Silent Aspiration After Coronary Artery Bypass Grafting

Transcription

1 Silent Aspiration After Coronary Artery Bypass Grafting O. Brewster Harrington, MD, John K. Duckworth, MD, Carey L. Starnes, BSN, Patricia White, MA, Lynn Fleming, MS, Stephen B. Kritchevsky, PhD, and Rexann Pickering, MSN Methodist Hospitals of Memphis and Department of Preventive Medicine, University of Tennessee, Memphis, Tennessee Background. Silent aspiration was recognized to be a more frequent complication at this hospital in patients who have had coronary artery bypass grafting than in the general surgical population. Methods. A case-control retrospective study covering a 4.5-year period was conducted to determine risk factors for pharyngeal dysfunction resulting in silent aspiration. Results. Significant predictors of silent aspiration were age, history of cerebral vascular disease, insulindependent diabetes mellitus, myocardial infarction, and chronic obstructive pulmonary disease. Intraaortic balloon pump and number of units of fresh-frozen plasma were the only independent intraoperative factors associated with silent aspiration in a model using continuous variables directly. Cold fibrillation was used in 7 of 53 study cases but no control patients, so it could not be modeled. Postoperative complications occurring with greater frequency included neurologic complications, adverse pulmonary outcomes, repeat surgical interventions, infection, and death. Using an Aspiration Risk Profile developed from the retrospective study, in a detailed prospective study of 10 patients, 3 of 4 patients with postoperative dysphagia had objective evidence of stroke. Conclusions. These findings suggest that postoperative coronary artery bypass graft dysphagia may be the result of intraoperative cerebral injury, and that careful postoperative clinical evaluation of coronary artery bypass graft patients with risk factors may result in early diagnosis of pharyngeal dysfunction with the goals of preventing silent aspiration and reducing morbidity, mortality, and hospital cost. (Ann Thorac Surg 1998;65: ) 1998 by The Society of Thoracic Surgeons ilent aspiration (SA) in the postoperative coronary S artery bypass graft (CABG) patient may result in significant postoperative morbidity, including reintubation, pulmonary failure, pneumonia, or death. Silent aspiration, a severe form of pharyngeal dysfunction, is the spilling of oral contents below the level of the true vocal cords during swallowing without the elicitation of gagging or coughing as the material passes through the pharynx [1]. Aspiration is silent in approximately 40% of patients with pharyngeal dysfunction who aspirate [2]. Pharyngeal dysfunction, which is a wellknown complication of stroke and other central nervous system diseases [3 6], has also been noted to be associated with tracheostomy and prolonged intubation [7 11]. In 1992, SA was recognized to be a more frequent complication at this hospital in patients who had undergone CABG than in the general surgical population. Although SA occurs in only a small percentage of CABG patients [12, 13], morbidity, length of hospital stay, and cost may be devastating if the process is not recognized early and treated with appropriate prophylactic measures [14]. To identify factors that may influence the occurrence of SA, a retrospective analysis of 53 recognized cases of SA Accepted for publication Jan 10, Address reprint requests to Dr Harrington, 1325 Eastmoreland, Suite 220, Memphis, TN among 5,777 post-cabg patients cared for at Methodist Hospitals of Memphis from 1989 through 1994 was conducted. The results were used to design a clinical profile to predict increased risk of developing pharyngeal dysfunction and possibly SA postoperatively. This model was termed aspiration risk profile (ARP). Subsequently, a small prospective study was initiated in an effort to validate the ARP, to rule out the possibility that swallowing dysfunction was present before the operation, and to elucidate the etiology of pharyngeal dysfunction in the CABG patient. Adoption and implementation of this, or a comparable ARP, could be used to predict the risk of surgically related pharyngeal dysfunction in the CABG patient. This increase in awareness should result in earlier diagnosis and treatment of SA, thus reducing morbidity, mortality, and cost. Material and Methods Both the retrospective and prospective studies were approved by the Institutional Review Board of Methodist Hospitals of Memphis. Informed consent was obtained from patients participating in the prospective study. Retrospective Study Population A retrospective analysis was conducted among the 5,777 CABG procedures performed at Methodist Hospitals of 1998 by The Society of Thoracic Surgeons /98/$19.00 Published by Elsevier Science Inc PII S (98)

