Improved Survival in Acute Kidney Injury After Cardiac Surgery

Improved Survival in Acute Kidney Injury After Cardiac Surgery

ORIGINAL INVESTIGATIONS Pathogenesis and Treatment of Kidney Disease Improved Survival in Acute Kidney Injury After Cardiac Surgery Charuhas V. Thakar...

279KB Sizes 17 Downloads 46 Views

ORIGINAL INVESTIGATIONS Pathogenesis and Treatment of Kidney Disease Improved Survival in Acute Kidney Injury After Cardiac Surgery Charuhas V. Thakar, MD, FASN,1,2 Sarah Worley, MS,3 Susana Arrigain, MA,3 Jean-Pierre Yared, MD,4 and Emil P. Paganini, MD1 Background: The overall incidence of acute kidney injury (AKI) or mortality after cardiac surgery is low, but mortality in patients with AKI remains high. Effects of factors such as change in comorbid disease burden, intraoperative factors, or postoperative complications on trends in the incidence of AKI and associated mortality after cardiac surgery were not examined. Study Design: Observational cohort study. Setting & Participants: 34,562 cardiac surgeries were performed from 1993 to 2002; only the first surgical procedure was considered (N ⫽ 33,217). Predictor, Outcomes, & Measurements: AKI was defined as a composite outcome of a 50% or greater decrease in postoperative glomerular filtration rate or requirement of dialysis (AKI-D). Mortality was defined as postoperative hospital mortality. We examined effects of the predictors AKI and year of surgery on mortality after accounting for preoperative risk factors and serious postoperative complications. Results: Between the first and second halves of the study period (1993 to 2002), the incidence of AKI increased from 5.1% to 6.6%, but the associated mortality rate decreased from 32% to 23% (P ⬍ 0.0001). Similarly, the incidence of AKI-D also increased from 1.5% to 2.0%, with a decrease in associated mortality from 61% to 49% (P ⬍ 0.01). In a risk-adjusted model, mortality in patients with AKI significantly decreased over time. Patients with AKI-D and with other organ system failures did not show improvement in survival over time. A preoperative history of congestive heart failure was associated significantly with a decrease in mortality risk over time, particularly in patients requiring dialysis. Limitations: Single-center, retrospective, observational cohort design. Conclusion: The incidence of AKI after cardiac surgery has increased over time. Although the adjusted risk of mortality decreased in patients with AKI without other postoperative complications, it is unchanged in those with multiorgan system failure. Am J Kidney Dis 50:703-711. © 2007 by the National Kidney Foundation, Inc. INDEX WORDS: Acute kidney injury; survival; cardiac surgery.

A

cute kidney injury (AKI) is one of the most serious complications occurring after cardiac surgery.1-3 Although overall postoperative mortality after cardiac surgery was relatively low, mortality in patients with severe AKI requiring dialysis exceeded 50%.4-12 Several observational studies reported the incidence, risk factors, and consequences of AKI after cardiac surgery.10,13-15 Conventional wisdom, based on the reported literature, states that: (1) the incidence of severe AKI after cardiac surgery is relatively low, and (2) the high mortality associated with postoperative AKI is unchanged during the past few decades.16,17 It is plausible that the comorbid disease burden of patients undergoing cardiac surgery has changed, along with the pattern of surgical procedures and postoperative care, including improvements in intensive care unit management and dialysis support. There are few data examining these factors for their effects on changing trends in incidence of AKI and mortal-

ity after cardiac surgery. Contributors to postoperative mortality in patients undergoing cardiac surgery include serious postoperative complications; moreover, it is well recognized that AKI after cardiac surgery commonly is associated with From the 1Department of Nephrology and Hypertension, Cleveland Clinic Foundation, Cleveland; 2Division of Nephrology and Hypertension, University of Cincinnati, Cincinnati; and Departments of 3Quantitative Health Sciences and 4 Cardiothoracic Anesthesiology, Cleveland Clinic Foundation, Cleveland, OH. Received December 6, 2006. Accepted in revised form July 10, 2007. Originally published online as doi: 10.1053/j.ajkd.2007.07.021 on September 20, 2007. Address correspondence to Charuhas V. Thakar, MD, FASN, Assistant Professor of Medicine, Division of Nephrology and Hypertension, University of Cincinnati College of Medicine, 231 Albert B. Sabin Way, MSB G-259, Cincinnati, OH 45267. E-mail: [email protected] © 2007 by the National Kidney Foundation, Inc. 0272-6386/07/5005-0004$32.00/0 doi:10.1053/j.ajkd.2007.07.021

