IDENTIFYING RISK FACTORS FOR POSTOPERATIVE CARDIOVASCULAR AND RESPIRATORY COMPLICATIONS AFTER MAJOR ORAL CANCER SURGERY

Transcription

1 ORIGINAL ARTICLE IDENTIFYING RISK FACTORS FOR POSTOPERATIVE CARDIOVASCULAR AND RESPIRATORY COMPLICATIONS AFTER MAJOR ORAL CANCER SURGERY Jasjit K. Dillon, BDS, MBBS, DDS, Stanley Y. Liu, DDS, Chirag M. Patel, DMD, Brian L. Schmidt, DDS, MD, PhD Department of Oral and Maxillofacial Surgery, University of California, San Francisco, San Francisco, California. Accepted 4 February 2010 Published online 19 May 2010 in Wiley Online Library (wileyonlinelibrary.com). DOI: /hed Abstract: Background. Surgical resection of oral cancer can be associated with significant postoperative cardiovascular and respiratory complications that require more sensitive predictors. Methods. All patients with oral squamous cell carcinoma treated from July 2005 to April 2008 were retrospectively reviewed. The Goldman Revised Cardiac Risk Index (GRCRI) was used to predict cardiovascular complications. Other evidence-based a priori predictors were applied in an h-fold cross-validation model. Results. Operating room (OR) time was an independent predictor of cardiovascular complications (odds ratio ¼ 1.54, p ¼.002, 95% confidence interval [CI] ¼ ) and respiratory complications (odds ratio ¼ 1.3, p ¼.06, 95% CI ¼ ) after multivariate adjustment. OR time and estimated blood loss predicted cardiovascular complications with 73% sensitivity. The GRCRI achieved 37% sensitivity. OR time and tracheostomy predicted respiratory complications with 75% sensitivity. Conclusions. The GRCRI was not prognostic for cardiovascular complications in patients with oral cancer. The most sensitive predictors for cardiovascular complications were OR time and estimated blood loss; for respiratory complications they were OR time and tracheostomy. VC 2010 Wiley Periodicals, Inc. Head Neck 33: , 2011 Keywords: Goldman criteria; oral cancer surgery; postoperative complications; cardiovascular complications; respiratory complications Correspondence to: B. L. Schmidt VC 2010 Wiley Periodicals, Inc. Oral cancer is the sixth most common cancer in the world. In the United States it accounts for 1 death every hour. 1 The long-term survival for patients with oral cancer has not changed over the last 5 decades. 2 As a result of advances in anesthesia, surgical techniques, and improved critical care, patients quality of life has significantly improved. 3,4 Primary treatment for oral cancer is surgical resection and neck dissection. In addition, microvascular free flap reconstruction is frequently required for advanced cancers. Oral cancer resection and reconstruction require several hours and are potentially associated with significant blood loss, necessitating aggressive fluid resuscitation. Postoperative complications and prolonged intensive care unit (ICU) stay are common in this patient group and have been reported to be as high as 20%. 4,5 Cardiovascular complications lead to significant mortality and morbidity. 6 Multiple international guidelines have been published to investigate and manage cardiovascular perioperative risk reduction. 7 Many cardiac and respiratory risk indices exist, but cardiac risk indices have been validated for use only in postoperative surgical patients. 8 The best-known cardiac risk index for predicting perioperative cardiovascular complications was produced by Goldman and colleagues. 9 This index was revised and validated in Although it has a predictive power of 77% for vascular surgery patients, it has not been formally used for patients undergoing major oral cancer surgery. Few studies of critical care management have outlined specific risk factors for cardiac and respiratory complications after oral cancer surgery. 4,11 13 To reduce the postoperative hospital costs, morbidity, and mortality in patients who have undergone major oral cancer surgery, we sought to determine predictors of postoperative cardiovascular and respiratory complications. We assessed the Goldman Revised Cardiac Risk Index (GRCRI) for its predictive power in our cohort (Table 1), and the strength of associations between cardiovascular and respiratory complications with patientrelated and surgery-related factors. MATERIALS AND METHODS Study Design. This retrospective study was performed at the University of California, San Francisco. All patients with biopsy-proven oral squamous cell carcinoma treated by the Oral and Maxillofacial Surgery Oncology Service from July 2005 to April Oral Cancer Surgery Complications HEAD & NECK DOI /hed January 2011

2 Table 1. Risk factors included in the Goldman Revised Cardiac Risk Index. Risk factor High-risk type of surgery Ischemic heart disease History of congestive heart failure History of cerebrovascular disease Insulin therapy for diabetes Preoperative serum creatinine >2.0 mg/dl Note: This Revised Cardiac Risk Index was constructed with each risk factor assigned 1 point. were included. The research project was approved by the Committee on Human Research. Data Collection. Departmental, hospital, operative, and anesthesia records were reviewed. Patient-related factors included age, weight, height, location of primary tumor and tumor stage, 14 respiratory disease, hypertension, ischemic heart disease (IHD), congestive heart failure, previous cardiac procedure, b-blocker use, hypercholesterolemia, tobacco and alcohol use, cerebrovascular accident, diabetes, preoperative creatinine, and American Society of Anesthesiology (ASA) class. 15,16 Surgery-related factors included primary surgical site, neck dissection, microvascular free flap reconstruction, tracheostomy, preoperative and postoperative hematocrit, arterial blood gas (ABG) readings, estimated blood loss (EBL), intraoperative fluids including blood products administered, and operating room (OR) time. Data were exported to STATA (version 9.2) for analysis. Outcome Definition. Cardiovascular complications were defined as myocardial infarction (MI), arrhythmia, and persistent hypertension necessitating treatment. Respiratory complications were defined as respiratory distress requiring re-intubation, respiratory distress not requiring re-intubation, and pneumonia. Complications were coded as binary variables. Data Analysis. Analyses were conducted using STATA (version 9.2). Descriptive analysis was performed to identify normality, linearity, and outliers. The strength of univariate associations was examined using logistic regression, with separate models for each predictor (determined a priori). Multiple logistic regression analysis was used to find independent associations between a priori predictors and postoperative cardiovascular and respiratory complications. Univariate predictors whose odds ratios suggest association with complications (p.10) were entered into the final multivariate model. We determined a priori that 2 predictors from patient-related factors and 2 predictors from surgeryrelated factors would be assessed for the strength of their associations with complications. Patient-related predictors for cardiovascular complications were age and IHD. Surgery-related predictors for cardiovascular complications were OR time and EBL. Patient-related predictors for respiratory complications were preexisting respiratory disease and IHD. Surgery-related predictors for respiratory complications were OR time and tracheostomy. These a priori predictors have been cited as contributors to postoperative complications. 4,17 The predictors were assessed for interactions and satisfaction of the linearity assumption. Finally, the multivariate models were assessed for prediction error via the h- fold cross-validation method, with data divided into 10 mutually exclusive subsets. Receiver operating characteristic curves (ROCs) were generated based on the h- fold cross-validation. RESULTS In all, 92 patients were reviewed (Table 2). There were no statistically significant differences between age and sex of patients presenting with and without postoperative complications. Fourofthe6indicesfortheGRCRIwereusedina multivariate logistic model to predict for postoperative cardiovascular complications in our cohort. The categories of major surgery and creatinine >2.0 were not scored in our cohort because oral cancer surgery is not considered a major surgery on the index, and none of our patients had a creatinine >2.0. The model did not reach statistical significance (p ¼.54), and when used as Table 2. Cohort characteristics. Characteristic Numerical value Total patients, n 92 Males, n 57 Females, n 35 Age, mean SD, y Cardiovascular complications, n 23 Age of patients with cardiovascular complications, y Respiratory complications, n 20 Age of patients with respiratory complications, mean SD, y Comorbidities Hypertension, % 53 Smoking, % 53 Alcohol, % 49 Hypercholesterolemia, % 33 Ischemic heart disease, % 22 Respiratory diseases, % 15 Tumor stage (TNM) I, % 47 II, % 19 III, % 4 IV, % 28 Tumor location Tongue, % 37 Mandibular gingival, % 31 Maxillary gingival, % 13 Floor of mouth, % 8 Palate, % 4 Buccal mucosa, % 5 Note: Characteristics of the cohort include comorbidities, tumor staging, and tumor location. Oral Cancer Surgery Complications HEAD & NECK DOI /hed January

3 patient-related factors using logistic regression, age was not associated with cardiovascular complications. The association was not significant with IHD. With surgeryrelated factors using univariate logistic analysis, only OR time was significantly associated with cardiovascular complications. After adjustment with EBL, OR time remained independently associated with cardiovascular complications (Table 3). When using the model with EBL and OR time, prediction error of the h-fold crossvalidation trial, equivalent to the area under the ROC curve, was 0.75 (Figure 1B). The pseudo R 2 is 0.20 and the sum of squared residuals is 18 for this multivariate model (R 2 ¼ 0.28 and chi-square ¼ 26 without h-fold cross-validation). Respiratory Complications Analysis. In univariate models for patient-related factors using logistic regression, tracheostomy and OR time had significant associations with respiratory complications. They were entered into the multivariate analysis. After adjustment, OR time remained independently associated with respiratory complications (Table 3). When this model was tested for its validity using the h-fold cross-validation trial, the area under the ROC curve was 0.73 (Figure 1C). The pseudo R 2 is 0.13 and the sum of squared residuals is 12 for this multivariate model (R 2 ¼ 0.19 and chi-square ¼ 17 without h-fold cross-validation). FIGURE 1. Receiver operating characteristic (ROC) curves depicting predictive power of models assessed in this study. (A) Cardiovascular complications predicted using Goldman Revised Cardiac Risk Index (GRCRI). (B) Cardiovascular complications predicted by operating room time and estimated blood loss. (C) Respiratory complications predicted by operating room time and tracheostomy. a prediction model with h-fold cross-validation, the resulting ROC curve had a negative predictive trend with a sensitivity of 37% (Figure 1A). Cardiovascular Complications Analysis. Before construction of logistic modeling, linear assumptions of age and EBL were confirmed by the LOWESS (locally weighted least squares) curve and quadratic term analysis. There were no significant interactions among age, OR time, and EBL. In single predictor models for DISCUSSION The purpose of this study was to determine predictors of postoperative cardiovascular and respiratory complications after major oral cancer surgery, with the aim of using this information to improve patient outcomes by reducing postoperative hospital costs, morbidity, and mortality. Our study population had a cardiovascular complication rate of 25% and a respiratory complication rate of 22%. Our rate of postoperative complications was higher than that of previous reports on patients with head and neck cancer, most likely because we defined the complications more broadly. 4,16 For example, no head and neck study to date has defined postoperative hypertension as a complication. Uncontrolled hypertension in the postoperative period is not only detrimental to the surgical site (eg, excessive bleeding or damage to the microvascular free flap), but also adds to patient morbidity by increasing the risk of perioperative ischemia and stroke. Goldman and colleagues 9 produced the most established cardiac risk index to predict for perioperative cardiovascular complications. The cohort consisted of 1001 patients. The scoring system was complex and based on 9 predictors, each given various weightings that allowed physicians to divide patients into 4 groups. This index did not take into account today s advances in critical care medicine, anesthesia, or surgery, and was subsequently validated only on 114 Oral Cancer Surgery Complications HEAD & NECK DOI /hed January 2011

4 Table 3. Strength of associations. Odds ratio, p value (95% CI) Predictor Univariate analysis Adjusted analysis Cardiovascular complications Age, every 10 years 1.00,.95 ( ) 1.00,.33 ( ) Ischemic heart disease 1.40,.58 ( ) 0.88,.88 ( ) Operating room time, every hour 1.40,.001* ( ) 1.50,.002* ( ) Estimated blood loss 1.00,.036* ( ) 1.00,.73 ( ) Respiratory complications Ischemic heart disease 1.90,.29 ( ) 2.90,.15 ( ) Operating room time, every hour 1.40,.003* ( ) 1.30,.06* ( ) Tracheostomy 6.60,.001* ( ) 3.00,.14 ( ) Abbreviation: CI, confidence interval (range). Note: Univariate and multivariate analyses of predictors with significant strengths of association to cardiovascular and respiratory complications. general surgery patients. Lee and Goldman produced a simpler index for patients undergoing nonurgent major noncardiac surgery known as the Goldman Revised Cardiac Risk Index. 9,18 These authors looked at 6 independent risk factors: IHD, congestive heart failure, cerebrovascular disease, high risk surgery (of which head and neck surgery was not classified), preoperative insulin treatment for diabetes mellitus, and preoperative creatinine >2 mg/dl (Table 1). Rather than applying weightings to the risk factors, the overall risk was classified in accord with the number of risk factors. Patients without any risk factors were assigned to the lowest risk class (I) and were found to have cardiovascular complication rates of 0.5% and 0.4% in the derivation and validation cohorts. Patients with 3 risk factors were assigned to the highest risk class (IV) and were found to have cardiovascular complication rates of 9.1% and 11%. The revised index outperformed the original Goldman index and had the most predictable outcome for postoperative cardiovascular complications in patients undergoing vascular surgery with a predictive value of 77%. 10 This is the first study in which the GRCRI was used to assess the postoperative cardiovascular risk specifically in patients with oral cancer. The GRCRI achieved a sensitivity of 37%, with a nonlinear relationship between score and complication; thus, we were unable to demonstrate any predictive value for the GRCRI for our heterogeneous cohort of patients undergoing major oral cancer surgery. Since the GRCRI did not result in a clinically acceptable sensitivity, we constructed a model based on the complications associated with oral cancer surgery. Given the small sample size and limited number of outcomes in our study, it was determined a priori that only 2 predictors from surgery-related factors and 2 predictors from patient-related factors would be assessed for their contributions in univariate and multivariate analyses. Prior to commencement of the study we did a search of the literature and looked at all the possible predictors for complications. Several predictors were consistent in the literature, and we restricted our model to the 2 most common surgeryrelated and patient-related factors for cardiovascular and respiratory complications. For cardiovascular complications the surgery factors were EBL and OR time and the patient factors were age and IHD. For respiratory complications the surgery factors were tracheostomy and OR time and the patient factors were respiratory diseases and IHD. We found that the surgery-related factors were more predictive of postoperative cardiovascular and respiratory complications than the patient-related factors. Based on the strength of associations we identified OR time and EBL to be predictive of cardiovascular complications. The sensitivity for these 2 factors was 75%, making them clinically suitable predictors for postoperative cardiovascular complications. For respiratory complications we identified OR time, IHD, and tracheostomy to be predictive. The sensitivity with these 3 factors was 73%. Therefore, using this multivariate analysis with a priori predictors our models achieved a sensitivity of 75% for cardiovascular complications and 73% for respiratory complications. The predictive accuracy of all our multivariate models was strengthened using the h-fold cross-validation. We found that of the surgery-related factors, OR time is most significantly associated with both cardiovascular and respiratory complications. Previous reports show that 8 hours is the threshold after which the complication rate increases significantly. 11 There are numerous reasons for the association between OR time and complications. Patient-related factors such as a prior history of head and neck radiation treatment is an example. The longer the OR time, the longer mechanical ventilation is required. Prolonged mechanical ventilation has both cardiovascular and respiratory consequences. Fluid management becomes more complex with increasing length of surgery and EBL. Although in our study EBL is not highly significant for cardiovascular complications on adjusted analysis, the univariate analysis and the adjusted OR time analysis suggest that it did influence the outcome, increasing the sensitivity from 44% to 54%. For OR time to be established as an independent predictor of postoperative complications, more surgeryrelated factors need to be considered in a multivariate analysis. Oral Cancer Surgery Complications HEAD & NECK DOI /hed January

