Study design

Systematic review. Surgical decompression for degenerative cervical myelopathy (DCM) is associated with perioperative complications, including difficulty or discomfort with swallowing (dysphagia) as well as changes in sound production (dysphonia). This systematic review aims to (1) outline how dysphagia and dysphonia are defined in the literature and (2) assess the quality of definitions using a novel 4-point rating system.

Methods

An electronic database search was conducted for studies that reported on dysphagia, dysphonia or other related complications of DCM surgery. Data extracted included study design, surgical details, as well as definitions and rates of surgical complications. A 4-point rating scale was developed to assess the quality of definitions for each complication.

Results

Our search yielded 2,673 unique citations, 11 of which met eligibility criteria and were summarized in this review. Defined complications included odynophagia (n = 1), dysphagia (n = 11), dysphonia (n = 2), perioperative swelling complications (n = 2), and soft tissue swelling (n = 3). Rates of dysphagia varied substantially (0.0%-50.0%) depending on whether this complication was patient-reported (4.4%); patient-reported using a modified Swallowing Quality of Life questionnaire (43.1%) or the Bazaz criteria (8.8%-50.0%); or diagnosed using an extensive protocol consisting of clinical assessment, a bedside swallowing test, evaluation by a speech and language pathologist and a modified barium swallowing test/fiberoptic endoscopy (42.9%). The reported incidences of dysphonia also ranged significantly from 0.6% to 38.0%.

Conclusion

There is substantial variability in reported rates of dysphagia and dysphonia due to differences in data collection methods, diagnostic strategies, and definitions. Consolidation of nomenclature will improve evaluation of the overall safety of surgery.

Although machine learning (ML) has shown promise across disciplines, out-of-sample generalizability is concerning. This is currently addressed by sharing multi-site data, but such centralization is challenging/infeasible to scale due to various limitations. Federated ML (FL) provides an alternative paradigm for accurate and generalizable ML, by only sharing numerical model updates. Here we present the largest FL study to-date, involving data from 71 sites across 6 continents, to generate an automatic tumor boundary detector for the rare disease of glioblastoma, reporting the largest such dataset in the literature (n = 6, 314). We demonstrate a 33% delineation improvement for the surgically targetable tumor, and 23% for the complete tumor extent, over a publicly trained model. We anticipate our study to: 1) enable more healthcare studies informed by large diverse data, ensuring meaningful results for rare diseases and underrepresented populations, 2) facilitate further analyses for glioblastoma by releasing our consensus model, and 3) demonstrate the FL effectiveness at such scale and task-complexity as a paradigm shift for multi-site collaborations, alleviating the need for data-sharing.