Using Machine Learning Models to Identify Factors Associated With 30-Day Readmissions After Posterior Cervical Fusions: A Longitudinal Cohort Study.

OBJECTIVE: Readmission rates after posterior cervical fusion (PCF) significantly impact patients and healthcare, with complication rates at 15%-25% and up to 12% 90-day readmission rates. In this study, we aim to test whether machine learning (ML) models that capture interfactorial interactions outperform traditional logistic regression (LR) in identifying readmission-associated factors. METHODS: The Optum Clinformatics Data Mart database was used to identify patients who underwent PCF between 2004-2017. To determine factors associated with 30-day readmissions, 5 ML models were generated and evaluated, including a multivariate LR (MLR) model. Then, the best-performing model, Gradient Boosting Machine (GBM), was compared to the LACE (Length patient stay in the hospital, Acuity of admission of patient in the hospital, Comorbidity, and Emergency visit) index regarding potential cost savings from algorithm implementation. RESULTS: This study included 4,130 patients, 874 of which were readmitted within 30 days. When analyzed and scaled, we found that patient discharge status, comorbidities, and number of procedure codes were factors that influenced MLR, while patient discharge status, billed admission charge, and length of stay influenced the GBM model. The GBM model significantly outperformed MLR in predicting unplanned readmissions (mean area under the receiver operating characteristic curve, 0.846 vs. 0.829; p < 0.001), while also projecting an average cost savings of 50% more than the LACE index. CONCLUSION: Five models (GBM, XGBoost [extreme gradient boosting], RF [random forest], LASSO [least absolute shrinkage and selection operator], and MLR) were evaluated, among which, the GBM model exhibited superior predictive performance, robustness, and accuracy. Factors associated with readmissions impact LR and GBM models differently, suggesting that these models can be used complementarily. When analyzing PCF procedures, the GBM model resulted in greater predictive performance and was associated with higher theoretical cost savings for readmissions associated with PCF complications.

Identification of Factors Associated With 30-day Readmissions After Posterior Lumbar Fusion Using Machine Learning and Traditional Models: A National Longitudinal Database Study.

STUDY DESIGN: A retrospective cohort study. OBJECTIVE: To identify the factors associated with readmissions after PLF using machine learning and logistic regression (LR) models. SUMMARY OF BACKGROUND DATA: Readmissions after posterior lumbar fusion (PLF) place significant health and financial burden on the patient and overall health care system. MATERIALS AND METHODS: The Optum Clinformatics Data Mart database was used to identify patients who underwent posterior lumbar laminectomy, fusion, and instrumentation between 2004 and 2017. Four machine learning models and a multivariable LR model were used to assess factors most closely associated with 30-day readmission. These models were also evaluated in terms of ability to predict unplanned 30-day readmissions. The top-performing model (Gradient Boosting Machine; GBM) was then compared with the validated LACE index in terms of potential cost savings associated with the implementation of the model. RESULTS: A total of 18,981 patients were included, of which 3080 (16.2%) were readmitted within 30 days of initial admission. Discharge status, prior admission, and geographic division were most influential for the LR model, whereas discharge status, length of stay, and prior admissions had the greatest relevance for the GBM model. GBM outperformed LR in predicting unplanned 30-day readmission (mean area under the receiver operating characteristic curve 0.865 vs. 0.850, P <0.0001). The use of GBM also achieved a projected 80% decrease in readmission-associated costs relative to those achieved by the LACE index model. CONCLUSIONS: The factors associated with readmission vary in terms of predictive influence based on standard LR and machine learning models used, highlighting the complementary roles these models have in identifying relevant factors for the prediction of 30-day readmissions. For PLF procedures, GBM yielded the greatest predictive ability and associated cost savings for readmission. LEVEL OF EVIDENCE: 3.

Comparing blood biomarkers to clinical decision rules to select patients suspected of traumatic brain injury for head computed tomography.

