Observer agreement in toenail disorders: implications for diagnosis and clinical research

Abstract

Background: Toenail disorders are frequent, especially onychomycosis. The interobserver variability of nail signs needs to be known before these signs can be confidently applied for diagnosis.

Objectives: To describe observer agreement in toenail findings as described by dermatologists in standard clinical practice, focusing on signs that could be useful for diagnosis of onychomycosis.

Methods: Prospective cross-sectional study in five dermatology departments. Eighty-six patients with abnormal toenails that could have onychomycosis as a differential diagnosis were independently examined by a pair of dermatologists using a predefined questionnaire, to describe the presence of 10 findings on previous history and 14 physical signs.

Results: Agreement was fine for previous history findings: it was very good (kappa > 0.81) for previous diagnosis of diabetes, smoking and use of public dressing rooms or swimming pools. Agreement was good (kappa 0.61-0.80) for immune suppression (drugs or cancer), previous diagnosis of fungal disease and worsening in the last year. It was moderate (kappa 0.41-0.60) for previous diagnosis of arterial disease, trauma induced by work or sports, and distal vs. proximal or lateral vs. central start of the lesion. Agreement was worse for physical signs: we found good agreement for the presence of the same disease in fingernails, abnormal plantar desquamation, deformity causing nail trauma, and subungual hyperkeratosis. It was moderate for the presence of nail destruction, tinea interdigitalis, onycholysis, and the type of material obtained by subungual curettage (dust vs. hard). Agreement was fair (kappa 0.21-0.40) for the presence of longitudinal or transverse striae, trachyonychia, pachyonychia, and change in colour of the nail plate. Pitting was too infrequent to allow for kappa calculation. Chance expected agreement was between 51% and 84% for all signs except pitting.

Conclusions: Agreement is adequate for most signs. It is low for the presence of longitudinal or transverse striae, trachyonychia, and change in colour of the nail plate. Pitting is rare in toenails.