American Journal of Kidney Diseases, Vol 50, No 5 (November), 2007: pp 703-711

703

704

Thakar et al

other organ system failure, such as the occurrence of serious infections, sepsis syndrome, and septic shock.15 However, it is unclear whether these factors could influence the changing trends in incidence of AKI or associated mortality. Two recent population-based studies derived from Medicare beneficiaries reported an increasing trend in incidence of AKI and a decrease in associated mortality in all hospitalized patients.18,19 These cross-sectional analyses were based on administrative databases and thus limited by the lack of patient-level data for clinical, comorbid, or laboratory variables. Although the observations raised an important question, there is limited applicability to a particular setting, such as cardiac surgery. We aim to examine temporal trends in the frequency of AKI and associated mortality in patients undergoing cardiac surgery. We also aim to study factors affecting these trends over time, including changes in comorbid disease burden, intraoperative factors, and postoperative complications. METHODS Study Design, Setting, and Participants We studied 33,217 patients who underwent cardiac surgery at the Cleveland Clinic Foundation, Cleveland, OH, between April 1993 and December 2002, as recorded in the database of the Department of Cardiothoracic Anesthesiology. This registry was approved by the Institutional Review Board to record perioperative information for cardiac surgery patients. There were 34,562 surgeries performed; for the purpose of this analysis, only the first surgical episode was considered. We excluded 1,540 patients from the analysis, including those requiring preoperative dialysis, preoperative extracorporeal membrane oxygenation, or preoperative tracheostomy or mechanical ventilation; heart transplant recipients; patients undergoing procedures for an automated implantable cardioverter-defibrillator, left ventricular assist device, or sternal procedure; and those with missing data. One hundred sixty-four patients met more than 1 criterion for exclusion, leaving 31,677 patients for analysis.

Definitions and Variables of Interest The primary outcome, AKI, was defined as the composite outcome of AKI requiring dialysis (AKI-D) or 50% or greater decrease in postoperative glomerular filtration rate (GFR) relative to baseline, but not requiring dialysis. Additionally, AKI-D was assessed as a subgroup analysis. Postoperative mortality was defined as hospital mortality. Preoperative GFR was estimated using baseline serum creatinine levels, whereas postoperative GFR was estimated based on peak serum creatinine levels during the immediate postoperative period. The 4-variable Modified

Diet in Renal Disease Study equation (National Kidney Foundation-Kidney Disease Outcomes Quality Initiative guidelines) was used to estimate GFR. The rationale for choosing these 2 definitions of AKI was based on clinical relevance and our prior validation of these definitions for their independent association with mortality after cardiac surgery.12 To analyze trends in incidence of AKI and mortality, we examined the following risk factors: age, sex, race, weight, preoperative renal function (evaluated as either continuous variables [serum creatinine in milligrams per deciliter or GFR in milliliters per minute per 1.73 m2] or a dichotomous variable [presence of preoperative chronic kidney disease, defined as preoperative GFR ⬍ 60 mL/min/1.73 m2]), serum albumin level (grams per deciliter), history of diabetes mellitus, chronic obstructive pulmonary disease, peripheral vascular disease, cerebrovascular disease, congestive heart failure (CHF), left ventricular dysfunction (ejection fraction ⬍ 35%), greater than 70% occlusion of the left main coronary artery, preoperative use of intra-aortic balloon pump, emergency surgery, and clinical severity score validated to predict morbidity and mortality after cardiac surgery.20 Intraoperative risk factors included duration of cardiopulmonary bypass and type of cardiac surgery, including coronary artery bypass graft surgery, valve surgery, combined coronary artery bypass graft and valve procedures, and other cardiac surgeries, such as ventricular aneurysm repair and septal defect repair. We also examined the effect of serious (nonrenal) postoperative complications on mortality, including cardiac morbidity, neurological morbidity, and serious infection, including sepsis syndrome and septic shock. A more detailed account of risk variables in the database, definitions of all risk factors, and postoperative complications was reported earlier.12,15 The rationale for selection of potential risk factors in the present analysis was based on a literature review of risk factors for mortality after cardiac surgery, our prior validation of the database, and clinical plausibility.

