Clin Exp Nephrol DOI 10.1007/s10157-013-0915-4
ORIGINAL ARTICLE
Comparison of Kidney Disease: Improving Global Outcomes and Acute Kidney Injury Network criteria for assessing patients in intensive care units Hibiki Shinjo • Waichi Sato • Enyu Imai • Tomoki Kosugi • Hiroki Hayashi • Kunihiro Nishimura • Kimitoshi Nishiwaki Yukio Yuzawa • Seiichi Matsuo • Shoichi Maruyama
•
Received: 14 June 2013 / Accepted: 11 November 2013 Ó Japanese Society of Nephrology 2013
Abstract Background The Kidney Disease: Improving Global Outcomes (KDIGO) group proposed to adopt the 48-h time window for the 0.3 mg/dL rise in serum creatinine (sCr) proposed by the Acute Kidney Injury Network (AKIN) group as a modification to the original risk, injury, failure, loss, and end-stage renal disease criteria, keeping the 7-day window for the 50 % increase in sCr from baseline. The present study evaluates the prevalence of acute kidney injury (AKI) and the accuracy of predicting mortality based on the KDIGO and AKIN criteria. Patients and methods We retrospectively studied a cohort of 2579 patients admitted to the intensive care unit of Nagoya University Hospital between 2005 and 2009. Results The total AKI prevalence was higher according to the KDIGO than to the AKIN criteria (38.4 versus 29.5 %). In-hospital mortality rates were higher among 238 patients classified as non-AKI by the AKIN but AKI by the KDIGO criteria than among those classified as non-AKI by H. Shinjo W. Sato (&) E. Imai T. Kosugi S. Matsuo S. Maruyama Departments of Nephrology, Nagoya University Graduate School of Medicine, 65 Tsurumai, Showa-ku, Nagoya, Aichi 466-8550, Japan e-mail: [email protected] H. Hayashi Y. Yuzawa Department of Nephrology, Fujita Health University School of Medicine, Toyoake, Japan K. Nishimura Department of Preventive Medicine and Epidemiology, National Cerebral and Cardiovascular Center Hospital, Suita, Japan K. Nishiwaki Departments of Anesthesiology, Nagoya University Graduate School of Medicine, Nagoya, Japan
both criteria (7.1 versus 2.7 %). Survival curves generated using KDIGO significantly differed among all stages, but not between AKIN stages I and II. Multivariate analysis showed that KDIGO criteria were better in a statistical model than the AKIN criteria according to the Akaike information criterion. Harrell’s C statistic was greater for the KDIGO than for the AKIN criteria. Conclusions The KDIGO criteria have improved sensitivity without compromising specificity for AKI and might predict mortality at least as well as the AKIN criteria. Keywords Acute kidney injury (AKI) Acute renal failure (ARF) Mortality Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Network (AKIN)
Introduction Acute renal failure (ARF) causes uremia and various electrolyte and fluid abnormalities, and it is defined as a rapid loss of kidney function over a period ranging from hours to days. ARF is considered to increase the mortality of critically ill patients [1, 2]. However, the wide variation in ARF definitions has caused difficulties in comparing results across studies and populations [3–7]. The Acute Dialysis Quality Initiative group proposed the risk, injury, failure, loss of kidney function, and end-stage kidney disease (RIFLE) classification in 2004 for acute kidney injury (AKI) [8], which is a frequent complication among critically ill patients admitted to the intensive care unit (ICU) [9]. These criteria for AKI were associated with mortality [10–14]. The RIFLE criteria included a recommendation to assume normal baseline renal function [backcalculation using modification of diet in renal disease
123
Clin Exp Nephrol
(MDRD) assuming a glomerular filtration rate (GFR) of 75 mL/min/1.73 m2] in the absence of baseline serum creatinine (sCr) values. However, such an assumption might overestimate the proportion of patients with AKI by ignoring those with extant chronic kidney disease (CKD) [15–17]. The Acute Kidney Injury Network (AKIN) group modified the RIFLE criteria in 2007 from five viewpoints [18]. The AKIN criteria proposed a time frame of 48 h for a diagnosis of AKI. AKIN stage I (risk in the RIFLE criteria) includes a slight increase of sCr of C0.3 mg/dL (C26.4 lmol/L) [19, 20]. A requirement for renal replacement therapy (RRT) was classified as AKIN stage III (failure in the RIFLE criteria). Outcome categories [loss and end-stage kidney disease (ESKD) in the RIFLE criteria] were eliminated and changes in GFR were excluded because increases in sCr and decreases in GFR in the definitions of RIFLE category (risk and failure) do not correspond mathematically. The risk category is defined as a 1.5-fold increase in sCr and a 25 % decrease in GFR, but a 1.5-fold increase in sCr corresponds to a one-third decrease in GFR. The failure category is defined as a threefold increase and a 75 % decrease, but a threefold increase corresponds to a decrease of two-thirds [21]. Compared with the RIFLE criteria, the AKIN criteria improved the sensitivity of AKI diagnosis, but did not significantly improve predictions of in-hospital mortality rates among critically ill patients [22–25]. In addition, the AKIN criteria did not improve sensitivity, robustness or predictive ability, because the introduction of a 48-h window for a diagnosis of AKI might miss patients with a slow increase in sCr that could not be identified within 48 h. The Kidney Disease: Improving Global Outcomes (KDIGO) working group modified the AKI criteria in 2012 [26]. This group proposed to adopt the 48-h time window for the 0.3 mg/dL rise in sCr proposed by the AKIN group as a modification to the original RIFLE criteria, keeping the 7-day window for the 50 % increase in sCr from baseline. The proposed baseline used the lowest sCr value recorded within 3 months of an event. The present study evaluates the prevalence of AKI and the accuracy of predicting in-hospital mortality among critically ill patients using the KDIGO and AKIN criteria for comparison.
Patients and methods Patients and procedures The Ethics Committee of Nagoya University School of Medicine approved this study (approval number 1), which proceeded according to the principles of the Declaration of Helsinki. Patients’ data were extracted from ICU and
123
Fig. 1 Study flowchart. Among 2579 critically ill patients admitted to ICU, we excluded those with ESKD on chronic hemodialysis or peritoneal dialysis, those who had undergone kidney transplant, those \16 years old, and those who spent\1 day in ICU. AKI acute kidney injury, AKIN Acute Kidney Injury Network, ESKD end-stage kidney disease, ICU intensive care unit, KDIGO Kidney Disease: Improving Global Outcomes
electronic hospital records. Laboratory data were extracted from our laboratory database. We matched data from the different sources and then manually organized the information. We retrospectively investigated a cohort of all adults moved to the ICU after admission to Nagoya University Hospital (n = 3350) between 27 May 2005 and 31 May 2009. We excluded patients with end-stage kidney disease (ESKD) on chronic hemodialysis, peritoneal dialysis, kidney transplants (n = 78), those younger than 16 years (n = 154) or who had stayed in the ICU for \1 day (n = 178), and considered only the first admission when patients were readmitted to the ICU during the study period (n = 360) (Fig. 1). We finally acquired data from 2579 patients. The end point of the study was all-cause (in-hospital) mortality. Systems for classifying disease severity The Acute Physiology and Chronic Health Evaluation (APACHE) II, Sequential Organ Failure Assessment (SOFA), and Simplified Acute Physiology Score (SAPS) II scores were calculated based on variables recorded during the first 24 h of ICU admission [27–29]. Nonrenal scores were calculated from the total score minus scores for renal parameters. Definitions of AKI We defined AKI according to the KDIGO and AKIN criteria, using sCr values during hospitalization. The
Clin Exp Nephrol Table 1 Comparison of Kidney Disease: Improving Global Outcomes, Acute Kidney Injury Network, and risk, injury, failure, loss of kidney function, and end-stage kidney disease criteria for acute kidney injury Diagnostic criteria for AKI: Abrupt reduction in kidney function defined as: Absolute increase in serum creatinine of either C0.3 mg/dL (KDIGO and AKIN criteria; within 48 h) or C50 % increase in serum creatinine within 48 h (AKIN criteria) or 7 days (KDIGO and RIFLE criteria) or reduced urine output KDIGO
Serum creatinine criteria
Urine output criteria
Stage 1
Increase to C150–199 % from baseline or increase in serum creatinine C0.3 mg/dL
\0.5 mL/kg/h for C6 h
Stage 2
Increase in serum creatinine to C200–299 % from baseline
\0.5 mL/kg/h for C12 h
Stage 3
Increase in serum creatinine to C300 % from baseline or serum creatinine C4.0 mg/ dL or initiation of renal replacement therapy or in patients \18 years, decrease in eGFR to \35 mL/min per 1.73 m2
