Original Investigation Acute Kidney Injury and Prognosis After Cardiopulmonary Bypass: A Meta-analysis of Cohort Studies John W. Pickering, PhD,1 Matthew T. James, MD, PhD, FRCPC,2 and Suetonia C. Palmer, MBChB, PhD1 Background: Robust estimates and sources of variation in risks of clinical outcomes for cardiopulmonary bypass (CPB)-associated acute kidney injury (AKI) are needed to inform clinical practice and policy. We aimed to assess whether the methods for defining acute kidney disease modify the estimated association of AKI with CPB. Study Design: Systematic review and meta-analysis. Setting & Population: Adults undergoing CPB. Selection Criteria for Studies: Cohort studies reporting adjusted associations between CPB-associated AKI and early mortality, later mortality, stroke, myocardial infarction, congestive heart failure, all-cause hospitalization, chronic kidney disease, end-stage kidney disease, bleeding complications, or perioperative infection. Predictors: CPB-associated AKI and renal replacement therapy. Outcomes: The primary outcome was early mortality (in-hospital or within 90 days of surgery) in studies reporting adjusted associations and secondary outcomes including total and cardiovascular mortality, major adverse cardiovascular events, rehospitalization, end-stage kidney disease, bleeding, and perioperative infection. Results: 46 studies with 47 unique cohorts comprising 242,388 participants were included. The pooled rate of CPB-associated AKI was 18.2%, and of renal replacement therapy, 2.1%. CPB-associated AKI was associated with early mortality (risk ratio [RR], 4.0; 95% CI, 3.1-5.2; crude mortality with CPB-associated AKI, 4.6%; without CPB-AKI, 1.5%) with considerable heterogeneity between studies (I 2 5 87%). The AKI definition did not modify prognostic estimates (P for subgroup analysis 5 0.9). When heterogeneity was fully accounted for using credibility ceilings, risks of early mortality were attenuated (RR, 2.2; 95% CI, 1.82.8) but remained high. Renal replacement therapy also was associated with early mortality (RR, 5.3; 95% CI, 3.4-8.1). CPB-associated AKI also was associated with long-term mortality (RR, 2.0; 95% CI, 1.7-2.3) and stroke (RR, 2.2; 95% CI, 1.1-4.5). No other outcomes were reported in more than 3 studies. Limitations: Unclear attrition from follow-up in most studies and variable adjustment for confounders across studies. Conclusions: CPB-associated AKI is associated with a more than 2-fold increase in early mortality regardless of AKI definition. Am J Kidney Dis. 65(2):283-293. ª 2015 by the National Kidney Foundation, Inc. INDEX WORDS: Cardiopulmonary bypass (CBP); cardiac surgery; coronary artery bypass graft; valve surgery; meta-analysis; acute kidney injury (AKI); acute renal failure (ARF); mortality; stroke.

I

schemic-reperfusion injury is common following cardiopulmonary bypass (CPB) and causes acute kidney injury (AKI).1 The importance of AKI as a risk factor for mortality following cardiac surgery is considered axiomatic; however, it has been argued that the previous absence of standardized definitions has made assessing the prognostic implications of CBP-associated AKI difficult. A robust understanding of prognosis associated with AKI is needed to enhance clinical decision making and provide a sound benchmark for epidemiologic and interventional research. Despite many published reports describing the association of CBP-associated AKI with adverse outcomes, there have been few attempts to systematically summarize the prognostic implications of AKI in this clinical setting. In a subgroup analysis within a global incidence study of AKI, Susantitaphong et al2 found that CBP-associated AKI was associated with an 8-fold increase in mortality based on results from 23 Am J Kidney Dis. 2015;65(2):283-293

studies. However, this study was not designed to assess the impact of the different AKI definitions on estimated prognosis, evaluate the impact of adjustment for other AKI risk factors, or examine the associations between AKI and other outcomes. Standardized AKI definitions, including first the RIFLE (risk, injury, failure, loss, end-stage renal From the 1Department of Medicine, University of Otago Christchurch, Christchurch, New Zealand; and 2Department of Medicine, University of Calgary, Calgary, Alberta, Canada. Received March 26, 2014. Accepted in revised form September 7, 2014. Originally published online November 4, 2014. Address correspondence to John W. Pickering, PhD, Department of Medicine, University of Otago Christchurch, PO Box 4345, Christchurch 8140, New Zealand. E-mail: john.pickering@ otago.ac.nz  2015 by the National Kidney Foundation, Inc. 0272-6386 http://dx.doi.org/10.1053/j.ajkd.2014.09.008 283

Pickering, James, and Palmer

disease) consensus definition in 20043 followed by the Acute Kidney Injury Network (AKIN)4 and the KDIGO (Kidney Disease: Improving Global Outcomes)5 definitions, facilitates prognostic analyses for AKI in various clinical settings, but it is unclear whether these definitions have different prognostic value. To address these knowledge gaps, we conducted a systematic review and meta-analysis to examine the prognostic implications of CPB-associated AKI for mortality and other clinical outcomes, including adverse cardiovascular and renal events. We specifically aimed to determine whether the association between CPB-associated AKI and mortality differed according to AKI definition.

METHODS Study Protocol This systematic review and meta-analysis was conducted using a prespecified protocol and according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA).6

Search Strategy and Selection Criteria We searched MEDLINE and EMBASE without language restriction for January 1, 2004, to June 2, 2014, using the following medical subject heading (MeSH) terms: acute kidney injury, renal failure, renal replacement therapy, dialysis, cardiac surgical procedure, cardiopulmonary bypass, coronary artery bypass, mortality, and death (see Item S1, available as online supplementary material, for search terms). We limited the search to studies published since the development of the first consensus AKI definition (RIFLE) in 2004.3 We included cohort studies if they reported data for the association between AKI after surgery supported by CPB and one of the prespecified clinical outcomes. We excluded studies in children and studies in which .25% of participants underwent off-pump cardio-artery bypass grafting. For mortality outcomes, we considered only studies with adjusted risk estimates.

Definition of Exposure We considered AKI as defined by KDIGO,5 AKIN,4 RIFLE,3 or the study investigators. We assigned a primary definition to each study, according to the available consensus definitions in reverse chronological order: KDIGO, AKIN, RIFLE, or study investigators, where reported in each publication. The primary definition was used in all analyses unless otherwise stated. We also obtained risk estimates for the outcome of renal replacement therapy (RRT) separately from AKI, when reported.

