A programme of research to set priorities and reduce uncertainties for the prevention and treatment of skin disease

What was already known about this topic?

• Many published studies have evaluated different interventions aimed at preventing eczema. Often, these have been incorporated into systematic reviews on the topic.

• However, there is no single source of information that compares all of the different interventions; thus, it is difficult to make comparisons and clinical recommendations.

What did this study add?

• This overview of systematic reviews presented the available evidence from Cochrane and non-Cochrane reviews on seven interventions for preventing eczema.

• There was no clear evidence that any of the interventions reviewed can reduce the incidence of eczema.

• In a subgroup of high-risk infants, there was a significant reduction in the incidence of eczema in those given prebiotics compared with no prebiotics.

• A similar reduction in eczema incidence was seen in high-risk infants who were exclusively breastfed for 6 months compared with those given non-breast milk liquids and/or solid food at 3–6 months, but this effect was not sustained past the age of 2 years.

• There were very few published data on the prevention of truly AE, that is, eczema associated with IgE sensitisation; data on probiotics versus no probiotics showed no significant difference.

• Adverse event data were available for infants only. The only adverse event to emerge from the review was that more infants receiving probiotics were spitting up (reflux/regurgitation) at 1 and 2 months of age than those not receiving probiotics.

• We were not able compare breastfeeding versus infant formula alone or the avoidance of pets and other aeroallergens from birth owing to an absence of published systematic reviews dealing with these topics that met the inclusion criteria.

Results of the search

Seven systematic reviews were included, containing 39 relevant trials (11,897 participants) (see Appendix 1). Six were Cochrane reviews covering the following interventions for preventing eczema: maternal dietary antigen avoidance,54 optimal duration of exclusive breastfeeding,55 hydrolysed protein formula,56 prebiotics,57 probiotics58 and soy formula. 59 One non-Cochrane review on omega-3 and -6 fatty acid supplementation was also included. 60 The included Cochrane review on optimal duration of exclusive breastfeeding compared prolonged exclusive breastfeeding with the introduction of complementary solid or liquid foods following a short period of exclusive breastfeeding. It did not compare breastfeeding with infant formula alone. Therefore, we sought to include a review on this topic, but none met our inclusion criteria. Similarly, none of the non-Cochrane reviews examining the avoidance of pets or other aeroallergens satisfied the inclusion criteria, so we were unable to assess these interventions.

Eczema

There was no strong evidence that any of the seven interventions reviewed can prevent the development of eczema (Table 1). However, there was weak evidence that some of the interventions may reduce the risk of developing eczema in some subgroups.

Exclusive breastfeeding for at least 6 months compared with introduction of non-breast milk liquids and/or solid food at 3–6 months did not significantly decrease the overall incidence of eczema. However, in a subgroup analysis of high-risk infants, exclusive breastfeeding for 6 months significantly decreased the risk of developing eczema by 60% (RR 0.40, 95% CI 0.21 to 0.78), an effect that was not significant beyond 2 years of age or in infants not selected for risk of developing allergic disease.

When prebiotics were compared with no prebiotics in infants not selected for risk, there was no overall effect on eczema prevention. However, in high-risk infants, use of prebiotics was found to significantly decrease risk of developing eczema by 58% (RR 0.42, 95% CI 0.21 to 0.84). One trial61 comparing different types of prebiotics found that a combination of polydextrose, galacto-oligosaccharide and lactulose compared with polydextrose and galacto-oligosaccharide alone significantly decreased the incidence of eczema by 78% in infants not selected for risk of developing allergic disease (RR 0.22, 95% CI 0.07 to 0.76).

Compared with cow’s milk formula, there was no clear benefit of prolonged feeding of hydrolysed milk formula (all types) or subgroup analyses of partially or extensively hydrolysed milk formula. Prolonged feeding with extensively versus partially hydrolysed milk formula showed no significant difference. Early short-term feeding of hydrolysed milk formula (all types) showed no benefit over human milk or cow’s milk formula. Soy milk formula versus cow’s milk formula also showed no significant benefit. Maternal antigen avoidance was no better than standard diet, omega-3 or -6 fatty acid supplementation was no better than placebo, and there was also no effect of probiotics compared with no probiotics. None of the above subgroup analyses was significant based on either age or risk level.

Atopic eczema

Although we pre-specified AE associated with IgE sensitisation as one of our two primary outcomes, only one comparison – probiotics versus no probiotics – provided data examining this outcome. There was no significant difference in the overall incidence of AE, nor in subgroup analyses of high-risk infants and infants not selected for risk.

Atopy

Data were available for three different comparisons that measured atopic sensitisation using skin prick tests to various allergens: exclusive breastfeeding for at least 6 months versus introduction of solids at 3–6 months, maternal dietary antigen avoidance versus standard diet and omega-3 fatty acid supplementation versus placebo. No significant differences were identified.

Adverse events

The only data available on adverse events were from infants, based on nine different comparisons of interventions. The only significant difference identified in adverse events was in the comparison of probiotics versus no probiotics. Parents reported that significantly more infants receiving probiotics were spitting up (reflux/regurgitation) at 1 and 2 months of age (RR 1.88, 95% CI 1.03 to 3.45, and RR 1.69, 95% CI 1.02 to 2.80); however, premature infants receiving probiotics experienced a 65% reduction in necrotising enterocolitis and/or death (RR 0.35, 95% CI 0.15 to 0.83).

Quality of the evidence

Significant numbers of trials and participants have been included in this overview, allowing reasonably robust conclusions to be drawn about the effectiveness of different interventions. A particular problem in some included reviews was the heterogeneity in the nature of the interventions (e.g. partially vs. extensively hydrolysed formula, and whey vs. casein formula) and population (e.g. early, late and prolonged formula feeding; exclusively breastfed, mixed breastfed or exclusively formula-fed; and high or normal risk for allergic disease) which makes it difficult to perform meta-analyses in such populations. Trials of prebiotic and probiotic interventions also suffered from large heterogeneity in the interventions and it should be noted that several probiotic trials have been published since the last systematic review on this subject,62–70 warranting an updated review.

The quality of the systematic reviews included in this overview was high: six reviews were Cochrane reviews and all seven reviews in this overview addressed most of the AMSTAR51 systematic review quality criteria and received overall ratings of ‘good’ (all included reviews are detailed in Appendix 1).

Overall completeness and applicability of evidence

Many of the interventions of interest that we identified prior to conducting this overview were captured in Cochrane and non-Cochrane systematic reviews. Comparisons examining exclusive breastfeeding for at least 6 months, hydrolysed formula, omega-3 fatty acids and probiotics contained significant numbers of trials and participants and were thus powered to detect important treatment effects. However, analyses of soy formula, maternal antigen avoidance and prebiotics included smaller numbers of participants and studies, so may have missed important treatment effects on the incidence of eczema.

Other areas for which more systematic reviews and trials are needed are interventions to promote normal immune development and interventions to correct skin barrier dysfunction. The latter is particularly important in view of the growing body of evidence for abnormalities of skin barrier function in eczema. Common loss-of-function mutations in the gene encoding the epidermal barrier protein filaggrin are a major predisposing factor for eczema34 and skin barrier dysfunction can be detected prior to eczema development. 48

There is a need for more complete reporting of adverse events within systematic reviews of interventions for eczema prevention. In addition, greater numbers of trials must report the effects of these interventions on eczema prevalence beyond infancy, particularly in normal-risk infants.

Limitations

This overview was undertaken using a consensus-based process informed by The Cochrane Collaboration expertise in preventing bias and provides a thorough review of several interventions for eczema prevention. A particular strength of the overview was collaboration with the authors of the 2010 NHS Evidence Skin Disorders Annual Evidence Update on Atopic Eczema, which ensured a thorough literature search and access to all published systematic reviews of interventions for preventing eczema. All analyses were defined a priori, such as the inclusion of reviews of non-randomised studies for two interventions (breastfeeding and pet avoidance) where RCTs are ethically difficult to perform.

The strength of some conclusions was limited by small numbers of trials or participants included in the relevant reviews and the analysis of probiotic effects was limited by the recent publication of many relevant trials62–70 that have not yet been incorporated into a Cochrane systematic review. A more recent meta-analysis incorporating many of these trials71 found that probiotic supplementation significantly reduces risk of developing eczema in infancy (RR 0.79, 95% CI 0.67 to 0.92), a finding supported by subsequent reviews. 72

It was not possible to extract data from the published systematic reviews on breastfeeding versus no breastfeeding and pet exposure in a form that could be incorporated in this overview. The systematic review of breastfeeding versus no breastfeeding73 did find a significant protective effect against eczema [odds ratio (OR) 0.70, 95% CI 0.50 to 0.99]; however, when one controversial trial was excluded from the analysis, this comparison became non-significant (OR 0.84, 95% CI 0.64 to 1.09). The review on pet exposure74 found that, in longitudinal studies, exposure to cats (OR 0.76, 95% CI 0.62 to 0.92), dogs (OR 0.68, 95% CI 0.53 to 0.87), or ‘any furry pets’ (OR 0.79, 95% CI 0.74 to 0.84) was associated with reduced risk of developing eczema. However, one trial noted that when adjustments were made for avoidance behaviour in participants allergic to pets, the protective effect of cats became non-significant.

We identified only one review which reported specifically on ‘atopic eczema’ as defined by the WAO recommendations as an outcome rather than ‘all eczema’. 2 Additionally, although we pre-specified AE associated with IgE sensitisation as one of our two primary outcomes, only one comparison – probiotics versus no probiotics – provided data examining this outcome. There was no significant difference in the overall incidence of AE, nor in subgroup analyses of high-risk infants and infants not selected for risk.

It is worth noting that very few of the systematic reviews evaluated atopy/allergic sensitisation as an outcome, despite it being reported in many of the included RCTs. Although this may be justified because atopy/allergic sensitisation is not a widely recognised disease entity, the primary prevention of atopy/allergic sensitisation may be associated with prevention of later onset allergic disease and is therefore worth including in future systematic reviews in this area. In addition, no data were identified for four of our pre-specified secondary outcomes: eczema severity, time to development of eczema, QoL and health-care utilisation. All are important parameters to measure in future trials and systematic reviews; for example, there is little gain in preventing eczema incidence if the absolute number of severe cases (which disproportionately account for most health-care costs) remains unchanged.

Agreements and disagreements with other studies or reviews

The results of this overview are broadly consistent with those of other published trials and reviews of RCTs, but are less in agreement with reviews and practice guidelines that include ‘all allergic disease’ or ‘allergy’ as primary outcomes or that include significant numbers of non-randomised trials. For example, a recent German taskforce on allergy prevention recommended exclusive breastfeeding for 4 months or, if not possible, the use of a hydrolysed formula for high-risk infants and general avoidance of pets. 75 These differences might stem from the inclusion of other allergy outcomes and a large number of non-randomised studies in the German taskforce recommendations. A recent RCT of hydrolysed formula for preventing allergic disease in high-risk infants supports our conclusion that further research is needed before recommending this intervention to prevent eczema. 76

Implications for clinical practice

This overview of systematic reviews on eczema prevention has not found any clear evidence that any of the main interventions reviewed can reduce the incidence of eczema. That does not mean to say that some interventions do not work; new, larger, well-conducted trials may indeed show a modest benefit in time. However, the current evidence is simply not strong enough to influence practice recommendations. Some interventions, such as the use of soy instead of cow’s milk, are unlikely to show a clinically useful benefit based on the lower 95% CIs and the pooled effect estimates. It may be considered that some of the interventions are unlikely to do much harm; however, it should be considered that adverse events were poorly reported in many of the included trials.

It is also worth noting that we have not been able to address some of the most commonly asked questions by parents (e.g. whether or not exclusive breastfeeding or owning a pet prevents eczema) owing to lack of appropriate data. This may partially reflect the ethical difficulties of conducting trials in these areas. Furthermore, some interventions such as ‘probiotics’ are simply too heterogeneous to consider as one intervention – some may work but some may be completely ineffective. This overview has found that the possible benefit of some interventions (such as exclusive breastfeeding or prebiotics) may only be present in infants born to families at high risk for allergic disease and that the magnitude of risk reduction is larger than the RRs from unselected populations. Again, caution has to be expressed as these results were based on only one trial each with significant limitations.

Implications for research

These findings indicate that there is little justification for further study of some proposed interventions for eczema prevention, such as omega-3 fatty acid supplementation and use of soy milk instead of cow’s milk. Other interventions, such as hydrolysed formula, prebiotics and probiotics, have shown inconclusive results and are worthy of further study. New interventions designed to enhance skin barrier function, such as intensive use of emollients, water softeners or avoidance of alkaline soaps, should also be explored and a separate overview of reviews examining interventions for preventing asthma and other allergic diseases is warranted.

What was already known about this topic?

• There are many published definitions for the diagnosis of AE. Many require the presence of signs or symptoms over a long period of time as a diagnostic criterion, but this is not appropriate for defining new cases.

• Interventions that delay rather than prevent the onset of eczema could still have a significant public health impact. Identification of such interventions requires more precise determination of the date of onset of eczema than is offered by cumulative incidence rates.

What did this study add?

• This study was a systematic review of how incident cases of eczema have been defined in previous primary prevention studies.

• Many different definitions have been used to define a new case of AE, with the Hanifin and Rajka criteria the most commonly applied. Many trials used a definition that was unique to that particular study.

• The Hanifin and Rajka criteria include presence of signs or symptoms over a long period of time as one of the major criteria, but only two studies stated how they dealt with this in relation to definition a new cases.

• We propose a minor modification to the UK Working Party Criteria to define an incident case to reduce the timescale from 1 year to 4 weeks for the presence for the presence of an itchy skin condition.

Included studies

Prospective, interventional, prevention trials published after 1980, which specified eczema as an outcome, were eligible. Studies whose primary outcome was not eczema, such as asthma prevention trials, but had eczema as a secondary outcome, were included. There were no age or language restrictions. Observational cohort studies were excluded as these studies most often use cumulative incidence as an outcome. Current validated definitions for the diagnosis of eczema are suitable for the measurement of cumulative incidence with a maximum precision of 1 year. We aimed to investigate incident case definitions that would capture cases as they occur in a primary prevention trial, enabling a time of disease onset to be determined.

Information sources

Studies were identified by searching electronic databases, scanning reference lists of articles and eczema reviews, and consultation with experts in the field. The search was applied to MEDLINE (1980–2012), Cochrane (1980–2012) and the last database search was run on 5 January 2011 (see Appendix 2). Multiple articles were identified through hand-searching, predominantly of reviews identified in search results, reviewing the literature and from the NHS Evidence mapping exercise of systematic reviews for eczema prevention. 82

Study selection

All database search results were entered into RefWorks (ProQuest LLC, MI, USA) in which duplicates were removed. Screening for eligibility was performed independently in a standardised manner by two reviewers (ELS and LEK) based on titles and abstracts. Discrepancies between reviewers were resolved by consensus. Full-text articles for all studies past screening were obtained. Full-text copies of studies with seemingly eligible titles but without abstracts were automatically included and scrutinised further for eligibility. All full-text articles were assessed for eligibility by one reviewer prior to data collection. Of note, if multiple publications of a larger longitudinal trial were identified, they were counted as only one definition unless the definitions significantly differed between publications, in which case they were kept separate.

Data collection process/items

Data collection was performed by one reviewer using a data extraction database created for the study. Information extracted from each trial included the type of intervention and the eczema definition used. One author (LEK) extracted the above data from the included studies and a second reviewer (ELS) double-checked all verbatim definitions that were considered to lack a true definition of eczema. Disagreements were resolved by discussion between the two reviewers and consultation with a third arbitrating author.

Summary measures

The primary outcome measured was the presence of any specified definition for eczema.

Study selection

The study selection procedure is summarised in the flow chart in Figure 2. A total of 108 articles met the criteria for inclusion. Seven of these were multiple publications from the same longitudinal study, so this definition was counted only once, resulting in 102 distinct studies for inclusion in the analysis.

FIGURE 2.

Study flow chart for showing screening process for screening studies related to defining new cases of eczema. AD, atopic dermatitis.

More than 80% of the included 102 studies evaluated a dietary intervention on either the mother or the infant, with infant formula as the most common eczema prevention strategy used. The other interventions were non-dietary allergen avoidance, vaccination and an emollient intervention (Figure 3).

FIGURE 3.

Interventions assessed in the 102 included studies relating to defining new cases of eczema.

Definitions of eczema

Of the 102 studies selected for further analysis of eczema definition, only 75 (73.5%) included some form of a description of the criteria used to diagnose eczema. The other 27 articles mentioned either a diagnosis made by a questionnaire that was unavailable for review (one study), a general morphological description of eczema (four studies), a physician or investigator diagnosis of ‘eczema’ without further elaboration of specified criteria (20 studies), or no description at all (two studies). 77

Of the 75 studies with reported disease criteria, the Hanifin and Rajka criteria were the most commonly used disease criteria (28 studies; Table 2). Of those studies that used the Hanifin and Rajka criteria (which includes chronic or chronically relapsing dermatitis as one of its four major diagnostic criteria), only two studies specified how they dealt with the anomaly of disease chronicity in relation to definition a new case. 84,85 The study by Laitinen et al. 85 required visible eczema to be present for at least 4 weeks at the 6- and 12-month visits and for at least 8 weeks at the 24- and 48-month visits. Arslanoglu et al. 84 required ‘symptoms’ to be present for at least 4 weeks to meet criteria for eczema.

Disease definitions cited in eczema prevention studies

A disease definition that was unique to that particular study (21 studies) was the second most commonly used disease definition. These reports did not cite a specific source and none described a scientific method or even more detailed empirical reasoning for justifying the choice of a novel definition of an incident case.

On closer examination of the 21 definitions that were unique to an individual study, most definitions included pruritus, the presence of visible eczema, and disease distribution requirements. Only 50% of these definitions had a time requirement, which ranged from requiring ‘chronic disease’ to 4 weeks of eczema needing to be present (data not shown).

Main findings

We found a large degree of variability in the methods used to define an incident case of eczema in relevant prevention studies, with one-quarter of studies failing to report any form of definition whatsoever. Of the studies reporting some form of definition of an incident case, many used a definition that was unique to that particular study, rendering comparisons between studies very difficult. In addition, most studies used eczema incident case definitions without strict time requirements. Combined, these findings demonstrate an urgent need for a standardised, valid and repeatable definition of an incident case of eczema in order to improve the ability to compare outcomes between studies and to allow more informative meta-analyses of prevention studies. Although no previous studies have examined incident case definitions for eczema, our results are consistent with the larger problem of the lack of standardised disease outcome measures in eczema research. For example, Schmitt et al. 175 found 20 different scoring systems for measuring the severity of eczema. Chamlin et al. 78 found 30 different definitions for defining an individual as high risk for developing eczema. In 2010 an international group called the Harmonising Outcome Measures for Eczema (HOME) initiative (see Chapter 2, Harmonising Outcome Measures for Eczema initiative) began the process of creating an accepted core group of outcome measures for eczema research. 176

Problems with existing definitions

The two most commonly used validated criteria found in our review were the Hanifin and Rajka criteria and the UK Working Party refinement of the Hanifin and Rajka criteria. Despite the potential usefulness of these criteria in reliably identifying cases of established eczema, as would be used for determining cumulative incidence over time in cohort studies,177 they were not designed for defining an incident case in prospective prevention studies. By defining an incident case in prevention trials, as opposed to a cumulative incidence, incidence rates can be more accurately calculated and a more accurate date of disease onset can be established. In order to qualify as a case of eczema using the Hanifin and Rajka criteria, at least three out of four of the following major criteria need to be fulfilled:

1. the presence of eczema

2. typical distribution

3. pruritus

4. a relapsing and remitting course.

The definition of ‘relapsing and remitting’ is not further defined and is left to the discretion of the investigator. The UK criteria state that a child must have an itchy skin condition in the past 12 months. These time requirements, although appropriate for diagnosing established cases of eczema, become problematic when diagnosing new onset eczema during the course of a prospective study.

Defining an incident case of eczema is not simply a question of noting the first time an eczematous rash appears in an infant because previous studies have shown that many forms of transient eczematous rashes occur often in infants, even in those children who do not eventually develop true eczema. 178 Thus, there is an urgent need for a standardised definition of an incident case of eczema that offers a satisfactory trade-off between overinclusion of transient eczematous eruptions of irritant and other aetiologies and overexclusion of genuine milder short-lived forms of eczema that still represent a health-care problem.