2 1600 HARRINGTON ET AL Ann Thorac Surg POST-CABG SILENT ASPIRATION 1998;65: Memphis from July 1, 1989, through August 11, Seventy-one of these patients were recognized to have manifested some degree of pharyngeal dysfunction identified by videofluoroscopic examination (VFE) [14, 15] of barium swallow or modified Evans blue dye test [16] in those with a tracheostomy; 53 of these patients with pharyngeal dysfunction experienced SA, as demonstrated by VFE. Videofluoroscopic examination was performed no sooner than 24 hours after extubation in those patients not requiring a tracheostomy. Patients requiring a tracheostomy and a modified Evans blue dye test initially later received a VFE except for 4 patients who were unable to receive a follow-up VFE because of a tracheostomy at time of discharge or death. Because the presence of a tracheostomy may contribute to pharyngeal dysfunction and aspiration, risk factors were analyzed both with and without the patients with tracheostomies. One hundred six control subjects were systematically sampled, after a random start, from a list of all non study subjects undergoing CABG operations during the same period as the study subjects [17]. These randomly selected control subjects were not matched to the study subjects because matching would have precluded the quantification of the strength of the matching variable(s) as a predictor of SA [18]. However, the risk factor relationships were also examined after restricting the analysis to those older than 63 years, the median age of the control subjects, in effect matching the restricted samples on age. Data Collection and Statistical Techniques The hospital medical records of study patients and control subjects were reviewed by a single data abstractor, an experienced cardiovascular surgical nurse. Information was collected covering every aspect of the patient s hospital experience: (1) demographics, (2) comorbidities, (3) medications (preoperative and intraoperative), (4) severity of disease, (5) surgical and anesthetic procedures, and (6) postsurgical experiences. For this study, a history of cerebral vascular disease included a history of transient ischemic attack or cerebral vascular accident. Aortic disease was defined as palpable aortic thickening or calcification. Chronic obstructive pulmonary disease accounted for 98% of the current pulmonary disease used in the models. The remaining 2% included other manifestations of pulmonary disease, such as pneumonia. Any variable significantly associated with SA at the 0.1 level was considered for further analysis. The data were abstracted using the Epi-Info System (version 5) [19]. Statistical analyses were performed using SAS (version 6.1) [20]. The Student s t test was used to compare the mean levels of continuous variables. For categorical variables, the cases and controls were compared using the 2 test, unless small sample sizes led to the use of Fisher s exact test. Odds ratios were calculated to quantify the strength of the association between the variable and case status. Logistic regression was then used to model the joint contribution of multiple variables to the prediction of case status. Two models were constructed, one for factors that were present preoperatively and another for factors associated with operative procedures. All variables that were related to post-cabg SA in a univariate mode were considered as candidates for the model. All multivariate models contained age. Other candidate variables were considered using a forward selection strategy [18]. To determine whether the associations differed by age, the models were also fit using only study patients older than 63 (the median age of the control group). The