Statistical Methods We compared 1993 to 1997 and 1998 to 2002 on categorical and continuous risk factors by using chi-square and Wilcoxon rank sum tests, respectively. To assess change over time in the association between AKI and mortality while adjusting for changes over time in characteristics of patients developing AKI and for other risk factors, we first created a propensity score for AKI and then fit a logistic regression model for mortality using the propensity score. The propensity score was defined as the estimated probability of AKI from a logistic regression model for AKI, including all preoperative and intraoperative risk factors and postoperative morbidities occurring before AKI (if the date of morbidity or AKI was unknown, we treated those morbidities as occurring before AKI). We then fit logistic regression models for mortality in 1,000 bootstrapped samples with the risk factors chosen for inclusion in the model by using stepwise selection and forcing AKI, date of surgery (treated as a continuous variable), and AKI–date of surgery interaction into the model. In addition to risk factors defined previously, variables used in this selection procedure included quadratic and cubic terms

Improved Survival in AKI After Cardiac Surgery for date of surgery, interactions between date of surgery terms and all other risk factors, and other interactions suggested by the data or clinical knowledge. Predictors used in the final model were AKI, date of surgery, AKI–date of surgery interaction, and any variables significant in at least 50% of the bootstrap runs. Two different multivariable models were analyzed. The first model used the composite definition of AKI and included only preoperative and intraoperative risk factors, omitting postoperative nonrenal complications. In the second model, a subgroup of AKI-D was analyzed, and the model included postoperative nonrenal complications along with preoperative and intraoperative risk factors. To assess the fits of the multivariable models, we used HosmerLemeshow test and area under the receiver operating characteristics curve. If the interaction between AKI and date of surgery was statistically significant, we concluded that the mortality rate in patients with AKI had changed at a different rate than the overall mortality rate. Results of models are presented as odds ratios (with their 95% confidence intervals) for mortality in a given year versus 1993. All tests were 2 tailed and performed at a significance level of 0.05. SAS 9.1 software (SAS Institute, Cary, NC) was used for analyses, and R 2.2.1 software (The R Foundation for Statistical Computing, Vienna, Austria) was used for plots.

705

decreased during the 2 periods (1993 to 1997, 32.2% versus 1998 to 2002, 23.2%; P ⬍ 0.0001). Similarly, mortality in the AKI-D subgroup was lower during the second half of the study period (49%) compared with the first half (61%; P ⬍ 0.01). Distributions of risk factors in patients with AKI during the 2 blocks of time are listed in Tables 1 and 2. The data indicate that generally, the prevalence of comorbid disease burden and frequency of high-risk cardiac surgical procedures in patients with AKI increased over time. Regarding nonrenal postoperative complications, 1,062 patients developed serious infections after cardiac surgery for an overall frequency of 3.4%; frequencies of cardiac and neurological morbidities were 1.9% (n ⫽ 596) and 2.1% (n ⫽ 670), respectively. The frequency of postoperative complications in patients with AKI did not change over time, except for the frequency of neurological complications.

RESULTS

Frequency of AKI, Mortality, and Nonrenal Morbidities

The demographic distribution for the cohort (n ⫽ 31,677) was 69.5% men (n ⫽ 22,012) and 31.5% women (n ⫽ 9,665). Racial categories, as recorded in the database, included white (89.1%; n ⫽ 28,230), black (4%; n ⫽ 1,264), and others (6.9%; n ⫽ 2,183). During the entire study period, 1,863 of 31,677 patients developed AKI for an overall frequency of 5.9%. A total of 555 patients (1.8%) developed severe AKI-D. The incidence of AKI increased from 5.1% (815 of 15,733 patients) in the first half of the study period (1993 to 1997) to 6.6% (1,048 of 15,944 patients) in the second half (1998 to 2002). For those requiring dialysis, the incidence increased from 1.5% (231 of 15,733 patients) in the first half to 2.0% (324 of 15,944 patients) in the second half of the decade, for a 33% increase over time. Trends in frequencies of AKI and associated mortality are shown in Fig 1. There were 698 deaths after cardiac surgery, for an overall postoperative mortality rate of 2.2%. The mortality rate in patients with AKI was 27.1% (505 of 1,863 patients). Of 555 patients who developed severe AKI-D, 300 (54%) died during the postoperative hospital stay. As shown in Fig 1, mortality in patients with AKI significantly