Proposed solution

Until more sophisticated validation studies can be performed, we suggest a modification of the UK Working Party criteria for eczema, adapted for prospective observational or interventional studies. This modification specifies a time frame that the eczema must be present in order to be considered as a case of eczema and allows for a diagnosis to be made even if the rash is treated early in its course. We propose the following definition, based on empirical reasoning considering the requirements of such a definition, and informed by previous studies that signal the best markers of true eczema. 178

A history of an itchy skin condition, which is either continuous or intermittent, lasting at least 4 weeks, plus three or more of the following:

1. a history of a rash in the skin creases (folds of elbows, behind the knees, fronts of ankles or around the neck), or on the extensor aspects of the forearms or lower legs

2. a personal history of asthma or hay fever or a history of atopic disease in a first-degree relative

3. a history of generally dry skin since birth

4. visible flexural dermatitis and/or visible dermatitis on the forearms or lower legs with absence of axillary involvement as defined by our online photographic protocol. 179,180

Visual confirmation of eczema diagnosis by a clinician, dermatology nurse or a research nurse suitably trained in recognising the symptoms of eczema is recommended. Clearly it is important to add the proviso that any infant fulfilling these criteria but who, on examination by a suitably trained health professional, is deemed to have a different skin disease, will be classified as not having eczema.

There are several benefits of our proposed definition, the strongest being that it is based on a current eczema definition that has undergone extensive scientific development assessing validity, repeatability and applicability. The only change added to the UK Working Party definition is the addition of a specified time requirement of 4 weeks. This time requirement should exclude most transient eczematous rashes that are typically irritant in nature and usually of little medical consequence. The use of an established eczema definition for incident eczema that is derived from one used for prevalent eczema allows for consistency in defining the public health burden of disease when assessed using different study designs. Another benefit of this proposed definition is that it does not allow a definition of eczema to be made based on the presence of facial eczema alone. Halkjaer et al. 178 found that 40% of children with facial eczema do not eventuate into chronic eczema. Finally, this definition allows for early treatment intervention during the course of a prospective study and does not require the disease to be untreated for a full 4 weeks if anti-inflammatory therapy is needed. Therefore, if a child develops significant eczema in the classic locations, it would be unethical to withhold treatment. Treatment can begin immediately if needed and, provided some degree of symptoms last for a 4-week period, a diagnosis of eczema will still be captured using this definition.

Very mild cases of new eczema treated immediately resulting in complete clearance will not be captured by such an approach, although it is debatable how many of these case are true eczema and it is probably wise to treat such new very mild cases with emollients alone until the disease declares itself. This approach is consistent with the recommendation to avoid diagnosing asthma in an infant based on a single episode of wheezing.

Particular strengths of this study include its systematic approach to the review of the literature and extensive searching of reference lists for prevention studies not found on the initial search. The findings are particularly relevant given the renewed interest in eczema prevention research and prevention by the National Eczema Association in the United States. 181

With regard to limitations, it is possible that other unpublished definitions of incident cases exist but these would not be detected by our search. In addition, the proposed eczema definition has not yet undergone extensive validity testing in prospective studies in the field. However, this limitation has to be tempered with the alternative practice to date, which has been to use unsuitable definitions or an array of poorly defined or completely undefined definitions. Our proposed definition is meant to be a starting point and we encourage those undertaking or designing new prospective studies of eczema to include it along with their preferred definitions so that knowledge of its utility and validity can be built up.

What was already known about this topic?

• Studies showing an association between loss-of-function filaggrin gene mutations and AE have generated interest in the potential for enhancing the skin barrier to prevent the development of eczema.

• Emollients can reduce the incidence of flares of existing eczema and there is some evidence from small pilot studies that emollients may have a preventative effect against development of eczema.

• A definitive RCT is needed but pilot work is required before this can go ahead.

What did this study add?

• Parents appear to be willing to participate in an emollient prevention trial. Those in the intervention group adhered to the emollient regimen and those in the control group were prepared to refrain from using emollients.

• We were able to establish that the liquid-paraffin based emollients, Doublebase^® gel (Dermal Laboratories Ltd) and Diprobase^® cream (Bayer Plc) are popular with parents and do not have any detrimental effects on skin barrier function.

• This pilot RCT provides the first signal from a RCT that daily full body emollient application from birth can prevent eczema.

• Caution should be shown in reading too much into the finding of reduced eczema at 6 months in this study because it might simply represent suppression of very mild eczema by the continued use of emollients.

• The results of the pilot study and emollient studies were used to support a successful bid to the NIHR HTA to conduct a definitive RCT of emollients for the prevention of eczema.

Importance of the skin barrier

Although previous eczema research has focused on the role of the immune system in atopic inflammation, the strong association between filaggrin mutations and AE (as well as atopic asthma and allergic rhinitis) has generated interest in the potential role of the skin barrier as the key early event leading to eczema development.

Dry skin is very common in eczema,183 even in the absence of known filaggrin loss-of-function mutations. A defective skin barrier allows water to be lost from the skin, resulting in a generally dry skin – one of the first abnormalities to be noticed in babies who eventually develop eczema, which supports the notion that the primary event in the development of eczema and atopy is a dysfunctional skin barrier. The skin barrier not only keeps useful things like water in, but also helps to keep out potentially harmful things such as irritants, bacteria and allergens. The use of harsh soap and detergents can raise the pH of the outer layers of the skin and disturb the fine balance of enzymes, proteins, lipids and micro-organisms on the skin surface. 184 A rise in pH leads to further breakdown of the skin barrier and is therefore a common pathway through which genetic and environmental factors influence skin barrier function. 184–186 Skin irritation from soaps and other wash products is worse in children with a pre-existing skin barrier defect (Figure 4).

FIGURE 4.

Model of skin barrier function. In healthy skin, the corneodesmosomes (iron rods) are intact throughout the stratum corneum. At the surface, the corneodesmosomes start to break down as part of the normal desquamation process, analogous to iron rods rusting (a). In an individual who is genetically predisposed to eczema, premature breakdown of the corneodesmosomes leads to enhanced desquamation, analogous to having rusty iron rods all the way down through the brick wall (b). If the iron rods are already weakened, an environmental agent, such as soap, can corrode them much more easily. The brick wall starts falling apart (c) and allows the penetration of allergens (d). Reprinted from J Allergy Clin Immunol, vol. 118, iss. 1, Cork MJ, Robinson DA, Vasilopoulos Y, Ferguson A, Moustafa M, MacGowan A, et al. New perspectives on epidermal barrier dysfunction in atopic dermatitis: gene–environment interactions, pp. 3–21, copyright 2006 with permission from Elsevier. 187

It is also possible that the skin is the primary organ for development of allergic sensitisation. Animal studies have suggested that IgE sensitisation may occur via the skin. 188 Even though allergens are too large to penetrate the skin directly, the defective skin barrier makes it easier for allergens to interact with skin cells such as Langerhans cells, which are responsible for initiating sensitisation. 189 The observation that mutations in the gene coding for the skin barrier protein filaggrin are associated with peanut allergy independently of eczema,190 and that the use of creams containing peanut oil on the skin during the first 6 months of life may be a linked to developing a peanut allergy,191 further support the notion that the skin might be a primary route of sensitisation for food allergies. If true, then the skin barrier is a target for prevention of not only eczema, but also for food allergy and progression to asthma and allergic rhinitis in the atopic march.

This evidence provides the basis for our proposal, to perform a primary prevention trial designed to enhance the skin barrier, by using emollients from birth, in an effort to prevent eczema and associated allergic diseases. It is unclear how long skin barrier enhancement needs to occur in order to produce possible lifelong benefits, but as allergic diseases occur in the first few months of life, and most sensitisation occurs in the first year of life, we have opted for a 12-month emollient intervention period.

Emollients and the skin barrier

Emollient (moisturiser) therapy improves the skin barrier function. An emollient improves skin hydration by trapping in water. Emollients have also been shown to play a role in preventing irritant occupational hand eczema. 192 Emollients have been shown in premature babies to reduce the incidence of skin inflammation,193 to reduce flares of eczema (secondary prevention) and to decrease the need for topical steroids. 194 Not all emollients are the same, as they vary in their consistency from greasy paraffin derivatives to lighter, water-based, creams.

Primary prevention and the NHS

Primary prevention is a highly desirable goal in a chronic disease such as eczema with no cure. Parents with experience of eczema are often anxious to know whether or not their future children will develop eczema and what they can do minimise the risk. 195 If primary prevention of eczema using a strategy of early skin barrier enhancement with simple low-cost emollients works, it would represent a significant cost saving for the NHS through reduced treatment and appointment costs, especially in those cases persisting into adulthood. Further cost savings would result if early skin barrier enhancement prevents sensitisation and associated food allergy, asthma or allergic rhinitis. Even if the frequency of eczema cannot be significantly reduced, a reduction in the severity distribution of eczema could reduce the distress to patients, the number of consultations in primary care and subsequent referrals to secondary care.

Other emollient prevention studies

An early case–control study conducted in Kenya and published in 1991 suggested that petroleum had a protective effect against the development of eczema,49 but this study has not been followed up with a definitive RCT. One RCT from Bangladesh has shown that barrier enhancement from sunflower oil and Aquaphor ointment may reduce serious infections in preterm babies. 196 A small Japanese pilot study197 [International Clinical Trials Registry Platform study identifier (ID): UMIN000004544] of 70 patients looking at emollients as a prevention strategy for eczema reported a significant difference between eczema onset between the proactive and reactive groups, and that this demonstrated that emollient use in early life can protect high-risk babies from developing eczema. Another small, short-term pilot study conducted in Japan randomised 71 babies at high risk of atopic disease to skin care instructions (including emollients) versus no instructions and found no difference in diagnosed eczema at 6 months. However, the group did show that positive reaction to skin prick tests was lower in the intervention group. 198 An open-label pilot study of emollient therapy from birth showed only 15% of high-risk infants developed eczema against an expected rate of 30–50%. 47 This study also showed that emollient therapy was a safe and acceptable intervention.

We are not aware of any other definitive trials under way to evaluate the prevention of eczema through barrier enhancement after searching trial registries (World Health Organization meta-register from inception to 18 December 2013). We did find one commercial study (n = 400) taking place in the USA and Canada (NCT01577628) that is evaluating a cosmetic moisturiser containing shea butter, paraffin, waxes and vegetable oils (Lipikar Balm AP, Cosmétique Active International) for the prevention of eczema. 199

This section describes the pilot work that has been conducted in order to design a large definitive RCT of emollient therapy from birth as an eczema prevention strategy. A pilot RCT was carried out to determine the feasibility of a large RCT, followed by a parent preference ranking exercise of emollients and mechanistic studies to look at the effects of emollients on the skin barrier to inform the choice of emollient(s) in the main RCT.

Methods

The full protocol is available. 200

Study design

This was a multicentre, multinational (UK and USA) two-arm parallel group, assessor-blind, randomised, controlled, pilot trial of 6 months’ duration. The intervention started within the first 3 weeks of birth.

Participants

Infants at high risk of developing eczema, defined as having a parent or full sibling that has (or had) doctor-diagnosed eczema, asthma or allergic rhinitis, were included. Infants needed to be in overall good health and the mother at least 16 years of age at delivery and capable of giving informed consent. If the mother had taken Lactobacillus rhamnosus supplements during pregnancy, they were excluded. Babies were excluded if they were born prior to 37 weeks’ gestation, or if they had a major congenital anomaly, hydrops fetalis, any immunodeficiency, a severe genetic skin disorder or a serious skin condition that would make the use of emollients inadvisable.

Intervention

Outcomes

The primary outcome for this pilot study was the proportion of eligible families who approached the study team who were willing to be randomised.

Secondary outcomes were:

• proportion of families eligible for the trial

• proportion of families accepting the initial invitation to participate

• percentage of early withdrawals

• cumulative incidence of eczema at 6 months

• age at onset of eczema and the proportion which are transient cases

• incidence of emollient-related adverse events

• proportion of families who found the interventions acceptable

• reported adherence with intervention

• amount of contamination in the control group as a result of increased awareness

• success of blinding of the assessor to the allocation status.

Filaggrin gene mutation testing was performed by the laboratory of Irwin McLean evaluating for the four most common mutations (R501X, R2447X, 2282del4, S3247X) using Taqman^® allelic discrimination assays (Thermo Fisher Scientific Inc., Waltham, MA, USA), as described previously. 204

An adaptation of the validated UK Working Party Criteria131 was used to measure new cases of eczema (see Defining a new case of eczema: a systematic review). These adaptations were a reduction in the duration of presence of symptoms from 1 year to 4 weeks and reflect the distribution pattern of the signs and symptoms of eczema in this young age group.

A history of an itchy skin condition for at least 2 days a week for the past 4 weeks plus three or more of:

1. A history of a rash in the skin creases (folds of elbows, behind the knees, fronts of ankles or around the neck), forearms or lower legs. Symptoms must have been present for at least 1 month either continuously or intermittently unless a topical anti-inflammatory therapy for symptom relief was required. If so, symptoms may have been present for a shorter duration.

2. A personal history of asthma or hay fever (seasonal allergic rhinitis) or a history of atopic disease in a first-degree relative.

3. A history of a generally dry skin since birth.

4. Visible flexural eczema as defined by a photographic protocol and/or visible eczema on the forearms or lower legs with absence of axillary involvement.

Visit schedule and randomisation

The duration of this trial was 6 months. Methods of identifying suitable families differed between the UK and the USA. In the UK, families were usually identified and screened during pregnancy. After the family had made contact with the study team and initial eligibility checks had been carried out by the co-ordinating centre, the research nurse carried out the screening and consent visit, usually at the family home. The baseline visit including randomisation then took place within 3 weeks of delivery, usually as a home visit. In the USA, families were identified by research nurses visiting the postnatal wards each day and approaching parents about the study directly. After giving parents time to consider the study, the research nurse returned to the family to take consent and randomise.

Infants were randomised in a 1 : 1 ratio using random block sizes to either the intervention or control group using a central web-based computer generated internet randomisation service provided by the Nottingham Clinical Trials Unit (NCTU). The allocation list was held by the NCTU and concealed from trial investigators and other trial staff. Allocation was only released to the research nurse by telephone once eligible participants’ details were irrevocably entered into the online database by the co-ordinating centre staff. Randomisation was stratified by the recruiting research nurse. In the case of multiple births, the first born was the index child.

The research nurse contacted parents by telephone at 10 days and 6 weeks, with a face-to-face visit at 12 weeks (usually at home in the UK and as a clinic visit in the USA). This was then followed by a further telephone call at 18 weeks and the final contact was a clinic visit at 24 weeks for an assessment by the dermatologist or dermatology specialist nurse, who conducted a blinded assessment of the skin. In addition to these scheduled contact points, parents were encouraged to contact the research nurse if they had any concerns about the child’s skin. If parents reported symptoms of eczema, then an unscheduled visit to the hospital to see the dermatologist was arranged so that presence of eczema could be confirmed.

Recruitment and setting

Recruitment took place in the UK and the USA between May 2010 and May 2011. In the UK, research nurses were based in three acute NHS hospital trusts (Nottingham University Hospitals, Derby Hospitals and United Lincolnshire Hospitals) and one GP surgery (The Surgery @ Wheatbridge, Chesterfield). In the USA, the study recruited in Oregon Health & Science University Hospital (Portland, OR, USA).

In Nottingham, a research nurse was able to work solely on this study, whereas for the other centres, the research nurses were involved in a number of different trials. Because this was a feasibility study, in the UK, participating families were recruited via a number of different routes in order to test the success of different strategies including invitation letters from GP, posters, leaflets, advertising and by involving health visitors and community and hospital midwives.

Blinding

It is not possible to blind parents in a trial of daily emollient application and it was anticipated that parents would want to discuss their baby’s skin with the nurse. Therefore, the research nurses who conducted the follow-up visits were not blinded to treatment allocation either. An independent outcome assessor who was blinded to treatment allocation performed the skin examinations and diagnosis of eczema. This was usually a dermatologist or a dermatology nurse specialist, although at the primary care site, this was a GP. The effectiveness of blinding was examined by asking the independent assessor to report whether or not they were aware of which group the family was in when they conducted the skin examination and, if unblinding had occurred, the reason why. The statistician was blinded to treatment group until the analysis was complete.

Approvals

The study was given ethical approval by the Nottingham Research Ethics Committee (REC) in the UK (reference 09/H0407/43) and the Institutional Review Board at Oregon Health and Science University in the USA, and approved by all participating institutions. The trial was registered at Current Controlled Trials (ISRCTN84854178) and ClinicalTrials.gov (NCT01142999).

Sample size

This was a pilot study and, therefore, not powered to establish clinical effectiveness of the intervention. The sample size was dictated by the time and resource available. One hundred families would provide an estimate within 10 percentage points for a 95% CI of the proportion willing to be randomised, assuming between 40% and 60% were willing to be randomised.

Data analysis

The primary clinical end point was the week 24 cumulative incidence of doctor-diagnosed eczema. Infants were classed as having developed eczema if either the investigator or another medically qualified person (e.g. a GP) judged that the infant had eczema on examination at any point during the 24-week intervention period.

The primary clinical end point analysis was conducted on an intention-to-treat (ITT) basis using a complete case approach. Multiple sensitivity analyses were performed for the primary clinical end point.

1. Missing data were imputed for infants with missing week 24 skin examination data because they were either lost to follow-up or withdrew during the 24-week intervention period (for reasons other than developed eczema). The following imputations were performed:

   1. Multiple imputation. First, a logistic regression model was created which included the strongest known predictors for eczema derived from a review of the literature. This model included filaggrin status, centre, family history of atopic disease and the assigned intervention group. Variables were then selected for the imputation based on the smallest Akaike information criterion.

   2. All those with missing data assumed to have developed eczema.

   3. All those with missing data assumed to have NOT developed eczema.

   4. All those with missing data in emollient arm assumed to have developed eczema and all those in the control arm assumed to have NOT developed eczema (worst case).

   5. All those with missing data in control arm assumed to have developed eczema and all those in the emollient arm assumed to have NOT developed eczema (best case).

2. Excluding any non-confirmed cases of eczema (i.e. not diagnosed by a study investigator).

Results

The combined results for the UK and USA participants are presented here for each outcome. However, owing to methodological differences between the UK and the USA, particularly around recruitment methods and emollient choices, the UK data (used to inform the design of the main UK-based RCT) are also presented separately.

Baseline characteristics

The baseline characteristics were broadly balanced across the two groups for eczema risk factors apart from filaggrin mutation status (Table 4).

Willingness to participate

The primary outcome of this pilot RCT was to determine the proportion of families willing to participate in a randomised trial of emollients for the prevention of eczema. A total of 295 families met the eligibility criteria, of which 124 (42%) were willing to be randomised, 64 to the intervention group and 60 to the control group as shown in Figure 5. Recruitment took place in five centres (Table 5).

FIGURE 5.

Patient flow in UK and US study populations (combined) for pilot emollient study.

In the UK, willingness to participate was higher than the overall study average. One hundred and sixty-eight families enquired about the trial, of which 121 were eligible after screening during pregnancy (Figure 6). Three families were subsequently not eligible owing to post-delivery exclusions. Of the 120 eligible families, 78 (65%) were randomised. Six families (5%) declined to participate after the baby had been born, despite agreeing to participate during pregnancy. The reasons given were that wider family circumstances meant it was no longer practical to take part or that they had changed their mind about being willing to be randomly allocated to either group.

FIGURE 6.

Patient flow in UK pilot emollient study.

The most common reason for excluding families at screening in the UK was that the baby was due to be born after the end of the planned recruitment period (n = 13), but this was only an issue towards the later part of the study as the deadline for randomisation approached. Seven (4%) families in the UK were excluded at screening because of a lack of family history of atopy.

Withdrawal rates

A total of 16 participants withdrew from the trial (12.9%); nine from the intervention arm and seven from the control arm. Nine out of the 16 withdrawals were lost to follow-up.

In the UK, the withdrawal rate was lower than the overall rate, with six (8%) participants withdrawing from the trial. Two participants were lost to follow-up after moving away from the area, three withdrew from the trial because of family circumstances or illness and one declined to give a reason for leaving the trial.

Incidence of eczema at 6 months

A statistically significant protective effect was seen with the use of daily emollient on the cumulative incidence of eczema at 6 months of age (Figure 7), with a relative risk reduction of 50% (RR of 0.50, 95% CI 0.28 to 0.90; p = 0.017). Sensitivity analysis using a multiple imputation approach for missing data from the 16 participants who withdrew or were lost to follow-up showed a similar protective effect of emollient therapy as the complete case results with an OR 0.31 (95% CI 0.13 to 0.75; p = 0.01). No effect modification was seen when results were stratified by filaggrin mutation status.

