Association of Modified Frailty Index Score With Perioperative Risk for Patients Undergoing Total Laryngectomy

Key Points

Question

Does the Modified Frailty Index accurately measure the risk for postoperative complications, length of hospitalization, and discharge disposition for patients undergoing total laryngectomy?

Findings

This population-based analysis of 595 patients in the American College of Surgeons National Surgical Quality Improvement Program data registry indicated that a high level of baseline frailty in patients undergoing total laryngectomy is associated with (1) a significantly increased risk for overall complications, (2) prolonged postoperative hospitalization, and (3) an increased likelihood of requiring postdischarge skilled care.

Meaning

Physiologic decline as measured by the Modified Frailty Index is a significant determinant of possible postoperative outcomes following total laryngectomy.

Abstract

Importance

Objective preoperative risk assessment tools, such as the Modified Frailty Index (mFI), may inform patient and physician decision making when considering total laryngectomy. Estimation of outcomes may help to set realistic expectations about recovery and outcomes and facilitate optimal resource management.

Objective

To evaluate the association between the mFI score as a measure of frailty and outcomes following total laryngectomy.

Design, Setting, and Participants

Retrospective evaluation using the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP), a risk- and case-mix–adjusted national quality assessment program. The ACS NSQIP database identified 595 patients who underwent total laryngectomy between 2006 and 2012. Patients were assessed for demographics and comorbidity and were stratified on the basis of calculated mFI score. Outcomes, including postoperative complications, length of hospitalization, and discharge destination, were evaluated as a function of increasing frailty using multivariable logistic regression and Cox proportional hazards regression models.

Main Outcomes and Measures

Risk of postoperative complications, length of hospitalization, and discharge disposition.

Results

After exclusion of patients who experienced significant deviation from standard care protocols and those with missing or incomplete data, 343 individuals were included in the analysis. Of these, 278 (81.0%) were men, and the mean age was 63 years (95% CI, 61.9-64.4 years). Increasing frailty resulted in a nonlinear but progressive rise in incidence of postoperative adverse events. Overall, 96 (28.0%) patients experienced a postoperative complication, and patients with an mFI score of 3 or higher were more likely to develop postoperative complications than were patients with an mFI score of 0 (50.0% vs 16.7%; OR, 3.83; 95% CI, 1.72- 8.51). Patients in the highest frailty group experienced a longer mean duration of hospitalization (14.2 vs 9.5 days; difference, 4.7; 95% CI, 1.3-8.1 days) and were more likely to require skilled care after discharge (33.3% vs 3.2%; difference, 30.1%; 95% CI, 7.4%-52.9%).

Conclusions and Relevance

An mFI score of 3 or higher is associated with increased risk for postoperative complications, longer hospitalization, and need for postdischarge skilled care following total laryngectomy. The mFI provides a personalized risk assessment to better inform patients, physicians, and payers when planning a total laryngectomy.

This population-based study evaluates use of the Modified Frailty Index in estimation of outcomes in patients undergoing total laryngectomy

Introduction

Decision making regarding appropriate treatment for laryngeal cancer is complex. Fundamental to this process is accurate determination of probable surgical outcomes, which depends on interdigitating factors that relate to the patient, disease process, and surgical team and health care system.

Quality improvement efforts have led to increased attention to preoperative risk stratification. Prediction models range from the gestalt or “eyeball test” to more structured assessment tools, such as the American Society of Anesthesiology score, Charlson Comorbidity Index, and American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) Surgical Risk Calculator. Each tool varies in level of objectivity, ease of use, and practical applicability. Although nominal or ordinal recognition of preoperative medical conditions as contributors toward perioperative risk is important, the identification of functional and physiologic decline (collectively measured as frailty) as a critical component of a person’s ability to withstand and recover from major surgery has garnered significant interest.

Frailty may be assessed by physical phenotype (eg, unintentional weight loss, loss of grip strength, walking speed, and other physical metrics) or multidomain assessments of accumulated deficits (eg, physical, cognitive, functional, and social realms). The Canadian Study of Health and Aging Frailty Index is the basis of the simplified Modified Frailty Index (mFI), which has been applied to diverse surgical settings, including colorectal, gynecologic, vascular, spine, and head and neck surgery.

This simple, objective, and predictive tool, based on 11 variables, provides a practical platform to gauge frailty during preoperative assessment. The variables used for mFI (Box) help to estimate clinical outcomes and quantify risk of complications based on accumulated deficits independent of age. The effect of frailty on mortality and severe complications (Clavien-Dindo grade IV complications) has been demonstrated in a variety of otolaryngology and selected major head and neck surgery procedures. The Clavien-Dindo measure includes 5 grades: grade I indicates any deviation from the normal postoperative course without the need for pharmacologic treatment or surgical, endoscopic, and radiologic interventions; grade V indicates death of the patient.

Box. The 11 Variables Included in Modified Frailty Index Assessment.

• History of diabetes

• Functional status (not independent at baseline)

• History of pneumonia or chronic obstructive pulmonary disease

• History of congestive cardiac failure

• History of myocardial infarction

• History of percutaneous coronary intervention, stent placement, or angina

• History of hypertension (requiring medical treatment)

• 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

We report on the association between mFI and outcomes for patients undergoing total laryngectomy with specific attention to overall morbidity, length of hospitalization, and postdischarge disposition. Patients who have undergone total laryngectomy are an excellent group to study since they typically have shared baseline risks (age and history of smoking and alcohol use) and receive interventions that are consistent in procedural and resource intensity. Improvements in preoperative risk assessment for this group may supplement selection of treatment strategy and perioperative management for patients in whom complications may be debilitating.

Methods

The ACS NSQIP is a national risk- and case-mix–adjusted database that collects deidentified information from over 700 diverse institutions. Surgical case reviewers record preoperative demographics and comorbidities, intraoperative variables, and 30-day outcomes through independent review of medical records. Since all information in the data set is deidentified by NSQIP prior to its availability to the investigator, this investigation remains exempt from review and approval by the institutional review board at University of Nebraska Medical Center as defined in its policies and procedures.

A total of 595 patients who underwent total laryngectomy, with or without a neck dissection (Current Procedural Terminology codes 31360 and 31365), between 2006 and 2012 were initially identified. Patients who underwent concurrent procedures that represented significant deviation in procedural complexity (eg, concurrent cardiothoracic procedures, exploratory laparotomy, and major vascular reconstruction and/or bypass) were excluded. Patients who underwent free tissue transfer to reconstruct pharyngeal defects were not included to limit the effect of additional surgical sites, added surgical time, and variation in postsurgical care that may complicate the comparison. Following elimination of patients with missing information (Figure), data from 343 patients were evaluated for baseline demographics, preoperative comorbidities, and American Society of Anesthesiology assessment status. The mFI was calculated by identification of variables listed in the Box, with a range of 0 to 11.

Figure. Patient Selection Algorithm.

CPT indicates Current Procedural Terminology; mFI, Modified Frailty Index.

Three hundred twenty-one patients (93.6%) had an mFI level between 0 and 3 (Table 1). Few patients demonstrated mFI values of 4, 5, and 6. No patients had mFI values higher than 6. To avoid statistical biases created by frailty groups with small numbers, patients were stratified into 4 cohorts based on mFI values of 0, 1, 2, and 3 or higher. Comparative analyses were performed for postoperative outcomes, including complications and length of hospitalization. Outcomes related to postoperative complications were adjusted for possible confounding variables. Additional subset analysis was performed to identify discharge disposition in patients with differing levels of frailty.

Table 1. Distribution of mFI Values Across Patients Undergoing Total Laryngectomy^a.

mFI Value	No. (%)
	Frequency	Cumulative Frequency
0	102 (29.7)	102 (29.7)
1	109 (31.8)	211 (61.5)
2	78 (22.7)	289 (84.3)
3	32 (9.3)	321 (93.6)
4	13 (3.8)	334 (97.4)
5	7 (2.0)	341 (99.4)
6	2 (0.6)	343 (100)

Abbreviation: mFI, Modified Frailty Index.

^aMinimum permissible value, 0; maximum permissible value, 11.

Statistical Analysis

Patients with an mFI value of 0 were identified as controls, and cohorts with higher mFI values were compared against this group. Pearson and Mantel-Haenszel χ² tests were used to determine differences in outcomes. Bivariate associations between the mFI and preoperative measures were evaluated using χ² tests for categorical measures and analysis of variance models for continuous measures, with effect sizes reported using Cramer V and η² with accompanying 95% CI, respectively. The bivariate effects of preoperative measures and mFI on outcomes, such as postsurgical complications, length of total hospital stay, and discharge to a facility other than home, were evaluated using logistic regression models and Cox proportional hazards regression models, with effect sizes reported as odds ratios (ORs) or hazard ratios and 95% CIs as appropriate. In addition, multivariable logistic regression and Cox proportional hazards regression models were used to evaluate the effect of mFI on outcomes controlling for preoperative measures with 95% CIs for OR and hazard ratios that did not include 1.0 in bivariate analyses. All analyses were conducted using SAS, version 9.3 (SAS Institute).

Results

A total of 343 patients who underwent a total laryngectomy, with or without a neck dissection, were included in the study. Of patients with race/ethnicity reported, 254 (82.2%) were white; the mean age was 63 years (95% CI, 61.9-64.4) overall (mean age was 58 [95% CI, 55.4-60.0], 64 [95% CI, 62.4-66.5], 64 [95% CI, 61.5-66.6], and 70 years [95% CI, 66.6-72.8] for mFI groups 0, 1, 2, and 3 or higher, respectively) (η² = 0.12). Most patients were men (278 [81.0%]), and this sex distribution persisted across individual mFI groups (proportion of men in mFI groups 0, 1, 2, and ≥3 was 71.6%, 86.2%, 82.1%, and 87.0%, respectively) (Cramer V, 0.16). The mean body mass index (calculated as weight in kilograms divided by height in meters squared) for the entire patient group was 24.4 (95% CI, 23.7-25.2). Body mass index (calculated as weight in kilograms divided by height in meters squared) was 24.3 (95% CI, 23.0-25.6), 22.8 (95% CI, 21.7-23.8), 25.0 (95% CI, 23.7-26.4), and 27.1 (95% CI, 24.1-30.0) for mFI groups 0, 1, 2, and 3 or higher, respectively (η² = 0.04). Overall, 58 (16.9%) patients had diabetes, and 174 (50.7%) patients had a history of smoking. Forty-nine (14.3%) patients in the study reported heavy alcohol use, 16 (4.7%) were corticosteroid dependent, and 57 (16.6%) exhibited a 10% or more weight loss over the preceding 6 months. A small number of patients had a history of chemotherapy (9 [2.6%]) or radiotherapy (13 [3.8%]), and 46 (13.4%) had undergone a surgical procedure within 30 days prior to undergoing a total laryngectomy. The mean operative duration for total laryngectomy in the patient cohort was 380 minutes, and no significant differences were observed between the groups when stratified by frailty.

The mFI scores ranged between 0 and 6 (Table 1). However, for most patients, mFI scores were clustered toward the low end of the observable range. A total of 321 (93.6%) patients had mFI values between 0 and 3. A total of 102 (29.7%) patients had mFI values of 0, and this group served as the control cohort. Cohorts with mFI values of 1, 2, and 3 or higher accounted for 109 (31.8%), 78 (22.7%), and 54 (15.7%) patients, respectively. Additional breakdown of variables contributing toward the composite mFI score are reported in the eTable in the Supplement.

Overall, 96 (28.0%) patients experienced a medical or surgical complication, and 29 (8.5%) patients experienced a severe (Clavien-Dindo grade IV) complication (Table 2). Patients undergoing a total laryngectomy spent a mean of 11.7 days in the hospital (95% CI, 10.6-12.7). In the subset of 130 patients for whom discharge disposition was recorded, 19 (14.6%) individuals required discharge to an intermediate skilled care facility.

Table 2. Outcomes by mFI.

Characteristic	Outcomes
	No. (%)a			Total Length of Hospital Stay, Mean (95% CI), d
	Any Complication	Clavien-Dindo Grade IV Complicationsb	Discharge to Skilled Care Facility
Overall patient cohort, No.	343	343	130	343
mFI category
0	17 of 102 (16.7)	4 of 102 (3.9)	1 of 31 (3.2)	9.5 (8.3-10.8)
1	32 of 109 (29.4)	11 of 109 (10.1)	8 of 43 (18.6)	12.7 (10.5-14.8)
2	20 of 78 (25.6)	6 of 78 (7.7)	3 of 35 (8.6)	11.3 (9.1-13.5)
≥3	27 of 54 (50.0)	8 of 54 (14.8)	7 of 21 (33.3)	14.2 (11.0-17.4)
Total	96 of 343 (28.0)	29 of 343 (8.5)	19 of 130 (14.6)	11.7 (10.6-12.7)

Abbreviation: mFI, Modified Frailty Index.

^aPercentages stated as proportion of patients in mFI category.

^bThe Clavien-Dindo measure includes 5 grades: grade I indicates any deviation from the normal postoperative course without the need for pharmacologic treatment or surgical, endoscopic, and radiologic interventions; grade V indicates death of the patient.

The risks for developing any perioperative medical or surgical complication, risk of Clavien-Dindo grade IV complications, likelihood of discharge to a skilled care facility, and proportional hazard for time to discharge from acute care hospitalization were studied as a function of preoperative variables, including mFI, sex, race, age, body mass index, and other factors. Multivariable analyses controlling for preoperative variables (Table 3) suggested that patients with mFI values of 3 or higher had an increased likelihood of developing any postoperative complication compared with the control group (OR, 3.83; 95% CI, 1.72-8.51). The incidence of severe complications that required intensive care management (Clavien-Dindo grade IV complications) did not differ significantly by increasing mFI category compared with the reference group (risk for severe complication with mFI≥3 vs 0: OR, 1.48; 95% CI, 0.37-5.97).

Table 3. Multivariable Associations of Preoperative Measures With Patient Outcomes.

Preoperative Measure	Patient Outcomesa
	OR (95% CI)			Total Length of Hospital Stay, Time to Discharge, HR (95% CI)
	Any Complication	Clavien-Dindo Grade IV Complicationb	Discharge to Skilled Care Facility
mFI value
1 vs 0	1.78 (0.89-3.55)	1.76 (0.50-6.18)	6.28 (0.73-53.80)	0.77 (0.57-1.04)
2 vs 0	1.46 (0.69-3.10)	1.20 (0.31-4.74)	3.04 (0.30-31.13)	0.82 (0.59-1.12)
≥3 vs 0	3.83 (1.72-8.51)	1.48 (0.37-5.97)	14.12 (1.56-127.62)	0.61 (0.42-0.87)
Race (other vs non-Hispanic white)				0.69 (0.51-0.93)
Age, y	1.03 (1.01-1.05)	1.09 (1.04-1.14)
BMI		1.05 (1.00-1.11)		1.01 (1.00-1.03)
Corticosteroid dependence (yes vs no)c			3.68 (0.67-20.17)
Antecedent (yes vs no)
Chemotherapyd	6.46 (1.44-28.99)	9.08 (1.74-47.52)
Radiotherapye
Surgeryf				0.69 (0.48-0.97)

Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); HR, hazard ratio; mFI, Modified Frailty Index; OR, odds ratio.

^aEmpty cells indicate test was not performed because the variable did not meet criteria for inclusion in the multivariable model.

^bThe Clavien-Dindo measure includes 5 grades: grade I indicates any deviation from the normal postoperative course without the need for pharmacologic treatment or surgical, endoscopic, and radiologic interventions; grade V indicates death of the patient.

^cCorticosteroid use for chronic condition.

^dChemotherapy for cancer 30 or fewer days preceding the operation.

^eRadiotherapy for cancer within the preceding 90 days.

^fSurgery within the preceding 30 days.

Other factors that were independently associated with an increased risk for development of any complication and severe complications included age (any complication: OR, 1.03; 95% CI, 1.01-1.05; severe complication: OR, 1.09; 95% CI, 1.04-1.14) and history of chemotherapy (any complication: OR, 6.46; 95% CI, 1.44-28.99; severe complication: OR, 9.08; 95% CI, 1.74-47.52) (Table 3).

With increasing mFI, there was a nonlinear elevation in length of hospital stay (Table 2). Patients with a high frailty profile (mFI≥3) demonstrated the greatest need for prolonged hospitalization and exceeded the mean length of stay in the control group by 4.7 days (14.2 vs 9.5; 95% CI, 1.3-8.1 days). The hazard ratio for time to discharge (predicting the likelihood of timely discharge) for patients in this high frailty group was 0.61 (95% CI, 0.42-0.87) (Table 3).

As reported in Table 2, patients in the highest frailty group were more likely to require skilled care after discharge (mFI≥3 vs 0; 33.3% vs 3.2%; difference of 30.1%; 95% CI, 7.4%-52.9%). The independent odds for patients with high frailty (mFI≥3) requiring skilled care following hospital discharge was 14.12 (95% CI, 1.56-127.62) (Table 3).

The overall incidence of mortality in the study population was low. There were only 4 deaths across the examined cohort, and the event rate was too low to detect meaningful differences between the various mFI groups. Similarly, there were no significant differences in the likelihood of wound complications based on differences in frailty scores.

Discussion

Several studies have validated the mFI as a reliable predictor of postoperative outcomes. Physiologic decline as measured by the mFI may be a key determinant of postoperative outcomes following total laryngectomy. Our study suggests that high baseline frailty in patients undergoing total laryngectomy is associated with a significantly increased risk for overall complications, prolonged postoperative hospitalization, and increased likelihood of requiring postdischarge skilled care.

Patients undergoing total laryngectomy experience significant changes in anatomy and physiology and are at risk for significant postoperative complications. These patients represent a group with significant comorbidities and heavy resource utilization. In addition, patients with laryngeal cancer may have alternative management strategies, such as radiotherapy and chemoradiotherapy, available to them. Therefore, identifying patients at increased risk for adverse postsurgical outcomes is especially critical.

In this study, 93.6% of patients had mFI values between 0 and 3, and the highest value observed was 6 on a scale of 0 to 11. This finding may reflect the realities of clinical practice in which patients who are too frail or ill to undergo surgery may be offered nonsurgical management strategies.

Patients with mFI values of 3 or higher who underwent total laryngectomy had a significantly higher overall likelihood of developing postoperative complications independent of preoperative variables, such as age, sex, body mass index, history of smoking and alcohol use, or prior exposure to radiotherapy or chemotherapy. However, frailty had no effect on the likelihood of Clavien-Dindo grade IV complications. Our findings are consistent with those of Adams and colleagues, who found an increase in overall complication rates for frail patients undergoing inpatient head and neck surgery. Another previous pooled analysis of major head and neck cancer operations found no significant effect of frailty on overall morbidity. Although that article did not describe the effect of frailty on overall risk for postoperative complications specific to total laryngectomy, a correlation with severe complications was reported. In the present study, we observed no significant association of frailty with Clavien-Dindo grade IV complications. This difference may be due to a more homogeneous population of patients who underwent total laryngectomy in our study. Furthermore, the exclusion of patients who underwent free flap reconstruction may have enriched the sample with patients who did not have a history of radiotherapy or chemotherapy, further augmenting this distinction. In our study, the total laryngectomy-specific incidence of Clavien-Dindo grade IV complications was low, and the effect of frailty on this set of complications may become more apparent with a larger pool of patients.

We found that more than 1 in 4 patients who underwent a total laryngectomy developed medical or surgical complications, most of which did not require intensive care unit management. The incidence of complications was significantly increased in patients with the highest frailty scores (mFI≥3). Thus, the cumulative burden of morbidity experienced by patients with poor reserves at baseline may translate to inability to salvage, increased cost, higher resource utilization, and overall poor outcomes.

The preoperative identification of frailty presents an opportunity for personalized counseling regarding anticipated risks, allowing more realistic expectations related to recovery. In its departure from using the same approach for all patients, mFI calculation may assist with the process of individualized risk assessment and may favorably affect preoperative decision making and postoperative recovery planning for patients and medical providers alike.

The data indicate that patients with increased frailty are likely to experience prolonged hospitalization. Although patients in all categories of frailty demonstrated lengthy inpatient stay, this difference was most pronounced for those with mFI values of 3 or higher. Measurement of the mFI in patients undergoing total laryngectomy quantifies the intuitive sense that patients with an increased risk of complications will likely require a longer hospitalization. Potential application of mFI may help to improve allocation of hospital-based resources, such as bed allotment. These findings may be of additional interest to health insurers who may otherwise underestimate frailty’s effect on inpatient utilization.

In addition, we found that 1 in 3 patients in the highest frailty group required discharge to an intermediate care facility, such as an acute care rehabilitation or skilled nursing home. Although data about eventual disposition and return to presurgical functional status of such patients are not available, these findings emphasize that patients with high baseline frailty have a significantly different course during postsurgical recovery. The data further underline how frailty may amplify the sequelae of periprocedural physiologic challenges. The discharge destination may have serious social implications (eg, displacement from home environment), psychological impact (eg, depression), and burdensome effects on the health care economy. Predicting the discharge destination proactively will improve discharge planning and reduce unintended hospital stay.

Limitations

We recognize that the retrospective nature of data has the potential for underrepresentation of frailty in the patient cohort if baseline comorbid conditions were inadequately documented in the medical record. However, the data-gathering mechanisms built into the ACS NSQIP maximize data accuracy and validity through independent review and accrual of data through careful review of clinical records from multiple sources, low rate of variance between data abstracters, and risk-adjusted nature of data collection. Further, due to the limitations of available information, the frailty data cannot be compared head to head with other measures of comorbidity, such as Adult Comorbidity Evaluation, which may independently provide prognostic information for patients with cancer.

Management of care for patients who are traditionally considered for total laryngectomy may be influenced by multiple factors that relate to underlying disease, treatment history, patient preferences, and expert recommendations from the treating medical clinicians. Available options may vary greatly between laryngeal-preserving surgery for patients with limited disease burden, laryngeal-preserving radiotherapy or chemoradiotherapy for patients with acceptable comorbidity profile and those for whom loss of natural voice is unacceptable, and primary total laryngectomy when disease burden or preexisting comorbidity (eg, chronic obstructive pulmonary disease or laryngeal dysfunction) precludes laryngeal preservation. Estimation of frailty may further aid decision making when generating recommendations for treatment.

In addition, patients who require free tissue transfer (due to treatment failure of laryngeal preservation strategies, postradiotherapy laryngeal dysfunction, or need for total pharyngectomy) present a profile associated with longer operative times, increased procedural complexity, and greater intensity of postoperative care. To prevent confounding factors that could emerge from a heterogeneous patient cohort, the present analysis focused on patients who underwent total laryngectomy and associated procedures without the use of free tissue transfer for reconstruction. The homogeneity in nature and complexity of interventions in this study improves the ability to generate clinical information specifically pertinent to patients considering total laryngectomy.

Clearly, frailty cannot and should not be the predominant determinant of any proposed management strategy but may better inform the patient and physicians when considering the risks and benefits of treatment choices. Systematic assessment of frailty may assist the identification and quantification of perioperative risks for patients with improved accuracy, reproducibility, and objectivity.

Conclusions

Frailty is one determinant of postoperative outcomes following total laryngectomy. Patients with high frailty (mFI≥3) have a significantly higher likelihood of overall complications, prolonged hospitalization, and need for skilled postdischarge care. These findings may aid in individualizing preoperative risk counseling, justify hospital-based resource allocation, and improve discharge planning for patients with higher baseline frailty.

Supplement.

eTable. Distribution of Baseline Comorbidities That May Contribute to mFI Score Across Patient Groups