To compare the double-Maddox rod test with other methods of measuring cyclodeviation

Retrospective cohort study.

We retrospectively identified 153 adults in a clinical practice with cyclodeviation assessed using double-Maddox rods, of whom 105 were also assessed using fusible synoptophore targets, 73 using nonfusible synoptophore targets, 118 using single-Maddox rod, and 43 using fundus photography. Relationships between double-Maddox rod and other tests were evaluated by calculating mean differences with 95% confidence intervals (CI), intraclass correlation coefficients (ICC), and Bland-Altman plots with linear regression.

Synoptophore cross-in-circle targets and the largest (of right or left) single-Maddox rod values were similar to double-Maddox values (mean differences: −1.2° and 0.1°, respectively; ICC: 0.79 and 0.82, respectively). Synoptophore house targets measured less excyclodeviation (mean difference: −2.7°; ICC: 0.71). Mean summed single-Maddox rod values were somewhat similar to double-Maddox values (mean difference: 1.5°; ICC: 0.85), but differences increased with greater cyclodeviation (r² = 0.2678; P < .001). Fundus photographs showed large, uncorrelated differences compared with double-Maddox rod test, when summing right and left eyes and when using the largest of right or left (mean differences: 12.2° and 6.2°; ICC: −0.02 and 0.21, respectively), and differences increased with greater cyclodeviation (r² = 0.4094; P < .001 and r² = .1143; P= .03, respectively).

There was good agreement between double-Maddox and the largest single- Maddox test values and synoptophore cross-in-circle targets but poorer agreement with other tests. Further study is needed to understand which measurements best reflect true cyclodeviation and relationships with symptoms.