FIGURE 7.

Cumulative incidence of eczema at 6 months in pilot emollient study (RR 0.50, 95% CI 0.28 to 0.90; p = 0.017).

Incidence of emollient-related adverse events

Three superficial cutaneous infections occurred in each group, all considered mild in nature. There were no reports of irritant or allergic contact eczema.

Acceptability of the intervention

To inform the decision regarding which emollient(s) to include in the main RCT, parents were offered a choice of emollient in this pilot. The most popular choice in both the UK and the USA was the cream/gel formulation, chosen by 43 (67.2%) of families (Table 6). Only a small number opted for the ointment, with just under one-quarter choosing the oil. The percentages choosing each type of emollient were very similar in the USA and the UK (data not shown).

The main reasons that the parents in the UK gave for choosing Doublebase gel were that it absorbed well and was not too greasy. Families chose sunflower seed oil mainly because it was a ‘natural product’ or because they also wanted to use if for baby massage. Some preferred the heavier feel of the 50 : 50 ointment. All reported that the emollient was ‘acceptable’ and none of the families withdrew from the study because they did not like the emollient.

Reported adherence with intervention

Parents reported good adherence to the intervention, with > 75% of parents in both countries applying the emollient daily throughout the intervention period (Table 7).

Contamination of the control group

The amount of emollient use by the control group during the 6-month intervention period was measured by asking parents whether or not they had regularly applied anything to their baby’s skin and extracting any emollients from the answers given. To reflect the nature of the intervention, contamination was defined as regular, generalised application of emollient for reasons other than the treatment of cradle cap, nappy rash or eczema. Eight control group families used emollients in a way that met this definition, making the overall contamination rate for the trial 8 out of 60 (13%).

In the UK, the contamination rate of the control group was lower, with only 3 out of 37 (8%) meeting our definition. None reported using emollient bath oils other than for the treatment of eczema. A high proportion of families reported using moisturisers in the weeks immediately following delivery and prior to being randomised into the trial (33/78, 42%). This was mainly due to community midwives and health visitors promoting the use of moisturisers for newborn baby dry skin.

Success of blinding of the assessor to the allocation status

A total of five skin examinations were carried out by an assessor who was aware which group the infant was in (all in the UK). In four of these cases, the infant had already been diagnosed with eczema and the only person available to perform the skin examination was already aware of this, having previously treated the infant. One family was not able to bring the child into hospital for an independent assessment so the research nurse, who was aware of the treatment allocation, visited the family home to conduct the skin examination.

Recruitment

In the UK, a number of different recruitment strategies were tested in order to establish the most effective methods for recruiting and retaining families for the main trial (Table 8). The highest number of enquiries was from families who had received a letter from their GP. Participating surgeries invited all pregnant women of > 12 weeks’ gestation who were registered with their practice. The 59 enquiries were generated from 935 letters sent out by 21 GP surgeries. A high proportion of enquiries also came from parents who had seen posters about the study in hospital, usually in the antenatal waiting rooms. Several other recruitment methods produced a smaller number of enquiries including newspaper advertising, hospital and community midwives, health visitors and antenatal classes.

In the USA, a different approach was taken to recruitment and a single method was employed. The research nurses visited the post-natal wards each day and approached mothers about the study. After being given sufficient time to consider their participation, they were screened and randomised while in hospital.

Discussion

Conclusion

This pilot study has shown that a RCT of emollients for the prevention of eczema in babies is feasible and that emollients as an intervention to prevent eczema look promising. These results have been used to inform the design of a definitive RCT (see Proposal for a randomised controlled trial to determine whether or not emollients can prevent eczema development).

Introduction

The intervention in any prevention study needs to be highly acceptable to participants, otherwise there is a high risk of poor adherence and low-quality data. Therefore, parental acceptability of the emollient is a particularly important issue for the main RCT. The pilot RCT clearly showed that parents preferred a cream/gel emollient rather than an oil or ointment. There is a lack of evidence regarding the effect of many of the commonly used emollients on the skin barrier, although it is reported that some emollients, including aqueous cream and olive oil, can adversely affect the skin barrier. 202,203 To inform the decision on which emollient(s) should be included in a large definitive RCT of emollients for preventing eczema, further data on the emollients were required.

Therefore, we undertook a UK parent preference study of potentially suitable emollients and the two most popular emollients were taken forward to undergo a series of tests to assess their effect on skin barrier function.

Choice of emollients to be included in the preference study

To avoid emollients with ingredients known to be potentially harmful, we excluded emollients that contained the emulsifier sodium lauryl sulfate, as this has been shown to damage the skin barrier. 203 We also avoided emollients that contain parabens or protein residue that could result in sensitisation (oat, soya or plant oils) which excluded some popular emollients, for example Aveeno^® (Johnson & Johnson Ltd) as it contains oatmeal. We excluded spray emollients as they contain propellant, the effects of which on the skin barrier are not understood, and formulations containing special additives, such as ceramides, because they are relatively expensive. The following four emollients were chosen: Doublebase gel, Diprobase cream, Oilatum^® Junior cream (Stiefel, GlaxoSmithKline Consumer Healthcare), and Diprobase ointment (Bayer Plc). Diprobase cream and Oilatum Junior cream both contain the emulsifier cetostearyl alcohol, which is a commonly found in emollients, and its inclusion in the main RCT would permit some degree of generalisation across other emollients. Doublebase gel, which was popular in the pilot RCT, contains isopropyl myristate as the emulsifier. Diprobase ointment absorbs quite well relative to other ointments so was included.

Parent emollient preference study

Effects of emollients on the skin barrier function

Summary of emollient choice studies

These studies allowed us to identify two emollients, Doublebase gel and Diprobase cream, that are popular with parents and do not damage the skin barrier and so can be recommended for use in the main RCT.

Summary of pilot studies

We have carried out a series of studies to enable a definitive RCT of emollients for the prevention of eczema to take place.

A pilot RCT, which demonstrated that parents were willing to participate, willing to adhere to the emollient regimen and found the intervention acceptable. Parents in the control group were prepared to refrain from using emollients. The pilot study also provided useful insights into the best ways to recruit to a main RCT:

• Although not powered to detect a clinical difference, the pilot study showed a 50% reduction in the chances of developing eczema in the emollient group (RR of 0.50, 95% CI 0.28 to 0.90; p = 0.017).

• A parent preference study of emollients showed that Doublebase gel and Diprobase cream were the most popular choices with parents.

• The effects of these two emollients were then assessed for their effects on skin barrier function and tests showed that they are suitable for inclusion in the main RCT.

The definitive RCT was funded by the NIHR HTA1 and started in November 2014.

Participants

Participants will be newborn babies at high risk of developing eczema (on the basis of having a parent or sibling with current or previous doctor-diagnosed asthma, eczema or allergic rhinitis). Family history of atopic disease is still the most consistent, most common and strongest risk factor for predicting the development of eczema; the risk of a child developing eczema increases 1.5–4 times when there is a family history of atopic disease. 207,208 Recruiting centres will be chosen to ensure a representative range of socioeconomic and ethnic group mixes to help with external validity.

Setting

This multicentre trial will be primarily research nurse-led and will take place in a total of eight centres. There will be four main study sites (in Nottingham, Sheffield, Bristol and London), each with a part-time trial-funded research nurse to recruit approximately 1000 families (250 per centre) over 2 years. This is comparable with the recruitment rates achieved in the pilot study.

Approximately 30 centres in the UK expressed an interest in being involved with this trial, but to keep the trial co-ordination manageable, we plan to only include a further four centres with Medicines for Children Research Network (MCRN) and Comprehensive Local Research Network (CLRN) research nurses to recruit the remaining 300 families. However, this high level of interest does mean that if recruitment is slower than expected, we are able to add in further centres. We have an excellent track record of recruiting into national trials in collaboration with CLRN/MCRN nurses.

We will recruit a cohort of pregnant women from the community for this prevention study. Our pilot study recruited successfully, ahead of target and allowed us to refine our recruitment strategies. We will work with the PCRN to identify GP surgeries to send invitation letters to pregnant women in their practice. We will display posters advertising the trial in hospitals (particularly in antenatal areas and delivery suites) and advertise in local media.

Inclusion criteria

• Participant (i.e. the newborn baby) must have a parent or sibling with a history of eczema, allergic rhinitis or asthma.

• Infant in overall good health.

• Mother at least 16 years of age at delivery and capable of giving informed consent.

Exclusion criteria

• Preterm birth (defined as birth prior to 37 weeks’ gestation).

• Sibling already randomised to this trial.

• Major congenital anomaly.

• Significant inflammatory skin disease at birth (except seborrhoeic eczema).

• Any immunodeficiency disorder or severe genetic skin disorder.

• Any condition that would make the use of emollients inadvisable or not possible.

Primary outcome

The primary outcome will be the proportion of infants with a diagnosis of eczema measured by 1-year period prevalence of eczema when the child reaches 2 years of age. Rationale: we will use standard diagnostic criteria for those < 4 years of age (the UK refinement of the Hanifin and Rajka diagnostic criteria)7 which are: must have an itchy skin condition in the last 12 months, plus at least three of:

1. history of flexural involvement

2. history of a generally dry skin

3. history of atopic disease in a first-degree relative

4. visible dermatitis as per photographic protocol. 209 These robust and validated criteria are applicable to a range of ethnic groups.

They are commonly used to confirm diagnosis of eczema in treatment trials and are the most widely used criteria for epidemiological studies. 177 To reflect the chronicity of eczema that may vary across seasons, the criteria refer to signs and symptoms present over the past year. A 1-year period prevalence at age 2 years has been chosen in order to exclude transient eczematous rashes that are common in the first year of life, most of which are probably irritant (saliva and nappy) and not true AE.

Secondary outcomes

1. Parental-reported 2-year cumulative incidence of eczema according to the UK refinement of the Hanifin and Rajka diagnostic criteria for eczema, not including the skin examination. Rationale: even though eczema in the first year will contain more ‘noise’, an incident measure may still be appropriate to reflect the ‘force’ of incidence as opposed to prevalent disease which may reflect factors of disease chronicity as well as incidence.

2. Point prevalence of eczema on skin examination (visible eczema as per photographic protocol) at 2 years. Rationale: a blinded outcome measure that takes less than 1 minute to perform and might reflect more persistent disease.

3. Time to onset of eczema (parental report of doctor diagnosis) and first topical corticosteroid prescription. Rationale: to determine whether or not emollients can delay the onset of eczema.

4. Severity of eczema; if the child has been diagnosed with eczema, the Eczema Area and Severity Index (EASI) will be carried out by the research nurse at 2 years and Patient-Oriented Eczema Measure (POEM) scale completed by the parents at each contact point. Rationale: any reduction in severity would mean significant health benefits and would reduce NHS costs.

5. Two-year cumulative incidence of parental-reported wheezing. 177 Rationale: early wheezing has been shown to be a predictor of allergies in later life.

6. A within-trial cost-effectiveness analysis will be conducted from the perspective of the NHS at 2 years, with a family perspective captured separately. Longer-term analysis will be conducted using economic modelling as appropriate. Health-care resource use and costs will be collected regularly throughout the trial and used to inform this analysis. Costs associated with both the emollient (estimated to be £60 per child) and other NHS care for the treatment of eczema will be collected. For the family perspective, time off school and work will be monitored, along with expenditure on skin care products and other eczema-related items. Rationale: to obtain an understanding of whether or not the proposed intervention offers value for money in the NHS and for families.

7. Effect of the intervention on QoL will be assessed using the following scales:

   1. parental/main carer health-related QoL at baseline and when the child is 2 years old, measured using the European Quality of Life-5 Dimensions-3 Levels (EQ-5D-3L)

   2. Infant’s Dermatitis Quality of Life Index (IDQOL) at 2 years old. 210 The IDQOL has been developed for use in children < 4 years and has previously been used in economic evaluations of interventions for eczema. 211

   3. an eczema-specific instrument for capturing utility in children with eczema at 2 years212 – this scale produces a utility score and has four health domains (activities, mood, comfort and sleep). The difference in utility between the two groups will be estimated by taking account of the difference in the number of children that develop eczema and any difference in eczema severity (children do not have eczema at the point of randomisation). Subsequently quality-adjusted life-year (QALY) estimates will inform the longer-term economic model based analysis. Rationale: to see whether or not any benefits in disease frequency are accompanied by improvements in QoL for the child and families.

8. Incidence of possible adverse reactions to the emollient including skin infections, occlusion, folliculitis and slipping owing to emollient use. Rationale: although extremely safe, a precise estimate of possible adverse effects is essential to balance benefits and harms.

Saliva samples will be collected from all participants for DNA extraction to allow genotyping for the most prevalent loss-of-function mutations in the gene encoding filaggrin. This will allow a stratified subgroup analysis to test the hypothesis that individuals with one or more filaggrin null mutations show the most significant reduction in AE incidence as a result of intensive emollient therapy.

Tertiary (long-term) outcomes

Long-term follow-up is important for a chronic disease like eczema. The main part of the trial, which includes the primary and all secondary outcomes, assesses the infants up to the age of 2 years. These data will be analysed and reported when complete. Children will continue to be followed up by questionnaire at 3, 4 and 5 years old to determine:

• 1-year period prevalence of doctor-diagnosed eczema and parental-reported eczema using the UK diagnostic criteria (without the physical signs for symptoms in the last year) at 3, 4 and 5 years old

• incidence of allergic rhinitis and asthma assessed by parental report of doctor diagnosis and parental questionnaire at 5 years old

• severity of eczema at 3, 4 and 5 years old (POEM score)

• parental/main carer health-related QoL when the child is 3, 4 and 5 years old using EQ-5D-3L.

Screening, consent and randomisation

Screening will mostly take place during the third trimester of pregnancy, when consent will be obtained. Once the family have notified the research nurse that the baby has been born safely, or the research nurse has contacted the maternity unit to establish this, and eligibility is confirmed, randomisation will take place. To ensure the research nurses remain blinded, the trial administrator will notify parents which group they have been assigned to. The emollient will then be posted to parents along with a booklet and a web link to a video demonstrating how to apply the emollient correctly. There is a need to start the intervention soon after birth (within a maximum 3 weeks of delivery) because it is thought that the stratum corneum is not mature at birth and because imbalances in the immune system [Th1 vs. T helper cell type 2 (Th2) shift] probably occur quite early at a critical period of development of the thymus gland. 213,214

Screening after delivery

If parents become aware of the trial after delivery, then providing they can start within 3 weeks of birth, they will be accepted onto the trial.

Assessment and follow up to 2 years old

Parents will be contacted by e-mail with secure web links for completion of questionnaires at 3, 6 and 18 months (or by post if no internet). Missing data and non-responders will be prompted by repeat e-mails and post, with resource intensive telephone calls as the last option. They will receive a telephone call from the trial administrator at 1 year and a home visit by the research nurse at 2 years. We will use the Medical Research Information Service data to get updates on the child’s health status. Parents in both groups will be advised that if the baby develops a skin problem at any point that they are concerned about then they should seek medical care. Any treatment of eczema or other skin problems will be through the child’s GP. At the 2-year visit, parents will be reminded not to reveal their allocation status to the research nurse until after the key skin examination and questions have been entered for the primary outcome. Research nurses will be asked to confirm their blinding status at the time of assessment. Table 10 shows details of the data collection schedule.

Long-term follow-up (when the child is 3, 4 and 5 years of age)

The long-term follow-up after the main trial of children at 3, 4 and 5 years of age will be conducted by web-based and postal questionnaire directly to the parents to keep the costs of this data collection to a minimum, while still obtaining key data on long-term outcomes.

Adherence

Parents in the intervention group will be asked at each contact how many containers of emollient they have used and also to report whether or not they have applied the emollient every day, most days, approximately every other day, once or twice a week or less than once a week.

Within-trial analysis

Patient-specific data collected during the trial will be used. The incremental cost per eczema case prevented, incremental cost per one point less improvement in IDQOL and incremental cost per QALY will be estimated. As cost data are inherently skewed, bootstrapping will be used to estimate CIs around the mean differences in costs and benefits. 215 An incremental cost-effectiveness analysis will be performed using accepted methods216 with data reported in a disaggregated way. Analysis of uncertainty will follow recommended practice [with results presented as cost-effectiveness acceptability curves (CEACs)].

Longer-term model-based analysis

If emollient intervention is found to be effective at 2 years, a longer-term economic model will be developed to model the economic costs and benefits of the intervention for a single birth cohort from birth to 16 years, in order to better inform policy recommendations.

An economic evaluation conducted alongside a trial is likely to provide a conservative estimate of cost-effectiveness because it captures only the short-term costs and benefits. Taking a longer time perspective is particularly important for preventative interventions, as, although the costs are incurred early (within this study in the first year of life) the benefits (if found) are likely to be felt over a number of years and are likely to be underestimated in a within-trial analysis.

Using a combination of trial data, published data and expert opinion, a decision analytic model will be developed, taking a NHS perspective. The model will be developed in accordance with current practice and guidelines. 215,217 Having identified the potential management pathway for the intervention, we will develop and verify the model structure (this will probably use a Markov chain approach). Future costs and benefits will be discounted in line with recommendations. Analysis of uncertainty will be undertaken using probabilistic sensitivity analysis with results presented as CEACs.

Sample size

There are data available on which to base an accurate estimation of the control group event rate; however, data on which to base the expected effect size are much more limited. A modified ITT analysis (complete cases only) in our pilot study showed a 50% reduction [43.4% developed eczema in the control group compared with 21.8% in the emollient group (OR 0.36, 95% CI 0.16 to 0.84; p < 0.0001)].

We have opted to assume a fairly conservative expected control event rate of 30%, rather than the approximately 40% seen in our pilot study, as this is more in line with the published data in this high-risk population. Our anticipated effect size is lower than that observed in our pilot study for five reasons: (1) the shorter intervention period and close contacts with participants in the pilot study might have enhanced treatment concordance, which would not be maintained in a national study with a longer (12-month) duration of the intervention; (2) eczema was only assessed at 6 months in the pilot, which might have included more transient cases; (3) it is possible that the halving of eczema in the pilot study was simply due to a direct treatment effect of emollients concealing mild eczema, which would not occur in a longer-term assessment at 2 years of age, when daily emollient application ceased at 1 year; (4) short-term efficacy studies typically overestimate treatment effects when tested in large-scale national trials with more heterogeneous populations and less research team contact; and (5) our participants will come from a slightly more diverse socioeconomic and ethnic background than in our pilot study.

Based on a control event rate of 30% and a conservative expected relative reduction of 30%, that is, a rate of 21% in the treatment group, with 90% power and allowing for 20% attrition, a total sample size of 1282 babies is required (Table 11). The calculation was performed using Stata/SE 11.2.

Ethics

We do not anticipate any issue with obtaining ethical approval for this study; the pilot study, which was very similar in design, was approved with no major issues. Ethical and trust/primary care trust approvals will be obtained before the study commences. There are a number of exclusion criteria that need to be established before randomisation, which will necessitate checking the records of the infant and mother; consent will be obtained to do this.

Risks and anticipated benefits

Emollients are widely used in babies in children and are generally very safe. However, it is possible the child may have a local skin reaction to the emollient or that occlusion may lead to folliculitis or skin infections. Babies will have slippery skin for a short time after the emollient is applied so parents/carers need to take extra care when handling the baby at this time. Parents will be made aware of these risks in the participant information and during the consent process, and incidence will be monitored throughout the trial.

We cannot guarantee any benefit as we do not know if this intervention is effective.

Informed consent

Written informed consent will be obtained from one of the parents before any study related procedure is undertaken.

Research governance

The University of Nottingham will sponsor this trial, which will be run according to good clinical practice and the University SOPs. An independent TSC and Data Monitoring Committee (DMC) will oversee the trial as described in the application form.

Confidentiality of data

All participant data will be handled and stored in accordance with the University of Nottingham SOPs and the Data Protection Act. 218 Archiving of the data shall be for a minimum of 7 years after the date of any related publication.

Potential sources of bias

We have thought carefully about measures to reduce potential for bias in this randomised trial. Selection bias will be reduced by allocation concealment; a web-based randomisation system will be used into which all data needed for randomisation are irrevocably entered prior to randomisation. Both groups will receive equal contact to reduce performance bias. To minimise information bias, we will use an outcome assessor (research nurse) who is blinded to treatment allocation for the primary outcome measure. Attrition bias will be reduced by regular contacts with participants and by doing an ITT analysis. We will register the trial to prevent any selective outcome reporting bias and we shall prevent publication bias by guaranteeing that all results will be published in the public domain regardless of the direction of results.

