Atopic dermatitis (AD), is one of the most common chronic inflammatory skin disorders among infants and children. AD is characterized by itching and recurrent eczematous lesions, and its incidence has increased worldwide over the past several decades. The current prevalence rate is 10–20% in infants and children (Weidinger and Novak, 2016). As the leading non-fatal medical skin disorder, AD imposes severe psychosocial burdens on pediatric patients and their families (Chamlin and Chren, 2010; Silverberg, 2016; Sidbury and Khorsand, 2017). AD is associated with high risks of allergy, asthma, and mental health issues (Sung et al., 2017). Infants and children with AD are typically treated with topical corticosteroids (TCS), antihistamines, and even antibiotics (Totri et al., 2017). However, these medications exert several adverse side effects, and AD symptoms may recur rapidly after treatment is stopped. Furthermore, long-term TCS use may trigger new-onset AD.

Probiotics is becoming increasingly attractive as a treatment option for some illnesses in children (Fuchs-Tarlovsky et al., 2016). Probiotics (live bacteria or yeasts) are not necessarily harmless, but they help to protect hosts from harmful bacteria (Mizock, 2015). When administered in adequate amounts, probiotics may play beneficial roles not only in the gastrointestinal tract but also in the gut–brain–skin axis (Ogden and Bielory, 2005; Dehingia et al., 2015; Huang et al., 2016; Huang and Hu, 2017). Several studies on the benefits of probiotics for pediatric AD patients have appeared over the past decades. In 2000, Pessi et al. reported that oral probiotics alleviated the clinical symptoms of gastrointestinal inflammation and AD (Pessi et al., 2000). Kirjavainen et al. (2003) reported lower Bacteroides counts in the fecal microflora of children with atopic eczema than in healthy infants and suggested that probiotics can be used to treat AD in children (Kirjavainen et al., 2003); however, some reports yielded contrasting results (Licari et al., 2015). For instance, Gruber et al. found that Lactobacillus rhamnosus strain GG (LGG) exerted no therapeutic effects in infants with mild-to-moderate AD (Gruber et al., 2007). Therefore, we systematically evaluated the effects of probiotics used to treat AD in children.

Overall, the data suggested an overall benefit of probiotics supplementation in children with AD, and age-specific sub-analyses showed that probiotics effectively reduce SCORAD values in children aged 1–18 years. Geography-specific sub-analyses showed that probiotics effectively reduced SCORAD values in Asia, while no effect was observed for Europe. Strain-specific sub-analyses indicated that Lactobacillus (LS), Lactobacillus fermentum (LF), and a probiotic mixture reduced SCORAD values in children with AD, while LGG and Lactobacillus plantarum (LP) showed no effect in children with AD.