INTRODUCTION: Traumatic brain injury (TBI) is a major public health concern in the U.S. Recommendations for patients admitted in the emergency department (ED) to receive head computed tomography (CT) scan are currently guided by various clinical decision rules. OBJECTIVE: To compare how a blood biomarker approach compares with clinical decision rules in terms of predicting a positive head CT in adult patients suspected of TBI. METHODS: We retrospectively identified patients transported to our emergency department and underwent a noncontrast head CT due to suspicion of TBI and who had blood samples available. Published thresholds for serum and plasma glial fibrillary acidic protein (GFAP), ubiquitin carboxyl-terminal hydrolase-L1 (UCH-L1), and serum S100ß were used to make CT recommendations. These blood biomarker-based recommendations were compared to those achieved under widely used clinical head CT decision rules (Canadian, New Orleans, NEXUS II, and ACEP Clinical Policy). RESULTS: Our study included 463 patients, of which 122 (26.3%) had one or more abnormalities presenting on head CT. Individual blood biomarkers achieved high negative predictive value (NPV) for abnormal head CT findings (88%-98%), although positive predictive value (PPV) was consistently low (25%-42%). A composite biomarker-based decision rule (GFAP+UCH-L1)'s NPV of 100% and PPV of 29% were comparable or better than those achieved under the clinical decision rules. CONCLUSION: Blood biomarkers perform at least as well as clinical rules in terms of selecting TBI patients for head CT and may be easier to implement in the clinical setting. A prospective study is necessary to validate this approach.

Association between Blood and Computed Tomographic Imaging Biomarkers in a Cohort of Mild Traumatic Brain Injury Patients.

The objective of this work was to analyze the relationships between traumatic brain injury (TBI) on computed tomographic (CT) imaging and blood concentration of glial fibrillary acidic protein (GFAP), ubiquitin C-terminal hydrolase-L1 (UCH-L1), and S100B. This prospective cohort study involved 644 TBI patients referred to Stanford Hospital's Emergency Department between November 2015 and April 2017. Plasma and serum samples of 462 patients were analyzed for levels of GFAP, UCH-L1, and S100B. Glial neuronal ratio (GNR) was calculated as the ratio between GFAP and UCH-L1 concentrations. Admission head CT scans were reviewed for TBI imaging common data elements, and performance of biomarkers for identifying TBI was assessed via area under the receiver operating characteristic curve (ROC). We also dichotomized biomarkers at established thresholds and estimated standard measures of classification accuracy. We assessed the ability of GFAP, UCH-L1, and GNR to discriminate small and large/diffuse lesions based on CT imaging using an ROC analysis. In our cohort of mostly mild TBI patients, GFAP was significantly more accurate in detecting all types of acute brain injuries than UCH-L1 in terms of area under the curve (AUC) values (p < 0.001), and also compared with S100B (p < 0.001). UCH-L1 and S100B had similar performance (comparable AUC values, p = 0.342). Sensitivity exceeded 0.8 for each biomarker across all different types of TBI injuries, and no significant differences were observed by type of injury. There was a significant difference between GFAP and GNR in distinguishing between small lesions and large/diffuse lesions in all injuries (p = 0.004, p = 0.007). In conclusion, GFAP, UCH-L1, and S100B show high sensitivity and negative predictive values for all types of TBI lesions on head CT. A combination of negative blood biomarkers (GFAP and UCH-L1) in a patient suspected of TBI may be used to safely obviate the need for a head CT scan. GFAP is a promising indicator to discriminate between small and large/diffuse TBI lesions.

Lumbar spine surgery reduces postoperative opioid use in the veteran population.

Background: The United States has been facing a worsening opioid epidemic over the past two decades. The veteran population represents a large and vulnerable group with a higher burden of mental health comorbidities. The purpose of this study was to analyze the impact of lumbar spine surgery on postoperative opioid usage in the United States veteran population. Methods: A retrospective cohort study was conducted using the Veterans Affairs Informatics and Computing Infrastructure database. Patients who underwent lumbar spine surgery were stratified into three groups by their preoperative opioid claims within 365 days of surgery. Postoperative cumulative morphine milligram equivalents (MME) were tracked for each group and the paired Wilcoxon signed rank test was used to compare cumulative preoperative MME (days -365-0) to cumulative postoperative MME (days 91-455). Results: At one year, 30.6% of patients in the high preoperative opioid group and 73.1% of patients in the low preoperative opioid group were no longer using opioids. In the opioid naive cohort, 10.0% of patients were still using opioids at one year. Among all patients, median cumulative postoperative MME was significantly less than median cumulative preoperative MME (P<0.001). High preoperative opioid usage of more than 3 claims was most significantly associated with continued postoperative opioid usage (odds ratio 12.55, P<0.001). From 2010 to 2020 the proportion of patients with preoperative opioid claims decreased (58.8% to 34.8%). Conclusions: In the veteran population, lumbar spine surgery was effective in getting 50% of patients who were on opioids preoperatively to discontinue opioids postoperatively. Even minimal exposure to opioids preoperatively resulted in a 2.69-time increase in risk of being on opioids at one year versus opioid naive patients. This study affirms that despite being a high-risk population, the veteran population has a similar response to lumbar spine surgery as the general population in regards to opioid dependence.

