Obesity is Not Associated with Perioperative Adverse Events in Patients Undergoing Complex Revision Surgery for the Thoracolumbar Spine

Study Design: Retrospective cohort study Objective: This study sought to determine the relationship between obesity and perioperative adverse events (AEs) with patients undergoing complex revision thoracolumbar spine surgery while controlling for psoas muscle index (PMI) as a confounding variable. Methods: Between May 2016 and February 2020, a retrospective analysis of individuals undergoing complex revision surgery of the thoracolumbar spine was performed at a single institution. Obesity was dened as BMI ≥ 30.0 kg/m 2 . PMI < 500 mm 2 /m 2 for males and < 412 mm 2 /m 2 for females were used to dene sarcopenia. A Spine Surgical Invasiveness Index (SSII) > 10 was used to dene complex revision surgery. A multivariable logistic regression model was used to ascertain the effects of sarcopenia, obesity, age, and gender on the likelihood of the occurrence of any AE. Results: The study included 114 patients. 54 patients were in the obese cohort and 60 patients in the non-obese cohort. There was not a signicant difference in perioperative outcomes of both the obese and non-obese patients. Multivariable analysis demonstrated that sarcopenic individuals had a signicantly higher likelihood for an AE than non-sarcopenic individuals (OR: 7.53, 95% CI: 3.05-18.60). Obesity did not have a signicant effect in predicting AEs. Conclusions: Obesity is not associated with perioperative AEs, 30-day readmission rates, 30-day reoperation rates, rate of discharge to a facility, or post-operative length of stay (LOS) among patients undergoing complex revision thoracolumbar spine surgery.


Introduction
Currently, over 100 million adults living in the United States are considered obese and overweight. 1 The NIH utilizes body mass index (BMI) as a parameter of categorizing weight status, calculating BMI using body weight in kilograms dived by height in meters squared. 2 Obesity is a chronic disease recognized clinically as having a BMI > 30 in adults and has association with other comorbidities such as Type 2 Diabetes, cardiovascular disease and dyslipidemia. 3 The impact of obesity on spinal surgeries remains unclear, and complications related to obesity have been reported in some studies. 4,5 Studies by Shamji et al. and Manson et al. found that obesity is associated with increased resources and perioperative transfusion requirements after elective thoracolumbar spine surgery; however, they reported no other associations between obesity and increased adverse events (AEs). 6,7 Another study by Yadla et al. investigating outcomes after primary elective thoracolumbar spine surgery found no associates between BMI and increased AEs. 8 Furthermore, a more recent study by Varshneya et al. reported that although obesity may be associated with overall health burdens outside of the perioperative environment, there is no signi cant relationship attributed between obesity and AEs after primary thoracolumbar deformity surgery. 9 Revision thoracolumbar spine surgeries have been associated with signi cantly higher perioperative AEs compared with primary surgeries. [10][11][12] An increase in surgical complexity has also been associated with higher perioperative AEs in spine surgery. 13 Therefore, identifying speci c prognostic indicators for complex revision thoracolumbar spine surgery is important in order to stratify pre-operative risk for these patients. A recent study by Hirase et al. demonstrated that sarcopenia, de ned by psoas muscle index (PMI) below 500 mm 2 /m 2 for males and 412 mm 2 /m 2 for females, is predictive of increased perioperative AEs among patients undergoing complex revision thoracolumbar spine surgery. 14 However, this study reported a signi cantly lower BMI among sarcopenic patients, indicating the presence of a possible confounding factor for predicting perioperative AEs within this patient cohort. Our understanding is that there are no previous studies that evaluate the effect of obesity status on patients undergoing complex revision thoracolumbar spine surgery. Thus, the purpose of this study was to elucidate the relationship between obesity and perioperative AEs among patients undergoing complex revision thoracolumbar spine surgery while controlling for PMI as a confounding variable.

Materials And Methods
The same patient cohort and similar methodologies were utilized from our previous publication by Hirase et al. 14 The local institutional review board approved the study procedure on April 13, 2020. Due to the retrospective observational nature of this study, the informed consent was waived.

Study Population
Between May 2016 and February 2020, a retrospective analysis was performed at a single institution of patients receiving complex revision thoracolumbar spine surgery by three board certi ed fellowshiptrained orthopaedic spine surgeons. The same inclusion and exclusion criteria for the same cohort of patients from our previous publication by Hirase et al. was applied for this study. 14 Any patients age 18 or above undergoing complex revision thoracolumbar spine surgery were included. Any history of prior surgical intervention of the same vertebral level was de ned as revision surgery. A Spine Surgical Invasiveness Index (SSII) > 10 was used to de ne complex surgery. 18 Patients lacking a pre-operative CT or MRI of the lumbar spine obtained at our facility within six months of surgery, poor image quality, preoperative MRI or CT acquired at any outside facilities, clinical evidence of L1 or L2 nerve root compression, a history of previous surgical treatment to or through the psoas muscle, and coronal deformity greater than 20 degrees were excluded as mentioned in our previous study. 14

