Systemic inaccuracies in the National Surgical Quality Improvement Program database: Implications for accuracy and validity for neurosurgery outcomes research

A machine learning-based approach for individualized prediction of short-term outcomes after anterior cervical corpectomy

STUDY DESIGN

A retrospective machine learning (ML) classification study for prognostic modeling after anterior cervical corpectomy (ACC).

PURPOSE

To evaluate the effectiveness of ML in predicting ACC outcomes and develop an accessible, user-friendly tool for this purpose.

OVERVIEW OF LITERATURE

Based on our literature review, no study has examined the capability of ML algorithms to predict major shortterm ACC outcomes, such as prolonged length of hospital stay (LOS), non-home discharge, and major complications.

METHODS

The American College of Surgeons' National Surgical Quality Improvement Program database was used to identify patients who underwent ACC. Prolonged LOS, non-home discharges, and major complications were assessed as the outcomes of interest. ML models were developed with the TabPFN algorithm and integrated into an open-access website to predict these outcomes.

RESULTS

The models for predicting prolonged LOS, non-home discharges, and major complications demonstrated mean areas under the receiver operating characteristic curve (AUROC) of 0.802, 0.816, and 0.702, respectively. These findings highlight the discriminatory capacities of the models: fair (AUROC >0.7) for differentiating patients with major complications from those without, and good (AUROC >0.8) for distinguishing between those with and without prolonged LOS and non-home discharges. According to the SHapley Additive Explanations analysis, single- versus multiple-level surgery, age, body mass index, preoperative hematocrit, and American Society of Anesthesiologists physical status repetitively emerged as the most important variables for each outcome.

CONCLUSIONS

This study has considerably enhanced the prediction of postoperative results after ACC surgery by implementing advanced ML techniques. A major contribution is the creation of an accessible web application, highlighting the practical value of the developed models. Our findings imply that ML can serve as an invaluable supplementary tool to stratify patient risk for this procedure and can predict diverse postoperative adverse outcomes.

Comparison of outcomes in small bowel surgery for Crohn's disease: a retrospective NSQIP review

INTRODUCTION

Despite advances in medical therapy, approximately 33% of Crohn's disease (CD) patients will need surgery within 5 years after initial diagnosis. Several surgical approaches to CD have been proposed including small bowel resection, strictureplasty, and combined surgery with resection plus strictureplasty. Here, we utilize the American College of Surgeons (ACS) national surgical quality registry (NSQIP) to perform a comprehensive analysis of 30-day outcomes between these three surgical approaches for CD.

METHODS

The authors queried the ACS-NSQIP database between 2015 and 2020 for all patients undergoing open or laparoscopic resection of small bowel or strictureplasty for CD using CPT and IC-CM 10. Outcomes of interest included length of stay, discharge disposition, wound complications, 30-day related readmission, and reoperation.

RESULTS

A total of 2578 patients were identified; 87% of patients underwent small bowel resection, 5% resection with strictureplasty, and 8% strictureplasty alone. Resection plus strictureplasty (combined surgery) was associated with the longest operative time (p = 0.002). Patients undergoing small bowel resection had the longest length of hospital stay (p = 0.030) and the highest incidence of superficial/deep wound infection (44%, p = 0.003) as well as the highest incidence of sepsis (3.5%, p = 0.03). Small bowel resection was found to be associated with higher odds of wound complication compared to combined surgery (OR 2.09, p = 0.024) and strictureplasty (1.9, p = 0.005).

CONCLUSION

Our study shows that various surgical approaches for CD are associated with comparable outcomes in 30-day related reoperation and readmission, or disposition following surgery between all three surgical approaches. However, small bowel resection displayed higher odds of developing post-operative wound complications.

Total Elbow Arthroplasty Versus Open Reduction and Internal Fixation for Distal Humerus Fractures: A Propensity Score Matched Analysis of 30-Day Postoperative Complications

Introduction

Open reduction and internal fixation (ORIF) is an established surgical procedure for distal humeral fractures; however, total elbow arthroplasty (TEA) has become an increasingly popular alternative for elderly patients with these injuries. Using a large sample of recent patient data, this study compares the rates of short-term complications between ORIF and TEA and evaluates complication risk factors.

Methods

Patients who underwent primary TEA or ORIF from 2012 to 2021 were identified by Current Procedural Terminology codes in the American College of Surgeons National Surgical Quality Improvement Program database. Propensity score matching controlled for demographic and comorbid differences. The rates of 30-day postoperative complications were compared.

Results

A total of 1539 patients were identified, with 1365 (88.7%) and 174 (11.3%) undergoing ORIF and TEA, respectively. Patients undergoing TEA were older on average (ORIF: 56.2 ± 19.8 years, TEA: 74.3 ± 11.0 years, P < .001). 348 patients were included in the matched analysis, with 174 patients in each group. TEA was associated with an increased risk for postoperative transfusion (OR = 6.808, 95% CI = 1.355 - 34.199, P = .020). There were no significant differences in any adverse event (AAE) between procedures (P = .259). A multivariate analysis indicated age was the only independent risk factor for the development of AAE across both groups (OR = 1.068, 95% CI = 1.011 - 1.128, P = .018).

Conclusion

The risk of short-term complications within 30-days of ORIF or TEA procedures are similar when patient characteristics are controlled. TEA, however, was found to increase the risk of postoperative transfusions. Risks associated with increasing patient age should be considered prior to either procedure. These findings suggest that long-term functional outcomes can be prioritized in the management of distal humerus fractures.

Frequency and Timing of Postoperative Complications After Outpatient Total Hip Arthroplasty

Background

Although there have been several studies describing risk factors for complications after outpatient total hip arthroplasty (THA), data describing the timing of such complications is lacking.

Methods

Patients who underwent outpatient or inpatient primary THA were identified in the 2012-2019 National Surgical Quality Improvement Program database. For 9 different 30-day complications, the median postoperative day of diagnosis was determined. Multivariable regressions were used to compare the risk of each complication between outpatient vs inpatient groups. Multivariable Cox proportional hazards modeling was used to evaluate the differences in the timing of each adverse event between the groups.