Home Use of a Percutaneous Wireless Intracortical Brain-Computer Interface by Individuals With Tetraplegia.

OBJECTIVE: Individuals with neurological disease or injury such as amyotrophic lateral sclerosis, spinal cord injury or stroke may become tetraplegic, unable to speak or even locked-in. For people with these conditions, current assistive technologies are often ineffective. Brain-computer interfaces are being developed to enhance independence and restore communication in the absence of physical movement. Over the past decade, individuals with tetraplegia have achieved rapid on-screen typing and point-and-click control of tablet apps using intracortical brain-computer interfaces (iBCIs) that decode intended arm and hand movements from neural signals recorded by implanted microelectrode arrays. However, cables used to convey neural signals from the brain tether participants to amplifiers and decoding computers and require expert oversight, severely limiting when and where iBCIs could be available for use. Here, we demonstrate the first human use of a wireless broadband iBCI. METHODS: Based on a prototype system previously used in pre-clinical research, we replaced the external cables of a 192-electrode iBCI with wireless transmitters and achieved high-resolution recording and decoding of broadband field potentials and spiking activity from people with paralysis. Two participants in an ongoing pilot clinical trial completed on-screen item selection tasks to assess iBCI-enabled cursor control. RESULTS: Communication bitrates were equivalent between cabled and wireless configurations. Participants also used the wireless iBCI to control a standard commercial tablet computer to browse the web and use several mobile applications. Within-day comparison of cabled and wireless interfaces evaluated bit error rate, packet loss, and the recovery of spike rates and spike waveforms from the recorded neural signals. In a representative use case, the wireless system recorded intracortical signals from two arrays in one participant continuously through a 24-hour period at home. SIGNIFICANCE: Wireless multi-electrode recording of broadband neural signals over extended periods introduces a valuable tool for human neuroscience research and is an important step toward practical deployment of iBCI technology for independent use by individuals with paralysis. On-demand access to high-performance iBCI technology in the home promises to enhance independence and restore communication and mobility for individuals with severe motor impairment.

The Neural Representation of Force across Grasp Types in Motor Cortex of Humans with Tetraplegia.

Intracortical brain-computer interfaces (iBCIs) have the potential to restore hand grasping and object interaction to individuals with tetraplegia. Optimal grasping and object interaction require simultaneous production of both force and grasp outputs. However, since overlapping neural populations are modulated by both parameters, grasp type could affect how well forces are decoded from motor cortex in a closed-loop force iBCI. Therefore, this work quantified the neural representation and offline decoding performance of discrete hand grasps and force levels in two human participants with tetraplegia. Participants attempted to produce three discrete forces (light, medium, hard) using up to five hand grasp configurations. A two-way Welch ANOVA was implemented on multiunit neural features to assess their modulation to force and grasp Demixed principal component analysis (dPCA) was used to assess for population-level tuning to force and grasp and to predict these parameters from neural activity. Three major findings emerged from this work: (1) force information was neurally represented and could be decoded across multiple hand grasps (and, in one participant, across attempted elbow extension as well); (2) grasp type affected force representation within multiunit neural features and offline force classification accuracy; and (3) grasp was classified more accurately and had greater population-level representation than force. These findings suggest that force and grasp have both independent and interacting representations within cortex, and that incorporating force control into real-time iBCI systems is feasible across multiple hand grasps if the decoder also accounts for grasp type.