Page 4/20
Electronic medical records were utilized retrospectively to obtain demographic data including age, gender, BMI, American Anesthesiologists' Society (ASA) class, comorbidities, indication for revision operation, baseline ambulation status, and presence of neurologic de cits. BMI ≥ 30.0 kg/m 2 was used to de ne obesity. Intraoperative data was acquired which included estimated blood loss (EBL) and operative time de ned as time of incision to post-operative dressing placement. SSII, a veri ed method of evaluation and comparison regarding spine surgery complexity was used, with a range of 0-48, which accounts for surgical approaches and the amount of decompressed, fused, and instrumented vertebral levels. 18 In regard to the included surgical population with revision surgeries, only the additional levels of fusion, instrumentation, or decompression were accounted for the SSII calculation as previously stated. 14 Each patients' primary surgery SSII score was also obtained.

Assessment Of Sarcopenia
PMI was used to analyze sarcopenia, which was calculated by measuring the total cross-sectional area (CSA) of the bilateral psoas muscles at the L3 vertebral body using the pre-operative T1 weighted MRI or CT normalized to body height 2 (mm 2 /m 2 ). The total CSA was measured using OsiriX DICOM Viewer software (Version 11.0, Bernex, Switzerland) by manual outlining the bilateral psoas muscles at the rst axial cut, done in the craniocaudal direction where both transverse processes are visible at the L3 level. 14,15 All of the images were obtained at a single institution with the same scanning protocols ensuring identical scanning thickness among all images analyzed. Three separate reviewers performed all measurements to improve interobserver reliability. Each measurement was acquired three times by all reviewers to improve intraobserver reliability. Intraclass correlation coe cient (ICC) was used to assess interobserver and intraobserver reliabilities, where an ICC above 0.90 signi es excellent agreement, between 0.75 and 0.90 signi es good agreement, between 0.5 and 0.75 signi es moderate agreement, and below 0.5 signi es poor agreement. 16 Each of the mean values obtained by the three reviewers was divided by the square of patient height to calculate the PMI as previously described. 14 To minimize the risk of bias, all reviewers were blinded to their respective measurements as well as to patient demographics and outcomes. Sarcopenia was de ned as PMI < 500 mm 2 /m 2 for males and < 412 mm 2 /m 2 for females as previously de ned. 14

Outcome Measures
Retrospective analysis of electronic medical records were used to review all perioperative outcomes. The primary outcome measures were perioperative AEs which included post-operative anemia that required transfusion, cardiac complication (cardiac arrest and myocardial infarction), sepsis, wound complication (wound dehiscence and deep wound infection), acute kidney injury (AKI), delirium, intra-operative dural tear, pneumonia, urinary tract infection (UTI), urinary retention, epidural hematoma, and deep vein thrombosis (DVT). The secondary outcome measures used were 30-day readmission rates, 30-day reoperation rates, in-hospital mortality rates, discharge disposition (home vs facility) and post-operative hospital length of stay (LOS). The number of days from surgery (or the last surgery if staged procedure) to discharge to either home or facility was used to de ned postoperative LOS.
Statistical Analysis SPSS statistical software (Version 25.0; SPSS, Inc, Chicago, IL) was utilized to perform data analysis. The Chi-Square or Fisher's exact test and continuous data was used to analyze categorical data and was further analyzed using Two-tailed student t-test. Continuous variables with non-normal distribution was analyzed using the Mann-Whitney U test. Statistical signi cance was set to p-value < 0.05. The odds ratio (OR) with 95% con dence interval (CI) was calculated for comparing perioperative outcomes. Post hoc power analysis with a two-tailed alpha of 0.05 was performed between obese and non-obese groups to evaluate the power of detecting differences between patients experiencing any perioperative. A multivariable logistic regression model was used to determine the effects of sarcopenia, obesity, age, and gender on the likelihood of the occurrence of any AE.

Results
In total, there were 166 patients that met the inclusion criteria and 52 were removed based on the exclusion criteria. Final analysis included 114 patients (mean age 60.1 ± 15.4 years, 45 males, 69 females). The overall mean PMI was 495.0 ± 182.9 mm 2 /m 2 . Interobserver and intraobserver reliabilities were considered excellent with ICC of 0.908 (95% CI 0.862-0.944) and 0.962 (95% CI 0.928-0.975), respectively. 54 patients were in the obese cohort and 60 patients in the non-obese cohort. The obese patients had a higher BMI and PMI compared to non-obese patients; otherwise, there were no signi cant differences in baseline demographics, comorbidities, presence of motor/sensory de cits, ambulatory status, indication for reoperation, or SSII between the two groups (Table 1). No signi cant difference between the perioperative outcomes was found among the obese and non-obese patients ( Table 2).  Using the prior de ned de nition of sarcopenia as PMI < 500 mm 2 /m 2 for males and < 412 mm 2 /m 2 for females, 15 patients were considered obese and sarcopenic and 34 patients were considered non-obese and sarcopenic. 14 No signi cant difference in perioperative outcomes was found between these patients (Table 3). 39 patients were considered obese and non-sarcopenic and 26 patients were considered non-  of the cases were correctly classi ed by the model, which also explained 29.6% (Nagelkerke R 2 ) of the variance in any AE. The model revealed that sarcopenic individuals had a signi cantly higher likelihood for an AE than non-sarcopenic individuals (OR: 7.53, 95% CI: 3.05-18.60). Obesity, age, and gender did not have a signi cant effect in predicting AEs ( Table 5). The post hoc power analysis between the obese and non-obese groups calculated a 95.2% power of detecting differences between patients experiencing any perioperative AEs.

