Interventions for the treatment of pityriasis versicolor

Joel TM Bamford; Rowena Natividad S Flores‐Genuino; Sujoy Ray; Michael Bigby; Martha Morales; Miguel Arkoncel; Ma. Assumpta Cecilia Realubit‐Serrano

doi:10.1002/14651858.CD011208

Interventions for the treatment of pityriasis versicolor

Authors' declarations of interest

Version published: 14 July 2014 Version history

https://doi.org/10.1002/14651858.CD011208

Collapse all Expand all

Abstract

This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:

To assess the effects and safety of topical and oral agents for pityriasis versicolor.

Background

Description of the condition

Definition

Pityriasis versicolor (PV), also known as tinea versicolor, is a chronic and benign superficial fungal skin infection caused by Malassezia yeasts, which are part of the many microscopic organisms that normally live on the skin. It is thought that the fungus produces either an enzyme that interferes with pigment production, light‐absorbing compounds that shield the skin from sunlight, or toxic compounds that destroy pigment‐producing cells (Gaitanis 2012). Pityriasis versicolor is not contagious; rather, it is an endogenous opportunistic infection.

Diagnosis

The condition is diagnosed by the appearance of light or dark spots on the skin that are 3 to 5 mm in diameter, round or oval‐shaped, and possibly covered with fine scales (Figure 1). The colour of the spots vary (hence the name "versicolor") from pink to brown. The marks may coalesce and affect extensive areas and may often persist on the skin even after the infection has been treated and cleared. The marks often occur on the upper chest and upper back, but can also involve the upper arms; neck; shoulders; and among children in the tropics, the face (Jena 2005).

Figure 1

Scaly hypopigmented lesions of pityriasis versicolor on the nape

Stretching or scraping the skin lesion helps to visualise the scales in active infection ('evoked scale sign') (Han 2009). Wood's lamp shows yellow to yellow‐green fluorescence in lesions with Malassezia furfur, but is rarely used. Confirmatory microscopic examination is done with skin scrapings or tape stripping treated with potassium hydroxide with or without blue‐black ink. The typical "spaghetti and meatballs" appearance is seen, referring to the cluster of spores and short hyphae between layers of the stratum corneum. Differential diagnosis of PV include pityriasis alba, seborrhoeic dermatitis, guttate psoriasis, tinea corporis, vitiligo, leprosy, or progressive macular hypomelanosis. A skin biopsy is generally not needed for the diagnosis of pityriasis versicolor, except in atypical cases. Fungal cultures and sensitivity testing of those cultures require special media that are not always readily available, so they are often used only for epidemiologic studies.

Epidemiology and risk factors

Pityriasis versicolor has a worldwide distribution, but it is more common in tropical countries, with prevalence ranging from 2.3% in the Philippines (Romero 1999) to 16.3% to 17.8% in Malawi (Msyamboza 2012; Ponnighaus 1996). There were 2.9 million visits to family doctors per year for pityriasis versicolor in a US national survey (Mellen 2004), which accounted for around 29% of all dermatomycoses seen at dermatology clinics (Ellabib 2002; Maaño 2011).

Pityriasis versicolor affects mainly adolescents and young adults aged 10 to 30 years (Dutta 2002; Framil 2011; Morais 2010), but may occur in individuals of all ages, including infants (Jena 2005) and the elderly (Kyriakis 2006; Ponnighaus 1996).

There is no conclusive evidence that either gender or any particular ethnic group is more susceptible. The inheritance pattern of PV is polygenic additive (He 2008). Since the heritability in first‐degree relatives of people with PV is only 22% to 48% (Hafez 1985; He 2008), other factors play a role. Environmental factors include heat and humidity in tropical climates and during the summer months in temperate climates. Host factors that are implicated include abnormal sebum composition, use of occlusive clothing, and application of oils (Gaitanis 2012). The use of oral immunosuppressants, such as ciclosporin and azathioprine, in renal transplant patients are independent risk factors for the development of PV (Gulec 2003). Biologic response modifiers, such as anti‐TNF (tumour necrosis factor) monoclonal antibodies (Balestri 2012) and etanercept (Levy 2008), have also been associated with the onset of PV. These cases are usually extensive and resistant to topical agents.

