Respiratory failure (RF), or prolonged mechanical ventilation,

Transcription

1 CARDIOTHORACIC ANESTHESIOLOGY: The Annals of Thoracic Surgery CME Program is located online at To take the CME activity related to this article, you must have either an STS member or an individual non-member subscription to the journal. Risk Factors and Survival in Patients With Respiratory Failure After Cardiac Operations Michael L. Bailey, MBBS, Sven M. Richter, MD, Daniel V. Mullany, FJFICM, Peter J. Tesar, FRACS, and John F. Fraser, PhD, MRCP Critical Care Research Group, The Prince Charles Hospital, Chermside, Queensland, Australia Background. Respiratory failure is a known complication of cardiac operations and contributes to postoperative morbidity and death. This study assessed the relevance of risk factors in the development of respiratory failure, defined as postoperative ventilation exceeding 48 hours, and looked at the effect of respiratory failure on short-term and long-term mortality rates. Methods. De-identified data for patients who underwent cardiac surgical procedures at The Prince Charles Hospital between January 2002 and December 2007 were collected prospectively and analyzed using logistic regression to identify significant risk factors associated with respiratory failure. Long-term mortality data were analyzed for patients who underwent operations between 1994 and 2005 using Kaplan-Meier survival curves. Results. The risk factor analysis included 7,440 patients. Identified risk factors for respiratory failure included critical preoperative state, neurologic dysfunction, poor left ventricular function, active endocarditis, chronic obstructive pulmonary disease, elevated preoperative creatinine, previous cardiac operation, and age. Survival was assessed in 18,488 patients and demonstrated increased short-term and long-term mortality rates when respiratory failure developed and increased mortality rates with increasing duration of respiratory failure. Conclusions. Respiratory failure is complication of cardiac operations associated with increased mortality and cost. Identification of patients at risk of respiratory failure may help select surgical candidates and aid resource planning and optimization. (Ann Thorac Surg 2011;92:1573 9) 2011 by The Society of Thoracic Surgeons Respiratory failure (RF), or prolonged mechanical ventilation, is one of the most expensive short-term complications after cardiac operations. The consequences of RF include increased intensive care unit stays [1], increased morbidity, and increased mortality rates [2, 3]. Patients with RF and ventilator dependency are more likely to undergo tracheostomy [4], which is itself related to increased 30-day and long-term death [5]. There is no universally accepted definition of RF after cardiac operations, which has been variously defined as a postoperative requirement for mechanical ventilation between 12 hours [6] and 7 days, and commonly as more than 48 hours [1, 3, 7, 8], more than 72 hours [2, 4], and more than 7 days [9]. Improved knowledge of risk factors for RF may assist in the identification of appropriate surgical candidates or management of resources for patients who may be expected to have more complications. Improved knowledge of risk factors may also improve informed consent and the decision of a patient to undergo an operation. Filsoufi and colleagues [2] analyzed a data set of 5,798 patients from which they identified demographic, vascular, and cardiac risk factors as independent predictors of Accepted for publication April 1, Address correspondence to Dr Fraser, Critical Care Research Group, The Prince Charles Hospital, Rode Rd, Chermside, QLD 4509, Australia; john_fraser@health.qld.gov.au. RF after cardiac operations. Reddy and associates [1] used the data of 12,662 consecutive patients to develop a logistic risk model for the prediction of prolonged ventilation after cardiac operations in adults [1]. Murthy and colleagues [4] looked at predictors of ventilator dependency in 12,777 patients, noting the importance of hemodynamic status and early postoperative events. As yet, there has been no analysis of Australasian data in this field. The European System for Cardiac Operative Risk Evaluation (EuroSCORE) model can be predictive of cardiac surgical