Systematic review of treatments for atopic eczema

What is atopic eczema?

Atopic eczema is a chronic inflammatory skin condition characterised by an itchy red rash that favours the skin creases such as folds of elbows or behind the knees. The eczema lesions themselves vary in appearance from collections of fluid in the skin (vesicles) to gross thickening of the skin (lichenification) on a background of poorly demarcated redness. Other features such as crusting, scaling, cracking and swelling of the skin can occur. 1 Atopic eczema is associated with other atopic diseases such as hay fever and asthma. People with atopic eczema also have a dry skin tendency, which makes them vulnerable to the drying effects of soaps. Atopic eczema typically starts in early life, with about 80% of cases starting before the age of 5 years. 2

Is atopic eczema ‘atopic’?

Although the word ‘atopic’ is used when describing atopic eczema, it should be noted that around 20% of people with otherwise typical atopic eczema are not atopic as defined by the presence of positive skin prick test reactions to common environmental allergens, or through blood tests, which detect specific circulating immunoglobulin E (IgE) antibodies. 3 The word ‘atopic’ in the term atopic eczema is simply an indicator of the frequent association with atopy and the need to separate this clinical phenotype from the ten or so other forms of eczema such as irritant, allergic contact, discoid, venous, seborrhoeic and photosensitive eczema, which have other causes and distinct patterns. The terms atopic eczema and atopic dermatitis are synonymous. The term atopic eczema or just eczema is frequently used in the UK, whereas atopic dermatitis is used more in North America. Much scientific energy has been wasted in debating which term should be used.

How is atopic eczema defined in clinical studies?

Very often, no definition of atopic eczema is given in clinical studies such as clinical trials. This leaves the reader guessing as to what sort of people were studied. Atopic eczema is a difficult disease to define as the clinical features are highly variable. This variability can be in the skin rash morphology (e.g. it can be dry and thickened or weeping and eroded), in place (e.g. it commonly affects the cheeks in infants and skin creases in older children) and time (it can be bright red one day and apparently gone in a couple of days). There is no specific diagnostic test that encompasses all people with typical eczema and which can serve as a reference standard. Diagnosis is therefore essentially a clinical one.

Up until the late 1970s, at least 12 synonyms for atopic eczema were in common usage in the dermatological literature, and it is not certain if physicians were all referring to the same disease when using these terms. A major milestone in describing the main clinical features of atopic eczema was the Hanifin and Rajka diagnostic criteria of 1980. 4 These are frequently cited in clinical trial articles, and they at least provide some degree of confidence that researchers are referring to a similar disease when using these features. It should be borne in mind however that these criteria were developed on the basis of consensus, and their validity and repeatability is unknown in relation to physician's diagnosis. 3 Some of the 30 or so minor features have since been shown not to be associated with atopic eczema, and many of the terms, which are poorly defined, probably mean something only to dermatologists. Scientifically developed refinements of the Hanifin and Rajka diagnostic criteria, mainly for epidemiological studies, have been developed by a UK working party, and these criteria have been widely used throughout the world. 5 According the these criteria,6 in order to qualify as a case of atopic eczema, the person must have:

• an itchy skin condition plus three or more of:

• past involvement of the skin creases, such as the bends of elbows or behind the knees

• personal or immediate family history of asthma or hay fever

• tendency towards a generally dry skin

• onset under the age of 2 years

• visible flexural dermatitis as defined by a photographic protocol.

Binary or continuous disease?

It is unclear whether atopic eczema is an ‘entity’ in itself or whether it is part of a continuum when considered at a population level. Some studies have suggested that atopic dermatitis score is distributed as part of a continuum. 3 Although it may be appropriate to ask the question: “How much atopic eczema does he/she have?” as opposed to “Does he/she have atopic eczema – yes or no?”, most population and clinical studies require a categorical cut-off point and tend to include well-defined and typical cases.

Is it all one disease?

It is quite possible that there are distinct subsets of atopic eczema, for example those cases associated with atopy and those who have severe disease with recurrent infections. Until the exact genetic and causative agents are known, it is wiser to consider the clinical disease as one condition. Perhaps sensitivity analyses can be done for those who are thought to represent distinct subsets (e.g. those who are definitely atopic with raised circulating IgE to allergens, or those with severe disease and associated asthma). 3

Age

Atopic eczema is commoner in childhood, particularly in the first 5 years of life. One study of 2365 patients who were examined by a dermatologist for atopic eczema in the town of Livingston, Scotland, suggested that atopic eczema is relatively rare over the age of 40, with a 1-year period prevalence of 0.2%. 10 Nevertheless, adults over 16 years made up 38% of all atopic eczema cases in that community. Adults also tend to represent a more persistent and severe subset of cases.

Severity distribution

Most cases of childhood eczema in any given community are mild. One recent study by Emerson and colleagues found that 84% of 1760 children aged 1–5 years from four urban and semi-urban general practices in Nottingham were mild, as defined globally by the examining physician, with 14% of cases in the moderate category and 2% in the severe category. 11 Disease severity was not the only determinant of referral for secondary care, however. This severity distribution was very similar to another recent population survey in Norway. 12

Economic costs

In financial terms, the cost of atopic eczema is potentially very large. One recent study of an entire community in Scotland estimated the mean personal cost to the patient at £25.90 over a 2-month period with the mean cost to the health service of £16.20. 13 If these results were extrapolated to the UK population, the annual personal costs to patients with atopic eczema based on lower prevalence estimates than recent studies suggest would be £297 million. The cost to the health service would be £125 million and the annual cost to society through lost working days would be £43 million making the total expenditure on atopic eczema £465 million per year (1995 prices). Another recent study from Australia found that the annual personal financial cost of managing mild, moderate and severe eczema was Aus$330,818 and $1255, respectively, which was greater than the costs associated with asthma in that study. 14

Genetics

There is strong evidence to suggest that genetic factors are important in the predisposition to atopic eczema. In addition to family studies, twin studies have shown a much higher concordance for monozygotic (85%) when compared with dizygotic twins (21%). 15 Preliminary work has suggested that a marker for IgE hyper-responsiveness might be located on chromosome 11q, but this has not been consistent. It is possible that the tendency to atopic eczema might be inherited independently from atopy.

Environment

While genetic factors are probably a very important factor for disease predisposition, there are numerous general and specific clues that point strongly to the crucial role of the environment on disease expression. 16 It is difficult to explain the large increase in atopic eczema prevalence over the past 30 years, for instance, in genetic terms. 9 It has been shown that atopic eczema is commoner in wealthier families. 17 It is unclear whether this positive social class gradient is a reflection of indoor allergen exposures or whether it reflects a whole constellation of other factors associated with ‘development’. Other studies have shown an inverse association between eczema prevalence and family size. 18 This observation led to the ‘hygiene hypothesis’, that is that children in larger families were protected from expressing atopy because of frequent exposure to infections. 19 Some evidence for this protective effect of infections on atopic eczema has been shown in relation to measles infection in Guinea Bissau. 20

Migrant studies also point strongly to the role of environmental factors in atopic eczema. It has been shown that 14.9% of black Caribbean children living in London develop atopic eczema (according to the UK diagnostic criteria) compared with only 5.6% for similar children living in Kingston, Jamaica. 21 Other migrant studies reviewed elsewhere have consistently recorded large differences in ethnic groups migrating from warmer climates to more prosperous cooler countries.

Further work has suggested that the tendency to atopy may be programmed at birth and could be related to factors such as maternal age. 22 The observation that many cases of atopic eczema improve spontaneously around puberty is also difficult to explain in genetic terms alone. 2 Specific risk factors for eczema expression in the environment are still not fully elucidated. Allergic factors such as exposure to house dust mite may be important but non-allergic factors such as exposure to irritants, bacteria and hard water may also be important. 23

How is care organised in the UK?

Most children with atopic eczema in the UK are probably managed by the primary care team. This includes advice from pharmacists, health visitors, practice nurses and family practitioners. About 4% of children with atopic eczema are referred to a dermatologist for further advice. 11

The quality of service provided by secondary care for eczema sufferers has recently been audited by the British Association of Dermatologists. Although most departments provided a high-quality service, some aspects of care, such as the administration of simple standardised record forms could be improved. 27,28

Compliance (or more correctly, concordance) seems to be a major cause of apparent treatment failures and a recent study suggested that this was often due to a poor understanding of the chronic nature of the disease, a fear of topical corticosteroids and the belief that all atopic eczema is caused by a specific allergy. A survey in Nottingham has found that most mothers worry that topical steroids cause adverse effects, though many were not able to distinguish between weak and strong ones. 29

The National Eczema Society is the UK's self-help organisation for atopic eczema sufferers and people with other forms of eczema. It has a well-organised information service and national network of activities geared to help eczema sufferers and their families. Sources of alternative care for atopic eczema sufferers abound in the community ranging from the highly professional to elaborate expensive diagnostic and therapeutic measures of dubious value.

An impossible task?

It is unrealistic to attempt to summarise the entire ‘treatments of atopic eczema’ into a single Cochrane-style systematic review, as such a task would take years and cover several volumes. Atopic eczema is a complex disease with at least 40 different treatment approaches and specific questions that can be asked of each treatment group. What is more realistic is to produce a ‘sketch’ or ‘map’ of RCTs of atopic eczema, to quantitatively summarise a few areas of conflicting studies where possible, and to qualitatively review the others in a form that would be helpful to clinicians and patients. Such an approach could also act as ‘seed reviews’ for subsequent, more detailed Cochrane systematic reviews.

A question or data-driven review?

The very broad-ranging scoping nature of this review implies that it cannot be hypothesis-driven. Even in just one area of atopic eczema management such as dietary prevention, there are at least six separate systematic reviews that can be asked of the available data:

• Does maternal avoidance of certain potentially allergenic foods prevent atopic eczema and if so, by how much in offspring at high risk (i.e. family history of atopy) versus those at normal risk?

• Does dietary manipulation in pregnancy reduce the severity of atopic eczema in offspring?

• Does exposing infants to allergens at an early stage of their immune development help by making them tolerant to substances that they will inevitably encounter in later life?

• Does exclusive breastfeeding protect against atopic eczema?

• Does prolonged breastfeeding with supplementation protect against atopic eczema?

• Does the early introduction of solids bring on atopic eczema?

Trying to answer similar questions for each of the 40 or so interventions used for the treatment of atopic eczema would be impossible in one short report.

This review is therefore unashamedly a data-driven one. It is a review that aims to map out what has been done in terms of RCTs in atopic eczema to date and to reflect and comment on the coverage of already researched areas in relation to questions that are commonly asked by physicians and their patients.

The authors are aware that there is a danger that a data-driven review can serve to amplify and perpetuate current trends in evaluating minor differences between a profusion of similar pharmacological products. The authors have mitigated against this inevitable hazard by drawing attention to gaps that have not been addressed when summarising the reported studies, and also by including a comprehensive section on ‘unanswered questions’ in chapter 14 of this report, based on the views of contemporary researchers, physicians and patients.

