Purchase this article with an account.
Dipesh Patel, Phillippa Cumberland, Isabelle Russell-Eggitt, Bronwen Walters, Jugnoo Rahi, ; How should we assess reliability of visual field assessment in children?. Invest. Ophthalmol. Vis. Sci. 2013;54(15):5675.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
In the OPTIC study, we are investigating a number of questions about kinetic and static perimetry in children using Goldmann, Octopus and Humphrey perimeters. Here we report how individual examiner-based assessment (EBA) compares with automatically generated 'reliability' indices (RI) used commonly in adult perimetry (fixation losses, false positive/negative rates).
To date 64 children aged 5-15 years, without ocular pathology that could cause a visual field defect i.e controls/normative subjects have undergone kinetic perimetry (Goldmann and Octopus) and static automated perimetry (SAP, using Humphrey SITA 24-2 FAST) using standard protocols with testing by one examiner. For each test, the examiner scored overall reliability (‘good’,’fair’ or ‘poor’) together with an assessment of the subjects’ comprehension of instructions, fatigue, fixation, response to visual and auditory stimuli and behaviour. Additionally, subjects self-rated the test difficulty (using a 5-point Likert scale), and duration of each phase of testing was recorded. The automatically generated SAP reliability indices (RI) were noted, with a test being recorded as ‘unreliable’ with ≥20% fixation losses or ≥15% false positives, as per conventional adult thresholds. We examined agreement, by age, between RI and EBA.
No significant agreement was found between EBA and ‘fixation losses’ as an RI (test for trend; p=0.887). EBA and ‘false positive’ measures demonstrated good agreement (p<0.001; Table 1). Analysis of EBA by age (2 groups; 5-8 and 9-15 years) showed 94% (17 of 18) of older children and 41% (19 of 46) of younger children achieved a ‘good’ rating (χ2, p<0.001). Only 1 child, aged 8, had difficulty understanding test instructions. Rest breaks were required to complete testing in 8.7% of younger participants but older children could complete the assessment without these.
In comparison to a qualitative examiner assessment, automated fixation loss measurements may have limited value in assessing reliability in children, with traditional thresholds used for adult testing erroneously under-estimating the reliability of paediatric perimetry. Where false positive rates are to be used to assess reliability, we advocate they are complimented by a qualitative assessment to avoid potentially informative perimetric tests being disregarded.
This PDF is available to Subscribers Only