In recent decades, rates of complex spine surgery have increased 15-fold,1 and total direct/indirect costs for spine care exceed $200 billion.2 Medicare costs for lumbar spine pathology are comparable to cardiovascular disease and higher than that of diabetes.3 A recent report found that spinal fusion is one of the most expensive operative procedures offered in the United States (US), when looking at aggregate hospital costs.4 While most patients undergoing spinal surgery report reduced pain and improved quality of life, postoperative decline, or persistent symptoms are seen in up to 20% of patients.5–7 A major driver of poor outcomes is medical and surgical complications, often requiring hospital readmission.8–10
Unplanned hospital readmission is an important healthcare quality indicator that places undue stress on Medicare and Medicaid, commercial insurers, hospitals, and patients.11 Given the substantial fiscal ramifications, hospitals are becoming increasingly penalized for excessive 30-day readmission rates.12 The Hospital Readmissions Reduction Program (HRRP)12 was created in 2012 to penalize hospitals for excessive readmissions, and in 2017, an estimated 80% of hospitals will suffer reduced payments.13 Given the high cost and variability of spinal pathology, readmission after spine surgery has potential to be a major driver of healthcare-wide financial penalties.14,15 Thus, it is crucial that providers develop the ability to predict those with a high likelihood of unplanned readmissions following spine surgery.
Among 32,000 Medicare recipients undergoing lumbar spinal surgery, 30-day readmission occurred in 7.8% of patients after decompression alone and 13.0% of patients after fusion.1 Among 695 readmissions from a multi-institutional registry, the most common reasons were wound problems (38.6%), pain (22.4%), and thromboembolic events (9.4%).16 A recent meta-analysis of 13 studies further confirmed that wound problems were the most common cause for readmission (39.3%).17 Previously described factors associated with readmission include older age, recent weight loss, lung disease, cancer history, kidney dysfunction, American Society of Anesthesiologists (ASA) Class 4, operative duration, and Medicare/Medicaid insurance.11,16,17 Despite robust sample sizes, most prior studies utilize large, billing-based administrative databases8,18 which are plagued by clinical inaccuracies and coding errors. More accurate single-center series exist but often lack generalizability, as geographic variations and hospital-bed availability can influence admission trends.8,19,20 Furthermore, valuable data are lost when readmission tracking is carried out only to 30-days11,21,22 rather than 90-days.19,20 Though influential factors have been identified, a cohesive predictive model is yet to be reported.
A recent report stated that the most common reasons for readmission were unavoidable.16 However, we hypothesize that the opportunity for improved care exists. Our aim was to develop predictive models for 3-month medical and surgical readmission after elective lumbar surgery, based on a multi-institutional, national spine registry. A predictive model has the potential to inform targeted, actionable interventions that can then be implemented during the preoperative, perioperative, and postoperative phases of care to decrease readmissions.
The Quality and Outcomes Database (QOD) registry is a multicenter, prospective database that enrolls patients undergoing spine surgery from 86 participating centers across 36 US states. Data for the current study were collected from patients undergoing elective lumbar spine surgery for degenerative disease. The QOD registry collects measures of surgical safety, patient-reported outcomes (PROs), and common postoperative hospital metrics. The QOD project is designated as a non-research, clinical quality improvement effort, and base on existing federal guidelines exempted from IRB review. Full enrollment information has been previously described.23
Inclusion and Exclusion Criteria
Patients undergoing elective lumbar spine surgery carried the following diagnoses: stenosis, spondylolisthesis, primary and revision disc herniation, symptomatic mechanical disc collapse, and adjacent segment disease. Exclusion criteria included spinal infection, tumor, traumatic fracture, deformity, recurrent multilevel stenosis, and neurological paralysis due to preexisting spinal disease. Patients less than 18 years and those who were incarcerated were also excluded.