Types of studies included in the review

Only RCTs of treatments for atopic eczema were included in the data summaries as other forms of evidence are associated with higher risks of bias. In order to be included as an RCT, a randomisation procedure was described, the study compared two or more treatments in human beings, and the study was prospective. In addition, the RCTs had to be concerned with therapeutic issues in relation to the prevention or treatment of atopic eczema. Thus, RCTs that involved evaluating cellular or biochemical responses of patients with atopic eczema after testing or injecting them with substances such as histamine were not included. Although they might inform future therapy, they were not therapeutic trials. Studies of possible increased incidence of drug adverse effects in atopic people compared with non-atopic people were also excluded. Studies also had to include at least one clinical outcome. Therefore, studies that only reported changes in blood tests or cellular mechanisms were excluded.

Study participants

Studies were included if participants were babies, children or adults who have atopic eczema (syn. atopic dermatitis) according to Hanifin and Rajka diagnostic criteria,4 or as diagnosed by a physician. Terms used to identify trial participants with definite, possible and definitely not atopic eczema are shown in Table 1. Those studies using terms in the ‘definitely not atopic eczema’ category such as allergic contact eczema were excluded. Those studies using terms in the ‘possible atopic eczema’ category, such as ‘childhood eczema’ were scrutinised by one of the authors and only included if the description of the participants clearly indicated atopic eczema (i.e. itching and flexural involvement).

Main outcome measures

Changes in patient-rated symptoms of atopic eczema such as itching (pruritus) or sleep loss were used where possible. Global severity as rated by patients or their physician were also sought. If these were not available, then global changes in composite rating scales using a published named scale, or where not possible, the author's modification of existing scales or new scales developed within the study were summarised. Adverse events were also included if reported. The selection of outcome measures was explored in more detail in a focus group of consumers held by one of the authors.

Secondary outcome measures

Secondary outcomes measures were changes in individual signs of atopic eczema as assessed by a physician, for example:

• erythema (redness)

• purulence (pus formation)

• excoriation (scratch marks)

• xerosis (skin dryness)

• scaling

• lichenification (thickening of the skin)

• fissuring (cracks)

• exudation (weeping serum from the skin surface)

• pustules (pus spots)

• papules (spots that protrude from the skin surface)

• vesicles (clear fluid or ‘water blisters’ in the skin)

• crusts (dried serum on skin surface)

• infiltration/oedema (swelling of the skin), and

• induration (a thickened feel to the skin).

Electronic searching

In order to retrieve all RCTs on atopic eczema treatments in accordance with inclusion criteria, a systematic and mainly electronic search strategy was carried out. The Cochrane Collaboration Handbook39 and the Centre for Reviews and Dissemination guidelines for systematic reviews37 were used as templates.

The following electronic databases have been searched:

• MEDLINE40 (1966 to end of 1999)

• EMBASE41 with its higher yield of non-English reports (1980 to end of 1999)

• The Cochrane Controlled Trials Register (CCTR)42 and

• The Cochrane Skin Group Specialised Trials Register. 43

Disease terms for atopic eczema (as a textword and MeSH term) are shown in appendix 1.

Possible trials were identified from each of the four databases by:

• MEDLINE (Index Medicus online): the Cochrane Collaboration ‘highly sensitive electronic search string’ for RCTs was used (appendix 1). Years 1966–December 1999 were searched and yielded over 3000 references using the disease search terms in appendix 1. An iterative approach was used with retrieved papers. Once trials on specific drug types were obtained, an additional MEDLINE search was carried out employing these specific drug terms (e.g. tacrolimus) or their developmental names (e.g. FK506) combined with a general skin search (appendix 1). References were checked for possible additional RCTs of atopic eczema. Review articles were also retrieved in hard copy form and references were checked for further RCTs.

• EMBASE (Excerpta Medica online): due to the different format of this database, an alternative search strategy was employed which was developed by the BMJ Publishing Group for its Clinical Evidence series (appendix 1). 44 Years 1980–December 1999, (the only years fully searchable on OVID), were searched. This yielded over 1000 references using the same eczema terms as for MEDLINE in Table 1. Trials that might have been on the EMBASE database from 1974 to 1979 would have been picked up by CCTR (see below), which has compiled its search of the entire EMBASE database since its inception in 1974.

• CCTR: the Cochrane Library, Issue 4, 1999 was searched for controlled trials within the CCTR section by exploding the disease-specific search terms separated by the boolean ‘AND’ with the advanced search option. These include clinical controlled trials (quasi randomisation) and RCTs (randomisation).

• Cochrane Skin Group Specialised Register: this was searched with the disease-specific terms and kind help of the Cochrane Skin Group Trials Search Coordinator.

Handsearching

As there are over 200 specialist dermatology journals and none specific to atopic eczema, separate handsearching was not done for this report. Some trials published in journals not listed in the main bibliographic databases or published within the body of a letter to the editor might therefore have been missed. However, results of handsearching of specialist dermatology journals by the Cochrane Skin Group are kept on the Cochrane Skin Group Specialised Register of trials, which was searched. This included results of handsearching the following dermatology journals as at July 2000:

• Acta Dermato-Venereologica Supplementum 1970–91

• Archives of Dermatology 1976–98

• British Journal of Dermatology 1991–97

• Clinical & Experimental Dermatology 1976–99

• Cutis 1967–99

• International Journal of Dermatology 1985–98

• Journal of Investigative Dermatology 1991–97

• Journal of the American Academy of Dermatology 1987–99.

In addition, conference proceedings of previous symposia such as the Atopic Dermatitis Symposium, held every 3 years (initially set up by Professor Georg Rajka), and all meeting abstracts for the annual meetings of the Society of Investigative Dermatology, European Academy of Dermatology and British Association of Dermatologists have been handsearched by one of the authors and the results made available to the Skin Group Specialised Register. Furthermore, one of the authors has been prospectively handsearching five dermatology journals (Clinical Experimental Dermatology, British Journal of Dermatology, Journal of the American Academy of Dermatology, Journal of Investigative Dermatology and Paediatric Dermatology) since January 1998, and any possible atopic eczema trials were accessed further by the team.

Other trial source

In addition to checking citations in retrieved RCTs and review articles, additional trials were sought by personal contact with atopic eczema researchers, and by writing to 37 pharmaceutical companies with a product or developing product in the area of atopic eczema.

Filtering

With the 3899 references yielded from the initial searches of MEDLINE, EMBASE and CCTR a filtering process began. This was carried out manually to assess whether the reference fitted the preliminary labels of ‘trial’ and ‘atopic eczema’.

Not all references had abstracts; therefore, ‘titles only’ had to be included as possible trials to avoid premature judgement. Where doubt existed from the abstract or title, the full paper was requested and scrutinised further by two of the authors. Papers labelled as ‘rejects’ were categorised with another label to specify why they were not suitable for inclusion. This was carried out by one reviewer and checked by a second reviewer in any cases of possible uncertainty.

All papers were catalogued on a specialised ProCite database. 45

Non-English studies

Studies published in non-English languages were screened by international colleagues (listed in the acknowledgement section) to see if they were possible RCTs with full data abstraction if this proved to be the case.

Study quality

Methodological quality of each study was assessed using a previously described scheme where the three potential sources of bias were evaluated,46 namely:

• the quality of the randomisation procedure

• the extent to which the primary analysis included all participants initially randomised (i.e. an intention-to-treat analysis)

• the extent to which those assessing the outcomes were aware of the treatments of those being assessed (blinding).

These three factors have been consistently shown to predict possible bias in effect estimates. 47 A descriptive component, rather than a score-based system, was used to quality rate the studies so that readers can see which aspects of the study reporting were deficient. Due to the sheer size of this scoping review, report authors were not blinded to the identity of the RCT authors when quality rating or data abstracting. Such blinding would have needed to be very thorough (to the point of having to conceal the interventions) as many of the RCTs are well known to one of the abstracting authors.

Quantitative data synthesis

Where pooling made sense clinically in terms of the interventions, study participants and common clinical outcomes, a meta-analysis was performed using both a fixed- and random-effects model depending on whether there was evidence of statistical heterogeneity. Odds ratios of improvement compared with the comparison intervention was used in the pooling exercises. The inverse of the variance of the outcome measures was used as weights for pooling the data from different trials. Sources of heterogeneity, such as differences in patients or formulation of interventions, were explored within the meta-analysis.

Atopic eczema prevention by allergen avoidance during pregnancy

Prevention of atopic eczema through allergen avoidance and dietary manipulation after birth

One Cochrane systematic review331 of maternal antigen avoidance during lactation pooled three studies53,61,332 and found some benefit on the prevention of atopic disease in offspring from maternal avoidance of allergenic foods while breastfeeding. Methodological shortcomings in all three trials (mainly loss of blinding) argue for caution in interpreting the results. RCTs that have examined the usefulness of manipulating the diet of lactating mothers and their infants after birth with a view to preventing atopic eczema56–61,63–67 are summarised in Table 4. A further trial332 has been excluded because no separate data on atopic eczema have been given. These studies share similar methodological concerns regarding disease definition and unmasking of blinding.

Other summary points are as follows.

• The Moore and colleagues63 study illustrates the difficulty of trying to randomise mothers to breastfeed.

• There is no evidence to support the use of soya milk as opposed to cows' milk supplementation to children as a means of preventing or delaying onset of atopic eczema.

• There is some evidence to support the use of extensively hydrolysed cows' milk formulae over regular cows' milk formulae in preventing atopic eczema in high-risk families, though the extent to which these unpalatable formulations can be taken up in practice is unclear.

• There is some evidence that maternal avoidance of allergenic foods during lactation may reduce the incidence of subsequent atopic eczema and other allergic disease.

• Some studies (e.g. Chandra et al. ,61 Marini et al. ,65 and Porch et al. ,57) mix up results of observational data with randomised participants in such a way as to render it difficult to make valid comparisons.

• Most of the studies refer to children born to families with atopic disease.

Results

Unfortunately few trials met our inclusion criteria and of the three that did, the methods of assessment were disparate. The response rate, defined as the proportion of patients who obtained at least a good response with treatment, allowed us to assess comparative efficacy. However, because of the disparate study designs, the estimates from the individual trials were not pooled.

Table 5 is a contingency table of the data from the three eligible trials. The estimated differences in response rates with once-daily versus more frequent applications, are also shown in Table 5 and Figure 3. It is clear that in none of the studies were more frequent applications superior to once-daily application. While the point estimates suggest that a small difference in favour of more frequent applications cannot be excluded, it is doubtful whether this is practically meaningful. This was also the conclusion of a recent review that considered this question in relation to all inflammatory dermatoses. 336

FIGURE 3.

Estimated risk difference for topical steroids

Koopmans and colleagues69 concluded that the proportion of patients who were cleared of eczema in the twice-daily group was higher than that in the once-daily group. Re-analysis of their data shows that this is the case, using doctors' assessment of clearance (rate difference –0.21; 95% confidence interval (CI) –0.36 to –0.06) but not the patients' assessments (rate difference 0.13; 95% CI –0.28 to 0.02). While the results of their trial do not exclude the possibility that Koopman and colleagues' claim may be true, theirs was the only trial that reported results in sufficient detail to test their claim.

