Criteria for decompensation in binocular vision.

Two groups of patients, one under 40 years of age and the other 40 years and over, were divided into subgroups according to whether they had symptoms associated with near vision. A method developed in 'signal detection theory' was used to determine whether a value could be found for heterophoria, or associated heterophoria, which would predict which patients had symptoms due to decompensated heterophoria in normal routine investigation. Receiver operator characteristic curves (ROC curves) were constructed to look for these cut-off values between patients with, and without, symptoms. No value for dissociated heterophoria was found that would discriminate between the symptomatic and asymptomatic patients. It is concluded that heterophoria measurement is not useful as a routine procedure. In the case of associated heterophoria, for the under 40 years age group, patients with a value of 1 prism dioptre or more are more likely to have symptoms than not, and one-third of patients with close work problems have a value of 2 prism dioptres or more. It was rare to find any asymptomatic patient with a value as high as this. In the 40 years and over age group those with an associated heterophoria of 2 prism dioptres or more are more likely to have symptoms than not. One-third of patients in the latter age group, with symptoms, have a value of 3 prism dioptres or more whilst it is rare to find an asymptomatic patient with a value as large as this.