Multivariate Analysis to Predict the Horizontal Anterior Chamber Diameter from Other Anterior Chamber Parameters Using Scheimpflug Imaging: Is There a Better Method Than Adding a Fixed Constant?

Abstract

ABSTRACTPurpose: To evaluate the predictive association between the horizontal anterior chamber diameter (HACD) and other measurements using Scheimpflug imaging, and to assess whether a method superior to using fixed constants can be proposed.Methods: This hospital-based study was performed in the Cornea and Refractive Surgery Services, New Medical Center (NMC) specialty hospital, Abu Dhabi. Initially, 100 candidates were included serially in the model building group (group1). All candidates underwent detailed evaluation and Scheimpflug imaging (CSO, Sirus, Italy). Subsequently, another 100 candidates were included serially in the validation group (group 2). Candidates in both groups underwent the same tests.Results: In group 1, the mean HACD was 12.25 ± 0.48 mm. This measurement correlated significantly with the horizontal visible iris diameter (HVID), anterior chamber depth, angle (ACA), and volume (ACV) (r = 0.2–0.7). The overall regression equation was HACD = 5.62 + 0.01 × ACV + 0.45 × HVID − 0.013 × ACA (adjusted R2 = 0.66, p < 0.001). Situation-based equations derived from the outcomes of group 1 were tested in group 2. Equation A included all factors found significant during model-building, Equation B included only non-volumetric significant factors, and situation C only included the HVID. Equations D through F used direct substitution by HVID plus a constant to predict HACD (the constant was 0.0, 0.5, and 1.0 for Equations D, E, and F, respectively).The predicted HACD (pHACD) was within ±0.5 mm of the actual HACD (aHACD) in 93%, 88%, 87%, 83%, 69%, and 16% case for Equations A, B, C, D, E, and F, respectively. The mean prediction error was −0.14 ± 0.27 mm (p = 0.08), −0.15 ± 0.28 mm (p = 0.03), −0.18 ± 0.30 mm (p = 0.001), −0.18 ± 0.32 mm (p < 0.001), 0.33 ± 0.35 mm (p < 0.001), and 0.83 ± 0.35 mm (p < 0.001) for Equations A, B, C, D, E, and F, respectively.Conclusion: Regression equations using Scheimpflug-derived anterior chamber parameters may predict HACD to varying degrees, depending on the input parameters. Adding 0.5 or 1.0 mm to the corneal diameter (CD) to estimate the HACD is not recommended.