Significant Reduction in Bone Density as Measured by Hounsfield Units in Patients with Ankylosing Spondylitis or Diffuse Idiopathic Skeletal Hyperostosis.

Background: Multisegmental pathologic autofusion occurs in patients with ankylosing spondylitis (AS) and diffuse idiopathic skeletal hyperostosis (DISH). It may lead to reduced vertebral bone density due to stress shielding. Methods: This study aimed to determine the effects of autofusion on bone density by measuring Hounsfield units (HU) in the mobile and immobile spinal segments of patients with AS and DISH treated at a tertiary care center. The mean HU was calculated for five distinct regions-cranial adjacent mobile segment, cranial fused segment, mid-construct fused segment, caudal fused segment, and caudal adjacent mobile segment. Means for each region were compared using paired-sample t-tests. Multivariable regression was used to determine independent predictors of mid-fused segment HUs. Results: One hundred patients were included (mean age 76 ± 11 years, 74% male). The mean HU for the mid-construct fused segment (100, 95% CI [86, 113]) was significantly lower than both cranial and caudal fused segments (174 and 108, respectively; both p < 0.001), and cranial and caudal adjacent mobile segments (195 and 115, respectively; both p < 0.001). Multivariable regression showed the mid-construct HUs were predicted by history of smoking (-30 HU, p = 0.009). Conclusions: HUs were significantly reduced in the middle of long-segment autofusion, which was consistent with stress shielding. Such shielding may contribute to the diminution of vertebral bone integrity in AS/DISH patients and potentially increased fracture risk.

Reliable Prediction of Discharge Disposition Following Cervical Spine Surgery With Ensemble Machine Learning and Validation on a National Cohort.

STUDY DESIGN: A retrospective cohort study. OBJECTIVE: The purpose of this study is to develop a machine learning algorithm to predict nonhome discharge after cervical spine surgery that is validated and usable on a national scale to ensure generalizability and elucidate candidate drivers for prediction. SUMMARY OF BACKGROUND DATA: Excessive length of hospital stay can be attributed to delays in postoperative referrals to intermediate care rehabilitation centers or skilled nursing facilities. Accurate preoperative prediction of patients who may require access to these resources can facilitate a more efficient referral and discharge process, thereby reducing hospital and patient costs in addition to minimizing the risk of hospital-acquired complications. METHODS: Electronic medical records were retrospectively reviewed from a single-center data warehouse (SCDW) to identify patients undergoing cervical spine surgeries between 2008 and 2019 for machine learning algorithm development and internal validation. The National Inpatient Sample (NIS) database was queried to identify cervical spine fusion surgeries between 2009 and 2017 for external validation of algorithm performance. Gradient-boosted trees were constructed to predict nonhome discharge across patient cohorts. The area under the receiver operating characteristic curve (AUROC) was used to measure model performance. SHAP values were used to identify nonlinear risk factors for nonhome discharge and to interpret algorithm predictions. RESULTS: A total of 3523 cases of cervical spine fusion surgeries were included from the SCDW data set, and 311,582 cases were isolated from NIS. The model demonstrated robust prediction of nonhome discharge across all cohorts, achieving an area under the receiver operating characteristic curve of 0.87 (SD=0.01) on both the SCDW and nationwide NIS test sets. Anterior approach only, age, elective admission status, Medicare insurance status, and total Elixhauser Comorbidity Index score were the most important predictors of discharge destination. CONCLUSIONS: Machine learning algorithms reliably predict nonhome discharge across single-center and national cohorts and identify preoperative features of importance following cervical spine fusion surgery.

The Predictors of Incidental Durotomy in Patients Undergoing Pedicle Subtraction Osteotomy for the Correction of Adult Spinal Deformity.