Multivariate Model: AKI This multivariable model assessed the relationship between mortality, date of surgery, and postoperative AKI (defined as composite outcome) after adjusting for the following risk factors: age, weight, preoperative chronic kidney disease, CHF, pulmonary vascular disease, chronic obstructive pulmonary disease, previous cardiac surgery, date of surgery squared, interactions between risk factors and between risk factors and surgery date, and propensity score for AKI. The adjusted risk of mortality associated with AKI significantly decreased over time (Fig 2). Table 3 lists odds ratios for mortality (for AKI and no AKI) in later years of the study period versus 1993, as predicted by the multivariate model (because of space limitations, alternate years were chosen as representative of the trend during the entire period). The model discriminated well between those who died and those who survived (area under the receiver operating characteristics curve, 0.95), but was calibrated poorly (HosmerLemeshow statistic, P ⫽ 0.0003). This may be a result of omitting from the model any postoperative morbidities, which are the factors most strongly associated with mortality.

706

Thakar et al

Figure 1. Trends in acute kidney injury (AKI) and associated mortality after cardiac surgery. Abbreviations: AKI-D, AKI requiring dialysis; GFR, glomerular filtration rate.

Multivariate Model: AKI-D This regression model assessed the relationship between mortality, date of surgery, and postoperative AKI-D, including the following additional risk factors: age, sex, preoperative chronic kidney disease, CHF, chronic obstructive pulmonary disease, serum albumin level, previous cardiac surgery, cardiopulmonary bypass time, type of surgery, nonrenal postoperative morbidity, date of surgery squared, interactions between risk factors and between risk factors and surgery date, and propensity score for AKI-D. The model fit was adequate (Hosmer-Lemeshow, P ⫽ 0.7) and prediction was good (area under the receiver operating characteristics curve ⫽ 0.95). In this model, there were 3 factors for which the association with mortality significantly changed over time: (1) AKI-D, (2) nonrenal postoperative morbidity, and (3) preoperative history of CHF. A combination of these 3 factors formed 8 mutually exclusive groups of patients that differed on

the rate at which the risk of mortality changed over time. In the risk-adjusted model (Table 4), mortality rates decreased significantly over time in 5 of the 8 groups of patients: patients with AKI-D only, CHF only, preoperative CHF and AKI-D, preoperative CHF and nonrenal morbidity, and AKI-D, CHF, and nonrenal morbidity. Table 4 lists odds ratios for each group for mortality in the later years of the study period versus 1993, as predicted by the multivariable model (because of space limitations, alternate years were chosen as representative of the trend during entire period). DISCUSSION

Major determinants of postoperative AKI or associated mortality after cardiac surgery may have changed over time, along with improvements in dialysis and intensive care unit care. However, there are limited data to examine the effects of these factors on trends in AKI and

Improved Survival in AKI After Cardiac Surgery

707

Table 1. Changes in Risk Factors in Patients With Acute Kidney Injury Over Time

Categorical Risk Factors

N ⫽ 1,863 (100%)

Women Race White and other Black Insulin-requiring diabetes COPD Peripheral vascular disease Cerebrovascular disease Ejection fraction ⬍ 35% Left main disease Congestive heart failure IABP use Prior surgery Emergency surgery Preoperative chronic kidney disease† Surgery type CABG* only (N ⫽ 16,646) Valve only (N ⫽ 8,158) Combined (N ⫽ 5,122) Other (N ⫽ 1,751) Cardiac morbidity Neurological morbidity Serious infection Death

761 (41)

1993-1997 (N ⫽ 815; 100%)

1998-2002 (N ⫽ 1,048; 100%)

318 (39)

443 (42.3)

1,751 (93) 112 (6) 609 (32.6) 251 (13.5) 365 (19.6) 486 (26.1) 355 (19.0) 145 (7.8) 841 (45.1) 97 (5.1) 594 (31.9) 202 (10.8) 778 (41.7)

771 (94.6) 44 (5.4) 254 (31.2) 97 (11.9) 132 (16.2) 192 (23.6) 122 (15) 79 (9.7) 316 (38.8) 47 (5.8) 271 (33.3) 98 (12) 381 (46.8)

980 (93.5) 68 (6.5) 355 (33.9) 154 (14.7) 233 (22.2) 294 (28.1) 233 (22.2) 66 (6.3) 525 (50.1) 50 (4.8) 323 (30.8) 104 (9.9) 397 (37.9)

