A systematic review of diagnostic tests based on radiologic measurements of structures in dogs and cats was done in order to reach generalizable conclusions about the value of making such measurements. Literature search was done using the ISI Web of Knowledge(SM) for studies in the subject category Veterinary sciences. Studies were eligible for inclusion that employed length, angle, area or volume measurements from radiographic, ultrasonographic, CT or MR images of dogs or cats as a diagnostic test for a naturally occurring condition, compared the results of imaging with a reference standard, included at least 10 subjects, and sufficient data that a 2 × 2 table of results could be constructed. Quality of studies was assessed using the QUADAS-2 tool. Twenty-six studies were found describing 40 tests that satisfied the inclusion criteria. Tests were radiographic in 22 (55%) instances and ultrasonographic in 18 (45%). Quality of studies was generally low, with a risk of bias in patient selection in 92% studies, performance of the index test in 73% studies, and patient flow in 42% studies. Median (range) number of subjects was 64 (20-305), sensitivity was 77% (38-99%), specificity was 82% (50-99%), positive likelihood ratio was 4.1 (1-103), and negative likelihood ratio was 0.29 (0.01-1). Two studies that compared accuracy of radiographic measurements to subjective image interpretation alone found no difference. Evidence is weak that radiologic measurements of structures in dogs and cats are useful for diagnosis, hence measurements should not be emphasized as a basis for diagnosis in either teaching or clinical imaging reports.
Keywords: diagnostic test accuracy; measurement; radiology; systematic reviews.
© 2014 American College of Veterinary Radiology.