Paper
No evidence that fenugreek is more effective than placebo as a galactagogue
Published Oct 21, 2020 · L. Grzeskowiak
Phytotherapy Research
3
Citations
0
Influential Citations
Abstract
Dear Editor, Khan, Wu, and Dolzhenko (2018) conducted a systematic review and network meta-analysis evaluating the effectiveness of fenugreek as a galactagogue. The authors original conclusion was that fenugreek was superior to placebo, but inferior inferior to Coleus amboinicus Lour and palm date. These findings, however, are the result of multiple data extraction errors making these conclusions invalid. The key issue relates to the incorrect extraction of data from two RCTs evaluating the use of fenugreek that reported the primary outcome as daily breast milk production. Khan et al. (2018) transformed the original values of daily breast milk production to breast milk production per expression episode by dividing the mean by 8 (assuming women all expressed 8 times per day) and adjusting the corresponding SD. This incorrect transformation procedure leads to an unintended manipulation of the variance estimate, artificially enhancing the actual precision of the original study findings. For example, taking the example of the study by Reeder, Legrand, and O'connor-Von (2013), the original mean and SD for daily breast milk production in the experimental group is 568 (± 419) ml/day. Khan et al. report in their meta-analysis a mean of 71 (±6.1) ml/expression. In this case the coefficient of variation for sample variance has gone from 73% (419/568 × 100) to just 9% (6.1/71 × 100). Similarly, the original study by Damanik, Wahlqvist, and Wattanapenpaiboon (2006) reports a mean and SD for the experimental group as 358.5 (±135.2) ml/day. In the meta-analysis by Khan et al. (2018) it is reported as being 44.81 (±2.1) ml/expression. In this case the coefficient of variation for sample variance has gone from 38% (135.2/358.5) to just 5% (2.1/44.81). Table 1 provides a comparison of the original findings from Khan et al. (2018) to those when the correct values are used in the metaanalysis. Of the original 15 comparisons performed, 3 change direction of effect and 6 change to being no longer statistically significant (due to greater imprecision in the effect size). That is, more than 50% of the original study findings by Khan et al. (2018) no longer hold, representing a critical flaw in their original manuscript. Systematic reviews and meta-analyses are essential tools for summarizing evidence accurately and reliably and it is disappointing that the review by Khan et al. (2018) misinforms readers as to the true
Full text analysis coming soon...