- Email: [email protected]
Impact of Preoperative Renal Dysfunction on In-hospital Mortality After Solitary Valve and Combined Valve and Coronary Procedures
Claudius Diez, MD, Peter Mohr, MD, Oliver Kuss, PhD, Bernd Osten, MD, Rolf-Edgar Silber, MD, and Hans-Stefan Hofmann, MD Departments of Cardiothoracic Surgery and Thoracic Surgery, University Regensburg, Regensburg, Departments of Internal Medicine II and Cardiothoracic Surgery, Martin Luther University Halle-Wittenberg, Germany, and Institute of Medical Epidemiology, Biostatistics, and Informatics, Halle (Saale), Germany
Background. Limited information exists on the influence of preoperative renal dysfunction on in-hospital mortality after valve and combined valve and coronary procedures. The impact of preoperative renal dysfunction on patient outcome was investigated. Methods. This was a retrospective observational study of 916 patients who underwent solitary valve or combined procedures. Primary outcome was in-hospital mortality. Preoperative estimated glomerular filtration rate (eGFR) was calculated with the abbreviated Modification of Diet in Renal Disease formula. Results. Independent predictors of death were prolonged stay in the intensive care unit (odds ratio [OR], 1.03; 95% confidence interval [CI], 1.01 to 1.05), preoperative atrial fibrillation (OR, 1.61; 95% CI, 1.02 to 2.54), chronic obstructive pulmonary disease (OR, 2.2; 95% CI,
1.06 to 4.55), and prolonged operation time (OR, 1.01; 95% CI, 1.00 to 1.01). Each unit of the eGFR (mL/min/1.73m2) above average exerted a renoprotective effect (OR, 0.97; 95% CI, 0.96 to 0.98). The final regression model showed no lack of fit (Hosmer-Lemeshow test, p ⴝ 0.38) and a good discrimination performance in a receiver operating characteristic analysis (area under the curve, 0.84; 95% CI, 0.80 to 0.88). The lower the preoperative eGFR rate, the longer the postoperative stay at the intensive care unit. Conclusions. Renal dysfunction is an important independent predictor of in-hospital mortality in adult patients after valve and combined valve and coronary procedures.
P
placement and CABG plus mitral valve replacements had mortality rates of 7.1% and 13%, respectively. In this study, we evaluated the impact of preoperative renal dysfunction on in-hospital mortality after isolated valve and combined valve operations among our highrisk patients.
reoperative renal dysfunction, defined as reduced estimated glomerular filtration rate (eGFR), is one of the most important risk factors for perioperative morbidity and mortality after cardiac operations. Several studies addressed this issue in patients with coronary artery bypass grafting (CABG) [1–3], but only a few studies examined effects of preoperative impaired renal function in patients with valve and combined procedures [4 – 6]. Furthermore, some reports indicated that proper risk stratification for valve and combined operations heavily depends on the scoring system that is used to calculate perioperative mortality [7–9]. In Germany, mortality rates in 2006 after isolated valve operations and concomitant coronary revascularization procedures ranged from 4.2% for single-valve operations to 17% for aortic and mitral valve replacement with CABG [10]. Isolated aortic valve replacement was associated with a mortality rate of 3.9%, whereas isolated mitral valve replacement had a mortality rate of 5.3%. Combination procedures of CABG with aortic valve re-
Accepted for publication Nov 17, 2008. Address correspondence to Dr Diez, University Hospital Regensburg, Department of Cardiothoracic Surgery, Franz-Josef-Strauß-Allee 11, Regensburg, D-93053, Germany; e-mail: [email protected].
© 2009 by The Society of Thoracic Surgeons Published by Elsevier Inc
(Ann Thorac Surg 2009;87:731– 6) © 2009 by The Society of Thoracic Surgeons
Patients and Methods Patients and Study Design The initial study sample comprised 994 patients who underwent isolated valve operations or combined valve and coronary artery procedures between January 1999 and June 2003 at the University Hospital Halle. Institutional Review Board approval was obtained in March 2003. The requirement of individual patient consent was waived because of the study’s retrospective design and the data collection from routine care. We excluded 78 patients because 17 patients were aged younger than 20 years, 18 patients had terminal renal failure and dialysis, 32 had insufficient documentation, and 11 were without preoperative creatinine measurement. The final study sample comprised 916 patients. Patients who died during the index hospitalization were defined as nonsurvivors. 0003-4975/09/$36.00 doi:10.1016/j.athoracsur.2008.11.055
ADULT CARDIAC
Impact of Preoperative Renal Dysfunction on In-hospital Mortality After Solitary Valve and Combined Valve and Coronary Procedures
732 ADULT CARDIAC
DIEZ ET AL RENAL DYSFUNCTION ON IN-HOSPITAL DEATH
Ann Thorac Surg 2009;87:731– 6
Table 1. Characteristics of Study Sample Categorized by Primary Outcome Variable Patients, No. Female gender Hypertension Diabetes mellitus Hypercholesterolemia COPD Pre-op renal dysfunction Atrial fibrillation Ejection fraction, mean ⫾ SD Peripheral arterial vascular disease Acute MI (⬍90 days) Unstable angina pectoris NYHA IV Critical pre-op state Previous cardiac operation Active endocarditis Thoracic aortic operation Age ⬎65 years Emergency operation Combined valve/coronary operation
Total, No. (%)a
Nonsurvivors, No. (%)b
916 440 (48) 576 (62.9) 282 (30.8) 403 (44) 69 (7.5) 114 (12.4) 254 (27.7) 0.60 ⫾ 0.157 127 (13.7) 39 (4.3) 29 (3.2) 124 (13.6) 20 (2.2) 85 (9.3) 78 (8.5) 59 (6.4) 509 (55.5) 85 (9.3) 412 (45)
125 76 (17.3) 85 (14.8) 55 (19.5) 56 (13.9) 14 (20.3) 14 (12.3) 53 (20.9) 0.61 ⫾ 0.166 26 (20.5) 9 (23.1) 6 (20.7) 35 (28.2) 9 (45) 22 (25.9) 10 (12.8) 18 (30.5) 98 (19.3) 20 (23.5) 76 (18.4)
OR (95% CI)
p Valuec
1.82 (1.24–2.67) 1.30 (0.87–1.94) 1.95 (1.33–2.87) 1.04 (0.71–1.51) 1.69 (0.91–3.14) 0.87 (0.48–1.58) 2.16 (1.46–3.19)
0.0027 0.23 0.0008e 0.92 0.10 0.78 0.0001e 0.52 0.025e 0.09e 0.27 ⬍0.0001e 0.0006e 0.0014e 1.00 0.0005e 0.0001 0.012e 0.0001e
d
1.79 (1.11–2.90) 1.97 (0.91–4.25) 1.68 (0.67–4.22) 3.09 (1.97–4.84) 5.50 (2.23–13.56) 2.47 (1.46–4.18) 0.92 (0.46–1.85) 3.08 (1.71–5.55) 3.36 (2.14–5.25) 2.10 (1.22–3.61) 2.14 (1.45–3.16)
a Unless otherwise indicated, the numbers represent the number (%) of all patients with this potential prognostic factor among the study b c Unless otherwise indicated, the number represent the proportion (%) of deceased patients within a certain group. Calculated with participants. d e The mean difference was 1.0 with a 95% CI of –2.0 to 4.0. The indicated variable has been included in the multivariate the Fisher exact test. logistic regression analysis.
CI ⫽ confidence interval; COPD ⫽ chronic obstructive pulmonary disease; Association; OR ⫽ odds ratio.
