Interventions for impetigo

Sander Koning; Renske van der Sande; Arianne P Verhagen; Lisette WA van Suijlekom‐Smit; Andrew D Morris; Christopher C Butler; Marjolein Berger; Johannes C van der Wouden

doi:10.1002/14651858.CD003261.pub3

. 2012 Jan 18;2012(1):CD003261. doi: 10.1002/14651858.CD003261.pub3

Interventions for impetigo

Sander Koning ¹, Renske van der Sande ¹, Arianne P Verhagen ¹, Lisette WA van Suijlekom‐Smit ², Andrew D Morris ³, Christopher C Butler ⁴, Marjolein Berger ^1,⁵, Johannes C van der Wouden ^1,^✉

Editor: Cochrane Skin Group⁶

PMCID: PMC7025440 PMID: 22258953

Abstract

Background

Impetigo is a common, superficial bacterial skin infection, which is most frequently encountered in children. There is no generally agreed standard therapy, and guidelines for treatment differ widely. Treatment options include many different oral and topical antibiotics as well as disinfectants. This is an updated version of the original review published in 2003.

Objectives

To assess the effects of treatments for impetigo, including non‐pharmacological interventions and 'waiting for natural resolution'.

Search methods

We updated our searches of the following databases to July 2010: the Cochrane Skin Group Specialised Register, the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE (from 2005), EMBASE (from 2007), and LILACS (from 1982). We also searched online trials registries for ongoing trials, and we handsearched the reference lists of new studies found in the updated search.

Selection criteria

Randomised controlled trials of treatments for non‐bullous, bullous, primary, and secondary impetigo.

Data collection and analysis

Two independent authors undertook all steps in data collection. We performed quality assessments and data collection in two separate stages.

Main results

We included 57 trials in the first version of this review. For this update 1 of those trials was excluded and 12 new trials were added. The total number of included trials was, thus, 68, with 5578 participants, reporting on 50 different treatments, including placebo. Most trials were in primary impetigo or did not specify this.

For many of the items that were assessed for risk of bias, most studies did not provide enough information. Fifteen studies reported blinding of participants and outcome assessors.

Topical antibiotic treatment showed better cure rates than placebo (pooled risk ratio (RR) 2. 24, 95% confidence interval (CI) 1.61 to 3.13) in 6 studies with 575 participants. In 4 studies with 440 participants, there was no clear evidence that either of the most commonly studied topical antibiotics (mupirocin and fusidic acid) was more effective than the other (RR 1.03, 95% CI 0.95 to 1.11).

In 10 studies with 581 participants, topical mupirocin was shown to be slightly superior to oral erythromycin (pooled RR 1.07, 95% CI 1.01 to 1.13). There were no significant differences in cure rates from treatment with topical versus other oral antibiotics. There were, however, differences in the outcome from treatment with different oral antibiotics: penicillin was inferior to erythromycin, in 2 studies with 79 participants (pooled RR 1.29, 95% CI 1.07 to 1.56), and cloxacillin, in 2 studies with 166 participants (pooled RR 1.59, 95% CI 1.21 to 2.08).

There was a lack of evidence for the benefit of using disinfectant solutions. When 2 studies with 292 participants were pooled, topical antibiotics were significantly better than disinfecting treatments (RR 1.15, 95% CI 1.01 to 1.32).

The reported number of side‐effects was low, and most of these were mild. Side‐effects were more common for oral antibiotic treatment compared to topical treatment. Gastrointestinal effects accounted for most of the difference.

Worldwide, bacteria causing impetigo show growing resistance rates for commonly used antibiotics. For a newly developed topical treatment, retapamulin, no resistance has yet been reported.

Authors' conclusions

There is good evidence that topical mupirocin and topical fusidic acid are equally, or more, effective than oral treatment. Due to the lack of studies in people with extensive impetigo, it is unclear if oral antibiotics are superior to topical antibiotics in this group. Fusidic acid and mupirocin are of similar efficacy. Penicillin was not as effective as most other antibiotics. There is a lack of evidence to support disinfection measures to manage impetigo.

Plain language summary

Interventions for the skin infection impetigo

Impetigo causes blister‐like sores. The sores can fill with pus and form scabs, and scratching can spread the infection. Impetigo is caused by bacteria. It is contagious and usually occurs in children. It is the most common bacterial skin infection presented by children to primary care physicians. Treatment options include topical antibiotics (antibiotic creams), oral antibiotics (antibiotics taken by mouth), and disinfectant solutions. There is no generally agreed standard treatment, and the evidence on what intervention works best is not clear.

We identified 68 randomised controlled trials comparing various treatments for impetigo. Altogether, these studies evaluated 26 oral treatments and 24 topical treatments, including placebo, and results were described for 5708 participants.

Overall, topical antibiotics showed better cure rates than topical placebo.

Two antibiotic creams, mupirocin and fusidic acid, are at least as effective as oral antibiotics where the disease is not extensive. There was no clear evidence that either of these most commonly studied topical antibiotics was more effective than the other.

Topical mupirocin was superior to the oral antibiotic, oral erythromycin.

We found that the oral antibiotic, oral penicillin, is not effective for impetigo, while other oral antibiotics (e.g. erythromycin and cloxacillin) can help.

It is unclear if oral antibiotics are superior to topical antibiotics for people with extensive impetigo.

There is a lack of evidence to suggest that using disinfectant solutions improves impetigo. When 2 studies with 292 participants were pooled, topical antibiotics were significantly better than disinfecting treatments.

Reported side‐effects for topical treatments were mild and low in frequency; the treatments sometimes resulted in itching, burning, or staining. Oral antibiotics produced gastrointestinal complaints, such as nausea and diarrhoea, in 2% to 30% of participants, depending upon the specific antibiotic.

Worldwide, bacteria causing impetigo show growing resistance rates for commonly used antibiotics. For a newly developed topical treatment, retapamulin, no resistance has yet been reported.

Background

Description of the condition

Biology and symptoms

Impetigo or impetigo contagiosa is a contagious superficial bacterial skin infection most frequently encountered in children. It is typically classified as either primary (e.g. direct bacterial invasion of previously normal skin), secondary, or common impetigo (where the infection is secondary to some other underlying skin disease that disrupts the skin barrier, such as scabies or eczema). Impetigo is also classified as bullous or non‐bullous impetigo. Bullous impetigo simply means that the skin eruption is characterised by bullae (blisters). The term 'impetigo contagiosa' is sometimes used to mean non‐bullous impetigo, and at other times it is used as a synonym for all impetigo.

Non‐bullous impetigo is the most common form of impetigo. The initial lesion is a thin‐walled vesicle on previously normal skin that rapidly ruptures. It then leaves superficial erosion covered with yellowish‐brown or honey‐coloured crusts. The crusts eventually dry, separate, and disappear, leaving a red mark that heals without scarring. The most frequently affected areas are the face and limbs. The lesions are sometimes painful. Usually, there are no systemic symptoms such as fever, malaise, or anorexia. Swelling of the lymph nodes draining the infected area of skin is common. It is believed that, in most cases, spontaneous resolution may be expected within two to three weeks without treatment but more prompt resolution occurs with adequate treatment. Diagnostic confusion can occur with a variety of skin disorders including shingles, cold sores, cutaneous fungal infections, and eczema (Hay 1998; Resnick 2000). Pyoderma is sometimes used as a synonym for impetigo in tropical countries. This is usually to denote streptococcal, as opposed to staphylococcal, impetigo.    Bullous impetigo is characterised by larger bullae or blisters that rupture less readily and can persist for several days. Usually there are fewer lesions and the trunk is affected more frequently than in non‐bullous impetigo. Diagnostic confusion can occur with thermal burns, blistering disorders (e.g. bullous pemphigoid), and Stevens Johnson syndrome.

Causes

Staphylococcus aureus (S. aureus) is considered to be the main bacterium that causes non‐bullous impetigo. However, Streptococcus pyogenes (S. pyogenes), or both S. pyogenes and S. aureus, are sometimes isolated from the skin. In moderate climates, staphylococcal impetigo is more common, whereas in warmer and more humid climates, the streptococcal form predominates. In moderate climates, the relative frequency of S. aureus infections has also changed with time (Dagan 1993). It was predominant in the 1940s and 1950s, after which Group A streptococci became more prevalent. In the past two decades, S. aureus has become more common again. Bullous impetigo is always caused by S. aureus.

Secondary impetigo may occur as a complication of many dermatological conditions (notably eczema). The eruption appears clinically similar to non‐bullous impetigo. Usually S. aureus is involved. The underlying skin disease may improve with successful treatment of the impetigo, and the converse may also be true.

Complications of non‐bullous impetigo are rare, but local and systemic spread of infection can occur that may result in cellulitis, lymphangitis, or septicaemia. Non‐infectious complications of S. pyogenes infection include guttate psoriasis, scarlet fever, and glomerulonephritis (an inflammation of the kidney that can lead to kidney failure). It is thought that most cases of glomerulonephritis result from streptococcal impetigo rather than streptococcal throat infection, and this has always been an important rationale for antibiotic treatment. The incidence of acute glomerulonephritis has declined rapidly over the last few decades. Baltimore 1985 stated that the risk of developing glomerulonephritis is not altered by treatment of impetigo; however, certain subtypes of Group A streptococci are associated with a much greater risk (Dillon 1979b).

Epidemiology

In the Netherlands, most people with impetigo consult their general practitioner and only approximately 1% of the cases are referred to a dermatologist (Bruijnzeels 1993). Although the incidence of impetigo in general practice has been declining, recent data show an increase in consultations for impetigo (Koning 2006; Van den Bosch 2007). Impetigo is still a common disease particularly in young children. It is the third most common skin disorder in children after dermatitis/eczema and viral warts (Bruijnzeels 1993; Dagan 1993; Mohammedamin 2006). Impetigo is the most common skin infection that is presented in general practice by children aged one to four years of age (Mohammedamin 2006). In British general practice, 2.8% of children aged 0 to 4 and 1.6% aged 5 to 15 consult their GP about impetigo each year (McCormick 1995). In the Netherlands in the late 1980s, the consultation rate was 1.7% of all children under 18 years of age; this increased to 2.1% in 2001 (Koning 2006). Peak incidence occurs between the ages of one and eight years (Koning 2006). In some tropical or developing countries the incidence of impetigo seems to be higher than elsewhere (Canizares 1993; Kristensen 1991).

Description of the intervention

Management options for impetigo include the following:

no pharmacological treatment, waiting for natural resolution, hygiene measures;
topical disinfectants (such as saline, hexachlorophene, povidone iodine, and chlorhexidine);
topical antibiotics (such as neomycin, bacitracin, polymyxin B, gentamycin, fusidic acid, mupirocin, retapamulin, or topical steroid/antibiotic combination); and
systemic antibiotics (such as penicillin, (flu)cloxacillin, amoxicillin/clavulanic acid, erythromycin, and cephalexin).

The aim of treatment includes resolving the soreness caused by lesions and the disease's unsightly appearance (especially on the face), as well as preventing recurrence and spread to other people. An ideal treatment should be effective, cheap, easy to use, and accepted by people. It should be free from side‐effects, and it should not contribute to bacterial resistance. For this reason, antibiotics should not have an unnecessarily broad spectrum (Espersen 1998; Smeenk 1999), and if a topical antibiotic is used, it should, preferably, not be one which may be needed for systemic use (Carruthers 1988; Smeenk 1999).