Background: Pedicle subtraction osteotomy (PSO) is a powerful tool for sagittal plane correction in patients with rigid adult spinal deformity (ASD); however, it is associated with high intraoperative blood loss and the increased risk of durotomy. The objective of the present study was to identify intraoperative techniques and baseline patient factors capable of predicting intraoperative durotomy. Methods: A tri-institutional database was retrospectively queried for all patients who underwent PSO for ASD. Data on baseline comorbidities, surgical history, surgeon characteristics and intraoperative maneuvers were gathered. PSO aggressiveness was defined as conventional (Schwab 3 PSO) or an extended PSO (Schwab type 4). The primary outcome of the study was the occurrence of durotomy intraoperatively. Univariable analyses were performed with Mann-Whitney U tests, Chi-squared analyses, and Fisher's exact tests. Statistical significance was defined by p < 0.05. Results: One hundred and sixteen patients were identified (mean age 61.9 ± 12.6 yr; 44.8% male), of whom 51 (44.0%) experienced intraoperative durotomy. There were no significant differences in baseline comorbidities between those who did and did not experience durotomy, with the exception that baseline weight and body mass index were higher in patients who did not suffer durotomy. Prior surgery (OR 2.73; 95% CI [1.13, 6.58]; p = 0.03) and, more specifically, prior decompression at the PSO level (OR 4.23; 95% CI [1.92, 9.34]; p < 0.001) was predictive of durotomy. A comparison of surgeon training showed no statistically significant difference in durotomy rate between fellowship and non-fellowship trained surgeons, or between orthopedic surgeons and neurosurgeons. The PSO level, PSO aggressiveness, the presence of stenosis at the PSO level, nor the surgical instrument used predicted the odds of durotomy occurrence. Those experiencing durotomy had similar hospitalization durations, rates of reoperation and rates of nonroutine discharge. Conclusions: In this large multisite series, a history of prior decompression at the PSO level was associated with a four-fold increase in intraoperative durotomy risk. Notably the use of extended (versus) standard PSO, surgical technique, nor baseline patient characteristics predicted durotomy. Durotomies occurred in 44% of patients and may prolong operative times. Additional prospective investigations are merited.

Angiographic cross-filling between inferior petrosal sinuses and alteration of adrenocorticotropic hormone sampling results for tumor localization in Cushing disease.

OBJECTIVE: Inferior petrosal sinus (IPS) sampling (IPSS) is a diagnostic procedure used to guide diagnostic localization of imaging-negative adrenocorticotropic hormone (ACTH)-secreting pituitary microadenomas. However, the efficacy of IPSS has been suboptimal at accurately lateralizing the adenoma, reducing surgical cure rates and leading to unintended pituitary dysfunction due to the added exploration. One rationale for the occasional imprecision is the existence of additional petrosal sinus collateral channels that connect the IPS bilaterally, which may lead to false localization results during sampling. The aim of this study was to explore a potential connection between normal anatomical variation in the angioarchitecture of the IPSs and the ACTH results obtained in subsequent IPSS tests. METHODS: A retrospective review was performed on all cases between 1998 and 2013 involving patients at a single institution who underwent IPSS for radiographically equivocal pituitary microadenomas. Cases were reviewed for tumor laterality noted on either operative or pathology reports, as well as the presence of angiographic evidence of cross-filling between the sinuses. In addition, ACTH levels from the right and left IPSs were documented at baseline and at 2, 5, and 10 minutes after corticotropin-releasing hormone (CRH) administration. A ratio of the change in ACTH levels measured at the time of maximal response (10 minutes) versus the levels measured at the initial response (2 minutes) was computed for each patient and compared between patients by their angiographic cross-filling status. RESULTS: There were 41 patients with a histopathologically confirmed right- or left-sided ACTH-secreting pituitary microadenoma who underwent preoperative IPSS. Among these patients, 28 (68%) showed angiographic evidence of cross-filling between the IPSs, and 13 showed no cross-filling. On average, ACTH levels increased by a factor of 3.91 ± 0.77 in the contralateral IPS in patients with angiographic cross-filling, compared with a factor increase of only 1.80 ± 0.27 in patients without cross-filling (p = 0.014). In comparison, ACTH levels increased by a factor of 2.01 ± 0.57 in the ipsilateral IPS in patients with cross-filling, and by 8.78 ± 7.30 in those without cross-filling (p = 0.373). CONCLUSIONS: The presence of angiographic cross-filling, suggestive of a greater degree of vascular channel networking between the right and left IPS, is a significant factor influencing the measured rates of change of ACTH in IPSS and may impact the specificity of this test to accurately determine microadenoma laterality in the preoperative setting.

Dopamine-Depleted Dopamine Transporter Knockout (DDD) Mice: Dyskinesia with L-DOPA and Dopamine D1 Agonists.

