Advertisement
American Journal of Kidney Diseases

Characteristics of and Outcomes After Dialysis-Treated Acute Kidney Injury 2009-2018: A Taiwanese Multicenter Study

Open AccessPublished:October 14, 2022DOI:https://doi.org/10.1053/j.ajkd.2022.08.022

      Abstract

      Rationale and objective

      Dialysis-requiring acute kidney injury (AKI) is increasingly common in intensive care units (ICUs) and is associated with poor outcomes. Few studies have explored the temporal trends in severity of acute illness at dialysis initiation, indications for dialysis, and their association with patient outcomes.

      Study Design

      Multicenter retrospective cohort study.

      Setting and participants

      9,535 adult patients admitted to the ICU and received their first dialysis treatment from Chang Gung Memorial Hospital system in Taiwan between 2009 and 2018.

      Exposure

      Calendar year.

      Outcomes

      ICU mortality and dialysis dependence among hospital survivors.

      Analytical approach

      The temporal trends over the study period were investigated using test statistics suited for continuous or categorical data. The association between the study year and the risk of mortality was analyzed using multivariable Cox regression with adjustment for relevant clinical variables, including the severity of acute illness, defined by the Sequential Organ Failure Assessment (SOFA) score.

      Main Results

      The mean SOFA score at dialysis initiation declined slightly from 14.0 in 2009 to 13.6 in 2018. There was no significant trend in the number of dialysis criteria that were fulfilled over time. Observed ICU mortality declined over time and appeared to be reverse J-shaped, with a substantial decline from 56.1% in 2009 to 46.3% in 2015 and a slight increase afterward. The risk of mortality was significantly reduced from 2013 to 2018 compared with 2009 in adjusted models. The declining trend in ICU mortality over time remained significant. Dialysis dependence among survivors increased, mainly in patients with CKD, from 36.8% in 2009 to 43.9% in 2018.

      Limitation

      Residual confounding from unmeasured factors over time such as severity of comorbidities, detailed medication interventions, and delivered dialysis dose.

      Conclusion

      We observed reductions in mortality among ICU patients with dialysis-requiring AKI between 2009 and 2018, even after adjusting for dialysis indication and severity of illness at dialysis initiation. However, dialysis dependence among survivors has increased over time, mainly in patients with pre-existing CKD.

      Graphical abstract

      Index words

      Introduction

      Acute kidney injury (AKI) is a common complication in intensive care units (ICUs) and significantly affects morbidity and mortality.
      • Hoste E.A.
      • Bagshaw S.M.
      • Bellomo R.
      • et al.
      Epidemiology of acute kidney injury in critically ill patients: the multinational AKI-EPI study.
      • Uchino S.
      • Kellum J.A.
      • Bellomo R.
      • et al.
      Acute renal failure in critically ill patients: a multinational, multicenter study.
      • Wu V.C.
      • Huang T.M.
      • Lai C.F.
      • et al.
      Acute-on-chronic kidney injury at hospital discharge is associated with long-term dialysis and mortality.
      • Gallagher M.
      • Cass A.
      • Bellomo R.
      • et al.
      Long-term survival and dialysis dependency following acute kidney injury in intensive care: extended follow-up of a randomized controlled trial.
      For example, the mortality rate in critically ill patients with severe dialysis-requiring AKI reported is 50–70%.
      • Uchino S.
      • Kellum J.A.
      • Bellomo R.
      • et al.
      Acute renal failure in critically ill patients: a multinational, multicenter study.
      ,
      • Gallagher M.
      • Cass A.
      • Bellomo R.
      • et al.
      Long-term survival and dialysis dependency following acute kidney injury in intensive care: extended follow-up of a randomized controlled trial.
      ,
      • Elseviers M.M.
      • Lins R.L.
      • Van der Niepen P.
      • et al.
      Renal replacement therapy is an independent risk factor for mortality in critically ill patients with acute kidney injury.
      Moreover, accumulating evidence shows that the incidence of dialysis-requiring AKI among critically ill patients is increasing, mainly due to the rising burden of the aging population and chronic comorbidities.
      • Lenihan C.R.
      • Montez-Rath M.E.
      • Mora Mangano C.T.
      • Chertow G.M.
      • Winkelmayer W.C.
      Trends in acute kidney injury, associated use of dialysis, and mortality after cardiac surgery, 1999 to 2008.
      The current medical management of severe AKI is primarily limited to supportive care and renal replacement therapy (RRT) if indicated, leading to perceptions that outcomes among patients with dialysis-requiring AKI have not improved. However, some studies have reported temporal reductions in mortality among hospitalized patients with dialysis-requiring AKI.
      • Brown J.R.
      • Rezaee M.E.
      • Hisey W.M.
      • Cox K.C.
      • Matheny M.E.
      • Sarnak M.J.
      Reduced Mortality Associated with Acute Kidney Injury Requiring Dialysis in the United States.
      ,
      • Harding J.L.
      • Li Y.
      • Burrows N.R.
      • Bullard K.M.
      • Pavkov M.E.
      US Trends in Hospitalizations for Dialysis-Requiring Acute Kidney Injury in People With Versus Without Diabetes.
      Recently published analyses of two studies demonstrated improved survival among critically ill ICU patients over time.
      • Wald R.
      • McArthur E.
      • Adhikari N.K.
      • et al.
      Changing incidence and outcomes following dialysis-requiring acute kidney injury among critically ill adults: a population-based cohort study.
      ,

      Miyamoto Y, Iwagami M, Aso S, et al. Temporal change in characteristics and outcomes of acute kidney injury on renal replacement therapy in intensive care units: analysis of a nationwide administrative database in Japan, 2007-2016. Crit Care. 2019;23(1):172. doi:10.1186/s13054-019-2468-8

      Although these two important population-based studies have endeavored to clarify the contribution of baseline characteristics, RRT modality, or varied admission diagnosis of ICU patients to improved outcomes over time, the lack of patients severity based on physiologic or laboratory measurements into consideration makes their findings somewhat inconclusive. Thus, it remains unclear whether the findings of mortality decline in critically ill patients with dialysis-requiring AKI over time are related to or independent of the severity or the timing of dialysis indication of enrolled patients.
      Therefore, this multicenter study aimed to examine the recent trends in the characteristics and outcomes of critically ill patients developing dialysis-requiring AKI from 2009 to 2018. Furthermore, we aimed to clarify the contribution of changing comorbidities, the severity of acute illness on ICU admission day and RRT initiation day, and dialysis indication on the changes in patient mortality over time.

      Methods

      Data source

      This retrospective cohort study was performed using the Chang Gung Research Database (CGRD). The CGRD contains anonymized multicenter electronic medical records collected prospectively from six branches of Chang Gung Memorial Hospital (CGMH) (two tertiary medical centers and four district hospitals located from northern to southern Taiwan), the largest medical care system representing 10.2% of annual hospitalizations in Taiwan.
      • Shao S.C.
      • Chan Y.Y.
      • Kao Yang Y.H.
      • et al.
      The Chang Gung Research Database-A multi-institutional electronic medical records database for real-world epidemiological studies in Taiwan.
      ,
      • Chan C.C.
      • Lee K.T.
      • Ho W.J.
      • Chan Y.H.
      • Chu P.H.
      Levosimendan use in patients with acute heart failure and reduced ejection fraction with or without severe renal dysfunction in critical cardiac care units: a multi-institution database study.
      The CGRD can offer researchers comprehensive medical records, including patient’s date of birth, sex, inpatient orders, date and dosage of prescriptions, procedure charge code, vital signs, laboratory data, and examination reports. Comorbidities were identified based on the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnostic code. This study was approved by the institutional review board of CGMH (IRB 201900835B0), and the need for individual informed consent was waived because of the lack of identifiable personal information.

