Machine learning-driven identification of novel patient factors for prediction of major complications after posterior cervical spinal fusion.

PURPOSE: Posterior cervical fusion is associated with increased rates of complications and readmission when compared to anterior fusion. Machine learning (ML) models for risk stratification of patients undergoing posterior cervical fusion remain limited. We aim to develop a novel ensemble ML algorithm for prediction of major perioperative complications and readmission after posterior cervical fusion and identify factors important to model performance. METHODS: This is a retrospective cohort study of adults who underwent posterior cervical fusion at non-federal California hospitals between 2015 and 2017. The primary outcome was readmission or major complication. We developed an ensemble model predicting complication risk using an automated ML framework. We compared performance with standard ML models and logistic regression (LR), ranking contribution of included variables to model performance. RESULTS: Of the included 6822 patients, 18.8% suffered a major complication or readmission. The ensemble model demonstrated slightly superior predictive performance compared to LR and standard ML models. The most important features to performance include sex, malignancy, pneumonia, stroke, and teaching hospital status. Seven of the ten most important features for the ensemble model were markedly less important for LR. CONCLUSION: We report an ensemble ML model for prediction of major complications and readmission after posterior cervical fusion with a modest risk prediction advantage compared to LR and benchmark ML models. Notably, the features most important to the ensemble are markedly different from those for LR, suggesting that advanced ML methods may identify novel prognostic factors for adverse outcomes after posterior cervical fusion.

Impact of the COVID-19 Pandemic Hiatus From Sports Activities on Injuries Observed Among Division I NCAA Athletes.

Background: The coronavirus disease 2019 (COVID-19) pandemic significantly disrupted athletic activities, including those in the Pacific 12 (Pac-12) Conference of the National Collegiate Athletic Association. It is currently unknown how the disruption in training and competition impacted athletes' risk of injury upon resumption of activities. Purpose: To describe and compare the rate, timing, mechanism, and severity of injuries among collegiate athletes across multiple sports in the Pac-12 Conference before and after the COVID-19 pandemic-associated hiatus of intercollegiate athletic activities. Study Design: Descriptive epidemiology study. Methods: Descriptive and injury data among intercollegiate athletes from both the season before the hiatus and the season after the hiatus were acquired from the Pac-12 Health Analytics Program database. Injury elements (timing of injury onset, injury severity, mechanism, recurrence, outcome, need for procedural intervention, and event segment during which the injury took place) were compared by time using the chi-square test and a multivariate logistic regression model. Subgroup analyses were performed on knee and shoulder injuries among athletes participating in sports with traditionally high rates of knee and shoulder injuries. Results: A total of 12,319 sports-related injuries across 23 sports were identified, with 7869 pre-hiatus injuries and 4450 post-hiatus injuries. There was no difference in the overall incidence of injury between the pre-hiatus and post-hiatus seasons. However, the proportion of noncontact injuries was higher in the post-hiatus season for football, baseball, and softball players, and the proportion of nonacute injuries in the post-hiatus season was higher among football, basketball, and rowing athletes. Finally, the overall proportion of injuries sustained by football players in the final 25% of competition or practice was higher in the post-hiatus season. Conclusion: Athletes competing in the post-hiatus season were observed to have higher rates of noncontact injuries and injuries sustained in the final 25% of competition. This study demonstrates that the COVID-19 pandemic has had varied effects on athletes from different sports, suggesting that many factors must be considered when designing return-to-sports programs for athletes after an extended absence from organized training.

Assessment of Characteristics and Methodological Quality of the Top 50 Most Cited Articles on Platelet-Rich Plasma in Musculoskeletal Medicine.