L-DOPA is the mainstay of treatment for Parkinson's disease (PD). However, over time this drug can produce dyskinesia. A useful acute PD model for screening novel compounds for anti-parkinsonian and L-DOPA-induced dyskinesia (LID) are dopamine-depleted dopamine-transporter KO (DDD) mice. Treatment with α-methyl-para-tyrosine rapidly depletes their brain stores of DA and renders them akinetic. During sensitization in the open field (OF), their locomotion declines as vertical activities increase and upon encountering a wall they stand on one leg or tail and engage in climbing behavior termed "three-paw dyskinesia". We have hypothesized that L-DOPA induces a stereotypic activation of locomotion in DDD mice, where they are unable to alter the course of their locomotion, and upon encountering walls engage in "three-paw dyskinesia" as reflected in vertical counts or beam-breaks. The purpose of our studies was to identify a valid index of LID in DDD mice that met three criteria: (a) sensitization with repeated L-DOPA administration, (b) insensitivity to a change in the test context, and (c) stimulatory or inhibitory responses to dopamine D1 receptor agonists (5 mg/kg SKF81297; 5 and 10 mg/kg MLM55-38, a novel compound) and amantadine (45 mg/kg), respectively. Responses were compared between the OF and a circular maze (CM) that did not hinder locomotion. We found vertical counts and climbing were specific for testing in the OF, while oral stereotypies were sensitized to L-DOPA in both the OF and CM and responded to D1R agonists and amantadine. Hence, in DDD mice oral stereotypies should be used as an index of LID in screening compounds for PD.

Opportunistic CT-Based Hounsfield Units Strongly Correlate with Biomechanical CT Measurements in the Thoracolumbar Spine.

STUDY DESIGN: Retrospective cohort study. OBJECTIVE: Hounsfield units (HUs) are known to correlate with clinical outcomes, no study has evaluated how they correlate with BCT and DXA measurements. SUMMARY OF BACKGROUND: Low bone mineral density (BMD) represents a major risk factor for fracture and poor outcomes following spine surgery. Dual-energy x-ray absorptiometry (DXA) can provide regional BMD measurements but has limitations. Opportunistic HUs provide targeted BMD estimates; however, they are not formally accepted for diagnosing osteoporosis in current guidelines. More recently, biomechanical computed tomography (BCT) analysis has emerged as a new modality endorsed by the International Society for Clinical Densitometry (ISCD) for assessing bone strength. METHODS: Consecutive cases from 2017-2022 at a single institution were reviewed for patients who underwent BCT in the thoracolumbar spine. BCT-measured vertebral strength, trabecular BMD, and the corresponding American College of Radiology (ACR) Classification were recorded. DXA studies within three months of the BCT were reviewed. Pearson Correlation Coefficients were calculated, and receiver-operating characteristic curves were constructed to assess the predictive capacity of HUs. Threshold analysis was performed to identify optimal HU values for identifying osteoporosis and low BMD. RESULTS: Correlation analysis of 114 cases revealed a strong relationship between HUs and BCT vertebral strength (r=0.69; P<0.0001; R2=0.47) and trabecular BMD (r=0.76; P<0.0001; R2=0.58). However, DXA poorly correlated with opportunistic HUs and BCT measurements. HUs accurately predicted osteoporosis and low BMD (Osteoporosis: C=0.95, 95% CI 0.89-1.00; Low BMD: C=0.87, 95% CI 0.79-0.96). Threshold analysis revealed that 106 and 122 HUs represent optimal thresholds for detecting osteoporosis and low BMD. CONCLUSION: Opportunistic HUs strongly correlated with BCT-based measures, while neither correlated strongly with DXA-based BMD measures in the thoracolumbar spine. HUs are easy to perform at no additional cost and provide accurate BMD estimates at non-instrumented vertebral levels across all ACR-designated BMD categories.

Robust prediction of nonhome discharge following elective anterior cervical discectomy and fusion using explainable machine learning.

PURPOSE: Predict nonhome discharge (NHD) following elective anterior cervical discectomy and fusion (ACDF) using an explainable machine learning model. METHODS: 2227 patients undergoing elective ACDF from 2008 to 2019 were identified from a single institutional database. A machine learning model was trained on preoperative variables, including demographics, comorbidity indices, and levels fused. The validation technique was repeated stratified K-Fold cross validation with the area under the receiver operating curve (AUROC) statistic as the performance metric. Shapley Additive Explanation (SHAP) values were calculated to provide further explainability regarding the model's decision making. RESULTS: The preoperative model performed with an AUROC of 0.83 ± 0.05. SHAP scores revealed the most pertinent risk factors to be age, medicare insurance, and American Society of Anesthesiology (ASA) score. Interaction analysis demonstrated that female patients over 65 with greater fusion levels were more likely to undergo NHD. Likewise, ASA demonstrated positive interaction effects with female sex, levels fused and BMI. CONCLUSION: We validated an explainable machine learning model for the prediction of NHD using common preoperative variables. Adding transparency is a key step towards clinical application because it demonstrates that our model's "thinking" aligns with clinical reasoning. Interactive analysis demonstrated that those of age over 65, female sex, higher ASA score, and greater fusion levels were more predisposed to NHD. Age and ASA score were similar in their predictive ability. Machine learning may be used to predict NHD, and can assist surgeons with patient counseling or early discharge planning.