      Study population

      We used the data from CGRD to identify adult patients admitted to an ICU and received their first RRT, either acute intermittent hemodialysis or continuous RRT (CRRT), while in the ICU from 2009 to 2018. The first episode of ICU admission receiving RRT for one patient during the study period was defined as the index hospitalization. Time zero was defined as the day initiating RRT during their ICU stay. The process for the inclusion and exclusion of the study cohort is shown in Figure 1. We excluded patients who had received renal transplantation or chronic dialysis before hospitalization. Patients who had received RRT before the ICU admission during the index hospitalization were also excluded. Of the 17,336 patients screened, 9,535 were included in our analysis.
      Figure thumbnail gr1
      Figure 1The flowchart for illustrating the inclusion and exclusion of the study patients.

      Exposure and Covariate Ascertainment

      The primary exposure of interest was the index year. Eligible patients were assigned to each calendar year group based on when they received their first RRT during their ICU stay. No patients were counted twice if their hospitalization period were across the 2 calendar years.
      The covariates we examined were age, sex, body mass index (BMI), nine pre-existing comorbidities, Charlson comorbidity index (CCI), baseline estimated glomerular filtration rates (eGFRs), initial diagnosis, source of initial ICU admission, mechanical ventilation, mechanical circulatory support (MCS), including intra-aortic balloon pump (IABP), extracorporeal membrane oxygenation (ECMO), and the vasopressor type and dose. Comorbidities included diabetes, hypertension, coronary artery disease, heart failure hospitalization, stroke, chronic obstructive pulmonary disease, liver cirrhosis, malignancy, and chronic kidney disease (CKD). Heart failure hospitalization and stroke were defined from hospitalization records prior to the index hospitalization. Except for CKD, other comorbidities were defined as having at least two outpatient visits or one inpatient stay before the index hospitalization. Baseline eGFRs were based on serum creatinine (Cr) levels measured 7–365 days before admission and calculated using the Chronic Kidney Disease Epidemiology Collaboration creatinine equation.
      • Levey A.S.
      • Stevens L.A.
      • Schmid C.H.
      • et al.
      A new equation to estimate glomerular filtration rate.
      The lowest serum Cr level was used if multiple measurements were available. We defined CKD as a baseline eGFR <60 mL/min/1.73 m2. We examined the severity of acute illness using the total sequential organ failure assessment (SOFA) score (on ICU admission day and RRT initiation day) and lactate concentrations. The SOFA score was calculated as a composite point across six organ systems (0–4 each organ), based on the ratio of partial arterial pressure of oxygen to fraction of inspired oxygen, mean arterial pressure and vasopressor dose, platelet count, Glasgow Coma Scale score, bilirubin, and Cr level. The SOFA score ranges from 0 to 24 points, with higher scores indicating more severe organ dysfunction. Laboratory values (blood urea nitrogen (BUN), Cr, potassium (K), arterial blood pH, and bicarbonate concentrations) were extracted from the database on the RRT initiation date to examine the number of criteria fulfilled. In addition, dates were obtained to calculate the duration from ICU admission to RRT initiation.

      Outcomes

      The primary outcomes were patient mortality in ICU, in-hospital, and 90-day mortality. All three primary outcomes started the follow up duration from the day of initiation of RRT during the ICU stay. For mortality in ICU, the follow up duration ended the day of death during ICU, or discharge from ICU, whichever came first. For in-hospital mortality, the follow up duration ended the day of death during hospitalization, or discharge from hospital, whichever came first. For 90-day mortality, the follow up duration ended the day of death, or 90 days after the day of initiation of RRT, whichever came first. Other outcome measures included length of stay in ICU and hospital and dialysis dependence among hospital survivors.

      Statistical Analysis

      The demographics results were summarized using descriptive statistics and presented by each index year. Continuous data were expressed as the mean ± standard deviation or median with interquartile ranges for the obviously skewed data. Categorical data were expressed as numbers with percentages (%) of each item. The temporal trends over the study period for continuous variables (i.e., age, CCI score), skewed continuous variables (i.e., lactate level, length of stay) and categorical variables with 2 categories (i.e., sex, mortality and the dummy variables for main cause of admission), categorical variables with more than 2 ordinal categories (number of fulfilled indications and CCI score grouping) were investigated using the linear contrast in the analysis of variance, Jonckheere-Terpstra test, Cochran–Armitage test, and Mantel-Haenszel test, respectively. The association between the study year (using year 2009 as the reference category) and the risks of mortality was analyzed using multivariable Cox proportional hazard model with adjustment of branch, age, sex, BMI, CCI score, baseline eGFR, main admission diagnosis, duration from ICU admission to RRT initiation, modality of RRT, the status of MCS and any vasopressor use, and SOFA scores at RRT initiation. The temporal trend across years on the mortality rate was also assessed using the linear contrast in the multivariable Cox model. A two-sided P value of <0.05 was considered statistically significant. Data analyses were conducted using SAS software version 9.4 (SAS Institute, Cary, NC).