757 (40.6) 413 (22.2) 499 (26.8) 194 (10.4) 324 (17.4) 208 (11.2) 629 (33.8) 505 (27.1)

404 (49.6) 151 (18.5) 176 (21.6) 84 (10.3) 149 (18.3) 112 (13.7) 283 (34.7) 262 (32.2)

353 (33.7) 262 (25.0) 323 (30.8) 110 (10.5) 175 (16.7) 96 (9.2) 346 (33.0) 243 (23.2)

P*

0.2 0.3

0.2 0.08 ⬍0.01 0.03 ⬍0.0001 ⬍0.01 ⬍0.0001 0.3 0.3 0.2 ⬍0.0001 ⬍0.001

0.4 ⬍0.01 0.4 ⬍0.0001

Abbreviations: COPD, chronic obstructive pulmonary disease; IABP, intra-aortic balloon pump; CABG, coronary artery bypass grafting. *Chi-square test. †Chronic kidney disease defined as estimated glomerular filtration rate less than 60 mL/min/1.73 m2.

mortality in a cardiac surgery setting. The present study indicates that the incidence of AKI after cardiac surgery, regardless of its definition, steadily increased over time. Increasing comorbid disease burden, including major risk factors for AKI, may explain this increased incidence. Conversely, mortality rates in patients with AKI decreased during the same periods. In a risk-

adjusted model, patients developing AKI showed a significant improvement in survival over time. Trends in AKI and mortality in hospitalized patients were reported by 2 recent cross-sectional studies derived from administrative databases, the National Inpatient Sample and the Medicare 5% Sample Beneficiary Standard Analytical file. Xue et al18 reported that the overall

Table 2. Comparison of Continuous Risk Factors for Acute Kidney Injury Risk Factors

Overall Sample

1993-1997

1998-2002

P*

Age (y) Weight (kg) Albumin (g/dL) Preoperative GFR CPB time (min) Clinical severity score Hospital length of stay (d)

68.6 (59-75.2) 79 (67-91) 3.9 (3.4-4.2) 66.1 (47.3-89.3) 125 (90-168) 6.0 (3.0-9.0) 20 (11-35)

68.1 (58.9-74.4) 77.0 (67-90) 3.9 (3.5-4.2) 63.2 (46.3-87.3) 133 (97-179) 6 (3.0-8.0) 22 (12-41)

69.2 (59.1-75.6) 80.0 (67.0-93.0) 3.8 (3.3-4.2) 71.4 (49.1-91.3) 118 (84-162) 6 (4.0-9.0) 19 (11-31)

0.02 ⬍0.01 ⬍0.01 ⬍0.01 ⬍0.0001 ⬍0.0001 ⬍0.01

Note: Values expressed as median (first quartile/25th percentile-third quartile/75th percentile). Glomerular filtration rate estimated by using the Modification of Diet in Renal Disease Study equation. Abbreviations: GFR, glomerular filtration rate; CPB, cardiopulmonary bypass. *Wilcoxon rank-sum test.

708

Thakar et al

Figure 2. Mortality risk in patients with acute kidney injury (AKI) over time: multivariable analysis.

rate of AKI was 23.8 cases/1,000 hospital discharges, with rates increasing at 11%/y (between 1992 and 2001). Older age, male sex, and black race were associated with an increasing incidence of AKI. There was a decreasing trend in mortality in patients with AKI during the same period. In a similar study of Medicare beneficiaries, Waikar et al19 reported that the incidence of AKI increased from 61 to 288 cases/100,000 population between 1988 and 2002. During the same time, the mortality rate in patients with AKI decreased from 40.4% to 20.3% (P ⬍ 0.001). Although these hypothesis-generating

studies were based on a large sample of patients, they were limited by the lack of patient-level data for clinical, comorbid, or laboratory variables. Results are based on post hoc assessment of the accuracy of International Classification of Diseases, Ninth Revision diagnosis codes as reported during a 15-year period and are subject to changing patterns of physician reporting, rather than a true reflection of consistent disease definitions. Furthermore, by examining the effect of AKI in all hospitalized patients, the data assume that thresholds of clinically relevant AKI or risk of mortality in patients with AKI are similar