Trials that evaluated oral steroids

Large treatment effects are observed in studies that compare systemic steroids versus placebo in terms of short-term disease response. We could not identify any longer-term studies evaluating the usefulness of oral prednisolone (continuous or intermittent) against other systemic agents such as cyclosporin in the treatment of severe atopic eczema.

Dilution of topical corticosteroids

This is widely practised in the UK on the basis that dilution might reduce adverse effects but maintain the same degree of efficacy. 338 We could find no RCT evidence to demonstrate this in patients with atopic eczema. The largest manufacturer of topical corticosteroids does not recommend dilution as there is a wide range of preparations currently available covering a wide range of potencies (Glaxo Wellcome, personal written communication, May 1998). Nevertheless, such a practice occurs with sufficient frequency to stimulate companies such as Glaxo Wellcome to provide ready diluted preparations such as Betnovate RD^™ 1:4 ointment of known stability, purity and compatibility. The possibility of compromising on stability, compatibility and microbiological purity following extemporaneous dilution of topical corticosteroids seems unnecessary with the current range of products of different strength. 339–341 It is recognised that widespread use of diluted topical corticosteroids reflects a perceived increased safety margin compared with undiluted products by prescribers and their patients, though such a belief may be a false reassurance unless there are clear data to say otherwise. Use of diluted products by prescribers may also indicate the need for a wider range of products towards the less potent end of the spectrum, particularly for the very young.

What is the most effective and safest way of using topical corticosteroids?

Atopic eczema is most commonly a condition that comes and goes, and constant use of topical corticosteroids is unnecessary and indeed undesirable in view of long-term local and systemic adverse effects. Thus, most practitioners probably use topical corticosteroids in short bursts followed by ‘holiday’ periods of emollients only. 342 Others may start with a potent preparation and then decrease to a lower potency preparation as the condition improves (the step-down approach). We could find no RCT evidence to throw light on these important issues. A pragmatic RCT set up to compare short bursts of a potent steroid versus longer-term use of a weak preparation is currently being conducted in Nottingham with support from Trent Regional R&D.

Benefits

Of the 23 evaluable patients after 4 weeks, infiltration, redness, skin thickening, scratch marks and dryness reduced by about 50% in both treatment groups. None of the differences were statistically significant. No statistically significant differences in parents preference for treatment was found. There was a statistically significant (p < 0.001) preference by patients or their parents for Clinitar cream compared with the crude coal tar cream, however, at the end of the study.

Harms

Four patients complained of stinging and itching, one on the site treated with Clinitar cream, two with coal tar cream and one with both. All were shown with patch testing to be an allergic reaction.

Comment

Although coal tar has been used for many years in atopic eczema, it is a pity not to discover any RCTs demonstrating its efficacy compared with vehicle alone. It is difficult to say from this study comparing two different preparations of coal tar whether both were vehicle effects or whether two active treatments were being compared. Both treatment groups improved equally during the study. Although cosmetic acceptability was higher for the Clinitar group, this was an unblinded assessment in patients who might have been eager to please the investigators. Method of randomisation was not described and no intention-to-treat analysis was performed. Further RCTs should compare the most cosmetically acceptable form of coal tar treatment versus vehicle alone and versus other common topical treatments such as topical steroids. There is a theoretical risk that tar is a possible carcinogen based on observational studies of occupational groups working with tar components.

Benefits

The Kantor and colleagues study154 compared the use of an oil-in-water emollient (Moisturel^™, not available in the UK) versus a water-in-oil emollient (Eucerin^™, Beiersdorf, UK) in 50 patients of all ages with symmetrical atopic eczema using a left/right comparison design for a period of 3 weeks. The study was split into two RCTs, which compared the two preparations in either a cream (Study 1) or lotion (Study 2). Test limbs affected by atopic eczema were treated once daily with the designated emollients and once daily with 2.5% hydrocortisone cream, and were assessed by an independent physician for redness, scaling/crusting, itching, burning/stinging and a global eczema severity on a scale of 0–3. Global severity reduced from 1.28 to 1.00 and 1.92 to 0.96 in the Eucerin and Moisturel cream groups. respectively (n = 25) and from 1.91 to 0.68 and 1.91 to 0.91 in the Eucerin and Moisturel lotion groups, respectively (n = 22). Although differences from baseline were statistically significant, there were no significant differences between the two emollients.

The Hanifin and colleagues study156 compared the effects of adding an emollient called Cetaphil^™ (manufactured by the study sponsor) applied three times daily to twice-daily application of 0.05% desonide lotion (a topical steroid) versus twice-daily topical desonide alone in 80 patients with atopic eczema for a 3-week period. The study was investigator-blinded only. Efficacy variables included seven symptoms and signs measured on a scale of 0–9, with a maximum score of 63. Global assessment by investigators was also recorded. At the end of 3 weeks, there was a 70% and 80% relative reduction in total score from a baseline of 24.23 and 24.13 for the desonide alone compared with the desonide/moisturiser side, respectively (p < 0.01). The sides treated by desonide alone showed complete clearing according to physician global evaluation in about 10% of people compared with 11% in the side treated with combination (data estimated from graph).

The Wilhelm and colleagues157 and Andersson and colleagues155 studies both evaluated the benefit of emollients containing urea preparations – a substance intended to improve water-binding capacity of the outer layer of skin. In the Wilhelm study, 80 patients with subacute atopic eczema and associated dry skin were randomised to apply a topical formulation containing 10% urea (manufactured by the study sponsors) versus the vehicle base as ‘placebo’ in a right/left forearm comparison for 4 weeks. Skin redness was improved in 70% of patients at the site of the 10% urea preparation compared with 30% on the vehicle site. Similar differences were noted for induration, but not for summary score. Measurement of outer skin moisture using a capacitance meter showed statistically increased hydration in the 10% urea group when compared with vehicle alone.

The Andersson and colleagues study155 compared a ‘new’ cream containing 5% urea as the active substance with an established licensed cream containing 4% urea and 4% sodium chloride in a parallel double-blind study of 48 adults with atopic eczema in Sweden. Patients were asked to apply the creams at least once daily for 30 days to dry, eczematous areas. Clinical disease severity measured by a physician on scale with a maximum score of 1600, showed a significant benefit for both creams, but with no statistical differences between the preparations. Actual data were not given and data were difficult to read from the figure as most data points were scrunched up towards the lower range of the large scale. Patient evaluation on a visual analogue scale (maximum of 14 meaning ‘no dry skin’) changed from 7.5 at baseline to 10 at Day 31 for those on the ‘new’ cream compared with a change from 7 at baseline to 9 at the end of treat-ment for the established cream (estimated from graph). There was no statistical difference between the two groups in terms of biometric measure-ments of water content or water loss through the outer layer of the skin.

Ammonium lactate is another substance thought to improve the water-binding capacity of the outer layer of the skin. This was the subject of Larregue and colleagues' study,153 comparing 6% ammonium lactate with its cream base only, in 46 children aged 6 months to 12 years with atopic dermatitis. The study was a within-person comparison of two symmetrical sites on patients recruited from France and Italy. Outcome measure included pruritus, and clinical objective measures included redness, dryness (xerosis), desquamation, lichenification, hyperkeratosis, and presence of papules. These were graded on a scale of 0–3. Intensity items were only partially reported in the results section, and suggested a reduction in lichenification, hyperkeratosis and dryness in both groups but slightly more in the ammonium lactate group. This was reported to be statistically significant at Day 15 for lichenification and for erythema at Day 30 (the final evaluation point of the study). Tolerance, as evaluated by the patients, was very similar in both groups. Results for itching symptoms were not given.

Harms

No adverse effects were reported in the first two studies with the exception of one patient in the Kantor and colleagues study154 experiencing a burning sensation when the oil-in-water emollient was applied. In the Hanifin and colleagues study,156 14% of the patients reported stinging or burning on the side treated with desonide compared with 12% on the side treated with a combination at week 1. Most patients (96% versus 4%) preferred the combination treatment. Transient burning was noticed in four and five patients when treated with urea and vehicle creams, respectively in the Wilhelm and Scholermann study. 157 No adverse effects were described in the Andersson and colleagues study. 155 Other adverse effects of emollients include an occlusion folliculitis on hair-bearing skin and accidents from slipping while climbing into the bath due to the use of emollient bath additives.

Comment

The first two studies were of extremely short duration and quality of reporting was generally poor with little description of randomisation method, limited blinding and no intention-to-treat analysis. The Kantor study failed to show any benefit of one emollient preparation over another (in the presence of a moderate potency topical steroid), and the Hanifin study suggested that regular use of an emollient with a topical steroid may result in a small increase in treatment response compared with a topical steroid alone. Very few physicians prescribe a course of topical steroids without an emollient, but it was good to see some RCT evidence that such a policy is justified. Neither study was designed to show a steroid-sparing effect for emollients.

The two studies on urea preparations showed a possible benefit of a urea-containing preparation compared with vehicle, and the other (which compared two preparations both containing urea in different concentrations) failed to show any benefit of a new preparation containing the higher concentration of urea. Quality of reporting on randomisation, blinding and intention-to-treat analysis was poor in both studies. Similar findings were present in the Larregue lactate study.

It is extremely disappointing, particularly in relation to the questions posed at the beginning of this section, to see a virtual absence of clinically useful RCT data on the use of emollients in atopic eczema. Many studies have been performed with emollients, but these have concentrated on ‘objective’ measurements such as transepidermal water loss and surface profilimetry. Despite their scientific ring, such measurements are often highly variable in terms of measurement reliability and more importantly, their clinical significance to a child with chronic atopic eczema is often completely unclear. In addition to measuring efficacy of emollients in treating mild atopic eczema lesions or the dry skin associated with atopic eczema, it is essential that future RCTs of emollients measure long-term tolerability, patient preferences and cosmetic acceptability as these are probably key determinants for successful long-term use.

The history of emollient use in atopic eczema is a good example of the inverse research law in dermatology, whereby the quality and quantity of evidence is inversely proportional to their frequency of use. 346 Emollients have become consecrated through usage and are firmly implanted in the treatment regimens of most European healthcare practitioners. Although emollients many have many beneficial actions,347 there is an urgent need to answer several basic questions about their use, preferably through industry-independent RCTs. Although other clinical and laboratory studies may continue to shed some light on the use of emollients in atopic eczema, RCTs are needed to address the unanswered clinical questions of most relevance to eczema sufferers and their carers. The top ten questions that need answering through RCTs with regard to use of emollients are as follows.

• Do some emollients have a useful therapeutic effect (with or without wet-wraps) for treating minor flares of atopic eczema when compared with other emollients or very mild topical steroids?

• Are some emollients effective in reducing itch and soreness associated with dry skin in atopic eczema compared with no emollients or other emollients?

• Do emollients have a topical steroid-sparing effect without loss of efficacy in the long-term management of atopic eczema?

• Does the regular use of emollients in between eczema flares treated by topical steroids help to reduce relapse rates?

• Does the use of emollients in children born to atopic parents reduce the incidence of allergen sensitisation and subsequent clinical disease?

• Does the regular use of emollients prevent painful cracks (fissures) on the hands of atopic eczema sufferers?

