KIDNEY DYSFUNCTION is a common

Transcription

1 One-Year Mortality in Critically Ill Patients by Severity of Kidney Dysfunction: A Population-Based Assessment Sean M. Bagshaw, MD, MSc, Garth Mortis, MD, Christopher J. Doig, MD, MSc, Tomas Godinez-Luna, MD, Gordon H. Fick, PhD, and Kevin B. Laupland, MD, MSc Background: Kidney dysfunction in the intensive care unit (ICU) results in increased morbidity, mortality, and health care costs; however, long-term mortality has not been described across strata of severity in kidney dysfunction. Methods: The primary objective is to describe and assess factors associated with 1-year mortality in critically ill patients stratified by severity of kidney dysfunction during admission to the ICU. Kidney dysfunction is defined by peak serum creatinine values and stratified by: (1) no dysfunction (creatinine < 1.7 mg/dl [<150 mol/l]), (2) mild dysfunction (creatinine, 1.7 to 3.4 mg/dl [150 to 299 mol/l]), (3) moderate dysfunction (creatinine > 3.4 mg/dl [> 300 mol/l]), (4) severe acute dysfunction requiring renal replacement therapy (acute renal failure), or (5) preexisting end-stage kidney disease. Population-based surveillance was of adult residents of the Calgary Health Region (population, 1 million) admitted to any multidisciplinary ICU and a cardiovascular surgery ICU from May 1, 1999, to April 30, Results: Of 5,693 admissions, 62% were men, median age was 64.9 years (interquartile range, 50.6 to 74.5 years), and mean Acute Physiology and Chronic Health Evaluation (APACHE) II score was (SD). Case fatality rates stratified by renal dysfunction were 17% (763 of 4,411), 47% (370 of 790), 48% (77 of 160), 64% (153 of 240), and 40% (37 of 92) for no, mild, and moderate dysfunction; severe acute renal failure; and end-stage kidney disease, respectively. By means of multivariate analysis, 1-year mortality was associated independently with advancing age, medical diagnosis, higher APACHE II score, and presence and severity of kidney dysfunction, although no difference was evident comparing those with mild to moderate dysfunction. End-stage kidney disease was not associated independently with 1-year mortality. Conclusion: Severity of kidney dysfunction in patients in the ICU is associated with an incremental increase in long-term mortality. Although patients classified with either mild or moderate kidney dysfunction had an increased risk for death, use of serum creatinine level alone was poor at discriminating long-term outcome, suggesting this measure alone should not be used for defining long-term prognosis. Am J Kidney Dis 48: by the National Kidney Foundation, Inc. INDEX WORDS: Kidney dysfunction; critical care; long-term mortality; population based. From the Departments of Critical Care Medicine, Community Health Sciences, Medicine, and Pathology and Laboratory Medicine, Calgary Health Region and University of Calgary, Alberta, Canada. Received February 21, 2006; accepted in revised form June 6, Originally published online as doi: /j.ajkd on July 28, Support: This study was funded by a grant from the Canadian Intensive Care Foundation. S.M.B. is supported by an Alberta Heritage Foundation for Medical Research Clinical Fellowship and a Royal College and Physicians and Surgeons of Canada Detweiler Traveling Fellowship. Potential conflicts of interest: None. Address reprint requests to Sean M. Bagshaw, MD, MSc, Department of Critical Care Medicine, Foothills Medical Centre, th St NW, Calgary, Alberta, Canada T2N 2T9. smbagsha@ucalgary.ca 2006 by the National Kidney Foundation, Inc /06/ $32.00/0 doi: /j.ajkd KIDNEY DYSFUNCTION is a common complication in critically ill patients and traditionally has been associated with a high burden of illness, increased morbidity, high mortality, and consumption of health care resources. 1-5 Although important advances have occurred in the understanding and treatment of acute renal failure (ARF) in critically ill patients, the current literature remains confusing because of a lack of consensus of an accepted and validated definition. 