Course and prognosis

Pityriasis versicolor may undergo remission in cooler weather, but it almost always recurs in hot weather (Bigby 2008). People with the condition are often confused and frustrated because of the persistence of light or dark marks on their skin, over weeks to months, even after adequate treatment and clearing of all fungus. Sun exposure should be avoided since it may worsen the colour contrast, as PV‐affected skin will fail to tan. Eventually, the contrast disappears, as the affected skin resumes normal pigment production.

Impact

Up to 48% of people with pityriasis versicolor may have itching in hot and humid climates (Salahi‐Moghaddam 2009). Although itching may be marked, it is the colour change that is primarily distressing. Although pityriasis versicolor is easily treated, relapse is common. Fifty‐three per cent of people adequately treated had one to four recurrences, whereas 14.7% had more than four relapsing episodes in one year (Framil 2011). Even when skin lesions are cleared of the fungus, distressing hypopigmentation may persist for months or years (28% to 47% of cases) (Salahi‐Moghaddam 2009), which people sometimes regard as treatment failure.

Despite its benign nature, pityriasis versicolor can lead to emotional distress, self‐consciousness, and social stigmatisation. Fear of contagion, unsightly appearance (especially in dark skin or tanned skin), and the recurrent and chronic nature of the disease all contribute to a lower quality of life (QOL). In a cross‐section of 43 people with pityriasis versicolor, a quality of life survey found depression and anxiety in 11 and 15 participants, respectively, although the study did not include controls for comparison (Kaymak 2008).

The cost of one treatment course varies widely, from USD three for a bottle of selenium sulfide shampoo to USD 26 for a four‐week once‐weekly dose of fluconazole tablets (Balestri 2012) to USD 105 for a seven‐day course of itraconazole tablets (Stratman 2010).

Description of the intervention

The goals of treatment of pityriasis versicolor are as follows:

to eradicate the hyphae and "restore the yeast's population dynamics to the commensal status" (Gaitanis 2012);
the disappearance of clinical signs (scaling and erythema) and symptoms (itch);
to lessen recurrences and increase time to recurrence after treatment; and
patient satisfaction and improved QOL.

There is a wide array of topical and oral treatments for PV, with different modes of action, drug classes, and dosing regimens. Localised PV is usually treated with topical drugs, while extensive, recalcitrant, or highly recurrent cases are given oral treatment. Convenience and higher compliance with oral medications may also be a consideration (Balestri 2012; Stratman 2010).

We followed Gupta 2005 to classify the antifungal drugs into specific antifungals that directly inhibit or kill fungi and non‐specific anti‐PV agents that work by shedding the keratinised skin layers bearing the fungi and other novel interventions.

Topical treatment

Specific antifungals
- Azoles. Topical azoles are mostly imidazoles (bifonazole, econazole, flutrimazole, ketoconazole, miconazole, fenticonazole, sulconazole, tioconazole) or triazoles (fluconazole). Topical imidazoles usually come in 1% to 2% concentration and in various formulations (shampoos, sprays, lotions, gels, creams, or powder). They may be used once or twice daily, as a single application, or up to as long as two to eight weeks. Triazoles are newer generation azoles, and consist of oral itraconazole and topical or oral fluconazole. Topical fluconazole is available as 2% shampoo and is used daily for five days.
- Allylamines (naftifine, terbinafine), benzyl amines (butenafine), and ciclopiroxolamine. Terbinafine is given as 1% solution, emulsion, or cream for one week. Ciclopiroxolamine is given as 0.1% solution for four to eight weeks or as a 1% cream twice daily for two weeks.
- Older non‐prescription agents, such as haloprogin, nystatin, tolnaftate, and zinc pyrithione (1% shampoo for two weeks).
Non‐specific antifungals (with keratolytic and other actions) include salicylic acid, selenium sulfide (2.5% lotion, cream, or shampoo for one week), sodium sulfacetamide, sodium thiosulphate, sulphur/salicylic acid, Whitfield's ointment (6% benzoic acid and 3% salicylic acid in an emulsifying ointment), tretinoin, adapalene, benzoyl peroxide (5% to 10% gel for three weeks), and propylene glycol. Laundry soaps, keratolytic herbal soaps (papaya, glycolic), and simply rubbing with a pumice stone or other rough and abrasive material have also been used, especially in developing countries.
Other novel interventions include herbal preparations (akapulco, lemon grass), cycloserine, and nitric oxide‐liberating cream.

