This cohort study evaluates a preoperative adverse event risk index in patients from the ACS NSQIP database who underwent surgery for head and neck cancer between 2006 and 2016.
Key Points
Question
To what extent are sociodemographic, clinical, and frailty-related characteristics associated with short-term postoperative adverse events in patients undergoing inpatient head and neck cancer surgery?
Findings
This cohort study of 31 399 operations registered in the American College of Surgeons NSQIP database demonstrates that multiple patient characteristics evaluated by the Head and Neck Surgery Risk Index (HNSRI) were independently associated with major adverse events or death on multiple regression analysis. The HNSRI, using all of these characterisitics, demonstrated a sensitivity of 80.1% (95% CI, 79.4-80.8) and specificity of 72.3% (95% CI, 70.3-74.2) regarding occurrence.
Meaning
The HNSRI might be used by clinicians to counsel patients awaiting head and neck cancer surgery and their families.
Abstract
Importance
Patients 65 years or older are the most frequent users of operative resources and are also the most vulnerable to postoperative adverse events (AEs). Frailty indices are increasingly being used for preoperative risk stratification within head and neck cancer surgery, but most models lack a multifactorial basis and cannot be directly applied to clinical practice. A practical risk index is needed for clinicians to gauge risk factors preoperatively.
Objective
To develop a preoperative risk index of short-term major postoperative AEs for patients undergoing head and neck cancer surgery.
Design
Cohort analysis of patients from multiple medical centers undergoing inpatient ablative or reconstructive head and neck cancer surgery and registered in the American College of Surgeons (ACS) National Surgical Quality Improvement Program (NSQIP) from 2006 to 2016.
Exposures
Inpatient ablative or reconstructive head and neck cancer surgery.
Main Outcomes and Measures
Sociodemographic, frailty-related, and surgical factors in the derivation cohort were evaluated using simple and multiple logistic regression. Risk factors were subsequently integrated into a preoperative head and neck surgery risk index (HNSRI) and compared with existing models using the validation cohort. A composite variable of major postoperative AEs was used, including death within 30 days of surgery.
Results
A total of 43 968 operations were found using the ACS NSQIP database. Of these, 12 569 cases were excluded as non–head and neck cancer or emergency surgery. Of the included 31 399 operations reviewed, the mean (SD) patient age was 56.9 (15.4) years, and 16 994 of the patients were women (54.1%). A total of 4556 (14.5%) patients had a major postoperative AE, and 209 (0.7%) died. Older age, male sex, smoking, anticoagulation, recent weight loss, functional dependence, free-tissue transfer, tracheotomy, duration of surgery, wound classification, anemia, leukocytosis, and hypoalbuminemia were independently associated with major AEs or death on multiple regression analysis (C statistic, 0.83). The area under the curve of the HNSRI to predict major AEs including death using the validation cohort (n = 15 699) was 0.84 (95% CI, 0.83-0.85) with a sensitivity of 80.1% (95% CI, 79.4%-80.8%) and specificity, 72.3% (95% CI, 70.3%-74.2%). The HNSRI outperformed existing risk models for prediction of AEs: delta C index of the HNSRI to the modified frailty index 11, 0.23 (95% CI, 0.22-0.25); the American Society of Anesthesiologists classification, 0.14 (95% CI, 0.13-0.16); and the ACS risk calculator, 0.02 (95% CI, 0.01-0.03).
Conclusions and Relevance
The proposed HNSRI demonstrated a high sensitivity and specificity for major postoperative AEs and death in the studied population. This risk index can be used to counsel patients awaiting head and neck cancer surgery.
Introduction
Head and neck cancer (HNC), including thyroid, salivary, and cutaneous disease, represented 11% of new cancer diagnoses among Canadians in 2016.1 Most cancers involve the upper aerodigestive tract, causing a disturbance in speech, swallowing, or breathing or disfigurement. Extensive surgical resection and radiotherapy remain the mainstay of treatment; however, these modalities are not without their own set of complications.2,3 Equally important, patients with HNC often have coexistent medical conditions and nutritional deficiencies, which may contribute to increased postoperative adverse events (AEs).2,3 Up to 40% of patients experience postoperative AEs, most within the first 30 days from surgery.4 Not only can major AEs pose significant health risks including death, such events also cause treatment delays and emotional distress to patients and their families.