patient death [10], but mortality rates are overestimated in the Australian cardiac surgical population due to population differences, including increased age, more comorbidities, and cardiac risk factors [11]. Therefore, the aim of this study was to review the characteristics and outcomes of patients with postoperative RF after cardiac operations. Patients and Methods This study was conducted at The Prince Charles Hospital (Brisbane, QLD, Australia). Ethical approval was granted for the collection of prospective and follow-up patient data, which is de-identified and stored in a database for auditing and research of patient outcomes. All patients aged 18 years or older who underwent coronary artery, valve, or thoracic aortic operations between January 2002 and December 2007 were included. The study excluded 2011 by The Society of Thoracic Surgeons /$36.00 Published by Elsevier Inc doi: /j.athoracsur

2 1574 BAILEY ET AL Ann Thorac Surg RESPIRATORY FAILURE AFTER CARDIAC OPERATIONS 2011;92: Abbreviations and Acronyms CABG coronary artery bypass graft CI confidence interval COPD chronic obstructive pulmonary disease EF ejection fraction EuroSCORE European System for Cardiac Operative Risk Evaluation ICU intensive care unit LOS length of stay MI myocardial infarction PVD peripheral vascular disease OR odds ratio RF respiratory failure SD standard deviation VSD ventricular septal defect patients who underwent thoracic organ transplant, ventricular assist device placement, pulmonary thromboendarterectomy, or adult congential heart operations. Collected data included patient and operative details, data for the calculation of the logistic EuroSCORE [10], preoperative ventilation status, initial ventilation time, reventilation time, hours spent in the intensive care unit, and date of death. Respiratory failure was defined as a patient requiring mechanical ventilation for a total exceeding 48 hours at any time in the postoperative period. Hours of noninvasive ventilation were not included. Long-term survival for patients with and without RF was assessed for those who underwent procedures between 1994 and These data were especially collected and maintained for patients undergoing operations at The Prince Charles Hospital as part of the cardiac surgical program. Further analysis of these data also looked at the effect of increasing durations of RF on long-term death. The censoring date was February 14, Preliminary data analysis consisted of summary and descriptive statistics, as reported in Tables 1 and 2. Continuous variables were described using mean and standard deviation (SD). Categoric variables were compared using proportions with a 2 test. Cutpoint values, if used, were based on commonly used intervals such as the EuroSCORE [10, 11]. Unless otherwise stated, the primary outcome measure was all-cause mortality. A value of p 0.05 was considered statistically significant. Survival was analyzed using Kaplan-Meier curves. with results shown in Figures 1 and 2 [12]. Univariate (Table 3) and multivariable logistic regression (Table 4) was used to assess risk factors for RF. The logistic regression excluded 32 patients who died within the first 3 days. This included only 1 patient who was ventilated for 50 hours, whereas no other patient required ventilation for more than 48 hours. Stepwise backwards elimination used the variables in Table 1. The likelihood ratio test was used to identify terms that had a significant input to the model as it was built up from univariate to multivariable. Model calibration and discrimination were checked using goodness-of-fit tests and the area under the receiver operating characteristic curve. Statistical analysis was done using Excel software (Microsoft Corp, Redmond WA), Epi Info (Centers for Disease Control and Prevention, Atlanta, GA), and Stata 9 software (StataCorp, College Station, TX). Results The sample was a mean age of 64.6 (SD, 12.1) years, 2683 (36.1%) were older than 70 years on the date of operation, and 28.5% were women. The in-hospital mortality rate was 1.6%. Extubation occurred in 68.2% of patients within the first 12 hours postoperatively. Mean initial ventilation time was 17.8 (SD, 53.7) hours. The mean reventilation time was 4.77 (SD, 55.4) hours, and exceeded 1 hour in 246 patients (3.3 %). Of