Data Collection and Variables We retrospectively reviewed patient medical records and collected the data in an Excel spreadsheet (Microsoft Corp, Redmond, WA). The primary outcome variable was inhospital mortality, defined as death during the index hospitalization. A secondary outcome variable was the postoperative length of stay at the intensive care unit (ICU). Preoperative risk factors such as gender, age, diabetes mellitus, and chronic obstructive pulmonary disease were defined as in the European System for Cardiac Operative Risk Evaluation (EuroSCORE). Preoperative serum creatinine (SCr) was measured in mol/L at the day of hospital admission and then converted to mg/dL (1 mg/dL ⫽ 88.4 mol/L). The eGFR was calculated with the abbreviated Modification of Diet in Renal Disease (MDRD) formula and expressed in mL/ min/1.73m2: MDRD-eGFR ⫽ 186.3 ⫻ SCr⫺1.154 ⫻ age⫺0.203 ⫻ 0.742 (if female) ⫻ 1.212 (if black)
MI ⫽ myocardial infarction;
NYHA ⫽ New York Heart
ascending aorta and right atrium or bicaval cannulation. A standard circuit with a hollow-fiber membrane oxygenator and a roller pump was used. The body temperature was kept at 37°C in 85% of the patients. Otherwise, mild hypothermia (32°C) was induced. Myocardial protection was achieved by antegrade crystalloid (45%) or warm blood cardioplegia. No statistical significant difference was noted in the outcomes of those patients with hypothermia and blood cardioplegia. All operations were performed by senior cardiothoracic surgeons. Cardiac anesthesia was performed according to the institution’s guidelines.
Table 2. Operative Procedures Among the Study Sample of 916 Patients
Procedure
Valve Combined Valve and Procedures, Coronary Procedures, No. (%) No. (%)
Surgical Technique
Total procedures Aortic valve replacement Mitral valve replacementa Aortic/mitral valve replacement Other valve proceduresb
All operations were performed with cardiopulmonary bypass (CPB) that was instituted by cannulation of the
a There were no mitral valve reconstructions during the observational b Pulmonary, tricuspid, triple valve. period.
For further evaluation and based on the current clinical practice guidelines for chronic kidney disease (CKD), patients were grouped into CKD classes according to their eGFR [11].
594 307 (61) 115 (23) 69 (13.7)
412 278 (67.5) 97 (23.5) 37 (9)
13 (2.3)
...
DIEZ ET AL RENAL DYSFUNCTION ON IN-HOSPITAL DEATH
733
Table 3. Perioperative Data Among the Study Sample of 916 Patients Variable Operation time, min Bypass time, min Cross-clamp time, min Distal anastomosesc,d Ventilation time, h ICU stay, d Logistic EuroSCORE, % Post-op temp dialysis SCr, mol/L Estimated GFRe
Survivors (n ⫽ 791) Median (IQR); No. (%)
Nonsurvivors (n ⫽ 125) Median (IQR); No. (%)
Difference (95% CI)
p-Valuea
210 (175–255) 132 (106–171) 83 (65–107) 3 13 (9–19) 3 (2–6) 5 (2.9–9.2) 73 (9.2) 86 (75–100) 73 (60–86)
295 (225–360) 190 (138–240) 111 (83–141) 3 19 (9–29) 9 (3–18) 10 (5.5–17.5) 60 (48) 95 (79–126) 58 (46–72)
85 (62–108) 58 (43–73) 28 (20–36) 0 6 (3.7–8.9) 6 (3.5–8.5) 5 (2.3–7.5) 0.39 (0.30–0.47) 9 (2.4–15.6) 15 (9.6–18.5)
⬍0.0001b ⬍0.0001 ⬍0.0001 0.78 ⬍0.0001 ⬍0.0001b ⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001
a b c Calculated with the Mann-Whitney U test. The indicated variable has been included in the multivariate logistic regression analysis. Number d e No range is shown. Calculated by the Modification of Diet in of anastomoses in patients with concomitant coronary artery bypass grafting. 2 Renal Disease formula; mL/min/1.73m .
CI ⫽ confidence interval; EuroSCORE ⫽ European System for Cardiac Operative Risk Evaluation; intensive care unit; IQR ⫽ interquartile range; SCr, serum creatinine.
Statistics Data were analyzed with the statistical software SPSS 15.0 (SPSS Inc, Chicago, IL) and SAS 9.1 (SAS Institute, Cary, NC). Differences between normally distributed continuous data were analyzed with the t test and presented as mean differences with the 95% confidence interval (CI). The Mann-Whitney U test was used for nonnormally distributed data. The CIs for the difference of two medians were calculated as proposed by Bonett and Price in 2002 [12]. Overall differences between more than two groups were analyzed with analysis of variance (ANOVA) or, where appropriate, with the Kruskal-Wallis test. The Fisher exact test was used for categoric variables in a 2 ⫻ 2 table. Logistic regression analysis helped to examine the relationship between potential risk factors and inhospital mortality. A conditional forward method was used with inclusion of clinically relevant variables with p ⬍ 0.25 from the univariate analysis as indicated in Table 1. Variables were retained when p ⬍ 0.05. In a second analysis, a stepwise forward method (Wald) was used to calculate the odds ratios using the different chronic
GFR ⫽ glomerular filtration rate;
ICU ⫽
kidney disease classes. The logistic EuroSCORE was calculated with the downloadable Excel spreadsheet (www.euroscore.org) and expressed with the 95% CI as proposed by Kuss and Börgermann [13]. SigmaPlot 10.0 software (Systat Software GmbH, Erkrath, Germany) was used for calculating the area under the curve (AUC) for the receiver operating characteristics (ROC) analysis and for calculating the differences between ROC curves. The CIs for these differences were determined by bootstrap [14]. All figures were created with SigmaPlot 10.0. A value of p ⬍ 0.05 was considered statistically significant. Data are shown as the
Table 4. Logistic Regression Model to Predict the Outcome Status “Nonsurvivor” After Operation Variable MDRD eGFRa ICU stay, days Atrial fibrillation COPD Operation time, min
OR (95% CI)
p Value
0.97 (0.96–0.98) 1.03 (1.01–1.05) 1.61 (1.02–2.54) 2.20 (1.07–4.56) 1.01 (1.01–1.01)
⬍0.001 ⬍0.001 0.039 0.033 ⬍0.001
a Modification of Diet in Renal Disease (MDRD) formula for the estimated glomerular filtration rate (eGFR) was included because it is the most widely used and is recommended by the National Kidney Foundation Kidney Disease Outcomes Quality Initiative chronic kidney disease guidelines [11].
CI ⫽ confidence interval; COPD ⫽ chronic obstructive pulmonary disease; ICU ⫽ intensive care unit; OR ⫽ odds ratio.
Fig 1. Receiver operator characteristic curve for death as predicted by the final logistic regression model. The area under the curve (AUC; black line) is 0.84, which translates into good discrimination of the final logistic regression model.
ADULT CARDIAC
Ann Thorac Surg 2009;87:731– 6
734 ADULT CARDIAC
DIEZ ET AL RENAL DYSFUNCTION ON IN-HOSPITAL DEATH
Ann Thorac Surg 2009;87:731– 6
Table 5. Estimated Preoperative Glomerular Filtration Rate and Postoperative Intensive Care Unit Stay Postoperative ICU Stay, Days MDRD-eGFR (mL/min/1.73m2)
Patients, No.
Mean ⫾ SD
Median (IQR)
p Value
27 235 654
17.4 ⫾ 21.7 9.3 ⫾ 12.3 5.9 ⫾ 9.1
7 (4–22) 5 (2–11) 3 (2–6)
⬍0.05
⬍30 30–59 ⱖ60
ICU ⫽ intensive care unit; IQR ⫽ interquartile range; MDRDeGFR ⫽ Modification of Diet in Renal Disease estimated glomerular filtration rate; SD ⫽ standard deviation.
mean and standard deviation (SD) or, where appropriate and indicated, as the median and interquartile range (IQR).
