Ultrasound gap measurement after acute Achilles rupture is reliable overall but uncertain near a 5—mm decision threshold

OBJECTIVE: Ultrasound gap distance has been proposed as a supportive triage input within protocol-based pathways after acute Achilles tendon rupture. We assessed intra-/inter-rater reliability of gap measurement and the extent of classification uncertainty around a 5-mm threshold. Retrospective observer-agreement study without a definitive reference standard, using protocol-acquired stored ultrasound cine loops from 30 clinically diagnosed complete acute ruptures in a prospectively collected cohort. METHODS: Retrospective observer-agreement study without a definitive reference standard, using protocol-acquired stored ultrasound cine loops from 30 clinically diagnosed complete acute ruptures in a prospectively collected cohort. Two blinded musculoskeletal radiologists re-reviewed stored cine loops and measured tendon-gap distance in two sessions each (≥ 4 weeks apart). We calculated intraclass correlation coefficients (ICC), Bland-Altman mean difference and 95% limits of agreement (LoA), and standard error of measurement (SEM). Agreement at the 5-mm protocol cut-point was assessed using Cohen's κ and Gwet's AC1. Using SEM, we derived 90% confidence zones indicating firm classification versus a grey zone where repeat measurement or second read may be considered. RESULTS: Intra-rater ICC(2,1) was 0.88 (95% CI 0.83-0.91) and 0.94 (0.82-0.97); SEMs were 1.57 and 1.19 mm. Inter-rater bias was + 1.12 mm with LoA - 3.85 to 6.10 mm. Agreement for study experimental pathway classification at the 5-mm cut-point (operative vs non-operative) was moderate (κ = 0.60; AC1 = 0.61), with 6/30 examinations classified differently. SEM-based 90% zones were ≤ 2.42 and ≥ 7.58 mm (rater 1) and ≤ 3.04 and ≥ 6.96 mm (rater 2); a conservative cross-rater rule suggested ≤ 2.05 or ≥ 7.95 mm. CONCLUSION: Gap measurement shows good relative reliability, but uncertainty near 5 mm can alter classification. Reporting LoA/SEM alongside ICC and flagging a grey zone may reduce misclassification in borderline cases.