Search Results

You are looking at 1 - 10 of 49 items for

  • Author or Editor: Eric Klineberg x
  • All content x
Clear All Modify Search
Restricted access

Michael P. Kelly, Michael A. Kallen, Christopher I. Shaffrey, Justin S. Smith, Douglas C. Burton, Christopher P. Ames, Virginie Lafage, Frank J. Schwab, Han Jo Kim, Eric O. Klineberg, Shay Bess, and the International Spine Study Group

OBJECTIVE

After using PROsetta Stone crosswalk tables to calculate Patient-Reported Outcomes Measurement Information System (PROMIS) Physical Function (PF) and Pain Interference (PI) scores, the authors sought to examine 1) correlations with Scoliosis Research Society–22r (SRS-22r) scores, 2) responsiveness to change, and 3) the relationship between baseline scores and 2-year follow-up scores in adult spinal deformity (ASD).

METHODS

PROsetta Stone crosswalk tables were used to converted SF-36 scores to PROMIS scores for pain and physical function in a cohort of ASD patients with 2-year follow-up. Spearman correlations were used to evaluate the relationship of PROMIS scores with SRS-22r scores. Effect size (ES) and adjusted standardized response mean (aSRM) were used to assess responsiveness to change. Linear regression was used to evaluate the association between baseline scores and 2-year follow-up scores.

RESULTS

In total, 425 (425/625, 68%) patients met inclusion criteria. Strong correlations (all |r| > 0.7, p < 0.001) were found between baseline and 2-year PROMIS values and corresponding SRS-22r domain scores. PROMIS-PI showed a large ES (1.09) and aSRM (0.88), indicating good responsiveness to change. PROMIS-PF showed a moderate ES (0.52) and moderate aSRM (0.69), indicating a moderate responsiveness to change. Patients with greater baseline pain complaints were associated with greater pain improvement at 2 years for both SRS-22r Pain (B = 0.39, p < 0.001) and PROMIS-PI (B = 0.45, p < 0.001). Higher functional scores at baseline were associated with greater average improvements in both SRS-22r Activity (B = 0.62, p < 0.001) and PROMIS-PF (B = 0.40, p < 0.001).

CONCLUSIONS

The authors found strong correlations between the SRS-22r Pain and Activity domains with corresponding PROMIS-PI and -PF scores. Pain measurements showed similar and strong ES and aSRM while the function measurements showed similar, moderate ES and aSRM at 2-year follow-up. These data support further exploration of the use of PROMIS–computer adaptive test instruments in ASD.

Full access

Blake N. Staub, Renaud Lafage, Han Jo Kim, Christopher I. Shaffrey, Gregory M. Mundis Jr., Richard Hostin, Douglas Burton, Lawrence Lenke, Munish C. Gupta, Christopher Ames, Eric Klineberg, Shay Bess, Frank Schwab, Virginie Lafage, and the International Spine Study Group

OBJECTIVE

Numerous studies have attempted to delineate the normative value for T1S−CL (T1 slope minus cervical lordosis) as a marker for both cervical deformity and a goal for correction similar to how PI-LL (pelvic incidence–lumbar lordosis) mismatch informs decision making in thoracolumbar adult spinal deformity (ASD). The goal of this study was to define the relationship between T1 slope (T1S) and cervical lordosis (CL).

METHODS

This is a retrospective review of a prospective database. Surgical ASD cases were initially analyzed. Analysis across the sagittal parameters was performed. Linear regression analysis based on T1S was used to provide a clinically applicable equation to predict CL. Findings were validated using the postoperative alignment of the ASD patients. Further validation was then performed using a second, normative database. The range of normal alignment associated with horizontal gaze was derived from a multilinear regression on data from asymptomatic patients.