Results Preoperative Data Of 916 patients, 504 underwent a solitary valve procedure and 412 had combined valve and coronary operations. Table 1 summarizes patient characteristics categorized by primary outcome after operation and odds ratios associated with poor outcome postoperatively. The median age difference between survivors (66 years) and nonsurvivors (71 years) was 5 years (95% CI, 3.90 to 6.10, p ⬍ 0.0001). Nonsurvivors had a slightly lower body mass index of 25.1 vs 26.5 kg/m2, with a median BMI difference of 1.4 kg/m2 (95% CI, 0.49 to 2.25 kg/m2; p ⫽ 0.009). Among several other variables, a critical preoperative state, operations on the thoracic aorta, New York Heart Association (NYHA) functional class IV, previous cardiac operation, age older than 65 years, and preoperative fibrillation were strongly associated with poor postoperative outcome. Female gender was also associated with poor outcome (Table 1).
Operative and Perioperative Data Table 2 summarizes operative procedures within each group. Almost 55% of all operations were solitary valve procedures, and 412 (45%) were combined valve and coronary operations. Perioperative data categorized by outcome are outlined in Table 3. Nonsurvivors had a significantly longer operation time, including longer CPB and aortic clamping times, a longer ICU stay, lower preoperative eGFR, and a higher risk of perioperative death as calculated by the logistic EuroSCORE.
Independent Risk Factors for In-Hospital Mortality The multivariate regression model was used to identify independent predictors for “nonsurvivor” outcome status after operation (Table 4). We identified four independent predictors of in-hospital mortality. An eGFR above average had a protective effect. Each unit of eGFR (mL/ min/1.72m2) reduced the odds for death by 3%. Each additional day at the ICU increased it by 3%. The length of operation time had a significant influence: every addi-
tional minute above the average increased the odds of death by 1.1%. Including aortic cross-clamping time or CPB time instead of operation time as a variable in the regression model did not lead to a higher odds ratio than the operation time itself. Interestingly, other significant variables from the univariate analysis, such as NYHA class IV or age older than 65 years, were found not to be included in our final model. The Hosmer-Lemeshow test was used to evaluate the goodness of fit, and the generated model showed a useful goodness of fit (final model: 2 ⫽ 8.60, p ⫽ 0.38). To evaluate the model’s discrimination performance between patients who survived and died, a ROC analysis was performed with the probabilities of death from the logistic regression analysis. The AUC was 0.84 (95% CI, 0.80 to 0.88; p ⬍ 0.0001), which translated into good discrimination of the final model (Fig 1). The primary reasons why patients did not survive the operation were severe low cardiac output in 34, postoperative myocardial infarction in 9, multisystem organ failure in 58, sepsis in 14, severe stroke in 4, severe pulmonary artery embolism in 3, and unknown cause in 3.
Preoperative eGFR and Postoperative ICU Stay Patients with severely impaired renal function (⬍30 mL/ min/1.73m2) experienced the longest ICU stay (Table 5). Patients with an eGFR of 60 mL/min/1.73m2 or higher had a median ICU stay of 3 days; however, there was a considerable interindividual variation.
Preoperative eGFR and Outcome Impaired preoperative renal function has been associated with increased mortality and risk of acute renal failure after cardiac procedures. Thus, we examined the association between the preoperative eGFR and primary postoperative outcome. The data are summarized in Table 6. The lower the eGFR values, the higher the actual and predicted mortalities. Because it is clinically important to detect patients with preoperative renal dysfunction, we assigned patients into three eGFR groups: (1) less than 30 mL/min/1.73m2, (2) 30 to 59 mL/min/1.73m2, and (3) 60 mL/min/1.73m2 or higher. For each group, we calculated odds ratios for in-hospital mortality in a logistic regression analysis, in which the group 60 mL/min/1.73m2 or higher repre-
Table 6. In-hospital Mortality, Preoperative Estimated Glomerular Filtration Rate, and Postoperative Dialysis MDRD-eGFR (mL/min/1.73m2) ⬍30 30–59 ⱖ60
Patients No.
Actual Mortality No. (%)
Post-op Temporary Dialysis No. (%)a
27 235 654
12 (44.4) 54 (23) 59 (9.0)
21 (77.7) 89 (37.8) 23 (3.5)
a Value represents number of patients with temporary dialysis/total number of patients in the indicated group.
MDRD-eGFR ⫽ Modification of Diet in Renal Disease estimated glomerular filtration rate.
sented the reference group with a normal eGFR. Our use of the MDRD formula resulted in a threefold increase (95% CI, 2.00 to 4.51) in risk of death among the 30 to 59 mL/min/1.73m2 eGFR group and an eightfold increase (95% CI, 3.60 to 18.0) for the less than 30 mL/min/1.73m2 eGFR group compared with the reference group.
Comment CKD is a worldwide major public health problem. In the United States, about 8% of the population has CKD, defined by an eGFR of less than 60 mL/min/1.73m2 [15]. CKD is explained by a higher prevalence of diabetes, hypertension, and higher body mass index. In addition, CKD remains an unrecognized condition in 80% to 90% of cases, and its presence is one of the most potent risk factors for cardiovascular disease [16]. Unfortunately, large population-based studies on renal dysfunction have not yet been conducted in Germany. Estimating the GFR instead of simply reporting the serum creatinine concentration has become the most widely used method for evaluating renal function today. The generation of creatinine is determined primarily by muscle mass and dietary intake, which accounts for the variations in the level of serum creatinine observed among different age, geographic, ethnic, and racial groups. Extrarenal elimination of creatinine may be increased at low levels of GFR. For these reasons, the relationship between the levels of serum creatinine and GFR varies substantially among persons and over time [17]. Several studies have shown that serum creatinine is an inadequate screening test for renal failure in elderly patients [18, 19]. This study adds to the evidence that preoperative renal dysfunction has an adverse effect on survival and the postoperative ICU stay of patients undergoing valve and combined valve and coronary operations. Although a number of studies have shown that preoperative renal dysfunction is a predictor of adverse outcome in patients undergoing CABG [20, 21], this study extends this finding to patients with solitary valve and combined valve and coronary procedures.
DIEZ ET AL RENAL DYSFUNCTION ON IN-HOSPITAL DEATH
735
regression model clearly indicated a renoprotective effect of an eGFR above average. Because patients in our study with an eGFR of less than 30 mL/min/1.73m2 had a mortality rate of up to 44%, one might question whether patients should undergo preoperative dialysis to reduce postoperative mortality. To our knowledge, no large study has examined the effect of a single preoperative dialysis in nondialysis-dependent patients on postoperative outcome; thus, one can only speculate about the effects. Although preoperative dialysis might lower increased serum creatinine levels to a normal range, an eGFR of less than 30 mL/min/1.73m2 usually reflects a long-term renal insufficiency and its sequelae on many organs. We doubt that dialysis would ameliorate the outcome. It is not clear how preoperative renal impairment contributes to postoperative mortality. Recent evidence, however, suggests that renal and cardiovascular conditions have been closely linked. An impaired preoperative eGFR may simply be a marker of a more advanced cardiovascular disease with increased levels of inflammatory mediators and hypercoagulability, endothelial dysfunction, arterial stiffness or calcification, and left ventricular hypertrophy [24 –28]. Another hypothesis considers renal dysfunction secondary to cardiac dysfunction. A decline in renal perfusion and activation of compensatory mechanisms such as the reninangiotensin-aldosterone-system occurs in patients with a reduced cardiac output.
Implications Our data suggest that cardiovascular surgeons should be very aware of subtle changes in the preoperative GFR as an independent risk factor for patients undergoing valve and combined procedures. The association between renal dysfunction and an increased incidence of postoperative complications such as prolonged ICU and hospital stay requires improved resource planning by those responsible for health care provision. A better understanding of mechanisms underlying progressive renal dysfunction and improved renal protection strategies during the operative period may ameliorate both in-hospital and late survival after cardiovascular operations.