Patients 65 years or older are the most frequent users of operative resources and are also the most vulnerable to postoperative AEs.5 Various authors have described an increased risk of severe morbidity or mortality for elderly patients, in particular those over 80 years undergoing major oncologic surgery.6,7 The increase in comorbid conditions and functional deficits in the elderly has prompted the development of strategies to stratify care in the preoperative setting. Of increasing importance is the contribution of frailty in surgical selection. Frailty is a multidimensional syndrome of decreased physiological reserves and poor response to stressors.8,9,10 Current risk models, including the modified frailty index (mFI)-11 and mFI-5 are based on the Canadian Study of Health and Aging.10,11,12 Nonetheless, these models do not account for age-related changes and surgical procedures and are thus not directly applicable to head and neck cancer care.10,11,12,13
The ability to accurately predict postoperative risk in the preoperative setting can guide patient counselling. Additionally, it may potentially identify modifiable risk factors that can mitigate risk and lower postoperative AEs. Such “prehabilitation” programs have already been introduced to other surgical specialties.14,15 Using the American College of Surgeons (ACS) National Surgical Quality Improvement Program (NSQIP) database, we identified patients who had undergone head and neck cancer surgery and developed a preoperative head and neck surgery risk index (HNSRI) for 30-day major postoperative AEs.
Methods
Study Population
The ACS NSQIP provides a standardized, multi-institutional database on 30-day postoperative AEs in patients undergoing surgery. The time of occurrence of each AE is available, up to 30 days postoperatively, permitting the use of robust models for the analysis. We included patients from the NSQIP database who underwent nonemergency, inpatient head and neck cancer surgery by otolaryngologists from 2006 to 2016. Head and neck cancer cases were based on Current Procedural Terminology codes (eTable 1 in the Supplement). Ethical approval was granted by the McGill University Health Centre research board (MP-37-2018-3568), waiving patient written informed consent for deidentified data.
Measurement of Risk Factors
Sociodemographic factors such as age, sex, smoking status, and alcohol consumption were noted. Smoking status was quantified as smoking within 1 year of surgery. Alcohol consumption was defined as greater than 2 drinks per day within 2 weeks of surgery. Medical comorbidities included cardiorespiratory (hypertension on medication, congestive heart failure, history of chronic obstructive pulmonary disease), neurologic (history of transient ischemic attack, cerebrovascular accident, delirium, coma, central nervous system tumor), renal (dialysis, acute renal failure preoperatively), hepatic (ascites, esophageal varices), metabolic (diabetes, long-term steroid use), hematologic (use of anticoagulation medications, recent blood transfusion), infectious (wound infection, preoperative sepsis), another surgery within 30 days of the index surgery, primary tumor site, postoperative diagnosis, surgery type, length of surgery, American Society of Anesthesiologists (ASA) class, preoperative laboratory results (white blood cell count, levels of hematocrit, creatinine, sodium, and bilirubin) and recent chemotherapy/radiotherapy. Nutritional factors included weight, body mass index (BMI), more than a 10% weight loss in the last 6 months, and preoperative albumin level. Continuous variables were classified into dichotomous variables based on clinically applicable cutoff values. Dyspnea (with exertion or at rest) and dependence level for activities of daily living were also analyzed.
Measurement of Outcomes
The primary outcome was any major postoperative AE including death within 30 days of surgery. Major postoperative AEs, as grouped by various authors, have included pulmonary embolism, acute kidney injury, cerebrovascular accident, coma, myocardial infarction, cardiac arrest, sepsis, septic shock, failure to wean off ventilator, reintubation, more than 4 blood transfusions, and return to the operating room.5,7,9,11,12 Any major postoperative AE or death was coded as a binary variable. Additionally, we evaluated any morbidity as defined by the ACS,16 including surgical site infection, wound disruption, pneumonia, unplanned intubation, pulmonary embolism, on a ventilator for more than 48 hours, progressive renal insufficiency, acute renal failure, urinary tract infection, stroke or cerebral vascular accident, cardiac arrest, myocardial infarction, deep venous thrombosis, systemic sepsis, pneumonia, cardiac event (cardiac arrest or myocardial infarction), and renal failure (progressive renal insufficiency or acute renal failure) within 30 days following surgery, or death within 30 days following surgery.16
Statistical Methods
A total of 43 968 operations were found using the ACS NSQIP database. Of these, 12 569 cases were excluded as non–head and neck cancer or emergency surgery. Half of remaining 31 399 patients meeting inclusion criteria were randomly assigned to the derivation cohort (n = 15 699), while the remaining half were assigned to the validation cohort (n = 15 700). The derivation cohort was used to develop the risk index.
