Long-term functional evolution after an acute kidney injury: a 10-year study

Transcription

1 Nephrol Dial Transplant (2008) 23: doi: /ndt/gfn398 Advance Access publication 15 July 2008 Original Article Long-term functional evolution after an acute kidney injury: a 10-year study Belén Ponte 1, Carmen Felipe 1,2, Alfonso Muriel 3, Maria Teresa Tenorio 1,4 and Fernando Liaño 1,4,5 1 Nephrology Department, Ramón y Cajal University Hospital Madrid, 2 Nuestra Señora de Sonsoles Hospital Avila, 3 Clinical Biostatistics Unit, Ramón y Cajal University Hospital Madrid, 4 Acute Renal Failure Madrid Consorcium and 5 Medicine Department of University of Alcala Alcala de Henares, Spain Abstract Background. Data on long-term effects of acute kidney injury (AKI) on renal function (RF) are scarce and factors implicated in the functional outcome are not established. Our aim was to investigate these aspects. Methods. At hospital discharge and annually for 10 years, we retrospectively reviewed RF of 187 patients surviving AKI. Glomerular filtration rates estimated with MDRD equation (egfr) and KDOQI stages were used to evaluate RF. Only 34.8% of patients had pre-existing renal dysfunction (KDOQI-3). Variables determining long-term RF were collected during AKI and at discharge and analysed with a regression model. Results. At discharge no patient necessitated dialysis, but egfr was lower than baseline (47.5 ± 23.3 ml/min/ 1.73 m 2 versus 75.8 ± 25.4 ml/min/1.73 m 2 ); 38.4% of survivors had recovered basal RF: 26% of those with previous normal RF and 61% of those in KDOQI-3, respectively. At 1 year, egfr increased to 61.9 ± 24.4 ml/min/1.73 m 2 and remained stable later. During an 8-year median follow-up (P25:2; P75:10), 31% improved RF, 50% remained stable and 19% deteriorated. In total only 46% (n = 82) definitively recovered RF. Finally, at the end of the study period 61.6% presented some degree of renal dysfunction: 40% of those with previous normal RF developed moderate severe renal dysfunction and 37% KDOQI-3 progressed into more severe renal failure. Only two patients needed dialysis. Regression model identified age, co-morbidities, discharge egfr and follow-up time as independent predictors of long-term RF. Conclusions. AKI carries implication for long-term RF even in patients without pre-existing renal dysfunction. Ageing, co-morbidities and RF at discharge are determinants of the long-term functional outcome. Keywords: acute kidney injury; chronic kidney disease; long-term outcome Introduction Acute tubular necrosis (ATN) is the most frequent cause of acute kidney injury (AKI) in hospitalized patients. In hospital and intensive care units (ICU) ATN accounts respectively for 38% and 76% cases of AKI [1,2]. The aetiology can be multifactorial. Ischaemia, nephrotoxicity (drugs or contrast) and sepsis are the leading precipitating factors [3]. There is some evidence that AKI per se affects patient survival, length of stay and hospital costs [4,5]. Despite recent progress in the management of AKI patients, mortality remains high ranging between 21% and 70% [6 8]. Introduction of continuous haemodiafiltration techniques has not increased the survival rate and there is no clear evidence that continuous replacement therapy is better than intermittent [9 11]. Recent cohort studies tend to show that despite all new therapeutic approaches the incidence of AKI is increasing whereas mortality is decreasing [12 16]. Recently, survival after 10 years was reported to be only 50% [17]. Data from other studies suggested that survival could vary from 15% to 72% depending on the setting and the period studied [6,18,19]. When evaluating the outcome of AKI not only survival but also recovery of renal function (RF) is a central issue. Unfortunately, data on long-term functional outcome are scarce and controversial [7,20 32]. Some of the few studies reported a good RF after AKI [22,28,29,31,32] while others described a high rate of dialysis-dependent patients [7,22,23,26,27,30]. Thus, it should not be assumed that renal recovery is always complete. Actually, the long-term effects of AKI or ATN on RF cannot be well defined because of the paucity of long-term follow-up studies. Moreover, the factors implicated in the functional long-term prognosis are not yet established. The purpose of our study was to analyse the long-term RF in all patients surviving an ATN episode and determine useful factors for estimating the functional outcome. Patients and methods Correspondence and offprint requests to: Fernando Liaño, Nephrology Department, Ramón y Cajal University Hospital, Madrid, Spain. Tel: ; Fax: ; lianof@yahoo.es All the data of this series referring to the patients, definitions, clinical management of the AKI episode and other C The Author [2008]. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved. For Permissions, please journals.permissions@oxfordjournals.org

2 3860 B. Ponte et al. Diet in Renal Disease (MDRD) study [34]. In order to standardize different levels of renal function, the five stages of the KDOQI classification for CKD [35] were used at baseline, discharge and during the follow-up. Even though the use of the MDRD equation for determining the RF in the early follow-up of the AKI patients could be controversial, we decided to compare the basal, discharge and last GFR values of each patient to analyse the degree of recovery of RF at discharge. Type of admission was classified into three categories: (1) medical (patients hospitalized for medical diseases not, at least initially, subsidiaries of surgery); (2) surgical and (3) trauma. Fig. 1. Schema of the population studied. aspects have been published elsewhere [17]. Hereafter, we summarize the main information related to this study. Selection criteria, population and study periods We performed an extended retrospective analysis of the data of all patients with a diagnosis of ATN in our hospital during 15 years. Patients with any kind of organ transplantation and those <15 years of age were previously excluded. There were 531 patients recorded. We excluded 118 patients for the reasons given in Figure 1. A baseline creatinine 1.4 mg/dl was used to define chronic kidney disease (CKD). From the remaining 413 cases: 226 died during their admission and 187 were discharged alive from the hospital. In this study, we focused on those 187 patients and analysed their RF as a cohort during 10 years. Definitions Acute kidney injury was considered when a sudden rise in serum creatinine concentration (SCr) to >2 mg/dl (177 µmol/l) was found in subjects with prior SCr <1.4 mg/dl. Functional severity of the AKI episode, in the patients fulfilling the previous criteria, was graded as R, I and F according to the RIFLE system [33]. Acute tubular necrosis was diagnosed when RF did not improve after the correction of possible pre-renal causes and when hepatorenal syndrome, vascular, interstitial, glomerular and obstructive aetiologies were excluded. Hereafter, AKI will be considered as synonymous with ATN. Although ATN aetiology is frequently multifactorial, we classified our patients into one of the four following aetiologic groups: (1) surgical, (2) nephrotoxic, (3) septic and (4) medical ATN in the case of patients, not included in the previous three groups, who had precipitating factors, isolated or combined, such as volume depletion, low cardiac output or arterial hypotension. Oliguria was defined as a urine output of <400 ml/24 h. RF at hospital discharge and during the follow-up was analysed using the glomerular filtration rate (GFR) estimated by a four-variable equation derived from the Modification of Variables recorded during the AKI episode (1) Sex and age; (2) co-morbid conditions (cardiac diseases, hepatic diseases, pulmonary obstructive diseases, tumours, diabetes mellitus, hypertension, acute cerebrovascular accident, peripheral arteriopathy, aortic aneurysm and AIDS); (3) type of admission; (4) ATN aetiologic groups; (5) clinical conditions required for calculating the individual severity index (ISI) [36] at the moment when the nephrologist saw the patient for the first time; (6) urine output: oliguric or non-oliguric; (7) need of dialysis; (8) SCr levels: at baseline, the highest value observed during admission and at discharge; (9) need of intensive care treatment and (10) the severity of AKI using RIFLE score. Patients follow-up: procedure for obtaining RF information The search to obtain the long-term outcome