Summary
The use of meta-analysis to evaluate probiotics in this condition is commendable. The abstract suggests that improvement may be real, but the evidence presented indicates that probiotics may be ineffective on careful reading. The abstract is lacking information needed to determine the quality of the meta-analysis and whether its results/suggestions should be considered in investment decisions.
Analysis
Several phrases need to be carefully considered. On first read the abstract suggests that the evidence supports probiotics, however closer examination suggests this may be incorrect.
1. The abstract says 4 of 9 RCTs 'suggest' statistically significant decrease in SCORAD. However, it does not say these studies show a statistical significant result. Why did the authors not state that 4 RCTs had P values < 0.05?
2. 2 RCTs were reported as statistically significant in children with IgE-associated AD. I assume that this conclusion is based on a subgroup analysis, the dangers of which I have reported on previously in GLG news (http://www.glgroup.com/News/Biostatistics--Subgroup-Analysis-in-Clinical-Trials-20281.html). If a subgroup analysis was done in these studies, was the analysis prespecified or post-hoc? How many IgE-associated AD children was in the subgroup analysis? If IgE-associated AD is common, did the other RCTs do the subgroup analysis and determine probiotics were not statistically significant?
3. The same issue as in point 2 applies to the RCT reporting statistical significance in food-sensitized children.
4. No statistical difference was reported in any biomarkers (e.g., TNF). If probiotics do not impact biomarkers what mechanism of action has been proposed for probiotics?
5. The authors state that, although not statistically significant, AD severity was lower in half of the RCTs in the probiotic group than the placebo group. The reverse is that AD severity was higher in half of the RCTs in the probiotic group than the placebo group. What importance does this statement have?
This study should be viewed with caution. The abstract suggests more positive evidence for probiotics than this biostatistician sees in the evidence.