Results

After outpatient THA, the median day of diagnosis for readmission was 12.5 (interquartile range 5-22), surgical site infection 15 (2-21), urinary tract infection 13.5 (6-19.5), deep vein thrombosis 13 (8-21), myocardial infarction 4.5 (1-7), pulmonary embolism 15 (8-25), sepsis 16 (9-26), stroke 2 (0-7), and pneumonia 6.5 (3-10). On multivariable regressions, outpatients had a lower relative risk (RR) of readmission (RR = 0.73), surgical site infection (RR = 0.72), and pneumonia (RR = 0.1), all P < .05. On multivariable cox proportional hazards modeling, there were no statistically significant differences in the timing of each complication between outpatient vs inpatient procedures (P > .05).

Conclusions

The timing of complications after outpatient THA was similar to inpatient procedures. Consideration should be given to lowering thresholds for diagnostic testing after outpatient THA for each complication during the at-risk time periods identified here. Although extremely rare, this is especially important for catastrophic adverse events, which tend to occur early after discharge.

Machine learning models on a web application to predict short-term postoperative outcomes following anterior cervical discectomy and fusion

BACKGROUND

The frequency of anterior cervical discectomy and fusion (ACDF) has increased up to 400% since 2011, underscoring the need to preoperatively anticipate adverse postoperative outcomes given the procedure's expanding use. Our study aims to accomplish two goals: firstly, to develop a suite of explainable machine learning (ML) models capable of predicting adverse postoperative outcomes following ACDF surgery, and secondly, to embed these models in a user-friendly web application, demonstrating their potential utility.

METHODS

We utilized data from the National Surgical Quality Improvement Program database to identify patients who underwent ACDF surgery. The outcomes of interest were four short-term postoperative adverse outcomes: prolonged length of stay (LOS), non-home discharges, 30-day readmissions, and major complications. We utilized five ML algorithms - TabPFN, TabNET, XGBoost, LightGBM, and Random Forest - coupled with the Optuna optimization library for hyperparameter tuning. To bolster the interpretability of our models, we employed SHapley Additive exPlanations (SHAP) for evaluating predictor variables' relative importance and used partial dependence plots to illustrate the impact of individual variables on the predictions generated by our top-performing models. We visualized model performance using receiver operating characteristic (ROC) curves and precision-recall curves (PRC). Quantitative metrics calculated were the area under the ROC curve (AUROC), balanced accuracy, weighted area under the PRC (AUPRC), weighted precision, and weighted recall. Models with the highest AUROC values were selected for inclusion in a web application.

RESULTS

The analysis included 57,760 patients for prolonged LOS [11.1% with prolonged LOS], 57,780 for non-home discharges [3.3% non-home discharges], 57,790 for 30-day readmissions [2.9% readmitted], and 57,800 for major complications [1.4% with major complications]. The top-performing models, which were the ones built with the Random Forest algorithm, yielded mean AUROCs of 0.776, 0.846, 0.775, and 0.747 for predicting prolonged LOS, non-home discharges, readmissions, and complications, respectively.

CONCLUSIONS

Our study employs advanced ML methodologies to enhance the prediction of adverse postoperative outcomes following ACDF. We designed an accessible web application to integrate these models into clinical practice. Our findings affirm that ML tools serve as vital supplements in risk stratification, facilitating the prediction of diverse outcomes and enhancing patient counseling for ACDF.

Predicting 30-Day Non-Seizure Outcomes Following Temporal Lobectomy with Personalized Machine Learning Models

BACKGROUND

Temporal lobe epilepsy is the most common reason behind drug-resistant seizures and temporal lobectomy (TL) is performed after all other efforts have been taken for a Temporal lobe epilepsy. Our study aims to develop multiple machine learning (ML) models capable of predicting postoperative outcomes following TL surgery.

METHODS

Data from the American College of Surgeons National Surgical Quality Improvement Program database identified patients who underwent TL surgery. We focused on 3 outcomes: prolonged length of stay (LOS), nonhome discharges, and 30-day readmissions. Six ML algorithms, TabPFN, XGBoost, LightGBM, Support Vector Machine, Random Forest, and Logistic Regression, coupled with the Optuna optimization library for hyperparameter tuning, were tested. Models with the highest area under the receiver operating characteristic (AUROC) values were included in the web application. SHapley Additive exPlanations was used to evaluate importance of predictor variables.

RESULTS

Our analysis included 423 patients. Of these patients, 111 (26.2%) experienced prolonged LOS, 33 (7.8%) had nonhome discharges, and 29 (6.9%) encountered 30-day readmissions. The top-performing models for each outcome were those built with the Random Forest algorithm. The Random Forest models yielded AUROCs of 0.868, 0.804, and 0.742 in predicting prolonged LOS, nonhome discharges, and 30-day readmissions, respectively.

CONCLUSIONS

Our study uses ML to forecast adverse postoperative outcomes following TL. We developed accessible predictive models that enhance prognosis prediction for TL surgery. Making ML models available for this purpose represents a significant advancement in shifting toward a more patient-centric, data-driven paradigm.

Personalized Prognosis with Machine Learning Models for Predicting In-Hospital Outcomes Following Intracranial Meningioma Resections

BACKGROUND

Meningiomas display diverse biological traits and clinical behaviors, complicating patient outcome prediction. This heterogeneity, along with varying prognoses, underscores the need for a precise, personalized evaluation of postoperative outcomes.

METHODS

Data from the American College of Surgeons National Surgical Quality Improvement Program database identified patients who underwent intracranial meningioma resections from 2014 to 2020. We focused on 5 outcomes: prolonged LOS, nonhome discharges, 30-day readmissions, unplanned reoperations, and major complications. Six machine learning algorithms, including TabPFN, TabNet, XGBoost, LightGBM, Random Forest, and Logistic Regression, coupled with the Optuna optimization library for hyperparameter tuning, were tested. Models with the highest area under the receiver operating characteristic (AUROC) values were included in the web application. SHapley Additive exPlanations were used to evaluate the importance of predictor variables.

RESULTS

Our analysis included 7000 patients. Of these patients, 1658 (23.7%) had prolonged LOS, 1266 (18.1%) had nonhome discharges, 573 (8.2%) had 30-day readmission, 253 (3.6%) had unplanned reoperation, and 888 (12.7%) had major complications. Performance evaluation indicated that the top-performing models for each outcome were the models built with LightGBM and Random Forest algorithms. The LightGBM models yielded AUROCs of 0.842 and 0.846 in predicting prolonged LOS and nonhome discharges, respectively. The Random Forest models yielded AUROCs of 0.717, 0.76, and 0.805 in predicting 30-day readmissions, unplanned reoperations, and major complications, respectively.