nominal type I error rate is The multivariate logistic model developed in the retrospective analysis was used to develop a score to index patient risk for SA. The point contributions were determined from the size of the -coefficient of the variable found to be significant in the multivariable model. Because of the predominance of SA among older CABG patients, the ARP was developed using the subsample of patients older than 63. The statistical model used to develop the ARP differs from the model predicting SA in the entire patient population in that in the entire patient population insulin-dependent diabetes mellitus (IDDM) was retained in the prediction model. Insulin-dependent diabetes mellitus was relatively uncommon in patients older than 63 years. Therefore, the prediction model used all forms of diabetes rather than just IDDM. This decision was supported by the fact that diabetes without respect to insulin dependence was a predictor of SA among patients older than 63 years. Prospective Study Population One hundred forty-one consecutive CABG patients older than 63 years at Methodist Hospitals of Memphis were evaluated preoperatively for this prospective study from January through August Sixty-one patients scored greater than 2 on the ARP profile. Ten of these 61 patients consented to additional examinations including carotid ultrasound, preoperative and postoperative detailed neurologic evaluation, and preoperative and postoperative cranial computed tomography and VFE. Results Retrospective Study PATIENT MATCHING. Descriptive statistics comparing patients and control subjects showed no significant difference in sex, race, weight, preoperative medication, use of internal thoracic artery as a graft, or number of vessels bypassed. Patients with SA were considerably older than control subjects (mean, 70.6 years versus 61.8 years), more likely to have had previous cardiac operation (28.3% versus 15.1%), and more likely to have undergone additional surgical procedures at the time of their CABG (13.2% versus 1.9%) (Table 1). RISK FACTORS. Age was the most important risk factor identified. Patients with SA ranged in age from 43 to 83 years with a mean of 70.6 years and a median of 72 years. Control subjects ranged in age from 35 to 81 years with a mean age of 61.8 years and a median of 63.5 years. The incidence of SA rose sharply at 64 years and older. The rate of SA among those older than 64 years was 5.6 times

3 Ann Thorac Surg HARRINGTON ET AL 1998;65: POST-CABG SILENT ASPIRATION 1601 Table 1. Comparison of Study Patients and Control Population Study Patients Control Subjects (n 106) p Value Male (%) Race (%) White Black Other Age (y) Mean Median Operative procedure (%) CABG alone CABG with Valve repair/ replacement Other procedures Previous cardiac operation (%) CABG coronary artery bypass graft. greater than among those younger than that age. The relative risk for developing SA for a 10-year difference in age was 2.61, p (95% confidence interval, 1.7 to 4.0). Other preoperative risk factors with both clinical and statistical significance were (1) history of cerebrovascular disease, (2) IDDM, (3) chronic obstructive pulmonary disease, (4) history of a myocardial infarction, and (5) congestive heart failure (Table 2). The adjusted model shows the effect of each factor on the risk of SA, adjusting for the effects of the other factors in the model. All factors except congestive heart failure remained significant predictors of SA after statistical adjustment. Restricting the population to those older than 63 years showed that IDDM was an important predictive factor primarily in younger patients and that a history of cerebrovascular disease was a particularly potent predictive factor in older patients. Results were analyzed separately, excluding those patients who received tracheostomies. This analysis revealed that the risk factors were the same regardless of the absence or presence of tracheostomy (data not shown). When IDDM as a possible risk factor was analyzed, 37.5% of patients with SA and younger than 63 years required insulin to control their diabetes. Only 1.9% of patients in the control group without SA and younger than 63 years required insulin. Intraoperative factors associated with a greater