Renal Dysfunction and In-Hospital Mortality
Strengths and Limitations
A persistent reduction in the eGFR to less than 60 mL/min/1.73m2 is defined as CKD [11], and despite an ongoing debate on a potential misclassification, large cohort studies showed that an eGFR of less than 60 mL/min/1.73m2 is associated with an increased risk of adverse outcomes of CKD [22]. Grouping patients into several CKD classes according to their estimated GFR provides clinicians with a straightforward classification of renal impairment. Our data show that the lower the eGFR, the higher the mortality and postoperative ICU stay. Our results are in line with two recently published studies [6, 23] that demonstrated a strong association between preoperative renal impairment and early and late mortality. In addition, the results of our logistic
This was a large contemporary German single-center analysis of preoperative renal dysfunction in patients undergoing valve and combined procedures. Data provided were directly collected from medical records, and the database is 98% complete for all analyzed fields. The database was revalidated by another coauthor before analysis. These procedures should have reduced errors, but did not eliminate them completely. This study has limitations. First, conclusions from a retrospective observational study are necessarily limited in their application. Second, we cannot provide longterm survival data because it is difficult to retrieve postoperative survival data in Germany. Third, observational bias, particularly for outcomes defined by clinical interventions, which are dependent on various treatment
ADULT CARDIAC
Ann Thorac Surg 2009;87:731– 6
736 ADULT CARDIAC
DIEZ ET AL RENAL DYSFUNCTION ON IN-HOSPITAL DEATH
thresholds used by different clinicians, are a common error found in database derived studies. Finally, our study is purely exploratory, especially by using automatic selection procedures, which are known to generally overestimate the influences of prognostic factors [29]. As such, we would like to see replications of our analyses in independent samples. We would like to thank the secretaries of the Department of Cardiothoracic Surgery at the University hospital Halle and all members of the Medical Record Archive for collecting and providing the files.
References 1. Cooper WA, O’Brien SM, Thourani VH, et al. Impact of renal dysfunction on outcomes of coronary artery bypass surgery: results from the Society of Thoracic Surgeons National Adult Cardiac Database. Circulation 2006;113:1063–70. 2. Hillis GS, Croal BL, Buchan KG, et al. Renal function and outcome from coronary artery bypass grafting: impact on mortality after a 2.3-year follow-up. Circulation 2006;113: 1056 – 62. 3. Devbhandari MP, Duncan AJ, Grayson AD, et al. Effect of risk-adjusted, non-dialysis-dependent renal dysfunction on mortality and morbidity following coronary artery bypass surgery: a multi-centre study. Eur J Cardiothorac Surg 2006;29:964 –70. 4. Ibanez J, Riera M, Saez de Ibarra JI, et al. Effect of preoperative mild renal dysfunction on mortality and morbidity following valve cardiac surgery. Interact Cardiovasc Thorac Surg 2007;6:748 –52. 5. Anderson RJ, O’Brien M, MaWhinney S, et al. Mild renal failure is associated with adverse outcome after cardiac valve surgery. Am J Kidney Dis 2000;35:1127–34. 6. Howell NJ, Keogh BE, Bonser RS, et al. Mild renal dysfunction predicts in-hospital mortality and post-discharge survival following cardiac surgery. Eur J Cardiothorac Surg 2008;34:390 –5; discussion 395. 7. Karthik S, Srinivasan AK, Grayson AD, et al. Limitations of additive EuroSCORE for measuring risk stratified mortality in combined coronary and valve surgery. Eur J Cardiothorac Surg 2004;26:318 –22. 8. Dewey TM, Brown D, Ryan WH, Herbert MA, Prince SL, Mack MJ. Reliability of risk algorithms in predicting early and late operative outcomes in high-risk patients undergoing aortic valve replacement. J Thorac Cardiovasc Surg 2008;135:180 –7. 9. Kawachi Y, Nakashima A, Toshima Y, Arinaga K, Kawano H. Risk stratification analysis of operative mortality in heart and thoracic aorta surgery: comparison between Parsonnet and EuroSCORE additive model. Eur J Cardiothorac Surg 2001;20:961– 6. 10. Gummert JF, Funkat A, Beckmann A, et al. Cardiac surgery in Germany during 2006: a report on behalf of the German Society for Thoracic and Cardiovascular Surgery. Thorac Cardiovasc Surg 2007;55:343–50. 11. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 2002;39:S1–266.
Ann Thorac Surg 2009;87:731– 6
12. Bonett DG, Price RM. Statistical inference for a linear function of medians: confidence intervals, hypothesis testing, and sample size requirements. Psychol Methods 2002;7:370 – 83. 13. Kuss O, Borgermann J. Prognoses from the logistic EuroSCORE are statistical estimates that require confidence intervals. Eur J Cardiothorac Surg 2005;27:1129; author reply 1129 –32. 14. Grunkemeier GL, Wu Y. Bootstrap resampling methods: something for nothing? Ann Thorac Surg 2004;77:1142– 4. 15. Coresh J, Selvin E, Stevens LA, et al. Prevalence of chronic kidney disease in the United States. JAMA 2007;298:2038 – 47. 16. Go AS, Chertow GM, Fan D, McCulloch CE, Hsu CY. Chronic kidney disease and the risks of death, cardiovascular events, and hospitalization. N Engl J Med 2004;351:1296 – 305. 17. Stevens LA, Coresh J, Greene T, Levey AS. Assessing kidney function--measured and estimated glomerular filtration rate. N Engl J Med 2006;354:2473– 83. 18. Swedko PJ, Clark HD, Paramsothy K, Akbari A. Serum creatinine is an inadequate screening test for renal failure in elderly patients. Arch Intern Med 2003;163:356 – 60. 19. Coresh J, Astor BC, McQuillan G, et al. Calibration and random variation of the serum creatinine assay as critical elements of using equations to estimate glomerular filtration rate. Am J Kidney Dis 2002;39:920 –9. 20. van de Wal RM, van Brussel BL, Voors AA, et al. Mild preoperative renal dysfunction as a predictor of long-term clinical outcome after coronary bypass surgery. J Thorac Cardiovasc Surg 2005;129:330 –5. 21. Zakeri R, Freemantle N, Barnett V, et al. Relation between mild renal dysfunction and outcomes after coronary artery bypass grafting. Circulation 2005;112:I270 –5. 22. Weiner DE, Tighiouart H, Stark PC, et al. Kidney disease as a risk factor for recurrent cardiovascular disease and mortality. Am J Kidney Dis 2004;44:198 –206. 23. Filsoufi F, Rahmanian PB, Castillo JG, Chikwe J, Carpentier A, Adams DH. Early and late outcomes of cardiac surgery in patients with moderate to severe preoperative renal dysfunction without dialysis. Interact Cardiovasc Thorac Surg 2008;7:90 –5. 24. Shlipak MG, Fried LF, Crump C, et al. Elevations of inflammatory and procoagulant biomarkers in elderly persons with renal insufficiency. Circulation 2003;107:87–92. 25. Blacher J, Safar ME, Guerin AP, Pannier B, Marchais SJ, London GM. Aortic pulse wave velocity index and mortality in end-stage renal disease. Kidney Int 2003;63:1852– 60. 26. London GM, Guerin AP, Marchais SJ, Metivier F, Pannier B, Adda H. Arterial media calcification in end-stage renal disease: impact on all-cause and cardiovascular mortality. Nephrol Dial Transplant 2003;18:1731– 40. 27. Raggi P, Boulay A, Chasan-Taber S, et al. Cardiac calcification in adult hemodialysis patients. A link between endstage renal disease and cardiovascular disease? J Am Coll Cardiol 2002;39:695–701. 28. Levin A, Thompson CR, Ethier J, et al. Left ventricular mass index increase in early renal disease: impact of decline in hemoglobin. Am J Kidney Dis 1999;34:125–34. 29. Babyak MA. What you see may not be what you get: a brief, nontechnical introduction to overfitting in regression-type models. Psychosom Med 2004;66:411–21.