Waiting for natural resolution could be acceptable if the natural history were known and benign. Impetigo is considered to be self‐limiting by many authors (Hay 1998; Resnick 2000). However, there are no robust data on the natural history of impetigo. Reported cure rates of placebo creams vary from 8% to 42% at 7 to 10 days (Eells 1986; Ruby 1973). Topical cleansing used to be advised in the 1970s as an alternative for antibiotic treatment, but this was later said to be no more effective than placebo (Dagan 1992). Guidelines and treatment advice often do not mention topical cleansing as a treatment because the main concern is preventing the spread of the infection to other children.

A choice has to be made between topical and systemic antibiotic treatment, although in some situations clinicians prescribe both topical and systemic antibiotics. An advantage of the use of topical antibiotics is that the drug can be applied where it is needed, avoiding systemic side‐effects such as gastrointestinal upset. Also, compliance may be better (Britton 1990).

The disadvantages of using topical antibiotics include the risks of developing bacterial resistance and sensitisation, e.g. developing an allergic contact dermatitis to one of the constituents of the topical preparation (Carruthers 1988; Smeenk 1999). This is especially common with the older antibiotics, such as gentamycin, bacitracin, and neomycin (Smeenk 1999). Some preparations (e.g. tetracycline) can cause staining of the skin and clothes.

Staphylococcal resistance against penicillin and erythromycin is common (Dagan 1992). Bacterial resistance against the newer topical antibiotics, such as mupirocin ointment and fusidic acid ointment, is increasing (Alsterholm 2010; de Neeling 1998). Another advantage of the newer topical antibiotics is that mupirocin is never, and fusidic acid not often, used systemically.

How the intervention might work

All treatment options listed above aim to either eradicate or prevent growth of the bacteria.

Why it is important to do this review

Guidelines concerning treatment vary widely ‐ some recommend oral antibiotic treatment, others local antibiotic treatment or even just disinfection in mild cases (Hay 1998; Resnick 2000) ‐ so clinicians have many treatment options. The evidence on what works best is not clear. There is potential conflict between what is in the best interest of the individual and what would best benefit the community in terms of cost and the increase in antibiotic resistance.

Objectives

To assess the effects of treatments for impetigo, including waiting for natural resolution.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised controlled trials.

Types of participants

We included people who have impetigo or impetigo contagiosa diagnosed by a medically trained person (and preferably confirmed by bacterial culture). We recorded whether or not bacterial culture was performed. The diagnosis could be either non‐bullous or bullous impetigo. Studies using a broader diagnostic category such as 'bacterial skin infections' or 'pyoderma' were eligible if a specific subgroup with impetigo could be identified, for which the results were separately described. Studies on secondary impetigo or impetiginised dermatoses were included.

Types of interventions

We included any program of topical or systemic (oral, intramuscular, or intravenous) treatment, including antibiotics, disinfectants, or any other intervention for impetigo, such as 'awaiting natural response'. We excluded studies that only compared different dosages of the same drug.

Types of outcome measures

Primary outcomes

1) Cure as defined by clearance of crusts, blisters, and redness as assessed by the investigator.

2) Relief of symptoms such as pain, itching, and soreness as assessed by participants.

Secondary outcomes

1) Recurrence rate.

2) Adverse effects such as pain, allergic sensitisation, and complications.

3) Development of bacterial resistance.

Search methods for identification of studies

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

Electronic searches

We updated our searches of the following databases on 27 July 2010:

the Cochrane Skin Group Specialised Register using the following search terms: (impetig* or pyoderma or ((staphylococc* or streptococc*) and skin and infection*)) and (therap* or treatment* or intervention*);
the Cochrane Central Register of Controlled Trials (CENTRAL) in The Cochrane Library using the search strategy in Appendix 1;
MEDLINE (from 2005 to the present) using the search strategy in Appendix 2;
EMBASE (from 2007 to the present) using the search strategy in Appendix 3; and
LILACS (Latin American and Caribbean Health Science Information database, from 1982 to the present) using the search strategy in Appendix 4.

Please note: The UK and US Cochrane Centres have an ongoing project to systematically search MEDLINE and EMBASE for reports of trials which are then included in the CENTRAL database. Searching has currently been completed in MEDLINE, from inception to 2004 and in EMBASE, from inception to 2006. Further searches of these two databases to cover the years not searched by the UK and US Cochrane Centres for CENTRAL were undertaken for this review as described above.

A final prepublication search for this review was undertaken on 16 August 2011. Although it has not been possible to incorporate RCTs identified through this search within this review, relevant references are listed under Studies awaiting classification. They will be incorporated into the next update of the review.

Ongoing Trials

We updated our searches of the following ongoing trials databases on 3 August 2010, using the terms 'impetigo' and 'pyoderma':

The metaRegister of Controlled Trials (www.controlled‐trials.com).
The US National Institutes of Health Ongoing TrialsRegister (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 Ongoing Skin Trials Register (www.nottingham.ac.uk/ongoingskintrials).

Searching other resources

Handsearching

We handsearched the Yearbook of Dermatology (1938 to 1966) and the Yearbook of Drug Therapy (1949 to 1966) for the pre‐PubMed era.

References from published studies

We checked references from published studies, including secondary review articles, for further studies.

Unpublished literature

We corresponded with authors and pharmaceutical companies to search for unpublished studies and grey literature.

Language

We did not apply any language restrictions.

Data collection and analysis

Selection of studies

Two authors (JCvdW and SK or RvdS) independently read all abstracts or citations of trials. If one of the authors thought the article might be relevant, a full copy of the article was acquired for further data collection. The reasons for exclusion were recorded for every excluded abstract or citation. Only full reports were included. Two authors independently screened all full‐copy articles (LvSS, SK, RvdS, JCvdW). The articles were selected according to the inclusion criteria. Reasons for exclusion were recorded on a specially‐designed registration form (see the 'Characteristics of excluded studies' table). In the case of doubt, the opinion of a third author was obtained. Many trials studied a range of (skin) infections including impetigo. Frequently, the results of the subgroup of impetigo participants were not reported separately. In these studies, provided they were published in the last 10 years, we contacted trial authors and asked them to provide the results of the subgroup of impetigo participants. We obtained data in this way in only two instances (Blaszcyk 1998; Claudy 2001).

Data extraction and management

Two authors (ADM and CCB), using a pre‐piloted data abstraction form, carried out the full data extraction. The form contained key elements such as time and setting of the study, participant characteristics, bacterial characteristics, type of interventions, outcomes, and side‐effects. We resolved disagreements with the help of a third author (SK).

For this update, RvdS and JCvdW carried out data extraction from newly included papers. When studies assessed outcome measures more than once, we included the assessment that was nearest to one week after the start of therapy. When studies had more than two arms and two of these arms were different dosages of the same drug, we combined these arms.

Assessment of risk of bias in included studies

Assessment of methodological quality

Two independent authors (JCvdW, RvdS and/or AV) assessed the methodological quality of all trials according to the updated guidelines (Higgins 2008). Because we could not read the Japanese study by Ishii 1977, this 'Risk of bias' table was completed by Tetsuri Matsumura. The two studies on which authors of this review were co‐authors (Koning 2003; Koning 2008) were assessed by other authors. The items that were addressed are shown in the 'Risk of bias' table. For feasibility reasons, the methodological quality assessment was not performed under masked conditions. There is no consensus over whether assessment should be done blinded for authors, institutions, journal, or publication year (Jadad 1998).

Unit of analysis issues

In the case of studies with more than two treatment arms, we deemed that pooling these studies under separate comparisons, without adjustment, would result in unit‐of‐analysis errors (overcounting). Should this have occurred, the problem was to be solved by dividing the group size by the number of comparisons.

Assessment of heterogeneity

We used the I² statistic to assess statistical heterogeneity, with I² statistic > 50% regarded as substantial heterogeneity.

Data synthesis

Where there was no statistical evidence of heterogeneity we used the fixed‐effect model to estimate effects. Otherwise, we used the random‐effects model. For dichotomous outcomes we reported risk ratios with 95% confidence intervals.

Sensitivity analysis

We prespecified the following factors for sensitivity analyses:

the quality of the studies;
whether there was observer blinding;
whether there was just a clinical diagnosis or bacterial swab confirmation;
primary versus secondary impetigo;
bullous versus non‐bullous; and
staphylococcal or streptococcal predominance.

During the update, we decided that an overall quality score per study was not useful. Furthermore, most trials were observer‐blind, took bacterial swabs, studied primary impetigo, and had staphylococcal predominance. Sensitivity analyses for these items were, therefore, not possible.

When we analysed the data we decided to consider the results for bullous and non‐bullous impetigo separately.

Results

Description of studies

Results of the search

Our initial search identified approximately 700 papers, 221 of which were selected for full copy reading. For this update, we identified more than 1000 additional papers. Two reviewers screened titles and abstracts, after which, approximately, 60 papers were studied in full copy.

Included studies

For the first version of the review we included 56 papers describing 57 trials. This update identified 12 additional studies, of which 2 were published before 2000 (Farah 1967; Ishii 1977). One study, which was previously included, was excluded because it turned out not to be a randomised trial (Park 1993), bringing the total number of included studies to 68. The lists of ongoing studies (Ongoing studies) and studies awaiting assessment (Studies awaiting classification) show studies that might be eligible for a future update of this review. Regarding the excluded studies, we only report on the most relevant ones (Excluded studies; Characteristics of excluded studies).

Most trials were reported in the English language. Four included studies were reported in Japanese, and one paper each was reported in Thai, Portuguese, Spanish, French, and Danish (some of these had abstracts and tables in English). Trials in Russian, Chinese, German, and French were among those that were excluded (not for language reasons). In instances where none of the authors were competent in the language of the paper, translators provided assistance.

We found an appreciable number of studies from the early 1940s (e.g. MacKenna 1945). These studies were often carried out in military populations, in which impetigo was a frequent disease at the time. These study reports did not meet the inclusion criteria of our review because of inadequate randomisation. The distribution of the included studies by decade is as follows: 1960s ‐ 1 study, 1970s ‐ 5 studies (7%), 1980s ‐ 31 studies (46%), 1990s ‐ 20 studies (29%), and 2000 to 2008 ‐ 11 studies (16%). Five included studies evaluating mupirocin were presented at an international symposium in 1984; we found no publication other than the conference proceedings for three of these (Kennedy 1985; Rojas 1985; Wainscott 1985). Two were published elsewhere as well (Eells 1986; Gould 1984).

Design

All studies were parallel group trials, but there were important design differences between the studies. As mentioned before, many trials included participants with infections other than impetigo, while some trials studied only impetigo. Ages of included participants differed widely, as some studies were carried out exclusively in either adults or children. The average age of study participants in trials that studied a range of skin infections was usually higher than in studies focusing on impetigo alone. With the exception of four studies (Faye 2007; Ishii 1977; Rice 1992; Vainer 1986), all studies performed bacteriological investigations. Although a number of studies explicitly stated that participants with a negative culture were excluded, other studies may also have excluded culture negative participants without reporting those exclusions. No study reported a predominantly streptococcal impetigo. The only studies not to report a preponderance of staphylococcal impetigo were Mertz 1989 and Ruby 1973 (carried out in Puerto Rico and Texas respectively).

Sample sizes

The 68 studies had a total of 5578 evaluable participants; this is an average of 82 participants and a median of 60.5 participants per study (see the 'Characteristics of included studies' tables). In 23 studies the number of participants with impetigo was less than 50; in 10 studies it was less than 20.

Setting

Twenty‐nine of the studies were carried out in North America (in 13 Canadian/Northern states, in 8 Southern states, in 8 multicentres), 15 in Europe, 9 in Central/South America, 10 in Asia, 1 in Africa, and 4 were worldwide multicentre trials. Most studies were carried out in hospital out‐patient clinics (paediatrics or dermatology, 60 studies), but some were carried out in general practice.

