Main findings and interpretation
The overriding finding from BEEP is a lack of any clear benefit from emollients for the prevention of eczema in high-risk children. The findings are remarkably consistent across the range of eczema outcomes used and time points. Results for the main outcomes are robust to sensitivity analyses that account for missing data. In subgroup analysis, there was also no evidence of a difference in the effect of emollients on eczema prevention according to genetic risk of atopic disease, for example, presence of filaggrin gene mutations. No benefit was seen for eczema severity (which could still have been useful in terms of reducing health burden), time to onset of eczema or associated allergic diseases including food allergy, asthma or hay fever. While disappointing, the findings from this study are very important in that they indicate that perhaps skin barrier enhancement using emollients is not an effective pragmatic strategy to prevent eczema and associated diseases, despite earlier interest in this approach. The null result is important because it releases carers from the inconvenience and financial cost of daily emollient application.
Data beyond 2 years
The BEEP study is perhaps a little unusual in that the primary outcome for the 5 years study was elicited at 2 years. The original idea of the 5-year outcome was mainly to collect outcomes on other allergic diseases such as asthma that are potentially difficult to diagnose under the age of 5 years. We also sought to see whether any protective benefits of eczema prevention were transient or sustained as it is unclear from studies on probiotics to prevent eczema whether the benefits diminish over time.96 The findings of no clinically useful eczema prevention at our primary outcome time point when study children were aged 2 years were upheld in parent-reported 3-, 4- and 5-year data of doctor-diagnosed eczema and other measures. We did not find any evidence of benefit for asthma or hay fever prevention, although we cannot exclude small degrees of benefit. We did not measure additional emollient-related safety outcomes beyond 2 years. None were reported spontaneously from participants for this period. Interestingly, moisturiser use over the body continued at a higher level in the intervention group than in the control group after the 12-month intervention period ceased. Such a ‘legacy’ effect might be due to routine or a genuine perceived benefit from parents, while some usage would have inevitably reflected the need to use emollients for recently diagnosed eczema.
Health economics
The economic evaluation results are consistent with the clinical findings of the study and finds that at 2 years the use of emollients to prevent eczema is unlikely to be cost-effective. The results at 5 years, if read uncritically, could be interpreted as suggesting the intervention is cost-effective over this longer period. However, incremental costs and effects were small in both the 2- and 5-year analyses. In addition, the costs in the 5-year analysis were largely driven by the small number of participants who incurred inpatient hospital costs due to wheezing and which were unequally distributed between the two study arms. Given the findings elsewhere in this report do not find evidence of benefit for asthma, or indeed hay fever, it seems unlikely these costs were related to the intervention. When they were excluded in a sensitivity analysis the CEA results suggest that the intervention is dominated by control (i.e. the intervention is both more costly and less effective than control).
Possible signals of harms?
Given that the combination of greasy emollients and water in the bathroom is a recipe for very slippery surfaces, one of our main concerns with the use of emollients was slippages. Thankfully, slippages were not common, and none were serious. More importantly, they were similar in intervention and control group and mainly due to non-study emollients for eczema or general skin-care use.
The finding of increased parental report of doctor-diagnosed skin infections in the intervention group is an area of concern. Although none were serious, they nevertheless represent a nuisance to the infant and parents and may involve healthcare contact. Interpretation of the skin infection data is challenging though as the range of skin infections was quite wide and not coherent with the sort of specific infections such as folliculitis due to S. aureus that one would normally associate with eczema or occlusion of the skin with emollients.39 Although we recorded parental report of doctor-diagnosed skin infection, we did not verify these reports or diagnoses with the doctors concerned and whether such infections were confirmed by laboratory investigations. Ideally, more specific investigation of skin infections should be undertaken in ongoing and future emollient prevention studies.