data has been previously described [17]. In summary, of the 187 patients discharged alive from our hospital, we have complete information of 177 cases including that of the 95 patients who died during the follow-up. This information was obtained from the patients medical records or from the data provided by their family doctors. We could not analyse the long-term outcome of only 10 patients lost to follow-up. SCr values were analysed at discharge and at 1 year; thereafter, they were recorded annually for 10 years. Due to the retrospective characteristics of the study, with the exception of SCr at discharge, the other values did not always correspond exactly to the above referred times. The yearly SCr values could have a variation in time of ± 3 months. When SCr figures were collected during an intercurrent admission, we recorded the lowest SCr of that admission. When SCr data were unavailable at a determined period, the data lacking was estimated in one of the following ways: (1) if the previous and posterior figures were stable, the missing figure was considered to be similar, and (2) if there was an increasing or decreasing figure, the mean value was estimated from the previous and posterior data. Statistical analysis Continuous variables were expressed as mean and standard deviation when distribution was normal. Categorical variables were expressed as absolute values and proportions. Comparisons were made using either the Mann Whitney,

3 One hundred and eighty-seven patients survived an ATN episode. Table 1 compares their general characteristics with those who died during admission. We observe that the type of admission, the aetiology and severity of AKI and fi- Long-term function after AKI 3861 Table 1. Patients characteristics initial population nally the clinical severity (ISI index and ICU admission) is significantly different between these two populations. In Deaths during the group of survivors : 51% (n = 95) had a medical admission, 52% (n = 97) had nephrotoxic ATN, only 39% admission, Discharged alive, Variables N = 226 (%) N = 187 (%) (n = 72) presented with oliguria and 31% (n = 57) needed Age during AKI 58.7 ± ± 16.2 dialysis. Their ISI score was lower, and 42% were admitted Gender to ICU. Female 76 (33.6) 63 (33.7) The baseline RF of those 187 discharged alive was: Male 150 (66.4) 124 (66.3) Scr 1.06 ± 0.23 mg/dl; GFR 75.8 ± 26.1ml/min/1.73 Admission type Medical 85 (37.6) 95 (50.8) m 2. Sixty-five patients had previous moderate renal dysfunction (KDOQI stage 3) while the others were con- Surgery 115 (50.9) 81 (43.3) Trauma 26 (11.5) 11 (5.9) sidered as having normal RF (KDOQI stages 1 and 2). AKI aetiology RF at discharge was: Scr 1.76 ± 0.77 mg/dl and GFR Nephrotoxicity 47 (20.8) 97 (51.9) Sepsis 57 (25.2) 26 (13.9) 47.5 ± 23.3 ml/min/1.73 m 2. At this moment, 53% patients Medical 66 (29.2) 32 (17.1) (n = 137) had renal impairment according to KDOQI classification. None of them needed chronic dialysis at discharge. Surgical 56 (24.8) 32 (17.1) Need of dialysis The rest had normal RF (Figure 2 a and b). Yes 142 (62.8) 57 (30.5) From this population 10 patients were lost during the No 84 (37.2) 130 (69.5) Oligo-anuria early follow-up. We analysed the long-term follow-up of Yes 161 (71.3) 72 (38.5) the remaining 177 patients. Their median follow-up was No 65 (28.7) 115 (61.5) 8 years (P25:2 years, P75:10 years). We divided the patients ICU into two groups according to their baseline renal function. Yes 166 (73.5) 80 (42.4) No 60 (26.5) 107 (57.6) A summary of their principal characteristics is shown in ISI index 0.61 ± ± 0.23 Table 2. Co-morbidities Comparing the two groups described in Table 2, we found Yes 180 (79.6) 152 (81.3) that: patients with previous renal dysfunction were older No 41 (18.1) 35 (18.7) (P < 0.001), had more co-morbidities (P = 0.001), but had Unknown 5 (2.2) a better rate of renal recovery at discharge (P < 0.001). In AKI: acute kidney injury; ICU: intensive care unit; ISI: individual severity contrast, in the group with previous normal RF there were index. more males (P = 0.004) and patients were treated more frequently in the ICU setting (P = 0.046). There were also more patients with severe AKI as defined by RIFLE criteria Student