A Nationwide Study Characterizing the Risk and Outcome Profiles of Multilevel Fusion Procedures in Neuromuscular Scoliosis Patients with Neurofibromatosis Type 1.

BACKGROUND: Spine abnormalities are a common manifestation of Neurofibromatosis Type 1 (NF1); however, the outcomes of surgical treatment for NF1-associated spinal deformity are not well explored. The purpose of this study was to investigate the outcome and risk profiles of multilevel fusion surgery for NF1 patients. METHODS: The National Inpatient Sample was queried for NF1 and non-NF1 patient populations with neuromuscular scoliosis who underwent multilevel fusion surgery involving eight or more vertebral levels between 2004 and 2017. Multivariate regression modeling was used to explore the relationship between perioperative variables and pertinent outcomes. RESULTS: Of the 55,485 patients with scoliosis, 533 patients (0.96%) had NF1. Patients with NF1 were more likely to have comorbid solid tumors (P < 0.0001), clinical depression (P < 0.0001), peripheral vascular disease (P < 0.0001), and hypertension (P < 0.001). Following surgery, NF1 patients had a higher incidence of hydrocephalus (0.6% vs. 1.9% P = 0.002), seizures (4.9% vs. 5.7% P = 0.006), and accidental vessel laceration (0.3% vs.1.9% P = 0.011). Although there were no differences in overall complication rates or in-hospital mortality, multivariate regression revealed NF1 patients had an increased probability of pulmonary (OR 0.5, 95%CI 0.3-0.8, P = 0.004) complications. There were no significant differences in utilization, including nonhome discharge or extended hospitalization; however, patients with NF1 had higher total hospital charges (mean -$18739, SE 3384, P < 0.0001). CONCLUSIONS: These findings indicate that NF1 is associated with certain complications following multilevel fusion surgery but does not appear to be associated with differences in quality or cost outcomes. These results provide some guidance to surgeons and other healthcare professionals in their perioperative decision making by raising awareness about risk factors for NF1 patients undergoing multilevel fusion surgery. We intend for this study to set the national baseline for complications after multilevel fusion in the NF1 population.

Comparative Analysis of Early Outcomes and Complications of PSO Among Neurosurgeons and Orthopedic Surgeons.

STUDY DESIGN: Retrospective comparative cohort study using the National Surgical Quality Improvement Program. OBJECTIVE: The aim of this study was to evaluate trends in the annual number of PSOs performed, describe the patient populations associated with each cohort, and compare outcomes between specialties.Summary of Background Data:Pedicle subtraction osteotomies (PSO) are complex and advanced spine deformity surgical procedures performed by neurosurgeons and orthopedic surgeons. Though both sets of surgeons can be equally qualified and credentialed to perform a PSO, it is possible that differences in training and exposure could translate into differences in patient management and outcomes. METHODS: Patients that underwent lumbar PSO from 2005 to 2014 in the American College of Surgeons-National Surgical Quality Improvement Program registry were identified. Relevant demographic, preoperative comorbidity, and postoperative 30-day complications were queried and analyzed. The data was divided into 2 cohorts consisting of those patients who were treated by neurosurgeons versus orthopedic surgeons. Additional data from the Scoliosis Research Society Morbidity and Mortality database was queried and analyzed for comparison. RESULTS: Demographic and comorbidity factors were similar between the neurosurgery and orthopedic surgery cohorts, except there were higher rates of hypertension among orthopedic surgeon-performed PSOs (65.66% vs. 48.67%, P =0.004). Except for 2012, in every year queried, orthopedic surgeons reported more PSOs than neurosurgeons. In patients who underwent lumbar fusion surgery, there was a higher rate of PSOs if the surgery was performed by an orthopedic surgeon (OR 1.7824, 95% CI: 1.4017-2.2665). The incidence of deep vein thrombosis after PSOs was higher for neurosurgery compared with orthopedic surgery (8.85% vs. 1.20%, P =0.004). However, besides deep vein thrombosis, there were no salient differences in surgical complication rates between neurosurgeon-performed PSOs and orthopedic surgeon-performed PSOs. CONCLUSIONS: The number of PSO procedures performed by neurosurgeons and orthopedic surgeons has increased annually. Differences in outcomes between neurosurgeons and orthopedic surgeons suggest an opportunity for wider assessment and alignment of adult spinal deformity surgery exposure and training across specialties.

Risk Factors for Delayed Cerebral Ischemia in Good-Grade Patients With Aneurysmal Subarachnoid Hemorrhage.