Skin sensitisation and food allergy outcomes; although funded separately, the results of this aspect of the BEEP warrant some brief discussion as food allergy and eczema are so inextricably linked.97 Our hope was that successful prevention or postponement, of eczema onset might lead to prevention of food allergy, since early-onset IgE-mediated food allergy is thought to be directly caused by early-onset eczema.98 BEEP trial findings showed no prevention or postponement of eczema, and a possible increase in IgE-mediated food allergy. Food allergy is less common than eczema, so the trial was not adequately powered to detect changes in food allergy prevalence. This means there is significant uncertainty about food allergy outcomes. While most food allergy outcomes, including the a priori defined main food allergy outcome, showed increased food allergy or sensitisation in the intervention group, CIs were wide and, for most analyses, included the possibility of no effect. If emollient application increases risk of IgE-mediated food allergy, then this may be an important mediating factor in the relationship between eczema and food allergy. Although no statistically significant increase in food allergy was seen across most estimates provided by the BEEP study (whether reported at 2, 3, 4 and 5 years or cumulative incidence), there is some uncertainty which ideally needs to be resolved by combining results with other studies. Stimulated by concerns of emollients possibly increasing sensitisation to food allergens and causing food allergy, some members of this team investigated the relationship between frequency of emollient use and food allergy in the enquiring about tolerance study population and found a dose–response relationship between frequency of emollient application and subsequent food allergy.99 Although this finding was based on observational data, the relationship between emollient use and transcutaneous sensitisation to foods clearly requires further investigation. Other work supports the concept of transcutaneous sensitisation to foods, and this process could potentially be enhanced by the use of lipid-rich emollient, by rubbing the skin and by environmental traces of food allergens on parent hands or infant skin.100 A European Union project called TRANS-FOODS, led by BEEP co-investigator Carsten Flohr, is building on these BEEP findings to try to understand how transcutaneous sensitisation to foods might be prevented.101
Although the possibility of harms as a result of emollients used in early life seems low, small degrees of harm affecting a large population of healthy children can add up to a considerable absolute burden and attention to such concerns become a significant issue for public health intervention studies on otherwise healthy children.
Equality, diversity and inclusion
The majority of participants were recruited via secondary care sites, with the remainder via GP surgeries. While sites were mainly in Central/Northern England and London, it is unlikely that any population were systematically excluded from taking part by virtue of where they lived in the country or how they were approached.
The majority of children had mothers of white ethnicity (see Table 4, 85%), which reflects the overall mix of ethnicity the UK.102 The high proportion of mothers (52%) or other first-degree relatives (82%) with a history of eczema reflect the inclusion criterion of ‘first degree relative with parental-reported doctor diagnosis of eczema, allergic rhinitis or asthma’.
No information was collected directly from mothers on their educational or economic status, but deprivation indices derived from participant’s home postcodes included a reasonable range (from 3 to 9).
We were conscious of the fact that severity assessment of eczema involves scoring the intensity of erythema (redness) which is traditionally underestimated in darker skin tones.103 During our face-to-face training that included tests images, we took measures to ensure that all research nurses were aware of the need to upgrade erythema scores by 1 point as per the EASI guidance by including several images of darker skins in the training material.104 The quality control test105 for determining the presence of visible flexural dermatitis which the nurses had to pass in order to undertake the physical assessments also included images of darker skin.
The interpretation of genetic test results is affected by the prevalence of genetic variants in different populations. The most prevalent FLG null mutations have been established in the white European population and other ethnicities have population-specific mutations.34,106,107 To maximise inclusivity in this study we welcomed DNA collection from all participants and all samples were genotyped.
The composition of the BEEP study team was diverse with a range of content expertise from medical (dermatology, paediatrics and primary care) and nursing backgrounds. The team ethos was to always include the important contribution of trial managers, co-ordinators and administrators in emergent publications, and allow opportunities for methodologists such as statisticians (who are rarely afforded an opportunity to be first author) to lead on some publications such as this monograph. The study also allowed the development of early career investigators (such as Maeve Kelleher) who then led on funded parallel projects such as the Cochrane Individual Patient Data meta-analysis of similar studies and also Stella Lartey, who is new to the UK, who benefited from having the opportunity to gain skills in undertaking an economic evaluation in a UK context.
Patient and public involvement
Our trial was unusual for a NIHR Health Technology Assessment (HTA) Programme trial as it was the only study that successfully applied to the specific HTA call at the time to undertake definitive trials as a result of feasibility work carried out in a NIHR Programme Grants for Applied Research (PGfAR) project. Most of the critically important PPI work for the BEEP trial was undertaken in a specific eczema prevention workstream in our PGfAR award entitled ‘Setting priorities and reducing uncertainties in the prevention and treatment of people with skin diseases’. PPI in this eczema prevention workstream defined the choice of intervention, duration of the intervention period and nature of follow-up and contact and is described briefly in the methods section of this report and in full elsewhere.42
Additional PPI work throughout the trial was critical in informing the way in which the add-on food allergy study and genetics substudy could be offered to participants resulting in both being offered as optional studies with a short additional consent procedure that resulted in excellent uptake. In addition, PPI colleagues helped to design, advertise and promote the study and to comment on and disseminate the study results as described in our methods section. It was also very beneficial to have a PPI member in the TSC as they provided an important and varied perspective throughout the trial, and in particular when disseminating the results to parents at 24 and 60 months in order to balance the positive aspects of the key study message (one less thing to do) versus the signal of possible harm (increased minor skin infections). There were no negative experiences of working with parents, carers and patients as PPI colleagues before and during the BEEP study. Although the whole team was disappointed that emollients did not prevent eczema in the BEEP study, our PPI colleagues were remarkably accepting and objective about the finding and enthusiastic about the need to disseminate our important findings.