Predicted Cost Savings Achieved by the Radiology Support, Communication and Alignment Network from Reducing Medical Imaging Overutilization in the Medicare Population.

OBJECTIVE: The Radiology Support, Communication and Alignment Network (R-SCAN) is a quality improvement program through which patients, referring clinicians, and radiologists collaborate to improve imaging appropriateness based on Choosing Wisely recommendations and ACR Appropriateness Criteria. R-SCAN was shown previously to increase the odds of obtaining an appropriate, higher patient or diagnostic value, imaging study. In the current study, we aimed to estimate the potential imaging cost savings associated with R-SCAN use for the Medicare population. MATERIAL AND METHODS: The R-SCAN data set was used to determine the proportion of appropriate and lesser value imaging studies performed, as well as the percent change in the total number of imaging studies performed, before and after an R-SCAN educational intervention. Using a separate CMS data set, we then identified the total number of relevant imaging studies and associated total costs using a 5% sample of Medicare beneficiaries in 2017. We applied R-SCAN proportions to the CMS data set to estimate the potential impact of the R-SCAN interventions across a broader Medicare population. RESULTS: We observed a substantial reduction in the costs associated with lesser value imaging in the R-SCAN cohort, totaling $260,000 over 3.5 months. When extrapolated to the Medicare population, the potential cost reductions associated with the decrease in lesser value imaging totaled $433 million yearly. CONCLUSION: If expanded broadly, R-SCAN interventions can result in substantial savings to the Medicare program.

Conventional versus stereotactic image guided pedicle screw placement during spinal deformity correction: a retrospective propensity score-matched study of a national longitudinal database.

PURPOSE/AIM: To compare complications, readmissions, revisions, and payments between navigated and conventional pedicle screw fixation for treatment of spine deformity. METHODS: The Thomson Reuters MarketScan national longitudinal database was used to identify patients undergoing osteotomy, posterior instrumentation, and fusion for treatment of spinal deformity with or without image-guided navigation between 2007-2016. Conventional and navigated groups were propensity-matched (1:1) to normalize differences between demographics, comorbidities, and surgical characteristics. Clinical outcomes and charges were compared between matched groups using bivariate analyses. RESULTS: A total of 4,604 patients were identified as having undergone deformity correction, of which 286 (6.2%) were navigated. Propensity-matching resulted in a total of 572 well-matched patients for subsequent analyses, of which half were navigated. Rate of mechanical instrumentation-related complications was found to be significantly lower for navigated procedures (p = 0.0371). Navigation was also associated with lower rates of 90-day unplanned readmissions (p = 0.0295), as well as 30- and 90-day postoperative revisions (30-day: p = 0.0304, 90-day: p = 0.0059). Hospital, physician, and total payments favored the conventional group for initial admission (p = 0.0481, 0.0001, 0.0019, respectively); however, when taking into account costs of readmissions, hospital payments became insignificantly different between the two groups. CONCLUSIONS: Procedures involving image-guided navigation resulted in decreased instrumentation-related complications, unplanned readmissions, and postoperative revisions, highlighting its potential utility for the treatment of spine deformity. Future advances in navigation technologies and methodologies can continue to improve clinical outcomes, decrease costs, and facilitate widespread adoption of navigation for deformity correction.

Assessment of the Radiology Support, Communication and Alignment Network to Reduce Medical Imaging Overutilization: A Multipractice Cohort Study.

PURPOSE: The aim of this study was to determine whether participation in Radiology Support, Communication and Alignment Network (R-SCAN) results in a reduction of inappropriate imaging in a wide range of real-world clinical environments. METHODS: This quality improvement study used imaging data from 27 US academic and private practices that completed R-SCAN projects between January 25, 2015, and August 8, 2018. Each project consisted of baseline, educational (intervention), and posteducational phases. Baseline and posteducational imaging cases were rated as high, medium, or low value on the basis of validated ACR Appropriateness Criteria®. Four cohorts were generated: a comprehensive cohort that included all eligible practices and three topic-specific cohorts that included practices that completed projects of specific Choosing Wisely topics (pulmonary embolism, adnexal cyst, and low back pain). Changes in the proportion of high-value cases after R-SCAN intervention were assessed for each cohort using generalized estimating equation logistic regression, and changes in the number of low-value cases were analyzed using Poisson regression. RESULTS: Use of R-SCAN in the comprehensive cohort resulted in a greater proportion of high-value imaging cases (from 57% to 79%; odds ratio, 2.69; 95% confidence interval, 1.50-4.86; P = .001) and 345 fewer low-value cases after intervention (incidence rate ratio, 0.45; 95% confidence interval, 0.29-0.70; P < .001). Similar changes in proportion of high-value cases and number of low-value cases were found for the pulmonary embolism, adnexal cyst, and low back pain cohorts. CONCLUSIONS: R-SCAN participation was associated with a reduced likelihood of inappropriate imaging and is thus a promising tool to enhance the quality of patient care and promote wise use of health care resources.

