CRD Summary: The authors concluded that for atopic dermatitis, topical calcineurin inhibitors were more effective than placebo. Pimecrolimus was as effective as mild but less effective than moderately potent topical corticosteroids. Tacrolimus was more effective than mild and as effective as moderately potent topical corticosteroids. As few data were available and the review had methodological problems, these conclusions appear questionable.

CRD Commentary: The objectives and inclusion criteria of the review were clear, but outcomes were not pre-specified as part of the inclusion criteria. Only one database was searched, so some studies may have been missed. The search was restricted by language and no specific attempts were made to retrieve unpublished studies, so the review was subject to language and publication biases. The reviewers acknowledged the potential for publication bias, but it did not appear that this was formally assessed. Study selection, validity assessment and data extraction were undertaken by a single reviewer without an independent check, which increased the risk of reviewer error and bias.

Relevant criteria were used to assess study validity, which was variable; less than half the studies were rated as highly generalisable. As the authors noted, nearly all studies were commercially sponsored. The statistical methods used to pool studies and assess for heterogeneity appeared appropriate in most respects, but where significant heterogeneity was detected it was not formally assessed or explored in the review. Although the authors stated that all age groups and grades of severity showed the same results, the possible impact of clinical heterogeneity between the studies was not formally tested. Most comparisons involved only one or two studies. The authors' statement regarding the equivalent effectiveness of pimecrolimus and mild topical corticosteroids did not appear to be supported by evidence. [A: The authors have sent a correction to the journal for this error in the published paper.]

As few data were available and the review had many methodological problems, the authors' conclusions appear questionable.