Outcome Measures The primary outcome was early mortality defined as in-hospital mortality or within 90 days. Secondary outcomes were total mortality, cardiovascular mortality, major adverse cardiovascular events (stroke, myocardial infarction, and congestive heart failure), all-cause hospitalization, chronic kidney disease (CKD), end-stage kidney disease requiring long-term RRT, bleeding complications, or perioperative infection.

Data Extraction Two reviewers (J.W.P. and S.C.P.) independently reviewed titles and abstracts of a randomly selected 10% sample of citations to reach consensus on eligible studies. When consensus was achieved, a single reviewer (J.W.P.) screened all remaining 284

citations. All potentially eligible citations then were examined in detail to identify studies that fulfilled the eligibility criteria. We extracted prevalence of CPB-associated AKI and risk estimates for outcomes (as adjusted risk ratios [RRs], hazard ratios, or odds ratios together with their 95% confidence intervals [CIs]). When only crude outcome data were provided for nonmortality outcomes, we calculated unadjusted RRs. We treated multiple reports of results from the same study as a single risk evaluation and entered data for disaggregated cohorts within studies into the analyses separately. We extracted prespecified descriptive statistics for the key clinical characteristics and outcomes for each included cohort.

Statistical Analysis We converted point estimates and their variances for each odds ratio to an RR when required using the techniques described in references.7,8 We assumed that hazard ratios reasonably approximated relative risks. We summarized RRs using random-effects meta-analysis and reported results as RR together with 95% CI.9 Heterogeneity was assessed by the I2 statistic, which provides an estimate of the proportion of variation in risk estimates among studies that is beyond that expected by chance.10 We considered I2 thresholds of ,25%, 25% to 75%, and .75% to represent low, moderate, and high heterogeneity, respectively.10 When individual study–adjusted risk estimates had been provided separately for CPB-associated AKI severity stages, we calculated an overall CPB-associated AKI risk estimate using a fixed-effects model. In sensitivity analyses, we used credibility ceilings to account for possible spurious precision of pooled risk estimates11 and additionally determined summary estimate risks for the primary outcome provided by the minimum credibility ceiling that generated I2 5 0%. We calculated the 95% prediction interval, which is the expected range of the effect in a new study.12 Potential sources of heterogeneity were explored using subgroup and univariate metaregression analyses and considered the following prespecified study-level demographic and clinical variables: age, sex, CKD, diabetes, ejection fraction, emergency surgery, radiocontrast use, intra-aortic pump use, operation type (coronary artery bypass grafting, valve surgery, or combination), CPB duration, cross-clamp time, preoperative creatinine level, offpump proportion, the biomarker used to identify CPB-associated AKI (plasma or serum creatinine, estimated glomerular filtration rate [eGFR], urine output, or any combination of the 3 surrogates), and World Health Organization global region.13 We adjudicated study risks of bias according to domains described in the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) statement,14 including specific items concerning AKI definition, adjustment for potential confounding factors, and clarity of reporting of outcome assessments and statistical analyses. We conducted all analyses using packages meta and metafor from R statistical software version 3.1.0 (R Foundation for Statistical Computing) and used the ceiling.data function (www.dhe.med.uoi.gr/assets/software/ceiling.txt).

RESULTS Description of Included Studies and Participants The systematic search identified 3,618 citations, of which 320 were retrieved for full-text examination after review by title and abstract. Overall, 46 studies reporting on 47 cohorts comprising 242,388 patients (Fig 1; Table 1) were eligible and included in the review. Sample size varied from 68 to 28,220 (median, 1,610) participants (Table 1). There were 45 cohorts adjusted for covariates reporting at least one Am J Kidney Dis. 2015;65(2):283-293

Meta-analysis of AKI after CPB

Figure 1. Flow diagram; the process to identify eligible studies. Note: Some studies report more than one acute kidney injury (AKI) definition or one outcome. Abbreviations: AKIN, Acute Kidney Injury Network definition; AMI, acute myocardial infarction; CPB, cardiopulmonary bypass; CHF, congestive heart failure; Hosp, hospitalization; RRT, renal replacement therapy; SI, study investigator definition; RIFLE, risk, injury, failure, loss, end-stage renal disease definition.

of early or late mortality outcomes. Data were extractable for the association between CPB-associated AKI and a prespecified clinical outcome in 36 cohorts comprising 200,129 participants and between RRT and a prespecified outcome in 12 cohorts (52,674 participants). The total incidence of CPB-associated AKI was 18.2% (cohort range, 1.0%-53.0%). The total incidence of RRT was 2.1% (range, 0.6%-23.2%). In total, 24.3% of participants had diabetes and 17.6% had preexisting CKD. Overall, 14.0% underwent emergency surgery, 58.0% underwent coronary artery bypass grafting only, 31.6% underwent cardiac valve surgery only, and 15.8% underwent combined valve and coronary artery bypass grafting surgery. Most cohorts were from Europe (n 5 24 [69.4% of participants]) and North America (n 5 14 [18.8%]). One cohort was from South or Central America (0.3%); 2, from the Eastern Mediterranean (0.5%); and 6, from the Western Pacific (3 from Oceania [10.1%] and 3 from Asia [0.9%]). Plasma creatinine level (29 cohorts), change in eGFR (3 cohorts), plasma creatinine level or eGFR (3 cohorts), and plasma creatinine level or urine output (2 cohorts) were used to define CPB-associated AKI. GFR was estimated by the MDRD (Modification of Am J Kidney Dis. 2015;65(2):283-293

Diet in Renal Disease) Study equation15 (5 cohorts) or Cockcroft-Gault equation16 (1 cohort). In 2 cohorts, AKI was considered present with eGFR reduction of 25%. In 8, 10, and 19 cohorts, respectively, AKI was defined by AKIN, RIFLE, and study investigators’ definitions. Risk of Bias The timing of AKI ascertainment was not defined clearly in many (46%) studies (Table S1). Furthermore, it was unclear whether AKI was determined according to the periods specified by AKIN and RIFLE criteria for many studies in which these definitions were used. Losses to follow-up were quantified in only 50% of the cohorts reporting postdischarge outcomes. Outcomes and study populations were clearly defined. Early Mortality The risk of early mortality (in-hospital or ,90 days) following CPB-associated AKI and adjusted for covariates was reported in 24 cohorts (94,141 participants; Table S2 contains details of adjusted-for covariates). Overall, 4.6% with AKI died and 1.5% without AKI died. There was an association between 285

286

Table 1. Included Cohorts

Study

N

Exposurea

Mean Age (y)

Male Sex

Ahmed40 (2009) Arnaoutakis41 (2007)