Background: The wide range of clinical applications and controversial scientific evidence associated with platelet-rich plasma (PRP) therapy in musculoskeletal medicine requires an examination of the most commonly cited studies within this field. Purpose: To identify the 50 most cited articles on PRP, assess their study design, and determine any correlations between the number of citations and level of evidence (LoE) or methodological quality. Study Design: Cross-sectional study. Methods: The Web of Science database was queried to identify the top 50 most cited articles on PRP in orthopaedic surgery. Bibliometric characteristics, number of citations, and LoE were recorded. Methodological quality was evaluated using the Modified Coleman Methodology Score (MCMS), Methodological Index for Non-randomized Studies (MINORS), and Minimum Information for Studies Evaluating Biologics in Orthopaedics (MIBO). The Pearson correlation coefficient and Spearman correlation coefficient (r S) were used to determine the degree of correlation between the number of citations or citation density and LoE, MCMS, MINORS score, and MIBO score. Student t tests were performed for 2-group comparisons. Results: The top 50 articles were published between 2005 and 2016 in 21 journals. The mean number of citations and citation density were 241 ± 94 (range, 151-625) and 23 ± 8, respectively, and the mean LoE was 2.44 ± 1.67, with 15 studies classified as LoE 1. The mean MCMS, MINORS score, and MIBO score were 66.9 ± 12.6, 16 ± 4.7, and 12.4 ± 3.7, respectively. No correlation was observed between the number of citations or citation density and LoE, MCMS, MINORS score, and MIBO score. A significant difference (P = .02) was noted in LoE in articles from the United States (3.56 ± 1.7) versus outside the United States (2 ± 1.5). Seven of the 8 in vivo studies were published between 2005 and 2010, whereas 19 of the 25 clinical outcome investigations were published between 2011 and 2016. Studies that were published more recently were found to significantly correlate with number of citations (r S = -0.38; P = .01), citation density (r S = 0.36; P = .01), and higher LoE (r S = 0.47; P = .01). Conclusion: The top 50 most cited articles on PRP consisted of high LoE and fair methodological quality. There was a temporal shift in research from in vivo animal studies toward investigations focused on clinical outcomes.

Reliability of MRI Detection of Kaplan Fiber Injury in Pediatric and Adolescent Patients with ACL Tears.

Background: While studies have described Kaplan fiber (KF) injury in up to 60% of adults with anterior cruciate ligament (ACL) tears, the incidence of KF injury in the pediatric and adolescent population remains unknown. Purpose: To (1) determine the reliability of using magnetic resonance imaging (MRI) to identify KF injury in the pediatric and adolescent population and (2) define the incidence of KF injury in these patients with acute ACL injuries. Study Design: Cohort study (diagnosis); Level of evidence, 3. Methods: The authors retrospectively identified patients ≤18 years of age who underwent ACL reconstruction for acute tears between 2013 and 2020. All preoperative MRI scans were reviewed independently and in a blinded fashion by 2 musculoskeletal radiologists, who noted the presence of the KF complex and any evidence of injury; interrater reliability was assessed. Patient characteristics, time from injury to MRI, laterality, and concomitant ligamentous or meniscal injuries were recorded, and associations between patient or injury characteristics and KF integrity on MRI were assessed. Results: In total, 51 patients (mean age, 14.9 years) met the inclusion criteria. Of these, 27 patients were female and 31 sustained an injury to the right knee. With respect to KF integrity, radiologist 1 visualized KF injury in 29% of patients, while radiologist 2 visualized KF injury in 35% of patients. In 12% of cases for radiologist 1 and 6% of cases for radiologist 2, KFs were unable to be visualized at all. The overall percentage agreement between the 2 radiologists was 76.5% with a kappa statistic of 0.57 (moderate agreement). There were no significant associations found between the presence of KF injury and patient age, sex, laterality, body mass index, concomitant ligamentous injury, or meniscal injury. However, visualization of KF injury on MRI was associated with a shorter time from index injury to MRI (15 days vs 23 days; P = .044). Conclusion: Approximately one-third of pediatric and adolescent patients who underwent ACL reconstruction were found to have KF injuries. Standard preoperative MRI scans can reliably be used to visualize KF injury in the majority of pediatric and adolescent patients with ACL tears, especially when the MRI is performed in the acute setting.

A Risk Calculator for the Prediction of C5 Nerve Root Palsy After Instrumented Cervical Fusion.

BACKGROUND: C5 palsy is a common postoperative complication after cervical fusion and is associated with increased health care costs and diminished quality of life. Accurate prediction of C5 palsy may allow for appropriate preoperative counseling and risk stratification. We primarily aim to develop an algorithm for the prediction of C5 palsy after instrumented cervical fusion and identify novel features for risk prediction. Additionally, we aim to build a risk calculator to provide the risk of C5 palsy. METHODS: We identified adult patients who underwent instrumented cervical fusion at a tertiary care medical center between 2013 and 2020. The primary outcome was postoperative C5 palsy. We developed ensemble machine learning, standard machine learning, and logistic regression models predicting the risk of C5 palsy-assessing discrimination and calibration. Additionally, a web-based risk calculator was built with the best-performing model. RESULTS: A total of 1024 patients were included, with 52 cases of C5 palsy. The ensemble model was well-calibrated and demonstrated excellent discrimination with an area under the receiver-operating characteristic curve of 0.773. The following features were the most important for ensemble model performance: diabetes mellitus, bipolar disorder, C5 or C4 level, surgical approach, preoperative non-motor neurologic symptoms, degenerative disease, number of fused levels, and age. CONCLUSIONS: We report a risk calculator that generates patient-specific C5 palsy risk after instrumented cervical fusion. Individualized risk prediction for patients may facilitate improved preoperative patient counseling and risk stratification as well as potential intraoperative mitigating measures. This tool may also aid in addressing potentially modifiable risk factors such as diabetes and obesity.