RESULTS

A total of 103 patients (mean age 54.7 years) were included. Analysis revealed a strong correlation between T1S and C0–7 lordosis (r = 0.886), C2–7 lordosis (r = 0.815), and C0–2 lordosis (r = 0.732). There was no significant correlation between T1S and T1S−CL. Linear regression analysis revealed that T1S−CL assumed a constant value of 16.5° (R2 = 0.664, standard error 2°). These findings were validated on the postoperative imaging (mean absolute error [MAE] 5.9°). The equation was then applied to the normative database (MAE 6.7° controlling for McGregor slope [MGS] between −5° and 15°). A multilinear regression between C2–7, T1S, and MGS demonstrated a range of T1S−CL between 14.5° and 26.5° was necessary to maintain horizontal gaze.

CONCLUSIONS

Normative CL can be predicted via the formula CL = T1S − 16.5° ± 2°. This implies a threshold of deformity and aids in providing a goal for surgical correction. Just as pelvic incidence (PI) can be used to determine the ideal LL, T1S can be used to predict ideal CL. This formula also implies that a kyphotic cervical alignment is to be expected for individuals with a T1S < 16.5°.

Free access

Christopher P. Ames, Justin S. Smith, Justin K. Scheer, Christopher I. Shaffrey, Virginie Lafage, Vedat Deviren, Bertrand Moal, Themistocles Protopsaltis, Praveen V. Mummaneni, Gregory M. Mundis Jr., Richard Hostin, Eric Klineberg, Douglas C. Burton, Robert Hart, Shay Bess, Frank J. Schwab, and the International Spine Study Group

Object

Cervical spine osteotomies are powerful techniques to correct rigid cervical spine deformity. Many variations exist, however, and there is no current standardized system with which to describe and classify cervical osteotomies. This complicates the ability to compare outcomes across procedures and studies. The authors' objective was to establish a universal nomenclature for cervical spine osteotomies to provide a common language among spine surgeons.

Methods

A proposed nomenclature with 7 anatomical grades of increasing extent of bone/soft tissue resection and destabilization was designed. The highest grade of resection is termed the major osteotomy, and an approach modifier is used to denote the surgical approach(es), including anterior (A), posterior (P), anterior-posterior (AP), posterior-anterior (PA), anterior-posterior-anterior (APA), and posterior-anterior-posterior (PAP). For cases in which multiple grades of osteotomies were performed, the highest grade is termed the major osteotomy, and lower-grade osteotomies are termed minor osteotomies. The nomenclature was evaluated by 11 reviewers through 25 different radiographic clinical cases. The review was performed twice, separated by a minimum 1-week interval. Reliability was assessed using Fleiss kappa coefficients.

Results

The average intrarater reliability was classified as “almost perfect agreement” for the major osteotomy (0.89 [range 0.60–1.00]) and approach modifier (0.99 [0.95–1.00]); it was classified as “moderate agreement” for the minor osteotomy (0.73 [range 0.41–1.00]). The average interrater reliability for the 2 readings was the following: major osteotomy, 0.87 (“almost perfect agreement”); approach modifier, 0.99 (“almost perfect agreement”); and minor osteotomy, 0.55 (“moderate agreement”). Analysis of only major osteotomy plus approach modifier yielded a classification that was “almost perfect” with an average intrarater reliability of 0.90 (0.63–1.00) and an interrater reliability of 0.88 and 0.86 for the two reviews.

Conclusions

The proposed cervical spine osteotomy nomenclature provides the surgeon with a simple, standard description of the various cervical osteotomies. The reliability analysis demonstrated that this system is consistent and directly applicable. Future work will evaluate the relationship between this system and health-related quality of life metrics.

Full access

Michael P. Kelly, Lukas P. Zebala, Han Jo Kim, Daniel M. Sciubba, Justin S. Smith, Christopher I. Shaffrey, Shay Bess, Eric Klineberg, Gregory Mundis Jr., Douglas Burton, Robert Hart, Alex Soroceanu, Frank Schwab, Virginie Lafage, and International Spine Study Group

OBJECT

The goal of this study was to examine the effectiveness of preoperative autologous blood donation (PABD) in adult spinal deformity (ASD) surgery.