Background. Limited information exists on the influence of preoperative renal dysfunction on in-hospital mortality after valve and combined valve and coronary procedures. The impact of preoperative renal dysfunction on patient outcome was investigated. Methods. This was a retrospective observational study of 916 patients who underwent solitary valve or combined procedures. Primary outcome was in-hospital mortality. Preoperative estimated glomerular filtration rate (eGFR) was calculated with the abbreviated Modification of Diet in Renal Disease formula. Results. Independent predictors of death were prolonged stay in the intensive care unit (odds ratio [OR], 1.03; 95% confidence interval [CI], 1.01 to 1.05), preoperative atrial fibrillation (OR, 1.61; 95% CI, 1.02 to 2.54), chronic obstructive pulmonary disease (OR, 2.2; 95% CI,
1.06 to 4.55), and prolonged operation time (OR, 1.01; 95% CI, 1.00 to 1.01). Each unit of the eGFR (mL/min/1.73m2) above average exerted a renoprotective effect (OR, 0.97; 95% CI, 0.96 to 0.98). The final regression model showed no lack of fit (Hosmer-Lemeshow test, p ⴝ 0.38) and a good discrimination performance in a receiver operating characteristic analysis (area under the curve, 0.84; 95% CI, 0.80 to 0.88). The lower the preoperative eGFR rate, the longer the postoperative stay at the intensive care unit. Conclusions. Renal dysfunction is an important independent predictor of in-hospital mortality in adult patients after valve and combined valve and coronary procedures.
P
placement and CABG plus mitral valve replacements had mortality rates of 7.1% and 13%, respectively. In this study, we evaluated the impact of preoperative renal dysfunction on in-hospital mortality after isolated valve and combined valve operations among our highrisk patients.
reoperative renal dysfunction, defined as reduced estimated glomerular filtration rate (eGFR), is one of the most important risk factors for perioperative morbidity and mortality after cardiac operations. Several studies addressed this issue in patients with coronary artery bypass grafting (CABG) [1–3], but only a few studies examined effects of preoperative impaired renal function in patients with valve and combined procedures [4 – 6]. Furthermore, some reports indicated that proper risk stratification for valve and combined operations heavily depends on the scoring system that is used to calculate perioperative mortality [7–9]. In Germany, mortality rates in 2006 after isolated valve operations and concomitant coronary revascularization procedures ranged from 4.2% for single-valve operations to 17% for aortic and mitral valve replacement with CABG [10]. Isolated aortic valve replacement was associated with a mortality rate of 3.9%, whereas isolated mitral valve replacement had a mortality rate of 5.3%. Combination procedures of CABG with aortic valve re-
Accepted for publication Nov 17, 2008. Address correspondence to Dr Diez, University Hospital Regensburg, Department of Cardiothoracic Surgery, Franz-Josef-Strauß-Allee 11, Regensburg, D-93053, Germany; e-mail: [email protected].
© 2009 by The Society of Thoracic Surgeons Published by Elsevier Inc
(Ann Thorac Surg 2009;87:731– 6) © 2009 by The Society of Thoracic Surgeons
Patients and Methods Patients and Study Design The initial study sample comprised 994 patients who underwent isolated valve operations or combined valve and coronary artery procedures between January 1999 and June 2003 at the University Hospital Halle. Institutional Review Board approval was obtained in March 2003. The requirement of individual patient consent was waived because of the study’s retrospective design and the data collection from routine care. We excluded 78 patients because 17 patients were aged younger than 20 years, 18 patients had terminal renal failure and dialysis, 32 had insufficient documentation, and 11 were without preoperative creatinine measurement. The final study sample comprised 916 patients. Patients who died during the index hospitalization were defined as nonsurvivors. 0003-4975/09/$36.00 doi:10.1016/j.athoracsur.2008.11.055
ADULT CARDIAC
Impact of Preoperative Renal Dysfunction on In-hospital Mortality After Solitary Valve and Combined Valve and Coronary Procedures
732 ADULT CARDIAC
DIEZ ET AL RENAL DYSFUNCTION ON IN-HOSPITAL DEATH
Ann Thorac Surg 2009;87:731– 6
Table 1. Characteristics of Study Sample Categorized by Primary Outcome Variable Patients, No. Female gender Hypertension Diabetes mellitus Hypercholesterolemia COPD Pre-op renal dysfunction Atrial fibrillation Ejection fraction, mean ⫾ SD Peripheral arterial vascular disease Acute MI (⬍90 days) Unstable angina pectoris NYHA IV Critical pre-op state Previous cardiac operation Active endocarditis Thoracic aortic operation Age ⬎65 years Emergency operation Combined valve/coronary operation
Total, No. (%)a
Nonsurvivors, No. (%)b
916 440 (48) 576 (62.9) 282 (30.8) 403 (44) 69 (7.5) 114 (12.4) 254 (27.7) 0.60 ⫾ 0.157 127 (13.7) 39 (4.3) 29 (3.2) 124 (13.6) 20 (2.2) 85 (9.3) 78 (8.5) 59 (6.4) 509 (55.5) 85 (9.3) 412 (45)
125 76 (17.3) 85 (14.8) 55 (19.5) 56 (13.9) 14 (20.3) 14 (12.3) 53 (20.9) 0.61 ⫾ 0.166 26 (20.5) 9 (23.1) 6 (20.7) 35 (28.2) 9 (45) 22 (25.9) 10 (12.8) 18 (30.5) 98 (19.3) 20 (23.5) 76 (18.4)
OR (95% CI)
p Valuec
1.82 (1.24–2.67) 1.30 (0.87–1.94) 1.95 (1.33–2.87) 1.04 (0.71–1.51) 1.69 (0.91–3.14) 0.87 (0.48–1.58) 2.16 (1.46–3.19)
0.0027 0.23 0.0008e 0.92 0.10 0.78 0.0001e 0.52 0.025e 0.09e 0.27 ⬍0.0001e 0.0006e 0.0014e 1.00 0.0005e 0.0001 0.012e 0.0001e
d
1.79 (1.11–2.90) 1.97 (0.91–4.25) 1.68 (0.67–4.22) 3.09 (1.97–4.84) 5.50 (2.23–13.56) 2.47 (1.46–4.18) 0.92 (0.46–1.85) 3.08 (1.71–5.55) 3.36 (2.14–5.25) 2.10 (1.22–3.61) 2.14 (1.45–3.16)
a Unless otherwise indicated, the numbers represent the number (%) of all patients with this potential prognostic factor among the study b c Unless otherwise indicated, the number represent the proportion (%) of deceased patients within a certain group. Calculated with participants. d e The mean difference was 1.0 with a 95% CI of –2.0 to 4.0. The indicated variable has been included in the multivariate the Fisher exact test. logistic regression analysis.
CI ⫽ confidence interval; COPD ⫽ chronic obstructive pulmonary disease; Association; OR ⫽ odds ratio.
