Risks of coronary artery bypass surgery in dialysis-dependent patients analysis of the 2001 National Inpatient Sample

Transcription

1 NDT Advance Access published February 13, 2007 Nephrol Dial Transplant (2007) 1 of 7 doi: /ndt/gfl835 Original Article Risks of coronary artery bypass surgery in dialysis-dependent patients analysis of the 2001 National Inpatient Sample David M. Charytan 1,2 and Richard E. Kuntz 2,3 1 Renal Division and 2 Division of Clinical Biometrics, Department of Medicine, Brigham and Women s Hospital, Boston, MA and 3 Medtronic Inc., Minneapolis, MN, USA Abstract Background. Dialysis patients have a high risk of cardiovascular death but may under-use coronary artery bypass grafting (CABG) because of the risk of peri-operative death. Whether operative mortality in dialysis patients has declined with contemporary techniques is uncertain. We undertook this study in order to compare peri-operative mortality in chronic dialysis (CD) and non-dialysis patients following CABG and to determine whether high levels of comorbidity in CD patients account for identified differences in operative risk. Methods. This study is a retrospective analysis of the 2001 National Inpatient Sample, a stratified probability sample of over seven million admissions in 33 states. Administrative data and ICD-9CM codes were used to identify dialysis patients, comorbidities, procedures and operative outcomes. Multivariable logistic regression was used to adjust for confounding. Results. In this study, non-dialysis patients and 635 dialysis patients underwent CABG. In-hospital death occurred in 11.1% of dialysis patients compared to 3.4% of non-dialysis patients. Rates of stroke, sepsis and pneumonia were also increased in dialysis patients. After adjustment for other surgical risk factors, the odds of in-hospital death were 3.38 ( , P < 0.001) times higher in dialysis than non-dialysis patients. Conclusions. Operative mortality in dialysis patients remains high despite recent advances in CABG surgery and is not explained by the high rates of comorbidity in dialysis patients. Because there is a very high risk of cardiovascular death without intervention, CABG may nevertheless be a life-saving therapy in CD patients. Randomized trials are needed to better define the optimal role of CABG in dialysis patients. Correspondence and offprint requests to: David Charytan, MD, MSc., Division of Clinical Biometrics, Brigham and Women s Hospital, 1 Brigham Circle, 3rd Floor, Boston, MA 02115, USA. dcharytan@partners.org Keywords: chronic dialysis; coronary artery bypass grafting; coronary artery disease; end-stage renal disease; mortality risk Introduction Coronary artery bypass grafting (CABG) reduces mortality in subjects with obstructive coronary artery disease and its increased use might reduce mortality rates in dialysis patients. In the general population, CABG is associated with an absolute reduction in 5-year mortality of 5.6% compared with medical therapy alone [1], and for patients with greater than average risks of cardiac death, such as those with three vessel disease or left main disease, the benefits of CABG are even greater [1 3]. Chronic dialysis (CD) patients nevertheless undergo coronary revascularization infrequently as few as 12%, undergo a percutaneous or surgical intervention following a myocardial infarction (MI) [4,5]. Whether this low use of CABG in this high-risk population stems from renalism [6], i.e. therapeutic nihilism towards patients on CD, or stems from appropriate scepticism about the merits of CABG in this population is unknown. In fact, scepticism about the benefits of CABG in the CD population may be justified. Dialysis patients have routinely been excluded from trials demonstrating the benefits of CABG [7 9] and they appear to have a high peri-operative mortality rate after CABG surgery [10,11]. Moreover, contemporary advances in the medical and percutaneous treatment of coronary artery disease may have lessened differences in longterm outcomes of medical vs surgical therapies while further highlighting the high short-term risks of surgical therapy in CD. Conversely, recent improvements in post-cabg complication rates [12], if they extend to the CD population, may have magnified the relative benefits of surgical vs medical therapy of coronary artery disease. Thus, whether the risks of CABG in CD patients actually outweigh its longterm benefits in CD patients i.e. whether CABG is ß The Author [2007]. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved. For Permissions, please journals.permissions@oxfordjournals.org

