Figures

• 

  Fig. 1. Flow chart for pooling population cohorts and exclusions made to form the population samples to be studied. Sample 2 is a subsample of Sample 1. All steps have been made for all cohorts but numbers after exclusions are illustrated only when the corresponding n is smaller due to them. MONICA Multinational Monitoring of trends and determinants in Cardiovascular disease, SHHEC Scottish Heart Health Extended Cohort, HF heart failure

• 

  Fig. 2. Conceptual model for mediation analysis. The effect of diabetes mellitus on the incidence of heart failure is divided to direct effect and mediator-driven effect. Covariates associated with the incidence of HF are depicted on the right. An individual’s age is taken into account by using age as the time-scale in the HF-free survival model. HF heart failure, BP blood pressure, MI myocardial infarction, AF atrial fibrillation, CRP C-reactive protein, nT-proBNP n-terminal atrial natriuretic peptide, type B, hs high-sensitivity assay, TnI troponin I

• 

  Fig. 3. Unadjusted cumulative HF incidence curves for diabetic and non-diabetic individuals in the study sample 1. Total number of participants in analyses is 94,011. A total of 652 (17.0%) and 5524 (6.1%) cases of incident heart failure were observed in 3834 and 90,177 individuals with and without diabetes, respectively. Shaded area represents the 95% confidence interval and ticks the censored subjects. Log rank P value < 0.001. Numbers of events, censored events and individuals at risk at selected points are provided in Additional file 1: Table S2. HF heart failure, HR hazard ratio

• 

  Fig. 4. Association between risk factors and incident heart failure in diabetic and non-diabetic individuals with each risk factor analyzed in an independent model while adjusting for classical risk factors of heart failure. Models are adjusted for sex, alcohol consumption, systolic blood pressure, BMI, baseline myocardial infarction, baseline atrial fibrillation and stratified by cohort. Total number of participants in analyses is 55,271. A total of 319 (12.9%) and 2175 (4.1%) cases of incident heart failure were observed in 2472 and 52,799 individuals with and without diabetes, respectively. HR hazard ratio, CI confidence interval, HF heart failure, BP blood pressure, MI myocardial infarction, CHD coronary heart disease, AF atrial fibrillation, CRP C-reactive protein, nT-proBNP n-terminal atrial natriuretic peptide, type B, hs high sensitivity assay, TnI troponin I

• 

  Fig. 5. Association between risk factors and incident heart failure in diabetic and non-diabetic individuals with all risk factors included in the same model while adjusting for classical risk factors of heart failure. Total number of participants in analyses is 55,271. A total of 319 (12.9%) and 2175 (4.1%) cases of incident heart failure were observed in 2472 and 52,799 individuals with and without diabetes, respectively. HR hazard ratio, CI confidence interval, HF heart failure, BP blood pressure, MI myocardial infarction, CHD coronary heart disease, AF atrial fibrillation, CRP C-reactive protein, nT-proBNP n-terminal atrial natriuretic peptide, type B, hs high sensitivity assay, TnI troponin I

• 

  Fig. 6. Direct and mediator-driven effects of diabetes mellitus on heart failure incidence. The 95% confidence intervals are reported in parentheses. The total effect of diabetes on heart failure risk (100%) is divided to the direct effect of diabetes and the combined mediator effect (see Fig. 2), with the individual mediators and their share of the mediation effect below. HF heart failure, BP blood pressure, MI myocardial infarction, AF atrial fibrillation, CRP C-reactive protein, nT-proBNP n-terminal atrial natriuretic peptide, type B, hs high-sensitivity assay, TnI troponin I