CONCLUSIONS

The study successfully demonstrated the potential of machine learning models in predicting short-term adverse postoperative outcomes after meningioma resections. This approach represents a significant step forward in personalizing the information provided to meningioma patients.

Development of personalized machine learning-based prediction models for short-term postoperative outcomes in patients undergoing cervical laminoplasty

PURPOSE

By predicting short-term postoperative outcomes before surgery, patients undergoing cervical laminoplasty (CLP) surgery could benefit from more accurate patient care strategies that could reduce the likelihood of adverse outcomes. With this study, we developed a series of machine learning (ML) models for predicting short-term postoperative outcomes and integrated them into an open-source online application.

METHODS

National surgical quality improvement program database was utilized to identify individuals who have undergone CLP surgery. The investigated outcomes were prolonged length of stay (LOS), non-home discharges, 30-day readmissions, unplanned reoperations, and major complications. ML models were developed and implemented on a website to predict these three outcomes.

RESULTS

A total of 1740 patients that underwent CLP were included in the analysis. Performance evaluation indicated that the top-performing models for each outcome were the models built with TabPFN and LightGBM algorithms. The TabPFN models yielded AUROCs of 0.830, 0.847, and 0.858 in predicting non-home discharges, unplanned reoperations, and major complications, respectively. The LightGBM models yielded AUROCs of 0.812 and 0.817 in predicting prolonged LOS, and 30-day readmissions, respectively.

CONCLUSION

The potential of ML approaches to predict postoperative outcomes following spine surgery is significant. As the volume of data in spine surgery continues to increase, the development of predictive models as clinically relevant decision-making tools could significantly improve risk assessment and prognosis. Here, we present an accessible predictive model for predicting short-term postoperative outcomes following CLP intended to achieve the stated objectives.

Interpretable machine learning models to predict short-term postoperative outcomes following posterior cervical fusion

By predicting short-term postoperative outcomes before surgery, patients who undergo posterior cervical fusion (PCF) surgery may benefit from more precise patient care plans that reduce the likelihood of unfavorable outcomes. We developed machine learning models for predicting short-term postoperative outcomes and incorporate these models into an open-source web application in this study. The American College of Surgeons National Surgical Quality Improvement Program database was used to identify patients who underwent PCF surgery. Prolonged length of stay, non-home discharges, and readmissions were the three outcomes that were investigated. To predict these three outcomes, machine learning models were developed and incorporated into an open access web application. A total of 6277 patients that underwent PCF surgery were included in the analysis. The most accurately predicted outcome in terms of the area under the receiver operating characteristic curve (AUROC) was the non-home discharges with a mean AUROC of 0.812, and the most accurately predicting algorithm in terms of AUROC was the LightGBM algorithm with a mean AUROC of 0.766. The following URL will take users to the open access web application created to provide predictions for individual patients based on their characteristics: https://huggingface.co/spaces/MSHS-Neurosurgery-Research/NSQIP-PCF. Machine learning techniques have a significant potential for predicting postoperative outcomes following PCF surgery. The development of predictive models as clinically useful decision-making tools may significantly improve risk assessment and prognosis as the amount of data in spinal surgery keeps growing. Here, we present predictive models for PCF surgery that are meant to accomplish the aforementioned goals and make them publicly available.

A quantitative assessment of the number of disease foci in papillary thyroid cancer

AIM

Multifocality is a frequent feature of papillary thyroid carcinoma (PTC). Its prognostic value is controversial although national guidelines recommend treatment intensification if present. However, multifocality is not a binary but discrete variable. This study aimed to examine the association between increasing number of foci and risk of recurrence following treatment.

METHODS

577 patients with PTC were identified with median follow-up of 61 months. Number of foci were taken from pathology reports. Log-rank test was used to assess significance. Multivariate analysis was performed and Hazard Ratios were calculated.

RESULTS

Of 577 patients, 206(35%) had multifocal disease and 36(6%) recurred. 133(23%), 89(15%) and 61(11%) had 3+, 4+ or 5+ foci respectively. The 5-year RFS stratified by number of foci was 95%v93% for 2+foci (p = 0.616), 95%v96% for 3+foci (p = 0.198) and 89%v96% for 4+foci (p = 0.022). The presence of 4 foci was associated with an over 2-fold risk of recurrence (HR 2.296, 95% CI 1.106-4.765, p = 0.026) although this was not independent of TNM staging. Of the 206 multifocal patients, 31(5%) had 4+foci as their sole risk factor for treatment intensification.

CONCLUSION

Although multifocality per se does not confer worse outcome in PTC, finding 4+foci is associated with worse outcome and could therefore be appropriate as a cut-off for treatment intensification. In our cohort, 5% of patients had 4+foci as a sole indication for treatment intensification, suggesting that such a cut off could impact clinical management.

Complications Related to Deep Brain Stimulation Lead Implantation: A Single-Surgeon Case Series

BACKGROUND

Deep brain stimulation (DBS) is the mainstay of surgical treatment for movement disorders, yet previous studies have shown widely varying complication rates. Given the elective nature of DBS surgery, minimizing surgical complications is imperative.

OBJECTIVE

To evaluate short-term and long-term complications related to DBS lead implantation surgeries performed by an experienced surgeon and provide an updated benchmark comparison for other DBS centers and alternative therapies.

METHODS

A retrospective chart review of patients who underwent DBS lead implantation surgery by a single surgeon at our institution between 2012 and 2020 was conducted. Demographic and clinical data including surgical complications were collected. A Kaplan-Meier survival analysis was used to evaluate the cumulative risk of lead revision or removal over time. Associations between patient characteristics and various complications were evaluated.

RESULTS

Four hundred fifty-one DBS leads were placed in 255 patients. Thirteen leads and 11 patients required revision. In total, 3.6% (95% CI [1.3%-5.9%]) of patients required revision at 1 year and 4.8% (95% CI [1.9%-7.6%]) at 5 years, with per-lead revision rates of 2.3% (95% CI [0.9%-3.6%]) and 3.3% (95% CI [1.5%-5.1%]), respectively. Less common diagnoses such as Tourette syndrome, post-traumatic tremor, and cluster headache trended toward association with lead revision or removal.

CONCLUSION

DBS performed by an experienced surgeon is associated with extremely low complication rates.

