We have carefully read the article published by de Mello et al.,1 and we have found several important issues to discuss.
Regarding the clinical topic, inclusion of studies was based on Rome III Criteria; however, we found that authors misclassified four out of the nine articles included in this systematic review: Castillejo et al.,2 actually used functional constipation (FC) criteria according to Rome II; Mozaffarpur et al.,3 included patients classified using both Rome III criteria and the Paris Consensus on Childhood Constipation Terminology Criteria; Loening-Baucke4 defined FC according to the Medical Position Statement of the North American Society for Pediatric Gastroenterology and Nutrition in 1999; finally, Kokke et al.,5 included patients according to Loening-Baucke.4 This unappropriated inclusion of studies might have led to different effect sizes due to clinical heterogeneity. The authors should have stated that they were going to include different Rome Criteria and afterwards, they should have conducted a sensibility or a subgroup analysis to identify the alteration in effect sizes.
Regarding the methods used to perform this systematic review, we have found some flaws in the following issues6:
- 1.
Protocols for systematic reviews and meta-analysis must be published in PROSPERO before performing the entire process and the publication of the final manuscript.
- 2.
The search strategy must be based on patient (population), intervention, and study design according to different manuscripts. This one was based on patient, outcome, and study design, which might lead to selection bias.
- 3.
Using two tools for assessing bias is unnecessary (RoB Cochrane tool and Jadad score). If the authors wanted to perform another analysis, assessing the quality of evidence using the GRADE tool would have been important and interesting.
- 4.
There was no appropriate description of the strategy to identify non-published literature, for example, Google Scholar, Open Grey database, clinical trials registries, and conferences, among others. This might have created publication bias.
Consequently, we suggest caution in the use of the findings when extrapolating results to clinical practice.
Conflicts of interestThe authors declare no conflicts of interest.
Please cite this article as: Velasco-Benítez CA, García-Perdomo HA. Some flaws that could change the effect sizes and limit the extrapolation of these results. J Pediatr (Rio J). 2019;95:374–5.