the overall group, 2,441 (32.8%) were initially ventilated for more than 12 hours. Total ventilation times exceeded 24 hours in 707 patients (9.5%), exceeded 48 hours in 392 (5.3%), and exceeded 72 hours in 279 (3.8 %). Risk Factors for RF Risk factors and odds ratios (OR) for the development of RF from the logistic regression are summarized in Tables 3 and 4. Female sex was not statistically significant as a risk factor in the univariate analysis. Unstable angina, pulmonary hypertension, postinfarct ventricular septal defect, and an aortic wall operation at the time of the procedure were not statistically significant in the multivariable analysis. Survival Analysis Kaplan-Meier curves for the 18,488 patients who underwent cardiac operations between 1994 and 2005 are shown in Figure 1. Within this group, complete follow-up data were available for 17,796 (119,126 person-years), and 3,353 deaths occurred. The analysis excluded patients who died within the first 48 hours postoperatively. The curves show that development of RF causes a significant early increase in death, which then remains increased until approximately 7 years postoperatively. In these first 7 years, patients who experienced RF have a hazard ratio of 3 to 4 for dying. After the first 7 years, the mortality rate of patients who presented with RF becomes similar to that for patients who did not. A further analysis (Fig 2) examined the effect of increasing duration of RF on death. Increasing duration of RF was associated with increasing short-term and long-term mortality rates. Comment The reported incidence of RF after cardiac operations varies between 5% and 20%, depending on the definition of RF and the procedures included in the analysis [1 4, 7 9]. In this study, the calculated incidence for RF was 5.3%, which is lower than that reported in most other studies. Filsoufi and colleagues [2] defined RF as ventila-

3 Ann Thorac Surg BAILEY ET AL 2011;92: RESPIRATORY FAILURE AFTER CARDIAC OPERATIONS Table 1. Characteristics of the Study Sample for the Years 2002 to 2007 Variable a No RF RF p Value b Excluded c Patients 7, Age, years 64.4 (12.1) 66.1 (13.0) d 70.0 (10.6) 70 years old 2, Female sex 1, Type of procedure e Isolated CABG 4, Isolated single valve 1, CABG single valve Complex Aortic wall operation Urgency of procedure e Elective 4, Urgent 2, Emergency Salvage Preoperative status Ventilation Vascular disease Previous cardiac operation Creatinine, mmol/l (0.050) (0.080) d (0.096) Creatinine 0.2 mmol/l COPD Neurologic dysfunction Active endocarditis Critical preoperative state Unstable angina Ejection fraction Ejection fraction , Recent myocardial infarction 1, Postinfarct VSD d 1 Pulmonary hypertension a Data are shown as mean (standard deviation) or number. b The p value using 2 test. c Patients excluded from logistic regression those who died within 3 days of the operation. d Fisher exact test where any group has fewer than 5 patients or Mann-Whitney U test for means. e The 2 calculation for group of procedures with reference procedure as the first in the list, excluding aortic wall surgery subanalysis. CABG coronary artery bypass grafting; COPD chronic obstructive pulmonary disease; RF respiratory failure; VSD ventricular septal defect. tion exceeding 72 hours and reported a 9.1% incidence of RF. Predictors of RF The multivariable model showed preoperative variables that independently predicted RF were critical preoperative state, neurologic dysfunction, ejection fraction (EF) of less than 0.30, active endocarditis, chronic obstructive pulmonary disease, preoperative creatinine exceeding 0.2 mmol/l, age older than 70, previous cardiac operations, and vascular disease. Complexity of procedure and urgency of procedure also increased the risk of RF. These Table 2. Outcomes of the Study Sample Years No RF RF p Value a Excluded In-hospital mortality, No. (%) 50 (0.7) 71 (18) Initial ICU LOS, mean (SD) hours 33.9 (25.1) 268 (294) (14.2) Initial ventilation time, mean (SD) hours 11.1 (6.93) 139 (196) (13.6) Repeat ventilation rate, No. (%) 92 (1.3) 154 (39) Hospital LOS, mean (SD) days 9.3 (5.3) 30 (23) (0.71) a The p value calculated using 2 or Mann-Whitney U test for means. ICU intensive care unit; LOS length of stay; RF respiratory failure; SD standard deviation.

