- Email: [email protected]
A Comparison of Acute Kidney Injury Classifications in Patients With Severe Sepsis and Septic Shock
CLINICAL INVESTIGATION
A Comparison of Acute Kidney Injury Classifications in Patients With Severe Sepsis and Septic Shock Won Young Kim, MD, Jin Won Huh, MD, PhD, Chae-Man Lim, MD, PhD, Younsuck Koh, MD, PhD and Sang-Bum Hong, MD, PhD
Abstract: Introduction: Acute kidney injury (AKI) is a common and serious complication in patients with severe sepsis and septic shock. AKI can be classified using the RIFLE (Risk, Injury, Failure, Loss and End-stage) and/or the Acute Kidney Injury Network (AKIN) criteria. The aim of this study was to determine which of the 2 systems better predicted 28-day mortality in patients with severe sepsis and septic shock. Methods: We retrospectively examined the medical records of all patients with severe sepsis and septic shock who were admitted to our medical intensive care unit (ICU) between January 2005 and December 2006 and compared the ability of the RIFLE and AKIN criteria to predict mortality. Results: Of the 326 patients admitted during the study period, 291 were included. AKI 48 and 72 hours after ICU admission was observed in 62.9% and 55.5% of patients, respectively, according to RIFLE criteria and in 65.6% and 59.2%, respectively, according to AKIN criteria. The overall 28-day mortality rate was 48.5%, and mortality was significantly associated with the severity of AKI, by both RIFLE and AKIN criteria. Both criteria had a similar ability to predict 28-day mortality 48 hours (0.65-69, P , 0.01) and 72 hours (0.66-70, P , 0.01) after ICU admission. Conclusions: The incidence of AKI in patients with sepsis and septic shock was slightly higher using AKIN than RIFLE criteria. However, using the AKIN criteria did not enhance the ability of AKI to predict 28-day mortality in such patients. Key Indexing Terms: Acute kidney injury; Acute renal failure; Critical illness; Outcome; Sepsis; RIFLE; AKIN. [Am J Med Sci 2012;344 (5):350–356.]
A
cute kidney injury (AKI) is a common and serious complication in patients with severe sepsis and septic shock and is associated with significant morbidity and mortality.1–4 Even a relatively modest impairment of renal function is considered an independent risk factor for mortality.5 The RIFLE (Risk, Injury, Failure, Loss and End-stage) classification, a consensus definition for AKI, was published by the Acute Dialysis Quality Initiative work group in 2004.6 These criteria categorize patients into those with 1 of 3 grades of AKI severity or 1 of 2 clinical outcomes based on glomerular filtration rate (GFR) and urine output (Table 1). More recently, the Acute Kidney Injury Network (AKIN) group, an international collaboration of nephrologists and intensivists, proposed a new system for the diagnosis and categorization of patients with AKI. This method, a modification of the RIFLE criteria, was designed to increase the sensitivity of classification.7 From the Pulmonary and Critical Care Medicine (JWH, C-ML, YK, S-BH) and Department of Emergency Medicine (WYK), Ulsan University College of Medicine, Asan Medical Center, Seoul, Korea. Submitted July 5, 2011; accepted in revised form November 16, 2011. This study was supported by a grant for the Korea Healthcare Technology R&D Project, Ministry for Health, Welfare & Family Affairs, Republic of Korea (A085068). Correspondence: Sang-Bum Hong, MD, PhD, Pulmonary and Critical Care Medicine, Ulsan University College of Medicine, Asan Medical Center, 86, Asanbyeongwon-gil, Songpa-gu, Seoul 138-736, Korea (E-mail: [email protected]).
350
Patients with a change in serum creatinine (SCr) concentration .0.3 mg/dL within 48 hours are regarded as having crossed an AKI threshold and are classified as AKIN stage 1, whereas patients receiving renal replacement therapy are classified as AKIN stage 3 (Table 1). In addition, the 2 clinical outcomes (Loss and End-stage) of the RIFLE criteria were removed from the AKIN criteria (Table 1). The RIFLE criteria have been widely validated and are regarded to constitute an acceptable, reliable, evidence-based system for defining AKI and have been shown to be of clinical relevance in diagnosing and classifying the severity of AKI.8–13 In contrast, the AKIN criteria may need further evaluation. Although a few studies have shown that the AKIN criteria do not improve on the ability of the RIFLE method to predict mortality, the comparative advantages and disadvantages of the 2 systems have not been determined.14–15 Moreover, few reports have assessed these criteria in patients with severe sepsis and septic shock, a setting in which AKI is common and markedly worsens outcomes. We therefore sought to determine the occurrence of AKI, as determined by RIFLE and AKIN criteria, 48 and 72 hours after admission to our medical intensive care unit (ICU), and to determine which of the 2 systems better predicted 28-day mortality in patients with severe sepsis and septic shock.
MATERIALS AND METHODS This retrospective cohort single-center study was conducted at the Asan Medical Center, a 2800-bed, Universityaffiliated, tertiary referral center in Seoul, Korea. Study subjects included all consecutive patients with severe sepsis and septic shock admitted to the medical ICU between January 2005 and December 2006. Diagnosis of severe sepsis and septic shock was based on the modified consensus criteria of the American College of Chest Physicians/Society of Critical Care Medicine.16 Patients were excluded if they were receiving chronic dialysis or their stay in the ICU was ,24 hours; for patients with multiple admissions, only the first admission was considered. This study was approved by the institutional review board of the Ethics Committee of the Asan Medical Center, which waived requirements for informed consent for our retrospective review of medical records. Data Collection Demographic and clinical data, including age, sex, symptoms, previous medical history, initial vital signs, blood results, 28-day course and diagnosis at admission, were retrieved from electronic hospital records. At this unit, SCr is determined at least once a day, and urine output is recorded hourly, for all patients. AKI was defined and classified by means of RIFLE criteria. Severity of illness was assessed using Acute Physiology and Chronic Health Evaluation II (APACHE II) and sequential organ failure assessment (SOFA) scores, which were calculated based on the worst variables recorded
The American Journal of the Medical Sciences
Volume 344, Number 5, November 2012
RIFLE Versus AKIN in Septic Shock
TABLE 1. Acute kidney injury, as determined by the RIFLE classification and AKIN criteria Serum Creatinine Criteria RIFLE classification Risk Injury Failure Loss ESKD Acute Kidney Injury Network (AKIN) criteria Stage 1 Stage 2 Stage 3
UO Criteria
Increase in serum Cr $1.5 X baseline or decrease in GFR $5% Increase in serum Cr $2.0 X baseline or decrease in GFR $50% Increase in serum Cr $3.0 X baseline or decrease in GFR $75% Persistent acute kidney injury $4 weeks Persistent need for RRT $4 weeks ESKD $3 months
,0.5 mL/kg/hr for $6 hr
Increase in serum Cr $0.3 mg/dL (26.5 /-mmol/L) or increase to 150%—199% (1.5–1.9–fold) from baseline Increase in serum Cr to 200%-299% (,2.0–2.9-fold) from baseline Increase in serum Cr to .300% ($3-fold) from baseline or serum Cr $4.0 mg/dL (44 /mmol/L) Treatment with RRT
,0.5 mL/kg/hr for .6 hr
,0.5 mL/kg/hr for $ 12 hr ,0.3 mL/kg/hr for $24 hr or anuria $12 hr
,0.5 mL/kg/hr for $ 12 hr ,0.3 mL/kg/hr for $24 hr or anuria .12 hr
RIFLE, Risk, Injury, Failure, Loss and End-stage; ESKD, end-stage kidney disease; GFR, glomerular filtration rate; RRT, renal replacement therapy; UO, urine output.