Descriptive statistics were used to compare predictor variables among patients with and without major postoperative AEs. The Mann-Whitney U test was used for continuous data, and χ2 testing for categorial data. Subsequently, the association of sociodemographic, frailty-related, and clinical variables with major postoperative AEs was evaluated using logistic regression. Risk factors were retained from the univariate analysis if the lower bounds of the 95% confidence interval (CI) was above 0.8 or deemed clinically significant based on the current literature. The final model was derived using the lowest Bayesian Information Criterion and predictive value using the C statistic. Beta coefficients from the final model were then used to derive the risk index based on a 10-fold multiple of the β coefficient rounded to the nearest unit. Interval validity was tested using the validation cohort. The fit and accuracy of the final model was performed using the Hosmer-Lemeshow test. The risk index was compared with existing models including the ACS Risk Calculator for morbidity and mortality as well as the mFI-11, mFI-5, and ASA using receiver operator characteristic curves. All data analysis was performed using R software (version 3.0) using libraries survminer, survival, MASS, survAUC, Hmisc, generalhoslem, pROC, and DescTools.
Results
Descriptive Statistics
Among the 31 399 patients meeting all inclusion criteria from records in the ACS NSQIP database, a variety of head and neck cancers were represented: mucosal, cutaneous, endocrine, skull base, and salivary operations were captured in the data set (Table 1). A total of 7635 (24.3%) postoperative AEs occurred within 30 days of surgery with 4556 (14.5%) major AEs including 209 (0.7%) deaths (eTable 2 in the Supplement). Most major AEs occurred during postoperative days 1 to 5.
Table 1. Patient Characteristics of the Derivation Seta.
| Characteristic | No Event (n = 13 368) | Minor Event (n = 405) | Major Event (n = 1813) | Death (n = 113) |
|---|---|---|---|---|
| Age, y | ||||
| <50 | 4271 (31.9) | 67 (16.5) | 299 (16.5) | 11 (9.7) |
| 50-59 | 3251 (24.3) | 76 (18.8) | 420 (23.2) | 17 (15.0) |
| 60-69 | 3149 (23.6) | 115 (28.4) | 523 (28.8) | 34 (30.1) |
| 70-79 | 1853 (13.9) | 99 (24.4) | 383 (21.2) | 27 (23.9) |
| 80-89 | 768 (5.7) | 42 (10.4) | 167 (9.2) | 17 (15.0) |
| ≥90 | 48 (0.4) | 5 (1.2) | 13 (0.7) | 5 (4.4) |
| Male sex | 5663 (42.3) | 232 (57.3) | 1164 (64.2) | 80 (70.8) |
| Social factors | ||||
| Current smoker | 2490 (18.6) | 117 (28.9) | 542 (29.9) | 27 (23.9) |
| Current EtOH user | 154 (1.2) | 8 (2.0) | 50 (2.8) | 4 (3.5) |
| Cardiorespiratory comorbidity | ||||
| Hypertension with medication | 5540 (41.4) | 217 (53.6) | 989 (54.6) | 80 (70.8) |
| Heart failure (recent exacerbation) | 57 (0.4) | 4 (1.0) | 26 (1.4) | 5 (4.4) |
| Dyspnea | 847 (6.3) | 36 (8.9) | 231 (12.7) | 29 (25.7) |
| COPD | 487 (3.6) | 33 (8.1) | 174 (9.6) | 15 (13.3) |
| Renal comorbidity | ||||
| Creatinine, mean (SE), mg/dL | 0.96 (0.008) | 0.94 (0.02) | 0.98 (0.02) | 1.21 (0.1) |
| Sodium, mean (SE), mmol/L | 139 (0.03) | 139 (0.16) | 139 (0.09) | 137 (0.4) |
| Acute renal failure | 18 (0.1) | 0 | 0 | 3 (2.7) |
| Dialysis | 119 (0.1) | 2 (0.5) | 24 (1.3) | 3 (2.7) |
| Hepatic comorbidity | ||||
| Ascites | 4 (0.02) | 1 (0.2) | 2 (0.1) | 1 (0.9) |
| Total bilirubin, mean (SE), mg/dL | 0.58 (0.004) | 0.56 (0.01) | 0.57 (0.01) | 0.70 (0.05) |
| Metabolic comorbidity | ||||
| Diabetes mellitus (noninsulin) | 1189 (8.9) | 45 (11.1) | 176 (9.7) | 10 (8.8) |
| Diabetes mellitus (insulin) | 504 (3.8) | 29 (7.2) | 118 (6.5) | 8 (7.1) |