Objective activity tracking in spine surgery: a prospective feasibility study with a low-cost consumer grade wearable accelerometer.

Patient-reported outcome measures (PROMs) are commonly used to estimate disability of patients with spinal degenerative disease. Emerging technological advances present an opportunity to provide objective measurements of activity. In a prospective, observational study we utilized a low-cost consumer grade wearable accelerometer (LCA) to determine patient activity (steps per day) preoperatively (baseline) and up to one year (Y1) after cervical and lumbar spine surgery. We studied 30 patients (46.7% male; mean age 57 years; 70% Caucasian) with a baseline activity level of 5624 steps per day. The activity level decreased by 71% in the 1st postoperative week (p < 0.001) and remained 37% lower in the 2nd (p < 0.001) and 23% lower in the 4th week (p = 0.015). At no time point until Y1 did patients increase their activity level, compared to baseline. Activity was greater in patients with cervical, as compared to patients with lumbar spine disease. Age, sex, ethnic group, anesthesia risk score and fusion were variables associated with activity. There was no correlation between activity and PROMs, but a strong correlation with depression. Determining activity using LCAs provides real-time and longitudinal information about patient mobility and return of function. Recovery took place over the first eight postoperative weeks, with subtle improvement afterwards.

Functional Mapping for Glioma Surgery: A Propensity-Matched Analysis of Outcomes and Cost.

OBJECTIVE: To compare clinical outcomes and payments between glioma resections with and without functional mapping. METHODS: The Thomas Reuters MarketScan national longitudinal database was used to identify patients undergoing resection of supratentorial primary malignant glioma with or without functional mapping between 2007 and 2016. Patients were stratified into mapped and unmapped (conventional) groups and subsequently propensity-matched based on demographics, clinical comorbidities, and surgical characteristics (i.e., use of stereotactic navigation, microscope, and intratumoral chemotherapy). Outcomes and charges were compared between matched groups using bivariate analyses. RESULTS: A total of 14,037 patients were identified, of whom 796 (6.0%) received functional mapping. Propensity matching (1:1) resulted in 796 mapped patients and 796 propensity-matched controls. Thirty-day postoperative rates of new-onset seizures, cerebral edema, hemorrhage, and neurologic deficits were significantly lower for the functional mapping group (all P < 0.05). Functional mapping was also associated with shorter hospital length of stay (P = 0.0144), lower 30-day rates of emergency department visits (P = 0.0001), and fewer reoperations (P = 0.0068). Total costs of initial admission were not significantly different between groups. CONCLUSIONS: Intraoperative functional mapping during glioma resection was associated with decreased complications, reoperations, emergency department visits, and shorter lengths of stay. Furthermore, total charges of mapped resections were not significantly different from those of conventional resections. These findings support the usefulness of functional mapping for resection of supratentorial primary malignant gliomas.

Neural Representation of Observed, Imagined, and Attempted Grasping Force in Motor Cortex of Individuals with Chronic Tetraplegia.

