- Email: [email protected]
Frailty and postoperative outcomes in patients undergoing surgery for degenerative spine disease
ARTICLE IN PRESS
The Spine Journal ■■ (2016) ■■–■■
Clinical Study
Frailty and postoperative outcomes in patients undergoing surgery for degenerative spine disease Alana M. Flexman, MDa,*, Raphaële Charest-Morin, MDb, Liam Stobart, MDa, John Street, MD PhDc, Christopher J. Ryerson, MD MASd a
Department of Anesthesiology, Pharmacology and Therapeutics, Vancouver General Hospital, University of British Columbia, Room 2449 JPP 899 West 12th Avenue, Vancouver, BC, Canada, V5Z 1M9 b Department of Orthopedic Surgery, Centre Hospitalier Universitaire de Québec, 1401 18e rue, Local B-2408, Québec, QC, Canada, G1J 1Z4 c Combined Neurosurgical and Orthopedic Spine Program, Department of Orthopedic Surgery, University of British Columbia, 818 West 10th Avenue, Vancouver, BC, Canada, V5Z 1M9 d Division of Respirology, Department of Medicine, St. Paul’s Hospital, University of British Columbia, Ward 8B, 1081 Burrard Street, Vancouver, BC, Canada, V6Z 1Y6 Received 22 December 2015; revised 31 May 2016; accepted 21 June 2016
Abstract
BACKGROUND CONTEXT: Frailty is defined as a state of decreased reserve and susceptibility to stressors. The relationship between frailty and postoperative outcomes after degenerative spine surgery has not been studied. PURPOSE: This study aimed to (1) determine prevalence of frailty in the degenerative spine population; (2) describe patient characteristics associated with frailty; and (3) determine the association between frailty and postoperative complications, mortality, length of stay, and discharge disposition. STUDY DESIGN: This is a retrospective analysis on a prospectively collected cohort from the National Surgical Quality Improvement Program (NSQIP). PATIENT SAMPLE: A total of 53,080 patients who underwent degenerative spine surgery between 2006 and 2012 were included in the study. OUTCOME MEASURES: A modified frailty index (mFI) with 11 variables derived from the NSQIP dataset was used to determine prevalence of frailty and its correlation with a composite outcome of perioperative complications as well as hospital length of stay, mortality, and discharge disposition. METHODS: After calculating the mFI for each patient, the prevalence and predictors of frailty were determined for our cohort. The association of frailty with postoperative outcomes was determined after adjusting for known and suspected confounders using multivariate logistic regression.
FDA device/drug status: Not applicable. Author disclosures: AMF: Grant: Department Internal Funds–Research Merit Award (C, Paid directly to the author), pertaining to the submitted manuscript; Speaking and/or Teaching Arrangements: Hospira, Inc (A, Paid directly to the author); Research Support (Investigator Salary, Staff/Materials): Department Internal Funds–Research Merit Award (C [Investigator Salary], Paid directly to the author); Grants: Canadian Anesthesiologists’ Society Research Grant (B, Paid directly to the institution/employer). RC-M: Nothing to disclose. LS: Nothing to disclose. JS: Consulting: DePuy Synthes (B, Paid directly to the author), Medtronic (B, Paid directly to the author); Speaking and/or Teaching Arrangements: Medtronic (Paid directly to the author); Trips/Travel: Medtronic (B, Paid directly to the institution/employer); Scientific Advisory Board/Other Office: DePuy Synthes (B, Paid directly to the author); Grants: Medtronic (E, Paid directly to the institution/employer); Fellowship Support: Medtronic (B, Paid directly to the institution/employer), outside the submitted work. CJR: Speaking and/or Teaching Arrangements: Boehringer Ingelheim (B, Paid directly to the author), InterMune/ Hoffmann La Roche (C, Paid directly to the author); Trips/Travel: InterMune (B, Paid directly to the author); Scientific Advisory Board/Other Office: InterMune/Hoffmann La Roche (B, Paid directly to the author), Boehringer Ingelheim (B, Paid directly to the author); Research Support (Investigator http://dx.doi.org/10.1016/j.spinee.2016.06.017 1529-9430/© 2016 Elsevier Inc. All rights reserved.
Salary, Staff/Materials): InterMune/Hoffmann La Roche (E, Paid directly to the institution/employer); Grants: InterMune/Hoffmann La Roche (F, Paid directly to the institution/employer), Boehringer Ingelheim (I, Paid directly to the institution/employer), outside the submitted work. The American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) and the hospitals participating in the ACS NSQIP are the source of the data used herein; they have not verified and are not responsible for the statistical validity of the data analysis or the conclusions derived by the authors. The authors have no conflicts of interest to disclose. This study did not receive funding. Dr. Alana Flexman is supported by a Vancouver Acute Department of Anesthesia Research Merit Award (Internal Funds). Dr. Christopher Ryerson is supported by a Career Investigator Award from the Michael Smith Foundation for Health Research. The disclosure key can be found on the Table of Contents and at www.TheSpineJournalOnline.com. * Corresponding author. Department of Anesthesiology, Pharmacology and Therapeutics, University of British Columbia, Vancouver General Hospital, Room 2449 JPP 899 West 12th Avenue, Vancouver, BC, Canada, V5Z 1M9. Fax +1 604 875 5209. E-mail address: [email protected] (A.M. Flexman).
ARTICLE IN PRESS 2
A.M. Flexman et al. / The Spine Journal ■■ (2016) ■■–■■
RESULTS: Frailty was present in 2,041 patients within the total population (4%) and in 8% of patients older than 65 years. Frailty severity increased with increasing age, male sex, African American race, higher body mass index, recent weight loss, paraplegia or quadriplegia, American Society of Anesthesiologists (ASA) score, and preadmission residence in a care facility. Frailty severity was an independent predictor of major complication (OR 1.15 for every 0.10 increase in mFI, 95%CI 1.09–1.21, p<.0005) and specifically predicted reoperation for postsurgical infection (OR 1.3, 95%CI 1.16–1.46, p<.0005). Prolonged length of stay and discharge to a new facility were also independently predicted by frailty severity (p<.0005). Frailty severity predicted 30-day mortality on unadjusted (OR 2.05, 95%CI 1.70–2.48, p<.0005) and adjusted analyses (OR 1.48, 95%CI 1.18–1.86, p<.0005). CONCLUSIONS: Frailty is an important predictor of postoperative outcomes following degenerative spine surgery. Preoperative recognition of frailty may be useful for perioperative optimization, risk stratification, and patient counseling. © 2016 Elsevier Inc. All rights reserved. Keywords:
Frailty; Degenerative spine disease; Morbidity; Outcomes; Risk stratification; Spine surgery
Introduction
Materials and methods
Frailty is defined as a state of increased vulnerability to poor resolution of homoeostasis after a stressor event, which increases the risk of adverse outcomes [1]. Frailty represents a state of weakened reserve against even minor stressors and may occur independent of and out of proportion to chronological age [1,2]. The prevalence of frailty in the nonsurgical hospital population is approximately 10%, although this varies with the population and method of frailty measurement [3]. The prevalence of frailty in community dwelling individuals increases with age, and even mild frailty is associated with worsening disability, admission to hospital, and death in the community-dwelling elderly population [4]. Frailty is more prevalent in the surgical population (42%–50%) compared with the nonsurgical elderly population (4%–10%), and is likely to be an important predictor of outcome in patients undergoing spine surgery [1]. Preoperative frailty has been independently associated with increased morbidity and mortality in several surgical populations [5–9]. These studies showed that frail surgical patients experienced higher rates of postoperative complications, length of stay, and mortality, and were more likely to be discharged to a care facility rather than home [7,8,10]. As the rate of complications after spine surgery is high [11] and the mean age of patients undergoing spine surgery is also increasing [12,13], frailty is likely to be increasingly common and relevant to this population. Despite these potential implications, little is known about the prevalence of frailty in the spine surgery population or on the impact of frailty on postoperative outcomes in this vulnerable population. Our study objectives were to (1) determine the prevalence of frailty in patients undergoing surgery for degenerative spine disease; (2) describe the patient characteristics associated with frailty; and (3) determine the relationship between frailty and postoperative complications, length of stay, discharge disposition, and mortality in this population.