than twofold relative risk of SA are shown in Table 3. In the multivariate models, only the use of the intraaortic balloon pump and elevated pump time remained significant predictors of SA. The models shown in Table 3 are based on dichotomized variables. In a model using continuous variables directly, time on the bypass pump was not a significant predictor of SA (odds ratio 1.0; p ), and the number of units of fresh-frozen plasma was associated with only a borderline significant increase in risk of SA (odds ratio 1.5/unit fresh-frozen plasma used; p 0.053). Once again results were analyzed separately, excluding patients who received tracheostomies; tracheostomy did not influence or modify the significance of risk factors (data not shown). Cold fibrillation, a surgical technique used to avoid clamping the aorta in those patients found by palpation to have severe ascending aortic atherosclerosis or calcification, was used in 7 of the 53 patients who had SA. However, each of these patients underwent ascending aorta or aortic arch cannulation. Notably, no control patients underwent cold fibrillation. Because there were no control patients needing this approach, the effect of cold fibrillation on SA could not be modeled. This should not be interpreted to mean that cold fibrillation is not an independent predictor of SA. However, models both including and excluding the 7 patients differed little (data not shown). POSTOPERATIVE COMPLICATIONS. Complications occurred with greater frequency in study cases as compared with controls (Table 4), particularly neurologic and pulmonary complications, repeat surgical intervention, and infec- Table 2. Comparison of Univariate and Odds Ratios for Preoperative s % Study Patients % Control Subjects (n 106) Odds Ratio (all patients) (patients 63 y) Age 63 y (p ) 8.6 (p )... Cerebrovascular disease (p 0.007) 4.5 (p 0.02) 8.7 (p 0.007) (history) IDDM (p 0.021) 5.1 (p 0.04) 2.0 (p 0.43) COPD (p 0.012) 3.6 (p 0.02) 4.1 (p 0.04) Myocardial infarction (p 0.013) 3.6 (p 0.005) 4.6 (p 0.007) (history) Congestive heart failure (p 0.024) 2.1 (p 0.302) 1.6 (p 0.56) a for age, history of cerebrovascular disease, IDDM, COPD (98% COPD, 2% other current pulmonary disease), history of myocardial infarction, and congestive heart failure. COPD chronic obstructive pulmonary disease; IDDM insulin-dependent diabetes mellitus.

4 1602 HARRINGTON ET AL Ann Thorac Surg POST-CABG SILENT ASPIRATION 1998;65: Table 3. Comparison of Univariate and Odds Ratios for Intraoperative s % Study Patients % Control Subjects (n 106) Odds Ratio (all patients) (patients 63 y) Cold fibrillation (p 0.001) model not run model not run Aortic disease (p 0.07) 1.6 (p ) 2.0 (p ) Use of IABP (p 0.002) 4.4 (p ) 4.9 (p 0.06) Hours of anesthesia (p 0.005) 1.4 (p ) 2.1 (p ) PRBC 1 unit transfused (p 0.025) 0.6 (p ) 0.4 (p ) FFP 2 units transfused (p 0.001) 3.6 (p 0.111) 5.4 (p ) Pump time 77 min (p 0.003) 2.6 (p 0.035) 2.3 (p ) a Model containing age, aortic disease, IABP, dichotomized hours of anesthesia, dichotomized units of PRBC, dichotomized units of FFP, and dichotomized pump time. FFP fresh-frozen plasma; IABP intraaortic balloon pump; PRBC packed red blood cells. tion. Ten (18.9%) had documented cerebral vascular accidents, and 7 (13.2%) had hypoxic encephalopathy based on additional documentation. Of the remaining 14 patients who had no definitive neurologic diagnosis, 7 (13.2%) had signs or symptoms such as confusion, delirium, tremors, or slurred speech, 6 (11.3%) demonstrated mild confusion, and 1 (1.9%) had Guillain-Barré syndrome by history with a postoperative recurrence. In total, 31 (58.5%) of the patients with SA were retrospectively observed to have clinical findings that could be classified as neurologic in nature. Of those study patients with adverse pulmonary outcomes, 30 (56.6%) were initially intubated more than 24 hours, 12 (22.6%) had pneumonia, 14 (26.4%) experienced adult respiratory distress syndrome, and 20 (37.7%) eventually required tracheostomies. Prospective Study VALIDATION OF THE ASPIRATION RISK PROFILE. The ARP developed using the case-control study is shown in Table 5. Based on the model, a score of 2 points on the ARP is consistent with approximately a 12% risk of developing SA. To test