Throughout the trial we have received positive feedback from the families and the level of engagement and enthusiasm is evidenced by the high retention rate, particularly for the longer-term follow-up at 36, 48 and 60 months. There were no negative experiences communicated to us from parents and carers who participated in the BEEP study which we suspect is due to the thorough PPI work and preparation undertaken in the proceeding Programme Grant for Applied Research.
Findings in context
The BEEP trial results contrast with the BEEP feasibility study92 where a signal on a preventive effect on eczema was found. It should be pointed out that the BEEP feasibility study was designed to test feasibility issues and was one-tenth of the size of the main BEEP trial. Reasons for the difference are unclear but could be due to chance or the fact that the outcome in the BEEP feasibility study was measured at 6 months when irritant dermatitis is more common.
Another large independent eczema prevention trial was conducted in Scandinavia (Preventing Atopic Dermatitis and ALLergies in children – PreventADALL) around the same time as BEEP and its primary outcome results for eczema were published alongside the BEEP trial in The Lancet in 2020. PreventADALL108 was a factorial cluster trial of 2397 newborns that included a skin-care intervention (bath additives and facial cream) and a food intervention (early complementary feeding of peanut, cow’s milk, wheat and egg) as well as both or neither with two primary outcomes: eczema at 12 months and food allergy at 36 months. That study found that neither the skin intervention nor food intervention reduced eczema development, with a risk difference of 3.1% (95% CI –0.3% to 6.5%) for the skin intervention and 1.0% (95% CI −2.1% to 4.1%) for food intervention, in favour of control. No safety concerns were identified.
The most important overall summary of contextual evidence to frame the BEEP study comes from the Cochrane living review of skin-care interventions to prevent eczema that includes traditional aggregate-based data and individual participant data (IPD), funded by the NIHR through the Research for Patient Benefit programme and a personal fellowship award to MK.109 Up to July 2020, the review identified 33 trials that included 25,827 participants. Information on one or more of the review’s specified outcomes was available in 17 studies that included 5823 participants. Eleven trials with 5217 participants, 10 of which provided IPD, were included in one or more meta-analysis. The review found that risk of eczema by 1–2 years of age was not decreased by interventions for skin barrier enhancement in infancy, with a pooled risk ratio from 7 trials and 3075 participants of 1.03, 95% CI 0.81 to 1.31 (classified as moderate-certainty evidence). Time to onset of eczema was also not earlier for those receiving the intervention, with a pooled hazard ratio from 9 trials and 3349 participants of 0.86, 95% CI 0.65 to 1.14 (classified as moderate-certainty evidence). Interestingly, the skin infections signal from BEEP was also found in other studies, with a pooled increased relative risk of skin infection over the intervention period from 6 trials and 2728 participants of 1.34, 95% CI 1.02 to 1.77 (classified as moderate-certainty evidence). Only one trial (the BEEP study) provided data on food allergy outcomes.
