[How Reliable are Meta-Analyses? An Appraisal Using the Example of Magnesium Tocolysis]

Crowther et al. 3 analysed the effectivity of magnesium tocolysis in preventing preterm birth. They conclude that there is no evidence for protection. In its latest guidelines, based on this Cochrane analysis, the German Association of Gynaecology and Obstetrics (DGGG) does not recommend any more the use of magnesium for tocolysis. Magnesium tocolysis is said neither to delay nor to prevent preterm birth. Moreover, magnesium could be responsible for increased mortality in infants. These conclusions are mostly based on the research of Mittendorf et al. 4. In a Cochrane study from 2014, which in principal was identical to the study mentioned above 3, Crowther et al. 6 confirm the previous findings and conclusions.

Having successfully applied magnesium tocolysis for many years, these surprising conclusions led us to review the soundness of the publications mentioned above. Combining the practical experience of many years with the results of a comprehensive literature retrieval, we finally contrasted this knowledge with the findings of the aforementioned publications that caused the DGGG to withdraw the recommendation for magnesium.

To draw binding consequences from a meta-analysis is possible only when stringent quality guidelines are observed. The studies that were included in the Cochrane review of Crowther et al. 3 are very heterogeneous and are not suitable for concluding on poor or even lacking effectiveness of magnesium tocolysis. Furthermore, the cases of infant deaths, as stated by Mittendorf et al. 4, are very unlikely caused by magnesium.

When including studies in a meta-analysis special attention has to be given to the relevance and unbiased selection of studies. To prevent any misjudgment, a thorough knowledge of the included studies seems essentiell. There is not sufficient evidence to withdraw the recommendation for applying magnesium tocolysis as a preventive measure to prevent preterm birth. In the sense of evidence-based medicine, long-standing, scientifically proven therapeutic success should be incorporated into the meta-analysis as well.