6 Numerous ARF definitions have been used in the literature, which makes comparisons across clinical studies of outcomes, therapies, and prevention problematic. Several epidemiological studies and clinical trials used as the case definition for ARF simple absolute or relative changes in serum creatinine levels as surrogates for changes in kidney function; however, use of serum creatinine level or other biochemical markers alone may be limited. 1,2,7-18 For example, the population-based study by Liano and Pascual 2 reported an incidence of 6.6/100,000 critically ill patients per year, with ARF defined as a serum creatinine level of 2.0 mg/dl or greater ( 177 mol/l) in subjects with prior normal kidney function or 50% or greater increase in those with preexisting chronic kidney disease. In a multicenter low- 402 American Journal of Kidney Diseases, Vol 48, No 3 (September), 2006: pp

2 MORTALITY BY SEVERITY OF KIDNEY DYSFUNCTION 403 dose dopamine trial, Bellomo et al 12 defined early kidney dysfunction as a serum creatinine level of 1.7 mg/dl or greater ( 150 mol/l), an increase greater than 0.9 mg/dl ( 80 mol/l) within 24 hours, or oliguria (urine output 0.5 ml/kg/h over 4 hours). In a multicenter study from 40 intensive care units (ICUs) in 16 countries, de Mendonça et al 7 reported that ARF developed in 24.7% of critically ill patients. ARF in this study was defined as a serum creatinine level of 3.4 mg/dl or greater ( 300 mol/l) or urine output less than 500 ml/d. These studies incorporated criteria for ARF based on absolute or relative changes in serum creatinine levels for several reasons. Specifically, these criteria are simple, objective, and easy to identify and measure. 1 Furthermore, these criteria are highly specific and ensure capture of critically ill patients with marked kidney dysfunction. 19 However, no study to date has assessed the performance and impact of serum creatinine level alone as a marker of severity of kidney dysfunction on long-term mortality. Likewise, the effect of different degrees of kidney dysfunction, defined by peak serum creatinine level not requiring renal replacement therapy, on long-term mortality has not been well described. We used absolute values of serum creatinine, need for acute renal replacement therapy, or presence of end-stage kidney disease requiring long-term renal replacement therapy to stratify kidney function in critically ill patients. Our primary objective in this study is to describe and assess factors associated with 1-year mortality outcome stratified by severity of kidney dysfunction during admission to the ICU in a large heterogeneous cohort of critically ill patients. METHODS Patient Population The Calgary Health Region provides hospital care to virtually all residents of the cities of Calgary, Airdrie, and surrounding nearby communities. 20 Adult critically ill patients in the Calgary Health Region are managed in closed ICUs by dedicated intensivists under the direction of the Department of Critical Care Medicine, University of Calgary, and the Calgary Health Region. The study population was restricted to adult ( 18 years) residents of the Calgary Health Region admitted to any of the 3 multidisciplinary ICUs or the cardiovascular surgery ICU in the 3 adult hospital sites from May 1, 1999, to April 30, The study protocol was approved by the Conjoint Health Research Ethics Board at the University of Calgary and Calgary Health Region before commencement. Study Protocol The study used a population-based surveillance cohort design. 21 As previously described, an intradepartmental database, the ICU Tracer database, a clinical research and departmental support database that prospectively and routinely records data for all patients admitted to adult ICUs in the Calgary Health Region, was used to capture data for all admissions to the study ICUs. 5 For each admission, peak serum creatinine levels and whether renal replacement therapy was performed during admission to the ICU were ascertained. Study Definitions Severity of kidney dysfunction was stratified by using peak serum creatinine level during ICU admission, similar to that reported in several previous clinical studies. 