2,054 267

4.2% w/ AKIb 48.3% w/ RIFLE

64.2 63.5

76.1%

Boyle21 (2006) Brown42 (2010) Che43 (2011) Dardashti44 (2014) Davoodi45 (2007) De Santo46 (2009) De Santo47 (2010) Del Duca48 (2007) Dell’Amore49 (2012)

756 4,931 1,056 5,742 707 1,047 1,424 649 285

5.8% w/ RRT 38.5% w/ AKIN 31.1% w/ AKIN 11.3% w/ RIFLE 1.0% w/ AKIb 38.1% w/ RIFLE 9.6% w/ AKIb 24.0% w/ AKIb 2.5% w/ RRT

56.9 66.2 57.3 67.3 58.0 63.2 61.9 65 82

80.2%

6.8% w/ AKIb

83

Faggian50 (2011)

251

DM

CKD

36% 12%

Off Pump

0% 0%

18%

4% 28.9% 7.4%

0% 0.52% 0% 3.8% 0% 0% 0% 0% 0%

54%

21%

8%

0%

85.1% 71.2% 53% 68.9% 61.1%

32% 15.9% 22% 36.6% 38.9% 16.4%

22%

Outcome

Long-term mortality Early mortality Early mortality; cardiac morbidityc Long-term mortality Early mortality Long-term mortality Early mortality Early mortality Early mortality Early mortality Early mortality; long-term mortality Early mortality

4,694

12.0% w/ RIFLE

66.6

79.6%

31%

30.4%

7.5%

Long-term mortality; stroke

Haase52 (2012) Habib22 (2005) Hei53 (2011) Holzmann54 (2013) Ivert55 (2014) ibid Jeppsson56 (2005)

920 1,760 68 23,584 28,220 27,709 151

19.5% w/ RIFLE 16.2% w/ RIFLE 30.9% w/ AKIb 11.7% w/ AKIN 0.6% w/ RRT 0.4% w/ RRT 23.2% w/ RRT

64.2 64 49.2 66.8 67 67 69

73.3% 69% 76.5% 79%

31% 34.1%

27.2%

0% 0% 0% 5.9% 5.2% 5.2% 0%

Early mortality Hospitalizationc Early mortality Stroke Early mortality Long-term mortality Long-term mortality

Karimi57 (2008) Karkouti32 (2009) Kohl58 (2013)

475 3,460 1,616

6.1% w/ AKIb 24.0% w/ RIFLE 10.6% w/ AKIb

60.4 65.2 66.7

64.6% 75% 65.3%

ibid Li59 (2011)

1,616 964

4.5% w/ RRT 19.3% w/ AKIN

66.7 68.9

65.3% 82.2%

25,665 843

53.0% w/ AKIb 17.2% w/ AKIb

67 63

72.7%

Liotta60 (2014) Loef19 (2005)

23%

63%

28.2% 23.1%

29.4%

23.1% 43% 24% 11.5%

0% 0% 0%

Early mortality Early mortality Early mortality

0% 0%

Early mortality Early mortality; long-term mortality Early mortality Early mortality; long-term mortality; strokec; AMIb

5.8% 0%

(Continued)

Isolated CABG Aortic arch surgery with deep hypothermic circulatory arrest Cardiac transplantation Cardiac surgery; not on dialysis Cardiac surgery; not on dialysis CABG; not emergency; excluded died within 7 d CABG with LVEF # 30% CABG; not on dialysis Valve surgery; not on dialysis Cardiac surgery; not on dialysis; all received contrast Valve surgery; very elderly Nonelective urgent or emergent cardiac surgery; age $ 80 y Isolated first-time CABG; not on dialysis; did not die in hospital Cardiac surgery; not on dialysis CABG; no preoperative kidney failure Open-heart surgery requiring ECMO CABG; not on dialysis CABG; not on dialysis CABG; not on dialysis; survived to 90 d Surgical repair of postinfarction ventricular septal rupture; survived first 30 d Open heart surgery with perioperative IABP support Cardiac surgery; not on dialysis Cardiac surgery; normal vs new protocol for glycemic control (2:1) Isolated CABG; not emergency surgery Isolated CABG Cardiac surgery; not on dialysis

Pickering, James, and Palmer

Am J Kidney Dis. 2015;65(2):283-293

Gallagher51 (2013)

Cohort Description

Study

Loef61 (2009) Lopez-Delgado62 (2013) Machado63 (2009) Mehta17 (2010) Melby64 (2007)

N

624 2940 817 10,415 245

Exposurea

Mean Age (y)

13.6% w/ AKIb 13.9% w/ AKIb

63.2 54.5

48.5% w/ AKIN 20.0% w/ AKIb

Male Sex

DM

Outcome

Cohort Description

11.5% 8.2%

0% 0%

Long-term mortality Early mortality

Isolated CABG; not on dialysis Cardiac surgery; not on dialysis

61 65

70% 69%

34.8% 34%

0% 0%

CABG; not on dialysis CABG; creatinine , 2 mg/dL; did not die within 7 d

11.8% w/ AKIb

83.6

47%

18%

0%

Early mortality Early mortality; strokec; AMIc; cardiac tamponadec Early mortality; long-term mortality Early mortality Early mortality Early mortality HF (first hospitalization)

3,109 6,465 2,634 24,018

5.1% w/ RIFLE 3.6% w/ RIFLE 3.1% w/ RRT 11.6% w/ AKIN

66.4 63.8 60 67

0% 100% 67.6% 79%

31.5% 24.8% 2.5% 23%

Rahmanian67 (2010) Rahmanian68 (2013) Ryde´n69 (2012) Ryde´n20 (2014) Saxena70 (2012)

6,641 5,318 7,594 27,929 21,534

1.5% w/ RRT 5.5% w/ RRT 13.8% w/ AKIN 13.0% w/ AKIN 3.7% w/ AKIb

63.9 68.9 65.5 67 65.84

62% 70.7% 80% 79% 77.9%

24% 28% 20% 23% 32.7%

Shapira71 (2006) Tamayo72 (2012) Tekumit73 (2010) Toumpoulis18 (2008)

115 1,610 137 4,140

19.1% w/ RRT 6.5% w/ AKIb 47.4% w/ AKIb 0.8% w/ RRT

66 67.8 62.63 64.8

76.5% 60.9% 57.7% 68.7%

42% 29.5% 25% 34.6%

Toumpoulis74 (2008)

1,376

2.1% w/ RRT

65

53.4%

Urso75 (2007) Vargas Hein76 (2006)

