Baseline Nail Thickness and Great Toenail Involvement Are Associated With Poor Response to Topical Antifungal Therapy for Onychomycosis: A Lesion‐Level Generalized Estimating Equations Analysis

ABSTRACT Topical antifungals are widely used for onychomycosis; however, lesion‐level morphological predictors of response have rarely been quantified while accounting for within‐patient correlation. We retrospectively studied patients with onychomycosis treated with topical luliconazole 5% nail solution or efinaconazole 10% solution between 2017 and 2021. Eighty lesions from 66 patients were identified. Treatment response was evaluated using lesion‐level generalized estimating equations (GEE) clustered by patient ID with an exchangeable working correlation structure and robust (sandwich) standard errors. Improvement rate (continuous) was analyzed using a linear GEE, and complete cure (yes/no) using a logistic GEE. Great toenail involvement was consistently associated with poor response, showing a lower improvement rate, β = −0.461 (standard error [SE] = 0.145; p = 0.0015), and markedly lower odds of complete cure, odds ratio (OR) 0.07 (95% confidence interval [CI] 0.01–0.37; p = 0.002). Greater baseline nail thickness was associated with lower odds of complete cure, OR 0.17 (95% CI 0.07–0.42; p < 0.001), although it was not significantly associated with improvement rate, β = −0.089 (SE = 0.058; p = 0.125). Use of luliconazole 5% nail solution (vs. efinaconazole 10% solution) was associated with a lower improvement rate, β = −0.310 (SE = 0.154; p = 0.044) and lower odds of complete cure, OR 0.07 (95% CI 0.01–0.60; p = 0.015); these associations should be interpreted cautiously because treatment allocation was not randomized. Sensitivity analyses using a linear mixed‐effects model showed consistent directions of effect. Great toenail involvement appears to be a robust predictor of poor response to topical therapy, and increased nail thickness may particularly hinder achieving complete cure; these findings warrant cautious interpretation given the retrospective, non‐randomized design.