METHODS

Patients undergoing single-stay ASD reconstructions were identified in a multicenter database. Patients were divided into groups according to PABD (either PABD or NoPABD). Propensity weighting was used to create matched cohorts of PABD and NoPABD patients. Allogeneic (ALLO) exposure, autologous (AUTO) wastage (unused AUTO), and complication rates were compared between groups.

RESULTS

Four hundred twenty-eight patients were identified as meeting eligibility criteria. Sixty patients were treated with PABD, of whom 50 were matched to 50 patients who were not treated with PABD (NoPABD). Nearly one-third of patients in the PABD group (18/60, 30%) did not receive any autologous transfusion and donated blood was wasted. In 6 of these cases (6/60, 10%), patients received ALLO blood transfusions without AUTO. In 9 cases (9/60, 15%), patients received ALLO and AUTO blood transfusions. Overall rates of transfusion of any type were similar between groups (PABD 70% [42/60], NoPABD 75% [275/368], p = 0.438). Major and minor in-hospital complications were similar between groups (Major PABD 10% [6/60], NoPABD 12% [43/368], p = 0.537; Minor PABD 30% [18/60], NoPABD 24% [87/368], p = 0.499). When controlling for potential confounders, PABD patients were more likely to receive some transfusion (OR 15.1, 95% CI 2.1-106.7). No relationship between PABD and ALLO blood exposure was observed, however, refuting the concept that PABD is protective against ALLO blood exposure. In the matched cohorts, PABD patients were more likely to sustain a major perioperative cardiac complication (PABD 8/50 [16%], NoPABD 1/50 [2%], p = 0.046). No differences in rates of infection or wound-healing complications were observed between cohorts.

CONCLUSIONS

Preoperative autologous blood donation was associated with a higher probability of perioperative transfusions of any type in patients with ASD. No protective effect of PABD against ALLO blood exposure was observed, and no risk of perioperative infectious complications was observed in patients exposed to ALLO blood only. The benefit of PABD in patients with ASD remains undefined.

Full access

Shayan Fakurnejad, Justin K. Scheer, Virginie Lafage, Justin S. Smith, Vedat Deviren, Richard Hostin, Gregory M. Mundis Jr., Douglas C. Burton, Eric Klineberg, Munish Gupta, Khaled Kebaish, Christopher I. Shaffrey, Shay Bess, Frank Schwab, Christopher P. Ames, and The International Spine Study Group

OBJECT

Three-column osteotomies (3COs) are technically challenging techniques for correcting severe rigid spinal deformities. The impact of these interventions on outcomes reaching minimum clinically important difference (MCID) or substantial clinical benefit (SCB) is unclear. The objective of this study was to determine the rates of MCID and SCB in standard health-related quality of life (HRQOL) measures after 3COs in patients with adult spinal deformity (ASD). The impacts of location of the uppermost instrumented vertebra (UIV) on clinical outcomes and of maintenance on sagittal correction at 2 years postoperatively were also examined.

METHODS

The authors conducted a retrospective multicenter analysis of the records from adult patients who underwent 3CO with complete 2-year radiographic and clinical follow-ups. Cases were categorized according to established radiographic thresholds for pelvic tilt (> 22°), sagittal vertical axis (> 4.7 cm), and the mismatch between pelvic incidence and lumbar lordosis (> 11°). The cases were also analyzed on the basis of a UIV in the upper thoracic (T1–6) or thoracolumbar (T9–L1) region. Patient-reported outcome measures evaluated preoperatively and 2 years postoperatively included Oswestry Disability Index (ODI) scores, the Physical Component Summary and Mental Component Summary (MCS) scores of the 36-Item Short Form Health Survey, and Scoliosis Research Society-22 questionnaire (SRS-22) scores. The percentages of patients whose outcomes for these measures met MCID and SCB were compared among the groups.