The following variables were collected from the QOD registry through the electronic medical record (EMR): age, sex, body mass index (BMI), ASA grade, history of surgery, history of diabetes, coronary artery disease (CAD), osteoporosis, depression, dominant presenting symptom (back, leg, both), motor deficit, symptom duration (< or >3 mo), preoperative ambulatory status, diagnosis, and surgery-specific variables including approach (anterior or posterior), surgical levels, arthrodesis, or presence/absence of interbody graft. Patient interviews were used to collect the following patient-reported variables: race, smoking status, anxiety, depression, education (less than high school, high school, 2-year college, 4-year college, post-college), employment status, type of occupation (sedentary, light manual labor, medium labor, heavy labor), workers’ compensation, liability insurance, health insurance (uninsured, private insurance, uninsured, Medicare, Medicaid, or Veterans Affairs/government), and baseline Oswestry Disability Index (ODI).
Three-month readmissions were prospectively captured through EMR review and self-reported outcome questionnaires. History and physicals, daily progress notes, laboratory and culture results, and discharge summaries were all included in the EMR review. Readmissions to the primary surgical facility were identified via EMR review and readmissions to outside facilities were recorded via patient-reported questionnaires. Patient visits to the emergency department that did not result in hospital admission were not counted as readmissions.
The following reasons for readmission were considered medical in nature: (1) deep venous thrombosis (DVT)/pulmonary embolism (PE), (2) fever of unknown origin, (3) infections (of unknown causes or from sources other than surgical site), (4) cardiovascular (including cardiogenic chest pain, myocardial infarction, congestive heart failure), (5) gastrointestinal, (6) central nervous system related (delirium, altered mental status, and stroke), (7) renal (urinary retention, acute kidney injury, hematuria and urinary tract infection [UTI]), (8) respiratory (shortness of breath, upper respiratory tract infection, chronic obstructive pulmonary disease [COPD], and pneumonia), (9) diabetes, (10) syncope, (11) drug-related issues (overdose/allergic reactions), and (12) general medical issues (skin rash, electrolyte imbalance, and mental health).
The following reasons were considered surgical in nature: (1) cerebrospinal fluid (CSF) leak/dural tear/pseudomeningocele, (2) disc re-herniation or disc herniation at another level, (3) instrumentation failure, (4) new neurologic deficit, (5) pain, (6) implant revision, (7) revision surgery, (8) hematoma/seroma/cyst at surgical site, (9) surgical site infection (SSI), and (10) wound dehiscence.
Median, interquartile ranges (IQR), mean and standard deviation for continuous variables and frequencies for categorical variables were calculated for prospectively collected data on patient and surgery-specific variables. Distinct multivariable logistic regression models were fitted for surgery-related and medical readmissions. The variables included in the models were age; sex; race; smoking status; employment; number of spinal levels; ASA grade; BMI; diabetes; osteoporosis; CAD; education level; depression; dominant symptom; motor deficit; symptom duration; workers’ compensation; liability insurance for disability; insurance type; ambulatory status; lumbar fusion, anterior, posterior or combined approach; and preoperative ODI score. We assumed smooth relationships for continuous variables and used restricted cubic regression splines with four knots for age and BMI, and three knots for preoperative ODI score. The remaining variables were included as binary or categorical. The effects of predictors on the likelihood of patients experiencing a readmission were reported as odd ratios (OR) and the corresponding confidence intervals (CI) were calculated using Wald statistics. The ORs were based on inter-quartile ranges for continuous variables. The importance of individual predictors on readmission status was determined using Wald chi-square values, from which the predictors’ respective degrees of freedom were subtracted. Ten multiple imputations were performed to handle missing covariates, using predictive mean matching as provided in the aregImpute function in R's Hmisc package.24,25
Internal validation of the model was performed using bootstrap resampling to evaluate the discrimination and calibration of the model. The model discrimination was measured by a concordance index (c-index). A c-index of value 1 suggests perfect discrimination by the model whereas 0.5 suggests random prediction. All statistical analyses were performed using the R statistical program (version 3.1.2), Vienna, Austria.26
A total of 33,674 elective lumbar surgery patients were included in this study, and 2079 (6.15%) reported at least one readmission during the 90-day postoperative period. Eight hundred fifty of these patients had a medical readmission while 1229 had a surgical readmission. The median age for the study cohort was 59 years (IQR 48–69 yr). Those with a medical readmission had a median age of 66 years, while those with a surgery-related readmission had a median age of 60 years. The median BMI for the study cohort was 29.68 (IQR 26.06–34.13). At the time of surgery, 19% (6176) were smokers, 18% (6160) had diabetes, and 11% (3676) had coronary artery disease (CAD). 23% (7703) had back-dominant symptoms and 32% (10,584) had leg-dominant symptoms. 86% (28,312) of patients were ambulatory before surgery. Table 1 summarizes patient-specific and surgery-specific variables for our study cohort. The most common reasons for medical readmission included cardiovascular, renal, and gastrointestinal, as seen in Figure 1. The most common reasons for surgical readmission included SSI, disc herniation, and uncontrollable pain, as seen in Figure 2.
Predictive Model for Medical Readmission
A multivariable logistic regression model was fit with medical readmission as the outcome of interest. The overall likelihood-ratio chi-square statistic was 366.83 (df = 45 and P < 0.001), and the c-index of the model was 0.681. The odds of medical readmission were significantly higher for older patients (P < 0.001), males versus females (P = 0.03), those with higher ASA grade (P < 0.001), those with diabetes (P = 0.04), and those with CAD (P = 0.002). African Americans had a greater odds of medical readmission than that of Caucasians (P = 0.02). The odds of readmission were higher in patients undergoing a fusion (P = 0.003) and in those with involvement of more spinal levels (P < 0.001). Patients undergoing anterior only or anterior–posterior procedures had a higher odds of medical readmission compared with those undergoing posterior only approach (P = 0.007). The odds were higher for patients who were unemployed due to disability, when compared with those who were employed prior to surgery (P = 0.04). Patients with a higher baseline ODI also had a higher odds of medical readmission within 3 months of surgery (P = 0.002). Finally, the odds of medical readmission were lower for patients with a smoking history (P = 0.04). Figure 3 summarizes the adjusted effects of these covariates, and Figure 4 demonstrates the relative importance of each covariate in explaining variation in medical readmissions. The most important drivers of medical readmission within 3 months were age, ASA grade, preoperative ODI score, number of involved surgical levels, fusion, surgical approach, diabetes, CAD, employment status, sex, and smoking status. Figure 5 demonstrates the calibration accuracy for our model. Internal validation of the model, using bootstrap-validation, yielded a corrected c-index of 0.663.
Predictive Model for Surgical Readmission
A separate multivariable logistic regression model was fit with surgical readmission as the outcome of interest. The overall likelihood-ratio chi-square statistic was 238.80 (df = 45 and P < 0.001), and the c-index of the model was 0.623. The odds of surgery-related readmission were significantly greater for patients with a higher BMI (P = 0.0002) and a higher ASA grade (3 vs. 1) (P = 0.0160). Females had a greater odds of surgical readmission compared with males (P = 0.0345), and the odds were greater for African Americans versus Caucasians (P = 0.0260). The odds were also greater for patients with severe depression (P = 0.0331). Moreover, patients with more involved spinal levels had greater odds of surgery-related readmission (P = 0.0082), as did those with anterior-only surgical approaches (P = 0.0261). Patients with 2 years of college had higher odds of readmission than those with less than a high school education (P = 0.0286), however this significant difference was not seen for patients with 4 years of college or post-college experience. Furthermore, patients with higher baseline ODI scores had higher odds of readmission (P < 0.001). Figure 6 summarizes the adjusted effects of these covariates, and Figure 7 demonstrates the relative importance of each covariate in explaining variation in surgical readmissions. The most important drivers of surgical readmission within 3 months are preoperative ODI score, BMI, ASA grade, surgical approach, number of levels, race, history of depression, and patient's sex. Figure 8 demonstrates the calibration accuracy for this model, for which the bootstrap-validated c-index is 0.603.