• Which emollients do children and adults prefer at different sites of their body?

• Does the pouring of an expensive emollient into a bath provide any additional benefit to having an ordinary bath and then applying an emollient directly to the skin afterwards?

• Does the use of regular emollients reduce the incidence and severity of secondary infection in atopic eczema?

• How common is clinically relevant sensitisation to emollient constituents such as lanolin?

It might be argued that because clinical opinion on the use of emollients is so deeply engrained that the position of clinical equipoise348 for testing emollients passed some 20 years ago. On the other hand, patients and the NHS spend a vast amount on emollients and trials would be justified on the basis of potential economic savings alone. It is understandable that it is not in the interests of the pharmaceutical industry to invest much into RCTs of the comparative efficacy of established emollients.

Benefits

Fourteen patients (mean age 16 years) with mild-to-moderate atopic eczema and symmetrical lesions were enrolled and three dropped out. One developed redness on both arms and subsequent patch testing confirmed a contact allergy to wool alcohol present in the ointment. At the end of 2 weeks, there was slight improvement for overall impression and global score compared with baseline and changes were virtually identical in the active and placebo group. None of the changes were statistically significant between the two groups.

Harms

No adverse effects were reported within this study apart from the contact allergy to wool alcohol. Patients given topical lithium should be monitored for lithium toxicity if large quantities are applied.

Comment

Although this was a very small study published in correspondence form only, the detailed results section is particularly informative as CIs around the treatment differences are presented. Because of the small standard deviations within patient differences, the study excluded moderate and large treatment effects. Method of randomisation and concealment of allocation and blinding was unclear and no intention-to-treat analysis was performed.

Benefits

In the adult study,161 215 patients with moderate-to-severe atopic eczema aged 13–60 years were randomised to a parallel group study of either 0.03%, 0.1% or 0.3% tacrolimus ointment or vehicle alone. No concurrent treatment except emollients were allowed and the study duration was 3 weeks. Compared with baseline, the percentage of patients with completely resolved or markedly improved skin lesions at the treatment site according to an overall assessment by a physician were 60%, 82%, 72% and 8% (data estimated from graphs) in the 0.03%, 0.1%, 0.3% tacrolimus ointment and vehicle groups, respectively. Results were similar for patients' overall assessments, though the data were not shown. Median percentage decrease in a summary score for redness, swelling and itch was markedly greater in the tacrolimus groups compared with vehicle (66%:83%) when compared with placebo (22.5%). All of the changes were statistically significant for comparisons of placebo versus the three tacrolimus concentrations, but there were no statistical differences between the three tacrolimus concentrations.

The tacrolimus ointment for children with atopic eczema study had the same four comparison groups as the adult study with 180 children of moderate-to-severe atopic eczema aged 7–16 years recruited from 18 study centres in North America. For the patients' (or parents') assessment of itching, the median percentage improvement from baseline to the end of the 3-week trial was 50.5% for vehicle, 88.7% for 0.03% tacrolimus, 73.6% for 0.1% tacrolimus, and 77.1% for 0.3% tacrolimus. Only the differences between vehicle and the different tacrolimus concentrations were statistically significant. Similar percentage improvements were seen for physician's global evaluation. The mean percentage improvement for a modified eczema area and severity index at the end of the treatment for each of the three tacrolimus groups (72% for the 0.03% preparation, 77% for the 0.1%, and 81% for the 0.3% preparation) was significantly better than the vehicle group (26%; p < 0.001).

Harms

There was a significantly higher sensation of burning at the site of application in the adult tacrolimus patients. Itching and redness were also reported in the adult study. Blood concentrations of tacrolimus remained very low throughout the adult study. No serious adverse effects were reported in the children's study, though again burning and increased itching at the site of application during the first 4 days of the study were reported in about a quarter of the patients. Only seven of the 254 blood samples evaluated contained more than 1 ng/ml of tacrolimus. To put this into context, increased toxicity in transplant patients has been associated with concentrations of greater than 20 ng/ml.

Another study of potential skin thinning of tacrolimus has been reported by Reitamo and colleagues,160 whereby the authors evaluated the potential of 0.1% tacrolimus ointment versus vehicle control versus betamethasone valerate (a potent topical corticosteroid) applied in a randomised order to apparently normal areas of skin on the abdomen under occlusion for 7 days. They found no evidence of skin thinning as measured by ultrasound in the tacrolimus group (compared with 8.8% median decrease in skin thickness relative to a vehicle control for betamethasone). They did not find any increase in chemical markers for collagen breakdown in the tacrolimus group whereas these were present in the steroid group.

Comment

Both of these RCTs were well reported with a good description of randomisation, blinding and both with intention-to-treat analyses. The flow diagram of the trial profile in the Boguniewicz and colleagues159 trial was particularly refreshing. Magnitude of the clinical benefit measured in a number of ways was moderately large in tacrolimus in both adults and children when compared with placebo, and little difference was observed between the different concentrations of tacrolimus. Although no more serious adverse effects have been reported to date, much larger numbers and experience with the drug is needed before topical tacrolimus can be declared safe for the widespread use of atopic eczema in the community. Both of the trials were of extremely short duration and longer-term trials are needed to evaluate the benefit of tacrolimus in managing the chronicity of atopic eczema. Both trials are placebo-controlled and therefore lack the crucial comparisons against short bursts of moderate-to-potent topical corticosteroids and also against the newly developed and very similar topical ascomycin derivatives. Such comparative trials should include cost-effectiveness, and they should also be pragmatic in order to capture the tolerability of the topical agents in the community, where transient burning may result in lower compliance than that in organised clinical trials. The lack of skin thinning is a potential advantage of tacrolimus, along with ascomycin, though the evidence that serious skin thinning occurs with correct use of topical corticosteroids in the community is lacking. Both of these RCTs were sponsored by the manufacturer, Fugisawa USA Inc.

Benefits

The van Leent and colleagues study of 1998162 randomised 34 patients with moderate atopic eczema to a double-blind placebo-controlled right/left comparison study. Topical 1% ascomycin cream was applied twice daily (n = 16) or once daily (n = 18) and compared with a corresponding placebo cream base for symmetrical lesions on the arm on one or other side. The trial duration was 21 days, and all participants were adults. At the end of 21 days, those in the twice-daily group showed a 71.9% decrease in baseline severity score compared with 10.3% in the placebo group (p < 0.001). Changes were also beneficial in the once-daily ascomycin group, but the magnitude was not so impressive with a 37.7% reduction in atopic dermatitis severity index in the active versus a 6.2% reduction of score in the placebo group (p = 0.002). Three out of 15 patients showed total clearance of their lesion in the twice-daily ascomycin group compared with none in the placebo group. None of the patients in the once-daily group (neither active nor placebo) showed any clearance of their lesions.

The other study has been published in abstract form only351 and preliminary results were shown at a recent satellite symposium sponsored by the manufacturers. The study randomised patients to one of four different concentrations of ascomycin, vehicle, or betamethasone 17-valerate cream. Results presented at the meeting suggested that the ascomycin cream was significantly more effective than the vehicle alone in a dose-dependent manner, but not as effective as the topical corticosteroid. Full publication of the study is awaited before further comments can be made.

Harms

None of the patients in the van Leent study were shown to have any skin irritation or other local adverse effects. Blood values of ascomycin remained low throughout the study. In another study published in abstract form only,352 ascomycin 1% cream did not cause any thinning of the skin in 16 healthy volunteers who applied the cream to their forearm for 4 weeks.

Comment

This was a well-reported study, though the method of randomisation and concealment of allocation in randomisation was not described. Blinding was well described and intention-to-treat analysis was performed. The study showed large treatment effects when presented in terms of relative treatment difference of a composite severity system when compared with placebo. The magnitude of the benefit was almost twice as large in the twice-daily group compared with the once-daily group, though the percentages are not directly comparable as they were not derived from the same randomised population. Although the atopic dermatitis severity index scores were unequivocally beneficial in the active group, it was disappointing not to see any patients' views on efficacy for pruritus incorporated in the main outcome measures. The small placebo effect of the cream is also unusual for atopic eczema trials, and possibly points to unblinding of the interventions. The fact that only three out of 16 patients showed total clearing of their eczema lesion on the arm after twice-daily continual application for 3 weeks is also slightly disappointing, bearing in mind the clearance potential of moderately potent topical steroids. Nevertheless, this study suggests that topical ascomycin derivatives are effective in moderately severe atopic eczema in adults, and is a welcome addition to the treatment modalities available. More comparative trials are needed, particularly against intermittent use of potent topical steroids and tacrolimus ointment in order to evaluate the place of ascomycin therapy in atopic eczema. Longer-term safety studies are also needed to ascertain rare but serious adverse effects from systemic absorption.

Comment

Results and discussion

A summary of the studies retrieved, their characteristics and the outcomes used are presented in Tables 10–17. The studies using oral SCG generally reported little or no beneficial effect for the drug compared with placebo. From those results it is probably safe to conclude that orally administered SCG is of little value in atopic eczema.

The results of trials of topical disodium cromoglycate (DSCG) are conflicting. Most of the studies that have reported positive effects were from the same laboratory84,194,201 and have used solutions of the drug rather than semi-solid formulations. While differences in product formulation may have accounted for the differences in observed responses, it is possible that the study populations may also have had an effect. One of the Kimata studies201 also included oxatomide in both the control and treatment arms so that treatment effects from his two studies cannot be quantitatively compared. The other two studies provide estimates which suggest that at 2 weeks, the itch scores decreased to a comparable extent with both DSCG and beclomethasone dipropionate (Table 18). The magnitude of change in itch scores obtained with those two drugs were somewhat higher than the difference seen between DSCG and placebo.

The study by Moore and colleagues198 also suggests that SCG was effective in atopic eczema with severity scores substantially reduced within 1 month of initiation of therapy. However, the individual symptom scores are not reported so that quantitative comparison with the Kimata studies is not possible.

Benefits

The study by Kemmett and Barnetson213 (abstract only) evaluated topical 4% nedocromil sodium cream versus matching placebo, in 32 atopic eczema patients over a 4-week period. There were no significant differences between treatment and control groups as determined by clinical assessment and IgE levels (no data given). Patient-assessed relief of itch, redness and weeping was recorded but no data were given.

The paper by van Bever and Stevens214 evaluated topical 4% nedocromil sodium cream versus vehicle only, in 26 adults and children with atopic eczema over a 4-week period. Patients and clinicians could not detect any difference between the two treatments as determined by daily score card for itch, sleep and overall severity of skin lesions and clinical examination for severity of skin lesions.

The study by Benton and colleagues215 evaluated oral nedocromil sodium 100 mg three times daily versus placebo, in 22 adults with moderate-to-severe atopic eczema over a 4-week period. Patient diary cards for itch, redness and weeping and clinician's overall opinion showed no significant differences between active treatment and placebo.

Harms

Benton and colleagues performed full blood counts and tests of renal and hepatic function each month to determine any drug toxicity. The authors report no abnormalities were found in the laboratory data, and the drug was well tolerated apart from one patient who developed persistent diarrhoea, which ceased on withdrawal of the drug. Van Bever and Stevens report 17 episodes of flaring of symptoms, nine were in the nedocromil sodium group. One other patient reported dryness of skin and another furunculosis.