The review failed to find any evidence that specific subgroups based on age, intervention duration, hereditary risk, FLG mutation, or classification of intervention type for risk of developing eczema are more likely to benefit from the skin-care interventions. The study also undertook a trial sequential analysis that indicates that further studies of similar interventions are unlikely to change the conclusion that emollients do not reduce eczema risk by more than 30%, although the sequential analysis was inconclusive for a RR reduction of 20%. The review also included studies which used more sophisticated emollients containing ceramides. The review concluded that skin-care interventions in early life are probably not effective for preventing eczema, and probably increase risk of skin infection, whereas the effects of such interventions on food allergy risk remain uncertain. The review also compared findings from aggregate studies versus those obtained from IPD and found that, although IPD did not significantly change the overall primary outcome risk estimates, certainty of evidence and safety outcomes were significantly different using IPD compared to aggregate data. In addition, subgroup and adherence analyses are only possible with IPD demonstrating the added value of using an IPD approach to meta-analysis.110
An update of the review has recently been accepted for publication in the Cochrane Library. This update was planned in order to include the food allergy outcome data from the largest similar trial, PreventADALL. At the timing of the updated review, no new relevant studies with eczema outcomes were published in time to be included, so the eczema outcome at 1–3 years did not change. For food allergy, skin-care interventions during infancy may increase the risk of IgE-mediated food allergy by 1–3 years of age (RR 2.53, 95% CI 0.99 to 6.49; low-certainty evidence; 976 participants, 1 trial) but may not change risk of allergic sensitisation to a food allergen by age 1–3 years (RR 1.05, 95% CI 0.64 to 1.71; low-certainty evidence; 1794 participants, 3 trials). Pre-planned sensitivity analysis assessing food allergy by oral food challenge or investigator assessment showed similar findings (RR 1.45, 95% CI 0.98 to 2.15; low certainty evidence; 2081 participants, 2 trials). The additional results from in this Cochrane review analysis consolidate the results for food allergy diagnosed in this way that were shown in the BEEP study.
Another systematic review on the same topic using aggregate data concluded that emollients do prevent eczema in high-risk children and that the significant protective benefit is only evident where emollients were used continuously to the point of eczema outcome assessment but not when treatment was stopped before eczema assessment.111 However, that review has been criticised for registering the review retrospectively with limited public record of pre-planned eligibility criteria, outcomes and analysis, including subgroup analyses.112 That review also failed to include three eligible trials that were included in the Cochrane review rendering it an unreliable source of evidence for informing clinical practice.
Following the update of the Cochrane IPD meta-analysis, the 3-year results for the Scandinavian PreventADALL study has been published.113 These showed that neither the skin-care intervention nor early complementary feeding of peanut, cow’s milk, wheat and egg from age 3 months had any influence on AD (eczema) diagnosis at 36 months. The study found early complementary feeding reduced food allergy (risk difference −1.6%, 95% CI −2.7% to −0.5%; OR 0.4, 95% CI 0.2 to 0.8) but not in the skin intervention group (risk difference 0.4%, 95% CI −0.6% to 1.5%; OR 1.3, 95% CI 0.7 to 2.3), with no evidence of an interaction effect (p = 0.98).
Study strengths and limitations
The strengths of the BEEP trial can be considered using the population, intervention, comparator, outcome framework. Participants were representative of babies born across the UK who would normally be considered at high risk of developing eczema and associated atopic disorders. The intervention was chosen to be acceptable in a previous feasibility study and was shown in pre-study experiments not to paradoxically damage the skin barrier as some emollients have been shown to do.114 The intervention was simple and cheap and adherence to the intervention was good, especially for study with only two face-to-face contacts with the study team (baseline and at 2 years). Participants in the control group followed standard best advice for infant skin care, and contamination rates with emollients were low (< 20% as specified in our stop/go criteria for study progression). Outcomes were robust and included family reported as well as objective outcomes on the presence of eczema and eczema severity with objective outcomes measured by trained assessors blinded to allocation status. Unblinding of intervention status prior to skin examination was minimal (12 in the intervention group and 6 in the control group). As some debate exists over the validity of diagnostic criteria for eczema in early life,115 a range of eczema outcomes were used that showed consistent lack of a preventive effect. Timing of the outcome was also conducted well away from the emollient intervention period in order to avoid any potential inflation of emollient benefit due to the mild anti-inflammatory effect of the emollient concealing mild eczema. The long duration of follow-up meant that any later benefits of emollient prevention could be picked up. Additional data from the add-on genetic study provided useful insights into whether filaggrin gene mutations are an important consideration for disease stratification in such studies. The add-on food allergy study also provided useful data on the feasibility and prevalence of SPT sensitivity to common food allergens in the UK community setting and resulted in a novel pragmatic method of confirming food allergy for those not able to travel to regional test centres for oral food challenges.59
In terms of study design, BEEP conforms to a relatively pragmatic design when considered using the PRECIS-2 (PRagmatic Explanatory Continuum Indicator Summary) tool.116 The large sample size had sufficient power to exclude a moderate reduction of eczema incidence. Various sensitivity analyses have confirmed the stability of the main conclusions. The effects of missing data on the study conclusions have also been thoroughly explored. The independence of the BEEP study from any commercial influence from emollient manufacturers is also a study strength. Risk of bias for the BEEP study has been rated independently in a Cochrane review109 as low risk of bias for randomisation process, deviations from intended interventions, missing outcome data, outcome measurement and selection of reported results for the eczema outcomes.
