[Skip to Content]
[Skip to Content Landing]
Figure 1.
Clavien-Dindo Grade IV Complications, Mortality, and Overall Surgical and Medical Morbidity for All Major Head and Neck Cancer Operations by the Modified Frailty Index
Clavien-Dindo Grade IV Complications, Mortality, and Overall Surgical and Medical Morbidity for All Major Head and Neck Cancer Operations by the Modified Frailty Index

CDIV indicates Clavien-Dindo grade IV.

Figure 2.
Clavien-Dindo Grade IV Complications for Major Head and Neck Cancer Operation Categories by the Modified Frailty Index
Clavien-Dindo Grade IV Complications for Major Head and Neck Cancer Operation Categories by the Modified Frailty Index

Linear regression trend lines are shown.

Table 1.  
Preoperative Patient Characteristics for All Major Head and Neck Cancer Operations by the Modified Frailty Index
Preoperative Patient Characteristics for All Major Head and Neck Cancer Operations by the Modified Frailty Index
Table 2.  
Major Head and Neck Cancer Operations by the Modified Frailty Index
Major Head and Neck Cancer Operations by the Modified Frailty Index
Table 3.  
Multivariable Regression Model of Clavien-Dindo Grade IV Complications
Multivariable Regression Model of Clavien-Dindo Grade IV Complications
1.
Siegel  R, Ma  J, Zou  Z, Jemal  A.  Cancer statistics, 2014.  CA Cancer J Clin. 2014;64(1):9-29.PubMedGoogle ScholarCrossref
2.
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.PubMedGoogle ScholarCrossref
3.
Corrêa  GT, Bandeira  GA, Cavalcanti  BG,  et al.  Analysis of ECOG performance status in head and neck squamous cell carcinoma patients: association with sociodemographical and clinical factors, and overall survival.  Support Care Cancer. 2012;20(11):2679-2685.PubMedGoogle ScholarCrossref
4.
Chang  PH, Yeh  KY, Huang  JS,  et al.  Pretreatment performance status and nutrition are associated with early mortality of locally advanced head and neck cancer patients undergoing concurrent chemoradiation.  Eur Arch Otorhinolaryngol. 2013;270(6):1909-1915.PubMedGoogle ScholarCrossref
5.
Bell  J.  Redefining disease.  Clin Med (Lond). 2010;10(6):584-594.PubMedGoogle ScholarCrossref
6.
Fried  LP, Tangen  CM, Walston  J,  et al; Cardiovascular Health Study Collaborative Research Group.  Frailty in older adults: evidence for a phenotype.  J Gerontol A Biol Sci Med Sci. 2001;56(3):M146-M156.PubMedGoogle ScholarCrossref
7.
Cesari  M, Gambassi  G, van Kan  GA, Vellas  B.  The frailty phenotype and the Frailty Index: different instruments for different purposes.  Age Ageing. 2014;43(1):10-12.PubMedGoogle ScholarCrossref
8.
Rockwood  K, Song  X, MacKnight  C,  et al.  A global clinical measure of fitness and frailty in elderly people . CMAJ. 2005;173(5):489-495.PubMedGoogle ScholarCrossref
9.
Uppal  S, Igwe  E, Rice  LW, Spencer  RJ, Rose  SL.  Frailty index predicts severe complications in gynecologic oncology patients.  Gynecol Oncol. 2015;137(1):98-101.PubMedGoogle ScholarCrossref
10.
Kolbe  N, Carlin  AM, Bakey  S, Louwers  L, Horst  HM, Rubinfeld  I.  Assessing risk of critical care complications and mortality in the elective bariatric surgery population using a modified frailty index.  Obes Surg. 2015;25(8):1401-1407.PubMedGoogle ScholarCrossref
11.
Tsiouris  A, Hammoud  ZT, Velanovich  V, Hodari  A, Borgi  J, Rubinfeld  I.  A modified frailty index to assess morbidity and mortality after lobectomy.  J Surg Res. 2013;183(1):40-46.PubMedGoogle ScholarCrossref
12.
Kolbe  N, Bakey  S, Louwers  L,  et al.  Predictors of Clavien 4 complications and mortality after necrosectomy: analysis of the NSQIP database.  J Gastrointest Surg. 2015;19(6):1086-1092.PubMedGoogle ScholarCrossref
13.
Melin  AA, Schmid  KK, Lynch  TG,  et al.  Preoperative frailty Risk Analysis Index to stratify patients undergoing carotid endarterectomy.  J Vasc Surg. 2015;61(3):683-689.PubMedGoogle ScholarCrossref
14.
Amrock  LG, Neuman  MD, Lin  HM, Deiner  S.  Can routine preoperative data predict adverse outcomes in the elderly? development and validation of a simple risk model incorporating a chart-derived frailty score.  J Am Coll Surg. 2014;219(4):684-694.PubMedGoogle ScholarCrossref
15.
Karam  J, Tsiouris  A, Shepard  A, Velanovich  V, Rubinfeld  I.  Simplified frailty index to predict adverse outcomes and mortality in vascular surgery patients.  Ann Vasc Surg. 2013;27(7):904-908.PubMedGoogle ScholarCrossref
16.
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.PubMedGoogle ScholarCrossref
17.
Mulvey  CL, Pronovost  PJ, Gourin  CG.  Hospital volume and failure to rescue after head and neck cancer surgery.  Otolaryngol Head Neck Surg. 2015;152(5):783-789.PubMedGoogle ScholarCrossref
18.
Saxton  A, Velanovich  V.  Preoperative frailty and quality of life as predictors of postoperative complications.  Ann Surg. 2011;253(6):1223-1229.PubMedGoogle ScholarCrossref
19.
Searle  SD, Mitnitski  A, Gahbauer  EA, Gill  TM, Rockwood  K.  A standard procedure for creating a frailty index.  BMC Geriatr. 2008;8:24.PubMedGoogle ScholarCrossref
20.
Rutledge  JW, Spencer  H, Moreno  MA.  Predictors for perioperative outcomes following total laryngectomy: a University HealthSystem Consortium discharge database study.  Otolaryngol Head Neck Surg. 2014;151(1):81-86.PubMedGoogle ScholarCrossref
21.
Reid  BC, Alberg  AJ, Klassen  AC, Koch  WM, Samet  JM.  The American Society of Anesthesiologists’ class as a comorbidity index in a cohort of head and neck cancer surgical patients.  Head Neck. 2001;23(11):985-994.PubMedGoogle ScholarCrossref
22.
Piccirillo  JF, Wells  CK, Sasaki  CT, Feinstein  AR.  New clinical severity staging system for cancer of the larynx: five-year survival rates.  Ann Otol Rhinol Laryngol. 1994;103(2):83-92.PubMedGoogle ScholarCrossref
23.
Singh  B, Bhaya  M, Stern  J,  et al.  Validation of the Charlson comorbidity index in patients with head and neck cancer: a multi-institutional study.  Laryngoscope. 1997;107(11, pt 1):1469-1475.PubMedGoogle ScholarCrossref
24.
Wang  JR, Habbous  S, Espin-Garcia  O,  et al.  Comorbidity and performance status as independent prognostic factors in patients with head and neck squamous cell carcinoma [published online December 18, 2014].  Head Neck.PubMedGoogle Scholar
25.
Paleri  V, Wight  RG, Silver  CE,  et al.  Comorbidity in head and neck cancer: a critical appraisal and recommendations for practice.  Oral Oncol. 2010;46(10):712-719.PubMedGoogle ScholarCrossref
26.
Datema  FR, Poldermans  D, Baatenburg de Jong  RJ.  Incidence and prediction of major cardiovascular complications in head and neck surgery.  Head Neck. 2010;32(11):1485-1493.PubMedGoogle ScholarCrossref
27.
Protière  C, Viens  P, Rousseau  F, Moatti  JP.  Prescribers’ attitudes toward elderly breast cancer patients: discrimination or empathy?  Crit Rev Oncol Hematol. 2010;75(2):138-150.PubMedGoogle ScholarCrossref
28.
Wong  J, Xu  B, Yeung  HN,  et al.  Age disparity in palliative radiation therapy among patients with advanced cancer.  Int J Radiat Oncol Biol Phys. 2014;90(1):224-230.PubMedGoogle ScholarCrossref
Original Investigation
July 2016

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

Author Affiliations
  • 1Department of Otolaryngology–Head and Neck Surgery, The Johns Hopkins University School of Medicine, Baltimore, Maryland
  • 2Department of Otolaryngology–Head and Neck Surgery, The University of Chicago Medicine, Chicago, Illinois
JAMA Otolaryngol Head Neck Surg. 2016;142(7):658-664. doi:10.1001/jamaoto.2016.0707
Abstract

Importance  Functional status and physiologic deficits independent of age are being recognized for surgical risk stratification. Frailty is expressed as a combination of decreased physiologic reserve and multisystem impairments distinct from normal aging processes.

Objective  To assess the predictive value of the Modified Frailty Index (mFI) for Clavien-Dindo grade IV (CDIV) (intensive care unit–level complications) and grade V (mortality) after major head and neck oncologic surgery.

Design, Setting, and Participants  Retrospective analysis of prospectively collected American College of Surgeons National Surgical Quality Improvement Program data. All major head and neck cancer operations data were obtained from the January 1, 2006, to December 31, 2013, American College of Surgeons National Surgical Quality Improvement Program databases. Fifteen variables composed a previously validated mFI, with higher mFIs identifying more frail patients. Clavien-Dindo grade IV and mortality were defined using a preexisting mapping scheme from the Canadian Study of Health and Aging. Multivariable logistic regression analyses were performed.

Main Outcomes and Measures  The primary outcome measures were Clavien-Dindo Grade IV critical care complications and Grade V complications (mortality). Second outcomes included morbidity, readmission, and reoperation.

Results  There were 1193 major head and neck operations in the American College of Surgeons National Surgical Quality Improvement Program databases, with 86 (7.2%) CDIV complications. The mean (SD) age of all patients was 63.4 (12.4) years, and 67.7% (807 of 1193) were male. Clavien-Dindo grade IV significantly increased from 4.6% (22 of 483) to 100% (1 of 1) from nonfrail to the frailest patients (R2 = 0.79, P < .001). Mortality increased with the mFI (but not significantly) from 0.8% (4 of 483) to 3.6% (2 of 55) (R2 = 0.46, P = .42). Overall morbidity was not significantly associated or correlated with the mFI. On cross tabulation, increases in the mFI led to more CDIV complications in patients undergoing glossectomy (P = .03), mandibulectomy (P = .02), or laryngectomy (P = .002). Patients undergoing pharyngectomy or esophagectomy did not have significant increases in CDIV complications by the mFI. The coefficients of determination for each category were R2 = 0.62 for glossectomy, R2 = 0.72 for mandibulectomy, R2 = 0.97 for laryngectomy, R2 = 0.94 for pharyngectomy, and R2 = 1.00 for esophagectomy. On multivariable analysis, the mFI was associated with CDIV complications (odds ratio, 1.65; 95% CI, 1.15-2.37) but not mortality (odds ratio, 0.78; 95% CI, 0.34-1.76).

Conclusions and Relevance  The mFI is predictive of postoperative critical care support after surgery for head and neck cancer. Specifically, increases in mFIs were strongly associated with CDIV complications for glossectomy, mandibulectomy, and laryngectomy. Classifying patients by their functional status using the mFI may help predict outcomes after head and neck oncologic surgery.

Introduction

In 2013, more than 42 000 new patients were diagnosed as having head and neck cancers.1 Many of these patients had comorbidities that significantly altered postoperative care and outcomes. Increasing age has been a main focus of previous surgical outcomes research; however, functional status and physiologic deficits independent of age are being recognized as more important for stratifying surgical risk.2 The Eastern Cooperative Oncology Group performance status score demonstrates this phenomenon as an independent predictor of overall survival3 and unexpected mortality in patients with advanced-stage head and neck squamous cell carcinoma treated with chemoradiotherapy.4

Frailty is expressed as a combination of decreased physiologic reserve and multisystem impairments distinct from normal aging processes.5 Two different models have quantified frailty, namely, the physical phenotypic6 and cumulative deficit7 models. Quiz Ref IDThe physical phenotypic model is originally described as including unintentional weight loss, subjective exhaustion, low grip strength as a proxy for weakness, reduced walking speed, and low physical activity levels.6 The cumulative deficit model is a comprehensive geriatric assessment based on an assumption that additive comorbidity effects produce an overall functional decline.7 The assessment is strictly objective, with true or false answers. The cumulative deficit model is more widely used owing to the data collection method and assessment.

Quiz Ref IDThe Modified Frailty Index (mFI) is a derivative of the Canadian Study of Health and Aging Frailty Index (CSHA-FI), which was created using the cumulative deficit model.8 The mFI has been validated in various surgical fields,915 including otolaryngology. Adams et al16 studied the mFI as a predictor of severe complications and mortality. However, their study combined all otolaryngology procedures, a heterogeneous surgical population ranging from tonsillectomy to total thyroidectomy, without indicating if major operations were included. To our knowledge, the mFI has not been used to focus on major head and neck cancer surgery, the population at greatest risk for intensive care unit (ICU)–level complications or mortality.17 Therefore, the objective of this study was to assess the predictive value of the mFI for Clavien-Dindo grade IV (CDIV) (ICU-level complications) and grade V (mortality) after major head and neck oncologic surgery.

Box Section Ref ID

Key Points

  • Question Does the Modified Frailty Index (mFI) predict intensive care unit–level complications after major head and neck oncologic surgery?

  • Findings In an analysis of prospectively collected American College of Surgeons National Surgical Quality Improvement Program data, the mFI was predictive of postoperative critical care support and in-hospital reoperation after surgery for head and neck cancer, with a strong correlation for glossectomy, mandibulectomy, and laryngectomy.

  • Meaning Classifying patients by their functional status through the mFI rather than on strict pathoanatomic comorbidities is important to predict postoperative outcomes in head and neck cancer.

Methods
Data Source and Patients

Data from all major head and neck cancer operations were obtained from the January 1, 2006, to December 31, 2013, American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) databases using Current Procedural Terminology coding (eTable in the Supplement). Exclusion criteria included age younger than 18 years and primary surgeons who were not otolaryngologists. The ACS-NSQIP is a nationally validated, risk-adjusted, prospectively collected database with more than 350 participating academic and nonacademic hospitals. A trained surgical clinical reviewer collects all data prospectively, and a surgeon champion supervises program implementation and ensures data reliability. The Johns Hopkins Institutional Review Board approved the ACS-NSQIP data use and analysis. All data were collected according to the ACS-NSQIP data use agreement as a quality assurance endeavor, with only deidentified data received.

Modified Frailty Index

Saxton and Velanovich18 created the mFI by matching 11 CSHA-FI variables to 15 ACS-NSQIP variables. The original CSHA-FI has 70 variables, but previous data have demonstrated validity for as few as 10 variables.19 The ACS-NSQIP variables that constitute the 11 mFI variables are history of diabetes mellitus, functional status index of at least 2, history of chronic obstructive pulmonary disease or pneumonia, history of congestive heart failure, history of myocardial infarction, history of hypertension requiring medication, history of peripheral vascular disease or ischemic rest pain, history of impaired sensorium, history of transient ischemic attack or cerebrovascular accident, history of cerebrovascular accident with neurologic deficit, and history of percutaneous coronary intervention, stenting, or angina. Quiz Ref IDA functional status refers to a patient’s independent ability to perform activities of daily living (ADLs) 30 days before surgery, including but not limited to eating, dressing, bathing, toileting, and ambulating. Functional status index of 1 indicates ADL independence, while an index of 2 signifies partial dependence on another individual for ADLs. A functional status index of 3 denotes total dependence.

The mFIs were calculated by dividing the number of variables each patient was positive for by the total number of variables (n = 11). For example, if a patient had a history of chronic obstructive pulmonary disease, congestive heart failure, and percutaneous coronary intervention, he or she received a score of 0.27 (3 of 11). All patients without any of the above factors were scored as 0.

Variables

The primary outcomes were CDIV and grade V (mortality) complications. The CDIV complications included unplanned intubations, pulmonary embolism, failure to wean off the ventilator more than 48 hours after surgery, acute renal failure, cardiac arrest requiring initiation of cardiopulmonary resuscitation, acute myocardial infarction, and severe sepsis or septic shock. Unplanned intubations required placement of an endotracheal tube or other similar breathing tube (laryngeal mask airway, nasotracheal tube, etc) and ventilator support during surgery or within 30 days after surgery that was not intended or planned. Indications for unplanned intubations included refractory hypotension, cardiac arrest, and inability to protect the airway. Acute renal failure was defined as a patient who did not require dialysis before surgery but had worsening renal function requiring hemodialysis, peritoneal dialysis, hemofiltration, hemodiafiltration, or ultrafiltration. Sepsis is considered severe if associated with organ or circulatory dysfunction. Documentation of clinical signs and symptoms of systemic inflammatory response syndrome or sepsis and organ or circulatory dysfunction was required. Organ dysfunction included but was not limited to oliguria, acute change in mental status, and acute respiratory distress. Circulatory dysfunction encompassed hypotension or requirement for inotropic or vasopressor agents.

Additional outcomes included unplanned reoperation, unplanned readmission, or any other complication, which was grouped as “morbidity.” Morbidity was defined as an aggregate of all available recorded 30-day postoperative negative outcomes, which included superficial and deep wound infection, organ space infection, wound dehiscence, pneumonia, unplanned intubation, pulmonary embolism, ventilation requirement for more than 48 hours, progressive renal insufficiency, acute renal failure, urinary tract infection, stroke or cerebrovascular accident, myocardial infarction, cardiac arrest requiring cardiopulmonary resuscitation, transfusion of more than 1 U of packed red blood cells up to 72 hours after surgery, graft or prosthesis failure, deep vein thrombosis or thrombophlebitis requiring anticoagulation, sepsis, and septic shock. All readmissions and reoperations were unplanned and occurred within 30 days.

Multivariable logistic regression analysis was performed in a standard stepwise approach, with inclusion of significant variables (P < .05) after univariable regression. Independent variables included for adjustment during multivariable analysis were age, sex, body mass index (BMI), smoking status, American Society of Anesthesiologists’ classification, wound classification, current wound infection, diabetes mellitus status, corticosteroid use for a chronic condition, operative time, history of previous operation within 30 days of surgery, operation year, and the consumption of more than 2 alcoholic drinks per day within the past 2 weeks.

The study population was described with general summary statistics. Group comparisons were made using the t test and Wilcoxon rank sum test for continuous variables, dependent on distribution, or the χ2 test for categorical variables, as appropriate (Table 1). Major head and neck operation categories were cross-tabulated using a χ2 analysis (Table 2). The CDIV complications, mortality, reoperation, readmission, and morbidity were analyzed in comparison with the mFI using the χ2 test. Results were graphed (Excel; Microsoft Corporation), with linear regression trend lines denoting coefficients of determination (R2 values) (Figure 1, Figure 2, and eFigure in the Supplement). To estimate the precision of the effect size, 95% CIs were calculated around R2. Results were considered significant if the observed P value was less than .05. Statistical analyses were performed with a software package (Stata, version SE 12; StataCorp LP).

Results

There were 1193 major head and neck operations in the ACS-NSQIP databases. Major head and neck operations included are listed in the eTable in the Supplement by their Current Procedural Terminology codes. There were 551 glossectomies, 394 laryngectomies, 191 mandibulectomies, 40 pharyngectomies, and 17 esophagectomies. There were 86 (7.2%) CDIV complications, 14 (1.2%) deaths, 119 (14.2%) unplanned reoperations (in 838 patients undergoing a second operation), and 61 (5.1%) unplanned readmissions.

All preoperative characteristics available in the ACS-NSQIP databases are listed in Table 1. The mean (SD) age of all patients was 63.4 (12.4) years, and 67.6% (807 of 1193) were male. The mean (SD) BMI (calculated as weight in kilograms divided by height in meters squared) was 26.0 (6.2). Thirty-three percent (394 of 1193) were smokers, 13.5% (94 of 697) had consumed more than 2 alcoholic drinks per day within the past 2 weeks, 4.3% (51 of 1193) used corticosteroids to control a chronic condition, 14.5% (173 of 1193) had diabetes, and 7.1% (49 of 694) had been operated on within the past 30 days. The mean (SD) operative time was 396.8 (240.8) minutes. Preoperative characteristics were similar for age, BMI, sex, race/ethnicity, smoking status, alcohol intake, corticosteroid use, and operative time at all mFIs. A history of recent surgery increased significantly with the mFI. The Charlson Comorbidity Index also increased significantly with the mFI. It increased in a stepwise fashion from 7.3% (35 of 483) of patients having an index of at least 3 with a 0 mFI to 44.4% (4 of 9) with a 0.36 mFI.

The median mFI was 0.09. As the mFI increased from 0 to 0.55, the percentages of patients in the 6 index categories were 40.5% (483 of 1193), 36.8% (439 of 1193), 17.3% (206 of 1193), 4.6% (55 of 1193), 0.8% (9 of 1193), and 0.1% (1 of 1193), respectively. Patients undergoing esophagectomy had the highest percentage of patients with a 0 mFI at 47.1% (8 of 17), with patients undergoing laryngectomy having the lowest 0 mFI percentage, at 34.0% (134 of 394). Patients undergoing laryngectomy had the highest mFIs, above 0.18. Most patients undergoing glossectomy or mandibulectomy had a 0 or 0.09 mFI. Patients undergoing pharyngectomy or esophagectomy had a greater percentage within the 0.18 and 0.27 mFI. Procedure groups differed significantly in patient numbers within each mFI (P = .01). Significant increases in the mFI led to more CDIV complications in patients undergoing glossectomy (P = .03), mandibulectomy (P = .02), or laryngectomy (P = .002). Patients undergoing pharyngectomy or esophagectomy did not have significant increases in CDIV complications by the mFI. All mFIs are listed in Table 2 for each major head and neck case by category.

There were 86 (7.2%) total CDIV complications, which increased significantly as the mFI increased. Clavien-Dindo grade IV increased from 4.6% (22 of 483) to 100% (1 of 1) from 0 to 0.55 mFI (P < .001). The R2 for a linear regression trend line was 0.79 (95% CI, 0.77-0.81) (P < .001). Mortality also increased with the increase in the mFI but not significantly (P = .42). Mortality increased from 0.8% (4 of 483) to 3.6% (2 of 55) when mFI increased from 0 to 0.27, with an R2 of 0.46 (95% CI, 0.42-0.50) (P = .42). Overall morbidity was not significantly associated on the χ2 test or correlated on linear regression with mFIs. From 0 to 0.36 mFI, overall morbidity changed from 20.3% (98 of 483) to 22.2% (2 of 9), with an R2 of 0.17 (95% CI, 0.12-0.22) (P = .12). These data are shown graphically in Figure 1.

As the primary outcome, CDIV complications were broken down by specific major head and neck surgery category (Figure 2). In general, increases in mFIs were associated with more CDIV complications in patients undergoing esophagectomy (R2 of 1.00; 95% CI, 1.00-1.00), laryngectomy (R2 of 0.97; 95% CI, 0.97-0.97), pharyngectomy (R2 of 0.94; 95% CI, 0.93-0.95), mandibulectomy (R2 of 0.72; 95% CI, 0.69-0.75), or glossectomy (R2 of 0.62; 95% CI, 0.58-0.65). Significant increases in mFIs led to more CDIV complications in patients undergoing glossectomy, mandibulectomy, or laryngectomy. Patients undergoing pharyngectomy or esophagectomy did not have significant increases in CDIV complications linked to the mFI scores.

There were 119 (14.2%) reoperations (in 838 patients undergoing a second operation) and 61 (5.1%) readmissions within 30 days (eFigure in the Supplement). The mFI moderately correlated with reoperation (R2 = 0.63) but not readmission (R2 = 0.31). However, on multivariable analysis, mFIs were not independent predictors of reoperation (P = .62) or readmission (P = .96).

In a multivariable logistic regression model, the mFI became the strongest predictor of CDIV complications (odds ratio [OR], 1.65; 95% CI, 1.15-2.37) (Table 3). Using the same model comparing higher mFIs with an mFI of 0 demonstrated the following odds: 0.09 mFI (OR, 1.34; 95% CI, 0.56-3.17), 0.18 mFI (OR, 2.37; 95% CI, 0.83-6.76), and 0.27 mFI (OR, 6.11; 95% CI, 1.66-22.42). Unplanned reoperation (OR, 1.87; 95% CI, 1.23-2.86) and overall morbidity (OR, 1.30; 95% CI, 1.02-1.65) were associated with mFIs. Neither unplanned readmission (OR, 1.15; 95% CI, 0.52-2.55) nor mortality (OR, 0.78; 95% CI, 0.34-1.76) was associated with mFIs.

Fourteen (1.2%) patients died. There were no deaths among patients undergoing pharyngectomy or esophagectomy. Three patients undergoing mandibulectomy, 3 patients undergoing laryngectomy, and 8 patients undergoing glossectomy died. For mandibulectomy deaths, one had an mFI of 0, while the other 2 had mFIs of 0.09. The 3 laryngectomy deaths had mFIs of 0.09, 0.18, and 0.27. Of 8 glossectomy deaths, 3 died at 0 mFI, 4 died at 0.09 mFI, and 1 died at 0.27 mFI.

Discussion

Our study demonstrated that the mFI is significantly associated with CDIV complications after head and neck oncology operations. When procedures were stratified, higher mFIs were strongly associated with CDIV complications for all procedures. Patients undergoing pharyngectomy or esophagectomy had increasing rates of CDIV complications with increasing mFIs, but they were not significant, which was most likely owing to statistical power, with 40 pharyngectomy cases and 17 esophagectomy cases. The coefficients of determination (R2 values) all exceeded 0.60 for each procedure, indicating that the mFI is strongly correlative of postoperative ICU-level complications. Laryngectomy, pharyngectomy, and esophagectomy were strongly correlative, with R2 exceeding 0.93, which leads us to conclude that the mFI is an accurate predictor of CDIV complications for all major head and neck cancer operations. Quiz Ref IDThe strongest predictor of CDIV complications was the mFI, with increasing operative time being the only other significant predictor.Quiz Ref ID For every incremental increase in the mFI, there was a 68.0% greater risk of CDIV complications. More than 5% (65 of 1193) of our cohort had mFIs exceeding 0.27, indicating that they had a 200.0% increased risk of an ICU-level complication before undergoing surgery. The mFI was more predictive than either age or American Society of Anesthesiologists’ classification, factors previously found to increase postoperative complications.20,21 In addition, the mFI was associated with unplanned reoperation and overall morbidity but not with unplanned readmissions or mortality. The strong correlation with reoperation (R2 = 0.63) but not readmission (R2 = 0.31) indicates that these patients were most likely reoperated on during their initial hospital admission. These data are important to the practicing surgeon to know which patients are at high risk of needing ICU care. Using a rapidly calculable preoperative mFI could allow teams to ensure that ICU beds are available if needed, a widely varying institutionally dependent practice.

Increasing mortality was associated with higher mFIs, with linear regression demonstrating an R2 of 0.46, indicating moderate correlation. Because our patient cohort had only 14 deaths and a small sample, the precision of the estimate of correlation was wide, suggesting an inability to draw definitive conclusions from the point estimate of 0.46. With more years of the ACS-NSQIP data, we predict that mortality will be associated with the mFI, as was found by Adams et al.16 However, they did not separate CDIV complications from mortality; therefore, it is not possible to ascertain if the mFI is predictive of CDIV alone, mortality alone, or both. In addition, they studied a heterogeneous mixture of all otolaryngology procedures, with the most common being total thyroidectomy, cervical lymphadenectomy, total thyroid lobectomy, parotid tumor excision, and partial thyroid lobectomy. We excluded these procedures and focused on what the literature has defined as major operations.

Other indexes, including the Kaplan-Feinstein Comorbidity Index,22 Charlson Comorbidity Index,2325 and Adult Comorbidity Evaluation 27 index,25,26 have shown decreased survival rates and increased complications with higher indexes in head and neck oncology cases. The mFI would be especially useful in head and neck oncology owing to its emphasis on functional status, which is more complicated than simply adding up comorbidities. Moving forward, our research demonstrates that functional descriptions of illness should be emphasized over strict pathoanatomic descriptions of disease. With frailty being more predictive of postoperative complications than age itself, focusing on a patient’s health status rather than age can help reduce age discrimination.27,28

There are several limitations to this study, most stemming from the retrospective analysis of prospectively collected ACS-NSQIP data. While only 30-day outcomes can be assessed and no conclusions can be drawn after 30 days, most CDIV complications occur within this window before discharge. The ACS-NSQIP has defined preoperative comorbidity variables, as outlined in the Variables subsection of the Methods section, and the mFI was applied to those data. Fifteen ACS-NSQIP variables were matched to 11 CSHA-FI variables, with previous literature demonstrating that the CSHA-FI can be modified to as few as 10 variables.19 The mFI is based on the cumulative deficit model, and comparisons with a physical frailty phenotype are not possible owing to the absence of unintentional weight loss, subjective exhaustion, low grip strength, reduced walking speed, and low physical activity levels as ACS-NSQIP variables. Finally, direct retrospective comparison with the Kaplan-Feinstein Comorbidity Index, Charlson Comorbidity Index, and Adult Comorbidity Evaluation 27 is not possible owing to underrepresentation of their variables in the ACS-NSQIP. Direct comparisons between comorbidity indexes and frailty are needed in the head and neck oncology literature.

Ascertaining surgical candidacy is a complex and crucial step in treating patients with head and neck cancer. Variables like tumor factors, patient parameters, and clinical judgment yield the recommendation on which patients base their treatment decisions. However, clinical judgment is subjective, prone to bias, and poorly replicable between surgeons. Using a data-driven index emphasizing functional status to predict complications and mortality allows the patient to make a more informed decision.

Conclusions

We demonstrated that the mFI is predictive of postoperative critical care support and in-hospital reoperation after major surgery for head and neck cancer. Specifically, higher mFIs were strongly associated with CDIV complications for glossectomy, mandibulectomy, and laryngectomy. Because of the severity of ICU-level complications, understanding their risk before a patient undergoes surgery is imperative. Risk knowledge can improve patient care by making informed consent more data driven and transparent and can ensure that appropriate resources are available after surgery.

Back to top
Article Information

Accepted for Publication: March 13, 2016.

Corresponding Author: Nishant Agrawal, MD, Department of Otolaryngology–Head and Neck Surgery, The University of Chicago Medicine, 5841 S Maryland Ave, Mail Code 1035, Chicago, IL 60637 (nagrawal1@surgery.bsd.uchicago.edu).

Published Online: May 12, 2016. doi:10.1001/jamaoto.2016.0707.

Author Contributions: Mr Abt and Dr Agrawal had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Abt, Koch, Agrawal.

Acquisition, analysis, or interpretation of data: Abt, Richmon, Eisele, Agrawal.

Drafting of the manuscript: Abt, Agrawal.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Abt.

Administrative, technical, or material support: Abt, Agrawal.

Study supervision: Richmon, Koch, Eisele.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

References
1.
Siegel  R, Ma  J, Zou  Z, Jemal  A.  Cancer statistics, 2014.  CA Cancer J Clin. 2014;64(1):9-29.PubMedGoogle ScholarCrossref
2.
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.PubMedGoogle ScholarCrossref
3.
Corrêa  GT, Bandeira  GA, Cavalcanti  BG,  et al.  Analysis of ECOG performance status in head and neck squamous cell carcinoma patients: association with sociodemographical and clinical factors, and overall survival.  Support Care Cancer. 2012;20(11):2679-2685.PubMedGoogle ScholarCrossref
4.
Chang  PH, Yeh  KY, Huang  JS,  et al.  Pretreatment performance status and nutrition are associated with early mortality of locally advanced head and neck cancer patients undergoing concurrent chemoradiation.  Eur Arch Otorhinolaryngol. 2013;270(6):1909-1915.PubMedGoogle ScholarCrossref
5.
Bell  J.  Redefining disease.  Clin Med (Lond). 2010;10(6):584-594.PubMedGoogle ScholarCrossref
6.
Fried  LP, Tangen  CM, Walston  J,  et al; Cardiovascular Health Study Collaborative Research Group.  Frailty in older adults: evidence for a phenotype.  J Gerontol A Biol Sci Med Sci. 2001;56(3):M146-M156.PubMedGoogle ScholarCrossref
7.
Cesari  M, Gambassi  G, van Kan  GA, Vellas  B.  The frailty phenotype and the Frailty Index: different instruments for different purposes.  Age Ageing. 2014;43(1):10-12.PubMedGoogle ScholarCrossref
8.
Rockwood  K, Song  X, MacKnight  C,  et al.  A global clinical measure of fitness and frailty in elderly people . CMAJ. 2005;173(5):489-495.PubMedGoogle ScholarCrossref
9.
Uppal  S, Igwe  E, Rice  LW, Spencer  RJ, Rose  SL.  Frailty index predicts severe complications in gynecologic oncology patients.  Gynecol Oncol. 2015;137(1):98-101.PubMedGoogle ScholarCrossref
10.
Kolbe  N, Carlin  AM, Bakey  S, Louwers  L, Horst  HM, Rubinfeld  I.  Assessing risk of critical care complications and mortality in the elective bariatric surgery population using a modified frailty index.  Obes Surg. 2015;25(8):1401-1407.PubMedGoogle ScholarCrossref
11.
Tsiouris  A, Hammoud  ZT, Velanovich  V, Hodari  A, Borgi  J, Rubinfeld  I.  A modified frailty index to assess morbidity and mortality after lobectomy.  J Surg Res. 2013;183(1):40-46.PubMedGoogle ScholarCrossref
12.
Kolbe  N, Bakey  S, Louwers  L,  et al.  Predictors of Clavien 4 complications and mortality after necrosectomy: analysis of the NSQIP database.  J Gastrointest Surg. 2015;19(6):1086-1092.PubMedGoogle ScholarCrossref
13.
Melin  AA, Schmid  KK, Lynch  TG,  et al.  Preoperative frailty Risk Analysis Index to stratify patients undergoing carotid endarterectomy.  J Vasc Surg. 2015;61(3):683-689.PubMedGoogle ScholarCrossref
14.
Amrock  LG, Neuman  MD, Lin  HM, Deiner  S.  Can routine preoperative data predict adverse outcomes in the elderly? development and validation of a simple risk model incorporating a chart-derived frailty score.  J Am Coll Surg. 2014;219(4):684-694.PubMedGoogle ScholarCrossref
15.
Karam  J, Tsiouris  A, Shepard  A, Velanovich  V, Rubinfeld  I.  Simplified frailty index to predict adverse outcomes and mortality in vascular surgery patients.  Ann Vasc Surg. 2013;27(7):904-908.PubMedGoogle ScholarCrossref
16.
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.PubMedGoogle ScholarCrossref
17.
Mulvey  CL, Pronovost  PJ, Gourin  CG.  Hospital volume and failure to rescue after head and neck cancer surgery.  Otolaryngol Head Neck Surg. 2015;152(5):783-789.PubMedGoogle ScholarCrossref
18.
Saxton  A, Velanovich  V.  Preoperative frailty and quality of life as predictors of postoperative complications.  Ann Surg. 2011;253(6):1223-1229.PubMedGoogle ScholarCrossref
19.
Searle  SD, Mitnitski  A, Gahbauer  EA, Gill  TM, Rockwood  K.  A standard procedure for creating a frailty index.  BMC Geriatr. 2008;8:24.PubMedGoogle ScholarCrossref
20.
Rutledge  JW, Spencer  H, Moreno  MA.  Predictors for perioperative outcomes following total laryngectomy: a University HealthSystem Consortium discharge database study.  Otolaryngol Head Neck Surg. 2014;151(1):81-86.PubMedGoogle ScholarCrossref
21.
Reid  BC, Alberg  AJ, Klassen  AC, Koch  WM, Samet  JM.  The American Society of Anesthesiologists’ class as a comorbidity index in a cohort of head and neck cancer surgical patients.  Head Neck. 2001;23(11):985-994.PubMedGoogle ScholarCrossref
22.
Piccirillo  JF, Wells  CK, Sasaki  CT, Feinstein  AR.  New clinical severity staging system for cancer of the larynx: five-year survival rates.  Ann Otol Rhinol Laryngol. 1994;103(2):83-92.PubMedGoogle ScholarCrossref
23.
Singh  B, Bhaya  M, Stern  J,  et al.  Validation of the Charlson comorbidity index in patients with head and neck cancer: a multi-institutional study.  Laryngoscope. 1997;107(11, pt 1):1469-1475.PubMedGoogle ScholarCrossref
24.
Wang  JR, Habbous  S, Espin-Garcia  O,  et al.  Comorbidity and performance status as independent prognostic factors in patients with head and neck squamous cell carcinoma [published online December 18, 2014].  Head Neck.PubMedGoogle Scholar
25.
Paleri  V, Wight  RG, Silver  CE,  et al.  Comorbidity in head and neck cancer: a critical appraisal and recommendations for practice.  Oral Oncol. 2010;46(10):712-719.PubMedGoogle ScholarCrossref
26.
Datema  FR, Poldermans  D, Baatenburg de Jong  RJ.  Incidence and prediction of major cardiovascular complications in head and neck surgery.  Head Neck. 2010;32(11):1485-1493.PubMedGoogle ScholarCrossref
27.
Protière  C, Viens  P, Rousseau  F, Moatti  JP.  Prescribers’ attitudes toward elderly breast cancer patients: discrimination or empathy?  Crit Rev Oncol Hematol. 2010;75(2):138-150.PubMedGoogle ScholarCrossref
28.
Wong  J, Xu  B, Yeung  HN,  et al.  Age disparity in palliative radiation therapy among patients with advanced cancer.  Int J Radiat Oncol Biol Phys. 2014;90(1):224-230.PubMedGoogle ScholarCrossref
×