Database introduction
The data of present study were extracted from a large critical care database— Medical Information Mart for Intensive Care III (MIMIC III) that was published by the Massachusetts Institute of Technology, with approval from the review boards of the Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center[14]. All patients in the database were de-identified for privacy protection, so the need for informed consent was waived. After successfully completing the National Institutes of Health (NIH) Web-based training course and the Protecting Human Research Participants examination (certification number 36211094), we were given the permission to extract data from MIMIC III.
Inclusion and exclusion criteria
We included adult patients receiving DEX for at least 6 hours during the first 48 hours after ICU admission. For patients who were admitted to ICU more than once, only the first ICU stay was considered. Patients who were younger than 18 years or those with known ESRD were excluded. In addition, patients who spent less than 48 hours in ICU were excluded.
AKI Definition
AKI was defined according to the Kidney Disease Improving Global Outcomes (KDIGO) criteria: an increase in serum creatinine (SCr) by ≥ 0.3 mg/dl ( ≥ 26.5 μmol/l) within 48 h, or an increase in SCr to ≥ 1.5 times baseline within the prior 7 days, or urine volume < 0.5 ml/kg/h for 6 hours[15]. AKI stages were also defined according to KDIGO criteria[15]. The minimum SCr value within 7 days before admission was used as the baseline SCr[16]. When pre-admission SCr value was not available, the first SCr after admission was used as the baseline SCr[17]. In patients with deficient or insufficient urine output measurements, only the SCr criterion was applied.
Data extraction
Data extracted from MIMIC III using Structured Query Language (SQL) with Navicat Premium (version 12.0.28) included the demographic characteristics, comorbidities, simplified acute physiology score Ⅱ(SAPSⅡ), nonrenal sequential organ failure assessment score (SOFA), SCr value and urine output, use of dexmedetomidine, other sedatives, opioid agents, vasopressors and inotropes, nephrotoxic drugs and mechanical ventilation. Sepsis was defined as life-threatening organ dysfunction caused by a dysregulated host response to infection (sepsis 3.0)[18]. In the present study, patients with suspected or documented infection plus an acute increase of ≥ 2 SOFA points were recorded as sepsis. Use of vasopressor and inotropes was defined as the use of any of these agents, including vasopressin, norepinephrine, epinephrine, dobutamine, dopamine, and phenylephrine within 48 h after ICU admission. Use of nephrotoxic drugs was defined as the use of vancomycin, aminoglycoside and amphotericin. The estimated glomerular filtration rate (eGFR) was calculated using the Modification of Diet in Renal Disease study formula (MDRD) [19].
Management of missing data
Variables with missing data are common in the MIMIC III database. The missing values were less than 10% for all variables in the present study (see additional file 1: Supplementary Table 1). Single imputation was used to impute the missing values including SCr, urine output and weight [20].
Outcomes
The primary endpoint was in-hospital mortality in the present study. The secondary endpoints included 90-day mortality, length of stay (LOS) in ICU, LOS in hospital and recovery of renal function. Recovery of renal function was defined as being discharged from ICU with serum creatinine below 1.5 times the baseline value and normal urine output (> 0.5 ml/kg/h).
Statistical analysis
Propensity score matching (PSM) analysis was used to minimize the effect of confounding factors. PSM was performed by a one-to-three greedy nearest neighbor matching algorithm using a caliper width of 0.01 without replacements. Variables including gender, age, ethnicity, admission type, chronic kidney disease (CKD), hypertension, diabetes mellitus (DM), chronic obstructive pulmonary disease (COPD), heart failure, liver disease, sepsis, SAPSII score, nonrenal SOFA score, SCr value and eGFR on admission, AKI stage, use of vasopressors and inotropes, vancomycin, aminoglycoside, amphotericin, opioid agents and other sedatives, and mechanical ventilation were selected to generate the propensity score. To evaluate the efficiency of PSM in reducing the differences between the two groups, standardized mean difference (SMD) was calculated. Finally, 324 matched pairs were generated and applied to the further analysis.
Continuous variables were expressed as median [interquartile range (IQR)]. Wilcoxon rank-sum test or Kruskal-Wallis test was used to identify the differences between groups. Categorical variables were expressed as the numbers and percentage and compared using the chi-square test or Fisher’s exact test as appropriate.
Cox regression was used to evaluate the impact of DEX administration on mortality adjusting for confounding variables which were selected based on p value <0.05 in univariate analysis. Linear regression was used to evaluate the impact of DEX administration on LOS in ICU and in hospital, and the hazard ratios (HR) was calculated using the formula HR =eβi. Logistic regression was used to evaluate the impact of DEX administration on recovery of renal function.
The statistical analysis was performed using Stata 16.0 (Stata Corp., College Station, TX, USA). A two-tailed test was performed, and p value <0.05 was considered statistically significant.