Comment

All three studies were randomised but method and concealment of randomisation was unclear, all described as double blind. The Kemmett and colleagues study was in abstract form only so little information was available and no data were given for results. The van Bever and Stevens study did not specify whether daily score card was patientor doctor-assessed, and no actual data were given. It was unclear how many people were enrolled in the Benton and colleagues study. The results do not show any evidence to support benefit of nedocromil sodium, though the studies were relatively small and over short periods of time.

Benefits

We located two RCTs reporting the use of ketotifen for atopic eczema216,217 one on adults and one on children. The study in children217 evaluated ketotifen 1–2 mg twice daily versus placebo on 42 atopic children with asthma and allergic rhinitis (15 had eczema) for a period of 4 months. Parent-assessed diary cards of asthma symptom scores plus night itch, day itch and redness of skin were the primary outcome measures. No statistically significant beneficial effect of ketotifen was shown in asthma, allergic rhinitis or eczema.

Falk216 evaluated ketotifen 1 mg twice daily versus placebo in 60 adults with atopic eczema over a 3-month period. The eczema was assessed for itch, sleep loss, erythema, lichenification and overall efficacy of treatment. Improvement of itch over baseline on a scale of 1–3 was 2.40 reduced to 1.20 for ketotifen (p < 0.01) versus 2.30 reduced to 1.60 for placebo (p < 0.05).

Harms

Apart from slight drowsiness no other adverse effects were reported.

Comment

The White and colleagues study217 was primarily evaluating ketotifen for asthma; however, 15 of the children also had eczema. Being a small sub-group the authors conclude the number of patients with eczema was too small for meaningful analysis. The method and concealment of randomisation were unclear, though the study was described as double-blind. Withdrawals or drop-outs were not mentioned.

The Falk study216 showed no appropriate test of differences between the two treatments and no standard errors were given, therefore the results are difficult to interpret. The method and concealment of randomisation were unclear, though the study was described as double-blind. Four dropped out, and there was no intention-to-treat analysis.

Benefits

Four RCTs218–221 that evaluated topical doxepin in atopic eczema patients were identified. A further study of weal response in atopic eczema patients was excluded because no atopic eczema outcomes were reported. 363 One trial364 was excluded as it was only published in abstract form with few data. Of the remaining four RCTs, two evaluated topical doxepin versus vehicle cream218,219 and the other two evaluated the possible additional benefit of topical doxepin to treatment with topical corticosteroids. 220,221 Statistical pooling of the Breneman and colleagues219 and Drake and colleagues221 studies was not possible because separate data on atopic eczema were not given in the latter study. Pooling was not attempted in the Berberian and colleagues220 and Drake and colleagues221 studies because different strengths of topical triamcinolone were used.

The study by Drake and colleagues218 evaluated topical 5% doxepin cream versus vehicle only, applied four times daily, in 270 patients with atopic eczema over a 7-day period. Relief of itch (as recorded by a physician) was reported in 85% of doxepin and 57% of vehicle-treated patients by Day 7. A statistically significant relief in itch as recorded by patients on a 100 mm visual analogue scale (where 0 = no relief and 100 = complete relief) was also noted in the doxepin versus vehicle groups (68.6 versus 54.6, respectively after 7 days with baseline of 0 for both groups). Physician-reported eczema severity was also reported to be better in the doxepin group, though no data were given.

The paper by Breneman and colleagues219 is more difficult to asses as the RCT reported within this paper included a mixture of 47 patients with atopic eczema and 49 with lichen simplex (another form of localised eczema), and results were not presented separately. There was no clinically or statistically significant difference in patient-assessed itch relief at the end of the 7-day RCT.

The study by Berberian and colleagues220 evaluated the possible additional benefit of adding 5% doxepin to commonly used topical steroids in an 8-day study of 349 patients with atopic eczema. Four groups were randomly allocated to 2.5% hydrocortisone, 0.1% triamcinolone acetonide, 2.5% hydrocortisone plus 5% doxepin or 0.1% triamcinolone acetonide plus 5% doxepin, applied four times daily. At the end of 8 days, the mean visual analogue scale value for patient-recorded itch in the doxepin/ hydrocortisone group versus hydrocortisone group was 77.8 and 68.3, respectively (where 100 = complete relief from itching). For the doxepin/triamcinolone versus triamcinolone groups, the relief scores were 94.9 versus 90.5, respectively (p < 0.05). Baseline scores were not reported. Statistically significant effects were noted in the groups containing the doxepin from Day 1 onwards, but the magnitude of these effects diminished with each consecutive day. Physicians' global evaluation of eczema severity at the end of the 8 days was not clinically or statistically significantly different.

The Drake and colleagues study221 was mainly a pharmacokinetic study comparing doxepin hydrochloride 5% cream alone with doxepin plus 0.025% triamcinolone acetonide in 24 adults for 7 days. Only limited efficacy data were given. The paper stated that there were no significant differences in severity of atopic eczema between the two treatments at any point throughout the course of treatment. Pruritus severity scores (one of six itching assessment methods used in this study) demonstrated statistically significant greater improvement in the doxepin/triamcinolone group at 8 days (p = 0.001), though actual data for this and the other pruritus outcomes were not given.

Harms

Transient stinging or burning was commoner in doxepin-treated patients (e.g. 16 versus three in the Drake study). Drowsiness was also a problem (37 for doxepin versus three in the vehicle-treated patients in the Drake study), resulting in 16 dropouts from the doxepin arm versus three in the vehicle arm. Somnolence was also noted in four out of 22 participants in the Drake study. 221

Comment

The quality of reporting in the studies was quite good – methods of randomisation (though not subsequent concealment of allocation), a description of blinding and an intention-to-treat analysis were present in all three studies above. The fact that many patients developed drowsiness and stinging resulting in differential drop-outs in study arms raises concerns regarding the success of blinding in these studies. All four RCTs were sponsored by the manufacturer and were conducted by the same group of US investigators. Two of the above studies suggest that there is some evidence that topical doxepin produces some additional relief for itching in the short term (24–48 hours) when compared with vehicle. Whether this initial relief is clinically useful is doubtful, particularly when problems of drowsiness are taken into account. These short-term effects on itch become clinically insignificant over a 1-week period. None of the studies have demonstrated a clinically useful benefit of doxepin on atopic eczema severity or control even over the very short 1-week assessment periods. The studies have also used outcome measures scales that include itch as a prominent feature, or they have used several methods to assess itch, thereby increasing the likelihood of demonstrating a significant difference between the groups for itch. Longer-term independent studies (i.e. at least 6 months) evaluating combinations of doxepin plus other commonly used topical agents in atopic eczema are needed.

Benefits

In this right/left comparison study of 28 days, a lesion of atopic eczema was rated as responding if the sum of its rating (a composite scale of signs and itching) decreased by at least 33% from baseline to the end of treatment. Of 32 evaluable patients, 78% were noted to respond on the active drug versus 75% compared with a vehicle (p = 0.614). Median changes of efficacy parameters from baseline to end of treatment were also very similar between the two groups.

Harms

Treatment was well tolerated except for one patient who experienced burning at the site of drug application (site not specified).

Comment

This is one of the few RCTs in atopic eczema to pre-specify a measure of ‘success’ of its complex efficacy ratings scale. Method of randomisation was unclear and no intention-to-treat analysis was performed. Although this study was under-powered, the complete lack of difference between active and vehicle argues against a large treatment effect.

Comment

Benefits

Forty-one of the 48 children in this parallel-group RCT study were evaluable at 4 weeks. Median skin severity score increased from 92.3 at the beginning of the trial of the pyridoxine group to 109.0 at the end of the 4-week period. In the placebo group, the median skin severity score fell from 125.5 to 77.0 at the end of the treatment period. The difference between the median change in skin scores was 29.2 (95% CIs ranging from a benefit of pyridoxine of 19.5 to a benefit of placebo of +85.0). There was no statistical difference for skin severity score, daytime itch or nocturnal itch score. With regard to parental observation, 16% in both groups felt that overall the skin was better (p = 0.95).

Harms

No serious adverse effects were described in the study, though one child developed a non-specific erythematous rash while taking pyridoxine, and another taking placebo was reported to be much more itchy than usual.

Comment

This was a well-reported study with method of randomisation and allocation concealment and blinding clearly described. No intention-to-treat analysis was performed and no adjustment of the different baseline scores was made. Some of the significant improvement of the placebo group could partly be due to a regression to the mean phenomenon. The study did not provide any evidence to support any benefit of pyridoxine in the treatment of atopic eczema despite the earlier favourable report.

Benefits

In the Czeizel study, data were available on 4122 pregnancies that ended in a live birth. Of these, 2090 were randomised to the ‘active’ multivitamin supplementation, compared with 2032 in the trace element supplementation group. A whole range of postnatal development factors were collected in the study, including a physical examination and study of medical records in 90% of the evaluated infants. At the end of the 17-month period, there were no significant differences in the occurrence of chronic diseases between the two groups, with the exception of atopic dermatitis and wheezy bronchitis. Fifteen out of 2090 receiving multivitamin supplementation had developed atopic eczema (four had a parent with atopic dermatitis) compared with four out of 2032 receiving trace element supplementation (none of these children's parents had atopic disease). The authors suggested that these unexpected findings may be a chance effect.

In the Fairris and colleagues study,257 60 adults with atopic eczema were randomised in a 12-week double-blind study to three groups taking either 600 μg of selenium alone, 600 μg of selenium plus 600 IU of vitamin E or a placebo. Using a severity assessment based on several skin signs at several body sites, mean severity score fell from 21.0 to 13.7 in the selenium only group, from 21.8 at baseline to 15.3 in the selenium plus vitamin E group, and from 20.4 to 14.5 in the placebo group. None of these differences were statistically significant. There was, however, a significant increase in concentration of selenium in whole blood of those taking selenium.

In the Hakakawa and Ogino study,258 59 participants with mild-to-moderate atopic eczema of the dry type were randomised to vitamin E (d-α-tocopherol) 100 mg plus vitamin B₂ (riboflavin butyrate 20 mg), or vitamin E 100 mg or vitamin B₂ alone for 4 weeks. Of the 49 evaluable participants, response as measured by physician-assessed overall usefulness and global rating, was greater in the combination vitamin group than in the single vitamin group.

Harms

The Hungarian study,256 in a sense, has detected a possible unwanted harm of an increase of allergic diseases in those given multivitamin supplements in early pregnancy. No adverse effects were reported in the studies of selenium and vitamin supplementation.

Comment

Although the Hungarian study is not really a therapeutic trial of an intervention in atopic eczema, it is nevertheless a large RCT that has found a statistical significant increase in atopic eczema and in wheezy bronchitis and asthma in those receiving multivitamin supplementation compared with trace element supplementation. The numbers of patients in these groups are very small and the results therefore may probably be due to chance. Nevertheless, these findings should be examined in other independent randomised birth cohort studies.