RESULTS

Data from 140 patients (101 women and 39 men) were included in the analysis; the average patient age was 57.3 ± 12.4 years (range 20–82 years). Of these patients, 94 had undergone only pedicle subtraction osteotomy (PSO) and 42 only vertebral column resection (VCR); 113 patients had a UIV in the upper thoracic (n = 63) orthoracolumbar region (n = 50). On average, 2 years postoperatively the patients had significantly improved in all HRQOL measures except the MCS score. For the entire patient cohort, the improvements ranged from 57.6% for the SRS-22 pain score MCID to 24.4% for the ODI score SCB. For patients undergoing PSO or VCR, the likelihood of their outcomes reaching MCID or SCB ranged from 24.3% to 62.3% and from 16.2% to 47.8%, respectively. The SRS-22 self-image score of patients who had a UIV in the upper thoracic region reached MCID significantly more than that of patients who had a UIV in the thoracolumbar region (70.6% vs 41.9%, p = 0.0281). All other outcomes were similar for UIVs of upper thoracic and thoracolumbar regions. Comparison of patients whose spines were above or below the radiographic thresholds associated with disability indicated similar rates of meeting MCID and SCB for HRQOL at the 2-year follow-up.

CONCLUSIONS

Outcomes for patients having UIVs in the upper thoracic region were no more likely to meet MCID or SCB than for those having UIVs in the thoracolumbar region, except for the MCID in the SRS-22 self-image measure. The HRQOL outcomes in patients who had optimal sagittal correction according to radiographic thresholds determined preoperatively were not significantly more likely to reach MCID or SCB at the 2-year follow-up. Future work needs to determine whether the Schwab preoperative radiographic thresholds for severe disability apply in postoperative settings.

Full access

Renaud Lafage, Ibrahim Obeid, Barthelemy Liabaud, Shay Bess, Douglas Burton, Justin S. Smith, Cyrus Jalai, Richard Hostin, Christopher I. Shaffrey, Christopher Ames, Han Jo Kim, Eric Klineberg, Frank Schwab, Virginie Lafage, and the International Spine Study Group

OBJECTIVE

The surgical correction of adult spinal deformity (ASD) often involves modifying lumbar lordosis (LL) to restore ideal sagittal alignment. However, corrections that include large changes in LL increase the risk for development of proximal junctional kyphosis (PJK). Little is known about the impact of cranial versus caudal correction in the lumbar spine on the occurrence of PJK. The goal of this study was to investigate the impact of the location of the correction on acute PJK development.

METHODS

This study was a retrospective review of a prospective multicenter database. Surgically treated ASD patients with early follow-up evaluations (6 weeks) and fusions of the full lumbosacral spine were included. Radiographic parameters analyzed included the classic spinopelvic parameters (pelvic incidence [PI], pelvic tilt [PT], PI−LL, and sagittal vertical axis [SVA]) and segmental correction. Using Glattes’ criteria, patients were stratified into PJK and noPJK groups and propensity matched by age and regional lumbar correction (ΔPI−LL). Radiographic parameters and segmental correction were compared between PJK and noPJK patients using independent t-tests.

RESULTS

After propensity matching, 312 of 483 patients were included in the analysis (mean age 64 years, 76% women, 40% with PJK). There were no significant differences between PJK and noPJK patients at baseline or postoperatively, or between changes in alignment, with the exception of thoracic kyphosis (TK) and ΔTK. PJK patients had a decrease in segmental lordosis at L4-L5-S1 (−0.6° vs 1.6°, p = 0.025), and larger increases in segmental correction at cranial levels L1-L2-L3 (9.9° vs 7.1°), T12-L1-L2 (7.3° vs 5.4°), and T11-T12-L1 (2.9° vs 0.7°) (all p < 0.05).