7,9,12,22 Thus, kidney dysfunction was stratified as no renal failure (serum creatinine 1.7 mg/dl [ 150 mol/l]), mild dysfunction (serum creatinine, 1.7 to 3.4 mg/dl [150 to 299 mol/l]), moderate dysfunction (serum creatinine 3.4 mg/dl [ 300 mol/l]), severe ARF with need for renal replacement therapy, 5 and the presence of end-stage kidney disease requiring long-term renal replacement therapy before ICU admission. Renal replacement therapy encompassed either continuous renal replacement therapy or intermittent hemodialysis. Severity of illness at ICU admission was assessed by using Acute Physiology and Chronic Health Evaluation (APACHE) II score. 23 Data Sources Basic ICU admission data, including demographics, clinical information, and laboratory data, were ascertained by the ICU Tracer database. Long-term mortality outcome status was obtained through linkage of the Death Registration Database maintained by Alberta Vital Statistics and the Alberta Health and Wellness registry. Alberta Health and Wellness maintains information for all residents of Alberta eligible for publicly funded health care coverage ( 99% of the population is included in this registry). The use of both information systems ensured completeness of the linkage process and ascertainment of mortality status at 1 year. 5 Data were exported from the source databases and linked by using Access 2003 (Microsoft Corp, Redmond, WA). Data Analysis Analysis was performed using Stata, version 8.2 (Stata Corp, College Station, TX). Normally or near-normally distributed variables are reported as mean SD and compared by using Student t-test or 1-way analysis of variance. Categorical data were compared by using Mantel-Haenszel or Fisher exact test, when appropriate. A multivariable logistic regression model was developed to assess independent factors associated with 1-year mortality. An initial model included selected variables thought to potentially confound the association of severity of kidney dysfunction and death at 1 year, including: (1) sex, (2) age, (3) severity of

3 404 BAGSHAW ET AL Table 1. Baseline Characteristics of Critically Ill Patients Stratified by Severity of Kidney Dysfunction Variable No Renal Dysfunction (n 4,411) Serum Creatinine 1.7 to 3.4 mg/dl (n 790) Serum Creatinine 3.4 mg/dl (n 160) Severe ARF (n 240) End-Stage Renal Disease/ Long-Term Renal Replacement Therapy (n 92) Age (y)* Male sex 2,724 (62) 504 (64) 108 (68) 140 (58) 57 (62) Admission type Medical 1,538 (35) 443 (56) 118 (74) 153 (64) 44 (48) Noncardiac surgical 1,169 (27) 225 (29) 35 (22) 56 (23) 26 (28) Cardiac surgical 1,697 (38) 119 (15) 6 (4) 31 (13) 22 (24) APACHE II NOTE. Values expressed as mean SD or number (percent). To convert creatinine in mg/dl to mol/l, multiply by *Overall 1-way analysis of variance, P Fisher exact test, P Fisher exact test, P Overall 1-way analysis of variance, P illness (APACHE II score), (4) admission type (medical, noncardiac surgical, or cardiac surgical), and (5) presence and severity of kidney dysfunction. Sequential elimination of variables was performed by means of the likelihood ratio method. Model calibration and discrimination were assessed by using the Hosmer-Lemeshow goodness-of-fit test and area under the receiver operating characteristic curve, respectively. Results are reported as odds ratios with 95% confidence intervals. Long-term survival outcome of critically ill patients by severity of kidney dysfunction is described graphically as crude Kaplan-Meier curves truncated at 1 year. Equality in the estimated survival function curves across groups was assessed by means of log-rank test. RESULTS During the study period, 5,693 adult residents of the Calgary Health Region had 6,762 admissions to a regional ICU. Sixty-two percent were men, median age was 64.9 years (interquartile range [IQR], 50.6 to 74.5 years), and mean APACHE II score was at ICU admission. Baseline patient characteristics stratified by severity of kidney dysfunction are listed in Table 1. Critically ill patients