA score (OR = 1.142, 95%CI was 1.105-1.246, P = 0.003), WBC (OR = 1.134, 95%CI was 1.054-1.221, P = 0.001) and BUN (OR = 1.010, 95%CI was 1.002-1.018, P = 0.018) were risk factors of death of DKD patients in ICU. ROC curve analysis showed that the area under ROC curve (AUC) of combination of risks factors of death was 0.706, the sensitivity was 61.6%, and the specificity was 73.2%.

Conclusions: In order to prevent DKD patients from getting worse in ICU, we should pay close attention to the blood biochemical indexes, especially the renal function indexes, and give timely treatment. At the same time, we should actively prevent the occurrence of complications such as infection and cardiovascular disease.