Risk difference plots (random effects). Pooled risk difference, −0.23
(95% confidence interval, −0.15 to −0.32).
Crawford F, Young P, Godfrey C, Bell-Syer SEM, Hart R, Brunt E, Russell I. Oral Treatments for Toenail OnychomycosisA Systematic Review. Arch Dermatol. 2002;138(6):811-816. doi:10.1001/archderm.138.6.811
DamianoAbeniMD, MPHRosamariaCoronaDSc, MDPaoloPasquiniMD, MPHMichael E.BigbyMDMoysesSzkloMD, MPH, DrPHHywelWilliamsMD
To identify and synthesize the evidence for the efficacy of oral treatments
for fungal infections of the toenails.
Systematic review of randomized controlled trials.
Oral treatments for dermatophyte infections of the toenails.
Main Outcome Measures
Cure confirmed by microscopy and culture results in patients with clinically
diagnosed fungal infections. Data relating to the clinical cure rates were
also extracted from the trials.
A pooled analysis of 2 trials comparing mycological cure rates from
continuous treatment with terbinafine (250 mg/d for 12 weeks) and continuous
treatment with itraconazole (200 mg/d for 12 weeks) found a statistically
significant difference in 11- and 12-month outcomes in favor of terbinafine
(risk difference, −0.23 [95% confidence interval, −0.32 to −0.15];
number needed to treat, 5 [95% confidence interval, 4 to 8]). An analysis
of clinical cure rates was not possible because of the diversity of definitions
used in researching the effectiveness of oral antifungal drugs for onychomycosis.
Only 3 trials gave a clear definition of clinical cure and presented data
for these outcomes.
There is good evidence that a continuous regimen of terbinafine (250
mg/d) for 3 months is the most effective oral treatment for fungally infected
toenails. Consensus among researchers evaluating oral antifungal drugs for
onychomycosis is needed to establish meaningful definitions of clinical cure.
Most trials were funded by the pharmaceutical industry; we found little independent
research, and this may have introduced bias to the review.
ORAL TREATMENT may be the only effective treatment for infected toenails.
A previous systematic review of topical compounds for fungal toenail infections
found little evidence of effectiveness for topical therapies.1
Griseofulvin, for many years the only oral therapy, is inexpensive but
has a protracted administration time. The newer drugs, azoles and allylamines,
have improved cure rates of nail infections but are vastly more expensive
than griseofulvin, and some general physicians believe that these resources
might be more beneficially used to treat potentially life-threatening conditions.2
The ingredient costs of oral antifungal drugs alone accounted for £30
million of the National Health Service (NHS) prescribing budget in 1998 (personal
communication, Department of Health, Statistics Division, Branch SD1E, September
1, 1999). Because 5% of UK adults older than 55 years have infected toenails,
the prescribing costs—coupled with consultation costs—represent
substantial NHS expenditure if all those affected sought treatment.3
Trials of oral antifungal drugs for toenail infections have described
a variety of cure rates for the drugs that are prescribed. The present systematic
review examines all available data from evaluations of these therapies and
includes a statistical summary of their clinical effectiveness.
We searched the following 5 databases up to March 2000: MEDLINE, EMBASE,
CINAHL, Bath Information and Data Services (BIDS), and the Cochrane Controlled
Trials Register. The MEDLINE search strategy is published elsewhere.4 The following 2 economic databases were searched up
to January 2000: NHS Economic Evaluation Database (NEED) and ECONLIT. Four
other databases were searched up to January 1997: CAB Health Abstracts, HEALTHSTAR,
and Database of Abstracts of Reviews of Effectiveness (DARE) but did not identify
any additional randomized clinical trials. The following 3 podiatry journals
not listed in these databases were searched manually: The
Foot, Journal of British Podiatric Medicine,
and British Journal of Podiatric Medicine and Surgery.
We obtained the Cochrane Skin Group's partial hand search of the British Journal of Dermatology. We searched the bibliographies of all
review articles identified and contacted all schools of podiatry in the United
Kingdom and pharmaceutical companies to identify unpublished or unlisted trials.
We considered all randomized clinical trials that evaluated oral treatments
for dermatophyte infections of the toenails. We included trials that used
microscopy and culture to confirm the presence of dermatophytes. We included
duplicate trials only once. We excluded trials that also evaluated the treatment
of fungal infections of the fingernails in which the foot-specific data were
not presented and trials that included patients with yeast and mold infections
of the toenails. Four reviewers working in pairs (F.C. and E.B.; R.H. and
S.E.M.B.S.) independently applied these criteria to each trial located. There
were no European language restrictions.