The BEEP study has also generated several useful spin-off publications that have contributed to thinking about the design and conduct of eczema prevention trials including how to define an incident case,6 an algorithm for diagnosing IgE-medicated food allergy,59 a SWAT to explore strategies for enhancing follow-up in RCTs,117 an overview of methodological considerations for eczema prevention studies,32 and the value of hyperlinear palms in evaluating FLG mutations.118
Study limitations include the risk that BEEP used the ‘wrong’ emollients. Not all emollients are the same. Some exhibit different physiological effects on skin barrier function in adults with eczema in terms of epidermal water loss, hydration, natural moisturising factor levels and response to irritant challenges.119 Alternatively, it may be that it is critically important to start emollients immediately after birth. Although we were pleased that families started reporting using emollient at a median of 11 days after birth (IQR 7–17 days), one single-centre industry-funded study120 has suggested preventive benefit of eczema at 12 months as a result of twice-daily specially formulated emollients applied for 8 weeks of life when started early (63% within 2 weeks). Perhaps the intervention was not intense enough and rather than an ‘advice to use’ study, an efficacy study whereby emollient application was supervised and measured accurately might have shown benefit. However, such a study would not have been practical and unlikely to be translated into everyday practice. It is also worth pointing out that based on strong user feedback from our BEEP feasibility study, parents wished to have a choice of emollients for the BEEP main study. This meant that the BEEP study used two emollients which might have had slightly different barrier enhancement properties.44 We were however unable to explore this in a meaningful sensitivity analysis as parents initially received both types of emollients and then chose which emollient they preferred when reordering. Perhaps we were wrong to ignore various forms of transient irritant eczemas that emerge in the first year of life during the intervention period, although other studies that have measured such skin findings at 3, 6 and 12 months have not found any benefit from emollients.108 As mentioned in the introduction, it is likely that eczema represents a range of distinct phenotypes with different disease trajectories and severity, but such variation is unlikely to affect the interpretation of this study which evaluates the prevention of new onset eczema that corresponds to a well-defined phenotype depicted by the UK diagnostic criteria for eczema. The strongest predictors of eczema onset and subsequent severity are family history of atopic disease and filaggrin gene mutations. Both were explored in subgroup analyses in this study with neither showing evidence of benefit in such subgroups. Although contamination in our control was considered low (< 20%), perhaps such a degree of contamination was sufficient to mask a small preventive effect. The results of the CACE analysis (with compliance defined as widespread emollient use at least 3 days per week) were consistent with the main analysis of no difference; however, the CIs for the CACE estimate were wider. Due to the pragmatic low-contact efficient study design, we inevitably used parental report for many of the outcomes such as skin infections and allergic diseases at 3, 4 and 5 years. Such unmasked outcome assessment is perhaps a potential source of information bias. We mitigated such bias by mainly asking for parental report of doctor-diagnosed skin infections or allergic disease. We would also argue that capturing parental report of such illnesses are important as not all families who are used to dealing with eczema will necessarily take a child with mild disease to a healthcare professional. Studies comparing parental report of the UK diagnostic criteria for eczema have also shown them to be equally valid when compared with research nurses’ assessed findings.121 The findings from parental reports were also consistent with findings from the research nurse assessors for eczema presence and eczema severity who were masked from intervention or control status at the 2-year primary outcome assessment. Information bias as a result of enthusiasm for the intervention is also perhaps more likely to have resulted in a positive benefit for eczema prevention, whereas the study showed consistent null effects regardless of the timing and choice of outcome. The trial was not powered to detect interaction effects in subgroup analysis meaning that there is considerable uncertainty in the interaction effects (shown through wide 95% CIs) between subgroups and the effect of the intervention including for presence of FLG mutations.
Although the spread of our centres was reasonable representative of English cities and towns, we could have been more inclusive by considering translation of written materials at the study outset. Although the proportion of mothers of white ethnicity in BEEP is broadly reflective of the UK, we could have done more to reach out to under-represented groups using such frameworks as INCLUDE that appeared towards the end of this study.122
Recommendations for clinical practice
Based on the findings of the BEEP study and the emerging evidence from other similar studies, summarised in an ongoing living systematic review that uses individual patient data meta-analysis, emollients cannot be recommended for primary prevention of eczema. Small benefits cannot be excluded, but the burden of applying whole-body emollients daily for the first year of life is considerable, and uncertainty exists about possible signals of harm in terms of skin infections and food allergy.