with any kidney dysfunction were significantly older (mean age, versus years; P ), had higher mean APACHE II scores ( versus ; P ), and were more likely to be medical, rather than surgical, patients (31.1% versus 13.9%; P 0.001) compared with those with no kidney dysfunction. There were no significant differences in age (P 0.12), sex (P 0.42), or APACHE II scores (P 0.25) between those with mild and moderate kidney dysfunction. For patients with severe ARF, median serum creatinine and urea levels to immediately before initiation of renal replacement therapy were 4.6 mg/dl (IQR, 3.0 to 5.8 mg/dl [405 mol/l; IQR, 265 to 515 mol/l) and 67 mg/dl (IQR, 45 to 218 mg/dl [24 mmol/l; IQR, 16 to 33 mmol/l), respectively. Renal replacement therapy was started at a median of 1 day (IQR, 0 to 3 days) and 4 days (IQR, 1 to 11 days) after ICU and hospital admission, respectively. Table 2. Crude Case-Fatality Rates Stratified Across Severity of Kidney Dysfunction During Admission to ICU Case Fatality* No Renal Failure (n 4,411) Serum Creatinine 1.7 to 3.4 mg/dl (n 790) Serum Creatinine 3.4 mg/dl (n 160) Severe ARF (n 240) End-Stage Renal Disease/ Long-Term Renal Replacement Therapy (n 92) ICU death 361 (8.2) 224 (28.4) 42 (26.3) 120 (50) 15 (16.3) Hospital death 592 (13.4) 316 (40) 65 (41) 143 (60) 31 (34) 1-Year death 763 (17.3) 370 (47) 77 (48) 153 (64) 37 (40) NOTE. Values expressed as number (percent). To convert creatinine in mg/dl to mol/l, multiply by *Overall tests by Fisher exact test, P for each of ICU, hospital, and 1-year case fatality.

4 MORTALITY BY SEVERITY OF KIDNEY DYSFUNCTION 405 Kaplan-Meier Estimates by Kidney Dysfunction Fig 1. Crude Kaplan- Meier survival curves for critically ill patients stratified by severity of kidney dysfunction. *Global assessment for difference in survival function. To convert creatinine in mg/dl to mol/l, multiply by Time (Days) No renal failure Creatinine63.4 mg/dl End-Stage Kidney Disease Creatinine mg/dl Severe ARF Crude 1-year case-fatality rates stratified by severity of kidney dysfunction were 17% for no renal failure, 47% for mild dysfunction, 48% for moderate dysfunction, 64% for severe ARF requiring renal replacement therapy, and 40% those with preexisting end-stage kidney disease, as listed in Table 2. There was evidence of significantly greater case-fatality rates for any kidney dysfunction compared with no dysfunction when assessed at ICU discharge (32.4% versus 8.2%; P 0.001), hospital discharge (44.0% versus 13.4%; P 0.001), and 1-year outcome (50.4% versus 17.3%; P 0.001). Of note, comparing patients with mild dysfunction with those with moderate dysfunction, there was no significant difference in ICU (P 0.63), hospital (P 0.93), or 1-year (P 0.80) case-fatality rates (Table 2). However, those with severe ARF had significantly greater ICU (50.0% versus 8.2%; P 0.001), hospital (60% versus 13.4%; P 0.001), and 1-year (64% versus 17.3%; P 0.001) casefatality rates compared with those with no dysfunction or any other strata of kidney dysfunction. Long-term crude survival curves stratified by severity of kidney dysfunction are shown in Fig 1. Table 3 lists the population at risk at baseline, 28 days, 90 days, and 6 months. There was a significant overall difference in survival function observed between levels of renal dysfunction, and each was different, with the exception of patients with mild dysfunction compared with those with moderate dysfunction (log rank P 0.55). A multivariable logistic regression model for outcome at 1 year is listed in Table 4. Model calibration was excellent, with an area under the receiver operating characteristic curve of 0.85 and no evidence of significant lack of fit by means of the goodness-fit-test result of P of Table 3. Number of Critically Ill Patients at Risk at Each Time Stratified by Severity of Kidney Dysfunction Follow-Up Baseline 28 Days 90 Days 6 Months No renal failure 4,411 3,874 3,752 3,695 Creatinine, 1.7 to 3.4 mg/dl Creatinine 3.4 mg/dl Severe ARF End-stage kidney disease NOTE. To convert creatinine in mg/dl to mol/l, multiply by 88.4.