Table 3. Adjusted Odds Ratios of Mortality in Patients With AKI Subgroup

AKI No AKI

1995 v 1993*

1997 v 1993

1999 v 1993

2001 v 1993

0.46 (0.33-0.65) P ⬍ 0.0001 0.80 (0.54-1.2) P ⫽ 0.2

0.32 (0.20-0.51) P ⬍ 0.0001 0.66 (0.39-1.1) P ⫽ 0.1

0.28 (0.17-0.47) P ⬍ 0.0001 0.54 (0.31-0.93) P ⫽ 0.03

0.31(0.19-0.51) P ⬍ 0.0001 0.43 (0.25-0.75) P ⬍ 0.01

Note: Composite outcome model for AKI. Values expressed as odds ratio ( 95% confidence interval). Abbreviation: AKI, acute kidney injury.

Improved Survival in AKI After Cardiac Surgery

709

Table 4. Adjusted Odds Ratios for Mortality Over Time Within Patient Subgroups Subgroup

AKI-D only CHF only AKI-D and CHF AKI-D and nonrenal morbidities CHF and nonrenal morbidities AKI-D, CHF, and nonrenal morbidities Nonrenal morbidities only No postoperative morbidities

1995 v 1993

1997 v 1993

1999 v 1993

2001 v 1993

0.36 (0.21-0.63) P ⬍ 0.0001 0.39 (0.27-0.57) P ⬍ 0.0001 0.17 (0.10-0.30) P ⬍ 0.0001 0.99 (0.65-1.5) P ⫽ 1.0 0.39 (0.27-0.57) P ⬍ 0.0001 0.47 (0.31-0.72) P ⬍ 0.0001 0.82 (0.56-1.2) P ⫽ 0.3 0.82 (0.55-1.2) P ⫽ 0.3

0.16 (0.07-0.40) P ⬍ 0.0001 0.24 (0.14-0.42) P ⬍ 0.0001 0.06 (0.02-0.14) P ⬍ 0.0001 0.99 (0.54-1.8) P ⫽ 1.0 0.25 (0.15-0.41) P ⬍ 0.0001 0.34 (0.18-0.64) P ⬍ 0.0001 0.71 (0.43-1.2) P ⫽ 0.2 0.71 (0.40-1.3) P ⫽ 0.2

0.08 (0.02-0.25) P ⬍ 0.0001 0.20 (0.11-0.37) P ⬍ 0.0001 0.02 (0.01-0.08) P ⬍ 0.0001 1.02 (0.50-2.1) P ⫽ 1.0 0.20 (0.11-0.35) P ⬍ 0.0001 0.32 (0.16-0.66) P ⬍ 0.01 0.63 (0.37-1.10) P ⫽ 0.1 0.63 (0.33-1.2) P ⫽ 0.2

0.04 (0.01-0.16) P ⬍ 0.0001 0.21 (0.11-0.41) P ⬍ 0.0001 0.01 (0.00-0.06) P ⬍ 0.0001 1.07 (0.48-2.4) P ⫽ 0.9 0.21 (0.12-0.36) P ⬍ 0.0001 0.39 (0.18-0.87) P ⬍ 0.01 0.57 (0.33-1.00) P ⫽ 0.05 0.57 (0.28-1.1) P ⫽ 0.1

Note: AKI, composite definition odds ratios calculated versus no AKI as reference group. Results based on AKI-D model Values expressed as odds ratio (confidence interval) Abbreviations: AKI, acute kidney injury; AKI-D, AKI requiring dialysis; CHF, congestive heart failure.

across different clinical settings. However, although the analyses raise an important question, application of these data to a particular clinical setting is relatively limited. We examined trends in AKI and mortality in the setting of cardiac surgery. Data indicated that such risk factors as peripheral vascular disease, decreased ejection fraction, history of CHF, and high-risk surgical procedures (valve surgery/ combined coronary artery bypass graft and valve procedures) were more prevalent in subjects with AKI during the second half of the decade than during the first. These differences in comorbid disease burden could be considered one of the contributors to the increasing incidence of AKI. The increasing incidence of AKI was present regardless of degree of severity of renal injury. However, contrary to the prevalent notion, the mortality rate in patients with AKI after cardiac surgery significantly decreased over time. It is well recognized that independent of AKI, mortality after cardiac surgery can be influenced by other nonrenal postoperative complications, such as cardiac complications, serious infections, sepsis syndrome, or septic shock.20-22 Hence, we analyzed the effect of AKI on mortality over time by using 2 separate models: (1) a model that tested the effect of AKI (composite definition) while accounting for preoperative and intraoperative risk factors, and (2) a subgroup