2 2of7 under-used or appropriately withheld in CD patients depends on accurate estimation of the peri-operative mortality rates in CD patients and the degree to which any increase in the risk of peri-operative death is associated with dialysis status rather than concurrent comorbid conditions. We therefore examined outcomes among CABGs performed in 2001 in order to derive contemporary estimates of the crude and adjusted risks of peri-operative death following CABG in patients with and without dialysis-dependent chronic renal failure. Subjects and methods Study population We used the 2001 National Inpatient Sample (NIS), a stratified probability sample of yearly state in-patient databases that was collected as part of the Healthcare Cost and Utilization Project. The sampling strategy of the NIS is designed to mimic a random sample of 20% of all US hospital discharges in a given year. Each discharge record within the NIS is weighted according to the survey s sampling design and represents approximately five similar discharges in the US population during that calendar year. In 2001, administrative data of over 7.4 million discharges from 1000 hospitals in 33 states were collected. Demographic data, diagnoses, procedures performed, length of stay, costs, discharge disposition and hospital characteristics are recorded. Within this population, we identified actual admissions for CABG and determined the dialysis status in each case. All analyses were conducted within this cohort. Definitions Table 1. Baseline characteristics of the study population according to dialysis status Condition Dialysis (n ¼ 635) Diagnoses and procedures were identified from ICD-9 CM diagnostic codes. Patients on CD were identified if they had either peritoneal dialysis or haemodialysis (54.98 and 39.95) during their hospitalization but did not carry the diagnosis of acute renal failure (584). Use of ICD-9 codes has been previously demonstrated to have an >90% sensitivity and negative predictive value for the exclusion of acute renal failure [13]. Patients with a principal diagnosis of acute myocardial infarction were identified by Coronary angiography ( and ), percutaneous coronary intervention (PCI, ), valve surgery (35.95, , ) and CABG ( ) were identified from procedure codes. Off-pump CABG was identified when codes for extracorporeal circulation auxiliary to open heart surgery (39.61) were absent. Cardiac catheterization included both diagnostic angiography and PCI. Stroke (947.02, 434, 436, 437.1, 437.9), pneumonia ( ) and sepsis (041, 038) were identified as events that might complicate CABG. Baseline comorbidities were identified with minor modifications of the methods described by Elixhauser et al. [14], as described below. This method was developed and validated within a subset of the NIS, and in administrative data sets, it is superior to the Charlson comorobidity score when used to predict in-hospital mortality in patients with coronary artery disease [15]. In brief, baseline comorbidities (Table 1) were identified from the presence of ICD-9 codes, Non-dialysis (n ¼ ) D. M. Charytan and R. E. Kuntz P-value Demographics Age <0.001 Female 223 (35.2) (30.2) Black 216 (23.3) 2885 (4.9) <0.001 Uninsured 4 (0.62) 1730 (2.23) <0.001 Medical history Congestive heart failure 253 (39.9) (18.9) <0.001 Arrhythmia 190 (29.9) (30.6) Valvular heart disease 101 (16.0) 9096 (11.7) 0.01 Peripheral vascular disease 94 (14.9) 7049 (9.1) <0.001 Neurological disorders 17 (2.6) 1099 (1.4) Chronic pulmonary disease 117 (18.4) (18.2) Diabetes 367 (57.7) (30.1) <0.001 Liver disease 12 (1.9) 386 (0.5) Cancer 22 (3.4) 4890 (6.4) Obesity 21 (3.3) 5889 (7.6) <0.001 Weight loss 21 (3.2) 536 (0.7) Alcohol abuse 4 (0.62) 1223 (1.6) Other risk factors Valve surgery 62 (9.8) 7853 (10.2) High volume centre 1535 (47.7) 3637 (47.3) 0.970* Medium volume centre 968 (30.1) (29.9) Low volume centre 717 (22.5) (22.8) Admission for myocardial infarction 130 (20.4) (20.2) Coronary angiography during hospitalization 294 (46.1) (57.6) <0.001 *P for trend across categories of surgical volume.