Oral treatment

Oral antifungal drugs consist of oral azoles, which are used off‐label for PV. Oral terbinafine is not used since it does not reach a therapeutic level in the stratum corneum.

Oral ketoconazole, the only oral imidazole for PV, is no longer U.S. Food and Drug Administration (FDA) or European Medicines Agency (EMA)‐approved for all superficial fungal skin infections because of its "significant risk for hepatotoxicity, adrenal insufficiency and drug interactions" (FDA 2013).

Oral itraconazole and fluconazole are newer generation triazoles, which have a broader spectrum of antifungal activity and better safety profile compared to oral ketoconazole. In particular, oral fluconazole has high bioavailability, minimal hepatic metabolism (thereby, less hepatotoxicity and drug‐drug interactions), and convenient once‐weekly dosing (Wolverton 2012). However, in vitro testing has shown that it has the highest minimum inhibitory concentration (MIC) among the azoles (Miranda 2007). Oral itraconazole is available as 100 mg capsules, and two capsules are given daily for five to seven days. Oral fluconazole is available as 50 mg, 100 mg, 150 mg, and 200 mg capsules and 10 mg/ml and 40 mg/ml suspension. It may be given as a 400 mg single dose, 300 mg weekly or biweekly for two to four weeks, or 150 mg weekly for two to four weeks, depending on response.

How the intervention might work

Specific antifungals
- Azoles are fungistatic because they inhibit fungal cell membrane formation.
- Allylamines and benzyl amines are both fungistatic and fungicidal, because they also inhibit an earlier enzyme in the pathway, squalene epoxidase, which disrupts the fungal cell membrane and destroys the fungus by causing accumulation of a toxic metabolite, squalene.
- Ciclopiroxolamine has a high affinity for trivalent metal cations, thus, inhibiting the metal‐dependent enzymes for peroxide degradation within the fungi (Subissi 2010).
- Zinc pyrithione increases copper influx, which inactivates the fungi (Reeder 2011).
- Nystatin is a polyene antibiotic that physicochemically interacts with fungal membrane sterols (Ghannoum 1999).
- Haloprogin is thought to inhibit oxygen uptake, thus, disrupting the yeast membrane.
- Tolnaftate is believed to inhibit squalene epoxidase.
Non‐specific antifungals
- Most are keratolytic agents, such as sodium thiosulphate, sulphur‐salicylic acid, selenium sulfide, and retinoids, which dissolve the intercellular lipid between the keratinised cells. Some (e.g. propylene glycol) are irritants.
Other novel interventions
- These include herbal preparations (akapulco, lemon grass), cycloserine, and nitric oxide‐liberating cream, which do not have well‐defined modes of action.

Why it is important to do this review

There are a wide variety of effective treatment options for PV, but there are no clear‐cut guidelines. The long list of drugs, both oral and topical (Drake 1996), with different drug classes, modes of action, dosage, and dosing regimens point to this clinical practice gap.

