 |
Introduction |
|---|
About 15% of the population of the United Kingdom have
fungal infections of the foot.1 The main treatments are
topical fungistatic or fungicidal preparations, some of which are
available over the counter. We report a systematic review of randomised controlled trials of topical antifungal treatments for dermatophyte infections of the skin and nails of the foot, which were designed to
assess efficacy and cost effectiveness.
| |
Methods |
|---|
Search strategy
We searched Medline, Embase, CINAHL, BIDS, the Cochrane Controlled
Trials Register, CAB-Health, Healthstar, DARE, the NHS Economic
Evaluation Database, and Econlit to December 1997. We searched by hand
Foot, the Journal of British Podiatric Medicine,
and the British Journal of Podiatric Medicine and Surgery; the last two are not listed in any databases. We obtained the details
of the Cochrane Skin Group's recent search by hand of the
British Journal of Dermatology. We searched the
bibliographies of all review papers identified. To identify unpublished
or unlisted trials, we contacted international pharmaceutical companies
and all schools of podiatry in the United Kingdom.
Selection criteria
We considered all randomised controlled trials that evaluated
topical treatments for fungal infections of the skin and nails of the
foot. For skin infections we included only trials that used microscopy
and culture to establish the presence of dermatophytes. For nail
infections we included only trials that used culture to establish the
presence of dermatophytes. We included duplicate trials only once. We
excluded trials covering sites other than the foot where data related
specifically to the foot could not be extracted. Two reviewers (RH and
SEMBS) independently applied these criteria to each trial located. No
language restrictions applied.
Data extraction and quality assessment
Both reviewers independently summarised the trials included, and
they appraised their quality of reporting using a structured data
extraction tool of 12 quality criteria. The criteria were: aims clearly
defined; prior sample size calculation reported; inclusion and
exclusion criteria defined; subjects blinded; method of randomisation
defined; baseline comparability of groups reported (age, sex, and
duration of complaint); interventions defined; outcome assessment
blinded; compliance assessed; and trial analysed by intention to
treat.2-4
Statistical analysis
For each trial we calculated the cure rates at follow up from the
reported mycological results
that is, negative results on microscopy,
and no growth of dermatophyte in culture. We then estimated the
difference, with 95% confidence intervals, in the proportion of
patients cured. To estimate differences between treatments we pooled
trials that evaluated similar interventions and controls. As there was
clear evidence of heterogeneity between trials (P<0.001), we used
random effects models. We calculated the relative risk of failure to
cure, with 95% confidence intervals.5
Economic analysis
We calculated cost effectiveness in four ways: cost per cure of
using an over the counter preparation; marginal cost (including general
practitioner's time) per cure of a podiatrist referring a patient to a
general practitioner for treatment with an allylamine rather than
providing an over the counter preparation; marginal cost per cure
(excluding general practitioner's time) of a general practitioner,
once consulted, prescribing an allylamine rather than an over the
counter preparation; and marginal cost per cure (excluding general
practitioner's time) of a general practitioner, once consulted,
treating first with an azole and reserving allylamines for treatment failures.
Identified trials of skin
We located 111 papers reporting trials of topical treatments
for fungal skin infections, and we included 65 (58.6%) of these
in our review.6-10 w1-w60 As
one of these papers8 reported three distinct trials, and
three papers7 w1 w3 each reported a pair of
distinct trials, we included 70 trials in all. Thirty one trials
in 29 papers6-10 w1-w24
compared a single active treatment with placebo (see website). Twenty
seven trials in 26 papers8 w2
w25-w48 compared two active treatments (see website).
Twelve trialsw49-w60 compared more than two
treatments within the same trial (see website). We excluded the
remaining 46 trials w61-w106 owing to duplicate
reporting, combining data from more than one anatomical site (for
example, hand and foot), or absence of mycological testing. Responses
from the authors of provisionally excluded trials enabled us to include
just one.w29 Responses from the
pharmaceutical industry identified no additional studies.
Identified trials of nails
Apart from one trial (already excluded) of topical treatments for
skin and nail infections,w99 we found seven
trials evaluating the efficacy of topical treatments for nails, and we
included two of these.w107 w108 The other five
trials reported combined data from fingernails and
toenails.w109-w113 We also excluded three trials
evaluating a combination of systemic and topical treatments for
infected nails.w114-w116
Quality assessment
The mean number of quality criteria met by the 72 included
trials (two for nail infections) was only 6.3 out of 12. Only 19 trials
reported the method of randomisation.9 w2 w17-w19 w22 w24 w28 w37
w40 w41 w44 w45 w49 w51 w54 w56 w60 w107 Blinded outcome
assessment was reported in only 10 trials.w8 w18 w22 w30 w31
w37 w41 w49 w54 w61 Only seven trials, however, did not report
blinding of participants.8 w17
w25 w26 w32 w47 w48
Included trials of skin
The 70 trials evaluated a variety of treatments.
Azoles were assessed in 46 trials, allylamines in 27, tolnaftate in
five,w30 w44 w56-w58 and undecenoic acid in four.w24
w56 w57 w60 Ciclopiroxolamine, which is available over the
counter in the United States and parts of Europe but not in the United
Kingdom, was evaluated in two trials reported in the same
paper.w2 Griseofulvin,10
haloprogen,w30 tea tree oil,w58 and
tolciclatew21 were each evaluated in a separate small
trial. Variotin and Whitfield's ointment were compared with each other
in one very small trial,w41 which followed patients for 24 weeks. As no other trial followed patients for more than 12 weeks, no
conclusions are possible about differences in relapse rates.
Azoles versus placebo
Meta-analysis of data from 17 trials comparing azoles with
placebos6-9 w1 w12-w20 w55 w59 estimated the
pooled relative risk of failure to cure as 0.54 (95% confidence
interval 0.42 to 0.68). The concentration of these drugs was generally
1% but 2% for miconazole.
Allylamines versus placebo
Meta-analysis of data from 12 trials comparing allylamines
(concentration of 1% in all) with placebos estimated the relative risk
of failure to cure as 0.30 (0.24 to 0.38).
Azoles versus allylamines
The efficacy of azoles seems to depend on the duration of
treatment. Ablon et al reported that 2 weeks of treatment with
oxiconazole cured only 27% of patients.w4 More rigorously
Bergstresser et al reported a randomised comparison showing that 1 week
of treatment with clotrimazole cured only 35% of patients whereas 4 weeks of treatment cured 70%.w53 Therefore we excluded
both Ablon's study and Bergstresser et al's group treated with azoles
for only 1 week from our meta-analysis of azoles versus allylamines.
This left 12 trials,w27 w29 w31-w33 w36-w40 w50 w53 which
included three azoles (bifonazole, clotrimazole, and miconazole) and
two allylamines (naftifine and terbinafine). Where stated the
concentration was 1% for all agents. The frequency of treatment was
once or twice daily, generally for 4 or more weeks. The pooled relative
risk of failure to cure of 0.88 (0.78 to 0.99) favoured allylamines and
was just significant (figure). There was, however, a statistically
significant difference between the relative risk estimates from papers
in English and other languages. Eight English language reportsw29
w31 w36 w37 w39 w40 w50 w53 favoured allylamines (0.79, 0.69 to
0.91), but four foreign language reportsw27 w32 w33 w38
showed no difference between the two drugs (1.01, 0.90 to
1.13).
Comparative trials of topical allylamines and topical azoles for
fungal skin infections of foot: relative risk plot summarised by random
effects model5
Other compounds
The three placebo controlled trials of 1% tolnaftatew56-w58 yielded a pooled relative risk of failure
to cure of 0.46 (0.17 to 1.22). The two small comparative trials showed
tolnaftate to be significantly worse than
haloprogenw30 and profoundly, though not
significantly, worse than clotrimazole.w43 The
four placebo controlled trials of 5% undecenoic acidthree with just
two groupsw24 w56 w57 and one that also compared
undecenoic acid with zinc undecenoatew60yielded a
pooled relative risk of failure to cure of 0.28 (0.11 to 0.74).
Ciclopiroxolamine was significantly better than placebo (0.14, 0.06 to 0.32) and better than clotrimazole although not significantly
(0.89, 0.72 to 1.10).w2
Included trials of nails
In the smaller trial, two amorolfine 5% nail lacquer formulations
with different vehicles both achieved a cure rate close to 90% after 6 weeks.w108 In the larger trial, clotrimazole solution and
tea tree oil both achieved a cure rate close to 10% after 6 months.w107
Costs and cost effectiveness
The table shows differences in cure rates and costs for the main
treatments and their cost effectiveness. So few trials compared azoles,
undecenoic acid, and tolnaftate directly that confidence intervals for
their costs per cure must be on the basis of placebo controlled trials.
Although there are only four such trials of undecenoic acid, their
findings are much more homogeneous than trials of other drugs.
Undecenoic acid therefore yields a significantly cheaper average cost
per cure when purchased over the counter than azoles, and a much
narrower confidence interval than tolnaftate. As azoles cost the NHS
about half the retail price, however, there is no significant
difference in NHS cost per cure between azoles and undecenoic acid,
regardless of general practitioner's time.
<
Athlete's foot is widespread and infectious. Initial treatment for most patients is a topical cream or ointment. Good evidence shows that allylamines, azoles, and undecenoic acid are efficacious compared with placebo. Unfortunately there is little evidence to assess tolnaftate against placebo or to compare azoles, undecenoic acid, and tolnaftate with each other.
More trials compared allylamines with azoles. The resulting meta-analysis suggests that allylamines are generally more efficacious, with a relative risk of failure to cure of 0.88 (0.78 to 0.99). This difference is more marked in trials reported in English (0.79, 0.69 to 0.91) but not significant in foreign language trials. This discrepancy could not be explained by differences in the quality of the trials, commercial sponsorship, or the patients' characteristics. This suggests that by including foreign papers in this review we have reduced English language biasw119 and improved the estimate of relative risk of failure to cure.
We detected no differences in efficacy between individual allylamines or individual azoles. As no trial reported the species obtained from patients who were not cured, we cannot draw conclusions about susceptibility to individual compounds to help clinical decision making. Without more evidence we recommend treatment with the least expensive compound within each class.
Although allylamines cure slightly more infections than azoles, they are available only on prescription at much greater cost. Hence the most cost effective strategy is to treat first with an azole or undecenoic acid and to reserve allylamines for treatment failures. To improve on this recommendation more direct comparisons are needed of undecenoic acid and tolnaftate with allylamines and azoles, and a large trial is needed to refine our estimate of the incremental benefit of allylamines over azoles.
Evidence about the efficacy of topical treatments for nail infections
is very sparse. Little can be concluded about the role of these agents
in curing infected toenails. Rigorous research is overdue.
In summary there is little evidence to differentiate between three
popular over the counter topical treatments for fungal skin infections.
Azoles may be more efficacious than tolnaftate. They seem, however, no
more efficacious than undecenoic acid. Evidence shows that allylamines
are slightly more efficacious than azoles and other over the counter
treatments but at much greater cost. Thus we recommend initial
treatment with azoles or undecenoic acid, and the use of allylamines
only if that fails.