Observer agreement in recording the clinical signs of nail disease and the accuracy of a clinical diagnosis of fungal and non-fungal nail disease.

Abstract

Onychomycosis is increasing in incidence. To date, no studies have examined the detection of abnormalities of the nail apparatus, nor the accuracy of a clinical diagnosis of onychomycosis and non-fungal nail disease. To assess the agreement between and within different groups of observers in detecting signs of nail disease, and to obtain information regarding clinical diagnostic skills. An observational study was performed. Nine observers, including dermatologists, mycologists, general practitioners and a dermatology clinical assistant, completed a questionnaire containing 21 clinical signs of nail disease during examination of nine patients, five with onychomycosis and four with non-fungal nail disease. Observers were additionally requested to suggest the most likely underlying diagnosis for the nail dystrophy. Mean pair observer agreement values were calculated for each of the clinical signs, between all observers and within groups of observers. The chance-corrected agreement index, kappa, was determined. From the clinical diagnoses given, the positive predictive value of a diagnosis of fungal and non-fungal nail disease was calculated. There was substantial between-observer agreement on only three clinical signs: abnormal nails on both hands, abnormal toenails and abnormal fingernails. More specific signs of nail disease such as onycholysis elicited weaker agreement. All observers showed accuracy in making a clinical diagnosis of fungal nail disease, with a mean positive predictive value of 0.91, compared with 0.77 for non-fungal nail disease. Our results showed that agreement between observers, in recording signs of nail disease, was generally poor. The clinical diagnosis of onychomycosis was highly likely to be correct, suggesting that other criteria are being employed by individuals in reaching the diagnosis.