3 Peri-CABG mortality in dialysis patients 3 of 7 which indicates that a specific comorbid condition was present during the admission and the absence of a DRG (diagnostic related group) code for that diagnosis. The absence of DRG coding is required because the use of a DRG suggests that a diagnosis was a cause of admission rather than a comorbid condition incidental to the reason for hospitalization. Valvular heart disease, arrhythmias and congestive heart failure might affect the risk of surgery regardless of whether they pre-date admission and these conditions were tabulated regardless of the DRG coding. Statistical analysis Univariate analyses were performed using Student s t-tests for continuous variables or 2 tests for binary variables. Relative risks were calculated with non-cd patients serving as the reference group. Statistical significance was defined as P<0.05. All analyses were performed using SAS-callable SUDAAN (RTI, Durham, N.C.) to account for the complex survey design and sampling weights. All numbers reported represent the numbers of actual discharges with a given characteristic observed in the dataset. Percentages, odds ratios and P-values were calculated using SUDAAN software to account for the sampling weights and survey design of the NIS and therefore differ slightly from values calculated without accounting for these crucial elements of the survey design [16]. Logistic regression was used to adjust for factors that might influence the use of coronary revascularization. Two logistic regression models were generated in order to explore the relationship between CD and peri-operative mortality. Model 1 was designed to include correlates of variables recognized as highly significant predictors of post- CABG mortality [17,18] and included age, gender, history of congestive heart failure, valvular heart disease, chronic lung disease, pericutaneous transluminal coronary angioplasty (PTCA) on current admission, history of arrhythmia, admission for myocardial infarction, diabetes, cerebrovascular disease, peripheral vascular disease, as well as the presence of CD, obesity or pathologic weight loss and concurrent valve surgery. Model 2 was further adjusted for race, liver disease, alcohol abuse, history of cancer, annual number of CABGs in the surgical centre and presence or absence of insurance coverage. These latter variables are considered to be moderate or level 2 predictors of peri-cabg mortality in standard guidelines on the use of CABG [17,18]. Dummy terms indicating missing values for race were generated and used in the regression models because race is not recorded by all states in the NIS. Goodness of fit was evaluated for both models using R 2 -values, 2 tests and likelihood ratio tests. Exclusion of the missing values had minimal effect on the results of these analyses (data not shown). Results Baseline characteristics Baseline characteristics of the study population are summarized in Table 1. Patients on CD were more likely to be black (23.3 vs 4.9%) or female (35.2 vs 30.2%) and were significantly younger than non-dialysis patients (62.9 vs 66.1 years at admission). As expected, the CD patients had a significantly greater degree of baseline comorbidity than non-dialysis-dependent patients. Patients with CD were more likely to have a history of congestive heart failure (39.9 vs 18.9%), valvular heart disease (16.0 vs 11.7%), peripheral vascular disease (14.9 vs 9.1%) and diabetes (57.7 vs 30.1%). Conversely, subjects without CD were more likely to have undergone a cardiac catheterization during the same hospitalization (57.6 vs 46.1%, P < 0.001). In both the dialysis and the non-dialysis groups, on-pump surgery was used more frequently than off-pump surgery, but off pump-operations were slightly more common in dialysis patients. Two hundred and fourteen CD patients (33.7%) had off-pump surgery compared with (25.6%) non-dialysis patients. Post-operative outcomes In-hospital outcomes are summarized in Table 2. On