Effects of Dexmedetomidine Administration on Outcomes in Critically Ill Patients with Acute Kidney Injury: A Propensity Score-Matching Analysis

Dexmedetomidine (DEX) had organ protection effects and could decrease mortality in animal models, but its association with mortality and length of stay (LOS) in ICU and hospital in critically ill patients was conicting. Whether acute kidney injury (AKI) subgroup of critically ill patients could benet from DEX was unknown. The present study aimed to evaluate the effects of DEX on clinical outcomes of critically ill patients with AKI. in-hospital and 90-day and ICU


Abstract
Background Dexmedetomidine (DEX) had organ protection effects and could decrease mortality in animal models, but its association with mortality and length of stay (LOS) in ICU and hospital in critically ill patients was con icting. Whether acute kidney injury (AKI) subgroup of critically ill patients could bene t from DEX was unknown. The present study aimed to evaluate the effects of DEX on clinical outcomes of critically ill patients with AKI.

Methods
Data were extracted from the Medical Information Mart for Intensive Care database (MIMIC ). Propensity score matching (PSM) analysis (1:3), cox proportional hazards model, linear regression and logistic regression model were used to assess the effect of DEX on clinical outcomes.
Conclusions DEX administration improved outcomes in critically ill patients with mild and moderate AKI and could be a good choice of sedation.

Background
Acute kidney injury (AKI) is characterized by abrupt decrease of kidney function, with increasing morbidity during the past decades. The morbidity and mortality of AKI in intensive care unit (ICU) patients are 50-59% and 14-51.7% respectively, of which 50% survivors suffered from irreversible declined renal function [1][2][3]. The etiologies of AKI in critically ill patients are diverse, including infection, ischemia, low cardiac output, toxins and so on [1]. Precise treatment is needed urgently, however few treatments have been proved to improve prognosis so far.
Dexmedetomidine (DEX) is a highly selective α2 receptor agonist and a widely used sedative in ICU. DEX still has the effects of antiin ammation, anti-oxidative stress, and reducing cell apoptosis. Previous studies con rmed that DEX could protect organs [4,5] and decreased mortality [6] in sepsis or ischemia-reperfusion rat models, but its association with mortality and length of stay (LOS) in ICU and hospital in critically ill patients was con icting [7][8][9][10][11]. Renal protection effect of DEX was observed both in animal models [5,12] and critically ill patients [8, 10,13].
However, whether the AKI subgroup of critically ill patients could bene t from DEX was unknown. The present study aimed to evaluate the effects of DEX on outcomes of critically ill patients with AKI, with outcomes de ned as in-hospital and 90-day mortality, length of stay (LOS) in ICU and in hospital, and recovery of renal function.

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 Page 3/15 of Technology and Beth Israel Deaconess Medical Center [14]. All patients in the database were de-identi ed 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 (certi cation 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 rst 48 hours after ICU admission. For patients who were admitted to ICU more than once, only the rst 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 De nition AKI was de ned 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 de ned 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 rst SCr after admission was used as the baseline SCr [17]. In patients with de cient or insu cient 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, simpli ed acute physiology score (SAPS ), 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 de ned 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 de ned 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 de ned as the use of vancomycin, aminoglycoside and amphotericin. The estimated glomerular ltration rate (eGFR) was calculated using the Modi cation 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 le 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 de ned 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-tothree 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 e ciency 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 =e βi . 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 signi cant.

Results
Baseline characteristics 9784 patients were included according to the inclusion and exclusion criteria, of which 326 patients were exposed to DEX for at least 6 hours during the rst 48 hours after ICU admission (Fig 1). After PSM, 324 patients who received DEX were matched with 972 patients who did not receive DEX.
The comparisons of the baseline characteristics between DEX group and non-DEX group were listed in Table 1. In unmatched cohort, there were signi cant differences in gender, age, admission types, hypertension, liver disease, sepsis, nonrenal SOFA score, SCr value and eGFR on admission, AKI stage, use of vasopressors and inotropes, vancomycin, opioid agents and other sedatives between DEX group and non-DEX group. After PSM, standardized differences for all variables were less than 10% and the propensity score distribution plot demonstrated successful propensity score matching between the two groups (Fig 2 and Fig 3). There was no signi cant difference between the two matched groups with regards to all covariates.

Subgroup analysis
As shown in Fig 4, DEX administration was associated with decreased in-hospital mortality and 90-day mortality in AKI stage 1 to 2 subgroups, but it did not decrease the mortality in AKI stage 3 subgroup. In Fig 5, DEX administration was associated with reduced LOS in ICU in AKI stage 1 subgroup, but it was not associated with reduced LOS in ICU in AKI stage 2 to 3 subgroups. DEX administration was associated with reduced LOS in hospital in AKI stage 1 to 2 subgroups, but it was not associated with reduced LOS in hospital in AKI stage 3 subgroup. DEX administration was not associated with the recovery of renal function in AKI stage 1 subgroup (HR 1.010, 95% CI 0.632-1.612, P = 0.968), AKI stage 2 subgroup (HR 1.320, 95% CI 0.675-2.582, P = 0.418) and AKI stage 3 subgroup (HR 1.762, 95% CI 0.553-5.620, P = 0.338).