the efficacy of the model, we applied the ARP prospectively to identify patients undergoing CABG who were at high risk for developing SA. Of the 141 patients Table 4. Comparison of Postoperative Experiences Experience Study Patients (%) Control Subjects (%) (n 106) p Value Intubated 24 h Reintubated Tracheostomy Pneumonia Adult respiratory distress syndrome Neurologic complications Percutaneous endoscopic gastrostomy Mortality evaluated for this study, 61 had an ARP score greater than 2.0. Of these 61 patients, 7 had pharyngeal dysfunction as documented by VFE postoperatively. Once pharyngeal dysfunction was demonstrated in this population, appropriate preventive measures were taken [14]. Clinical SA did not develop in any of the patients. Of the 80 patients with an ARP score less than 2.0, none experienced postoperative pharyngeal dysfunction. The ARP model had a sensitivity of 100% for pharyngeal dysfunction (95% confidence interval, 95% to 100%) and a moderate specificity of 60% (95% confidence interval, 51% to 68%). EVALUATION OF PATIENTS AT HIGH RISK FOR SILENT ASPIRATION. Ten of the 61 patients scoring greater than 2.0 on the ARP agreed to be enrolled in a prospective study to evaluate possible mechanisms in the development of pharyngeal dysfunction. Four of the 10 patients had abnormal or worsening pharyngeal function postoperatively as demonstrated by VFE. In 3 of these 4 patients there were accompanying neurologic changes. Two of these patients had new cerebral infarcts noted on the postoperative cranial computed tomographic scan and in 1 patient a cerebral vascular accident was diagnosed clinically, but no evidence was seen on the postoperative cranial computed tomographic scan. The fourth patient had worsening pharyngeal function postoperatively demonstrated only by vallecular pooling, but no cause could be identified. Table 5. Aspiration Risk Profile Point Value a Age y 0.0 Age 75 y 1.0 Diabetes 1.0 Chronic obstructive pulmonary disease 1.5 Myocardial infarction (history) 1.5 Cerebrovascular disease (history) 2.3 a Add point value s for characteristics applicable to patient. If total point value is greater than 2.0, evaluate for pharyngeal dysfunction preoperatively and postoperatively.

5 Ann Thorac Surg HARRINGTON ET AL 1998;65: POST-CABG SILENT ASPIRATION 1603 Comment Pharyngeal dysfunction after cardiac operation may be a life-threatening complication [12, 21]. Hogue and colleagues [12] described swallowing dysfunction as a significant complication of CABG, with an incidence of pulmonary aspiration of 3.6%. The aspiration was silent or unaccompanied by a cough reflex in 22% of these patients. This compares with an average incidence of 1.8 per 100 procedures from January 1992 through July 1994 at this institution. Hogue and associates [12] identified age, duration of intubation, and intraoperative use of transesophageal echocardiography as major independent risk factors for SA. They found no association with prior cerebral vascular accident [12]. Our retrospective study confirmed many of their findings, but also isolated additional factors associated with SA, namely IDDM, a history of myocardial infarction, chronic obstructive pulmonary disease, and a history of cerebrovascular disease. The most prominent historical risk factor was history of cerebral vascular accident. We also found that the intraoperative variables associated with SA were pump time, length of anesthesia, units of packed red blood cells transfused, and units of fresh-frozen plasma transfused. Our results are consistent with the findings of Hogue and coinvestigators [12] that time on an intraaortic balloon pump is also a risk factor. Only use of an intraaortic balloon pump and pump time, however, remained significant in the multivariate analysis. Unlike the study of Hogue and coworkers, none of our patients underwent transesophageal echocardiography. We realize that the specificity of the ARP is low; however, the sensitivity is high. We identified 100% of those patients in whom pharyngeal dysfunction developed. By instituting appropriate preventive measures [14], we were able to reduce the incidence of SA from 1.8% to 0%, resulting in a significant decrease in morbidity, mortality, and cost for those patients. In a detailed