4 1576 BAILEY ET AL Ann Thorac Surg RESPIRATORY FAILURE AFTER CARDIAC OPERATIONS 2011;92: Fig 1. Unadjusted survival is shown with and without respiratory failure (1994 to 2005). risk factors have been previously identified [1, 2, 6, 13, 14], and our findings further support the veracity of these data and their relevance for an Australasian population. Critical preoperative state was one of the strongest preoperative predictors of RF (OR, 5.1), but the relevance remains unclear because the definition of critical preoperative state in the study data includes preoperative ventilation and may therefore explain the strong predictive value. Further analysis of the data may be required to exclude patients with preoperative RF from the data set. Neurologic dysfunction is a strong predictive factor (OR, 3.2). Previous cerebrovascular accident has been used in other analyses [1, 2] but has not been reported as a significant risk factor for RF. In a previous study that evaluated the 12-year predictive value of the EuroSCORE, neurologic dysfunction was not significant in predicting overall or cardiac death, but the effect may be limited because only 9 patients with neurologic dysfunction were included in the study sample [18]. Several investigators have described poor left ventricular function (commonly defined as EF 0.30) as a risk factor for RF in coronary artery bypass graft (CABG) operations [1, 2, 6]. The ORs for poor left ventricular function in the literature range from 1.4 to 2.2. Cislaghi and colleagues [6] found that a functional classification of heart failure (New York Heart Association class III or Fig 2. Unadjusted survival is shown by duration of postoperative mechanical ventilation.

5 Ann Thorac Surg BAILEY ET AL 2011;92: RESPIRATORY FAILURE AFTER CARDIAC OPERATIONS Table 3. Univariate Logistic Regression Analysis of the Study Sample for the Years 2002 to 2007 (N 7408) a Variable OR (95% CI) p Value b Age 70 years ( ) Female sex ( ) Type of procedure Isolated CABG ( ) Isolated single valve Nonconvergent CABG single valve ( ) Complex ( ) Aortic wall operation ( ) Urgency of procedure Elective ( ) Urgent ( ) Emergency ( ) Salvage Nonconvergent Preoperative status Ventilation ( ) Vascular disease ( ) Previous cardiac operation ( ) Creatinine, mmol/l ( ) Creatinine 0.2 mmol/l ( ) COPD ( ) Neurologic dysfunction ( ) Active endocarditis ( ) Critical preoperative state ( ) Unstable angina ( ) EF ( ) EF ( ) Recent MI ( ) Postinfarct VSD ( ) Pulmonary hypertension ( ) a Excluding those who died within the first 3 days. b Calculated for 95% CI. CABG coronary artery bypass grafting; CI confidence interval; COPD chronic obstructive pulmonary disease; EF ejection fraction; MI myocardial infarction; OR odds ratio; VSD ventricular septal defect. higher) was associated with increased risk of RF (OR, 1.6) as an independent variable from poor EF in their logistic regression model. The multivariable analysis showed an OR of 3.1, suggesting a similar or slightly stronger correlation. The Australian population also has also been noted to have a higher prevalence of poor left ventricular function [11]. Active endocarditis may result in embolization, heart failure, and neurologic complications, including embolic stroke, acute encephalopathy, or cerebral hemorrhage, therefore leading to an increased incidence of such events. With an OR of 2.6, the predictive value almost matches neurologic dysfunction. Previous cardiac operation was a risk for RF (OR, 1.7). This is comparable with similar studies that showed an OR of 1.7 [2] and 2.4 [1]. The EuroSCORE definition of vascular disease is more inclusive of vascular diseases processes than peripheral vascular disease (PVD) [11], which has been considered in other studies. Patients with PVD have higher rates of complications than patients without PVD [19]. This study showed that vascular disease increases the risk of RF (OR, 1.6), but other studies have shown a higher predictive value for PVD by comparison (OR, 1.7 to 1.9) [1, 2]. Chronic obstructive pulmonary disease (COPD) has been associated with adverse outcomes after CABG [20]. Reddy and colleagues [1] identified forced expiratory volume in 1 second of less than 0.7 as an independent risk factor after cardiac operations for the development of RF (OR, 1.7). Cislaghi and colleagues [6] identified COPD as a risk of developing RF (OR, 1.5). In the Prince Charles Hospital patient group, COPD was associated with an increased risk of RF (OR, 2.2), but the Australian population has a higher incidence of COPD [11]. Elevated levels of creatinine have been associated with poor outcomes after myocardial infarction [15], CABG [16], and valvular heart operations [17]. The cutoff value for preoperative creatinine of more than 0.2 mmol/l has been used to denote impaired renal function in the EuroSCORE model [10, 11]. In this instance, creatinine exceeding 0.2 mmol/l was associated with an OR of 2.2. Chronic renal failure has been shown to have an OR of 1.6 [6]. Age has previously been reported to be a independent predictor of RF [1, 2, 6]. In this study population, age Table 4. Multivariable Logistic Regression Analysis of the Study Sample for the Years 2002 to 2007 (N 7408) a Variable OR (95% CI) p Value b Age 70 years ( ) Complexity (compared with CABG) Single valve ( ) Valve CABG ( ) Complex ( ) Urgency (compared with elective) Urgent ( ) Emergency ( ) Salvage ( ) Preoperative status Vascular disease ( ) Previous cardiac operation ( ) Creatinine, 0.2 mmol/l ( ) COPD ( ) Neurologic dysfunction ( ) Active endocarditis ( ) Critical preoperative state ( ) EF (compared with EF 0.50) EF ( ) EF ( ) a Excluding those who died within the first 3 days. b Calculated for 95% CI CABG coronary artery bypass grafting; CI confidence interval; COPD chronic obstructive pulmonary disease; EF ejection fraction; OR odds ratio.