100 2,683

26.0% w/ AKIb 9.5% w/ RRT

82.1 67

3.1% w/ RRT

81.5

355

Off Pump

76.6%

Mitter65 (2010) Mitter65 (2010) Olsson66 (2007) Olsson23 (2013)

Zingone77 (2009)

CKD

30% 18.8%

18%

3.2% 3.3% 0% 5.9% 0% 0% 0.4% 8.1%

2.4%

0% 0% 0% 0%

17.7%

5%

0%

52% 68.3%

9%

17%

0% 0%

53.2%

19.4%

3.4%

6.7%

4.96%

Early mortality Early mortality Early mortality; stroke Long-term mortality; AMI Early mortality; long-term mortality Long-term mortality Early mortality Long-term mortality Early mortality; long-term mortality; stroke Early mortality; long-term mortality Early mortality Early mortality Early mortality

Aortic valve replacement; age . 80 y Cardiac surgery; not on dialysis; women aged .30 y Cardiac surgery; not on dialysis; men aged .30 y Operations on proximal thoracic aorta Isolated CABG; not on dialysis; eGFR . 15; survived until 30 d Cardiac surgery; not on dialysis Cardiac surgery; not on dialysis Isolated CABG; not on dialysis; survived .48 h CABG; no MI within 7 d; no previous cardiac surgery Isolated CABG CABG with LVEF # 30% Cardiac surgery Elective isolated CABG Isolated CABG Valve surgery Aortic valve replacement; age . 80 y Cardiac surgery; ICU . 1 d; not on dialysis; no liver failure; no preoperative IABP use Cardiac surgery; age . 80 y

287

Abbreviations and definitions: AKI, acute kidney injury; AKIN, Acute Kidney Injury Network; AMI, acute myocardial infarction; CABG, coronary artery bypass grafting; CHF, congestive heart failure; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; DM, diabetes mellitus; ECMO, extracorporeal membrane oxygenation; eGFR, estimated glomerular filtration rate (in mL/min/1.73 m2); IABP, intra-aortic balloon pump; ICU, intensive care unit; LVEF, left ventricular ejection fraction; MI, myocardial infarction; MODS, multiple organ dysfunction syndrome; NYHA, New York Heart Association classification; PVD, peripheral vascular disease; RRT, renal replacement therapy; RIFLE, risk, injury, failure, loss, end-stage renal disease. a RIFLE and AKIN denote individuals who met these respective definitions of AKI. b Study investigator’s definition of AKI. c Outcome not adjusted for covariates

Meta-analysis of AKI after CPB

Am J Kidney Dis. 2015;65(2):283-293

Table 1 (Cont’d). Included Cohorts

Pickering, James, and Palmer

Figure 2. Early mortality risk following cardiopulmonary bypass– associated acute kidney injury (primary outcome). Abbreviations: AKIN, Acute Kidney Injury Network definition; CI, confidence interval; RIFLE, risk, injury, failure, loss, end-stage renal disease definition.

CPB-associated AKI and early mortality (RR, 4.0; 95% CI, 3.1-5.2; Fig 2), although there was high-level heterogeneity among studies (I2 5 87%). When the heterogeneity statistic (I2) was reduced to zero using a credibility ceiling of 4.0%, the strength of the association of CPB-associated AKI with early mortality was attenuated, but significant (RR, 2.2; 95% CI, 1.82.8; Fig S1). This is likely to be a conservative figure. The study with the highest weight also had the lowest risk estimate, perhaps because it excluded those who died in the first 7 days postsurgery.17 Sensitivity analysis excluding this study, and after application of a credibility ceiling of 0.65% to reduce I2 to zero, increased the risk estimate to 3.0 (95% CI, 2.4-3.6).

RRT similarly was associated with early mortality (12 cohorts, 54,175 participants; RR, 5.3 [95% CI, 3.4-8.1]; high heterogeneity, I2 5 81%; Fig 3). When I2 was reduced to zero by applying a credibility ceiling of 0.15%, the association with early mortality again was attenuated, but marked (RR, 3.8; 95% CI, 2.9-5.0; Fig S2). Data for early mortality according to CPBassociated AKI severity were provided in 6 cohorts (16,364 participants). Compared to no AKI, there was a dose-response relationship with increasing severity of AKI (unadjusted): RRs for stages 1 to 3, respectively, of 2.4 (95% CI, 1.4-5.6), 7.0 (95% CI, 2.222.6), and 17.2 (95% CI, 6.1-48.6; Fig S3).

Figure 3. Early mortality risk following renal replacement therapy. Abbreviation: CI, confidence interval. 288

Am J Kidney Dis. 2015;65(2):283-293

Meta-analysis of AKI after CPB

Figure 4. Risk of stroke following cardiopulmonary bypass–associated acute kidney injury. *Nonadjusted outcomes. Abbreviation: CI, confidence interval.

Long-term Mortality There was an association between CPB-associated AKI and long-term mortality in 12 cohorts (69,797 participants; RR, 2.0 [95% CI, 1.7-2.3]; high heterogeneity, I2 5 74%; Fig S4). When I2 was reduced to zero by applying a credibility ceiling of 0.15%, the association with long-term mortality persisted (RR, 1.8; 95% CI, 1.6-2.1). Similarly, RRT was associated with long-term mortality (5 cohorts, 33,625 participants; RR, 2.1 [95% CI, 1.6-2.9]; low heterogeneity, I2 5 6.4%; Fig S5). Stroke There was an association between CPB-associated AKI and postoperative in-hospital stroke (5 cohorts, 47,131 participants; RR, 2.2 [95% CI, 1.1-4.5]; very high heterogeneity, I2 5 86%; Fig 4). Results from 2 of 5 cohorts were adjusted for covariates. When I2 was reduced to zero by applying a credibility ceiling of 8.3%, increased risk of stroke persisted (RR, 1.3; 95% CI, 1.1-1.7). One study reported on the association between RRT and stroke (adjusted RR, 4.6; 95% CI, 1.5-13.6).18 Other Outcomes Three studies reported an association between CPB-associated AKI and acute myocardial infarction (RRs of 1.7 [95% CI, 0.5-5.5],19 4.8 [95% CI, 1.515.7],17 and 1.6 [95% CI, 1.4-1.7]).20 One study each reported an unadjusted association of AKI or RRT with cardiac tamponade (RR, 2.1; 95% CI, 1.6-2.8) post-AKI,17 cardiac morbidity (RR, 1.7; 95% CI, 0.93.2) post-RRT,21 all-cause hospitalization (RR, 10.4; 95% CI, 4.8-22.8) post-AKI,22 and heart failure (RR, 1.8; 95% CI,1.6-2.0) post-AKI.23 No studies reported associations between CPB-associated AKI and CKD or end-stage renal disease outcomes. Investigation Into Sources of Heterogeneity and Metaregression Different definitions of AKI did not influence the association with mortality (P 5 0.9). Stratification of studies according to the definition of AKI showed no difference in RR of mortality when AKI was defined by AKIN, RIFLE, or study investigators (Fig 2). Use of plasma creatinine level, eGFR, urine output, or the Am J Kidney Dis. 2015;65(2):283-293

combination of any of them to diagnose AKI did not modify associations (P 5 0.8). The high heterogeneity among studies precluded meaningful assessment for small-study effects (publication bias). No demographic feature, clinical characteristic, or study-level variable modified the risk of early mortality (Table S3).