      Results

      The trend in the baseline characteristics of ICU patients with dialysis-requiring AKI
      Table 1 shows the baseline characteristics of the study population by year. During the 10-year study period, 9,535 patients experienced dialysis-requiring AKI in the ICU. The mean age was 66.5 ± 15.4 years, and 62.4% of patients were males. The most common comorbidities were diabetes (42.2%), hypertension (57.0%), and CKD (63.3%). There were several significant changes in the baseline characteristics over the study year. Compared with patients in the early years, patients in more recent years had a higher BMI and a higher prevalence of comorbidities such as diabetes (40.6% in 2009 to 47.2% in 2018, annual change 0.81%, P <0.001), hypertension, and CKD (Figure S1). Further, the CCI score increased significantly over this period (P for trend = 0.003), with the proportion of CCI score >8 increasing from 20.0% in 2009 to 25.5% in 2018 (Figure S2).
      Table 1Baseline characteristics of the study population by year
      Variable2009201020112012201320142015201620172018P trendaΔ/yb
      Number of patients7798989979699789681022946990988-
      Age, years65.7 ± 16.067.1 ± 15.565.7 ± 16.267.7 ± 14.967.0 ± 15.466.7 ± 15.465.6 ± 15.365.8 ± 15.267.7 ± 14.866.2 ± 14.80.9530.003
      Male475 (61.0)561 (62.5)640 (64.2)589 (60.8)597 (61.0)626 (64.7)649 (63.5)601 (63.5)606 (61.2)604 (61.1)0.943-0.01
      Body mass index, kg/m225.1 ± 4.925.1 ± 5.324.6 ± 5.124.9 ± 5.125.4 ± 5.125.3 ± 5.225.9 ± 5.325.7 ± 5.125.5 ± 5.426.0 ± 5.5<0.0010.12
      Comorbidity
      Diabetes316 (40.6)340 (37.9)388 (38.9)397 (41.0)421 (43.0)406 (41.9)444 (43.4)409 (43.2)436 (44.0)466 (47.2)<0.0010.81
      Hypertension414 (53.1)471 (52.4)536 (53.8)545 (56.2)571 (58.4)569 (58.8)587 (57.4)533 (56.3)606 (61.2)605 (61.2)<0.0010.92
      Coronary artery disease202 (25.9)258 (28.7)267 (26.8)301 (31.1)320 (32.7)296 (30.6)296 (29.0)236 (24.9)238 (24.0)272 (27.5)0.120-0.25
      Heart failure hospitalization181 (23.2)210 (23.4)232 (23.3)237 (24.5)279 (28.5)239 (24.7)231 (22.6)213 (22.5)242 (24.4)240 (24.3)0.9410.01
      Stroke106 (13.6)145 (16.1)136 (13.6)137 (14.1)132 (13.5)116 (12.0)140 (13.7)123 (13.0)119 (12.0)128 (13.0)0.042-0.25
      COPD137 (17.6)154 (17.1)150 (15.0)180 (18.6)164 (16.8)171 (17.7)149 (14.6)154 (16.3)136 (13.7)152 (15.4)0.027-0.30
      Liver cirrhosis157 (20.2)172 (19.2)175 (17.6)150 (15.5)152 (15.5)156 (16.1)156 (15.3)139 (14.7)134 (13.5)127 (12.9)<0.001-0.72
      Malignancy283 (36.3)295 (32.9)309 (31.0)332 (34.3)316 (32.3)353 (36.5)383 (37.5)318 (33.6)349 (35.3)379 (38.4)0.0150.42
      CKDc471 (60.5)545 (60.7)617 (61.9)589 (60.8)628 (64.2)610 (63.0)648 (63.4)628 (66.4)666 (67.3)637 (64.5)0.0040.67
      eGFR, ml/min/1.73m
      • Uchino S.
      • Kellum J.A.
      • Bellomo R.
      • et al.
      Acute renal failure in critically ill patients: a multinational, multicenter study.
      52.8 ± 39.153.9 ± 38.751.0 ± 38.952.1 ± 38.749.3 ± 37.249.2 ± 38.049.6 ± 37.047.3 ± 37.446.1 ± 36.748.1 ± 36.5<0.001-0.74
      CCI score4.9 ± 3.44.6 ± 3.34.6 ± 3.44.6 ± 3.25.0 ± 3.54.8 ± 3.44.8 ± 3.54.7 ± 3.54.9 ± 3.55.2 ± 3.70.0030.04
      CCI score grouping0.017
      0-1133 (17.1)161 (17.9)196 (19.7)183 (18.9)160 (16.4)166 (17.1)189 (18.5)187 (19.8)177 (17.9)172 (17.4)0.003
      2-3153 (19.6)217 (24.2)237 (23.8)209 (21.6)211 (21.6)194 (20.0)228 (22.3)211 (22.3)220 (22.2)191 (19.3)-0.17
      4-5191 (24.5)207 (23.1)231 (23.2)220 (22.7)227 (23.2)254 (26.2)209 (20.5)217 (22.9)219 (22.1)190 (19.2)-0.36
      6-7146 (18.7)137 (15.3)126 (12.6)173 (17.9)168 (17.2)157 (16.2)167 (16.3)126 (13.3)150 (15.2)183 (18.5)0.004
      ≥8156 (20.0)176 (19.6)207 (20.8)184 (19.0)212 (21.7)197 (20.4)229 (22.4)205 (21.7)224 (22.6)252 (25.5)0.52
      Main cause of admission
      Coronary artery disease42 (5.4)50 (5.6)55 (5.5)75 (7.7)80 (8.2)69 (7.1)71 (6.9)72 (7.6)69 (7.0)81 (8.2)0.0070.25
      Infection183 (23.5)211 (23.5)255 (25.6)248 (25.6)255 (26.1)260 (26.9)278 (27.2)249 (26.3)271 (27.4)217 (22.0)0.4650.12
      Neurologic disorders24 (3.1)32 (3.6)37 (3.7)31 (3.2)22 (2.2)31 (3.2)34 (3.3)30 (3.2)35 (3.5)26 (2.6)0.571-0.04
      Cardiovascular surgeries44 (5.6)58 (6.5)74 (7.4)61 (6.3)57 (5.8)54 (5.6)55 (5.4)52 (5.5)64 (6.5)62 (6.3)0.572-0.05
      Other non-CVS surgeries127 (16.3)139 (15.5)147 (14.7)163 (16.8)140 (14.3)149 (15.4)155 (15.2)156 (16.5)143 (14.4)164 (16.6)0.8850.01
      Trauma and burn36 (4.6)29 (3.2)47 (4.7)39 (4.0)49 (5.0)29 (3.0)36 (3.5)36 (3.8)42 (4.2)48 (4.9)0.9410.01
      Others323 (41.5)379 (42.2)382 (38.3)352 (36.3)375 (38.3)376 (38.8)393 (38.5)351 (37.1)366 (37.0)390 (39.5)0.09-0.30
      Abbreviations: Δ/y, annual change; CKD, chronic kidney disease; eGFR, estimated Glomerular filtration rate; CCI, Charlson’s Comorbidity Index; CVS, cardiovascular surgery.
      Data were presented as mean ± standard deviation or frequency (percentage).
      aThe linear contrast in the analysis of variance for continuous variable, Cochran–Armitage test for categorical variables with 2 categories, and Mantel-Haenszel test on the variable with more than 2 ordinal categories (CCI score grouping), respectively.
      bthe slope estimate of a simple linear regression of the proportion (or the mean) of each category per 1 calendar year.
      cdefined as eGFR <60 ml/min/1.73 m2.
      ECMO use increased over the study period (3.6% in 2009 to 8.0% in 2018), while the mechanical ventilation and vasopressors both decreased (Table 2). Furthermore, substantial shifts in vasopressor use during the decade wherein recent years have seen a significant decline in the use of dopamine (53.3% in 2009 to 11.4% in 2018, annual change -4.29%, P <0.001) but a notable increase in norepinephrine (42.1% in 2009 to 55.8% in 2018, annual change 1.59%, P <0.001).
      Table 2Treatment, severity of acute illness and laboratory data of the study population by year
      Variable2009201020112012201320142015201620172018P trendaΔ/yb
      Treatment during ICU
      Mechanical ventilation678 (87.0)794 (88.4)838 (84.1)796 (82.1)821 (83.9)780 (80.6)840 (82.2)780 (82.5)793 (80.1)815 (82.5)<0.001-0.65
      ECMO28 (3.6)45 (5.0)56 (5.6)60 (6.2)55 (5.6)55 (5.7)53 (5.2)74 (7.8)69 (7.0)79 (8.0)<0.0010.36
      IABP12 (1.5)15 (1.7)30 (3.0)33 (3.4)39 (4.0)30 (3.1)28 (2.7)32 (3.4)24 (2.4)30 (3.0)0.1230.09
      Vasopressor at RRT
      Any vasopressor535 (68.7)585 (65.1)639 (64.1)615 (63.5)634 (64.8)621 (64.2)658 (64.4)607 (64.2)607 (61.3)613 (62.0)0.005-0.49
      Dopamine415 (53.3)407 (45.3)425 (42.6)366 (37.8)329 (33.6)288 (29.8)247 (24.2)221 (23.4)177 (17.9)113 (11.4)<0.001-4.29
      Norepinephrine328 (42.1)402 (44.8)446 (44.7)451 (46.5)509 (52.0)528 (54.5)560 (54.8)529 (55.9)527 (53.2)551 (55.8)<0.0011.59
      Epinephrine234 (30.0)246 (27.4)251 (25.2)223 (23.0)218 (22.3)219 (22.6)231 (22.6)210 (22.2)207 (20.9)217 (22.0)<0.001-0.78
      Dobutamine58 (7.4)69 (7.7)73 (7.3)75 (7.7)69 (7.1)64 (6.6)69 (6.8)66 (7.0)60 (6.1)70 (7.1)0.193-0.12
      Duration between ICU admission to RRT initiation, day3 [2, 9]4 [2, 9]3 [2, 7]3 [2, 8]3 [2, 6]3 [2, 7]3 [2, 7]3 [2, 8]3 [2, 7]3 [2, 7]0.078-0.13
      CRRT as initial RRT modality209 (26.8)219 (24.4)250 (25.1)203 (20.9)275 (28.1)284 (29.3)295 (28.9)274 (29.0)292 (29.5)298 (30.2)<0.0010.70
      SOFA score
      at ICU admission12.4 ± 3.312.0 ± 3.412.0 ± 3.411.7 ± 3.412.2 ± 3.312.2 ± 3.212.3 ± 3.312.2 ± 3.112.1 ± 3.112.1 ± 3.10.6790.01
      at dialysis date14.0 ± 3.413.8 ± 3.213.7 ± 3.213.5 ± 3.313.7 ± 3.213.6 ± 3.113.6 ± 3.313.8 ± 3.213.4 ± 3.113.6 ± 3.20.027-0.03
      Dialysis indication
      BUN >100 mg/dL253 (32.5)274 (30.5)292 (29.3)287 (29.6)296 (30.3)306 (31.6)327 (32.0)326 (34.5)334 (33.7)333 (33.7)0.0090.44
      Creatinine >4 mg/dL455 (58.4)476 (53.0)525 (52.7)513 (52.9)540 (55.2)547 (56.5)591 (57.8)584 (61.7)593 (59.9)590 (59.7)<0.0010.79
      Hyperkalemia (K ≥6.5)42 (5.4)52 (5.8)79 (7.9)75 (7.7)70 (7.2)100 (10.3)77 (7.5)66 (7.0)93 (9.4)56 (5.7)0.1590.13
      Severe acidemia (pH ≤7.2)270 (34.7)287 (32.0)356 (35.7)308 (31.8)241 (24.6)239 (24.7)224 (21.9)178 (18.8)201 (20.3)190 (19.2)<0.001-2.04
      Acute pulmonary edema40 (5.1)30 (3.3)45 (4.5)33 (3.4)38 (3.9)28 (2.9)46 (4.5)33 (3.5)48 (4.8)40 (4.0)0.988-0.001
      Number of dialysis indication fulfilled0.059
      1307 (39.4)341 (38.0)337 (33.8)352 (36.3)378 (38.7)347 (35.8)401 (39.2)368 (38.9)370 (37.4)358 (36.2)0.02
      2216 (27.7)235 (26.2)291 (29.2)269 (27.8)266 (27.2)287 (29.6)288 (28.2)301 (31.8)298 (30.1)299 (30.3)0.41
      ≥3102 (13.1)100 (11.1)121 (12.1)101 (10.4)89 (9.1)94 (9.7)93 (9.1)70 (7.4)94 (9.5)81 (8.2)-0.49
      Laboratory data
      BUN, mg/dL64.2 ± 46.059.7 ± 41.561.4 ± 42.661.4 ± 42.764.3 ± 42.466.9 ± 46.862.3 ± 43.765.7 ± 44.266.3 ± 42.264.4 ± 43.40.0040.66
      Creatinine, mg/dL4.0 ± 3.03.8 ± 3.04.0 ± 3.24.0 ± 3.44.2 ± 3.14.3 ± 3.34.2 ± 3.34.5 ± 3.74.4 ± 3.44.3 ± 3.5<0.0010.07
      Lactate, mg/dL55.4 [22.3,114.3]48.4 [20.5,102.0]45.5 [19.5,118.9]51.3 [20.4,109.7]39.6 [17.8,99.9]43.9 [19.8,103.0]35.6 [17.3, 94.7]35.5 [14.5, 84.6]33.9 [15.2, 91.4]42.2 [17.1, 99.5]<0.001-1.64
      Peak CRP within 7 days, mg/dL12.1 [4.8, 21.1]12.2 [5.0, 21.6]12.6 [5.4, 21.6]13.0 [6.0, 21.6]12.4 [5.8, 20.7]13.0 [5.9, 22.4]12.5 [5.7, 21.4]12.5 [6.1, 22.8]13.0 [6.1, 21.2]12.9 [5.7, 23.4]0.0430.61
      Abbreviations: Δ/y,annual change; ICU, intensive care unit; ECMO, extracorporeal membrane oxygenation; IABP, intra-aortic balloon pump; RRT, renal replacement therapy; IHD, intermittent hemodialysis; CRRT, continuous renal replacement therapy; SOFA, Sequential Organ Failure Assessment; BUN, blood urea nitrogen; CRP, C-Reactive protein.
      Data were presented as mean ± standard deviation, median [interquartile range] or frequency (percentage).
      aThe linear contrast in the analysis of variance for continuous variables, Jonckheere-Terpstra test for skewed continuous variables, Cochran–Armitage test for categorical variables with 2 categories, and Mantel-Haenszel test on the variable with more than 2 ordinal categories (number of dialysis indication fulfilled), respectively.
      bthe slope estimate of a simple linear regression of the proportion (or the mean) of each category per 1 calendar year.

