Applying Estimation and Robust Methods to Compare the Intraocular Pressure Measured by Tonometers in the Eye Dynamics and Engineering Network (EDEN) Aqueous Humor Dynamics (AHD) Clinical Trial

Abstract

Purpose : To compare IOP measurements from four different tonometers taken on 121 healthy participants during the Eye Dynamics and Engineering Network (EDEN) Aqueous Humor Dynamics (AHD) clinical trial.

Methods : Studies of agreement examine the distance (difference) between readings made by different devices (or observers) measuring the same quantity (or subjects). Four tonometers (iCare™ rebound tonometer (RT), Ocular response analyzer Goldmann-correlated (GC), pneumatonometer (PN), and Goldmann applanation tonometer (GAT)) were used for gathering baseline IOP in the AHD study. A recommended set of comparisons for tonometer IOP readings taken within the same visit were analyzed including Bland-Altman Limits of Agreement (LoA), Lin’s Concordance Correlation Coefficient (CCC), Intraclass Correlation Coefficients (ICC), Kendall’s coefficient of concordance (W), robust equivalence (TOST-two one-sided tests) in tandem with null hypothesis significance testing, percent of observations within 2 mmHg, and Gardner-Altman estimation plots using bootstrap-coupled estimation.

Results : Within visit IOP was compared between tonometers (paired observations) at baseline for both eyes. The effect sizes (mean difference, md) for RT, GC, and GAT were below any clinically important difference (e.g., RT-GAT md ± 95% CI = {-0.33, 0.45, 1.23} mmHg, B-A 95% LoA = {-3.95, 4.84} mmHg). IOP ranged from 7-21 mmHg as measured by GAT. We also compared the mean IOP by RT and GC without proparacaine to the mean IOP with proparacaine; the mean change in IOP was a decrease by 1.2 ± 1.9 mmHg OD and a decrease by 1.1 ± 1.7 mmHg OS that was statistically significant for both eyes (p < 0.0001). The md in IOP between PN-GAT md ± 90% CI = {1.7, 2.1, 2.6} mmHg OD and {2, 2.4 ,2.9} mmHg OS (p < 0.0001).

Conclusions : Although one cannot conclude that the devices are statistically equivalent, we found that PN reads greater than 2 mmHg higher than either RT, GC, or GAT. Application of topical proparacaine results in a statistical lower IOP when compared to no proparacaine.

This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.

 

An example estimation plot for OD eyes as measured by RT and PN. The slopegraph depicts pairs of observations, the effect size (Δ, mean difference) is indicated by the dot, and the resampled distribution of the effect size with a 95% CI is illustrated by the thick vertical line.