during the first 24 hours after ICU admission.17,18 Patients were followed up 18 months after hospital discharge by telephonic interview; when necessary, information on patient survival or date of death was obtained from the hospital registry office. The primary outcome of this study was 28-day mortality, and the secondary outcome was mortality at 18 months. The occurrence of AKI 48 and 72 hours after ICU admission was determined using the RIFLE and AKIN criteria.6,7 The RIFLE and AKIN Criteria Patients were categorized on SCr or urine output or both; the criteria that led to the worst classification were used. Clinical charts were reviewed for previous creatinine values. Baseline renal function was defined as the lowest known creatinine value during the preceding 3 months. For patients without known previous creatinine, the baseline creatinine was estimated using the simplified modification of diet in renal disease (MDRD) formula, assuming a GFR of 75 mL/min/1.73 m2.19 In this study, the MDRD formula was applied for 61 patients to estimate baseline creatinine concentration. AKIN criteria were calculated using SCr at ICU admission as reference, with staging based on any increase within 48 hours, 72 hours with at least 2 measurements being necessary. Patients were classified using the RIFLE and AKIN criteria at 48 and 72 hours after ICU admission. Patients were considered as having newly developed AKI who develop AKI at least 24 hours or more after admission if they did not have AKI on ICU admission. Statistical Analysis Continuous variables are reported as means with standard deviations and categorical variables as percentages. The primary analysis compared hospital survivors with nonsurvivors. All variables were tested for normal distribution using the Kolmogorov-Smirnov test. Student’s t test was used to compare the means of normally distributed continuous variables, whereas the Mann-Whitney U test was used to compare non-continuous variables. The x2 and McNemar tests were used to compare categorical variables. In addition, the x2 test for Ó 2012 Lippincott Williams & Wilkins
trend (x 2 for trend) was used to compare 28-day mortality prediction according to the RIFLE and AKIN criteria. Calibration of the model was evaluated by goodness-of-fit testing using the Hos-mer-Lemeshow statistic. Discrimination capability (ie, the ability of a model to differentiate between patients who died and those who survived) was evaluated using the area under a receiver operating characteristic curve (AuROC).20 AuROCs were compared using a nonparametric approach. In addition, AuROC analysis was used to calculate cutoff values, sensitivity, specificity and overall correctness. Finally, cutoff points were calculated from the best Youden index (sensitivity + specificity — 1). Data are presented as odds ratios (ORs) with 95% confidence intervals (CIs). Logistic regression models were used to evaluate the association of each RIFLE and AKIN class/stage with 28-day mortality. Cumulative survival curves as a function of time were generated using the Kaplan-Meier method and compared using the log-rank test. Cox’s regression analysis was used to determine the association of each AKI category with mortality, followed by calculation of an adjusted hazard ratio (HR) and the 95% CI. A 2-sided P value ,0.05 was considered statistically significant. All statistical analyses were performed using SPSS version 13.0 for Windows (SPSS, Chicago, EL).
RESULTS During the study period, 326 patients were admitted with severe sepsis and septic shock; of these, 35 were excluded from the study, including 19 who stayed in the ICU for ,24 hours and 16 who were receiving chronic dialysis. The remaining 291 patients consisted of 198 men (68%) and 93 women (32%), with a mean age of 61.1 6 14.4 years. The demographic and clinical characteristics of survivors and nonsurvivors are shown in Table 2. Ninety-seven patients received renal replacement therapy. A history of cancer and mechanical ventilatory support was significantly more prevalent in nonsurvivors than in survivors (Table 2). Not surprisingly, nonsurviving patients had higher APACHE II and SOFA scores.
351
Kim et al
TABLE 2. Patient demographic and clinical characteristics Survivors (n 5 150) Age (years) Gender (M/F) Height (cm) Weight (kg) Reasons (%) Pneumonia Hepatobiliary Intra-abdominal Urinary tract infection Others Cancer Mechanical ventilation (%) Creatinine (mg/dL) Urea (mmol/L) APACHE II score day 1 SOFA score day 1 RIFLE classification (%) 48 hr 72 hr AKIN criteria (%) 48 hr 72 hr
Nonsurvivors (n 5 141)
All patients (n 5 291)
61.7 6 13.6 97/53 162.3 6 8.8 60.2 611.2
62.5 6 14.3 101/40 163.2 6 8.6 59.0 6 10.7
61.1 6 14.4 198/93 162.8 6 8.7 59.6 6 10.7
0.64 0.20 0.33 0.37
64 (42.7) 28 (18.7) 10 (6.7) 15 (10.0) 33 (22.0) 38 (25.3) 104 (69.3) 1.98 6 1.80 33.6 6 28.5 22.4 6 7.9 11.0 6 3.4
75 (53.2) 13 (9.2) 19 (13.5) 5 (3.5) 29 (20.6) 54 (38.3) 136 (96.5) 2.12 6 2.93 37.2 6 27.5 26.7 6 7.6 12.9 6 3.8
139 (47.8) 41 (14.1) 29 (10.0) 15 (5.2) 62 (21.3) 92 (31.6) 240 (82.5) 2.08 6 2.73 35.0 6 28.3 24.5 6 7.8 11.9 6 3.7
0.22 0.28 0.27 0.24 0.68 0.02 0.00 0.63 ,0.00 ,0.00
76 (50.7) 60 (41.1)
107 (75.5) 87 (73.1)
183 (62.9) 147 (55.5)
,0.00 ,0.00
81 (54.0) 65 (44.5)
110 (78.0) 92 (77.3)
191 (65.6) 157 (59.2)
,0.00 ,0.00
p
RIFLE, Risk, Injury, Failure, Loss and End-stage; Reasons, reasons for admission; UTI, urinary tract infection; APACHE, Acute Physiology and Chronic Health Evaluation, first day; SOFA, sequential organ failure assessment, first day.
Incidence of AKI According to the RIFLE classification, AKI occurred in 62.9% of patients 48 hours after ICU admission, and 21.0%, 21.6% and 20.3% of these patients were classified as having Risk, Injury or Failure, respectively (Table 3). Forty-four patients who did not have AKI at ICU admission developed AKI over 72 hours. Compared with patients who did not develop AKI, those who developed AKI during their stay in the ICU were more severely ill at admission, as shown by a significantly higher APACHE II score (25.1 versus 21.8, P 5 0.03) and SOFA score (12.5 versus 10.9, P 5 0.04). Using the AKIN criteria, we found that AKI 48 hours after ICU admission occurred in 65.6% of patients, including 22.8% in stage 1,
17.9% in stage 2 and 25.1% in stage 3 (Table 3). The incidence of AKI 72 hours after ICU admission is also summarized in Table 3. About 3% to 4% more patients were found to have AKI by the AKIN than by the RIFLE criteria (P 5 0.18; P 5 0.13). Patient Mortality The overall 28-day mortality rate was 48.5%. Mortality stratified by RIFLE and AKIN criteria is shown in Table 3. Mortality was significantly higher in patients with AKI than in those without AKI, defined by either the RIFLE or AKIN criteria. However, there was no significant difference in 28-day mortality rates when AKI was defined by the RIFLE and AKIN
TABLE 3. Incidence and 28-day mortality stratified by RIFLE anc AKIN criteria RIFLE Incidence, n (%) Mortality, n (%) AKIN RIFLE 48 hr No AKI Risk Injury Failure Any category RIFLE 72 hr No AKI Risk Injury Failure Any category
108 61 63 59 183
(37.1) (21.0) (21.6) (20.3) (62.9)
34 28 44 35 107
(31.5) (45.9) (69.8) (59.3) (58.5)
118 36 55 56 147
(44.5) (13.6) (20.8) (21.1) (55.5)
32 16 34 37 87
(27.1) (44.4) (61.8) (67.3) (59.2)
AKIN 48 hr No AKI Stage 1 Stage 2 Stage 3 Any category AKIN 72 hr No AKI Stage 1 Stage 2 Stage 3 Any category
Incidence, n (%)
Mortality, n (%)
100 66 52 73 191
(34.4) (22.8) (17.9) (25.1) (65.6)
31 28 36 46 110
(31.0) (42.4) (69.2) (63.0) (57.6)
108 45 40 72 157
(40.8) (17.0) (15.1) (27.2) (59.2)
27 20 24 48 92
(25.0) (44.4) (60.0) (66.7) (58.6)
RIFLE, Risk, Injury, Failure, Loss and End-stage; AKIN, Kidney Injury Network.