      The trend in the severity of acute illness and dialysis indications

      The severity of acute illness at ICU admission and at the time of RRT and laboratory data, including lactate and C-reactive protein concentrations, are summarized by year in Table 2. The median time between ICU admission and RRT initiation was consistently 3 days. The proportion of patients using CRRT as the initial modality continued to rise during the study period from 24.4% in 2010 to 30.2% in 2018. The severity of acute illness at ICU admission assessed by total SOFA scores did not show a significant change over the study year (approximately 12.2, P = 0.679). However, SOFA scores at RRT initiation declined slightly from 14.0 in 2009 to 13.6 in 2018 (annual change -0.03, P = 0.027). The median lactate level, another marker of illness severity, significantly decreased over time (55.4 mg/dl in 2009 to 42.2 mg/dl in 2018, annual change -1.64 mg/dl, P <0.001, Figure S3).
      Table 2 also provides the dialysis indications of the patients at the time of RRT during the 10-year study period. There were modest but significantly increased levels of BUN and Cr at the time of RRT over this period (annual change 0.66 mg/dl and 0.07 mg/dl, P =0.004 and <0.001, respectively). The proportion of severe hyperkalemia (K>6.5 mmol/L) and acute pulmonary edema did not significantly change over time. Only the proportion of severe acidemia (pH <7.2) significantly declined from 34.7% in 2009 to 19.2% in 2018 (annual change -2.04%, P <0.001). However, there was no significant trend in the number of dialysis criteria fulfilled over time (P = 0.059).