This study utilized a prospective, national registry to develop and internally validate predictive models for 90-day medical and surgical readmission after elective lumbar spine surgery. In our cohort, 6.15% of patients were readmitted in that time period—2.5% for medical and 3.6% for surgical reasons, respectively. Our multivariable regression models revealed numerous patient- and surgery-related variables that were significantly associated with readmission, and while roughly half of them were shared between both etiologies, many variables were unique to one or the other. This suggests that patients who are readmitted for medical versus surgical reasons may in fact represent distinct clinical populations.
The following five factors were found to be associated with both medical and surgical readmissions: increased number of levels, anterior approach, higher ASA grade, sex, and baseline ODI. Of note, single-institution and administrative database studies have consistently reported ASA grade, operative duration, Medicare/Medicaid, age, and pulmonary disease as significant factors for readmission.17 However, the granularity and scale of the QOD has provided us with additional insights. While men were more likely to be admitted for medical reasons, women were more likely to be admitted for surgical reasons. One possible explanation for this is that men can often carry more (but un-diagnosed) comorbidities than women, while women may be more cautious about wound healing and have a lower threshold to return to the hospital. Literature in other fields of surgery also cite an increased risk of surgical readmission for females—possible explanations include reduced postoperative support at home (due to increased longevity compared with men), and increased anesthesia requirements that contribute to adverse events after surgery.27 The association of increased levels with readmission is understandable, given the longer operative duration and anesthetic time, and anterior lumbar procedures likely incur greater risk due to the manipulation of abdominal and vascular structures. Patients with higher ASA grades are more likely to be readmitted given their increased medical complexity, but the rationale for the influence of baseline ODI is less clear. One possibility, supported by the literature, is that patients with higher preoperative disability may be less likely to mobilize in the immediate postoperative period, and that this tendency makes them more apt to return to the hospital for medical reasons such as DVT.28 Patients who stay on their back and do not mobilize are also less likely to have their wounds monitored regularly, which could increase the risk of wound complications.
Specifically for medical readmissions, we found that age, CAD, diabetes, employment, fusion, and smoking were significant factors. The role of comorbidities such as CAD and diabetes is expected. After all, postoperative hyperglycemia alone is a known risk factor for any post-surgical wound infection, even in those without diabetes.29–31 The link between fusion and medical readmission is also understandable, given that fusion procedures are generally longer and therefore increase the risk of cardiovascular strain and immobility. The importance of employment status also has a likely explanation—patients who are unemployed due to disability are more likely to have severe and debilitating lumbar pathology, leading to more invasive and longer surgeries, as well as prolonged postoperative stays with all the attendant risks. Surprisingly, smoking was a protective factor against medical readmission. This counterintuitive effect has in fact been seen elsewhere in the literature—smoking has been associated with a decreased risk of postoperative infection and sepsis in adult deformity patients.32 One possibility is that smoking is associated with certain confounding variables, not currently tracked in QOD, which happen to reduce the risk of medical readmission.
