Search

Abstract. Objectives:. With more than 300,000 patients per year in the United States alone, hip fractures are one of the most common injuries occurring in the elderly. The incidence is predicted to rise to 6 million cases per annum worldwide by 2050. Many fracture registries have been established, serving as tools for quality surveillance and evaluating patient outcomes. Most registries are based on billing and procedural codes, prone to under-reporting of cases. Deep learning (DL) is able to interpret radiographic images and assist in fracture detection; we propose to conduct a DL-based approach intended to autocreate a fracture registry, specifically for the hip fracture population. Methods:. Conventional radiographs (n = 18,834) from 2919 patients from Massachusetts General Brigham hospitals were extracted (images designated as hip radiographs within the medical record). We designed a cascade model consisting of 3 submodules for image view classification (MI), postoperative implant detection (MII), and proximal femoral fracture detection (MIII), including data augmentation and scaling, and convolutional neural networks for model development. An ensemble model of 10 models (based on ResNet, VGG, DenseNet, and EfficientNet architectures) was created to detect the presence of a fracture. Results:. The accuracy of the developed submodules reached 92%–100%; visual explanations of model predictions were generated through gradient-based methods. Time for the automated model-based fracture–labeling was 0.03 seconds/image, compared with an average of 12 seconds/image for human annotation as calculated in our preprocessing stages. Conclusion:. This semisupervised DL approach labeled hip fractures with high accuracy. This mitigates the burden of annotations in a large data set, which is time-consuming and prone to under-reporting. The DL approach may prove beneficial for future efforts to autocreate construct registries that outperform current diagnosis and procedural codes. Clinicians and researchers can use the developed DL approach for quality improvement, diagnostic and prognostic research purposes, and building clinical decision support tools.

Background:. Patient-reported outcome measures (PROMs) and, specifically, the Patient-Reported Outcomes Measurement Information System (PROMIS), are increasingly utilized for clinical research, clinical care, and health-care policy. However, completion of these outcome measures can be inconsistent and challenging. We hypothesized that sociodemographic variables are associated with the completion of PROM questionnaires. The purposes of the present study were to calculate the completion rate of assigned PROM forms and to identify sociodemographic and other variables associated with completion to help guide improved collection efforts. Methods:. All new orthopaedic patients at a single academic medical center were identified from 2016 to 2020. On the basis of subspecialty and presenting condition, patients were assigned certain PROMIS forms and legacy PROMs. Demographic and clinical information was abstracted from the electronic medical record. Bivariate analyses were performed to compare characteristics among those who completed assigned PROMs and those who did not. A multivariable logistic regression model was created to determine which variables were associated with successful completion of assigned PROMs. Results:. Of the 219,891 new patients, 88,052 (40%) completed all assigned PROMs. Patients who did not activate their internet-based patient portal had a 62% increased likelihood of not completing assigned PROMs (odds ratio [OR], 1.62; 95% confidence interval [CI], 1.58 to 1.66; p < 0.001). Non-English-speaking patients had a 90% (OR, 1.90; 95% CI, 1.82 to 2.00; p < 0.001) increased likelihood of not completing assigned PROMs at presentation. Older patients (≥65 years of age) and patients of Black race had a 23% (OR, 1.23; 95% CI, 1.19 to 1.27; p < 0.001) and 24% (OR, 1.24; 95% CI, 1.19 to 1.30; p < 0.001) increased likelihood of not completing assigned PROMs, respectively. Conclusions:. The rate of completion of PROMs varies according to sociodemographic variables. This variability could bias clinical outcomes research in orthopaedic surgery. The present study highlights the need to uniformly increase completion rates so that outcomes research incorporates truly representative cohorts of patients treated. Furthermore, the use of these PROMs to guide health-care policy decisions necessitates a representative patient distribution to avoid bias in the health-care system. Level of Evidence:. Prognostic Level III. See Instructions for Authors for a complete description of levels of evidence.

Background: Laminoplasty (LP) and laminectomy and fusion (LF) are utilized to achieve decompression in patients with symptomatic degenerative cervical myelopathy (DCM). Comparative analyses aimed at determining outcomes and clarifying indications between these procedures represent an area of active research. Accordingly, we sought to compare inpatient opioid use between LP and LF patients and to determine if opioid use correlated with length of stay. Methods: Sociodemographic information, surgical and hospitalization data, and medication administration records were abstracted for patients >18 years of age who underwent LP or LF for DCM in the Mass General Brigham (MGB) health system between 2017 and 2019. Specifically, morphine milligram equivalents (MME) of oral and parenteral pain medication given after arrival in the recovery area until discharge from the hospital were collected. Categorical variables were analyzed using chi-squared analysis or Fisher exact test when appropriate. Continuous variables were compared using Independent samples t tests and Mann-Whitney U tests. Results: One hundred eight patients underwent LF, while 138 patients underwent LP. Total inpatient opioid use was significantly higher in the LF group (312 vs. 260 MME, p=.03); this difference was primarily driven by higher postoperative day 0 pain medication requirements. Furthermore, more LF patients required high dose (>80 MME/day) regimens. While length of stay was significantly different between groups, with LF patients staying approximately 1 additional day, postoperative day 0 MME was not a significant predictor of this difference. When operative levels including C2, T1, and T2 were excluded, the differences in total opioid use and average length of stay lost significance. Conclusions: Inpatient opioid use and length of stay were significantly greater in LF patients compared to LP patients; however, when constructs including C2, T1, T2 were excluded from analysis, these differences lost significance. Such findings highlight the impact of operative extent between these procedures. Future studies incorporating patient reported outcomes and evaluating long-term pain needs will provide a more complete understanding of postoperative outcomes between these 2 procedures.