All reviewers independently summarized the included trials and appraised
their quality of reporting based on items from published checklists.4- 6 The 12-quality criteria
included the following: aims clearly defined, prior sample size calculation
reported, inclusion and exclusion criteria defined, subjects blinded, method
of randomization defined, baseline comparability of groups reported (age,
sex, and duration of complaint), interventions defined, outcome assessment
blinded, compliance assessed, and trial analyzed by intention to treat.
Reports were scrutinized for cure rates from mycological investigations
(microscopy and culture) and clinical cure rates. The definition(s) of clinical
cure was noted, and data that could be used in a reanalysis (absolute numbers
or means and a measure of variance) sought.
For each trial we calculated the cure rates at follow-up (3, 6, 9, and
12 months) from the reported mycological results, with "cure" defined as negative
results on microscopy and no growth of dermatophyte in culture. We also collected
data regarding clinical cure. We estimated the difference in the proportion
of patients cured with 95% confidence intervals in the unpaired proportions
that constitute the risk difference.7 We also
calculated the numbers needed to treat for those comparisons that were statistically
significantly different. To estimate differences between treatments, we pooled
trials that evaluated similar interventions and controls. Because there was
clear evidence of heterogeneity between trials (P<.001
for the Q-combinability test [part of the DerSimonian-Laird random effects
analysis], which is known to have low power) we used a random effects model.8 Finally, we combined the evidence of direct "head-to-head"
We identified 50 trials evaluating treatment efficacy,9- 58 32 of which we included9- 40 (Table 1,
Table 2, Table 3,
Table 4, and Table 5).
Eighteen trials were excluded for the following reasons: duplicate reports,42,43,46,48- 50,52,53 combined hand and feet data or combined with topical data,41,47,51,54,55,57 no mycology assessment,56 protocol deviation,45 and data not clearly presented.44,58
We found 6 placebo-controlled trials of terbinafine and itraconazole
that presented outcomes at 12 weeks11- 16:
3 trials evaluating itraconazole vs placebo (N = 433) found that itraconazole
had greater effectiveness,11- 13 and 3 trials evaluating terbinafine vs placebo (N = 337) found that terbinafine
was more effective.14- 16 We regarded outcomes at 3 months as clinically irrelevant; a more clinically
meaningful assessment of the treatment were outcomes at 9 months.
Only 2 trials of direct comparisons between treatment with itraconazole
(200 mg/d) and terbinafine (250 mg/d) gave data on outcomes at 11 and 12 months.
These were pooled in a meta-analysis using a random effects model, which showed
a risk difference in favor of terbinafine (−0.23 ([95% CI, −0.15
to −0.32]) (N = 501) (Figure 1).30,40 Both trialists found terbinafine
to produce clinically significant improvements in the length of the unaffected
nail in great toenails.
Two studies compared different regimens of itraconazole with terbinafine.17,38 Tosti et al38 compared intermittent treatment with itraconazole (400 mg/d) with intermittent
treatment with terbinafine (500 mg/d) and continuous treatment with terbinafine
(250 mg/d) (N = 60), while Evans and Sigurgeirsson17 compared 12- and 16-week regimens of continuous terbinafine (250 mg/d) with
intermittent itraconazole (400 mg/d) for 12 and 16 weeks (N = 421). The resultant
data from these 2 trials lie within the confidence intervals for the dose-finding
analysis, which suggest no advantage in higher or prolonged dosages.
Only 1 trial investigated the use of terbinafine in a dosing schedule.18 Alpsoy et al18 did
not detect a difference in cure rates when comparing a continuous regimen
of terbinafine (79%) with an intermittent regimen (74%), but the trial population
was small (N = 47).
Three trials evaluated the value of continuous and intermittent dosage
schedules of itraconazole. Havu et al21 found
that the cure rate was not significantly different between 3 months of continuous
treatment with itraconazole (200 mg/d) and an intermittent (1 week of treatment
in every 4 [1:4]) regimen of 400 mg/d (N = 121). De Doncker et al20 evaluated 2 different intermittent schedules of itraconazole
of 3 and 4 months (400 mg/d; 1:4 weeks) and found that the cure rates at 24
weeks were 64% for patients who received the shorter treatment and 72% for
those who received 4 months of treatment.20 The trial had a small number of patients (N = 50), and no significant difference
was detected between the treatments. Shemer et al10 also did not find that a higher long-term cure rate was associated with a
200-mg/d continuous regimen of itraconazole compared with either of the 2
intermittent regimens (1:4 weeks) for 12 or 16 weeks (N = 64).10
Three studies comparing the effectiveness of itraconazole and griseofulvin
produced poor cure rates.28,29,36 Two studies with small numbers of patients (N = 80) did not detect a difference
in patient outcomes with griseofulvin (500 mg/d) and itraconazole (100 mg/d)
taken for 24 to 36 weeks: Walsoe et al29 found
that none of their 19 patients were cured with either drug while the 61 patients
in the trial by Piepponen et al28 had cure
rates of 30% in the griseofulvin arm and 36% in the itraconazole arm. Korting
et al36 compared 2 different dosages of griseofulvin
(660 mg/d and 990 mg/d) with 100-mg/d dosage of itraconazole taken for 18
months but detected no difference in the cure rates (N = 108). The cure rates
were 6% in both griseofulvin arms and 8% in the itraconazole arm.
Two studies compared terbinafine (250 mg/d) and griseofulvin (1000 mg/d)
for 6 to 12 months of treatment and concluded that the 250-mg/d dosage of
terbinafine is the superior treatment at 12- and 18-month outcomes.32,34 Faergenann et al33 reported that 500 mg of griseofulvin taken daily for 1 year had a significantly
poorer cure rate than 250 mg of terbinafine taken daily for 3 months.33 The cure rates were 84% in the terbinafine arm compared
with 45% in the griseofulvin arm.
Poor cure rates also occurred after 6 to 11 months of treatment with
ketoconazole (200 mg/d) and treatments with griseofulvin (1000 mg/d and 500
mg/d).26,27 These small trials
did not detect differences between the cure rates of the 3 different treatments.
We included 2 dose-finding studies evaluating fluconazole. Ling et al24 compared treatment with fluconazole (450 mg/wk) for
4, 6, and 9 months vs placebo. Cure rates were 61% for 9 months of treatment,
with significantly lower rates for shorter treatment times (4 months, 34%).
Scher et al25 detected improved cure rates
associated with higher dosages of 150 mg, 300 mg, and 450 mg of fluconazole
(once weekly) were compared with placebo, and the 450-mg/wk dosage produced
the highest cure rate (62%) after a maximum of 12 months.
A great deal of variation in the definition of clinical cure was found
between the included trial reports. Table
6 gives the various definitions stated in the methods sections of
each of the included trials. Table 6
also specifies whether the clinical cure data presented in the reports are
complete or incomplete.
The reviewers found deviations from the stated methods in most trial reports for clinical cure data. Sometimes deviations occurred when undefined clinical cure data were presented.13,31,37,39 In most reports the reverse was true: authors stated their intention to evaluate the effect of the drug of interest on certain clinical signs and symptoms but failed to present separate data for them. Instead, an estimate of "clinical success" was made without explicit reference to any individual clinical feature.9,12,20- 23,25,26,29,30,32 Some authors simply failed to present data for all intended outcomes.15,36,38 The use of line graphs, means without measures of variance, and P values to present clinical cure data hampered the production of a coherent data summary of clinical cure rates.13,16,18,19,28,34,40
Only Ling et al,24 Evans and Sigurgeirsson,17 and Gupta et al11 presented data in absolute numbers for clinical cures they defined a priori. These authors evaluated different drugs. Ling et al24 found that people who took fluconazole, 450 mg once weekly for 9 months, had the highest percentage of clinically normal nail with complete regrowth of healthy tissue at 6-month follow-up (37%). Evans and Sigurgeirsson17 found that a higher proportion of people who took terbinafine, 250 mg/d for 16 weeks, had 100% clear toenail and at least 5 mm of unaffected nail growth compared with people randomized to either 250 mg/d of terbinafine for 12 weeks or a 1 week in 3- or 4-intermittent cycle of itraconazole (400 mg/d). In both these trials the clinical cure rates were consistent with the mycological cure rates. Gupta et al11 compared intermittent itraconazole treatment (400 mg/d) with placebo and found that the proportion of patients who were clear of all signs of infection or markedly improved reflected the rates determined by mycological investigation. However, the proportion deemed clinically cured in the placebo arm was much lower than the proportion found to be cured using the mycological outcome criteria (1% and 28%, respectively).
Trichophyton rubrum was the most commonly identified infecting organism in all the included studies, with proportions reported to be from 68% to 100% of the identified fungi. Others species included Trichophyton mentagrophytes, Trichophyton tonsurans, Trichophyton interdigital, Trichophyton soudanense, and Epidermophyton floccosum. The review included trials with patients who had dermatophyte infection only; therefore, we cannot make conclusions about the efficacy of oral antifungals in the treatment of nondermatophyte onychomycosis.
The average score for all trials included in the review was 6.7 of 12.0. Blinded outcome assessment was only reported in 1 study,34 and the method used to conceal the random allocation from the researchers was reported in 4.17- 19,32 In 2 trials the inclusion/exclusion criteria were unclear,19,34 but all of the included trials re ported clear aims. Only half of the included trials reported comparable populations of patients at baseline for the duration of infection.13,16- 22,24,25,30,31,33,37
With the exception of 1 trial,9 all reported adverse events. The frequency of adverse events was not significantly different between treatment and placebo arms for terbinafine, itraconazole, and fluconazole.12- 15,22- 26 The data from dose-finding studies do not suggest that shorter treatment times (including intermittent regimens) result in fewer reported adverse events. No trials with placebo controls were identified for evaluations of griseofulvin and ketoconazole.
We found 32 randomized evaluations of terbinafine, itraconazole, griseofulvin, fluconazole, and ketoconazole that met our inclusion criteria. No difference in outcomes was detected between terbinafine (250 mg/d) and itraconazole (400 mg/d) at 3 months. However, a pooled analysis of mycological cure rates taken at 11 and 12 months (Figure 1) showed that terbinafine (250 mg/d) was more effective than itraconazole (400 mg/d) in the treatment of fungally infected toenails in the longer term.
The 1999 trial by Evans and Sigurgeirsson17 was not included in the meta-analysis because it compared continuous dosages of terbinafine (250 mg/d) with intermittent regimens of itraconazole (400 mg/d). The cure rates achieved in the terbinafine arms of the trial, relative to the itraconazole arms, are consistent with the meta-analysis in Figure 1 and provide evidence that continuous treatment with terbinafine (250 mg/d) is significantly more effective than an intermittent regimen of itraconazole (400 mg/d). The small trial by Alpsoy et al18 and data from the arm of the trial evaluating intermittent and continuous dosages of terbinafine in the study by Tosti et al38 both suggest that high levels of effectiveness can be achieved using a intermittent regimen of terbinafine, but this needs to be evaluated in a large randomized controlled trial.
There is no evidence that intermittent regimens of itraconazole produce statistically different cure rates from continuous schedules. Nor is there evidence that intermittent regimens or shorter treatment times result in fewer reported adverse events.
There was no evidence of significantly different rates of effectiveness in direct comparisons between itraconazole and griseofulvin or between itraconazole and ketoconazole, but the sample sizes in these trials were small. Fluconazole also produced only modest cure rates after particularly prolonged treatment times.
A large variety of signs and symptoms are used in dermatology research to establish the clinical cure rates of drugs used in the management of toenail onychomycosis. The lack of standardization of definitions for clinical cure, together with fairly arbitrary methods of data collection and presentation, render these outcomes meaningless in a systematic review of the literature. Future researchers of effective oral treatments for onychomycosis should clearly define the outcomes of interest, specify the mode of measurement, and ensure a clear presentation of the data. Consensus among the dermatology community about the most important signs and symptoms of clinical cure and methods of measurement for onychomycosis would be particularly helpful.
Twenty-two trials included in this systematic review were supported by pharmaceutical companies. All produced data to endorse the use of the sponsor's product, and it is possible that the conclusions of the present systematic review are compromised by a publication bias. The small number of trials in the meta-analysis for 11- and 12-month outcomes (Figure 1) make it difficult to use a funnel plot, but commercial influence and small sample sizes are 2 features frequently associated with such bias. This only serves to reinforce the need for a trial amnesty in which unpublished data are made available, otherwise independently funded research may be the only route to unbiased estimates of the effects of oral antifungal drugs.
Based on the mycological cure rates in our systematic review, it appears that a continuous regimen of terbinafine (250 mg/d) is the most effective oral therapy for the long-term management of fungally infected toenails. A standardization of methods regarding the collection and presentation of data regarding complete clinical cure is required to allow future researchers to compare the effect of oral antifungal agents on clinical outcomes in the treatment of onychomycosis. A consensus between clinicians and researchers regarding the best way to collect and present those secondary outcomes is needed.
Accepted for publication January 30, 2002.
A cooperative effort of the Clinical Epidemiology Unit of the Istituto Dermopatico dell'Immacolata–Istituto di Ricovero e Cura a Carattere Scientifico (IDI-IRCCS) and the Archives of Dermatology
Corresponding author and reprints: Fay Crawford, PhD, The Dental Health Services Research Unit, The University of Dundee, Park Place, Dundee DD1 4MR, Scotland (e-mail: email@example.com).