Participants

Only three studies exclusively addressed participants with bullous impetigo (Dillon 1983; Ishii 1977; Moraes Barbosa 1986). Seven trials included both bullous and non‐bullous impetigo participants (Barton 1989; Ciftci 2002; Dagan 1992; Koning 2008; Kuniyuki 2005; Oranje 2007; Pruksachat 1993). Three studies on secondary impetigo were included (Fujita 1984; Rist 2002; Wachs 1992). Three other trials included both primary and secondary impetigo participants (Faye 2007; Gonzalez 1989; Tamayo 1991). Thirty‐nine trials studied impetigo alone whereas 29 trials studied participants with a range of (usually skin) infections, impetigo being 1 of them. This was the typical study design when a new antibiotic was studied. This type of study design imposed problems in retrieving outcome data as the outcomes were often presented for all the participants together. We included these studies only if the main outcome measure was presented separately for the subgroup of impetigo participants.

Interventions

The 68 trials evaluated 50 different treatments (26 oral treatments and 24 topical treatments ‐ both including placebo). The systemic treatments that were studied were all administered orally (tablets). A total of 74 different comparisons were made. Some comparisons were made in several studies; some studies made more than one comparison. Sixty‐eight comparisons were made only once. Six different comparisons were made in more than 1 trial, especially when topical mupirocin was studied (topical mupirocin versus oral erythromycin was considered in 10 studies, mupirocin versus fusidic acid was considered in 4 studies, mupirocin versus placebo was considered in 3 studies). For each of these comparisons we pooled the outcomes of the different studies (see Data and analyses).

The most common type of comparison was between 2 different oral antibiotic treatments (29 studies including duplicates). Cephalosporins (15 studies) and macrolide antibiotics, especially erythromycin and azithromycin (9 studies), were most often involved. A topical antibiotic treatment was compared with an oral antibiotic treatment in 22 studies. Nineteen of these comparisons contained erythromycin, mupirocin, or both.

Only two trials studied antiseptic or disinfecting treatments (Christensen 1994; Ruby 1973).

Only seven placebo controlled trials were found (Eells 1986; Gould 1984; Ishii 1977; Koning 2003; Koning 2008; Rojas 1985; Ruby 1973). The latter is the only trial that compared an oral treatment with placebo.

Three studies had three arms but the treatment in two of these were different dosages of the same drug (Blaszcyk 1998; Bucko 2002a;Bucko 2002b). We combined these arms. Nine other studies had more than two arms but with different treatments: three arms (Bass 1997;Demidovich 1990;Dux 1986;Rodriguez‐Solares 1993;Vainer 1986;Wachs 1976), four arms (Kuniyuki 2005;Moraes Barbosa 1986), and five arms (Ruby 1973). Only two of the comparisons in these multiple‐arm studies could be pooled with other studies: erythromycin versus penicillin V from Demidovich 1990, and mupirocin versus erythromycin from Dux 1986. For this reason we refrained from adjusting for multiple treatment comparisons.

Outcomes

Cure as assessed by investigator was our main outcome measure. This was often not defined. Researchers sometimes combined the categories 'cured' and 'improved' and presented those participants as one group. The length of follow‐up varied widely, and it was sometimes not even specified; however, we tried to retrieve the data for follow up as close as possible to seven days after the start of treatment. The development of bacterial resistance to the study drug was reported in only 10 studies.

Excluded studies

One hundred and sixty‐five of the studies did not meet the inclusion criteria for the first version of the review, and 33 more were excluded when updating the review (see the 'Characteristics of excluded studies' tables). The most common reasons included the following: the study was not about impetigo, the outcomes of impetigo participants were not reported separately, or studies were not randomised.

Studies awaiting classification

In the previous version of this review, four studies were awaiting classification. For this update two of these studies were included (Ciftci 2002;Claudy 2001) and two were excluded (Liu 1986; Parish 2000).

Ten studies that were found during the update process are listed in the 'Characteristics of studies awaiting classification' tables, as are a further 6 studies that were identified at the prepublication search. We are currently unable to include or exclude these due to insufficient information about them. We hope to fully incorporate them into future updates of this review.

Ongoing studies

Seven studies that were found during the update process are listed in the 'Characteristics of ongoing studies' tables. These will be fully incorporated into future updates of this review when they are completed.

Risk of bias in included studies

For many of the items that were assessed, the studies did not provide enough information (Figure 1; Figure 2).

Methodological quality summary: review authors' judgements about each methodological quality item for each included study.

Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.

Sequence generation

Fourteen of the studies reported an adequate generation of the randomisation scheme. All other papers did not report on this item.

Allocation

All but two of the included studies were described as randomised as this was a selection criterion. For two papers in Japanese, this was unclear, and these papers were given the benefit of the doubt (see Figure 1). Most papers did not describe the method of randomisation in detail, so the method could not be judged as appropriate. Only 19 of the 68 studies provided information on allocation concealment. In most cases (18 of 19), treatment allocation was considered to be concealed.

Blinding

In many cases it was not clear whether the participant, the caregiver, or the outcome assessor were blinded. A total of 15 studies were considered to be adequately blinded (see Figure 1). In 24 studies, at least 1 party was considered not to be blinded. In 29 papers, the information was insufficient to judge blinding.

Inclusion and exclusion criteria of the trials

In 10 of our included studies, the inclusion and exclusion criteria of the trial were not specified in more detail than saying 'patients with impetigo' (see Figure 1).

Incomplete outcome data

In some studies, high numbers lost to follow up were recorded. Thirty‐four studies either included an intention‐to‐treat analysis or had fewer than 10% dropouts balanced between groups. For some other studies, an intention‐to‐treat analysis could be calculated from the data presented in the study.

Effects of interventions

Primary outcomes: 1) clinical cure

The first primary outcome was clinical cure (or improvement) as assessed by the investigator. When this was assessed more than once, we only included the assessment that was nearest one week from commencement of treatment.

Under the following two main headings ('non‐bullous impetigo' and 'bullous impetigo') we have grouped all studies that either included only primary impetigo, combined primary and secondary impetigo, or did not specify whether participants had primary or secondary impetigo. The third heading 'secondary impetigo' addresses all studies that focused exclusively on secondary impetigo (see Background for an explanation).

(a) Non‐bullous impetigo

(i) Topical antibiotics

Topical antibiotics versus placebo (six studies, four comparisons)

Overall topical antibiotics showed better cure rates or more improvement than placebo (pooled risk ratio (RR) 2.24, 95% CI 1.16 to 3.13 using a random‐effects model, I² = 53%) (see Analysis 1.1). This result was consistent for mupirocin (RR 2.21, 95% CI 1.59 to 3.05; 3 studies ‐ Eells 1986; Gould 1984; Rojas 1985) (see Analysis 1.1), fusidic acid (RR 4.42, 95% CI 2.39 to 8.17; 1 study ‐ Koning 2003) (see Analysis 1.1), and retapamulin (RR 1.64, 95% CI 1.30 to 2.07; 1 study ‐ Koning 2008) (see Analysis 1.1). In one small study (Ruby 1973), bacitracin did not show a significant difference in cure rate compared with placebo (RR 3.71, 95% CI 0.16 to 85.29) (see Analysis 1.1).

1.1 — Comparison 1 Non‐bullous impetigo: topical (Top) antibiotic (Ab) vs placebo (P), Outcome 1 Cure/improvement.

Topical antibiotic versus another topical antibiotic (14 studies, 15 comparisons)

Only one topical antibiotic showed superiority over another topical antibiotic ‐ in a single study: gentamycin over neomycin (RR 1.43, 95% CI 1.03 to 1.98; Farah 1967) (see Analysis 2.1). Also from a single study, the difference between retapamulin over fusidic acid was not statistically significant (RR 1.05, 95% CI 1.00 to 1.11; Oranje 2007) (see Analysis 2.1). There were 12 different comparisons: 4 studies (Gilbert 1989; Morley 1988; Sutton 1992; White 1989) compared mupirocin with fusidic acid (RR 1.03, 95% CI 0.95 to 1.11) (see Analysis 2.1), and the remaining 11 were all only represented by a single study.

2.1 — Comparison 2 Non‐bullous impetigo: topical (Top) antibiotic (Ab) vs another topical (Top) antibiotic (Ab), Outcome 1 Cure/improvement.

Topical antibiotics versus oral (systemic) antibiotics (16 studies, 17 comparisons)

Pooling 10 studies which compared mupirocin with oral erythromycin showed significantly better cure rates, or more improvement, with mupirocin (RR 1.07, 95% CI 1.01 to 1.13) (see Analysis 3.1). However, no significant differences were seen between mupirocin and dicloxacillin (Arredondo 1987), cephalexin (Bass 1997), or ampicillin (Welsh 1987). Bacitracin was significantly worse than oral cephalexin in one small study (Bass 1997), but no difference was seen between bacitracin and erythromycin (Koranyi 1976), or penicillin (Ruby 1973).

3.1 — Comparison 3 Non‐bullous impetigo: topical (Top) antibiotic (Ab) vs oral (Or) antibiotic (Ab), Outcome 1 Cure/improvement.

A sensitivity analysis on the influence of blinding the outcome assessor on the comparison of mupirocin versus erythromycin (10 studies) revealed that there was no clear relationship between blinding of the outcome assessor and the outcome.

Pooling the 2 studies with observer blinding (Britton 1990; Dagan 1992) showed high heterogeneity (I² statistic = 79%) and resulted in a non‐significant difference between the 2 drugs (random‐effects model, RR 1.12, 95% CI 0.86 to 1.46) (see Analysis 3.2).

3.2 — Comparison 3 Non‐bullous impetigo: topical (Top) antibiotic (Ab) vs oral (Or) antibiotic (Ab), Outcome 2 Cure/improvement.

Topical antibiotics versus disinfecting treatment (two studies)

In one study (Ruby 1973), no statistically significant difference in cure/improvement was seen when bacitracin was compared to hexachlorophene (RR 3.71, 95% CI 0.16 to 85.29) (see Analysis 4.1). In another study (Christensen 1994), there was a tendency for fusidic acid cream to be more effective than hydrogen peroxide, but this just failed to reach statistical significance (RR 1.14, 95% CI 1.00 to 1.31) (see Analysis 4.1). When the 2 studies were pooled, topical antibiotics were significantly better than disinfecting treatments (fixed‐effect model, RR 1.15, 95% 1.01 to 1.32, I² statistic 0%) (see Analysis 4.1).

4.1 — Comparison 4 Non‐bullous impetigo: topical (Top) antibiotic (Ab) vs disinfecting treatments (Dt), Outcome 1 Cure/improvement.

Topical antibiotic versus antifungal (one study)

Only one study compared a topical antibiotic to an antifungal, comparing topical mupirocin to topical terbinafine (Ciftci 2002). No statistical difference was seen (RR 1.39, 95% CI 0.98 to 1.96) (see Analysis 5.1).

5.1 — Comparison 5 Non‐bullous impetigo: topical (Top) antibiotic (Ab) vs antifungal (Af), Outcome 1 Cure.

Topical antibiotic + oral antibiotic vs topical antibiotic + oral antibiotic (one study, three comparisons)

In a four‐armed study, three arms addressed the following combinations of a topical antibiotic and an oral antibiotic: topical tetracycline combined with oral cefdinir compared to topical tetracycline combined with oral minomycin, topical tetracycline combined with oral cefdinir compared to topical tetracycline combined with oral fosfomycin, and topical tetracycline combined with oral minomycin compared to topical tetracycline combined with oral fosfomycin (Kuniyuki 2005). None of the three comparisons showed a statistically significant difference (see Analysis 6.1).