We performed a retrospective analysis of the prospectively collected American College of Surgeons National Surgical Quality Improvement Program (NSQIP) database. Study population Data were extracted from the American College of Surgeons NSQIP database following approval from our Institutional Clinical Research Ethics Board (H12-03433). The NSQIP database is a prospective multicenter database of adult patients that has been described previously [14]. Highquality data collection and entry are ensured through rigorous training of the clinical reviewer as well as inter-rater reliability audits to ensure accurate and consistent data entry. Cases are sampled using an 8-day cycle to ensure an even case mix, and operating room logs are audited to ensure representative sampling of cases. The most recent year included in this study (2012) includes 543,885 cases from 374 centers. We identified patients undergoing spine surgery for degenerative spine conditions in the NSQIP database from 2006 to 2012. We first identified patients undergoing any type of spine surgery by selecting all common procedural terminology (CPT) codes related to the spine. We used International Classification of Diseases and CPT codes to further restrict our population to spine surgery patients with a primary diagnosis of degenerative disease (Supplementary Appendix, Table S1). We excluded patients who did not receive a general anesthetic. Predictor variables We calculated a modified frailty score and modified frailty index (mFI) for each patient using a previously described method [9,15]. The mFI is a simplified form of the Canadian Study of Health and Aging Frailty Index [16,17]. The Canadian Study of Health and Aging Frailty Index is based on the theory of “accumulating deficits” and strongly correlates with overall mortality in community-dwelling adults
ARTICLE IN PRESS A.M. Flexman et al. / The Spine Journal ■■ (2016) ■■–■■
[16,17]. The mFI incorporates 11 NSQIP variables (deficits) and similarly predicted postoperative morbidity and mortality in a broad surgical population [9]. The modified frailty score was calculated based on the number of deficits present preoperatively, including dependent functional status, diabetes mellitus, lung problems, congestive heart failure, myocardial infarction, cardiac problems, hypertension, impaired sensorium, prior transient ischemic attack, history of stroke, and peripheral vascular disease (Table 1). The mFI was calculated as the modified frailty score (ie the number of deficits present) divided by 11, thus providing an index with a range of 0 to 1. We categorized patients as not frail (mFI=0), pre-frail (mFI>0 and <0.21), and frail (≥0.21) based on previous data defining frailty as an index greater than 0.21 [18]. Additional candidate predictor variables for major complications included patient- and procedure-specific variables. Patient-specific predictor variables included age, sex, race, American Society of Anesthesiologists’ (ASA) classification, body mass index, recent weight loss, paraplegia, quadriplegia, and preoperative residence in a care facility. Patients older than 90 years are listed as 90+ in the NSQIP database to maintain de-identification; these 119 patients (0.2% of the total cohort) were assigned an age of 90 for the purposes of the analysis. Procedure-specific predictor variables included year of operation, emergency surgery, surgical site, and surgical approach. Surgical variables (eg site, level, presence of fusion, or multiple levels) were determined using CPT codes (Supplementary Appendix, Table S2). We estimated sur-
3
gical complexity using the standardized work relative value unit (RVU) as previously described [19], and also considered whether surgery encompassed multiple vertebral levels and whether a spinal fusion was performed. Outcome variables The primary outcome variable was a composite of major complications within 30 days of surgery (Table 2). Complications were considered major if classified as grade 2 or above using the validated Clavien-Dindo classification system of postoperative complications [20]. Patients were assigned either 0 (no complication) or 1 (the presence of any complication). Secondary outcomes included hospital length of stay, discharge to a facility that was not home (available for 2011– 2012 only), and death within 30 days of surgery. Statistical analysis Data are described as mean (standard deviation) or median (interquartile range [IQR]). The prevalence of frailty was determined both in the overall population and in those over 65 years of age. The association of mFI with baseline patientand procedure-specific variables was determined using a Wilcoxon rank sum test or Spearman correlation coefficient for dichotomous and continuous predictor variables, respectively. The association of mFI with postoperative outcomes was determined using logistic regression analyses. The mFI was first used as a single predictor variable in unadjusted
Table 1 Components of the CSHA-FI and NSQIP modified frailty index [9] CSHA-FI variable
NSQIP variable
Changes in everyday activity –Problems getting dressed –Problems with bathing –Problems with personal grooming –Problems cooking –Problems going out alone History of diabetes mellitus
Functional status –Partially dependent OR –Totally dependent
1,783 (3.4)
Diabetes mellitus-non-insulin OR Diabetes mellitus-insulin History of severe COPD Current pneumonia Congestive heart failure History of myocardial infarction in the past 6 months Previous percutaneous coronary intervention Previous cardiac surgery History of angina Hypertension requiring medication Impaired sensorium
7,928 (15.1)
Lung problems –Respiratory problems Congestive heart failure Myocardial infarction Cardiac problems
Arterial hypertension Clouding or delirium –History relevant to cognitive impairment or loss –Family history relevant to cognitive impairment Cerebrovascular problems History of stroke Decreased peripheral pulses
History of transient ischemic attack History of cerebrovascular accident or stroke with neurologic deficit History of revascularization for peripheral vascular disease OR Rest pain or gangrene
Number (%)
1,936 (3.7) 88 (0.2) 35 (0.1) 2,347 (4.5)
25,422 (48.3) 39 (0.1)
603 (1.1) 314 (0.6) 285 (0.5)
Abbreviations: CSHA-FI, Canadian Study of Health and Aging Frailty Index; NSQIP, National Surgical Quality Improvement Program; COPD, chronic obstructive pulmonary disease. Adapted from Ref. [9].
ARTICLE IN PRESS 4
A.M. Flexman et al. / The Spine Journal ■■ (2016) ■■–■■
Table 2 Components of the major complication outcome variable
Characteristics of frailty
Variable description
Incidence (%)*
Return to the operating room within 30 days Unplanned reintubation Failure to wean from a ventilator >48 hours Cardiac arrest Myocardial infarction Sepsis or septic shock Stroke Pulmonary embolism Deep vein thrombosis requiring treatment Acute renal failure requiring dialysis Urinary tract infection Pneumonia Wound dehiscence Deep incisional surgical site infection Organ-space surgical site infection Total
1,361 (2.6) 211 (0.4) 171 (0.3) 72 (0.1) 104 (0.2) 344 (0.7) 53 (0.1) 182 (0.4) 271 (0.5) 65 (0.1) 661 (1.3) 272 (0.5) 97 (0.2) 249 (0.5) 106 (0.2) 2,881 (5.5)
Abbreviations: NSQIP, National Surgical Quality Improvement Program. * Complications are not mutually exclusive (eg patients with reoperation because of a surgical site infection would be included in both categories). The “Total” represents the number of patients with at least one complication.
analysis for each outcome. A stepwise multivariate analysis with backward elimination was then used to determine the independent association of frailty index with each outcome, adjusting for prespecified patient- and procedure-specific variables, as listed above. Variables that were independently associated with the outcome were retained in the final multivariate model. Patients with missing variables were not included in the final model. Model performance was described using the area under the receiver operating characteristic (AUROC) curve and the goodness of fit test, where poor fit is indicated by p<.05. A two-sided p value <.05 was considered significant for all analyses. All data analysis was performed using STATA 11.0 (StataCorp, College Station, TX, USA). Results Study population The study population included a total of 53,145 patients who underwent spine surgery between 2006 and 2012, inclusive (Table 3). A total of 474 patients were excluded, as they did not receive a general anesthetic, leaving 52,671 patients for the analysis. The number of spine surgery procedures per year increased from 414 in 2006 to 20,205 in 2012. The mean age was 56.1 years, 52% were male, and 82% of patients were white. The majority of patients were ASA class 2 or 3, and 98.7% of surgical procedures were elective or semielective. Lumbosacral procedures with a posterior approach and cervical procedures with an anterior approach accounted for the majority of procedures (61% and 25%, respectively), and 43% of the procedures included a spinal fusion. Less than 1% of data was missing from each variable with the exception of race and RVU (both missing 3.8%).
The mean and median frailty scores were 0.07 (standard deviation 0.08) and 0.09 (IQR 0–0.09), respectively (Fig. 1). The maximum mFI was 0.73 (frailty score of 8), which was present in one patient. The prevalence of individual components of frailty is shown in Table 2. Hypertension was the most common individual component of frailty, present in 48% of the population. A total of 2,041 patients (4%) were frail with an mFI of ≥0.21 (frailty score≥3 points), with the prevalence increasing to 8% in patients over 65 years old. The mFI was lower in 2011 and 2012 than in 2007– 2010 (p<.0005). An mFI of ≥0.21 was present in 6.0% of the cohort in 2007–2010 compared with 2.9% of the cohort in 2011–2012. The mFI was associated with several patient- and procedurespecific variables (Table 4). Severity of frailty increased with increasing age, male sex, African American race, higher body mass index, recent weight loss, paraplegia or quadriplegia, ASA score, and preadmission residence in a care facility. The mean age increased with increasing frailty score: from 49 years
Table 3 Characteristics of the study population Variable Patient variables Age, years Male sex Race American Indian or Alaska native Asian Black or African American Native Hawaiian or Pacific Islander White Other or unknown Body mass index, kg/m2 Weight loss Paraplegia Quadriplegia ASA classification 1 2 3 4–5 Preoperative care facility Surgical variables Location Lumbosacral (posterior) Lumbosacral (anterior) Thoracic (posterior) Thoracic (anterior) Cervical (posterior) Cervical (anterior) Multiple level Spinal fusion Emergency Work relative unit (mean, SD)
Value N=52,671 56.1 (14.5) 27,495 (52.3%) 239 (0.5%) 783 (1.5%) 3,859 (7.3%) 122 (0.2%) 43,133 (81.9%) 4,535 (8.6%) 30.1 (6.5) 187 (0.4%) 1,051 (2.0%) 186 (0.4%) 3,146 (6.0%) 28,590 (54.3%) 19,792 (37.6%) 1,084 (2.1%) 183 (0.4%)
30,524 (61.0%) 2,536 (5.1%) 663 (1.3%) 145 (0.3%) 3,942 (7.9%) 12,242 (24.5%) 2,558 (4.9%) 22,726 (43.2%) 657 (1.3%) 17.9 (5.0)
Abbreviations: ASA, American Society of Anesthesiologists; kg, kilograms; m, meters; SD, standard deviation. Values are for number (%) unless otherwise indicated.
ARTICLE IN PRESS A.M. Flexman et al. / The Spine Journal ■■ (2016) ■■–■■
5
or cervical procedures, and was lower in patients who had lumbosacral or cervical procedures with an anterior approach.
Impact of frailty on postoperative outcomes
Fig. 1. Distribution of frailty score and index in patients undergoing spine surgery.
in patients with a frailty index of 0 to 67 years in patients with a frailty index of ≥0.21 (Fig. 2). Compared with lumbosacral procedures with a posterior approach, frailty index was higher in patients who underwent thoracic and posteri-
The median postoperative length of stay was 2 days (IQR 1–3 days). A postoperative complication occurred in 2,881 patients (5.5%) (Table 2). The complication rate increased slightly to 6.5% in patients undergoing more complex procedures (ie top quartile of RVU). The most common major complication was the need to return for a second surgical procedure, which occurred in 1,361 patients (2.6%). Patients with a postoperative complication had an OR of 15.6 (95%CI 14.1– 17.3) for a prolonged hospital stay, with a median stay of 4 days (IQR 2–8 days) in patients with a major complication compared with 2 days (IQR 1–3 days) in patients without a major complication (p<.0005). The discharge destination was reported in 35,037 patients in 2011 and 2012 (years when this variable was available), including 3,455 patients (9.9%) who were admitted from home and discharged to a new care facility. The primary composite end point of major complications was predicted by the mFI (unadjusted OR 1.58 per 0.10 increase in the mFI, 95%CI 1.51–1.64, p<.0005). The mFI remained an independent predictor of major complications
Table 4 Correlates of frailty Frailty index Variable Patient variables Age, y Male sex Black or African American race* BMI, kg/m2 Weight loss Paraplegia Quadriplegia ASA score 1 2 3 4–5 Preoperative care facility Surgical variables Location Lumbosacral (posterior) Lumbosacral (anterior) Thoracic (posterior) Thoracic (anterior) Cervical (posterior) Cervical (anterior) Multiple level Spinal fusion Emergency Work relative unit
0
>0 and <0.21
≥0.21
p value
48.9 (13.3) 50.8% 5.1% 28.8 (6.0) 0.3% 1.6% 0.1%
61.9 (12.5) 53.0% 9.2% 31.2 (6.6) 0.4% 2.1% 0.4%
67.3 (10.3) 60.1% 10.3% 32.1 (7.7) 1.0% 5.2% 2.3%
<.0005 <.0005 <.0005 <.0005 .002 <.0005 <.0005 <.0005
94.8% 59.4% 21.9% 9.2% 0.1%
5.1% 40.0% 70.1% 67.0% 0.4%
0.2% 0.7% 8.0% 23.7% 2.5%
45.2% 53.8% 32.9% 46.2% 40.1% 50.7% 3.8% 43.8% 1.3% 17.8 (5.0)
50.9% 44.2% 58.0% 46.2% 53.2% 46.1% 5.6% 42.5% 1.1% 18.0 (4.9)
3.9% 2.0% 9.1% 7.6% 6.7% 3.2% 7.5% 41.5% 1.9% 18.5 (5.2)
<.0005 <.0005
<.0005 <.0005 .61 <.0005
Abbreviations: BMI, body mass index; ASA, American Society of Anesthesiologists; kg, kilograms; m, meters. Data shown are mean (standard deviation) or n (%) for patients with the corresponding frailty index. The p values are shown for the association of frailty index with each variable. * Race was dichotomized to Black or African American versus other because all other races possessed similar frailty scores.
ARTICLE IN PRESS 6
A.M. Flexman et al. / The Spine Journal ■■ (2016) ■■–■■
Fig. 2. Association of frailty score and index with age.
on multivariate analysis (OR 1.15, 95%CI 1.09–1.21, p<.0005). A serious surgical site infection (deep or organ space infection) occurred in 0.67% of the total population, accounting for 76% of patients returning for a second operation. Frailty was a strong predictor of infection on unadjusted analysis (OR 1.43, 95% CI 1.28–1.60, p<.0005), and remained an independent predictor of infection when adjusting for potential confounders (OR 1.15, 95%CI 1.00–1.33, p=.04). Prolonged length of stay and discharge to a new facility were each predicted by the mFI on both unadjusted and adjusted analyses (p<.0005; Table 5). Several additional patient and procedural variables predicted each complication on adjusted analyses, although discrimination was modest for prediction of major complications (AUROC curve 0.69), and calibration was poor for prediction of prolonged length of stay (goodness of fit test p=.008). Prediction of discharge to a new facility had good discrimination (AUROC curve 0.83) and calibration (goodness of fit test p=.04). Death occurred in 89 patients (0.2% of the total cohort) within 30 days of surgery, with a median time to death of 10 days (IQR 5–17 days). mFI was a significant predictor of death on unadjusted and adjusted analyses (p<.005). The OR for death was 2.05 (95%CI 1.69–2.47) for every 0.10 increase in frailty score on unadjusted analysis and 1.44 (95%CI 1.15– 1.81) on adjusted analysis. Additional predictors of death on adjusted analysis included increasing age, paraplegia or quadriplegia, recent weight loss, and surgery other than a lumbosacral procedure with a posterior approach. These variables predicted 30-day mortality with good discrimination (AUROC curve 0.79) and calibration (goodness of fit test p=.19). Discussion Our findings demonstrate that frailty is present in approximately 4% of the spine surgery population, and is twice as common in those over 65 years of age. Frailty is an important predictor of postoperative outcomes in patients undergoing surgery for degenerative spine disease. We show that frailty
independently predicts major postoperative complications, prolonged length of stay, discharge to a higher level of care, and 30-day mortality. These findings have important implications in risk stratification, perioperative planning, and counseling of patients who present with degenerative spine disease, particularly as these procedures are overwhelmingly elective. Frailty has been measured in several ways, including the mFI used in our study. The original frailty index uses a checklist of comorbidities and functional limitations and has been extensively described and validated in both communitydwelling and hospitalized populations to accurately predict outcomes [16,17]. Other commonly used measures of frailty are the Short Physical Performance Battery [21], the Fried Frailty Phenotype [4], and the Edmonton Frail Scale [22]. Unlike the frailty index, these measures incorporate physical performance, although it remains unclear which definition of frailty is optimal, particularly in patients with spine pathology. It is possible that other components of frailty may be more relevant to this population, such as the presence of osteopenia or low muscle mass (sarcopenia). We chose to use the mFI given its previous validation in a broad surgical population [9] and the ease of application to the NSQIP database. We identified multiple populations with an increased prevalence of frailty, particularly including patients at high risk of functional dependence. For example, patients undergoing cervical and thoracic procedures had a higher prevalence of frailty than patients with lumbosacral disease. This is likely due to the frequent association of cervical and thoracic degenerative spine disease with myelopathy and the consequent functional dependence. Frailty was similarly more common in patients with paraplegia, quadriplegia, and preoperative nursing home residence. We also found that African American race is an independent risk factor for frailty. This is consistent with a previous study [23]; however, the reason for this finding remains unknown [24]. We identified a lower prevalence of frailty in 2011–2012 than in earlier years (6.1% in 2007–2010 and 2.9% in 2011–2012). The reduction in
ARTICLE IN PRESS A.M. Flexman et al. / The Spine Journal ■■ (2016) ■■–■■
7
Table 5 Impact of frailty on outcomes Variable Unadjusted analysis Frailty index* Frailty score Adjusted analysis Frailty index* Patient variables Age, y* Male sex Black or African American race BMI, kg/m2 Weight loss Paraplegia Quadriplegia ASA classification 1 2 3 4–5 Preoperative care facility Surgical variables Emergency
Major complications
Length of stay
Discharge disposition
1.58 (1.51–1.64) p<.0005 1.51 (1.46–1.57) p<.0005
1.89 (1.81–1.99) p<.0005 1.79 (1.71–1.87) p<.0005
2.29 (2.19–2.39) p<.0005 2.12 (2.04–2.21) p<.0005
1.15 (1.09–1.22) p<.0005
1.27 (1.19–1.35) p<.0005
1.32 (1.24–1.40) p<.0005
1.20 (1.16–1.24) p<.0005 0.84 (0.77–0.91) p<.0005 1.45 (1.27–1.65) p<.0005 1.01 (1.01–1.02) p<.0005 2.18 (1.38–3.44) p=.001 1.48 (1.19–1.85) p=.001 4.61 (3.15–6.74) p<.0005
1.28 (1.22–1.33) p<.0005 0.82 (0.74–0.90) p<.0005 1.81 (1.55–2.11) p<.0005
2.01 (1.93–2.09) p<.0005 0.62 (0.57–0.67) p<.0005 1.98 (1.72–2.27) p<.0005 1.02 (1.01–1.02) p<.0005 2.51 (1.48–4.26) p=.001 3.46 (2.40–4.98) p<.0005 6.37 (3.21–12.67) p<.0005
reference 1.20 (0.92–1.55) p=.17 1.81 (1.39–2.36) p<.0005 3.01 (2.18–4.16) p<.0005 1.73 (1.14–2.62) p=.01
1.17 (0.79–1.73) p=.44 2.31 (1.55–3.45) p<.0005 4.59 (2.94–7.17) p<.0005 2.01 (1.30–3.10) p=.002
1.60 (1.01–2.53) p=.04 2.93 (1.85–4.64) p<.0005 5.02 (3.05–8.25) p<.0005 not applicable
1.48 (1.34–1.64) p<.0005
2.92 (2.17–3.93) p<.0005 1.02 (1.01–1.04) p<.0005 1.67 (1.34–2.07) p<.0005 2.00 (1.71–2.34) p<.0005
4.10 (3.00–5.61) p<.0005 1.04 (1.03–1.06) p<.0005 1.45 (1.20–1.75) p<.0005 1.57 (1.36–1.80) p<.0005
reference 1.40 (1.18–1.66) p<.0005 2.09 (1.65–2.64) p<.0005 2.20 (1.36–3.56) p=.001 1.22 (1.07–1.40) p=.004 0.51 (0.44–0.58) p<.0005
1.68 (1.38–2.04) p<.0005 4.04 (3.19–5.13) p<.0005 3.68 (2.29–5.92) p<.0005 1.56 (1.32–1.83) p<.0005 0.54 (0.46–0.64) p<.0005
1.12 (0.93–1.34) p=.24 4.41 (3.49–5.57) p<.0005 1.57 (0.85–2.90) p=.15 1.61 (1.41–1.85) p<.0005 0.34 (0.29–0.39) p<.0005
2.42 (1.86–3.15) p<.0005
Work relative unit Multiple level Spinal fusion Location Lumbosacral (posterior) Lumbosacral (anterior) Thoracic (posterior) Thoracic (anterior) Cervical (posterior) Cervical (anterior)
1.88 (1.07–3.31) p=.03 1.83 (1.44–2.32) p<.0005 4.37 (2.94–6.49) p<.0005
Abbreviations: BMI, body mass index; ASA, American Society of Anesthesiologists; kg, kilograms; m, meters. Data are shown as adjusted OR (95% CI) and p value. Empty cells represent variables that were not independently associated with the stated outcome. * Odds ratios are reported for each decade increase in age and each 0.10 increase in frailty index.
frailty prevalence over time could be due to changes in the population undergoing spine surgery (eg changes in surgical indication that result in operation on milder forms of disease in younger patients), however, this change
is more likely due to the rapid expansion of the NSQIP database to include nonacademic or community hospitals that provide service to less complex patients with fewer comorbidities.
ARTICLE IN PRESS 8
A.M. Flexman et al. / The Spine Journal ■■ (2016) ■■–■■
Although direct comparison with the widely used ASA score was not performed, the results suggest that the mFI may have superior prognostic ability. The ASA score is subjective with at best moderate inter-rater reliability [25]. Although the ASA and mFI both measure a similar construct, the ASA score provides limited discrimination in our population as 92% of our patients were ASA 2 or 3. The mFI, in contrast, may be more exhaustive and allows greater discrimination in populations with frequent and multiple comorbidities. Although the mFI is more challenging to use clinically given the large number of variables, our results suggest that these differences correlate with meaningful differences in outcome, and the mFI may be useful in future investigations of risk prediction. The overall rate of complications observed in our cohort (5.5%) is lower than most previous reports that have suggested an overall complication rate of 5%–19.3% for cervical procedures and 3.7%–12.8% for lumbar surgery [26]. This variation is likely due to the definitions used, the population studied, and the method of data collection. For example, a recent prospective study reported an adverse event rate of 73.5%; however, multiple minor adverse events were considered complications in this study (eg electrolyte abnormalities), and the study population included oncology and trauma patients who are at increased risk of postoperative complications [11]. Conversely, the lower overall surgical complexity in the NSQIP database may underestimate the complication rates seen in some centers. This is likely a minor effect, however, because the complication rate in our study remained low (6.6%) in patients with higher surgical complexity (ie within the top quartile of RVU). Deep incisional or organ space infection occurred in 0.67% of spine surgery patients, compared with previous reports that range from 0.7% to 12.9% [27–29]. Our infection rate was likely lower than previous estimates because we did not include superficial infections. Despite the low overall rate of infection, patients with a surgical site infection frequently required repeat surgery to manage this complication. The need for reoperation because of a surgical site infection was strongly predicted by the presence of frailty, indicating that frail patients likely had weakened reserve that may have predisposed to surgical site infection and to the inability to conservatively manage such infections without surgical intervention. The NSQIP database is a large, prospective, international cohort that has rigorous and validated quality control testing. These features allowed us to analyze rare but clinically relevant outcomes, adjust for multiple potential confounders, perform subgroup sensitivity analyses, and extrapolate these findings to a large population of patients. Our study has some limitations. Hospitals that participate in the NSQIP program may have important differences in case mix, patient volume, and practice style compared with non-NSQIP hospitals, limiting the generalizability of our results [30]. In addition, the NSQIP database included most relevant variables for our analysis; however, some variables that would be relevant to the evaluation of frailty in spine surgery were not available (eg
preoperative cognitive function, dementia), and these potential predictors require further study. Finally, our predictive model should be validated on an independent population given the risk of over-fitting our model to this dataset. Other methods of frailty assessment should be considered in future prospective studies; however, these alternative measurements may be difficult to perform or have less validity in the spine population (eg gait speed, hand grip strength) [4]. Conclusions In summary, we provide novel evidence that frailty is an important predictor of clinically relevant outcomes in patients undergoing surgery for degenerative spine disease. We were able to identify specific populations with an increased prevalence of frailty and show that frailty predicts major complications, prolonged length of stay, discharge to a new facility, and 30-day mortality. The impact of frailty on risk stratification, postoperative outcomes, and resource consumption will become increasingly important as the mean age of patients undergoing spine surgery continues to increase. Additional prospective studies are needed to further validate the role of frailty in this population and to evaluate whether modifying frailty can improve outcomes. Supplementary material Supplementary material related to this article can be found at http://dx.doi.org/10.1016/j.spinee.10.1016/j.spinee.2016 .06.017. References [1] Iqbal J, Denvir M, Gunn J. Frailty assessment in elderly people. Lancet 2013;381:1985–6. [2] Partridge JS, Harari D, Dhesi JK. Frailty in the older surgical patient: a review. Age Ageing 2012;41:142–7. [3] Collard RM, Boter H, Schoevers RA, Oude Voshaar RC. Prevalence of frailty in community-dwelling older persons: a systematic review. J Am Geriatr Soc 2012;60:1487–92. [4] Fried LP, Tangen CM, Walston J, et al. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci 2001;56:M146–56. [5] 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;27:904–8. [6] Kim SW, Han HS, Jung HW, et al. Multidimensional frailty score for the prediction of postoperative mortality risk. JAMA Surg 2014;149:633–40. [7] Makary MA, Segev DL, Pronovost PJ, et al. Frailty as a predictor of surgical outcomes in older patients. J Am Coll Surg 2010;210:901–8. [8] Saxton A, Velanovich V. Preoperative frailty and quality of life as predictors of postoperative complications. Ann Surg 2011;253:1223–9. [9] Velanovich V, Antoine H, Swartz A, Peters D, Rubinfeld I. Accumulating deficits model of frailty and postoperative mortality and morbidity: its application to a national database. J Surg Res 2013;183:104–10. [10] Revenig LM, Canter DJ, Taylor MD, et al. Too frail for surgery? Initial results of a large multidisciplinary prospective study examining preoperative variables predictive of poor surgical outcomes. J Am Coll Surg 2013;217:665–70.e1.
ARTICLE IN PRESS A.M. Flexman et al. / The Spine Journal ■■ (2016) ■■–■■ [11] Street JT, Lenehan BJ, DiPaola CP, et al. Morbidity and mortality of major adult spinal surgery. A prospective cohort analysis of 942 consecutive patients. Spine J 2012;12:22–34. [12] Cowan JA Jr, Dimick JB, Wainess R, Upchurch GR Jr, Chandler WF, La Marca F. Changes in the utilization of spinal fusion in the United States. Neurosurgery 2006;59:15–20 discussion 15–20. [13] Deyo RA, Gray DT, Kreuter W, Mirza S, Martin BI. United States trends in lumbar fusion surgery for degenerative conditions. Spine 2005;30:1441–5 discussion 6–7. [14] Mashour GA, Shanks AM, Kheterpal S. Perioperative stroke and associated mortality after noncardiac, nonneurologic surgery. Anesthesiology 2011;114:1289–96. [15] 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;183:40–6. [16] Mitnitski AB, Graham JE, Mogilner AJ, Rockwood K. Frailty, fitness and late-life mortality in relation to chronological and biological age. BMC Geriatr 2002;2:1. [17] Mitnitski AB, Mogilner AJ, Rockwood K. Accumulation of deficits as a proxy measure of aging. ScientificWorldJournal 2001;1:323– 36. [18] Rockwood K, Song X, Mitnitski A. Changes in relative fitness and frailty across the adult lifespan: evidence from the Canadian National Population Health Survey. CMAJ 2011;183:E487–94. [19] Mabry CD, McCann BC, Harris JA, et al. The use of intraservice work per unit of time (IWPUT) and the building block method (BBM) for the calculation of surgical work. Ann Surg 2005;241:929–38 discussion 38–40. [20] Dindo D, Demartines N, Clavien PA. Classification of surgical complications: a new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann Surg 2004;240:205–13.
9
[21] Guralnik JM, Simonsick EM, Ferrucci L, et al. A short physical performance battery assessing lower extremity function: association with self-reported disability and prediction of mortality and nursing home admission. J Gerontol 1994;49:M85–94. [22] Rolfson DB, Majumdar SR, Tsuyuki RT, Tahir A, Rockwood K. Validity and reliability of the Edmonton Frail Scale. Age Ageing 2006;35:526–9. [23] Hirsch C, Anderson ML, Newman A, et al. The association of race with frailty: the cardiovascular health study. Ann Epidemiol 2006;16:545–53. [24] Kaufman JS, Cooper RS. RE: Hirsch C, Anderson ML, Newman A, Kop W, Jackson S, Gottdiener J, et al, for the Cardiovascular Health Study Research Group. The association of race with frailty: the Cardiovascular Health Study. Ann Epidemiol. 2006;16:545–553. Ann Epidemiol 2007;17:157–8. [25] Sankar A, Johnson SR, Beattie WS, Tait G, Wijeysundera DN. Reliability of the American Society of Anesthesiologists physical status scale in clinical practice. Br J Anaesth 2014;113:424–32. [26] Dekutoski MB, Norvell DC, Dettori JR, Fehlings MG, Chapman JR. Surgeon perceptions and reported complications in spine surgery. Spine 2010;35(9 Suppl.):S9–21. [27] Moucha CS, Clyburn TA, Evans RP, Prokuski L. Modifiable risk factors for surgical site infection. Instr Course Lect 2011;60:557–64. [28] Olsen MA, Mayfield J, Lauryssen C, et al. Risk factors for surgical site infection in spinal surgery. J Neurosurg 2003;98(2 Suppl.):149– 55. [29] Roberts FJ, Walsh A, Wing P, Dvorak M, Schweigel J. The influence of surveillance methods on surgical wound infection rates in a tertiary care spinal surgery service. Spine 1998;23:366–70. [30] Etzioni DA, Wasif N, Dueck AC, et al. Association of hospital participation in a surgical outcomes monitoring program with inpatient complications and mortality. JAMA 2015;313:505–11.
The Spine Journal ■■ (2016) ■■–■■
Clinical Study
Frailty and postoperative outcomes in patients undergoing surgery for degenerative spine disease Alana M. Flexman, MDa,*, Raphaële Charest-Morin, MDb, Liam Stobart, MDa, John Street, MD PhDc, Christopher J. Ryerson, MD MASd a
Department of Anesthesiology, Pharmacology and Therapeutics, Vancouver General Hospital, University of British Columbia, Room 2449 JPP 899 West 12th Avenue, Vancouver, BC, Canada, V5Z 1M9 b Department of Orthopedic Surgery, Centre Hospitalier Universitaire de Québec, 1401 18e rue, Local B-2408, Québec, QC, Canada, G1J 1Z4 c Combined Neurosurgical and Orthopedic Spine Program, Department of Orthopedic Surgery, University of British Columbia, 818 West 10th Avenue, Vancouver, BC, Canada, V5Z 1M9 d Division of Respirology, Department of Medicine, St. Paul’s Hospital, University of British Columbia, Ward 8B, 1081 Burrard Street, Vancouver, BC, Canada, V6Z 1Y6 Received 22 December 2015; revised 31 May 2016; accepted 21 June 2016
Abstract
BACKGROUND CONTEXT: Frailty is defined as a state of decreased reserve and susceptibility to stressors. The relationship between frailty and postoperative outcomes after degenerative spine surgery has not been studied. PURPOSE: This study aimed to (1) determine prevalence of frailty in the degenerative spine population; (2) describe patient characteristics associated with frailty; and (3) determine the association between frailty and postoperative complications, mortality, length of stay, and discharge disposition. STUDY DESIGN: This is a retrospective analysis on a prospectively collected cohort from the National Surgical Quality Improvement Program (NSQIP). PATIENT SAMPLE: A total of 53,080 patients who underwent degenerative spine surgery between 2006 and 2012 were included in the study. OUTCOME MEASURES: A modified frailty index (mFI) with 11 variables derived from the NSQIP dataset was used to determine prevalence of frailty and its correlation with a composite outcome of perioperative complications as well as hospital length of stay, mortality, and discharge disposition. METHODS: After calculating the mFI for each patient, the prevalence and predictors of frailty were determined for our cohort. The association of frailty with postoperative outcomes was determined after adjusting for known and suspected confounders using multivariate logistic regression.
FDA device/drug status: Not applicable. Author disclosures: AMF: Grant: Department Internal Funds–Research Merit Award (C, Paid directly to the author), pertaining to the submitted manuscript; Speaking and/or Teaching Arrangements: Hospira, Inc (A, Paid directly to the author); Research Support (Investigator Salary, Staff/Materials): Department Internal Funds–Research Merit Award (C [Investigator Salary], Paid directly to the author); Grants: Canadian Anesthesiologists’ Society Research Grant (B, Paid directly to the institution/employer). RC-M: Nothing to disclose. LS: Nothing to disclose. JS: Consulting: DePuy Synthes (B, Paid directly to the author), Medtronic (B, Paid directly to the author); Speaking and/or Teaching Arrangements: Medtronic (Paid directly to the author); Trips/Travel: Medtronic (B, Paid directly to the institution/employer); Scientific Advisory Board/Other Office: DePuy Synthes (B, Paid directly to the author); Grants: Medtronic (E, Paid directly to the institution/employer); Fellowship Support: Medtronic (B, Paid directly to the institution/employer), outside the submitted work. CJR: Speaking and/or Teaching Arrangements: Boehringer Ingelheim (B, Paid directly to the author), InterMune/ Hoffmann La Roche (C, Paid directly to the author); Trips/Travel: InterMune (B, Paid directly to the author); Scientific Advisory Board/Other Office: InterMune/Hoffmann La Roche (B, Paid directly to the author), Boehringer Ingelheim (B, Paid directly to the author); Research Support (Investigator http://dx.doi.org/10.1016/j.spinee.2016.06.017 1529-9430/© 2016 Elsevier Inc. All rights reserved.
Salary, Staff/Materials): InterMune/Hoffmann La Roche (E, Paid directly to the institution/employer); Grants: InterMune/Hoffmann La Roche (F, Paid directly to the institution/employer), Boehringer Ingelheim (I, Paid directly to the institution/employer), outside the submitted work. The American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) and the hospitals participating in the ACS NSQIP are the source of the data used herein; they have not verified and are not responsible for the statistical validity of the data analysis or the conclusions derived by the authors. The authors have no conflicts of interest to disclose. This study did not receive funding. Dr. Alana Flexman is supported by a Vancouver Acute Department of Anesthesia Research Merit Award (Internal Funds). Dr. Christopher Ryerson is supported by a Career Investigator Award from the Michael Smith Foundation for Health Research. The disclosure key can be found on the Table of Contents and at www.TheSpineJournalOnline.com. * Corresponding author. Department of Anesthesiology, Pharmacology and Therapeutics, University of British Columbia, Vancouver General Hospital, Room 2449 JPP 899 West 12th Avenue, Vancouver, BC, Canada, V5Z 1M9. Fax +1 604 875 5209. E-mail address: [email protected] (A.M. Flexman).
ARTICLE IN PRESS 2
A.M. Flexman et al. / The Spine Journal ■■ (2016) ■■–■■
RESULTS: Frailty was present in 2,041 patients within the total population (4%) and in 8% of patients older than 65 years. Frailty severity increased with increasing age, male sex, African American race, higher body mass index, recent weight loss, paraplegia or quadriplegia, American Society of Anesthesiologists (ASA) score, and preadmission residence in a care facility. Frailty severity was an independent predictor of major complication (OR 1.15 for every 0.10 increase in mFI, 95%CI 1.09–1.21, p<.0005) and specifically predicted reoperation for postsurgical infection (OR 1.3, 95%CI 1.16–1.46, p<.0005). Prolonged length of stay and discharge to a new facility were also independently predicted by frailty severity (p<.0005). Frailty severity predicted 30-day mortality on unadjusted (OR 2.05, 95%CI 1.70–2.48, p<.0005) and adjusted analyses (OR 1.48, 95%CI 1.18–1.86, p<.0005). CONCLUSIONS: Frailty is an important predictor of postoperative outcomes following degenerative spine surgery. Preoperative recognition of frailty may be useful for perioperative optimization, risk stratification, and patient counseling. © 2016 Elsevier Inc. All rights reserved. Keywords:
Frailty; Degenerative spine disease; Morbidity; Outcomes; Risk stratification; Spine surgery
Introduction
Materials and methods
Frailty is defined as a state of increased vulnerability to poor resolution of homoeostasis after a stressor event, which increases the risk of adverse outcomes [1]. Frailty represents a state of weakened reserve against even minor stressors and may occur independent of and out of proportion to chronological age [1,2]. The prevalence of frailty in the nonsurgical hospital population is approximately 10%, although this varies with the population and method of frailty measurement [3]. The prevalence of frailty in community dwelling individuals increases with age, and even mild frailty is associated with worsening disability, admission to hospital, and death in the community-dwelling elderly population [4]. Frailty is more prevalent in the surgical population (42%–50%) compared with the nonsurgical elderly population (4%–10%), and is likely to be an important predictor of outcome in patients undergoing spine surgery [1]. Preoperative frailty has been independently associated with increased morbidity and mortality in several surgical populations [5–9]. These studies showed that frail surgical patients experienced higher rates of postoperative complications, length of stay, and mortality, and were more likely to be discharged to a care facility rather than home [7,8,10]. As the rate of complications after spine surgery is high [11] and the mean age of patients undergoing spine surgery is also increasing [12,13], frailty is likely to be increasingly common and relevant to this population. Despite these potential implications, little is known about the prevalence of frailty in the spine surgery population or on the impact of frailty on postoperative outcomes in this vulnerable population. Our study objectives were to (1) determine the prevalence of frailty in patients undergoing surgery for degenerative spine disease; (2) describe the patient characteristics associated with frailty; and (3) determine the relationship between frailty and postoperative complications, length of stay, discharge disposition, and mortality in this population.
We performed a retrospective analysis of the prospectively collected American College of Surgeons National Surgical Quality Improvement Program (NSQIP) database. Study population Data were extracted from the American College of Surgeons NSQIP database following approval from our Institutional Clinical Research Ethics Board (H12-03433). The NSQIP database is a prospective multicenter database of adult patients that has been described previously [14]. Highquality data collection and entry are ensured through rigorous training of the clinical reviewer as well as inter-rater reliability audits to ensure accurate and consistent data entry. Cases are sampled using an 8-day cycle to ensure an even case mix, and operating room logs are audited to ensure representative sampling of cases. The most recent year included in this study (2012) includes 543,885 cases from 374 centers. We identified patients undergoing spine surgery for degenerative spine conditions in the NSQIP database from 2006 to 2012. We first identified patients undergoing any type of spine surgery by selecting all common procedural terminology (CPT) codes related to the spine. We used International Classification of Diseases and CPT codes to further restrict our population to spine surgery patients with a primary diagnosis of degenerative disease (Supplementary Appendix, Table S1). We excluded patients who did not receive a general anesthetic. Predictor variables We calculated a modified frailty score and modified frailty index (mFI) for each patient using a previously described method [9,15]. The mFI is a simplified form of the Canadian Study of Health and Aging Frailty Index [16,17]. The Canadian Study of Health and Aging Frailty Index is based on the theory of “accumulating deficits” and strongly correlates with overall mortality in community-dwelling adults
ARTICLE IN PRESS A.M. Flexman et al. / The Spine Journal ■■ (2016) ■■–■■
[16,17]. The mFI incorporates 11 NSQIP variables (deficits) and similarly predicted postoperative morbidity and mortality in a broad surgical population [9]. The modified frailty score was calculated based on the number of deficits present preoperatively, including dependent functional status, diabetes mellitus, lung problems, congestive heart failure, myocardial infarction, cardiac problems, hypertension, impaired sensorium, prior transient ischemic attack, history of stroke, and peripheral vascular disease (Table 1). The mFI was calculated as the modified frailty score (ie the number of deficits present) divided by 11, thus providing an index with a range of 0 to 1. We categorized patients as not frail (mFI=0), pre-frail (mFI>0 and <0.21), and frail (≥0.21) based on previous data defining frailty as an index greater than 0.21 [18]. Additional candidate predictor variables for major complications included patient- and procedure-specific variables. Patient-specific predictor variables included age, sex, race, American Society of Anesthesiologists’ (ASA) classification, body mass index, recent weight loss, paraplegia, quadriplegia, and preoperative residence in a care facility. Patients older than 90 years are listed as 90+ in the NSQIP database to maintain de-identification; these 119 patients (0.2% of the total cohort) were assigned an age of 90 for the purposes of the analysis. Procedure-specific predictor variables included year of operation, emergency surgery, surgical site, and surgical approach. Surgical variables (eg site, level, presence of fusion, or multiple levels) were determined using CPT codes (Supplementary Appendix, Table S2). We estimated sur-
3
gical complexity using the standardized work relative value unit (RVU) as previously described [19], and also considered whether surgery encompassed multiple vertebral levels and whether a spinal fusion was performed. Outcome variables The primary outcome variable was a composite of major complications within 30 days of surgery (Table 2). Complications were considered major if classified as grade 2 or above using the validated Clavien-Dindo classification system of postoperative complications [20]. Patients were assigned either 0 (no complication) or 1 (the presence of any complication). Secondary outcomes included hospital length of stay, discharge to a facility that was not home (available for 2011– 2012 only), and death within 30 days of surgery. Statistical analysis Data are described as mean (standard deviation) or median (interquartile range [IQR]). The prevalence of frailty was determined both in the overall population and in those over 65 years of age. The association of mFI with baseline patientand procedure-specific variables was determined using a Wilcoxon rank sum test or Spearman correlation coefficient for dichotomous and continuous predictor variables, respectively. The association of mFI with postoperative outcomes was determined using logistic regression analyses. The mFI was first used as a single predictor variable in unadjusted
Table 1 Components of the CSHA-FI and NSQIP modified frailty index [9] CSHA-FI variable
NSQIP variable
Changes in everyday activity –Problems getting dressed –Problems with bathing –Problems with personal grooming –Problems cooking –Problems going out alone History of diabetes mellitus
Functional status –Partially dependent OR –Totally dependent
1,783 (3.4)
Diabetes mellitus-non-insulin OR Diabetes mellitus-insulin History of severe COPD Current pneumonia Congestive heart failure History of myocardial infarction in the past 6 months Previous percutaneous coronary intervention Previous cardiac surgery History of angina Hypertension requiring medication Impaired sensorium
7,928 (15.1)
Lung problems –Respiratory problems Congestive heart failure Myocardial infarction Cardiac problems
Arterial hypertension Clouding or delirium –History relevant to cognitive impairment or loss –Family history relevant to cognitive impairment Cerebrovascular problems History of stroke Decreased peripheral pulses
History of transient ischemic attack History of cerebrovascular accident or stroke with neurologic deficit History of revascularization for peripheral vascular disease OR Rest pain or gangrene
Number (%)
1,936 (3.7) 88 (0.2) 35 (0.1) 2,347 (4.5)
25,422 (48.3) 39 (0.1)
603 (1.1) 314 (0.6) 285 (0.5)
Abbreviations: CSHA-FI, Canadian Study of Health and Aging Frailty Index; NSQIP, National Surgical Quality Improvement Program; COPD, chronic obstructive pulmonary disease. Adapted from Ref. [9].
ARTICLE IN PRESS 4
A.M. Flexman et al. / The Spine Journal ■■ (2016) ■■–■■
Table 2 Components of the major complication outcome variable
Characteristics of frailty
Variable description
Incidence (%)*
Return to the operating room within 30 days Unplanned reintubation Failure to wean from a ventilator >48 hours Cardiac arrest Myocardial infarction Sepsis or septic shock Stroke Pulmonary embolism Deep vein thrombosis requiring treatment Acute renal failure requiring dialysis Urinary tract infection Pneumonia Wound dehiscence Deep incisional surgical site infection Organ-space surgical site infection Total
1,361 (2.6) 211 (0.4) 171 (0.3) 72 (0.1) 104 (0.2) 344 (0.7) 53 (0.1) 182 (0.4) 271 (0.5) 65 (0.1) 661 (1.3) 272 (0.5) 97 (0.2) 249 (0.5) 106 (0.2) 2,881 (5.5)
Abbreviations: NSQIP, National Surgical Quality Improvement Program. * Complications are not mutually exclusive (eg patients with reoperation because of a surgical site infection would be included in both categories). The “Total” represents the number of patients with at least one complication.
analysis for each outcome. A stepwise multivariate analysis with backward elimination was then used to determine the independent association of frailty index with each outcome, adjusting for prespecified patient- and procedure-specific variables, as listed above. Variables that were independently associated with the outcome were retained in the final multivariate model. Patients with missing variables were not included in the final model. Model performance was described using the area under the receiver operating characteristic (AUROC) curve and the goodness of fit test, where poor fit is indicated by p<.05. A two-sided p value <.05 was considered significant for all analyses. All data analysis was performed using STATA 11.0 (StataCorp, College Station, TX, USA). Results Study population The study population included a total of 53,145 patients who underwent spine surgery between 2006 and 2012, inclusive (Table 3). A total of 474 patients were excluded, as they did not receive a general anesthetic, leaving 52,671 patients for the analysis. The number of spine surgery procedures per year increased from 414 in 2006 to 20,205 in 2012. The mean age was 56.1 years, 52% were male, and 82% of patients were white. The majority of patients were ASA class 2 or 3, and 98.7% of surgical procedures were elective or semielective. Lumbosacral procedures with a posterior approach and cervical procedures with an anterior approach accounted for the majority of procedures (61% and 25%, respectively), and 43% of the procedures included a spinal fusion. Less than 1% of data was missing from each variable with the exception of race and RVU (both missing 3.8%).
The mean and median frailty scores were 0.07 (standard deviation 0.08) and 0.09 (IQR 0–0.09), respectively (Fig. 1). The maximum mFI was 0.73 (frailty score of 8), which was present in one patient. The prevalence of individual components of frailty is shown in Table 2. Hypertension was the most common individual component of frailty, present in 48% of the population. A total of 2,041 patients (4%) were frail with an mFI of ≥0.21 (frailty score≥3 points), with the prevalence increasing to 8% in patients over 65 years old. The mFI was lower in 2011 and 2012 than in 2007– 2010 (p<.0005). An mFI of ≥0.21 was present in 6.0% of the cohort in 2007–2010 compared with 2.9% of the cohort in 2011–2012. The mFI was associated with several patient- and procedurespecific variables (Table 4). Severity of frailty increased with increasing age, male sex, African American race, higher body mass index, recent weight loss, paraplegia or quadriplegia, ASA score, and preadmission residence in a care facility. The mean age increased with increasing frailty score: from 49 years
Table 3 Characteristics of the study population Variable Patient variables Age, years Male sex Race American Indian or Alaska native Asian Black or African American Native Hawaiian or Pacific Islander White Other or unknown Body mass index, kg/m2 Weight loss Paraplegia Quadriplegia ASA classification 1 2 3 4–5 Preoperative care facility Surgical variables Location Lumbosacral (posterior) Lumbosacral (anterior) Thoracic (posterior) Thoracic (anterior) Cervical (posterior) Cervical (anterior) Multiple level Spinal fusion Emergency Work relative unit (mean, SD)
Value N=52,671 56.1 (14.5) 27,495 (52.3%) 239 (0.5%) 783 (1.5%) 3,859 (7.3%) 122 (0.2%) 43,133 (81.9%) 4,535 (8.6%) 30.1 (6.5) 187 (0.4%) 1,051 (2.0%) 186 (0.4%) 3,146 (6.0%) 28,590 (54.3%) 19,792 (37.6%) 1,084 (2.1%) 183 (0.4%)
30,524 (61.0%) 2,536 (5.1%) 663 (1.3%) 145 (0.3%) 3,942 (7.9%) 12,242 (24.5%) 2,558 (4.9%) 22,726 (43.2%) 657 (1.3%) 17.9 (5.0)
Abbreviations: ASA, American Society of Anesthesiologists; kg, kilograms; m, meters; SD, standard deviation. Values are for number (%) unless otherwise indicated.
ARTICLE IN PRESS A.M. Flexman et al. / The Spine Journal ■■ (2016) ■■–■■
5
or cervical procedures, and was lower in patients who had lumbosacral or cervical procedures with an anterior approach.
Impact of frailty on postoperative outcomes
Fig. 1. Distribution of frailty score and index in patients undergoing spine surgery.
in patients with a frailty index of 0 to 67 years in patients with a frailty index of ≥0.21 (Fig. 2). Compared with lumbosacral procedures with a posterior approach, frailty index was higher in patients who underwent thoracic and posteri-
The median postoperative length of stay was 2 days (IQR 1–3 days). A postoperative complication occurred in 2,881 patients (5.5%) (Table 2). The complication rate increased slightly to 6.5% in patients undergoing more complex procedures (ie top quartile of RVU). The most common major complication was the need to return for a second surgical procedure, which occurred in 1,361 patients (2.6%). Patients with a postoperative complication had an OR of 15.6 (95%CI 14.1– 17.3) for a prolonged hospital stay, with a median stay of 4 days (IQR 2–8 days) in patients with a major complication compared with 2 days (IQR 1–3 days) in patients without a major complication (p<.0005). The discharge destination was reported in 35,037 patients in 2011 and 2012 (years when this variable was available), including 3,455 patients (9.9%) who were admitted from home and discharged to a new care facility. The primary composite end point of major complications was predicted by the mFI (unadjusted OR 1.58 per 0.10 increase in the mFI, 95%CI 1.51–1.64, p<.0005). The mFI remained an independent predictor of major complications
Table 4 Correlates of frailty Frailty index Variable Patient variables Age, y Male sex Black or African American race* BMI, kg/m2 Weight loss Paraplegia Quadriplegia ASA score 1 2 3 4–5 Preoperative care facility Surgical variables Location Lumbosacral (posterior) Lumbosacral (anterior) Thoracic (posterior) Thoracic (anterior) Cervical (posterior) Cervical (anterior) Multiple level Spinal fusion Emergency Work relative unit
0
>0 and <0.21
≥0.21
p value
48.9 (13.3) 50.8% 5.1% 28.8 (6.0) 0.3% 1.6% 0.1%
61.9 (12.5) 53.0% 9.2% 31.2 (6.6) 0.4% 2.1% 0.4%
67.3 (10.3) 60.1% 10.3% 32.1 (7.7) 1.0% 5.2% 2.3%
<.0005 <.0005 <.0005 <.0005 .002 <.0005 <.0005 <.0005
94.8% 59.4% 21.9% 9.2% 0.1%
5.1% 40.0% 70.1% 67.0% 0.4%
0.2% 0.7% 8.0% 23.7% 2.5%
45.2% 53.8% 32.9% 46.2% 40.1% 50.7% 3.8% 43.8% 1.3% 17.8 (5.0)
50.9% 44.2% 58.0% 46.2% 53.2% 46.1% 5.6% 42.5% 1.1% 18.0 (4.9)
3.9% 2.0% 9.1% 7.6% 6.7% 3.2% 7.5% 41.5% 1.9% 18.5 (5.2)
<.0005 <.0005
<.0005 <.0005 .61 <.0005
Abbreviations: BMI, body mass index; ASA, American Society of Anesthesiologists; kg, kilograms; m, meters. Data shown are mean (standard deviation) or n (%) for patients with the corresponding frailty index. The p values are shown for the association of frailty index with each variable. * Race was dichotomized to Black or African American versus other because all other races possessed similar frailty scores.
ARTICLE IN PRESS 6
A.M. Flexman et al. / The Spine Journal ■■ (2016) ■■–■■
Fig. 2. Association of frailty score and index with age.
on multivariate analysis (OR 1.15, 95%CI 1.09–1.21, p<.0005). A serious surgical site infection (deep or organ space infection) occurred in 0.67% of the total population, accounting for 76% of patients returning for a second operation. Frailty was a strong predictor of infection on unadjusted analysis (OR 1.43, 95% CI 1.28–1.60, p<.0005), and remained an independent predictor of infection when adjusting for potential confounders (OR 1.15, 95%CI 1.00–1.33, p=.04). Prolonged length of stay and discharge to a new facility were each predicted by the mFI on both unadjusted and adjusted analyses (p<.0005; Table 5). Several additional patient and procedural variables predicted each complication on adjusted analyses, although discrimination was modest for prediction of major complications (AUROC curve 0.69), and calibration was poor for prediction of prolonged length of stay (goodness of fit test p=.008). Prediction of discharge to a new facility had good discrimination (AUROC curve 0.83) and calibration (goodness of fit test p=.04). Death occurred in 89 patients (0.2% of the total cohort) within 30 days of surgery, with a median time to death of 10 days (IQR 5–17 days). mFI was a significant predictor of death on unadjusted and adjusted analyses (p<.005). The OR for death was 2.05 (95%CI 1.69–2.47) for every 0.10 increase in frailty score on unadjusted analysis and 1.44 (95%CI 1.15– 1.81) on adjusted analysis. Additional predictors of death on adjusted analysis included increasing age, paraplegia or quadriplegia, recent weight loss, and surgery other than a lumbosacral procedure with a posterior approach. These variables predicted 30-day mortality with good discrimination (AUROC curve 0.79) and calibration (goodness of fit test p=.19). Discussion Our findings demonstrate that frailty is present in approximately 4% of the spine surgery population, and is twice as common in those over 65 years of age. Frailty is an important predictor of postoperative outcomes in patients undergoing surgery for degenerative spine disease. We show that frailty
independently predicts major postoperative complications, prolonged length of stay, discharge to a higher level of care, and 30-day mortality. These findings have important implications in risk stratification, perioperative planning, and counseling of patients who present with degenerative spine disease, particularly as these procedures are overwhelmingly elective. Frailty has been measured in several ways, including the mFI used in our study. The original frailty index uses a checklist of comorbidities and functional limitations and has been extensively described and validated in both communitydwelling and hospitalized populations to accurately predict outcomes [16,17]. Other commonly used measures of frailty are the Short Physical Performance Battery [21], the Fried Frailty Phenotype [4], and the Edmonton Frail Scale [22]. Unlike the frailty index, these measures incorporate physical performance, although it remains unclear which definition of frailty is optimal, particularly in patients with spine pathology. It is possible that other components of frailty may be more relevant to this population, such as the presence of osteopenia or low muscle mass (sarcopenia). We chose to use the mFI given its previous validation in a broad surgical population [9] and the ease of application to the NSQIP database. We identified multiple populations with an increased prevalence of frailty, particularly including patients at high risk of functional dependence. For example, patients undergoing cervical and thoracic procedures had a higher prevalence of frailty than patients with lumbosacral disease. This is likely due to the frequent association of cervical and thoracic degenerative spine disease with myelopathy and the consequent functional dependence. Frailty was similarly more common in patients with paraplegia, quadriplegia, and preoperative nursing home residence. We also found that African American race is an independent risk factor for frailty. This is consistent with a previous study [23]; however, the reason for this finding remains unknown [24]. We identified a lower prevalence of frailty in 2011–2012 than in earlier years (6.1% in 2007–2010 and 2.9% in 2011–2012). The reduction in
ARTICLE IN PRESS A.M. Flexman et al. / The Spine Journal ■■ (2016) ■■–■■
7
Table 5 Impact of frailty on outcomes Variable Unadjusted analysis Frailty index* Frailty score Adjusted analysis Frailty index* Patient variables Age, y* Male sex Black or African American race BMI, kg/m2 Weight loss Paraplegia Quadriplegia ASA classification 1 2 3 4–5 Preoperative care facility Surgical variables Emergency
Major complications
Length of stay
Discharge disposition
1.58 (1.51–1.64) p<.0005 1.51 (1.46–1.57) p<.0005
1.89 (1.81–1.99) p<.0005 1.79 (1.71–1.87) p<.0005
2.29 (2.19–2.39) p<.0005 2.12 (2.04–2.21) p<.0005
1.15 (1.09–1.22) p<.0005
1.27 (1.19–1.35) p<.0005
1.32 (1.24–1.40) p<.0005
1.20 (1.16–1.24) p<.0005 0.84 (0.77–0.91) p<.0005 1.45 (1.27–1.65) p<.0005 1.01 (1.01–1.02) p<.0005 2.18 (1.38–3.44) p=.001 1.48 (1.19–1.85) p=.001 4.61 (3.15–6.74) p<.0005
1.28 (1.22–1.33) p<.0005 0.82 (0.74–0.90) p<.0005 1.81 (1.55–2.11) p<.0005
2.01 (1.93–2.09) p<.0005 0.62 (0.57–0.67) p<.0005 1.98 (1.72–2.27) p<.0005 1.02 (1.01–1.02) p<.0005 2.51 (1.48–4.26) p=.001 3.46 (2.40–4.98) p<.0005 6.37 (3.21–12.67) p<.0005
reference 1.20 (0.92–1.55) p=.17 1.81 (1.39–2.36) p<.0005 3.01 (2.18–4.16) p<.0005 1.73 (1.14–2.62) p=.01
1.17 (0.79–1.73) p=.44 2.31 (1.55–3.45) p<.0005 4.59 (2.94–7.17) p<.0005 2.01 (1.30–3.10) p=.002
1.60 (1.01–2.53) p=.04 2.93 (1.85–4.64) p<.0005 5.02 (3.05–8.25) p<.0005 not applicable
1.48 (1.34–1.64) p<.0005
2.92 (2.17–3.93) p<.0005 1.02 (1.01–1.04) p<.0005 1.67 (1.34–2.07) p<.0005 2.00 (1.71–2.34) p<.0005
4.10 (3.00–5.61) p<.0005 1.04 (1.03–1.06) p<.0005 1.45 (1.20–1.75) p<.0005 1.57 (1.36–1.80) p<.0005
reference 1.40 (1.18–1.66) p<.0005 2.09 (1.65–2.64) p<.0005 2.20 (1.36–3.56) p=.001 1.22 (1.07–1.40) p=.004 0.51 (0.44–0.58) p<.0005
1.68 (1.38–2.04) p<.0005 4.04 (3.19–5.13) p<.0005 3.68 (2.29–5.92) p<.0005 1.56 (1.32–1.83) p<.0005 0.54 (0.46–0.64) p<.0005
1.12 (0.93–1.34) p=.24 4.41 (3.49–5.57) p<.0005 1.57 (0.85–2.90) p=.15 1.61 (1.41–1.85) p<.0005 0.34 (0.29–0.39) p<.0005
2.42 (1.86–3.15) p<.0005
Work relative unit Multiple level Spinal fusion Location Lumbosacral (posterior) Lumbosacral (anterior) Thoracic (posterior) Thoracic (anterior) Cervical (posterior) Cervical (anterior)
1.88 (1.07–3.31) p=.03 1.83 (1.44–2.32) p<.0005 4.37 (2.94–6.49) p<.0005
Abbreviations: BMI, body mass index; ASA, American Society of Anesthesiologists; kg, kilograms; m, meters. Data are shown as adjusted OR (95% CI) and p value. Empty cells represent variables that were not independently associated with the stated outcome. * Odds ratios are reported for each decade increase in age and each 0.10 increase in frailty index.
frailty prevalence over time could be due to changes in the population undergoing spine surgery (eg changes in surgical indication that result in operation on milder forms of disease in younger patients), however, this change
is more likely due to the rapid expansion of the NSQIP database to include nonacademic or community hospitals that provide service to less complex patients with fewer comorbidities.
ARTICLE IN PRESS 8
A.M. Flexman et al. / The Spine Journal ■■ (2016) ■■–■■
Although direct comparison with the widely used ASA score was not performed, the results suggest that the mFI may have superior prognostic ability. The ASA score is subjective with at best moderate inter-rater reliability [25]. Although the ASA and mFI both measure a similar construct, the ASA score provides limited discrimination in our population as 92% of our patients were ASA 2 or 3. The mFI, in contrast, may be more exhaustive and allows greater discrimination in populations with frequent and multiple comorbidities. Although the mFI is more challenging to use clinically given the large number of variables, our results suggest that these differences correlate with meaningful differences in outcome, and the mFI may be useful in future investigations of risk prediction. The overall rate of complications observed in our cohort (5.5%) is lower than most previous reports that have suggested an overall complication rate of 5%–19.3% for cervical procedures and 3.7%–12.8% for lumbar surgery [26]. This variation is likely due to the definitions used, the population studied, and the method of data collection. For example, a recent prospective study reported an adverse event rate of 73.5%; however, multiple minor adverse events were considered complications in this study (eg electrolyte abnormalities), and the study population included oncology and trauma patients who are at increased risk of postoperative complications [11]. Conversely, the lower overall surgical complexity in the NSQIP database may underestimate the complication rates seen in some centers. This is likely a minor effect, however, because the complication rate in our study remained low (6.6%) in patients with higher surgical complexity (ie within the top quartile of RVU). Deep incisional or organ space infection occurred in 0.67% of spine surgery patients, compared with previous reports that range from 0.7% to 12.9% [27–29]. Our infection rate was likely lower than previous estimates because we did not include superficial infections. Despite the low overall rate of infection, patients with a surgical site infection frequently required repeat surgery to manage this complication. The need for reoperation because of a surgical site infection was strongly predicted by the presence of frailty, indicating that frail patients likely had weakened reserve that may have predisposed to surgical site infection and to the inability to conservatively manage such infections without surgical intervention. The NSQIP database is a large, prospective, international cohort that has rigorous and validated quality control testing. These features allowed us to analyze rare but clinically relevant outcomes, adjust for multiple potential confounders, perform subgroup sensitivity analyses, and extrapolate these findings to a large population of patients. Our study has some limitations. Hospitals that participate in the NSQIP program may have important differences in case mix, patient volume, and practice style compared with non-NSQIP hospitals, limiting the generalizability of our results [30]. In addition, the NSQIP database included most relevant variables for our analysis; however, some variables that would be relevant to the evaluation of frailty in spine surgery were not available (eg
preoperative cognitive function, dementia), and these potential predictors require further study. Finally, our predictive model should be validated on an independent population given the risk of over-fitting our model to this dataset. Other methods of frailty assessment should be considered in future prospective studies; however, these alternative measurements may be difficult to perform or have less validity in the spine population (eg gait speed, hand grip strength) [4]. Conclusions In summary, we provide novel evidence that frailty is an important predictor of clinically relevant outcomes in patients undergoing surgery for degenerative spine disease. We were able to identify specific populations with an increased prevalence of frailty and show that frailty predicts major complications, prolonged length of stay, discharge to a new facility, and 30-day mortality. The impact of frailty on risk stratification, postoperative outcomes, and resource consumption will become increasingly important as the mean age of patients undergoing spine surgery continues to increase. Additional prospective studies are needed to further validate the role of frailty in this population and to evaluate whether modifying frailty can improve outcomes. Supplementary material Supplementary material related to this article can be found at http://dx.doi.org/10.1016/j.spinee.10.1016/j.spinee.2016 .06.017. References [1] Iqbal J, Denvir M, Gunn J. Frailty assessment in elderly people. Lancet 2013;381:1985–6. [2] Partridge JS, Harari D, Dhesi JK. Frailty in the older surgical patient: a review. Age Ageing 2012;41:142–7. [3] Collard RM, Boter H, Schoevers RA, Oude Voshaar RC. Prevalence of frailty in community-dwelling older persons: a systematic review. J Am Geriatr Soc 2012;60:1487–92. [4] Fried LP, Tangen CM, Walston J, et al. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci 2001;56:M146–56. [5] 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;27:904–8. [6] Kim SW, Han HS, Jung HW, et al. Multidimensional frailty score for the prediction of postoperative mortality risk. JAMA Surg 2014;149:633–40. [7] Makary MA, Segev DL, Pronovost PJ, et al. Frailty as a predictor of surgical outcomes in older patients. J Am Coll Surg 2010;210:901–8. [8] Saxton A, Velanovich V. Preoperative frailty and quality of life as predictors of postoperative complications. Ann Surg 2011;253:1223–9. [9] Velanovich V, Antoine H, Swartz A, Peters D, Rubinfeld I. Accumulating deficits model of frailty and postoperative mortality and morbidity: its application to a national database. J Surg Res 2013;183:104–10. [10] Revenig LM, Canter DJ, Taylor MD, et al. Too frail for surgery? Initial results of a large multidisciplinary prospective study examining preoperative variables predictive of poor surgical outcomes. J Am Coll Surg 2013;217:665–70.e1.
ARTICLE IN PRESS A.M. Flexman et al. / The Spine Journal ■■ (2016) ■■–■■ [11] Street JT, Lenehan BJ, DiPaola CP, et al. Morbidity and mortality of major adult spinal surgery. A prospective cohort analysis of 942 consecutive patients. Spine J 2012;12:22–34. [12] Cowan JA Jr, Dimick JB, Wainess R, Upchurch GR Jr, Chandler WF, La Marca F. Changes in the utilization of spinal fusion in the United States. Neurosurgery 2006;59:15–20 discussion 15–20. [13] Deyo RA, Gray DT, Kreuter W, Mirza S, Martin BI. United States trends in lumbar fusion surgery for degenerative conditions. Spine 2005;30:1441–5 discussion 6–7. [14] Mashour GA, Shanks AM, Kheterpal S. Perioperative stroke and associated mortality after noncardiac, nonneurologic surgery. Anesthesiology 2011;114:1289–96. [15] 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;183:40–6. [16] Mitnitski AB, Graham JE, Mogilner AJ, Rockwood K. Frailty, fitness and late-life mortality in relation to chronological and biological age. BMC Geriatr 2002;2:1. [17] Mitnitski AB, Mogilner AJ, Rockwood K. Accumulation of deficits as a proxy measure of aging. ScientificWorldJournal 2001;1:323– 36. [18] Rockwood K, Song X, Mitnitski A. Changes in relative fitness and frailty across the adult lifespan: evidence from the Canadian National Population Health Survey. CMAJ 2011;183:E487–94. [19] Mabry CD, McCann BC, Harris JA, et al. The use of intraservice work per unit of time (IWPUT) and the building block method (BBM) for the calculation of surgical work. Ann Surg 2005;241:929–38 discussion 38–40. [20] Dindo D, Demartines N, Clavien PA. Classification of surgical complications: a new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann Surg 2004;240:205–13.
9
[21] Guralnik JM, Simonsick EM, Ferrucci L, et al. A short physical performance battery assessing lower extremity function: association with self-reported disability and prediction of mortality and nursing home admission. J Gerontol 1994;49:M85–94. [22] Rolfson DB, Majumdar SR, Tsuyuki RT, Tahir A, Rockwood K. Validity and reliability of the Edmonton Frail Scale. Age Ageing 2006;35:526–9. [23] Hirsch C, Anderson ML, Newman A, et al. The association of race with frailty: the cardiovascular health study. Ann Epidemiol 2006;16:545–53. [24] Kaufman JS, Cooper RS. RE: Hirsch C, Anderson ML, Newman A, Kop W, Jackson S, Gottdiener J, et al, for the Cardiovascular Health Study Research Group. The association of race with frailty: the Cardiovascular Health Study. Ann Epidemiol. 2006;16:545–553. Ann Epidemiol 2007;17:157–8. [25] Sankar A, Johnson SR, Beattie WS, Tait G, Wijeysundera DN. Reliability of the American Society of Anesthesiologists physical status scale in clinical practice. Br J Anaesth 2014;113:424–32. [26] Dekutoski MB, Norvell DC, Dettori JR, Fehlings MG, Chapman JR. Surgeon perceptions and reported complications in spine surgery. Spine 2010;35(9 Suppl.):S9–21. [27] Moucha CS, Clyburn TA, Evans RP, Prokuski L. Modifiable risk factors for surgical site infection. Instr Course Lect 2011;60:557–64. [28] Olsen MA, Mayfield J, Lauryssen C, et al. Risk factors for surgical site infection in spinal surgery. J Neurosurg 2003;98(2 Suppl.):149– 55. [29] Roberts FJ, Walsh A, Wing P, Dvorak M, Schweigel J. The influence of surveillance methods on surgical wound infection rates in a tertiary care spinal surgery service. Spine 1998;23:366–70. [30] Etzioni DA, Wasif N, Dueck AC, et al. Association of hospital participation in a surgical outcomes monitoring program with inpatient complications and mortality. JAMA 2015;313:505–11.