The study of selenium and vitamin E,257 while small, probably excluded moderate-to-large treatment effects. A method of randomisation was described (unusual for such an early study), though no intention-to-treat analysis was performed.

The short Japanese study258 on the use of combination of vitamin E and B₂ is difficult to interpret in the absence of a placebo-controlled study of either compound. The validity of the study is also threatened by difficulties in blinding and post hoc subgroup analysis of dry skin subtypes at different time intervals.

Benefits

At the end of the 8-week period, the mean combined disease severity score increased from 36.1 at baseline to 48.7 in the zinc group compared with a change from 34.6 at baseline to 39.3 in the placebo group. Erythema score, surface area score and use of steroids, emollients and antihistamines were also almost identical in both groups and none were statistically significantly different.

Harms

No adverse effects were reported in this small RCT.

Comment

Despite the usual reservations on lack of description of the randomisation process and no intention-to-treat analysis, this small RCT illustrates the importance of an unbiased comparison of possible treatment benefits when compared with earlier enthusiastic claims of benefit. Even though the trial was quite small, the complete lack of treatment benefit along with small standard errors, argues against missing any moderate-to-large treatment effects.

Benefits

The first small RCT by Colloff and colleagues262 evaluated the daily use of natamycin (a spray used to kill house dust mites) versus matched placebo spray with and without vacuum cleaning in a parallel group study for 4 months in 20 young adults with atopic eczema. They demonstrated that it was the vacuum cleaning and not the natamycin spray that had a significant impact on reducing house dust mite numbers. There was no significant clinical improvement in those who had been allocated to natamycin versus placebo. The mean symptom score (maximum score 288) in the natamycin and vacuum group changed from 55.2 at baseline to 38.6 at 4 months compared with 45.2 to 35.8 in the group with no natamycin and no daily vacuuming.

A second small, but important RCT was conducted by Tan and colleagues263 in 1996 with duplicated publication in 1998 and again in 1999. Tan and colleagues randomised 60 patients (30 adults and 30 children) for a total of 6 months to an intensive dust mite eradication regimen comprising Gore-Tex^® (Intervent, UK) bedding covers, benzyltannate spray to kill mites and denature their allergens, and a high filtration vacuum cleaner, or to a control group of plain cotton bedcovers, placebo spray and a standard upright vacuum cleaner with a poor filtration performance. One trained nurse applied the bedcovers and spray each week, and participants were encouraged to vacuum bedrooms daily. They showed a dramatic and very similar reduction in concentration of house dust mite major allergen (Der p1) in bedroom carpets in both the active and placebo treatment groups at the end of 6 months. Disease activity, as recorded in terms of surface area involvement and a composite severity score (maximum score 108) measured at one point at the end of the 6 months, reduced by a small amount in both groups, but more so in the active group. The mean reduction in scores for the active and placebo groups were 12.6 and 4.2 units, respectively. Those in the active treatment group were more severe to begin with, and so an analysis of covariance was conducted to allow for baseline scores and initial house dust mite antigen levels. This showed a mean difference of 4.2 in change of score (95% CI 1.7 to 6.7 units; p = 0.008) between the two treatments. Further analysis also suggested that it was changes in the mattress and carpet dust in the bedroom that mediated much of the treatment effect. Subgroup analysis suggested that only children had a clinically and statistically significant benefit, and that there was no correlation between clinical improvement and positive skin prick tests at the study outset.

Another small study in Japan by Endo and colleagues261 evaluated the potential benefit of intensive vacuum cleaning in the rooms of 30 children with atopic eczema for a total of 12 months. Both groups were visited every 3 weeks by a team of mite specialists who either cleaned room floors, mattresses and quilts very thoroughly and encouraged parents to clean in the same way in-between visits, versus a less intensive clean (vacuum suction power reduced to 50%) with similar cleaning in-between visits. Parents were thus unblinded to the intervention. A statistically significant decrease in mite numbers in favour of the intensive cleaning group was only noted for the room floors. Clinical scores, as evaluated by a physician blind to treatment allocation, were significantly improved in the active group compared with baseline but not in the control group. Clinical scores were given in graphical form only and the appropriate statistical test of mean difference between the two treatments was not reported.

Another unblinded RCT by Nishioka and colleagues260 evaluated the benefit of encasing quilts and mattresses in microfine fibres versus simple cleaning measures alone in 57 Japanese infants with atopic eczema who were not allergic to house dust mite as determined by blood tests at the study outset. After 1 year, they found that 31% of children in the encasement group compared with 63% in the control group had serological evidence of house dust mite sensitivity (p < 0.02). The authors did not report any outcomes on atopic eczema disease activity in that paper. Correspondence with the authors suggests that there were no differences between the two groups at the end of the study for the clinical outcomes, though there was a reduction in topical corticosteroid requirement in the intervention group.

Harms

None of the studies reported any adverse events of the anti-house dust mite treatments. This does not necessarily imply that none occurred. The imposition of daily vacuuming for a long period has a cost in terms of time for parents and sufferers, as does the purchase of a high filtration vacuum cleaner, impermeable mattress covers and mite sprays.

Comment

It is a pity that so few studies on house dust mite avoidance have been performed on atopic eczema. Those that have been done tend to be small and difficult to generalise in the absence of more pragmatic studies. In none of the studies was the method of randomisation and concealment reported, and no intention-to-treat analyses were performed (though drop-outs were quite low). The validity of blinding in the Tan study in unclear due to the use of very different vacuum cleaners. It is also unclear which method of reducing house dust mites is the most efficient as studies have tended to use several measures at once. Data from the Tan and colleagues263 study suggest that just vacuuming with an ordinary household cleaner achieves similar and massive reductions of house dust mite antigen levels in bedroom carpets to those achieved by the more expensive high filtration vacuum cleaners that are advertised in patient support group magazines. The Tan and colleagues study suggests a definite benefit for a range of intensive measures to reduce house dust mite levels around the home, but the clinical relevance of the small changes in scores observed is difficult to determine in the absence of patients' evaluations or outcome measures that capture the chronicity of disease over the entire 6-month period. The Endo and colleagues261 study suggested some clinical benefit with frequent intensive vacuum cleaning, but this could not be related to reduction in house dust mite numbers. On the basis of the Tan study, the Endo study could be criticised for comparing two active treatments, and that a comparison of frequent vacuuming versus ‘normal cleaning’ would have been better given the fact that any form of active intervention for intensive or high-filtration vacuuming is difficult to blind.

Both Tan and Endo are to be commended for conducting such long-term studies. Further such studies in other populations, separating the different interventions for reducing house dust mite, are needed. It is important that such trials are as pragmatic as possible to determine which groups respond best, which interventions are the most cost-effective and whether the laborious interventions are sustainable in less-motivated people.

Benefit

The first small study by Glover and Atherton264 evaluated the use of a tyrosine-adsorbed extract of house dust mite in 26 children with atopic eczema who were house dust mite positive on skin-prick testing. In the first part of the study, children were randomly allocated to weekly active or placebo injections for 8 months. Clinical scores improved dramatically in both groups, but there were no obvious differences between the groups. In the second part of the study, the 13 children who had been allocated active treatment in the first part of the study were offered a further 6-month period of monthly injections. The seven who accepted were then randomly allocated to receive active treatment or placebo. Redness and skin thickening scores (but not surface damage) deteriorated more in the control group and these differences were statistically significant.

Another study by Galli and colleagues265 looked at the possible benefit of a mixture of house dust mite allergens given in an oral suspension three times weekly in children with atopic eczema who were sensitised to house dust mite. Three groups were compared, one non-randomised group of children with concurrent asthma and or rhinitis (n = 26), and two groups with exclusive atopic eczema who were randomised to oral hyposensitisation (n = 16) or no specific treatment other than ‘conventional therapy’ and measures to reduce house dust mite (n = 18). Comparison of change in clinical scores between the two randomised groups did not reveal a statistically significant or clinically relevant improvement in the active group.

A further study by Wen and colleagues266 was conducted in Shanghai with allergenic extracts of house dust mite manufactured at that university. In that study, 56 patients with atopic eczema (mean age 24.8 years) were randomly allocated to weekly injections of local allergenic extract (18 patients), a partially purified extract (20 patients) and normal saline placebo (18 patients) for 1 year. Clinical scores (unspecified) were reduced in all three groups and possibly more so in the two active groups, though no statistical tests were performed. The data was presented in graphical form only.

Harms

Apart from local discomfort at injection sites (similar in each group), no adverse effects were reported in either small study. Caution is needed based on the rare but potentially life-threatening hazard of an anaphylactic reaction when desensitisation has been done with bee sting or hay fever allergy. Wen and colleagues state in their paper that allergenic extracts of house dust mite have been used for 20 years for treating and diagnosing mite allergy in China and that there have been no recorded deaths due to anaphylaxis. Weekly injections for children are also painful and require attendance at a healthcare facility.

Comment

It is possible that the lack of statistical significance in the Glover study was due to lack of power or a large placebo effect due to the injections. Improved treatment concordance and ancillary care could also explain the impressive improvements in those having placebo injections. Similarly, the lack of obvious benefit in the Galli study could be due to inadequate numbers, other concurrent treatments such as reduction in house dust mite measures, or that the oral allergen hyposensitisation therapy was inactive via the gut route. It is difficult to make any further judgement on the Wen paper due to the scant methodological details provided.

Benefits

After a 1-month washout using their normal detergent, 26 adults with mild-to-moderate atopic eczema (mean age 25 years) were randomised in a double-blind crossover study to receive either a trial detergent containing enzyme concentrations reflecting the highest quantity in commercial enzyme-enriched detergents or a visually identical detergent without enzymes as control, for a 1-month period followed by a further month with the opposite detergent. Topical steroids were permitted during the study and weighed. In the 25 patients completing the trial, there was no hint of difference between the active detergent and the control in terms of SCORAD score (29 on active, 29 on control, with 95% CIs for the mean difference extending from –4 to +5 on a scale of 108), usage of topical steroid (44 g/month in active versus 43 g/month on control), patient-reported itch (1.3 versus 1.3), or patient-reported eczema activity (1.4 versus 1.4).

Harms

None of the patients had contact dermatitis to enzymes when patch tested at the end of the study, and there was no evidence of specific blood IgE against any of the enzymes.

Comment

Although this study was small, the virtual absence of any differences between the enzyme and nonenzyme detergents and the corresponding narrow CIs provide convincing evidence of a lack of harmful effect. The study was not sponsored by industry.

Benefits

In the Diepgen 1990 study,269 55 patients with atopic eczema were compared with 31 control patients without atopic eczema and randomised to one of four poncho-like shirts of varying fibre roughness. The intensity of itching or discomfort due to repeated wearing of these shirts was evaluated by means of a points system, whereby 10 equals a maximum comfort and 1 equals maximum discomfort. At the end of the study, those wearing cotton reported a comfort score of 8.4 compared with 7.3, 3.6 and 3.3 for the other textile shirts in increasing order of weight and fibre roughness (estimated from graph). The difference between the cotton and other fibres was significant only for the latter two groups.

