Clavicle nonunion (pseudarthrosis) is an uncommon but clinically important complication of clavicle fracture that may lead to persistent pain, limitation of shoulder function, prolonged work disability, and impaired quality of life. Surgical repair is frequently proposed, most commonly using plate fixation with or without bone grafting. However, because the condition is relatively rare, available evidence is largely based on retrospective series and practice varies regarding key technical decisions (e.g., when to add bone graft, how to manage bone defects or loss of length, and how to approach septic versus aseptic nonunion). The PSEUDO-CLAV study has been designed to describe "real-world" management in a single tertiary center and to estimate healing and recovery outcomes after surgical treatment, using routinely collected data.
Overall design and setting PSEUDO-CLAV is a single-center, retrospective, observational cohort/case-series study conducted in the Orthopaedic and Trauma Surgery Department at CHU de Brest. The study reuses data generated during routine care and does not introduce additional visits, tests, or changes in management.
The study period covers patients surgically treated for clavicle nonunion between 01 January 2000 and 31 December 2024, chosen to ensure adequate follow-up for outcome assessment at the time of analysis. Approximately 30 cases are expected based on local incidence and feasibility estimates.
The study scope includes aseptic and septic clavicle nonunions treated surgically, reflecting the full spectrum of cases managed at the center.
Data sources and study procedures (no additional patient burden)
All data originate from the care pathway and are extracted from routine clinical systems, including:
medical records (history, comorbidities, perioperative and follow-up documentation), operative reports and perioperative documentation (surgical strategy and intraoperative details), imaging and radiology systems (e.g., radiographs and CT where available), and laboratory/microbiology results when relevant (including samples collected as part of standard care).
No participant is asked to perform additional tasks. There is no direct intervention and typically no direct contact for study purposes, because the project is based on reuse of existing information.
Key clinical concepts assessed
The dataset is structured to capture the patient journey from the initial fracture through nonunion diagnosis and surgical repair, with an emphasis on variables that may plausibly influence healing and recovery. Examples of routinely captured domains include:
baseline patient characteristics (e.g., age/sex, relevant comorbidities and treatments that may affect bone healing), features of the initial fracture and prior management (including previous procedures on the same clavicle), preoperative nonunion characterization (clinical status, radiographic appearance, and any measurable bone defect), infection work-up and classification when suspected (clinical signs, inflammatory markers, microbiology, and operative sampling), surgical strategy at index nonunion surgery (type of osteosynthesis, approach, debridement, restoration of length, graft type when used, and perioperative antibiotics), and follow-up information from routine visits (pain, functional scores when documented, range of motion, return to work/sport, imaging findings, complications, and any re-operations).
Outcome assessment approach (as captured in routine care) Healing is assessed using clinical and radiographic information as documented in the medical record and imaging archive. Functional outcomes are assessed using validated shoulder outcome measures when available in the chart (e.g., Constant and/or DASH), alongside other clinically relevant elements routinely recorded in follow-up (pain scores, range of motion, return to activity, and patient satisfaction). Complications are collected and categorized as early and late events, and re-interventions are documented with indication and timing.
Because follow-up schedules may vary over the long inclusion period, the protocol defines pragmatic time windows to harmonize assessment-particularly around the standard post-operative checkpoints used in routine practice (e.g., early postoperative review, intermediate visits, and the evaluation around six months).
Data management, quality control, and bias mitigation
To reduce information bias inherent in retrospective studies, the protocol specifies:
predefined operational definitions (e.g., for nonunion and for outcome assessment), a structured case report form / data dictionary approach for consistent abstraction, double review of a random sample (≥10%) of records/imaging to check data quality and resolve discrepancies, and transparent accounting of case selection (e.g., documenting exclusions and incomplete follow-up where applicable).
The study acknowledges key methodological limitations (retrospective design, small sample size, monocentric practice patterns, temporal evolution of implants/techniques across 2000-2024) and addresses them primarily through careful description, stratified/exploratory analyses, and explicit discussion of generalizability.
Statistical analysis plan (high level)
Analyses are primarily descriptive and exploratory, appropriate for the expected sample size. The protocol plans:
descriptive statistics for baseline characteristics and management patterns, estimation of proportions (e.g., healing rate) with 95% confidence intervals, time-to-event methods for time to documented healing (e.g., Kaplan-Meier; exploratory Cox modeling), exploratory group comparisons using tests suited to small samples (e.g., Fisher's exact test for categorical data; Mann-Whitney or t-test for continuous data depending on distribution), and exploratory multivariable modeling for selected outcomes where feasible (e.g., Cox for time to healing; linear regression for functional scores), interpreted cautiously given sample size.
Analyses are conducted locally using R (≥4.3), with verification steps in Python where helpful for cross-checking key results.
Ethics, privacy, and legal basis PSEUDO-CLAV is conducted under a framework for reuse of health data for research, consistent with MR004 as documented in the protocol.
Data protection measures include:
extraction limited to variables necessary for the study (data minimization), pseudonymization of the research dataset (no direct identifiers in the analysis file), restricted access on secure CHU infrastructure (password-protected study files; controlled access to linkage information when needed for information/opt-out handling), no external transfer outside CHU Brest except via approved secure internal mechanisms if required, and dissemination only in aggregated, non-identifying form. As this is a retrospective reuse of data, patients are informed through the center's information pathway and have the right to object (opt out / non-opposition framework); a minimum delay between information and analysis is planned to allow exercise of this right. The DPO contact point is provided for privacy-related questions and objections.
Expected impact and dissemination By providing an exhaustive description of surgically treated clavicle nonunion cases in this center-including operative strategies, complications, re-interventions, and functional recovery-PSEUDO-CLAV aims to support local practice improvement (e.g., clarifying when grafting is used, harmonizing follow-up windows, and informing prevention of reoperation). The protocol anticipates dissemination through scientific communication and publication, with results reported in a manner that does not permit identification of individuals.
Timeline (planned) The protocol outlines a staged process including patient information/non-opposition handling, retrospective data abstraction from records and imaging, database lock, analysis, and reporting, with planned archival and retention consistent with institutional requirements.