6 1578 BAILEY ET AL Ann Thorac Surg RESPIRATORY FAILURE AFTER CARDIAC OPERATIONS 2011;92: older than 70 had an OR of 1.7. Filsoufi and colleagues [2] calculated an OR of 1.6 for increasing age. Reddy and colleagues [1] used three categories of age, which clearly showed the OR to increase with age (2.2 to 5.5 for age groups 65 to 75 and 80 years). Cislaghi and colleagues [6] found age older than 65 had OR of 1.3. Female sex has variously been reported to be a significant predictor of RF [2, 13, 21] or a nonsignificant predictor of RF [1, 22]. This study found that female sex was not a risk factor for the development of RF after cardiac operations. The type of procedure was clearly an independent risk factor for RF. Compared with isolated CABG, isolated valve (OR, 1.7), CABG plus single-valve operation (OR, 2.9), and complex procedure (OR, 5.1) all showed increased risk for RF. The risk of RF increased with the complexity of the procedure. Mitral valve operations have previously been associated with an increased risk of RF (OR, 2.2) [1], as has combined CABG and single-valve replacement (OR, 1.5) [2]. An increasing risk with increased urgency was also identified when urgent (OR, 1.5), emergency (OR, 5.9), or salvage (OR, 14.8) procedures were compared with elective procedures. Urgent operation has been associated with a risk of RF (OR, 1.6), similar to the findings of this study, whereas emergency operation has been associated as a risk factor with a lower OR of 2.1 than presently identified [1]. The Australian population has also been noted to be more likely to require emergency operations [11]. Death After RF High in-hospital mortality rates of 15% to 20% have been reported after RF [2, 7]. Patients without RF have a higher probability of discharge from the intensive care unit, cardiac surgical wards, and rehabilitation wards, with higher hospital survival [6]. In the study population of 7,440 patients, RF was associated with an in-hospital mortality rate of 18% compared with 0.7% in the group without RF. The analysis of the long-term survival data showed a significant increase in death in patients when RF develops. Increasing duration of RF is also associated with increasing death. These increases in death persist well beyond the perioperative period and may also be associated with the increasing levels of comorbidities in patients at higher risk of developing RF. Limitations The analysis did not exclude patients with preoperative ventilation, which may have been a confounding variable in the evaluation of risk factors for the development of RF. The effect of this would most likely be to overestimate the ORs for the development of RF. Several of the variables that were predictive of RF became nonsignificant in the multivariable analysis. However, it may be that a higher-powered study is required to confirm the significance of these variables. In particular, pulmonary hypertension may become significant. Conclusion Multiple studies have consistently shown that RF is a complication of cardiac operations. The development of RF results in higher short-term and long-term mortality rates. Likewise, the duration of RF has been shown to increase patient mortality rates. The identification of patients at risk of RF has the potential to improve the selection of surgical candidates, improve resource allocation, and improve informed consent. This may suggest that the risk of RF should be emphasized in patients with strong or multiple risk factors as part of their informed consent. It may also be possible to schedule their procedure on a Thursday or Friday, thereby allowing better use of an intensive care unit over the weekend when fewer elective cases are scheduled. Finally, although not yet undertaken