6.1 — Comparison 6 Non‐bullous impetigo: topical (Top) antibiotic (Ab) + oral (Or) antibiotic (Ab) vs topical (Top) antibiotic (Ab) + oral (Or) antibiotic (Ab), Outcome 1 Cure.

Topical antibiotic versus topical antibiotic + oral antibiotic (one study, three comparisons)

The fourth arm of the study described under the previous heading (Kuniyuki 2005) was tetracycline. None of the comparisons with the other three treatments (see above) showed a statistically significant difference (see Analysis 7.1).

7.1 — Comparison 7 Non‐bullous impetigo: topical (Top) antibiotic (Ab) vs topical (Top) antibiotic (Ab) + oral (Or) antibiotic (Ab), Outcome 1 Cure.

(ii) Oral antibiotics

Oral antibiotics versus placebo (one study)

A single study (Ruby 1973) found no significant difference between oral penicillin and placebo (RR 7.74, 95% CI 0.43 to 140.26) (see Analysis 8.1).

8.1 — Comparison 8 Non‐bullous impetigo: oral (Or) antibiotics (Ab) vs placebo (P), Outcome 1 Cure/improvement.

Oral antibiotic versus another oral antibiotic: cephalosporin versus another antibiotic (six studies)

All comparisons consisted of single studies (or arms of a single study); only one comparison ‐ cephalexin versus penicillin ‐ showed a significant difference (Demidovich 1990) (see Analysis 9.1).

9.1 — Comparison 9 Non‐bullous impetigo: oral (Or) antibiotic (Ab) (cephalosporin) vs another oral (Or) antibiotic (Ab), Outcome 1 Cure/improvement.

Oral antibiotic versus another oral antibiotic: one cephalosporin versus another cephalosporin (seven studies)

No significant differences were seen between cephalexin and cefadroxil (Hains 1989), cefdinir (Giordano 2006; Tack 1997; Tack 1998); cefaclor and cefdinir (Arata 1989a), or cefditoren and cefadroxil (Bucko 2002b). Cefditoren turned out to be less effective than cefuroxime (Bucko 2002a) (see Analysis 10.1).

10.1 — Comparison 10 Non‐bullous impetigo: oral (Or) cephalosporin vs other oral (Or) cephalosporin, Outcome 1 Cure/improvement.

Oral antibiotic versus another oral antibiotic: macrolides (erythromycin, azithromycin, clindamycin) versus penicillins (penicillin V, dicloxacillin, amoxacillin, cloxacillin, flucloxacillin) (seven studies)

In two studies (Barton 1987; Demidovich 1990), erythromycin showed a better cure rate or more improvement than penicillin (pooled fixed‐effect model, RR 1.29, 95% CI 1.07 to 1.56, I² statistic 0%) (see Analysis 11.1). The other five comparisons consisted of single studies, and they did not show significant differences between macrolides and penicillins.

11.1 — Comparison 11 Non‐bullous impetigo: oral (Or) macrolide vs penicillin, Outcome 1 Cure/improvement.

Oral antibiotic versus another oral antibiotic: macrolide versus another macrolide (one study)

In a single study (Daniel 1991a), no difference in cure rate or improvement was seen between azithromycin and erythromycin (RR 1.18, 95% CI 0.88 to 1.58) (see Analysis 12.1).

12.1 — Comparison 12 Non‐bullous impetigo: oral (Or) macrolide vs another oral (Or) macrolide, Outcome 1 Cure/improvement.

Oral antibiotic versus another oral antibiotic: penicillin versus other oral antibiotics (including other penicillins) (four studies)

In 1 study (Dagan 1989), amoxicillin plus clavulanic acid showed a better cure rate than amoxicillin alone (RR 1.40, 95% CI 1.04 to 1.89) (see Analysis 13.1), but when amoxicillin plus clavulanic acid was compared with fleroxacin in another study (Tassler 1993), no significant difference was seen (RR 1.14, 95% CI 0.80 to 1.62) (see Analysis 13.1). Cloxacillin was significantly superior to penicillin in 2 studies (Gonzalez 1989; Pruksachat 1993) although these studies were statistically heterogeneous (I² statistic 57%) (pooled RR 1.59, 95% CI 1.21 to 2.08) (see Analysis 13.1).

13.1 — Comparison 13 Non‐bullous impetigo: oral (Or) penicillin vs other oral (Or) antibiotic (Ab) (including penicillin), Outcome 1 Cure/improvement.

Other comparisons of oral antibiotics (two studies)

In two studies (Arata 1989b; Claudy 2001), no difference in cure rates/improvement could be detected between lomefloxacin and norfloxacin nor between (oral) fusidic acid and pristinamycin (see Analysis 14.1).

14.1 — Comparison 14 Non‐bullous impetigo: other comparisons of oral (Or) antibiotics (Ab), Outcome 1 Cure/improvement.

Oral antibiotics versus disinfecting treatments (one study)

In a single small study (Ruby 1973), no difference in cure rates/improvement could be detected between penicillin and hexachlorophene (RR 7.74, 95% CI 0.43 to 140.26) (see Analysis 15.1).

15.1 — Comparison 15 Non‐bullous impetigo: oral (Or) antibiotics (Ab) vs disinfecting treatments (Dt), Outcome 1 Cure/improvement.

(iii) Disinfecting treatments

Disinfecting treatments versus placebo (one study)

In a single small study (Ruby 1973), no participants in either the hexachlorophene (n = 11) or placebo group (n = 13) showed cure or improvement. Comparisons of disinfecting treatments with antibiotics are given above.

(b) Bullous impetigo

(i) Topical antibiotics

Topical antimicrobial versus placebo (one study)

In one study (Ishii 1977), topical Eksalbe simplex (a drug containing killed Eschelichia, Staphylococcus, Streptococcus, and Pseudomonas) was compared to placebo. The active drug turned out to be superior (cure/improvement RR 2.30, 95% CI 1.10 to 4.79) (see Analysis 16.1).

16.1 — Comparison 16 Bullous impetigo: topical (Top) antimicrobial vs placebo (P), Outcome 1 Cured/improved after 3 to 4 days.

Topical antibiotics versus other topical antibiotics (one study, three comparisons)

In a small study (Moraes Barbosa 1986), fusidic acid was significantly more effective than both neomycin/bacitracin (RR 10.00, 95% CI 1.51 to 66.43) (see Analysis 17.1) and chloramphenicol (RR 5.00, 95% CI 1.38 to 18.17) (see Analysis 17.1). In the same study, no difference was detected between chloramphenicol and neomycin/bacitracin (RR 2.00, 95% CI 0.21 to 19.23) (see Analysis 17.1).

17.1 — Comparison 17 Bullous impetigo: topical (Top) antibiotic (Ab) vs another topical (Top) antibiotic (Ab), Outcome 1 Cure/improvement.

Topical antibiotics versus oral antibiotics (one study, three comparisons)

The same study (Moraes Barbosa 1986) showed that neomycin/bacitracin was significantly less effective than oral erythromycin (RR 0.14 95% CI 0.02 to 0.99) (see Analysis 18.1). There was no significant difference between either erythromycin and fusidic acid (RR 1.43, 95% CI 0.83 to 2.45) (see Analysis 18.1) or chloramphenicol (RR 0.29, 95% CI 0.07 to 1.10) (see Analysis 18.1).

18.1 — Comparison 18 Bullous impetigo: topical (Top) antibiotic (Ab) vs oral (Or) antibiotic (Ab), Outcome 1 Cure/improvement.

(ii) Oral antibiotics

Oral antibiotic versus another oral antibiotic (one study)

No significant difference was seen between cephalexin and dicloxacillin (Dillon 1983; RR 1.17, 95% CI 0.95 to 1.45) (see Analysis 19.1).

19.1 — Comparison 19 Bullous impetigo: oral (Or) antibiotic (Ab) vs another oral (Or) antibiotic (Ab), Outcome 1 Cure/improvement.

(c) Secondary impetigo

(i) Topical antibiotics

Topical antibiotic versus oral antibiotic (one study)

No significant difference was seen between mupirocin and cephalexin (Rist 2002) (see Analysis 20.1).

20.1 — Comparison 20 Secondary impetigo: topical (Top) antibiotic (Ab) vs oral (Or) antibiotic (Ab), Outcome 1 Cure/improvement.

Antibiotic versus steroid versus antibiotic plus steroid (one study)

In a three‐armed study (Wachs 1976), the comparisons of betamethasone with gentamycin alone or with betamethasone plus gentamycin did not show significant differences (see Analysis 21.1 and Analysis 22.1). The combination of betamethasone and gentamycin cream was significantly more effective than gentamycin alone (RR 2.43, 95% CI 1.29 to 4.57) (see Analysis 23.1).

21.1 — Comparison 21 Secondary impetigo: steroid (S) vs antibiotic (Ab), Outcome 1 Cure/improvement.

22.1 — Comparison 22 Secondary impetigo: steroid (S) + antibiotic (Ab) vs steroid (S), Outcome 1 Cure/improvement.

23.1 — Comparison 23 Secondary impetigo: steroid (S) + antibiotic (Ab) vs antibiotic (Ab), Outcome 1 Cure/improvement.

(ii) Oral antibiotics

In a very small study, no significant difference was detected between cephalexin and enoxacin (Fujita 1984) (see Analysis 24.1).

24.1 — Comparison 24 Secondary impetigo: oral (Or) antibiotic (Ab) vs another oral (Or) antibiotic (Ab), Outcome 1 Cure/improvement.

Primary outcomes: 2) relief of symptoms

The second primary outcome was relief of symptoms, such as pain, itching, and soreness, as assessed by study participants. Although some studies asked about overall satisfaction, acceptability, or treatment preference (McLinn 1988;Rice 1992;Rist 2002; Sutton 1992; White 1989), only one study asked participants to rate their symptoms at follow‐up (Giordano 2006). However, this was a study addressing not only impetigo but other skin infections as well, and results for this outcome were not reported for impetigo separately.

Secondary outcomes: 1) recurrence rate

No relevant data were provided by any study for this outcome.

Secondary outcomes: 2) adverse effects

(i) Topical antibiotics

The trials included in this review usually reported few, if any, side‐effects from topical antibiotics (see Table 1). The studies comparing mupirocin, bacitracin, and placebo reported none (Eells 1986; Ruby 1973). The study that compared fusidic acid to placebo recorded more side‐effects in the placebo group (Koning 2003). Three of 4 studies comparing mupirocin with fusidic acid recorded side‐effects: minor skin side‐effects were reported for mupirocin by 10 out of 368 participants (3%) and for fusidic acid by 4 out of 242 participants (2%). The study that compared retapamulin to placebo found more itching in the group treated with retapamulin (7% vs 1%; P = 0.17) (Koning 2008). In the other study of retapamulin, this side‐effect was reported in less than 1% of cases (Oranje 2007). Most other trials comparing topical antibiotics reported no side‐effects or reported minor skin side‐effects in low numbers (less than 5% of participants).