      Changes in patient mortality and dialysis dependence

      As shown in Table 3, the median ICU and hospital stay lengths were similar over time at approximately 13 and 27 days, respectively. The observed ICU mortality declined over time (annual change -0.81%, P for linear trend <0.001). However, it appeared to be reverse J-shaped (P for quadratic term = 0.001), with a substantial decline from 56.1% in 2009 to 46.3% in 2015 and a slight increase in the following 3 years. A similar trend was observed for in-hospital mortality and 90-day mortality. After adjusting for baseline characteristics, interventions, and SOFA scores at RRT initiation, the risk of mortalities was significantly reduced from 2013 to 2018 compared with 2009. The declining trend of ICU mortality over time remained significant (P for linear trend <0.001, Table 4), with the most significant reduction in 2015 (adjusted hazard ratio, 0.81; 95% confidence interval, 0.70–0.92). However, the decline did not persist in the recent 3 years (2016–2018). Similar declining trends of in-hospital mortality and 90-day mortality were observed in the adjusted models (P for linear trend = 0.001 and 0.004, respectively).
      Table 3Length of ICU, hospital stay and mortality data of the study population by year
      Variable2009201020112012201320142015201620172018P trendaΔ/yb
      Length of stay, day
      ICU13 [6, 26]13 [6, 28]12 [5, 26]13 [6, 27]12 [6, 25]12 [6, 26]13 [6, 26]13 [6, 29]12 [6, 26]13 [6, 28]0.166-0.06
      Hospital24 [13, 50]27 [13, 52]28 [13, 55]28 [13, 53]26 [13, 53]27 [13, 50]27 [14, 52]28 [13, 52]26 [14, 48]26 [13, 50]0.993-0.39
      Mortality
      ICU mortality437 (56.1)504 (56.1)514 (51.6)479 (49.4)475 (48.6)472 (48.8)473 (46.3)473 (50.0)460 (46.5)495 (50.1)<0.001-0.81
      In-hospital mortality489 (62.8)579 (64.5)611 (61.3)563 (58.1)553 (56.5)545 (56.3)555 (54.3)546 (57.7)559 (56.5)575 (58.2)<0.001-0.75
      Day 90 mortality515 (66.1)613 (68.3)640 (64.2)593 (61.2)581 (59.4)594 (61.4)608 (59.5)591 (62.5)599 (60.5)610 (61.7)<0.001-0.62
      Dialysis dependence among hospital survivors
      Total90 (31.0)111 (34.8)102 (26.4)152 (37.4)127 (29.9)160 (37.8)167 (35.8)137 (34.3)154 (35.7)155 (37.5)0.0140.67
      Non-CKD16 (18.0)19 (20.9)16 (15.8)22 (19.5)14 (11.7)22 (18.2)27 (20.0)23 (24.2)24 (21.2)15 (16.0)0.5610.25
      CKD74 (36.8)92 (40.4)86 (30.2)130 (44.4)113 (37.0)138 (45.7)140 (42.2)114 (37.4)130 (40.9)140 (43.9)0.0450.66
      Abbreviations: Δ/y, annual change; ICU, intensive care unit; CKD, chronic kidney disease.
      Data were presented as median [interquartile range] or frequency (percentage).
      aJonckheere-Terpstra test for skewed continuous variables and Cochran–Armitage test for categorical variables, respectively.
      bthe slope estimate of a simple linear regression of the proportion (or the mean) of each category per 1 calendar year.
      Table 4Mortality data across the study period (using year 2009 as the reference) with adjustment of covariates*
      Mortality / Statistics2009201020112012201320142015201620172018P trend
      ICU
      Crude HR (95% CI)Reference1.002 (0.88–1.14)0.94 (0.83–1.07)0.87 (0.77–0.99)0.84 (0.73–0.95)0.86 (0.76–0.98)0.81 (0.71–0.92)0.82 (0.72–0.94)0.82 (0.72–0.93)0.85 (0.74–0.96)<0.001
      Adjusted HR (95% CI)Reference1.10 (0.96–1.25)0.997 (0.87–1.14)0.94 (0.82–1.08)0.84 (0.74–0.96)0.85 (0.74–0.97)0.81 (0.70–0.92)0.82 (0.71–0.94)0.84 (0.73–0.96)0.82 (0.72–0.94)<0.001
      In-hospital
      Crude HR (95% CI)Reference1.000 (0.89–1.13)0.91 (0.80–1.02)0.85 (0.75–0.96)0.82 (0.72–0.92)0.83 (0.74–0.94)0.80 (0.71–0.91)0.85 (0.76–0.96)0.85 (0.75–0.95)0.87 (0.77–0.98)<0.001
      Adjusted HR (95% CI)Reference1.07 (0.94–1.21)0.94 (0.83–1.07)0.92 (0.81–1.04)0.81 (0.71–0.92)0.83 (0.73–0.95)0.79 (0.70–0.90)0.86 (0.75–0.97)0.86 (0.76–0.98)0.86 (0.76–0.98)0.001
      90-day
      Crude HR (95% CI)Reference1.01 (0.89–1.13)0.89 (0.79–1.002)0.82 (0.73–0.93)0.79 (0.70–0.88)0.84 (0.74–0.94)0.78 (0.69–0.87)0.84 (0.75–0.95)0.79 (0.71–0.89)0.82 (0.73–0.92)
      Adjusted HR (95% CI)Reference1.07 (0.95–1.20)0.94 (0.83–1.06)0.89 (0.79–1.01)0.79 (0.70–0.89)0.84 (0.75–0.96)0.79 (0.70–0.89)0.86 (0.76–0.97)0.83 (0.74–0.94)0.84 (0.75–0.95)0.004
      Abbreviations: ICU, intensive care unit; HR, hazard ratio; CI, confidence interval.
      * Adjusted for branch, age, sex, BMI, main diagnosis category, CCI’s total score, eGFR, duration from ICU admission to renal replacement therapy initiation, renal replacement therapy modality (IHD or CRRT), SOFA scores at RRT, IABP, ECMO and use of any vasopressor.
      There was an increasing trend in the proportion of dialysis dependence among survivors at hospital discharge, from 31% in 2009 to 37.5% in 2018 (annual change 0.67%, P = 0.014). However, when stratified by baseline CKD status, the rise in dialysis dependence among survivors over time was only observed in patients with CKD (from 36.8% in 2009 to 43.9% in 2018, P = 0.045) but not in patients without CKD (at approximately 18.5%, P = 0.561) (Table 3).

      Discussion

      In this multicenter study of ICU patients developing dialysis-requiring AKI treated in a universal health care system, we observed declining patient mortality over the past decade despite the increasing comorbidities of these patients. Furthermore, we observed a modest decrease in the severity of acute illness at RRT initiation and a similar threshold of RRT initiation over the study year, as demonstrated by lower total SOFA scores, lower median lactate levels, and less severe academia. The adjusted model of decreasing linear trends in mortality remained significant, although the decline did not persist in the recent 3 years. Additionally, our study showed an increasing trend in dialysis dependence among hospital survivors, particularly in patients with CKD.
      The present study results were consistent with previous studies regarding a reduction in mortality over time following dialysis-requiring AKI but had several unique aspects.
      • Harding J.L.
      • Li Y.
      • Burrows N.R.
      • Bullard K.M.
      • Pavkov M.E.
      US Trends in Hospitalizations for Dialysis-Requiring Acute Kidney Injury in People With Versus Without Diabetes.
      • Wald R.
      • McArthur E.
      • Adhikari N.K.
      • et al.
      Changing incidence and outcomes following dialysis-requiring acute kidney injury among critically ill adults: a population-based cohort study.

      Miyamoto Y, Iwagami M, Aso S, et al. Temporal change in characteristics and outcomes of acute kidney injury on renal replacement therapy in intensive care units: analysis of a nationwide administrative database in Japan, 2007-2016. Crit Care. 2019;23(1):172. doi:10.1186/s13054-019-2468-8

      First, our patients represent a group of ICU patients with high comorbidities, which complement the previous two large-scale studies. In a nationwide cohort study in Japan from 2007 to 2016, Miyamoto et al. found an increasing trend in the CCI score of ICU patients developing dialysis-requiring AKI, and their crude in-hospital mortality declined from 44.9% to 36.1%.