5 406 BAGSHAW ET AL Table 4. Logistic Regression Model for 1-Year Mortality by Severity of Kidney Dysfunction Variable Odds Ratio 95% Confidence Interval P Severity of renal dysfunction* Serum creatinine, 1.7 to 3.4 mg/dl Serum creatinine 3.4 mg/dl Severe ARF End-stage kidney disease Age (/y) APACHE II score (/point) Admission type Noncardiac surgical Cardiac surgical NOTE. Area under receiver operating characteristic curve, 0.85; goodness of fit, P To convert creatinine in mg/dl to mol/l, multiply by *Baseline indicator variable: no renal failure. Baseline indicator variable: medical admission type. In a sensitivity analysis, the mild (creatinine, 1.7 to 3.4 mg/dl [150 to 299 mol/l]) and moderate (creatinine 3.4 mg/dl [ 300 mol/l]) dysfunction categories were collapsed and the analysis was repeated. This resulted in an odds ratio for death at 1 year for mild or moderate dysfunction of 1.30 (95% confidence interval, 1.1 to 1.6; P 0.006). There was no significant change in the odds ratios for the other model covariates or in model calibration or fit. DISCUSSION To our knowledge, this study is the first population-based assessment of long-term outcome in critically ill patients stratified by severity of kidney dysfunction. In general, these results confirm that any kidney dysfunction in critically ill patients is associated with decreased long-term survival. This study further highlights how the use of absolute measures of serum creatinine alone may be insufficient to discriminate ICU, hospital, or long-term outcome. This study further clarifies the additional impact of severe ARF requiring renal replacement therapy in critically ill patients on ICU, hospital, and long-term outcomes. The institution of renal replacement therapy not only results in an escalation in management and expenditure of resources, but also is associated with a significantly greater risk for death. 3-5 Numerous prior epidemiological studies and clinical trials used as case definition for ARF absolute or relative changes in serum creatinine levels to assess outcome 7-11,22,24-29 ; however, this study shows the potential limitations of measures of serum creatinine alone. Likewise, using our data, when severity of kidney dysfunction was stratified according to the definition used by Liano and Pascual, 2 there was no significant difference in survival for those with a serum creatinine level of 1.7 to 3.4 mg/dl (150 to 299 mol/l) compared with those with a serum creatinine level of 3.4 mg/dl or greater ( 300 mol/l), respectively (log-rank P 0.28). Whether failure to find a difference in long-term mortality with the definitions used in this study is a limitation of the definition of ARF alone or instead reflects no true difference in long-term mortality remains uncertain. Thus, these results would argue for the need for and support the incorporation of a widely accepted and validated definition of kidney dysfunction in critically ill patients. Recent consensus recommendations for defining and categorizing ARF, the Risk, Injury, Failure, Loss, End-stage kidney disease (RIFLE) classification, were published; however, this classification has not yet been validated prospectively with such long-term clinical outcomes as mortality or renal recovery. 6 However, the proposed RIFLE classification incorporates several important factors that aim to reduce redundancy and confusion in the field of critical care nephrology. Importantly, the RIFLE classification incorporates several measures of kidney function, considers severity of dysfunction, and takes into account baseline function, absolute change from baseline, and the presence of preexisting chronic kidney disease, if present.