Background A subset of good-grade patients with aneurysmal subarachnoid hemorrhage (aSAH) develop delayed cerebral ischemia (DCI) that may cause permanent disabilities after aSAH. However, little is known about the risk factors of DCI among this specific patient group. Methods and Results We obtained a multinational cohort of good-grade (Glasgow Coma Scale 13-15 on admission) patients with aSAH by pooling patient data from 4 clinical trials and 2 prospective cohort studies. We collected baseline data on lifestyle-related factors and the clinical characteristics of aSAHs. By calculating fully adjusted risk estimates for DCI and DCI-related poor outcome, we identified the most high-risk patient groups. The pooled study cohort included 1918 good-grade patients with aSAH (median age, 51 years; 64% women), of whom 21% and 7% experienced DCI and DCI-related poor outcome, respectively. Among men, patients with obesity and (body mass index ≥30 kg/m2) thick aSAH experienced most commonly DCI (33%) and DCI-related poor outcome (20%), whereas none of the normotensive or young (aged <50 years) men with low body mass index (body mass index <22.5 kg/m2) had DCI-related poor outcome. In women, the highest prevalence of DCI (28%) and DCI-related poor outcome (13%) was found in patients with preadmission hypertension and thick aSAH. Conversely, the lowest rates (11% and 2%, respectively) were observed in normotensive women with a thin aSAH. Conclusions Increasing age, thick aSAH, obesity, and preadmission hypertension are risk factors for DCI in good-grade patients with aSAH. These findings may help clinicians to consider which good-grade patients with aSAH should be monitored carefully in the intensive care unit.

Deploying deep learning models on unseen medical imaging using adversarial domain adaptation.

The fundamental challenge in machine learning is ensuring that trained models generalize well to unseen data. We developed a general technique for ameliorating the effect of dataset shift using generative adversarial networks (GANs) on a dataset of 149,298 handwritten digits and dataset of 868,549 chest radiographs obtained from four academic medical centers. Efficacy was assessed by comparing area under the curve (AUC) pre- and post-adaptation. On the digit recognition task, the baseline CNN achieved an average internal test AUC of 99.87% (95% CI, 99.87-99.87%), which decreased to an average external test AUC of 91.85% (95% CI, 91.82-91.88%), with an average salvage of 35% from baseline upon adaptation. On the lung pathology classification task, the baseline CNN achieved an average internal test AUC of 78.07% (95% CI, 77.97-78.17%) and an average external test AUC of 71.43% (95% CI, 71.32-71.60%), with a salvage of 25% from baseline upon adaptation. Adversarial domain adaptation leads to improved model performance on radiographic data derived from multiple out-of-sample healthcare populations. This work can be applied to other medical imaging domains to help shape the deployment toolkit of machine learning in medicine.

The Impact of Obstructive Sleep Apnea on Clinical, Perioperative, and Cost Outcomes in Patients Who Underwent Posterior Cervical Decompression and Fusion: A Single-Center Retrospective Analysis From 2008 to 2016.

BACKGROUND: Obstructive sleep apnea (OSA) is a pervasive problem that can result in diminished neurocognitive performance, increased risk of all-cause mortality, and significant cardiovascular disease. While previous studies have examined risk factors that influence outcomes following cervical fusion procedures, to our knowledge, no study has examined the cost or outcome profiles for posterior cervical decompression and fusion (PCDF) procedures in patients with OSA. METHODS: All cases at a single institution between 2008 and 2016 involving a PCDF were included. The primary outcome was prolonged extubation, defined as an extubation that took place outside of the operating room. Secondary outcomes included admission to the intensive care unit (ICU), complications, extended hospitalization, nonhome discharge, readmission within 30 and 90 days, emergency room visit within 30 and 90 days, and higher total costs. RESULTS: We reviewed 1191 PCDF cases, of which 93 patients (7.81%) had a history of OSA. At the univariate level, patients with OSA had higher rates of ICU admissions (33.3% vs 16.8%, P < 0.0001), total complications (29.0% vs 19.0%, P = 0.0202), and respiratory complications (12.9% vs 6.6%, P = 0.0217). Multivariate regression analyses revealed no difference in the odds of a prolonged extubation (P = 0.4773) and showed that history of OSA was not predictive of higher costs. However, a significant difference was observed in the odds of having an ICU admission (P = 0.0046). CONCLUSION: While patients with sleep apnea may be more likely to be admitted to the ICU postoperatively, OSA status a lone is not a risk factor for poor primary and secondary clinical outcomes following posterior cervical fusion procedures. CLINICAL RELEVANCE: Various deformities of the cervical spine can exert extraluminal forces that partially collapse or obstruct the airway, thereby predisposing patients to OSA; however, no study has examined the cost or outcome profiles for PCDF procedures in patients with OSA. Therefore, this investigation highlights the ways in which OSA influences the risks, outcomes, and costs following PCDF using medical data from an institutional registry.