average, costs of CABG were 1.6-fold higher in dialysis patients, perhaps because they were hospitalized for 6 days longer than in subjects not on dialysis. CD patients were significantly more likely to experience a major complication following CABG. In-hospital death was approximately 3 times more frequent in CD patients than in non-dialysis patients (11.1 vs 3.4%, P<0.001). Data on in-hospital mortality was missing in 63 patients. Univariate predictors of in-hospital death Unadjusted associations with in-hospital death are summarized in Table 3. CD status and peri-operative Table 2. Outcomes after CABG according to dialysis status Outcome Dialysis (n ¼ 635) Non-dialysis (n ¼ ) Odds ratio (95% CI) P-value In-hospital death 70 (11.1) 2587 (3.4) 3.52 ( ) <0.001 Total charges (dollars) <0.001 Length of stay (days) <0.001 Stroke 35 (5.5) 1838 (2.4) 2.38 ( ) <0.001 Sepsis 52 (8.1) 1946 (2.5) 3.34 ( ) <0.001 Pneumonia 34 (5.3) 2566 (3.3) 1.63 ( ) 0.029

4 4of7 Table 3. Univariate predictors of in-hospital mortality after coronary artery bypass surgery D. M. Charytan and R. E. Kuntz Condition Dead (n ¼ 2657) Alive (n ¼ ) RR (95% CI) P-value Demographics Age ( ) a <0.001 Female 1188 (44.8) (29.7) 1.51 ( ) <0.001 Black 58 (6.2) 2875 (5.0) 1.01 ( ) Uninsured 38 (1.45) 1696 (2.25) 1.01 ( ) Medical history Pre-operative dialysis dependence 70 (2.7) 565 (0.75) 3.52 ( ) <0.001 Congestive heart failure 1145 (43.0) (18.2) 2.37 ( ) <0.001 Arrhythmia 1013 (38.1) (30.3) 1.26 ( ) <0.001 Valvular heart disease 652 (24.5) 8533 (11.3) 2.17 ( ) <0.001 Peripheral vascular disease 233 (8.8) 6903 (9.2) 0.95 ( ) Neurological disorders 158 (5.9) 958 (1.3) 4.65 ( ) <0.001 Chronic pulmonary disease 609 (22.7) (18.0) 1.26 ( ) <0.001 Diabetes 590 (22.4) (30.6) 0.73 ( ) <0.001 Liver disease 40 (1.5) 358 (0.48) 3.21 ( ) <0.001 Cancer 115 (4.3) 4792 (6.4) 0.68 ( ) <0.001 Obesity 66 (2.4) 5843 (7.7) 0.32 ( ) <0.001 Weight loss 101 (3.8) 476 (0.64) 5.90 ( ) <0.001 Alcohol abuse 29 (1.1) 1196 (1.6) 0.70 ( ) Other risk factors Low volume hospital 661 (24.3) (22.7) 1.14 ( ) 0.15 Valve surgery 709 (26.8) 7199 (9.6) 2.78 ( ) <0.001 Admission for MI b 849 (32.0) (19.7) 1.62 ( ) <0.001 Coronary angiography 1636 (61.4) (57.4) 1.07 ( ) a OR for age >60 vs 60. b Myocardial infarction. mortality were strongly associated. The relative risk of in-hospital death in subjects on dialysis was 3.52 (95% CI: ). Other important predictors included congestive heart failure (RR 2.37, 95% CI: ), valvular heart disease (RR 2.17, 95% CI: ), neurological disorders (RR 4.65, 95% CI: ), and valve surgery (RR 2.78, 95% CI: ). Diabetic patients were somewhat less likely to die following surgery compared to nondiabetic patients. Among non-dialysis patients 562 of (2.4%) diabetic patients died compared with 2024 of (3.8%) non-diabetic patients. In the subgroup of 367 diabetic CD patients, there were 28 deaths (7.7%), compared with 42 deaths out of 268 non-diabetic CD patients (15.7%). Adjusted risk of in-hospital death Two logistic regression models were generated to adjust for confounding by comorbid conditions. CD status and in-hospital death remained strongly associated after adjustment for comorbidity. In Model 1, the adjusted odds of peri-operative death in subjects on CD was 3.38 (95% CI: ). This risk was similar before and after additional adjustment for cancer, alcohol abuse, liver disease, race, insurance status and surgical volume in Model 2 (Tables 4 and 5). The risk of death in patients with CD was not significantly altered by the presence of baseline comorobidities such as congestive heart failure (P for interaction ¼ 0.80), diabetes (P ¼ 0.14), peripheral vascular disease (P ¼ 0.67) or admission for myocardial infarction (P ¼ 0.92). Discussion In this study, we assessed the rates of in-hospital death and major morbidity in patients hospitalized for CABG. In a large, contemporary, national database, we found that patients on CD had a 3-fold higher risk of in-hospital death and were more likely to