Introducing a Novel Combined Acetabuloplasty and Chondroplasty Technique for the Treatment of Developmental Dysplasia of the Hip.

BACKGROUND: The aim of the treatment of developmental dysplasia of the hip (DDH) is to maintain a concentric reduction. We describe a novel approach to treat DDH that involves improvement of cartilaginous acetabular coverage, involves the preservation of the secondary ossification center of the acetabulum, and is adjunctive to early open reduction. METHODOLOGY: Thirty-nine children (40 hips) aged six to 18 months were included in the study. Open reduction with chondroplasty was performed during the same surgery. Patients were followed up for 15 years with both clinical and radiological assessments. At the final follow-up, all patients were graded as good or excellent according to Severin's classification. RESULTS: The mean age at reduction was 11.9 months (range: 8-16). The mean preoperative acetabular index (AI) was 43.43 (range: 40-48). After the operation, mean AI decreased to 16.97 (P < 0.0001, 95% confidence interval (CI) = 16.24-17.70). AI improved significantly during growth (mean AI changes 13.50, P < 0.0001, 95% CI = 12.65-14.34). The mean lateral center-edge (CE) angle at skeletal maturity was 32.94° (SD = 4.16°). Mild avascular necrosis (AVN) was observed in two hips with involvement of the epiphysis and was of Kalamchi grade 1. CONCLUSION: Chondroplasty in conjunction with open reduction can yield a concentric reduction with improved acetabular coverage that facilitates acetabular remodeling that is sustained until skeletal maturity. Prompt correction through this procedure may help to improve the development of the hip and lead to near normal function as demonstrated by improved mean AI and Severin scores at the last follow-up. With low complication and reoperation rates, this procedure could be considered as a surgical treatment option for DDH in patients between the age of six and 18 months.

Prediction of Major Complications and Readmission After Lumbar Spinal Fusion: A Machine Learning-Driven Approach.

BACKGROUND: Given the significant cost and morbidity of patients undergoing lumbar fusion, accurate preoperative risk-stratification would be of great utility. We aim to develop a machine learning model for prediction of major complications and readmission after lumbar fusion. We also aim to identify the factors most important to performance of each tested model. METHODS: We identified 38,788 adult patients who underwent lumbar fusion at any California hospital between 2015 and 2017. The primary outcome was major perioperative complication or readmission within 30 days. We build logistic regression and advanced machine learning models: XGBoost, AdaBoost, Gradient Boosting, and Random Forest. Discrimination and calibration were assessed using area under the receiver operating characteristic curve and Brier score, respectively. RESULTS: There were 4470 major complications (11.5%). The XGBoost algorithm demonstrates the highest discrimination of the machine learning models, outperforming regression. The variables most important to XGBoost performance include angina pectoris, metastatic cancer, teaching hospital status, history of concussion, comorbidity burden, and workers' compensation insurance. Teaching hospital status and concussion history were not found to be important for regression. CONCLUSIONS: We report a machine learning algorithm for prediction of major complications and readmission after lumbar fusion that outperforms logistic regression. Notably, the predictors most important for XGBoost differed from those for regression. The superior performance of XGBoost may be due to the ability of advanced machine learning methods to capture relationships between variables that regression is unable to detect. This tool may identify and address potentially modifiable risk factors, helping risk-stratify patients and decrease complication rates.

Predictors of Medical Malpractice Outcomes After Spine Surgery: A Comprehensive Analysis From 2010 to 2019.