6 MORTALITY BY SEVERITY OF KIDNEY DYSFUNCTION 407 Another important finding in this study is the minimum duration of observation required before evidence of a stable survival rate in critically ill patients with kidney dysfunction. Few studies assessed long-term mortality in critically ill patients with kidney dysfunction. 3,5,30,31 Results presented here would suggest that a minimum of 90 days be used for ascertainment of such important clinical outcomes as mortality in future clinical studies because less than 2% of mortality events occurred after this period in our study. Furthermore, follow-up to 90 days would permit the opportunity for determination of additional important information on morbidity outcomes after critical illness, such as renal recovery and renal replacement therapy independence, hospital discharge location, and quality of life. 3,31,32 This study also shows lower long-term mortality in surgical patients, in particular, those undergoing cardiac surgery, compared with those admitted for medical indications (case fatality, 5% versus 39%; P 0.001). This casefatality rate after cardiac surgery has been described similarly and likely is attributable to the routine and elective postoperative admission of such patients. 33 Likewise, the apparent magnitude of decrease in long-term mortality by means of multivariate analysis in this study is accounted for in part by the low overall event rate in these patients. Finally, our study finds that critically ill patients with end-stage kidney disease had a mortality rate at 1 year similar to those with no kidney dysfunction after adjustment for age, severity of illness, and admission type. Conversely, 3 prior small cohort studies suggested that critically ill patients with end-stage kidney disease have an increased risk for death compared with those with no kidney dysfunction. 10,34,35 In these studies, estimated ICU and hospital mortality rates were 11% to 28% and 34% to 38%, respectively. Uchino et al 34 reported similar hospital mortality in 37 patients with end-stage kidney disease matched by severity of illness to those with severe ARF requiring renal replacement therapy. In a single multidisciplinary ICU, Manhes et al 35 reported a 6-month mortality rate of 48% in 92 patients with end-stage kidney disease admitted predominantly for septic indications. Two of these studies proposed that the severity of illness and organ dysfunction scoring systems APACHE II and Simplified Acute Physiology Score II reliably predict hospital mortality; however, in our study, APACHE II scores of patients with endstage kidney disease were considerably higher despite similar hospital mortality. 23,36 Therefore, APACHE II and Simplified Acute Physiology Score II scores are likely to underestimate hospital mortality, as recently suggested in the Beginning and Ending Supportive Therapy for the Kidney Study. 1 The apparent better outcome in our study for patients with end-stage kidney disease may be accounted for by a larger sample size and adjustment for severity of illness and admission type. Also, the end-stage kidney disease population in this study had a high burden of illness, and a greater proportion underwent cardiac surgery. However, given these findings, further investigations into the outcome of critically ill patients with end-stage kidney disease are warranted. Our study has notable limitations that warrant discussion. First, although this is a multicenter population-based study, it represents a single health region. Therefore, although we believe results of this study are an unbiased and valid assessment, they may differ from other health regions. However, demographic characteristics and outcomes in this study are similar to those reported in previous studies. 1,22 Second, we used peak serum creatinine level or need for short- or long-term renal replacement therapy during admission to the ICU to define and stratify kidney dysfunction. We were not able to incorporate additional measures of kidney dysfunction, such as the presence of azotemia (defined by an increase in blood urea nitrogen level) or oliguria, such discharge laboratory values as hemoglobin or albumin level, and the presence of preexisting chronic kidney disease or other comorbidities for all patients. Likewise, we were unable to incorporate the recently proposed RIFLE criteria because they were published after completion of our study. 6 In addition, use of an absolute serum creatinine level of 1.7 mg/dl or greater ( 150 mol/l) may have resulted in omission of patients with genuine kidney dysfunction not requiring renal replacement therapy. However, peak serum creatinine level alone was used by design principally to address how this measure would perform for discriminating long-term outcome, as suggested

7 408 by previous studies. 1,2,7-9 Because of evidence of nonproportional hazards, we were unable to conduct a valid Cox proportional hazards analysis and adjust for covariates by using time-to-event data without violation of model assumptions. Crude survival curves stratified by severity of kidney dysfunction are presented and thus require cautious interpretation. One plausible explanation for nonproportional hazards may be evident upon assessment of crude survival curves, for which early in the course of ICU admission, patients with mild or moderate dysfunction had lower survival compared with those with severe ARF receiving acute renal replacement therapy, as shown in Fig 1. This early difference may relate to such factors as physician decision to offer and initiate renal replacement therapy in selected patients; however, this remains speculative. Finally, this study is not a randomized controlled trial, but rather a retrospective assessment of prospectively collected data. Also, we were unable to collect and incorporate data for other potentially important variables to assess in multivariate analysis, such as the proportion of patients with sepsis syndrome and/or septic shock and chronic kidney disease other than those with end-stage kidney disease requiring long-term renal replacement therapy. However, our study is strengthened by use of a population-based cohort design, incorporation of a relatively large sample size, capture of all ICU admissions during the study period, and complete ascertainment of outcome at 1 year. 21 In summary, we conducted a large populationbased cohort study of critically ill patients stratified by severity of kidney dysfunction with longterm follow-up. In this study, we show in a cohort of nonspecific critically ill patients that any kidney dysfunction is associated with increased risk for long-term death, with the exception of patients with end-stage kidney disease, who had outcomes similar to those with no kidney dysfunction. Additionally, this study confirms that absolute measures of serum creatinine are poor for classifying long-term outcome and highlights the limitations of inference from serum creatinine level alone. These results draw attention to the importance of adopting accepted and validated measures of kidney function in critically ill patients, such as the RIFLE criteria BAGSHAW ET AL proposed by the Acute Dialysis Quality Initiative Workgroup. 6 Finally, this study strengthens the observation that determination of morbidity or mortality outcomes in critically ill patients with any kidney dysfunction ideally should occur after a minimum follow-up of 90 days to ensure capture of all clinically important events. REFERENCES 1. Uchino S, Kellum JA, Bellomo R, et al: Acute renal failure in critically ill patients: A multinational, multicenter study. JAMA 294: , Liano F, Pascual J: Epidemiology of acute renal failure: A prospective, multicenter, community-based study. Madrid Acute Renal Failure Study Group. Kidney Int 50: , Korkeila M, Ruokonen E, Takala J: Costs of care, long-term prognosis and quality of life in patients requiring renal replacement therapy during intensive care. Intensive Care Med 26: , Manns B, Doig CJ, Lee H, et al: Cost of acute renal failure requiring dialysis in the intensive care unit: Clinical and resource implications of renal recovery. Crit Care Med 31: , Bagshaw SM, Laupland KB, Doig CJ, et al: Prognosis for long-term survival and renal recovery in critically ill patients with severe acute renal failure: A population-based study. Crit Care 9:R700-R709, Bellomo R, Ronco C, Kellum JA, Mehta RL, Palevsky P: Acute renal failure Definition, outcome measures, animal models, fluid therapy and information technology needs: The Second International Consensus Conference of the Acute Dialysis Quality Initiative (ADQI) Group. Crit Care 8:R204-R212, de Mendonça A, Vincent JL, Suter PM, et al: Acute renal failure in the ICU: Risk factors and outcome evaluated by the SOFA score. Intensive Care Med 26: , Brivet F, Kleinknecht D, Loirat P, Landais P: Acute renal failure in intensive care units Causes, outcome, and prognostic factors of hospital mortality: A prospective, multicenter study. Crit Care Med 24: , Guerin C, Girard R, Selli JM, Perdrix JP, Ayzac L: Initial versus delayed acute renal failure in the intensive care unit. A