Evaluation of a Role for Virtual Neurosurgical Education for Medical Students Over 2 Years of a Global Pandemic.

BACKGROUND: Subinternships are critical experiences for medical students applying into neurosurgery to acquire knowledge of the field and network with colleagues. During the coronavirus disease 2019 pandemic, in-person rotations were suspended for 2020 and reduced for 2021. In 2020, our department developed a neurosurgical course to address this need. The course was continued in 2021, enabling assessment of student perceptions as the pandemic progresses. METHODS: The virtual course consisted of weekly 1-hour seminars over a 3- to 4-month period. Prior to starting, participants were sent a comprehensive survey assessing their backgrounds, experiences, and confidences in core concepts across neurosurgical subdisciplines. Participants also completed postcourse surveys assessing the course's value and their confidence in the same topics. Responses from students completing both precourse and postcourse surveys were included, analyzed in pairwise fashion, and compared across course years. RESULTS: Students shared similar baseline characteristics in terms of demographics, educational background, and exposure to neurosurgery prior to the course. In the 2020 and 2021 cohorts, quality ratings for presentations were favorable for all seminars, and participants reported significantly increased confidence in core topics across all neurosurgical disciplines after the course (2020: 3.36 ± 0.26, P < 0.0001; 2021: 3.56 ± 0.93, P = 0.005). Most participants felt the course would remain useful following the pandemic in both the 2020 (96.9%) and 2021 (100.0%) cohorts. CONCLUSIONS: Survey results suggest that the course adds value for students seeking a basic didactic curriculum to supplement their education, and perhaps, an online curriculum for medical students would still be beneficial going forward as in-person rotations resume.

Pragmatic Prediction of Excessive Length of Stay After Cervical Spine Surgery With Machine Learning and Validation on a National Scale.

BACKGROUND: Extended postoperative hospital stays are associated with numerous clinical risks and increased economic cost. Accurate preoperative prediction of extended length of stay (LOS) can facilitate targeted interventions to mitigate clinical harm and resource utilization. OBJECTIVE: To develop a machine learning algorithm aimed at predicting extended LOS after cervical spine surgery on a national level and elucidate drivers of prediction. METHODS: Electronic medical records from a large, urban academic medical center were retrospectively examined to identify patients who underwent cervical spine fusion surgeries between 2008 and 2019 for machine learning algorithm development and in-sample validation. The National Inpatient Sample database was queried to identify cervical spine fusion surgeries between 2009 and 2017 for out-of-sample validation of algorithm performance. Gradient-boosted trees predicted LOS and efficacy was assessed using the area under the receiver operating characteristic curve (AUROC). Shapley values were calculated to characterize preoperative risk factors for extended LOS and explain algorithm predictions. RESULTS: Gradient-boosted trees accurately predicted extended LOS across cohorts, achieving an AUROC of 0.87 (SD = 0.01) on the single-center validation set and an AUROC of 0.84 (SD = 0.00) on the nationwide National Inpatient Sample data set. Anterior approach only, elective admission status, age, and total number of Elixhauser comorbidities were important predictors that affected the likelihood of prolonged LOS. CONCLUSION: Machine learning algorithms accurately predict extended LOS across single-center and national patient cohorts and characterize key preoperative drivers of increased LOS after cervical spine surgery.

Robust Prediction of Non-home Discharge After Thoracolumbar Spine Surgery With Ensemble Machine Learning and Validation on a Nationwide Cohort.

BACKGROUND: Delays in postoperative referrals to rehabilitation or skilled nursing facilities contribute toward extended hospital stays. Facilitating more efficient referrals through accurate preoperative prediction algorithms has the potential to reduce unnecessary economic burden and minimize risk of hospital-acquired complications. We develop a robust machine learning algorithm to predict non-home discharge after thoracolumbar spine surgery that generalizes to unseen populations and identifies markers for prediction. METHODS: Retrospective electronic health records were obtained from our single-center data warehouse (SCDW) to identify patients undergoing thoracolumbar spine surgeries between 2008 and 2019 for algorithm development and internal validation. The National Inpatient Sample (NIS) database was queried to identify thoracolumbar surgeries between 2009 and 2017 for out-of-sample validation. Ensemble decision trees were constructed for prediction and area under the receiver operating characteristic curve (AUROC) was used to assess performance. Shapley additive explanations values were derived to identify drivers of non-home discharge for interpretation of algorithm predictions. RESULTS: A total of 5224 cases of thoracolumbar spine surgeries were isolated from the SCDW and 492,312 cases were identified from NIS. The model achieved an AUROC of 0.81 (standard deviation [SD] = 0.01) on the SCDW test set and 0.77 (SD = 0.01) on the nationwide NIS data set, thereby demonstrating robust prediction of non-home discharge across all diverse patient cohorts. Age, total Elixhauser comorbidities, Medicare insurance, weighted Elixhauser score, and female sex were among the most important predictors of non-home discharge. CONCLUSIONS: Machine learning algorithms reliably predict non-home discharge after thoracolumbar spine surgery across single-center and national cohorts and identify preoperative features of importance that elucidate algorithm decision-making.