\0.3 mL/kg/h for C24 h or anuria for C12 h
AKIN
Serum creatinine criteria
Urine output criteria
Stage I
Increase to C150–200 % from baseline or increase in serum creatinine C0.3 mg/dL
\0.5 mL/kg/h for C6 h
Stage II
Increase in serum creatinine to [200–300 % from baseline
\0.5 mL/kg/h for C12 h
Stage III
Increase in serum creatinine to[300 % from baseline or serum creatinine C4.0 mg/ dL with an acute increase of at least 0.5 mg/dL or initiation of renal replacement therapy
\0.3 mL/kg/h for C24 h or anuria for C12 h
RIFLE
GFR criteria
Urine output criteria
Risk
Serum creatinine 9 1.5 or GFR decrease [25 %
\0.5 mL/kg/h for C6 h
Injury
Serum creatinine 9 2 or GFR decrease [50 %
\0.5 mL/kg/h for C12 h
Failure
Serum creatinine 9 3 or GFR decrease [75 % or serum creatinine C4.0 mg/dL with an acute increase of at least 0.5 mg/dL
\0.3 mL/kg/h for C24 h or anuria for C12 h
Loss
Persistent acute renal failure = complete loss of kidney function [4 weeks
ESKD
End-stage kidney disease [3 months
AKI acute kidney injury, KDIGO Kidney Disease: Improving Global Outcomes, AKIN Acute Kidney Injury Network, RIFLE risk, injury, failure, loss of kidney function, and end-stage kidney disease, GFR glomerular filtration rate, ESKD end-stage kidney disease
diagnosis of AKI for all the patients was analyzed using both criteria, that is, an abrupt (within 48 h) reduction in kidney function currently defined as an absolute increase in sCr of C0.3 mg/dL or a reduction in urine output (documented oliguria of \0.5 mL/kg/h for C6 h). In addition, AKI was defined as a C50 % (1.5-fold from baseline) increase in sCr within 48 h by the AKIN criteria and within 7 days by the KDIGO criteria. The AKIN criteria strictly applied a moving window of 48 h [25]. The KDIGO criteria used the lowest value during the 3 months before ICU admission as baseline sCr (Table 1). Due to the absence of information about 6- or 12-h urine volumes, we determined KDIGO and AKIN stages using only sCr criteria. Statistical analysis Continuous variables are expressed as mean ± standard deviation (SD) or as median [interquartile range (IQR)]. Means were compared using Student’s t test when normally distributed, and the Mann–Whitney U test otherwise. Categorical variables were analyzed using the Chi-square test and are presented as ratios (%) of number of patients. To
correct for differences in patient characteristics, we included age, gender, reason for ICU admission, type of ICU admission, nonrenal SOFA score, duration of mechanical ventilation (MV), and AKIN or KDIGO stage. The nonrenal SOFA score was selected as a covariate to control for multicollinearity between the AKI classification and scoring systems that included values for kidney insufficiency such as the APACHE II and SOFA scores. The statistical significance of differences in the prevalence of AKI between KDIGO and AKIN stages was assessed using McNemar’s test. Associations between the AKIN and KDIGO criteria with mortality were evaluated using univariate and multivariate Cox proportional hazards analysis. The Akaike information criterion (AIC) is an index for evaluating the merit of a statistical model and was used to balance the complexity of models and the goodness of fit with the data. A model with lower AIC is better balanced [30]. To compare the predictability of mortality between two criteria, discrimination was also evaluated using Harrell’s C index. A model with perfect predictive capacity (sensitivity and specificity of 100 %) has a Harrell’s C statistic of 1.00; Harrell’s C index is a statistic similar to
123
Clin Exp Nephrol Table 2 Baseline characteristics of patients Entire cohort (n = 2579)
Hospital survivors (n = 2397)
Hospital nonsurvivors (n = 182)
P value*
\0.001
Demographics Age, years
63.2 ± 13.8
62.9 ± 13.8
67.6 ± 12.8
Male sex
1708 (66.2 %)
1598 (66.7 %)
110 (60.4 %)
Body mass index, kg/m2
22.2 ± 3.7
22.2 ± 3.6
22.1 ± 4.1
0.666
475 (18.4 %)
438 (18.3 %)
37 (20.3 %)
0.490
0.70 (0.56–0.86) 82.3 (64.1–100.3)
0.70 (0.56–0.85) 83.0 (65.1–100.3)
0.79 (0.58–1.1) 68.7 (45.3–100.6)
0.001 \0.001 \0.001
Diabetes mellitus
0.087
Baseline renal function Baseline serum creatinine, mg/dL Baseline eGFR, mL/min/1.73 m2 C90
990 (38.4 %)
934 (39.0 %)
56 (30.8 %)
60–89
1064 (41.3 %)
1015 (42.3 %)
49 (26.9 %)
30–59
422 (16.4 %)
366 (15.3 %)
56 (30.8 %)
15–29
71 (2.8 %)
58 (2.4 %)
13 (7.1 %)
32 (1.2 %)
24 (1.0 %)
8 (4.4 %)
Elective surgery
1917 (74.3 %)
1876 (78.3 %)
41 (22.5 %)
Emergency surgery
322 (12.5 %)
295 (12.3 %)
27 (14.8 %)
Nonsurgical
340 (13.2 %)
226 (9.4 %)
114 (62.6 %)
Cardiovascular
993 (38.5 %)
947 (39.5 %)
46 (25.3 %)
Gastrointestinal
64 (2.5 %)
61 (2.5 %)
3 (1.6 %)
Hepatic
79 (3.1 %)
73 (3.0 %)
6 (3.3 %)
Malignancy Neurological
964 (37.4 %) 261 (10.1 %)
912 (38.0 %) 234 (9.8 %)
52 (28.6 %) 27 (14.8 %)
Pulmonary
90 (3.5 %)
61 (2.5 %)
29 (15.9 %)
Other
128 (5.0 %)
109 (4.5 %)
19 (10.4 %)
4 (2–6)
4 (2–6)
7 (5–10)
\15
\0.001
ICU admission type
\0.001
Reason for ICU admission
Risk adjustment SOFA
\0.001
SOFA nonrenal score
4 (2–6)
4 (2–6)
7 (5–10)
\0.001
APACHE II
9 (7–12)
9 (7–12)
19 (13–24)
\0.001
APACHE II nonrenal score
9 (6–11)
8 (6–11)
17 (12–24)
\0.001
SAPS II
28 (23–35)
27 (22–34)
51 (39–63)
\0.001
SAPS II nonrenal score
27 (22–32)
26 (21–31)
44 (34–54)
\0.001
Renal replacement therapy
80 (3.1 %)
53 (2.2 %)
27 (14.8 %)
\0.001
Hospital length of stay (days)
31 (21–52)
31 (21–51)
24 (8–59)
\0.001
ICU length of stay (days)
1 (1–2)
1 (1–2)
3 (1–10)
\0.001
Duration of mechanical ventilation (days)
1 (0–1)
1 (0–1)
5 (1–17)
\0.001 \0.001
None
226 (8.8 %)
214 (8.9 %)
12 (6.6 %)
0–1
1828 (70.9 %)
1777 (74.1 %)
51 (28.0 %)
2–6
305 (11.8 %)
261 (10.9 %)
44 (24.2 %)
C7
220 (8.5 %)
145 (6.0 %)
75 (41.2 %)
Continuous variables presented as mean ± standard deviation when normally distributed or as median (interquartile interval) when not normally distributed. Categorical variables are presented as number (percentage) Nonrenal score was calculated from total score minus renal parameter scores APACHE Acute Physiology and Chronic Health Evaluation, eGFR estimated glomerular filtration rate, ICU intensive care unit, SOFA Sequential Organ Failure Assessment, SAPS Simplified Acute Physiology Score * P for hospital survivors compared with nonsurvivors
123
Clin Exp Nephrol Table 3 Prevalence and in-hospital mortality of acute kidney injury stratified by the Kidney Disease: Improving Global Outcomes and Acute Kidney Injury Network criteria Prevalence, n (%)
Table 4 Cross-tabulation of patients classified by Kidney Disease: Improving Global Outcomes versus Acute Kidney Injury Network criteria
In-hospital mortality, n (%)
AKIN Non-AKI
KDIGO Non-AKI
1588 (61.6)
44 (2.8)
Stage 1 Stage 2
639 (24.8) 159 (6.2)
44 (6.9) 28 (17.6)
Stage 3
193 (7.5)
66 (34.2)
All AKI
991 (38.4)
138 (13.9)
1818 (70.5)
60 (3.3)
535 (20.7)
55 (10.3)
Stage I Stage II
87 (3.4)
18 (20.7)
Stage III
139 (5.4)
49 (35.3)
All AKI
761 (29.5)
122 (16.0)
Patient baseline characteristics and univariate analysis of hospital survivors and nonsurvivors During the study period, 3350 critically ill patients were admitted to the ICU; data from 2579 (77.0 %) patients were analyzed (Fig. 1). The median duration of followup after ICU admission was 2.72 years (0.55–4.08 years, IQR). Table 2 summarizes the baseline characteristics of the patients [mean age (±SD) 63.2 (13.8) years; male 66.2 %]. The percentages of patients with diabetes mellitus, chronic kidney disease, and who underwent elective surgery were 18.4, 20.4, and 74.3 %, respectively. Table 2 also compares the baseline characteristics and physiological variables between in-hospital survivors and nonsurvivors. Values of mean age and median baseline sCr were higher for nonsurvivors. Nonsurvivors were more likely to be admitted for nonsurgical reasons
Total (KDIGO)
n
1580
8
0
0
1588
a
2.7 %
12.5 %
0%
0%
2.8 %
Stage 1 n
208
421
9
1
639
a
5.8 %
7.1 %
22.2 %
0%
6.9 %
Stage 2 n
23
85
49
2
159
a
13.0 %
16.5 %
20.4 %
50.0 %
17.6 %
Stage 3 n
7
21
29
136
193
a
28.6 %
47.6 %
20.7 %
35.3 %
34.2 %
535
87
139
2579
10.3 %
20.7 %
35.3 %
7.1 %
Total (AKIN) n 1818
Results
Stage III
Non-AKI
AKI acute kidney injury, AKIN Acute Kidney Injury Network, KDIGO Kidney Disease: Improving Global Outcomes
the area under the receiver operating characteristic curve [31, 32]. A two-tailed P value of \0.05 was considered significant. Hospital survival across groups was analyzed using the Chi-square and Kaplan–Meier methods, and differences between groups were tested using the log-rank test. We used STATA version 11 (Stata Corp LP, College Station, TX, USA) to evaluate survival data and diagnostic accuracy from receiver operating characteristic (ROC) curves and Harrell’s C index, and all other data were statistically analyzed using SPSS (BM SPSS statistics 19) software.