      Miyamoto Y, Iwagami M, Aso S, et al. Temporal change in characteristics and outcomes of acute kidney injury on renal replacement therapy in intensive care units: analysis of a nationwide administrative database in Japan, 2007-2016. Crit Care. 2019;23(1):172. doi:10.1186/s13054-019-2468-8

      The proportion of CCI scores of more than four in Miyamoto’s study is approximately 15%, representing a low to moderate comorbid population. However, our study showed that approximately 60% of patients had CCI scores of more than four. In a Canadian cohort of 21234 ICU patients with dialysis-requiring AKI, 90-day mortality declined from 50% in 1996 to 2000 to 45% in 2006 to 2010.
      • Wald R.
      • McArthur E.
      • Adhikari N.K.
      • et al.
      Changing incidence and outcomes following dialysis-requiring acute kidney injury among critically ill adults: a population-based cohort study.
      They reported that 60% of patients had a CCI score of <2. Second, in addition to the information about the proportion of mechanical ventilation or vasopressor use, the use of severity of illness score, such as SOFA scores, and laboratory data, including lactate level, allowed us to quantify better the severity of patients’ condition at initiating RRT in the ICU. We still observed a significant decreasing mortality trend despite a slightly declining total SOFA score at RRT initiation over time. Finally, few studies have tried to look at the effect of the threshold of dialysis indications on patient outcomes over time, although the issue of optimal criteria for initiating RRT in critically ill patients remains controversial.
      • Barbar S.D.
      • Clere-Jehl R.
      • Bourredjem A.
      • et al.
      Timing of Renal-Replacement Therapy in Patients with Acute Kidney Injury and Sepsis.
      • Bagshaw S.M.
      • Wald R.
      • Adhikari N.K.J.
      • et al.
      Timing of Initiation of Renal-Replacement Therapy in Acute Kidney Injury.
      • Gaudry S.
      • Hajage D.
      • Martin-Lefevre L.
      • et al.
      Comparison of two delayed strategies for renal replacement therapy initiation for severe acute kidney injury (AKIKI 2): a multicentre, open-label, randomised, controlled trial.
      • Karvellas C.J.
      • Farhat M.R.
      • Sajjad I.
      • et al.
      A comparison of early versus late initiation of renal replacement therapy in critically ill patients with acute kidney injury: a systematic review and meta-analysis.
      One may hypothesize that lower thresholds to start RRT over time may potentially enroll more “less severe” patients who are more likely to survive.
      • Wald R.
      • McArthur E.
      • Adhikari N.K.
      • et al.
      Changing incidence and outcomes following dialysis-requiring acute kidney injury among critically ill adults: a population-based cohort study.
      Our analysis did not conclude that dialysis criteria were the main contributor to the declining mortality. The proportion of severe hyperkalemia and acute pulmonary edema did not change over time except for the significant decline in the proportion of severe academia. Our study revealed a modest increase in the levels of BUN and Cr at the time of RRT over this period, even limiting the patients without CKD, suggesting that more liberal RRT initiation is less likely in the study period.
      In our study, the total SOFA score at ICU admission did not change over time. Although the significance of the decline in mortality gradually attenuated after the inclusion of several critical interventions and the severity at RRT initiation, they remained statistically significant. The explanation for better patient mortality is likely multifactorial, and we completely agree with Wald et al.,
      • Wald R.
      • McArthur E.
      • Adhikari N.K.
      • et al.
      Changing incidence and outcomes following dialysis-requiring acute kidney injury among critically ill adults: a population-based cohort study.
      who suggested that advances in critical care management contribute to improved outcomes. Population-based or nationwide cohorts have found improved mortality over the past decades for critically ill patients with acute myocardial infarction or heart failure,
      • Woolridge S.
      • Alemayehu W.
      • Kaul P.
      • et al.
      National trends in coronary intensive care unit admissions, resource utilization, and outcomes.
      severe sepsis and septic shock,
      • Kadri S.S.
      • Rhee C.
      • Strich J.R.
      • et al.
      Estimating Ten-Year Trends in Septic Shock Incidence and Mortality in United States Academic Medical Centers Using Clinical Data.
      ,
      • Kaukonen K.M.
      • Bailey M.
      • Suzuki S.
      • Pilcher D.
      • Bellomo R.
      Mortality related to severe sepsis and septic shock among critically ill patients in Australia and New Zealand, 2000-2012.
      and those who undergo cardiac surgery.
      • Lenihan C.R.
      • Montez-Rath M.E.
      • Mora Mangano C.T.
      • Chertow G.M.
      • Winkelmayer W.C.
      Trends in acute kidney injury, associated use of dialysis, and mortality after cardiac surgery, 1999 to 2008.
      ,
      • Zimmerman J.E.
      • Kramer A.A.
      • Knaus W.A.
      Changes in hospital mortality for United States intensive care unit admissions from 1988 to 2012.
      This finding is valuable if these medical advances and improved ICU care may also bring survival benefits to those patients with dialysis-requiring AKI. For example, our study showed a substantial shift in vasopressor use over the study period. This could be the result of increasing awareness of the adverse effects of dopamine from several critical studies of vasopressors in 2010, and the concept of optimal vasopressor management has been effectively disseminated.
      • Patel G.P.
      • Grahe J.S.
      • Sperry M.
      • et al.
      Efficacy and safety of dopamine versus norepinephrine in the management of septic shock.
      ,
      • De Backer D.
      • Biston P.
      • Devriendt J.
      • et al.
      Comparison of dopamine and norepinephrine in the treatment of shock.
      Although several trials of RRT-related intervention have not been shown to improve outcomes, they still convey some critical knowledge that help us provide better care for critically ill patients with dialysis-requiring AKI. To this point, various evidence-based guidance has emerged regarding dialysis catheter and anticoagulation management, selection of RRT modality, optimal RRT timing, and dose intensity, and they are sometimes required to adapt to the changing clinical needs of critically ill patients.
      • Truche A.S.
      • Darmon M.
      • Bailly S.
      • et al.
      Continuous renal replacement therapy versus intermittent hemodialysis in intensive care patients: impact on mortality and renal recovery.
      • Bagshaw S.M.
      • Darmon M.
      • Ostermann M.
      • et al.
      Current state of the art for renal replacement therapy in critically ill patients with acute kidney injury.
      • Karkar A.
      • Ronco C.
      Prescription of CRRT: a pathway to optimize therapy.
      Thus, it is likely that increased awareness of potential drawbacks with intensive RRT, such as hypophosphatemia
      • Maynar Moliner J.
      • Honore P.M.
      • Sánchez-Izquierdo Riera J.A.
      • Herrera Gutiérrez M.
      • Spapen H.D.
      Handling continuous renal replacement therapy-related adverse effects in intensive care unit patients: the dialytrauma concept.
      ,
      • Sharma S.
      • Kelly Y.P.
      • Palevsky P.M.
      • Waikar S.S.
      Intensity of Renal Replacement Therapy and Duration of Mechanical Ventilation: Secondary Analysis of the Acute Renal Failure Trial Network Study.
      and subtherapeutic levels of antibiotics
      • Sharma S.
      • Kelly Y.P.
      • Palevsky P.M.
      • Waikar S.S.
      Intensity of Renal Replacement Therapy and Duration of Mechanical Ventilation: Secondary Analysis of the Acute Renal Failure Trial Network Study.
      , more timely RRT intervention based on patient’s potential demand and renal capacity
      • Ostermann M.
      • Dickie H.
      • Barrett N.A.
      Renal replacement therapy in critically ill patients with acute kidney injury--when to start.
      ,
      • Macedo E.
      • Mehta R.L.
      Timing of dialysis initiation in acute kidney injury and acute-on-chronic renal failure.
      , or more skillful management of RRT-induced hemodynamic instability
      • Douvris A.
      • Zeid K.
      • Hiremath S.
      • et al.
      Mechanisms for hemodynamic instability related to renal replacement therapy: a narrative review.
      , may additively reduce the complication of RRT and possibly increase the chances of survival in our study population.
      Although patient mortality declined over the study period, there was an increasing trend of dialysis dependence among hospital survivors from 31.0% in 2009 to 37.5% in 2018. This higher rate of dialysis dependence, compared with around 26% in Wald et al.’s study
      • Wald R.
      • McArthur E.
      • Adhikari N.K.
      • et al.
      Changing incidence and outcomes following dialysis-requiring acute kidney injury among critically ill adults: a population-based cohort study.
      and approximately 18%–19% in Miyamoto et al.’s study

