Terrible Triad of the Elbow: Proposal and Reliability Validation of a Novel CT-Based Alphanumeric Classification System

Background/Objectives: The terrible triad (TT) of the elbow combines elbow dislocation with radial-head and coronoid fractures. Existing systems classify either the individual osseous components or the broader spectrum of elbow fracture-dislocations, requiring multiple labels to describe the combined TT morphology. We propose a novel combined alphanumeric CT-based classification that integrates radial-head and coronoid morphology into one TT-specific code. The rationale is not merely terminological: the two axes correspond to osseous features routinely considered during preoperative planning and may therefore provide a compact, clinically oriented description in addition to standardising communication and research stratification. Methods: Radial-head morphology was coded 1–3 using CT-derived displacement, comminution and radial-neck involvement, while coronoid morphology was coded A–C using tip-fragment height and basal extension. Among 167 consecutive TT patients treated between 2012 and 2023, 63 had a qualifying post-reduction CT and met all eligibility criteria. Three observers with different levels of expertise independently classified anonymised CT datasets twice, three weeks apart, with randomised case order at T2. Interobserver agreement among all three raters was assessed with Fleiss’ κ; pairwise Cohen’s κ and exact percentage agreement were secondary analyses, and intra-observer agreement was assessed with Cohen’s κ. Results: For the combined alphanumeric code, Fleiss’ κ increased from 0.708 (95% CI 0.584–0.815) at T1 to 0.817 (0.702–0.913) at T2, with exact three-rater agreement of 71.4% and 84.1%, respectively. Radial-head Fleiss’ κ was 0.761 (0.615–0.888) at T1 and 0.821 (0.682–0.940) at T2; coronoid κ was 0.746 (0.602–0.864) and 0.807 (0.676–0.912). Intra-observer κ for the combined code ranged from 0.753 to 0.886. The final descriptive distribution was 1B 2/63 (3.2%), 2A 5/63 (7.9%), 2B 8/63 (12.7%), 3A 19/63 (30.2%), 3B 27/63 (42.9%) and 3C 2/63 (3.2%); no 1A, 1C or 2C cases were present. Conclusions: The novel combined CT-based alphanumeric system offers a compact, reproducible descriptor of paired osseous morphology. Its added value lies in integrating established fracture concepts into one TT-specific code with prespecified CT rules. In addition to standardising communication and research stratification, the paired descriptor may provide a clinically intuitive framework for treatment planning by simultaneously conveying radial-head complexity and coronoid morphology. It is not intended to replace assessment of ligamentous injury, dynamic stability or intraoperative findings, and treatment-guiding and prognostic validity require dedicated outcome-based studies. Level of evidence: IV (reliability study).