Stage II
KDIGO
AKIN Non-AKI
Stage I
a
3.3 %
Italics indicate patients judged as same degree of AKI by both classification systems AKI acute kidney injury, AKIN Acute Kidney Injury Network, KDIGO Kidney Disease: Improving Global Outcomes, n number of patients a
In-hospital mortality rate (%) of the respective group
and were more critically ill, and a higher proportion remained on mechanical ventilation for an extended period. Prevalence of AKI, in-hospital mortality, and crosstabulation according to AKI stratified by KDIGO and AKIN criteria The prevalence of AKI according to the KDIGO and AKIN criteria was 991 (38.4 %) and 761 (29.5 %), respectively. Among the 991 patients defined by the KDIGO criteria as AKI, 24.8, 6.2, and 7.5 % had stage 1, 2, and 3, respectively. Among the 761 patients defined by the AKIN criteria, 20.7, 3.4, and 5.4 % had stage I, II, and III, respectively (Table 3). The KDIGO criteria significantly increased the number of patients judged as AKI in all stages compared with the AKIN criteria (P \ 0.001). Table 4 cross-tabulates the stages classified by the KDIGO and AKIN criteria. Two hundred and thirty-eight (9.2 %) patients were classified as AKI by the KDIGO criteria but as non-AKI by the AKIN criteria, and 87 % of them were classified as KDIGO stage 1. Serum creatinine increased in this group from within 48 h to 7 days. These patients had
123
Clin Exp Nephrol Table 5 Risk factors for long-term mortality in patients (univariate analysis) HR
95 % CI
P value
Age
1.03
1.02–1.04
\0.001
Male sex
0.77
0.57–1.03
0.081
Body mass index
0.99
0.94–1.04
0.703
Diabetes mellitus
1.14
0.78–1.66
0.46
CKD
3.19
2.38–4.28
\0.001
1 0.83
Reference 0.56–1.12
0.329 \0.001
about 2.5-fold higher in-hospital mortality rates than those without AKI according to both criteria (7.1 versus 2.7 %, respectively), and were similar for KDIGO stage 1 and AKIN stage I (7.1 %). On the other hand, the in-hospital mortality rate was relatively high (12.5 %) among patients who were classified as non-AKI by the KDIGO criteria but AKI by the AKIN criteria. However, it was difficult to draw any conclusion since there were only 8 (0.3 %) patients in this group.
eGFR C90 60–89 30–59
2.6
1.79–3.77
15–29
3.71
2.03–6.78
\0.001
5.47
2.39–12.93
\0.001
Elective surgery
1
Reference
Emergency surgery
4.19
\15 ICU admission type
Nonsurgical
23.08
2.54–6.91
\0.001
15.74–33.84
\0.001
Reason for ICU admission Cardiovascular
1
Reference
Gastrointestinal
1.01
0.31–3.16
0.977
Hepatic
1.62
0.69–3.79
0.266
Malignancy
1.1
0.74–1.63
0.653
Neurological
2.27
1.41–3.65
0.001
Pulmonary Other
8.25 3.47
5.18–13.14 2.04–5.93
\0.001 \0.001
1.41
1.34–1.47
\0.001
Risk adjustment SOFA nonrenal score APACHE II nonrenal score
1.22
1.19–1.24
\0.001
SAPS II nonrenal score
1.11
1.12–1.16
\0.001
Duration of mechanical ventilation (days) None
1
Reference
0–1
0.51
0.27–0.99
2–6
2.94
1.55–5.56
C7
7.14
3.88–13.13
\0.001
2.44
2.16–2.74
\0.001
KDIGO (per stage)
0.038 0.001
Non-AKI
1
Reference
Stage 1
2.52
1.66–3.82
\0.001
Stage 2
6.64
4.14–10.67
\0.001
Stage 3
14.42
9.84–21.13
\0.001
2.32 1
2.06–2.61 Reference
\0.001
Stage I
3.22
2.24–4.65
\0.001
Stage II
6.78
4.00–11.48
\0.001
Stage III
12.83
8.79–18.73
\0.001
AKIN (per stage) Non-AKI
AKI acute kidney injury, APACHE Acute Physiology and Chronic Health Evaluation, AKIN Acute Kidney Injury Network, CI confidence interval, CKD chronic kidney disease, HR hazard ratio, ICU intensive care unit, KDIGO Kidney Disease: Improving Global Outcomes, SAPS Simplified Acute Physiology Score, SOFA Sequential Organ Failure Assessment
123
Univariate analysis for mortality judged by the KDIGO and AKIN criteria Univariate Cox proportional hazards models revealed the following as risk factors for increased mortality: age, chronic kidney disease, emergency surgery, nonsurgical admission, neurological admission, pulmonary admission, SOFA, APACHE II, SAPS II, prolonged mechanical ventilation, and KDIGO and AKIN stages (Table 5). Nonsurgical patients had high hazard ratio, because these patients had higher severity scores (SOFA, APACHE II, and SAPS II) in ICU than surgical patients. Risk factors for mortality using a multivariate Cox proportional hazards model and survival curves within KDIGO and AKIN stages We analyzed risk factors for mortality using multivariate Cox proportional hazards models 1, 2, and 3 selected by the Akaike information criterion (AIC) in which KDIGO and AKIN stages were considered as continuous variable, categorical variable, and stepwise variable selection, respectively. After adjustment for covariates [age, sex, body mass index (BMI), diabetes mellitus, CKD, ICU admission type, reason for ICU admission, nonrenal SOFA score, and duration of mechanical ventilation], the KDIGO and AKIN stages remained independently associated with mortality. Age, BMI, neurological and pulmonary patients, nonsurgical patients, emergency surgical patients, and nonrenal SOFA score were considered significant. Hazard ratios (HRs) and 95 % CIs for mortality versus non-AKI increased at every stage in models 2 and 3. Comparing the goodness of fit and complexity of the three models by AIC, the KDIGO criteria represented a more accurate predictor than the AKIN criteria (AIC value in KDIGO versus AKIN: model 1, 1181.03 versus 1191.07; model 2, 1183.97 versus 1193.31; model 3, 1175.65 versus 1184.64; Table 6). During follow-up after ICU admission, 417 (16.2 %) patients died. Figure 2a shows the survival curves for all patients with all stages of AKI according to the AKIN criteria. Stages I and II did not differ significantly
Clin Exp Nephrol Table 6 Predictive ability for long-term mortality by multivariate Cox proportional hazards models for Kidney Disease: Improving Global Outcomes and Acute Kidney Injury Network criteria
Model 1 Model 2
Model 3
KDIGO
HR
95 % CI
P value
AIC
AKIN
HR
95 % CI
P value
AIC
Per stage
1.69
1.37–2.09
\0.001
1181.03
Per stage
1.52
1.22–1.90
\0.001
1191.07
1183.97
Non-AKI
1
Reference
Stage I
2.08
1.19–3.63
Non-AKI
1
Reference
Stage 1
2.31
1.19–4.43
Stage 2
3.61
1.70–7.68
Stage 3
5.48
2.69–11.17
Non-AKI
1
Reference
Stage 1 Stage 2
2.46 3.94
1.12–3.75 1.88–8.27
Stage 3
5.98
2.99–11.94
0.012 0.001 \0.001 1175.65 0.012 0.001 \0.001
1193.31 0.011
Stage II
3.16
1.43–7.00
0.005
Stage III
3.71
1.82–7.57
\0.001
Non-AKI
1
Reference
Stage I Stage II
2.15 3.08
1.24–3.73 1.40–6.77
0.006 0.005
Stage III
3.92
1.99–7.71
\0.001
1184.64
Multivariate cox proportional hazards model adjusted for age, sex, BMI, diabetes mellitus, CKD, ICU admission type, reason for ICU admission, SOFA nonrenal score, and duration of mechanical ventilation. Model 1: KDIGO and AKIN stage considered as continuous variable. Model 2: KDIGO and AKIN stage considered as categorical variable. Model 3: KDIGO and AKIN stage considered as stepwise variable selection, by Akaike information criterion. Age, BMI, neurological and pulmonary cases, nonsurgical case, emergency surgery case, and SOFA nonrenal score were considered significant. Smaller AIC model was considered as a better prediction model AKIN Acute Kidney Injury Network, CI confidence interval, CKD chronic kidney disease, ICU intensive care unit, KDIGO Kidney Disease: Improving Global Outcomes, HR hazard ratio, AIC Akaike information criterion, SOFA Sequential Organ Failure Assessment
Table 7 Comparison between Kidney Disease: Improving Global Outcomes and Acute Kidney Injury Network criteria by Harrell’s C statistic of Cox model prediction Harrell’s C
P value
95 % CI
AKIN stage (continuous)
0.905
\0.001
0.873 to 0.936
KDIGO stage (continuous)
0.912
\0.001
-0.007
0.149
Model 1
Difference
0.882 to 0.941 -0.017 to 0.003
Model 2 AKIN stage (categorical)
0.908
\0.001
KDIGO stage (categorical)
0.915
\0.001
Difference
0.007
0.249
0.877 to 0.938 0.887 to 0.942 -0.005 to 0.018
Model 3 AKIN stage
0.908
\0.001
KDIGO stage
0.915
\0.001
Difference
0.007
0.249
0.877 to 0.938 0.887 to 0.942 -0.005 to 0.018
AKIN Acute Kidney Injury Network, KDIGO Kidney Disease: Improving Global Outcomes, CI confidence interval
Fig. 2 Long-term survival of patients classified according to Acute Kidney Injury Network (a) or Kidney Disease: Improving Global Outcomes criteria (b). AKI acute kidney injury, AKIN Acute Kidney Injury Network, KDIGO Kidney Disease: Improving Global Outcomes
(P = 0.116), whereas all stages in the KDIGO criteria differed significantly (Fig. 2b). The ability to predict mortality did not differ significantly between the KDIGO and AKIN criteria as assessed by Harrell’s C statistic of the Cox model, although the absolute value of Harrell’s C index was greater for the KDIGO than for the AKIN criteria (Table 7).