Data Collection and Variables We retrospectively reviewed patient medical records and collected the data in an Excel spreadsheet (Microsoft Corp, Redmond, WA). The primary outcome variable was inhospital mortality, defined as death during the index hospitalization. A secondary outcome variable was the postoperative length of stay at the intensive care unit (ICU). Preoperative risk factors such as gender, age, diabetes mellitus, and chronic obstructive pulmonary disease were defined as in the European System for Cardiac Operative Risk Evaluation (EuroSCORE). Preoperative serum creatinine (SCr) was measured in mol/L at the day of hospital admission and then converted to mg/dL (1 mg/dL ⫽ 88.4 mol/L). The eGFR was calculated with the abbreviated Modification of Diet in Renal Disease (MDRD) formula and expressed in mL/ min/1.73m2: MDRD-eGFR ⫽ 186.3 ⫻ SCr⫺1.154 ⫻ age⫺0.203 ⫻ 0.742 (if female) ⫻ 1.212 (if black)
MI ⫽ myocardial infarction;
NYHA ⫽ New York Heart
ascending aorta and right atrium or bicaval cannulation. A standard circuit with a hollow-fiber membrane oxygenator and a roller pump was used. The body temperature was kept at 37°C in 85% of the patients. Otherwise, mild hypothermia (32°C) was induced. Myocardial protection was achieved by antegrade crystalloid (45%) or warm blood cardioplegia. No statistical significant difference was noted in the outcomes of those patients with hypothermia and blood cardioplegia. All operations were performed by senior cardiothoracic surgeons. Cardiac anesthesia was performed according to the institution’s guidelines.
Table 2. Operative Procedures Among the Study Sample of 916 Patients
Procedure
Valve Combined Valve and Procedures, Coronary Procedures, No. (%) No. (%)
Surgical Technique
Total procedures Aortic valve replacement Mitral valve replacementa Aortic/mitral valve replacement Other valve proceduresb
All operations were performed with cardiopulmonary bypass (CPB) that was instituted by cannulation of the
a There were no mitral valve reconstructions during the observational b Pulmonary, tricuspid, triple valve. period.
For further evaluation and based on the current clinical practice guidelines for chronic kidney disease (CKD), patients were grouped into CKD classes according to their eGFR [11].
594 307 (61) 115 (23) 69 (13.7)
412 278 (67.5) 97 (23.5) 37 (9)
13 (2.3)
...
DIEZ ET AL RENAL DYSFUNCTION ON IN-HOSPITAL DEATH
733
Table 3. Perioperative Data Among the Study Sample of 916 Patients Variable Operation time, min Bypass time, min Cross-clamp time, min Distal anastomosesc,d Ventilation time, h ICU stay, d Logistic EuroSCORE, % Post-op temp dialysis SCr, mol/L Estimated GFRe
Survivors (n ⫽ 791) Median (IQR); No. (%)
Nonsurvivors (n ⫽ 125) Median (IQR); No. (%)
Difference (95% CI)
p-Valuea
210 (175–255) 132 (106–171) 83 (65–107) 3 13 (9–19) 3 (2–6) 5 (2.9–9.2) 73 (9.2) 86 (75–100) 73 (60–86)
295 (225–360) 190 (138–240) 111 (83–141) 3 19 (9–29) 9 (3–18) 10 (5.5–17.5) 60 (48) 95 (79–126) 58 (46–72)
85 (62–108) 58 (43–73) 28 (20–36) 0 6 (3.7–8.9) 6 (3.5–8.5) 5 (2.3–7.5) 0.39 (0.30–0.47) 9 (2.4–15.6) 15 (9.6–18.5)
⬍0.0001b ⬍0.0001 ⬍0.0001 0.78 ⬍0.0001 ⬍0.0001b ⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001
a b c Calculated with the Mann-Whitney U test. The indicated variable has been included in the multivariate logistic regression analysis. Number d e No range is shown. Calculated by the Modification of Diet in of anastomoses in patients with concomitant coronary artery bypass grafting. 2 Renal Disease formula; mL/min/1.73m .
CI ⫽ confidence interval; EuroSCORE ⫽ European System for Cardiac Operative Risk Evaluation; intensive care unit; IQR ⫽ interquartile range; SCr, serum creatinine.
Statistics Data were analyzed with the statistical software SPSS 15.0 (SPSS Inc, Chicago, IL) and SAS 9.1 (SAS Institute, Cary, NC). Differences between normally distributed continuous data were analyzed with the t test and presented as mean differences with the 95% confidence interval (CI). The Mann-Whitney U test was used for nonnormally distributed data. The CIs for the difference of two medians were calculated as proposed by Bonett and Price in 2002 [12]. Overall differences between more than two groups were analyzed with analysis of variance (ANOVA) or, where appropriate, with the Kruskal-Wallis test. The Fisher exact test was used for categoric variables in a 2 ⫻ 2 table. Logistic regression analysis helped to examine the relationship between potential risk factors and inhospital mortality. A conditional forward method was used with inclusion of clinically relevant variables with p ⬍ 0.25 from the univariate analysis as indicated in Table 1. Variables were retained when p ⬍ 0.05. In a second analysis, a stepwise forward method (Wald) was used to calculate the odds ratios using the different chronic
GFR ⫽ glomerular filtration rate;
ICU ⫽
kidney disease classes. The logistic EuroSCORE was calculated with the downloadable Excel spreadsheet (www.euroscore.org) and expressed with the 95% CI as proposed by Kuss and Börgermann [13]. SigmaPlot 10.0 software (Systat Software GmbH, Erkrath, Germany) was used for calculating the area under the curve (AUC) for the receiver operating characteristics (ROC) analysis and for calculating the differences between ROC curves. The CIs for these differences were determined by bootstrap [14]. All figures were created with SigmaPlot 10.0. A value of p ⬍ 0.05 was considered statistically significant. Data are shown as the
Table 4. Logistic Regression Model to Predict the Outcome Status “Nonsurvivor” After Operation Variable MDRD eGFRa ICU stay, days Atrial fibrillation COPD Operation time, min
OR (95% CI)
p Value
0.97 (0.96–0.98) 1.03 (1.01–1.05) 1.61 (1.02–2.54) 2.20 (1.07–4.56) 1.01 (1.01–1.01)
⬍0.001 ⬍0.001 0.039 0.033 ⬍0.001
a Modification of Diet in Renal Disease (MDRD) formula for the estimated glomerular filtration rate (eGFR) was included because it is the most widely used and is recommended by the National Kidney Foundation Kidney Disease Outcomes Quality Initiative chronic kidney disease guidelines [11].
CI ⫽ confidence interval; COPD ⫽ chronic obstructive pulmonary disease; ICU ⫽ intensive care unit; OR ⫽ odds ratio.
Fig 1. Receiver operator characteristic curve for death as predicted by the final logistic regression model. The area under the curve (AUC; black line) is 0.84, which translates into good discrimination of the final logistic regression model.
ADULT CARDIAC
Ann Thorac Surg 2009;87:731– 6
734 ADULT CARDIAC
DIEZ ET AL RENAL DYSFUNCTION ON IN-HOSPITAL DEATH
Ann Thorac Surg 2009;87:731– 6
Table 5. Estimated Preoperative Glomerular Filtration Rate and Postoperative Intensive Care Unit Stay Postoperative ICU Stay, Days MDRD-eGFR (mL/min/1.73m2)
Patients, No.
Mean ⫾ SD
Median (IQR)
p Value
27 235 654
17.4 ⫾ 21.7 9.3 ⫾ 12.3 5.9 ⫾ 9.1
7 (4–22) 5 (2–11) 3 (2–6)
⬍0.05
⬍30 30–59 ⱖ60
ICU ⫽ intensive care unit; IQR ⫽ interquartile range; MDRDeGFR ⫽ Modification of Diet in Renal Disease estimated glomerular filtration rate; SD ⫽ standard deviation.
mean and standard deviation (SD) or, where appropriate and indicated, as the median and interquartile range (IQR).