352
Volume 344, Number 5, November 2012
RIFLE Versus AKIN in Septic Shock
TABLE 4. Ability of the RIFLE and AKIN criteria to predict in 28-day mortality by separate logistic regression models RIFLE Odds Ratio (95% CI) P AKIN Odds Ratio (95% CI) RIFLE 48 hr No AKI Risk Injury Failure RIFLE 72 hr No AKI Risk Injury Failure
1a 1.85 (0.9723.53) 5.04 (2.5729.89) 3.17 (1.6426.14)
0.06 ,0.01 ,0.01
1a 2.15 (0.9724.66) 4.35 (2.2128.58) 5.23 (2.64210.40)
0.04 ,0.01 ,0.01
AKIN 48 hr No AKI Stage 1 Stage 2 Stage 3 AKIN 72 hr No AKI Stage 1 Stage 2 Stage 3
P
1a 1.64 (0.8623.13) 5.01 (2.42210.35) 3.79 (2.0127.17)
0.13 ,0.01 ,0.01
1a 2.40 (1.1624.99) 4.50 (2.0929.70) 6.00 (3.12211.56)
0.02 ,0.01 ,0.01
a
Reference category. RIFLE, Risk, Injury, Failure, Loss and End-stage; AKIN, Acute Kidney Injury Network; CI, confidence interval.
TABLE 5. Calibration and discrimination of the RIFLE and AKIN scoring methods in predicting in 28-day mortality x2a df P AUROC 95% CI p RIFLE 48 hr RIFLE 72 hr AKIN 48 hr AKIN 72 hr
6.116 1.277 4.480 1.132
2 2 2 2
0.047 0.528 0.106 0.568
0.65 0.69 0.66 0.70
0.59–0.71 0.62–0.75 0.60–0.72 0.63–0.76
,0.01 ,0.01 ,0.01 ,0.01
a
Hosmer-Lemeshow goodness-of-fit test. df degrees of freedom; AUROC, area under a receiver operation characteristic curve; RIFLE, Risk, Injury, Failure, Loss and End-stage; AKIN, Acute Kidney Injury Network.
methods (P 5 0.83; P 5 0.79). The ability of these criteria to predict 28-day mortality was determined by logistic regression models (Table 4). We found that AKI, as defined by the chosen parameters, was independently associated with 28-day mortality, after adjusting for age and sex. The ORs for RIFLE classifications 72 hours after ICU admission, compared with those of non-AKI patients, were 2.15 (P 5 0.05) for RIFLE-R, 4.35 (P , 0.01) for RIFLE-I and 5.23 (P , 0.01) for RIFLE-F. Similarly, when we compared AKIN classifications, we found that, compared with AKIN stage 0, the ORs for AKIN stages 1,
2 and 3 were 2.40 (P 5 0.02), 4.50 (P , 0.01) and 6.00 (P , 0.01), respectively, 72 hours after ICU admission (Table 4). Table 5 shows the goodness-of-fit, as measured by HosmerLemeshow x 2 values, for predicted mortality risk, and the predictive accuracy of the RIFLE and AKIN criteria. Results of AuROC analysis, with 95% CIs, are shown in Table 5 and Figure 1, confirming that both the RIFLE and AKIN criteria could distinguish between survivors and nonsurvivors. AuROC analysis showed that use of the AKIN criteria was no better in predicting 28-day mortality than were the RIFLE criteria in patients with severe sepsis and septic shock. To determine the optimal cutoff points of both methods, we measured predictive sensitivities and specificities, using the Youden index (Table 6). The 28-day mortality rates differed significantly in patients classified as RIFLE-R and AKIN-1 (P , 0.01). Cox’s regression analysis, adjusted for age and sex, showed that each classification independently predicted mortality 48 and 72 hours after ICU admission (Table 7). Cumulative survival rates differed significantly (P , 0.01) between patients classified as non-AKI and those classified as RIFLE-R, REFLE-I and RIFLE-F, at 24 and 72 hours after ICU admission (Table 7). Figure 2 also shows the large stepwise increments in HR for mortality relative to AKI category according to the 2 classification systems, determined 48 and 72 hours after ICU admission (Figure 2).
FIGURE 1. The area under the receiver operating characteristic (AuROC) curve analyses for 291 patients with severe sepsis and septic shock based on their RIFLE classification and AKIN criteria within the first 24 hours of ICU admission, initial 48 and 72 hours after ICU admission.
Ó 2012 Lippincott Williams & Wilkins
353
Kim et al
TABLE 6. Subsequent 28-day mortality predicted after ICU admission Youden index Sensitivity (%) Specificity (%) RIFLE 48 hr No AKI Riska Injury Failure RIFLE 72 hr No AKI Riska Injury Failure
0.252 0.274 0.088 0.000
75.9 56.0 24.8 0.0
49.3 71.3 84.0 1.0
0.320 0.323 0.181 0.000
73.1 59.7 31.1 0.0
58.9 72.6 87.0 1.0
AKIN 48 hr No AKI Stage 1a Stage 2 Stage 3 AKIN 72 hr No AKI Stage 1a Stage 2 Stage 3
Youden index
Sensitivity (%)
Specificity (%)
0.240 0.295 0.146 0.000
78.0 58.2 32.6 0.0
46.0 71.3 82.0 1.0
0.328 0.331 0.239 0.000
77.3 60.5 40.3 0.0
55.5 72.6 83.6 1.0
Value giving the best Youden index (5sensitivity + specificity – 1). ICU, intensive care unit; RIFLE, Risk, Injury, Failure, Loss and End-stage; AKIN, Acute Kidney Injury Network.