CONCLUSIONS

Although achievement of an optimal sagittal alignment is the goal of realignment surgery, dramatic lumbar corrections appear to increase the risk of PJK. This study was the first to demonstrate that patients who developed PJK underwent kyphotic changes in the L4–S1 segments while restoring LL at more cranial levels (T12–L3). These findings suggest that restoring lordosis at lower lumbar levels may result in a decreased risk of developing PJK.

Full access

Dominic Maggio, Tamir T. Ailon, Justin S. Smith, Christopher I. Shaffrey, Virginie Lafage, Frank Schwab, Regis W. Haid Jr., Themistocles Protopsaltis, Eric Klineberg, Justin K. Scheer, Shay Bess, Paul M. Arnold, Jens Chapman, Michael G. Fehlings, Christopher Ames, AOSpine North America, and International Spine Study Group

OBJECT

The associations among global spinal alignment, patient-reported disability, and surgical outcomes have increasingly gained attention. The assessment of global spinal alignment requires standing long-cassette anteroposterior and lateral radiographs; however, spine surgeons routinely rely only on short-segment imaging when evaluating seemingly isolated lumbar pathology. This may prohibit adequate surgical planning and may predispose surgeons to not recognize associated pathology in the thoracic spine and sagittal spinopelvic malalignment. The authors used a case-based survey questionnaire to evaluate if including long-cassette radiographs led to changes to respondents' operative plans as compared with their chosen plan when cases contained standard imaging of the involved lumbar spine only.

METHODS

A case-based survey was distributed to AOSpine International members that consisted of 15 cases of lumbar spine pathology and lumbar imaging only. The same 15 cases were then shuffled and presented a second time with additional long-cassette radiographs. Each case required participants to select a single operative plan with 5 choices ranging from least to most extensive. The cases included 5 “control” cases with normal global spinal alignment and 10 “test” cases with significant sagittal and/or coronal malalignment. Mean scores were determined for each question with higher scores representing more invasive and/or extensive operative plans.

RESULTS

Of 712 spine surgeons who started the survey, 316 (44%) completed the entire series, including 68% of surgeons with spine fellowship training and representation from more than 40 countries. For test cases, but not for control cases, there were significantly higher average surgical invasiveness scores for cases presented with long-cassette radiographs (4.2) as compared with those cases with lumbar imaging only (3.4; p = 0.002). The addition of long-cassette radiographs resulted in 82.1% of respondents recommending instrumentation up to the thoracic spine, a 23.2% increase as compared with the same cases presented with lumbar imaging only (p = 0.008).

CONCLUSIONS

This study demonstrates the importance of maintaining a low threshold for performing standing long-cassette imaging when assessing seemingly isolated lumbar pathology. Such imaging is necessary for the assessment of spinopelvic and global spinal alignment, which can be important in operative planning. Deformity, particularly positive sagittal malalignment, may go undetected unless one maintains a high index of suspicion and obtains long-cassette radiographs. It is recommended that spine surgeons recognize the prevalence and importance of such deformity when contemplating operative intervention.

Free access

Taemin Oh, Justin K. Scheer, Justin S. Smith, Richard Hostin, Chessie Robinson, Jeffrey L. Gum, Frank Schwab, Robert A. Hart, Virginie Lafage, Douglas C. Burton, Shay Bess, Themistocles Protopsaltis, Eric O. Klineberg, Christopher I. Shaffrey, Christopher P. Ames, and the International Spine Study Group

OBJECTIVE

Patients with adult spinal deformity (ASD) experience significant quality of life improvements after surgery. Treatment, however, is expensive and complication rates are high. Predictive analytics has the potential to use many variables to make accurate predictions in large data sets. A validated minimum clinically important difference (MCID) model has the potential to assist in patient selection, thereby improving outcomes and, potentially, cost-effectiveness.