A Machine Learning-Based Online Prediction Tool for Predicting Short-Term Postoperative Outcomes Following Spinal Tumor Resections

Background: Preoperative prediction of short-term postoperative outcomes in spinal tumor patients can lead to more precise patient care plans that reduce the likelihood of negative outcomes. With this study, we aimed to develop machine learning algorithms for predicting short-term postoperative outcomes and implement these models in an open-source web application. Methods: Patients who underwent surgical resection of spinal tumors were identified using the American College of Surgeons, National Surgical Quality Improvement Program. Three outcomes were predicted: prolonged length of stay (LOS), nonhome discharges, and major complications. Four machine learning algorithms were developed and integrated into an open access web application to predict these outcomes. Results: A total of 3073 patients that underwent spinal tumor resection were included in the analysis. The most accurately predicted outcomes in terms of the area under the receiver operating characteristic curve (AUROC) was the prolonged LOS with a mean AUROC of 0.745 The most accurately predicting algorithm in terms of AUROC was random forest, with a mean AUROC of 0.743. An open access web application was developed for getting predictions for individual patients based on their characteristics and this web application can be accessed here: huggingface.co/spaces/MSHS-Neurosurgery-Research/NSQIP-ST. Conclusion: Machine learning approaches carry significant potential for the purpose of predicting postoperative outcomes following spinal tumor resections. Development of predictive models as clinically useful decision-making tools may considerably enhance risk assessment and prognosis as the amount of data in spinal tumor surgery continues to rise.

Association of age, sex and race with prescription of anti-osteoporosis medications following low-energy hip fracture in a retrospective registry cohort

BACKGROUND

Initiation of anti-osteoporosis medications after hip fracture lowers the risk of subsequent fragility fractures. Historical biases of targeting secondary fracture prevention towards certain groups may result in treatment disparities. We examined associations of patient age, sex and race with anti-osteoporosis medication prescription following hip fracture.

METHODS

A cohort of patients with a hip fracture between 2016-2018 was assembled from the American College of Surgeons National Surgical Quality Improvement Program registry. Patients on anti-osteoporosis medications prior to admission were excluded. Multivariable logistic regression was used to determine adjusted associations between patient age, sex and race and their interactions with prescription of anti-osteoporosis medications within 30 days of surgery.

RESULTS

In total, 12,249 patients with a hip fracture were identified with a median age of 82 years (IQR: 73-87), and 67% were female (n = 8,218). Thirty days postoperatively, 26% (n = 3146) of patients had been prescribed anti-osteoporosis medication. A significant interaction between age and sex with medication prescription was observed (p = 0.04). Male patients in their 50s (OR:0.75, 95%CI:0.60-0.92), 60s (OR:0.81, 95%CI:0.70-0.94) and 70s (OR:0.89, 95%CI:0.81-0.97) were less likely to be prescribed anti-osteoporosis medication compared to female patients of the same age. Patients who belonged to minority racial groups were not less likely to receive anti-osteoporosis medications than patients of white race.

INTERPRETATION

Only 26% of patients were prescribed anti-osteoporosis medications following hip fracture, despite consensus guidelines urging early initiation of secondary prevention treatments. Given that prescription varied by age and sex, strategies to prevent disparities in secondary fracture prevention are warranted.

Machine Learning in the Management of Lateral Skull Base Tumors: A Systematic Review

The application of machine learning (ML) techniques to otolaryngology remains a topic of interest and prevalence in the literature, though no previous articles have summarized the current state of ML application to management and the diagnosis of lateral skull base (LSB) tumors. Subsequently, we present a systematic overview of previous applications of ML techniques to the management of LSB tumors. Independent searches were conducted on PubMed and Web of Science between August 2020 and February 2021 to identify the literature pertaining to the use of ML techniques in LSB tumor surgery written in the English language. All articles were assessed in regard to their application task, ML methodology, and their outcomes. A total of 32 articles were examined. The number of articles involving applications of ML techniques to LSB tumor surgeries has significantly increased since the first article relevant to this field was published in 1994. The most commonly employed ML category was tree-based algorithms. Most articles were included in the category of surgical management (13; 40.6%), followed by those in disease classification (8; 25%). Overall, the application of ML techniques to the management of LSB tumor has evolved rapidly over the past two decades, and the anticipated growth in the future could significantly augment the surgical outcomes and management of LSB tumors.

Machine Learning for the Orthopaedic Surgeon: Uses and Limitations

➤

Machine learning is a subset of artificial intelligence in which computer algorithms are trained to make classifications and predictions based on patterns in data. The utilization of these techniques is rapidly expanding in the field of orthopaedic research.

➤

There are several domains in which machine learning has application to orthopaedics, including radiographic diagnosis, gait analysis, implant identification, and patient outcome prediction.

➤

Several limitations prevent the widespread use of machine learning in the daily clinical environment. However, future work can overcome these issues and enable machine learning tools to be a useful adjunct for orthopaedic surgeons in their clinical decision-making.

The accuracy of hip fracture data entered into the national surgical quality improvement program (NSQIP) database

PURPOSE

Internal validation studies of National Surgical Quality Improvement Program (NSQIP) registry data have reported potential inaccuracies. The purpose of this study was to determine the accuracy of hip fracture CPT codes and complications entered into NSQIP for a single participating center.

METHODS

A retrospective study identified patients with a hip fracture CPT code from NSQIP data at a single institution over a two-year period. CPT codes included 27235 (percutaneous fixation of femoral neck fracture (Perc FNFX)), 27236 (open treatment of femoral neck fracture, internal fixation/prosthetic replacement (Open FNFX)), 27244 (open treatment of inter/peri/subtrochanteric femoral fracture with plate (Plate ITFX)), 27245 (treatment of inter/peri/subtrochanteric femoral fracture, with intramedullary implant (IMN ITFX)), and 27125 (hemiarthroplasty (HA)). The institutional medical record was reviewed to determine the accuracy of CPT code and 30-day complication data entered into the registry.

RESULT

12.8% (n = 20/156) of patients had an inaccurate CPT code. The proportion of inaccurate CPT codes varied significantly by procedure: Plate ITFX (76.9%), Open FNFX (13.8%), IMN ITFX (7.0%), and HA (0%) (p < 0.0001). A total of 82 complications were identified in 66 patients via the medical record. 43.9% (n = 36/82) of these complications were not documented in the NSQIP data. The proportion of missing complications varied significantly by type: renal (100%), UTI (53.8%), infection (50%), bleeding (30%), death (25%), respiratory (25%), cardiac (0%), stroke (0%), and VTE (0%) (p < 0.0001).