1. Adverse events.

Study	Adverse events: nature and number or percentage by treatment group
Arata 1989a;Arata 1989b	mainly gastrointestinal: cefdinir 9/142, cefaclor 4/145
Arredondo 1987	mupirocin: nil reported  dicloxacillin: abdominal pain 1/ 31, vomiting 2/31
Barton 1987	not reported
Barton 1988	abdominal pain: erythromycin 1/ 49, dicloxacillin 1/51  vomiting + rash: dicloxacillin 1/51
Barton 1989	gastrointestinal: erythromycin 8/48, mupirocin 4/49
Bass 1997	not reported
Beitner 1996	diarrhoea: cefadroxil 14/327, flucloxacillin 87/324 (all participants)  severe (stomach ache/rash/fever/vomiting): cefadroxil 14/327, flucloxacillin 2/234 (all participants)
Blaszcyk 1998	mainly gastro‐intestinal (half of which were considered treatment‐related): clindamycin 150 mg (19%), clindamycin 300 mg (17%), dicloxacillin 10% (all participants)
Britton 1990	minor gastrointestinal: 11 total, equally divided
Bucko 2002a;Bucko 2002b	unclear and not specified for impetigo participants
Christensen 1994	led to withdrawal: skin irritation 1, burning 1, blistering 1 (all fusidic acid ‐ hydrogen peroxide: 0) mild SE: fusidic acid 9, hydrogen peroxide 13
Ciftci 2002	burning, stinging, itching: 1 in each group rash: 1 in terbinafine group
Claudy 2001	upper gastrointestinal: fusidic acid 6.8% vs pristinamycin 11.6% lower gastrointestinal: 2.5% vs 16.7% hypersensibility: 1.9% vs 5.8%
Dagan 1989	vomiting: amoxicillin 1, amoxicillin and clavulanic acid (augmentin) 0 diarrhoea: amoxicillin 1, amoxicillin and clavulanic acid (augmentin) 0
Dagan 1992	gastrointestinal: erythromycin 11/47, mupirocin 4/51
Daniel 1991a;Daniel 1991b	no subgroup data
Demidovich 1990	nil reported
Dillon 1983	not reported
Dux 1986	pruritus: mupirocin 1/78 nausea and abdominal pain: erythromycin 1/50, cloxacillin 0/20 (all participants)
Eells 1986	not reported
Esterly 1991	mupirocin: nil reported erythromycin: stomach pain and nausea 1/20, vomiting and irritability 1/20, hysterical attacks 1/20
Farah 1967	not reported
Faye 2007	diarrhoea: amoxicillin 2/64 vs erythromycin 11/65
Fujita 1984	mainly gastrointestinal: enoxacin 11/113, cephalexin 4/110 (all participants)
Gilbert 1989	nil reported
Ginsburg 1978	1 child removed from cefadroxil group because of vomiting; no other SE reported
Giordano 2006	diarrhoea: cefdinir 10% vs cephalexin 4% nausea: cefdinir 3% vs cephalexin 6% vaginal mycose of females: cefdinir 3% vs cephalexin 6%
Goldfarb 1988	mild diarrhoea: amupirocin 0/30, erythromycin 5/30
Gonzalez 1989	not reported
Gould 1984	not reported
Gratton 1987	mostly gastrointestinal: erythromycin 8/29 mupirocin: nil reported
Hains 1989	nil reported
Ishii 1977	nil reported
Jaffe 1985	mild diarrhoea: Augmentin® 2/18, cefaclor 5/16 (all participants)
Jaffe 1986	mild staining: hydrocortisone + potassium hydroxyquinoline sulphate cream 2/24, 1% hydrocortisone + 2% miconazole nitrate cream 0/24
Kennedy 1985	nil reported
Kiani 1991	mainly gastrointestinal: azithromycin 30/182, cephalexin: 20/184 Withdrawn: azithromycin 5 (4 gastrointestinal; 1 dizziness and somnolence), cephalexin 1(euphoria) (all participants)
Koning 2003	mainly pain and burning due to povidone iodine: fusidic acid 7/76, placebo 19/80
Koning 2008	any: retapamulin 15/139 vs placebo 2/71 application site pruritis: 9 vs 1
Koranyi 1976	mild abdominal cramps: erythromycin 2/15, bacitracin 0/15
Kuniyuki 2005	not reported
McLinn 1988	gastrointestinal: mupirocin 0/30, erythromycin 6/30
Mertz 1989	nil reported
Montero 1996	mild skin side‐effects: azythromycin 3/100, cefaclor 2/100
Moraes Barbosa 1986	not reported
Morley 1988	all local skin reactions: sodium fusidate 2/191, mupirocin 12/163 (all participants)
Nolting 1988	mild burning: sulconazole 0/32, miconazole 1/34
Oranje 2007	local irritation: retapamulin 6/346 vs sodium fusidate 0/173
Pruksachat 1993	not reported
Rice 1992	stomach ache/diarrhoea/vomiting/itching/burning (%): erythromycin 24/10/7/5/0, mupirocin 2/2/0/12/10
Rist 2002	diarrhoea: mupirocin 2/82 vs cephalexin 3/77
Rodriguez‐Solares 1993	gastrointestinal: azithromycin 2/25, dicloxacillin/flucloxacillin 2/14
Rojas 1985	nausea/vomiting: mupirocin 0/52, vehicle 1/52
Ruby 1973	not reported
Sutton 1992	local: fusidic acid 2/104, mupirocin 4/97
Tack 1997	mainly gastrointestinal: cefdinir 16%, cephalexin 11% (all participants)
Tack 1998	no subgroup data was available; it included only participants that had pathogens susceptible to both study drugs
Tamayo 1991	nil reported
Tassler 1993	mainly gastrointestinal: fleroxacin 17%, amoxicillin/clavunalate 21% (all participants)
Vainer 1986	total 3% skin rash: fusidic acid 1/43 burning and itching: tetracycline/polymyxin B ointment and neomycin/bacitracin ointment both 1/44 and 1/41 respectively
Wachs 1976	not reported
Wainscott 1985	nil reported
Welsh 1987	nil reported
White 1989	minor itching or burning: mupirocin 6/263, fusidic acid 2/127 (all participants)
Wilkinson 1988	rash: mupirocin 0/24, neomycin 1/26 (all participants)

Date	Event	Description
7 March 2012	Amended	The lead author's contact details have been updated.
8 November 2011	New citation required but conclusions have not changed	A substantial amount of new information has been added in the form of 12 newly included studies.
8 November 2011	New search has been performed	New search for studies
29 July 2011	Feedback has been incorporated	In response to peer reviewers' comments, the following major changes were implemented: (1) removed sumscore for risk of bias items; (2) dropped intention to treat analysis as separate risk of bias item; (3) provided more precise information on subjective assessment of symptoms; (4) made a separate table for adverse events.
4 August 2010	Amended	When finalizing the update, new searches were run (2009‐July 2010), resulting in the addition of eight papers to the list of Studies awaiting assessment.
23 February 2009	New citation required and conclusions have changed	New search (2002‐2008), 12 new trials found, one trial previously included discarded. Tables with outcomes of methodological assessments replaced by 'Risk of bias' tables. New author added.
3 October 2008	Amended	Converted to new review format.
2 September 2004	New search has been performed	Minor update
4 January 2003	Amended	New studies found but not yet included or excluded
27 November 2002	New citation required and conclusions have changed	Substantive amendment

Study	Sponsor (product)
Barton 1987	Fleur de Lis Foundation
Barton 1988	Warner‐Lambert Corporation
Barton 1989	Warner‐ Lambert Corporation
Beitner 1996	Bristol‐Myers Squibb (cefadroxil)
Blaszcyk 1998	Pharmacia & Upjohn Asia (clindamycin)
Britton 1990	US Navy Bureau of Medicine and Surgery Clinical Investigation Program
Bucko 2002a, Bucko 2002b	TAF Pharmaceutical Products (cefditoren)
Daniel 1991a; Daniel 1991b	Pfizer Central Research (azithromycin)
Dillon 1983	Eli Lilly Research (cephalexin)
Giordano 2006	Abott Laboratories (cefdinir)
Goldfarb 1988	Beecham Laboratories (mupirocin)
Hains 1989	Bristol‐Myers Squibb (cefadroxil)
Jaffe 1985	Beecham Laboratories (amoxicillin+clavulanic acid)
Koning 2003	Dutch College of General Practitioners
Koning 2008	GlaxoSmithKline (retapamulin)
Mertz 1989	Beecham Laboratories (mupirocin)
Oranje 2007	GlaxoSmithKline (retapamulin)
Rist 2002	GlaxoSmithKline (mupirocin)
Sutton 1992	Leo Laboratories (fusidic acid)
Tack 1997	Parke‐Davis pharmaceutical research (cefdinir)
Tack 1998	Parke‐Davis pharmaceutical research (cefdinir)
Wainscott 1985	Beecham Pharmaceuticals (mupirocin)
White 1989	Beecham Pharmaceuticals (mupirocin)

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Cure/improvement	6	575	Risk Ratio (M‐H, Random, 95% CI)	2.24 [1.61, 3.13]
1.1 Mupirocin	3	173	Risk Ratio (M‐H, Random, 95% CI)	2.18 [1.58, 3.00]
1.2 Fusidic acid	1	156	Risk Ratio (M‐H, Random, 95% CI)	4.42 [2.39, 8.17]
1.3 Bacitracin	1	36	Risk Ratio (M‐H, Random, 95% CI)	3.71 [0.16, 85.29]
1.4 Retapamulin	1	210	Risk Ratio (M‐H, Random, 95% CI)	1.64 [1.30, 2.07]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Cure/improvement	14		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.1 Mupirocin vs rifamycin	1	17	Risk Ratio (M‐H, Fixed, 95% CI)	1.72 [0.96, 3.07]
1.2 Mupirocin vs neomycin	1	32	Risk Ratio (M‐H, Fixed, 95% CI)	1.29 [0.98, 1.71]
1.3 Mupirocin vs bacitracin	1	16	Risk Ratio (M‐H, Fixed, 95% CI)	2.57 [0.97, 6.80]
1.4 Mupirocin vs chlortetracycline	1	14	Risk Ratio (M‐H, Fixed, 95% CI)	1.11 [0.78, 1.59]
1.5 Mupirocin vs polymyxin B/neomycin	1	8	Risk Ratio (M‐H, Fixed, 95% CI)	1.06 [0.56, 2.01]
1.6 Fusidic acid vs neomycin/bacitracin	1	84	Risk Ratio (M‐H, Fixed, 95% CI)	0.92 [0.66, 1.27]
1.7 Fusidic acid vs tetracycline/polymyxin B	1	87	Risk Ratio (M‐H, Fixed, 95% CI)	1.06 [0.75, 1.52]
1.8 Retapamulin vs fusidic acid	1	517	Risk Ratio (M‐H, Fixed, 95% CI)	1.05 [1.00, 1.11]
1.9 Sulcanozol vs miconazole	1	66	Risk Ratio (M‐H, Fixed, 95% CI)	5.31 [0.66, 43.04]
1.10 Hydrocortisone + hydroxyquinoline vs hydrocortisone + miconazole	1	43	Risk Ratio (M‐H, Fixed, 95% CI)	2.06 [0.89, 4.76]
1.11 Gentamycin vs neomycin	1	128	Risk Ratio (M‐H, Fixed, 95% CI)	1.43 [1.03, 1.98]
1.12 Mupirocin vs fusidic acid	4	440	Risk Ratio (M‐H, Fixed, 95% CI)	1.03 [0.95, 1.11]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Cure/improvement	2	292	Risk Ratio (M‐H, Fixed, 95% CI)	1.15 [1.01, 1.32]
1.1 Bacitracin vs hexachlorophene	1	36	Risk Ratio (M‐H, Fixed, 95% CI)	3.71 [0.16, 85.29]
1.2 Fusidic acid vs hydrogen peroxide	1	256	Risk Ratio (M‐H, Fixed, 95% CI)	1.14 [1.00, 1.31]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Cure	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
1.1 Mupirocin vs terbinafine	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Cure/improvement	7	363	Risk Ratio (M‐H, Fixed, 95% CI)	1.06 [0.98, 1.15]
1.1 Erythromycin vs penicillin V	2	79	Risk Ratio (M‐H, Fixed, 95% CI)	1.29 [1.07, 1.56]
1.2 Erythromycin vs dicloxacillin	1	58	Risk Ratio (M‐H, Fixed, 95% CI)	1.03 [0.94, 1.13]
1.3 Erythromycin vs amoxicillin	1	129	Risk Ratio (M‐H, Fixed, 95% CI)	1.00 [0.89, 1.13]
1.4 Azithromycin vs cloxacillin	1	16	Risk Ratio (M‐H, Fixed, 95% CI)	1.4 [0.57, 3.43]
1.5 Azithromycin vs flucloxacillin/dicloxacillin	1	39	Risk Ratio (M‐H, Fixed, 95% CI)	0.84 [0.61, 1.16]
1.6 Clindamycin vs dicloxacillin	1	42	Risk Ratio (M‐H, Fixed, 95% CI)	1.01 [0.80, 1.27]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Cured/improved after 3 to 4 days	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
1.1 Eksalb vs placebo	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]

