Purchase this article with an account.
J. M. Holmes, M. Melia, D. L. Chandler, S. P. Christansen, Pediatric Eye Disease Investigator Group; Defining Stability and Instability in Esotropia. Invest. Ophthalmol. Vis. Sci. 2008;49(13):1804.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The timing of surgery in childhood esotropia is controversial; some surgeons operate as soon as possible, whereas others wait to demonstrate a stable angle of misalignment, each approach intended to optimize stereoacuity and motor outcomes. Nevertheless, the question of "how to define stable misalignment?" has not been rigorously studied.
We used our recent estimate of alternate prism cover test measurement error (95% limit of agreement of 8PD) to conduct 10,000 Monte Carlo simulations for subjects with "no ocular alignment change" over 4 visits, where the only changes in ocular alignment would be sampled from a distribution of measurement errors. Additional simulations were conducted for a "defined ocular alignment change" (10,000 simulations each, for changes of 5PD, 10PD, 15PD, and 20PD) over 4 visits, where the changes in ocular alignment were modeled as the sum of measurement error and actual change. The number of visits (4) was chosen to parallel an ongoing study of the course of esotropia. We then estimated sensitivities and specificities for specific pragmatic classification rules for stability (all 4 measurements within 0PD, 5PD, 10PD or 15PD) and for instability (at least 2 measurements differ by 10PD or more, 15PD or more, or 20PD or more).
The rules: "within 0PD over 4 visits," "within 5PD," "within 10PD," and "within 15PD," had sensitivity for stability (no real change) of 5%, 43%, 84% and 98% respectively. These rules had false positive rates for stability (a real 10PD change) of 1%, 15%, 53% and 85% respectively. For defining instability, the rules; "at least 10PD between 2 measurements," "at least 15PD" and "at least 20PD," had sensitivities to detect a real 10PD change of 84%, 46% and 14% respectively, and false positive rates for instability (a real 0PD change) of 53%, 13% and 1% respectively.
For 4 esotropia measurements over time, a "within 5PD" rule seems reasonable for defining stability, since higher thresholds had unacceptable false positive rates. The "at least 15PD between 2 measurements" rule for seems reasonable for defining instability, since lower thresholds had unacceptable false positives rates. If neither the rule for stability or instability is met, then the misalignment would be classified as uncertain. These data will be used in future analyses to define rules that can be applied to an undefined number of prospective measurements in clinical practice.
This PDF is available to Subscribers Only