a
DISCUSSION We evaluated the incidence of AKI using the 2 recently proposed staging systems, RIFLE and AKIN, in 326 consecutive patients with severe sepsis and septic shock who had been admitted to the medical ICU. In addition, we compared the ability of these classification systems to predict 28-day mortality. We found that the incidence of AKI in patients with severe sepsis and septic shock was similar when determined by RIFLE and AKIN criteria (62.9% versus 65.6%). In comparison, the incidences of AKI in similarly classified patients have been reported to be 35.5% and 50.4%.14,15 In contrast to these earlier studies, we included only patients with severe sepsis and septic shock, the leading contributing factors to AKI in critically ill patients and those that are generally predictive of poor clinical prognosis.21 Although the AKIN criteria were designed to improve the sensitivity and reproducibility of the RIFLE system for predicting AKI, the AKIN criteria did not affect the sensitivity or predictive ability of the RIFLE criteria in critically ill patients.14 A comparison of the RIFLE and AKIN parameters in patients receiving stem cell transplantation found that the RIFLE criteria were more sensitive than the AKIN criteria in identifying patients with AKI,22 suggesting that the RIFLE criteria are better able to predict 28-day mortality in at least some study populations. In patients with severe sepsis and septic shock, although we found that the AKIN criteria identified 2.7% and 3.7% more patients with some degree of AKI than did RIFLE criteria, at 48 and
72 hours, respectively, after ICU admission, there were no statistically significant differences in 28-day mortality rates. Increasing RIFLE and AKIN classifications 48 and 72 hours after ICU admission were associated with increased mortality. These results are in agreement with those evaluating AKI in other study populations.14,22 We found that both the RIFLE and AKIN criteria precisely predicted hospital mortality and long-term prognosis in patients with severe sepsis and septic shock, and no differences between the 2 classification systems 48 and 72 hours after ICU admission were evident. Although previous studies evaluated the incidence of AKI 24 and 48 hours after ICU admission, and the abilities of the 2 systems to predict 28-day mortality at the 24- or 48-hour observation point, we evaluated the incidence of AKI 72 hours after ICU admission and the ability of AKI at this time to predict 28-day mortality. Some ICU patients may show delayed onset of AKI, occurring .48 hours or even later after ICU admission. Limitations This study had several limitations, including a retrospective design, relatively small patient numbers and evaluation of patients from only a single center. Moreover, baseline SCr was not measured in 61 patients, compelling us to estimate baseline data in such patients using the MDRD equation, a drawback that is common in studies such as ours, attributable to the emergency nature of ICU admissions. However, this may have
TABLE 7. Cox’s regression analyses of the ability of the RIFLE and AKIN criteria to predict in 28-day mortality Hazard Ratio Hazard Ratio (95% CI) P (95% CI) RIFLE 48 hr No AKI Risk Injury Failure RIFLE 72 hr No AKI Risk Injury Failure
1 1.70 (1.03–2.81) 3.33 (2.13–5.22) 2.48 (1.55–3.98)
0.04 ,0.01 ,0.01
1 2.04 (1.12–3.72) 3.13 (1.93–5.07) 3.61 (2.24–5.80)
0.02 ,0.01 ,0.01
AKIN No Stage Stage Stage AKIN No Stage Stage Stage
48hr AKI 1 2 3 72hr AKI 1 2 3
P
1 1.57 (0.94–2.61) 3.25 (2.01–5.27) 2.88 (1.82–4.55)
0.08 ,0.01 ,0.01
1 2.24 (1.26–4.00) 3.24 (1.87–5.63) 4.12 (2.56–6.61)
,0.01 ,0.01 ,0.01
CI, confidence interval; RIFLE, Risk, Injury, Failure, Loss and End-stage; AKIN, Acute Kidney Injury Network.
354
Volume 344, Number 5, November 2012
RIFLE Versus AKIN in Septic Shock
FIGURE 2. Cumulative survival rate for 291 patients with severe sepsis and septic shock based on their RIFLE classification (A and B) and AKIN stage (C and D).
led to the misclassification of some patients. Finally, all of our patients had been admitted to a medical (thus nonsurgical) ICU, limiting the applicability of our findings to more heterogeneous populations or surgical patients.
CONCLUSION In summary, we retrospectively compared the ability of the RIFLE and AKIN criteria to predict AFI in patients with severe sepsis and septic shock. We found that these 2 classification systems had the same capacity to predict 28-day mortality rates. The incidence of AKI, as determined by AKIN, was slightly higher than that measured by RIFLE. However, the 2 criteria showed no clinically important differences when used to assess patient outcomes over a 72-hour observation period. ACKNOWLEDGMENT We are grateful to Eun-mi Cho for many hours of assistance in gathering data from medical records. REFERENCES 1. Angus DC, Linde-Zwirble WT, Lidicker J, et al. Epidemiology of severe sepsis in the United States: analysis of incidence, outcome, and associated costs of care. Crit Care Med 2001;29:1303–10. 2. Engel C, Brunkhorst FM, Bone HG, et al. Epidemiology of sepsis in Germany: results from a national prospective multicenter study. Intensive Care Med 2007;33:606–18. 3. Metnitz PG, Krenn CG, Steltzer H, et al. Effect of acute renal failure requiring renal replacement therapy on outcome in critically ill patients. Crit Care Med 2002;30:2051–8.
Ó 2012 Lippincott Williams & Wilkins
4. Joannidis M, Metnitz PG. Epidemiology and natural history of acute renal failure in the ICU. Crit Care Clin 2005;21:239–49. 5. Lassnigg A, Schmidlin D, Mouhieddine M, et al. Minimal changes of serum creatinine predict prognosis in patients after cardiothoracic surgery: a prospective cohort study. J Am Soc Nephrol 2004;15: 1597–605. 6. Bellomo R, Ronco C, Kellum JA, et al. 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. 7. Mehta RL, Kellum JA, Shah SV, et al. Acute kidney injury network: report of an initiative to improve outcomes in acute kidney injury. Crit Care 2007;11:R31. 8. Ostermann M, Chang RW. Acute kidney injury in the intensive care unit according to RIFLE. Crit Care Med 2007;35:1837–43; quiz 52. 9. Gibney N, Hoste E, Burdmann EA, et al. Timing of initiation and discontinuation of renal replacement therapy in AKI: unanswered key questions. Clin J Am Soc Nephrol 2008;3:876–80. 10. Bellomo R, Kellum JA, Ronco C. Defining and classifying acute renal failure: from advocacy to consensus and validation of the RIFLE criteria. Intensive Care Med 2007;33:409–13. 11. Ricci Z, Cruz D, Ronco C. The RIFLE criteria and mortality in acute kidney injury: a systematic review. Kidney Int 2008;73:538–46. 12. Hoste EA, Schurgers M. Epidemiology of acute kidney injury: how big is the problem? Crit Care Med 2008;36:S146–51. 13. Hoste EA, Clermont G, Kersten A, et al. RIFLE criteria for acute kidney injury are associated with hospital mortality in critically ill patients: a cohort analysis. Crit Care 2006;10:R73.
355
Kim et al
14. 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.
On behalf of the Working Group on Sepsis-Related Problems of the European Society of Intensive Care Medicine. Intensive Care Med 1996;22:707–10.
15. Lopes JA, Fernandes P, Jorge S, et al. 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.
19. National Kidney Foundation. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 2002;39:Sl-266.
16. Bone RC, Balk RA, Cerra FB, et al. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. The ACCP/SCCM Consensus Conference Committee. American College of Chest Physicians/Society of Critical Care Medicine. Chest 1992;101: 1644–55. 17. Knaus WA, Draper EA, Wagner DP, et al. APACHE II: a severity of disease classification system. Crit Care Med 1985;13:818–29. 18. Vincent JL, Moreno R, Takala J, et al. The SOFA (sepsis-related organ failure assessment) score to describe organ dysfunction/failure.
356
20. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988;44:837–45. 21. Bagshaw SM, Uchino S, Bellomo R, et al. Septic acute kidney injury in critically ill patients: clinical characteristics and outcomes. Clin J Am Soc Nephrol 2007;2:431–9. 22. Ando M, Mori J, Ohashi K, et al. A comparative assessment of the RIFLE, AKIN and conventional criteria for acute kidney injury after hematopoietic SCT. Bone Marrow Transplant 2010;45:1427–34.