METHODS

The present study was a retrospective analysis of a multiinstitutional database of patients with ASD. Inclusion criteria were as follows: age ≥ 18 years, radiographic evidence of ASD, 2-year follow-up, and preoperative Oswestry Disability Index (ODI) > 15. Forty-six variables were used for model training: demographic data, radiographic parameters, surgical variables, and results on the health-related quality of life questionnaire. Patients were grouped as reaching a 2-year ODI MCID (+MCID) or not (−MCID). An ensemble of 5 different bootstrapped decision trees was constructed using the C5.0 algorithm. Internal validation was performed via 70:30 data split for training/testing. Model accuracy and area under the curve (AUC) were calculated. The mean quality-adjusted life years (QALYs) and QALYs gained at 2 years were calculated and discounted at 3.5% per year. The QALYs were compared between patients in the +MCID and –MCID groups.

RESULTS

A total of 234 patients met inclusion criteria (+MCID 129, −MCID 105). Sixty-nine patients (29.5%) were included for model testing. Predicted versus actual results were 50 versus 40 for +MCID and 19 versus 29 for −MCID (i.e., 10 patients were misclassified). Model accuracy was 85.5%, with 0.96 AUC. Predicted results showed that patients in the +MCID group had significantly greater 2-year mean QALYs (p = 0.0057) and QALYs gained (p = 0.0002).

CONCLUSIONS

A successful model with 85.5% accuracy and 0.96 AUC was constructed to predict which patients would reach ODI MCID. The patients in the +MCID group had significantly higher mean 2-year QALYs and QALYs gained. This study provides proof of concept for using predictive modeling techniques to optimize patient selection in complex spine surgery.

Full access

Justin K. Scheer, Peter G. Passias, Alexandra M. Sorocean, Anthony J. Boniello, Gregory M. Mundis Jr., Eric Klineberg, Han Jo Kim, Themistocles S. Protopsaltis, Munish Gupta, Shay Bess, Christopher I. Shaffrey, Frank Schwab, Virginie Lafage, Justin S. Smith, Christopher P. Ames, and The International Spine Study Group

OBJECT

A high prevalence of cervical deformity (CD) has been identified among adult patients with thoracolumbar spinal deformity undergoing surgical treatment. The clinical impact of this is uncertain. This study aimed to quantify the differences in patient-reported outcomes among patients with adult spinal deformity (ASD) based on presence of CD prior to treatment.

METHODS

A retrospective review was conducted of a multicenter prospective database of patients with ASD who underwent surgical treatment with 2-year follow-up. Patients were grouped by the presence of preoperative CD: 1) cervical positive sagittal malalignment (CPSM) C2–7 sagittal vertical axis ≥ 4 cm; 2) cervical kyphosis (CK) C2–7 angle > 0; 3) CPSM and CK (BOTH); and 4) no baseline CD (NONE). Health-related quality of life (HRQOL) scores included the Physical Component Summary and Mental Component Summary (PCS and MCS) scores of the 36-Item Short Form Health Survey (SF-36), Oswestry Disability Index (ODI), Scoliosis Research Society-22 questionnaire (SRS-22), and minimum clinically important difference (MCID) of these scores at 2 years. Standard radiographic measurements were conducted for cervical, thoracic, and thoracolumbar parameters.

RESULTS

One hundred eighty-two patients were included in this study: CPSM, 45; CK, 37; BOTH, 16; and NONE, 84. Patients with preoperative CD and those without had similar baseline thoracolumbar radiographic measurements and similar correction rates at 2 years. Patients with and without preoperative CD had similar baseline HRQOL and on average both groups experienced some HRQOL improvement. However, those with preoperative CPSM had significantly worse postoperative ODI, PCS, SRS-22 Activity, SRS-22 Appearance, SRS-22 Pain, SRS-22 Satisfaction, and SRS-22 Total score, and were less likely to meet MCID for ODI, PCS, SRS-22 Activity, and SRS-22 Pain scores with the following ORs and 95% CIs: ODI 0.19 (0.07–0.58), PCS 0.17 (0.06–0.47), SRS-22 Activity 0.23 (0.09–0.62), SRS-22 Pain 0.20 (0.08–0.53), and SRS-22 Appearance 0.34 (0.12–0.94). Preoperative CK did not have an effect on outcomes. Interestingly, despite correction of the thoracolumbar deformity, 53.3% and 51.4% of patients had persistent CPSM and persistent CK, respectively.