Methods	Time NR; Japan; range of infections (impetigo 13/265)
Participants	Age 15 to 82 years M/F 150/115 (all participants) Mainly S.aureus
Interventions	A: cefdinir 100 mg, 3 td  B: cefaclor 250 mg, 3 td
Outcomes	Outcomes of the trial 1) 10 days, excellent/good/poor
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information was available in the abstract.
Allocation concealment (selection bias)	Unclear risk	Insufficient information was available in the abstract.
Blinding (performance bias and detection bias)   patient	Unclear risk	Quote: "....double‐blind." Comment: There was unclear blinding of the outcome assessor and caregiver. The participant was probably blinded (see also Figure 2). The test drug packages also included placebo capsules.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	35/300 participants were omitted in the analysis: 16/147 in the cefdinir group (8 due to no or delayed visit to hospital, others for several reasons), 19/153 in the cefaclor group (8 due to no or delayed visit to hospital, others for several reasons) (see table 2).
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	Unclear risk	There was no baseline imbalance.
Randomised?	Unclear risk	Insufficient information was available in the abstract.
Were both inclusion and exclusion criteria specified?	Unclear risk	Insufficient information was available in the abstract.

Methods	Time NR; Japan; range of skin infections (including impetigo 18/259)
Participants	All ages M/F 162/97 Mainly S.aureus (data for all participants)
Interventions	A: lomefloxacin 200 mg, 3 td  B: norfloxacin 200 mg, 3 td
Outcomes	Outcomes of the trial 1) 7 days, cured/improved
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information was available in the abstract.
Allocation concealment (selection bias)	Unclear risk	This was unclear.
Blinding (performance bias and detection bias)   patient	Unclear risk	Quote: "...a double‐blind clinical trial." It was unclear who was blinded (and how). The outcome assessor and caregiver were probably not blinded. The participant was probably blinded (see Figure 1 Dosing schedule).
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	33/291 participants were omitted in the analysis: 15/147 in the NY‐198 group, 17/144 in the norfloxacin group. There was insufficient information in the abstract and figures.
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	Unclear risk	There was no baseline imbalance.
Randomised?	Low risk	Quote: "...were randomly allocated to one of the two drugs."
Were both inclusion and exclusion criteria specified?	Unclear risk	Inclusion (quote): "...skin and soft tissue infections, patients > 15 years". There was no exclusion criteria.

Methods	Time NR; Mexico city, Mexico; range of skin infections (including impetigo 55/61)
Participants	Average age 7 years M/F 30/31 S.aureus 67%
Interventions	A: mupirocin ointment 2%, 3 td, 5 to 10 days  B: dicloxacillin 250 mg, 4 td, 5 to 10 days
Outcomes	Outcomes of the trial 1) 10 days, cure
Notes	Open trial
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information was available.
Allocation concealment (selection bias)	Unclear risk	This was not mentioned in the article.
Blinding (performance bias and detection bias)   patient	High risk	Quote: "In an open trial..." Participants received capsules or ointment. Neither the participant, caregiver, nor outcome assessor were blinded.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	3/61 participants were omitted in the analysis: 2/29 in the mupirocin group, 1/32 in the dicloxacillin group. Reasons for being non‐evaluable for clinical outcome were not specified (but this was a small %).
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	High risk	Baseline imbalance: more severe impetigo in the mupirocin group (9/32 vs 3/29, Table 1). There was no data on compliance.
Randomised?	Low risk	Quote: "After obtaining informed consent, patients were randomly divided into two treatment groups."
Were both inclusion and exclusion criteria specified?	Low risk	Quote: "...pediatric patients with skin infections of sufficient severity to require treatment with a antibiotic." Quote: "Patients who...were excluded from the trial."

Methods	June to August 1986; Missouri, USA; outpatients; only impetigo
Participants	Children (age NR) M/F 29/32 S.aureus 35/65, Streptococcus 2/65, both: 30% PE
Interventions	A: penicillin V 50 mg/kg/day in 4 dd, 10 ds  B: erythromycin 40 mg/kg/day in 4 dd, 10 ds
Outcomes	Outcomes of the trial 1) 7 days, failure
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information was available about sequence generation.
Allocation concealment (selection bias)	Low risk	Quote: "The patients were assigned to receive either erythromycin or penicillin in a random, double‐blind fashion by a pharmacist." Comment: Participants and investigators enrolling participants could not foresee assignment.
Blinding (performance bias and detection bias)   patient	Unclear risk	See above ‐ it was not specified how this was done. It is unclear whether the caregiver, participant, or outcome assessor was blinded.
Incomplete outcome data (attrition bias)   All outcomes	High risk	42/71 participants were omitted in the analysis ‐ reasons and numbers were not specified for each group (6 due to negative culture, 21 not evaluable for effectiveness (not further specified), 6 due to no ascertained compliance, 3 due to no growth of S. aureus alone, 6 withS. aureus alone but not available for follow‐up). 14 were left for analysis in the erythromycin group and 15 in the penicillin group.
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	Unclear risk	Compliance and baseline comparability was unclear.
Randomised?	Low risk	Quote: "The patients were assigned to receive either erythromycin or penicillin in a random, double‐blind fashion by a pharmacist."
Were both inclusion and exclusion criteria specified?	Low risk	Quote: "All patients examined in the outpatient department between June and August 1986 with primarily non bullous impetigo were asked to participate in the study if they were not receiving antibiotics at the time of being seen at CGH, had not taken antibiotics during the preceding week..."

Methods	June to August 1987; Missouri, USA; outpatients; only impetigo
Participants	2 months to 16 years M/F 55/45 S. aureus 46/100, S. pyogenes 9/100, both 25/199 PE
Interventions	A: erythromycin 40 mg/kg/day in 4 dd, 10 ds  B: dicloxacillin 25 mg/kg /day in 4 dd, 10 ds
Outcomes	Outcomes of the trial 1) 5 to 7 days, cure + improved
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information was available.
Allocation concealment (selection bias)	Low risk	Quote: "Participants were randomly assigned in a double‐blind manner by the hospital pharmacist to receive..." Hence, participants and investigators enrolling participants could not foresee assignment.
Blinding (performance bias and detection bias)   patient	Unclear risk	See above ‐ not specified how and who was blinded.
Incomplete outcome data (attrition bias)   All outcomes	High risk	41/100 participants were omitted in analysis ‐ not specified for each group (12/100 were lost to follow up, but not stated from which group). 29 were left in the erythromycin group and 30 left in the dicloxacillin group.
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	Unclear risk	Group assignment of non‐compliant participants was unclear.
Randomised?	Low risk	Quote: "Participants were randomly assigned in a double‐blind manner by the hospital pharmacist to receive..."
Were both inclusion and exclusion criteria specified?	Low risk	Quote: "During the months of June, July and August, 1987, 100 children with impetigo, from whom informed consent was obtained, were consecutively enrolled in the study." Quote: "Exclusion criteria included..."

Methods	June to August 1988; Missouri, USA; outpatients; only impetigo
Participants	3 months to 16 years M/F 49/48 S. aureus 80% PNE
Interventions	A: erythromycin 40 mg/kg/day in 3 dd, 7 days  B: mupirocin ointment 2%, 3 td, 7 days
Outcomes	Outcomes of the trial 1) 4 to 7 days, cured + improved
Notes	14% bullous
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information was available.
Allocation concealment (selection bias)	Unclear risk	This was not mentioned in the article.
Blinding (performance bias and detection bias)   patient	High risk	Participant and caregiver were not blinded because they received either capsules or ointment. It is not mentioned in the article whether the outcome assessor was blinded (probably not, because the caregiver and participant were not blinded)
Incomplete outcome data (attrition bias)   All outcomes	Low risk	1(/97) participant was omitted in the analysis, specified: 1/48 in the erythromycin group (lost to follow up), 0/49 in the mupirocin group.
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	Unclear risk	Compliance was not reported.
Randomised?	Low risk	Quote: "Participants were randomly assigned to receive either..."
Were both inclusion and exclusion criteria specified?	Low risk	Quote: "Children over 6 weeks of age with a clinical diagnosis of impetigo were invited to participate in the study. Exclusion criteria included..."

Methods	Time NR; Honolulu, Hawaii; hospital outpatients; only impetigo
Participants	Average age 3.8 years Sex NR S. aureus 41/48 PNE
Interventions	3 arms:  A: cephalexin 50 mg/kg/day in 3 dd + placebo ointment, 10 days  B: mupirocin ointment 2%, 3 td + liquid oral placebo  C: bacitracin ointment 500 units/g, 3 td + liquid oral placebo
Outcomes	Outcomes of the trial 1) 8 to 10 days, cure
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "...clinical pharmacist assigned them by a table of random numbers to one of the three treatment groups".
Allocation concealment (selection bias)	Low risk	See above and the quote: "The clinician, patients and their parents were not aware of which of the three treatment regimens they were assigned." Comment: Central allocation ‐ participants and investigators enrolling participants could not foresee assignment.
Blinding (performance bias and detection bias)   patient	Low risk	Quote: "The clinician, patients and their parents were not aware of which of the three treatment regimens they were assigned." Comment: The outcome assessor, participant, and caregiver were all blinded.
Incomplete outcome data (attrition bias)   All outcomes	High risk	6/32 participants were omitted in the analysis: 0/10 in the cephalexin group, 5/12 in the mupirocin group, 1/10 in the bacitracin group (missing Imbalance for missing data).
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	Unclear risk	There was a baseline imbalance for size and type of lesion. Compliance was assessed in only 17 participants.
Randomised?	Low risk	Quote: "...clinical pharmacist assigned them by a table of random numbers to one of the three treatment groups."
Were both inclusion and exclusion criteria specified?	Low risk	Quoted from the referred article Demidovich: "Children presenting with impetigo to our clinic were eligible for the study. Exclusion criteria were..."

Methods	December 1992 to November 1994; 25 centres, Sweden; outpatients; range of skin infections (impetigo 60/327)
Participants	Age range 3 to 80 years S. aureus 86% of 327, Streptococcus 14% of 327 Included only participants with bacteria sensitive to both drugs PE
Interventions	A: cefadroxil 40 mg/kg/day, 10 days  B: flucloxacillin tablets 750 mg, 2 td, or susp 30 to 50 mg/kg/day in 2 to 3 dd, 10 days
Outcomes	Outcomes of the trial 1) 10 to 12 days, cure/improved/failed
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information was available.
Allocation concealment (selection bias)	Unclear risk	The method of concealment was not described.
Blinding (performance bias and detection bias)   patient	High risk	Quote: "...single blinded." Quote: "Statistical analysis was performed blinded." Comment: The participant, outcome assessor, and caregiver were probably not blinded because participants in both groups did not receive the same administrations of study drugs daily.
Incomplete outcome data (attrition bias)   All outcomes	High risk	334/661 participants were missing mainly due to a lack of a bacterial culture sensitive to both drugs, and 351/661 were omitted from the primary analysis. 33 impetigo participants were included in the primary analysis. Exact reasons for not being evaluable and group assignment were not reported. 19/661 were omitted in the "ITT‐analysis".
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	Unclear risk	Quote: "The randomization produced two comparable groups of patients with no differences in known prognostic factors." Comment: There was no compliance data.
Randomised?	Low risk	Quote: "In this prospective single‐blind comparative and randomized multicentre trial..."
Were both inclusion and exclusion criteria specified?	Low risk	Table 1: Inclusion and exclusion criteria for the subjects participating in the study.