Contamination of the control group is clearly a potential risk for this study and the effect would be to diminish the estimation of the true magnitude of potential benefit. In an individually randomised study, the patient information leaflet could, even if written in a neutral and balanced way, promote the perception in the control group that emollients from birth are ‘good’ for preventing eczema. However, our pilot study of 124 families has allowed us to assess potential sources of contamination. We were open and honest with parents and explained that we truly do not know whether or not emollients are beneficial. We showed that the number of parents in the control group who frequently applied emollients to their baby was very low (8%) and that it is possible to have a relatively clean control group in this type of trial. We plan to repeat the approaches that were successful in the pilot and to introduce additional measures to help mitigate against any potential contamination. We will actively monitor the rate and degree of contamination throughout the trial. As indicated in our stopping criteria, if emollient use in controls exceeds 25% at 6 months, we will discuss the implications with the Data Monitoring and Ethics Committee and if necessary stop the trial on their advice through the TSC.

Systematic reviews of eczema prevention

Systematic reviews of eczema prevention continue to be done to a high standard by groups such as those working in The Cochrane Collaboration. Nonetheless, our overview of systematic reviews for the prevention of eczema should ideally be regularly updated so that HCPs, patients, policy-makers and guideline developers continue to have a document in which they can examine all of the commonly tried eczema prevention strategies. Those conducting systematic reviews and overviews of such reviews need to take heed of the large variation in the way dietary interventions, allergen avoidance and probiotics delivery can vary between studies. Care should be taken in judging whether or not such studies are sufficiently similar to combine in a meta-analysis. The ideal standard comparator in future systematic reviews should be normal care (no treatment). In the absence of a ‘no-treatment’ group, it is difficult to say whether or not a lack of significant difference between the effectiveness of dietary interventions is owing to neither being effective or both being effective. Systematic reviews need to use clear definitions of eczema and, when possible, undertake subgroup analyses for those with true atopy (i.e. eczema plus sensitisation to common allergens). Long-term outcomes, including disease severity, and other allergic diseases such as asthma need to be included as important outcomes in such systematic reviews. Furthermore, future systematic reviews on this topic should register the review plan on the Prospective Register of Systematic Reviews (PROSPERO) database222 in order to reduce unwanted duplication of research.

Future randomised controlled trials of eczema prevention

The key implications of the work reported here for future RCTs of eczema prevention are straightforward.

• Ensure that the population targeted is clearly identified as being high risk (family history of allergic disease) or normal risk (general population).

• Ensure that the intervention is sufficiently well described to permit others to replicate the study. The study should clearly define whether the intervention was given prenatally, postnatally or both.

• The comparator for a prevention study should ideally be no treatment to clearly establish the effectiveness of any prevention strategy.

• An incident case definition should not include lengthy disease chronicity as one of its diagnostic criteria.

• Care should be taken in diagnosing eczema confidently in the first year of life because many transient forms of irritant and seborrhoeic eczema can occur in that period.

• Outcomes for prevention studies should ideally be measured up to 5 years after birth as initial reductions in eczema incidence may be lost over time.

• Eczema severity should also be measured in such studies in order to ensure that the fraction of eczema cases prevented are the clinically important ones.

International collaboration on a ‘family’ of eczema prevention studies

We encourage others working on eczema/allergic disease prevention studies to contact this team so that learning may be shared. This applies especially to interventions designed to prevent eczema through barrier enhancement from birth, so that common outcomes can be used to permit meta-analysis within the context of systematic reviews, or prospective meta-analysis if appropriate.

What was already known about this topic?

• There have been many systematic reviews of eczema treatments published, mostly for single interventions, as well as an NIHR HTA systematic review of eczema treatments in 2000. 224

• Since 2000, there has been no similar overarching systematic review of eczema treatments.

What did this study add?

• An up-to-date summary of the RCT evidence for or against eczema treatments.

• Clear signposting of eczema treatments which require more research, with specific recommendations about future research and how this could be prioritised.

• An overview of the quality of reporting and risk of bias in eczema treatment trial reports.

Main findings

In addition to the 254 RCTs identified in the original 2000 review, this update includes an additional 287 new RCTs, making 541 RCTs in total covering 92 different interventions for treating AE.

Trial reporting was generally poor, with ‘unclear’ categories dominating the assessments: randomisation method (2% high, 36% low and 62% unclear risk of bias), allocation concealment (3% high, 15% low and 82% unclear risk of bias) and blinding or masking of the intervention (17% high, 26% low and 57% unclear risk of bias). Only 22 (8%) of trials were considered to be at low risk of bias for all three quality criterion.

Fourteen interventions or treatment approaches were felt to have reasonable evidence of benefit. These include the use of topical corticosteroids and topical calcineurin inhibitors, both for the treatment of active AE, and as intermittent proactive (maintenance) therapy for the prevention of AE flares. Other interventions including Atopiclair^TM emollient, ultraviolet (UV) light therapy, azathioprine and ciclosporin (Neoral^®, Novartis Pharmaceuticals), all had reasonable evidence of benefit compared with placebo/vehicle. Similarly, RCT evidence suggested that education may be beneficial, although the exact components of a successful programme are unclear.

There was evidence of no clinically useful benefit for the following treatment comparisons: topical corticosteroids containing antibiotics for non-infected eczema; emollient with furfuryl palmitate in children with unspecified eczema severity; cipamfylline cream, in adults with eczema on the arms of unspecified severity; Mycobacterium vaccae vaccine in children with moderate to severe eczema; probiotics for treating established eczema in children whose disease severity was not clearly described; ion exchange water softening devices, in children with moderate to severe eczema; and dietary supplements rich in linoleic acid, such as evening primrose oil and borage oil in children and adults with eczema of unspecified severity.

The trial evidence was not clear enough to make recommendations for the use of emollients to reduce severity of eczema and prevent flares or to reduce the need for other eczema treatments; topical corticosteroids in combination with antibiotics for infected eczema; wet wraps on top of topical corticosteroids; antiseptic bath additives; topical antifungals; other topical treatments such as WBI-1001 cream; topical coal tar; topical vitamin B12; vitreoscilla filiformis lysate cream; oral treatments including antihistamines, prednisolone; methotrexate; montelukast; mycophenolate mofetil; immunotherapy (desensitisation); omalizumab; mepolizumab; oral pimecrolimus; oral naltrexone; autologous blood therapy; tandospirone citrate; full spectrum light therapy; excimer laser; intravenous immunoglobulin, specialised clothing (silk or synthetic fibres with or without antibiotics); environmental interventions such as house dust mite reduction; staying in a different climate; different approaches to organisation of care such as additional visits to the doctors or nurse led clinics; support groups; E-Health management; prebiotics, dietary restrictions and synbiotics; complementary therapies such as Chinese herbal treatment; hypnotherapy; massage therapy; aromatherapy; acupuncture; acupressure; other herbal treatments; psychological therapies such as stress reduction techniques and biofeedback; and balneotherapy (salt baths).

There was a complete absence of RCT evidence for dilution of topical corticosteroids, order of application of topical corticosteroids and emollients, impregnated bandages (zinc or ichthammol paste bandages), modified bathing habits (non-antiseptic bath additives, soap avoidance, frequency of bathing) and the role of routine allergy testing.

Changes in the evidence base since the previous review in 2000

Topical calcineurin inhibitors, educational interventions, oral azathioprine and Atopiclair emollient have entered the category of ‘reasonable evidence of benefit’ since the previous review in 2000.

Some interventions have now been tested sufficiently to suggest that they are not clinically useful. These include topical corticosteroids containing antibiotics for eczema that is not overtly infected, probiotics, ion-exchange water softeners and supplements rich in linoleic acid.

As in the previous review, many dietary, non-pharmacological, complementary and other topical or systemic interventions have been investigated in small and generally poorly reported trials, resulting in inconclusive findings.

Clinical relevance of the new evidence

Outcomes

There has been a modest improvement in the number of trials, including patient-reported outcome measures (PROMs), although the results were often not well reported. The move towards more trials using the same core outcome sets as encouraged by the HOME initiative (www.homeforeczema.org)225 can only be beneficial for future clinical interpretation and evidence syntheses.

Limitations of this review

Although there is always a chance that a mainly electronic search of reference databases will miss certain RCTs owing to misclassification, we have attempted to mitigate this risk by searching both the main bibliographic databases (MEDLINE and EMBASE) and several smaller, specialist, databases (CINAHL, AMED and LILACS). Blinding of the review team to the authors of the trials was not practically possible, which may have resulted in bias when summarising qualitative aspects of the results. However, we used a standardised approach to summarising the data and in the assessment of risk of bias, and we attempted to make clear distinctions between what the studies found and our own interpretation of study findings. Given the very wide scope of this review and heterogeneous nature of participants, interventions and outcomes, it has not been possible to undertake detailed meta-analysis for single interventions. These will hopefully be conducted in much narrower intervention-specific Cochrane systematic reviews. As with all systematic reviews, the evidence presented will become out of date quite rapidly for some topics and readers are directed to our Global Resource of EczemA Trials (GREAT) database226 for newly published RCTs of eczema.

The decision to include only RCTs could imply that there is no research at all for treatments without RCTs. This is, of course, not the case and many of the treatments that are lacking in RCT evidence have been studied using uncontrolled designs. Rare treatment adverse effects could also have been missed.

Our classification of treatment options into categories such as ‘evidence of benefit to support’ is not the same as a positive recommendation for widespread use or otherwise, as that is the remit of guideline developers and depends on factors such as magnitude of benefit, adverse effects, how the treatment compares with existing active treatments, as well as factors such as availability and cost.

Implications for research

Implications for health care

The evidence base of RCTs for eczema has accelerated since the HTA systematic review conducted in 2000 and many commissioners, guideline developers, HCPs and patients can now refer to this report for a rapid summary of relevant evidence to guide everyday decisions in the treatment of eczema. In addition to the established approach for treating eczema flares with topical corticosteroids, evidence supports the value of a proactive approach for maintaining eczema remission, through the use of twice weekly topical corticosteroids, topical tacrolimus or pimecrolimus. Educational approaches have also emerged as a potentially promising intervention, although further work is needed to establish the most important components of the intervention and the most cost-effective ways of delivering education in different health settings.

Equally important is the understanding that some interventions now have sufficient evidence to suggest little or no benefit for eczema patients. These include the use of topical corticosteroids containing antibiotics when used for the management of non-infected eczema, probiotics, ion-exchange water softeners and supplements rich in linoleic acid (e.g. borage oil, evening primrose oil).

What was already known about this topic?

• Searching for RCTs is a lengthy and detailed process that is repeated by researchers around the world and uses up research time.

• There are hundreds of RCTs on eczema treatment across many different journals in many different languages.

What did this study add?

• A freely available online database holding key information on RCTs of eczema treatments, including interventions compared, numbers randomised, inclusion and exclusion criteria, and outcomes.

• The ability to browse the GREAT database by treatment category or treatment and search specifically using keywords across the whole database, or within a specific field.

• Provision of a resource that can prevent duplication of effort in searching for RCTs and systematic reviews of eczema treatments. The GREAT database will be particularly useful for people writing clinical guidelines or patient information resources, systematic reviewers and health researchers interested in eczema treatments, and for HCPs and patients seeking easy access to research evidence.

Construction and content

To be included in the GREAT database, a trial has to fulfil the following broad criteria.

• The trial report must include a clear mention of randomisation.

• The participants must have established eczema either diagnosed by a clinician or using the Hanifin and Rajka criteria232 or the UK Working Party criteria,131 or criteria very close to these.

• There must be prospective allocation of people with eczema to two or more interventions.

• The trial must measure at least one efficacy outcome.

Search strategy

To ensure that as many RCTs as possible were identified, we searched the Cochrane Central Register of Controlled Trials (CENTRAL) in The Cochrane Library and the Cochrane Skin Group Specialised Register, which includes hand-searching of dermatology conference proceedings, as well as MEDLINE (via Ovid), EMBASE (via Ovid) (see Appendix 4), LILACS, AMED and CINAHL databases from the beginning of 2000 to the end of 2010.

The searches for MEDLINE and EMBASE were based on the Cochrane highly sensitive search strategy for RCTs,231 combined with a list of known terms for eczema, intended to be as comprehensive as possible.

What was already known about this topic?

• The GREAT database holds detailed information about RCTs of eczema treatments.

What did this study add?

• The GREAT database has a fairly high level of sensitivity.

• The sensitivity of the GREAT database can be improved further, most probably through dual-filtering of references.

• The GREAT database is a valuable resource for the eczema community, although researchers conducting specific systematic reviews will still need to employ hand-searching techniques and direct contact with triallists for unpublished data.

Identifying systematic reviews of eczema treatments

The mapping of systematic reviews of AE housed at the Centre of Evidence Based Dermatology240,241 has records of reviews from 2000 onwards. This map of systematic reviews was mined independently by two authors for all Cochrane systematic reviews and non-Cochrane reviews involving eczema treatments. Any disagreements about which systematic reviews to include as per the pre-specified inclusion criteria were resolved by the arbiter before the list was finalised.

Identifying included and excluded trials from systematic reviews

Two authors independently scrutinised the included systematic reviews to obtain the list of citations of included studies. Any disagreements between the two authors about the included and excluded studies were resolved by the arbiter. Excluded studies were defined as those cited as such in the review text or a table(s). Any trials not in English and not in the GREAT database had the appropriate data extracted by a researcher with the required language skills, if further information was required.

Identifying trials from the Global Resource of EczemA Trials database

Two authors independently attempted to find all the included and excluded trials previously identified from the systematic reviews, using the search facility and exploration of the GREAT database via the browse and filter facilities. When an included or excluded trial was not found in the GREAT database, the reasons for this were explored and recorded.

The reported inclusion criteria for each systematic review were used to identify any additional trials in the GREAT database that could have been eligible for inclusion in the review, but were not in the list of included studies of that review. When a trial was identified in the GREAT database that was not in the included studies list of one or more of the systematic reviews the reasons for this were explored, consulting the full trial paper and GREAT if necessary. Any disagreements between the two author’s results were resolved by the independent arbiter (JI) before the results were analysed.

Summary of results

A search of the mapping of eczema systematic reviews240 revealed 59 eczema treatment systematic reviews published since 2000. Of these, six were Cochrane reviews242–247 and were included in the study. Five non-Cochrane systematic reviews235–239 with the most recent search end date and which matched all other inclusion criteria were included in the study and 48 were excluded. Details of the excluded reviews are available in the map of systematic reviews of AE. 240

The six Cochrane reviews included 72 RCTs on eczema treatments and no other studies. All of the reviews searched The Cochrane Library, MEDLINE, and EMBASE; three searched the AMED;243,244,246 one searched the CINHAL;246 two searched LILACS;243,244 three searched PsycINFO;243–245 and two searched the ISI Web of Science. 243,244 Four242,243,245,246 of the six reviews searched various other databases and repositories such as ongoing trials databases, the Food and Drug Administration, European Medicines Agency, and Chinese databases. The Cochrane reviews included in this study covered topical calcineurin inhibitors, antibacterials, dietary interventions, Chinese herbal medicine, psychological and educational interventions and probiotics.

The five non-Cochrane reviews included 33 RCTs on eczema treatments out of a total of 85 studies that included RCTS, cohort studies and controlled trials on other dermatoses. Four reviews searched The Cochrane Library,235–237,239 four searched MEDLINE,235–238 three searched EMBASE,235–237 one searched Current Contents,238 one searched HomInform238 and two searched PubMed. 236,238 Four235,237–239 out of five of the reviews undertook manual searching of the reference lists of included trials, reviews and other sources such as textbooks and guidelines. The non-Cochrane reviews included in this study covered homeopathy, azathioprine, topical calcineurin inhibitors and topical corticosteroids.

Assessment of the trials missing from the GREAT database

The GREAT database did not contain all the included trials for four237,242–244 of the 11235–239,242–247 included systematic reviews. There were 17 out of 105 trials in total (Table 13) that were not found in the GREAT database. The reasons for this are discussed below.

Three trials248–250 did not fit the GREAT database inclusion criteria primarily owing to not giving an adequate description of randomisation having occurred.

One of the trials did not fulfil the GREAT inclusion criteria as the trial was not aiming to treat eczema, only to try to reduce bacterial numbers. 251

One of the trials was not included in the GREAT database as it has only been published as a conference abstract, which did not report results from the trial. 252 Although abstracts of trials are added to the GREAT database if there is no full trial report found, this abstract was not added as it was not reporting the completed trial results.

Five trials identified through a pharmaceutical company trials database253 as being unpublished did not contain enough information in the trial records provided to be certain whether or not these trials appear in the GREAT database (Novartis Pharmaceuticals Corporation, Basel, Switzerland, 2005 and 2006).

Seven trials out of 88 (8%) were identified from systematic reviews and fitted the GREAT inclusion criteria, but were not present in the GREAT database. 252,254–258 Two trials were included in a review paper,257 two were additional citations for a trial that was already included in the GREAT database,254,255 and one252 was published as a conference abstract and did not report trial results.

Trials found in the Global Resource of EczemA Trials database that potentially should have been included in systematic reviews

For one of the included Cochrane reviews on antistaphylococcal interventions,243 three trials233,259,260 were identified using the GREAT database that do not appear as included studies or on the excluded studies list for the review. This study’s authors are reasonably sure from the trial reports and the reported inclusion criteria of the review, that the three trials should have been included in this review. It is possible that the review authors managed to gather additional information about these trials which resulted in them being excluded; however, as the trials are not listed as excluded studies when they were so close to inclusion, this is less likely.

Strengths and weaknesses

Using published systematic reviews to validate the database helps potential users to make informed decisions for ‘real world’ uses of the GREAT database. The disadvantage with comparing systematic reviews against the GREAT database is apparent when the inclusion criteria of the review and the database differ, leading to an apparent lowering of the sensitivity of the database.

Many treatments that the GREAT database covers have not been assessed at all during this validation as only a selection of systematic reviews were used. It is clear from the results of this validation work that the sensitivity of the database differs depending on the treatment being reviewed.

Reviewing the GREAT database against all the Cochrane eczema treatment systematic reviews available has ensured that a complete comparison of the reviews with a very rigorous methodology. The main weakness with this comparison is that authors of Cochrane reviews are expected to contact trial authors for information not present in the trial report. The GREAT database methodology does not call for this to happen owing to practical constraints such as time.

Implications for the future development of the Global Resource of EczemA Trials database

One of the most obvious areas for improvement of the database highlighted by the validation is dual-filtering of references in order to reduce errors of omission of eczema treatment RCTs. This is when two authors independently assess each reference against the inclusion criteria and then any references for which the two authors disagree are resolved, involving a third author if necessary. Dual-filtering has recently become part of the GREAT database methodology and should further improve the sensitivity of the database.

When relevant additional research questions are being addressed, such as bacterial load, then the GREAT database does not currently cater for these. Other possible additions to the GREAT database could be the interrogation of unpublished trials databases, searching of additional specialist databases, such as PsycINFO or Chinese databases, or adding all citations found for each trial, including leaving citations for conference abstracts in the database. Widening the inclusion criteria of the GREAT database beyond its current scope and whether or not to add to the process of searching and filtering for the GREAT database requires careful consideration as to its feasibility, impact and resource use.

Implications for the use of the Global Resource of EczemA Trials database

The sensitivity of the GREAT database is fairly high, but there is room for improvement. The level of true errors of omission owing to faults in the filtering process for the GREAT database was low (5%) but needs to be improved to reassure users as to the utility of the database.

It is clear that, currently, for many research questions and when broader inclusion criteria than those for the GREAT database are needed, additional specific searches of specialist databases, such as PsycINFO, and unpublished trials registers are needed in addition to searching the GREAT database.

It is important to be aware that the GREAT database offers much more than a standard bibliographic database, which only provides citations and abstracts of articles. The potential time savings resulting from the GREAT database cannot be quantified within the scope of this validation study and should also be assessed in order to measure the potential benefits of this resource. The GREAT database is also a resource that is specifically designed to be accessible for all the potential users, such as guideline writers, patient information development, methodologists, health-care practitioners and patients.

The GREAT database is sufficiently sensitive and is a resource that is quick and simple to interrogate for most users, particularly for those looking for the best evidence for specific treatments that have not been included in a full systematic review, or for people looking for new trial evidence published after the systematic review search end date. The more the GREAT database is used and added to on an iterative basis, the better it will be.

What was already known about this topic?

