Copyright @ 200 by the Shock Society. Unauthorized reproduction of this article is prohibited.9

RIFLE CLASSIFICATION FOR PREDICTING IN-HOSPITAL MORTALITYIN CRITICALLY ILL SEPSIS PATIENTS

Yung-Chang Chen,* Chang-Chyi Jenq,* Ya-Chung Tian,* Ming-Yang Chang,*Chan-Yu Lin,* Chih-Cheng Chang,† Horng-Chyuan Lin,† Ji-Tseng Fang,*

Chih-Wei Yang,* and Shu-Min Lin†

Departments of *Nephrology and †Thoracic Medicine, Chang Gung Memorial Hospital,Chang Gung University College of Medicine, Taipei, Taiwan

Received 01 Apr 2008; first review completed 15 Apr 2008; accepted in final form 17 Apr 2008

ABSTRACT—Severe sepsis and septic shock, often complicated by acute kidney injury (AKI), are the most commoncauses of mortality in noncoronary intensive care units (ICUs). This study investigates the outcomes of critically ill patientswith sepsis and elucidates the association between prognosis and risk of renal failure, injury to the kidney, failure of kidneyfunction, loss of kidney function, and end-stage renal failure (RIFLE) classification. A total of 121 sepsis patients wereadmitted to ICU from June 2003 to January 2004. Forty-seven demographic, clinical, and laboratory variables wereprospectively recorded for post hoc analysis as predictors of survival on the first day of ICU admission. Overall in-hospitalmortality rate was 47.9%. Mortality was significantly associated (chi-square for trend; P G 0.001) with RIFLE classification.Septic shock, RIFLE category, and number of organ system failures on the first day of ICU admission were independentpredictors of hospital mortality according to forward conditional logistic regression. The severity of RIFLE classificationcorrelated with organ system failure number and Acute Physiology and Chronic Health Evaluation (APACHE) II to IV andsequential organ failure assessment scores. Cumulative survival rates at 6-month follow-up after hospital dischargesignificantly (P G 0.05) differed between non-AKI versus RIFLE injury, non-AKI versus RIFLE failure (RIFLE-F), and RIFLErisk versus RIFLE F. At 6-month follow-up, full recovery of renal function was noted in 85% of surviving patients with AKI(RIFLE risk, RIFLE injury, and RIFLE-F). In conclusion, these findings are consistent with a role for RIFLE classification inaccurately predicting in-hospital mortality and short-term prognosis in ICU sepsis patients.

KEYWORDS—Acute renal failure, systemic inflammatory response syndrome, organ system failure, ICU, prognosis

INTRODUCTION

Sepsis is a constant concern in intensive care unit (ICU)

patients not only because of its high incidence but also

owing to the high mortality associated with this condition (1,

2). Sepsis is also a well-known risk factor for acute renal

failure (ARF); 35% to 50% of ARF cases in ICUs are

attributable to sepsis (3Y5). Mortality in this subgroup of

ARF patients is considerably higher than that in other ARF

subgroups (4, 5).

Acute kidney injury (AKI) is well recognized for its impact

on the ICU patient outcome. A literature review reveals as

many as 35 different definitions of ARF in critically ill

patients (6). The numerous definitions cause clinical con-

fusion and complicate data comparison (7, 8). The acronym

RIFLE, for Brisk of renal failure, injury to the kidney, failure

of kidney function, loss of kidney function, and end-stage

renal failure,[ was first proposed by the Acute Dialysis

Quality Initiative (ADQI) group in an attempt to standardize

the definition of ARF (9). To date, RIFLE classification has

been used to assess critically ill patients on renal replacement

therapy, cardiac surgery patients, heterogeneous ICU patients,

heterogeneous populations of hospitalized patients, and

unique populations such as patients requiring extracorporeal

membrane oxygenation for postcardiotomy cardiogenic shock

(10Y15).

Investigators and physicians with limited medical resources

require a reliable procedure for risk stratification and

monitoring of ARF patients during practice and clinical trials

of new treatments. This investigation focuses on the following

goals: (1) comparing the recently defined ARF (RIFLE

classification), organ system failure (OSF) number (16), and

general ICU prognostic models [Acute Physiology and

Chronic Health Evaluation (APACHE) II (17), APACHE III

(18), APACHE IV (19), and Sequential Organ Failure

Assessment (SOFA) (20)] for predicting in-hospital mortality

in critically ill patients with sepsis and (2) evaluating in-

hospital mortality predictors in ICU sepsis patients. All

assessed scores were those observed on the first day of ICU

admission.