STUDY DESIGN: Retrospective review of spine surgery malpractice cases. OBJECTIVES: The aim was to compare medical malpractice outcomes among different types of spine surgery and identify predictors of litigation outcomes. SUMMARY OF BACKGROUND DATA: Spine surgery is highly litigious in the United States with data suggesting favorable outcomes for defendant surgeons. However, factor specific data and explanations for plaintiff verdicts are lacking. METHODS: Westlaw legal database was queried for spine surgery malpractice outcomes from 2010 to 2019. Clinical data, reasons for litigation, and legal outcomes were tabulated. Statistical analysis was performed to identify factors associated with litigation outcomes. RESULTS: A total of 257 cases were identified for inclusion. There were 98 noninstrumented and 148 instrumented cases; 110 single-level and 99 multilevel; 83 decompressions, 95 decompression and fusions, and 47 fusion only. In all, 182 (71%) resulted in a defendant verdict, 44 (17%) plaintiff verdict, and 31 (12%) settlement. Plaintiff verdicts resulted in payouts of $2.03 million, while settlements resulted in $1.11 million (P=0.34). Common reasons for litigation were intraoperative error, hardware complication, and improper postoperative management. Cases were more likely to result for the plaintiff if postoperative cauda equina syndrome (55% vs. 26%, P<0.01), a surgical site infection (46% vs. 27%, P=0.03), or other catastrophic injury (40% vs. 26%, P=0.03) occurred. Higher monetary awards were associated with multi versus single-level (median: $2.61 vs. $0.92 million, P=0.01), improper postoperative management cited (median: $2.29 vs. $1.12 million, P=0.04), and permanent neurological deficits ($2.29 vs. $0.78 million, P<0.01). Plaintiff payouts were more likely if defendant specialty was neurosurgery versus orthopedic surgery (33% vs. 18%, P=0.01). CONCLUSIONS: Spine surgery is a litigious field with multiple factors associated with outcomes. Efforts to reduce intraoperative errors and complications may improve patient care and decrease the risk of litigation.

Steep effort discounting of a preferred reward over a freely-available option in prolonged methamphetamine withdrawal in male rats.

RATIONALE: Drug addiction can be described as aberrant allocation of effort toward acquiring drug, despite associated costs. It is unclear if this behavioral pattern results from an overvaluation of reward or to an altered sensitivity to costs. OBJECTIVE: Present experiments assessed reward sensitivity and effortful choice in rats following 1 week of withdrawal from methamphetamine (mAMPH). METHODS: Rats were treated with either saline or an escalating dose mAMPH regimen, then tested after a week without the drug. In experiment 1, rats were given a free choice between water and various concentrations of sucrose solution to assess general reward sensitivity. In experiment 2, rats were presented with a choice between lever-pressing for sucrose pellets on a progressive ratio schedule or consuming freely-available chow. RESULTS: In experiment 1, we found no differences in sucrose preference between mAMPH- and saline-pretreated rats. In experiment 2, when selecting between two options, mAMPH-pretreated rats engaged in less lever-pressing for sucrose pellets (p < 0.01) and switched from this preferred reward to the chow sooner than saline-pretreated rats (p < 0.05). This effect was not consistent with general reward devaluation or loss of motivation. CONCLUSIONS: These findings demonstrate that mAMPH exposure and withdrawal lead to steeper discounting of reward value by effort, an effect that is consistent with the effect of mAMPH on discounting by delay, and which may reflect an underlying shared mechanism.

Post-training depletions of basolateral amygdala serotonin fail to disrupt discrimination, retention, or reversal learning.

In goal-directed pursuits, the basolateral amygdala (BLA) is critical in learning about changes in the value of rewards. BLA-lesioned rats show enhanced reversal learning, a task employed to measure the flexibility of response to changes in reward. Similarly, there is a trend for enhanced discrimination learning, suggesting that BLA may modulate formation of stimulus-reward associations. There is a parallel literature on the importance of serotonin (5HT) in new stimulus-reward and reversal learning. Recent postulations implicate 5HT in learning from punishment. Whereas, dopaminergic involvement is critical in behavioral activation and reinforcement, 5HT may be most critical for aversive processing and behavioral inhibition, complementary cognitive processes. Given these findings, a 5HT-mediated mechanism in BLA may mediate the facilitated learning observed previously. The present study investigated the effects of selective 5HT lesions in BLA using 5,7-dihydroxytryptamine (5,7-DHT) vs. infusions of saline (Sham) on discrimination, retention, and deterministic reversal learning. Rats were required to reach an 85% correct pairwise discrimination and single reversal criterion prior to surgery. Postoperatively, rats were then tested on the (1) retention of the pretreatment discrimination pair, (2) discrimination of a novel pair, and (3) reversal learning performance. We found statistically comparable preoperative learning rates between groups, intact postoperative retention, and unaltered novel discrimination and reversal learning in 5,7-DHT rats. These findings suggest that 5HT in BLA is not required for formation and flexible adjustment of new stimulus-reward associations when the strategy to efficiently solve the task has already been learned. Given the complementary role of orbitofrontal cortex in reward learning and its interconnectivity with BLA, these findings add to the list of dissociable mechanisms for BLA and orbitofrontal cortex in reward learning.