Volume 344, Number 5, November 2012
A Comparison of Acute Kidney Injury Classifications in Patients With Severe Sepsis and Septic Shock Won Young Kim, MD, Jin Won Huh, MD, PhD, Chae-Man Lim, MD, PhD, Younsuck Koh, MD, PhD and Sang-Bum Hong, MD, PhD
Abstract: Introduction: Acute kidney injury (AKI) is a common and serious complication in patients with severe sepsis and septic shock. AKI can be classified using the RIFLE (Risk, Injury, Failure, Loss and End-stage) and/or the Acute Kidney Injury Network (AKIN) criteria. The aim of this study was to determine which of the 2 systems better predicted 28-day mortality in patients with severe sepsis and septic shock. Methods: We retrospectively examined the medical records of all patients with severe sepsis and septic shock who were admitted to our medical intensive care unit (ICU) between January 2005 and December 2006 and compared the ability of the RIFLE and AKIN criteria to predict mortality. Results: Of the 326 patients admitted during the study period, 291 were included. AKI 48 and 72 hours after ICU admission was observed in 62.9% and 55.5% of patients, respectively, according to RIFLE criteria and in 65.6% and 59.2%, respectively, according to AKIN criteria. The overall 28-day mortality rate was 48.5%, and mortality was significantly associated with the severity of AKI, by both RIFLE and AKIN criteria. Both criteria had a similar ability to predict 28-day mortality 48 hours (0.65-69, P , 0.01) and 72 hours (0.66-70, P , 0.01) after ICU admission. Conclusions: The incidence of AKI in patients with sepsis and septic shock was slightly higher using AKIN than RIFLE criteria. However, using the AKIN criteria did not enhance the ability of AKI to predict 28-day mortality in such patients. Key Indexing Terms: Acute kidney injury; Acute renal failure; Critical illness; Outcome; Sepsis; RIFLE; AKIN. [Am J Med Sci 2012;344 (5):350–356.]
A
cute kidney injury (AKI) is a common and serious complication in patients with severe sepsis and septic shock and is associated with significant morbidity and mortality.1–4 Even a relatively modest impairment of renal function is considered an independent risk factor for mortality.5 The RIFLE (Risk, Injury, Failure, Loss and End-stage) classification, a consensus definition for AKI, was published by the Acute Dialysis Quality Initiative work group in 2004.6 These criteria categorize patients into those with 1 of 3 grades of AKI severity or 1 of 2 clinical outcomes based on glomerular filtration rate (GFR) and urine output (Table 1). More recently, the Acute Kidney Injury Network (AKIN) group, an international collaboration of nephrologists and intensivists, proposed a new system for the diagnosis and categorization of patients with AKI. This method, a modification of the RIFLE criteria, was designed to increase the sensitivity of classification.7 From the Pulmonary and Critical Care Medicine (JWH, C-ML, YK, S-BH) and Department of Emergency Medicine (WYK), Ulsan University College of Medicine, Asan Medical Center, Seoul, Korea. Submitted July 5, 2011; accepted in revised form November 16, 2011. This study was supported by a grant for the Korea Healthcare Technology R&D Project, Ministry for Health, Welfare & Family Affairs, Republic of Korea (A085068). Correspondence: Sang-Bum Hong, MD, PhD, Pulmonary and Critical Care Medicine, Ulsan University College of Medicine, Asan Medical Center, 86, Asanbyeongwon-gil, Songpa-gu, Seoul 138-736, Korea (E-mail: [email protected]).
350
Patients with a change in serum creatinine (SCr) concentration .0.3 mg/dL within 48 hours are regarded as having crossed an AKI threshold and are classified as AKIN stage 1, whereas patients receiving renal replacement therapy are classified as AKIN stage 3 (Table 1). In addition, the 2 clinical outcomes (Loss and End-stage) of the RIFLE criteria were removed from the AKIN criteria (Table 1). The RIFLE criteria have been widely validated and are regarded to constitute an acceptable, reliable, evidence-based system for defining AKI and have been shown to be of clinical relevance in diagnosing and classifying the severity of AKI.8–13 In contrast, the AKIN criteria may need further evaluation. Although a few studies have shown that the AKIN criteria do not improve on the ability of the RIFLE method to predict mortality, the comparative advantages and disadvantages of the 2 systems have not been determined.14–15 Moreover, few reports have assessed these criteria in patients with severe sepsis and septic shock, a setting in which AKI is common and markedly worsens outcomes. We therefore sought to determine the occurrence of AKI, as determined by RIFLE and AKIN criteria, 48 and 72 hours after admission to our medical intensive care unit (ICU), and to determine which of the 2 systems better predicted 28-day mortality in patients with severe sepsis and septic shock.
MATERIALS AND METHODS This retrospective cohort single-center study was conducted at the Asan Medical Center, a 2800-bed, Universityaffiliated, tertiary referral center in Seoul, Korea. Study subjects included all consecutive patients with severe sepsis and septic shock admitted to the medical ICU between January 2005 and December 2006. Diagnosis of severe sepsis and septic shock was based on the modified consensus criteria of the American College of Chest Physicians/Society of Critical Care Medicine.16 Patients were excluded if they were receiving chronic dialysis or their stay in the ICU was ,24 hours; for patients with multiple admissions, only the first admission was considered. This study was approved by the institutional review board of the Ethics Committee of the Asan Medical Center, which waived requirements for informed consent for our retrospective review of medical records. Data Collection Demographic and clinical data, including age, sex, symptoms, previous medical history, initial vital signs, blood results, 28-day course and diagnosis at admission, were retrieved from electronic hospital records. At this unit, SCr is determined at least once a day, and urine output is recorded hourly, for all patients. AKI was defined and classified by means of RIFLE criteria. Severity of illness was assessed using Acute Physiology and Chronic Health Evaluation II (APACHE II) and sequential organ failure assessment (SOFA) scores, which were calculated based on the worst variables recorded
The American Journal of the Medical Sciences
Volume 344, Number 5, November 2012
RIFLE Versus AKIN in Septic Shock
TABLE 1. Acute kidney injury, as determined by the RIFLE classification and AKIN criteria Serum Creatinine Criteria RIFLE classification Risk Injury Failure Loss ESKD Acute Kidney Injury Network (AKIN) criteria Stage 1 Stage 2 Stage 3
UO Criteria
Increase in serum Cr $1.5 X baseline or decrease in GFR $5% Increase in serum Cr $2.0 X baseline or decrease in GFR $50% Increase in serum Cr $3.0 X baseline or decrease in GFR $75% Persistent acute kidney injury $4 weeks Persistent need for RRT $4 weeks ESKD $3 months
,0.5 mL/kg/hr for $6 hr
Increase in serum Cr $0.3 mg/dL (26.5 /-mmol/L) or increase to 150%—199% (1.5–1.9–fold) from baseline Increase in serum Cr to 200%-299% (,2.0–2.9-fold) from baseline Increase in serum Cr to .300% ($3-fold) from baseline or serum Cr $4.0 mg/dL (44 /mmol/L) Treatment with RRT
,0.5 mL/kg/hr for .6 hr
,0.5 mL/kg/hr for $ 12 hr ,0.3 mL/kg/hr for $24 hr or anuria $12 hr
,0.5 mL/kg/hr for $ 12 hr ,0.3 mL/kg/hr for $24 hr or anuria .12 hr
RIFLE, Risk, Injury, Failure, Loss and End-stage; ESKD, end-stage kidney disease; GFR, glomerular filtration rate; RRT, renal replacement therapy; UO, urine output.