have non-fatal complications after bypass surgery than non-dialysis patients. The odds of death were virtually unchanged after adjustment for important surgical risk factors suggesting that the increase in risk in dialysis patients is independent of other comorbidities and is largely attributable to the presence of end-stage renal disease itself. These findings provide important information about the risks of CABG in CD patients and imply that data on the long-term benefits of CABG derived from trials conducted in the general population should be extrapolated to the dialysis population with extreme caution. Our data are consistent with and add to earlier studies of CABG in dialysis patients. Liu et al. [11] examined mortality rates in consecutive patients undergoing CABG in northern New England between 1992 and 1997, and found that dialysis patients had operative mortality rates of 12.2% and a multivariableadjusted odds of death 3.1 times higher than non-dialysis patients. In a study of dialysis patients

5 Peri-CABG mortality in dialysis patients 5 of 7 Table 4. Adjusted risks of peri-operative mortality: Model 1 Table 5. Adjusted risks of peri-operative mortality: Model 2 Condition Adjusted OR 95% CI P-value Condition Adjusted OR 95% CI P-value Unadjusted risk of death Dialysis dependence <0.001 Model 1 Dialysis dependence <0.001 Congestive heart failure <0.001 Valvular heart disease <0.001 Valve surgery <0.001 Peripheral vascular disease Arrhythmias Diabetes <0.001 Female < years old years old < years <0.001 Obesity <0.001 Pathological weight loss <0.001 Chronic lung disease Neurological disorders <0.001 Admission for myocardial <0.001 infarction Neurological disorders 17 (2.6) 1099 (1.4) <0.001 Coronary angiography during hospitalization hospitalized for a first coronary revascularization between 1995 and 1998, Herzog et al. [10] found that the overall in-hospital mortality rate was 8.6%. Our results are thus consistent with these earlier studies [19 21] as well as the finding that the risk of serious peri-operative morbidity and mortality is inversely related to pre-operative glomerular filtration rates [19,22 25]. Our study extends these findings by demonstrating that the risk of peri-operative death has not diminished with the decreasing levels of anaemia and increased dialysis adequacy of more recent dialysis cohorts [26]. Our results further demonstrate that mortality rates in dialysis patients remain high despite the increased availability of minimally invasive and off-pump surgical techniques used in approximately one-third of the CD patients in our study. Finally, with 635 CD and non-dialysis patients included in our analysis, the current analysis is one of the largest to compare post-cabg outcomes in dialysis and non-dialysis patients. The 7.7% absolute increase in in-hospital mortality rates that we observed in dialysis patients actually exceeds the 5.6% absolute reduction in 5-year mortality rates seen in pivotal-trials comparing CABG and medical therapy in the general population [1]. Nevertheless, observational studies suggest that CABG is an effective therapy for coronary artery disease in CD patients [27,28] and our findings of a 3-fold increase in the risk of peri-operative mortality sheds light on the appropriate role of CABG in this high-risk population. The absolute benefits of CABG in CD patients are likely to depend upon three factors: (i) the degree to which CABG s effect on the incidence of cardiac arrest, MI or congestive failure is similar in dialysis Unadjusted risks of death Dialysis dependence <0.001 Model 2 Dialysis dependence <0.001 Congestive heart failure <0.001 Valvular heart disease Valve surgery <0.001 Peripheral vascular disease Arrhythmias Diabetes <0.001 Female < years old years old years old <0.001 Obesity <0.001 Pathological weight loss <0.001 Chronic lung disease Neurological disorders <0.001 Admission for <0.001 myocardial infarction Coronary angiography during hospitalization Cancer Uninsured Alcohol abuse Medium volume centre High volume centre Black race Liver disease <0.001 Model 2 includes all parameters in Model 1 with further adjustment for liver disease, cancer, alcohol abuse, number of annual CABGs in the surgical centre, race and insurance status (see text for full details). Dialysis dependence and other variables with significant change between Model 1 and Model 2 are shown in bold. and non-dialysis patients i.e. the degree to which the relative risk reduction (RR) is preserved, (ii) the absolute risk of death without bypass surgery and (iii) the risk of peri-operative death following bypass. The first two factors are likely to remain uncertain in the absence of randomized trials of CABG surgery within the CD population. As shown in Figure 1, the high absolute mortality rates in dialysis patients of 60% at 5 years [5] result in a situation in which the net long-term benefit of CABG compared with medical therapy is likely to be highly dependent on the degree to which the high peri-operative risk in the CD population is offset by a preservation (or increase) of CABG s usual cardiovascular benefits. These considerations emphasize the need for well-designed randomized trials of medical therapy vs CABG in CD patients to better determine the optimal role for CABG in this high-risk population. Our results should be interpreted cautiously and within the context of the NIS s design. Administrative data may under-report comorbid conditions [29,30] and this could lead to overestimation of the risk attributable to CD status. However, the strength of the associations we observed, the use of an adjustment method validated for the NIS [14] and the recording of up to 15 diagnoses for each admission in the NIS,

6 6of7 Fig. 1. Change in overall-mortality at 5 years in dialysis patients treated with CABG vs medical therapy. Surgical mortality is assumed to be 11.1%. The effect of CABG varies according to baseline risk of death at 5 years and the relative effect of CABG on the long-term risk of cardiovascular death. suggest that under-reporting of comorbid conditions is unlikely to have qualitatively affected our results. Additionally, because coding data has a poor sensitivity for the detection of pre-dialysis chronic kidney disease [31], we were forced to include patients with non-dialysis-dependent chronic kidney disease within the control group. Since post-cabg mortality rates are high in pre-dialysis chronic kidney disease [22], our results are most likely a conservative estimate of the increase in post-cabg death associated with CD. In contrast to previous studies [12,17], we found that diabetic patients were less likely than non-diabetic patients to die prior to hospital discharge. Since we adjusted for presence of diabetes and since CD patients were more likely to have diabetes than non-cd patients, these findings are unlikely to account for the high mortality rate in CD patients. Nevertheless, the finding of a higher mortality rate in non-diabetic CABG patients does merit comment. Patients with diabetes were more likely to require CD or have a history of typical diabetes-related conditions than non-diabetics (data not shown). Misclassification of diabetic status is thus unlikely to explain this finding. Whether the lower risk of death in diabetics is due to random error, contemporary improvements in post-surgical outcomes in diabetic patients or selective referral of relatively low-risk diabetic patients for CABG is uncertain and merits further study. Finally, we cannot exclude the possibility that dialysis patients are referred for CABG only when their coronary disease becomes so advanced that it is otherwise untreatable. If this were true, the risk of peri-operative death might not be attributable to dialysis per se but would instead reflect residual confounding by prevailing practice patterns. This interpretation is consistent with the fact that the dialysis patients in our sample were more likely to be black or female characteristics that typically bias against referral for revascularization [32,33]. It is also consistent with our own findings that CD patients are only half as likely to be referred for revascularization D. M. Charytan and R. E. Kuntz or angiography after an admission for myocardial infarction [34]. If late referral of dialysis patients is indeed responsible for a significant portion of their in-hospital mortality after CABG, then our analysis may actually underestimate the potential benefits of earlier, more frequent use of CABG. In this case, an even stronger argument exists for the increased referral of dialysis patients for surgical bypass. In summary, we compared the rates of peri-cabg mortality in dialysis and non-dialysis patients in a large, national database and found that dialysis patients are more than three times as likely to die before leaving the hospital. This risk is not accounted for by comorbid conditions, and it is likely the result of either uraemia itself or an altered threshold for referring uraemic patients for bypass surgery. Despite this high risk of peri-operative death, CABG is likely to reduce mortality in specific subgroups of dialysis patients, and in the absence of randomized trials should be considered for dialysis patients with high-risk coronary artery disease. Acknowledgement. DK This work was supported by NIH grant T32 Conflict of interest statement. We have had no involvements that might raise the question of bias in the work reported or in the conclusions, implications or opinions stated. References 1. Yusuf S, Zucker D, Peduzzi P et al. Effect of coronary artery bypass graft surgery on survival: overview of 10-year results from randomised trials by the Coronary Artery Bypass Graft Surgery Trialists Collaboration. Lancet 1994; 344: Myers WO, Schaff HV, Gersh BJ et al. Improved survival of surgically treated patients with triple vessel coronary artery disease and severe angina pectoris. A report from the Coronary Artery Surgery Study (CASS) registry. J Thorac Cardiovasc Surg 1989; 97: VA Coronary Artery Bypass Surgery Cooperative Study Group. Eleven-year survival in the Veterans Administration randomized trial of coronary bypas surgery for stable angina. The Veterans Administration Coronary Artery Bypass Surgery Coooperative Study Group. N Engl J Med 1984; 311: Chertow GM, Normand SL, Silva LR, McNeil BJ. Survival after acute myocardial infarction in patients with end-stage renal disease: results from the cooperative cardiovascular project. Am J Kidney Dis 2000; 35: United States Renal Data System. USRDS 2002 Annual Data Report: Atlas of End-Stage Renal Disease in the United States, Bethesda, MD, Chertow GM, Normand SL, McNeil BJ. Renalism : inappropriately low rates of coronary angiography in elderly individuals with renal insufficiency. J Am Soc Nephrol 2004; 15: Serruys PW, Unger F, Sousa JE et al. Comparison of coronaryartery bypass surgery and stenting for the treatment of multivessel disease. N Engl J Med 2001; 344: Henderson RA, Pocock SJ, Clayton TC et al. Seven-year outcome in the RITA-2 trial: coronary angioplasty versus medical therapy. J Am Coll Cardiol 2003; 42: Henderson RA, Pocock SJ, Sharp SJ et al. Long-term results of RITA-1 trial: clinical and cost comparisons of coronary angioplasty and coronary-artery bypass grafting. Randomised intervention treatment of angina. Lancet 1998; 352:

7 Peri-CABG mortality in dialysis patients 7 of Herzog CA, Ma JZ, Collins AJ. Comparative survival of dialysis patients in the United States after coronary angioplasty, coronary artery stenting, and coronary artery bypass surgery and impact of diabetes. Circulation 2002; 106: Liu JY, Birkmeyer NJ, Sanders JH et al. Risks of morbidity and mortality in dialysis patients undergoing coronary artery bypass surgery. Northern New England Cardiovascular Disease Study Group. Circulation 2000; 102: Ferguson TB, Jr, Hammill BG, Peterson ED, DeLong ER, Grover FL. A decade of change risk profiles and outcomes for isolated coronary artery bypass grafting procedures, : a report from the STS National Database Committee and the Duke Clinical Research Institute. Society of Thoracic Surgeons. Ann Thorac Surg 2002; 73: ; discussion Waikar SS, Wald R, Chertow GM et al. Validity of International Classification of Diseases, Ninth Revision, Clinical Modification Codes for Acute Renal Failure. J Am Soc Nephrol 2006; 17(6): Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care J 1998; 36: Southern DA, Quan H, Ghali WA. Comparison of the Elixhauser and Charlson/Deyo methods of comorbidity measurement in administrative data. Med Care 2004; 42: Lemeshow S, Cook ED. Practical considerations in the analysis of complex sample survey data. Rev Epidemiol Sante Publique 1999; 47: Eagle KA, Guyton RA, Davidoff R et al. ACC/AHA guidelines for coronary artery bypass graft surgery: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee to Revise the 1991 Guidelines for Coronary Artery Bypass Graft Surgery). American College of Cardiology/American Heart Association. J Am Coll Cardiol 1999; 34: Jones RH, Hannan EL, Hammermeister KE et al. Identification of preoperative variables needed for risk adjustment of shortterm mortality after coronary artery bypass graft surgery. The Working Group Panel on the Cooperative CABG Database Project. J Am Coll Cardiol 1996; 28: Hannan EL, Kilburn H,Jr, O Donnell JF, Lukacik G, Shields EP. Adult open heart surgery in New York State. An analysis of risk factors and hospital mortality rates. JAMA 1990; 264: Owen CH, Cummings RG, Sell TL, Schwab SJ, Jones RH, Glower DD. Coronary artery bypass grafting in patients with dialysis-dependent renal failure. Ann Thorac Surg 1994; 58: Rostand SG, Kirk KA, Rutsky EA, Pacifico AD. Results of coronary artery bypass grafting in end-stage renal disease. Am J Kidney Dis 1988; 12: Anderson RJ, O Brien M, MaWhinney S et al. Renal failure predisposes patients to adverse outcome after coronary artery bypass surgery. VA Cooperative Study #5. Kidney Int 1999; 55: Higgins TL, Estafanous FG, Loop FD, Beck GJ, Blum JM, Paranandi L. Stratification of morbidity and mortality outcome by preoperative risk factors in coronary artery bypass patients. A clinical severity score. JAMA 1992; 267: Kurki TS, Kataja M. Preoperative prediction of postoperative morbidity in coronary artery bypass grafting. Ann Thorac Surg 1996; 61: O Connor GT, Plume SK, Olmstead EM et al. Multivariate prediction of in-hospital mortality associated with coronary artery bypass graft surgery. Northern New England Cardiovascular Disease Study Group. Circulation 1992; 85: NIH. USRDS 2003 Annual Data Report: atlas of end-stage renal disease in the United States Herzog CA, Ma JZ, Collins AJ. Long-term outcome of dialysis patients in the United States with coronary revascularization procedures. Kidney Int 1999; 56: Szczech LA, Reddan DN, Owen WF et al. Differential survival after coronary revascularization procedures among patients with renal insufficiency. Kidney Int 2001; 60: Powell H, Lim LL, Heller RF. Accuracy of administrative data to assess comorbidity in patients with heart disease. An Australian perspective. J Clin Epidemiol 2001; 54: Quan H, Parsons GA, Ghali WA. Validity of information on comorbidity derived rom ICD-9-CCM administrative data. Med Care 2002; 40: Winkelmayer WC, Schneeweiss S, Mogun H, Patrick AR, Avorn J, Solomon DH. Identification of individuals with CKD from Medicare claims data: a validation study. Am J Kidney Dis 2005; 46: Ayanian JZ, Epstein AM. Differences in the use of procedures between women and men hospitalized for coronary heart disease. N Engl J Med 1991; 325: Jha AK, Fisher ES, Li Z, Orav EJ, Epstein AM. Racial trends in the use of major procedures among the elderly. N Engl J Med 2005; 353: Charytan D, Mauri L, Agarwal A, Servoss S, Scirica B, Kuntz RE. The use of invasive cardiac procedures after acute myocardial infarction in long-term dialysis patients. Am Heart J 2006; 152: Received for publication: Accepted in revised form: