Bhattacharyya N, Fried MP. Benchmarks for Mortality, Morbidity, and Length of Stay for Head and Neck Surgical Procedures. Arch Otolaryngol Head Neck Surg. 2001;127(2):127-132. doi:10.1001/archotol.127.2.127
To determine benchmarks and factors that influence morbidity, mortality, and length of stay (LOS) for head and neck surgical procedures.
A specific database for head and neck surgical procedures was extracted from the National Hospital Data Survey database for 1995 through 1997. Records of inpatient admissions for patients undergoing head and neck surgery were examined to determine demographics, incidence of medical complications, mortality, and LOS. Data were statistically examined for the influence of medical complications on mortality and LOS.
A total of 3932 patients underwent head and neck procedures. The overall medical morbidity and mortality rates were 5.65% and 2.98%, respectively. The presence of a major medical complication increased the odds of death by 5.65 (P<.001). Postoperative pneumonia was the most common medical complication (3.26%) and was associated with a mortality rate of 10.94% (odds ratio for mortality, 4.4). Acute myocardial infarction and stroke were rare (combined incidence, 1.86%) and were not statistically associated with increased mortality. Procedures that involved the esophagus carried the highest mortality rate (8.38%). The overall mean LOS was 6.15 days. This increased to 17.7 days when a major medical complication occurred (P<.001). Increasing age was associated with increased rates of medical complications and mortality (P<.001).
Medical complications in patients undergoing head and neck surgical procedures are associated with increased mortality rates and longer LOS. Mortality rates in head and neck surgery are low but may be improved by preventing medical complications.
AS PART of the ongoing trend toward ambulatory surgery, many otolaryngology procedures have shifted from the inpatient arena to the outpatient setting. The remaining procedures performed in the inpatient setting, therefore, tend to reflect more difficult or extensive procedures or a more ill patient population. Many of the procedures performed in the realm of head and neck oncology qualify on both fronts. Most patients with head and neck tumors present in the later decades of life and have risk factors for cardiopulmonary or vascular disease, usually in the setting of tobacco and/or alcohol use. This places them at higher risk for anesthesia, surgery, and postoperative complications. Preoperative counseling for these patients should include a discussion of such potential complications, and accurate counseling requires knowledge of the incidence and relative risks for such events. Several published reports1- 5 have examined the incidence of various comorbidities and postoperative complications for patients with head and neck tumors. However, most of these studies are conducted within a single academic medical center and focus on specific patient populations or a select few complications. To determine reliable benchmarks, a large database of head and neck surgical procedures would need to be examined, sampling a wide range of institutions, thereby eliminating institutional and selection biases.
The National Hospital Discharge Survey (NHDS) is an annual survey conducted by the National Center for Health Care Statistics, which is a public agency charged with tracking health care utilization on a yearly basis across the United States. From the aggregate group of all US hospitals, representative institutions are randomly selected to provide data for the NHDS. A sample of discharge records from this national sample of nonfederal hospitals from all 50 states and the District of Columbia are reviewed, and data are collected both manually and electronically by trained staff members. The sample contains representative proportions of inpatient hospital admissions spanning all services, diagnoses, hospital sizes, and geographic locations. Community hospitals, teaching hospitals, and tertiary care centers are all sampled. Patients undergoing ambulatory surgery are excluded. The data collected include age, sex, ethnic background, geographic region, length of stay (LOS), discharge status, and up to 7 diagnoses and 4 procedures. Procedures and diagnoses are coded according to the system of the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM).6 This data set has been frequently used to determine benchmarks for incidence of disease, rates of surgery, and annual trends. In addition, it has been used to investigate mortality and morbidity in both surgical and medical patient populations.7- 9 In the present study, the NHDS data were adapted to encompass a standard battery of head and neck surgical procedures to determine benchmarks for morbidity, mortality, and LOS in head and neck surgery.
The NHDS databases for each of the years 1995 through 1997 were combined into a composite database, consisting of more than 750 000 non-newborn inpatient admissions. A list of head and neck surgical procedure codes was compiled that directly reflected the head and neck procedures section of the American Board of Otolaryngology's operative log, which is completed by each graduating otolaryngology resident. The 1997 version of the operative log was used. From the NHDS database, cases were extracted by matching any 1 of 4 procedure codes for each patient to any of the head and neck surgical procedure codes derived from the operative log. This formed a database in which each patient had at least 1 head and neck surgical procedure during their admission. A second more selective database was constructed in which the primary procedure was a head and neck surgical procedure. Further data processing and statistical analyses were conducted for each of these 2 databases.
The data were examined using the SPSS statistical package system (SPSS Inc, Chicago, Ill). From the ICD-9-CM diagnosis codes, categorical variables that reflected the following comorbid medical complications were constructed: pulmonary embolism, acute myocardial infarction, acute cerebrovascular accident, pneumonia, and acute renal failure. The term complication used herein refers to one of these "medical" morbidities incurred during the admission. The morbidity rate refers to the rate of occurrence of 1 or more of these complications in the population.
Descriptive statistics for the patient population were computed. In addition, statistics for LOS, incidence of morbidities during hospitalization, discharge status, and mortality were computed for the entire database of head and neck surgical procedures. Head and neck procedures were subclassified according to primary surgical site, roughly paralleling the American Joint Committee on Cancer staging system. Surgical site groupings are listed in Table 1. The incidences of acute morbidities (complications) listed herein were tabulated. The Pearson χ2 statistic was used to determine associations between morbidities and mortality. Where appropriate, odds ratios for mortality were calculated. Univariate analysis of variance (ANOVA) was used to determine the influence of type of procedure, site of procedure, and morbidities on LOS.
The NHDS database for the combined years 1995 through 1997 was composed of 780 830 inpatient admissions, excluding newborns. After extraction according to procedure code, the database for patients who had 1 or more head and neck procedures during their admission consisted of 3932 admissions (aggregate group). The more selective database for patients whose primary procedure was a head and neck intervention resulted in 3039 patients (primary procedure group). These groups constituted 0.50% and 0.39% of the total NHDS inpatient admissions, respectively.
A total of 4591 head and neck procedures were performed in 3932 admitted patients, for an average of 1.17 head and neck procedures per patient. A total of 77 different head and neck surgical procedures were encountered in the database. For example, the aggregate group included 570 unilateral thyroid lobectomies, 352 parathyroidectomies, 96 glossectomies, 126 total laryngectomies, 289 unilateral neck dissections, and 63 mandibulectomies. The distributions of procedures by surgical site for the aggregate group are depicted in Table 1.
For the aggregate group, the mean age was 51.9 years (SD, 20.47 years). The male-female ratio was 1:1.28. For the aggregate group, the mortality rate was 2.98% (117 of 3932 patients). A total of 222 patients incurred a medical complication for a morbidity rate of 5.65%. A total of 12.16% of patients who experienced a complication also died, whereas only 2.43% of patients without a major complication died (P<.001, χ2). The distributions of major complications and associated mortality are displayed in Table 2. The mean LOS was 6.15 days (SD, 10.39 days) and ranged from 1 to 245 days. The mean LOS was 5.46 days (SD, 8.96 days) in the absence of a major complication vs 17.70 days (SD, 20.72 days) when a major complication occurred (P<.001, ANOVA). Similarly, the LOS was 15.42 days (SD, 17.29 days) when the outcome was death vs 5.87 days (SD, 9.97 days) when the patient was discharged alive (P<.001, ANOVA). Patients with a fatal outcome had a higher mean age (60.79 years) than patients without a fatal outcome (51.65 years, P<.001, ANOVA). Similarly, increasing age was associated with a higher medical complication rate (P<.001).
In the primary procedure group, 3039 patients underwent a primary head and neck surgical procedure. The distributions of procedures by surgical site for the primary procedure group are depicted in Table 1. The overall mortality rate for this group was 2.11%. The major morbidity rate was 3.55%. A total of 12.96% of patients who experienced a concomitant complication died, whereas 1.71% of patients without a major complication died (P<.001, χ2). The distributions of major complications and associated mortality for the primary procedure group are displayed in Table 3. For the primary procedure group, procedures involving the esophagus carried the highest mortality at 8.4% (P<.001). The average LOS was 4.79 days (SD, 7.97 days). The LOS was significantly higher for patients with a major morbidity than for those without (16.19 vs 4.65 days, respectively; P<.001, ANOVA). Similarly, LOS was increased for patients who died during admission when compared with those who did not (11.15 vs 4.65 days, respectively; P<.001). The LOS benchmarks for procedures according to surgical site are listed in Table 4.
Patients undergoing head and neck oncologic surgical procedures are different from typical patients undergoing other otolaryngologic procedures. These patients tend to be older with a history of tobacco and/or alcohol abuse. These same risk factors that are implicated in the development of head and neck tumors also impart risk for coronary artery disease, chronic obstructive pulmonary disease, peripheral vascular disease, and stroke. In addition, the increasing use of chemotherapy and radiation therapy may contribute to poorer nutritional status and technically more challenging surgery. Furthermore, because free flap reconstruction for postoperative defects has become more prevalent, these procedures tend to be longer and more involved.10,11 Therefore, critical assessment of perioperative surgical risk must include consideration of these medical risk factors and their associated potential complications.
Several published reports have focused on various perioperative complications from head and neck surgical procedures. These may be divided into local complications, such as wound infection or fistula formation, and distant complications, such as myocardial infarction, stroke, or pulmonary embolism. We have chosen to examine our database to determine the incidence and mortality for various typically encountered perioperative medical morbidities. Understanding the incidence of such complications is important in preoperative counseling for patients, the determination of proper monitoring in the postoperative setting, and allocation of resources for patients with head and neck cancer.12 In many cases, patients are more concerned with the implications of anesthesia rather than the immediate postsurgical outcome from their extirpative procedure. This likely stems from the widespread knowledge that chronic obstructive pulmonary disease and coronary artery disease increase anesthetic risk.1
Because head and neck cancer only accounts for approximately 3% of new cancers diagnosed each year, the absolute number of treated patients is small.13 Furthermore, the increasing interest in organ preservation strategies has perhaps further decreased surgical volumes in head and neck oncology. Therefore, it is difficult to assemble a large, prospective series of patients that can be examined to determine the incidence of perioperative medical complications. Studies in the otolaryngology literature have therefore focused primarily on assessing the incidence and outcomes for individual uncommon medical complications that arise in a series of treated patients. Data can be extrapolated from other surgical disciplines with some success, but head and neck procedures alone are associated with increased perioperative risk compared with other surgical procedures.14
The database constructed for this study represents a very large patient population who underwent head and neck surgical procedures. Patients have been extracted from a national database, which includes a wide variety of medical institutions, from all geographic areas of the United States. The NHDS data have been previously used to determine the incidence of various procedures, the prevalence of disease, and year-to-year trends in surgery.7 In general, NHDS data are thought to accurately reflect disease and surgical mortality but less accurately reflect morbidity.9 The NHDS data collection system consists of a random sampling of these institutions, carefully chosen to represent an accurate cross section of medical practice in the United States. Therefore, treatment, selection, and institutional biases are less likely to play a role. The NHDS data extracted in this study differ significantly from the patient populations in most of the head and neck surgical literature in that almost all of the previously published literature reflects experience from single academic institutions. This introduces an inherent bias when attempting to determine benchmarks, since teaching hospitals have a well-documented lower mortality rate and different patient populations.15 Therefore, we consider these data suitable benchmarks for the incidence of medical complications in head and neck surgery.
We examined an aggregate group and a primary procedure group to ensure that all head and neck procedures would be captured. For example, a patient with an esophageal cancer may also have cervical metastasis. The primary procedure for this patient may actually be either the neck dissection or the esophagectomy, depending on the coding method. Although looking at admissions for which the primary procedure was a head and neck procedure is probably the most accurate, it would miss patients who undergo head and neck procedures in conjunction with other services, such as thoracic surgery or neurosurgery. By examining both databases, both a conservative and liberal extreme can be analyzed, and a check for internal consistency is provided.
Our data indicate that the incidence of major medical complications is not negligible in head and neck surgery. Patients should be counseled that there is a significant chance of medical complications that may affect the postoperative course after head and neck surgery. Postoperative pneumonia is an especially notable risk and carries with it a major impact on overall mortality. This is likely because these patients often have a history of smoking with underlying chronic obstructive pulmonary disease, placing them at higher risk for both pneumonia and its complications.3 Our data suggest that prevention and prompt treatment of postoperative pneumonia should have a significant impact on mortality after head and neck surgery and decrease overall LOS.
Interestingly, we found that acute myocardial infarction and acute cerebrovascular accident are both uncommon complications after head and neck surgery. Furthermore, the mortality rate occurring with these complications was surprisingly small. This is likely due to advances in preoperative preparation with optimization of cardiac status and more intensive monitoring in the postoperative period.16 Therefore, although patients should be counseled about these 2 risks, they probably do not have a major impact on immediate perioperative outcomes. Other studies have identified a significant increased risk for myocardial infarction, specifically in patients with head and neck cancer.1,17 The risk of perioperative stroke is small in head and neck surgery. Causes of stroke may include excessive handling of the carotid vessels, hypoperfusion, or patient positioning.4 Even given the proximity of these procedures to the carotid artery, and the likelihood of concurrent carotid vascular disease, the risk of stroke is small. However, appropriate precautions should be taken in this group of patients, who may be at increased risk.
Acute renal failure is associated with a very high likelihood of mortality in both patient groups. Although its overall incidence is small, acute renal failure probably reflects one component of multiple organ system failure in a patient with a poor postoperative outcome and, therefore, is highly correlated with death. The extremely low incidence of pulmonary embolism is likely due to increased awareness and standard perioperative prophylaxis against deep venous thrombosis. Also, most patients undergoing head and neck surgery are ambulatory soon after surgery, decreasing their risk for deep venous thrombosis. Moreano et al18 found a similar incidence of 0.4% of pulmonary embolism among patients undergoing head and neck surgery in a single institution study.
Aside from the mortality implications of perioperative medical complications, such complications have a definite impact on LOS, as our data demonstrate. Such impact needs to be considered in preoperative counseling for patients regarding expected LOS. In addition, minimizing such complications will likely decrease LOS, conserving valuable resources.
Mortality from otolaryngologic procedures is rare. However, it would seem that patients undergoing head and neck surgical procedures do manifest a nonnegligible mortality rate, approximating 2% to 3%. This mortality rate holds true for most of the operative sites, with operations on the esophagus having a significantly higher mortality rate. Other authors19 have reported higher mortality and morbidity rates for esophageal procedures as well. Mortality rates approximating 15% to 20% for esophageal procedures are not uncommon, especially when esophageal resection is involved. Looking at the esophageal procedures subset, esophageal resection and endoscopic resection and destruction of esophageal lesions carried higher mortality rates. These procedures may be performed by thoracic surgeons and gastroenterologists, as well as by otolaryngologists. Nonetheless, these procedures were part of the operative log for the American Board of Otolaryngology and are therefore included in the data. Such procedures should be approached with caution, regardless of the specialist performing the procedure.
Notably, procedures that involved the salivary glands, face and sinonasal region, and the thyroid and parathyroid glands had exceedingly low mortality rates. Interestingly, these sites are often affected by tumors other than squamous cell carcinoma, which may help explain the low mortality rates encountered with these sites. Furthermore, procedures that involve these sites are often shorter and do not involve entering the cervical portion of the upper aerodigestive tract. Procedures that involve the pharynx have been associated with increased complication rates.20
Length of stay has become the focus of an increasing number of health care utilization studies. To conserve resources and provide more efficient care, substantial efforts have been directed at reducing LOS. Part of that effort includes the shift from the inpatient to the outpatient arena for many surgical procedures. For example, our database does not include noninterventional endoscopic procedures, such as direct laryngoscopy, esophagoscopy, or bronchoscopy, as part of a staging endoscopy. First, these patients are not likely to be admitted. Second, these procedures are not listed as part of the American Board of Otolaryngology head and neck procedure list used to extract the data.
Our data clearly indicate that patients undergoing head and neck surgical procedures have an LOS that approximates 1 week. Procedures involving any part of the upper aerodigestive tract that is involved with deglutition (pharynx, larynx, or esophagus) are associated with a substantial increase in LOS. In addition, increasing patient age was also found to be associated with an increased LOS. This may be due to additional comorbidities that elderly patients have. However, several studies2,21 have documented that head and neck surgery in elderly patients is not significantly associated with additional risk. Our data demonstrate that increasing age is associated with additional morbidity in head and neck surgery. Finally, the occurrence of a medical complication significantly increased the LOS by almost 3-fold. This suggests that diligence in preventing medical complications should assist in decreasing LOS for patients undergoing head and neck surgery. Additional effort should be directed at decreasing LOS for patients undergoing head and neck surgical procedures, when appropriate. Several studies22,23 have emerged showing that the use of critical pathways in head and neck surgery may decrease LOS and provide more cost-efficient care.
One limitation of the present study is that it fails to capture patients who undergo surgery, are discharged from the hospital, and then experience a significant complication after discharge. These patients may return to the hospital of origin or be treated at a second hospital for their complication. To fully measure extended perioperative complications, one should include this group of patients, which may approximate an estimated 5% to 8% of patients, up through 30 days after discharge. Because our methods overlook this group of patients, it is likely that some of the morbidity and mortality estimates may in fact be higher when one considers 30-day postoperative morbidity and mortality rates.
Medical morbidity and mortality rates for head and neck surgery are relatively low (5.65% and 2.98%, respectively). Pulmonary complications are the most frequently encountered postoperative morbidity.
When a major medical complication occurs, both the odds for death and the LOS increase significantly (increase in odds, 5.65, and a 3-fold increase in LOS).
Contrary to other reports, increasing age is associated with increased rates of medical complications and mortality in head and neck surgery.
Accepted for publication July 13, 2000.
Corresponding author and reprints: Neil Bhattacharyya, MD, Division of Otolaryngology, Brigham and Women's Hospital, Harvard Medical School, 333 Longwood Ave, Boston, MA 02115.