| Long-term steroid use | 375 (2.8) | 22 (5.4) | 95 (6.3) | 11 (9.7) |
| Hematologic comorbidity | ||||
| Anticoagulation | 215 (1.6) | 13 (3.2) | 93 (5.1) | 12 (10.6) |
| Recent transfusion | 15 (0.1) | 1 (0.2) | 21 (1.2) | 2 (1.8) |
| White blood cell count, mean (SE), ×109/L | 7.3 (0.02) | 7.8 (0.15) | 7.9 (0.08) | 9.6 (0.54) |
| Hematocrit, mean (SE), % | 40.4 (0.04) | 39.6 (0.24) | 37.6 (0.14) | 36.1 (0.55) |
| Platelets, mean (SE), ×109/L | 246 (0.63) | 250 (4.3) | 261 (2.4) | 265 (12.2) |
| Other comorbidity | ||||
| Disseminated cancer | 505 (3.8) | 20 (4.9) | 152 (8.4) | 24 (21.2) |
| Wound infection | 200 (1.5) | 22 (5.4) | 138 (7.6) | 12 (10.6) |
| SIRS/shock | 74 (0.6) | 11 (2.7) | 62 (3.4) | 11 (9.7) |
| Sarcopenia markers | ||||
| Weight, mean (SE), kg | 180 (0.41) | 178 (2.37) | 170 (1.14) | 164 (4.41) |
| Underweight (BMI <18.0) | 258 (1.9) | 16 (4.0) | 131 (7.2) | 15 (13.3) |
| Normal (BMI 18.0-25.9) | 3590 (26.9) | 122 (30.1) | 645 (35.6) | 39 (34.5) |
| Overweight (BMI 26.0-29.9) | 4313 (32.3) | 124 (30.6) | 555 (30.6) | 36 (31.9) |
| Obese class 1 (BMI 30.0-34.9) | 2730 (20.4) | 71 (17.5) | 258 (14.2) | 11 (9.7) |
| Obese class 2 (BMI 35.0-39.9) | 1320 (9.9) | 44 (10.9) | 114 (6.3) | 8 (7.1) |
| Obese class 3 (BMI ≥40) | 1040 (7.8) | 24 (5.9) | 94 (5.2) | 3 (2.7) |
| Weight loss (recent) | 236 (1.8) | 17 (4.2) | 189 (10.4) | 19 (16.8) |
| Albumin, mean (SE), g/dL | 4.1 (0.004) | 3.8 (0.03) | 3.8 (0.01) | 3.4 (0.07) |
| Functional dependence | 191 (1.4) | 13 (3.2) | 88 (4.9) | 16 (14.2) |
| Primary cancer site | ||||
| Oral cavity | 1051 (7.9) | 67 (16.5) | 372 (20.5) | 20 (18.7) |
| Salivary | 1694 (12.7) | 59 (14.6) | 108 (6.0) | 9 (8.0) |
| Neck dissection | 1829 (13.9) | 60 (14.8) | 275 (15.2) | 32 (28.3) |
| Larynx | 471 (3.5) | 50 (12.3) | 237 (13.1) | 13 (11.5) |
| Pharynx | 265 (2.0) | 14 (3.5) | 70 (3.9) | 3 (2.7) |
| Craniofacial | 317 (2.4) | 10 (2.5) | 60 (3.3) | 2 (1.8) |
| Endocrine | 7026 (52.6) | 88 (21.7) | 303 (16.7) | 18 (15.9) |
| Integumentary | 59 (0.4) | 3 (0.7) | 7 (0.4) | 0 |
| Reconstruction | 607 (4.7) | 52 (12.8) | 378 (20.8) | 15 (13.3) |
| Other | 49 (0.4) | 2 (0.5) | 3 (0.2) | 1 (0.9) |
| Free-tissue transfer | 776 (5.8) | 75 (18.5) | 678 (37.3) | 34 (30.1) |
| Operative time, mean (SD), min | 199.0 (1.3) | 312 (10) | 422 (2) | 368 (23) |
| Tracheostomy | 429 (3.2) | 50 (12.8) | 368 (20.3) | 17 (15.0) |
Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); COPD, chronic obstructive pulmonary disease; EtOH, ethanol; SE, standard error; SIRS, systemic inflammatory response syndrome.
Unless otherwise indicated, data are reported as number (percentage) of patients.
Logistic Regression Analysis
In the derivation cohort, patients having major postoperative AEs were older, male, underweight, and smokers (eTable 3 in the Supplement). Most of these patients also had cardiorespiratory, renal, metabolic, nutritional, or hematologic comorbidities. Patients undergoing free-tissue transfer surgery, tracheotomy, surgery lasting more than 4 hours, or clean-contaminated or contaminated surgery were more likely to have major AEs. Clinically related risk factors were combined to form interactions; however, no significant interactions with sufficient prevalence were established. Risk factors in the multiple regression analysis included age (categorized by decade above 50 years), sex, hypertension, dyspnea, chronic steroid use, anticoagulation use, leukocytosis, recent weight loss, hypoalbuminemia, functional status, tracheotomy, free-tissue transfer, wound classification, and operative time (categorized as <4 hours, 4-8 hours, >8 hours) with a C statistic of 0.83 (95% CI, 0.82-0.84).
Head and Neck Surgery Risk Index
The scores used in the HNSRI were based on the β coefficients of the 16 independent risk factors from the multiple regression function (Table 2). Subsequently determined risk factors were classified into 6 groups with differential incidence of major AEs and death for clinical applicability (Table 3). The area under the curve (AUC) of the HNSRI using the derivation cohort was 0.83 (95% CI, 0.82-0.84) with a sensitivity of 77.3% (95% CI, 75.3%-79.2%) and a specificity of 76.3% (95% CI, 75.6%-77.0%) using a cutoff of 18. Using the validation cohort, the HNSRI produced an AUC of 0.84 (95% CI, 0.83-0.85) with a sensitivity of 80.1% (95% CI, 79.4%-80.8%) and a specificity of 72.3% (95% CI, 70.3%-74.2%). The Figure compares the accuracy of the HNSRI with that of the mFI-11, mFI-5, and ASA class to predict major postoperative AEs and death; the HNSRI has the largest AUC (0.82; 95% CI, 0.82-0.83; P < .001). The sensitivity and specificity of the HNSRI to predict any postoperative morbidity (excluding death) was 74.1% (95% CI, 72.3%-75.9%) and 77.7% (95% CI, 76.7%-78.1%), respectively. The risk of major postoperative AEs based on HNSRI score is detailed in Table 4. The Hosmer-Lemeshow goodness of fit test showed no significant difference between cohorts with acceptable calibration curves for the derivation and validation cohorts (eFigure 1 in the Supplement). The delta C indices for the HNSRI compared with the mFI-11, ASA class, and ACS risk calculator were 0.23 (95% CI, 0.22-0.25), 0.14 (95% CI, 0.13-0.16) and 0.02 (95% CI, 0.01-0.03), respectively.
Table 2. Final Model for Predicting Major Adverse Events or Death.
| Risk Factor | Prevalence (%) | PAF (%) | β Coefficient | Odds Ratio (95% CI) | Scorea |
|---|---|---|---|---|---|
| Age, y | |||||
| 50-59 | 24.0 | 3.3 | 0.129 | 1.14 (0.88-1.47) | 1 |
| 60-69 | 24.3 | 7.0 | 0.274 | 1.31 (1.03-1.69) | 3 |
| 70-79 | 15.0 | 8.0 | 0.447 | 1.58 (1.21-2.07) | 4 |
| 80-89 | 6.3 | 3.0 | 0.396 | 1.49 (1.05-2.09) | 4 |
| ≥90 | 0.5 | 0.05 | 1.146 | 3.15 (1.14-8.30) | 11 |
| Male sex | 45.5 | 12.0 | 0.265 | 1.30 (1.10-1.55) | 3 |
| Hypertension (with medication) | 43.5 | 12.6 | 0.285 | 1.33 (1.12-1.58) | 3 |
| Dyspnea | 7.3 | 3.4 | 0.392 | 1.48 (1.16-1.88) | 4 |
| Long-term steroid use | 3.2 | 1.6 | 0.410 | 1.51 (1.08-2.09) | 4 |
| Anticoagulation | 2.1 | 1.5 | 0.827 | 2.29 (1.57-3.31) | 8 |
| Current smoker | 2.2 | 1.5 | 0.521 | 1.68 (1.16-2.42) | 5 |
| Leukocytosis (WBC >11.5 ×109/L) | 8.2 | 5.8 | 0.542 | 1.72 (1.35-2.20) | 5 |
| Anemia (HCT <35%) | 10.6 | 15.1 | 0.987 | 2.68 (2.12-3.25) | 10 |
| Hypoalbuminemia (<3.5 g/dL) | 4.9 | 3.1 | 0.507 | 1.66 (1.33-2.06) | 5 |
| Weight loss | 2.9 | 2.9 | 0.708 | 2.03 (1.51-2.72) | 7 |
| Functional lossb | 2.0 | 1.6 | 0.591 | 1.81 (1.23-2.63) | 6 |
| Surgical time, h | |||||
| 4-8 | 21.7 | 22.9 | 0.864 | 2.37 (1.93-2.91) | 9 |
| >8 | 11.0 | 31.7 | 1.652 | 5.22 (4.06-6.71) | 17 |
| Free-tissue transfer | 10.0 | 9.1 | 0.696 | 2.00 (1.59-2.53) | 7 |
| Tracheostomy | 5.5 | 4.3 | 0.600 | 1.82 (1.42-2.34) | 6 |
| Wound class | |||||
| Clean-contaminated | 30.7 | 16.0 | 0.481 | 1.62 (1.34-1.95) | 5 |
| Contaminated | 1.3 | 2.5 | 1.094 | 2.99 (1.84-4.80) | 11 |
| Dirty | 0.8 | 2.6 | 1.462 | 4.31 (2.45-7.57) | 15 |
Abbreviations: HCT, hematocrit; OR, odds ratio; PAF, population attributable fraction; WBC, white blood cell count.
Score derived from multiplying the β coefficient by 10.
Functional loss includes dependence on others for activities of daily living or living in a nursing facility prior to surgery.
Table 3. HNSRI for Major Adverse Events and Death in the Derivation and Validation Cohorts.
| Preoperative Risk Score | Derivation Set | Validation Set | ||||||
|---|---|---|---|---|---|---|---|---|
| Total Patients, No. | Event or Death, No. (%) | Odds Ratio (95% CI) | AUC (95% CI) | Total Patients, No. | Event or Death, No. (%) | Odds Ratio (95% CI) | AUC (95% CI) | |
| 0-10 | 8054 | 235 (2.9) | 1 [Reference] | 0.83 (0.82-0.84) | 8012 | 208 (2.6) | 1 [Reference] | 0.84 (0.83-0.85) |
| 11-20 | 3482 | 277 (8.0) | 2.88 (2.41-3.44) | 3554 | 324 (9.1) | 3.76 (3.15-4.51) | ||
| 21-30 | 1990 | 378 (19.0) | 7.80 (6.58-9.27) | 2018 | 376 (18.6) | 8.59 (7.20-10.3) | ||
| 31-40 | 1278 | 487 (38.1) | 20.5 (17.3-24.4) | 1228 | 483 (39.3) | 24.3 (20.4-29.1) | ||
| 41-50 | 634 | 349 (55.0) | 40.7 (33.3-50.0) | 626 | 347 (55.4) | 46.7 (37.9-57.6) | ||
| ≥51 | 261 | 200 (76.6) | 109 (80.1-150) | 262 | 182 (69.5) | 85.4 (63.7-115) | ||
Abbreviations: AUC, area under the curve; HNSRI, Head and Neck Surgery Risk Index.
Figure. Receiver Operator Characteristic Curves of Different Risk Models for Postoperative Major Adverse Events Within 30 Days of Surgery.
ACS indicates the American College of Surgeons risk calculator; ASA, American Society of Anesthesiologists classification; HNSRI, Head and Neck Surgery Risk Index (the present study); mFI-5, modified frailty index 5; and mFI-11, modified frailty index 11.
Table 4. Risk of a Major Postoperative Adverse Event Within 30 Days of Surgery Using the HNSRIa.
| Total HNSRI Score | Probability of an Event Within 30 Days |
|---|---|
| 0-5 | 0.020 |
| 6-10 | 0.040 |
| 11-15 | 0.068 |
| 16-20 | 0.108 |
| 21-25 | 0.160 |
| 26-30 | 0.226 |
| 31-35 | 0.361 |
| 36-40 | 0.419 |
| 41-45 | 0.507 |
| 46-50 | 0.632 |
| 51-55 | 0.676 |
| 56-60 | 0.777 |
| ≥61 | 0.795 |
Abbreviation: HNSRI, Head and Neck Surgery Risk Index.
This table reports the associated probability of a major postoperative adverse event including death within 30 days of surgery based on HNSRI score.
Discussion
We found that roughly 14.5% of inpatient head and neck cancer surgical procedures were complicated by a major postoperative AE within the first 30 days following surgery. The HNSRI predicted major postoperative events including death with acceptable accuracy. Despite including 16 parameters, the HNSRI outperformed other NSQIP-derived risk tools to predict morbidity in head and neck cancer patients undergoing surgery.16,17,18,19,20 The AUC for the HNSRI was larger than that of the mFI-5, mFI-11, and the ACS risk calculator to predict any postoperative AE (morbidity and mortality) within 30 days of surgery. The C statistic for the ACS Universal Risk Calculator is in line with a previous large data comparison.17 The HNSRI predictive capability for major AEs including death cannot be directly compared with that of the ACS risk calculator because a combined risk of major morbidity and mortality is not calculated in the ACS calculator.
Limitations
Several limitations restrict the generalizability of this study. As detailed in Table 1, patients who died within 30 days of surgery had significantly different preoperative characteristics than those who survived. It is possible that the patients who died represent a different population enrolled for head and neck cancer surgery than those who survived. Also, direct comparison could not be made with other externally validated indices, such as the Charlson comorbidity index, given lack of available data.21 Furthermore, the study data originate from a surgical cohort, and patients eligible for head and neck cancer surgery but not operated on were not included. Also, risk factors specific to patients with head and neck cancer could not be included, such as TNM stage, p16 status, Eastern Cooperative Oncology Group performance status, and previous radiotherapy (beyond 30 days before surgery). Outcome measures were limited to 30 days and based on common surgical postoperative AEs. These may not include AEs relevant to head and neck cancer surgery including free-tissue transfer failure, feeding-tube dependence, and oncologic outcome.22 Finally, clustering of cases based on location and medical center could not be performed.
Conclusions
Herein, we report the development and evaluation of a practical risk index that might be used by head and neck surgeons, anesthesiologists and oncologists to better counsel patients awaiting head and neck cancer surgery and their families preoperatively. By stratifying patients into groups based on risk score, clinicians can identify vulnerable patients who may require additional care postoperatively. In like manner, this model may bring to light potentially reversible risk factors, which can mitigate postoperative risk. The calculated population attributable fractions for potentially reversible risk factors include anemia, hypoalbuminemia, weight loss, current smoking, and dyspnea. If these factors are causal and could be reversed in the preoperative setting, roughly 29% of major postoperative AEs could potentially be avoided. While this is likely an overestimation, reduced morbidity may be achieved by directly addressing these issues. For instance, recent weight loss, hypoalbuminemia, and anemia, which are particularly relevant in the head and neck cancer population, may be targeted by intensive preoperative nutrition supplementation.2 Nutritional programs have already shown benefit in other disease states.14,15 Additionally, surgical risk factors (operative time and free-tissue transfer) may potentially be substituted with acceptable alternative locoregional reconstructive strategies to mitigate risk. Prospective studies are needed to determine if addressing potentially modifiable risk factors can lower the risk of AEs in this population.
eTable 1. Head and Neck Surgery Current Procedural Terminology Codes
eTable 2. Postoperative Adverse Events
eTable 3. Simple and Multiple Logistic Regression
eFigure 1. Calibration Curves for the Head and Neck Surgery Risk Index
References
- 1.Canadian Cancer Society’s Advisory Committee on Cancer Statistics Canadian Cancer Statistics 2016. Toronto, Canada: Canadian Cancer Society; 2016. [Google Scholar]
- 2.Dort JC, Farwell DG, Findlay M, et al. Optimal perioperative care in major head and neck cancer surgery with free flap reconstruction: a consensus review and recommendations from the Enhanced Recovery After Surgery Society. JAMA Otolaryngol Head Neck Surg. 2017;143(3):292-303. doi: 10.1001/jamaoto.2016.2981 [DOI] [PubMed] [Google Scholar]
- 3.Cramer JD, Patel UA, Samant S, Smith SS. Postoperative complications in elderly patients undergoing head and neck surgery: opportunities for quality improvement. Otolaryngol Head Neck Surg. 2016;154(3):518-526. doi: 10.1177/0194599815618204 [DOI] [PubMed] [Google Scholar]
- 4.Suh JD, Sercarz JA, Abemayor E, et al. Analysis of outcome and complications in 400 cases of microvascular head and neck reconstruction. Arch Otolaryngol Head Neck Surg. 2004;130(8):962-966. doi: 10.1001/archotol.130.8.962 [DOI] [PubMed] [Google Scholar]
- 5.Khavanin N, Mlodinow A, Kim JY, Ver Halen JP, Antony AK, Samant S. Assessing safety and outcomes in outpatient versus inpatient thyroidectomy using the NSQIP: a propensity score matched analysis of 16,370 patients. Ann Surg Oncol. 2015;22(2):429-436. doi: 10.1245/s10434-014-3785-4 [DOI] [PubMed] [Google Scholar]
- 6.Xie X, Young J, Kost K, McGregor M. KTP 532 nm laser for laryngeal lesions: a systematic review. J Voice. 2013;27(2):245-249. doi: 10.1016/j.jvoice.2012.11.006 [DOI] [PubMed] [Google Scholar]
- 7.Fleisher LA, Pasternak LR, Herbert R, Anderson GF. Inpatient hospital admission and death after outpatient surgery in elderly patients: importance of patient and system characteristics and location of care. Arch Surg. 2004;139(1):67-72. doi: 10.1001/archsurg.139.1.67 [DOI] [PubMed] [Google Scholar]
- 8.Kulminski AM, Ukraintseva SV, Kulminskaya IV, Arbeev KG, Land K, Yashin AI. Cumulative deficits better characterize susceptibility to death in elderly people than phenotypic frailty: lessons from the Cardiovascular Health Study. J Am Geriatr Soc. 2008;56(5):898-903. doi: 10.1111/j.1532-5415.2008.01656.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Al-Refaie WB, Parsons HM, Henderson WG, et al. Major cancer surgery in the elderly: results from the American College of Surgeons National Surgical Quality Improvement Program. Ann Surg. 2010;251(2):311-318. doi: 10.1097/SLA.0b013e3181b6b04c [DOI] [PubMed] [Google Scholar]
- 10.Saxton A, Velanovich V. Preoperative frailty and quality of life as predictors of postoperative complications. Ann Surg. 2011;253(6):1223-1229. doi: 10.1097/SLA.0b013e318214bce7 [DOI] [PubMed] [Google Scholar]
- 11.Adams P, Ghanem T, Stachler R, Hall F, Velanovich V, Rubinfeld I. Frailty as a predictor of morbidity and mortality in inpatient head and neck surgery. JAMA Otolaryngol Head Neck Surg. 2013;139(8):783-789. doi: 10.1001/jamaoto.2013.3969 [DOI] [PubMed] [Google Scholar]
- 12.Subramaniam S, Aalberg JJ, Soriano RP, Divino CM. New 5-factor modified frailty index using American College of Surgeons NSQIP Data. J Am Coll Surg. 2018;226(2):173-181.e8. doi: 10.1016/j.jamcollsurg.2017.11.005 [DOI] [PubMed] [Google Scholar]
- 13.Gani F, Canner JK, Pawlik TM. Use of the modified frailty index in the American College of Surgeons National Surgical Improvement Program database: highlighting the problem of missing data. JAMA Surg. 2017;152(2):205-207. doi: 10.1001/jamasurg.2016.3479 [DOI] [PubMed] [Google Scholar]
- 14.Carli F, Charlebois P, Stein B, et al. Randomized clinical trial of prehabilitation in colorectal surgery. Br J Surg. 2010;97(8):1187-1197. doi: 10.1002/bjs.7102 [DOI] [PubMed] [Google Scholar]
- 15.Mayo NE, Feldman L, Scott S, et al. Impact of preoperative change in physical function on postoperative recovery: argument supporting prehabilitation for colorectal surgery. Surgery. 2011;150(3):505-514. doi: 10.1016/j.surg.2011.07.045 [DOI] [PubMed] [Google Scholar]
- 16.Bilimoria KY, Liu Y, Paruch JL, et al. Development and evaluation of the universal ACS NSQIP surgical risk calculator: a decision aid and informed consent tool for patients and surgeons. J Am Coll Surg. 2013;217(5):833-42.e1, 3. doi: 10.1016/j.jamcollsurg.2013.07.385 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Abt NB, Richmon JD, Koch WM, Eisele DW, Agrawal N. Assessment of the predictive value of the modified frailty index for Clavien-Dindo grade IV critical care complications in major head and neck cancer operations. JAMA Otolaryngol Head Neck Surg. 2016;142(7):658-664. doi: 10.1001/jamaoto.2016.0707 [DOI] [PubMed] [Google Scholar]
- 18.Abt NB, Xie Y, Puram SV, Richmon JD, Varvares MA. Frailty index: intensive care unit complications in head and neck oncologic regional and free flap reconstruction. Head Neck. 2017;39(8):1578-1585. doi: 10.1002/hed.24790 [DOI] [PubMed] [Google Scholar]
- 19.L’Esperance HE, Kallogjeri D, Yousaf S, Piccirillo JF, Rich JT. Prediction of mortality and morbidity in head and neck cancer patients 80 years of age and older undergoing surgery. Laryngoscope. 2018;128(4):871-877. doi: 10.1002/lary.26858. [DOI] [PubMed] [Google Scholar]
- 20.Vosler PS, Orsini M, Enepekides DJ, Higgins KM. Predicting complications of major head and neck oncological surgery: an evaluation of the ACS NSQIP surgical risk calculator. J Otolaryngol Head Neck Surg. 2018;47(1):21. doi: 10.1186/s40463-018-0269-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Kheterpal S, Tremper KK, Heung M, et al. Development and validation of an acute kidney injury risk index for patients undergoing general surgery: results from a national data set. Anesthesiology. 2009;110(3):505-515. doi: 10.1097/ALN.0b013e3181979440 [DOI] [PubMed] [Google Scholar]
- 22.Lewis CM, Aloia TA, Shi W, et al. Development and feasibility of a specialty-specific National Surgical Quality Improvement Program (NSQIP): the head and neck-reconstructive surgery NSQIP. JAMA Otolaryngol Head Neck Surg. 2016;142(4):321-327. doi: 10.1001/jamaoto.2015.3608 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable 1. Head and Neck Surgery Current Procedural Terminology Codes
eTable 2. Postoperative Adverse Events
eTable 3. Simple and Multiple Logistic Regression
eFigure 1. Calibration Curves for the Head and Neck Surgery Risk Index
