Statistical reporting of metabolomics data: experience from a high-throughput NMR platform and epidemiological applications

Abstract

Abstract Introduction Meta-analysis is the cornerstone of robust biomedical evidence. Objectives We investigated whether statistical reporting practices facilitate metabolomics meta-analyses. Methods A literature review of 44 studies that used a comparable platform. Results Non-numeric formats were used in 31 studies. In half of the studies, less than a third of all measures were reported. Unadjusted P-values were missing from 12 studies and exact P-values from 9 studies. Conclusion  Reporting practices can be improved. We recommend (i) publishing all results as numbers, (ii) reporting effect sizes of all measured metabolites and (iii) always reporting unadjusted exact P-values.