123
Clin Exp Nephrol
Discussion We compared the KDIGO and AKIN criteria using a retrospective study of 2579 ICU admissions to investigate the prevalence of AKI and the accuracy of predicting in-hospital mortality among critically ill patients in Japan. The AKIN criteria have some deficiencies. The proposed time frame of 48 h for a diagnosis of progressive AKI might result in missing patients in whom sCr slowly increases after 48 h [33]. When applied to patients undergoing cardiac surgery without correction of the sCr for fluid balance, the AKIN criteria might lead to overdiagnosis of AKI [25]. The KDIGO group recently proposed to adopt the 48-h time window for the 0.3 mg/dL rise in sCr proposed by the AKIN group as a modification to the original RIFLE criteria, keeping the 7-day window for the 50 % increase in sCr from baseline. The proposed baseline used the lowest sCr value recorded within 3 months of an event. The number of patients classified as all stages of AKI significantly increased when selected by the KDIGO criteria with a 7-day window, compared with the AKIN criteria that propose a 48-h window for a diagnosis of AKI [26]. Based on these criteria, 238 patients (9.2 %) who were judged as non-AKI based on the AKIN criteria were diagnosed with AKI based on the KDIGO criteria. The inhospital mortality rate was higher among patients with AKI that slowly progressed within 7 days than among those judged as non-AKI by both criteria (7.1 versus 2.7 %). The trade-off theory states that sensitivity increases as specificity decreases. However, there was no significant difference in Harrell’s C between the two criteria. Thus, the KDIGO criteria might decrease the mortality of patients with AKI without overlooking any AKI due to improved sensitivity without compromising specificity. Progress through the severity of AKI stages according to both criteria accompanied an increase in in-hospital mortality. However, the survival ratio did not significantly differ between AKIN stages I and II. These results are consistent with previous findings, and AKIN staging might be needed to reevaluate stage I and II stratification [34, 35]. Differences in hazard ratios were greater at every stage in the KDIGO than in the AKIN criteria, indicating that the KDIGO criteria had clearer prognostic staging. The KDIGO criteria also resulted in a more effective statistical model than the AKIN criteria, because the KDIGO criteria had a smaller AIC than the AKIN criteria in models with continuous variable, categorical variable, and stepwise variable selection. Therefore, these results suggest that stratification for mortality risk is more effective by the KDIGO than by the AKIN criteria. Our study was of a relatively large cohort, but conducted retrospectively at a single medical center, and surgical patients were the majority in our hospital. Thus,
123
generalization from these findings must be limited. We did not include hourly urine output data during hospitalization, because collecting such information is not routine clinical practice outside the ICU. Collecting retrospective sCr data was easier than determining hourly urine output, and sCr criteria seemed to predict mortality more accurately than urine output. As suggested by a multicenter study of AKI classified by the RIFLE criteria in 2,164 patients in an ICU [36], sCr seemed to predict mortality more effectively than urine output. In[60 % of patients with AKI, the creatinine criteria led to a worse RIFLE class or AKIN stage than urine output [23]. However, urine output data represent an established indicator of AKI [1, 37], and a prospective study should use urine output data according to the KDIGO criteria. In conclusion, the KDIGO criteria, with improved sensitivity without compromised AKI specificity, might predict mortality accurately as well as the AKIN criteria. Acknowledgments This study was supported in part by a Grant-inAid for Progressive Renal Diseases Research, Research on Intractable Disease, from the Ministry of Health, Labour and Welfare of Japan. Conflict of interest interest exists.
The authors have declared that no conflict of
References 1. Mehta RL, Pascual MT, Soroko S, Savage BR, Himmelfarb J, Ikizler TA, Paganini EP, Chertow GM. Spectrum of acute renal failure in the intensive care unit: the PICARD experience. Kidney Int. 2004;66:1613–21. 2. Uchino S, Kellum JA, Bellomo R, Doig GS, Morimatsu H, Morgera S, Schetz M, Tan I, Bouman C, Macedo E, Gibney N, Tolwani A, Ronco C. Acute renal failure in critically ill patients: a multinational, multicenter study. JAMA. 2005;294:813–8. 3. Kellum JA, Levin N, Bouman C, Lameire N. Developing a consensus classification system for acute renal failure. Curr Opin Crit Care. 2002;8:509–14. 4. Bellomo R, Kellum JA, Ronco C. Defining acute renal failure: physiological principles. Intensive Care Med. 2004;30:33–7. 5. Bellomo R, Kellum J, Ronco C. Acute renal failure: time for consensus. Intensive Care Med. 2001;27:1685–8. 6. Hoste EA, Kellum JA. Acute kidney injury: epidemiology and diagnostic criteria. Curr Opin Crit Care. 2006;12:531–7. 7. Singbartl K, Kellum JA. AKI in the ICU: definition, epidemiology, risk stratification, and outcomes. Kidney Int. 2012;81(9):819–25. 8. Bellomo R, Ronco C, Kellum JA, Mehta RL, Palevsky P. Acute renal failure—definition, outcome measures, animal models, fluid therapy and information technology needs: the Second International Consensus Conference of the Acute Dialysis Quality Initiative (ADQI) Group. Crit Care. 2004;8:R204–12. 9. Negi S, Shigematsu T. Current therapeutic strategies for acute kidney injury. Clin Exp Nephrol. 2012;16:672–8. 10. Hoste EA, Clermont G, Kersten A, Venkataraman R, Angus DC, De Bacquer D, Kellum JA. RIFLE criteria for acute kidney injury are associated with hospital mortality in critically ill patients: a cohort analysis. Crit Care. 2006;10:R73.
Clin Exp Nephrol 11. Ostermann M, Chang RW. Acute kidney injury in the intensive care unit according to RIFLE. Crit Care Med. 2007;35:1837–43 (quiz 1852). 12. Bagshaw SM, George C, Dinu I, Bellomo R. A multi-centre evaluation of the RIFLE criteria for early acute kidney injury in critically ill patients. Nephrol Dial Transplant. 2008;23:1203–10. 13. Ricci Z, Cruz D, Ronco C. The RIFLE criteria and mortality in acute kidney injury: a systematic review. Kidney Int. 2008;73:538–46. 14. Clec’h C, Gonzalez F, Lautrette A, Nguile-Makao M, GarrousteOrgeas M, Jamali S, Golgran-Toledano D, Descorps-Declere A, Chemouni F, Hamidfar-Roy R, Azoulay E, Timsit JF. Multiplecenter evaluation of mortality associated with acute kidney injury in critically ill patients: a competing risks analysis. Crit Care. 2011;15(3):R128. 15. Bagshaw SM, Uchino S, Cruz D, Bellomo R, Morimatsu H, Morgera S, Schetz M, Tan I, Bouman C, Macedo E, Gibney N, Tolwani A, Oudemans-van Straaten HM, Ronco C, Kellum JA. A comparison of observed versus estimated baseline creatinine for determination of RIFLE class in patients with acute kidney injury. Nephrol Dial Transplant. 2009;24:2739–44. 16. Siew ED, Matheny ME, Ikizler TA, Lewis JB, Miller RA, Waitman LR, Go AS, Parikh CR, Peterson JF. Commonly used surrogates for baseline renal function affect the classification and prognosis of acute kidney injury. Kidney Int. 2010;77:536–42. 17. Pickering JW, Endre ZH. Back-calculating baseline creatinine with MDRD misclassifies acute kidney injury in the intensive care unit. Clin J Am Soc Nephrol. 2010;5:1165–73. 18. Mehta RL, Kellum JA, Shah SV, Molitoris BA, Ronco C, Warnock DG, Levin A. Acute kidney injury network: report of an initiative to improve outcomes in acute kidney injury. Crit Care. 2007;11:R31. 19. Lassnigg A, Schmidlin D, Mouhieddine M, Bachmann LM, Druml W, Bauer P, Hiesmayr M. Minimal changes of serum creatinine predict prognosis in patients after cardiothoracic surgery: a prospective cohort study. J Am Soc Nephrol. 2004;15:1597–605. 20. Chertow GM, Burdick E, Honour M, Bonventre JV, Bates DW. Acute kidney injury, mortality, length of stay, and costs in hospitalized patients. J Am Soc Nephrol. 2005;16:3365–70. 21. Pickering JW, Endre ZH. GFR shot by RIFLE: errors in staging acute kidney injury. Lancet. 2009;373:1318–9. 22. Bagshaw SM, George C, Bellomo R. A comparison of the RIFLE and AKIN criteria for acute kidney injury in critically ill patients. Nephrol Dial Transplant. 2008;23:1569–74. 23. Lopes JA, Fernandes P, Jorge S, Goncalves S, Alvarez A, Costa e Silva Z, Franca C, Prata MM. Acute kidney injury in intensive care unit patients: a comparison between the RIFLE and the acute kidney injury network classifications. Crit Care. 2008;12:R110.
24. Joannidis M, Metnitz B, Bauer P, Schusterschitz N, Moreno R, Druml W, Metnitz PG. Acute kidney injury in critically ill patients classified by AKIN versus RIFLE using the SAPS 3 database. Intensive Care Med. 2009;35:1692–702. 25. Englberger L, Suri RM, Li Z, Casey ET, Daly RC, Dearani JA, Schaff HV. Clinical accuracy of RIFLE and acute kidney injury network (AKIN) criteria for acute kidney injury in patients undergoing cardiac surgery. Crit Care. 2011;15:R16. 26. Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Work Group. KDIGO clinical practice guideline for acute kidney injury. Kidney Int. 2012;Suppl 2:1–138. 27. Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of disease classification system. Crit Care Med. 1985;13:818–29. 28. Le Gall JR, Lemeshow S, Saulnier F. A new Simplified Acute Physiology Score (SAPS II) based on a European/North American multicenter study. JAMA. 1993;270:2957–63. 29. Vincent JL, Moreno R, Takala J, Willatts S, De Mendonca A, Bruining H, Reinhart CK, Suter PM, Thijs LG. The SOFA (sepsis-related organ failure assessment) score to describe organ dysfunction/failure. On behalf of the Working Group on sepsisrelated problems of the European Society of intensive care medicine. Intensive Care Med. 1996;22:707–10. 30. Akaike H. A new look at the statistical model identification. IEEE Trans Autom Control. 1974;19:716–23. 31. Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med. 1996;15:361–87. 32. Harrell FE Jr, Lee KL, Califf RM, Pryor DB, Rosati RA. Regression modelling strategies for improved prognostic prediction. Stat Med. 1984;3:143–52. 33. Ostermann M, Chang RW. Challenges of defining acute kidney injury. QJM. 2011;104:237–43. 34. Ostermann M, Chang R. Correlation between the AKI classification and outcome. Crit Care. 2008;12:R144. 35. Mandelbaum T, Scott DJ, Lee J, Mark RG, Malhotra A, Waikar S, Howell MD, Talmor DS. Outcome of critically ill patients with acute kidney injury using the acute kidney injury network criteria. Crit Care Med. 2011;39:2659–64. 36. Cruz DN, Bolgan I, Perazella MA, Bonello M, de Cal M, Corradi V, Polanco N, Ocampo C, Nalesso F, Piccinni P, Ronco C. North East Italian Prospective Hospital Renal Outcome Survey on Acute Kidney Injury (NEiPHROS-AKI): targeting the problem with the RIFLE criteria. Clin J Am Soc Nephrol. 2007;2:418–25. 37. Lameire N, Van Biesen W, Vanholder R. Acute renal failure. Lancet. 2005;365:417–30.