Unique to surgical readmissions, we found that BMI, depression, and African American race were significant factors. The literature shows that patients with higher BMI often require longer operative time and experience more perioperative complications.33–35 It is likely that obese patients are at increased risk for disc re-herniations, as well as wound healing issues due to the added difficulty of keeping dependent incisional sites aerated and clean.36,37 The link between depression and surgical readmission is more complex. Adogwa et al 38 report a similar association in their analyses, and one possible explanation is that depressed patients are less motivated to care for their wounds or adhere to postoperative activity guidelines (such as the avoidance of bending/lifting twisting). This could increase their risk for surgical site infections and implant failure. We also know that depression and pain are highly correlated, and so depressed patients may be more likely to have a surgical readmission due to inadequate pain control.39 Our model also revealed African American ethnicity as a risk factor for surgical readmission. This is an association that is corroborated by the literature, and may be due to a link with poorer insurance coverage and socioeconomic status, which can lead to readmissions that might otherwise have been prevented with a reliable primary care provider.16 Of note, our modeling suggested that patients with 2-year college degrees were more likely to have surgical readmissions than those without a high school degree. However, those with 4 years of college and those with post-college experience did not have higher odds of surgical readmission, as compared with those with less than a high school education. The inconsistent finding for patients with 2-year college degrees was likely due to the small number of patients in that category. The existing literature links improved education with better outcomes in general.40
Armed with robust predictors generalizable to nearly many different types of spine surgery practices, providers may be able to use the models presented here to improve care. After all, we know that reducing “preventable readmissions” represents an opportunity to improve the value of care.41 While sterility can always be optimized, it is likely that SSI and wound issues will never be completely eliminated. One study of over 100,000 general spine surgery cases reported a 2.1% local infection rate.42 However, medical complications such as pain management and exacerbation of known comorbidities can proactively be addressed. Furthermore, stratification of time to readmission is a growing area of research. A subset analysis of QOD data across eight centers found that 80% of readmissions occurred within the first 4 weeks, and 46% within 14 days.43 Week 1 readmissions were for pain and early wound issues, week 2 included mostly medical complications, and weeks 3 to 4 constituted SSI, disc recurrence, and hardware failures.43 Risk scores can be created using this information, and the models we present here, to predict the probability of readmission.44 Communication and expectations can be optimized prior to and after surgery, as seen in a small study of pediatric patients who received scheduled post-discharge phone calls.45 One can easily imagine a phone call from a surgeon's office 1 week after a lumbar fusion, reassuring the patient and family that intermittent back and leg pain is part of the normal healing process. Interventions such as this, targeted to patients at high risk for readmission based on our models, may significantly improve patient knowledge and simultaneously reduce readmissions.
The current predictive modeling study is not without limitation. A weakness of any predictive modeling study is the uncertainty inherent in the decision-making for the number and type of variables to be included. Though many different types of patient and surgical data were included, it is possible that unaccounted variables may play a significant role in predicting readmission. The reported c-statistics of 0.663 and 0.603 reflects discrimination, but room for improvements exists. Model discrimination on an absolute risk basis was modest because of the low incidence of readmission. The 90th percentile of estimated risk was 0.076 to 0.0623 for medical readmission and 0.016 to 0.068 for surgery-related readmission. However, the large sample size from diverse centers across the country allows these data to be generalizable to many patients, despite the limitations of predictive modeling. Another limitation is that the reasons for readmission were categorized into a single entity, whereas in reality, reasons for readmission can be multi-factorial. This study did not track the lengths of surgeries or estimated blood loss; however, the number of levels involved, and the inclusion of a fusion procedure were accounted for in the regression analysis, which both likely correlate with those variables. Differences in structural pathologies of the degenerative disease may impact a patient's risk for readmission; however, the models in this study did not account for the diagnostic classification. Moreover, due to the size and reach of the QOD database, readmission practice was not standardized, and heterogeneity in individual hospital and surgeon practice may vary. Another limitation is that we were able to confirm (via EMR review) the veracity of, and identify the reason for, readmissions when they occurred at the QOD sites themselves, but not when they occurred outside those sites.
Here we present internally validated predictive models for medical and surgical readmission after elective lumbar spine surgery. Factors predictive of any readmission included increasing levels, anterior approach, higher ASA grade, sex, and baseline ODI. Factors specific to medical readmission included increasing age, CAD, diabetes, lack of employment, fusion, and smoking, while factors specific to surgical readmission included increasing BMI, depression, and African American race. These findings set the stage for targeted interventions that have the potential to reduce unnecessary readmissions, and also suggest that medical and surgical readmissions be treated as distinct clinical events.Key PointsHere we aim to develop predictive models for 3-month medical and surgical readmission after elective lumbar surgery, based on a multi-institutional, national spine registry.This study retrospectively analyzes prospectively collected data. The QOD was queried for patients undergoing elective lumbar surgery for degenerative diseases.In this study we present internally validated predictive models for medical and surgical readmission after elective lumbar spine surgery.These findings set the stage for targeted interventions with a potential to reduce unnecessary readmissions, and also suggest that medical and surgical readmissions be treated as distinct clinical events.
1. Deyo RA, Mirza SK, Martin BI, et al. Trends, major medical complications, and charges associated with surgery for lumbar spinal stenosis in older adults. JAMA
2. Chen E, Tong KB, Laouri M. Surgical treatment patterns among Medicare beneficiaries newly diagnosed with lumbar spinal stenosis. Spine J
3. Martin BI, Deyo RA, Mirza SK, et al. Expenditures and health status among adults with back and neck problems. JAMA
5. Martin BI, Mirza SK, Comstock BA, et al. Reoperation rates following lumbar spine surgery and the influence of spinal fusion procedures. Spine (Phila Pa 1976)
6. Rajaee SS, Kanim LE, Bae HW. National trends in revision spinal fusion in the USA: patient characteristics and complications. Bone Joint J
7. Ong KL, Auerbach JD, Lau E, et al. Perioperative outcomes, complications, and costs associated with lumbar spinal fusion in older patients with spinal stenosis and spondylolisthesis. Neurosurg Focus
8. McCormack RA, Hunter T, Ramos N, et al. An analysis of causes of readmission after spine surgery. Spine (Phila Pa 1976)
9. Papadopoulos EC, Girardi FP, Sandhu HS, et al. Outcome of revision discectomies following recurrent lumbar disc herniation. Spine (Phila Pa 1976)
10. Chotai S, Parker SL, Sivaganesan A, et al. Effect of complications within 90 days on patient-reported outcomes 3 months and 12 months following elective surgery for lumbar degenerative disease. Neurosurg Focus
11. Kim BD, Smith TR, Lim S, et al. Predictors of unplanned readmission in patients undergoing lumbar decompression: multi-institutional analysis of 7016 patients. J Neurosurg Spine
12. Mancuso CA, Duculan R, Stal M, et al. Patients’ expectations of cervical spine surgery. Spine (Phila Pa 1976)
13. Medicare Payment Advisory Commission (U.S.). Report to the Congress: Medicare Payment Policy. 2017; March 2017.
14. Davis MA, Onega T, Weeks WB, et al. Where the United States spends its spine dollars: expenditures on different ambulatory services for the management of back and neck conditions. Spine (Phila Pa 1976)
15. Weiss AJ (Truven Health Analytics), Elixhauser AA, Andrews RM (AHRQ). Characteristics of Operating Room Procedures in U.S. Hospitals, 2011. HCUP Statistical Brief #170; 2014. Agency for Healthcare Research and Quality, Rockville, MD.
16. Pugely AJ, Martin CT, Gao Y, et al. Causes and risk factors for 30-day unplanned readmissions
after lumbar spine surgery. Spine (Phila Pa 1976)
17. Bernatz JT, Anderson PA. Thirty-day readmission rates in spine surgery: systematic review and meta-analysis. Neurosurg Focus
18. Modhia U, Takemoto S, Braid-Forbes MJ, et al. Readmission rates after decompression surgery in patients with lumbar spinal stenosis among Medicare beneficiaries. Spine (Phila Pa 1976)
19. Akamnonu C, Cheriyan T, Goldstein JA, et al. Ninety-day readmissions
after degenerative cervical spine surgery: a single-center administrative database study. Int J Spine Surg
20. Akamnonu C, Cheriyan T, Goldstein JA, et al. Unplanned hospital readmission after surgical treatment of common lumbar pathologies: rates and causes. Spine (Phila Pa 1976)
21. McGirt MJ, Sivaganesan A, Asher AL, et al. Prediction model for outcome after low-back surgery: individualized likelihood of complication, hospital readmission, return to work, and 12-month improvement in functional disability. Neurosurg Focus
22. Lovecchio F, Hsu WK, Smith TR, et al. Predictors of thirty-day readmission after anterior cervical fusion. Spine (Phila Pa 1976)
23. Asher AL, McGirt MJ, Glassman SD, et al. Regulatory considerations for prospective patient care registries: lessons learned from the National Neurosurgery Quality and Outcomes Database. Neurosurg Focus
24. Harrell FE Jr, Harrell MFE Jr. Package ‘Hmisc’. Vienna, Austria: R Foundation for Statistical Computing; 2017.
25. Harrell FE Jr. Regression Modeling Strategies: With Applications to Linear Models, Logistic and Ordinal Regression, and Survival Analysis. New York City, New York: Springer; 2015.
26. Team RC. R: A language and environment for statistical computing [Internet], 2014. Vienna, Austria; 2017.
27. Chen SL, Kuo IJ, Kabutey NK, et al. Perioperative risk factors for hospital readmission after elective endovascular aortic aneurysm repair. J Vasc Surg
28. Tominaga H, Setoguchi T, Tanabe F, et al. Risk factors for venous thromboembolism after spine surgery. Medicine (Baltimore)
29. Richards JE, Hutchinson J, Mukherjee K, et al. Stress hyperglycemia and surgical site infection in stable nondiabetic adults with orthopedic injuries. J Trauma Acute Care Surg
30. Richards JE, Kauffmann RM, Zuckerman SL, et al. Relationship of hyperglycemia and surgical-site infection in orthopaedic surgery. J Bone Joint Surg Am
31. Stauber MN, Aberer F, Oulhaj A, et al. Early hyperglycemia after initiation of glucocorticoid therapy predicts adverse outcome in patients with acute graft-versus-host disease. Biol Blood Marrow Transplant
32. Zuckerman SL, Lakomkin N, Stannard BP, et al. Incidence and predictive factors of sepsis following adult spinal deformity surgery. Neurosurgery
2017; doi: 10.1093/neuros/nyx578.
33. Kalanithi PA, Arrigo R, Boakye M. Morbid obesity increases cost and complication rates in spinal arthrodesis. Spine (Phila Pa 1976)
34. Burks CA, Werner BC, Yang S, et al. Obesity is associated with an increased rate of incidental durotomy in lumbar spine surgery. Spine (Phila Pa 1976)
35. Yagi M, Patel R, Boachie-Adjei O. Complications and unfavorable clinical outcomes in obese and overweight patients treated for adult lumbar or thoracolumbar scoliosis with combined anterior/posterior surgery. J Spinal Disord Tech
36. Jackson KL, Devine JG. The effects of obesity on spine surgery: a systematic review of the literature. Global Spine J
37. Drazin D, Ugiliweneza B, Al-Khouja L, et al. Treatment of recurrent disc herniation: a systematic review. Cureus
38. Adogwa O, Elsamadicy AA, Mehta AI, et al. Association between baseline affective disorders and 30-day readmission rates in patients undergoing elective spine surgery. World Neurosurg
39. Amaral V, Marchi L, Martim H, et al. Influence of psychosocial distress in the results of elective lumbar spine surgery. J Spine Surg
40. Chotai S, Devin CJ, Archer KR, et al. Effect of patients’ functional status on satisfaction with outcomes 12 months after elective spine surgery for lumbar degenerative disease. Spine J
41. Dawes AJ, Sacks GD, Russell MM, et al. Preventable readmissions
to surgical services: lessons learned and targets for improvement. J Am Coll Surg
42. Smith JS, Shaffrey CI, Sansur CA, et al. Rates of infection after spine surgery based on 108,419 procedures: a report from the Scoliosis Research Society Morbidity and Mortality Committee. Spine (Phila Pa 1976)
43. Asher. The Institute for Healthcare Improvement – NeuroPoint Alliance Cooperative Quality Improvement Project: Using National Registries to Design Continuous Quality Improvement Protocols; 2018. AANS/CNS Spine Section, Orlando, FL.
44. Burke RE, Schnipper JL, Williams MV, et al. The HOSPITAL score predicts potentially preventable 30-day readmissions
in conditions targeted by the Hospital Readmissions
Reduction Program. Med Care
45. Flippo R, NeSmith E, Stark N, et al. Reduction of 30-day preventable pediatric readmission rates with postdischarge phone calls utilizing a patient- and family-centered care approach. J Pediatr Health Care