• Authors of RCTs are now compelled to register a protocol of the trial in one of the publicly available trials registration databases, preferably before the trial starts.

• Selective outcome reporting bias, when outcomes originally planned are either not reported or altered when reported in the final published trial report.

What did this study add?

• Not many RCTs of eczema treatments had been registered at all and even fewer had been properly registered (before trial recruitment began).

• Of those that were registered, it was difficult to assess outcome reporting bias using trial registration records, as they mostly lack the required detail about the primary or other outcomes.

• Only 5% of the eczema RCTs examined were registered a priori, gave sufficient detail about the primary outcome to judge adherence to the original registration record, and reported the primary outcome in line with the original registration (or included an explanation as to why this had changed).

Searching for published eczema trials

We used a convenience sample of all published eczema trials that have been captured in the GREAT database. 226 This open-access resource brings together information on all RCTs of eczema treatments published since the beginning of 2000 in order to facilitate future methodological research and systematic reviews. To identify randomised trials to include in the GREAT database the following resources are searched: MEDLINE, EMBASE and AMED (via Ovid), CINAHL (via EBSCOhost) and LILACS. The search strategy for MEDLINE uses the following subject terms: eczema, AE, AD, infantile eczema, childhood eczema, neurodermatitis and besnier’s Prurigo, combined with the Cochrane Highly Sensitive Search Strategy for identifying randomised trials in MEDLINE (sensitivity maximising version, 2008 revision). 269 Search strategies for the other resources are modelled on the MEDLINE strategy. The Cochrane Library270 and the Cochrane Skin Group Specialised register are also searched using the same eczema terms. The search runs from the beginning of 2000 and is updated regularly.

Inclusion criteria for the present study

Allowing 1.5 years from the ICMJE policy of trial registration coming into effect, we included all eczema treatment RCTs contained in the GREAT database that had been published between 1 January 2007 and 31 July 2011. Whether or not these trials contained safety data was not relevant to this study and so did not form part of the inclusion or exclusion criteria. We excluded RCTs that were published in abstract form only. If both an abstract and a full paper were found, only the full paper was included. (Details of all trials included in this study can be found in the full publication.)262

Identifying trial registration

If the trial registration ID number was stated in the final publication, we used this number to find the trial registration. When this number was not stated, we tried to obtain the registration record using the World Health Organization International Clinical Trial Registry Platform267 which searches a number of different trial registries throughout the world (Box 1). Two people searched for trial registration independently (AB and HN) using a combination of key words: ‘eczema’ or ‘atopic dermatitis’ plus keywords describing the trial’s interventions such as ‘pimecrolimus’ or ‘probiotics’. All registration records that met the search criteria were reviewed to find the one that matched the published trial using key information supplied in the record [e.g. the name of principal investigator (PI), funding source and trial design]. As a rule, we did not spend more than 15–20 minutes looking for each record, on the grounds that trial registration should be easy and quick to identify.

Distinguishing between ‘registered’ and ‘properly registered’ trials

All eczema treatment trials in the sample from the GREAT database, registered on the World Health Organization International Clinical Trial Registry Platform were included in the number of ‘registered’ trials. This provides the overall level of trial registration for eczema trials, regardless of whether or not they fulfilled the guidelines for trial registration correctly. In addition, we identified a ‘properly registered’ subset of these trials which could be used to assess outcome reporting bias. We defined ‘properly registered’ trials as being those trials for which the primary outcome was explicitly stated in the registration document and for which the trial was registered no later than the end of the study (last patient visit). In cases when the start of participant enrolment was stated but an end date was not recorded, we considered a trial to be properly registered if the registration was lodged within 12 months of the study start date. Only ‘properly registered’ trials were included in the evaluation of outcome reporting bias.

To take into account the amendments and possible changes that could have taken place after initial trial registration but prior to data lock, we used the website provided on the World Health Organization trial registration webpage to go to the source record for each of the ‘properly registered’ trials in the primary register and looked for additional information or updates regarding the trial.

Comparison between registered and non-registered trials

We compared registered and non-registered trials with respect to several characteristics including number of participants randomised, funding and quality of reporting. The quality of reporting was assessed using The Cochrane Collaboration’s Risk of Bias Tool272 in four different domains: randomisation method, allocation concealment, blinding (of participants, personnel and outcome assessors), and completeness of outcome data. Each trial was assessed for high, low or unclear risk of bias for each of these domains using the revised terms suggested in the updated version of the Cochrane Handbook272 to improve interpretation of the risk of bias.

Comparison of primary outcome between publication and registration among properly registered trials

For each properly registered trial we reviewed and compared primary outcomes reported in the publication with those stated in the registration. Both the primary outcome and the time frame for analysis were recorded. We chose to limit the review of outcome bias to the primary outcome for this first study owing to the importance of primary outcomes in determining the success or otherwise of a study and because detailed analysis of secondary outcomes was beyond the scope of our available resources.

We defined a discrepancy as follows:

• when the primary outcome in the published report was different from that in the registration

• when the time frame for assessing the primary outcome in the published report was different from the registration.

Sample size calculation

Sample size was determined by the number of available trials that had been published since the introduction of the ICMJE guidelines. We also estimated that we would need around 100 studies in order to estimate the proportion of registered studies (which we hypothesised would be around 50%) to within 10 percentage points. All eczema RCTs published in the last 4.5 years were included.

Data assessment

The following comparisons were made:

1. The proportion of trials registered.

2. The proportion of trials properly registered versus the proportion registered in any form.

3. The proportion of registered trials with the registration number stated in the final publication.

4. The average sample size of registered trials versus non-registered trials.

5. The proportion of trials that declared a funding source.

6. The proportion of registered versus non-registered trials with a placebo comparator.

7. The number of trials with a low/unclear/high risk of bias for each of the four domains for registered trials versus non-registered trials.

8. Number of properly registered trials with a missing time frame for the primary outcome.

9. Number of properly registered trials with a vague and/or unclear primary outcome.

Statistical analysis

We used frequency and percentages for categorical variables, and mean with standard deviation (SD) or median and interquartile ranges (IQRs) for continuous variables, as appropriate. Proportions were compared using chi-squared tests and Fisher exact tests where appropriate. Trends over time were assessed using chi-squared tests for trend. Continuous variables were compared using t-tests and Mann–Whitney U-tests for non-parametric data. A value of p < 0.05 (two tailed) was considered statistically significant. All statistics were calculated using Statistical Package for the Social Sciences (SPSS) version 16 (SPSS Inc., Chicago, IL, USA) except chi-squared tests for trend, which were calculated using StatsDirect (StatsDirect, Altrincham, UK). The protocol for this study was posted on the Centre of Evidence Based Dermatology, University of Nottingham, website273 at the beginning of the study in September 2011, and prior to data analysis.

Levels of registration of eczema treatment trials

Of the remaining 109 included studies, only 37 (34%) had been registered in any form, as shown in the study flow chart in Figure 12. In 9 out of these 37 trials, the trial registration was more than a year after the study had been completed and published (the longest time interval was almost 3 years after study completion) and so could not be defined as ‘properly registered’. Only 20 (54%) out of the 37 registered studies indicated their trial registration number in the final trial publication.

Among the 37 studies with a registered protocol, 18 studies (49%) had ‘properly registered’ according to our definition. Although there has been an increase in trial registration from 2007 onwards, the proportion of ‘properly registered’ trials was still low, reaching a maximum of 29% (4/14) in the first half of 2011. The increasing trend in the proportion of registered eczema treatment trials was significant (p = 0.003) (Figure 13).

FIGURE 13.

Registry of eczema RCT trials.

Differences between trials with and without registration

Although registered trials had a larger median sample size of 70 people (IQR 37.5–192) compared with 60 people (IQR 30–104.5) for trials without registration, this difference was not statistically significant (p = 0.405). Non-registered trials were significantly less likely to specify their funding source in the subsequent publication than registered trials (3% vs. 30%; p = 0.001) and a post hoc analysis looking at choice of comparator showed that non-registered trials were no more likely to include a placebo comparator than registered trials (70% vs. 57%; p = 0.188).

Overall, the differences in the estimated risk of bias were not statistically significant, with the exception of allocation concealment domain, for which registered trials scored marginally better: 24% (9/37) were at low risk of bias compared with 8% (6/72) for non-registered trials (p = 0.04; Figure 14).

FIGURE 14.

Risk of bias for eczema RCT trials. (a) Proportion of trials at risk of bias: registered trials; and (b) proportion of trials at risk of bias: non-registered trials.

Outcome reporting in properly registered trials and their registration records

The most common discrepancy was missing time frame information for measurement of the primary outcome: in 8 out of 18 trials, registration records did not state the timing of the primary outcome measurement. In addition, the description of the primary outcome was often vague and unclear, which occurred in 5 out of the 18 ‘properly registered’ trials (Table 15). For example, a primary outcome might be described as ‘efficacy in treating exacerbations’ or ‘stable remission’, without further detail as to how this was to be measured or assessed.

Assessment of outcome reporting bias among properly registered trials

The lack of detail and clarity of reporting for the outcomes of registered trials both from the registration record and the published trial report makes the objective of assessing outcome reporting bias in eczema treatment trials difficult. One study introduced a new statistically significant primary outcome that was not present in the registration record (original primary outcome was not significant),275 one study changed the scale measuring eczema severity from Severity Scoring of Atopic Dermatitis (SCORAD) index to Six Area, Six Sign Atopic Dermatitis (SASSAD),282 and one study changed the time frame for assessment of the primary outcome from 26 weeks to 4 weeks. 289 None of these changes resulted in a significant result being reported.

Main findings

Our study has shown that the proportion of published eczema RCTs registered on an approved trial registry has increased from 19% in 2007 to 57% in 2011. The fact that over half of the RCTs published in 2011 were registered is cause for some optimism. However, the proportion of trials that were ‘properly registered’ (lodging details of the proposed trial design prior to the end of the study, or within 12 months of the recorded trial start date and indicating the primary outcome in the registration) was considerably lower. Taking into account that our definition of ‘properly registered’ was a very generous one, this is a particularly sobering result. Overall, out of the 109 eczema RCTs examined, only five trials were registered a priori, gave sufficient detail about the primary outcome to judge adherence to the original registration record and reported the primary outcome in line with the original registration (or included an explanation as to why this had changed). That only 5% of recent eczema trials were registered correctly and with enough detail to assess outcome reporting bias for the primary outcome, coupled with the observation that non-registered or incompletely registered studies fail to highlight the shortcomings of such omissions, is bad science and a potential waste of resources. 38,291

Even when investigators had pre-registered their trial, 46% failed to include details of their trial registration in the published trial report, suggesting that investigators and journal editors do not yet appreciate the importance of such information in their trial reports. The revised version of the CONSORT statement for guidance on the reporting of RCTs now explicitly recommends that details of trial registration be reported fully and this should improve trial reporting in the future. 268

The fact that some investigators chose to pre-register their trial could be an indicator of trial quality and this was explored using the four key domains known to be associated with high risk of bias. 272 In this sample of eczema RCTs, there was a suggestion that trial quality might be improved in registered trials, but this was only significant for the domain of allocation concealment and is possibly limited by the modest sample size of our survey.

With regard to the hypothesis that eczema trials are subject to outcome reporting bias, we found some evidence of discrepancies between trial registration and trial reports; however, these were generally as a result of unclear and non-specific trial registration information rather than clear signs of biased reporting. In general, the number of correctly registered trials was so low that firm conclusions are difficult. Greater efforts could certainly be made to provide more detail regarding the primary outcome and its timing within trial registry records. It is hoped that reviews such as this one will be helpful in informing triallists of the importance and relevance of detailed information being lodged in the trial registries. It is also important that changes to the registration record subsequent to trial commencement are fed into the trial registries in a timely fashion, in order to maintain transparency throughout. The concept of threaded publications, enabling the tracking of clinical research studies from inception and the linking of all resulting publications including the raw data when these are available, is another emerging tool to ensure consistency between what was planned and what was done. 292

Comparison with other studies

Overall, the rates for registration of eczema RCTs were lower than those reported for other surveys of general medical journals and larger specialities. Mathieu et al. 293 looked at RCTs indexed in 2008 in 10 general medicine and specialty journals (cardiology, rheumatology and gastroenterology) and out of the 323 included trials, 45.5% were adequately registered, 13.9% trials were registered after the completion of study and 12% were registered with no or unclear description of the primary outcome.

A systematic review of studies examining the impact of selective outcome reporting in RCTs identified five studies and compared the trial publication with trial registration record. It found that between 40% and 62% of trials had at least one primary outcome that had been changed, introduced or omitted. 263 These rates are higher than the rates found in the current study, which might be a reflection of improvements in trial reporting subsequent to the introduction of the ICMJE policy. Alternatively, it is possible that higher rates of outcome reporting bias would have been found if more of the eczema trials had been registered, thus allowing a comparison of the published results with the trial registration. It is conceivable that, to date, only the better quality eczema trials have been registered appropriately.

Strengths and limitations of our study

This study used the GREAT database to identify and assess all eczema trials published over the last four and a half years. Using this global collection of appraised eczema trials allowed for the speedy completion of this review and the easy identification of all relevant RCTs that had been identified through a sensitive search of several bibliographic databases. Including all trials on a particular topic, rather than just those reported in specific journals or as a random sample of published trials, meant that our study is less open to selection bias. Eczema is one of the most common skin conditions and is commonly researched. More than 250 RCTs of eczema treatments have been published in the last 10 years alone. 226 It is possible that for other common skin conditions, the proportion of trials registered and possible outcome reporting bias may be different, which makes the results of this study difficult to generalise to the field of dermatology research as a whole.

Our definition of a properly registered trial as one that was registered before the end of the study was a generous one, given that the ICMJE requires that all clinical trials need to be registered in an approved publicly accessible clinical trial register before patient recruitment begins. It is likely that the bottom line figure of only 5% of all recent eczema trials that have been properly registered would have been even worse if this more stringent definition had been applied, although in many cases it was impossible for us to tell when recruitment began and when the study ended. On the point of timing, it is worth noting that some triallists registered their trials after the study had been completed, which defeats the whole purpose of prospective trial registration as a tool to prevent inappropriate post hoc analysis. Our sample size was relatively small, which affected the overall precision of our estimates of ‘properly registered’ RCTs.

Finally, this review limited its scope to an evaluation of the primary outcomes assessed in trials. Others have reported considerable outcome reporting bias for secondary outcomes, particularly in relation to the preferential reporting of statistically significant secondary outcomes. 294

Implications of our findings

The introduction of mandatory trial as required by the Journal of Investigative Dermatology is a great opportunity to improve the quality and truthfulness of trial reporting, which will in turn lead to better clinical decision-making by reducing the prevalence of spurious and misleading results and reduction in research wastage. 38 Although publication bias of whole studies is relatively well known, the phenomena of selective outcome reporting bias within studies is possibly less well understood within the dermatology clinical community. 295

Ideally, deviations from the trial registration record should be described in the published articles so that readers can interpret the results in full knowledge of the changes made. For the full benefits of trial registration to be realised, it is important that all investigators, funders, journal editors, peer reviewers, readers and the public play an active role in making full use of trial registration information and highlighting the need for transparent trial reporting. Journal referees and readers of clinical trials especially should make more use of scrutinising trial registries to note if a clinical trial has truly been registered prospectively and that the outcomes reported in the paper are consistent with what was planned. Those conducting systematic reviews are also in a good position to check on and comment on trial registration of included trials and to assess whether or not selective outcome reporting bias was likely to have occurred.

What was already known about this topic?

• The high number of scales used to measure the severity of eczema means that it is difficult to compare the results of different trials in a meta-analysis.

• The aim of HOME is to recommend a set of core outcome measures to be included in all eczema trials.

What did this study add?

• Two working meetings were held at which a combination of formal presentations and nominal group techniques were used to progress the HOME project.

• Outcome measures must meet the Outcome Measures in Rheumatoid Arthritis Clinical Trials (OMERACT) filter to be included in the core set. The filter comprises truth (validity), discrimination (sensitivity to change and responsiveness) and feasibility (ease of use, costs, time to perform and interpret).

• The four domains of clinical signs, symptoms, a long-term measure that captures flares and remissions and health-related quality of life (HRQoL) should all be measured by the core outcome set.

• The clinical signs of eczema should be assessed by an investigator and both the intensity and extent of signs should be measured.

• The following four clinical signs of eczema are essential for the scale to achieve content validity: erythema, excoriation, oedema/papulation and lichenification.

• Consensus was achieved that EASI be recommended as the core outcome measure for the clinical signs of eczema.

Methods

Results

A total of 43 individuals from 10 countries participated in the HOME II consensus study. The panel comprised an approximately equal number of males (n = 23; 53%) and females (n = 20; 47%). The majority of participants (n = 30; 71%) were from Europe (UK, n = 10; the Netherlands, n = 10; Germany, n = 5; France, n = 3; and Sweden, n = 2). The panel included six participants from Japan, three from Brazil, two from Israel, one from the USA and one from Australia. Clinical experts were the largest stakeholder group (n = 29; 67%). The panel included five consumers, five methodologists and one pharmaceutical industry representative.

Discussion

Forty-three individuals from five continents representing clinicians, patients, methodologists and the pharmaceutical industry participated in the HOME II consensus study.

The previously described307 core outcome domains of symptoms, clinical signs and long-term control of flares were confirmed by the panel. An additional core outcome domain of HRQoL was identified with broad consensus. It thus appears that the four core outcome domains of symptoms, clinical signs, long-term control of flares and QoL should be measured consistently in future AE trials. Therefore, it is clear that more than one of the currently available instruments will need to be included in the core outcome set as none cover all four domains (Table 18).

The OMERACT filter310 as a quality requirement to recommend outcome measures by the HOME group has been adapted with full consensus of the HOME II panel. In order to be recommended, outcome measures need to be valid, to discriminate between different disease states and need to be feasible in terms of ease of use, costs, performance and interpretation.

The HOME initiative will continuously review new and emerging outcome measures as they are published, such as the EASIdig (EASI score established from the digital images), a recently published digital tool for the digital assessment of severity and extent of AE, that has been shown to correlate with SCORAD index and EASI scores. 311

The panel highlighted the need to design and/or validate adequate instruments that meet the OMERACT filter to measure the four core outcome domains symptoms, clinical signs, long-term control and QoL.

Methods

Results

Consensus statement

Identification of the best instrument to measure clinical signs of atopic eczema in clinical trials

The HOME roadmap (Figure 15) has been developed as guidance on how to identify appropriate instruments to measure each of the four core outcome domains of AE. This section describes in detail how we applied the HOME roadmap for the core domain ‘clinical signs of atopic eczema’. Instruments for other domains have not yet been agreed by the HOME initiative.

FIGURE 15.

The HOME roadmap. a, For trials the scope should generally be global.

Stage 1 of the HOME roadmap

In 2007, a systematic review175 identified 20 named instruments to assess the severity of AE. Another systematic review336 summarised outcome measures in all 791 clinical trials on interventions for AE published in 1985–2010 and identified SCORAD index303 and EASI as the two most widely used instruments to measure clinical signs of AE.

Stage 2 of the HOME roadmap

To establish the extent and quality of testing of the identified instruments, we conducted a systematic review330 of inauguration and validation studies on all published instruments to measure clinical signs of AE. Out of the 16 different instruments identified to assess clinical signs of AE, only the EASI and the SCORAD index were identified as valid and reliable.

Based on predefined quality criteria also suggested by other outcomes, research initiatives such as COSMIN,337 the EASI has adequate validity, responsiveness, internal consistency, intraobserver reliability and intermediate interobserver reliability. The SCORAD index has adequate validity, responsiveness, interobserver reliability and unclear intraobserver reliability. 330 The SASSAD326 severity score and the TIS315 score fulfil some quality criteria, but the performance in other required measurement properties is unclear. 330 However, both SASSAD and TIS only have intermediate content validity (SASSAD does not capture oedema/population and TIS does not include BSA involved), which is a disadvantage compared with EASI and SCORAD index. 330 The clinical signs of AE are assessed by a physician in the EASI, SCORAD index, SASSAD and TIS. The POEM305 scale is a reliable and responsive instrument to assess the signs and symptoms of AE by the patient, but was not developed as a tool to capture objective AE signs. 330

Stages 3–5 of the HOME roadmap

At the HOME III meeting in San Diego on 6–7 April 2013,338 the international panel of 56 individuals from 10 countries including Asia, Europe, South America and the USA representing stakeholders such as consumers, dermatologists, nurses, methodologists and pharmaceutical industry specified the content required to adequately and completely assess ‘clinical signs’ of AE.

