Fig. 2: Predicted log hazard function for all-cause, cardiovascular diseases (CVD), and cancer mortality in relation to plasma chemerin concentration.

Results are based on sex-adjusted Cox proportional hazard regression models with restricted cubic splines for chemerin (three knots located at the 5th, 50th, 95th percentile). The study cohort was included as a stratification variable and participants’ age was used as time scale (unit: years). Spline functions and regression models were calculated for the whole study population as well as for participants aged < 60 years, 60–69 years, and ≥ 70 years, separately. Cause-specific hazards were modeled for each event separately and the respective competing events were considered as censored cases (cause-specific hazard function). The dotted lines colored in grey represent the highest tertile (109 ng/mL) of chemerin concentration in the overall population.