Duration of anesthesia vs risk of any postoperative complication.
Farwell DG, Reilly DF, Weymuller, Jr EA, Greenberg DL, Staiger TO, Futran NA. Predictors of Perioperative Complications in Head and Neck Patients. Arch Otolaryngol Head Neck Surg. 2002;128(5):505-511. doi:10.1001/archotol.128.5.505
Copyright 2002 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2002
Patients with significant medical problems requiring major otolaryngology procedures are at high risk for both medical and surgical complications.
To identify risk factors associated with perioperative complications in medically compromised patients undergoing major otolaryngology procedures.
Ninety-three consecutive patients with significant comorbid medical illnesses (eg, diabetes, hypertension) undergoing major head and neck surgical procedures were referred to a medical consultation center for preoperative assessment and medical management. Patient and surgical characteristics as well as perioperative complications were identified and recorded. Univariate and multivariate analyses were performed to determine which characteristics were associated with complications.
Thirty-two patients (34%) had postoperative complications. Twenty-six patients (28%) had serious medical complications, and 18 (19%) had surgical complications. No deaths occurred in the study population. On univariate analysis, the factors associated with all complications included history of hepatitis, flap reconstruction, oncologic surgery, preoperative radiation therapy, preoperative gastrostomy placement, intraoperative transfusion, anesthesia time (≥8 hours), and those with greater intraoperative fluid replacement and estimated blood losses. Only anesthesia time (≥8 hours) remained independently significant on multivariate analysis. A history of hepatitis and prolonged anesthesia time were the only independent predictors of medical complications. The only independent predictor of surgical complications was the volume of intraoperative fluid administered.
Prolonged anesthesia times of 8 hours or more, a history of hepatitis, and large-volume intraoperative fluid resuscitations predicted adverse outcomes. Special care must be taken in counseling these patients preoperatively and in caring for them during their operative and postoperative course.
MAJOR otolaryngology procedures are often complicated by challenging anatomy, complex reconstructions, and long surgical procedures. Additionally, many patients undergoing these procedures have comorbid medical conditions that complicate their care and may predispose them to perioperative complications. These patients may frequently have a history of tobacco and/or alcohol abuse. As such, they may be at risk for significant cardiac and pulmonary disease. Proper assessment of the risk of perioperative complications is an important part of the preoperative planning and counseling, as these comorbidities often have significant impact on the perioperative course.
Different authors have evaluated a variety of patient and surgical factors associated with adverse perioperative events.1- 5 Within the head and neck literature, several authors have evaluated complications and noted the importance of comorbidities in predicting postoperative complications.6,7 These have included radiation exposure, operative time, hypertension, alcohol abuse, and the Charlson comorbidity score.6- 12 Piccirillo13 and Singh et al14 recently demonstrated that medical comorbidities are an independent predictor of survival in patients with head and neck cancer. However, the specific risk of complications in patients with significant medical comorbidities undergoing major otolaryngology procedures remains incompletely explored. The aim of this study was to examine this issue more completely, ultimately allowing us to better predict which patients will develop perioperative complications.
The study was performed at the University of Washington Medical Center, Seattle. All otolaryngology–head and neck surgery patients 18 years and older with significant medical problems were referred to the medical consultation service for preoperative evaluation. Study participants provided written informed consent and the project was approved by our institutional review committee.
One hundred forty-one consecutive patients considered at risk for perioperative complications were referred preoperatively to the medical consultation center for evaluation prior to otolaryngology–head and neck surgery from October 17, 1995, to June 9, 1997. These patients represented approximately 6% of the otolaryngology–head and neck surgery patients undergoing surgical procedures. Patients were referred for preoperative evaluation by the surgical team or anesthesiologists when there were concerns about medical comorbidities. These concerns included active or presumed cardiovascular, respiratory, or endocrine diseases; significant deconditioning; and advanced age. Of those patients, 134 (95%) consented to participate in a study designed to examine the impact of self-reported exercise tolerance on the risk of serious postoperative complications. Our study population is a subgroup of the otolaryngology patients in that study.3
Of the eligible patients, 26 (19%) were determined to have minor procedures and were excluded from the study. Minor procedures were associated with minimal risk and included biopsies, panendoscopy, vocal cord medializations, and nasal reconstructions. Of the remaining 108 patients having major procedures, 15 patients (14%) had their procedure canceled, leaving 93 patients to form our study population. Patients whose procedures were canceled generally opted for alternative therapy (chemotherapy/radiation therapy) or no treatment.
Each patient was evaluated and managed preoperatively and postoperatively by a team of physicians that included an attending otolaryngologist and internist. Outside records were sought and preoperative testing was ordered as clinically indicated. There was no additional testing specifically performed as part of this study.
The patient's medical problems, habits, and exercise tolerance were prospectively recorded by the internal medicine team and cataloged in a computer database. Information about the nature of the surgical procedure, reconstruction technique used, estimated intraoperative blood loss, intraoperative transfusions, and preoperative radiation or chemotherapy was collected retrospectively by review of the medical record. The American Society of Anesthesiologists (ASA) classification was assigned by the attending anesthesiologist on the day of surgery. The duration of anesthesia and ASA classification were obtained from the department of anesthesia's existing quality improvement database or the anesthesia record.
Prior to beginning the study, a list of 26 serious perioperative medical complications and their definitions was established.3 They included those of a cardiovascular, pulmonary, neurologic, infectious, and miscellaneous nature. Only events that caused the patient to require additional therapy, placed the patient at increased risk, or prolonged the length of stay were included. Minor complications, such as urinary tract infections and atelectasis, were excluded. Medical complications were recorded from entry into the preanesthesia area until the time of discharge from the hospital. Two attending internists blinded to the study design and results of the preoperative evaluation reviewed each record to assure that it met the predefined criteria. In instances when the 2 internists disagreed on the diagnosis, a third physician adjudicated the case. A second category of surgical complications was devised after the surgical procedures had been completed. These included wound breakdown, fistula formation, flap donor and recipient site complications, flap failure, wound hematomas, and the need for additional (unexpected) procedures. These complications were identified by review of the medical record by one of us (D.G.F.). Both immediate and delayed surgical complications were recorded. Delayed complications, such as wound breakdown, were often not identified until patients' outpatient otolaryngology clinic visits. Follow-up data were available for all patients.
Statistical analysis was performed with SAS, version 6.12, statistical software (SAS Institute Inc, Cary, NC). Values are reported as mean ± SD. We considered P≤.05 to be significant. The χ2 tests, including test for trend for ordinal data, or the Fisher exact test was used to compare categorical variables between groups. All P values shown are for χ2 tests, unless otherwise specified. The Wilcoxon rank sum test was used to compare continuous variables between groups. Odds ratios estimating the risk of any perioperative complication were calculated using logistic regression for each variable found to be significant on univariate analysis. To determine if these variables predicted perioperative complications independent of one another, we also added them to a multivariable logistic regression model in a stepwise manner.
Ninety-three patients, referred to the medical consultation service for their comorbid diseases, underwent major otolaryngology procedures. The vast majority (84%) of these cases were oncology related (Table 1). The mean ± SD age of the study population was 65.8 ± 13.0 years. The majority of patients were male (66%) and white (91%). Most had smoked at least 20 pack-years (65%) and 13% had a history of active alcohol abuse. The majority of patients (55%) were ASA class III.
Of the 78 oncology-related cases, nearly half received a flap reconstruction as part of their surgical procedure (n = 36) (Table 1). There were 12 radial forearm, 6 fibula, 5 iliac crest, 3 rectus, 1 scapula, 1 jejunum, and 1 latissimus free flaps used. An additional 13 pectoralis rotational flaps were used in this study population. Completing the data set were patients undergoing sinus and endoscopic skull-base procedures (n = 9), otologic procedures (n = 3), uvulopalatopharyngoplasty (n = 2), and a cricopharyngeal myotomy (n = 1). The majority of patients undergoing sinus and endoscopic skull-base procedures had significant pulmonary problems (Samter triad) prompting the preoperative involvement of the medical consult service.
Medical complications occurred in 26 patients (Table 2). These included 7 cardiac, 17 pulmonary, 13 neurologic, 5 serious infections (defined as deep surgical site infection, wound abscess, bacteremia, or sepsis), and 8 miscellaneous events. Fifteen patients had more than 1 medical complication. There were no deaths in the study population. Thirteen patients were unexpectedly transferred to the intensive care unit or the cardiac telemetry ward. One patient required activation of the hospital emergency response system (code blue).
Eighteen patients (19%) had a surgical complication. These included 12 patients with wound breakdowns, 4 with oral cutaneous fistulas, 4 hematomas, 4 donor and 4 flap recipient site complications, and 12 patients who required an additional unanticipated procedure.
On univariate analysis (χ2), the risk for a patient to develop any complication (medical and surgical combined) was correlated with a history of hepatitis, oncologic surgery, undergoing a flap reconstruction, preoperative radiation therapy, preoperative gastrostomy placement, intraoperative transfusion, smoking within the previous 6 weeks, and an anesthesia time of 8 hours or more (P = .05, χ2) (Table 1). Using univariate regression analysis, those same factors, with the exception of preoperative radiation therapy, were found to be significantly associated with all complications (Table 3). All patients who had a gastrostomy tube placed preoperatively had a complication, making regression analysis of that group (n = 3) impossible. Upon multivariate analysis, only an anesthesia time of 8 hours or more was an independent predictor of all complications. (Table 3). After 8 hours of anesthesia, the complication rate increased dramatically and rose to 100% after 16 hours of anesthesia (Figure 1).
Patients with any complication received more intraoperative fluid and more intraoperative units of blood, had a greater estimated blood loss, and had longer procedures (Table 1). However, after correction for the duration of surgery, none of these factors was significantly associated with all complications.
Patients with any complication had longer lengths of stay (14.2 ± 11.2 vs 4.7 ± 3.9 days; Wilcoxon rank sum, P<.001), had longer stays in the intensive care unit (6.1 ± 8.5 vs 0.7 ± 1.1 days; Wilcoxon rank sum, P<.001), required more days of ventilator support (3.0 ± 7.9 vs 0.07 ± 0.3 days; Wilcoxon rank sum, P<.001), and had a longer period until tracheostomy decannulation (15.5 ± 25.3 vs 3.1 ± 13.3 days; Wilcoxon rank sum, P<.001).
Advanced age, ASA classification, poor exercise tolerance, history of smoking 20 pack-years, alcohol abuse, preoperative chemotherapy, weight, and the other patient and surgical characteristics were not associated with an increased risk for all complications.
When evaluating medical complications, a history of hepatitis, undergoing a flap reconstruction, preoperative gastrostomy placement, intraoperative transfusion, and a prolonged anesthesia time (≥8 hours) were significant by univariate analysis (P<.05, χ2) (Table 4). Patients with medical complications received a greater amount of intraoperative fluid (5.7 ± 2.6 vs 3.4 ± 2.6 L; Wilcoxon rank sum, P<.001), received more units of blood intraoperatively (0.61 ± 0.92 vs 0.16 ± 0.59 U; Wilcoxon rank sum, P = .006), and had greater estimated intraoperative blood loss (0.50 ± 0.42 vs 0.32 ± 0.38 L; Wilcoxon rank sum, P = .012). Upon multivariate analysis a history of hepatitis and a prolonged anesthesia time of 8 hours or more were the only independent predictors of medical complications (Table 4).
A history of hepatitis was an independently significant predictor of adverse outcomes despite the fact that only 5 patients had a medical diagnosis of a history of hepatitis. Three of these patients had a history of remote active hepatitis and jaundice; 1 patient had evidence of cirrhosis. The exact etiology of hepatitis was not confirmed by serologic testing in these patients. Of those 5 patients, 4 had complications including myocardial ischemia, delirium (2 patients), acute renal insufficiency, wound infections (2 patients), pneumonia, alcohol withdrawal, prolonged postoperative ventilation, hypoxia, and a fall.
On subgroup analysis (data not shown), alcohol use (>150 mL per week) was associated with both alcohol withdrawal and infectious complications (P<.05), but otherwise not associated with any specific medical complications. Smoking was nearly significant in its association with an adverse medical complication (Fisher exact test, P = .056). Additionally, congestive heart failure, advanced age, and diabetes were not associated with medical complications.
Factors associated with surgical complications on univariate analysis included oncologic surgery, preoperative radiation therapy, performance of a flap, smoking within the last 6 weeks, intraoperative transfusion, and anesthesia time of 8 hours or more (P≤.05) (Table 5). Additionally, patients with surgical complications received more intraoperative fluid (6.6 ± 2.5 vs 3.4 ± 2.5 L; Wilcoxon rank sum, P<.001), had greater estimated intraoperative blood loss (0.57 ± 0.35 vs 0.32 ± 0.39 L; Wilcoxon rank sum, P<.001), and received more units of blood intraoperatively (0.61 ± 0.92 vs 0.16 ± 0.59 U; Wilcoxon rank sum, P<.001). However, after multivariable analysis, greater intraoperative fluid administration was the only independent predictor of surgical complications (Table 5).
Patients undergoing major head and neck surgical procedures frequently have significant comorbidities that increase their risk for perioperative complications. It is imperative to accurately counsel these patients preoperatively on the risks inherent to the procedure and the risks associated with their medical comorbidities. In our study population we have identified a number of factors that are associated with an increased risk for both medical and surgical complications. In particular, patients undergoing prolonged procedures (≥8 hours), patients with a history of hepatitis, and those who received greater volumes of fluid intraoperatively were at higher risk for postoperative complications. Interestingly, several factors commonly believed to predict adverse outcomes, such as advanced age and diabetes, did not prove significant in our evaluation. It is possible that the highly select nature of our referred population may have contributed to advanced age not being significant. These patients were approximately 20 years older than the nonreferred patients. However, when controlling for other comorbidities associated with advanced age, age itself did not prove significant.
The ASA classification system, considered by many to be one of the best methods for assessing patient risk, did not perform well in our patient population. As such, our findings are in agreement with Arriaga et al.5 However, this more likely reflects the fact that the majority of our patients had medical comorbidities making them ASA class II or III. This skewed the data, making it difficult to establish significance.
Our finding of an independent risk of medical complications in patients with a history of hepatitis is useful information that was not readily available in the head and neck surgery literature. While there were few patients in our study population who had a diagnosis of hepatitis, our data are compelling in that those patients need to be carefully counseled as to their increased risk. Given the small numbers of patients in our study with hepatitis, further evaluation of this risk factor in other populations should be considered.
The relationship between the duration of the procedure and risk has been long debated. Haljamae15 has shown that the duration of anesthesia influences the incidence of postoperative complications but suggests that this incidence might reflect the severity of the underlying disease and the extent of the surgery performed instead of some unique characteristic of a prolonged exposure to anesthetics. Likewise, longer operations are associated with more extensive disease, which might be the cause of the higher complication rate in long anesthesia times. Schusterman and Horndeski16 performed subgroup analysis on their retrospective evaluation of flap reconstructions and were not able to show anesthesia time as a predictor of complications. However, the Singh et al6 study and our data clearly show that the duration of the anesthesia time is a powerful predictor of adverse events after correcting for other patient and surgical factors. For procedures lasting less than 8 hours, the risk of complications remains low at our institution.
When trying to predict which patients are at higher risk for surgical complications, we have shown that greater estimated blood loss, oncologic surgery, preoperative radiation therapy, performance of a flap, recent smoking, intraoperative transfusions, excessive fluid administration, and prolonged procedures are all significant on univariate analysis. On multivariate analysis, surgical complications were associated only with the volume of intraoperative fluid replacement. This is in contrast to all complications and medical complications, where the duration of anesthesia proved independently significant but the volume of fluid administered did not.
Evaluating the interactions between prolonged anesthesia times, blood loss, intraoperative transfusion, and fluid replacement is challenging. While it is clear that intravenous fluid replacement and anesthesia time are closely linked, it is important to coordinate a plan with the anesthesiologist for judicious fluid replacement. The mechanism of overhydration causing surgical complications is not immediately clear. In previous retrospective studies at our institution, prolonged anesthesia was associated with complications while fluid replacement was not.2 While there has been little published in head and neck surgery literature on the effect of fluid replacement and complications, Nishi et al17 have suggested that fluid restriction may reduce the risk of complications in esophagectomy patients. One could hypothesize that tissue edema increases wound tension and compromises healing. This is an area that will require future research to fully understand the effect of fluid replacement on surgical complications.
Undergoing a flap reconstruction was also associated with complications of both medical and surgical types. When controlling for other predictors, however, a flap reconstruction did not prove to be an independent predictor on multivariate analysis. It is likely that the true causes of increased complications are the additional time required to perform the flap and the observation that larger tumors, which require the flap reconstructions, tend to present in sicker patients.
In an effort to reduce the complication rate from flap reconstructions, efforts should be undertaken to efficiently perform the surgery to minimize the surgical time. Utilizing 2 teams to perform the oncologic resection and the flap harvest can dramatically reduce operative time and potentially the need for excessive fluid administration. Having a coordinated operative team with anesthesiologists and nurses committed to efficiently starting and proceeding through the case is also important in minimizing anesthesia times. To help achieve that goal, we have created a task force to improve on-time starts and to shorten operative times.
Despite the high complication rate in our study (34%), it is important to note that we had no deaths. We have coordinated the care of these patients with a team of attending internists who assist us in the preoperative workup and through the perioperative treatment. It is our opinion that involving them in the patient's care has dramatically improved our outcomes. With these new data on perioperative predictors, we hope to be able to more accurately counsel patients and improve our ability to plan treatments and prevent these complications.
Accepted for publication October 26, 2001.
This work was presented at the Fifth International Conference on Head and Neck Cancer, San Francisco, Calif, July 30, 2000.
Corresponding author and reprints: D. Gregory Farwell, MD, University of Washington, Department of Otolaryngology–Head and Neck Surgery, Box 356515, Seattle, WA 98195-6515.