Objectives

The objectives of the study were to assess agreement of subjective radiographic left atrial enlargement (LAE) between readers, to compare subjective radiographic LAE with echocardiography, and to assess the accuracy and reliability of commonly used Roentgen signs for LAE.

Animals

One hundred one dogs with thoracic radiographs and echocardiography performed on the same day at a veterinary teaching hospital were retrospectively reviewed.

Methods

Thoracic radiographs were reviewed by two cardiologists, two radiologists, and two small animal rotating interns. Radiographs were evaluated for the subjective presence and severity of LAE and for seven Roentgen signs. Echocardiographic LA size was evaluated objectively by the left atrial-to-aortic root ratio and LA volume indexed to body weight. Interreader agreement of radiographic LAE evaluation and agreement between radiographic LAE assessment and echocardiographic LAE were assessed by linearly weighted kappa and intraclass correlation coefficient.

Results

Interreader agreement for the presence and degree of radiographic LAE ranged from moderate to substantial. The agreement between subjective radiographic LAE and echocardiographic LAE was moderate. Single Roentgen signs had poor to fair correlation with echocardiographic LAE.

Conclusions

The present study found that interreader agreement for radiographic classification of LAE was substantial among specialists and moderate for all readers. Subjective, global assessment of radiographic LAE is better than using any single Roentgen sign.