Advertisement

Predictive models to assess risk of extended length of stay in adults with spinal deformity and lumbar degenerative pathology: development and internal validation

Open AccessPublished:October 28, 2022DOI:https://doi.org/10.1016/j.spinee.2022.10.009

      Abstract

      BACKGROUND CONTEXT

      Postoperative recovery after adult spinal deformity (ASD) operations is arduous, fraught with complications, and often requires extended hospital stays. A need exists for a method to rapidly predict patients at risk for extended length of stay (eLOS) in the preoperative setting.

      PURPOSE

      To develop a machine learning model to preoperatively estimate the likelihood of eLOS following elective multi-level lumbar/thoracolumbar spinal instrumented fusions (≥3 segments) for ASD.

      STUDY DESIGN/SETTING

      Retrospectively from a state-level inpatient database hosted by the Health care cost and Utilization Project.

      PATIENT SAMPLE

      Of 8,866 patients of age ≥50 with ASD undergoing elective lumbar or thoracolumbar multilevel instrumented fusions.

      OUTCOME MEASURES

      The primary outcome was eLOS (>7 days).

      METHODS

      Predictive variables consisted of demographics, comorbidities, and operative information. Significant variables from univariate and multivariate analyses were used to develop a logistic regression-based predictive model that use six predictors. Model accuracy was assessed through area under the curve (AUC), sensitivity, and specificity.

      RESULTS

      Of 8,866 patients met inclusion criteria. A saturated logistic model with all significant variables from multivariate analysis was developed (AUC=0.77), followed by generation of a simplified logistic model through stepwise logistic regression (AUC=0.76). Peak AUC was reached with inclusion of six selected predictors (combined anterior and posterior approach, surgery to both lumbar and thoracic regions, ≥8 level fusion, malnutrition, congestive heart failure, and academic institution). A cutoff of 0.18 for eLOS yielded a sensitivity of 77% and specificity of 68%.

      CONCLUSIONS

      This predictive model can facilitate identification of adults at risk for eLOS following elective multilevel lumbar/thoracolumbar spinal instrumented fusions for ASD. With a fair diagnostic accuracy, the predictive calculator will ideally enable clinicians to improve preoperative planning, guide patient expectations, enable optimization of modifiable risk factors, facilitate appropriate discharge planning, stratify financial risk, and accurately identify patients who may represent high-cost outliers. Future prospective studies that validate this risk assessment tool on external datasets would be valuable.

      Keywords

      Introduction

      Adult spinal deformity (ASD) is a highly prevalent condition with a definitive negative impact on health-related quality of life [
      • Ames CP
      • Scheer JK
      • Lafage V
      • Smith JS
      • Bess S
      • Berven SH
      • et al.
      Adult spinal deformity: epidemiology, health impact, evaluation, and management.
      ,
      • Safaee MM
      • Scheer JK
      • Ailon T
      • Smith JS
      • Hart RA
      • Burton DC
      • et al.
      Predictive modeling of length of hospital stay following adult spinal deformity correction: analysis of 653 patients with an accuracy of 75% within 2 days.
      ]. Such patients have greater functional limitations and pain than patients with other chronic conditions, even when compared with age-matched controls [
      • Pellisé F
      • Vila-Casademunt A
      • Ferrer M
      • Domingo-Sàbat M
      • Bagó J
      • Pérez-Grueso FJ
      • et al.
      Impact on health related quality of life of adult spinal deformity (ASD) compared with other chronic conditions.
      ,
      • Theis J
      • Gerdhem P
      • Abbott A.
      Quality of life outcomes in surgically treated adult scoliosis patients: a systematic review.
      ]. When expectant management and physical therapy fails to provide relief, elective surgical intervention can be pursued [
      • Terran J
      • McHugh BJ
      • Fischer CR
      • Lonner B
      • Warren D
      • Glassman S
      • et al.
      Surgical treatment for adult spinal deformity: projected cost effectiveness at 5-year follow-up.
      ]. Although surgical intervention can provide considerable benefit, postoperative recovery after ASD operations is arduous, fraught with complications, and often requires extended hospital stays, and rehabilitation [
      • Campbell PG
      • Yadla S
      • Nasser R
      • Malone J
      • Maltenfort MG
      • Ratliff JK.
      Patient comorbidity score predicting the incidence of perioperative complications: assessing the impact of comorbidities on complications in spine surgery.
      ,
      • Reis RC
      • de Oliveira MF
      • Rotta JM
      • Botelho RV
      Risk of complications in spine surgery: a prospective study.
      ,
      • Uribe JS
      • Deukmedjian AR
      • Mummaneni PV
      • Fu KM
      • Mundis Jr, GM
      • Okonkwo DO
      • et al.
      Complications in adult spinal deformity surgery: an analysis of minimally invasive, hybrid, and open surgical techniques.
      ].
      Measurement of eLOS can serve as a composite reflection of the postoperative course for ASD patients, with extended stay associated with increased risk of hospital-acquired infections, medical complications, and readmissions [
      • Barnett AG
      • Page K
      • Campbell M
      • Martin E
      • Rashleigh-Rolls R
      • Halton K
      • et al.
      The increased risks of death and extra lengths of hospital and ICU stay from hospital-acquired bloodstream infections: a case-control study.
      ,
      • Hauck K
      • Zhao X
      How dangerous is a day in hospital? A model of adverse events and length of stay for medical inpatients.
      ,
      • Kim RB
      • Wilkerson C
      • Karsy M
      • Joyce E
      • Rolston JD
      • Couldwell WT
      • et al.
      Prolonged length of stay and risk of unplanned 30-day readmission after elective spine surgery: propensity score-matched analysis of 33,840 patients.
      ]. The resulting retention of patients in postacute care settings can result in significant administrative challenges due to the ensuing disruption of patient flow and bed shortages, limiting access to care [
      • Toh HJ
      • Lim ZY
      • Yap P
      • Tang T.
      Factors associated with prolonged length of stay in older patients.
      ]. Furthermore, eLOS has also been identified as one of the top predictors of catastrophic costs, of over $ 100,000 for ASD patients [
      • Ames CP
      • Smith JS
      • Gum JL
      • Kelly M
      • Vila-Casademunt A
      • Burton DC
      • et al.
      Utilization of predictive modeling to determine episode of care costs and to accurately identify catastrophic cost nonwarranty outlier patients in adult spinal deformity surgery: a step toward bundled payments and risk sharing.
      ]. Hence, there exists a significant interest in predicting which patients will have eLOS following surgery for ASD.
      Although numerous studies have identified independent risk factors associated with eLOS, few are specific to ASD patients, major discrepancies exist on importance of selected risk factors [
      • Horn SR
      • Passias PG
      • Bortz CA
      • Pierce KE
      • Lafage V
      • Lafage R
      • et al.
      Predicting extended operative time and length of inpatient stay in cervical deformity corrective surgery.
      ,
      • Joshi RS
      • Lau D
      • Haddad AF
      • Deviren V
      • Ames CP
      Risk factors for determining length of intensive care unit and hospital stays following correction of cervical deformity: evaluation of early severe adverse events.
      ,
      • Lovecchio F
      • Steinhaus M
      • Elysee JC
      • Huang A
      • Ang B
      • Lafage R
      • et al.
      Factors associated with short length of stay after long fusions for adult spinal deformity: initial steps toward developing an enhanced recovery pathway.
      ]. Additionally, assessment of risk through a combination of many significant variables, each with respective odds ratio or relative risk, can make preoperative evaluation difficult. Therefore, a need exists for a method to rapidly predict patients at risk for eLOS in the preoperative setting. Thus, the goal of this study is to develop a machine learning model to preoperatively estimate the likelihood of eLOS for patients with elective multilevel lumbar/thoracolumbar spinal instrumented fusions for spinal deformity.