Methods	Period NR; multicentre; Europe, Latin America, Asia; range of skin infections (impetigo 42/539)
Participants	16 to 70 years (all participants) PNE
Interventions	A: clindamycin caps 150 mg, 4 td  B: clindamycin caps 300 mg, 2 td  C: dicloxacillin caps 250 mg, 4 td
Outcomes	Outcomes of the trial 1) 7 days, cure
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information was available.
Allocation concealment (selection bias)	Unclear risk	Quote: "Drug supplies were masked." Comment: Insufficient information was available.
Blinding (performance bias and detection bias)   patient	Low risk	Quote: "Drug supplies were masked." Quote: "Patients in all groups received four administrations of study drugs daily." Comment: The outcome assessor, participant, and caregiver were probably all blinded.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	48/588 were omitted in the analysis: 16/196 in the clindamycin caps 150 mg group, 19/198 in the clindamycin caps 300 mg group, 20/194 in the dicloxacillin caps group. Proportions of participants who did not complete the study medication and reasons were ˜ similar (table II). There were not only impetigo participants.
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	Unclear risk	Compliance data was provided (table II) and well‐balanced. The distribution of baseline characteristics was not provided.
Randomised?	Low risk	Quote: "This prospective, double mask, randomized study..."
Were both inclusion and exclusion criteria specified?	Low risk	Quote: "Patients were selected based on..." Quote: "Patients were ineligible if..."

Methods	October 1988 to October 1989; Portsmouth, Virginia, USA; outpatients; only impetigo
Participants	2 months to 12 years M/F 27/17 S. aureus 26/48 PNE
Interventions	A: erythromycin 40 mg/kg/day in 4 dd + placebo cream  B: mupirocin ointment 2%, 3 td + placebo susp
Outcomes	Outcomes of the trial 1) 10 days, cured + improved
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Using a random numbers table, the hospital pharmacist randomly assigned each patient to one of the groups."
Allocation concealment (selection bias)	Low risk	Quote: "Using a random numbers table, the hospital pharmacist randomly assigned each patient to one of the groups." Comment: central allocation ‐ pharmacy‐controlled.
Blinding (performance bias and detection bias)   patient	Low risk	Quote: "The child group assignment was not known to parents or investigators." Quote: "...assigned each patient to one of two groups: orally administered erythromycin plus topically applied placebo (erythromycin group) or orally administered placebo plus topically applied mupirocin (mupirocin group)." Comment: The outcome assessor, participant, and caregiver were probably all blinded.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	6/54 participants were omitted in the analysis: 2/24 in the mupirocin group, 4/30 in the erythromycin group. Participants not completing the study were left out of the analysis. Reasons for not completing the study were not specified for each group. 3 were lost to follow up, 2 dropped out when misdiagnosis was suspected, and 1 was removed because of S. pyogenes pharyngitis. < 20% withdrawals and numbers were balanced.
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	High risk	Baseline characteristics were imbalanced (sex, severity), and compliance was also skewed.
Randomised?	Low risk	Quote: "Using a random numbers table, the hospital pharmacist randomly assigned each patient to one of the groups."
Were both inclusion and exclusion criteria specified?	Low risk	Quote: "Children aged 12 years and younger with the clinical diagnosis of impetigo..." Quote: "We excluded..."

Methods	Unlear, around 2000; US, multicentre; ambulatory setting; range of skin infections (including impetigo 58/857)
Participants	12 to 93 years M/F 427/430 S.aureus 525/1685, S.pyogenes 53/1685 (including Bucko 2002b) PNE
Interventions	A: cefditoren 200 mg, 2 td, 10 days B: cefditoren 400 mg, 2 td, 10 days C: cefuroxime 250 mg, 2 td, 10 days
Outcomes	Outcomes of the trial 1) 7 to 14 days, cured or improved
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	This was not reported.
Allocation concealment (selection bias)	Low risk	Quote: "Study‐drug containers were dispensed in numeric sequence at each investigative site as patients were enrolled to ensure random assignment."
Blinding (performance bias and detection bias)   patient	Low risk	Quote: "...double‐blind, double‐dummy..." Quote: "Patients' evaluability and outcomes were assessed under blinded conditions". The outcome assessor, caregiver, and participant were all blinded.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	For only impetigo: 2/58 missing impetigo participants in total ‐ 0/19 in the cefditoren 200 mg group, 2/21 in the cefditoren 400 mg group, 0/18 in the cefuroxime 250 mg group. Reasons for missing participants were not specified (but a small % were non‐evaluable)
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	Unclear risk	There was no compliance data and no baseline imbalance.
Randomised?	Low risk	Quote: "Patients were randomized."
Were both inclusion and exclusion criteria specified?	Low risk	Quote: "Eligible patients included..." Quote: "Study exclusion criteria included..."

Methods	Unclear, around 2000; US; multicentre; ambulatory setting; range of skin infections (including impetigo 74/828)
Participants	12 to 95 years M/F 428/400 S.aureus 525/1685, S.pyogenes 53/1685 (including Bucko 2002a) PNE
Interventions	A: Cefditoren 200 mg, 2 td, 10 days B: Cefditoren 400 mg, 2 td, 10 days C: Cefadroxil 500 mg, 2 td, 10 days
Outcomes	Outcomes of the trial 1) 7 to 14 days, cured or improved
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	This was not reported.
Allocation concealment (selection bias)	Low risk	Quote: "Study‐drug containers were dispensed in numeric sequence at each investigative site as patients were enrolled to ensure random assignment."
Blinding (performance bias and detection bias)   patient	Low risk	Quote: "...double‐blind, double‐dummy..." Quote: "Patients' evaluability and outcomes were assessed under blinded conditions". Comment: The outcome assessor, caregiver, and participant were all blinded.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	For only impetigo: 4/74 missing participants ‐ 1/27 in the cefditoren 200 mg group, 0/25 in the cefditoren 400 mg group, 3/22 in the cefadroxil 500 mg group. Reasons for missing participants were not specified (but a small % were non‐evaluable)
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	Unclear risk	There was no compliance data and no baseline imbalance.
Randomised?	Low risk	Quote: "Patients were randomized..."
Were both inclusion and exclusion criteria specified?	Low risk	Quote: "Eligible patients included..." Quote: "Study exclusion criteria included..."

PERMALINK

Interventions for impetigo

Sander Koning

Renske van der Sande

Arianne P Verhagen

Lisette WA van Suijlekom‐Smit

Andrew D Morris

Christopher C Butler

Marjolein Berger

Johannes C van der Wouden

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Background

Description of the condition

Biology and symptoms

Causes

Epidemiology

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Primary outcomes

Secondary outcomes

Search methods for identification of studies

Electronic searches

Ongoing Trials

Searching other resources

Handsearching

References from published studies

Unpublished literature

Language

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Assessment of methodological quality

Unit of analysis issues

Assessment of heterogeneity

Data synthesis

Sensitivity analysis

Results

Description of studies

Results of the search

Included studies

Design

Sample sizes

Setting

Participants

Interventions

Outcomes

Excluded studies

Studies awaiting classification

Ongoing studies

Risk of bias in included studies

1.

2.

Sequence generation

Allocation

Blinding

Inclusion and exclusion criteria of the trials

Incomplete outcome data

Effects of interventions

Primary outcomes: 1) clinical cure

(a) Non‐bullous impetigo

(i) Topical antibiotics

Topical antibiotics versus placebo (six studies, four comparisons)

1.1. Analysis.

Methods	Time NR; Sweden, Germany, UK; Outpatients (Germany) and GP (UK), both (Sweden); only impetigo
Participants	3 + years M/F 131/125 S.aureus 199/256, S.pyogenes 21/256, both 36/256 PE
Interventions	A: hydrogen peroxide cream 1% (Microcid), 2 to 3 td, max 21 days  B: fusidic acid cream gel 2%, 2 to 3 td, max 21 days
Outcomes	Outcomes of the trial 1) evaluation time NR, cure
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "...randomized (in blocks of four)." The process for selecting the blocks was not specified.
Allocation concealment (selection bias)	Unclear risk	Quote: "The tubes were put into identical paper boxes, to keep the trials blind." Comment: Insufficient information was available. The tubes may have been different.
Blinding (performance bias and detection bias)   patient	Unclear risk	Quote: "The tubes were put into identical paper boxes, to keep the trials blind." There was incomplete blinding ‐ participants were probably not blinded (see above), and blinding with regard to the outcome assessor and caregiver is unclear.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	135/391 participants were omitted in the analysis because they were culture negative (not specified per group); 11/156 participants in the M‐group and 3/156 in the F‐group were withdrawn due to deterioration of their impetigo (statistically significant), 3/156 in the F‐group and 0/156 in the M‐group were withdrawn due to adverse events (irritation of the skin, burning, and blistering). All participants fulfilling the prespecified requirement of bacteriologically‐verified impetigo were analysed.
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	Unclear risk	There were no data on baseline comparability and compliance.
Randomised?	Low risk	Quote: "...randomized (in blocks of four)."
Were both inclusion and exclusion criteria specified?	Low risk	Quote: "...were included" Quote: "Patients were not allowed... prior to start of study..."

Methods	1999; Turkey; hospital outpatient department; only impetigo
Participants	Age 10 to 132 months M/F 32/16 S. aureus around 70%
Interventions	A: topical mupirocin 2% 3td for 10 days B: topical terbinafine 1% 3td for 10 days
Outcomes	Outcomes of the trial 1) 10 days, cure
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information was available.
Allocation concealment (selection bias)	Unclear risk	Insufficient information was available.
Blinding (performance bias and detection bias)   patient	High risk	Quote: "...mupirocin group was instructed to use Bactroban 2% ointment and terbinafine group was instructed to use Lamisil 1% cream topically three times daily for ten days". The outcome assessor, caregiver, and participant were not blinded.
Incomplete outcome data (attrition bias)   All outcomes	High risk	14/62 participants were not analysed: 6/31 were missing in the mupirocin group, 8/31 were missing in the terbinafine group. Quote: "At the end of the treatment, 25 participants in the mupirocin group and 23 participants in the terbinafine group were considered eligible" . > 20% missing.
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	High risk	Quote: "The group had similar features except for the time from appearance of lesions to hospital admission." There was a mean of 5.44 in the mupirocin group versus 6.78 in the terbinafine group.
Randomised?	Low risk	Quote: "...in a randomized fashion."
Were both inclusion and exclusion criteria specified?	Low risk	Quote: "...were excluded." Quote: "....children, less than 12 years old, presenting with impetigo to..."

