Color Dissociation Artifacts in Double Maddox Rod Cyclodeviation Testing

Abstract

The double Maddox rod test, based on a red Maddox rod in front of one eye and a clear Maddox rod in front of the other, is used to measure cyclodeviation, typically in patients with superior oblique muscle pareses. Discrepant results between the double Maddox rod test and other torsion measures, and reports of "paradoxic" cyclodeviation in the normal eye of some patients with superior oblique paresis, suggest the two-color format of the double Maddox rod test may produce artifactual torsion measures. Forty patients with superior oblique paresis were tested twice using the double Maddox rod test, reversing the red and white Maddox rods between eyes for the second test, and 18 were tested further with same-color red or clear Maddox rods in front of both eyes. With the standard double Maddox rod test, 33 (83%) of 40 patients localized their cyclodeviation to the eye viewing through the red Maddox rod, irrespective of laterality of the paresis or fixation preference. In all 33 patients, laterality of the perceived torsion changed between eyes when testing was repeated with red and white Maddox rods interchanged between eyes. With same-color Maddox rods before both eyes, 17 (94%) of 18 patients localized extorsion to the paretic eye. There was 7.6:1 ratio of luminance transmission and a 1.6:1 ratio of grating spatial frequency bandpass in the plano meridian between the clear and red Maddox rods, which appear to be responsible for the double Maddox rod test artifact. The traditional double Maddox rod test may produce artifactual cyclodeviation measurements. An alternative version of the test, based on same-color Maddox rods in front of both eyes, is proposed. The relatively high spatial frequency bandpass characteristics of the plano meridian of the Maddox rod (as high as 20/25 Snellen equivalent resolution through the clear Maddox rod) also suggests double Maddox rod testing should be conducted in a dark room to avoid biases from visual environment cues.