123
ORIGINAL ARTICLE
Comparison of Kidney Disease: Improving Global Outcomes and Acute Kidney Injury Network criteria for assessing patients in intensive care units Hibiki Shinjo • Waichi Sato • Enyu Imai • Tomoki Kosugi • Hiroki Hayashi • Kunihiro Nishimura • Kimitoshi Nishiwaki Yukio Yuzawa • Seiichi Matsuo • Shoichi Maruyama
•
Received: 14 June 2013 / Accepted: 11 November 2013 Ó Japanese Society of Nephrology 2013
Abstract Background The Kidney Disease: Improving Global Outcomes (KDIGO) group proposed to adopt the 48-h time window for the 0.3 mg/dL rise in serum creatinine (sCr) proposed by the Acute Kidney Injury Network (AKIN) group as a modification to the original risk, injury, failure, loss, and end-stage renal disease criteria, keeping the 7-day window for the 50 % increase in sCr from baseline. The present study evaluates the prevalence of acute kidney injury (AKI) and the accuracy of predicting mortality based on the KDIGO and AKIN criteria. Patients and methods We retrospectively studied a cohort of 2579 patients admitted to the intensive care unit of Nagoya University Hospital between 2005 and 2009. Results The total AKI prevalence was higher according to the KDIGO than to the AKIN criteria (38.4 versus 29.5 %). In-hospital mortality rates were higher among 238 patients classified as non-AKI by the AKIN but AKI by the KDIGO criteria than among those classified as non-AKI by H. Shinjo W. Sato (&) E. Imai T. Kosugi S. Matsuo S. Maruyama Departments of Nephrology, Nagoya University Graduate School of Medicine, 65 Tsurumai, Showa-ku, Nagoya, Aichi 466-8550, Japan e-mail: [email protected] H. Hayashi Y. Yuzawa Department of Nephrology, Fujita Health University School of Medicine, Toyoake, Japan K. Nishimura Department of Preventive Medicine and Epidemiology, National Cerebral and Cardiovascular Center Hospital, Suita, Japan K. Nishiwaki Departments of Anesthesiology, Nagoya University Graduate School of Medicine, Nagoya, Japan
both criteria (7.1 versus 2.7 %). Survival curves generated using KDIGO significantly differed among all stages, but not between AKIN stages I and II. Multivariate analysis showed that KDIGO criteria were better in a statistical model than the AKIN criteria according to the Akaike information criterion. Harrell’s C statistic was greater for the KDIGO than for the AKIN criteria. Conclusions The KDIGO criteria have improved sensitivity without compromising specificity for AKI and might predict mortality at least as well as the AKIN criteria. Keywords Acute kidney injury (AKI) Acute renal failure (ARF) Mortality Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Network (AKIN)
Introduction Acute renal failure (ARF) causes uremia and various electrolyte and fluid abnormalities, and it is defined as a rapid loss of kidney function over a period ranging from hours to days. ARF is considered to increase the mortality of critically ill patients [1, 2]. However, the wide variation in ARF definitions has caused difficulties in comparing results across studies and populations [3–7]. The Acute Dialysis Quality Initiative group proposed the risk, injury, failure, loss of kidney function, and end-stage kidney disease (RIFLE) classification in 2004 for acute kidney injury (AKI) [8], which is a frequent complication among critically ill patients admitted to the intensive care unit (ICU) [9]. These criteria for AKI were associated with mortality [10–14]. The RIFLE criteria included a recommendation to assume normal baseline renal function [backcalculation using modification of diet in renal disease
123
Clin Exp Nephrol
(MDRD) assuming a glomerular filtration rate (GFR) of 75 mL/min/1.73 m2] in the absence of baseline serum creatinine (sCr) values. However, such an assumption might overestimate the proportion of patients with AKI by ignoring those with extant chronic kidney disease (CKD) [15–17]. The Acute Kidney Injury Network (AKIN) group modified the RIFLE criteria in 2007 from five viewpoints [18]. The AKIN criteria proposed a time frame of 48 h for a diagnosis of AKI. AKIN stage I (risk in the RIFLE criteria) includes a slight increase of sCr of C0.3 mg/dL (C26.4 lmol/L) [19, 20]. A requirement for renal replacement therapy (RRT) was classified as AKIN stage III (failure in the RIFLE criteria). Outcome categories [loss and end-stage kidney disease (ESKD) in the RIFLE criteria] were eliminated and changes in GFR were excluded because increases in sCr and decreases in GFR in the definitions of RIFLE category (risk and failure) do not correspond mathematically. The risk category is defined as a 1.5-fold increase in sCr and a 25 % decrease in GFR, but a 1.5-fold increase in sCr corresponds to a one-third decrease in GFR. The failure category is defined as a threefold increase and a 75 % decrease, but a threefold increase corresponds to a decrease of two-thirds [21]. Compared with the RIFLE criteria, the AKIN criteria improved the sensitivity of AKI diagnosis, but did not significantly improve predictions of in-hospital mortality rates among critically ill patients [22–25]. In addition, the AKIN criteria did not improve sensitivity, robustness or predictive ability, because the introduction of a 48-h window for a diagnosis of AKI might miss patients with a slow increase in sCr that could not be identified within 48 h. The Kidney Disease: Improving Global Outcomes (KDIGO) working group modified the AKI criteria in 2012 [26]. This group proposed to adopt the 48-h time window for the 0.3 mg/dL rise in sCr proposed by the AKIN group as a modification to the original RIFLE criteria, keeping the 7-day window for the 50 % increase in sCr from baseline. The proposed baseline used the lowest sCr value recorded within 3 months of an event. The present study evaluates the prevalence of AKI and the accuracy of predicting in-hospital mortality among critically ill patients using the KDIGO and AKIN criteria for comparison.
Patients and methods Patients and procedures The Ethics Committee of Nagoya University School of Medicine approved this study (approval number 1), which proceeded according to the principles of the Declaration of Helsinki. Patients’ data were extracted from ICU and
123
Fig. 1 Study flowchart. Among 2579 critically ill patients admitted to ICU, we excluded those with ESKD on chronic hemodialysis or peritoneal dialysis, those who had undergone kidney transplant, those \16 years old, and those who spent\1 day in ICU. AKI acute kidney injury, AKIN Acute Kidney Injury Network, ESKD end-stage kidney disease, ICU intensive care unit, KDIGO Kidney Disease: Improving Global Outcomes
electronic hospital records. Laboratory data were extracted from our laboratory database. We matched data from the different sources and then manually organized the information. We retrospectively investigated a cohort of all adults moved to the ICU after admission to Nagoya University Hospital (n = 3350) between 27 May 2005 and 31 May 2009. We excluded patients with end-stage kidney disease (ESKD) on chronic hemodialysis, peritoneal dialysis, kidney transplants (n = 78), those younger than 16 years (n = 154) or who had stayed in the ICU for \1 day (n = 178), and considered only the first admission when patients were readmitted to the ICU during the study period (n = 360) (Fig. 1). We finally acquired data from 2579 patients. The end point of the study was all-cause (in-hospital) mortality. Systems for classifying disease severity The Acute Physiology and Chronic Health Evaluation (APACHE) II, Sequential Organ Failure Assessment (SOFA), and Simplified Acute Physiology Score (SAPS) II scores were calculated based on variables recorded during the first 24 h of ICU admission [27–29]. Nonrenal scores were calculated from the total score minus scores for renal parameters. Definitions of AKI We defined AKI according to the KDIGO and AKIN criteria, using sCr values during hospitalization. The
Clin Exp Nephrol Table 1 Comparison of Kidney Disease: Improving Global Outcomes, Acute Kidney Injury Network, and risk, injury, failure, loss of kidney function, and end-stage kidney disease criteria for acute kidney injury Diagnostic criteria for AKI: Abrupt reduction in kidney function defined as: Absolute increase in serum creatinine of either C0.3 mg/dL (KDIGO and AKIN criteria; within 48 h) or C50 % increase in serum creatinine within 48 h (AKIN criteria) or 7 days (KDIGO and RIFLE criteria) or reduced urine output KDIGO
Serum creatinine criteria
Urine output criteria
Stage 1
Increase to C150–199 % from baseline or increase in serum creatinine C0.3 mg/dL
\0.5 mL/kg/h for C6 h
Stage 2
Increase in serum creatinine to C200–299 % from baseline
\0.5 mL/kg/h for C12 h
Stage 3
Increase in serum creatinine to C300 % from baseline or serum creatinine C4.0 mg/ dL or initiation of renal replacement therapy or in patients \18 years, decrease in eGFR to \35 mL/min per 1.73 m2