DISCUSSION This meta-analysis including 242,000 patients at risk of AKI after CPB highlights the prognostic importance of AKI. Adults experiencing CPBassociated AKI experienced markedly increased premature mortality following surgery. Importantly, differing consortia- or guideline-based definitions for AKI provided similar prognostic information about CPB-associated AKI for early death in this clinical setting. Markedly increased risks of death with CPBassociated AKI persisted even when inconsistencies among contributing studies were accounted for, and there was a clear dose-response relationship between severity of kidney injury and prognosis. Death was considerably more frequent in patients who required RRT after surgery. In addition to the prognostic importance for mortality, kidney injury also was increased consistently and markedly with long-term mortality and stroke. The reported prognosis of CPB-associated AKI appeared unrelated to demographic, clinical, or surgical features. To our knowledge, this is the first meta-analysis to look at the differing prognostic implications of AKI definitions in any setting. The first consensus definition (RIFLE) authors emphasized that a scheme should be sensitive and predictive of relevant clinical outcomes, including mortality.3 The RIFLE criteria defined AKI by a 25% reduction in GFR (this was equated with a 50% increase in creatinine level when it should have been a 33.3% reduction in GFR equated with a 50% increase in creatinine level24) or low urine output.3 The AKIN definition introduced in 20074 retained the urine output criterion of the RIFLE definition, dropped the change in GFR per se, and modified the creatinine definition such that a small absolute increase of 0.3 mg/dL also could diagnose AKI. The AKIN definition also reduced the time over which AKI was to be diagnosed from 7 days to 48 hours. The recent KDIGO definition combined RIFLE and AKIN by allowing the 289

Pickering, James, and Palmer

relative change of 50% in creatinine level to be over 7 days, whereas absolute change is restricted to 48 hours.5 None of the studies we report used the KDIGO definition. Even with one or more of these definitions in existence, many studies in this meta-analysis identified AKI according to their own definitions. Despite these differences, we did not observe differences in prognostic importance based on AKI definitions. This may be due to concealment of such differences by the heterogeneity of surrogate markers used to define AKI and in timing of ascertainment of AKI. We found limited descriptions of the timing over which AKI was assessed. The recent KDIGO definition provides clear guidelines for the timing of AKI ascertainment and severity staging, which should aid comparisons in future studies. Nonetheless, our systematic review demonstrates that the concerns over the heterogeneity of definitions used to define CPB-associated AKI in epidemiologic studies should not overshadow the message of its prognostic importance. Several cohort studies have compared the prognostic implications of AKIN and RIFLE definitions in other clinical settings. In an intensive care unit cohort, Joannidis et al25 demonstrated that RIFLE is a more robust predictor of 30-day mortality than AKIN. However, in a multicenter cohort comprising 57 intensive care units, Bagshaw et al26 found no prognostic difference between RIFLE and AKIN. Roy et al27 found that KDIGO, RIFLE, and AKIN did not differ in predicting 30-day mortality, RRT, or readmission with heart failure in a cohort with acute decompensated heart failure. Our meta-analytic findings support the conclusion that there is consistency of prognostic estimates across AKI definitions. None of the studies we found reported the postdischarge development of CKD. However, we note that in recent years, CPB-associated AKI has been associated with increased risk of CKD,28-30 highlighting that it may have long-term prognostic relevance. Related to this, we found that CPB-associated AKI also was associated with elevated risk of longterm mortality. Although we believe this is the most comprehensive meta-analysis of kidney injury prognosis after CPB using robust methods, our findings should be interpreted in light of their limitations. First, we identified a high degree of heterogeneity in prognostic estimates among studies. However, we accounted for this heterogeneity by applying the credibility ceiling that reduced the I2 metric to 0% to minimize the potential for spuriously precise estimates and produce robust minimum estimates of risk. The credibility ceiling is the assumed likelihood for each study that the effect is in a particular direction and not null or in the other direction.11 If we were to assume larger values for the credibility ceiling, our estimates of 290

relative risk would not be lowered because they are already maximally attenuated at I2 of 0%. However, relative risk estimates would be greater if we assumed an even smaller credibility ceiling. Although credibility ceilings are already low (,5%), it is possible that the estimates of risk are underestimated by this meta-analysis. Second, many studies did not report loss to follow-up; therefore, loss-to-follow-up bias is possible, particularly for long-term mortality. Third, the study did not set out to quantify all possible risk factors for AKI in this cohort. Therefore, there are some (eg, previous cardiac surgery, chronic obstructive pulmonary disease,31 surgical re-exploration, red blood cell transfusions and preoperative hemoglobin level,32 and race33) that do not appear in the analysis, yet have been shown in individual studies to influence outcomes. Fourth, not all studies analyzed each outcome, which raises the possibility of reporting bias if the studies not reporting the outcomes differ systematically from those that do. Fifth, insufficient data were available to stratify pooled rates of AKI by the absence or presence of surgical factors or comorbid conditions. Patient-level data would be necessary to answer such questions as whether different surgical factors, recent exposure to coronary angiography in unstable patients, or other factors contribute to different risks of mortality in urgent cases.34 Sixth, the number of studies examining nonmortality outcomes, such as stroke, was small. As a result, the 95% prediction interval (the range for risk in a new study) is large, meaning that although further studies are likely to reach similar conclusions about associations between CPB-associated AKI and mortality, they may change the pooled RRs for stroke considerably. If we take the lower bound of the 95% CI for the RR for early mortality after application of a credibility ceiling of 4.0% (ie, a conservative approach with lower-bound RR of 1.8) to be the true RR for CPB-associated AKI–associated early mortality and assume the AKI incidence of 22.9% and overall mortality of 2.4% from this early-mortality metaanalysis, elimination of CPB-associated AKI could prevent 37 deaths for every 10,000 operations involving CPB. This may be as high as 118 (assuming an RR of 5.2, upper bound of the CI from the metaanalysis prior to application of credibility ceiling). This meta-analysis has a number of practical implications. The robustness of the association between CPB-associated AKI and mortality suggests that further observational studies examining this association are not needed. Epidemiologic studies instead should move toward improving risk assessment tools that would enable patients and clinicians to better appreciate the risks of CPB-associated AKI. As a first step, we suggest that epidemiologic studies focus on preoperative risk factors for mortality and Am J Kidney Dis. 2015;65(2):283-293

