Precision in automated refraction

Forty-six eyes were examined with automated refraction with the Nidek 1000-AR autorefractometer to determine how large variation there was in readings of each patient, under dry conditions (without cycloplegia) and in atropine and cyclopentolate cycloplegia. Likewise, the differences between methods of cycloplegia were analysed with regard to sphere, cylinder power, and axis. Overall the variation in each set of measurements was greatest for the spherical component, and a larger variation was found in the youngest age group. Variation in cylinder power and axis was small. Cycloplegics had a significant influence on the spherical component of automated refraction, and a mean difference of 0.76D was found between atropine and dry readings, and 0.23D between atropine and cyclopentolate readings. The differences between cycloplegic and dry readings in cylinder power and axis were insignificant. A regression model relating spherical power of dry and cyclopentolate automated refraction was developed, and the predictive power of this equation was tested.