\0.3 mL/kg/h for C24 h or anuria for C12 h
AKIN
Serum creatinine criteria
Urine output criteria
Stage I
Increase to C150–200 % from baseline or increase in serum creatinine C0.3 mg/dL
\0.5 mL/kg/h for C6 h
Stage II
Increase in serum creatinine to [200–300 % from baseline
\0.5 mL/kg/h for C12 h
Stage III
Increase in serum creatinine to[300 % from baseline or serum creatinine C4.0 mg/ dL with an acute increase of at least 0.5 mg/dL or initiation of renal replacement therapy
\0.3 mL/kg/h for C24 h or anuria for C12 h
RIFLE
GFR criteria
Urine output criteria
Risk
Serum creatinine 9 1.5 or GFR decrease [25 %
\0.5 mL/kg/h for C6 h
Injury
Serum creatinine 9 2 or GFR decrease [50 %
\0.5 mL/kg/h for C12 h
Failure
Serum creatinine 9 3 or GFR decrease [75 % or serum creatinine C4.0 mg/dL with an acute increase of at least 0.5 mg/dL
\0.3 mL/kg/h for C24 h or anuria for C12 h
Loss
Persistent acute renal failure = complete loss of kidney function [4 weeks
ESKD
End-stage kidney disease [3 months
AKI acute kidney injury, KDIGO Kidney Disease: Improving Global Outcomes, AKIN Acute Kidney Injury Network, RIFLE risk, injury, failure, loss of kidney function, and end-stage kidney disease, GFR glomerular filtration rate, ESKD end-stage kidney disease
diagnosis of AKI for all the patients was analyzed using both criteria, that is, an abrupt (within 48 h) reduction in kidney function currently defined as an absolute increase in sCr of C0.3 mg/dL or a reduction in urine output (documented oliguria of \0.5 mL/kg/h for C6 h). In addition, AKI was defined as a C50 % (1.5-fold from baseline) increase in sCr within 48 h by the AKIN criteria and within 7 days by the KDIGO criteria. The AKIN criteria strictly applied a moving window of 48 h [25]. The KDIGO criteria used the lowest value during the 3 months before ICU admission as baseline sCr (Table 1). Due to the absence of information about 6- or 12-h urine volumes, we determined KDIGO and AKIN stages using only sCr criteria. Statistical analysis Continuous variables are expressed as mean ± standard deviation (SD) or as median [interquartile range (IQR)]. Means were compared using Student’s t test when normally distributed, and the Mann–Whitney U test otherwise. Categorical variables were analyzed using the Chi-square test and are presented as ratios (%) of number of patients. To
correct for differences in patient characteristics, we included age, gender, reason for ICU admission, type of ICU admission, nonrenal SOFA score, duration of mechanical ventilation (MV), and AKIN or KDIGO stage. The nonrenal SOFA score was selected as a covariate to control for multicollinearity between the AKI classification and scoring systems that included values for kidney insufficiency such as the APACHE II and SOFA scores. The statistical significance of differences in the prevalence of AKI between KDIGO and AKIN stages was assessed using McNemar’s test. Associations between the AKIN and KDIGO criteria with mortality were evaluated using univariate and multivariate Cox proportional hazards analysis. The Akaike information criterion (AIC) is an index for evaluating the merit of a statistical model and was used to balance the complexity of models and the goodness of fit with the data. A model with lower AIC is better balanced [30]. To compare the predictability of mortality between two criteria, discrimination was also evaluated using Harrell’s C index. A model with perfect predictive capacity (sensitivity and specificity of 100 %) has a Harrell’s C statistic of 1.00; Harrell’s C index is a statistic similar to
123
Clin Exp Nephrol Table 2 Baseline characteristics of patients Entire cohort (n = 2579)
Hospital survivors (n = 2397)
Hospital nonsurvivors (n = 182)
P value*
\0.001
Demographics Age, years
63.2 ± 13.8
62.9 ± 13.8
67.6 ± 12.8
Male sex
1708 (66.2 %)
1598 (66.7 %)
110 (60.4 %)
Body mass index, kg/m2
22.2 ± 3.7
22.2 ± 3.6
22.1 ± 4.1
0.666
475 (18.4 %)
438 (18.3 %)
37 (20.3 %)
0.490
0.70 (0.56–0.86) 82.3 (64.1–100.3)
0.70 (0.56–0.85) 83.0 (65.1–100.3)
0.79 (0.58–1.1) 68.7 (45.3–100.6)
0.001 \0.001 \0.001
Diabetes mellitus
0.087
Baseline renal function Baseline serum creatinine, mg/dL Baseline eGFR, mL/min/1.73 m2 C90
990 (38.4 %)
934 (39.0 %)
56 (30.8 %)
60–89
1064 (41.3 %)
1015 (42.3 %)
49 (26.9 %)
30–59
422 (16.4 %)
366 (15.3 %)
56 (30.8 %)
15–29
71 (2.8 %)
58 (2.4 %)
13 (7.1 %)
32 (1.2 %)
24 (1.0 %)
8 (4.4 %)
Elective surgery
1917 (74.3 %)
1876 (78.3 %)
41 (22.5 %)
Emergency surgery
322 (12.5 %)
295 (12.3 %)
27 (14.8 %)
Nonsurgical
340 (13.2 %)
226 (9.4 %)
114 (62.6 %)
Cardiovascular
993 (38.5 %)
947 (39.5 %)
46 (25.3 %)
Gastrointestinal
64 (2.5 %)
61 (2.5 %)
3 (1.6 %)
Hepatic
79 (3.1 %)
73 (3.0 %)
6 (3.3 %)
Malignancy Neurological
964 (37.4 %) 261 (10.1 %)
912 (38.0 %) 234 (9.8 %)
52 (28.6 %) 27 (14.8 %)
Pulmonary
90 (3.5 %)
61 (2.5 %)
29 (15.9 %)
Other
128 (5.0 %)
109 (4.5 %)
19 (10.4 %)
4 (2–6)
4 (2–6)
7 (5–10)
\15
\0.001
ICU admission type
\0.001
Reason for ICU admission
Risk adjustment SOFA
\0.001
SOFA nonrenal score
4 (2–6)
4 (2–6)
7 (5–10)
\0.001
APACHE II
9 (7–12)
9 (7–12)
19 (13–24)
\0.001
APACHE II nonrenal score
9 (6–11)
8 (6–11)
17 (12–24)
\0.001
SAPS II
28 (23–35)
27 (22–34)
51 (39–63)
\0.001
SAPS II nonrenal score
27 (22–32)
26 (21–31)
44 (34–54)
\0.001
Renal replacement therapy
80 (3.1 %)
53 (2.2 %)
27 (14.8 %)
\0.001
Hospital length of stay (days)
31 (21–52)
31 (21–51)
24 (8–59)
\0.001
ICU length of stay (days)
1 (1–2)
1 (1–2)
3 (1–10)
\0.001
Duration of mechanical ventilation (days)
1 (0–1)
1 (0–1)
5 (1–17)
\0.001 \0.001
None
226 (8.8 %)
214 (8.9 %)
12 (6.6 %)
0–1
1828 (70.9 %)
1777 (74.1 %)
51 (28.0 %)
2–6
305 (11.8 %)
261 (10.9 %)
44 (24.2 %)
C7
220 (8.5 %)
145 (6.0 %)
75 (41.2 %)
Continuous variables presented as mean ± standard deviation when normally distributed or as median (interquartile interval) when not normally distributed. Categorical variables are presented as number (percentage) Nonrenal score was calculated from total score minus renal parameter scores APACHE Acute Physiology and Chronic Health Evaluation, eGFR estimated glomerular filtration rate, ICU intensive care unit, SOFA Sequential Organ Failure Assessment, SAPS Simplified Acute Physiology Score * P for hospital survivors compared with nonsurvivors
123
Clin Exp Nephrol Table 3 Prevalence and in-hospital mortality of acute kidney injury stratified by the Kidney Disease: Improving Global Outcomes and Acute Kidney Injury Network criteria Prevalence, n (%)
Table 4 Cross-tabulation of patients classified by Kidney Disease: Improving Global Outcomes versus Acute Kidney Injury Network criteria
In-hospital mortality, n (%)
AKIN Non-AKI
KDIGO Non-AKI
1588 (61.6)
44 (2.8)
Stage 1 Stage 2
639 (24.8) 159 (6.2)
44 (6.9) 28 (17.6)
Stage 3
193 (7.5)
66 (34.2)
All AKI
991 (38.4)
138 (13.9)
1818 (70.5)
60 (3.3)
535 (20.7)
55 (10.3)
Stage I Stage II
87 (3.4)
18 (20.7)
Stage III
139 (5.4)
49 (35.3)
All AKI
761 (29.5)
122 (16.0)
Patient baseline characteristics and univariate analysis of hospital survivors and nonsurvivors During the study period, 3350 critically ill patients were admitted to the ICU; data from 2579 (77.0 %) patients were analyzed (Fig. 1). The median duration of followup after ICU admission was 2.72 years (0.55–4.08 years, IQR). Table 2 summarizes the baseline characteristics of the patients [mean age (±SD) 63.2 (13.8) years; male 66.2 %]. The percentages of patients with diabetes mellitus, chronic kidney disease, and who underwent elective surgery were 18.4, 20.4, and 74.3 %, respectively. Table 2 also compares the baseline characteristics and physiological variables between in-hospital survivors and nonsurvivors. Values of mean age and median baseline sCr were higher for nonsurvivors. Nonsurvivors were more likely to be admitted for nonsurgical reasons
Total (KDIGO)
n
1580
8
0
0
1588
a
2.7 %
12.5 %
0%
0%
2.8 %
Stage 1 n
208
421
9
1
639
a
5.8 %
7.1 %
22.2 %
0%
6.9 %
Stage 2 n
23
85
49
2
159
a
13.0 %
16.5 %
20.4 %
50.0 %
17.6 %
Stage 3 n
7
21
29
136
193
a
28.6 %
47.6 %
20.7 %
35.3 %
34.2 %
535
87
139
2579
10.3 %
20.7 %
35.3 %
7.1 %
Total (AKIN) n 1818
Results
Stage III
Non-AKI
AKI acute kidney injury, AKIN Acute Kidney Injury Network, KDIGO Kidney Disease: Improving Global Outcomes
the area under the receiver operating characteristic curve [31, 32]. A two-tailed P value of \0.05 was considered significant. Hospital survival across groups was analyzed using the Chi-square and Kaplan–Meier methods, and differences between groups were tested using the log-rank test. We used STATA version 11 (Stata Corp LP, College Station, TX, USA) to evaluate survival data and diagnostic accuracy from receiver operating characteristic (ROC) curves and Harrell’s C index, and all other data were statistically analyzed using SPSS (BM SPSS statistics 19) software.