MATERIALS AND METHODS

Patient information and data collectionThe local institutional review board approved this study and waived the

requirement for informed consent. This study enrolled 121 critically illpatients with sepsis. Records were collected for patients admitted to a 24-bedmedical ICU at a tertiary care referral center between June 2003 and January2004. The population of patients was recruited from the emergency depart-ment and medical wards. They were transferred to the medical ICU oncesepsis with organ failure was found and included in the study when shockdeveloped during their stay in the ICU. Additionally, patients who haddeveloped septic shock in the emergency department or medical wards wereincluded in this study if they had been transferred to the medical ICU within4 h. Exclusion criteria included ages younger than 18 years, hospital stay lessthan 24 h, or any history of the following: renal replacement therapy for

139

SHOCK, Vol. 31, No. 2, pp. 139Y145, 2009

Address reprint requests to Shu-Min Lin, MD, Chang Gung Memorial Hospital.

199 Tung Hwa North Road, Taipei, 105 Taiwan. E-mail: smlin100@gmail.com.

This study was supported by the Chang Gung Medical Research Fund, Chang

Gung Memorial Hospital, Linkou, Taiwan (grant no. CMRPG 361391), and in part

by National Science Council of Taiwan (grant no. NSC 96-2314-B-182A-035-

MY3).

DOI: 10.1097/SHK.0b013e31817d419e

Copyright � 2009 by the Shock Society

Copyright @ 200 by the Shock Society. Unauthorized reproduction of this article is prohibited.9

end-stage renal disease patients; pregnancy; acute cerebral or vascular eventsuch as acute coronary syndrome, acute pulmonary edema, status asthmaticus,or cardiac dysrhythmias (as a primary diagnosis); seizure; drug overdose; burninjury; trauma requiring emergent surgical treatment; uncured cancer;immunosuppression; or do-not-resuscitate status. Readmitted patients werealso excluded.

Post hoc analysis of a prospectively accumulated database examined thefollowing variables on the first day of ICU admission: demographic data,RIFLE category, OSF number, and APACHEs II to IV and SOFA scores.Length of hospitalization and in-hospital outcome were also recorded. Follow-up at 6 months after hospital discharge was performed via telephoneinterview. When necessary, the hospital registry office provided informationregarding patient survival or date of death.

DefinitionsSevere sepsis and septic shock were defined according to modified

American College of Chest Physicians and Society of Critical Care Medicineconsensus criteria (16). Patients with proven or suspected infection, two ormore systemic inflammatory response syndrome criteria, and an infection-induced organ dysfunction were classified as having severe sepsis. Septicshock was diagnosed when the systolic arterial blood pressure remained lessthan 90 mmHg despite adequate fluid resuscitation. Illness severity wasassessed by the following scoring systems: APACHEs II to IV and SOFAscores. A spreadsheet obtained from www.criticaloutcomes.cerner.com wasused to predict APACHE IV hospital mortality rates. Diagnosis at ICUadmission and length of hospitalization before ICU admission were recordedas primary reasons for intensive care and duration of hospitalization beforeICU admission, respectively (19). The worst physiological and biochemicalvalues on day of initial ICU admission were recorded. When the patient wasparalyzed or sedated, neurological scoring was not performed because thepatient was not considered to be in neurological failure. For inpatients whowere incubated but not sedated, the best verbal response was determinedaccording to clinical judgment (21). The number of organ failures wasrecorded within 24 h of ICU admission. The definition of organ failure wasaccording to the consensus committee of the American College of ChestPhysicians and Society of Critical Care Medicine: respiratory failure, need formechanical ventilation, cardiovascular failure, systolic blood pressure equal toor less than 90 mmHg or MAP equal to or less than 60 mmHg for 1 h despite

fluid bolus, renal failure, low urine output (e.g. G0.5 mL kgj1 hj1), increasedcreatinine level (Q50% increase from baseline) or need for acute dialysis,hematological failure, low platelet count (G100,000 mmj3) or prothrombintime/activated partial thromboplastin time exceeding the upper limit of normal,metabolic failure, low pH with high lactate (e.g. pH G7.30 and plasma lactategreater than the upper limit of normal), hepatic failure, liver enzymes greaterthan two times the upper limit of normal, central nervous system failure, alteredconsciousness, and reduced Glasgow Coma Scale score (16).

The ADQI group first proposed the RIFLE system at the second ADQIconference in Vicenza, Italy, in May 2002. Baseline serum creatinine (SCr)concentration was determined first during hospitalization. The modification ofdiet in renal disease formula was applied for 11 patients to estimate baselineSCr concentrations because those patients were admitted directly to an ICU,and their SCr concentrations at admission were unknown (9). Patients weregrouped according to RIFLE classification into risk, injury, and failurecategories. No patients met the criteria for loss or end-stage renal diseasecategories. A simple model for mortalityVnon-AKI (0 points), RIFLE risk(RIFLE-R; 1 point), RIFLE injury (RIFLE-I; 2 points), and RIFLE failure(RIFLE-F; 3 points) for day 1 of ICU admissionVwas used (22Y24). Fullrenal recovery was defined as SCr fell to less than threshold.