There was broad consensus that the instrument used should be scored by a physician and rate both the intensity and extent of the AE signs. Intensity should be assessed for the following signs: erythema, excoriation, oedema/papulation and lichenification.

The panel further reached consensus that there are no other essential signs to be included in an instrument. As highlighted by the systematic review (stage 2), only the objective SCORAD index and EASI include the four essential signs and performed adequately in validation studies. 17 Using a nominal group technique involving small group work, followed by whole panel discussion, the EASI was identified as being preferable to the SCORAD index to be used for the assessment of clinical signs. The main reasons for this choice were that (1) it only includes the four essential signs, thus avoiding redundancy; (2) assessment of severity of AE signs is assessed at multiple body sites, rather than at a single representative site for each sign, which can be problematic when using the SCORAD index; and (3) that the extent of AE lesions has sufficient weighting (50% of total score) in the EASI, but is not given enough importance in SCORAD index (approximately 20% of total score). In the final consensus voting patients, physicians, methodologists and pharmaceutical industry representatives agreed to recommend EASI as the core outcome for the assessment of clinical signs to be used in all future AE trials. In the final voting, 90% of the panel indicated that the EASI should be included into the core set of outcomes domains to measure clinical signs of AE, 7% voted for the objective SCORAD index, and 2% were unsure. 322

What was already known about this topic?

• There is uncertainty regarding the effectiveness of many of the commonly used treatments for eczema.

• Eczema research to date has largely been driven by the priorities of the pharmaceutical industry and academia.

What did this study add?

• By means of an explicit process, we identified the top 14 uncertainties in eczema treatment that are important to patients who have the disease, their carers and the HCPs who treat them.

• These prioritised treatment uncertainties can be used to guide researchers and funding bodies in their decisions regarding future investment of time and resources in eczema research.

Stage 1: initiation

We identified potential partner organisations and individuals through a process of peer knowledge and consultation, using the Steering Group members’ networks and through the James Lind Alliance’s existing register of affiliates. We were careful to ensure representation of the following groups: people who have had eczema; carers of people who have had eczema; and doctors, nurses and professionals allied to medicine with clinical experience of eczema.

Organisations/groups approached at this stage and who agreed to participate included National Eczema Society; Nottingham Support Group for Carers of Children with Eczema; Skin Care Campaign; BAD; British Dermatological Nursing Group; Queen’s Nursing Institute; Royal College of Nursing; Royal Pharmaceutical Society; UK Dermatology Clinical Trials Network (UKDCTN); the HOME Initiative; MCRN; Cochrane Skin Group; Primary Care Dermatology Society; Society for Academic Primary Care Dermatology Research Specialist Interest Group; and Royal College of General Practitioners. Potential partners were informed about the PSP, invited to participate and encouraged to affiliate with the James Lind Alliance.

The James Lind Alliance and Steering Group assessed organisations for any potential conflict of interest that might have led to unacceptable bias if they were to participate in the PSP. 341 Parties in which a potential conflict of interest was identified were not invited to participate.

Stage 2: consultation – collection of treatment uncertainties

The aim of this stage was to collect treatment uncertainties using the following question: ‘What question(s) about eczema treatments would you like to see answered by research?’ To enable people from a wide geographical area to take part, we collected data using online and paper surveys. 342 This methodology has the additional advantage of avoiding interviewer bias343 and interview effects. 19 We carried out the online survey (SurveyMonkey^®, SurveyMonkey Inc., Palo Alto, CA, USA) and sent a paper copy to participants on request. For practical and ethical reasons, we asked the carers of children with eczema to complete a survey on their child’s behalf. As eczema is such a common condition, if a participant reported that they were both a HCP and a patient or carer, then they were categorised as being a HCP for the summary statistics.

Participants submitted a maximum of five treatment uncertainties. The survey questionnaire was designed to capture each participant’s uncertainty about the effects of treatments for eczema. The survey text and accompanying information sheet were checked for readability and ease of understanding by lay members of the Steering Group. We provided broad treatment categories (e.g. topical, systemic, non-drug treatments) in order to guide participants in submitting uncertainties. Completion of the survey was considered to imply consent to participation in the prioritisation process. Any uncertainties registered on Database of Uncertainties about the Effects of Treatments (DUETs) but not represented in the list of uncertainties generated through the survey, were added to the list of uncertainties.

The survey was advertised through a combination of direct e-mails and newsletters to members of the partner organisations and through links on relevant websites. Individuals with eczema who had been in contact with the Centre for Evidence Based Dermatology and the James Lind Alliance were contacted directly and invited to participate.

Stage 3: collation of treatment uncertainties

The aim of this stage was to review the treatment uncertainties gathered in the consultation stage and to create a list of ‘indicative uncertainties’ (i.e. those that represent a number of similar uncertainties submitted in the survey). We first removed any non-questions (e.g. statements or comments), uncertainties not related to treatment of eczema (e.g. aetiology) or uncertainties not relating to eczema. We then reviewed the remaining uncertainties and, when appropriate, combined any similar uncertainties to create ‘indicative uncertainties’, which we refined into a standard format. 344 An example of several treatment uncertainties and the indicative uncertainty into which they were refined is given in Box 4. Any indicative uncertainties that arose from single treatment uncertainties were removed prior to the ranking stage.

Stage 4: ranking

The aim of this stage was to rank the indicative uncertainties into those that both patients and HCPs felt were important. The list of indicative uncertainties was converted into a draft survey, which was user tested by members of the Steering Group and representatives of the UKDCTN patient panel. We sent a link to the online survey (or paper copy) to all those who responded to the first survey outlined in stage 2, asking participants to select a maximum of 10 uncertainties that they considered to be the most important.

Results of the ranking exercise were then reviewed by the Steering Group and an agreed priority list established based on rank order when ordered by frequency of votes. In the event of a large difference between the number of patients and HCPs taking part, it was agreed that the ranked lists would be examined both separately and combined for all participants.

Stage 5: workshop to develop research questions

The aim of the workshop was to review the results of the prioritisation exercise and to start to develop research questions on the basis of the prioritised uncertainties.

A selection of participants, drawn from partner organisations and individuals that had contributed to the prioritisation process, were invited to take part. In addition to these patients and HCPs, researchers with experience in the design and conduct of applied clinical research were also invited to attend for this final stage, as it was felt that their expertise would be crucial in the development of well-formulated and feasible research questions. Participants were asked to complete a declaration of interests, including disclosure of relationships with for-profit organisations.

Participants were divided into four discussion groups. Each group was led by an independent facilitator and comprised broadly equal numbers of patients, HCPs and researchers. Group members received summary information about each indicative uncertainty, including details of the original wording of the treatment uncertainties (as submitted by the participants during the consultation stage) in order to provide contextual information about the topic when developing potential research questions. Each of the groups was asked to address up to four treatment uncertainties and it was anticipated that each broad area of uncertainty would potentially result in several focussed research questions.

A standard pro forma, based on a participants, intervention, comparator, outcome (PICO) format, was used to guide discussions.

Consultation and collation stages

A total of 493 participants (341 patients/carers; 132 HCPs; 20 unclear) submitted a total of 1070 uncertainties in the initial consultation stage. After removal of submissions that were either non-questions, uncertainties not related to treatment of eczema or uncertainties not relating to eczema, 718 uncertainties remained (including 61 from the UK DUETs – 55 from systematic reviews and 6 from patients). These were represented by 65 indicative uncertainties, across 17 treatment categories. Thirteen out of the 65 indicative uncertainties were removed because only one person submitted them, therefore 52 indicative uncertainties progressed to the ranking stage. The Steering Group agreed that this was a suitably representative and manageably sized list to take forward to the next stage of the process.

Ranking stage

A total of 514 individuals participated in this survey, of whom 399 (78%) were patients/carers and 106 (21%) were HCPs and 9 were unclear (1%). The Steering Group noted the imbalance between patient/carer and clinician respondents and although this was felt to reflect an achievement in ensuring patient/carer participation, they also recognised the importance of representing the views of HCPs adequately. Consequently, the ranked uncertainties were examined as both a single combined list and for the two groups separately.

Four uncertainties featured in the top 10 lists for both patients/carers and HCPs and these were immediately prioritised as being important (hereafter referred to as ‘shared priorities’). The remaining uncertainties from the two top 10 lists were assessed. Some uncertainties were combined at this stage as they were very similar and the Steering Group wanted to ensure that individual topics did not dominate the final shortlist at the expense of other important topics. Three uncertainties relating to diet were combined in this way from the patients’ list to form the single indicative uncertainty: ‘What is the role of diet in treating eczema: exclusion diets and nutritional supplements?’ Two uncertainties relating to education were combined from the HCPs’ list to create the uncertainty: ‘Which is more effective in the management of eczema: education programmes, GP care, nurse-led care, dermatologist-led care or multidisciplinary care?’ After these revisions had been made, the next five highest ranked uncertainties from the patients/carers’ priorities and the HCPs’ priorities were added to the four shared priorities, giving a shortlist of 14 priority area (for future research) (Table 28).

Workshop and resulting research questions

The 1-day workshop took place in January 2012 and involved 40 participants (11 patients, 17 HCPs, 7 researchers, 4 facilitators and 1 observer). Following brief presentations about the prioritisation process and the aims of the day, participants went into their groups to discuss the priority areas.

Overall, 6 out of the 14 prioritised uncertainties were discussed during the day and these resulted in 13 possible research questions. Details of the research questions arising from the four shared priority topics were given most consideration and these are listed in Table 28. The proposed research questions are intended to be a broad reflection of how the priority topics might usefully be interpreted and addressed, as viewed by the workshop participants on the day. They are not intended to be exhaustive in their scope or to be fully developed in terms of detail and breadth.

What was already known about this topic?

• Silk clothing for eczema has been available on prescription in the UK since 2009. It is thought that the clothing may help eczema by maintaining a constant humidity next to the skin, reducing the number of bacteria on the skin and by the prevention of scratching. However, the existing trial evidence to support its use is limited to small trials with inconclusive findings.

What did this study add?

• This trial assesses the clinical effectiveness and cost-effectiveness of silk therapeutic clothing for the long-term management of eczema in children.

Pilot and feasibility work

An internal pilot RCT will be conducted over the first 6 months of trial recruitment. Progress with recruitment and retention will be monitored monthly by the Trial Management Group. If progress is below target, strategies will be implemented to improve progress. We have also pre-specified the following stop/go criteria to be assessed by the TSC at 6 months (target recruitment = 75 participants by 6 months) (Table 29).

Adherence in wearing the clothing will be assessed in the intervention group only and will act as a trigger for concern if participants report using the clothing less than 50% of the time.

Setting and target population

Intervention group: silk clothing plus standard care

The medical device under investigation is a knitted, sericin-free silk therapeutic garment with a Conformité Européene mark for use in eczema (DermaSilk and DreamSkin). Sericin is a protein that coats the outside of silk fibres and has the potential to cause allergic reactions. Medical-grade silk (such as silk used for stitches during operations) has the sericin removed for this reason. The silk clothing will be worn at night and, when possible, during the day. Both brands will be included in the trial in order to limit commercial advantage to one particular company. Participants will receive three sets of pyjamas (long-sleeved vest and leggings or all-in-one body suits depending on the age of the child) (Figure 16). This will give sufficient garments to allow for washing between uses. The same washing instructions will be provided for both brands of clothing (40 °C machine wash or hand wash, normal washing powder can be used). Adherence in wearing the garments will be collected using participants’ diaries and follow-up questionnaire.

FIGURE 16.

Child wearing silk clothing.

The clothing will have product labels removed and be supplied using trial-specific packaging prior to distribution from the co-ordinating centre. Clothing will be replaced as required during the 6-month RCT (e.g. because they are worn out or because the child has grown). Once the trial follow-up is complete at 6 months, the garments will not be replaced. Two months after the end of the trial, parents will be surveyed about continued use of clothing and willingness of GPs to prescribe them. It will be explained to participants at this stage that it is not possible for the trial team to recommend use of the garments or to insist that their GPs prescribe them.

Participants allocated to the silk clothing will continue to use their standard eczema care (as described in Control group: standard care).

Control group: standard care

Standard advice about what clothing to use for a child with eczema will be provided (avoid wool and wear cool, loose, clothing – especially cotton and linen), but specific products will not be recommended. All participants (active and control groups) will continue with their standard eczema care in line with NICE guidance. 357 No efforts will be made to intervene or change a child’s standard eczema care unless the research nurse thinks that the skin may be infected. In this case, the research nurse will contact the child’s GP surgery by telephone and arrange for appropriate treatment of the infection to be initiated.

Participants will be asked not to start new eczema treatments (other than antibiotics for infection) unless medically indicated and will be discouraged from routinely using wet or dry wrap dressings. Any changes to a child’s treatment regimen, or use of wet/dry wrapping will be recorded, but the child will remain in the trial.

If a child is currently using ‘specialist’ cotton clothing (e.g. special sleep suits with built-in mittens), the use of these garments will be recorded, but will not be grounds for exclusion. However, participants in the control group will be asked to refrain from using prescription clothing (including silk and garments used for wet wrapping) during the trial. Once the primary outcome has been collected at 6 months, children in the control group will be given a supply of silk clothing.

Outcomes

Outcome measures chosen are in line with the four domain recommendations made by the HOME initiative.

What was already known about this topic?

• A Cochrane systematic review of interventions for vitiligo, published in 2006, identified only 19 eligible RCTs, covering eight broad treatment categories. Most were small and of relatively poor quality.

• The review found some evidence to support existing therapies for vitiligo, but the usefulness of the findings were confounded by the different designs and outcome measurements as well as the lack of QoL measures and adverse effect reporting in the source studies.

What did this study add?

• There has been a marked increase in the number of RCTs assessing interventions for vitiligo; 38 new studies were identified for inclusion in this updated review.

• The quality of design of vitiligo RCTs has improved somewhat over the past few years, but many RCTs are still poorly designed and reported.

• In vitiligo RCTs, outcomes are assessed in many different ways, making intertrial comparisons very difficult and meta-analysis almost impossible. Future RCTs should aim to focus on a core set of outcomes, ideally including patient-centred outcomes.

• Light therapies, either alone or in combination with other treatments, were the most commonly assessed interventions. However, studies are needed to establish the dosage, safety and long-term efficacy of NB-UVB monotherapy in treating vitiligo.

• Combination therapies, particularly those that include a light source, show some promise in treating vitiligo, but larger studies are needed.

• Topical tacrolimus appears to have a similar effect to topical corticosteroids, but with a better safety profile and warrants further study to establish whether or not this intervention is a viable, cost-effective alternative.

• Randomised controlled trials assessing psychological interventions and cosmetic camouflage are lacking.

Causes

There are several hypotheses for the cause of vitiligo, none of which fully explains the disease. Several factors, including autoimmune, biochemical, genetic, neuronal and environmental, may interact to contribute to its development. 370–372 Although evidence indicates an important role for genetic factors,373–375 it also seems that triggers such as trauma to the skin, hormonal changes and psychological distress may need to be present for the disease to become apparent. 376 A number of observations suggest that vitiligo is an autoimmune disease, such as its association with other autoimmune diseases (e.g. pernicious anaemia, thyroid disorders, diabetes mellitus and Addison’s disease377) and the presence of antibodies directed against melanocytes in affected people.

Impact

Although neither life-threatening nor symptomatic (except that depigmented patches burn easily when exposed to the sun), vitiligo can be cosmetically and psychologically devastating,378 resulting in reduced self-esteem,379 poor body image,380 difficulties in sexual relationships381 and feelings of social isolation and stigmatisation. 382 Some people with vitiligo experience discrimination in employment, particularly in jobs where they have to deal with the public. 383 The effect of vitiligo on QoL is often overlooked,384 although more recent studies have started to address this. 385

Typical treatments

As yet, there is no cure for vitiligo nor any effective method of stopping its spread. Several interventions have been used in the treatment of vitiligo, including pharmacological agents (e.g. topical, intralesional or systemic corticosteroids; topical and systemic immunomodulators; topical calcineurin inhibitors; and vitamin D analogues); various forms of phototherapy [e.g. UV light A (UVA), NB-UVB and broadband UV light B (UVB), psoralen and ultraviolet light A (PUVA), various forms of laser, and monochromatic excimer light]; surgical procedures (grafting, melanocyte transplantation, micropigmentation); cosmetic measures (depigmentation, cosmetic camouflage, fake tan); complementary therapies; and psychological interventions. Many published studies describe combination therapies, usually combining a light source with another form of treatment.

Why it was important to do this review

When the original systematic review was published in 2006,366 there were only 19 eligible published RCTs and no systematic review covering all available interventions for vitiligo. The 2010 update367 was needed to assess new interventions, to highlight the gaps in research and the need for better designed and powered studies, thus informing clinical decisions and future research priorities. 344 The findings of the 2010 update are summarised in this chapter.

Having published the updated review in 2010, we have also embarked on an additional update, the initial findings of which are included in this chapter, although formal analysis has not yet been completed.

Criteria for study selection

The review included all RCTs of interventions for vitiligo, including participants of all ages and with any type of vitiligo.

Primary outcomes of interest were:

• QoL, measured using a validated tool

• percentage repigmentation of vitiliginous skin.

Secondary outcomes of interest were:

• cessation of spread of vitiligo or stabilisation of the disease, defined as:

  • no increase in the size of individual vitiligo patches measured objectively

  • no new lesions appearing, despite no improvement in existing patches resulting from treatment

• long-term permanence of repigmentation resulting from treatment (at least 2 years of follow-up)

• adverse effects.

Search methods for identification of studies

Full details of the search strategies for this review can be found in the full version, accessible from The Cochrane Library. 367

Electronic searches for published trials were carried out in the following databases: Cochrane Skin Group Specialised Register; CENTRAL in The Cochrane Library; MEDLINE; EMBASE; PsycINFO; CINAHL; LILACS; AMED and ISI Web of Knowledge.

Ongoing trials were identified using a number of online registers. The bibliographies of all included and excluded studies and reviews were also checked to identify further trials.

For all search methods, studies in other languages were included if they met the criteria and could be translated into English.

Data collection

Two of the reviewers checked the titles and abstracts identified in the search, independently assessed the full text of all studies of possible relevance and decided which trials met the inclusion criteria. Any disagreements were resolved by discussion and a decision made by consensus.

One reviewer extracted data from all the new studies, which were also allocated randomly to four of the other reviewers who independently extracted data. One of three reviewers checked the data extraction forms for discrepancies, ensuring that they did not check their own forms. When discrepancies could not be solved by reference to the text of the studies, differences were resolved by consensus.

At least two reviewers independently assessed risk of bias for the new studies identified in the updated search. The following aspects of studies were considered:

• the method of generation of the randomisation sequence

• the method of allocation concealment

• who was blinded or not blinded

• how many participants dropped out of the study overall and whether or not participants were analysed in the groups to which they were originally randomised (ITT).

Data analysis

Details of the SAP can be found in the full version of the review. 367

We expressed the results as RR with 95% CIs for dichotomous outcomes and difference in means with 95% CI for continuous outcomes.

When there were multiple intervention groups within a trial, pairwise comparisons were made of similar active interventions versus no treatment, placebo or another active intervention. No crossover trials were identified in the review. Internally-controlled trials were analysed using appropriate techniques for paired designs (e.g. for continuous outcomes using Wilcoxon signed-rank test or paired t-test; or for dichotomous data using McNemar’s test) when available and were not pooled with studies of other designs. However, when paired data could not be extracted from the papers, we presented the data using non-paired methods. If participant dropout led to missing data, we conducted an ITT analysis. Trial authors of studies were contacted to provide missing statistics such as SD. Statistical heterogeneity was assessed using I². Data were synthesised using meta-analysis techniques if I² was < 80%.

Publication bias was not tested as insufficient data were available for similar types of interventions.

Data synthesis

For studies with a similar type of active intervention, a meta-analysis was performed to calculate a weighted treatment effect across trials, using a random-effects (DerSimonian and Laird) model. When it was not possible to perform a meta-analysis, the data were summarised for each trial. If raw data could not be extracted, we extracted the results from appropriate statistical analyses presented in the paper and reported these in the review. We considered p < 0.05 as statistically significant. Data relating to adverse effects were described qualitatively as reported in each trial. Subgroup analyses were not conducted as no substantial heterogeneity (I² > 50%) existed between studies for the primary outcome.

Insufficient numbers of studies were included in the meta-analysis for sensitivity analyses to be feasible.