Hippocrates (460–370) stated that “All diseases begin in the gut”, which is the earliest suggestion that bacteria affect health (Hippocrates, 2002). Metchnikoff, known as the father of probiotics (Gordon, 2016), proposed that colonic bacteria afforded health benefits in aging adults. In recent decades, probiotics that aid in the resolution of pediatric atopic eczema have been investigated. Viljanen et al. explored probiotic effects on pediatric atopic eczema/dermatitis syndrome but found no significant difference between the treatment and control groups (Viljanen et al., 2005). Passeron et al. compared probiotics and prebiotics and found that both significantly improved AD manifestations in children (Passeron et al., 2006). Brouwer et al. evaluated the clinical and immunological effects of Lactobacillus rhamnosus (LR) supplementation in a hydrolyzed formula given to children with AD but found no significant effect (Brouwer et al., 2006). The cited authors suggested that the discrepancies between their results and those of other trials were likely attributable to differences in treatment timing and the strains used. Sistek et al. conducted a 12-week trial in the UK and found that a combination of LR and Bifidobacteria lactis (BL) improved AD symptoms in food-sensitive children (Sistek et al., 2006). At roughly the same time, a prospective German study by Folster-Holst et al. yielded insufficient evidence to make the conclusion that LGG is an effective treatment for moderate-to-severe AD in infants (Folster-Holst, 2010). Gruber et al. also found that LGG had no therapeutic effect in such patients (Gruber et al., 2007). Despite these discouraging findings, Gerasimov et al. reported that Lactobacillus acidophilus DDS-1 and Bifidobacterium lactis UABLA-12 afforded significant clinical improvements in children with moderate-to-severe AD (Gerasimov et al., 2010). Similarly, Wu et al. showed that Lactobacillus salivarius (LS) exerted short-term beneficial effects in patients with moderate-to-severe AD (Wu et al., 2012). Drago et al. suggested that such effects may be attributable to restoration of the altered intestinal microbiota (Drago et al., 2011). In contrast, Gore et al. found that LS exerted no beneficial effects on eczema when given as an adjunct to basic topical treatment (Gore et al., 2012). Several reports have examined the effects of other bacterial strains on AD in children. Supplementation with LPCJLP 133, Lactobacillus paracasei, and LF was reported to be effective. The discrepancies described above may be attributable to differences in the strains used, the study areas, and/or the ethnicities of the subjects. Several groups have performed meta-analyses to evaluate the effectiveness of probiotics on AD. Da Costa Baptista et al. reviewed all published trials and reported that the biological effects observed in most trials suggest that probiotic adjuvant treatments are of benefit for AD (da et al., 2013). The cited review, although comprehensive, did not report total MDs or 95%CIs. Chang performed a meta-analysis of studies in which either prebiotics or probiotics were given and reported that synbiotics may be useful to treat AD (Chang et al., 2017). However, the focus was on synbiotics rather than probiotics. Szajewska et al. stressed the need for data on individual probiotic strains rather than on probiotics in general (Szajewska and Mrukowicz, 2003; Szajewska et al., 2015). Ogden et al. suggested probiotics as a complementary approach to the treatment and prevention of pediatric AD (Ogden and Bielory, 2005). They concluded that probiotics should be an active area of investigation, considering the role of gut microbiota in altered immune responses in atopic patients. However, the authors did not perform a meta-analysis to obtain further details about the treatment effects of probiotics. Kim et al. reviewed 25 RCTs on the effects of probiotics in the treatment of AD in patients of all ages. They observed significant differences in SCORAD values favoring probiotics over the control group in children 1–18 years old and in adults, whereas no favorable effects were seen in infants < 1 year old (Kim et al., 2014). We found that probiotics were efficacious in children aged 1–18 years (MD, −4.50; 95%CI, −7.45 to −1.54) and showed strong efficacy in Asia but not in Europe; furthermore, LGG had no effects on AD whereas LS, LF, LP, and a mixture of strains showed beneficial effects. Our findings are in agreement with those of Lee et al., who concluded that the evidence for probiotics as a useful treatment of AD in children is convincing. However, the cited authors reviewed only trials published before 2008, whereas we included later trials to afford greater insight. The differences may be because we included only RCTs involving children under the age of 18 years and those that reported MDs. Some RCTs presented values other than MDs, including the study by Kim et al. (2010), and some presented the results as figures, rendering calculations impossible. We contacted the corresponding authors but did not receive useful replies. Thus, we excluded those studies. Third, some of the included studies had small sample sizes, which may affect the reliability and validity of the conclusions. Thus, our overall results are affected by these issues, and the data were highly heterogeneous. These topics require further attention. Also, in the subgroup analyses, children with AD may have different gut microbiota profiles from those of normal children. Thus, probiotics supplementation in children < 1 year old and 1–18 years old may promote a healthier gut microbiota profile, boosting their immune response. People from different areas have different dietary structures and gut bacterial compositions. Dehingia et al. compared gut bacterial diversity between Indian populations and worldwide data (Dehingia et al., 2015). Zhang et al. also suggested that a phylogenetically diverse gut microbiota at the genus level may be commonly shared by distinctive healthy populations, which may explain the diversity of the effects of probiotics across people from different countries (Zhang et al., 2015). The above discussion is of importance to physicians, dermatologists, and other public healthcare workers who deal with diverse ethnic populations.

To the best of our knowledge, there are no previous reports on the effects of different probiotic strains on AD in children. In our meta-analysis, all trials involving LGG and one trial involving LP showed no effects, while two studies confirmed the beneficial effects of LP on AD (MD, −0.70; 95%CI, −2.30 to 0.90; P = 0.39). This discrepancy may be associated with differences in dosages, the timing and duration of intervention, and sample sizes, and further trials are required to clarify this point.

Our meta-analysis had certain limitations. First, we attempted to minimize heterogeneity and publication bias, but significant heterogeneity among trials remained evident. Differences in study samples, study populations, and intervention methods contributed to the heterogeneity. For example, some of the included studies had small sample sizes, compromising the reliability and validity of the conclusions. In addition, the RCTs were performed in various countries, thus, the subjects differed among RCTs in terms of their genetic make-up and microbial exposure, which in turn are associated with varying responses to the same probiotic. Also, we excluded some RCTs from this meta-analysis, and fewer studies included will reduce the confidence associated with the data interpretation and increase heterogeneity and publication bias. Finally, we cannot draw robust conclusions as to which probiotic strain/mixture should be given to children with AD and which population(s) would receive maximum benefit from such treatment.

Our present work demonstrated that probiotics may have the potential to decrease SCORAD values in children with AD. However, the findings presented here must be generalized with caution because of heterogeneity. The results are a source of optimism with regard to the management of AD in children. More adequately powered RCTs using standardized measurements are necessary to assess which species of probiotics and dosages and what treatment periods are most efficacious for children with AD.

RH, MS, and XC proposed the idea of this study and designed the study; RH, HN, MS, and JL conducted data screening and performed quality assessment; RH and MS used RevMan software to assess the data and performed the statistical analysis and gave the explanations of the statistical results. RH drafted the initial manuscript. JL, JZ, and XC critically reviewed and revised the manuscript.