Statistical analysisDescriptive statistics were expressed as mean T SE. Primary analysis

compared hospital survivors with nonsurvivors. All variables were tested fornormal distributions using the Kolmogorov-Smirnov test. The Student t testwas applied to compare the means of continuous variables and normallydistributed data; otherwise, the Mann-Whitney U test was used. Categoricaldata were tested using the chi-square test. The ANOVA test with Scheffe testpost hoc for numerical values and the chi-square test for trends were used toassess categorical data associated with RIFLE classifications. Correlations ofpaired group variables were assessed by linear regression using Pearsonanalysis. Finally, risk factors were assessed by univariate analysis, andvariables that were statistically significant (P G 0.05) in the univariate analysiswere included in multivariate analysis by applying a multiple logisticregression based on forward elimination of data.

Calibration was assessed by Hosmer-Lemeshow goodness-of-fit test (Cstatistic) to compare the number of observed and predicted deaths in risk groupsfor the entire range of death probabilities. Discrimination was assessed using the

TABLE 1. Patient demographic data and clinical characteristics

All patients, n = 121 Survivors, n = 63 Nonsurvivors, n = 58 P

Age, yrs 68 T 1 69 T 2 68 T 2 NS (0.563)

Sex, M/F 81/40 42/21 39/19 NS (1.000)

Length of ICU stay, d 15 T 1 15 T 2 16 T 2 NS (0.645)

Body temperature ICU admission, -C 37.0 T 0.1 36.7 T 0.1 37.2 T 0.1 0.009

PaO2/FIO2 ICU first day, mmHg 254 T 13 262 T 16 245 T 21 NS (0.518)

AaDO2, ICU first day, mmHg 199 T 20 172 T 17 227 T 37 NS (0.171)

GCS, ICU first day, points 10 T 1 11 T 1 9 T 1 0.039

MAP, ICU admission, mmHg 79 T 2 80 T 2 77 T 2 NS (0.375)

SCr, ICU first day, mgdLj1 1.9 T 0.2 1.8 T 0.2 1.9 T 0.2 NS (0.812)

Arterial HCO3j, ICU first day 23.2 T 0.6 24.4 T 0.9 21.8 T 0.9 0.034

Bilirubin, ICU first day, mgdLj1 1.4 T 0.2 1.0 T0.2 1.8 T0.4 0.049

Albumin, ICU first day, g Lj1 2.5 T 0.1 2.6 T1 2.4 T1 0.024

Hemoglobin, ICU first day, gdLj1 10.3 T 0.2 10.7 T0.3 9.8 T0.3 0.047

Septic shock, ICU first day (yes/no) 93/28 39/24 54/4 G0.001

RIFLE 1.0 T 0.1 0.7 T 0.1 1.4 T 0.2 G0.001

OSF no. 2.8 T 0.1 2.2 T 0.1 3.4 T 0.2 G0.001

APACHE II, ICU first day, mean T SE 20 T 1 18 T 1 23 T 1 0.002

APACHE III, ICU first day, mean T SE 74 T 2 65 T 3 84 T 4 G0.001

APACHE IV, ICU first day, mean T SE 38.6 T 1.4 32.9 T 1.6 44.8 T 2.3 G0.001

SOFA, ICU first day, mean T SE 6.8 T 0.3 6.0 T 0.3 7.7 T0.4 0.001

AaDO2 indicates alveolar-arterial oxygen tension difference; F, female; FIO2, fraction of inspired oxygen; GCS, Glasgow Coma Scale M, male; NS, notsignificant; PaO2, arterial partial pressure of oxygen.

140 SHOCK VOL. 31, NO. 2 CHEN ET AL.

Copyright @ 200 by the Shock Society. Unauthorized reproduction of this article is prohibited.9

area under a receiver operating characteristic curve (AUROC). Areas under twoAUROC curves were compared by a nonparametric approach. Area under areceiver operating characteristic curve analysis was also conducted to estimatethe cutoff values, sensitivity, specificity, overall correctness, and positive andnegative predictive values. Finally, cutoff points were calculated by determiningthe best Youden index (sensitivity + specificity j 1).

Cumulative survival curves as a function of time were generated byKaplan-Meier approach and compared by log-rank test. All statistical

tests were two-tailed; a value of P G 0.05 was considered statisticallysignificant.