      Miyamoto Y, Iwagami M, Aso S, et al. Temporal change in characteristics and outcomes of acute kidney injury on renal replacement therapy in intensive care units: analysis of a nationwide administrative database in Japan, 2007-2016. Crit Care. 2019;23(1):172. doi:10.1186/s13054-019-2468-8

      , may reflect a higher proportion of CKD and higher severity of acute illness in our study population. Contrary to the results of Miyamoto et al.’s study, we did not observe an increasing trend in dialysis dependence in patients without CKD, which is consistently around 18%. The discrepancy is unclear, but raises the possibility that dialysis dependence and death represent two competitive risks. Thus, the likelihood of becoming dialysis-dependent is decreased in the population with relatively high mortality. Further studies are required to explore this issue.
      Our data showed that the declining mortality trends did not continue and even slightly reversed in 2016–2018, especially in-hospital and 90-day mortality. This might be explained by potentially including more patients with more devastating conditions that could not reflect by single SOFA scores. Another possibility is whether the improvement in survival is harder to achieve in dialysis-requiring critically ill patients with more complex comorbidities or extremely high comorbidities, e.g., underlying malignancy or CCI score >8. The cumulative effect of multiple comorbidities on an individual may vary with the combination and severity of the disorder.
      • Sullivan M.K.
      • Jani B.D.
      • Lees J.S.
      • et al.
      Multimorbidity and the risk of major adverse kidney events: findings from the UK Biobank cohort.
      In the future, investigations among ICU patients with dialysis-requiring AKI should be undertaken to identify certain combinations of comorbidities associated with a heightened mortality risk. Further studies incorporating serial measurements of SOFA scores are also needed to better characterize the severity of illness in this population. These data will allow us to better understand the basis for the observed declining mortality through contemporary treatment strategies.
      The strengths of this study are the large, multicenter study and complete data collection at ICU admission and RRT initiation. Another strength of this study is that it is from a universal electronic medical records system, which provided data that are more consistent, and accuracy could be checked immediately compared with a nationwide registry database. However, this study has some limitations. First, our data comes from six centers in one country, potentially limiting external generalizability. Moreover, only patients with first-episode dialysis-requiring AKI after ICU admission were enrolled, and the results may not be extrapolated to all ICU patients undergoing dialysis. Potential heterogeneity of patients and regular care among different branches does exist. However, we have adjusted for the “branch” effect, and the results did not change. Second, comorbidity data were only collected from the CGMH system; this may have resulted in some miscoding. We also had no access to the mortality data of the patients who had survived until hospital discharge but possibly later died at other hospitals, which could have underestimated the 90-day mortality. However, any such misclassification would be a random effect and was likely to be nondifferential among each index year. Third, we used diagnostic codes rather than validated clinical criteria to ascertain comorbidities other than CKD. Fourth, trends in mortality might vary according to the etiologies of critically ill patients. The relatively small number of individual causes of admission prevented adequate subgroup analysis, and thus, the applicability of our data to a specific type of disease with dialysis-requiring AKI is limited. Finally, we cannot exclude the possibility of residual confounding in the analysis, such as severity of baseline comorbidities, detailed medication intervention, delivered dose of RRT, or strategy of fluid management.
      In conclusion, this multicenter study demonstrated a decline in patient mortality among critically ill patients with dialysis-requiring AKI from 2009 to 2018. In addition, after adjustment for baseline characteristics and physiologic measurements of the severity of illness, the adjusted mortality remained significant. Although these findings are encouraging, the burden of mortality and dialysis dependence remains high, especially in patients with CKD. Future work should focus on strategies to mitigate the risk of kidney injury and enhance renal recovery.

      Acknowledgements

      We would like to thank all the nursing and support staff of the intensive care unit and dialysis care team of Chang Gung Memorial Hospital system for all the effort they have put in. The authors also thank Alfred Hsing-Fen Lin for his assistance in statistical analysis.
      Peer Review: Received March 12, 2022 Evaluated by 2 external peer reviewers, with direct editorial input from a Statistics/Methods Editor, an Associate Editor, and the Editor-in-Chief. Accepted in revised form August 17, 2022.

      References

        • Hoste E.A.
        • Bagshaw S.M.
        • Bellomo R.
        • et al.
        Epidemiology of acute kidney injury in critically ill patients: the multinational AKI-EPI study.
        Intensive Care Med. 2015; 41: 1411-1423https://doi.org/10.1007/s00134-015-3934-7
        • Uchino S.
        • Kellum J.A.
        • Bellomo R.
        • et al.
        Acute renal failure in critically ill patients: a multinational, multicenter study.
        JAMA. 2005; 294: 813-818https://doi.org/10.1001/jama.294.7.813
        • Wu V.C.
        • Huang T.M.
        • Lai C.F.
        • et al.
        Acute-on-chronic kidney injury at hospital discharge is associated with long-term dialysis and mortality.
        Kidney Int. 2011; 80: 1222-1230https://doi.org/10.1038/ki.2011.259
        • Gallagher M.
        • Cass A.
        • Bellomo R.
        • et al.
        Long-term survival and dialysis dependency following acute kidney injury in intensive care: extended follow-up of a randomized controlled trial.
        PLoS Med. 2014; 11e1001601https://doi.org/10.1371/journal.pmed.1001601
        • Elseviers M.M.
        • Lins R.L.
        • Van der Niepen P.
        • et al.
        Renal replacement therapy is an independent risk factor for mortality in critically ill patients with acute kidney injury.
        Crit Care. 2010; 14: R221https://doi.org/10.1186/cc9355
        • Lenihan C.R.
        • Montez-Rath M.E.
        • Mora Mangano C.T.
        • Chertow G.M.
        • Winkelmayer W.C.
        Trends in acute kidney injury, associated use of dialysis, and mortality after cardiac surgery, 1999 to 2008.
        Ann Thorac Surg. 2013; 95: 20-28https://doi.org/10.1016/j.athoracsur.2012.05.131
        • Brown J.R.
        • Rezaee M.E.
        • Hisey W.M.
        • Cox K.C.
        • Matheny M.E.
        • Sarnak M.J.
        Reduced Mortality Associated with Acute Kidney Injury Requiring Dialysis in the United States.
        Am J Nephrol. 2016; 43: 261-270https://doi.org/10.1159/000445846
        • Harding J.L.
        • Li Y.
        • Burrows N.R.
        • Bullard K.M.
        • Pavkov M.E.
        US Trends in Hospitalizations for Dialysis-Requiring Acute Kidney Injury in People With Versus Without Diabetes.
        Am J Kidney Dis. 2020; 75: 897-907https://doi.org/10.1053/j.ajkd.2019.09.012
        • Wald R.
        • McArthur E.
        • Adhikari N.K.
        • et al.
        Changing incidence and outcomes following dialysis-requiring acute kidney injury among critically ill adults: a population-based cohort study.
        Am J Kidney Dis. 2015; 65: 870-877https://doi.org/10.1053/j.ajkd.2014.10.017
      1. Miyamoto Y, Iwagami M, Aso S, et al. Temporal change in characteristics and outcomes of acute kidney injury on renal replacement therapy in intensive care units: analysis of a nationwide administrative database in Japan, 2007-2016. Crit Care. 2019;23(1):172. doi:10.1186/s13054-019-2468-8