Meta-analysis of AKI after CPB

CPB-associated AKI. Currently available risk scores, such as the Thakar score,31 appear to both be pragmatic and have reasonable discrimination for prediction of dialysis-requiring CPB-associated AKI, but only a moderate area under the curve for prediction of the condition.35 Kidney-specific injury biomarkers represent one tool that may enhance such models.36 A further implication of this meta-analysis is that the choice of AKI definition to apply clinically or in research studies largely becomes a pragmatic one. For example, if the expected hospital stay is fewer than 7 days, the AKIN definition may be the most appropriate. We found no difference among cohorts that included urine output in the definition and those that did not. When urine output collection and monitoring are likely to be invasive or expensive in terms of clinician time, daily monitoring of serum creatinine level may suffice. However, we note that urine output criteria, at least in the intensive care unit, are poorly prognostic of mortality and may yet be refined.37-39 In summary, this systematic review and metaanalysis demonstrates a strong and consistent association between CPB-associated AKI and early and long-term mortality and stroke. It highlights the prognostic importance of CPB-associated AKI regardless of the definition used and provides a synthesis of data underlying the clinical relevance of CPB-associated AKI that should motivate future experimental studies of interventions for this condition.

ACKNOWLEDGEMENTS We thank the multiple authors who responded to our requests for additional information. Support: None. Financial Disclosure: Dr Pickering has undertaken consultancy for AMPharma. The other authors declare that they have no relevant financial interests. Contributions: Research idea and study design: JWP, SCP, MJ; data acquisition: JWP; data analysis/interpretation: JWP, SCP, MJ; statistical analysis: JWP. Each author contributed important intellectual content during manuscript drafting or revision and accepts accountability for the overall work by ensuring that questions pertaining to the accuracy or integrity of any portion of the work are appropriately investigated and resolved. SCP takes responsibility that this study has been reported honestly, accurately, and transparently; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned have been explained.

SUPPLEMENTARY MATERIAL Table S1: Quality assessment. Table S2: Covariate adjustments. Table S3: Summary of meta-regression. Figure S1: Credibility ceiling plots of risk of early mortality with CBP-associated AKI and I2. Figure S2: Credibility ceiling plots of risk of early mortality following RRT and I2. Figure S3: Early mortality risk following CBP-associated AKI stages. Figure S4: Long-term mortality risk following CBP-associated AKI. Am J Kidney Dis. 2015;65(2):283-293

Figure S5: Long-term mortality risk following RRT. Item S1: MEDLINE and EMBASE search strategy. Note: The supplementary material accompanying this article (http://dx.doi.org/10.1053/j.ajkd.2014.09.008) is available at www.ajkd.org

REFERENCES 1. Uchino S, Kellum J, Bellomo R, et al. Acute renal failure in critically ill patients—a multinational, multicenter study. JAMA. 2005;294:813-818. 2. Susantitaphong P, Cruz DN, Cerda J, et al. World incidence of AKI: A meta-analysis. Clin J Am Soc Nephrol. 2013;8(9):14821493. 3. Bellomo R, Ronco C, Kellum JA, Mehta RL, Palevsky PM; Acute Dialysis Quality Initiative 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. Crit Care. 2004;8(4):R204-R212. 4. Mehta RL, Kellum JA, Shah SV, et al. Acute Kidney Injury Network: report of an initiative to improve outcomes in acute kidney injury. Crit Care. 2007;11(2):R31. 5. KDIGO. Clinical practice guideline for acute kidney injury section 2: AKI definition. Kidney Int Suppl. 2012;1:19-36. 6. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. Plos Med. 2009;6(7):e1000100. 7. Zhang J, Yu KF. What’s the relative risk? A method of correcting the odds ratio in cohort studies of common outcomes. JAMA. 1998;280(19):1690-1691. 8. Miettinen O. Estimability and estimation in case-referent studies. Am J Epidemiol. 1976;103(2):226-235. 9. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7(3):177-188. 10. Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003;327(7414): 557-560. 11. Salanti G, Ioannidis JPA. Synthesis of observational studies should consider credibility ceilings. J Clin Epidemiol. 2009;62(2): 115-122. 12. Riley RD, Higgins JPT, Deeks JJ. Interpretation of random effects meta-analyses. BMJ. 2011;342:d549. 13. World Health Organization. WHO definition of region groupings. Geneva, Switzerland: World Health Organization; 2014. 14. Vandenbroucke JP, Elm Von E, Altman DG, et al. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration. Plos Med. 2007;4(10):e297. 15. Levey A, Bosch J, Lewis J, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Ann Intern Med. 1999;130(6):461-470. 16. Cockcroft D, Gault M. Prediction of creatinine clearance from serum creatinine. Nephron. 1976;16(1):31-41. 17. Mehta RH, Honeycutt E, Patel UD, et al. Impact of recovery of renal function on long-term mortality after coronary artery bypass grafting. Am J Cardiol. 2010;106(12):1728-1734. 18. Toumpoulis IK, Anagnostopoulos CE, Chamogeorgakis TP, et al. Impact of early and delayed stroke on in-hospital and long-term mortality after isolated coronary artery bypass grafting. Am J Cardiol. 2008;102(4):411-417. 19. Loef BG, Epema AH, Smilde TD, et al. Immediate postoperative renal function deterioration in cardiac surgical patients 291

