Monotherapy for Toenail Onychomycosis

A Systematic Review and Network Meta-analysis

A.K. Gupta; K.A. Foley; R.R. Mays; N.H. Shear; V. Piguet


The British Journal of Dermatology. 2020;182(2):287-299. 

In This Article

Materials and Methods

Systematic Review

This systematic review and meta-analysis was conducted in agreement with the modified 32-item Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) extension statement for NMA 2015.[15] The selection of databases, eligibility criteria, outcomes of the review and analytic methods were defined a priori in an internal protocol and registered on PROSPERO under the code CRD42018086912 (Table S1; see Supporting Information). Studies were eligible for inclusion if they were RCTs of treatment for toenail onychomycosis in adults 18 years or older and published in the English language. RCTs could be multiple-arm trials utilizing placebo, vehicle, or sham-controls, active comparators (e.g. antifungal treatment) or both. Cross over and within-participant designs were not included. Onychomycosis was required to be diagnosed mycologically, as determined by study investigators. Oral, topical or device-based monotherapies and any combination of therapies were eligible for inclusion.

Outcomes of interest were mycological cure, complete cure, treatment success, adverse events and QoL. Mycological cure was defined as negative results from either potassium hydroxide (KOH) microscopy or periodic acid–Schiff stain (PAS), and a negative culture; complete cure was defined as mycological cure plus 100% clear nail; and treatment success was defined as mycological cure plus ≤ 10% nail involvement. Safety parameters were considered and defined as number of patients with at least one adverse event. QoL was defined as patient-reported improvement in symptoms or QoL, as defined by study investigators. Endpoints as close to 48–52 weeks as possible were selected. For more details, see the PROSPERO register file and PRISMA NMA 2015 checklist (Table S1, Table S2; see Supporting Information). Given the study characteristics of included studies (e.g. adults, outpatient setting, consistent dosage within treatments, outcome endpoints), treatments included in the NMA are exchangeable and meet the assumption of transitivity.[16]

Databases Searched and Study Identification

Details regarding the databases searched and study identification for this review are provided in Appendix S1 (see Supporting Information). Briefly, literature searches of PubMed, Scopus, OVID (Embase and Medline), and were performed on 28 March 2018 using the search terms 'onychomycosis' or 'tinea unguium' with no date limitations (K.A.F.). Two authors (K.A.F. and R.R.M.) screened titles and abstracts, reviewed the full text and extracted data from eligible studies. A sample search strategy can be found in Figure S1 (see Supporting Information).

Data Extraction

Treatment effects were evaluated based on the intention-to-treat cure rates whenever possible (number of patients with cure). In order to ensure data consistency and clinical relevance, decisions were made as to what treatments and treatment regimens were to be included in the NMA. Oral ketoconazole was excluded because it is no longer indicated for onychomycosis due to hepatotoxicity concerns.[17–19] Fluconazole doses of 150, 300 and 450-mg weekly for 9–12 months were collapsed as these have been shown to have equivalent efficacy.[20] Clinically relevant doses of oral terbinafine and itraconazole were included: 250 mg terbinafine (500-mg pulse) and 200 mg itraconazole (400-mg pulse) for 3–4 months. Topical treatments were to be used for at least 9 months.

Quality of Evidence and Risk-of-bias Assessment

Details regarding the quality of evidence and risk-of-bias assessment for this review are provided in Appendix S1 (see Supporting Information).[21] The Cochrane Collaboration's tool was used with two authors (K.A.F. and R.R.M.) issuing judgements of low, high or unclear risk of bias. Individual studies were assessed for bias associated with randomization and allocation concealment procedures, blinding of participants, personnel and outcome assessors, attrition data and selecting reporting.[21] The quality of evidence was assessed using the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) approach (risk of bias, inconsistency, indirectness, imprecision and publication bias) for direct comparisons of treatments (e.g. from meta-analyses).

Meta-analysis of Direct Treatment Effects

Data extracted from trials were combined by a random-effects model, with effect sizes expressed as odds ratios (ORs) of achieving each outcome in the treatment arm vs. the control/comparator arm. Total effect size was calculated by the Mantel–Haenszel method and heterogeneity was evaluated with the I 2 statistic. Statistical analysis was performed with RevMan 5·3 (, with two-tailed P-values < 0·05 considered significant. Forest plots and funnel plots were obtained for each outcome analysed and are included in the supplementary material.

Network Meta-analysis

NMA was used to make mixed comparisons among the therapeutic options and to rank treatments, using the program Aggregate Data Drug Information Software (ADDIS) version 2 program, which is software designed to run NMA from a Bayesian framework (Appendix S1; see Supporting Information).[22] Inconsistency between direct and indirect evidence in the network was analysed using difference of means. A P-value < 0·05 indicated significant inconsistency between the direct and indirect evidence in the network.[23]