MIS-TLIF Procedure is Improving With Experience: Systematic Review of the Learning Curve Over the Last Decade.

STUDY DESIGN: This was a systematic review. OBJECTIVE: This review evaluates the minimally invasive transforaminal lumbar interbody fusions (MIS-TLIF) learning curve in the literature and compares outcomes during and after completing the curve. SUMMARY OF BACKGROUND DATA: MIS-TLIF are performed for various spine conditions. Proponents cite improved clinical outcomes while critics highlight the steep learning curve to attain proficiency. METHODS: Literature searches on Medline and Embase utilized relevant subject headings and keywords. Manuscripts reporting learning curve statistics were included. Monotonic trends of operative duration were assessed with Mann-Kendall nonparametric testing. RESULTS: Nine studies met inclusion criteria. Number of patients ranged from 26 to 150 (average 83.2, median of 86). Commonly reported metrics included number of procedures to complete the curve, operative duration, blood loss, ambulation time, length of stay, complication rate, follow-up visual analogue scale (VAS) for back and leg pain, and fusion rate. Various methods were employed to determine number of cases to complete the curve, all involving operative duration. Number of cases ranged from 14 to 44. A significant negative trend for operative duration of cases during the learning curve (τ=-0.733, P =0.039) was found over the years that studies were published. Initial complication rates varied from 6.8% to 23.8%. Initial VAS-back and VAS-leg ranged from 0.8 to 2.9 and 0.5 to 2.3, respectively. While definitions of "good" fusion varied, fusion rates meeting Bridwell grade I or II during the learning curve ranged from 84.0% to 95.2%. CONCLUSIONS: Surgeons in their learning curve have become faster at the MIS-TLIF procedure. Clinical outcomes including postoperative pain and fusion rates showed satisfactory results, but surgeons learning the procedure should take measures to minimize complications in early cases, such as utilizing novel navigation technology or supervision from more experienced surgeons. Learning curve research methodology could benefit from standardization.

Neurosurgical Performance in the First 2 Years of Merit-Based Incentive Payment System: A Descriptive Analysis and Predictors of Receiving Bonus Payments.

BACKGROUND: The merit-based incentive payment system (MIPS) program was implemented to tie Medicare reimbursements to value-based care measures. Neurosurgical performance in MIPS has not yet been described. OBJECTIVE: To characterize neurosurgical performance in the first 2 years of MIPS. METHODS: Publicly available data regarding MIPS performance for neurosurgeons in 2017 and 2018 were queried. Descriptive statistics about physician characteristics, MIPS performance, and ensuing payment adjustments were performed, and predictors of bonus payments were identified. RESULTS: There were 2811 physicians included in 2017 and 3147 in 2018. Median total MIPS scores (99.1 vs 90.4, P < .001) and quality scores (97.9 vs 88.5, P < .001) were higher in 2018 than in 2017. More neurosurgeons (2758, 87.6%) received bonus payments in 2018 than in 2017 (2013, 71.6%). Of the 2232 neurosurgeons with scores in both years, 1347 (60.4%) improved their score. Reporting through an alternative payment model (odds ratio [OR]: 32.3, 95% CI: 16.0-65.4; P < .001) and any practice size larger than 10 (ORs ranging from 2.37 to 10.2, all P < .001) were associated with receiving bonus payments. Increasing years in practice (OR: 0.99; 95% CI: 0.982-0.998, P = .011) and having 25% to 49% (OR: 0.72; 95% CI: 0.53-0.97; P = .029) or ≥50% (OR: 0.48; 95% CI: 0.28-0.82; P = .007) of a physician's patients eligible for Medicaid were associated with lower rates of bonus payments. CONCLUSION: Neurosurgeons performed well in MIPS in 2017 and 2018, although the program may be biased against surgeons who practice in small groups or take care of socially disadvantaged patients.