RESULTS

Subject characteristics

Between June 2003 and January 2004, 121 critically ill

patients treated for sepsis at a single medical ICU were

TABLE 2. Bacterial studies, chronic coexisting conditions, complications, and OSF

All patients(n = 121), n (%)

Hospital survivors(n = 63), n (%)

Hospital nonsurvivors(n = 58), n (%) P

Bacterial studies

Blood culture positive 27 (22) 18 (29) 9 (16) NS (0.085)

Sputum culture positive 82 (68) 43 (68) 39 (67) NS (0.905)

Urine culture positive 20 (17) 12 (19) 8 (14) NS (0.437)

Fungus culture positive 10 (8) 5 (8) 5 (9) NS (1.000)

Presence of Acinetobacter baumannii 50 (41) 32 (51) 19 (33) NS (0.065)

Presence of multiple drug-resistant bacteria 47 (39) 19 (30) 28 (48) NS (0.061)

All cultures negative 26 (21) 12 (19) 14 (24) NS (0.515)

Chronic coexisting conditions

Diabetes mellitus 34 (28) 18 (29) 16 (28) NS (0.904)

Hypertension 44 (36) 25 (40) 19 (33) NS (0.429)

Congestive heart failure 16 (13) 5 (8) 11 (19) NS (0.074)

Coronary arterial disease 10 (8) 7 (11) 3 (5) NS (0.327)

Arrhythmia 21 (17) 6 (10) 15 (26) 0.018

Neurological disease 30 (25) 16 (25) 14 (24) NS (0.873)

COPD 13 (11) 7 (11) 6 (10) NS (1.000)

History of malignancy 35 (29) 17 (27) 18 (31) NS (0.690)

Complications

ARDS 14 (12) 6 (10) 8 (14) NS (0.463)

Nosocomial pneumonia 46 (380) 23 (37) 23 (40) NS (0.722)

Cardiopulmonary edema 11 (9) 6 (10) 5 (9) NS (1.000)

DIC 27 (22) 7 (11) 20 (34) 0.002

Gastrointestinal bleeding 18 (15) 5 (8) 13 (22) 0.025

Organ system failure

Central nervous system failure 29 (24) 10 (16) 19 (33) 0.030

Cardiovascular failure 64 (53) 28 (44) 36 (62) NS (0.052)

Respiratory failure 116 (96) 58 (92) 58 (100) NS (0.058)

Hepatic failure 18 (15) 6 (10) 12 (21) NS (0.085)

Renal failure 47 (39) 17 (27) 30 (52) 0.005

Hematological failure 19 (16) 8 (13) 11 (19) NS (0.344)

Metabolic failure 18 (15) 9 (14) 9 (16) NS (1.000)

ARDS indicates acute respiratory distress syndrome; COPD, chronic obstructive pulmonary disease; DIC, disseminated intravascular coagulation; NS, not significant.

TABLE 3. RIFLE classification for AKI

RIFLE n Hospital mortality, % "-Coefficient Standard error Odds ratios, 95% CI P

Non-AKI 53 34 0 0 1 (reference) V

Risk 32 40.6 0.285 0.462 1.330 (0.538 Y 3.292) NS (0.537)

Injury 19 73.7 1.695 0.596 5.444 (1.692 Y 17.519) 0.004

Failure 17 76.5 1.844 0.641 6.319 (1.799 Y 22.203) 0.004

CI indicates confidence interval; NS, not significant.

SHOCK FEBRUARY 2009 IN-HOSPITAL MORTALITY IN SEPSIS PATIENTS 141

Copyright @ 200 by the Shock Society. Unauthorized reproduction of this article is prohibited.9

enrolled in the study. Patient median age was 73 years; 81

(67%) were men, and 40 (33%) were women. Overall in-

hospital mortality was 47.9% (58/121), and 6-month mortality

rate was 59.5% (72/121). Table 1 lists patient demographic

data and clinical characteristics of both in-hospital survivors

and nonsurvivors. Table 2 presents bacterial studies, chronic

coexisting conditions, complications, and OSF. Twenty-six

(21%) patients were negative in cultures for blood, sputum,

urine, and fungus.

Hospital mortality, short-term prognosis, and severity ofillness scoring systems

Hospital mortality was 34% (18/53) for non-AKI patients,

40.6% (13/32) for RIFLE-R, 73.7% (14/19) for RIFLE-I, and

76.5% (13/17) for RIFLE-F (chi-square for trend; P G 0.001).

In all patients, RIFLE severity correlated with mortality. Odds

ratios for RIFLE criteria were 1.33 (P = 0.537) for RIFLE-R

versus non-AKI, 5.444 (P = 0.004) for RIFLE-I versus non-

AKI, and 6.319 (P = 0.004) for RIFLE-F versus non-AKI

(Table 3).

To assess calibration, Table 4 lists goodness of fit, as

measured by the Hosmer-Lemeshow chi-square analysis of

predicted mortality risk, the predictive accuracy of the RIFLE

category, OSF number, APACHEs II to IV, and SOFA scores

for these patients. Table 4 also compares the discriminatory

power of the six scoring systems. The AUROC analysis

verified the OSF number had the best discriminatory power.