Hybrid kinetic and kinematic intracortical brain-computer interfaces (iBCIs) have the potential to restore functional grasping and object interaction capabilities in individuals with tetraplegia. This requires an understanding of how kinetic information is represented in neural activity, and how this representation is affected by non-motor parameters such as volitional state (VoS), namely, whether one observes, imagines, or attempts an action. To this end, this work investigates how motor cortical neural activity changes when three human participants with tetraplegia observe, imagine, and attempt to produce three discrete hand grasping forces with the dominant hand. We show that force representation follows the same VoS-related trends as previously shown for directional arm movements; namely, that attempted force production recruits more neural activity compared to observed or imagined force production. Additionally, VoS-modulated neural activity to a greater extent than grasping force. Neural representation of forces was lower than expected, possibly due to compromised somatosensory pathways in individuals with tetraplegia, which have been shown to influence motor cortical activity. Nevertheless, attempted forces (but not always observed or imagined forces) could be decoded significantly above chance, thereby potentially providing relevant information towards the development of a hybrid kinetic and kinematic iBCI.

Conventional Versus Stereotactic Image-guided Pedicle Screw Placement During Posterior Lumbar Fusions: A Retrospective Propensity Score-matched Study of a National Longitudinal Database.

STUDY DESIGN: Retrospective 1:1 propensity score-matched analysis on a national longitudinal database between 2007 and 2016. OBJECTIVE: The aim of this study was to compare complication rates, revision rates, and payment differences between navigated and conventional posterior lumbar fusion (PLF) procedures with instrumentation. SUMMARY OF BACKGROUND DATA: Stereotactic navigation techniques for spinal instrumentation have been widely demonstrated to improve screw placement accuracies and decrease perforation rates when compared to conventional fluoroscopic and free-hand techniques. However, the clinical utility of navigation for instrumented PLF remains controversial. METHODS: Patients who underwent elective laminectomy and instrumented PLF were stratified into "single level" and "3- to 6-level" cohorts. Navigation and conventional groups within each cohort were balanced using 1:1 propensity score matching, resulting in 1786 navigated and conventional patients in the single-level cohort and 2060 in the 3 to 6 level cohort. Outcomes were compared using bivariate analysis. RESULTS: For the single-level cohort, there were no significant differences in rates of complications, readmissions, revisions, and length of stay between the navigation and conventional groups. For the 3- to 6-level cohort, length of stay was significantly longer in the navigation group (P < 0.0001). Rates of readmissions were, however, greater for the conventional group (30-day: P = 0.0239; 90-day: P = 0.0449). Overall complications were also greater for the conventional group (P = 0.0338), whereas revision rate was not significantly different between the 2 groups. Total payments were significantly greater for the navigation group in both the single level and 3- to 6-level cohorts (P < 0.0001). CONCLUSION: Although use of navigation for 3- to 6-level instrumented PLF was associated with increased length of stay and payments, the concurrent decreased overall complication and readmission rates alluded to its potential clinical utility. However, for single-level instrumented PLF, no differences in outcomes were found between groups, suggesting that the value in navigation may lie in more complex procedures. LEVEL OF EVIDENCE: 3.

Socioeconomic Predictors of Pituitary Surgery.

BACKGROUND: There exists a lack of data on the effect of socioeconomic status (SES) on outcomes for pituitary tumors, which have been associated with significant morbidity. The goal of this population-level study is to investigate the role of SES on receiving treatment and survival in patients with pituitary tumors. METHODS: The Surveillance, Epidemiology, and End Results (SEER) program database from the National Cancer Institute was used to identify patients diagnosed with pituitary tumors between 2003 and 2012. SES was determined using a validated composite index. Race was categorized as Caucasian and non-Caucasian. Treatment received included surgery, radiation, and radiation with surgery. Odds of receiving surgery and survival probability were analyzed using multivariate logistic regression and Cox proportional hazards model, respectively. RESULTS: A total of 25,802 patients with pituitary tumors were identified for analysis. High SES tertile (odds ratio (OR) = 1.095; 95% confidence interval (CI) [1.059, 1.132]) and quintile (OR = 1.052; 95% CI [1.031, 1.072]) were associated with higher odds of receiving surgery (p<0.0001). Caucasian patients had higher odds of receiving surgery when compared to non-Caucasian patients (OR = 1.064; 95% CI [1.000, 1.133]; p<0.05). Neither SES nor race were significant predictors of survival probability. CONCLUSION: Socioeconomic status and race were found to be associated with higher odds of receiving surgery for pituitary tumors, and thus serve as independent predictors of surgical management. Further studies are required to investigate possible causes for these findings.