Retrospective study of onychomycosis patients treated with ciclopirox 8% HPCH and oral antifungals applying artificial intelligence to electronic health records

Abstract

We conducted a multicenter retrospective analysis of 408 patients diagnosed with onychomycosis who attended three tertiary care Spanish hospitals. The study was conducted to assess the clinical characteristics and outcomes of onychomycosis patients undertaken combined treatment with ciclopirox 8% hydroxypropyl chitosan (CPX 8% HPCH) nail lacquer and oral antifungal agents (terbinafine, itraconazole, and fluconazole). Data were extracted and analyzed using the EHRead^® technology based on natural language processing and machine learning and using SNOMED CT terminology. The mean age was 51.1 years, with 54.4% females. Repeated nail trauma was the most common risk factor (7.1%). More than half of the patients (56.6%) underwent fungal culture. Terbinafine, itraconazole, and fluconazole were used in 67.7%, 20.8%, and 11.5% of patients, respectively. Treatment synchronicity distribution revealed that 59.1% of patients started concomitant treatment with CPX 8% HPCH and oral antifungals, 27.9% had initial oral antifungals, and 13% had initial nail lacquer. The response to treatment (positive response 15.7%, presumed positive 59.8%) was unrelated to treatment synchronicity or type of antifungal agent. Erythema (5.6%), diarrhea (4.9%), and fever (4.2%) were the most frequently registered potential adverse events. These findings provide valuable insights for physicians regarding the management of patients with onychomycosis in daily practice.

Keywords: Artificial intelligence; Ciclopirox 8% HPCH; Fluconazole; Itraconazole; Machine learning; Onychomycosis; Real-world data; Terbinafine.

Fig. 1

Flow chart of the study population and subgroups by antifungal agent and treatment. synchronicity.

Fig. 2

Time to positive response in patients who registered positive response within the study period for the combination of oral antifungal agents and CPX 8% HPCH nail lacquer. Note that due to curve stabilization, X-axis was truncated at 30 months. Only patients with an event (positive response) occurring after the index date were included. Accordingly, 13 patients were excluded. Patients were censored at CPX 8% HPCH discontinuation or end of follow-up. Dotted line marks 5.08 months which is the median time to response for patients included in this figure.

Fig. 3

Time to CPX 8% HPCH and oral antifungal combination discontinuation in the entire study population. X-axis truncated at 24 months due to curve stabilization and Y-axis truncated at 0.13 given no events appeared beyond this point. Only patients with a discontinuation or censoring occurring after the index date were included. Accordingly, 26 patients were excluded. Patients were censored at the end of follow-up.

Fig. 4

Time to CPX 8% HPCH and oral antifungal combination discontinuation in patients with registered positive response. Note that due to curve stabilization, X-axis was truncated at 18 months. Only patients with a discontinuation or censoring occurring after the index date were included. Accordingly, 9 patients were excluded. Patients were censored at the end of the follow-up. Dotted line marks 3.21 months which is the median time to discontinuation for patients included in this figure.