Stage II
KDIGO
AKIN Non-AKI
Stage I
a
3.3 %
Italics indicate patients judged as same degree of AKI by both classification systems AKI acute kidney injury, AKIN Acute Kidney Injury Network, KDIGO Kidney Disease: Improving Global Outcomes, n number of patients a
In-hospital mortality rate (%) of the respective group
and were more critically ill, and a higher proportion remained on mechanical ventilation for an extended period. Prevalence of AKI, in-hospital mortality, and crosstabulation according to AKI stratified by KDIGO and AKIN criteria The prevalence of AKI according to the KDIGO and AKIN criteria was 991 (38.4 %) and 761 (29.5 %), respectively. Among the 991 patients defined by the KDIGO criteria as AKI, 24.8, 6.2, and 7.5 % had stage 1, 2, and 3, respectively. Among the 761 patients defined by the AKIN criteria, 20.7, 3.4, and 5.4 % had stage I, II, and III, respectively (Table 3). The KDIGO criteria significantly increased the number of patients judged as AKI in all stages compared with the AKIN criteria (P \ 0.001). Table 4 cross-tabulates the stages classified by the KDIGO and AKIN criteria. Two hundred and thirty-eight (9.2 %) patients were classified as AKI by the KDIGO criteria but as non-AKI by the AKIN criteria, and 87 % of them were classified as KDIGO stage 1. Serum creatinine increased in this group from within 48 h to 7 days. These patients had
123
Clin Exp Nephrol Table 5 Risk factors for long-term mortality in patients (univariate analysis) HR
95 % CI
P value
Age
1.03
1.02–1.04
\0.001
Male sex
0.77
0.57–1.03
0.081
Body mass index
0.99
0.94–1.04
0.703
Diabetes mellitus
1.14
0.78–1.66
0.46
CKD
3.19
2.38–4.28
\0.001
1 0.83
Reference 0.56–1.12
0.329 \0.001
about 2.5-fold higher in-hospital mortality rates than those without AKI according to both criteria (7.1 versus 2.7 %, respectively), and were similar for KDIGO stage 1 and AKIN stage I (7.1 %). On the other hand, the in-hospital mortality rate was relatively high (12.5 %) among patients who were classified as non-AKI by the KDIGO criteria but AKI by the AKIN criteria. However, it was difficult to draw any conclusion since there were only 8 (0.3 %) patients in this group.
eGFR C90 60–89 30–59
2.6
1.79–3.77
15–29
3.71
2.03–6.78
\0.001
5.47
2.39–12.93
\0.001
Elective surgery
1
Reference
Emergency surgery
4.19
\15 ICU admission type
Nonsurgical
23.08
2.54–6.91
\0.001
15.74–33.84
\0.001
Reason for ICU admission Cardiovascular
1
Reference
Gastrointestinal
1.01
0.31–3.16
0.977
Hepatic
1.62
0.69–3.79
0.266
Malignancy
1.1
0.74–1.63
0.653
Neurological
2.27
1.41–3.65
0.001
Pulmonary Other
8.25 3.47
5.18–13.14 2.04–5.93
\0.001 \0.001
1.41
1.34–1.47
\0.001
Risk adjustment SOFA nonrenal score APACHE II nonrenal score
1.22
1.19–1.24
\0.001
SAPS II nonrenal score
1.11
1.12–1.16
\0.001
Duration of mechanical ventilation (days) None
1
Reference
0–1
0.51
0.27–0.99
2–6
2.94
1.55–5.56
C7
7.14
3.88–13.13
\0.001
2.44
2.16–2.74
\0.001
KDIGO (per stage)
0.038 0.001
Non-AKI
1
Reference
Stage 1
2.52
1.66–3.82
\0.001
Stage 2
6.64
4.14–10.67
\0.001
Stage 3
14.42
9.84–21.13
\0.001
2.32 1
2.06–2.61 Reference
\0.001
Stage I
3.22
2.24–4.65
\0.001
Stage II
6.78
4.00–11.48
\0.001
Stage III
12.83
8.79–18.73
\0.001
AKIN (per stage) Non-AKI
AKI acute kidney injury, APACHE Acute Physiology and Chronic Health Evaluation, AKIN Acute Kidney Injury Network, CI confidence interval, CKD chronic kidney disease, HR hazard ratio, ICU intensive care unit, KDIGO Kidney Disease: Improving Global Outcomes, SAPS Simplified Acute Physiology Score, SOFA Sequential Organ Failure Assessment
123
Univariate analysis for mortality judged by the KDIGO and AKIN criteria Univariate Cox proportional hazards models revealed the following as risk factors for increased mortality: age, chronic kidney disease, emergency surgery, nonsurgical admission, neurological admission, pulmonary admission, SOFA, APACHE II, SAPS II, prolonged mechanical ventilation, and KDIGO and AKIN stages (Table 5). Nonsurgical patients had high hazard ratio, because these patients had higher severity scores (SOFA, APACHE II, and SAPS II) in ICU than surgical patients. Risk factors for mortality using a multivariate Cox proportional hazards model and survival curves within KDIGO and AKIN stages We analyzed risk factors for mortality using multivariate Cox proportional hazards models 1, 2, and 3 selected by the Akaike information criterion (AIC) in which KDIGO and AKIN stages were considered as continuous variable, categorical variable, and stepwise variable selection, respectively. After adjustment for covariates [age, sex, body mass index (BMI), diabetes mellitus, CKD, ICU admission type, reason for ICU admission, nonrenal SOFA score, and duration of mechanical ventilation], the KDIGO and AKIN stages remained independently associated with mortality. Age, BMI, neurological and pulmonary patients, nonsurgical patients, emergency surgical patients, and nonrenal SOFA score were considered significant. Hazard ratios (HRs) and 95 % CIs for mortality versus non-AKI increased at every stage in models 2 and 3. Comparing the goodness of fit and complexity of the three models by AIC, the KDIGO criteria represented a more accurate predictor than the AKIN criteria (AIC value in KDIGO versus AKIN: model 1, 1181.03 versus 1191.07; model 2, 1183.97 versus 1193.31; model 3, 1175.65 versus 1184.64; Table 6). During follow-up after ICU admission, 417 (16.2 %) patients died. Figure 2a shows the survival curves for all patients with all stages of AKI according to the AKIN criteria. Stages I and II did not differ significantly
Clin Exp Nephrol Table 6 Predictive ability for long-term mortality by multivariate Cox proportional hazards models for Kidney Disease: Improving Global Outcomes and Acute Kidney Injury Network criteria
Model 1 Model 2
Model 3
KDIGO
HR
95 % CI
P value
AIC
AKIN
HR
95 % CI
P value
AIC
Per stage
1.69
1.37–2.09
\0.001
1181.03
Per stage
1.52
1.22–1.90
\0.001
1191.07
1183.97
Non-AKI
1
Reference
Stage I
2.08
1.19–3.63
Non-AKI
1
Reference
Stage 1
2.31
1.19–4.43
Stage 2
3.61
1.70–7.68
Stage 3
5.48
2.69–11.17
Non-AKI
1
Reference
Stage 1 Stage 2
2.46 3.94
1.12–3.75 1.88–8.27
Stage 3
5.98
2.99–11.94
0.012 0.001 \0.001 1175.65 0.012 0.001 \0.001
1193.31 0.011
Stage II
3.16
1.43–7.00
0.005
Stage III
3.71
1.82–7.57
\0.001
Non-AKI
1
Reference
Stage I Stage II
2.15 3.08
1.24–3.73 1.40–6.77
0.006 0.005
Stage III
3.92
1.99–7.71
\0.001
1184.64
Multivariate cox proportional hazards model adjusted for age, sex, BMI, diabetes mellitus, CKD, ICU admission type, reason for ICU admission, SOFA nonrenal score, and duration of mechanical ventilation. Model 1: KDIGO and AKIN stage considered as continuous variable. Model 2: KDIGO and AKIN stage considered as categorical variable. Model 3: KDIGO and AKIN stage considered as stepwise variable selection, by Akaike information criterion. Age, BMI, neurological and pulmonary cases, nonsurgical case, emergency surgery case, and SOFA nonrenal score were considered significant. Smaller AIC model was considered as a better prediction model AKIN Acute Kidney Injury Network, CI confidence interval, CKD chronic kidney disease, ICU intensive care unit, KDIGO Kidney Disease: Improving Global Outcomes, HR hazard ratio, AIC Akaike information criterion, SOFA Sequential Organ Failure Assessment
Table 7 Comparison between Kidney Disease: Improving Global Outcomes and Acute Kidney Injury Network criteria by Harrell’s C statistic of Cox model prediction Harrell’s C
P value
95 % CI
AKIN stage (continuous)
0.905
\0.001
0.873 to 0.936
KDIGO stage (continuous)
0.912
\0.001
-0.007
0.149
Model 1
Difference
0.882 to 0.941 -0.017 to 0.003
Model 2 AKIN stage (categorical)
0.908
\0.001
KDIGO stage (categorical)
0.915
\0.001
Difference
0.007
0.249
0.877 to 0.938 0.887 to 0.942 -0.005 to 0.018
Model 3 AKIN stage
0.908
\0.001
KDIGO stage
0.915
\0.001
Difference
0.007
0.249
0.877 to 0.938 0.887 to 0.942 -0.005 to 0.018
AKIN Acute Kidney Injury Network, KDIGO Kidney Disease: Improving Global Outcomes, CI confidence interval
Fig. 2 Long-term survival of patients classified according to Acute Kidney Injury Network (a) or Kidney Disease: Improving Global Outcomes criteria (b). AKI acute kidney injury, AKIN Acute Kidney Injury Network, KDIGO Kidney Disease: Improving Global Outcomes
(P = 0.116), whereas all stages in the KDIGO criteria differed significantly (Fig. 2b). The ability to predict mortality did not differ significantly between the KDIGO and AKIN criteria as assessed by Harrell’s C statistic of the Cox model, although the absolute value of Harrell’s C index was greater for the KDIGO than for the AKIN criteria (Table 7).
