Lateromedial and oblique radiographs detect most fetlock pathologies as effectively as a full series in horses.

Abstract

Background: Radiographic protocols for the metacarpo-/tarsophalangeal joint during pre-purchase examinations (PPE) vary internationally, but their impact on pathology detection remains unclear. Optimising imaging protocols is essential to balance diagnostic accuracy with workflow efficiency and radiation exposure.

Objectives: To evaluate the effect of different radiographic view combinations on fetlock pathology detection and observer agreement in a PPE context; hypothesising that detection rates vary with view selection.

Study design: Retrospective observational study.

Methods: Two observers reviewed fetlock radiographic series using four view combinations: lateromedial (LM) alone, LM and dorsopalmar/plantar (LM/DP), LM and oblique projections (LM/OB), and the full series (FULL). McNemar's chi-squared test assessed detection differences; Cohen's kappa evaluated intra- and inter-observer agreement, and diagnostic parameters were calculated for reduced views relative to the full series.

Results: A total of 673 fetlock series were reviewed. The LM/OB combination showed no significant difference in detecting most pathologies compared to the full series. Observer agreement was generally highest with the full series. Intra-observer agreement was highest for LM/OB, except for subchondral bone changes in the proximal phalanx, where LM/DP performed better; though overall agreement was low. Lateromedial projections reliably detected fragmentation, sesamoid fractures, and osseous cyst-like lesions. Oblique views were superior for sesamoid bone changes.

Main limitations: Retrospective design, selection bias, and lack of gold standard confirmation.

Conclusions: While LM/OB offers a practical compromise in many PPE scenarios, certain pathologies, particularly subchondral bone changes, may require additional views. A tailored approach based on age, discipline, and clinical risk may optimise diagnostic yield.