To assess the values of selected cutoff points for predicting

hospital mortality, the sensitivity, specificity, overall correct-

ness of prediction, and positive and negative predictive values

were all determined. The OSF number revealed the best

Youden index and highest overall correctness of prediction.

Hospital mortality rates differed significantly (P G 0.001) less

than and more than cutoffs for RIFLE-I of RIFLE classifica-

tions, 3 OSF numbers, 20 APACHE II points, 80 APACHE III

points, 40 APACHE IV points, and 8 SOFA points.

Next, RIFLE classification, OSF number, APACHEs II to

IV, and SOFA scores were examined for correlations. Table 5

lists correlations between scoring systems used on the first

day of ICU admission. The six scores in Table 5 exhibited

positive correlations (all r 9 0.36; P G 0.001). Figures 1, A to

D show OSF number, APACHEs II and IV, and SOFA scores

according to RIFLE classification on ICU day 1. In all

patients, RIFLE classification correlated with those scores.

Cumulative survival rates differed significantly (P G 0.05)

for non-AKI versus RIFLE-I and RIFLE-F and for RIFLE-R

versus RIFLE-F (Fig. 2). In 6-month survivors, full renal

recovery was achieved in 12 (92.3%) patients in the RIFLE-R

group, 4 (80%) in the RIFLE-I group, and 1 (50%) in the

RIFLE-F group. Three (15%) patients did not fully recover.

Risk factors for hospital mortality

Univariate analysis identified 16 (Table 6) of 47 variables

(Tables 1 and 2) as prognostically valuable. Multivariate

analysis identified the following variables as independent

prognostic significance: septic shock, RIFLE score, and OSF

number (Table 6). Regression coefficients of these variables

were used to calculate a logit of death for each patient as

follows:

The logarithm of odds of death = j1.491 + 0.424 � septic

shock (with = 1/without = 0) + 0.596 � RIFLE classification

(non-AKI = 0/RIFLE-R = 1/RIFLE-I = 2/RIFLE-F = 3) +

0.424 � OSF number.

DISCUSSION

Overall in-hospital mortality rate was 47.9%, which is

consistent with previous studies indicating the poor prognosis

TABLE 4. Calibration and discrimination for the scoring methods in predicting hospital mortality

Calibration Discrimination

Goodness of fit, chi-square df P AUROC T SE 95% CI P

RIFLE 5.238 2 0.073 0.678 T 0.049 0.581 Y 0.775 0.001

OSF no. 2.046 3 0.563 0.742 T 0.045 0.655 Y 0.829 G0.001

APACHE II 4.608 8 0.799 0.660 T 0.050 0.562 Y 0.758 0.002

APACHE III 12.350 8 0.136 0.701 T 0.048 0.606 Y 0.795 G0.001

APACHE IV 5.307 8 0.724 0.710 T 0.047 0.618 Y 0.802 G0.001

SOFA 4.657 7 0.702 0.669 T 0.049 0.573 Y 0.765 0.001

CI indicates confidence interval; df, degree of freedom.

FIG. 1. Organ system failure number (A), APACHEs III and IV (B, C), and SOFA (D) scores according to RIFLE classification on ICU day 1 (*P G 0.05vs. non-AKI; †P G 0.05 vs. RIFLE-R; ‡P G 0.05 vs. RIFLE-I).

142 SHOCK VOL. 31, NO. 2 CHEN ET AL.

Copyright @ 200 by the Shock Society. Unauthorized reproduction of this article is prohibited.9

of septic patients (25Y27). This investigation revealed that the

predictors of septic shock, RIFLE category, and OSF number

are significantly associated with in-hospital mortality in

critically ill patients with sepsis (Table 6). During septic

shock, global tissue hypoxia caused by an imbalance between

systemic oxygen delivery and oxygen demand results in renal

tubular necrosis, multiple organ failure, and increased mortal-

ity rate (28Y30).

The pathogenesis of experimental ARF secondary to

endotoxemia is the compensation for impaired hemody-

namics by up-regulation of vasoconstrictor systems and renal

vasoconstriction (31Y33). Although ARF is an independent

predictor of mortality, the leading causes of death associated

with ARF are nonrenal complications, typically those related

to multiorgan dysfunction. The RIFLE score on ICU day 1

was an independent predictor of hospital mortality, whereas

SCr concentrations on admission or on ICU day 1 and urine

output on ICU day 1 were not predictive of mortality. The

FIG. 2. Cumulative survival rate for 121 critically ill patients withsepsis based on their RIFLE classification.

