Excessive resident duty hours (RDH) are a recognized issue with implications for physician well-being and patient safety. A major component of the RDH concern is on-call duty. While considerable work has been done to reduce resident call workload, there is a paucity of research in optimizing resident call scheduling. Call coverage is scheduled manually rather than demand-based, which generally leads to over-scheduling to prevent a service gap. Machine learning (ML) has been widely applied in other industries to prevent such issues of a supply-demand mismatch. However, the healthcare field has been slow to adopt these innovations. As such, the aim of this study was to use ML models to 1) predict demand on orthopaedic surgery residents at a level I trauma centre and 2) identify variables key to demand prediction.

Daily surgical handover emails over an eight year (2012-2019) period at a level I trauma centre were collected. The following data was used to calculate demand: spine call coverage, date, and number of operating rooms (ORs), traumas, admissions and consults completed. Various ML models (linear, tree-based and neural networks) were trained to predict the workload, with their results compared to the current scheduling approach. Quality of models was determined by using the area under the receiver operator curve (AUC) and accuracy of the predictions. The top ten most important variables were extracted from the most successful model.

During training, the model with the highest AUC and accuracy was the multivariate adaptive regression splines (MARS) model, with an AUC of 0.78±0.03 and accuracy of 71.7%±3.1%. During testing, the model with the highest AUC and accuracy was the neural network model, with an AUC of 0.81 and accuracy of 73.7%. All models were better than the current approach, which had an AUC of 0.50 and accuracy of 50.1%. Key variables used by the neural network model were (descending order): spine call duty, year, weekday/weekend, month, and day of the week.

This was the first study attempting to use ML to predict the service demand on orthopaedic surgery residents at a major level I trauma centre. Multiple ML models were shown to be more appropriate and accurate at predicting the demand on surgical residents as compared to the current scheduling approach. Future work should look to incorporate predictive models with optimization strategies to match scheduling with demand in order to improve resident well being and patient care.

The primary objective is to compare revision rates for lumbar disc replacement (LDR) and fusion at the same or adjacent levels in Ontario, Canada. The secondary objectives include acute complications during hospitalization and in 30 days, and length of hospital stay.

A population-based cohort study was conducted using health administrative databases including patients undergoing LDR or single level fusion between October 2005 to March 2018. Patients receiving LDR or fusion were identified using physician claims recorded in the Ontario Health Insurance Program database. Additional details of surgical procedure were obtained from the Canadian Institute for Health Information hospital discharge abstract. Primary outcome measured was presence of revision surgery in the lumbar spine defined as operation greater than 30 days from index procedure. Secondary outcomes were immediate/ acute complications within the first 30 days of index operation.

A total of 42,024 patients were included. Mean follow up in the LDR and fusion groups were 2943 and 2301 days, respectively. The rates of revision surgery at the same or adjacent levels were 4.7% in the LDR group and 11.1% in the fusion group (P=.003). Multivariate analysis identified risk factors for revision surgery as being female, hypertension, and lower surgeon volume. More patients in the fusion group had dural tears (p<.001), while the LDR group had more “other” complications (p=.037). The LDR group had a longer mean hospital stay (p=.018).

In this study population, the LDR group had lower rates of revision compared to the fusion group. Caution is needed in concluding its significance due to lack of clinical variables and possible differences in indications between LDR and posterior decompression and fusion.

Single level discectomy (SLD) is one of the most commonly performed spinal surgery procedures. Two key drivers of their cost-of-care are duration of surgery (DOS) and postoperative length of stay (LOS). Therefore, the ability to preoperatively predict SLD DOS and LOS has substantial implications for both hospital and healthcare system finances, scheduling and resource allocation. As such, the goal of this study was to predict DOS and LOS for SLD using machine learning models (MLMs) constructed on preoperative factors using a large North American database.

The American College of Surgeons (ACS) National Surgical and Quality Improvement (NSQIP) database was queried for SLD procedures from 2014-2019. The dataset was split in a 60/20/20 ratio of training/validation/testing based on year. Various MLMs (traditional regression models, tree-based models, and multilayer perceptron neural networks) were used and evaluated according to 1) mean squared error (MSE), 2) buffer accuracy (the number of times the predicted target was within a predesignated buffer), and 3) classification accuracy (the number of times the correct class was predicted by the models). To ensure real world applicability, the results of the models were compared to a mean regressor model.

A total of 11,525 patients were included in this study. During validation, the neural network model (NNM) had the best MSEs for DOS (0.99) and LOS (0.67). During testing, the NNM had the best MSEs for DOS (0.89) and LOS (0.65). The NNM yielded the best 30-minute buffer accuracy for DOS (70.9%) and ≤120 min, >120 min classification accuracy (86.8%). The NNM had the best 1-day buffer accuracy for LOS (84.5%) and ≤2 days, >2 days classification accuracy (94.6%). All models were more accurate than the mean regressors for both DOS and LOS predictions.