s t-test or chi-square tests when appropriate. Analyses were first performed using SPSS version 12.0 for descriptions and functional evolution. In order to identify the effect of various factors implicated in the functional outcome over the entire period studied, we had to analyse repeated measurements of estimated GFR using generalized estimating equation (GEE) models with STATA 9.0 [37,38]. Cox regression models were inappropriate for the purpose of our study and could not be used. All parameters previously described were first included as explanatory variables in the univariate analysis of change in GRF (calculated by the MDRD equation) in the whole population. Only significant variables (P < 0.05) were then included in the multivariate analysis. Multinomial variables were coded as dummy when necessary. A backward strategy was used to find the final model that allowed us to estimate the final GFR for each patient at a determined time. Analysis of subgroups divided according to their baseline KDOQI (P = 0.04) although statistics were not significant when categories R and I were considered together (P = 0.13). The need of dialysis, the length of hospital admission, the type of admission, the aetiology of AKI, the incidence of oliguria and the ISI score were similar in both groups. Only 38.4% patients (n = 68) recovered their baseline RF at discharge and 20.3% (n = 36) during the follow-up period. Half of the patients kept the same KDOQI stage as at discharge. Details of the changes in RF are given in Figures 2 and Figure 3. Finally, at the end of the study period 61.1% patients (n = 108) had some degree of renal dysfunction (Figure 2c), but only two patients were started on chronic dialysis in this period. Figure 4 shows the GFR evolution of the whole population studied during the follow-up period. The mean GFR which was 47.3 ± 22.5 ml/min/1.73 m 2 at discharge increased significantly during the first year (GFR 61.9 ± 24.4 ml/min/1.73 m 2 ; P < 0.001) and generally remained stage was not performed due to the small sample size in each subgroup. The adequacy of the model was tested using the analysis of the residuals. stable thereafter (GFR 59.0 ± 23.2 ml/min/1.73 m 2 at Results 10 years). Figure 5a and b shows the GFR evolution in patients with or without previous renal impairment. In both groups, there was a significant initial improvement at 1 year, but RF then remained stable with a slight tendency to decrease during the follow-up. Despite a relatively good functional evolution, more patients died during the follow-up in the group with previous renal dysfunction (71.0% versus 44.3%, P = 0.001). To understand the functional outcome, we looked for conditioning factors that could explain the variations in the

4 3862 B. Ponte et al. Table 2. Characteristics of studied patients, discharged alive from hospital according to their baseline renal function Baseline renal function Normal (KDOQI Moderate dysfunction Variables 1 2), N = 115 (%) (KDOQI 3), N = 62 (%) Total, N = 177 (%) Age during AKI 52.2 ± ± ± 16 Gender Female 31 (27) 30 (48.4) 61 (34.5) Male 84 (73) 32 (51.6) 116 (65.5) Baseline GFR (ml/min/1.73 m 2 ) 88.5 ± ± ± 25.4 Hospital stay (days) median (P25; P75) 42 (26; 70) 51 (37; 75) 45 (30; 71) Admission type Medical 63 (54.8) 29 (46.8) 92 (52) Surgery 45 (39.1) 33 (53.2) 78 (44) Trauma 7 (6.1) 7 (4) AKI aetiology Nephrotoxicity 60 (52.2) 33 (53.3) 93 (52.5) Sepsis 15 (13) 9 (14.5) 24 (13.6) Medical 20 (17.4) 11 (17.7) 31 (17.5) Surgical 20 (17.4) 9 (14.5) 29 (16.4) Need of dialysis Yes 41 (35.7) 14 (22.6) 55 (31.1) No 74 (64.3) 48 (77.4) 122 (68.9) Functional ATN severity Risk 2 (1.7) 6 (9.7) 8 (4.5) Injury 28 (24.3) 17 (27.4) 45 (25.4) Failure 85 (73.9) 39 (62.9) 124 (70.1) Oligo-anuria Yes 49 (42.6) 19 (30.6) 68 (38.4) No 66 (57.4) 43 (69.4) 109 (61.6) ICU Yes 55 (47.8) 20 (32.3) 75 (42.4) No 60 (52.2) 42 (67.7) 102 (57.6) ISI index 0.29 ± ± ± 0.23 Discharge creatinine (mg/dl) median (P25; P75) 1.4 (1.2;1.9) 1.8 (1.5; 2.3) 1.6 (1.2;2) Discharge GFR (ml/min/1.73 m 2 ) 54.6 ± ± ± 22.5 Recovery basai KDOQI stage at discharge Yes 30 (26.1) 38 (61.3) 68 (38.4) No 85 (73.9) 24 (38.7) 109 (61.6) Renal function at discharge Stage KDOQI 1 10 (8.7) 10 (5.6) Stage KDOQI 2 37 (32.2) 37 (20.9) Stage KDOQI 3 56 (48.7) 38 (61.3) 94 (53.1) Stage KDOQI 4 10 (8.7) 20 (32.3) 30 (16.9) Stage KDOQI 5 2 (1.7) 4 (6.5) 6 (3.4) Co-morbidities Yes 89 (77.4) 60 (96.8) 149 (84.2) No 26 (22.6) 2 (3.2) 28 (15.8) Death during the follow-up Yes 51 (44.3) 44 (71) 95 (53.7) No 64 (55.7) 18 (29) 82 (46.3) AKI: acute kidney injury; ICU: intensive care unit; ISI: individual severity index; GFR: glomerular filtration rate. P < 0.05: moderate renal dysfunction versus normal renal function at baseline. GFR over the 10 years studied in all patients. A total of 1331 serum creatinine values could be analysed and were used to estimate GFR. Of the variables described in material and methods: age during AKI, sex, existence of co-morbidities, the need of dialysis, presence of oliguria, type of admission, estimated GFR at discharge, recovery of baseline RF at discharge and time after discharge (years) were significant in the univariate analysis. These variables were included in the multivariate analysis. ATN aetiology, its length or severity, the need of ICU admission and ISI score were not found to be relevant. With a backward strategy, we found finally that age, existence of co-morbidities, GFR at discharge and follow-up time could predict GFR at any time during the follow-up (Table 3). Discussion Recovery of RF after AKI is an important clinical issue that should be evaluated as persistent chronic renal dysfunction can influence future morbidity and mortality. Recovery of RF is in most studies defined as non-dependence on renal replacement therapy, which is probably too simplistic [39].

5 Long-term function after AKI 3863 Fig. 2. Evolution of renal function according to KDOQI staging. Fig. 3. Functional evolution according to baseline KDOQI stages at discharge and during follow-up. Data from studies including all aetiologies of AKI are difficult to interpret as functional prognosis is known to be worse in some types of glomerulonephritis or vasculitis [1]. Depending on the AKI aetiology, 3 41% of patients could have end-stage renal disease and 3 24% could need chronic dialysis after discharge [24]. In our series, 46% patients (n = 82) definitively recovered their basal RF (KDOQI stage) during the follow-up. However, 40% (n = 46) of those with a previous normal renal function developed moderate-to-severe renal dysfunction. From those known to have previously moderate chronic renal failure, 37% (n = 23) progressed into severe or end-stage renal failure. Additionally, only 1.1% (two patients) became dependent on dialysis during the analysed period: one had baseline normal function (GRF 63.3 ml/min/1.73 m 2 ) and the other CKD (GRF 56.4 ml/min/1.73 m 2 ). Both of them had a GFR <45 ml/min/1.73 m 2 at discharge. This rate of long-term renal replacement therapy is in agreement with the percentage of 1 3% found in other studies where only patients having ATN as the cause of AKI were included [1,21,32]. When patients having other forms of AKI were included this range increased to % [7,22,25 28,31]. This suggests that other forms of AKI may have a worse functional prognosis than ATN. We would also expect patients having acute-on-chronic AKI to have a worse evolution than AKI patients having previous normal function [7,26,28,31], but we could not find studies comparing outcome as endpoint in both

6 3864 B. Ponte et al. Fig. 4. GFR evolution during follow-up in all patients. settings. Nevertheless, some studies reported a greater risk of irreversible renal failure in patients with pre-existing renal impairment [23,40 42]. In our study surprisingly the percentage of patients with deterioration in their RF was higher in patients with previous normal function (63.4% versus 37.0%), but patients with previous CKD developed more severe or end-stage renal failure (37% versus 5%). However, as post-discharge mortality is imbalanced and CKD patients died more frequently (71% versus 44.3%) than patients without CKD, a survivor bias could explain this unexpected finding. Our data suggest that the natural long-term functional outcome of ATN probably has two phases. During the first one, which extended to 1 year after discharge, the RF improved dramatically as seen in other series [20,21]. During the