Fig. 3
Mediation of hsCRP by change in HbA1c and/or body weight. Panel (a) shows mediation of hsCRP by HbA1c, panel (b) by body weight, and panel (c) by HbA1c and body weight combined, all according to trial and with pooled semaglutide arms versus comparators. The indirect effect is the effect of semaglutide versus comparator on change in hsCRP that was caused by its effect through change on HbA1c and/or BW. The direct effect is the effect of semaglutide versus comparator on change in hsCRP independent of change in HbA1c and/or BW. The total effect was calculated by adding the indirect effect to the direct effect. The proportion mediated was calculated by dividing the indirect effect by the total effect × 100. Data from the ‘on-treatment without rescue medication’ period. Mediation was analyzed with mixed models for repeated measurements, adjusted for relevant mediators at baseline: HbA1c (panel a), BW (panel b), BW and HbA1c (panel c); and furthermore adjusted for additional covariates at baseline: log(hsCRP), age, sex, eGFR (Chronic Kidney Disease–Epidemiology Collaboration 2009), body mass index, statins (yes/no), insulin treatment (PIONEER 5 analyses only), metformin treatment (SUSTAIN 3 and PIONEER 5 analyses only), and HbA1c (panel b only). Mediator data were only taken from the visits where hsCRP was assessed for estimation of the direct effect. BW body weight; CI confidence interval; eGFR estimated glomerular filtration rate; HbA1c glycated hemoglobin; hsCRP high-sensitivity C-reactive protein