CONCLUSION

Hip fracture CPT codes and 30-day complication data entered into the NSQIP registry were frequently inaccurate. Studies incorporating NSQIP data should acknowledge these potential limitations of the registry, and future research to validate NSQIP orthopedic data across procedures and institutions is necessary.

LEVEL OF EVIDENCE

LEVEL III: Diagnostic study.

Status epilepticus after intracranial neurosurgery: incidence and risk stratification by perioperative clinical features

OBJECTIVE

Status epilepticus (SE) is associated with significant mortality, cost, and risk of future seizures. In one of the first studies of SE after neurosurgery, the authors assess the incidence, risk factors, and outcome of postneurosurgical SE (PNSE).

METHODS

Neurosurgical admissions from the MarketScan Claims and Encounters database (2007 through 2015) were assessed in a longitudinal cross-sectional sample of privately insured patients who underwent qualifying cranial procedures in the US and were older than 18 years of age. The incidence of early (in-hospital) and late (postdischarge readmission) SE and associated mortality was assessed. Procedural, pathological, demographic, and anatomical covariates parameterized multivariable logistic regression and Cox models. Multivariable logistic regression and Cox proportional hazards models were used to study the incidence of early and late PNSE. A risk-stratification simulation was performed, combining individual predictors into singular risk estimates.

RESULTS

A total of 197,218 admissions (218,217 procedures) were identified. Early PNSE occurred during 637 (0.32%) of 197,218 admissions for cranial neurosurgical procedures. A total of 1045 (0.56%) cases of late PNSE were identified after 187,771 procedure admissions with nonhospice postdischarge follow-up. After correction for comorbidities, craniotomy for trauma, hematoma, or elevated intracranial pressure was associated with increased risk of early PNSE (adjusted OR [aOR] 1.538, 95% CI 1.183-1.999). Craniotomy for meningioma resection was associated with an increased risk of early PNSE compared with resection of metastases and parenchymal primary brain tumors (aOR 2.701, 95% CI 1.388-5.255). Craniotomies for infection or abscess (aHR 1.447, 95% CI 1.016-2.061) and CSF diversion (aHR 1.307, 95% CI 1.076-1.587) were associated with highest risk of late PNSE. Use of continuous electroencephalography in patients with early (p < 0.005) and late (p < 0.001) PNSE rose significantly over the study time period. The simulation regression model predicted that patients at high risk for early PNSE experienced a 1.10% event rate compared with those at low risk (0.07%). Similarly, patients predicted to be at highest risk for late PNSE were significantly more likely to eventually develop late PNSE than those at lowest risk (HR 54.16, 95% CI 24.99-104.80).

CONCLUSIONS

Occurrence of early and late PNSE was associated with discrete neurosurgical pathologies and increased mortality. These data provide a framework for prospective validation of clinical and perioperative risk factors and indicate patients for heightened diagnostic suspicion of PNSE.

Wide range of applications for machine-learning prediction models in orthopedic surgical outcome: a systematic review

Background and purpose - Advancements in software and hardware have enabled the rise of clinical prediction models based on machine learning (ML) in orthopedic surgery. Given their growing popularity and their likely implementation in clinical practice we evaluated which outcomes these new models have focused on and what methodologies are being employed.Material and methods - We performed a systematic search in PubMed, Embase, and Cochrane Library for studies published up to June 18, 2020. Studies reporting on non-ML prediction models or non-orthopedic outcomes were excluded. After screening 7,138 studies, 59 studies reporting on 77 prediction models were included. We extracted data regarding outcome, study design, and reported performance metrics.Results - Of the 77 identified ML prediction models the most commonly reported outcome domain was medical management (17/77). Spinal surgery was the most commonly involved orthopedic subspecialty (28/77). The most frequently employed algorithm was neural networks (42/77). Median size of datasets was 5,507 (IQR 635-26,364). The median area under the curve (AUC) was 0.80 (IQR 0.73-0.86). Calibration was reported for 26 of the models and 14 provided decision-curve analysis.Interpretation - ML prediction models have been developed for a wide variety of topics in orthopedics. Topics regarding medical management were the most commonly studied. Heterogeneity between studies is based on study size, algorithm, and time-point of outcome. Calibration and decision-curve analysis were generally poorly reported.

The association of preoperative blood markers with postoperative readmissions following arthroplasty

AIMS

While preoperative bloodwork is routinely ordered, its value in determining which patients are at risk of postoperative readmission following total knee arthroplasty (TKA) and total hip arthroplasty (THA) is unclear. The objective of this study was to determine which routinely ordered preoperative blood markers have the strongest association with acute hospital readmission for patients undergoing elective TKA and THA.

METHODS

Two population-based retrospective cohorts were assembled for all adult primary elective TKA (n = 137,969) and THA (n = 78,532) patients between 2011 to 2018 across 678 North American hospitals using the American College of Surgeons National Quality Improvement Programme (ACS-NSQIP) registry. Six routinely ordered preoperative blood markers - albumin, haematocrit, platelet count, white blood cell count (WBC), estimated glomerular filtration rate (eGFR), and sodium level - were queried. The association between preoperative blood marker values and all-cause readmission within 30 days of surgery was compared using univariable analysis and multivariable logistic regression adjusted for relevant patient and treatment factors.

RESULTS

The mean TKA age was 66.6 years (SD 9.6) with 62% being females (n = 85,163/137,969), while in the THA cohort the mean age was 64.7 years (SD 11.4) with 54% being female (n = 42,637/78,532). In both cohorts, preoperative hypoalbuminemia (< 35 g/l) was associated with a 1.5- and 1.8-times increased odds of 30-day readmission following TKA and THA, respectively. In TKA patients, decreased eGFR demonstrated the strongest association with acute readmission with a standardized odds ratio of 0.75 per two standard deviations increase (p < 0.0001).

CONCLUSION

In this population level cohort analysis of arthroplasty patients, low albumin demonstrated the strongest association with acute readmission in comparison to five other commonly ordered preoperative blood markers. Identification and optimization of preoperative hypoalbuminemia could help healthcare providers recognize and address at-risk patients undergoing TKA and THA. This is the most comprehensive and rigorous examination of the association between preoperative blood markers and readmission for TKA and THA patients to date. Cite this article: Bone Jt Open 2021;2(6):388-396.