CONCLUSIONS

Patients with thoracolumbar deformity without preoperative CD are likely to have greater improvements in HRQOL after surgery than patients with concomitant preoperative CD. Cervical positive sagittal alignment in adult patients with thoracolumbar deformity is strongly associated with inferior outcomes and failure to reach MCID at 2-year follow-up despite having similar baseline HRQOL to patients without CD. This was the first study to assess the impact of concomitant preoperative cervical malalignment in adult patients with thoracolumbar deformity. These results can help surgeons educate patients at risk for inferior outcomes and direct future research to identify an etiology and improve patient outcomes. Investigation into the etiology of the baseline cervical malalignment may be warranted in patients who present with thoracolumbar deformity.

Full access

Justin K. Scheer, Justin S. Smith, Frank Schwab, Virginie Lafage, Christopher I. Shaffrey, Shay Bess, Alan H. Daniels, Robert A. Hart, Themistocles S. Protopsaltis, Gregory M. Mundis Jr., Daniel M. Sciubba, Tamir Ailon, Douglas C. Burton, Eric Klineberg, Christopher P. Ames, and The International Spine Study Group

OBJECTIVE

The operative management of patients with adult spinal deformity (ASD) has a high complication rate and it remains unknown whether baseline patient characteristics and surgical variables can predict early complications (intraoperative and perioperative [within 6 weeks]). The development of an accurate preoperative predictive model can aid in patient counseling, shared decision making, and improved surgical planning. The purpose of this study was to develop a model based on baseline demographic, radiographic, and surgical factors that can predict if patients will sustain an intraoperative or perioperative major complication.

METHODS

This study was a retrospective analysis of a prospective, multicenter ASD database. The inclusion criteria were age ≥ 18 years and the presence of ASD. In total, 45 variables were used in the initial training of the model including demographic data, comorbidities, modifiable surgical variables, baseline health-related quality of life, and coronal and sagittal radiographic parameters. Patients were grouped as either having at least 1 major intraoperative or perioperative complication (COMP group) or not (NOCOMP group). An ensemble of decision trees was constructed utilizing the C5.0 algorithm with 5 different bootstrapped models. Internal validation was accomplished via a 70/30 data split for training and testing each model, respectively. Overall accuracy, the area under the receiver operating characteristic (AUROC) curve, and predictor importance were calculated.

RESULTS

Five hundred fifty-seven patients were included: 409 (73.4%) in the NOCOMP group, and 148 (26.6%) in the COMP group. The overall model accuracy was 87.6% correct with an AUROC curve of 0.89 indicating a very good model fit. Twenty variables were determined to be the top predictors (importance ≥ 0.90 as determined by the model) and included (in decreasing importance): age, leg pain, Oswestry Disability Index, number of decompression levels, number of interbody fusion levels, Physical Component Summary of the SF-36, Scoliosis Research Society (SRS)–Schwab coronal curve type, Charlson Comorbidity Index, SRS activity, T-1 pelvic angle, American Society of Anesthesiologists grade, presence of osteoporosis, pelvic tilt, sagittal vertical axis, primary versus revision surgery, SRS pain, SRS total, use of bone morphogenetic protein, use of iliac crest graft, and pelvic incidence–lumbar lordosis mismatch.

CONCLUSIONS

A successful model (87% accuracy, 0.89 AUROC curve) was built predicting major intraoperative or perioperative complications following ASD surgery. This model can provide the foundation toward improved education and point-of-care decision making for patients undergoing ASD surgery.