Pickering, James, and Palmer predicts in-hospital mortality and long-term survival. J Am Soc Nephrol. 2005;16(1):195-200. 20. Rydén L, Ahnve S, Bell M, et al. Acute kidney injury after coronary artery bypass grafting and long-term risk of myocardial infarction and death. Int J Cardiol. 2014;172(1):190-195. 21. Boyle JM, Moualla S, Arrigain S, et al. Risks and outcomes of acute kidney injury requiring dialysis after cardiac transplantation. Am J Kidney Dis. 2006;48(5):787-796. 22. Habib RH, Zacharias A, Schwann TA, et al. Role of hemodilutional anemia and transfusion during cardiopulmonary bypass in renal injury after coronary revascularization: implications on operative outcome. Crit Care Med. 2005;33(8):17491756. 23. Olsson D, Sartipy U, Braunschweig F, Holzmann MJ. Acute kidney injury following coronary artery bypass surgery and long-term risk of heart failure. Circ Heart Fail. 2013;6(1):83-90. 24. Pickering JW, Endre ZH. GFR shot by RIFLE: errors in staging acute kidney injury. Lancet. 2009;373(9672):1318-1319. 25. Joannidis M, Metnitz B, Bauer P, et al. Acute kidney injury in critically ill patients classified by AKIN versus RIFLE using the SAPS 3 database. Intensive Care Med. 2009;35(10):1692-1702. 26. Bagshaw SM, George C, Bellomo R; ANZICS Database Management Committee. A comparison of the RIFLE and AKIN criteria for acute kidney injury in critically ill patients. Nephrol Dial Transplant. 2008;23(5):1569-1574. 27. Roy AK, Mc Gorrian C, Treacy C, et al. A comparison of traditional and novel definitions (RIFLE, AKIN, and KDIGO) of acute kidney injury for the prediction of outcomes in acute decompensated heart failure. Cardiorenal Med. 2013;3(1):26-37. 28. Coca SG, Yusuf B, Shlipak MG, Garg AX, Parikh CR. Long-term risk of mortality and other adverse outcomes after acute kidney injury: a systematic review and meta-analysis. Am J Kidney Dis. 2009;53(6):961-973. 29. Leung KCW, Tonelli M, James MT. Chronic kidney disease following acute kidney injury—risk and outcomes. Nat Rev Nephrol. 2012;9(2):77-85. 30. Ishani A, Nelson D, Clothier B, et al. The magnitude of acute serum creatinine increase after cardiac surgery and the risk of chronic kidney disease, progression of kidney disease, and death. Arch Intern Med. 2011;171(3):226-233. 31. Thakar C, Arrigain S, Worley S, Yared J, Paganini EP. A clinical score to predict acute renal failure after cardiac surgery. J Am Soc Nephrol. 2005;16(1):162-168. 32. Karkouti K, Wijeysundera DN, Yau TM, et al. Acute kidney injury after cardiac surgery: focus on modifiable risk factors. Circulation. 2009;119(4):495-502. 33. Zeng X, McMahon GM, Brunelli SM, Bates DW, Waikar SS. Incidence, outcomes, and comparisons across definitions of AKI in hospitalized individuals. Clin J Am Soc Nephrol. 2014;9(1):12-20. 34. Mehta RH, Honeycutt E, Shaw LK, et al. Clinical and angiographic correlates of short- and long-term mortality in patients undergoing coronary artery bypass grafting. Am J Cardiol. 2007;100(10):1538-1542. 35. Kiers HD, van den Boogaard M, Schoenmakers MCJ, et al. Comparison and clinical suitability of eight prediction models for cardiac surgery-related acute kidney injury. Nephrol Dial Transplant. 2013;28(2):345-351. 36. Zappitelli M. Preoperative prediction of acute kidney injury—from clinical scores to biomarkers. Pediatr Nephrol. 2013;28(8):1173-1182. 37. Mehta RL. Acute kidney injury: urine output in AKI—the canary in the coal mine? Nat Rev Nephrol. 2013;9(10):568-570.

292

38. Ralib AM, Pickering JW, Shaw GM, Endre ZH. The urine output definition of acute kidney injury is too liberal. Crit Care. 2013;17(3):R112. 39. Mcilroy DR, Argenziano M, Farkas D, Umann T, Sladen RN. Incorporating oliguria into the diagnostic criteria for acute kidney injury after on-pump cardiac surgery: impact on incidence and outcomes. J Cardiothorac Vasc Anesth. 2013;27(6): 1145-1152. 40. Ahmed WA, Tully PJ, Baker RA, Knight JL. Survival after isolated coronary artery bypass grafting in patients with severe left ventricular dysfunction. Ann Thorac Surg. 2009;87(4):1106-1112. 41. Arnaoutakis GJ, Bihorac A, Martin TD, et al. RIFLE criteria for acute kidney injury in aortic arch surgery. J Thorac Cardiovasc Surg. 2007;134(6):1554-1560; discussion 15601561. 42. Brown JR, Kramer RS, Coca SG, Parikh CR. Duration of acute kidney injury impacts long-term survival after cardiac surgery. Ann Thorac Surg. 2010;90(4):1142-1149. 43. Che M, Li Y, Liang X, et al. Prevalence of acute kidney injury following cardiac surgery and related risk factors in Chinese patients. Nephron Clin Pract. 2011;117(4):C305-C311. 44. Dardashti A, Ederoth P, Algotsson L, Bronden B, Bjursten H. Incidence, dynamics, and prognostic value of acute kidney injury for death after cardiac surgery. J Thorac Cardiovasc Surg. 2014;147(2):800-807. 45. Davoodi S, Hossein Ahmadi S, Omran A, Hesameddin Abbasi S. Early outcome of coronary artery bypass grafting in patients with severe left ventricular dysfunction. J Tehran Heart Cent. 2007;2(3):167-172. 46. De Santo L, Romano G, Corte Della A, et al. Preoperative anemia in patients undergoing coronary artery bypass grafting predicts acute kidney injury. J Thorac Cardiovasc Surg. 2009;138(4): 965-970. 47. De Santo LS, Romano G, Galdieri N, et al. RIFLE criteria for acute kidney injury in valvular surgery. J Heart Valve Dis. 2010;19(1):139-147. 48. Del Duca D, Iqbal S, Rahme E, Goldberg P, de Varennes B. Renal failure after cardiac surgery: timing of cardiac catheterization and other perioperative risk factors. Ann Thorac Surg. 2007;84(4):1264-1271. 49. Dell’Amore A, Aquino TM, Pagliaro M, Lamarra M, Zussa C. Aortic valve replacement with and without combined coronary bypass grafts in very elderly patients: early and long-term results. Eur J Cardiothorac Surg. 2012;41(3):491-498. 50. Faggian G, Milano AD, Santini F, et al. Urgent cardiac surgery in octogenarians. Eur Surg. 2011;43(2):90-95. 51. Gallagher S, Jones DA, Lovell MJ, et al. The impact of acute kidney injury on midterm outcomes after coronary artery bypass graft surgery: a matched propensity score analysis. J Thorac Cardiovasc Surg. 2014;147(3):989-995. 52. Haase M, Bellomo R, Story D, et al. Effect of mean arterial pressure, haemoglobin and blood transfusion during cardiopulmonary bypass on post-operative acute kidney injury. Nephrol Dial Transplant. 2012;27(1):153-160. 53. Hei F, Lou S, Li J, et al. Five-year results of 121 consecutive patients treated with extracorporeal membrane oxygenation at Fu Wai Hospital. Artif Organs. 2011;35(6):572-578. 54. Holzmann MJ, Rydén L, Sartipy U. Acute kidney injury and long-term risk of stroke after coronary artery bypass surgery. Int J Cardiol. 2013;168(6):5405-5410. 55. Ivert T, Holzmann MJ, Sartipy U. Survival in patients with acute kidney injury requiring dialysis after coronary artery bypass grafting. Eur J Cardiothorac Surg. 2014;45(2):312-317.

