Fig. 2From: Higher thyrotropin leads to unfavorable lipid profile and somewhat higher cardiovascular disease risk: evidence from multi-cohort Mendelian randomization and metabolomic profilingSecond stage associations between TSH and 52 metabolomic markers using multivariable and Mendelian randomization analyses on Nightingale platform and multivariable analysis on Bruker platform. Point estimates represent the standardized change in metabolomic marker concentration per standard deviation change in TSH; error bars indicate 95% confidence intervals. Multivariable analyses were adjusted for age, sex, body mass index, and smoking, Mendelian randomization analyses are inverse variance-weighted (IVW) estimate. Hollow effect estimates refer to associations with p value > 0.05Back to article page