TABLE 6. Prognostically significant variables for hospital mortality

Parameter "-Coefficient Standard error Odds ratios, 95% CI P

Univariate logistic regression

Body temperature 0.518 0.203 1.679 (1.129 Y 2.498) 0.011

GCS j0.086 0.042 0.917 (0.844 Y 0.996) 0.040

Arterial HCO3j, ICU first day j0.060 0.029 0.941 (0.889 Y 0.997) 0.038

Albumin, ICU first day j0.753 0.343 0.471 (0.241 Y 0.922) 0.028

Septic shock, ICU first day 2.117 0.580 8.308 (2.668 Y 25.868) G0.001

RIFLE classification, ICU first day 0.681 0.191 1.976 (1.3458 Y 2.875) G0.001

OSF no., ICU first day 0.746 0.173 2.108 (1.501 Y 2.962) G0.001

APACHE II, ICU first day 0.077 0.027 1.080 (1.025 Y 1.138) 0.004

APACHE III, ICU first day 0.033 0.009 1.034 (1.016 Y 1.052) G0.001

APACHE IV, ICU first day 0.036 0.009 1.036 (1.017 Y 1.056) G0.001

SOFA, ICU first day 0.229 0.073 1.257 (1.089 Y 1.451) 0.002

Arrhythmia 1.198 0.524 3.314 (1.188 Y 9.247) 0.022

DIC 1.438 0.487 4.211 (1.622 Y 10.933) 0.003

Gastrointestinal bleeding j1.209 0.562 0.288 (0.099 Y 0.899) 0.032

Central nervous system failure j0.949 0.444 0.387 (0.162 Y 0.925) 0.033

Renal failure 1.064 0.387 2.899 (1.358 Y 6.187) 0.006

Multivariate logistic regression

Septic shock 1.733 0.647 5.658 (1.593 Y 20.095) 0.007

RIFLE 0.596 0.226 1.815 (1.166 Y 2.825) 0.008

OSF no. 0.424 0.192 1.527 (1.047 Y 2.227) 0.028

Constant j1.491 0.587 0.225 0.011

CI indicates confidence interval; DIC, disseminated intravascular coagulation; GCS, Glasgow Coma Scale.

TABLE 5. Correlation between scoring systems on the first dayof ICU admission

ScoresOSF

score, rAPACHE

II, rAPACHE

III, rAPACHE

IV, rSOFA,

r

RIFLE 0.363 0.397 0.443 0.422 0.458

OSF no. V 0.423 0.526 0.539 0.573

APACHE II V 0.634 0.613 0.476

APACHE III V 0.911 0.634

APACHE IV V 0.656

Six scores revealed positive correlations with each other (all P G 0.001).

SHOCK FEBRUARY 2009 IN-HOSPITAL MORTALITY IN SEPSIS PATIENTS 143

Copyright @ 200 by the Shock Society. Unauthorized reproduction of this article is prohibited.9

RIFLE criteria are various levels of AKI severity (risk, injury,

and failure). A trend toward significantly increased mortality

as RIFLE score increased was apparent in all patients (risk,

1.330; injury, 5.444; failure, 6.319; Table 3). Renal dysfunc-

tion in this study was significantly related to OSF number and

APACHEs II to IV, and SOFA scores, and may have

contributed to mortality (Fig. 1 and Table 5). The analytical

results in this study demonstrate that RIFLE criteria precisely

predict hospital mortality and short-term prognosis (Fig. 2

and Table 3) in this subset of critically ill patients with sepsis.

Excluding patients who died within 6 months, full recovery of

renal function was very high in those with AKI (85%).

Although few studies have reported renal recovery in patients

with AKI, some have shown that most patients recover

adequate renal function (34).

Previous studies show that endotoxemia induces a rapid

formation of microthrombi in hepatic circulation and sub-

sequently causes multiple fibrin clots, resulting in focal areas

of hypoperfusion, tissue necrosis, and multiple organ dys-

functions (35Y37). Schwilk et al. (38) reported mortality rates

of 12%, 38%, 72%, 90%, and 100% in patients with one to

five organ failures, respectively. This study revealed mortality

rates of 15.8%, 35.1%, 61.3%, 60%, 78.6%, and 100% in

patients with one to six organ failures, respectively. The

prognosis of critically ill patients with sepsis is apparently

related to the number of separate organ failures. Mortality

rates in this study substantially increased when the number of

failed organs is three or more.

Despite the promising analytical results of this study,

several important limitations should be recognized. First, this

study was conducted at a single institution; consequently, the

results may not be directly extrapolated to other patient

populations. Second, scoring was performed only on the first

day of ICU admission. Sequential application of these scoring

systems (for example, daily or weekly) may reflect the

dynamic aspects of clinical diseases and thus provide more

complete data for mortality risk. Third, the predictive

accuracy of logistic regression models is also subject to

limitations. Finally, use of the prognostic instruments in

patients already admitted to ICUs, rather than as a preadmis-

sion screening tool, may have skewed measurement results.