during the first 24 hours after ICU admission.17,18 Patients were followed up 18 months after hospital discharge by telephonic interview; when necessary, information on patient survival or date of death was obtained from the hospital registry office. The primary outcome of this study was 28-day mortality, and the secondary outcome was mortality at 18 months. The occurrence of AKI 48 and 72 hours after ICU admission was determined using the RIFLE and AKIN criteria.6,7 The RIFLE and AKIN Criteria Patients were categorized on SCr or urine output or both; the criteria that led to the worst classification were used. Clinical charts were reviewed for previous creatinine values. Baseline renal function was defined as the lowest known creatinine value during the preceding 3 months. For patients without known previous creatinine, the baseline creatinine was estimated using the simplified modification of diet in renal disease (MDRD) formula, assuming a GFR of 75 mL/min/1.73 m2.19 In this study, the MDRD formula was applied for 61 patients to estimate baseline creatinine concentration. AKIN criteria were calculated using SCr at ICU admission as reference, with staging based on any increase within 48 hours, 72 hours with at least 2 measurements being necessary. Patients were classified using the RIFLE and AKIN criteria at 48 and 72 hours after ICU admission. Patients were considered as having newly developed AKI who develop AKI at least 24 hours or more after admission if they did not have AKI on ICU admission. Statistical Analysis Continuous variables are reported as means with standard deviations and categorical variables as percentages. The primary analysis compared hospital survivors with nonsurvivors. All variables were tested for normal distribution using the Kolmogorov-Smirnov test. Student’s t test was used to compare the means of normally distributed continuous variables, whereas the Mann-Whitney U test was used to compare non-continuous variables. The x2 and McNemar tests were used to compare categorical variables. In addition, the x2 test for Ó 2012 Lippincott Williams & Wilkins
trend (x 2 for trend) was used to compare 28-day mortality prediction according to the RIFLE and AKIN criteria. Calibration of the model was evaluated by goodness-of-fit testing using the Hos-mer-Lemeshow statistic. Discrimination capability (ie, the ability of a model to differentiate between patients who died and those who survived) was evaluated using the area under a receiver operating characteristic curve (AuROC).20 AuROCs were compared using a nonparametric approach. In addition, AuROC analysis was used to calculate cutoff values, sensitivity, specificity and overall correctness. Finally, cutoff points were calculated from the best Youden index (sensitivity + specificity — 1). Data are presented as odds ratios (ORs) with 95% confidence intervals (CIs). Logistic regression models were used to evaluate the association of each RIFLE and AKIN class/stage with 28-day mortality. Cumulative survival curves as a function of time were generated using the Kaplan-Meier method and compared using the log-rank test. Cox’s regression analysis was used to determine the association of each AKI category with mortality, followed by calculation of an adjusted hazard ratio (HR) and the 95% CI. A 2-sided P value ,0.05 was considered statistically significant. All statistical analyses were performed using SPSS version 13.0 for Windows (SPSS, Chicago, EL).
RESULTS During the study period, 326 patients were admitted with severe sepsis and septic shock; of these, 35 were excluded from the study, including 19 who stayed in the ICU for ,24 hours and 16 who were receiving chronic dialysis. The remaining 291 patients consisted of 198 men (68%) and 93 women (32%), with a mean age of 61.1 6 14.4 years. The demographic and clinical characteristics of survivors and nonsurvivors are shown in Table 2. Ninety-seven patients received renal replacement therapy. A history of cancer and mechanical ventilatory support was significantly more prevalent in nonsurvivors than in survivors (Table 2). Not surprisingly, nonsurviving patients had higher APACHE II and SOFA scores.
351
Kim et al
TABLE 2. Patient demographic and clinical characteristics Survivors (n 5 150) Age (years) Gender (M/F) Height (cm) Weight (kg) Reasons (%) Pneumonia Hepatobiliary Intra-abdominal Urinary tract infection Others Cancer Mechanical ventilation (%) Creatinine (mg/dL) Urea (mmol/L) APACHE II score day 1 SOFA score day 1 RIFLE classification (%) 48 hr 72 hr AKIN criteria (%) 48 hr 72 hr
Nonsurvivors (n 5 141)
All patients (n 5 291)
61.7 6 13.6 97/53 162.3 6 8.8 60.2 611.2
62.5 6 14.3 101/40 163.2 6 8.6 59.0 6 10.7
61.1 6 14.4 198/93 162.8 6 8.7 59.6 6 10.7
0.64 0.20 0.33 0.37
64 (42.7) 28 (18.7) 10 (6.7) 15 (10.0) 33 (22.0) 38 (25.3) 104 (69.3) 1.98 6 1.80 33.6 6 28.5 22.4 6 7.9 11.0 6 3.4
75 (53.2) 13 (9.2) 19 (13.5) 5 (3.5) 29 (20.6) 54 (38.3) 136 (96.5) 2.12 6 2.93 37.2 6 27.5 26.7 6 7.6 12.9 6 3.8
139 (47.8) 41 (14.1) 29 (10.0) 15 (5.2) 62 (21.3) 92 (31.6) 240 (82.5) 2.08 6 2.73 35.0 6 28.3 24.5 6 7.8 11.9 6 3.7
0.22 0.28 0.27 0.24 0.68 0.02 0.00 0.63 ,0.00 ,0.00
76 (50.7) 60 (41.1)
107 (75.5) 87 (73.1)
183 (62.9) 147 (55.5)
,0.00 ,0.00
81 (54.0) 65 (44.5)
110 (78.0) 92 (77.3)
191 (65.6) 157 (59.2)
,0.00 ,0.00
p
RIFLE, Risk, Injury, Failure, Loss and End-stage; Reasons, reasons for admission; UTI, urinary tract infection; APACHE, Acute Physiology and Chronic Health Evaluation, first day; SOFA, sequential organ failure assessment, first day.
Incidence of AKI According to the RIFLE classification, AKI occurred in 62.9% of patients 48 hours after ICU admission, and 21.0%, 21.6% and 20.3% of these patients were classified as having Risk, Injury or Failure, respectively (Table 3). Forty-four patients who did not have AKI at ICU admission developed AKI over 72 hours. Compared with patients who did not develop AKI, those who developed AKI during their stay in the ICU were more severely ill at admission, as shown by a significantly higher APACHE II score (25.1 versus 21.8, P 5 0.03) and SOFA score (12.5 versus 10.9, P 5 0.04). Using the AKIN criteria, we found that AKI 48 hours after ICU admission occurred in 65.6% of patients, including 22.8% in stage 1,
17.9% in stage 2 and 25.1% in stage 3 (Table 3). The incidence of AKI 72 hours after ICU admission is also summarized in Table 3. About 3% to 4% more patients were found to have AKI by the AKIN than by the RIFLE criteria (P 5 0.18; P 5 0.13). Patient Mortality The overall 28-day mortality rate was 48.5%. Mortality stratified by RIFLE and AKIN criteria is shown in Table 3. Mortality was significantly higher in patients with AKI than in those without AKI, defined by either the RIFLE or AKIN criteria. However, there was no significant difference in 28-day mortality rates when AKI was defined by the RIFLE and AKIN
TABLE 3. Incidence and 28-day mortality stratified by RIFLE anc AKIN criteria RIFLE Incidence, n (%) Mortality, n (%) AKIN RIFLE 48 hr No AKI Risk Injury Failure Any category RIFLE 72 hr No AKI Risk Injury Failure Any category
108 61 63 59 183
(37.1) (21.0) (21.6) (20.3) (62.9)
34 28 44 35 107
(31.5) (45.9) (69.8) (59.3) (58.5)
118 36 55 56 147
(44.5) (13.6) (20.8) (21.1) (55.5)
32 16 34 37 87
(27.1) (44.4) (61.8) (67.3) (59.2)
AKIN 48 hr No AKI Stage 1 Stage 2 Stage 3 Any category AKIN 72 hr No AKI Stage 1 Stage 2 Stage 3 Any category
Incidence, n (%)
Mortality, n (%)
100 66 52 73 191
(34.4) (22.8) (17.9) (25.1) (65.6)
31 28 36 46 110
(31.0) (42.4) (69.2) (63.0) (57.6)
108 45 40 72 157
(40.8) (17.0) (15.1) (27.2) (59.2)
27 20 24 48 92
(25.0) (44.4) (60.0) (66.7) (58.6)
RIFLE, Risk, Injury, Failure, Loss and End-stage; AKIN, Kidney Injury Network.