Results Preoperative Data Of 916 patients, 504 underwent a solitary valve procedure and 412 had combined valve and coronary operations. Table 1 summarizes patient characteristics categorized by primary outcome after operation and odds ratios associated with poor outcome postoperatively. The median age difference between survivors (66 years) and nonsurvivors (71 years) was 5 years (95% CI, 3.90 to 6.10, p ⬍ 0.0001). Nonsurvivors had a slightly lower body mass index of 25.1 vs 26.5 kg/m2, with a median BMI difference of 1.4 kg/m2 (95% CI, 0.49 to 2.25 kg/m2; p ⫽ 0.009). Among several other variables, a critical preoperative state, operations on the thoracic aorta, New York Heart Association (NYHA) functional class IV, previous cardiac operation, age older than 65 years, and preoperative fibrillation were strongly associated with poor postoperative outcome. Female gender was also associated with poor outcome (Table 1).
Operative and Perioperative Data Table 2 summarizes operative procedures within each group. Almost 55% of all operations were solitary valve procedures, and 412 (45%) were combined valve and coronary operations. Perioperative data categorized by outcome are outlined in Table 3. Nonsurvivors had a significantly longer operation time, including longer CPB and aortic clamping times, a longer ICU stay, lower preoperative eGFR, and a higher risk of perioperative death as calculated by the logistic EuroSCORE.
Independent Risk Factors for In-Hospital Mortality The multivariate regression model was used to identify independent predictors for “nonsurvivor” outcome status after operation (Table 4). We identified four independent predictors of in-hospital mortality. An eGFR above average had a protective effect. Each unit of eGFR (mL/ min/1.72m2) reduced the odds for death by 3%. Each additional day at the ICU increased it by 3%. The length of operation time had a significant influence: every addi-
tional minute above the average increased the odds of death by 1.1%. Including aortic cross-clamping time or CPB time instead of operation time as a variable in the regression model did not lead to a higher odds ratio than the operation time itself. Interestingly, other significant variables from the univariate analysis, such as NYHA class IV or age older than 65 years, were found not to be included in our final model. The Hosmer-Lemeshow test was used to evaluate the goodness of fit, and the generated model showed a useful goodness of fit (final model: 2 ⫽ 8.60, p ⫽ 0.38). To evaluate the model’s discrimination performance between patients who survived and died, a ROC analysis was performed with the probabilities of death from the logistic regression analysis. The AUC was 0.84 (95% CI, 0.80 to 0.88; p ⬍ 0.0001), which translated into good discrimination of the final model (Fig 1). The primary reasons why patients did not survive the operation were severe low cardiac output in 34, postoperative myocardial infarction in 9, multisystem organ failure in 58, sepsis in 14, severe stroke in 4, severe pulmonary artery embolism in 3, and unknown cause in 3.
Preoperative eGFR and Postoperative ICU Stay Patients with severely impaired renal function (⬍30 mL/ min/1.73m2) experienced the longest ICU stay (Table 5). Patients with an eGFR of 60 mL/min/1.73m2 or higher had a median ICU stay of 3 days; however, there was a considerable interindividual variation.
Preoperative eGFR and Outcome Impaired preoperative renal function has been associated with increased mortality and risk of acute renal failure after cardiac procedures. Thus, we examined the association between the preoperative eGFR and primary postoperative outcome. The data are summarized in Table 6. The lower the eGFR values, the higher the actual and predicted mortalities. Because it is clinically important to detect patients with preoperative renal dysfunction, we assigned patients into three eGFR groups: (1) less than 30 mL/min/1.73m2, (2) 30 to 59 mL/min/1.73m2, and (3) 60 mL/min/1.73m2 or higher. For each group, we calculated odds ratios for in-hospital mortality in a logistic regression analysis, in which the group 60 mL/min/1.73m2 or higher repre-
Table 6. In-hospital Mortality, Preoperative Estimated Glomerular Filtration Rate, and Postoperative Dialysis MDRD-eGFR (mL/min/1.73m2) ⬍30 30–59 ⱖ60
Patients No.
Actual Mortality No. (%)
Post-op Temporary Dialysis No. (%)a
27 235 654
12 (44.4) 54 (23) 59 (9.0)
21 (77.7) 89 (37.8) 23 (3.5)
a Value represents number of patients with temporary dialysis/total number of patients in the indicated group.
MDRD-eGFR ⫽ Modification of Diet in Renal Disease estimated glomerular filtration rate.
sented the reference group with a normal eGFR. Our use of the MDRD formula resulted in a threefold increase (95% CI, 2.00 to 4.51) in risk of death among the 30 to 59 mL/min/1.73m2 eGFR group and an eightfold increase (95% CI, 3.60 to 18.0) for the less than 30 mL/min/1.73m2 eGFR group compared with the reference group.
Comment CKD is a worldwide major public health problem. In the United States, about 8% of the population has CKD, defined by an eGFR of less than 60 mL/min/1.73m2 [15]. CKD is explained by a higher prevalence of diabetes, hypertension, and higher body mass index. In addition, CKD remains an unrecognized condition in 80% to 90% of cases, and its presence is one of the most potent risk factors for cardiovascular disease [16]. Unfortunately, large population-based studies on renal dysfunction have not yet been conducted in Germany. Estimating the GFR instead of simply reporting the serum creatinine concentration has become the most widely used method for evaluating renal function today. The generation of creatinine is determined primarily by muscle mass and dietary intake, which accounts for the variations in the level of serum creatinine observed among different age, geographic, ethnic, and racial groups. Extrarenal elimination of creatinine may be increased at low levels of GFR. For these reasons, the relationship between the levels of serum creatinine and GFR varies substantially among persons and over time [17]. Several studies have shown that serum creatinine is an inadequate screening test for renal failure in elderly patients [18, 19]. This study adds to the evidence that preoperative renal dysfunction has an adverse effect on survival and the postoperative ICU stay of patients undergoing valve and combined valve and coronary operations. Although a number of studies have shown that preoperative renal dysfunction is a predictor of adverse outcome in patients undergoing CABG [20, 21], this study extends this finding to patients with solitary valve and combined valve and coronary procedures.
DIEZ ET AL RENAL DYSFUNCTION ON IN-HOSPITAL DEATH
735
regression model clearly indicated a renoprotective effect of an eGFR above average. Because patients in our study with an eGFR of less than 30 mL/min/1.73m2 had a mortality rate of up to 44%, one might question whether patients should undergo preoperative dialysis to reduce postoperative mortality. To our knowledge, no large study has examined the effect of a single preoperative dialysis in nondialysis-dependent patients on postoperative outcome; thus, one can only speculate about the effects. Although preoperative dialysis might lower increased serum creatinine levels to a normal range, an eGFR of less than 30 mL/min/1.73m2 usually reflects a long-term renal insufficiency and its sequelae on many organs. We doubt that dialysis would ameliorate the outcome. It is not clear how preoperative renal impairment contributes to postoperative mortality. Recent evidence, however, suggests that renal and cardiovascular conditions have been closely linked. An impaired preoperative eGFR may simply be a marker of a more advanced cardiovascular disease with increased levels of inflammatory mediators and hypercoagulability, endothelial dysfunction, arterial stiffness or calcification, and left ventricular hypertrophy [24 –28]. Another hypothesis considers renal dysfunction secondary to cardiac dysfunction. A decline in renal perfusion and activation of compensatory mechanisms such as the reninangiotensin-aldosterone-system occurs in patients with a reduced cardiac output.
Implications Our data suggest that cardiovascular surgeons should be very aware of subtle changes in the preoperative GFR as an independent risk factor for patients undergoing valve and combined procedures. The association between renal dysfunction and an increased incidence of postoperative complications such as prolonged ICU and hospital stay requires improved resource planning by those responsible for health care provision. A better understanding of mechanisms underlying progressive renal dysfunction and improved renal protection strategies during the operative period may ameliorate both in-hospital and late survival after cardiovascular operations.