Prediction calculator for nonroutine discharge and length of stay after spine surgery

BACKGROUND CONTEXT

Following spine surgery, delays in referral to rehabilitation facilities leads to increased length of hospital stay (LOS), increases costs, more risk of hospital acquired complications, and decreased patient satisfaction.

PURPOSE

We sought to create a prediction calculator to determine the expected LOS after spine surgery and identify patients most likely to need postoperative nonhome discharge. The goal would be to facilitate earlier referral to rehabilitation and thereby ultimately shorten LOS, reduce costs, and improve patient satisfaction.

STUDY DESIGN

Retrospective.

PATIENT SAMPLE

We retrospectively reviewed all adult patients who underwent spine surgery for all indications between January and June 2018.

OUTCOME MEASURES

Length of stay and discharge disposition.

METHODS

Demographic variables, insurance status, baseline comorbidities, narcotic use, operative characteristics, as well as postoperative length of stay and discharge disposition data were collected. Univariable and multivariable analyses were performed to identify independent predictors of LOS and discharge disposition.

RESULTS

Two hundred fifty-seven patients were included. Mean age was 59 years, 46% were females, and 52% had private insurance vs 7% with Medicaid and 41% with Medicare. The most commonly performed procedure was lumbar fusion (31.9%). Mean LOS after surgery was 4.8 days and 18% had prolonged LOS >7 days. Age, insurance type, marriage status, and surgical procedure were significantly associated with LOS and discharge disposition. The final model had an area under the curve of 89% with good discrimination. A web based calculator was developed: https://jhuspine1.shinyapps.io/RehabLOS/ CONCLUSIONS: This study established a novel pilot calculator to identify those patients most likely to be discharged to rehabilitation facilities and to predict LOS after spine surgery. Our calculator had a high predictive accuracy of 89% compared to others in the literature. With validation this tool may ultimately facilitate streamlining of the postoperative period to shorten LOS, optimize resource utilization, and improve patient care.

Quality of British and American Nationwide Quality of Care and Patient Safety Benchmarking Programs: Case Neurosurgery

BACKGROUND

Multiple nationwide outcome registries are utilized for quality benchmarking between institutions and individual surgeons.

OBJECTIVE

To evaluate whether nationwide quality of care programs in the United Kingdom and United States can measure differences in neurosurgical quality.

METHODS

This prospective observational study comprised 418 consecutive adult patients undergoing elective craniotomy at Helsinki University Hospital between December 7, 2011 and December 31, 2012.We recorded outcome event rates and categorized them according to British Neurosurgical National Audit Programme (NNAP), American National Surgical Quality Improvement Program (NSQIP), and American National Neurosurgery Quality and Outcomes Database (N2QOD) to assess the applicability of these programs for quality benchmarking and estimated sample sizes required for reliable quality comparisons.

RESULTS

The rate of in-hospital major and minor morbidity was 18.7% and 38.0%, respectively, and 30-d mortality rate was 2.4%. The NSQIP criteria identified 96.2% of major but only 38.4% of minor complications. N2QOD performed better, but almost one-fourth (23.2%) of all patients with adverse outcomes, mostly minor, went unnoticed. For NNAP, a sample size of over 4200 patients per surgeon is required to detect a 50.0% increase in mortality rates between surgeons. The sample size required for reliable comparisons between the rates of complications exceeds 600 patients per center per year.

CONCLUSION

The implemented benchmarking programs in the United Kingdom and United States fail to identify a considerable number of complications in a high-volume center. Health care policy makers should be cautious as outcome comparisons between most centers and individual surgeons are questionable if based on the programs.

Reporting Methodology of Neurosurgical Studies Utilizing the American College of Surgeons-National Surgical Quality Improvement Program Database: A Systematic Review and Critical Appraisal

BACKGROUND

Use of large databases such as the American College of Surgeons-National Surgical Quality Improvement Program (ACS-NSQIP) has become increasingly common in neurosurgical research.

OBJECTIVE

To perform a critical appraisal and evaluation of the methodological reporting for studies in neurosurgical literature that utilize the ACS-NSQIP database.

METHODS

We queried Ovid MEDLINE, EMBASE, and PubMed databases for all neurosurgical studies utilizing the ACS-NSQIP. We assessed each study according to number of criteria fulfilled with respect to Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement, REporting of studies Conducted using Observational Routinely-collected Health Data (RECORD) Statement, and Journal of American Medical Association–Surgical Section (JAMA-Surgery) Checklist. A separate analysis was conducted among papers published in core and noncore journals in neurosurgery according to Bradford's law.

RESULTS

A total of 117 studies were included. Median (interquartile range [IQR]) scores for number of fulfilled criteria for STROBE Statement, RECORD Statement, and JAMA-Surgery Checklist were 20 (IQR:19-21), 9 (IQR:8-9), and 6 (IQR:5-6), respectively. For STROBE Statement, RECORD Statement, and JAMA-Surgery Checklist, item 9 (potential sources of bias), item 13 (supplemental information), and item 9 (missing data/sensitivity analysis) had the highest number of studies with no fulfillment among all studies (56, 68, 50%), respectively. When comparing core journals vs noncore journals, no significant difference was found (STROBE, P = .94; RECORD, P = .24; JAMA-Surgery checklist, P = .60).

CONCLUSION

While we observed an overall satisfactory reporting of methodology, most studies lacked mention of potential sources of bias, data cleaning methods, supplemental information, and external validity. Given the pervasive role of national databases and registries for research and health care policy, the surgical community needs to ensure the credibility and quality of such studies that ultimately aim to improve the value of surgical care delivery to patients.

Predicting discharge placement after elective surgery for lumbar spinal stenosis using machine learning methods

PURPOSE

An excessive amount of total hospitalization is caused by delays due to patients waiting to be placed in a rehabilitation facility or skilled nursing facility (RF/SNF). An accurate preoperative prediction of who would need a RF/SNF place after surgery could reduce costs and allow more efficient organizational planning. We aimed to develop a machine learning algorithm that predicts non-home discharge after elective surgery for lumbar spinal stenosis.

METHODS

We used the American College of Surgeons National Surgical Quality Improvement Program to select patient that underwent elective surgery for lumbar spinal stenosis between 2009 and 2016. The primary outcome measure for the algorithm was non-home discharge. Four machine learning algorithms were developed to predict non-home discharge. Performance of the algorithms was measured with discrimination, calibration, and an overall performance score.