352
Volume 344, Number 5, November 2012
RIFLE Versus AKIN in Septic Shock
TABLE 4. Ability of the RIFLE and AKIN criteria to predict in 28-day mortality by separate logistic regression models RIFLE Odds Ratio (95% CI) P AKIN Odds Ratio (95% CI) RIFLE 48 hr No AKI Risk Injury Failure RIFLE 72 hr No AKI Risk Injury Failure
1a 1.85 (0.9723.53) 5.04 (2.5729.89) 3.17 (1.6426.14)
0.06 ,0.01 ,0.01
1a 2.15 (0.9724.66) 4.35 (2.2128.58) 5.23 (2.64210.40)
0.04 ,0.01 ,0.01
AKIN 48 hr No AKI Stage 1 Stage 2 Stage 3 AKIN 72 hr No AKI Stage 1 Stage 2 Stage 3
P
1a 1.64 (0.8623.13) 5.01 (2.42210.35) 3.79 (2.0127.17)
0.13 ,0.01 ,0.01
1a 2.40 (1.1624.99) 4.50 (2.0929.70) 6.00 (3.12211.56)
0.02 ,0.01 ,0.01
a
Reference category. RIFLE, Risk, Injury, Failure, Loss and End-stage; AKIN, Acute Kidney Injury Network; CI, confidence interval.
TABLE 5. Calibration and discrimination of the RIFLE and AKIN scoring methods in predicting in 28-day mortality x2a df P AUROC 95% CI p RIFLE 48 hr RIFLE 72 hr AKIN 48 hr AKIN 72 hr
6.116 1.277 4.480 1.132
2 2 2 2
0.047 0.528 0.106 0.568
0.65 0.69 0.66 0.70
0.59–0.71 0.62–0.75 0.60–0.72 0.63–0.76
,0.01 ,0.01 ,0.01 ,0.01
a
Hosmer-Lemeshow goodness-of-fit test. df degrees of freedom; AUROC, area under a receiver operation characteristic curve; RIFLE, Risk, Injury, Failure, Loss and End-stage; AKIN, Acute Kidney Injury Network.
methods (P 5 0.83; P 5 0.79). The ability of these criteria to predict 28-day mortality was determined by logistic regression models (Table 4). We found that AKI, as defined by the chosen parameters, was independently associated with 28-day mortality, after adjusting for age and sex. The ORs for RIFLE classifications 72 hours after ICU admission, compared with those of non-AKI patients, were 2.15 (P 5 0.05) for RIFLE-R, 4.35 (P , 0.01) for RIFLE-I and 5.23 (P , 0.01) for RIFLE-F. Similarly, when we compared AKIN classifications, we found that, compared with AKIN stage 0, the ORs for AKIN stages 1,
2 and 3 were 2.40 (P 5 0.02), 4.50 (P , 0.01) and 6.00 (P , 0.01), respectively, 72 hours after ICU admission (Table 4). Table 5 shows the goodness-of-fit, as measured by HosmerLemeshow x 2 values, for predicted mortality risk, and the predictive accuracy of the RIFLE and AKIN criteria. Results of AuROC analysis, with 95% CIs, are shown in Table 5 and Figure 1, confirming that both the RIFLE and AKIN criteria could distinguish between survivors and nonsurvivors. AuROC analysis showed that use of the AKIN criteria was no better in predicting 28-day mortality than were the RIFLE criteria in patients with severe sepsis and septic shock. To determine the optimal cutoff points of both methods, we measured predictive sensitivities and specificities, using the Youden index (Table 6). The 28-day mortality rates differed significantly in patients classified as RIFLE-R and AKIN-1 (P , 0.01). Cox’s regression analysis, adjusted for age and sex, showed that each classification independently predicted mortality 48 and 72 hours after ICU admission (Table 7). Cumulative survival rates differed significantly (P , 0.01) between patients classified as non-AKI and those classified as RIFLE-R, REFLE-I and RIFLE-F, at 24 and 72 hours after ICU admission (Table 7). Figure 2 also shows the large stepwise increments in HR for mortality relative to AKI category according to the 2 classification systems, determined 48 and 72 hours after ICU admission (Figure 2).
FIGURE 1. The area under the receiver operating characteristic (AuROC) curve analyses for 291 patients with severe sepsis and septic shock based on their RIFLE classification and AKIN criteria within the first 24 hours of ICU admission, initial 48 and 72 hours after ICU admission.
Ó 2012 Lippincott Williams & Wilkins
353
Kim et al
TABLE 6. Subsequent 28-day mortality predicted after ICU admission Youden index Sensitivity (%) Specificity (%) RIFLE 48 hr No AKI Riska Injury Failure RIFLE 72 hr No AKI Riska Injury Failure
0.252 0.274 0.088 0.000
75.9 56.0 24.8 0.0
49.3 71.3 84.0 1.0
0.320 0.323 0.181 0.000
73.1 59.7 31.1 0.0
58.9 72.6 87.0 1.0
AKIN 48 hr No AKI Stage 1a Stage 2 Stage 3 AKIN 72 hr No AKI Stage 1a Stage 2 Stage 3
Youden index
Sensitivity (%)
Specificity (%)
0.240 0.295 0.146 0.000
78.0 58.2 32.6 0.0
46.0 71.3 82.0 1.0
0.328 0.331 0.239 0.000
77.3 60.5 40.3 0.0
55.5 72.6 83.6 1.0
Value giving the best Youden index (5sensitivity + specificity – 1). ICU, intensive care unit; RIFLE, Risk, Injury, Failure, Loss and End-stage; AKIN, Acute Kidney Injury Network.