      Methods

      Source of data

      Data were acquired retrospectively on ASD patients from state-level inpatient database hosted by the Health care cost and Utilization Project [

      HCUP State Inpatient Databases (SID). Healthcare Cost and Utilization Project (HCUP). 2005-2013. Agency for Healthcare Research and Quality R, MD. www.hcup-us.ahrq.gov/sidoverview.jsp. Accessed July 14, 2022.

      ]. Data were derived from both public and private health care institutions in the states of California, Florida, Nebraska, New York, North Carolina, and Utah from the period of 2005–2013. No patient identifiers were gathered throughout data collection. Data included patient demographic variables, comorbidities, operative information, and LOS measured in days for each patient.

      Participants, sample size, and missing data

      Inclusion criteria were ASD patients with age ≥50 years undergoing elective multilevel spine fusions (3 levels) to the lumbar or thoracolumbar regions (Fig. 1). Cases of malignancy, trauma, or infection were excluded. Patients with unknown LOS, discharge against medical advice, or missing data were also excluded. Inclusion and exclusion criteria were applied by utilization of International Classification of Diseases, Volumes 9 codes (ICD-9) [

      World Health Organization. (2004). ICD-9: international statistical classification of diseases and related health problems: ninth revision neWHO.

      ].

      Predictors and outcomes

      The primary outcome was whether a patient had eLOS, defined as >7 days. Preoperative variables consisted of demographics, insurance status, comorbidities, and operative variables. Demographics included age allocated into ranges (50–59, 60–69, 70–79, 80+), sex (male, female), race/ethnicity (White, Hispanic, Black, Asian, Native American/Other), and health care institution type (Academic vs. Non-Academic). Insurance type was designated as either public (Medicare/Medicaid), private (Commercial), or other (Self-Pay/No Insurance). Comorbidities included Charlson's Comorbidity Index (CCI: 1, 2, 3, 4), as well as the individual comorbid conditions used to calculate the CCI [
      • Charlson ME
      • Pompei P
      • Ales KL
      • MacKenzie CR
      A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.
      ]. Substance abuse variables were captured in terms of smoking history, alcohol abuse, and drug abuse. Mental health variables consisting of anxiety and depression were also acquired. Operative variables included whether the procedure was a revision (yes/no), surgical approach (posterior alone vs. anterior and posterior), surgical region (lumbar vs. lumbar and thoracic), and number of levels instrumented/fused (3–7 levels vs. ≥8 levels).

      Statistical analysis

      Chi-Square and Fisher's exact tests were used for univariate analysis to determine the association between predictor variables and an eLOS, with generation of odds ratios and corresponding 95% confidence intervals (CI), and p-values. The p-values <.05 were considered statistically significant. Multivariate analyses were then conducted on all significant variables through binary logistic regression.

      Development and validation of predictive models

      In development of machine learning models, the cohorts were separated into 80% derivation and 20% validation groups. For the derivation group, all significant variables from multivariate analysis were used to develop a saturated logistic regression model. The model was then tested on the validation group to predict the probability of eLOS, with generation of Area Under the Receiver Operating Curve (AUC) and corresponding 95% CIs to assess model performance.
      Following development of the saturated logistic model, a simplified model with retention of diagnostic accuracy was the aim. Least absolute and selection operator (LASSO) was used to identify the most important variables in the saturated predictive model. Variables with the highest LASSO coefficients were then sequentially entered into a stepwise logistic model, in order of highest magnitude LASSO coefficient. The corresponding model AUC was calculated for each additional variable added, and inclusion of additional variables was stopped when AUC failed to increase by more than 0.5%.
      To create a clinically applicable tool with estimation of predictive probability of eLOS, beta coefficients for the simplified logistic model were determined. In addition to AUC, model characteristics such as sensitivity, specificity, positive predictive value, and negative predictive value were calculated at varying thresholds.

      Software/tools used

      MATLAB version 2020b was used to conduct all statistical analyses and predictive model development [

      Matlab [Computer Software]. Version 2020b. Natick MM. 2020.

      ].