Randomised controlled trials included in this review

A total of 57 studies (3139 participants) met the inclusion criteria (19 studies from the original review, 38 new studies). Of these, 18 were ‘within-participant, left/right comparison studies’, one was a ‘left/right comparison study’, three were ‘within-participant studies’ and the rest were parallel group studies. All included studies allocated either participants or bilaterally symmetrical lesions to treatment groups in a random manner. Twenty-nine of the studies were placebo controlled and the rest used active comparators. The number of participants in each study varied from 8 to 596. The majority of studies (n = 38) recruited < 50 participants and most (n = 40) included participants with symmetrical vitiligo. All but one of the studies386 included both male and female participants and three studies recruited only children. 387–389 Most of the studies were carried out in Asia (n = 26) or Europe (n = 19), with fewer undertaken in the Americas (n = 10) or Africa (n = 2). 367

A total of seven studies were excluded. 390–396 The reasons are documented in the full version of the review (see ‘Characteristics of excluded studies’ table367). In addition, 13 studies were omitted on the basis that they were awaiting classification and five were ongoing at the time of the review. 367

Interventions

This update found many studies assessing interventions not described in the original review.

• New topical interventions included pimecrolimus,397,398 tacalcitol,399–401 5-fluorouracil (5-FU),402 topical lactic acid,403 catalase/dismutase404 and an antioxidant and mitochondrial stimulating cream. 405

• New oral interventions included Zengse pill with oral cobamamide plus topical psoralea tincture,406 Polypodium leucotomos,407,408 levamisole (Ergamisol^®, Janssen Pharmaceutica),409 minipulses of betamethasone,410 azathioprine,411 oral antioxidants412 and phenylalanine. 405

• New light therapies included monochromatic excimer light,400,413 broadband UVB410,414 and erbium-doped yttrium aluminium garnet laser. 402

• New surgical interventions included minipunch and split skin grafts415 and transplantation of autologous melanocytes. 416,417

• There was also one new study assessing psychological interventions. 418

The use of a light source, either as monotherapy or in combination with oral or topical photoactive chemicals (i.e. psoralens or khellin), or with other interventions, is the most common method used in practice. 419 Light therapies were also the most common intervention assessed in the included studies (42 studies, 13 of which were in the original review and 29 new studies). Two new studies assessed topical steroids as monotherapy. 404,420

Outcomes

All studies assessed repigmentation as an outcome. No two studies used exactly the same method of measuring the amount of repigmentation, but the majority rated the percentage repigmentation in categories (e.g. 0–25%, 25–50%, 50–75%, > 75%). Many authors made an attempt to define the amount of repigmentation that was considered ‘successful’ but opinions varied considerably. Repigmentation was usually assessed visually by a clinician, using photographs of the lesions. However, some studies used planimetry (tracing lesions on transparent paper) and some analysed photographs of lesions using morphometry analysis.

Small numbers of studies measured cessation of spread of vitiligo and/or improvement in QoL, the latter using Skindex-29, the Dermatology Life Quality Index (DLQI), or the Children’s Dermatology Life Quality Index (CDLQI). The duration of the studies varied widely, from 3 weeks to 3 years. Within these periods, some studies continued treatment until an optimal response was achieved.

Risk of bias in included studies

To assess risk of bias in the included studies, we considered the generation of the randomisation sequence, allocation concealment, blinding and losses to follow-up. The method of generation of the randomisation sequence was deemed adequate in 32 of the studies.

Although some studies were described as ‘double-blind’ or implied double-blinding, not all of them provided details as to how blinding was maintained. Many studies were within-participant comparisons of different interventions so it was not possible for the participants (and sometimes the clinicians) to be blinded. Blinding of participants was also not possible in some other studies because two different modalities of intervention were being assessed (e.g. oral vs. topical interventions). In seven studies, participants, clinicians and assessors were all blinded. In 14 studies, blinding of participants, clinicians or outcome assessors was not stated. 367

Overall, 16.6% of all study participants included in the review were lost to follow-up. In 23 of the included studies, data were analysed on an ITT basis. This was either because there were no losses to follow-up (16 studies) or because data from dropouts were included in an explicit ITT analysis. For two of the studies it was unclear whether or not an ITT analysis had been performed. 410,421

Effects of interventions

Not all of the 57 included studies reported on the outcomes of interest for this review. In general, the extent of variation in the ways in which vitiligo was measured, as well as the number of combination interventions assessed, prevented data pooling. Analyses for this review were therefore based primarily on individual studies, with the exception of one meta-analysis.

Many of the trials had control arms in which none of the participants improved. This resulted in large RRs with very large associated 95% CIs, exacerbated by the relatively small participant numbers in many of the trials. Analyses for this review were therefore based primarily on individual studies, with the exception of one meta-analysis of two studies assessing the combination of tacrolimus ointment and 308-nm excimer laser367 (Figure 17).

FIGURE 17.

Analyses reporting statistically significant treatment differences on the outcome of percentage repigmentation > 75%. df, degrees of freedom; M–H, Mantel–Haenszel.

Primary outcomes

Secondary outcomes

Implications for practice

Owing to the small numbers of participants and heterogeneity of design of the studies in this review, it is difficult to make firm recommendations for clinical practice. However, a number of implications may be considered:

• There is moderate evidence for the use of topical corticosteroids, although long-term use is likely to lead to adverse effects. Topical non-steroidal immunomodulators may be a promising alternative, although the theoretical risk of skin cancer must be considered for long-term combination use with phototherapy.

• Most of the studies that assessed combination interventions employed light therapies (UVA, PUVA or UVB). In general, combination interventions were superior to monotherapies.

• There is limited evidence for the benefit of oral G. biloba.

• There is limited evidence for the benefit of split-thickness skin grafts plus PUVAsol. Surgical therapies can be effective for small areas in people with stable disease (bearing in mind the risk of adverse effects with suction blister grafts).

• There is a lack of formal evidence supporting the use of cosmetic camouflage, depigmentation (in severe cases) or psychological interventions.

Implications for research

Since the original review was published in 2006, there has been an increase in the number and quality of RCTs assessing interventions for vitiligo. However, it is still not clear which are the best interventions for this condition. Until the causes of vitiligo are better understood, treatments will continue to be based on the many theories that exist for this disease. Although there has been a noticeable increase in basic scientific research, more investment is needed so that new discoveries in experimental models can be systematically advanced to preclinical and clinical stages.

Summarised below are the recommendations that emerged from this update of the review. Many of the recommendations relate to the design and reporting of future clinical trials for vitiligo and are explored in more detail in a separate publication. 431

• Standardised methodologies for describing and classifying vitiligo and for assessing the effect of interventions need to be developed and used by trial investigators.

• Study design should take into account variations in participant features, including skin colour, patient age or duration of the disease, extent and type of vitiligo, as well as site of affected areas.

• In order to assess the efficacy of an intervention, trials should last for at least 1 year.

• The following outcomes should be considered in the design of future studies:

  • cessation of spread of vitiligo (ideally with 2-year follow-up)

  • patient-centred outcomes (e.g. HRQoL measures).

• Studies are needed to establish the dosage, safety and long-term efficacy of NB-UVB monotherapy in treating vitiligo.

• Topical tacrolimus appears to have a similar effect to topical corticosteroids, but with a better safety profile; however, this finding must be replicated in further studies to establish it as a viable, cost-effective alternative.

• The combination of two active interventions (especially those incorporating some form of phototherapy) have shown promise and require further investigation.

• A greater number of well-conducted RCTs of complementary or psychological interventions are required.

• Studies are needed on the use of cosmetic camouflage, particularly with regard to the long-term effect on QoL.

Update the review beyond 2010

In addition to the review commissioned as part of this Programme Grant, we have also undertaken a further update of the Cochrane review432 and the results should be available in 2014. We have identified a further 39 RCTs for this latest update, giving a total of 96 RCTs to be included in the review. Many of these (n = 41) are within-participant studies and most have < 50 participants.

New interventions assessed by RCTs in this update include transplantation of autologous cultured or non-cultured melanocytes, autologous non-cultured hair follicle outer root sheath cell suspension, autologous non-cultured epidermal cell suspension, carbon dioxide laser, helium neon laser, broadband UVA, tetrahydrocucurminoid cream, and alpha-tocopherol (vitamin E) plus NB-UVB. As in the previous update, many of the new studies assess interventions in combination, usually including a form of light therapy. Nine of the new RCTs assess surgical interventions, reflecting an increasing interest in this area of treatment. Just over half of the studies report our primary outcome of > 75% repigmentation and more than half report adverse events.

There has been a huge increase in the number of vitiligo trials in recent years, which is reflected in the large number of registered ongoing studies (n = 44). Two of these studies are investigating afamelanotide (Scenesse^®, Clinuvel), which was one of the top 10 research uncertainties identified in the vitiligo PSP (see Vitiligo Priority Setting Partnership).

What was already known about this topic?

The pharmaceutical and medical technology industries and academia play an essential role in developing new treatments.

• However, it is increasingly recognised that patients and HCPs have a key role to play in identifying important areas for research.

• The priorities of industry and academics are not necessarily the same as those of patients and clinicians.

• It is essential, therefore, that researchers and funding bodies are aware of the needs of patients and clinicians.

What did this study add?

The research areas identified provide direction for future research activity by highlighting the questions of importance to patients and HCPs.

Stage 1: initiation (January 2009 to February 2009)

The aim of this stage was to establish the vitiligo PSP by raising awareness and identifying and engaging potential stakeholders. The following professional bodies and patient support groups were approached during this stage: BAD; UKDCTN; NHS Evidence-Skin Disorders; Cochrane Skin Group; British Dermatological Nursing Group; Changing Faces; British Red Cross Camouflage Service; Skin Care Campaign; Primary Care Dermatology Society; Vitiligo European Task Force; and British Association of Skin Camouflage and Vitiligo Society. Individual researchers, dermatologists, specialist nurses and psychologists with a special interest in vitiligo were also informed. Our research group, called the Steering Group, included 12 members with knowledge and interest in vitiligo.

Stage 2: consultation (March 2009 to August 2009)

The aim of this stage was to collect treatment uncertainties. An online and paper survey was undertaken that encouraged patients and clinicians to submit their questions about the treatment of vitiligo. Paper copies of the questionnaire were sent to the Vitiligo Society (n = 1268) and to BAD (n = 835). E-mails were sent to members of the UKDCTN (n = 500) and details of the project (with links to the online survey) were advertised on the websites and in the newsletters of the relevant organisations listed above. Additional treatment uncertainties were identified from existing sources of current evidence: updated Cochrane systematic review367 and the BAD guidelines for diagnosis and management of vitiligo. 434

Stage 3: collation (August 2009 to December 2009)

Submitted uncertainties were reviewed by members of the Steering Group. Questions on the aetiology, natural history and prevention of the disease (uncertainties not related to treatment) were excluded at this stage. The treatment uncertainties were then collated into groups representing an area of uncertainty. A structured format was then used to refine these uncertainties in order to aid comparison of different questions in the next stage (Table 31). For example, the refined uncertainty ‘How effective is UVB light therapy when combined with creams or ointments in treating vitiligo?’ includes submitted uncertainties on combination of NB-UVB with a number of different topical agents (such as corticosteroids, calcineurin inhibitors and vitamin D analogues).

Stage 4: ranking (January 2010 to February 2010)

The aim of the ranking exercise was to identify the 25 uncertainties that the PSP felt were the most important priorities for future research. The ranking exercise was open to all people who gave contact details during the consultation. It was also advertised via websites and newsletters of relevant organisations, as per the consultation stage. In addition, advertisements and articles were placed in The Voice magazine (GV Media Group Ltd, London) for black and ethnic minorities, the British Dermatological Nursing Group magazine435 (British Dermatological Nursing Group, Belfast) and the bulletin of The Primary Care Dermatology Society435 (The Primary Care Dermatology Society, Hatfield) to target specific groups that had been under-represented during the consultation stage.

Participants were asked to vote for up to three uncertainties they felt were most important (three individual votes), either online436 or using paper questionnaires. The order in which uncertainties appeared on the survey was randomised in order to guard against response bias.

Stage 5: final prioritisation workshop (March 2010)

The aim of this final stage was to identify the top 10 most important treatment uncertainties for vitiligo by creating consensus through a face-to-face workshop of HCPs and patients. Selected participants of previous stages of the vitiligo PSP attended this workshop. Efforts were made to ensure that equal numbers of patients and HCPs attended. The workshop was a full-day event, held at the London offices of BAD on 25 March 2010. Further details of the methods used during the vitiligo PSP are outlined in the James Lind Alliance guidebook. 437

Ethics

This project was approved by the Medical School REC, University of Nottingham, UK (ethics reference number G/2/2009).

Statistical methods

We aimed for a minimum of 100 participants in the consultation and the ranking exercise and for 20 participants for the final prioritisation workshop. This sample size was estimated on the basis of previous James Lind Alliance PSPs438 and determined by the time frame available for the vitiligo PSP. Data from all stages were stored and analysed in Microsoft Access^® 2007 (Microsoft Corporation, Redmond, WA, USA) and processed by the Steering Group members.

Implications for research

The identified uncertainties provide a steer for future research activity by guiding researchers and funding bodies to questions of importance to patients and HCPs. At this point it is important to remember that the vitiligo PSP aimed to identify treatment uncertainties. These are then used as reference to inform future research questions as developed by individual research teams. It is entirely possible that one treatment uncertainty will result in several related research questions.

Reflections on the process

It is possible to offer recommendations for future PSPs, based on the key challenges faced during this process.

• Information about the existing research evidence for the different treatments should be presented in a patient-friendly format at the beginning of the workshop. This would allow all participants to engage in the process more effectively, regardless of their background, experience or levels of expertise.

• Most of the uncertainties (during consultation) were broad and non-specific, or did not specify the comparator, the duration of treatment or the population, whereas ones from the research community were often very focused. For the purposes of priority setting, the uncertainties should be reasonably broad so that all issues are easily understood. These broad uncertainties will necessarily be developed into more detailed research questions by individual research groups. It is encouraging that some funding bodies are now reflecting this by issuing ‘broad’ themed calls in targeted areas.

Impact of the priority setting partnership

The topics prioritised here cover a broad range of interventions including topical therapies, light therapy, psychological support and new and emerging therapies. These research suggestions have been distributed to relevant funding bodies in the UK and will be used to inform future research funding priorities. The NIHR HTA funding stream was proactive in picking up these research priorities and put out a commissioned call designed to address two of the suggested topic areas (uncertainties 4 and 8 as listed in Table 32). This has led to a proposed trial assessing the use of handheld NB-UVB units, with and without topical corticosteroid, which is due to start in June 2014 [see Proposal for a trial light therapy and topical corticosteroids for the treatment of vitiligo (HI-Light)]. Another important research priority on psychological interventions for the management of vitiligo has been picked up by a group of trainee dermatologists who are currently working up a project proposal through the UKDCTN’s Trainee Network.

This PSP has also stimulated industry partners to look again at this disease and new research programmes have been started looking at the role of afamelanotide (a hormone that is thought to stimulate pigmentation in the skin) for the treatment of vitiligo.

Based on our findings, we recommend the following:

• More research on the effectiveness and safety of systemic immunosuppressants for the treatment of vitiligo such as methotrexate (Maxtrex^®, Pharmacia) and ciclosporin. Research in this field would potentially contribute to our knowledge about the aetiology of the disease, which is believed to have a strong autoimmune component372,377 (which would be addressed in uncertainty 1 as listed in Table 32).

• Evaluation of currently available and widely used treatments such as topical corticosteroids and calcineurin inhibitors in a head-to-head RCT (which would be addressed in uncertainty 7 as listed in Table 32).

• Evaluation of the clinical effectiveness and safety of NB-UVB. More detailed information is needed to answer questions such as ‘should the first line treatment for vitiligo be topical agents (topical steroids or calcineurin inhibitors) or more systemic intervention such as NB-UVB’. A factorial trial design could evaluate all three treatment options in one trial (which would be addressed in uncertainties 3 and 8 as listed in Table 32).

• Evaluation of psychological interventions by conducting a systematic review of the current literature and substantial pilot and exploratory qualitative work prior to progressing to a full RCT. Together with active treatments, psychological interventions are believed to be of great importance. More evidence is needed to establish the role of psychological support as monotherapy (which would be addressed in uncertainty 2 as listed in Table 32) as well as in combination with other treatments for vitiligo (which would be addressed in uncertainty 9 as listed in Table 32).

• Evaluation of innovative treatments such as afamelanotide and pseudocatalase, which seem to be important and promising to both clinicians and patients (which would be addressed in uncertainties 6 and 10 as listed in Table 32).

• Evaluation of NB-UVB combination therapies with topical agents, which also reflects the current research trend, shown by the Cochrane systematic review, that combination treatments seem to be more effective than monotherapies367 (which would be addressed in uncertainty 4 as listed in Table 32).

• More research into the pathophysiology and the aetiology of the disease based on the interest expressed by clinicians and patients on exploration of potential effectiveness of gene therapy and stem cells (which would be addressed in uncertainties 5 and 12 as listed in Table 32).

In recommending the above, we are not commenting on the legitimacy of the interventions that have been prioritised, but reporting what clinicians and patients identified as important research topics in order to meet their needs.

What was already known about this topic?

• There is a lack of consensus as to the best outcome measures for use in the vitiligo research, which makes it difficult to compare results of RCTs and to perform a meta-analysis.

• Patients’ views are rarely considered on this issue.

What did this study add?

• This study provided a new perspective by combining a survey of the views of patients, clinicians and researchers with data published in RCTs.

• Although repigmentation was measured in almost all trials, it was measured using 48 different scales in 54 trials.

• In the absence of an internationally agreed core outcome set, we propose that repigmentation, cosmetic acceptability, global assessment of disease, QoL, maintenance of repigmentation, stabilisation of vitiligo and side effects should be included in all future vitiligo trials.

• Further work is required to establish how best to measure these outcomes.

Systematic review of outcome measures in randomised controlled trials of treatments for vitiligo

A review was conducted of the outcome measures used in RCTs published in English and included in the updated Cochrane systematic review published in 2010. 367 Studies published in other languages were included if they met the criteria and could be translated into English. The review included RCTs published up to November 2009. All of the outcomes reported in the eligible RCTs were retrieved and the scales and measures used to monitor these were collated. Information on outcome assessors was also collected.

Two researchers extracted the data from each eligible study using a standardised pro forma. In the event of any discrepancies, a third researcher adjudicated.

Survey of the most desirable outcomes for vitiligo identified by patients and clinicians

A survey was conducted as part of the previously described vitiligo PSP. 344 Patients and HCPs were asked about their views on what outcomes should be measured in future vitiligo trials. The survey was open for submissions between January and March 2009.

Ethics

The survey was approved by the Medical School REC, University of Nottingham, UK (ethics reference number G/2/2009).

Statistical methods

Data were stored and analysed in Microsoft Excel^® 2007 (Microsoft Corporation, Redmond, WA, USA).

Systematic review of outcome measures in randomised controlled trials of treatments for vitiligo

A total of 57 RCTs were evaluated, three of which were excluded as published English translations were not available. Only 22.2% (12/54) of trials had clearly stated the primary outcome measures in the abstract or the main text. The majority of these trials defined the primary outcome as repigmentation (10/12; 83.3%). Other primary outcomes included the size of target lesions (1/12; 8.3%) and the number of new vitiliginous lesions (1/12; 8.3%).

In total, 25 different outcomes were reported in 54 RCTs on the treatment of vitiligo (Table 33). The most frequently reported outcome was repigmentation which was measured in 96.3% (52/54) of trials, followed by side effects (85.2%; 46/54), number of treatment sessions or time (31.5%; 17/54) and cumulative dose to reach repigmentation (22.2%; 12/54).

Repigmentation

Although it was the most frequently reported outcome in the RCTs, repigmentation was measured using a great variety of scales (Table 34) including grades, scores (e.g. 0–4), categories (e.g. poor to excellent, partial to complete), quartiles and other quintiles (e.g. 0–24, 25–50, 51–74, 75–100), percentages (e.g. 0–40, 40–60, 60–100) and mean difference in lesion size. Five trials (9.3%) used more than one scale to measure repigmentation.

In total, repigmentation was measured using 48 different scales in 54 eligible trials. Although 29.6% (16/54) of the trials used quartiles, 14 different scales were created including differences in the definition of quartiles and the names of the corresponding categories. For example, Kumaran et al. 420 and Bhatnagar et al. 444 both used the ‘0–24%, 25–50%, 50–75%, 76–100%’ quartiles, but Kumaran et al. 420 reported moderate improvement at 25–50% repigmentation and Bhatnagar et al. 444 at 50–75% repigmentation of vitiliginous lesions. The definition of excellent repigmentation or success varied from trial to trial and included values from ‘any repigmentation’ to 100% repigmentation of vitiliginous lesions.