Am J Kidney Dis. 2015;65(2):283-293

Meta-analysis of AKI after CPB 56. Jeppsson A, Liden H, Johnsson P, Hartford M, Rådegran K. Surgical repair of post infarction ventricular septal defects: a national experience. Eur J Cardiothorac Surg. 2005;27(2):216-221. 57. Karimi A, Movahedi N, Salehiomran A, Marzban M, Hesameddin Abbasi S, Yazdanifard P. Mortality in open heart surgery with intraaortic balloon pump support. Asian Cardiovasc Thorac Ann. 2008;16(4):301-304. 58. Kohl BA, Hammond MS, Ochroch EA. Implementation of an intraoperative glycemic control protocol for cardiac surgery in a high-acuity academic medical center: an observational study. J Clin Anesth. 2013;25(2):121-128. 59. Li S-Y, Chen J-Y, Yang W-C, Chuang C-L. Acute Kidney Injury Network classification predicts in-hospital and long-term mortality in patients undergoing elective coronary artery bypass grafting surgery. Eur J Cardiothorac Surg. 2011;39(3):323-328. 60. Liotta M, Olsson D, Sartipy U, Holzmann MJ. Minimal changes in postoperative creatinine values and early and late mortality and cardiovascular events after coronary artery bypass grafting. Am J Cardiol. 2014;113(1):70-75. 61. Loef BG, Epema AH, Navis G, Ebels T, Stegeman CA. Postoperative renal dysfunction and preoperative left ventricular dysfunction predispose patients to increased long-term mortality after coronary artery bypass graft surgery. Br J Anaesth. 2009;102(6): 749-755. 62. Lopez-Delgado JC, Esteve F, Torrado H, et al. Influence of acute kidney injury on short- and long-term outcomes in patients undergoing cardiac surgery: risk factors and prognostic value of a modified RIFLE classification. Crit Care. 2013;17(6):R293. 63. Machado M de N, Miranda RC, Takakura IT, et al. Acute kidney injury after on-pump coronary artery bypass graft surgery. Arq Bras Cardiol. 2009;93(3):247-252. 64. Melby SJ, Zierer A, Kaiser SP, et al. Aortic valve replacement in octogenarians: risk factors for early and late mortality. Ann Thorac Surg. 2007;83(5):1651-1656; discussion 16561657. 65. Mitter N, Shah A, Yuh D, et al. Renal injury is associated with operative mortality after cardiac surgery for women and men. J Thorac Cardiovasc Surg. 2010;140(6):1367-1373. 66. Olsson C, Eriksson N, Ståhle E, Thelin S. Surgical and long-term mortality in 2634 consecutive patients operated on the proximal thoracic aorta. Eur J Cardiothorac Surg. 2007;31(6): 963-969.

Am J Kidney Dis. 2015;65(2):283-293

67. Rahmanian PB, Adams DH, Castillo JG, Carpentier A, Filsoufi F. Predicting hospital mortality and analysis of long-term survival after major noncardiac complications in cardiac surgery patients. Ann Thorac Surg. 2010;90(4):1221-1229. 68. Rahmanian PB, Kröner A, Langebartels G, Özel O, Wippermann J, Wahlers T. Impact of major non-cardiac complications on outcome following cardiac surgery procedures: logistic regression analysis in a very recent patient cohort. Interact Cardiovasc Thorac Surg. 2013;17(2):319-326; discussion 326-327. 69. Rydén L, Ahnve S, Bell M, Hammar N, Ivert T, Holzmann MJ. Acute kidney injury following coronary artery bypass grafting: early mortality and postoperative complications. Scand Cardiovasc J. 2012;46(2):114-120. 70. Saxena A, Dinh D, Smith JA, Shardey G, Reid CM, Newcomb AE. Sex differences in outcomes following isolated coronary artery bypass graft surgery in Australian patients: analysis of the Australasian Society of Cardiac and Thoracic Surgeons cardiac surgery database. Eur J Cardiothorac Surg. 2012;41(4): 755-762. 71. Shapira OM, Hunter CT, Anter E, et al. Coronary artery bypass grafting in patients with severe left ventricular dysfunction— early and mid-term outcomes. J Card Surg. 2006;21(3):225-232. 72. Tamayo E, Alvarez FJ, Martinez-Rafael B, et al. Ventilatorassociated pneumonia is an important risk factor for mortality after major cardiac surgery. J Crit Care. 2012;27(1):18-25. 73. Tekumit H, Uzun K, Cenal AR, Tataroglu C, Polat A, Akinci E. Determinants of mortality in patients requiring prolonged intensive care unit stay after elective isolated on-pump coronary artery bypass grafting surgery. Kardiol Pol. 2010;68(3): 257-262. 74. Toumpoulis IK, Chamogeorgakis TP, Angouras DC, Swistel DG, Anagnostopoulos CE, Rokkas CK. Independent predictors for early and long-term mortality after heart valve surgery. J Heart Valve Dis. 2008;17(5):548-556. 75. Urso S, Sadaba R, Greco E, et al. One-hundred aortic valve replacements in octogenarians: outcomes and risk factors for early mortality. J Heart Valve Dis. 2007;16(2):139-144. 76. Vargas Hein O, Birnbaum J, Wernecke KD, Konertz W, Jain U, Spies C. Three-year survival after four major post-cardiac operative complications. Crit Care Med. 2006;34(11):2729-2737. 77. Zingone B, Gatti G, Rauber E, et al. Early and late outcomes of cardiac surgery in octogenarians. Ann Thorac Surg. 2009;87(1):71-78.

293