Validity of self-reported nail counts in patients with onychomycosis: A retrospective pilot analysis

Onychomycosis (fungal infection of the toenails or fingernails) is common, affecting up to 8% of the general population. Telephone and mailed surveys conducted to study this disease are usually completed via patient self-report. The validity of the counts of diseased nails reported by the patients participating in these surveys has not been established.

The aim of this study was to assess the interrater agreement between patient and health care professional (HCP) counts of affected nails in patients with onychomycosis.

Patient and HCP counts of infected toenails and fingernails corresponding to a total of 567 patients originating from 3 different clinical trials were retrospectively analyzed. All these patients were initially classified as mycologically positive for toenail onychomycosis, and all 3 trials used identical self-administered questions regarding counts of infected toenails and fingernails. The level of agreement between patient versus HCP counts of infected toenails and fingernails (based on collapsed and uncollapsed counts) was measured using the quadratically weighted kappa statistic.

Thirty-eight percent (213/554, 95% confidence interval [CI] = 35% to 43%) of the 554 patients with self-reported and HCP counts of infected toenails had exact agreement; for these 554 patients, quadratically weighted kappa = 0.47 (95% CI: 0.40 to 0.54). The highest agreement rates-79, 62, and 50%-correspond to the patients with HCP counts of affected toenails equaling 1, 2, and 10, respectively; that is, HCP counts at the extremes. The agreement rates ranged from 3% to 31% for the remaining patients with HCP counts between 3 and 9. Patients tended to undercount affected toenails, irrespective of the number of affected toenails counted by HCPs. No statistically significant association was found between agreement rates and whether or not the patients had been previously treated for toenail onychomycosis (P = .91). A marginally nonsignificant association of increasing agreement rates with increasing disease duration was observed (P = .06). Among the patients who reported one or more affected fingernails (n = 122), percent agreement was similar to that of toenails (41%, 95% CI: 33.0 to 50.0%).

This pilot study is limited by its retrospective design, its homogeneous group of patients (predominantly white, male veterans), and the fact that information on intraobserver and interobserver reliability was not obtained from all of the 3 clinical trials, the sources for this analysis.

Patients tend to undercount affected nails as compared to HCPs. Reliance on self-reported counts in epidemiological surveys may lead to an inaccurate, underestimate of disease severity.