second phase, the RF remained stable or slightly decreased over 10 years. This pattern was similar in patients with previous normal RF or previous CKD. To our knowledge, this is the first time that these characteristics in the long-term evolution of ATN have been reported. This finding should emphasize the necessity of a 1-year followup (at least) after an AKI episode to evaluate the risk of the long-term renal impairment. In the multivariate analysis, we found that the variables predicting the GFR of the whole population were age at the AKI episode, existence of co-morbidities, GFR at discharge and time of follow-up after discharge. Unfortunately, the small sample population did not allow us to look for functional prognosis factors according to the basal KDOQI stage. It is no surprise to discover that not only the age, but also the length of time after an episode of ATN has an impact on the long-term outcome. It is well known that ageing is an important factor conditioning RF both in the general population and in AKI patients [25,43,44]. Some reports suggest that even children who present an AKI have a progressive decline in RF in adulthood [45]. The existence of co-morbidities, particularly cardiovascular diseases, is known to be a risk factor for developing AKI [12]. In another article, the presence of co-morbidities was not associated with in-hospital mortality but contributed to the long-term mortality [17]. We have now found that co-morbidity is also an important factor for the long-term renal function. This aspect has been previously reported by Bagshaw [39]. Unfortunately, we could not evaluate the effect of different co-morbid pathologies due to the small number of patients in each subgroup. Recently, Shiffl did not find an association between comorbidity and partial recovery of renal function, but the follow-up period of his study was only 1 year [32]. Using a regression model we were able to estimate GFR at any time during follow-up. In our model, GFR at discharge was one of the predictors of long-term GFR: the higher the GFR at discharge, the higher we expect it to be later. Interestingly, the severity of the AKI, its length, the presence of oliguria and the need of dialysis or ICU admission did not have any effect on the long-term RF in our model. Similarly, in a previous study those variables did not have any effect on the mortality [17]. The long-term evolution seems to be essentially conditioned by general factors from the individual human being rather than by the acute illness. Although this study is retrospective, unicentric and only included a relatively small number of patients, it contributes to the understanding of long-term effects of AKI in the general population. The extended length of the period studied is particularly interesting. Previous studies giving some data on functional evolution had a maximum follow-up of 5 years and only 5 of them had more than 100 patients included after hospital discharge [22,25,28,30,32]. Most of them analysed factors that could explain the mortality but not the potential factors, other than age, that could influence the long-term functional evolution [21,46]. The main characteristics of the patients discharged alive in our study were the following: (1) 100% had a clinical diagnosis of ATN as the cause of AKI; (2) both ICU (42.4%) and non-icu patients were included; (3) 30% required renal replacement therapy during admission and (4) only 34.8% had previous CKD (KDOQI stage 3) as initially severe forms of CKD were excluded to avoid bias in the functional outcome. This represents a good sample of the subset of patients with an AKI episode discharged alive from hospital. This population has clearly different characteristics from the subset of AKI patients dying in hospital (table 1), and our results may still be optimistic when considering critically ill patients surviving AKI in the ICU. Nowadays, according to some long-term studies [28,30,31] ICU patients seem to have a high rate of dialysis dependence and probably the rate of CKD, if adequately defined and analysed, would be higher. Hospitalization secondary to AKI has increased dramatically during the last few years [13 16]. There is growing evidence that AKI can have chronic effects in clinical situations as well as in animal models [47]. We should thus expect a burden of chronic renal diseases with an increase in costs secondary to AKI within a few years. The slight