multicenter prospective epidemiological study. Rhone- Alpes Area Study Group on Acute Renal Failure. Am J Respir Crit Care Med 161: , Clermont G, Acker CG, Angus DC, et al: Renal failure in the ICU: Comparison of the impact of acute renal failure and end-stage renal disease on ICU outcomes. Kidney Int 62: , Carbonell N, Blasco M, Ferreres J, et al: Sepsis and SOFA score: Related outcome for critically ill renal patients. Clin Nephrol 62: , Bellomo R, Chapman M, Finfer S, Hickling K, Myburgh J: Low-dose dopamine in patients with early renal dysfunction: A placebo-controlled randomised trial. Australian and New Zealand Intensive Care Society (ANZICS) Clinical Trials Group. Lancet 356: , 2000

8 MORTALITY BY SEVERITY OF KIDNEY DYSFUNCTION Yegenaga I, Hoste E, Van Biesen W, et al: Clinical characteristics of patients developing ARF due to sepsis/ systemic inflammatory response syndrome: Results of a prospective study. Am J Kidney Dis 43: , Van Biesen W, Yegenaga I, Vanholder R, et al: Relationship between fluid status and its management on acute renal failure (ARF) in intensive care unit (ICU) patients with sepsis: A prospective analysis. J Nephrol 18:54-60, Mazul-Sunko B, Zarkovic N, Vrkic N, et al: Proatrial natriuretic peptide (1-98), but not cystatin C, is predictive for occurrence of acute renal insufficiency in critically ill septic patients. Nephron Clin Pract 97:c103-c107, The National Cooperative Dialysis Study. Kidney Int Suppl 23:S1-S123, Paganini E, Depner T, Wensley D: The Acute Dialysis Quality Initiative Part III: Solute control (treatment dose). Adv Ren Replace Ther 9: , Morelli A, Ricci Z, Bellomo R, et al: Prophylactic fenoldopam for renal protection in sepsis: A randomized, double-blind, placebo-controlled pilot trial. Crit Care Med 33: , Couchoud C, Pozet N, Labeeuw M, Pouteil-Noble C: Screening early renal failure: Cut-off values for serum creatinine as an indicator of renal impairment. Kidney Int 55: , Calgary Health Region: Population estimates of the Calgary Health Region. Available at: Accessed May 22, Laupland K: Population-based epidemiology of intensive care: Critical importance of ascertainment of residency status. Crit Care 8:R431-R436, 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 Suppl 66:S16-S24, Knaus WA, Draper EA, Wagner DP, Zimmerman JE: APACHE II: A severity of disease classification system. Crit Care Med 13: , Groeneveld A, Tran D, van der Meulen J, Nauta J, Thijs L: Acute renal failure in the medical intensive care unit: Predisposing, complicating factors and outcome. Nephron 59: , Hoste EA, Lameire NH, Vanholder RC, et al: Acute renal failure in patients with sepsis in a surgical ICU: Predictive factors, incidence, comorbidity, and outcome. J Am Soc Nephrol 14: , Mehta RL, Pascual MT, Soroko S, et al: Spectrum of acute renal failure in the intensive care unit: The PICARD experience. Kidney Int 66: , Neveu H, Kleinknecht D, Brivet F, Loirat P, Landais P: Prognostic factors in acute renal failure due to sepsis. Results of a prospective multicentre study. Nephrol Dial Transplant 11: , Rasmussen H, Pitt E, Ibels L, McNeill D: Prediction of outcome in acute renal failure by discriminant analysis of clinical variables. Arch Intern Med 145: , Sural S, Sharma RK, Singhal MK, et al: Acute renal failure in an intensive care unit in India Prognostic factors and outcome. J Nephrol 12: , Spurney RF, Fulkerson WJ, Schwab SJ: Acute renal failure in critically ill patients: Prognosis for recovery of kidney function after prolonged dialysis support. Crit Care Med 19:8-11, Morgera S, Kraft AK, Siebert G, Luft FC, Neumayer HH: Long-term outcomes in acute renal failure patients treated with continuous renal replacement therapies. Am J Kidney Dis 40: , Sukantarat KT, Burgess PW, Williamson RC, Brett SJ: Prolonged cognitive dysfunction in survivors of critical illness. Anaesthesia 60: , Chertow GM, Levy EM, Hammermeister KE, Grover F, Daley J: Independent association between acute renal failure and mortality following cardiac surgery. Am J Med 104: , Uchino S, Morimatsu H, Bellomo R, Silvester W, Cole L: End-stage renal failure patients requiring renal replacement therapy in the intensive care unit: Incidence, clinical features, and outcome. Blood Purif 21: , Manhes G, Heng AE, Aublet-Cuvelier B, et al: Clinical features and outcome of chronic dialysis patients admitted to an intensive care unit. Nephrol Dial Transplant 20: , Le Gall J, Lemeshow S, Saulnier F: A new Simplified Acute Physiology Score (SAPS II) based on a European/ North American multicenter study. JAMA 270: , 1993