Extended Data Fig. 2: Drug–metabolite associations in model 1 versus model 2.
The drugs with at least one significant metabolite association in baseline model (model 1) by linear regression are shown. The first letter of the ATC code precedes the drug name, to identify different categories. Sample sizes of the drug users and non-users in model 1 (age and sex adjusted) and model 2 (age, sex, BMI and smoking adjusted) are shown following drug names, respectively. Dark red, positive significant associations in model 1 (P-value < 1.9 × 10−5); light red, positive nonsignificant associations in model 1 (P-value ≥ 1.9 × 10−5); dark blue, negatively significant associations in model 1 (P-value < 1.9 × 10−5); light blue, negatively nonsignificant associations in model 1 (P-value ≥ 1.9 × 10−5). Asterisks in boxes denote that neither direction nor significance status were different between model 1 and model 2 (P-value < 1.9 × 10−5). Two-tailed tests were used.