Renal Dysfunction and In-Hospital Mortality
Strengths and Limitations
A persistent reduction in the eGFR to less than 60 mL/min/1.73m2 is defined as CKD [11], and despite an ongoing debate on a potential misclassification, large cohort studies showed that an eGFR of less than 60 mL/min/1.73m2 is associated with an increased risk of adverse outcomes of CKD [22]. Grouping patients into several CKD classes according to their estimated GFR provides clinicians with a straightforward classification of renal impairment. Our data show that the lower the eGFR, the higher the mortality and postoperative ICU stay. Our results are in line with two recently published studies [6, 23] that demonstrated a strong association between preoperative renal impairment and early and late mortality. In addition, the results of our logistic
This was a large contemporary German single-center analysis of preoperative renal dysfunction in patients undergoing valve and combined procedures. Data provided were directly collected from medical records, and the database is 98% complete for all analyzed fields. The database was revalidated by another coauthor before analysis. These procedures should have reduced errors, but did not eliminate them completely. This study has limitations. First, conclusions from a retrospective observational study are necessarily limited in their application. Second, we cannot provide longterm survival data because it is difficult to retrieve postoperative survival data in Germany. Third, observational bias, particularly for outcomes defined by clinical interventions, which are dependent on various treatment
ADULT CARDIAC
Ann Thorac Surg 2009;87:731– 6
736 ADULT CARDIAC
DIEZ ET AL RENAL DYSFUNCTION ON IN-HOSPITAL DEATH
thresholds used by different clinicians, are a common error found in database derived studies. Finally, our study is purely exploratory, especially by using automatic selection procedures, which are known to generally overestimate the influences of prognostic factors [29]. As such, we would like to see replications of our analyses in independent samples. We would like to thank the secretaries of the Department of Cardiothoracic Surgery at the University hospital Halle and all members of the Medical Record Archive for collecting and providing the files.
References 1. Cooper WA, O’Brien SM, Thourani VH, et al. Impact of renal dysfunction on outcomes of coronary artery bypass surgery: results from the Society of Thoracic Surgeons National Adult Cardiac Database. Circulation 2006;113:1063–70. 2. Hillis GS, Croal BL, Buchan KG, et al. Renal function and outcome from coronary artery bypass grafting: impact on mortality after a 2.3-year follow-up. Circulation 2006;113: 1056 – 62. 3. Devbhandari MP, Duncan AJ, Grayson AD, et al. Effect of risk-adjusted, non-dialysis-dependent renal dysfunction on mortality and morbidity following coronary artery bypass surgery: a multi-centre study. Eur J Cardiothorac Surg 2006;29:964 –70. 4. Ibanez J, Riera M, Saez de Ibarra JI, et al. Effect of preoperative mild renal dysfunction on mortality and morbidity following valve cardiac surgery. Interact Cardiovasc Thorac Surg 2007;6:748 –52. 5. Anderson RJ, O’Brien M, MaWhinney S, et al. Mild renal failure is associated with adverse outcome after cardiac valve surgery. Am J Kidney Dis 2000;35:1127–34. 6. Howell NJ, Keogh BE, Bonser RS, et al. Mild renal dysfunction predicts in-hospital mortality and post-discharge survival following cardiac surgery. Eur J Cardiothorac Surg 2008;34:390 –5; discussion 395. 7. Karthik S, Srinivasan AK, Grayson AD, et al. Limitations of additive EuroSCORE for measuring risk stratified mortality in combined coronary and valve surgery. Eur J Cardiothorac Surg 2004;26:318 –22. 8. Dewey TM, Brown D, Ryan WH, Herbert MA, Prince SL, Mack MJ. Reliability of risk algorithms in predicting early and late operative outcomes in high-risk patients undergoing aortic valve replacement. J Thorac Cardiovasc Surg 2008;135:180 –7. 9. Kawachi Y, Nakashima A, Toshima Y, Arinaga K, Kawano H. Risk stratification analysis of operative mortality in heart and thoracic aorta surgery: comparison between Parsonnet and EuroSCORE additive model. Eur J Cardiothorac Surg 2001;20:961– 6. 10. Gummert JF, Funkat A, Beckmann A, et al. Cardiac surgery in Germany during 2006: a report on behalf of the German Society for Thoracic and Cardiovascular Surgery. Thorac Cardiovasc Surg 2007;55:343–50. 11. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 2002;39:S1–266.
Ann Thorac Surg 2009;87:731– 6
12. Bonett DG, Price RM. Statistical inference for a linear function of medians: confidence intervals, hypothesis testing, and sample size requirements. Psychol Methods 2002;7:370 – 83. 13. Kuss O, Borgermann J. Prognoses from the logistic EuroSCORE are statistical estimates that require confidence intervals. Eur J Cardiothorac Surg 2005;27:1129; author reply 1129 –32. 14. Grunkemeier GL, Wu Y. Bootstrap resampling methods: something for nothing? Ann Thorac Surg 2004;77:1142– 4. 15. Coresh J, Selvin E, Stevens LA, et al. Prevalence of chronic kidney disease in the United States. JAMA 2007;298:2038 – 47. 16. Go AS, Chertow GM, Fan D, McCulloch CE, Hsu CY. Chronic kidney disease and the risks of death, cardiovascular events, and hospitalization. N Engl J Med 2004;351:1296 – 305. 17. Stevens LA, Coresh J, Greene T, Levey AS. Assessing kidney function--measured and estimated glomerular filtration rate. N Engl J Med 2006;354:2473– 83. 18. Swedko PJ, Clark HD, Paramsothy K, Akbari A. Serum creatinine is an inadequate screening test for renal failure in elderly patients. Arch Intern Med 2003;163:356 – 60. 19. Coresh J, Astor BC, McQuillan G, et al. Calibration and random variation of the serum creatinine assay as critical elements of using equations to estimate glomerular filtration rate. Am J Kidney Dis 2002;39:920 –9. 20. van de Wal RM, van Brussel BL, Voors AA, et al. Mild preoperative renal dysfunction as a predictor of long-term clinical outcome after coronary bypass surgery. J Thorac Cardiovasc Surg 2005;129:330 –5. 21. Zakeri R, Freemantle N, Barnett V, et al. Relation between mild renal dysfunction and outcomes after coronary artery bypass grafting. Circulation 2005;112:I270 –5. 22. Weiner DE, Tighiouart H, Stark PC, et al. Kidney disease as a risk factor for recurrent cardiovascular disease and mortality. Am J Kidney Dis 2004;44:198 –206. 23. Filsoufi F, Rahmanian PB, Castillo JG, Chikwe J, Carpentier A, Adams DH. Early and late outcomes of cardiac surgery in patients with moderate to severe preoperative renal dysfunction without dialysis. Interact Cardiovasc Thorac Surg 2008;7:90 –5. 24. Shlipak MG, Fried LF, Crump C, et al. Elevations of inflammatory and procoagulant biomarkers in elderly persons with renal insufficiency. Circulation 2003;107:87–92. 25. Blacher J, Safar ME, Guerin AP, Pannier B, Marchais SJ, London GM. Aortic pulse wave velocity index and mortality in end-stage renal disease. Kidney Int 2003;63:1852– 60. 26. London GM, Guerin AP, Marchais SJ, Metivier F, Pannier B, Adda H. Arterial media calcification in end-stage renal disease: impact on all-cause and cardiovascular mortality. Nephrol Dial Transplant 2003;18:1731– 40. 27. Raggi P, Boulay A, Chasan-Taber S, et al. Cardiac calcification in adult hemodialysis patients. A link between endstage renal disease and cardiovascular disease? J Am Coll Cardiol 2002;39:695–701. 28. Levin A, Thompson CR, Ethier J, et al. Left ventricular mass index increase in early renal disease: impact of decline in hemoglobin. Am J Kidney Dis 1999;34:125–34. 29. Babyak MA. What you see may not be what you get: a brief, nontechnical introduction to overfitting in regression-type models. Psychosom Med 2004;66:411–21.