Intraoperative Navigation in Spine Surgery: Effects on Complications and Reoperations.

INTRODUCTION: Intraoperative navigation during spine surgery improves pedicle screw placement accuracy. However, limited studies have correlated the use of navigation with clinical factors, including operative time and safety. In the present study, we compared the complications and reoperations between surgeries with and without navigation. METHODS: The National Surgical Quality Improvement Project database was queried for posterior cervical and lumbar fusions and deformity surgeries from 2011 to 2018 and divided by navigation use. Patients aged >89 years, patients with deformity aged <25 years, and patients undergoing surgery for tumors, fractures, infections, or nonelective indications were excluded. The demographics and perioperative factors were compared via univariate analysis. The outcomes were compared using multivariable logistic regression adjusting for age, sex, body mass index, American Society of Anesthesiologists class, surgical region, and multiple treatment levels. The outcomes were also compared stratifying by revision status. RESULTS: Navigation surgery patients had had higher American Society of Anesthesiologists class (P < 0.0001), more multiple level surgeries (P < 0.0001), and longer operative times (P < 0.0001). The adjusted analysis revealed that navigated lumbar surgery had lower odds of complications (odds ratio [OR], 0.82; 95% confidence interval [CI], 0.77-0.90; P < 0.0001), blood transfusion (OR, 0.79; 95% CI, 0.72-0.87; P < 0.0001), and wound debridement and/or drainage (OR, 0.66; 95% CI, 0.44-0.97; P = 0.04) compared with non-navigated lumbar surgery. Navigated cervical fusions had increased odds of transfusions (OR, 1.53; 95% CI, 1.06-2.23; P = 0.02). Navigated primary fusion had decreased odds of complications (OR, 0.91; 95% CI, 0.85-0.98; P = 0.01). However, no differences were found in revisions (OR, 0.89; 95% CI, 0.69-1.14; P = 0.34). CONCLUSIONS: Navigated surgery patients experienced longer operations owing to a combination of the time required for navigation, more multilevel procedures, and a larger comorbidity burden, without differences in the incidence of infection. Fewer complications and wound washouts were required for navigated lumbar surgery owing to a greater proportion percentage of minimally invasive cases. The combined use of navigation and minimally invasive surgery might benefit patients with the proper indications.

Rescue therapy for vasospasm following aneurysmal subarachnoid hemorrhage: a propensity score-matched analysis with machine learning.

OBJECTIVE: Rescue therapies have been recommended for patients with angiographic vasospasm (aVSP) and delayed cerebral ischemia (DCI) following subarachnoid hemorrhage (SAH). However, there is little evidence from randomized clinical trials that these therapies are safe and effective. The primary aim of this study was to apply game theory-based methods in explainable machine learning (ML) and propensity score matching to determine if rescue therapy was associated with better 3-month outcomes following post-SAH aVSP and DCI. The authors also sought to use these explainable ML methods to identify patient populations that were more likely to receive rescue therapy and factors associated with better outcomes after rescue therapy. METHODS: Data for patients with aVSP or DCI after SAH were obtained from 8 clinical trials and 1 observational study in the Subarachnoid Hemorrhage International Trialists repository. Gradient boosting ML models were constructed for each patient to predict the probability of receiving rescue therapy and the 3-month Glasgow Outcome Scale (GOS) score. Favorable outcome was defined as a 3-month GOS score of 4 or 5. Shapley Additive Explanation (SHAP) values were calculated for each patient-derived model to quantify feature importance and interaction effects. Variables with high SHAP importance in predicting rescue therapy administration were used in a propensity score-matched analysis of rescue therapy and 3-month GOS scores. RESULTS: The authors identified 1532 patients with aVSP or DCI. Predictive, explainable ML models revealed that aneurysm characteristics and neurological complications, but not admission neurological scores, carried the highest relative importance rankings in predicting whether rescue therapy was administered. Younger age and absence of cerebral ischemia/infarction were invariably linked to better rescue outcomes, whereas the other important predictors of outcome varied by rescue type (interventional or noninterventional). In a propensity score-matched analysis guided by SHAP-based variable selection, rescue therapy was associated with higher odds of 3-month GOS scores of 4-5 (OR 1.63, 95% CI 1.22-2.17). CONCLUSIONS: Rescue therapy may increase the odds of good outcome in patients with aVSP or DCI after SAH. Given the strong association between cerebral ischemia/infarction and poor outcome, trials focusing on preventative or therapeutic interventions in these patients may be most able to demonstrate improvements in clinical outcomes. Insights developed from these models may be helpful for improving patient selection and trial design.