The 1995 study by Diepgen and colleagues270 (published in a German textile journal) evaluated seven different garments on 20 atopic eczema patients with mild-to-moderate disease (average age 25.3 years) with and without a ‘sweat test’ designed to lower the itch threshold. The garments included cotton, and polyester garments made with different fibre roughness, yarn roughness and fabric weaves. The study was a randomised crossover study (Diepgen T, personal oral communication, January 2000), with each garment worn under standardised conditions on 4 consecutive days. Comfort, as assessed on a visual analogue scale, was statistically significantly higher for warp-knits compared with jersey knits, but no different for cotton and polyester of fine fibre construction (assessed by scanning electron microscopy). Garment comfort in all groups was reduced after sweating.

In the Seymour study,271 cloth nappies were compared with cellulose core nappies versus cellulose core nappies containing absorbent gelling material in 85 babies aged less than 20 months of age who had atopic eczema and who were recruited by an advertising campaign. Average grade of eczema on the body as well as degree of nappy rash was scored by an independent dermatologist. At the end of the 26-week period, there was no clinical or statistical difference between the different nappy types for overall grade of atopic eczema. Nappy rash, however, was significantly less in the group using cellulose nappies with absorbent gelling material, compared with the others at 26 weeks and throughout the trial (p < 0.05).

Harms

No adverse effects are reported in these small studies, though specialised cotton clothing for atopic eczema sufferers is more expensive than other synthetic fibres.

No specific adverse effects were reported in the trial of different nappies, though nappy rash itself (the main efficacy outcome of this study) could be considered an adverse effect, which is desirable to prevent.

Comment

The studies by Diepgen and colleagues in 1990 and 1995 tested the hypothesis that cotton clothing is best for atopic eczema sufferers. The success of blinding in both trials is under question in view of the different roughness of the various shirt fibres. Magnitude of effects were not stated in the Diepgen 1995 paper, and it is possible that the study could have missed small differences between cotton and polyester fabrics comfort. The purported need for specialised clothing can result in considerable increased economic burden to eczema families and to the State. The two RCTs both suggest that there is nothing special about cotton for atopic eczema sufferers apart from smooth fibres. Other synthetic fibres can be constructed with similar smooth fibres using yarns and fabric construction that is just as comfortable for atopic eczema sufferers. It would be wise to repeat such studies in the UK and elsewhere before implementing policy decisions, and public knowledge of the availability and cost of cotton alternatives would be an advantage to eczema sufferers.

It was good also to locate an RCT evaluating different nappy types in atopic eczema. It was unclear in this study if the group with atopic eczema simply wearing a cloth nappy were randomised in the same way as the other two groups and whether statistical comparisons were made to the control population who were not part of the same randomised group. The study, nevertheless, suggests that nappy rash is less severe in atopic infants who wear nappies with gel absorbent material, though there appears to be no benefit for the general eczema control elsewhere. There was no evidence to support any benefit of conventional disposable nappies over cloth nappies, though the study may have lacked power to demonstrate small differences.

Benefits

One-hundred patients with mild-to-severe atopic eczema aged 15 years and over were randomised to either deep-sea water or physiological saline. Both waters were sterilised and heated to 65 degrees centigrade then sprayed onto the body before home bathing and washed away after 10 minutes. Treatment was daily for 1 week. Doctor's global evaluation and several other skin signs reduced by only a small amount (∼15%) in each group after 1 week, and there were no clinical or statistical differences in the change in scores between each group.

Harms

No adverse effects were reported in this study. There were five drop-outs for reasons not related to the treatment.

Comment

Quality of reporting was good in the Adachi study, but they compared two types of salt that could have been equally active. We agree with the author's conclusions that possibly a longer contact with sea water is necessary. The study duration was also very short (1 week). The uptake of an intervention that consists of spraying each other with a concentrated salt solution is also likely to be limited in the UK and possibly quite costly to patients. It is a pity that a simple RCT comparing salt baths versus ordinary baths has not been done. Even an RCT comparing the effect of regular versus irregular bathing would be informative given the different strongly held views of frequency of bathing recommendations in children with atopic eczema. One recent non-randomised study388 from Japan has compared the benefit of sea water therapy at a beach with and without dolphins to encourage children with severe atopic eczema to enjoy the sea water. This needs to be tested further in an RCT.

Benefits

Fifty consecutive patients aged 4 months to 6 years 2 months, with atopic eczema of varying severity were randomised to either conventional treatment by a dermatologist or the same treatment plus a single session by a nurse (‘a nurse lesson’), which included general information about atopic eczema and environmental control, information and demonstration of topical treatment and also a discussion of realistic expectations. The study lasted for 3 months and was unblinded. At the end of the 3-month evaluation period, mean eczema score (maximum score 96) had fallen from 26.4 at baseline to 7.1 in the group given standard dermatological care plus education compared with 21.3 at baseline to 10.8 for conventional care alone (p < 0.05). This comparison was not adjusted for baseline scores, which were different. Each score also showed a statistically significant reduction in favour of the education group compared with the standard dermatological care, and hydrocortisone consumption was significantly greater in the intervention group.

Harms

No adverse effects were reported in this study. Those patients in the intervention group had to attend one additional session with the nurse.

Comment

This RCT suggests a modest benefit from a single nurse education session following standard dermatological care. The authors are to be applauded for randomising the treatment groups, though the lack of blinding and failure to perform an intention-to- treat analysis limits the study quality. The study nevertheless suggests a modest benefit to a single session of nurse education, though the component of the ‘package’ of the nurse education that conferred the most benefit is unclear. It is possible for instance that most of the benefit could be due to increased and appropriate use of hydrocortisone, which tends to be underused in the community because of inappropriate fear of adverse effects. Further RCTs in other countries that use a similar educational package and blinded outcome measures are needed.

Benefits

Thirty-six children with atopic dermatitis admitted as inpatients to a high-altitude specialist treatment centre for atopic eczema in Davo, Switzerland were randomised according to sex, age and severity of disease to receive sham (placebo) or active treatment with the bioresonance apparatus. The bioresonance was conducted exactly as described in the specific literature and by a qualified BIT therapist for at least 4 weeks. Blinding was obtained by a specially designed switchbox operated by an engineer who kept its randomisation code outside the clinic in a sealed envelope. Patients were allowed their normal treatment with creams, emollients and dietary restrictions as required throughout the study. For the short-term outcome (at least 4 weeks), total disease severity score in the active group had fallen from 39.8 to 27.3 at the end of the study period compared with a fall from 35.3 to 26.6 at the end of the sham treatment (p = 0.23). No difference was observed for the sleep score, though pruritus score was slightly improved in the actively treated group (p = 0.12). There were no clinically or statistically significant differences between the two treatment groups in a number of immunological markers in the blood and long-term outcomes measured 8 months after the treatment.

Harms

No adverse effects were described in the study, though the therapy can attract a considerable financial cost outside of health services.

Comment

This study was very carefully reported and authors were meticulous in giving the bioresonance a ‘fair test’ by following the intervention as meticulously as possibly. The study was blinded, and randomisation and concealment of allocation were well described, though an intention-to-treat analysis was not performed (4 out of 36 children dropped out). The results do not show any evidence to support benefit of bioresonance therapy. Although the study was relatively small, the study was powered to exclude a 35% benefit of relevant treatment benefit of bioresonance therapy in the sensitive COSTA scoring method. Therefore, although small benefits cannot be excluded by the study, the study failed to show evidence of any moderate or large treatment effects of this mode of treatment. It may be argued that the treatment was tried in a situation where marked treatment benefits were already occurring as a result of inpatient stay at high altitude (with very low house dust mite levels), and ideally a similar RCT could be performed in a more usual outpatient setting.

Benefit

In the Melin and colleagues study,276 17 patients with atopic eczema aged 19–41 years were randomised into two groups. One group was treated with hydrocortisone cream alone and the other group was treated with the cream plus two sessions of habit-reversal treatment during the first week of the treatment period of 28 days. The study was unblinded. At the end of the assessment period, there was a 67% mean reduction in global eczema score in the habit-reversal plus hydrocortisone cream group compared with 37% score reduction in the hydrocortisone only group (p < 0.05). Total score of self-assessed annoyance was also markedly reduced in the active versus comparator groups. Mean percentage reduction of scratching episodes per day was 79% in the habit-reversal and hydrocortisone group compared with 49% in the hydrocortisone only group (p < 0.01). In the later study conducted by the same team,106 45 patients (mean age 24.8 years) were randomised in a parallel fashion to four groups for a period of 5 weeks:

• application of hydrocortisone cream for the entire 5-week period

• application of betamethasone valerate (a strong topical steroid) for 3 weeks followed by hydrocortisone for the remaining 2 weeks

• application of hydrocortisone plus habit-reversal for the 5-week period

• application of betamethasone plus habit-reversal for the first 3 weeks followed by hydrocortisone plus habit-reversal for the remaining 2 weeks.

The study was unblinded. Results are reported more fully in the section on topical corticosteroids (see chapter 4 and appendix 3). The authors reported significant differences between the behaviour therapy groups and those taking steroids alone for total skin status. Scratching was reduced by 65% in the hydrocortisone only group, 74% in the betnovate followed by hydrocortisone group, 88% in the hydrocortisone plus habit-reversal group, and 90% in the betnovate and hydrocortisone and habit-reversal groups (statistics not presented).

The study by Ehlers and colleagues in 1995275 evaluated the use of an autogenic training as a form of relaxation therapy (ATP) versus a cognitive-behavioural treatment (BT), versus a standard dermatological educational programme (DE) versus combined DE and BT (DEBT). One hundred and thirteen patients attending an outpatient clinic in Germany were randomised to these four groups and were also compared with an additional standard medical treatment group who were not part of the random assignment. Investigators were blinded as to the group allocation. The intervention was for 3 months and patients were followed-up for 1 year in order to evaluate disease relapse. At the end of 1 year, mean skin severity lesion score dropped from 29.5 to 28.8 in the DE group, 33.7 to 19.8 in the ATP group, 31.0 to 20.7 in the BT group, and 35.4 to 25.8 in the DEBT group. There were no significant differences in mean severity of itching between the four randomised groups. DEBT led to significantly larger improvement in global skin severity than DE alone and this was also accompanied by significant reductions in topical steroid use.

Harms

In the Ehlers study, the behavioural approaches required 12 weekly group sessions of 1.5 to 2 hours each with a group size of between five and seven patients. No adverse events were reported in any of the trials, though some of the drop-outs could possibly be related to the fact that extra visits were needed for the behavioural technique.

Comment

The Ehlers study275 was clearly reported and included an assessment of patient expectation of treatment benefit. No intention-to-treat analysis was performed, but drop-outs were low (nine out of 113 at 3 months). Over 14 outcome measures were reported in the study, which introduces the possibility of multiple hypothesis testing. The authors also performed statistical tests in comparison to a non-randomised control group, which may not be justified. Nevertheless, the magnitude of improvement for those receiving behavioural techniques in addition to their standard dermatological care (which included topical corticosteroids) was moderately large, and carried more weight than the Melin and Noren studies because assessments were made by an investigator blinded to the treatment allocation. The combination of habit-reversal plus judicious use of topical corticosteroids seems an attractive one and evidence from two RCTs supports its use. Both RCTs were unblinded, and conducted in the hands of enthusiasts. The generalisability of these findings to other populations should be determined in further RCTs, using objective assessment methods that are clinically meaningful, and conducted by investigators blinded to group allocation. The magnitude of the benefits in these unblinded studies were considerable, particularly when these were above that which could be expected with topical corticosteroids.