      Results

      Participants

      Inclusion criteria was met for 8,866 patients, 22.5% (n=1,994) of whom had an eLOS. (Table 1). The median age for eLOS patients was 66 years (Q1–Q3: 60–74), compared with a median of 68 years (Q1–Q3: 61–74) for patients with non-eLOS. Male patients consisted of 31.9% of the cohort. Most operations involved only the lumbar spine (81.4%) with utilization of a posterior approach (78.6%). A substantial number of patients had a CCI≥4 (36.9%) with all individuals in the study having a CCI score ≥1. Common comorbidities were HTN (65.5%), COPD (23.0%), and hypothyroidism (19.0%). Of the patients with eLOS, 58.2% were discharged to a postacute care facility and 41.8% were discharged to home, whereas the patients with non-eLOS, 40.8% were discharged to a postacute care facility and 59.2% were discharged to home (discharge location p<.001).
      Table 1Baseline data of patients with respect to length of stay following operations for multi-level lumbar instrumented fusions
      VariableEntire cohortLength of stay 7 d (%)Length of stay>7 d (%)p
      N%N%N%
      Population8,866(100.0%)6,872(77.5%)1994(22.5%)
      Age - Median (Q1, Q3)68(61, 74)68(61, 74)66(60, 74)<.001
       50–591,847(20.8%)1,362(73.7%)485(26.3%)
       60–693,259(36.8%)2,506(76.9%)753(23.1%)
       70–792,909(32.8%)2,326(80.0%)583(20.0%)
       ≥80851(9.6%)678(79.7%)173(20.3%)
      Patient sex<.001
       Male2,832(31.9%)2,262(79.9%)570(20.1%)
       Female6,034(68.1%)4,610(76.4%)1,424(23.6%)
      Race.008
       White7,544(85.1%)5,848(77.5%)1,696(22.5%)
       Asian420(4.7%)308(73.3%)112(26.7%)
       Black211(2.4%)153(72.5%)58(27.5%)
       Hispanic99(1.1%)83(83.8%)16(16.2%)
       Native American or Multiracial592(6.7%)480(81.1%)112(18.9%)
      Surgical approach
       Posterior6,970(78.6%)5,755(82.6%)1,215(17.4%)<.001
       Anterior and posterior (Combined)1,896(21.4%)1,117(58.9%)779(41.1%)
      Region of surgery<.001
       Lumbar only7,218(81.4%)5,921(82.0%)1,297(18.0%)
       Lumbar and thoracic1,648(18.6%)951(57.7%)697(42.3%)
      Revision surgery1,983(22.4%)1,428(72.0%)555(28.0%)<.001
      Vertebral levels<.001
       3–7 levels7,841(88.4%)6,303(80.4%)1,538(19.6%)
       ≥8 levels1,025(11.6%)569(55.5%)456(44.5%)
      Institutional type<.001
       Nonacademic6,864(82.9%)5,385(78.5%)1,479(21.5%)
       Academic1,418(17.1%)980(69.1%)438(30.9%)
      Insurance type<.001
       Public5,727(64.6%)4,512(78.8%)1,215(21.2%)
       Private2,596(29.3%)1,993(76.8%)603(23.2%)
       Other543(6.1%)367(67.6%)176(32.4%)
      Charlson's Comorbidity Index (CCI).009
       CCI (1)1,066(12.0%)800(75.0%)266(25.0%)
       CCI (2)2,073(23.4%)1,616(78.0%)457(22.0%)
       CCI (3)2,458(27.7%)1,955(79.5%)503(20.5%)
       CCI (≥4)3,269(36.9%)2,501(76.5%)768(23.5%)
      Comorbidities
       COPD2,037(23.0%)1,521(74.7%)516(25.3%)<.001
       CHF491(5.5%)309(62.9%)182(37.1%)<.001
       Hemiplagia/Paraplegia201(2.3%)135(67.2%)66(32.8%)<.001
       Past myocardial infarction566(6.4%)420(74.2%)146(25.8%).052
       Renal disease439(5.0%)302(68.8%)137(31.2%)<.001
       Rheumatic disease636(7.2%)480(75.5%)156(24.5%).201
       Hypertension5,809(65.5%)4,546(78.3%)1,263(21.7%).020
       Malnutrition162(1.8%)85(52.5%)77(47.5%)<.001
       Coronary artery disease1,532(17.3%)1,184(77.3%)348(22.7%).817
       Hypothyroidism1,686(19.0%)1,300(77.1%)386(22.9%).195
       Osteoporosis1,344(15.2%)980(72.9%)364(27.1%)<.001
       Diabetes (DMII)
       No DMII7,282(82.1%)5,623(77.2%)1,659(22.8%)<.001
       Controlled DMII1,470(16.6%)1,161(79.0%)309(21.0%)
       Uncontrolled DMII114(1.3%)88(77.2%)26(22.8%)
      Substance abuse
       Smoking history2,778(31.3%)2,164(77.9%)614(22.1%).350
       Alcohol abuse207(2.3%)139(67.1%)68(32.9%).003
       Drug abuse264(3.0%)172(65.2%)92(34.8%)<.001
      Mental health
       Anxiety1,062(12.0%)796(75.0%)266(25.0%).033
       Depression2,056(23.2%)1,538(74.8%)518(25.2%).008
      Discharge disposition



      <.001
       Home4,904(55.3%)4,070(83.0%)834(17.0%)
       Postacute care facility3,962(44.7%)2,802(70.7%)1,160(29.3%)
      COPD, chronic obstructive pulmonary disease; CHF, congestive heart failure; MI, myocardial infarction.

      Univariate and multivariate analyses

      Results from univariate and multivariate analysis are displayed in Table 2. Preoperative variables significantly associated with increased likelihood of eLOS in the multivariate analysis included: combined anterior and posterior surgical approach (OR=3.59, 95% CI: 3.19–4.04, p<.001), surgery at both lumbar and thoracic regions (OR=2.49, 95% CI: 2.15–2.89, p<.001), ≥8 level fusion (OR=1.83, 95% CI: 1.54–2.17, p<.001), academic institution (OR=1.56, 95% CI: 1.36-1.79, p<.001), self-pay/no insurance status (OR=1.62, 95% CI: 1.30-2.03, p<.001), congestive heart failure (OR=2.09, 95% CI: 1.69–2.58, p<.001), hemiplegia/paraplegia (OR=1.43, 95% CI: 1.03-1.99, p=.034), malnutrition (OR=2.39, 95% CI: 1.70–3.37, p<.001), alcohol abuse (OR=1.75, 95% CI: 1.27–2.42, p=.001), and drug abuse (OR=1.41, 95% CI: 1.05–1.88, p=.021). Only one variable, male sex (OR=0.88, 95% CI: 0.78–1.00, p=.045), was significantly associated with a decreased likelihood of eLOS.
      Table 2Univariate and multivariate analyses
      VariableUnivariate testsMultivariate tests
      OR95% CIpOR95% CIp
      Age (Continuous)
       50–59Ref
       60–690.840.74–0.96.0120.980.84–1.14.814
       70–790.700.61–0.81<.0010.940.78–1.14.541
       ≥800.720.59–0.87.0011.080.85–1.38.518
      Patient sex
       FemaleRef
       Male0.820.73–0.91<.0010.880.78–1.00.045
      Race
       WhiteRef
       Hispanic1.251.00–1.57.0481.271.00–1.61.054
       Black1.310.96–1.78.095
       Asian0.660.39–1.14.146
       Native American/Other1.130.82–1.56.447
      Surgical approach
       PosteriorRef
       Anterior and posterior (Combined)3.302.95 - 3.68<.0013.593.19– 4.04<.001
      Region of surgery
       Lumbar onlyRef
       Lumbar and thoracic3.352.98 - 3.75<.0012.492.15–2.89<.001
      Revision surgery1.471.31 - 1.65<.0011.0390.91–1.18.559
      Vertebral levels
       3–7 LevelsRef
       ≥8 Levels3.282.87–3.76<.0011.831.54–2.17<.001
      Institutional type
       NonacademicRef
       Academic1.631.43–1.85<.0011.561.36–1.79<.001
      Insurance type
       PublicRef
       Private1.781.47–2.15<.0011.000.86–1.15.956
       Other1.121.01–1.26.0421.621.30– 2.03<.001
      Charlson's Comorbidity Index (CCI)
       CCI (1)Ref
       CCI (2)0.850.72–1.01.073
       CCI (3)0.770.65–0.92.0030.960.84–1.09.507
       CCI (≥4)0.920.79–1.08.341
      Comorbidities
       COPD1.231.09–1.38.0011.070.94–1.21.318
       CHF2.131.76–2.58<.0012.091.69–2.58<0.001
       Hemiplegia/Paraplegia1.711.27–2.30.0011.431.03–1.99.034
       Past MI1.211.00–1.48.054
       Renal disease1.601.30–1.98.0001.611.28–2.04<.001
       Rheumatic disease1.130.94–1.36.200
       Hypertension0.880.80–0.98.0210.930.83–1.04.212
       Malnutrition3.212.35–4.38<.0012.391.70–3.37<.001
       Coronary artery disease1.020.89–1.16.814
       Hypothyroidism1.030.91–1.17.674
       Osteoporosis1.341.18–1.53<.0011.191.03–1.39.021
      Diabetes (DMII)
       No DMIIRef
       Controlled DMII0.900.79–1.04.151
       Uncontrolled DMII1.000.65–1.561.000
      Substance abuse
       Smoking history0.970.87–1.08.565
       Alcohol abuse1.711.27–2.30.0011.751.27–2.42.001
       Drug abuse1.881.46–2.44<.0011.411.05–1.88.021
      Mental health
       Anxiety1.181.01–1.36.0340.860.72–1.01.073
       Depression1.221.08–1.37.0011.030.90–1.17.705
      COPD, chronic obstructive pulmonary disease; CHF, congestive heart failure; MI, myocardial infarction.

      Saturated model development

      Data from 80% of patients the cohort (n=7,093) were used to train the machine learning models with validation on 20% (n=1,773). All significant variables from multivariate analysis were used in the development of the saturated logistic regression predictive model (AUC=0.77, 95% CI: 0.74–0.80). The ROC is displayed in Fig. 2.
      Fig 2
      Fig. 2Receiver Operating Curve (ROC) for logistic regression predictive model for extended length of stay. The AUC was 0.76 (95% CI: 0.73-0.79).

      Model specification: simplified predictive model

      LASSO regression identified seven variables as relevant to the predictive model (most to least important): combined anterior and posterior approach, surgery to both lumbar and thoracic regions, ≥8 level fusion, malnutrition, CHF, academic institution, and renal disease. Stepwise logistic regression, with generation of corresponding AUC for each added variables, is shown in Fig. 3. Peak AUC of 0.76 (95% CI: 0.73–0.79) was reached with six of the seven variables. The addition of renal disease increased AUC by only 0.2% and was hence excluded from the model. For use as a predictive calculator to predict eLOS likelihood, beta coefficients for the simplified logistic model were determined (Supplementary Table 1). For each component of the simplified model, ORs and 95% CIs for each model component were also derived (Table 3).
      Fig 3
      Fig. 3Stepwise Logistic Regression. Each curve represents a logistic predictive model using one additional variable. For example, the black curve represents a predictive model only using surgical approach (combined anterior and posterior), whereas the light-blue curve represents model using surgical approach (combined anterior and posterior), surgical region (lumbar and thoracic), and number of interspaces instrumented/fused (8+).
      Table 3Characteristics of final logistic regression model for extended length of stay
      Logistic model componentOR95% CIp
      Surgical approach (combined anterior and posterior)3.372.96–3.84<.001
      Surgical region (Lumbar+Thoracic)2.422.06–2.84<.001
      # Interspaces (8+)1.781.47–2.16<.001
      Malnutrition2.721.87–3.95<.001
      Congestive heart failure2.001.59–2.52<.001
      Academic facility1.561.34–1.81<.001

      Model performance

      Performance characteristics such as sensitivity, specificity, positive predictive value, and negative predictive value are shown in Table 4 for each corresponding predictive probability threshold. At a threshold of 0.18, the simplified model produced a sensitivity of 0.77 and a specificity of 0.68. The cutoff threshold can be adjusted based on the acceptable risk tolerance of the health care team.
      Table 4Predictive model characteristics depending on threshold level
      ThresholdSensitivitySpecificityPositive predictive valueNegative predictive value
      0.140.820.590.370.92
      0.160.820.590.370.92
      0.180.770.680.420.91
      0.200.740.690.420.90
      0.220.700.720.430.89
      0.240.700.720.430.89
      0.260.620.770.450.87
      0.280.610.780.450.87
      0.300.610.780.460.87
      0.320.390.890.510.83
      0.340.380.900.530.83

      Discussion

      Model interpretation

      The aim of this study was to develop a machine learning model to preoperatively predict eLOS in patients undergoing elective multilevel lumbar or thoracolumbar instrumented fusion for a diagnosis of ASD. The model created used six essential preoperative patient variables (combined anterior and posterior approaches, thoracic+lumbar regions, >8 instrumented/fused levels, malnutrition, congestive heart failure, academic institution) and produced a diagnostic AUC of 0.76, with a sensitivity of 77% and specificity of 68% at a selected threshold of 0.18. Given that the saturated logistic model utilizing all significant variables had an AUC of 0.77, the goal was met in creation of a simplified preoperative model that kept diagnostic accuracy. With the provided beta coefficients, clinicians can easily used the model preoperatively within the clinical setting to facilitate rapid risk assessment.
      Overall, the associations determined between significant predictive variables and eLOS agree with the literature. For example, male sex has been widely associated with reduced LOS and lower readmission rates in patients undergoing surgery for lumbar degenerative pathology [
      • Ilyas H
      • Golubovsky JL
      • Chen J
      • Winkelman RD
      • Mroz TE
      • Steinmetz MP
      Risk factors for 90-day reoperation and readmission after lumbar surgery for lumbar spinal stenosis.
      ,
      • Zeidan M
      • Goz V
      • Lakomkin N
      • Spina N
      • Brodke DS
      • Spiker WR
      Predictors of readmission and prolonged length of stay after cervical disc arthroplasty.
      ]. Combined anterior and posterior approaches, surgery to both lumbar and thoracic regions, and a greater number of fused interspaces (≥8) have been proven to increase chances postoperative infection risk, medical complications, and LOS in ASD patients [
      • Blumberg TJ
      • Woelber E
      • Bellabarba C
      • Bransford R
      • Spina N
      Predictors of increased cost and length of stay in the treatment of postoperative spine surgical site infection.
      ,
      • Cizik AM
      • Lee MJ
      • Martin BI
      • Bransford RJ
      • Bellabarba C
      • Chapman JR
      • et al.
      Using the spine surgical invasiveness index to identify risk of surgical site infection: a multivariate analysis.
      ,
      • Le HV
      • Wick JB
      • Lafage R
      • Kelly MP
      • Kim HJ
      • Gupta MC
      • et al.
      Surgical factors and treatment severity for perioperative complications predict hospital length of stay in adult spinal deformity surgery.
      ,
      • Schwartz DA
      • Hui X
      • Schneider EB
      • Ali MT
      • Canner JK
      • Leeper WR
      • et al.
      Worse outcomes among uninsured general surgery patients: does the need for an emergency operation explain these disparities?.
      ]. Certain comorbidities, such as CHF, renal disease, malnutrition, and osteoporosis, have been well documented as significant mortality risks following elective spine surgery [
      • Lee MJ
      • Konodi MA
      • Cizik AM
      • Bransford RJ
      • Bellabarba C
      • Chapman JR.
      Risk factors for medical complication after spine surgery: a multivariate analysis of 1,591 patients.
      ,
      • Soroceanu A
      • Burton DC
      • Oren JH
      • Smith JS
      • Hostin R
      • Shaffrey CI
      • et al.
      Medical complications after adult spinal deformity surgery: incidence, risk factors, and clinical impact.
      ,
      • Elsamadicy AA
      • Adogwa O
      • Vuong VD
      • Sergesketter A
      • Reddy G
      • Cheng J
      • et al.
      Impact of alcohol use on 30-day complication and readmission rates after elective spinal fusion (>/=2 levels) for adult spine deformity: a single institutional study of 1,010 patients.
      ,
      • Han L
      • Han H
      • Liu H
      • Wang C
      • Wei X
      • He J
      • et al.
      Alcohol abuse and alcohol withdrawal are associated with adverse perioperative outcomes following elective spine fusion surgery.
      ]. Although the association between alcohol use disorder and poor perioperative outcome is mixed for ASD patients, our current findings confirm that alcohol abuse disorder could be a significant risk factor [
      • Elsamadicy AA
      • Adogwa O
      • Vuong VD
      • Sergesketter A
      • Reddy G
      • Cheng J
      • et al.
      Impact of alcohol use on 30-day complication and readmission rates after elective spinal fusion (>/=2 levels) for adult spine deformity: a single institutional study of 1,010 patients.
      ,
      • Han L
      • Han H
      • Liu H
      • Wang C
      • Wei X
      • He J
      • et al.
      Alcohol abuse and alcohol withdrawal are associated with adverse perioperative outcomes following elective spine fusion surgery.
      ].
      Although prior literature is sparse, a limited number of studies have derived predictive models for eLOS in ASD patients from large patient databases, such as the National Surgical Improvement Program Database and NSQIP databases. Such studies have produced fairly accurate AUCs, typically between 0.65 and 0.80 [
      • Etzel CM
      • Veeramani A
      • Zhang AS
      • McDonald CL
      • DiSilvestro KJ
      • Cohen EM
      • et al.
      Supervised machine learning for predicting length of stay after lumbar arthrodesis: a comprehensive artificial intelligence approach.
      ,

      Zhang AS, Veeramani A, Quinn MS, Alsoof D, Kuris EO, Daniels AH. Machine 2021;10(18)doi:10.3390/jcm10184074

      ]. Current predictive calculators, including the ACS NSQIP Surgical Risk Calculator, have provided an applicable tool that can estimate LOS based on over 20 patient variables [
      • Basques BA
      • Fu MC
      • Buerba RA
      • Bohl DD
      • Golinvaux NS
      • Grauer JN.
      Using the ACS-NSQIP to identify factors affecting hospital length of stay after elective posterior lumbar fusion.
      ]. However, no study has conducted stratification or exclusion patients with trauma, malignancy, or infection – cases which represent vastly different patient and complication profiles when compared with elective surgery [
      • Watanabe M
      • Sakai D
      • Matsuyama D
      • Yamamoto Y
      • Sato M
      • Mochida J.
      Risk factors for surgical site infection following spine surgery: efficacy of intraoperative saline irrigation.
      ]. Moreover, tools such as the ACS NSQIP calculator provide no option for users to specify patient diagnosis and are not specific to spine patients. Additionally, studies which have captured more comprehensive and granular information on patient comorbidities are often based on a single institution and lack broader applicability [
      • Safaee MM
      • Scheer JK
      • Ailon T
      • Smith JS
      • Hart RA
      • Burton DC
      • et al.
      Predictive modeling of length of hospital stay following adult spinal deformity correction: analysis of 653 patients with an accuracy of 75% within 2 days.
      ,
      • Lubelski D
      • Ehresman J
      • Feghali J
      • Tanenbaum J
      • Bydon A
      • Theodore N
      • et al.
      Prediction calculator for nonroutine discharge and length of stay after spine surgery.
      ,
      • Figueroa RL
      • Zeng-Treitler Q
      • Kandula S
      • Ngo LH.
      Predicting sample size required for classification performance.
      ].

      Implications

      A critical implication of the preoperative predictive model is the potential to predict financial risk. The relation between eLOS and costs of care is significant and represents a driving factor behind the need for accurate preoperative assessment. For example, for spine deformity patients, a single additional day in the hospital can incur over $ 10,000 in insurance charges and over $5,000 in hospital costs, accompanied with significant associated financial risks of returning to the operating room within 90 days [
      • Boylan MR
      • Riesgo AM
      • Chu A
      • Paulino CB
      • Feldman DS
      Costs and complications of increased length of stay following adolescent idiopathic scoliosis surgery.
      ]. Moreover, the direct cost per day in the hospital is significantly greater than that of a postacute care rehabilitation facility [
      • Elsamadicy AA
      • Koo AB
      • Kundishora AJ
      • Chouairi F
      • Lee M
      • Hengartner AC
      • et al.
      Impact of patient and hospital-level risk factors on extended length of stay following spinal fusion for adolescent idiopathic scoliosis.
      ,
      • Theologis AA
      • Lau D
      • Dalle-Ore C
      • Tsu A
      • Deviren V
      • Ames CP.
      Costs and utility of post-discharge acute inpatient rehabilitation following adult spinal deformity surgery.
      ]. Of note, whereas a portion of patients require an eLOS due need for management of perioperative complications, a substantial amount of eLOS patients reside longer in the hospital due to delays in the discharge transfer process to a rehabilitation or skilled nursing facility [
      • New PW
      • Andrianopoulos N
      • Cameron PA
      • Olver JH
      • Stoelwinder JU.
      Reducing the length of stay for acute hospital patients needing admission into inpatient rehabilitation: a multicentre study of process barriers.
      ]. Thus, such patients represent cost outliers as they incur high costs associated with both eLOS and rehabilitation.
      Under reimbursement models such as Bundled Payments for Care Improvement Initiative (BPCI), where reimbursement is fixed for the duration of care, eLOS can cause catastrophic financial loss and inability for the hospital to sustain surgical spine care [
      • Dietz N
      • Sharma M
      • Alhourani A
      • Ugiliweneza B
      • Wang D
      • Nuño MA
      • et al.
      Bundled payment models in spine surgery: current challenges and opportunities, a systematic review.
      ,
      • Miller HD.
      From volume to value: better ways to pay for health care.
      ,
      • Siddiqi A
      • White PB
      • Murphy W
      • Terry D
      • Murphy SB
      • Talmo CT.
      Cost savings in a surgeon-directed BPCI program for total joint arthroplasty.
      ]. Ensuring financial viability of elective surgeries is important to ensure that hospital systems can continue to operate. Therefore, a key utility of the predictive calculator derived in this study is to stratify financial risk and accurately identify patients who may represent high-cost outliers.
      Usage of the predictive calculator to predict patients at risk of eLOS may also aid in alleviating hospital bed shortages and therefore improve patient access to care [
      • Toh HJ
      • Lim ZY
      • Yap P
      • Tang T.
      Factors associated with prolonged length of stay in older patients.
      ]. Administrative teams may establish an acceptable predictive model risk tolerance depending on the hospital's typical space availability. The predictive calculator threshold can be changed depending on the most recent occupancy of the postacute care unit and the acceptable risk tolerance as determined by the health care team. With the predictive calculator in hand, clinicians can ensure greater transparency with patients, better manage postoperative expectations, and have additional tools in the shared-decision making process on the risks and benefits of surgery [
      • O'Donnell FT
      Preoperative Evaluation of the Surgical Patient.
      ].

      Strengths and limitations

      A key strength of this study is that it uses a large cohort size with patients from multiple health care institutions in different states within the United States whereas retaining a sufficient granularity of patient information. The usage of large sample size in training the models is critical for robust machine learning and predictive model development [
      • Figueroa RL
      • Zeng-Treitler Q
      • Kandula S
      • Ngo LH.
      Predicting sample size required for classification performance.
      ]. Moreover, application of inclusion criteria to focus only on elective procedures for ASD provides a reasonable control against confounding conditions, which few prior studies have done. The predictive model only used six variables, with retention of diagnostic accuracy when compared with the fully saturated model, and is preferable to other models that require every single feature of the patient's risk profile. The resulting predictive calculator, with corresponding beta coefficients, can be easily applied in the clinical setting to rapidly facilitate preoperative identification of adult patients at risk for eLOS following spinal deformity surgery. Future prospective studies that validate the risk assessment tool on an external dataset would be valuable. As additional data become available, the relative contribution of each variable to the prediction of eLOS can be modified for improved accuracy.
      Key limitations of this study include the lack of additional variables that could influence the likelihood of eLOS. For example, social variables such as education level, income, and at-home support have been widely associated with postoperative outcomes in spine surgery [
      • Adogwa O
      • Elsamadicy AA
      • Vuong VD
      • Mehta AI
      • Vasquez RA
      • Cheng J
      • et al.
      Effect of social support and marital status on perceived surgical effectiveness and 30-day hospital readmission.
      ,
      • Yap ZL
      • Summers SJ
      • Grant AR
      • Moseley GL
      • Karran EL.
      The role of the social determinants of health in outcomes of surgery for low back pain. A systematic review and narrative synthesis.
      ]. Further, patients on high-dose narcotics preoperatively often require longer recovery times and were not able to be identified and assessed within this study [
      • Hardy N
      • Zeba F
      • Ovalle A
      • Yanac A
      • Nzugang-Noutonsi C
      • Abadier M
      • et al.
      Association of prescription opioid use on mortality and hospital length of stay in the intensive care unit.
      ]. Furthermore, whereas no studies have reported the effect of postoperative pain on eLOS in patients with ASD, the patient's subjective readiness to be discharged from the hospital could possibly influence the length of hospital stay. Future studies could use pain scores and document narcotic use to determine any association with eLOS.
      Delays in the referral process to a postoperative care facility due to administrative barriers can also result in eLOS [
      • Lubelski D
      • Ehresman J
      • Feghali J
      • Tanenbaum J
      • Bydon A
      • Theodore N
      • et al.
      Prediction calculator for nonroutine discharge and length of stay after spine surgery.
      ]. Unfortunately, identification of patients with a prolonged discharge referral process was not feasible in this study, as the dataset does not allow one to assess and/or differentiate between reasons for an eLOS (ie, secondary to additional postoperative management needed versus complication versus waiting for transfer to rehabilitation/SNF). Of note, eLOS and discharge disposition were interdependent outcomes, as discharge location was significant for eLOS on univariate analysis. However, discharge disposition was not factored into the predictive model development because the primary goal of this study was to use only variables available preoperatively.
      We also acknowledge that because the composition of our cohort is dependent on the accuracy of the ICD codes queried, it is possible that patients with purely degenerative pathology were treated in this cohort. Also unavailable were information on the prevalence of individual diagnoses and granular information on etiologies of the ASD patients that comprised our cohort. Furthermore, we realize that information on the patient's condition before surgery, such as ambulatory status, neurological function, preoperative narcotic usage, and preoperative living situation, can influence eLOS, and were not used in this study. Although more granular data with additional risk factors may be attained from a single institution, building a predictive model from one institution with a more limited patient cohort size would likely be overly specific to that singular location and lack broader generalizability. The predictive model we have presented is the first to use high-volume patient data from multiple health care institutions for ASD patients specifically, and hence represents a foundational tool that can be improved in the future as more granular patient data become available.

      Conclusion

      In this study of 8,866 ASD patients, a predictive calculator was created that can facilitate preoperative identification of adults at risk for eLOS following elective multilevel lumbar/thoracolumbar spinal instrumented fusions for ASD. The predictive calculator built used six essential preoperative predictors (surgical approach, surgical region, levels fused, malnutrition, congestive heart failure, and institution type). With a diagnostic AUC of 0.76, this predictive calculator will ideally enable clinicians to improve preoperative planning, guide patient expectations, enable optimization of modifiable risk factors, facilitate appropriate discharge planning, stratify financial risk, and accurately identify patients who may represent high-cost outliers. Future prospective studies that validate this risk assessment tool on external datasets would be valuable.

