[A study on the accuracy of the gaze-tilt test in predicting the postoperative head posture improvement in patients with congenital superior oblique palsy].

To investigate the accuracy of the gaze-tilt test in predicting the postoperative head posture improvement in patients with congenital superior oblique palsy (CSOP). A diagnostic test was conducted, enrolling children aged 3 to 6 years diagnosed with unilateral CSOP and scheduled for unilateral inferior oblique weakening surgery at Tianjin Eye Hospital between May 2023 and January 2024. Preoperatively, the single eye cover test and the gaze-tilt test were used to assess the patients' compensatory head posture improvement. The overall predictive accuracy, predictive accuracy for different head position improvement outcomes, and error rates (false positive rate and false negative rate) of the two methods were compared by evaluating the predicted outcomes against the actual head postures at postoperative day 1 and month 1. The paired -test, paired Wilcoxon signed-rank test, and paired chi-square test were used for statistical analysis. Thirty-two children were included in the final analysis, consisting of 17 males and 15 females. The preoperative head tilt angle was 77.50°(75.00°, 80.75°). The head tilt angles at postoperative day 1 and month 1 were 88.00°(82.64°, 89.99°) and 88.90°(87.35°, 90.00°), respectively. The gaze-tilt test predicted the postoperative head position tilt angle of 87.00°(86.00°, 89.07°), which was significantly different from the angle of 83.00°(80.27°, 87.30°) predicted by the single eye cover test (<0.05). Compared with the actual tilt position angle on postoperative day 1, the gaze-tilt test showed no statistically significant difference (>0.05), but the single eye cover test showed significant difference (<0.05). The predictive accuracy of the complete head position improvement was significantly higher for the gaze-tilt test [postoperative day 1: 86.36%(19/22), postoperative month 1: 81.48% (22/27)] compared to the single eye cover test [postoperative day 1: 50.00% (11/22), postoperative month 1: 44.44% (12/27)] (both <0.05). The false negative rate of the gaze-tilt test was lower than that of the single eye cover test. The prediction error of the gaze-tilt test (postoperative day 1:-0.53°±4.17°, postoperative month 1: 1.67°±4.26°) was smaller than that of the single eye cover test (postoperative day 1: 2.81°±4.61°, postoperative month 1: 5.01°±4.90°), with statistically significant differences (both <0.05). Compared to the single eye cover test, the gaze-tilt test demonstrated higher accuracy and smaller predictive errors in predicting the postoperative head posture improvement in patients with CSOP.