A non‐Cochrane systematic review (Hu 2010), which included 93 controlled trials, concluded that "most topical and systemic treatments used for PV are effective compared to placebo" but "randomised controlled clinical trials are needed to establish relative efficacy". However, this review pooled randomised controlled trials (RCTs) and non‐RCTs together and did not perform formal tests for heterogeneity, nor subgroup analyses to explain possible heterogeneity. Sensitivity analysis was also not done to test the robustness of the results. Methodological quality was assessed using a summary quality scoring system rather than 'Risk of bias' assessment in relation to outcomes, clinical cure was not reported as an outcome measure, and the timing of assessment for mycologic cure ranged from seven days to two months. They concluded that "data suggest that longer durations of treatment and higher concentrations of active agents produce greater cure rates" (Hu 2010). However, these may have more side‐effects, lower patient compliance, and higher cost. Some clinical trials suggest that topical agents may be equally effective or more effective than oral medications (del Palacio‐Hernanz 1989; Silva 1998). "Choosing oral therapy over topical therapy when topical therapy is reasonable" may be a practice gap (Stratman 2010). Also, comparative efficacy and safety of prescription versus non‐prescription treatments must be known.

Thus, there is a need to search for and pool the evidence to answer these practice gaps, which we aim to do in this review.

Objectives

To assess the effects and safety of topical and oral agents for pityriasis versicolor.

Methods

Criteria for considering studies for this review

Types of studies

We will include only randomised controlled trials whose aim is treatment of pityriasis versicolor. For cross‐over studies, we will include only the first phase.

Types of participants

We will include studies with children and adults, men and women, of any ethnic background, of any socioeconomic status, who are healthy or immunocompromised, and with any PV severity (localised, moderate, or extensive), with a clinical diagnosis of PV.

We will include studies even if there is no mycologic confirmation through potassium hydroxide or Wood's light examination.

Studies that include participants under a broader diagnostic category, such as 'superficial fungal skin infections' or 'superficial dermatomycoses' will be eligible, if they describe results for those with PV separately.

Types of interventions

All topical and oral agents used for the treatment of PV compared with placebo, vehicle, no treatment, other topical or oral agents, or same agent but different formulation, concentration, dose, frequency, or duration.

Comparisons will include the following:

any topical agent versus placebo, vehicle, or no treatment;
any topical agent versus any oral agent;
any topical agent versus another topical agent;
any oral agent versus placebo, vehicle, or no treatment;
any oral agent versus another oral agent; or
any different formulation, concentration, dose, frequency, or duration of the same topical or oral agent.

Types of outcome measures

Timing of outcome assessments

When interventions are given for different durations, we will consider the longest treatment period as the point of reference. We will define the post‐treatment period to start on the day following the last day of treatment, or if the dosing is once‐weekly, on the 8th day after the dose.
For mycologic cure, clinical cure, patient satisfaction, and QOL, the primary time point for measurement will be three to five weeks after treatment ends.
Secondary time points will be > five weeks to six months (medium‐term) and > six months to 12 months (long‐term).
For adverse effects, assessment will be any time during treatment or during follow‐up. For recurrence, the primary time point will be three to six months after treatment, with secondary time points at > six to 12 months after treatment.
If a study reports more than one time point, we will choose the one nearest to the primary time point.

Primary outcomes

Proportion of participants with mycological cure, defined as disappearance of short hyphae ("spaghetti and meatballs" appearance suggestive of Malassezia spp) on microscopic examination of potassium hydroxide mounts of skin scrapings or tape stripping, assessed by a physician or trained technician.
Proportion of participants with adverse effects (serious, e.g. hepatotoxicity or requiring withdrawal of treatment, and minor, e.g. not requiring withdrawal of treatment) (US‐FDA 2013).

Secondary outcomes

Proportion of participants with clinical cure, defined as absence of the signs (scale and erythema), assessed by a physician or trained investigator. We will not assess pigmentary change as a sign of clinical cure as it often remains long after microscopic cure.
Participant‐reported cure based on cessation of symptoms, such as itching and disappearance of rash.
Quality of life score, as assessed by participant or proxy, using a validated scale, such as Dermatology Life Quality Index (Finlay 1994), Children's Dermatology Life Quality Index (Lewis‐Jones 1995), Skindex (Chren 2001), or Skindex‐Teen (Smidt 2010).
Proportion of participants who are satisfied with the treatment (ease of use, compliance with the treatment, convenience, acceptability, preference, etc).
Time to recurrence.

Search methods for identification of studies

We aim to identify all relevant RCTs regardless of language or publication status (published, unpublished, in press, or in progress).

