Reliability and agreement analysis among 5 subjective methods of ocular torsion measurement

To evaluate the reliability in the measurement of ocular torsion and the agreement of the following subjective methods: double Maddox test, synoptophore, Gracis torsionometer, Harms screen and Awaya cyclodeviation test.

Patients with vertical strabismus acquired in adulthood were recruited and ocular torsion was measured with the 5 methods described on three occasions. As a gold standard test does not exist, the one that obtained the best repeatability data was chosen as the reference test.

25 patients were included in the study. The repeatability of each test was studied: double Maddox test (ICC = 0.783, CV = 29.33%), synoptophore (ICC = 0.976, CV = 6.71%), Gracis torsionometer (ICC = 0.937, CV = 20.10%), Harms screen (ICC = 0.962, CV = 11.86%) and Awaya test (ICC = 0.987, CV = 52.58%). The reference test to compare the agreement was the synoptophore. Statistically significant differences were found when comparing the ocular torsion ranges between the synoptophore and the Gracis torsionometer (p = 0.008) and between the synoptophore and the Awaya test (p = 0.02).

The double Maddox test, the synoptophore, the Gracis torsionometer, and the Harms screen are reliable methods with good reproducibility indices. Among them, the synoptophore is the most consistent method. The Awaya test did not show good reliability. The bilateral Maddox test, the Gracis torsionometer, and the Harms screen were methods with good agreement with the synoptophore, which was determined as the reference test. The Awaya test did not show good agreement with the synoptophore.