        • Shao S.C.
        • Chan Y.Y.
        • Kao Yang Y.H.
        • et al.
        The Chang Gung Research Database-A multi-institutional electronic medical records database for real-world epidemiological studies in Taiwan.
        Pharmacoepidemiol Drug Saf. 2019; 28: 593-600https://doi.org/10.1002/pds.4713
        • Chan C.C.
        • Lee K.T.
        • Ho W.J.
        • Chan Y.H.
        • Chu P.H.
        Levosimendan use in patients with acute heart failure and reduced ejection fraction with or without severe renal dysfunction in critical cardiac care units: a multi-institution database study.
        Ann Intensive Care. 2021; 11: 27https://doi.org/10.1186/s13613-021-00810-y
        • Levey A.S.
        • Stevens L.A.
        • Schmid C.H.
        • et al.
        A new equation to estimate glomerular filtration rate.
        Ann Intern Med. 2009; 150: 604-612https://doi.org/10.7326/0003-4819-150-9-200905050-00006
        • Barbar S.D.
        • Clere-Jehl R.
        • Bourredjem A.
        • et al.
        Timing of Renal-Replacement Therapy in Patients with Acute Kidney Injury and Sepsis.
        N Engl J Med. 2018; 379: 1431-1442https://doi.org/10.1056/NEJMoa1803213
        • Bagshaw S.M.
        • Wald R.
        • Adhikari N.K.J.
        • et al.
        Timing of Initiation of Renal-Replacement Therapy in Acute Kidney Injury.
        N Engl J Med. 2020; 383: 240-251https://doi.org/10.1056/NEJMoa2000741
        • Gaudry S.
        • Hajage D.
        • Martin-Lefevre L.
        • et al.
        Comparison of two delayed strategies for renal replacement therapy initiation for severe acute kidney injury (AKIKI 2): a multicentre, open-label, randomised, controlled trial.
        Lancet. 2021; 397: 1293-1300https://doi.org/10.1016/s0140-6736(21)00350-0
        • Karvellas C.J.
        • Farhat M.R.
        • Sajjad I.
        • et al.
        A comparison of early versus late initiation of renal replacement therapy in critically ill patients with acute kidney injury: a systematic review and meta-analysis.
        Crit Care. 2011; 15: R72https://doi.org/10.1186/cc10061
        • Woolridge S.
        • Alemayehu W.
        • Kaul P.
        • et al.
        National trends in coronary intensive care unit admissions, resource utilization, and outcomes.
        Eur Heart J Acute Cardiovasc Care. 2020; 9: 923-930https://doi.org/10.1177/2048872619883400
        • Kadri S.S.
        • Rhee C.
        • Strich J.R.
        • et al.
        Estimating Ten-Year Trends in Septic Shock Incidence and Mortality in United States Academic Medical Centers Using Clinical Data.
        Chest. 2017; 151: 278-285https://doi.org/10.1016/j.chest.2016.07.010
        • Kaukonen K.M.
        • Bailey M.
        • Suzuki S.
        • Pilcher D.
        • Bellomo R.
        Mortality related to severe sepsis and septic shock among critically ill patients in Australia and New Zealand, 2000-2012.
        JAMA. 2014; 311: 1308-1316https://doi.org/10.1001/jama.2014.2637
        • Zimmerman J.E.
        • Kramer A.A.
        • Knaus W.A.
        Changes in hospital mortality for United States intensive care unit admissions from 1988 to 2012.
        Crit Care. 2013; 17: R81https://doi.org/10.1186/cc12695
        • Patel G.P.
        • Grahe J.S.
        • Sperry M.
        • et al.
        Efficacy and safety of dopamine versus norepinephrine in the management of septic shock.
        Shock. 2010; 33: 375-380https://doi.org/10.1097/SHK.0b013e3181c6ba6f
        • De Backer D.
        • Biston P.
        • Devriendt J.
        • et al.
        Comparison of dopamine and norepinephrine in the treatment of shock.
        N Engl J Med. 2010; 362: 779-789https://doi.org/10.1056/NEJMoa0907118
        • Truche A.S.
        • Darmon M.
        • Bailly S.
        • et al.
        Continuous renal replacement therapy versus intermittent hemodialysis in intensive care patients: impact on mortality and renal recovery.
        Intensive Care Med. 2016; 42: 1408-1417https://doi.org/10.1007/s00134-016-4404-6
        • Bagshaw S.M.
        • Darmon M.
        • Ostermann M.
        • et al.
        Current state of the art for renal replacement therapy in critically ill patients with acute kidney injury.
        Intensive Care Med. 2017; 43: 841-854https://doi.org/10.1007/s00134-017-4762-8
        • Karkar A.
        • Ronco C.
        Prescription of CRRT: a pathway to optimize therapy.
        Ann Intensive Care. 2020; 10: 32https://doi.org/10.1186/s13613-020-0648-y
        • Maynar Moliner J.
        • Honore P.M.
        • Sánchez-Izquierdo Riera J.A.
        • Herrera Gutiérrez M.
        • Spapen H.D.
        Handling continuous renal replacement therapy-related adverse effects in intensive care unit patients: the dialytrauma concept.
        Blood Purif. 2012; 34: 177-185https://doi.org/10.1159/000342064
        • Sharma S.
        • Kelly Y.P.
        • Palevsky P.M.
        • Waikar S.S.
        Intensity of Renal Replacement Therapy and Duration of Mechanical Ventilation: Secondary Analysis of the Acute Renal Failure Trial Network Study.
        Chest. 2020; 158: 1473-1481https://doi.org/10.1016/j.chest.2020.05.542
        • Ostermann M.
        • Dickie H.
        • Barrett N.A.
        Renal replacement therapy in critically ill patients with acute kidney injury--when to start.
        Nephrol Dial Transplant. 2012; 27: 2242-2248https://doi.org/10.1093/ndt/gfr707
        • Macedo E.
        • Mehta R.L.
        Timing of dialysis initiation in acute kidney injury and acute-on-chronic renal failure.
        Semin Dial. 2013; 26: 675-681https://doi.org/10.1111/sdi.12128
        • Douvris A.
        • Zeid K.
        • Hiremath S.
        • et al.
        Mechanisms for hemodynamic instability related to renal replacement therapy: a narrative review.
        Intensive Care Med. 2019; 45: 1333-1346https://doi.org/10.1007/s00134-019-05707-w
        • Sullivan M.K.
        • Jani B.D.
        • Lees J.S.
        • et al.
        Multimorbidity and the risk of major adverse kidney events: findings from the UK Biobank cohort.
        Clin Kidney J. 2021; 14: 2409-2419https://doi.org/10.1093/ckj/sfab079