in this study, it may be possible to develop a scoring system derived from identified risk factors that can aid in predicting RF and subsequent resource requirements. We acknowledge Carole L. Foot for obtaining long-term mortality data and Adrian Barnett for assistance with statistical analysis. References 1. Reddy S, Grayson A, Griffiths E, Pullan D, Rashid A. Logistic risk model for prolonged ventilation after adult cardiac surgery. Ann Thorac Surg 2007;82: Filsoufi F, Rahmanian P, Castillo J, Chikwe J, Adams D. Predictors of early and late outcomes of respiratory failure in contemporary cardiac surgery. Chest 2008;133: Hein O, Birnbaum J, Wernecke K, Konertz W, Jain U, Spies C. Three-year survival after four major post-cardiac operative complications. Crit Care Med 2006;34: Murthy S, Arroliga A, Walts P, et al. Ventilatory dependency after cardiovascular surgery. J Thorac Cardiovasc Surg 2007; 134: Walts P, Murthy S, Arroliga A, et al. Tracheostomy after cardiovascular surgery: an assessment of long-term outcome. J Thorac Cardiovasc Surg 2006;131: Cislaghi F, Condemi AM, Corona A. Predictors of prolonged mechanical ventilation in a cohort of 5123 cardiac surgical patients. Eur J Anaesthesiol 2009;26: Kollef M, Wragge T, Prasque C. Determinants of mortality and multiorgan dysfunction in cardiac surgery patients requiring prolonged mechanical ventilation. Chest 1995;107: Spivack S, Shinozaki T, Albertini J, Deane R. Preoperative prediction of postoperative respiratory outcome: coronary artery bypass grafting. Chest 1996;109: Papparlardo F, Frano A, Landoni G, Cardano P, Zangrillo A, Alfieri O. Long-term outcome and quality of life of patients requiring prolonged mechanical ventilation after cardiac surgery. Eur J Cardiothorac Surg 2004;25: Nashef S, Roques F, Michel P, Gauducheau E, Lemeshow S, Salamon R. European system for cardiac operative risk evaluation (EuroSCORE). Eur J Cadiothorac Surg 1999;16: Yap C, Mohajeri M, Ihle B, Wilson A, Goyal S, Yii M. Validation of EuroSCORE model in an Australian patient population. ANZ J Surg 2005;75: Rao S, Schoenfeld D. Survival methods. Circulation 2008;115: Branca P, McGaw P, Light R. Factors associated with prolonged mechanical ventilation following coronary artery bypass surgery. Chest 2001;119: O Rourke D, Quinton H, Piper W, et al. Survival in patients with peripheral vascular disease after percutaneous coronary intervention and coronary artery bypass graft surgery. Ann Thorac Surg 2004;78:

7 Ann Thorac Surg BAILEY ET AL 2011;92: RESPIRATORY FAILURE AFTER CARDIAC OPERATIONS 15. Foley R, Parfrey P. Cardiac disease in chronic uremia: clinical outcome and risk factors. Adv Ren Replace Ther 1997;4: Hirose H, Amano A, Takahashi A, Nagano N. Coronary artery bypass grafting for patients with non-dialysis dependent renal dysfunction (serum creatinine 2.0 mg/dl). Eur J Cardiothorac Surg 2001;20: Rankin J, Hammill B, Ferguson T, et al. Determinants of operative mortality in valvular heart surgery. J Thorac Cardiovasc Surg 2006: Biancari F, Kangasniemi O, Luukkonen J, et al. EuroSCORE predicts immediate and late outcome after coronary artery bypass surgery. Ann Thorac Surg 2006;82: Birkmeyer J, Quinton H, O Connor N, et al. The effect of peripheral vascular disease on in-hospital mortality rates with coronary artery bypass surgery. J Vasc Surg 1995;21: Samuels L, Kaufman M, Morris R, et al. Coronary artery bypass grafting in patients with COPD. Chest 1998;113: Shroyer A, Commbs L, Peterson E, et al. The Society of Thoracic Surgeons: 30-day operative mortality and morbidity risk models. Ann Thorac Surg 2003;75: Canver C, Chanda J. Intraoperative and postoperative risk factors for respiratory failure after coronary bypass. Ann Thorac Surg 2003;74:853 7.