Discussion
In this study, the use of BMI as a predictor of perioperative AEs among patients undergoing complex thoracolumbar spine surgery was investigated. The multivariable logistic regression model con rmed that sarcopenic individuals were at a signi cantly higher likelihood to experience AE compared to nonsarcopenic individuals; however, there were no associations between BMI and perioperative AEs, 30-day readmission rates, 30-day re-operation rates, rate of discharge to a facility, or post-operative LOS among patients undergoing complex revision thoracolumbar spine surgery.
The relationship between obesity and perioperative AEs has been an area of debate over the past decade.
Various studies have shown a signi cantly higher risk of AEs among obese patients after thoracolumbar surgery. [17][18][19][20][21] The most recent study by Passias et al. determined that obese patients with a prior bariatric surgery had a signi cantly lower complication rate after thoracolumbar spine surgery compared with obese patients that did not undergo bariatric surgery. 21 Our ndings contradict these series of studies as our analysis demonstrated no signi cant difference in post-operative AEs between obese and non-obese patients. There are likely multiple reasons that our study had different results. First, our study performed a multivariable analysis with sarcopenia, measured by PMI, which demonstrated a signi cantly higher association with post-operative AEs. None of the prior studies that investigated the association between obesity and AEs included sarcopenia as a potential confounding variable. Second, our patient cohort consisted solely of complex revision cases, which is a particularly unique population with high complication rates at baseline. From our study analysis, it may be speculated that amongst these complex patients, pre-operative sarcopenia, and the associated debility and frailty, is a much better predictor for post-operative complications compared with the risks associated with obesity. Furthermore, studies have shown that although bariatric surgery is effective at lowering the BMI, this procedure may also induce sarcopenia post-operatively. 22,23 Thus, with the known risks of sarcopenia shown within our study, it may be inadvisable to recommend bariatric surgery to improve BMI prior to undergoing complex revision thoracolumbar spine surgery.
A particularly important process among patients undergoing complex revision surgeries known to have high perioperative complications is preoperative risk strati cation. Our previous study by Hirase et al. identi ed sarcopenia measured by PMI as a predictor of perioperative AEs among patients undergoing complex revision thoracolumbar spine surgery. 14 Our multivariable analysis within this study con rmed these ndings and also demonstrated that obesity is not associated with perioperative AEs within this patient population. This combination of ndings will speci cally assist spine surgeons during preoperative counseling and evaluation on two fronts. First, pre-operative optimization and overall conditioning to increase PMI among sarcopenic patients may be bene cial to preventing post-operative AE. Second, obese patients may not bene t from aggressive weight loss prior to these surgeries, particularly, as studies have shown that improper weight loss methods may lead to sarcopenia. 24,25 Our study has several limitations. Data accuracy is contingent on charting accuracy due to the retrospective nature of the study and may be susceptible to certain selection bias. Our results obtained from a single-center data may not be completely re ective of outcomes from other institutions due to variations in surgical technique or management. Our study also consisted of a relatively small patient cohort that may have led to underpowering to detect certain associations; however, our post hoc power analysis demonstrated that the absence of observed difference is unlikely due to lack of power.
Furthermore, there may have been small discrepancies in inter-scan agreement between CT and MRI scans used to obtain the PMI; however, this is a previously established method with studies showing good inter-scan reliability with an ICC of 0.821. 14,26 Thus, to increase the power of the study, patients were included if they received either a CT or an MRI pre-operatively. Furthermore, although this study demonstrates that obesity is not a predictor of post-operative AEs within this surgical population, we cannot directly conclude that weight loss will not be bene cial in preventing post-operative AEs.
Therefore, further studies that investigate the external validity of this study may be bene cial prior to application in practice.
In spite of the aformentioned limitations, to the best of our knowledge, this is the largest study examining the relationship between obesity and perioperative outcomes among patients receiving complex revision thoracolumbar spine surgery. This study will serve to assist spine surgeons with additional information during pre-operative counseling and evaluation regarding risks and bene ts associated with sarcopenia and obesity. For patients undergoing complex revision thoracolumbar spine surgery, our ndings may suggest that the bene ts of weight loss among obese patients may not outweigh the risk of inducing sarcopenia prior to the operation.

Conclusions
Obesity is not associated with perioperative AEs, 30-day readmission rates, 30-day re-operation rates, rate of discharge to a facility, or post-operative length of stay (LOS) among patients undergoing complex revision thoracolumbar spine surgery.