      Appendix. Supplementary materials

      References

        • Ames CP
        • Scheer JK
        • Lafage V
        • Smith JS
        • Bess S
        • Berven SH
        • et al.
        Adult spinal deformity: epidemiology, health impact, evaluation, and management.
        Spine Deform. 2016; 4: 310-322https://doi.org/10.1016/j.jspd.2015.12.009
        • Safaee MM
        • Scheer JK
        • Ailon T
        • Smith JS
        • Hart RA
        • Burton DC
        • et al.
        Predictive modeling of length of hospital stay following adult spinal deformity correction: analysis of 653 patients with an accuracy of 75% within 2 days.
        World Neurosurg. 2018; 115: e422-e427https://doi.org/10.1016/j.wneu.2018.04.064
        • Pellisé F
        • Vila-Casademunt A
        • Ferrer M
        • Domingo-Sàbat M
        • Bagó J
        • Pérez-Grueso FJ
        • et al.
        Impact on health related quality of life of adult spinal deformity (ASD) compared with other chronic conditions.
        Eur Spine J. 2015; 24: 3-11https://doi.org/10.1007/s00586-014-3542-1
        • Theis J
        • Gerdhem P
        • Abbott A.
        Quality of life outcomes in surgically treated adult scoliosis patients: a systematic review.
        Eur Spine J. 2015; 24: 1343-1355https://doi.org/10.1007/s00586-014-3593-3
        • Terran J
        • McHugh BJ
        • Fischer CR
        • Lonner B
        • Warren D
        • Glassman S
        • et al.
        Surgical treatment for adult spinal deformity: projected cost effectiveness at 5-year follow-up.
        Ochsner J. Spring. 2014; 14: 14-22
        • Campbell PG
        • Yadla S
        • Nasser R
        • Malone J
        • Maltenfort MG
        • Ratliff JK.
        Patient comorbidity score predicting the incidence of perioperative complications: assessing the impact of comorbidities on complications in spine surgery.
        J Neurosurg Spine. 2012; 16: 37-43https://doi.org/10.3171/2011.9.SPINE11283
        • Reis RC
        • de Oliveira MF
        • Rotta JM
        • Botelho RV
        Risk of complications in spine surgery: a prospective study.
        Open Orthop J. 2015; 9: 20-25https://doi.org/10.2174/1874325001509010020
        • Uribe JS
        • Deukmedjian AR
        • Mummaneni PV
        • Fu KM
        • Mundis Jr, GM
        • Okonkwo DO
        • et al.
        Complications in adult spinal deformity surgery: an analysis of minimally invasive, hybrid, and open surgical techniques.
        Neurosurg Focus. 2014; 36: E15https://doi.org/10.3171/2014.3.FOCUS13534
        • Barnett AG
        • Page K
        • Campbell M
        • Martin E
        • Rashleigh-Rolls R
        • Halton K
        • et al.
        The increased risks of death and extra lengths of hospital and ICU stay from hospital-acquired bloodstream infections: a case-control study.
        BMJ Open. 2013; 3e003587https://doi.org/10.1136/bmjopen-2013-003587
        • Hauck K
        • Zhao X
        How dangerous is a day in hospital? A model of adverse events and length of stay for medical inpatients.
        Med Care. 2011; 49: 1068-1075https://doi.org/10.1097/MLR.0b013e31822efb09
        • Kim RB
        • Wilkerson C
        • Karsy M
        • Joyce E
        • Rolston JD
        • Couldwell WT
        • et al.
        Prolonged length of stay and risk of unplanned 30-day readmission after elective spine surgery: propensity score-matched analysis of 33,840 patients.
        Spine (Phila Pa 1976). 2020; 45: 1260-1268https://doi.org/10.1097/BRS.0000000000003520
        • Toh HJ
        • Lim ZY
        • Yap P
        • Tang T.
        Factors associated with prolonged length of stay in older patients.
        Singapore Med J. 2017; 58: 134-138https://doi.org/10.11622/smedj.2016158
        • Ames CP
        • Smith JS
        • Gum JL
        • Kelly M
        • Vila-Casademunt A
        • Burton DC
        • et al.
        Utilization of predictive modeling to determine episode of care costs and to accurately identify catastrophic cost nonwarranty outlier patients in adult spinal deformity surgery: a step toward bundled payments and risk sharing.
        Spine (Phila Pa 1976). 2020; 45: E252-E265https://doi.org/10.1097/BRS.0000000000003242
        • Horn SR
        • Passias PG
        • Bortz CA
        • Pierce KE
        • Lafage V
        • Lafage R
        • et al.
        Predicting extended operative time and length of inpatient stay in cervical deformity corrective surgery.
        J Clin Neurosci. 2019; 69: 206-213https://doi.org/10.1016/j.jocn.2019.07.064
        • Joshi RS
        • Lau D
        • Haddad AF
        • Deviren V
        • Ames CP
        Risk factors for determining length of intensive care unit and hospital stays following correction of cervical deformity: evaluation of early severe adverse events.
        J Neurosurg Spine. 2020; : 1-12https://doi.org/10.3171/2020.6.SPINE20826
        • Lovecchio F
        • Steinhaus M
        • Elysee JC
        • Huang A
        • Ang B
        • Lafage R
        • et al.
        Factors associated with short length of stay after long fusions for adult spinal deformity: initial steps toward developing an enhanced recovery pathway.
        Global Spine J. 2021; 11: 866-873https://doi.org/10.1177/2192568220941448
      1. HCUP State Inpatient Databases (SID). Healthcare Cost and Utilization Project (HCUP). 2005-2013. Agency for Healthcare Research and Quality R, MD. www.hcup-us.ahrq.gov/sidoverview.jsp. Accessed July 14, 2022.

      2. World Health Organization. (2004). ICD-9: international statistical classification of diseases and related health problems: ninth revision neWHO.

        • Charlson ME
        • Pompei P
        • Ales KL
        • MacKenzie CR
        A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.
        J Chronic Dis. 1987; 40: 373-383https://doi.org/10.1016/0021-9681(87)90171-8
      3. Matlab [Computer Software]. Version 2020b. Natick MM. 2020.