RESULTS

We included 28,600 patients with a median age of 67 (interquartile range 58-74). The non-home discharge rate was 18.2%. Our final model consisted of the following variables: age, sex, body mass index, diabetes, functional status, ASA class, level, fusion, preoperative hematocrit, and preoperative serum creatinine. The neural network was the best model based on discrimination (c-statistic = 0.751), calibration (slope = 0.933; intercept = 0.037), and overall performance (Brier score = 0.131).

CONCLUSIONS

A machine learning algorithm is able to predict discharge placement after surgery for lumbar spinal stenosis with both good discrimination and calibration. Implementing this type of algorithm in clinical practice could avert risks associated with delayed discharge and lower costs. These slides can be retrieved under Electronic Supplementary Material.

Little Insights from Big Data: Cerebrospinal Fluid Leak After Skull Base Surgery and the Limitations of Database Research

BACKGROUND

Cerebrospinal fluid (CSF) leak is a frustrating complication of skull base surgery. Published methodologies using national surgical databases to assess CSF leak have not accounted for variability between skull base operations.

OBJECTIVE

Our goal was to attempt the development of a novel framework for adapting big data techniques to skull base surgery and assess the reliability of corresponding data manipulations.

METHODS

A retrospective nested case-control analysis was performed using patients from the National Surgical Quality Improvement Program (NSQIP) registry, 2012-2015. Current Procedural Terminology and International Classification of Diseases, Ninth Revision codes identified possible skull base operations, which were systematically grouped by anatomic location. Meningioma, schwannoma, pituitary adenoma, and trigeminal neuralgia (TN) were included.

RESULTS

Of 2918 patients, 84 (2.9%) were readmitted/reoperated on within 30 days for CSF leak. Operations involving the anterior fossa, both middle/posterior fossas in 1 approach, or the orbitocranial zygomatic approach were significantly associated with CSF leak, as were schwannomas and meningiomas in any location (8.5%, 3.1%, 10.2%, 4.1%, and 3.0%; all P < 0.0001). Multivariate analysis of only middle/posterior fossa lesions identified schwannoma (odds ratio [OR], 2.7; 95% confidence interval [CI], 1.3-5.6; P = 0.008), TN (OR, 5.4; 95% CI, 2-14.7; P = 0.008), chronic obstructive pulmonary disease (OR, 3.9; 95% CI, 1.1-14; P = 0.03), and increased operative time (OR, 4.0; 95% CI, 1.7-9.5; P = 0.009) as significant CSF leak risk factors.

CONCLUSIONS

Based on NSQIP data analyzed using a rational skull base/anatomic framework, risk factors for postoperative CSF leak include chronic obstructive pulmonary disease, operative time, anterior fossa meningioma, and middle/posterior fossa schwannoma or TN. Although databases such as NSQIP can be extensively manipulated to generate surrogate results that may provide limited insight, applications beyond their design should be approached carefully.

Development of a machine learning algorithm predicting discharge placement after surgery for spondylolisthesis

PURPOSE

We aimed to develop a machine learning algorithm that can accurately predict discharge placement in patients undergoing elective surgery for degenerative spondylolisthesis.

METHODS

The National Surgical Quality Improvement Program (NSQIP) database was used to select patients that underwent surgical treatment for degenerative spondylolisthesis between 2009 and 2016. Our primary outcome measure was non-home discharge which was defined as any discharge not to home for which we grouped together all non-home discharge destinations including rehabilitation facility, skilled nursing facility, and unskilled nursing facility. We used Akaike information criterion to select the most appropriate model based on the outcomes of the stepwise backward logistic regression. Four machine learning algorithms were developed to predict discharge placement and were assessed by discrimination, calibration, and overall performance.

RESULTS

Nine thousand three hundred and thirty-eight patients were included. Median age was 63 (interquartile range [IQR] 54-71), and 63% (n = 5,887) were female. The non-home discharge rate was 18.6%. Our models included age, sex, diabetes, elective surgery, BMI, procedure, number of levels, ASA class, preoperative white blood cell count, and preoperative creatinine. The Bayes point machine was considered the best model based on discrimination (AUC = 0.753), calibration (slope = 1.111; intercept = - 0.002), and overall model performance (Brier score = 0.132).

CONCLUSION

This study has shown that it is possible to create a predictive machine learning algorithm with both good accuracy and calibration to predict discharge placement. Using our methodology, this type of model can be developed for many other conditions and (elective) treatments. These slides can be retrieved under Electronic Supplementary Material.

Variation in Coding Practices for Vestibular Schwannoma Surgery

Introduction  Nationwide databases are frequently used resources for assessing practice patterns and clinical outcomes. However, analyses based on billing codes may be limited by the inconsistent application of current procedural terminology (CPT) codes to specific operations. We investigated the variability among commonly used CPT codes for vestibular schwannomas resection and sought to identify factors that underlie this variation. Methods  The surgical procedure for 274 cases of vestibular schwannoma resections from two institutions was reviewed and classified as retrosigmoid, translabyrinthine, or middle fossa approaches. We then assessed the CPT codes assigned to each case and analyzed their association with surgical approach, surgeons involved, the coding specialty, and year of surgery. We further compared the incidence of CPT codes assigned for vestibular schwannoma surgeries in the American College Surgeons National Surgical Quality Improvement Program (NSQIP) database from 2010 to 2014. Results  The majority (65%) of vestibular schwannoma resections within the institutional cohort were billed with skull base approach and/or excision codes, whereas 76% of cases in NSQIP were associated with a single craniotomy for tumor code. The use of skull base codes over the past decade increased within our institutional cohort but remained relatively stable within NSQIP. CPT codes did not consistently reflect the operative approaches for vestibular schwannomas. Conclusion  We observed significant variability in coding patterns for vestibular schwannoma surgeries within institutions, surgical practices, and national databases. These results call for discretion in interpretation of data from aggregated billing code-based nationwide databases and suggests a role for institutional standardization of CPT assignments for the same approaches.

Room for "quality" improvement? Validating National Surgical Quality Improvement Program-Pediatric (NSQIP-P) appendectomy data

INTRODUCTION

Accurate data are essential for the validity of clinical registries. This study aimed to validate NSQIP-P data, assess representativeness, and evaluate risk-adjusted predictive ability at a single institution.