123
Clin Exp Nephrol
Discussion We compared the KDIGO and AKIN criteria using a retrospective study of 2579 ICU admissions to investigate the prevalence of AKI and the accuracy of predicting in-hospital mortality among critically ill patients in Japan. The AKIN criteria have some deficiencies. The proposed time frame of 48 h for a diagnosis of progressive AKI might result in missing patients in whom sCr slowly increases after 48 h [33]. When applied to patients undergoing cardiac surgery without correction of the sCr for fluid balance, the AKIN criteria might lead to overdiagnosis of AKI [25]. The KDIGO group recently proposed to adopt the 48-h time window for the 0.3 mg/dL rise in sCr proposed by the AKIN group as a modification to the original RIFLE criteria, keeping the 7-day window for the 50 % increase in sCr from baseline. The proposed baseline used the lowest sCr value recorded within 3 months of an event. The number of patients classified as all stages of AKI significantly increased when selected by the KDIGO criteria with a 7-day window, compared with the AKIN criteria that propose a 48-h window for a diagnosis of AKI [26]. Based on these criteria, 238 patients (9.2 %) who were judged as non-AKI based on the AKIN criteria were diagnosed with AKI based on the KDIGO criteria. The inhospital mortality rate was higher among patients with AKI that slowly progressed within 7 days than among those judged as non-AKI by both criteria (7.1 versus 2.7 %). The trade-off theory states that sensitivity increases as specificity decreases. However, there was no significant difference in Harrell’s C between the two criteria. Thus, the KDIGO criteria might decrease the mortality of patients with AKI without overlooking any AKI due to improved sensitivity without compromising specificity. Progress through the severity of AKI stages according to both criteria accompanied an increase in in-hospital mortality. However, the survival ratio did not significantly differ between AKIN stages I and II. These results are consistent with previous findings, and AKIN staging might be needed to reevaluate stage I and II stratification [34, 35]. Differences in hazard ratios were greater at every stage in the KDIGO than in the AKIN criteria, indicating that the KDIGO criteria had clearer prognostic staging. The KDIGO criteria also resulted in a more effective statistical model than the AKIN criteria, because the KDIGO criteria had a smaller AIC than the AKIN criteria in models with continuous variable, categorical variable, and stepwise variable selection. Therefore, these results suggest that stratification for mortality risk is more effective by the KDIGO than by the AKIN criteria. Our study was of a relatively large cohort, but conducted retrospectively at a single medical center, and surgical patients were the majority in our hospital. Thus,
123
generalization from these findings must be limited. We did not include hourly urine output data during hospitalization, because collecting such information is not routine clinical practice outside the ICU. Collecting retrospective sCr data was easier than determining hourly urine output, and sCr criteria seemed to predict mortality more accurately than urine output. As suggested by a multicenter study of AKI classified by the RIFLE criteria in 2,164 patients in an ICU [36], sCr seemed to predict mortality more effectively than urine output. In[60 % of patients with AKI, the creatinine criteria led to a worse RIFLE class or AKIN stage than urine output [23]. However, urine output data represent an established indicator of AKI [1, 37], and a prospective study should use urine output data according to the KDIGO criteria. In conclusion, the KDIGO criteria, with improved sensitivity without compromised AKI specificity, might predict mortality accurately as well as the AKIN criteria. Acknowledgments This study was supported in part by a Grant-inAid for Progressive Renal Diseases Research, Research on Intractable Disease, from the Ministry of Health, Labour and Welfare of Japan. Conflict of interest interest exists.
The authors have declared that no conflict of
References 1. Mehta RL, Pascual MT, Soroko S, Savage BR, Himmelfarb J, Ikizler TA, Paganini EP, Chertow GM. Spectrum of acute renal failure in the intensive care unit: the PICARD experience. Kidney Int. 2004;66:1613–21. 2. Uchino S, Kellum JA, Bellomo R, Doig GS, Morimatsu H, Morgera S, Schetz M, Tan I, Bouman C, Macedo E, Gibney N, Tolwani A, Ronco C. Acute renal failure in critically ill patients: a multinational, multicenter study. JAMA. 2005;294:813–8. 3. Kellum JA, Levin N, Bouman C, Lameire N. Developing a consensus classification system for acute renal failure. Curr Opin Crit Care. 2002;8:509–14. 4. Bellomo R, Kellum JA, Ronco C. Defining acute renal failure: physiological principles. Intensive Care Med. 2004;30:33–7. 5. Bellomo R, Kellum J, Ronco C. Acute renal failure: time for consensus. Intensive Care Med. 2001;27:1685–8. 6. Hoste EA, Kellum JA. Acute kidney injury: epidemiology and diagnostic criteria. Curr Opin Crit Care. 2006;12:531–7. 7. Singbartl K, Kellum JA. AKI in the ICU: definition, epidemiology, risk stratification, and outcomes. Kidney Int. 2012;81(9):819–25. 8. Bellomo R, Ronco C, Kellum JA, Mehta RL, Palevsky P. Acute renal failure—definition, outcome measures, animal models, fluid therapy and information technology needs: the Second International Consensus Conference of the Acute Dialysis Quality Initiative (ADQI) Group. Crit Care. 2004;8:R204–12. 9. Negi S, Shigematsu T. Current therapeutic strategies for acute kidney injury. Clin Exp Nephrol. 2012;16:672–8. 10. Hoste EA, Clermont G, Kersten A, Venkataraman R, Angus DC, De Bacquer D, Kellum JA. RIFLE criteria for acute kidney injury are associated with hospital mortality in critically ill patients: a cohort analysis. Crit Care. 2006;10:R73.
Clin Exp Nephrol 11. Ostermann M, Chang RW. Acute kidney injury in the intensive care unit according to RIFLE. Crit Care Med. 2007;35:1837–43 (quiz 1852). 12. Bagshaw SM, George C, Dinu I, Bellomo R. A multi-centre evaluation of the RIFLE criteria for early acute kidney injury in critically ill patients. Nephrol Dial Transplant. 2008;23:1203–10. 13. Ricci Z, Cruz D, Ronco C. The RIFLE criteria and mortality in acute kidney injury: a systematic review. Kidney Int. 2008;73:538–46. 14. Clec’h C, Gonzalez F, Lautrette A, Nguile-Makao M, GarrousteOrgeas M, Jamali S, Golgran-Toledano D, Descorps-Declere A, Chemouni F, Hamidfar-Roy R, Azoulay E, Timsit JF. Multiplecenter evaluation of mortality associated with acute kidney injury in critically ill patients: a competing risks analysis. Crit Care. 2011;15(3):R128. 15. Bagshaw SM, Uchino S, Cruz D, Bellomo R, Morimatsu H, Morgera S, Schetz M, Tan I, Bouman C, Macedo E, Gibney N, Tolwani A, Oudemans-van Straaten HM, Ronco C, Kellum JA. A comparison of observed versus estimated baseline creatinine for determination of RIFLE class in patients with acute kidney injury. Nephrol Dial Transplant. 2009;24:2739–44. 16. Siew ED, Matheny ME, Ikizler TA, Lewis JB, Miller RA, Waitman LR, Go AS, Parikh CR, Peterson JF. Commonly used surrogates for baseline renal function affect the classification and prognosis of acute kidney injury. Kidney Int. 2010;77:536–42. 17. Pickering JW, Endre ZH. Back-calculating baseline creatinine with MDRD misclassifies acute kidney injury in the intensive care unit. Clin J Am Soc Nephrol. 2010;5:1165–73. 18. Mehta RL, Kellum JA, Shah SV, Molitoris BA, Ronco C, Warnock DG, Levin A. Acute kidney injury network: report of an initiative to improve outcomes in acute kidney injury. Crit Care. 2007;11:R31. 19. Lassnigg A, Schmidlin D, Mouhieddine M, Bachmann LM, Druml W, Bauer P, Hiesmayr M. Minimal changes of serum creatinine predict prognosis in patients after cardiothoracic surgery: a prospective cohort study. J Am Soc Nephrol. 2004;15:1597–605. 20. Chertow GM, Burdick E, Honour M, Bonventre JV, Bates DW. Acute kidney injury, mortality, length of stay, and costs in hospitalized patients. J Am Soc Nephrol. 2005;16:3365–70. 21. Pickering JW, Endre ZH. GFR shot by RIFLE: errors in staging acute kidney injury. Lancet. 2009;373:1318–9. 22. Bagshaw SM, George C, Bellomo R. A comparison of the RIFLE and AKIN criteria for acute kidney injury in critically ill patients. Nephrol Dial Transplant. 2008;23:1569–74. 23. Lopes JA, Fernandes P, Jorge S, Goncalves S, Alvarez A, Costa e Silva Z, Franca C, Prata MM. Acute kidney injury in intensive care unit patients: a comparison between the RIFLE and the acute kidney injury network classifications. Crit Care. 2008;12:R110.
24. Joannidis M, Metnitz B, Bauer P, Schusterschitz N, Moreno R, Druml W, Metnitz PG. Acute kidney injury in critically ill patients classified by AKIN versus RIFLE using the SAPS 3 database. Intensive Care Med. 2009;35:1692–702. 25. Englberger L, Suri RM, Li Z, Casey ET, Daly RC, Dearani JA, Schaff HV. Clinical accuracy of RIFLE and acute kidney injury network (AKIN) criteria for acute kidney injury in patients undergoing cardiac surgery. Crit Care. 2011;15:R16. 26. Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Work Group. KDIGO clinical practice guideline for acute kidney injury. Kidney Int. 2012;Suppl 2:1–138. 27. Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of disease classification system. Crit Care Med. 1985;13:818–29. 28. Le Gall JR, Lemeshow S, Saulnier F. A new Simplified Acute Physiology Score (SAPS II) based on a European/North American multicenter study. JAMA. 1993;270:2957–63. 29. Vincent JL, Moreno R, Takala J, Willatts S, De Mendonca A, Bruining H, Reinhart CK, Suter PM, Thijs LG. The SOFA (sepsis-related organ failure assessment) score to describe organ dysfunction/failure. On behalf of the Working Group on sepsisrelated problems of the European Society of intensive care medicine. Intensive Care Med. 1996;22:707–10. 30. Akaike H. A new look at the statistical model identification. IEEE Trans Autom Control. 1974;19:716–23. 31. Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med. 1996;15:361–87. 32. Harrell FE Jr, Lee KL, Califf RM, Pryor DB, Rosati RA. Regression modelling strategies for improved prognostic prediction. Stat Med. 1984;3:143–52. 33. Ostermann M, Chang RW. Challenges of defining acute kidney injury. QJM. 2011;104:237–43. 34. Ostermann M, Chang R. Correlation between the AKI classification and outcome. Crit Care. 2008;12:R144. 35. Mandelbaum T, Scott DJ, Lee J, Mark RG, Malhotra A, Waikar S, Howell MD, Talmor DS. Outcome of critically ill patients with acute kidney injury using the acute kidney injury network criteria. Crit Care Med. 2011;39:2659–64. 36. Cruz DN, Bolgan I, Perazella MA, Bonello M, de Cal M, Corradi V, Polanco N, Ocampo C, Nalesso F, Piccinni P, Ronco C. North East Italian Prospective Hospital Renal Outcome Survey on Acute Kidney Injury (NEiPHROS-AKI): targeting the problem with the RIFLE criteria. Clin J Am Soc Nephrol. 2007;2:418–25. 37. Lameire N, Van Biesen W, Vanholder R. Acute renal failure. Lancet. 2005;365:417–30.
123