Electronic searches

We will search the following databases for relevant trials:

the Cochrane Skin Group Specialised Register;
the Cochrane Central Register of Controlled Trials (CENTRAL) in The Cochrane Library;
MEDLINE via OVID (from 1946);
Embase via OVID (from 1974);
LILACS (Latin American and Caribbean Health Science Information database, from 1982);
HERDIN NeON Database (Health Research and Development Information Network, from 1971) using the following search terms: tinea versicolor or tinea flava or tinea alba or pityriasis versicolor);
Philippine Index Medicus (from 1975) using the following search terms: tinea versicolor or tinea flava or tinea alba or pityriasis versicolor;
Wanfang Data (Chinese Medicine Premier) (www.wanfangdata.com);
China Knowledge Resource Integrated Database (CKRID) (www.global.cnki.net);
Chinese Biomedical Literature Database (CBM); and
Chinese Scientific Journals Database (CSJD)/VIP Information.

We have devised a draft search strategy for RCTs for MEDLINE (OVID), which is displayed in Appendix 1. This will be used as the basis for search strategies for the other databases listed.

Trials registers

We will search the following trials registers.

The metaRegister of Controlled Trials (www.controlled‐trials.com).
The US National Institutes of Health Ongoing Trials Register (www.clinicaltrials.gov).
The Australian New Zealand Clinical Trials Registry (www.anzctr.org.au).
The World Health Organization International Clinical Trials Registry Platform (www.who.int/trialsearch).
The EU Clinical Trials Register (www.clinicaltrialsregister.eu/).
Philippine Health Research Registry (www.registry.healthresearch.ph).

Searching other resources

Adverse effects

We will not perform a separate search for adverse effects of interventions for pityriasis versicolor. However, we will examine data on adverse effects from our included studies.

Grey literature

We will search for unpublished trials by contacting researchers, organisations, and pharmaceutical companies in the field.

Reference lists

We will check the bibliographies of included studies and review articles cited as references for further references to relevant trials.

Data collection and analysis

We plan to include at least one 'Summary of findings' table in our review using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) profiler software. In this, we will summarise the primary outcomes of clinical cure and adverse effects for the most important comparisons: topical versus oral treatment, best topical treatment, and best oral treatment. If we feel there are several major comparisons or that our findings need to be summarised for different populations, we will include further 'Summary of findings' tables.

Selection of studies

We will merge all electronic searches and remove duplicate records. Two authors will check the titles and abstracts identified from the searches. Two authors will independently decide which trials meet the inclusion criteria using a pre‐piloted data extraction form (DEF), and they will resolve any disagreements by discussion between them or a third author. Of electronic search results, if the title or abstract is unclear, we will obtain the full text of potentially relevant studies for assessment. If the full report still yields unclear or insufficient information, we will contact the study authors. A single failed inclusion criterion will be enough to exclude the trial. Two authors will also compare studies and reports by author names, location, details of the interventions, and number and baseline data of participants in order to detect multiple articles or reports from the same study. They will also record the excluded studies and the reasons for exclusion.

Data extraction and management

We will collect all information from studies relevant to this review in an electronic form. Two authors will independently extract data from every report using a pre‐piloted DEF, which will include the following.

General information (corresponding author's complete contact information (phone, fax, or e‐mail; mailing address), complete citation with any identifying number)
Methods (study design, single or multiple sites, country, setting)
Participants (study population; baseline characteristics, such as age, sex, and extent of disease; initial or a recurrent problem; inclusion and exclusion criteria; diagnostic method; number randomised into each treatment group; number lost to follow up; and withdrawals for each group, plus reasons)
Interventions (type of agent, route of administration, dosage, cumulative dosage (for oral), frequency, duration of treatment, co‐intervention, contamination, integrity of intervention, compliance)
Outcome measures including harms reported in the study (methods and timing of assessment; outcome definition; person who made assessment; unit of measurement, upper and lower limit of scales, validation of tools)
'Risk of bias' assessment (random sequence generation, allocation concealment, blinding of participants and caregivers, blinding of outcome assessment, incomplete outcome data, selective outcome reporting, other bias (pharmaceutical industry funding, conflict of interest))
Data and analysis (for each of our primary and secondary outcomes of interest that are reported: number of participants with outcome and total number of randomised participants (per group); number of missing participants and the reasons; number of participants moved from other group and the reasons; unit of analysis; statistical methods used and appropriateness; reanalysis, if needed and possible to perform)

If there is insufficient or unclear information, we will contact the study authors. The two authors will resolve any disagreements through discussion or by consulting a third author. Two authors will independently enter data and compare results, resolving any disagreements by double‐checking entries and discussion or by consulting a third author.

Assessment of risk of bias in included studies

Two authors will independently assess the risk of bias in included studies using The Cochrane Collaboration's tool that includes domains for the following biases: selection, performance, detection, attrition, and reporting. We will assess risk of bias as 'high', 'low', or 'unclear' (as per section 8.8.2.2 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011)), recorded for each study in the 'Risk of bias' tables in Review Manager (RevMan), and summarised in the 'Risk of bias' summary table and graph.

We will consider a study 'low risk' when all domains are low risk and will imply that bias is unlikely to seriously alter the results. It will be 'unclear' when at least one domain has unclear risk and will imply that there may be bias that raises doubts about the results. A 'high risk' study is one wherein at least one domain is rated 'high risk' and will imply that bias seriously weakens confidence in the results.

We will contact authors to clarify any unclear study information. We will resolve any disagreement in assessment of risk of bias through discussion or by consulting a third author.

Measures of treatment effect

We will compute risk ratios (RR) and corresponding 95% confidence intervals (CI) for dichotomous outcomes (clinical cure, mycologic cure, recurrences, adverse effects). We will compute mean differences (MD) and standard deviations (SD) for continuous outcomes (such as duration to recurrence, quality of life score, or participant satisfaction score) that used the same scale. We will compute standardised mean differences (SMD) and standard deviations (SD) for continuous outcomes that use different scales.

Unit of analysis issues

The unit of analysis will be the participant. In case the participant is not the unit of randomisation, such as in cluster randomisation, we will adjust for clustering using recommended methods described in theCochrane Handbook of Systematic Reviews for Intervention, section 16.3, to account for the cluster design.

In case of studies with more than two treatment groups, if there are independent pair‐wise comparisons that are relevant to the review, we will include them as if they were from different studies. To avoid double‐counting in studies with multiple, correlated comparisons (i.e. when the comparisons have an intervention group, and thus participants, in common), we will follow the recommended method (section 16.5.4, Cochrane Handbook for Systematic Reviews of Interventions), which is to 'combine groups to create a single pair‐wise comparison (combine all relevant intervention groups into a single group and to combine all relevant control groups into a single group)'.

For dichotomous outcomes, both the sample sizes and the numbers of participants with events can be aggregated across groups. For continuous outcomes, means and standard deviations can be combined using methods described in section 7.7.3.8 of the Cochrane Handbook for Systematic Reviews of Interventions.

Dealing with missing data

We will do intention‐to‐treat analysis, by analysing participants in the group they were initially randomised into, despite the treatment they actually received.

In case of missing data, we will try to request it from the original investigators. In case we cannot obtain missing data, we will use available case analysis as the main analysis for both dichotomous and continuous outcomes.

If there are missing standard deviations for continuous data, we will attempt to look for statistics that allow calculation or estimation of the standard deviation (SD) (e.g. confidence intervals, standard errors, t values, P values, F values). If graphs are provided, we will attempt to extract the desired data. If missing data are only a small proportion (5%) of the total data, we will impute using the highest SD from other similar studies in the meta‐analysis for a conservative estimate. We will conduct sensitivity analysis by imputing the average SD from other similar studies. If missing data are greater than 30% of the total data, we will exclude the study from the analysis. If missing data are greater than 15% of the total data, we will consider the study high risk of bias in the 'Incomplete outcome reporting' item in the 'Risk of bias' table.