7 Long-term function after AKI 3865 Fig. 5. (a) GFR evolution in patients with previous normal GFR (KDOQI 1 2). (b) GFR evolution in patients with previous moderate dysfunction (KDOQI 3). Table 3. Final multiple regression model to predict GFR during follow-up Variables Coefficient P-value 95% CI GFR at discharge 0.45 < ; 0.56 Age during AKI (years) 0.48 < ; 0.33 Co-morbidities (if any) < ; 5.84 Time after discharge (years) 0.96 < ; 0.83 Constant < ; decrease in mortality that we started to observe could also increase the number of cases needing chronic dialysis. Consequently, patients at higher risk of developing chronic renal disease after an ATN episode should be identified and closely monitored to prevent any complications or further deterioration of the renal function. The long-term effects of AKI are still unclear because of the paucity of long-term follow-up studies [46]; however, our study suggests that the view that all patients recover a normal RF after an ATN is probably too simplistic [48]. In fact, AKI should definitely no longer be viewed as an acute and short-term illness. Acknowledgement. We would like to thank Mary Harper for her help in the English version of this article. Conflict of interest statement. None declared. Preliminary results of this study have been presented at the ASN congress in San Francisco in References 1. Liaño F, Pascual J. Madrid Acute Renal Failure Study Group. Epidemiology of acute renal failure: a prospective, multicenter, community based study. Kidney Int 1996; 50: Lameire N, Van Biesen W, Vanholder R. Acute renal failure. Lancet 2005; 365: Esson ML, Schrier RW. Diagnosis and treatment of acute tubular necrosis. Ann Intern Med 2002; 137: Levy EM, Viscoli CM, Horwitz RI. The effect of acute renal failure on mortality. A cohort analysis. JAMA 1996; 275: Chertow GM, Burdick E, Honour M et al. Acute kidney injury, mortality, length of stay, and costs in hospitalized patients. J Am Soc Nephrol 2005; 16: Turney JH, Ellis CH, Parsons FM. Obstetric acute renal failure Br J Obstet Gynecol 1989; 96: Chertow GM, Christiansen CL, Clearly PD et al. Prognostic stratification in critically ill patients with acute renal failure requiring dialysis. Arch Intern Med 1995; 155: Liaño F, Junco E, Pascual J et al. The Madrid Acute Renal Failure Study Group. The spectrum of acute renal failure in the intensive care unit compared with that seen in other settings. Kidney Int Suppl 1998; 66: S16 S24 9. Cho KC, Himmerlfalb J, Paganini E et al. Survival by dialysis modality in critically ill patients with acute kidney injury. J Am Soc Nephrol 2006; 17: Mehta RL, McDonald B, Gabbai FB et al. for the Collaborative Group for Treatment of ARF in the ICU. A randomized trial of continuous versus intermittent dialysis for acute renal failure. Kidney Int 2001; 60: Vinsonneau C, Camus C, Combes A et al. for the Hemodiafe Study Group. Continuous venovenous haemodiafiltration versus intermittent haemodialysis for acute renal failure in patients with multiple-organ dysfunction syndrome: a multicentre randomised trial. Lancet 2006; 368: Liangos O, Wald R, O Bell JW et al. Epidemiology and outcome of acute renal failure in hospitalized patients: a national survey. Clin J Am Soc Nephrol 2006; 1: Waikar SS, Curhan GC, Wald R et al. Declining mortality with acute renal failure, 1988 to J Am Soc Nephrol 2006; 17: Xue JL, Daniels F, Star RA et al. Incidence and mortality of acute renal failure in medicare beneficiaries, 1992 to J Am Soc Nephrol 2006; 17: Tariq A, Khan I, Simpson W et al. Incidence and outcome in acute kidney injury: a comprehensive population-based study. J Am Soc Nephrol 2007; 18:

8 3866 B. Ponte et al. 16. Bagshaw SM, George C, Bellomo R for the ANZICS Database Management Committee. Changes in the incidence and outcome for early acute kidney injury in a cohort of Australian intensive care units. Crit Care 2007; 11: R Liaño F, Felipe C, Tenorio MT et al. Long term outcome of acute tubular necrosis: a contribution to its natural history. Kidney Int 2007; 71: Lameire N, Hoste E, Van Loo A et al. Pathophysiology, causes and prognosis of acute renal failure in the elderly. Ren Fail 1996; 18: Stevens PE, Tamimi NA, Al-Hasani MK et al. Non specialist management of acute renal failure. QJMed2001; 94: Hall JW, Johnson WJ, Maher FT et al. Immediate and long term prognosis of acute renal failure. Ann Intern Med 1970; 73: Kjellstrand CM, Ebben J, Davin T. Time of death, recovery of renal function, development of chronic renal failure and need for chronic hemodialysis in patients with acute tubular necrosis. Trans Am Soc Artif Intern Organs 1981; 27: Bonomini V, Stefoni S, Vangelista A. Long-term patient and renal prognosis in acute renal failure. Nephron 1984; 36: 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 1991; 19: Brivet FG, Kleinknecht DJ, Loirat P et al. The French Study Group on Acute Renal Failure: acute renal failure in intensive care units causes, outcome, and prognostic factors of hospital mortality: a prospective, multicenter study. Crit Care Med 1996; 24: Bandhari S, Turney JH. Survivors of acute renal failure who do not recover renal function. QJMed1996; 89: Mc Carthy T. Prognosis of patients with acute renal failure in the intensive care unit: a tale of two eras. Mayo Clin Proc 1996; 71: Korkeika 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 2000; 26: Morgera S, Kraft AK, Siebert G et al. Long term outcome in acute renal failure patients treated with continuous renal replacement therapies. Am J Kidney Dis 2002; 2: 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 2005; 9: R700 R Lins RL, Elseviers MM, Daelemans R. Severity scoring and mortality 1 year after acute renal failure. Nephrol Dial Transplant 2006; 21: Saudan P, Triverio PA, Pugin J et al. Long term prognosis in critically ill patients with acute renal failure treated by continuous renal replacement therapy. J AM Soc Nephrol 2006; 17(abstracts issue): TH-PO Schiffl H. Renal recovery from acute tubular necrosis requiring renal replacement therapy: a prospective study in critically ill patients. Nephrol Dial Transplant 2006; 21: Bellomo R, Ronco C, Kellum JA et al. for the ADQI workgroup. 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. Critical Care 2004; 8: R24 R Levey AS, Coresh J, Greene T, et al. for the Chronic Kidney Disease Epidemiology Collaboration. Using standardized serum creatinine values in the Modification of Diet in Renal Disease Study equation for estimating glomerular filtration rate. Ann Intern Med 2006; 145: National Kidney Foundation. K/DOQI clinical practice guidelines for chronic kidney disease: evolution, classification and stratification. Am JKidneyDis2002; 39(Suppl 1): S76 S Liaño F. Severity of acute renal failure: the need of measurement. Nephrol Dial Transplant 1994; 9(Suppl 4): Diggle PJ, Liang KY, Zeger SL. Analysis of Longitudinal Data.New York: Oxford University Press, StataCorp: xtgee. Estimate panel-data models using GEE. Sata Reference Manuel: Release 5.0, Vol 1. College Station, TX, Stata Press, 1997, Bagshaw SM. Epidemiology of renal recovery after acute renal failure. Curr Opin Crit Care 2006; 12: Noble JS, MacKirdy FN, Donaldson SI et al. Renal and respiratory failure in Scottish ICUs. Anaesthesia 2001; 56: 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 2002; 62: Uehlinger DE, Jakob SM, Ferrari P et al. Comparison of continuous and intermittent renal replacement therapy for acute renal failure. Nephrol Dial Transplant 2005; 20: Gentric A, Cledes J. Immediate and long-term prognosis in acute renal failure in the elderly. Nephrol Dial Transplant 1991; 6: Pascual J, Liano F, Ortuno J. The elderly with acute renal failure. J Am Soc Nephrol 1995; 6: Askenazi D, Feig D, Graham N. 3 5 year longitudinal follow-up of pediatric patients after acute renal failure. Kidney Int 2006; 69: Bagshaw SM. The long-term outcome after acute renal failure. Curr Opin Crit Care 2006; 12: Basile DP. The endothelial cell in ischemic acute kidney injury: implications for acute and chronic renal function. Kidney Int 2007; 72: Block CA, Schoolwerth AC. The epidemiology and outcome of acute renal failure and the impact on chronic kidney disease. Semin Dial 2006; 19: Received for publication: Accepted in revised form: