Is the number of ideal cardiovascular health metrics in midlife associated with lower risk of cancer? Evidence from 3 European prospective cohorts

Figures

• 

  Figure 1:. Hazard ratios (HRs) for incident cancer per 1 increment in ideal cardiovascular health metric. Hazard ratios and 95% confidence intervals (CIs) were estimated by Cox proportional hazards models stratified by year of birth (5-yr intervals) and using age as the timescale. Hazard ratios were adjusted for sex, marital/cohabitation status, education and occupation at baseline. In NutriNet-Santé, results were additionally adjusted for alcohol use and family history of cancer. In Whitehall II, results were additionally adjusted for alcohol use and race/ethnicity. The overall HR and CI was estimated using random-effects meta-analysis.

• 

  Figure 2:. Hazard ratios (HRs) per 1 increment in ideal cardiovascular health metric for site-specific cancer — A) lung cancer, B) breast cancer, C) prostate cancer, D) colon cancer. Hazard ratios and 95% confidence intervals (CIs) were estimated by Cox proportional hazards models stratified by year of birth (5-yr intervals) and using age as the timescale. Hazards ratios were adjusted for sex, marital/cohabitation status, education and occupation at baseline. In NutriNet-Santé, results were additionally adjusted for alcohol use and family history of cancer. In Whitehall II, results were additionally adjusted for alcohol use and race/ethnicity. The overall HR and CI was estimated using random-effects meta-analysis.

• 

  Figure 3:. Hazard ratios (HRs) for incident cancer per 1 increment in ideal cardiovascular health metric excluding the smoking metric and adjusted for smoking status. Hazard ratios and 95% confidence intervals (CIs) were estimated by Cox proportional hazards models stratified by year of birth (5-yr intervals) and using age as the timescale. Hazards ratios were adjusted for sex, marital/cohabitation status, education, occupation at baseline and smoking metric. In NutriNet-Santé, results were additionally adjusted for alcohol use and family history of cancer. In Whitehall II, results were additionally adjusted for alcohol use and race/ethnicity. The overall HR and CI was estimated using random-effects meta-analysis.