In conclusion, this study revealed a 6-month mortality rate

of 59.5% in critically ill patients with sepsis. The short-term

prognosis for these patients is poor. This study also elucidated

the predictors of septic shock, RIFLE classification, and OSF

number that are independently associated with in-hospital

mortality. The RIFLE criteria classified 56.2% of ICU septic

patients with some degree of ARF. After discharge from the

hospital, most of those who developed AKI fully recovered

renal function. The analytical results of this study suggest that

RIFLE category accurately predicts in-hospital mortality and

short-term prognosis in this subset of patients.

REFERENCES1. Schrier RW, Wang W: Acute renal failure and sepsis. N Engl J Med 351:

159Y169, 2004.

2. Ronco C, Kellum JA, Bellomo R, House AA: Potential interventions in sepsis-

related acute kidney injury. Clin J Am Soc Nephrol 3:531Y544, 2008.

3. Brivet FG, Kleinknecht DJ, Loirat P, Landais PJ: Acute renal failure in

intensive care unitsVcauses, outcome, and prognostic factors of hospital

mortality; a prospective, multicenter study. French Study Group on Acute

Renal Failure. Crit Care Med 24:192Y198, 1996.

4. Liano F, Junco E, Pascual J, Madero R, Verde E: The spectrum of acute renal

failure in the intensive care unit compared with that seen in other settings. The

Madrid Acute Renal Failure Study Group. Kidney Int 66:S16YS24, 1998.

5. Fang JT, Tsai MH, Tian YC, Jenq CC, Lin CY, Chen YC, Lien JM, Chen PC,

Yang CW: Outcome predictors and new score of critically ill cirrhotic patients

with acute renal failure. Nephrol Dial Transplant 23:1961Y1969, 2008.

6. Kellum JA, Levin N, Bouman C, Lameire N: Developing a consensus

classification system for acute renal failure. Curr Opin Crit Care 8:509Y514,

2002.

7. Schrier RW, Wang W, Poole B, Mitra A: Acute renal failure: definitions,

diagnosis, pathogenesis, and therapy. J Clin Invest 114:5Y14, 2004.

8. Lameire N, Van Biesen W, Vanholder R: Acute renal failure. Lancet 365:

417Y430, 2005.

9. Bellomo R, Ronco C, Kellum JA, Mehta RL, Palevsky P: Acute Dialysis

Quality Initiative workgroup: acute renal failureVdefinition, outcome meas-

ures, animal models, fluid therapy and information technology needs: the

Second International Consensus Conference of the Acute Dialysis Quality

Initiative (ADQI) Group. Crit Care 8:R204YR212, 2004.

10. Bellomo R, Kellum JA, Ronco C: Defining and classifying acute renal failure:

from advocacy to consensus and validation of the RIFLE criteria. IntensiveCare Med 33:409Y413, 2007.

11. Uchino S, Bellomo R, Goldsmith D, Bates S, Ronco C: An assessment of the

RIFLE criteria for acute renal failure in hospitalized patients. Crit Care Med34:1913Y1917, 2006.

12. 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 10:R73,

2006.

13. Ostermann M, Chang RW: Acute kidney injury in the intensive care unit

according to RIFLE. Crit Care Med 35:1837Y1843, 2007.

14. 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. NephrolDial Transplant 23:1203Y1210, 2008.

15. Ricci Z, Cruz D, Ronco C: The RIFLE criteria and mortality in acute kidney

injury: a systematic review. Kidney Int 73:538Y546, 2008.

16. American College of Chest Physicians/Society of Critical Care Medicine

Consensus Conference: Definitions for sepsis and organ failure and guide-

lines for the use of innovative therapies in sepsis. Crit Care Med 20:

864Y874, 1992.

17. Knaus WA, Draper EA, Wagner DP, Zimmerman JE: APACHE II: a severity

of disease classification system. Crit Care Med 13:818Y824, 1985.

18. Knaus WA, Wagner DP, Draper EA, Zimmerman JE, Bergner M, Bastos PG:

The APACHE III prognostic system: risk prediction of hospital mortality for

critically ill hospitalized adults. Chest 100:1619Y1636, 1991.

19. Zimmerman JE, Kramer AA, McNair DS, Malila FM, Shaffer VL: Intensive

care unit length of stay: benchmarking based on Acute Physiology and Chronic

Health Evaluation (APACHE) IV. Crit Care Med 34:2517Y2529, 2006.

20. 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 Sepsis-Related Problems of the European Society of

Intensive Care Medicine. Intensive Care Med 22:707Y710, 1996.

21. Tsai MH, Chen YC, Ho YP, Fang JT, Lien JM, Chiu CT, Liu NJ, Chen PC:

Organ system failure scoring system can predict hospital mortality in critically

ill cirrhotic patients. J Clin Gastroenterol 37:251Y257, 2003.

