The accuracy of the amblyopia treatment study visual acuity testing protocol

J AAPOS. 2003 Dec;7(6):406-12. doi: 10.1016/j.jaapos.2003.07.009.

Abstract

Purpose: To study the accuracy of the newly proposed Amblyopia Treatment Study (ATS) visual acuity testing protocol for 3- to 6-year-old children. Because no "gold standard" is available for acuity testing in pediatric patients, accuracy was evaluated using computer simulations based on a psychometric model.

Methods: Monte Carlo simulations of ATS acuity data were generated using a psychometric model that accounts for true acuity, noise in the visual system, and the rate of inadvertent misses. We varied true acuity from 20/15 to 20/400 (-0.1 to 1.3 logMAR). Visual system noise was represented by the slope beta of the psychometric function and ranged from 1 (noisy) to 8 (not noisy). The rate of inadvertent misses ranged from 0% to 10%. Accuracy of the ATS protocol was evaluated in terms of precision, bias, and stimulus range limitations. The same model was fitted to experimental ATS acuity data, thus allowing us to study the distributions of acuity, visual system noise, and level of attentiveness in 126 children ages 3 to <7 years.

Results: For conditions with little noise in the visual system (beta > 2), precision was well within 0.1 logMAR (corresponding to one line on a logMAR letter chart), except for acuities worse than 1.2 logMAR, and decreased to 0.15 to 0.2 logMAR for beta = 1. Bias was negligible, except in noisy conditions, where the ATS protocol tended to overestimate acuity by one line at the poor end of the true acuity range and underestimate acuity at the good end of the true acuity range. Effects of the rate of inadvertent misses were small. Fits to the real data showed a wide range of slope parameters, but only 11% had beta < or = 2. The rate of inadvertent misses was < or = 2% in 89% of cases.

Conclusion: The simulations suggest that the ATS protocol offers an accurate method for assessing visual acuity in children in the range of 3 to 6 years of age with both precision and bias within 0.1 logMAR for typical values of the psychometric parameters.

Publication types

  • Comparative Study

MeSH terms

  • Amblyopia / diagnosis*
  • Amblyopia / therapy
  • Bias
  • Child
  • Child, Preschool
  • Clinical Protocols
  • Computer Simulation
  • Female
  • Humans
  • Likelihood Functions
  • Male
  • Psychometrics
  • Reproducibility of Results
  • Vision Tests / instrumentation
  • Vision Tests / methods*
  • Visual Acuity*