        • Ilyas H
        • Golubovsky JL
        • Chen J
        • Winkelman RD
        • Mroz TE
        • Steinmetz MP
        Risk factors for 90-day reoperation and readmission after lumbar surgery for lumbar spinal stenosis.
        J Neurosurg Spine. 2019; 31: 20-26https://doi.org/10.3171/2019.1.SPINE18878
        • Zeidan M
        • Goz V
        • Lakomkin N
        • Spina N
        • Brodke DS
        • Spiker WR
        Predictors of readmission and prolonged length of stay after cervical disc arthroplasty.
        Spine (Phila Pa 1976). 2021; 46: 487-491https://doi.org/10.1097/BRS.0000000000003839
        • Blumberg TJ
        • Woelber E
        • Bellabarba C
        • Bransford R
        • Spina N
        Predictors of increased cost and length of stay in the treatment of postoperative spine surgical site infection.
        Spine J. 2018; 18: 300-306https://doi.org/10.1016/j.spinee.2017.07.173
        • Cizik AM
        • Lee MJ
        • Martin BI
        • Bransford RJ
        • Bellabarba C
        • Chapman JR
        • 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: 335-342https://doi.org/10.2106/JBJS.J.01084
        • Le HV
        • Wick JB
        • Lafage R
        • Kelly MP
        • Kim HJ
        • Gupta MC
        • et al.
        Surgical factors and treatment severity for perioperative complications predict hospital length of stay in adult spinal deformity surgery.
        Spine (Phila Pa 1976). 2022; 47: 136-143https://doi.org/10.1097/BRS.0000000000004122
        • Schwartz DA
        • Hui X
        • Schneider EB
        • Ali MT
        • Canner JK
        • Leeper WR
        • et al.
        Worse outcomes among uninsured general surgery patients: does the need for an emergency operation explain these disparities?.
        Surgery. 2014; 156: 345-351https://doi.org/10.1016/j.surg.2014.04.039
        • Lee MJ
        • Konodi MA
        • Cizik AM
        • Bransford RJ
        • Bellabarba C
        • Chapman JR.
        Risk factors for medical complication after spine surgery: a multivariate analysis of 1,591 patients.
        Spine J. 2012; 12: 197-206https://doi.org/10.1016/j.spinee.2011.11.008
        • Soroceanu A
        • Burton DC
        • Oren JH
        • Smith JS
        • Hostin R
        • Shaffrey CI
        • et al.
        Medical complications after adult spinal deformity surgery: incidence, risk factors, and clinical impact.
        Spine (Phila Pa 1976). 2016; 41: 1718-1723https://doi.org/10.1097/BRS.0000000000001636
        • Elsamadicy AA
        • Adogwa O
        • Vuong VD
        • Sergesketter A
        • Reddy G
        • Cheng J
        • et al.
        Impact of alcohol use on 30-day complication and readmission rates after elective spinal fusion (>/=2 levels) for adult spine deformity: a single institutional study of 1,010 patients.
        J Spine Surg. 2017; 3: 403-410https://doi.org/10.21037/jss.2017.08.12
        • Han L
        • Han H
        • Liu H
        • Wang C
        • Wei X
        • He J
        • et al.
        Alcohol abuse and alcohol withdrawal are associated with adverse perioperative outcomes following elective spine fusion surgery.
        Spine (Phila Pa 1976). 2021; 46: 588-595https://doi.org/10.1097/BRS.0000000000003868
        • Etzel CM
        • Veeramani A
        • Zhang AS
        • McDonald CL
        • DiSilvestro KJ
        • Cohen EM
        • et al.
        Supervised machine learning for predicting length of stay after lumbar arthrodesis: a comprehensive artificial intelligence approach.
        J Am Acad Orthop Surg. 2021; https://doi.org/10.5435/JAAOS-D-21-00241
      4. Zhang AS, Veeramani A, Quinn MS, Alsoof D, Kuris EO, Daniels AH. Machine 2021;10(18)doi:10.3390/jcm10184074

        • Basques BA
        • Fu MC
        • Buerba RA
        • Bohl DD
        • Golinvaux NS
        • Grauer JN.
        Using the ACS-NSQIP to identify factors affecting hospital length of stay after elective posterior lumbar fusion.
        Spine (Phila Pa 1976). 2014; 39: 497-502https://doi.org/10.1097/BRS.0000000000000184
        • Watanabe M
        • Sakai D
        • Matsuyama D
        • Yamamoto Y
        • Sato M
        • Mochida J.
        Risk factors for surgical site infection following spine surgery: efficacy of intraoperative saline irrigation.
        J Neurosurg Spine. 2010; 12: 540-546https://doi.org/10.3171/2009.11.SPINE09308
        • Lubelski D
        • Ehresman J
        • Feghali J
        • Tanenbaum J
        • Bydon A
        • Theodore N
        • et al.
        Prediction calculator for nonroutine discharge and length of stay after spine surgery.
        Spine J. 2020; 20: 1154-1158https://doi.org/10.1016/j.spinee.2020.02.022
        • Figueroa RL
        • Zeng-Treitler Q
        • Kandula S
        • Ngo LH.
        Predicting sample size required for classification performance.
        BMC Med Inform Decis Mak. 2012; 12: 8https://doi.org/10.1186/1472-6947-12-8
        • Boylan MR
        • Riesgo AM
        • Chu A
        • Paulino CB
        • Feldman DS
        Costs and complications of increased length of stay following adolescent idiopathic scoliosis surgery.
        J Pediatr Orthop B. 2019; 28: 27-31https://doi.org/10.1097/BPB.0000000000000543
        • Elsamadicy AA
        • Koo AB
        • Kundishora AJ
        • Chouairi F
        • Lee M
        • Hengartner AC
        • et al.
        Impact of patient and hospital-level risk factors on extended length of stay following spinal fusion for adolescent idiopathic scoliosis.
        J Neurosurg Pediatr. 2019; : 1-7https://doi.org/10.3171/2019.5.PEDS19161
        • Theologis AA
        • Lau D
        • Dalle-Ore C
        • Tsu A
        • Deviren V
        • Ames CP.
        Costs and utility of post-discharge acute inpatient rehabilitation following adult spinal deformity surgery.
        Spine Deform. 2021; 9: 817-822https://doi.org/10.1007/s43390-020-00251-w
        • New PW
        • Andrianopoulos N
        • Cameron PA
        • Olver JH
        • Stoelwinder JU.
        Reducing the length of stay for acute hospital patients needing admission into inpatient rehabilitation: a multicentre study of process barriers.
        Intern Med J. 2013; 43: 1005-1011https://doi.org/10.1111/imj.12227
        • Dietz N
        • Sharma M
        • Alhourani A
        • Ugiliweneza B
        • Wang D
        • Nuño MA
        • et al.
        Bundled payment models in spine surgery: current challenges and opportunities, a systematic review.
        World Neurosurg. 2019; 123: 177-183https://doi.org/10.1016/j.wneu.2018.12.001
        • Miller HD.
        From volume to value: better ways to pay for health care.
        Health Aff (Millwood). 2009; 28: 1418-1428https://doi.org/10.1377/hlthaff.28.5.1418
        • Siddiqi A
        • White PB
        • Murphy W
        • Terry D
        • Murphy SB
        • Talmo CT.
        Cost savings in a surgeon-directed BPCI program for total joint arthroplasty.
        Surg Technol Int. 2018; 33: 319-325
        • O'Donnell FT
        Preoperative Evaluation of the Surgical Patient.
        Mo Med. 2016; 113: 196-201
        • Adogwa O
        • Elsamadicy AA
        • Vuong VD
        • Mehta AI
        • Vasquez RA
        • Cheng J
        • et al.
        Effect of social support and marital status on perceived surgical effectiveness and 30-day hospital readmission.
        Global Spine J. 2017; 7: 774-779https://doi.org/10.1177/2192568217696696
        • Yap ZL
        • Summers SJ
        • Grant AR
        • Moseley GL
        • Karran EL.
        The role of the social determinants of health in outcomes of surgery for low back pain. A systematic review and narrative synthesis.
        Spine J. 2021; https://doi.org/10.1016/j.spinee.2021.11.013
        • Hardy N
        • Zeba F
        • Ovalle A
        • Yanac A
        • Nzugang-Noutonsi C
        • Abadier M
        • et al.
        Association of prescription opioid use on mortality and hospital length of stay in the intensive care unit.
        PLoS One. 2021; 16 (PMID: 33886667; PMCID: PMC8061930)e0250320https://doi.org/10.1371/journal.pone.0250320