Abstract
Purpose: :
The Kay Picture Test was introduced 25 years ago and has since been used widely in both clinical and research settings. However, some pictures are dated and with the introduction of computerised tests and presentation of pictures in logMAR format a redesign of the test was required.
Methods: :
Twenty five pictures were drawn by a graphic design artist in a 10 x 10 square with the line width and spacing subtending1min arc (one tenth of the square). Each picture touches the boundary of the square on all four sides, allowing easier correct spacing in a crowded format. Visual acuity was measured at 6 metres via computer presentation, using an interleaved staircase of 25 pictures plus 5 ETDRS letters and 8 Llandolt C rings. The staircase incorporated 8 reversals and a measurement was calculated from an average of the last 6 reversals.
Results: :
Fifty adult subjects with normal acuity (≤0.2 logMAR) were recruited. Comparison of the acuity level of each picture compared to the average of the 5 letters and the average of the 8 Llandolt rings showed that the picture acuity was on average lower for the pictures compared to letters (-0.08 logMAR, range of mean±sd: 0.01±0.09 to -0.18±0.04) and also compared to Llandolt rings (-0.14 logMAR, range -0.06±0.09 to -0.25±0.06). The variability of the 5 letter scores was 0.07logMAR (average across subjects of the maximum variability within subjects) and 0.14logMAR for the Llandolt rings. Restricting the variation across pictures to 0.14logMAR eliminates three pictures (flower, man and umbrella). Reducing the variability to the level of the letter variability eliminated a further 8 pictures. Bland Altman analysis comparing the ETDRS letters to the Llandolt rings showed a mean bias of 0.07 with the 95% limits of agreement at -0.03 to 0.16. Six pictures had either the same or smaller limits of agreement when compared to the letters.
Conclusions: :
Based on these findings, three pictures have been removed from the current selection. The next phase of testing will determine which pictures are most commonly recognised by young children presented initially in isolation and subsequently with crowded bars. The final selection of pictures will incorporate those with the maximum recognition and minimum variability.