a
DISCUSSION We evaluated the incidence of AKI using the 2 recently proposed staging systems, RIFLE and AKIN, in 326 consecutive patients with severe sepsis and septic shock who had been admitted to the medical ICU. In addition, we compared the ability of these classification systems to predict 28-day mortality. We found that the incidence of AKI in patients with severe sepsis and septic shock was similar when determined by RIFLE and AKIN criteria (62.9% versus 65.6%). In comparison, the incidences of AKI in similarly classified patients have been reported to be 35.5% and 50.4%.14,15 In contrast to these earlier studies, we included only patients with severe sepsis and septic shock, the leading contributing factors to AKI in critically ill patients and those that are generally predictive of poor clinical prognosis.21 Although the AKIN criteria were designed to improve the sensitivity and reproducibility of the RIFLE system for predicting AKI, the AKIN criteria did not affect the sensitivity or predictive ability of the RIFLE criteria in critically ill patients.14 A comparison of the RIFLE and AKIN parameters in patients receiving stem cell transplantation found that the RIFLE criteria were more sensitive than the AKIN criteria in identifying patients with AKI,22 suggesting that the RIFLE criteria are better able to predict 28-day mortality in at least some study populations. In patients with severe sepsis and septic shock, although we found that the AKIN criteria identified 2.7% and 3.7% more patients with some degree of AKI than did RIFLE criteria, at 48 and
72 hours, respectively, after ICU admission, there were no statistically significant differences in 28-day mortality rates. Increasing RIFLE and AKIN classifications 48 and 72 hours after ICU admission were associated with increased mortality. These results are in agreement with those evaluating AKI in other study populations.14,22 We found that both the RIFLE and AKIN criteria precisely predicted hospital mortality and long-term prognosis in patients with severe sepsis and septic shock, and no differences between the 2 classification systems 48 and 72 hours after ICU admission were evident. Although previous studies evaluated the incidence of AKI 24 and 48 hours after ICU admission, and the abilities of the 2 systems to predict 28-day mortality at the 24- or 48-hour observation point, we evaluated the incidence of AKI 72 hours after ICU admission and the ability of AKI at this time to predict 28-day mortality. Some ICU patients may show delayed onset of AKI, occurring .48 hours or even later after ICU admission. Limitations This study had several limitations, including a retrospective design, relatively small patient numbers and evaluation of patients from only a single center. Moreover, baseline SCr was not measured in 61 patients, compelling us to estimate baseline data in such patients using the MDRD equation, a drawback that is common in studies such as ours, attributable to the emergency nature of ICU admissions. However, this may have
TABLE 7. Cox’s regression analyses of the ability of the RIFLE and AKIN criteria to predict in 28-day mortality Hazard Ratio Hazard Ratio (95% CI) P (95% CI) RIFLE 48 hr No AKI Risk Injury Failure RIFLE 72 hr No AKI Risk Injury Failure
1 1.70 (1.03–2.81) 3.33 (2.13–5.22) 2.48 (1.55–3.98)
0.04 ,0.01 ,0.01
1 2.04 (1.12–3.72) 3.13 (1.93–5.07) 3.61 (2.24–5.80)
0.02 ,0.01 ,0.01
AKIN No Stage Stage Stage AKIN No Stage Stage Stage
48hr AKI 1 2 3 72hr AKI 1 2 3
P
1 1.57 (0.94–2.61) 3.25 (2.01–5.27) 2.88 (1.82–4.55)
0.08 ,0.01 ,0.01
1 2.24 (1.26–4.00) 3.24 (1.87–5.63) 4.12 (2.56–6.61)
,0.01 ,0.01 ,0.01
CI, confidence interval; RIFLE, Risk, Injury, Failure, Loss and End-stage; AKIN, Acute Kidney Injury Network.
354
Volume 344, Number 5, November 2012
RIFLE Versus AKIN in Septic Shock
FIGURE 2. Cumulative survival rate for 291 patients with severe sepsis and septic shock based on their RIFLE classification (A and B) and AKIN stage (C and D).
led to the misclassification of some patients. Finally, all of our patients had been admitted to a medical (thus nonsurgical) ICU, limiting the applicability of our findings to more heterogeneous populations or surgical patients.
CONCLUSION In summary, we retrospectively compared the ability of the RIFLE and AKIN criteria to predict AFI in patients with severe sepsis and septic shock. We found that these 2 classification systems had the same capacity to predict 28-day mortality rates. The incidence of AKI, as determined by AKIN, was slightly higher than that measured by RIFLE. However, the 2 criteria showed no clinically important differences when used to assess patient outcomes over a 72-hour observation period. ACKNOWLEDGMENT We are grateful to Eun-mi Cho for many hours of assistance in gathering data from medical records. REFERENCES 1. Angus DC, Linde-Zwirble WT, Lidicker J, et al. Epidemiology of severe sepsis in the United States: analysis of incidence, outcome, and associated costs of care. Crit Care Med 2001;29:1303–10. 2. Engel C, Brunkhorst FM, Bone HG, et al. Epidemiology of sepsis in Germany: results from a national prospective multicenter study. Intensive Care Med 2007;33:606–18. 3. Metnitz PG, Krenn CG, Steltzer H, et al. Effect of acute renal failure requiring renal replacement therapy on outcome in critically ill patients. Crit Care Med 2002;30:2051–8.
Ó 2012 Lippincott Williams & Wilkins
4. Joannidis M, Metnitz PG. Epidemiology and natural history of acute renal failure in the ICU. Crit Care Clin 2005;21:239–49. 5. Lassnigg A, Schmidlin D, Mouhieddine M, et al. Minimal changes of serum creatinine predict prognosis in patients after cardiothoracic surgery: a prospective cohort study. J Am Soc Nephrol 2004;15: 1597–605. 6. Bellomo R, Ronco C, Kellum JA, et al. 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. 7. Mehta RL, Kellum JA, Shah SV, et al. Acute kidney injury network: report of an initiative to improve outcomes in acute kidney injury. Crit Care 2007;11:R31. 8. Ostermann M, Chang RW. Acute kidney injury in the intensive care unit according to RIFLE. Crit Care Med 2007;35:1837–43; quiz 52. 9. Gibney N, Hoste E, Burdmann EA, et al. Timing of initiation and discontinuation of renal replacement therapy in AKI: unanswered key questions. Clin J Am Soc Nephrol 2008;3:876–80. 10. Bellomo R, Kellum JA, Ronco C. Defining and classifying acute renal failure: from advocacy to consensus and validation of the RIFLE criteria. Intensive Care Med 2007;33:409–13. 11. Ricci Z, Cruz D, Ronco C. The RIFLE criteria and mortality in acute kidney injury: a systematic review. Kidney Int 2008;73:538–46. 12. Hoste EA, Schurgers M. Epidemiology of acute kidney injury: how big is the problem? Crit Care Med 2008;36:S146–51. 13. Hoste EA, Clermont G, Kersten A, et al. RIFLE criteria for acute kidney injury are associated with hospital mortality in critically ill patients: a cohort analysis. Crit Care 2006;10:R73.
355
Kim et al
14. 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.
On behalf of the Working Group on Sepsis-Related Problems of the European Society of Intensive Care Medicine. Intensive Care Med 1996;22:707–10.
15. Lopes JA, Fernandes P, Jorge S, et al. 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.
19. National Kidney Foundation. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 2002;39:Sl-266.
16. Bone RC, Balk RA, Cerra FB, et al. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. The ACCP/SCCM Consensus Conference Committee. American College of Chest Physicians/Society of Critical Care Medicine. Chest 1992;101: 1644–55. 17. Knaus WA, Draper EA, Wagner DP, et al. APACHE II: a severity of disease classification system. Crit Care Med 1985;13:818–29. 18. Vincent JL, Moreno R, Takala J, et al. The SOFA (sepsis-related organ failure assessment) score to describe organ dysfunction/failure.
356
20. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988;44:837–45. 21. Bagshaw SM, Uchino S, Bellomo R, et al. Septic acute kidney injury in critically ill patients: clinical characteristics and outcomes. Clin J Am Soc Nephrol 2007;2:431–9. 22. Ando M, Mori J, Ohashi K, et al. A comparative assessment of the RIFLE, AKIN and conventional criteria for acute kidney injury after hematopoietic SCT. Bone Marrow Transplant 2010;45:1427–34.
Volume 344, Number 5, November 2012