Methods	Time NR; France; ambulatory setting (dermatology outpatient departments); range of skin infections (including impetigo 53/334)
Participants	All participants: age > 18 years M/F 206/128 S aureus: 162/334; S pyogenes 34/334
Interventions	A: oral fusidic acid 2 x 250 mg 2 td for 7.5 days B: oral pristinamycin 2 x 500 mg 2 td for 10 days
Outcomes	Outcomes of the trial 1) 11 days, cured and improved
Notes	Outcome data for impetigo participants was provided by the author (personal communication).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information was available.
Allocation concealment (selection bias)	Unclear risk	Insufficient information was available.
Blinding (performance bias and detection bias)   patient	Unclear risk	Quote: "Afin de garantir le double insu, chaque patient recevait le traitement dont 2.5 jours de placebo". [To ensure double blinding, each patient received a placebo for 2.5 days]. The participants were blinded, but blinding is unclear with regard to the caregiver and outcome assessor.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	313/334 participants were analysed (< 10% not in analysis). There is no data for impetigo participants.
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	Unclear risk	There were no compliance data and no baseline comparison.
Randomised?	Low risk	Quote: "Une etude multicentrique, prospective, randomisée..." [A randomised, prospective, multicentre study...]
Were both inclusion and exclusion criteria specified?	Low risk	Quote: "Tout patient ambulatoire, âgé de plus de 18 ans, avec une pyodermite superficielle nécessitant une antibiothérapie orale et ayant donné son consentement éclairé pouvait être inclus dans l'essai à condition de ne presenter aucun des critères d'exclusion suivants." [Most ambulatory participants, older than the 18 years old, with a superficial pyoderma requiring oral antibiotics and with given informed consent could be included in the study provided there were none of the following exclusion criteria present.]

Methods	May to October 1987; Negev region, Israel; outpatients; only impetigo
Participants	6 months to 9 years Sex NR S. aureus 37/51, S. pyogenes 14/51 PE
Interventions	A: amoxicillin trihydrate syrup 40 mg/kg/day, in 3 dd, 10 days  B: amoxicillin/clavulanic acid syrup 40 + 10 mg/kg/day, in 3 dd, 10 days
Outcomes	Outcomes of the trial 1) 5 days, cure + improved
Notes	There was missing data from the first follow‐up measurement for 4/26 participants in the amoxicillin trihydrate syrup group and 3/25 participants in the amoxicillin/clavulanic acid syrup group.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information was available.
Allocation concealment (selection bias)	Unclear risk	Insufficient information was available.
Blinding (performance bias and detection bias)   patient	Unclear risk	Quote: "...in a double‐blind fashion". It is unclear whether the outcome assessor, participant, or caregiver were blinded.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	7/52 (< 20%) participants were omitted in the analysis after 5 days.
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	High risk	There was a baseline imbalance for lymphadenopathy > 20%. There were no compliance data.
Randomised?	Low risk	Quote: "After obtaining the cultures, patients were randomized to..."
Were both inclusion and exclusion criteria specified?	High risk	Quote: "We included..." Exclusion criteria was not mentioned.

Methods	July 1989 to October 1990; Negev region, Israel; outpatients; only impetigo (bullous and non‐bullous)
Participants	< 16 years M/F 56/46 S. aureus 90/102, streptococci 1/3 of participants PNE
Interventions	A: erythromycin susp 50 mg/kg/day 3 td + placebo ointment, 7 days  B: mupirocin ointment 2% 3 td + oral placebo susp, 7 days
Outcomes	Outcomes of the trial 1) 7 days, failed
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information was available.
Allocation concealment (selection bias)	Low risk	Quote: "The randomized code was prepared by Beecham Pharmaceutical and was not known to the investigators until after the raw data were tabulated."
Blinding (performance bias and detection bias)   patient	Low risk	Quote: "The randomized code was prepared by Beecham Pharmaceutical and was not known to the investigators until after the raw data were tabulated." The erythromycin group received a placebo ointment and the mupirocin group received an oral placebo suspension. The outcome assessor, caregiver, and participant were probably all blinded.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	13/102 participants were omitted in the analysis: 8/51 in the erythromycin group (1 due to side‐effects, 7 due to refusal to continue treatment or to return for the follow‐up visit), 5/51 missing in the mupirocin group (all due to refusal to continue treatment or to return for the follow‐up visit).
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	Unclear risk	There were age and sex differences at baseline (table 1), although they were not significant. There were no compliance data.
Randomised?	Low risk	Quote: "...were randomized into two groups."
Were both inclusion and exclusion criteria specified?	Low risk	Quote: "Infants... were enrolled." Quote: "Excluded groups were..."

Methods	1987 to 1991; Belgium, France, FRG, Netherlands, Norway, UK; setting unclear; range of skin infections (including impetigo 69/308)
Participants	16 to 80 years All participants: S. aureus 195/308, streptococci 59/308 PNE
Interventions	A: azithromycin 250 mg twice (day 1),once daily (day 2 to 5), 5 days  B: erythromycin 500 mg 4 td, 7 days
Outcomes	Outcomes of the trial 1) 11 to 16 days, cured
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Patients were allocated to treatment with azithromycin or erythromycin in a 1:1 ratio using a randomization list."
Allocation concealment (selection bias)	Unclear risk	See above ‐ it is unclear whether participants and investigators enrolling participants could foresee assignment.
Blinding (performance bias and detection bias)   patient	High risk	Participants in both groups did not receive the same administrations of study drugs daily. The outcome assessor was likely to also be the caregiver, so probably all 3 were not blinded.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	The number of impetigo participants not included in analysis was small and well‐balanced (1 vs 2).
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	Unclear risk	There were no compliance data. Baseline characteristics were well‐balanced.
Randomised?	Low risk	Quote: "Patients were allocated to treatment with azithromycin or erythromycin in a 1:1 ratio using a randomization list."
Were both inclusion and exclusion criteria specified?	Low risk	Quote: "In order to be included..." Quote: "Exclusion criteria were..."

Methods	1987 to 1989; Belgium, Germany, Ireland, UK; setting unclear; range of skin infections (including impetigo 17/323)
Participants	Adults 17 to 90 years All participants: S aureus 158/323, streptococci 41/323 PNE
Interventions	A: azithromycin 250 mg twice (day 1),once daily (day 2 to 5), 5 days  B: cloxacillin 500 mg, 4 td, 7 days
Outcomes	Outcomes of the trial 1) 11 to 16 days, cured/improved/failed
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Using a presupplied randomization list patients were allocated to receive azithromycin or cloxacillin in the ratio of 2:1."
Allocation concealment (selection bias)	Unclear risk	See above ‐ it is unclear whether participants and investigators enrolling participants could foresee assignment.
Blinding (performance bias and detection bias)   patient	High risk	Participants in both groups did not receive the same administrations of study drugs daily. The outcome assessor, caregiver, and participant were probably not blinded.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Only 1 impetigo participant was not in the analysis.
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	Unclear risk	There were no compliance data. Baseline characteristics were comparable.
Randomised?	Low risk	Quote: "Using a presupplied randomization list patients were allocated to receive azithromycin or cloxacillin in the ratio of 2:1."
Were both inclusion and exclusion criteria specified?	Low risk	Quote: "In order to be included..." Quote: "Exclusion criteria were..."

Methods	Time NR; Honolulu, Hawaii; outpatients; only impetigo
Participants	5 months to 15 years, average 3 years S. aureus 45/73, GABHS 6/73, both 14/73 PNE
Interventions	A: penicillin V 40 to 50 mg/kg/day in 3 dd, 10 days  B: cephalexin 40 to 50 mg/kg/day in 3 dd, 10 days  C: erythromycin 30 to 40 mg/kg/day in 3 dd, 10 days
Outcomes	Outcomes of the trial 1) 8 to 10 days, failed
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information was available.
Allocation concealment (selection bias)	Low risk	Quote: "The pharmacist randomly assigned them to one of three treatment regimens." Central allocation ‐ participants could not foresee assignment.
Blinding (performance bias and detection bias)   patient	Unclear risk	Quote: "Patients were reevaluated...by one of the authors, both of whom were blinded to the treatment each child was receiving." Comment: Participants were probably not blinded. The caregiver and outcome assessor were probably blinded.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	2/75 participants were omitted in the analysis: 2 participants were lost to follow up (not further specified).
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	Unclear risk	Quote: "There was not a significant difference in disease severity among treatment groups." Compliance in both groups was comparable, but low.
Randomised?	Low risk	Quote: "The pharmacist randomly assigned them to one of three treatment regimens."
Were both inclusion and exclusion criteria specified?	Low risk	Quote: "Children presenting with impetigo to our pediatric clinic were eligible for the study. Exclusion criteria were..."

Methods	1980 summer/fall; Alabama, USA; outpatients; only impetigo (bullous impetigo 57/70)
Participants	Average age 3.2 years MF 41/37 S. aureus: 64/70 PNE
Interventions	A: cephalexin 50 mg/kg/day in 2 dd (> 20 kg: 500 mg 2 td)  B: dicloxacillin 15 mg/kg/day in 4 dd (> 40 kg: 125 mg 4 td)
Outcomes	Outcomes of the trial 1) Prompt cure
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Patients were randomly assigned, according to a standard table, to receive..." (Referring to a standard table.)
Allocation concealment (selection bias)	Unclear risk	See above ‐ it is unclear whether participants and investigators enrolling participants could foresee assignment.
Blinding (performance bias and detection bias)   patient	High risk	Participants in both groups did not receive the same administrations of study drugs daily. The outcome assessor, caregiver, and participant were probably not blinded.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	8/78 participants were omitted in the analysis: 5 vs 3 participants failed to return or, with a negative culture, were not included in the analysis (< 20% and balanced).
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	Unclear risk	Quote: "Preference was given to patients with skin infections typical of staphylococcal bullous impetigo." Comment: Furthermore, there were no baseline differences, and compliance was not reported.
Randomised?	Low risk	Quote: "Patients were randomly assigned, according to a standard table, to receive..."
Were both inclusion and exclusion criteria specified?	Low risk	Quote: "The criterion for enrolment was..." Quote: "...were excluded."

Methods	Time NR; Toronto, Canada; setting unclear; range of skin infections (including impetigo 36/149)
Participants	Average age 22 years M/F 81/68 Bacterial culture results unclear PNE
Interventions	A: mupirocin ointment 2%, 3 td, 7 days  B: erythromycin 250 mg, 4 td, 7 days  C: cloxacillin 250 mg, 4 td, 7 days
Outcomes	Outcomes of the trial 1) 7 days, cure/improved/failure. Clocacillin: no participants with impetigo allocated
Notes	2 cases of secondary impetigo, both in the mupirocin group, were excluded from the results presented here.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information about the sequence generation process was available, and there was unexpected distribution (78 vs 50 vs 20).
Allocation concealment (selection bias)	Unclear risk	Quote: "...were randomized into two treatment groups by each investigator." Comment: It is unclear whether participants and investigators enrolling participants could foresee assignment.
Blinding (performance bias and detection bias)   patient	Unclear risk	Quote: "...single‐blind". Comment: It is not clear who was blinded and how this was done. Also, participants in both groups did not receive the same administrations of study drugs daily. Participants were probably not blinded. The blinding of outcome assessor and caregiver is unclear.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	1 (/149) participant was omitted in the analysis: 1/79 in the mupirocin group due to an infected cyst (not included in analysis), 0/50 in the erythromycin group, 0/20 in the cloxacillin group.
Selective reporting (reporting bias)	Unclear risk	This was unclear.
Other bias	Unclear risk	Compliance was not reported. There was a large age difference between groups (mean 22 vs 31 years), unknown for impetigo participants.
Randomised?	Low risk	Quote: "In each section of the study, patients with primary or secondary skin infections were randomized into two treatment groups."
Were both inclusion and exclusion criteria specified?	Low risk	Quote: "Patients with primary and secondary skin infections that were severe enough were included in three parallel‐study groups." Quote: "Patient who did not..."