METHODS

A prospective appendectomy-specific pediatric surgery research database (RD) maintained by clinical researchers was compared to the NSQIP-P data for appendectomies performed in 2016 at a tertiary children's hospital. NSQIP-P sampled data collected by trained surgical clinical reviewers (SCRs) were compared to matched RD patients. Both datasets used NSQIP-P definitions. Using χ², datasets were compared by patient demographics, disease severity (simple vs. complicated), and outcomes.

RESULTS

458 appendectomies for acute appendicitis were performed in 2016, of which 250 (55%) were abstracted by SCRs and matched to RD patients. Patient demographics were similar between datasets. Disease severity (NSQIP-P:50% complicated vs RD:31% complicated) and composite morbidity (NSQIP-P:6.0% vs RD:14.4%) were significantly different (both p < 0.01). Demographics and outcomes were similar between matched (n = 250) and unsampled patients in the RD (n = 208). NSQIP-P's risk-adjusted predicted morbidity was significantly lower than morbidity observed in all (n = 458) RD patients (NSQIP-P:9.9% vs RD:14.2%, p < 0.01).

CONCLUSIONS

Though constituting a representative sample, NSQIP-P appendectomy data were inconsistent with department data. Discrepancies appear to be the result of underreporting of outcome variables and disease misclassification.

TYPE OF STUDY

Retrospective comparative review.

LEVEL OF EVIDENCE

Level III.

Safety Outcomes Following Spine and Cranial Neurosurgery: Evidence From the National Surgical Quality Improvement Program

BACKGROUND

The American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) was used to establish predictors for 30-day postoperative complications following spine and cranial neurosurgery.

MATERIALS AND METHODS

The ACS-NSQIP participant use files were queried for neurosurgical cases between 2005 and 2015. Prevalence of postoperative complications following neurosurgery was determined. Nested multivariable logistic regression analysis was used to identify demographic, comorbidity, and perioperative characteristics associated with any complication and mortality for spine and cranial surgery.

RESULTS

There were 175,313 neurosurgical cases (137,029 spine, 38,284 cranial) identified. A total of 23,723 (13.5%) patients developed a complication and 2588 (1.5%) patients died. Compared with spine surgery, cranial surgery had higher likelihood of any complication (22.2% vs. 11.1%; P<0.001) and mortality (4.8% vs. 0.5%; P<0.001). In multivariable analysis, cranial surgery had 2.73 times higher likelihood for mortality compared with spine surgery (95% confidence interval, 2.46-3.03; P<0.001), but demonstrated lower odds of any complication (odds ratio, 0.93; 95% confidence interval, 0.90-0.97; P<0.001). There were 6 predictors (race, tobacco use, dyspnea, chronic obstructive pulmonary disease, chronic heart failure, and wound classification) significantly associated with any complication, but not mortality. Paradoxically, tobacco use had an unexplained protective effect on at least one complication or any complication. Similarly, increasing body mass index was protective for any complication and mortality, which suggests there may be a newly observed "obesity paradox" in neurosurgery.

CONCLUSIONS

After controlling for demographic characteristics, preoperative comorbidities, and perioperative factors, cranial surgery had higher risk for mortality compared with spine surgery despite lower risk for other complications. These findings highlight a discrepancy in the risk for postoperative complications following neurosurgical procedures that requires emphasis within quality improvement initiatives.

Short-term outcomes following posterior cervical fusion among octogenarians with cervical spondylotic myelopathy: a NSQIP database analysis

BACKGROUND CONTEXT

Degenerative changes in the cervical spine occur in an age-dependent manner. As the US population continues to age, the incidence of age-dependent, multilevel, degenerative cervical pathologies is expected to increase. Similarly, the average age of patients with cervical spondylotic myelopathy (CSM) will likely trend upward. Posterior cervical fusion (PCF) is often the treatment modality of choice in the management of multilevel cervical spine disease. Although outcomes following anterior cervical fusion for degenerative disease have been studied among older patients (aged 80 years and older), it is unknown if these results extend to octogenarian patients undergoing PCF for the surgical management of CSM.

PURPOSE

The present study aimed to quantify surgical outcomes following PCF for the treatment of CSM among the octogenarian patient population compared with patients younger than 80 years old.

STUDY DESIGN/SETTING

This was a retrospective study that used the National Surgical Quality Improvement Program (NSQIP).

PATIENT SAMPLE

The sample included patients aged 60-89 who had CSM and who underwent PCF from 2012 to 2014.

OUTCOME MEASURES

The outcome measures were multimorbidity, prolonged length of stay (LOS), discharge disposition (to home or skilled nursing/rehabilitation facility), 30-day all-cause readmission, and 30-day reoperation.

METHODS

The NSQIP database was queried for patients with CSM (International Classification of Disease, Ninth Revision, Clinical Modification code 721.1) aged 60-89 who underwent PCF (Current Procedural Terminology code 22600) from 2012 to 2014. Cohorts were defined by age group (60-69, 70-79, 80-89). Data were collected on gender, race, elective or emergent status, inpatientor outpatient status, where patients were admitted from (home vs. skilled nursing facility), American Society of Anesthesiologists class, comorbidities, and single- or multilevel fusion. After controllingfor these variables, logistic regression analysis was used to compare outcome measures in the different age groups.

RESULTS

A total of 819 patients with CSM who underwent PCF (416 aged 60-69, 320 aged 70-79, and 83 aged 80-89) were identified from 2012 to 2014. Of the PCF procedures, 79.7% were multilevel. There were no significant differences in the odds of multimorbidity, prolonged LOS, readmission, or reoperation when comparing octogenarian patients with CSM with patients aged 60-69 or 70-79. Patients aged 60-69 and 70-79 were significantly more likely to be discharged to home than patients over 80 (odds ratio [OR] 4.3, 95% confidence interval [CI] 1.8-10.4, p<.0001, and OR 2.7, 95% CI 1.1-6.4, p=.0005, respectively).

CONCLUSIONS

Compared with patients aged 60-69 and 70-79, octogenarian patients with CSM were significantly more likely to be discharged to a location other than home following PCF. After controlling for patient comorbidities and demographics, 80- to 89-year-old patients with CSM who underwent PCF did not differ in other outcomes when compared with the other age cohorts. These results can improve preoperative risk counseling and surgical decision-making.