analysis of AKI-D that accounted for preoperative and intraoperative risk factors along with nonrenal postoperative complications. Although both models discriminated well, the model fit for the second model was better than for the first. By using multivariable analysis, the adjusted risk of mortality in patients with AKI decreased over time. Improved survival also was evident in a subgroup analysis for AKI-D, but without other postoperative complications. We interpret the data to indicate that although AKI was associated with improved survival during the study period, mortality in patients who developed severe AKI-D and other organ system failure is unchanged over time. Reasons for decreasing mortality in patients with AKI are not completely clear. One can be cautiously optimistic that changes in practices of dialysis care, including initiation of dialysis therapy, dose of dialysis, or use of different modalities of renal replacement therapy, may have contributed to the improved outcomes in these patients. These reasons are speculative because data regarding these variables were not recorded in the registry and hence could not be examined for their effect on mortality trend. However, it should be noted that in the absence of standardized objective criteria to initiate dialysis therapy or a dose or modality of choice in AKI, availability of such data would still be met

710

with inherent biases. Regarding the preoperative comorbid disease burden, although the frequency of patients with CHF was greater in the second half of the study, mortality in these patients was lower, more so in those who developed postoperative AKI-D. This effect was independent of other confounding risk factors. We interpret these observations to suggest that improvements in the preoperative management of patients with CHF may be linked to improved postoperative AKI outcomes. There are certain limitations to this analysis. The report involves experience from a single center. However, the data represent a very large cohort of patients recorded in a meticulously maintained prospective registry, which allowed us to perform trend analyses with patient-level data. The cohort also is well represented by differences in demographic distribution and includes all major risk factors for both AKI and mortality, thus increasing its generalizability. Without the availability of a standardized definition of AKI or objective criteria to initiate dialysis therapy, it is inherent that practices of initiation of dialysis therapy may have changed over time. Thus, examining the subgroup of AKI-D is involved with certain amount of bias while performing a trend analysis. Additionally, it should be noted that the risk of mortality associated with varying degrees of AKI follows a nonlinear relationship, and postoperative morbidities other than AKI also are associated strongly with mortality. To account for these limitations, we present data classified according to 2 different degrees of severity of AKI, as well as with or without the effect of postoperative complications. It is possible that subjective definitions of such risk factors as diabetes or CHF may have changed over time and cannot be accounted for in the database. Additionally, the scope of the database does not record patient outcomes after hospital discharge, including the location of disposition (eg, home versus nursing home). Thus, outcome data in our study are restricted to duration of hospitalization. This emphasizes the issue of lack of longitudinal data in the setting of AKI, a prevalent deficiency in the literature, and the need to maintain prospective data for a longer follow-up in a large cohort of patients. However, despite some of its limitations, the present observations suggest important clinical implications.

Thakar et al

In addition, the cardiac surgery setting is considered an important “clinical research model” to examine strategies of prevention, early diagnosis, and treatment of AKI. The overall incidence of postoperative AKI remains relatively low, which necessitates preoperative risk stratification of patients for inclusion into clinical trials. The present study indicates that future clinical trials related to prevention or treatment of AKI need to consider the changing patterns of frequency of AKI and associated mortality, along with factors influencing these trends. The incidence of AKI after cardiac surgery, regardless of its definition, increased over time. The increasing frequency of major risk factors of AKI is one of the likely contributors to the increasing incidence of AKI. Conversely, postoperative mortality in patients with AKI decreased over time. Patients developing AKI showed a significant decrease in mortality. Persistently high mortality in patients with postoperative multiorgan system failure along with severe AKI-D remains a therapeutic challenge. ACKNOWLEDGEMENTS The authors acknowledge the assistance of personnel from the Department of Cardiothoracic Anesthesiology in maintaining the database. Support: Dr Thakar is supported by a career development award (Merit Review Entry Program) from the Department of Veteran’s Affairs and National Institute of Health award DK071802-01A. Financial Disclosure: None.