5 There are several limitations to this study. The small number of outcomes limits this analysis to only a handful of a priori predictors. There needs to be increased power if other potential predictors are to be included in the model, along with more stringent definitions of predictors and outcomes that will increase precision (shorter 95% CI). 19 Second, we were also limited by the retrospective nature of the study, depending entirely on the data that had been entered into the patient records. Despite these limitations, many of which are inevitable in a retrospective chart review study, we believe that the central conclusions remain valid. We were not able to show that the GRCRI is predictive of postoperative complications in our heterogeneous oral cancer cohort. Our study shows that surgery-related factors are more predictive than patient-related factors, suggested by our findings with EBL and OR time. We acknowledge that there needs to be additional analysis involving confounders of EBL and OR time, to establish whether these factors can be used as surrogate markers of postoperative complications in patients with oral cancer. REFERENCES 1. Kademani D. Oral cancer. Mayo Clin Proc 2007;82: Bell RB, Kademani D, Homer L, Dierks EJ, Potter BE. Tongue cancer: is there a difference in survival compared with other subsites in the oral cavity? J Oral Maxillofac Surg 2007;65: Young CW, Pogrel MA, Schmidt BL. Quality of life in patients undergoing segmental mandibular resection and staged reconstruction with nonvascularized bone grafts. J Oral Maxillofac Surg 2007;65: Buitelaar DR, Balm AJ, Antonini N, van Tinteren H, Huitink JM. Cardiovascular and respiratory complications after major head and neck surgery. Head Neck 2006;28: Clark JR, McCluskey SA, Hall F, et al. Predictors of morbidity following free flap reconstruction for cancer of the head and neck. Head Neck 2007;29: Fleisher LA, Beckman JA, Brown KA, et al. ACC/AHA 2007 Guidelines on perioperative cardiovascular evaluation and care for noncardiac surgery. Executive Summary, J Am Coll Cardiol 2007:e Fleisher LA, Beckman JA, Brown KA, et al. ACC/AHA 2007 guidelines on perioperative cardiovascular evaluation and care for noncardiac surgery: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Revise the 2002 Guidelines on Perioperative Cardiovascular Evaluation for Noncardiac Surgery) developed in collaboration with the American Society of Echocardiography, American Society of Nuclear Cardiology, Heart Rhythm Society, Society of Cardiovascular Anesthesiologists, Society for Cardiovascular Angiography and Interventions, Society for Vascular Medicine and Biology, and Society for Vascular Surgery. J Am Coll Cardiol 2007;50:e159 e Lawrence VA, Dhanda R, Hilsenbeck SG, Page CP. Risk of pulmonary complications after elective abdominal surgery. Chest 1996;110: Goldman L, Caldera DL, Nussbaum SR, et al. Multifactorial index of cardiac risk in noncardiac surgical procedures. N Engl J Med 1977;297: Lee TH, Marcantonio ER, Mangione CM, et al. Derivation and prospective validation of a simple index for prediction of cardiac risk of major noncardiac surgery. Circulation 1999;100: Farwell DG, Reilly DF, Weymuller EA Jr, Greenberg DL, Staiger TO, Futran NA. Predictors of perioperative complications in head and neck patients. Arch Otolaryngol Head Neck Surg 2002;128: Pelczar BT, Weed HG, Schuller DE, Young DC, Reilley TE. Identifying high-risk patients before head and neck oncologic surgery. Arch Otolaryngol Head Neck Surg 1993;119: Rao MK, Reilley TE, Schuller DE, Young DC. Analysis of risk factors for postoperative pulmonary complications in head and neck surgery. Laryngoscope 1992;102: O sullivan B, Shah J. New TNM staging criteria for head and neck tumors. Semin Surg Oncol 2003;21: Wolters U, Wolf T, Stutzer H, Schroder T. ASA classification and perioperative variables as predictors of postoperative outcome. Br J Anaesth 1996;77: de Cassia Braga Ribeiro K, Kowalski LP. APACHE II, POSSUM, and ASA scores and the risk of perioperative complications in patients with oral or oropharyngeal cancer. Arch Otolaryngol Head Neck Surg 2003;129: Ong SK, Morton RP, Kolbe J, Whitlock RM, McIvor NP. Pulmonary complications following major head and neck surgery with tracheostomy: a prospective, randomized, controlled trial of prophylactic antibiotics. Arch Otolaryngol Head Neck Surg 2004;130: Goldman L. Cardiac risk in noncardiac surgery: an update. Anesth Analg 1995;80: Magder LS, Hughes JP. Logistic regression when the outcome is measured with uncertainty. Am J Epidemiol 1997;146: Oral Cancer Surgery Complications HEAD & NECK DOI /hed January 2011