Comment

Generally the studies were quite small and poorly reported. Blinding was likely to have become unmasked in placebo-controlled trials due to the obvious tanning on one half of the body, along with mild burning and marked treatment effects. Although the right/left body comparison design had its limitations in terms of blinding, the lack of effect on placebo-treated body halves argues against a systemic effect of ultraviolet light treatment. Treatment effects were generally large and of rapid onset (i.e. within 1–2 weeks). Future studies should consider using a simple parallel group design and they should be of longer duration in order to capture duration of remissions.

Harms

Treatment with ultraviolet light usually ties patients to twice- or thrice-weekly visits to hospitals. Mild degrees of skin redness and burning are also common short-term adverse effects. Occasionally more severe burning may occur. There is no direct information on the long-term risk of skin cancer in atopic eczema patients undergoing ultraviolet light treatment. Although data on cohorts of psoriasis patients undergoing PUVA suggest that cancer risk only increases after around 250 treatment sessions, extrapolating from these studies to another inflammatory disease with a generally younger patient population raises some concerns. This is particularly so for melanoma skin cancer where it is thought that most risk is acquired from ultraviolet light in the first 20 years of life. Specific cohort studies of different modalities of ultraviolet light treatment of atopic eczema treatment are recommended.

Benefits

Of the 23 evaluable patients at 4 months, there was a statistically significant improvement in disease intensity index in those receiving active injections. Disease intensity was measured on a complex scale, which multiplied extent versus six physical signs. This reduced from 1000 to 612 in the active and from 1000 to 859 in the placebo group. Percentage mean reduction in itching was not so large, reducing from 3.3 to 2.2 in the active group and from 3.3 to 2.6 in the placebo group.

Harms

Three patients in the active group developed a delayed-type inflammatory action at the injection site and itching increased within 24 hours after injection in four patients on active and on two patients receiving placebo therapy.

Comment

Method of randomisation and concealment of allocation was unclear in this study, though blinding was probably successful. The ‘beneficial’ treatment effects are difficult to interpret given the use of an invalidated exploded scale and absence of a patient's perspective. The modest benefit demonstrated in this study needs to be replicated elsewhere before the intervention can be recommended as a treatment option.

Statistical methods

Rate differences were estimated for categorical variables and differences in means for continuous variables. The Mantel–Haenszel type method of Greenland and Robins399 was used to estimate the pooled rate difference for all strata under the assumption of a fixed-effects model. A CI for the pooled risk difference was calculated using the Greenland–Robins variance formula. 399 The Q (‘combinability’) statistic is given with its associated probability on k (number of strata) minus one degree of freedom. This has low power as a strict test of homogeneity. There are no comprehensive rules on when to use random effects and when to use fixed effects models; debate continues in the statistical community. 400 However, when this was significant at the 0.05 level, a random effects analysis was applied. 401 The pooled mean effect estimate was calculated using weights calculated as the inverse of the variance for each study.

In the trials, estimates of error were not reported in a consistent manner. When standard errors were reported for both the treatment and control arms, both were used in the calculations. When only the pooled estimate was given, the variances for the treatment and control arms were assumed to be the same. When quantitative values were reported in graphical form only, these were read electronically from the scanned images and the responses and their associated variances were estimated by linear interpolation as described by Poolsup and colleagues. 402

Results

Of the twelve eligible reports of RCTs (Tables 31–36), two dealt with topical cyclosporin294,297 and ten with oral cyclosporin. 286–293,295,296

Topical application

The two reports294,297 on the use of topical cyclosporin gave conflicting results. While in a study of 20 patients, De Prost and colleagues297 reported positive results based on assessments of pruritus, erythema, oozing, crusts, xerosis and lichenification, De Rie and colleagues294 reported no benefit in a study of seven patients with atopic eczema. Both studies used an intrasubject design with randomised applications to comparable lesions. Neither provided any justification for sample size chosen. On the basis of the available evidence, the efficacy of cyclosporin in atopic eczema has yet to be rigorously evaluated.

Oral therapy

Of the ten reports of RCTs of oral cyclosporin therapy, those by Zonneveld and colleagues288 and Zurbriggen and colleagues289 were comparisons of dose schedule and formulations, respectively, and therefore provided little evidence on the value of cyclosporin. One of the trials291 adopted a randomised parallel group design with 23 patients receiving cyclosporin and 23 placebo. The study showed that both disease severity and area involvement were reduced by cyclosporin treatment (5 mg/kg/day) by week 6. At the end of the 6-week trial, 15 of 19 of the patients on cyclosporin compared with six of 19 patients on placebo reported at least a moderate improvement. The estimated rate difference was 39% (95% CI 10 to 62). One small study by Harper and colleagues286 of continuous versus intermittent cyclosporin A in children, suggested that more consistent control is achieved with continuous therapy when given over a 1-year period. This longer-term study did not demonstrate any clinically significant change in kidney function (serum creatinine) and blood pressure in either group.

Three of the remaining six reports were of the same trial293,295,298 with most of the outcome data reported in the Sowden and colleagues report. 295 None of the studies compared cyclosporin with an active agent, with all three remaining RCTs being comparisons of cyclosporin with placebo. 290,292,295 Those studies gave some poolable data because of similarities in study design and the use of a consistent visual metric analogue scale for scoring itch (Table 36). However, because of the crossover design used in all three trials, only the first phase of each study was used so that our estimates were not confounded by carry-over effects. Table 37 shows the estimated mean difference in itch scores for the cyclosporin group compared with the placebo group for each of the three studies as well as the pooled estimate at the end of the first period.

Irrespective of whether a fixed- or random-effects model was used, the pooled estimate showed that cyclosporin was effective in relieving eczematous itch compared with placebo as shown in Figures 4 (fixed effects) and 5 (random effects).

Generally positive results were reported for sleep loss, area involvement, reduction in steroid use and erythema.

The small study remaining by Miranda and colleagues287 compared oral cyclosporin A with oral transfer factor for 6 months and found no statistical differences in a range of outcomes between the two. No intention-to-treat analysis was performed. Three patients in the cyclosporin A group reported excess hair growth (hypertrichosis).

Comment

There is little doubt that cyclosporin is effective for the treatment of atopic eczema when compared with placebo but that continued use is necessary for the prevention of relapse, which is rapid once therapy is discontinued. Clinical scores return close to baseline values within 8 weeks. 292 Adverse effects of the drug, notably on the liver and kidneys, suggest that long-term treatment is not justifiable. Even in short-term trials, cases of hypertension and elevations of serum bilirubin and creatinine have been reported. 291 Reducing dose schedules or prolongation of treatment-free intervals have yielded unconvincing results with obvious poorer control of the disease and modest decrease in drug exposure. As recommended by Zaki and colleagues,397 cyclosporin should be reserved for the short-term treatment of refractory disease, and even then its superiority over oral steroid therapy, a substantially cheaper alternative, is as yet untested.

Benefits

Thirty-six patients aged 13–64 years with atopic eczema were randomised to levamisole hydrochloride or placebo according to body weight with topical triamcinolone as co-treatment. Of 26 evaluable patients at the end of the 6-month trial, there were no clinical or statistically significant differences in patients' objective improvement, frequency of infections, physician prediction of active treatment, clinical scores, or immunological markers such as IgE changes. In the active groups, six of 11 patients noticed improvement compared with six of 15 in the placebo group. Mean percentage improvement in a composite sign score (not defined) was 44% in the levamisole group and 16% in the placebo group.

Harms

One patient developed urticaria and another developed nausea and vomiting while taking levamisole.

Comments

The quality of this study was surprisingly good for such an early publication with a clear description of the blinding process and testing for success of blinding. Despite the lack of intention-to-treat analysis and very small sample size, there is not a hint of any benefit of levamisole in this study or in the Alomar study. The placebo benefit in the studies was quite remarkable in that they were greater than those on active treatment. The authors conclude that the study emphasises the need for randomised double-blind comparisons for new drugs such as levamisole in view of the previous claimed excellent results from uncontrolled studies. Although the study was under-powered to miss small treatment effects, it is unlikely that levamisole has any long-term moderate-to-large treatment benefit in atopic eczema.

Benefits

Based on an intention-to-treat population analysis, 57% of those on the experimental PAF antagonist ‘responded’ (response not defined by authors) compared with 61% on placebo at the end of the treatment period. Based on the evaluable population of 36 patients, 18 of the experimental group showed marked improvement or total clearing on the site treated with the solution compared with 17 of the 36 for the site treated with placebo. Other parameters of atopic dermatitis severity showed similar lack of difference apart from a transient statistically significant benefit for itching at Day 14 but not on Day 28.

Harms

Fourteen out of 15 patients complained of skin dryness and burning immediately after application of the treatment. Data were not presented separately for the active versus vehicle treatment. There was one case of possible contact dermatitis related to the trial medication and another who developed severe erythema.

Comment

There appears to be no benefit from the PAF antagonist in this small study, although it lacks the power to exclude a potentially useful benefit. Randomisation, concealment of allocation, and blinding was poorly described, though an intention-to-treat analysis was performed.

Benefits

In a 12-week multicentre study of 83 patients with severe atopic eczema aged between 2 and 65 years, daily subcutaneous injections of recombinant interferon-gamma at a dose of 50 μg/m² was compared with placebo injections with topical corticosteroids continued as co-treatment. All patients were instructed to take oral acitaminophen (an analgesic) before and after injections to protect against headaches and aching limbs associated with interferon therapy. Those randomised to active treatment were significantly older than those on placebo. At the end of the assessment period, 45% of 40 patients on active treatment compared with 21% of 43 on placebo were reported as having greater than 50% global improvement (p = 0.016) as judged by a physician who was blinded to treatment allocation. Corresponding figures for the proportion of patients with more than 50% patient/parent-reported global improvement was 53% and 21% for active versus placebo treatment, respectively (p = 0.002). Other physical signs, such as redness and scratch marks, were statistically significantly reduced by about 30% in the active group, and a similar magnitude of improvement was seen for induration, itching, dryness and lichenification, though these were not statistically significant. Response was greatest in younger patients. Serum IgE levels did not fall significantly. Another more recent study by Jang and colleagues303 has compared high- versus low-dose interferon gamma (1.5 versus 0.5 million units/m²) versus placebo given subcutaneously three times week to patients with recalcitrant atopic dermatitis aged 18–42 years for 12 weeks. At the end of the evaluation period, clinical improvement measured by means of a composite of different physical signs and surface area, was markedly better in the two interferon gamma groups compared with placebo (p < 0.05), but not between the two interferon-gamma dose groups, except perhaps with a more rapid benefit in the first 6 weeks in the higher dose group. A host of other immunological markers were assessed.