Discussion
DEX inhibits the release of norepinephrine from the locus coeruleus and competitively binds to α2 receptors, which can relieve sympathetic excitement storms and anxiety, and has a mild analgesic and sedative effect. Since it does not act on the midbrain reticular ascending system and GABA receptors, patients sedated with DEX are more likely to wake up and respiratory inhibition is less common. Besides sedative effect, DEX has the effects of anti-in ammation, anti-oxidative stress, and reducing cell apoptosis [21].

Page 5/15
In sepsis or ischemia-reperfusion rat models, DEX had protective effects on heart [22], lung [4], liver [23], kidney [5] and intestine [24] by inhibiting the in ammatory responses and hence reduced the mortality of model rats [6]. However, its association with mortality and LOS in ICU and in hospital in critically ill patients was con icting [7][8][9][10][11]. Some studies indicated that DEX administration could decrease the mortality of patients [7,8] and reduce the LOS in ICU [9,10] and in hospital [11]. However, other studies showed that DEX had nothing to do with mortality [25][26][27] and LOS in hospital [11,25,26]. The effect of DEX might be related to age. One study showed that early application of DEX failed to decrease the 90-day mortality in mechanically ventilated patients [28], and then strati ed analysis showed that DEX decreased the 90-day mortality in elderly patients, but increased the 90-day mortality rate in young patients [29]. In this retrospective cohort study of critically ill patients with AKI, we selected two matched groups receiving and not receiving DEX for at least 6 hours in the rst 48 hours after ICU admission. Our results demonstrated DEX administration was associated signi cantly with decreased in-hospital and 90-day mortality and reduced LOS in ICU and in hospital in critically ill patients with AKI. The results were robust after adjusting for confounding variables.
Subgroup analysis showed DEX administration was associated with decreased in-hospital and 90-day mortality, reduced LOS in hospital in patients with AKI stage 1 to 2, but the above associations did not exist in patients with AKI stage 3. DEX administration was associated with reduced LOS in ICU in patients with AKI stage 1, but the above association did not exist in patients with AKI stage 2 to 3. The results seemed to indicate that patients with severe renal impairment could not bene t from DEX. The underlying reason was unknown. DEX is mainly metabolized by the liver and excreted by the kidney. Patients with severely impaired renal function cannot effectively and timely eliminate the metabolites of DEX. One previous study showed that the effective sedation time of DEX in patients with severe impaired renal function (Ccr < 30ml/min) was longer than that in the control group [21]. Therefore, the excretion disorder of DEX in patients with AKI stage 3 might offset its bene t.
Renal protection effect of DEX was con rmed in animal models [12,30]. In patients after heart and lung surgeries [7,13] and in patients with severe sepsis [10], the renal protective effect of DEX was also observed. However, DEX administration was not associated with the recovery of renal function in critically ill patients with AKI in this study. Recovery of renal function was de ned as being discharged from ICU with serum creatinine below 1.5 times the baseline value and normal urine output (> 0.5 ml/kg/h), which was a short-term renal outcome indicator. We did not observe the long-term effect of DEX administration on kidney in critically ill patients with AKI. Moreover, the dose of DEX was not considered, which might impact the effect.
This study was the rst to evaluate the effect of DEX administration on clinical outcomes in critically ill patients with AKI. But there are still a few limitations in this study. First, although we have incorporated many confounding factors and made a PSM to reduce bias, there were still some possible confounding factors not incorporated because of the missing data such as body mass index (BMI).
Second, heterogeneity of treatment strategies might affect the outcomes of patients. Finally, this study was a single-centered retrospective study, and multicenter randomized controlled trials were needed to con rm the conclusions.

Conclusions
DEX administration decreased mortality and reduced the length of ICU stay and hospital stay in critically ill patients with mild and moderate AKI, but the bene cial effect did not exist in critically ill patients with severe AKI. DEX administration could not improve shortterm renal outcome, while its effect on long-term renal outcome needs further investigation. In summary, DEX administration could improve clinical outcomes in critically ill patients with AKI, and it could be a good choice of sedation.

Declarations
Ethics approval and consent to participate The establishment of this database was approved by the Massachusetts Institute of Technology (Cambridge, MA) and Beth Israel Deaconess Medical Center (Boston, MA). Consent was obtained for data extraction. Therefore, ethics approval and informed consent were both waived for this study.

Consent for publication
Not applicable.

Availability of data and materials
The datasets presented in this study are available in the MIMIC database (https://physionet.org/content/mimiciii/1.4/).

Competing interests
All the authors declare that they have no competing interests.  Data are presented as n (%), or median (interquartile range). DEX, dexmedetomidine; CKD, chronic kidney diseases; COPD, chronic obstructive pulmonary disease; ICU, intensive care unit; SAPS , simpli ed acute physiology score ; SOFA, sequential organ failure assessment; eGFR, estimated glomerular ltration rate a eGFR was calculated using MDRD formula c Recovery of renal function was de ned as being discharged from ICU with serum creatinine below 1.5 times the baseline value and normal urine output (> 0.5 ml/kg/h). Impact of DEX administration on the recovery of renal function was evaluated using logistic regression model. Standardized bias before and after PSM. PSM: propensity-score matching  Mortality by stage of AKI between Non-DEX group and DEX group after PSM. DEX: dexmedetomidine PSM: propensity-score matching