REFERENCES 1. Abel RM, Wick J, Beck CH Jr, Buckley MJ, Austen WG: Renal dysfunction following open-heart operations. Arch Surg 108:175-177, 1974 2. Chertow GM, Levy EM, Hammermeister KE, Grover F, Daley J: Independent association between acute renal failure and mortality following cardiac surgery. Am J Med 104:343-348, 1998 3. Thakar CV, Worley S, Arrigain S, Yared J-P, Paganini EP: Influence of renal dysfunction on mortality after cardiac surgery: Modifying effect of preoperative renal function. Kidney Int 67:1112-1119, 2005 4. Davis K, Kennedy JW, Kemp HG Jr, Judkins MP, Gosselin AJ, Killip T: Complications of coronary arteriography from the Collaborative Study of Coronary Artery Surgery (CASS). Circulation 59:1105-1112, 1979 5. Anderson RP: Mortality rates and risk factors in coronary bypass surgery. Ann Thorac Surg 47:489-490, 1989 6. Grover FL, Hammermeister KE, Burchfiel C: Initial report of the Veterans Administration Preoperative Risk Assessment Study for Cardiac Surgery. Ann Thorac Surg 50:12-26, 1990

Improved Survival in AKI After Cardiac Surgery 7. Hammermeister KE, Burchfiel C, Johnson R, Grover FL: Identification of patients at greatest risk for developing major complications at cardiac surgery. Circulation 82:380-389, 1990 8. Tu JV, Jaglal SB, Naylor CD: Multicenter validation of a risk index for mortality, intensive care unit stay, and overall hospital length of stay after cardiac surgery. Steering Committee of the Provincial Adult Cardiac Care Network of Ontario. Circulation 91:677-684, 1995 9. Conlon PJ, Stafford-Smith M, White WD, et al: Acute renal failure following cardiac surgery. Nephrol Dial Transplant 14:1158-1162, 1999 10. Chertow GM, Lazarus JM, Christiansen CL, et al: Preoperative renal risk stratification. Circulation 95:878-884, 1997 11. Mangano CM, Diamondstone LS, Ramsay JG, Aggarwal A, Herskowitz A, Mangano DT: Renal dysfunction after myocardial revascularization: Risk factors, adverse outcomes, and hospital resource utilization. The Multicenter Study of Perioperative Ischemia Research Group. Ann Intern Med 128:194-203, 1998 12. Thakar CV, Liangos O, Yared JP, et al: ARF after open-heart surgery: Influence of gender and race. Am J Kidney Dis 41:742-751, 2003 13. Fortescue EB, Bates DW, Chertow GM: Predicting acute renal failure after coronary bypass surgery: Crossvalidation of two risk-stratification algorithms. Kidney Int 57:2594-2602, 2000 14. Thakar CV, Arrigain S, Worley S, Yared J-P, Paganini EP: A clinical score to predict acute renal failure after cardiac surgery. J Am Soc Nephrol 16:162-1685, 2005

711 15. Thakar CV, Yared JP, Worley S, Cotman K, Paganini EP: Renal dysfunction and serious infections after openheart surgery. Kidney Int 64:239-246, 2003 16. Siegel NJ, Shah SV: Acute renal failure: Directions for the next decade. J Am Soc Nephrol 14:2176-2177, 2003 17. Rosner MH, Okusa MD: Acute kidney injury associated with cardiac surgery. Clin J Am Soc Nephrol 1:19-32, 2006 18. Xue JL, Daniels F, Star RA, et al: Incidence and mortality of acute renal failure in Medicare beneficiaries, 1992 to 2001. J Am Soc Nephrol 17:1135-1142, 2006 19. Waikar SS, Curhan GC, Wald R, McCarthy EP, Chertow GM: Declining mortality in patients with acute renal failure, 1988 to 2002. J Am Soc Nephrol 17:1143-1150, 2006 20. Higgins TL, Estafanous FG, Loop FD, et al: ICU admission score for predicting morbidity and mortality risk after coronary artery bypass grafting. Ann Thorac Surg 64:1050-1058, 1997 21. Higgins TL, Estafanous FG, Loop FD, Beck GJ, Blum JM, Paranandi L: Stratification of morbidity and mortality outcome by preoperative risk factors in coronary artery bypass patients. A clinical severity score. JAMA 267:2344-2348, 1992 22. Kennedy JW, Kaiser GC, Fisher LD, et al: Clinical and angiographic predictors of operative mortality from the Collaborative Study of Coronary Artery Surgery (CASS). Circulation 63:793-802, 1981