[Skip to Content]
[Skip to Content Landing]
Views 513
Citations 0
Comment & Response
March 13, 2018

Meta-analysis in Research on Nutrition—Reply

Author Affiliations
  • 1Physicians Committee for Responsible Medicine, Washington, DC
  • 2Department of Nutrition, Harvard T.H. Chan School of Public Health, Boston, Massachusetts
JAMA. 2018;319(10):1050-1051. doi:10.1001/jama.2017.21684

In Reply A thorough systematic review does provide a helpful overview of the literature. However, meta-analysis is a separate step, and study findings can be buried in a poorly planned meta-analysis.

The meta-analysis by Chowdhury et al1 illustrates the challenges. As we noted in our Viewpoint, it combined findings from heterogeneous populations with very different diet patterns. In making a case for the relative weights given to the Oxford Vegetarian Study and the Malmö Diet and Cancer Study in the Chowdhury et al meta-analysis, Dr Thornley and colleagues incorrectly reported the sizes of their data sets. There were 20 440 (not 28 098) participants reported for Malmö and 10 802 (not 1100) in Oxford.1 As noted by the authors of the Malmö study, the high range of saturated fat intake made their study of limited value for testing the association between saturated fat intake and cardiovascular disease, and saturated fat was not compared with other sources of calories.2