22. Lin CY, Chen YC, Tsai FC, Tian YC, Jenq CC, Fang JT, Yang CW: RIFLE

classification is predictive of short-term prognosis in critically ill patients with

acute renal failure supported by extracorporeal membrane oxygenation.

Nephrol Dial Transplant 21:2867Y2873, 2006.

23. Jenq CC, Tsai MH, Tian YC, Lin CY, Yang C, Liu NJ, Lien JM, Chen YC,

Fang JT, Chen PC, et al.: RIFLE classification can predict short-term prognosis

in critically ill cirrhotic patients. Intensive Care Med 33:1921Y1930, 2007.

24. Lin CY, Tsai FC, Tian YC, Jenq CC, Chen YC, Fang JT, Yang CW:

Evaluation of outcome scoring systems for patients on extracorporeal

membrane oxygenation. Ann Thorac Surg 84:1256Y1262, 2007.

25. Lin SM, Huang CD, Lin HC, Liu CY, Wang CH, Kuo HP: A modified

goal-directed protocol improves clinical outcomes in intensive care unit

patients with septic shock: a randomized controlled trial. Shock 26:

551Y557, 2006.

144 SHOCK VOL. 31, NO. 2 CHEN ET AL.

Copyright @ 200 by the Shock Society. Unauthorized reproduction of this article is prohibited.9

26. Moreno RP, Metnitz B, Adler L, Hoechtl A, Bauer P, Metnitz PG: SAPS 3

Investigators: sepsis mortality prediction based on predisposition, infection and

response. Intensive Care Med 34:496Y504, 2008.

27. Routsi C, Pratikaki M, Sotiropoulou C, Platsouka E, Markaki V, Paniara O,

Vincent JL, Roussoss C: Application of the sequential organ failure assessment

(SOFA) score to bacteremic ICU patients. Infection 35:240Y244, 2007.

28. Lin SM, Frevert CW, Kajikawa O, Wurfel MM, Ballman K, Mongovin S,

Wong VA, Selk A, Martin TR: Differential regulation of membrane CD14

expression and endotoxin-tolerance in alveolar macrophages. Am J Respir CellMol Biol 31:162Y170, 2004.

29. Samransamruajkit R, Hiranrat T, Prapphal N, Sritippayawan S, Deerojanawong

J, Poovorawan Y: Levels of protein C activity and clinical factors in early

phase of pediatric septic shock may be associated with the risk of death. Shock28:518Y523, 2007.

30. Sakr Y, Vincent JL, Schuerholz T, Filipescu D, Romain A, Hjelmqvist H,

Reinhart K: Early- versus late-onset shock in European intensive care units.

Shock 28:636Y643, 2007.

31. Poole B, Wang W, Chen YC, Zolty E, Falk S, Mitra A, Schrier R: Role of

heme oxygenaseY1 in endotoxemic acute renal failure. Am J Physiol RenalPhysiol 289:F1382YF1385, 2005.

32. Kadoi Y, Goto F: Effects of selective iNOS inhibition on systemic

hemodynamics and mortality rate on endotoxic shock in streptozotocin-

induced diabetic rats. Shock 28:602Y609, 2007.

33. Versteilen AM, Heemskerk AE, Groeneveld AB, van Wijhe M, van

Lambalgen AA, Tangelder GJ: Mechanisms of the urinary concentration

defect and effect of desmopressin during endotoxemia in rats. Shock 29:

217Y222, 2008.

34. Ali T, Khan I, Simpson W, Prescott G, Townend J, Smith W, Macleod A:

Incidence and outcomes in acute kidney injury: a comprehensive population-

based study. J Am Soc Nephrol 18:1292Y1298, 2007.

35. Faust SN, Levin M, Harrison OB, Goldin RD, Lockhart MS, Kondaveeti S,

Laszik Z, Esmon CT, Heyderman RS: Dysfunction of endothelial protein C

activation in severe meningococcal sepsis. N Engl J Med 345:408Y416, 2001.

36. Derikx JP, Poeze M, van Bijnen AA, Buurman WA, Heineman E: Evidence

for intestinal and liver epithelial cell injury in the early phase of sepsis. Shock28:544Y548, 2007.

37. Lendemans S, Peszko A, Oberbeck R, Schmitz D, Husain B, Burkhard M,

Schade FU, Flohe S: Microcirculatory alterations of hepatic and mesenteric

microcirculation in endotoxin tolerance. Shock 29:223Y231, 2008.

38. Schwilk B, Wiedeck H, Stein B, Reinelt H, Treiber H, Bothner U:

Epidemiology of acute renal failure and outcome of haemodiafiltration in

intensive care. Intensive Care Med 23:1204Y1211, 1997.

SHOCK FEBRUARY 2009 IN-HOSPITAL MORTALITY IN SEPSIS PATIENTS 145