Trials assessed repigmentation by clinical assessment only (17/54; 31.5%) or in combination with other methods, such as digital images, paper tracing or planimetry (29/54; 53.7%). Six trials (11.1%) used only image-based methods such as digital images, planimetry and image analysis of reflected UV photographs in assessing repigmentation. Only 3.7% trials (2/54) incorporated patient-assessed repigmentation.

Pattern of repigmentation and colour matching

Pattern of repigmentation was reported in 16.7% (9/54) of trials. This was usually briefly mentioned in the discussion section or the results section only. Perifollicular and peripheral/perifollicular patterns were mentioned in 3.7% (2/54) and 9.3% (5/54) of the trials, respectively. A total of 4% of the trials reported all patterns (perifollicular, marginal and diffuse) of repigmentation (Table 35). Change in colour and colour matching of vitiliginous lesions was assessed in 5.6% (3/54) and 1.9% (1/54) trials, respectively. These were assessed by clinicians only,444,458 a combination of clinical assessment and photographs461 or by using digital images. 404 Patient assessment of colour matching was not included in any of the trials (see Table 35). Response to the treatment in light versus dark skin types was reported in one trial (1.9%) and was measured clinically.

Cessation of disease activity

Seven RCTs (13.0%) measured the cessation of spreading of vitiligo during treatment period; only four (7.4%) stated the scale used.

Stability of gained repigmentation

Two trials (3.7%) assessed stability of gained repigmentation and development of new lesions. The timescales of the assessment of this outcome were 1 year and 20 weeks after the completion of the treatment.

Patients’ opinion on treatment efficacy

Only 16.7% of studies (9/54) reported patients’ opinions regarding treatment efficacy and degree of satisfaction with the treatment including cosmetic acceptability of the results. In particular, 7.4% of trials (4/54) assessed patient satisfaction with the treatment by asking them directly using a questionnaire or using a VAS. However, details of which questions were asked were generally lacking. Cosmetic acceptability of the results and patient global assessment were monitored in 1.9% (1/54) and 7.4% (4/54), respectively. The latter was measured in three different ways (see Table 35). Tolerability of the treatment was briefly reported in one trial (1.9%).

Quality of life

Five trials (9.3%) assessed QoL of vitiligo patients. Different questionnaires were used DLQI (3/54; 5.6%), Skindex-29 (2/54; 3.7%) and World Health Organization QoL questionnaire (1/54; 1.9%).

General health, stress, body image and clinician global assessment

One trial17 reported several patient-centred outcomes: stress, body image, self-esteem and general patient’s health. Clinician global assessment was measured using three different scales in three trials (5.5%).

Side effects and harms

Various side effects and harms, such as erythema, blistering, graft failure, Koebner phenomenon and hyperpigmentation around vitiliginous lesions, were reported in 46 out of the 54 eligible trials (85.2%). In nine trials, parameters such as full blood count, kidney and liver function tests were measured for the purpose of monitoring possible treatment side effects only and, therefore, were included in this category of outcomes. Side effects and harms were generally reported briefly in the results or discussion section of the article with limited information on frequency or severity.

Specific blood, skin parameters and other outcomes

Histopathology of vitiliginous lesions (3/54; 5.6%), presence of cytokines, proliferation and CD expression of T-lymphocytes (3/54; 5.6%), concentration of epidermal hydrogen peroxide (1/54; 1.9%), and catalase activity and lipid saturation (1/54; 1.9%) were measured as an indicator of effectiveness of compared treatment.

Number of treatments to gain repigmentation, cumulative dose and time to suction blister formation were sometimes reported in trials when relevant to delivery of the intervention (see Table 35).

Survey of the most desirable outcomes for vitiligo identified by patients and clinicians

Eighty-seven per cent (401/461) of vitiligo PSP participants suggested at least one outcome (response rate 87%). In total, 438 suggestions were made and 23.5% (103/438) of these were excluded as non-relevant, such as questions on inheritance and trigger factors for vitiligo.

Of the remaining 335 suggestions, 68% of responses were from patients, 25% from clinicians and 7% did not specify. More women responded than men (53% women; 30% men and 17% did not specify) and most were aged 30–60 years (8%, < 30 years; 50%, 30–60 years; 25%, > 60 years; 17% did not specify).

The most popular outcome among all responders was repigmentation of vitiliginous lesions (68%), followed by cessation of spreading of the disease (15%), QoL (6%), maintenance of repigmentation (4%) and other outcomes such as camouflage effectiveness and depigmentation.

Repigmentation of vitiliginous lesions

Statements such as ‘cosmetically acceptable repigmentation rather than percentage’, ‘normal-looking skin’, ‘90–100% repigmentation’ and ‘quick and long lasting (at least 2 years)’ were made by survey responders. Repigmentation on visible areas such as face and hands was also considered important (Figure 18).

FIGURE 18.

Desirable outcomes for (a) clinicians; and (b) patients.

Cessation of spread of vitiligo

Nineteen per cent of patients found cessation of the disease important compared with only 5% of clinicians (see Figure 18).

Quality of life

Nearly equal percentages of patients (9%) and clinicians (8%) found the QoL of vitiligo patients to be an important outcome. Statements such as ‘to reduce the stress and embarrassment, psychological impact, support, coping strategies and self-confidence, psychological support for children with vitiligo’ were reported (see Figure 18).

Other submissions included depigmentation and camouflage improvement.

Provisional recommendations for a vitiligo core outcome set

We believe that the findings of this review and survey are a good starting point for creating international consensus over a core outcome set for vitiligo trials.

In the interim, we propose that the following outcomes should be considered in future trials of vitiligo treatments (Table 36).

Recommendations for future vitiligo trial design

Guidelines for designing and reporting clinical trials on vitiligo have suggested that patient-centred outcomes should be incorporated into the design of future trials of vitiligo treatments, along with QoL, percentage of repigmentation, permanence of gained repigmentation and the arrest of the progression of the disease. 431 The results of this study support such a recommendation.

Although QoL has been reported in only 9% of trials, it is considered an important outcome from a patient’s point of view and was also proposed as a primary outcome in the guidelines for designing and reporting clinical trials for vitiligo. 431

To overcome the issue of different body sites responding differently to treatments, it has been suggested that studies perform stratified analyses based on body sites. 431 Similarly, revised international classifications for the stability of disease recommends that disease stability (or activity) be assessed for individual lesions rather than a person overall, as this may fluctuate by site of involvement. 441

Strengths and weaknesses of the study

The survey responders included patients with vitiligo, their parents/guardians, dermatologists and other HCPs with an interest in vitiligo. As a result, the proposed outcomes reflect a broad spectrum of views.

A convenience sample of RCTs included in the updated Cochrane systematic review was chosen based on resources and time limitations. It is possible that more recent trials have been more consistent in including patient-reported outcomes and QoL scales in line with recent recommendations. 431

Implications for research

To our knowledge, this study is the first to systematically summarise outcomes reported in RCTs of vitiligo treatments and to compare these outcomes with perspectives from patients and clinicians. It is encouraging that repigmentation appears to be a good candidate for a core outcome domain in future vitiligo trials. However, it is still unclear how this (and other outcomes) should be measured and there is no consensus over what constitutes a treatment success. Future research is needed to address how best to capture treatment success from a patient’s perspective.

An international consensus initiative to define an agreed set of core outcomes for use in future clinical trials of vitiligo treatments is now under way. Further details of this project are reported in Gaining consensus over outcomes for use in future vitiligo trials.

What was already known about this topic?

• Systematic reviews of intervention trials are hampered by a lack of consistent reporting of trial outcomes.

• Development of core outcome sets can help to prevent variation in practice and can facilitate pooled analysis of data across trials.

• An internationally agreed core outcome set for vitiligo trials is not currently available.

What did this study add?

• International agreement has been reached over the core outcome domains to be included in future vitiligo trials.

• Future trials should include:

  • repigmentation

  • side effects

  • maintenance of gained repigmentation as essential core outcomes.

• A further four items were recommended for inclusion, but were not considered to be essential:

  • cosmetic acceptability of the results

  • QoL

  • cessation of spreading of vitiligo

  • tolerability or burden of treatment.

• Further work is needed to establish how best to measure these core domains.

Study design

An international web-based electronic Delphi (e-Delphi) consensus exercise was conducted between January 2013 and September 2013. A total of 101 self-selecting participants took part from 25 countries. Participants included HCPs with an interest in vitiligo (dermatologists, n = 51; nurse, n = 1); patients with vitiligo or their carers (n = 32); and others interested stakeholders (researcher, n = 1; representatives of regulatory agencies, n = 2; journal editors, n = 14).

The e-Delphi process included three rounds, each taking approximately 2–4 weeks to complete. The first round was conducted using a questionnaire distributed by e-mail, but subsequent rounds were conducted using Survey Monkey software. Reminders were sent by e-mail 7–10 days prior to the submission deadline.

Identification of participants

Potential participants were identified through the IFPCS. Representatives of each subgroup of the IFPCS, that is, Asian, Japanese, European and Pan-American societies, contacted their local patient support groups in order to identify patients willing to participate in this process. In addition, representatives identified regulatory bodies and journal editors and invited them to participate.

All communication took place electronically and participants were given clear deadlines for taking part in each round of the three rounds. No financial incentives were provided to any of the participants and completion of round one was taken as consent to participate.

Participants were strongly encouraged to continue participation throughout all three rounds, as this was important for maintaining the validity of the consensus approach.

Preparation of the initial survey

A list of possible outcomes to be capture in vitiligo trials was compiled on the basis of our systematic review of outcomes measures reported (see Systematic review of outcome measures) and from the patient and clinician surveys reported in Survey of UK clinical practice. In addition, demographic information about the participants was captured including country and experience of vitiligo (patient, parent/guardian, dermatologist, editor, regulatory body, researcher). Copies of the survey questions used are appended (see Appendix 5).

Outcomes included in the initial survey were:

• repigmentation

• cosmetically acceptable repigmentation

• global assessment of the disease

• QoL

• maintenance of gained repigmentation

• cessation of spread

• side effects and harms

• self-esteem

• psychological impact.

Consensus criteria

Each participant was asked to rate the importance of the individual items in the questionnaire using a five-point Likert scale (very important, important, less important, not important and not sure).

In the first round, participants were also asked to list additional outcomes they considered to be potentially relevant, but which had not been included on the original list.

Consensus was defined as having being achieved if at least 75% of participants in any two categories of responders (HCPs/patients/others) were in agreement. Responses from each round were collated and analysed by VE prior to circulation of the subsequent round.

For round 2, the additional outcomes identified during round one were added as additional questions and outcomes for which there was consensus that the item was ‘not important’ as a core outcome were removed. Details of items removed at each stage were provided and participants had the opportunity to agree or disagree with that decision.

Participants received feedback on the group’s opinion for each domain (modal response and percentages for each category) and were then asked to answer the questionnaire again in the light of this feedback, with the aim of moving closer to consensus.

For the final round, participants were presented with the selected core outcome domains from round two and were asked to identify which of the outcomes were ‘essential’ in the core outcome set and which were ‘recommended’.

• Essential: must be relevant to all interventions for vitiligo and reported in all clinical trials for the treatment of vitiligo.

• Recommended: should be reported if relevant to the intervention being tested and trial design used.

Free-text answers were summarised and organised thematically.

Sample size and justification

We initially aimed to involve approximately 12 participants from each stakeholder group (clinicians, patients and regulatory bodies/journal editors), resulting in a total sample size of 34 participants. However, the final sample size was dictated by the number of responses to the initial e-mail invitation. No-one who expressed an interest in taking part was prevented from doing so.

Implications for research

The identification of core outcome domains is just the first step in developing an agreed core outcome set for vitiligo trials. Work is now required to establish the best way of defining each of these outcomes (including appropriate time frames) and to identify the most suitable outcome instruments that are standardised and well validated. In defining recommended instruments for capturing each of the seven domains, work will now take place following the roadmap outlined by the HOME initiative and using guidance by the COSMIN initiative. 464

All seven outcome domains will be collected in the HI-Light trial [see Proposal for a trial light therapy and topical corticosteroids for the treatment of vitiligo (HI-Light)], although it is possible that the specific instruments selected will not be compatible with those selected through future iterations of this international initiative. Nevertheless, by working closely with groups around the world, it is to be hoped that agreement over provisional instruments can be defined quite quickly while further validation work is under way.

This study represents an important first step for improving the quality of future vitiligo trials. Through consensus over a core outcome set, the ability to use trial findings in meta-analyses will be improved, thus facilitating better use of trial data and allowing systematic reviewers to combine data to generate clinically relevant findings.

What was already known about this topic?

• Patient-rates outcome measures are not used widely in vitiligo trials.

• The most commonly used outcome measure in vitiligo trials to date has been percentage repigmentation; however, from the point of view of people with vitiligo, this is not a very meaningful measure.

• Previous work has suggested that a measure of cosmetic acceptability of the treatment response might be more meaningful to people with vitiligo.

What did this study add?

• People with vitiligo do not like the term ‘cosmetic acceptability’, as it is impersonal and suggests that vitiligo is merely a cosmetic problem. How noticeable the vitiligo is after treatment is a more meaningful concept.

• Using an online survey and online discussion groups, we have developed a patient-rated outcome measure to assess the ‘noticeability’ of vitiligo after treatment.

Online survey

Online discussion groups

Sample size and participant selection

The sample size for this study was dictated by the time and resources available. However, we aimed to include at least 100 participants in the survey and 18–20 participants in the discussion groups in order to gather a broad selection of views.

For the discussion groups, purposive sampling was used to ensure diversity in terms of ethnicity and age within the groups. Using Statistical Product and Service Solutions (SPSS) 21 software (IBM Corporation, Armonk, NY, USA), potential participants who had responded to the online survey were split into three groups: parents/carers of those with vitiligo (n = 13), people with vitiligo aged 17–45 years (n = 45) and those aged ≥ 46 years (n = 76). Participants’ ages were taken into consideration when forming the discussion groups given the potentially greater familiarity with technology and ‘text speak’ in younger participants. 474 Invitations were sent to all parents/carers, plus a random selection from other age groups. All potential participants from non-white ethnic backgrounds were invited, as well as a random selection of those from white backgrounds. This was to enable discussion of treatment for vitiligo in the context of different skin types.

Statistical analysis

Survey results were analysed using SPSS statistics software. Results were presented descriptively and responses to open questions were analysed thematically by a researcher (ST) and checked by a second researcher (JB) for agreement. This allowed for comparison between themes emerging from open and closed questions to be made, as well as allowing for ranking of themes by popularity overall for use in the discussion groups.

Because the main objective of the discussion groups was to seek consensus regarding the most appropriate wording of the question to ask people about the response of vitiligo to treatment, the results were not formally coded in the manner of a typical discussion group. Instead, consensus reached during each discussion group was summarised at the end of each session and presented back to the participants of the group for approval. These are summarised below (see Summary of areas of consensus), with examples of key comments made by participants. Two researchers independently checked the copy of each discussion group thread to ensure that all relevant points had been adequately identified and extracted.

Although more formal methods of consensus development (e.g. Delphi and Nominal Group Technique) are the focus of much academic consideration, informal consensus groups such as those employed here are commonly employed in health-care settings. 476 To counter some of the criticisms that informal mechanisms for reaching consensus lack ‘control’, ‘focus’ and ‘scientific credibility’,476 here data are handled and analysed systematically following an adapted version of template analysis. 477,478 Template analysis utilises a hierarchical model to organise text in order to aid interpretation. In this case, each point of consensus was taken as a higher-level organising code within which to summarise and consider the group discussion. Therefore, for each point of agreement, the discussion that led towards this was considered and coded to reflect those factors that contributed towards the consensus and those that were a barrier. This process was completed for each discussion group with a final template constructed to include all statements for which agreement spanned the different groups. This mode of analysis provides greater depth in understanding consensus, both mapping when agreement occurred and charting the process and factors which generated it.

Online survey

Participants

In total, 202 responses were received. Of these, 165 (82%) were included in analyses and, of these, 154 (76%) were fully completed surveys (Figure 19).

FIGURE 19.

Participant flow diagram for vitiligo survey.

Responses were excluded for the following reasons:

• If the survey had been completed from the same internet protocol (IP) address more than once:

  • the first completely filled survey was included and the rest excluded.

  • if multiple surveys were completed fully, only the first was included.

• The same two rules applied for duplicate e-mail addresses given.

• If the survey had not been completed past question 1, it was excluded.

The only exception to these exclusion criteria was when e-mail addresses and demographic responses indicated that two different individuals had responded from the same IP address, so both sets of responses were included.

Baseline characteristics of the survey participants are summarised in Table 39.

TABLE 39 - Demographic/other characteristics of vitiligo survey respondents

More than one response could be given.

Characteristic	Online survey (N = 165), n (%)	Online discussion groups (N = 12), n (%)
Responses completed on behalf of
Themselves	149 (90.3)	12 (100)
Other
Total number of responses for those responding on behalf of others	14 (8.5)	0 (0)
Child with vitiligo	13 (92.9)	0 (0)
Spouse with vitiligo	1 (7.1)	0 (0)
Themselves and other(s)	2 (1.2)	0 (0)
Unknown	0 (0)	0 (0)
Age (years)
< 5	1 (0.6)	0 (0)
5–16	10 (6.1)	1 (8.3)
17–30	17 (10.3)	2 (16.7)
31–45	43 (26)	3 (25)
46–65	55 (33.3)	3 (25
> 65	29 (17.6)	3 (25)
Unknown	10 (6.1)	0 (0)
Ethnicity
White British	117 (70.9)	9 (75)
White Irish	1 (0.6)	0 (0)
Other white background	15 (9.1)	0 (0)
Any other mixed background	1 (0.6)	0 (0)
Indian/British Indian	9 (5.5)	1 (8.3)
Pakistani/British Pakistani	1 (0.6)	0 (0)
Bangladeshi/British Bangladeshi	2 (1.2)	1 (8.3)
Caribbean/British Caribbean	1 (0.6)	1 (8.3)
African/British African	1 (0.6)	0 (0)
Other	7 (4.2)	0 (0)
Unknown	10 (6.1)	0 (0)
Country of residence
UK	135 (81.8)	11 (91.7)
USA	12 (7.3)	1 (8.3)
Europe (excluding UK)	3 (1.8)	0 (0)
Australia	1 (0.6)	0 (0)
Asia	1 (0.6)	0 (0)
Dual: UK and other	2 (1.2)	0 (0)
Unknown	11 (6.7)	0 (0)
Duration of diagnosis(years)
6–12 months	3 (1.9)	0 (0)
1–2 years	3 (1.9)	0 (0)
2–5 years	12 (7.7)	2 (16.7)
5–10 years	26 (16.8)	1 (8.3)
> 10 years	111 (71.6)	9 (75)
Percentage of skin affected (estimate)
0–10	37 (23.3)	1 (8.3)
10–25	49 (30.8)	4 (33.3)
25–50	33 (20.8)	5 (41.7)
50–80	24 (15.1)	2 (16.7)
> 80	16 (10.1)	0 (0)
Unknown	6 (3.6)	0 (0)
Area(s) of skin affecteda
Face/neck	129 (78.2)	10 (83.3)
Body	110 (66.7)	7 (58.3)
Arms	113 (68.5)	8 (66.7)
Legs	109 (66.1)	8 (66.7)
Hands	127 (77)	10 (83.3)
Feet	113 (68.5)	9 (75)
I’d rather not say	2 (1.2)	0 (0)
Other (responses included: under arms, genitalia and hair)	35 (21.2)	2 (16.7)
Treatments used previouslya
No treatment	57 (34.5)	2 (16.7)
Topical corticosteroid	41 (24.8)	3 (25)
Tacrolimus	46 (27.9)	6 (50)
Pimecrolimus (Elidel®, Meda Pharmaceuticals Ltd)	6 (3.6)	0 (0)
Vitamin D derived cream or ointment (e.g. calcipotriol)	5 (3)	2 (16.7)
UVB	33 (20)	5 (41.7)
PUVA	20 (12.1)	2 (16.7)
Not sure	1 (0.6)	0 (0)
Other (responses Included diet changes and alternative therapies)	28 (17)	3 (25)

The majority of participants were aged between 31 and 65 years of age and had had vitiligo for > 10 years. One hundred and thirty three (80.6%) of those completing the survey were from white ethnic backgrounds and 135 (81.8%) were from the UK.