We successfully demonstrated that MLMs can be used to accurately predict the DOS and LOS of SLD based on preoperative factors. This big-data application has significant practical implications with respect to surgical scheduling and inpatient bedflow, as well as major implications for both private and publicly funded healthcare systems. Incorporating this artificial intelligence technique in real-time hospital operations would be enhanced by including institution-specific operational factors such as surgical team and operating room workflow.

Prolonged length of stay (LOS) is a significant contributor to the variation in surgical health care costs and resource utilization after elective spine surgery. The primary goal of this study was to identify patient, surgical and institutional variables that influence LOS. The secondary objective is to examine variability in institutional practices among participating centers.

This is a retrospective study of a prospectively multicentric followed cohort of patients enrolled in the CSORN between January 2015 and October 2020. A logistic regression model and bootstrapping method was used. A survey was sent to participating centers to assessed institutional level interventions in place to decrease LOS. Centers with LOS shorter than the median were compared to centers with LOS longer than the median.

A total of 3734 patients were included (979 discectomies, 1102 laminectomies, 1653 fusions). The median LOS for discectomy, laminectomy and fusion were respectively 0.0 day (IQR 1.0), 1.0 day (IQR 2.0) and 4.0 days (IQR 2.0). Laminectomy group had the largest variability (SD=4.4, Range 0-133 days). For discectomy, predictors of LOS longer than 0 days were having less leg pain, higher ODI, symptoms duration over 2 years, open procedure, and AE (p< 0.05). Predictors of longer LOS than median of 1 day for laminectomy were increasing age, living alone, higher ODI, open procedures, longer operative time, and AEs (p< 0.05). For posterior instrumented fusion, predictors of longer LOS than median of 4 days were older age, living alone, more comorbidities, less back pain, higher ODI, using narcotics, longer operative time, open procedures, and AEs (p< 0.05). Ten centers (53%) had either ERAS or a standardized protocol aimed at reducing LOS.

In this study stratifying individual patient and institutional level factors across Canada, several independent predictors were identified to enhance the understanding of LOS variability in common elective lumbar spine surgery. The current study provides an updated detailed analysis of the ongoing Canadian efforts in the implementation of multimodal ERAS care pathways. Future studies should explore multivariate analysis in institutional factors and the influence of preoperative patient education on LOS.

Spinal stenosis is a condition resulting in the compression of the neural elements due to narrowing of the spinal canal. Anatomical factors including enlargement of the facet joints, thickening of the ligaments, and bulging or collapse of the intervertebral discs contribute to the compression. Decompression surgery alleviates spinal stenosis through a laminectomy involving the resection of bone and ligament. Spinal decompression surgery requires appropriate planning and variable strategies depending on the specific situation. Given the potential for neural complications, there exist significant barriers to residents and fellows obtaining adequate experience performing spinal decompression in the operating room. Virtual teaching tools exist for learning instrumentation which can enhance the quality of orthopaedic training, building competency and procedural understanding. However, virtual simulation tools are lacking for decompression surgery. The aim of this work was to develop an open-source 3D virtual simulator as a teaching tool to improve orthopaedic training in spinal decompression.

A custom step-wise spinal decompression simulator workflow was built using 3D Slicer, an open-source software development platform for medical image visualization and processing. The procedural steps include multimodal patient-specific loading and fusion of Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) data, bone threshold-based segmentation, soft tissue segmentation, surgical planning, and a laminectomy and spinal decompression simulation. Fusion of CT and MRI elements was achieved using Fiducial-Based Registration which aligned the scans based on manually placed points allowing for the identification of the relative position of soft and hard tissues. Soft tissue segmentation of the spinal cord, the cerebrospinal fluid, the cauda equina, and the ligamentum flavum was performed using Simple Region Growing Segmentation (with manual adjustment allowed) involving the selection of structures on T1 and/or T2-weighted scans. A high-fidelity 3D model of the bony and soft tissue anatomy was generated with the resulting surgical exposure defined by labeled vertebrae simulating the central surgical incision. Bone and soft tissue resecting tools were developed by customizing manual 3D segmentation tools. Simulating a laminectomy was enabled through bone and ligamentum flavum resection at the site of compression. Elimination of the stenosis enabled decompression of the neural elements simulated by interpolation of the undeformed anatomy above and below the site of compression using Fill Between Slices to reestablish pre-compression neural tissue anatomy.

The completed workflow allows patient specific simulation of decompression procedures by staff surgeons, fellows and residents. Qualitatively, good visualization was achieved of merged soft tissue and bony anatomy. Procedural accuracy, the design of resecting tools, and modeling of the impact of bone and ligament removal was found to adequately encompass important challenges in decompression surgery.

This software development project has resulted in a well-characterized freely accessible tool for simulating spinal decompression surgery. Future work will integrate and evaluate the simulator within existing orthopaedic resident competency-based curriculum and fellowship training instruction. Best practices for effectively teaching decompression in tight areas of spinal stenosis using virtual simulation will also be investigated in future work.