For dichotomous outcomes, we will do sensitivity analysis based on the 'best case' and 'worst case' scenarios, which are as follows. For the 'best case scenario', we will consider the missing outcomes for the intervention as cures and those for the control as failures, while we will consider the opposite for the 'worst case' scenario. If the conclusion does not change, we will consider results from the main analysis as robust. We will also consider a change in the direction of benefit due to missing data as a high risk of bias in the 'Risk of bias' assessment table.

Assessment of heterogeneity

We will assess heterogeneity using visual inspection of the forest plots. If confidence intervals have little overlap, it will suggest that studies are heterogenous. We will also compute the Chi² test for heterogeneity at 10% level of significance. We will consider a P value of < 0.10 to mean significant heterogeneity. We will also compute the I² statistic value, and if the I² statistic is > 50%, will assess heterogeneity to be significant, and if the I² statistic is > 75%, we will assess heterogeneity to be substantial. We will use the random‐effects model (DerSimonian and Laird method) as the main analysis, while we will use the fixed‐effect model (Mantel‐Haenszel method) as a sensitivity analysis. In addition, if significant heterogeneity exists, we will undertake subgroup analysis to determine the possible causes of heterogeneity.

Assessment of reporting biases

We will include non‐English language trials to reduce language bias. We will search for unpublished trials and ongoing or recently completed trials using trial registers and by contacting authors, organisations, and pharmaceutical companies, to reduce publication bias and time lag bias. If there are more than 10 trials, we will construct a funnel plot. If asymmetry exists, it may indicate publication bias, although it may also be due to heterogeneity and poor methodological quality of trials. We will assess selective reporting bias in the 'Risk of bias' tables.

Data synthesis

We will enter data into RevMan. We will pool data for studies that are sufficiently homogenous in types of participants, interventions, and outcomes (clinical homogeneity) and in design and risk of bias (methodological homogeneity). We will use a random‐effects model since we assume heterogeneity in intervention effects due to varied interventions. We will perform fixed‐effect model analysis as a sensitivity analysis. If meta‐analysis of studies is not possible, we will describe data qualitatively.

Subgroup analysis and investigation of heterogeneity

We will undertake subgroup analyses for the primary outcomes if there are at least two studies in at least one subgroup. We will compare the following subgroups.

Single dose versus longer duration: Single dose regimens seem to be less effective than multiple doses over several days or weeks (Hu 2010).
Pulsed dose versus continuous daily dose: Pulsed dose of ketoconazole has been found to be less effective than daily dose (Hu 2010).
Low cumulative dose versus high cumulative dose: Serious adverse effects may increase with higher doses.
Immunosuppressed versus non‐selected participants: Immunosuppressed patients may require higher doses or longer duration of treatment.
Non‐extensive versus extensive PV. Extensive PV is defined as four or more body regions affected (Morais 2010) using the rule of nines for estimating extent of burns or if > 30% body surface area (BSA) is involved. Patients with extensive PV may be immunosuppressed or due to the higher fungal load, may require higher doses or longer duration of treatment.
Tropical climate versus non‐tropical climate, as defined according to the updated world Köppen‐Geiger climate classification map. Tropical climate is found in countries that have part of their land mass between the Tropics of Cancer and Capricorn and where the temperature of the coldest month is greater than 18 °C (Peel 2007). People in tropical climates may have more extensive PV.

Sensitivity analysis

Aside from available case analysis, which will be the main analysis, we will test the robustness of our results by performing the following sensitivity analyses:

excluding trials with high risk or unclear risk of bias;
doing worst and best case scenario analysis for dichotomous outcomes ‐ imputing the worst outcome for the intervention and the best outcome for the control (worst case scenario analysis);
imputing the best outcome for the intervention and the worst outcome for the control (best case scenario analysis);
imputing the average SD (based on available statistics from other similar studies) for missing SD values;
excluding studies that are outliers;
using the fixed‐effect method of analysis compared to the random‐effects method; and
excluding pharmaceutical industry‐sponsored studies, which we will define as those initiated by the pharmaceutical industry or where investigators were employees or paid consultants of the companies. We will not exclude studies where the pharmaceutical companies provided only the medications.