study of adverse cerebral outcomes after CABG, Roach and associates [22] found an incidence of 6.1%. Interestingly, the factors we found to be significant for development of SA, specifically age, a history of cerebral vascular disease, diabetes, pulmonary disease, and use of an intraaortic balloon pump, were all found by Roach and coworkers to be significant risk factors for type I cerebral outcomes in CABG patients. In our detailed prospective study of 10 patients who had an ARP of greater than 2, 3 of the 4 with postoperative pharyngeal dysfunction had objective evidence of stroke. In this study we evaluated pharyngeal function before the operation. This confirmed that swallowing dysfunction was not present before the operation. These findings suggest that a primary cause of post- CABG pharyngeal dysfunction is intraoperative cerebral injury. Careful preoperative and postoperative clinical evaluation of CABG patients with significant risk factors for stroke may result in early diagnosis of pharyngeal dysfunction and achieve significant reduction in morbidity, mortality, and hospital cost. This research was financially supported by Methodist Hospitals of Memphis and the Methodist Hospitals Foundation. We thank the Memphis Radiological Professional Corporation, and the departments of Speech Therapy, Nursing Education, and Nursing Service at Methodist Hospitals of Memphis for their support. We thank all of those who served as advisors in the development of this study: R. Neal Aguillard, MD, Tulio E. Bertorini, MD, James R. Galyean III, MD, W. Michael Leppert, MD, Mary R. McCalla, MD, and Wendy Sorgen, CCC/SP. Tulio E. Bertorini, MD, served as both an advisor and editorial assistant. References 1. Horner J, Massey EW. Managing dysphagia: special problems in patients with neurologic disease. Dysphagia 1991;89: Logemann, JA. Evaluation and treatment of swallowing disorders. San Diego, CA: College-Hill Press, Horner J, Massey EW. Silent aspiration following stroke. Neurology 1988;38: Kirshner HS. Causes of neurogenic dysphagia. Dysphagia 1989;3: Brin MF, Younger D. Neurologic disorders and aspiration. Otolaryngol Clin North Am 1988;21: Gordon C, Hewer RL, Wade DT. Dysphagia in acute stroke. Br Med J 1987;295: Elpern EH, Scott MG, Petro L, Ries MH. Pulmonary aspiration in mechanically ventilated patients with tracheostomies. Chest 1994;105: Devita MA, Speirer-Rundback L. Swallowing disorders in patients with prolonged orotracheal intubation or tracheostomy tubes. Crit Care Med 1990;18: Bonanno PC. Swallowing dysfunction after tracheostomy. Ann Surg 1971;174: Cameron JL, Reynolds J, Zuidema GD. Aspiration in patients with tracheostomies. Surg Gynecol Obstet 1973;136: Burgess GE, Cooper JR, Marino RJ, Peuler MJ, Warriner RA. Laryngeal competence after tracheal extubation. Anesthesiology 1979;51: Hogue CW, Lappas GD, Creswell LL, et al. Swallowing dysfunction after cardiac operations: associated adverse outcomes and risk factors including intraoperative transesophageal echocardiography. J Thorac Cardiovasc Surg 1995;110: Kritchevsky SB, Harrington OB, Starnes C, White P, Fleming L. Silent aspiration after coronary artery by-pass surgery: a case-control study [Abstract]. Am J Epidemiol 1996;143:S Logemann JA. The role of the speech language pathologist in the management of dysphagia. Otolaryngol Clin North Am 1988;21: Linden P, Siebens AA. Dysphagia: predicting laryngeal penetration. Arch Phys Med Rehabil 1983;64: Thompson-Henry S, Braddock B. The modified Evan s blue dye procedure fails to detect aspiration in the tracheostomized patient: five case reports. Dysphagia 1995;101: Kish, L. Survey sampling. New York: J. Wiley & Sons, 1965: Kleinbaum DG, Kupper LL, Morgenstern H. Epidemiologic research. Belmont, CA: Lifetime Learning Publications, 1982; Dean AG, Dean JA, Burten AH, Dicker RC. Epi-Info Version 5: a word processing, database, and statistics program for epidemiology on microcomputers. Stone Mountain, GA: USD, Inc, SAS Procedures Guide, Version 6, 3rd ed. Cary, NC: SAS Institute, Inc, Forshag MS, Cooper AD Jr. Postoperative care of the thoracotomy patient. Clin Chest Med 1992;13: Roach, GW, Kanchuger, M, Mangano, CM et al. Adverse cerebral outcomes after coronary bypass surgery. N Engl J Med 1996;335: