Body mass index predicts risk of complications in lumbar spine surgery based on surgical invasiveness

ARTICLE IN PRESS

The Spine Journal ■■ (2017) ■■–■■

Clinical Study

Body mass index predicts risk of complications in lumbar spine surgery based on surgical invasiveness Olivia J. Bono, BAa, Gregory W. Poorman, BAa, Norah Foster, MDa, Cyrus M. Jalai, BAa, Samantha R. Horn, BAa, Jonathan Oren, MDa, Alexandra Soroceanu, MDb, Subaraman Ramachandran, MDa, Taylor E. Purvis, BAc, Deeptee Jain, MDd, Shaleen Vira, MDa, Bassel G. Diebo, MDe, Breton Line, BSMEa, Daniel M. Sciubba, MDc, Themistocles S. Protopsaltis, MDa, Aaron J. Buckland, MBBS FRACSa, Thomas J. Errico, MDa, Virginie Lafage, PhDf, Shay Bess, MDa, Peter G. Passias, MDa,* a

Department of Orthopaedic Surgery, NYU Langone Orthopaedic Hospital, 301 E. 17th St, New York, NY 10003, USA b Department of Orthopaedic Surgery, University of Calgary, 3330 Hospital Dr NW, Calgary, AB T2N 4N1, Canada c Department of Neurosurgery, The Johns Hopkins University School of Medicine, 733 N. Broadway, Baltimore, MD 21205, USA d Depatment of Orthopaedic Surgery, University of California, 3333 California St, San Francisco, CA, 94118, USA e Department of Orthopaedic Surgery, SUNY Downstate Medical Center, 450 Clarkson Ave, Brooklyn, NY 11203, USA f Spine Service, Hospital for Special Surgery, 535 E. 70th St, New York, NY 10021, USA Received 8 November 2016; revised 19 October 2017; accepted 9 November 2017

Abstract

BACKGROUND CONTEXT: Obesity as a comorbidity in spine pathology may increase the risk of complications following surgical treatment. The body mass index (BMI) threshold at which obesity becomes clinically relevant, and the exact nature of that effect, remains poorly understood. PURPOSE: Identify the BMI that independently predicts risk of postoperative complications following lumbar spine surgery. STUDY DESIGN/SETTING: Retrospective review of the National Surgery Quality Improvement Program (NSQIP) years 2011–2013. PATIENT SAMPLE: A total of 31,763 patients were undergoing arthrodesis, discectomy, laminectomy, laminoplasty, corpectomy, or osteotomy of the lumbar spine. OUTCOME MEASURES: Complication rates. METHODS: The patient sample was categorized preoperatively by BMI according to the World Health Organization stratification: underweight (BMI <18.5), normal overweight (BMI 20.0–29.9), obesity class 1 (BMI 30.0–34.9), 2 (BMI 35.0–39.9), and 3 (BMI≥40). Patients were dichotomized based on their position above or below the 75th surgical invasiveness index (SII) percentile cutoff into low-SII and high-SII. Differences in complication rates in BMI groups were analyzed by Bonferroni analysis of variance (ANOVA) method. Multivariate binary logistic regression evaluated relationship between BMI and complication categories in all patients and in high-SII and low-SII surgeries.

FDA device/drug status: Not applicable. Author disclosures: OJB: Nothing to disclose. GWP: Nothing to disclose. NF: Nothing to disclose. CMJ: Nothing to disclose. SRH: Nothing to disclose. JO: Nothing to disclose. AS: Grants: DePuy Synthes paid to International Spine Study Group Foundation (I), outside the submitted work. SR: Nothing to disclose. TEP: Nothing to disclose. DJ: Nothing to disclose. SV: Nothing to disclose. BGD: Nothing to disclose. BL: Nothing to disclose. DMS: Consulting: Medtronic (D), outside the submitted work. TSP: Consulting: Medicrea International (B); Grants: Zimmer Spine (A, Paid directly to institution/employer), outside the submitted work. AJB: Nothing to disclose. TJE: Consulting: K2M (C); Speaking and/or Teaching Arrangements: K2M (C); Grants: Medtronic (B), Pfizer (E), Paradigm Spine (B), Fastenetix (F), outside the submitted work. VL: Stock Ownership: Nemaris Inc; Speaking and/or Teaching Arrangements: Nemaris Inc (B), DePuy Synthes (B), Medicrea (B), NuVasive (B); Board of Directors: Nemaris Inc; Grants: https://doi.org/10.1016/j.spinee.2017.11.015 1529-9430/© 2017 Elsevier Inc. All rights reserved.

SRS (D, Paid directly to institution/employer), NIH (D, Paid directly to institution/employer), outside the submitted work. SB: Royalties: Pioneer (B); Consulting: K2M (B), NuVasive (none), Inovasis (none), Allosource (B); Grants: K2M (D, Paid directly to institution/employer), NuVasive (D, Paid directly to institution/employer), Inovasis (F, Paid directly to institution/ employer), DePuy Synthes (F, Paid directly to institution/employer), outside the submitted work. PGP: Consulting: Medicrea (B); Speaking and/or Teaching Arrangements: Zimmer Biomet (A), outside the submitted work. The disclosure key can be found on the Table of Contents and at www.TheSpineJournalOnline.com. * Corresponding author. Division of Spinal Surgery, Departments of Orthopaedic and Neurological Surgery, NYU Langone Medical Center – Orthopaedic Hospital, 301 East 17th St, New York, NY 10003, USA. Tel.: (516) 357 8777; fax: (516) 357 0087. E-mail address: [email protected] (P.G. Passias)

ARTICLE IN PRESS 2

O.J. Bono et al. / The Spine Journal ■■ (2017) ■■–■■

RESULTS: Controlling for baseline difference in SII, Charlson Comorbidity Index (CCI) score, diabetes, hypertension, and smoking, complications significantly increased at a BMI of 35 kg/m2. The odds ratios for any complication (odds ratio [OR] [95% confidence interval {CI}]; obesity 2: 1.218 [1.020–1.455]; obesity 3: 1.742 [1.439–2.110]), infection (obesity 2: 1.335 [1.110–1.605]; obesity 3: 1.685 [1.372–2.069]), and surgical complication (obesity 2: 1.622 [1.250–2.104]; obesity 3: 2.798 [2.154–3.634]) were significantly higher in obesity classes 2 and 3 relative to the normaloverweight cohort (all p<.05). CONCLUSION: There is a significant increase in complications, specifically infection and surgical complications, in patients with BMI≥35 following lumbar spine surgery, with that rate further increasing with BMI≥40. © 2017 Elsevier Inc. All rights reserved. Keywords:

Complications; Lumbar; Obesity; Predictor; Risk; Spine

Introduction Obesity has been found to be an independent risk factor for comorbidities such as hypertension, diabetes, and cardiovascular disease [1,2]. Current literature suggests that obese patients, representing nearly 35% of the United States population, need an altered surgical treatment protocol for spine pathologies [3–5]. Additionally, recent literature suggests that obesity contributes to low back pain and greater rates of degenerative spine pathology, which as a result, contributes to a relatively high prevalence of obesity in patients undergoing elective lumbar surgery [5]. Therefore, given the effect of excessive weight on the development of lumbar pathology, a high proportion of patients undergoing elective lumbar spine surgery are obese [6]. In fact, according to the National Surgery Quality Improvement Program (NSQIP) data, almost 80% of patients undergoing spine surgery are overweight or obese [7]. The role of obesity on complications following spinal procedures has been thoroughly investigated [8–10]. Factors associated with higher complications following lumbar spine procedures in obese patients include the associated medical comorbidities, longer surgical duration, and increased blood loss. In a recent study, Seicean et al. reported increased perioperative complications in 49,314 patients undergoing elective lumbar procedures with body mass index (BMI)≥30 kg/m2 using the NSQIP database. However, after controlling for medical comorbidities, the authors found that the BMI threshold increased to >40 kg/m2 for higher complications, readmissions, and non-routine discharges [11]. In addition to the medical comorbidities, another important factor that may influence complication rates in obese patients undergoing spine surgeries is the invasiveness of the surgical procedure. However, there is no literature on the effect of surgical invasiveness in obese patients. Given previous findings that proved surgical invasiveness is associated with surgical site infection and increased operation time, it would be beneficial to dichotomize patients based on high or low surgical invasiveness given that invasiveness has been proven to significantly affect outcome [12,13]. Although it is widely regarded that obesity is correlated with complications, it is not entirely clear which complications predominate. This may be related to the high prevalence

of “obese” patients as defined by the World Health Organization (BMI greater than 30 kg/m2), which confounds the results of many series. Marquez-Lara et al. found that in 24,196 lumbar spine procedures, patients with BMI≥25 demonstrated higher risk of deep vein thrombosis, pulmonary embolism, and superficial wound infections compared with normal-weight patients, whereas morbidly obese patients (BMI≥40) had additional risk of postoperative renal failure, sepsis, and urinary tract infection [14]. Because of lack of consistent findings, there is a need for a thorough examination of obesity’s effect on complications in lumbar spine surgeries, and to define a threshold at which body mass index becomes a comorbidity. The primary objective of this study was to report the effects of obesity on specific complications following lumbar spine surgery based on the level of surgical invasiveness. Secondarily, we sought to determine a threshold value at which obesity significantly impacts complications. We hypothesized that obesity is an independent risk factor for complications following lumbar spine surgery and that there is an identifiable BMI that predicts risk of postoperative complications. Materials and Methods National Surgery Quality Improvement Program (NSQIP) The American College of Surgeons NSQIP was implemented in 1991 by the US Department of Veterans Affairs to model patient risk-adjusted outcomes for patient factors and operative procedure. Patient data are collected by a Surgical Clinical Reviewer auditing participating hospitals. It samples all patient populations undergoing general, spinal, or epidural anesthesia. Patient data collected include primary diagnosis in the form of an International Classification of Diseases, Ninth Revision, Clinical Modification code, procedure as indicated by Current Procedural Terminology (CPT) coding, 70 preoperative risk factors, 11 variables related to the surgery, and 24 postoperative outcomes. Complications, outcomes, and any further procedures are recorded for 30 days after enrollment. More detailed information on the NSQIP design is available at www.facs.org/quality-programs/acs-nsqip [15]. Institutional review board approval is exempt from studies using de-identified NSQIP data.

ARTICLE IN PRESS O.J. Bono et al. / The Spine Journal ■■ (2017) ■■–■■

Patient Population NSQIP data collected between 2011 and 2013 were analyzed. Inclusion criteria for the current analysis were patients over 18 years of age undergoing arthrodesis, discectomy, laminectomy, laminoplasty, corpectomy, or osteotomy of the lumbar spine as defined by CPT codes (available in Supplementary material Appendix A). Patients with the following diagnoses were excluded from analysis: ascites, pre-operative acute renal failure, preoperative dialysis, disseminated cancer, or open wound or wound infection. Data Analysis Patients meeting inclusion and exclusion criteria were categorized by BMI according to the World Health Organization stratification: underweight (BMI<18.5), normaloverweight (BMI 20.0–29.9), obesity classes 1 (BMI 30.0–34.9), 2 (BMI 35.0–39.9), and 3 (BMI≥40) [16]. To objectively calculate the surgical invasiveness of the procedure for each patient, we used the surgical invasiveness index (SII) proposed by Mirza et al. [12]. The SII is calculated by the sum, across all vertebral levels, of 6 possible interventions on each operated vertebra: anterior decompression, anterior fusion, anterior instrumentation, posterior decompression, posterior fusion, and posterior instrumentation with a score of 1 given to each procedure performed. The invasiveness grading was modified to allow for inclusion of procedures that were felt to also represent invasiveness—we added five additional possible interventions: score of 9 for 3 column osteotomy, score of 3 for revision procedures, a score of 2 for fixation to the pelvis, a score of 2 for interbody device, and a score of 1 for Smith Peterson osteotomy (Supplementary material Appendix B). Patients were dichotomized based on the 75th SII percentile cutoff into Low-SII and High-SII groups. Differences in complications rates across BMI groups were analyzed by Bonferroni analysis of variance method. Demographic information collected included age, gender, race, BMI, American Society of Anesthesiologists grade, and Charlson Comorbidity Index (CCI) score, measured using standardized International Classification of Diseases, Ninth Revision protocol (Supplementary material Appendix C). Postoperative outcomes assessed included length of stay (LOS), mortality, and readmission within 30 days. Mean operative variables assessed were operative time, SII, levels fused, levels decompressed, interbody device use, and three-column osteotomy use. We performed analysis of increasing BMI as a risk factor for different complications by category as delineated by the NSQIP dataset (cardiopulmonary, infection, airway, surgical, medical, and “any”), as well as individually (Supplementary material Appendix D). The individual complications analyzed included: medical complications such as renal failure, deep vein thrombosis, pneumonia, pulmonary embolism, sepsis, unplanned intubation, cardiac arrest, myocardial infarction, coma, blood transfusions, occurrences

3

on ventilator>48 hours, iatrogenic nerve injury, and stroke; surgical complications such as wound disruption, graft failure, deep and superficial surgical site infection (SSI). Complications were categorized according to the organ-system involvement as: airway (unplanned intubation and ventilator>48 hours), cardiopulmonary (cardiac arrest, pulmonary embolism, deep vein thrombosis, unplanned intubation, ventilator>48 hours, pneumonia), and infection (deep SSI, superficial SSI, pneumonia, urinary tract infection, sepsis, organ or space SSI). Multivariate binary logistic regression controlling for baseline difference in SII, CCI, diabetes, hypertension, and smoking was used to measure the relationship between BMI and complication in all patients, and then separately in high-SII and low-SII surgeries. Odds ratios were reported as (odds ratio [95% CI]), with significance set at p<.05. Results Patient demographics and associated comorbidities We identified 31,763 eligible surgical patients (51% men, 49% women) aged 18 and older undergoing lumbar spine surgery in the years 2011–2013, out of which 23,509 patients (74%) were categorized in the low-SII group (mean SII=1.77), and 8,254 patients (26%) in the high-SII group (mean SII=11.73). The stratification by BMI revealed that 305 patients were underweight (BMI<18.5), 17,206 patients were normal-overweight (BMI 20.0–29.9), 8,055 patients were in obesity class 1 (BMI 30.0–34.9), 3,888 patients were in obesity class 2 (BMI 35.0–39.9), and 2,309 patients were in obesity class 3 (BMI≥40). The CCI and American Society of Anesthesiologists classification for all obesity classes were greater than the normal-overweight cohort (all p<.001) (Table 1). Obesity classes 1, 2, and 3 had higher incidence of diabetes and hypertension, and lower percentages of smokers compared with the normal-overweight cohort, all p<.01. All obesity classes had more levels decompressed (obesity 1: 1.269, p<.001; obesity 2: 1.271, p<.001; obesity 3: 1.257; p=.042), and longer operative times compared with normal-overweight patients (p<.001) (Table 2). Obesity class 2 patients underwent higher invasive surgical procedures compared with normal-overweight patients (obesity 2: 4.515; normaloverweight: 4.225; p=.008). Overall medical and surgical complications Without consideration of surgical invasiveness, obesity class 1 independently predicted 4 of the 11 complications analyzed in multivariate analysis; obesity class 2 independently predicted 6 of 11 complications, including any complication, infection, surgical, renal insufficiency, renal failure, and superficial SSI; and obesity class 3 independently predicted 9 of 11 complications (Table 3). Of the complication categories that we created, the odds ratios for any complication(obesity 2: 1.218; obesity 3: 1.742), infection

ARTICLE IN PRESS O.J. Bono et al. / The Spine Journal ■■ (2017) ■■–■■

4

Table 1 Demographics of patients undergoing lumbar spinal surgery stratified by obesity

CCI Gender (F) Age (y) White Black Asian ASA Diabetes Smoker HTN LOS (d) Readmission Death

Underweight N=305

p

Normal weight N=17,206

Obesity class 1 N=8,055

p

Obesity class 2 N=3,888

p

Obesity class 3 N=2.309

p

0.1311 73.11% 58.93 92.4% 3.8% 3.8% 2.2984 4.26% 31.15% 34.75% 3.2951 5.05% 0.00%

1.000 <0.001 1.000 <.001 1.000 1.000 1.000 0.078 0.018 0.015 1.000 1.000 1.000

0.1409 0.4892 58.5093 92.5% 4.8% 2.6% 2.2640 9.68% 23.70% 43.75% 2.7781 4.77% 0.09%

0.2324 45.68% 58.56 92.2% 6.6% 1.1% 2.4141 18.14% 19.69% 57.74% 2.9985 5.28% 0.21%

<.001 <.001 1.000 <.001 <.001 .630 <.001 <.001 <.001 <.001 .073 1.000 .229

0.3063 53.10% 56.94 90.5% 8.9% 0.7% 2.5529 25.57% 19.65% 62.86% 3.1034 6.04% 0.23%

<.001 <.001 <.001 <.001 .001 1.000 <.001 <0.001 <.001 <.001 .026 .852 .425

0.3690 62.49% 54.24 88.9% 10.9% 0.2% 2.7579 31.62% 18.80% 67.30% 3.3590 6.29% 0.22%

<.001 <.001 <.001 <.001 .012 1.000 <.001 <.001 <.001 <.001 <.001 1.000 1.000

ASA, American Society of Anesthesiologists grade; CCI, Charlson Comorbidity Index; HTN, hypertension; LOS, length of stay.

(obesity 2: 1.335; obesity 3: 1.685), and surgical complication (obesity 2: 1.622; obesity 3: 2.798) were all significantly higher in obesity classes 2 and 3 relative to the normaloverweight cohort (all p<.05). Additionally, the rates of both superficial and deep incisional surgical site

infections increased with increasing obesity class (both p<.001). Rates for reoperation within 30 days (normal: 0.7%, obesity 1: 0.6%, obesity 2: 0.8%, obesity 3: 1.0%) also increased with obesity (all p<.001).

Table 2 Clinical data of patients undergoing lumbar spinal surgery stratified by obesity

SII Anterior approach Posterior approach Approach combined Levels fused Levels decompressed Interbody use 3-CO* Operative time (min)

Underweight

p

Normal weight

Obesity class 1

p

Obesity class 2

p

Obesity class 3

p

4.430 0.066 0.839 0.039 1.761 1.043 5.175 0.007 148.731

1.000 1.000 1.000 1.000 1.000 .022 1.000 1.000 1.000

4.225 0.050 0.850 0.043 1.588 1.200 4.853 0.003 153.404

4.378 0.044 0.860 0.041 1.624 1.269 4.871 0.003 165.347

.199 .272 .229 1.000 1.000 <.001 1.000 1.000 <.001

4.515 0.044 0.861 0.045 1.696 1.271 4.858 0.002 174.172

.008 1.000 .667 1.000 .108 <.001 1.000 1.000 <.001

4.297 0.036 0.867 0.032 1.588 1.257 4.815 0.000 176.825

1.000 .021 .223 .093 1.000 .042 1.000 .139 <.001

SII, surgical invasiveness index. * 3-CO, three-column osteotomy includes pedicle subtraction osteotomy, vertebral column resection, and anterior wedge osteotomy. Bold values indicate statistical significance, set to p < .05.

Table 3 Multivariate analysis on change in risk of complications by obesity group relative to “normal” BMI in all patients All patients OR [95% CI]

Underweight (BMI<18.5)

Obesity class 1 (BMI 30.0 – 34.99)

Obesity class 2 (BMI 35.0 – 39.99)

Obesity class 3 (BMI≥40)

Any Cardiopulmonary Wound disruption Infection Surgical Medical Renal insufficiency Renal failure Airway DVT Superficial surgical site infection

0.615 [0.272–1.389] 0.545 [0.135–2.210] — 0.661 [0.292–1.495] 0.513 [0.127–2.080] 0.765 [0.313–1.868] — — — — 0.487 [0.068–3.505]

1.094 [0.948–1.264] 1.263 [1.019–1.566] 0.897 [0.496–1.625] 1.146 [0.984–1.335] 1.146 [0.911–1.443] 1.051 [0.885–1.249] 3.140 [1.308–7.537] 2.147 [0.651–7.076] 1.541 [1.007–2.359] 1.351 [0.936–1.950] 1.389 [1.029–1.876]

1.218 [1.020–1.455] 1.219 [0.925–1.607] 1.057 [0.519–2.154] 1.335 [1.110–1.605] 1.622 [1.250–2.104] 0.990 [0.790–1.240] 2.936 [1.050–8.206] 5.491 [1.770–17.036] 1.364 [0.787–2.362] 1.480 [0.939–2.334] 1.811 [1.278–2.565]

1.742 [1.439–2.110] 1.420 [1.032–1.954] 1.994 [1.016–3.916] 1.685 [1.372–2.069] 2.798 [2.154–3.634] 1.259 [0.978–1.619] 3.866 [1.309–11.418] 9.933 [3.236–30.490] 1.209 [0.604–2.419] 1.939 [1.166–3.224] 3.446 [2.456–4.836]

DVT, deep vein thrombosis. Bold values indicate statistical significance, set to p < .05.

ARTICLE IN PRESS O.J. Bono et al. / The Spine Journal ■■ (2017) ■■–■■

5

Table 4 Multivariate analysis on change in risk of complications by obesity group relative to “normal” BMI for bottom 75th and top 25th percentile surgical invasiveness index (SII) Bottom 75th percentile SII OR [95% CI]

Underweight (BMI<18.5)

Obesity class 1 (BMI 30.0–34.99)

Obesity class 2 (BMI 35.0–39.99)

Obesity class 3 (BMI≥40)

Any complication Cardiopulmonary Infection Wound disruption Surgical Medical Renal insufficiency Renal failure Airway DVT Superficial surgical site infection

0.395 [0.097–1.599] — 0.578 [0.183–1.822] — 0.428 [0.059–3.080] 0.326 [0.045–2.341] — — — — —

1.273 [1.055–1.536] 1.304 [0.976–1.742] 1.230 [1.010–1.498] 0.893 [0.407–1.959] 1.256 [0.939–1.680] 1.217 [0.966–1.534] 7.839 [2.217–27.722] 9.871 [1.180–82.559] 1.606 [0.896–2.879] 1.296 [0.790–2.127] 1.486 [1.025–2.154]

1.403 [1.114–1.767] 1.149 [0.779–1.694] 1.329 [1.043–1.694] 0.969 [0.362–2.598] 1.671 [1.195–2.336] 1.185 [0.881–1.595] 5.967 [1.411–25.241] 5.740 [0.510–64.636] 0.897 [0.365–2.205] 1.392 [0.741–2.616] 1.768 [1.138–2.747]

2.011 [1.572–2.572] 1.661 [1.099–2.509] 1.674 [1.282–2.186] 2.762 [1.233–6.188] 2.785 [1.989–3.899] 1.523 [1.097–2.113] 5.495 [1.087–27.775] 22.713 [2.591–199.125] 0.479 [0.112–2.045] 3.121 [1.744–5.585] 2.924 [1.879–4.550]

Top 25th percentile SII OR [95% CI]

Underweight (BMI<18.5)

Obesity class 1 (BMI 30.0–34.99)

Obesity class 2 (BMI 35.0–39.99)

Obesity class 3 (BMI≥40)

Any Cardiopulmonary Wound disruption Infection Surgical Medical Renal insufficiency Renal failure Airway DVT Superficial Surgical Site Infection

0.869 [0.313–2.414] 1.123 [0.270–4.670] — 0.792 [0.246–2.548] 0.651 [0.089–4.756] 1.188 [0.426–3.310] — — — — 1.419 [0.191–10.531]

0.855 [0.681–1.073] 1.179 [0.855–1.625] 0.861 [0.348–2.131] 1.001 [0.785–1.276] 0.960 [0.660–1.399] 0.846 [0.652–1.099] 0.353 [0.041–3.044] — 1.435 [0.768–2.680] 1.380 [0.799–2.383] 1.191 [0.714–1.987]

0.980 [0.740–1.298] 1.274 [0.857–1.893] 1.123 [0.398–3.172] 1.316 [0.989–1.752] 1.519 [1.003–2.299] 0.772 [0.544–1.095] 1.207 [0.225–6.480] 5.860 [1.621–21.188] 1.815 [0.889–3.702] 1.535 [0.791–2.976] 1.847 [1.045–3.266]

1.460 [1.071–1.990] 1.178 [0.709–1.958] 1.084 [0.303–3.881] 1.743 [1.260–2.411] 2.879 [1.896–4.370] 1.010 [0.678–1.505] 2.809 [0.626–12.609] 7.240 [1.727–30.349] 2.008 [0.874–4.613] 0.618 [0.186–2.054] 4.509 [2.645–7.688]

DVT, deep vein thrombosis.Bold values indicate statistical significance, set to p < .05

Complications stratified on the basis of surgical invasiveness In the low-SII group, obesity class 1 independently predicted 5 of the 11 complications, obesity class 2 independently predicted 5 of 11 complications, and obesity class 3 independently predicted 10 of 11 complications (Table 4). The odds ratios for any complication (obesity 1: 1.273; obesity 2: 1.403; obesity 3: 2.011), infection (obesity 1: 1.230; obesity 2: 1.329; obesity 3: 1.674), renal insufficiency (obesity 1: 7.839; obesity 2: 5.967; obesity 3: 5.495), and superficial SSI (obesity 1: 1.486; obesity 2: 1.768; obesity 3: 2.924) were significantly higher for all obesity classes (all p<.05). Surgical complication (obesity 2: 1.671; obesity 3: 2.785; all p<.05) was greater for both obesity classes 2 and 3. Similarly, in the high-SII group, obesity class 2 independently predicted 3 of 11 complications analyzed in multivariate analysis, and obesity class 3 independently predicted 5 of 11 complications. The odds ratios for surgical complication (obesity 2: 1.519; obesity 3: 2.879), renal failure (obesity 2: 5.860; obesity 3: 7.240), and superficial SSI (obesity 2: 1.847; obesity 3: 4.509) were significantly higher in obesity classes 2 and 3 compared with the normal-overweight cohort (all p<.05) (Table 4).

Postoperative outcomes Obesity class 2 (3.10 days) and class 3 (3.36 days) had significantly longer LOS compared with the normal-overweight cohort (2.78 days) (p<.05) (Table 1). Considering invasiveness, average LOS in high-SII procedures was 4.73 days, and LOS in low-SII procedures was 2.54 days; however, for both SII groups, LOS did not differ among BMI cohorts (p>.05). Mortality is also presented in Table 1; there were no significant differences among BMI groups for mortality (p>.05). Finally, readmission within 30 days was measured, and no significant difference was found among BMI groups (p>.05) (Table 1). Discussion Historically, the consensus on the effect of BMI on complications following lumbar surgery has been controversial. Although some studies suggest that there is a positive correlation between increasing BMI and complication occurrence, others provide conflicting evidence [17,18]. One of the potential reasons for this controversy is that several prior studies did not find clear disparities in complications because they did not stratify patients based on obesity class, rather they took all “obese” patients together under the definition of

ARTICLE IN PRESS 6

O.J. Bono et al. / The Spine Journal ■■ (2017) ■■–■■

BMI>30. The potential for notable differences between a patient with BMI 30–35 and a patient with BMI>40 exists; this is not accounted for in an analysis that looks only at nonobese (BMI<30) versus obese (BMI≥30) patients. In this study comparing 31,763 normal-overweight, underweight, and obese patients, we found that at obesity class 2 (BMI≥35) there is a significant increase in postoperative lumbar spine surgery complications, especially any complication, infection, and surgical complications, given the difference in the number of significantly higher rates of complications between obesity class 1 and obesity class 2 (Table 3). Additionally, the odds ratios for complications in obesity class 2 relative to the normal-overweight group were greater than the values of odds ratios for complications in obesity class 1 relative to the normal-overweight group. Buerba et al. performed a retrospective analysis of a cohort of 10,387 prospectively collected patients undergoing lumbar spine surgery using NSQIP and found that high BMI results in higher complication rates; most importantly, they concluded that there is a BMI threshold at obesity class 3 (BMI≥40) [17]. However, we found that complication risk significantly increases at obesity class 2 (BMI 35.0–39.9). Our contrasting results are perhaps a reflection of the increased sample size of 31,763 patients, more than three times the size of Buerba et al.’s patient cohort. This disparity in sample size is likely because of our inclusion of an even larger subset of lumbar surgeries, which included more than the fusions, discectomies, and decompressions that Buerba et al. evaluated. Additionally, our NSQIP data were collected from 2011 to 2013, whereas Buerba et al. collected NSQIP data from 2005 to 2010, making our study potentially more applicable to today’s surgical environment. De la Garza-Ramos et al. investigated a smaller patient cohort of 732 patients undergoing lumbar fusion and found that although most of the complications following lumbar surgery occurred in obese patients with BMI≥30, complications began to rise in overweight patients with BMI≥25 [3]. This study also had a small sample size compared with the larger database studies and did not further divide the “obese” category into different classes of BMI range. Obesity has been found to be positively correlated with blood loss, surgery duration, SSI, and venous thromboembolisms, in addition to perioperative complications [6,12,14]. We found that across all groups, risk of infection, surgical complications, renal insufficiency, renal failure, and superficial SSI were significantly higher for both obesity classes 2 and 3 than for normal-overweight patients. Our results align with previously reported data [8,10,19]. In a recent study using the Medicare database, Puvanesarajah et al. reported that morbidly obese patients had increased risk of major medical complications, wound complications, and 30-day readmissions [5]. Although multiple studies have investigated the effect of surgical invasiveness on risk of surgical complications, no study has looked at the effect of SII specifically in obese patients undergoing lumbar spine surgery compared with nonobese

patients [12,20]. We found that in low-SII cases, all obesity classes were associated with significantly higher rates of infection, whereas in high-SII cases, only obesity class 3 had higher rates of infection complications. However, obesity classes 2 and 3 in both SII groups had higher rates of surgical complications and superficial SSI. It is important to note that when looking at obesity classes regardless of SII, we found significant increases in all complications, infection-related complications, and surgical complication categories for obesity class 2, with that rate further increasing for obesity class 3. This modified definition of significant BMI elevation that poses additional surgical risk to a patient is of critical clinical importance given the overwhelming prevalence of general obesity in the population. Such data will help guide the preoperative decision making process between the surgeon and the patient. It will also encourage surgeons and health-care systems to be more aware of the increased risk of a multitude of adverse events in patients with BMI≥35 regardless of the invasiveness of the procedure. Because of the increased morbidity risks in obese patients, multiple studies have demonstrated that weight loss before spine surgery leads to significantly fewer complications [3,21]. However, the potential for mitigating these surgical risks by reduction of only one obesity class, below a BMI of 35 kg/m2, may encourage patients to achieve realistic weight loss goals before surgery to optimize outcomes—because previous goals of normalizing BMI are perhaps unrealistic and extremely difficult to achieve given contemporary standards. These results should be interpreted with caution especially because the literature has not defined what an “acceptable” complication rate would be. It is important to note that previous literature has documented that obese patients, despite increased risk of adverse events, report improvements from spine surgery as measured by patient-derived outcome measures and satisfaction scores [22,23]. Further study is needed to more intricately assess the risk-benefit ratio of surgical intervention in this patient population. Limitations This study is primarily limited by the retrospective nature of NSQIP, which creates the possibility of selection and treatment bias [3]. The design of NSQIP limits the study, for it is a large database with little granularity, and it includes only 30-day follow-up data. As a result, there is a lack of sustained follow-up to capture complications (both mid- and longterm complications) and patient reported outcome measures, which hinders our ability to study the overall effects of the proposed comorbidities and treatment and warrants further investigation. Additionally, NSQIP does not provide any information about the severity of patients’ spinal pathology and alignment, or their disability, which likely plays a role in clinical outcomes. Furthermore, the lack of radiographic data forces us to rely exclusively on CPT codes. The use of a large non–surgeon-derived data to make specific clinical recommendations and conclusions based on technical aspects of

ARTICLE IN PRESS O.J. Bono et al. / The Spine Journal ■■ (2017) ■■–■■

surgery also warrants further study using a more specific and granular dataset to confirm the present findings. Lastly, the study may be limited by the use of BMI to categorize the patients, given the argument that BMI is not the best indicator of obesity because of differences in body habitus [24]. Conclusion There is a significant increase in complications, most possibly infection and surgical complications, in patients with BMI≥35 following lumbar spine surgery. Surgeons can use this information preoperatively to counsel patients in obesity classes 2 and 3 on their risk of complications. Additionally, encouragement of preoperative weight loss by even limited amounts, such as reduction by a single BMI class, can potentially mitigate obese patients’ surgical risk. Prospective studies are necessary to further assess the benefits of this potential positive intervention. Supplementary material Supplementary material related to this article can be found at https://doi.org/10.1016/j.spinee.2017.11.015. References [1] Jackson KL, Devine JG. The effects of obesity on spine surgery: a systematic review of the literature. Global Spine J 2016;6:394–400. [2] Khaodhiar L, McCowen KC, Blackburn GL. Obesity and its comorbid conditions. Clin Cornerstone 1999;2:17–28. [3] De la Garza-Ramos R, Bydon M, Abt N, et al. The impact of obesity on short and long-term outcomes after lumbar fusion. Spine 2014; 40:56–61. [4] Ogden CL, Carroll MD, Kit BK, et al. Prevalence of childhood and adult obesity in the United States, 2011–2012. JAMA 2014;311:806–14. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid =4770258&tool=pmcentrez&rendertype=abstract. [Internet]. [5] Puvanesarajah V, Cancienne JM, Pehlivan H, et al. Morbid obesity and lumbar fusion in patients over 65 years of age. Spine 2016;42:122–7. Available at: http://content.wkhealth.com/linkback/openurl?sid =WKPTLP:landingpage&an=00007632-900000000-96034. [Internet]. [6] Marawar S, Girardi FP, Sama AA, et al. National trends in anterior cervical fusion procedures. Spine 2010;35:1454–9. Available at: http://www.ncbi.nlm.nih.gov/pubmed/20216341. Accessed October 13, 2014. [Internet]. [7] Owens RK, Djurasovic M, Onyekwelu I, et al. Outcomes and revision rates in normal, overweight and obese patients five years after lumbar fusion. Spine J 2016;16:1178–83. Available at: http://dx.doi.org/10.1016/ j.spinee.2016.06.005. [Internet]. Elsevier Inc. [8] Djurasovic M, Bratcher KR, Glassman SD, et al. The effect of obesity on clinical outcomes after lumbar fusion. Spine 2008;33:1789–92. Available at: http://www.ncbi.nlm.nih.gov/pubmed/18628712. [Internet]. [9] Pereira BJA, De Holanda CVM, Ribeiro CAA, et al. Impact of body mass index in spinal surgery for degenerative lumbar spine disease. Clin Neurol Neurosurg 2014;127:112–15. Available at: http://dx.doi.org/ 10.1016/j.clineuro.2014.09.016. [Internet]. Elsevier B.V. [10] Higgins DM, Mallory GW, Planchard RF, et al. Understanding the impact of obesity on short-term outcomes and in-hospital costs after instrumented spinal fusion. Neurosurgery 2015;78:127–32. [11] Seicean A, Alan N, Seicean S, et al. Impact of increased body mass index on outcomes of elective spinal surgery. Spine 2014;39:1520–30. Available at: http://www.ncbi.nlm.nih.gov/pubmed/24859584. Accessed July 14, 2014. [Internet].

7

[12] Mirza SK, Deyo R, Heagerty PJ, et al. Development of an index to characterize the “invasiveness” of spine surgery: validation by comparison to blood loss and operative time. Spine 2008;33:2651–61. discussion 2662. [13] Cizik AM, Lee MJ, Martin BI, et al. Using the spine surgical invasiveness index to identify risk of surgical site infection: a multivariate analysis. J Bone Joint Surg 2012;94:335–42. American Vol. Department of Orthopaedics and Sports Medicine, University of Washington Medical Center, 1959 N.E. Pacific Street, Box 356500, Seattle, WA 98195-6500, USA. [email protected]. [14] Marquez-Lara A, Nandyala S V, Sankaranarayanan S, et al. Body mass index as a predictor of complications and mortality after lumbar spine surgery. Spine 2014;39:798–804. Available at: http:// www.ncbi.nlm.nih.gov/pubmed/24480950. Accessed April 18, 2014. [Internet]. [15] American College of Surgeons. ACS NSQIP User Guide [Internet]. National Surgical Quality Improvement Program (NSQIP); 2013. Available at: http://site.acsnsqip.org/wp-content/uploads/2013/ 10/ACSNSQIP.PUF_.UserGuide.2012.pdf. Accessed April 1, 2014. [Internet]. [16] World Health Organization (WHO). Obesity: preventing and managing the global epidemic. Report of a WHO consultation. World Health Organ Tech Rep Ser 2000;894(i–xii):1–253. [17] Buerba RA, Fu MC, Gruskay JA, et al. Obese Class III patients at significantly greater risk of multiple complications after lumbar surgery: an analysis of 10,387 patients in the ACS NSQIP database. Spine J 2013;14:2008–18. Available at: http://www.ncbi.nlm.nih.gov/pubmed/ 24316118. Accessed July 14, 2014. [Internet]. Elsevier Inc. [18] Yadla S, Malone J, Campbell PG, et al. Obesity and spine surgery: reassessment based on a prospective evaluation of perioperative complications in elective degenerative thoracolumbar procedures. Spine J 2010;10:581–7. Available at: http://www.ncbi.nlm.nih.gov/pubmed/ 20409758. Accessed July 14, 2014. [Internet]. Elsevier Inc. [19] Pereira BJ, de Holanda CV, Ribeiro CA, et al. Impact of body mass index in spinal surgery for degenerative lumbar spine disease. Clin Neurol Neurosurg 2014;127:112–15. Department of Neurosurgery, Center of Neurology and Neurosurgery Associates (CENNA), Hospital Beneficencia Portuguesa de Sao Paulo, Sao Paulo, Brazil; Division of Neurosurgery, Department of Medical Sciences, School of Medicine, University Nove de Julho Elsevier B.V. [20] Cizik AM, Lee MJ, Martin BI, et al. Using the spine surgical invasiveness index to identify risk of surgical site infection: a multivariate analysis. J Bone Joint Surg Am 2012;94:335–42. Available at: http://jbjs.org/content/94/4/335.abstract. [Internet]. [21] Melissas J, Volakakis E, Hadjipavlou A. Low-back pain in morbidly obese patients and the effect of weight loss following surgery. Obes Surg 2003;13:389–93. Available at: http://www.ncbi.nlm.nih.gov/ pubmed/12841899/nhttp://download.springer.com/static/pdf/ 411/art%253A10.1381%252F096089203765887714.pdf?auth66 =1383404335_c820c3c1110fb7d2d197d87dab96d772&ext=.pdf. [Internet]. [22] Rihn JA, Radcliff K, Hilibrand AS, et al. Does obesity affect outcomes of treatment for lumbar stenosis and degenerative spondylolisthesis? Analysis of the Spine Patient Outcomes Research Trial (SPORT). Spine 2012;37:1933–46. Rihn, J.A., Rothman Institute, Department of Orthopedic Surgery, Thomas Jefferson University, Philadelphia, PA 19107, United States. [23] Rihn JA, Kurd M, Hilibrand AS, et al. The influence of obesity on the outcome of treatment of lumbar disc herniation: analysis of the Spine Patient Outcomes Research Trial (SPORT). J Bone Joint Surg Am 2013;95:1–8. Available at: http://www.pubmedcentral.nih.gov/ articlerender.fcgi?artid=3528022&tool=pmcentrez&rendertype=abstract. Accessed November 5, 2014. [Internet]. [24] Rothman KJ. BMI-related errors in the measurement of obesity. Int J Obes (Lond) 2008;32(suppl 3):S56–9.