The use of emollients for eczema prevention should not be confused with use of emollients for eczema treatment. Emollients should continue to be used as part of standard treatment for eczema along with topical corticosteroids and calcineurin inhibitors and systemic therapies for more severe disease. In addition to treating the symptoms of dry skin associated with eczema, emollients may contribute to reducing flares (secondary prevention).123 A recent NIHR HTA study suggests that there is little difference between the efficacy of emollient types (creams, ointments, gels or lotions) for eczema treatment in the community but patient and carer preference for each formulation varies.124
Recommendations for future research – specific and general
Apart from the PreventADALL study, emollient prevention studies to date have focused on barrier enhancement in early life to prevent eczema in high-risk children on the basis that the preventive fraction is likely to be higher in this group when compared to others. The assumption that potential preventive benefits are likely to be higher in high-risk babies may be untrue, as it is possible that babies born to high-risk families inherit such a strong deviation in immunological responses that any potential benefit of barrier enhancement is overcome by the inherited prevailing immunological dysfunction. Studies evaluating emollients for the prevention of eczema in all (including lower risk children) may be useful to provide a complete the picture. Such a study (the CASCADE trial)125 is underway in the USA recruiting around 1250 infant/parent dyads from community settings in Oregon, Colorado, Wisconsin and North Carolina. Future studies should describe and define the phenotype of eczema that develops in the infants participating in prevention studies using well-established diagnostic criteria for defining incident cases6 and by describing the severity and trajectory of such cases once they develop. Many transient irritant forms of eczema may develop in the first year of life most of which probably do not develop into eczema,7 so care should be taken in following up such cases to see if they evolve into persistent eczema. Candidate emollients tested in future studies should ideally be tested for their skin barrier properties on infant skin. Measurement of eczema should also be done well away from the intervention period, for example, 1 year after the intervention has ceased as in the BEEP study, in order to avoid masking any potential mild eczema by the use of emollients.
With around 33 similar studies published or in progress, we suggest that enough studies on emollients for eczema prevention in high-risk families have now been set up and/or completed. Large independent high-quality studies are always welcome, but the addition of a lot of smaller studies is unlikely to help as indicated by the trial sequential analysis illustrated in the ongoing Cochrane IPD review. The priority must be to try and include all existing eczema prevention studies in that Cochrane review in some form – preferably individual patient data rather than aggregate data. We predict that there could be a degree of publication bias from some studies funded by companies with a vested interest in showing a positive outcome that claim that specially formulated emollients containing ceramides or natural moisturising factors are useful in preventing eczema. Although such emollients have not appeared to work so far, we cannot exclude the possibility that such newer specialised formulations could work, and we look forward to seeing the authors share their individual patient data of all registered and non-registered trials with the living Cochrane IPD review on this topic. Some degree of heterogeneity of effects between studies is expected since not all emollients are the same. Other studies should also link their choice of emollient to functional studies of the skin barrier. Attention to measuring and pooling potential harms of daily emollients during infancy should be encouraged in all similar future studies, with more refined collection and verification of the types and severity of skin infection. Experimental studies on the role of whole-body emollients as a vehicle for inducing skin sensitisation should also be considered.
In general terms, investing in research to prevent eczema and associated diseases is important given the burden of disease in the UK and worldwide and concerns about increasing prevalence. The variation in eczema prevalence between different environmental settings and variable genetic penetrance indicates that there are important environmental determinants, which may be modifiable. Given the disappointing results of BEEP and other trials on the strategy of skin barrier enhancement in early life using moisturisers, future eczema prevention research might concentrate on strategies to reduce potentially harmful interventions such as frequent bathing, exposure to hard water and/or use of wash products126 or future molecular mechanisms led by personalised medicine to improve skin barrier formation and function.127 Other areas warranting further investigation are interventions which might influence early immune development, including nutritional and microbial interventions during pre or postnatal life.128 There is already some supportive evidence for probiotics during pregnancy/lactation – however the data are heterogeneous and potentially selectively reported, so the probiotic evidence may potentially be clarified through a high-quality individual patient data meta-analysis.128