Association of circulating fibroblast growth factor 21 levels with all-cause and cardiovascular mortality: The multi-ethnic study of atherosclerosis.

BACKGROUND: Fibroblast growth factor 21 (FGF21) levels are often elevated in cardiovascular disease (CVD). However, no study has assessed its association with cardiovascular and all-cause mortality in a population free of clinically evident CVD. METHODS: A total of 5543 Multi-Ethnic Study of Atherosclerosis (MESA) participants (mean age 62.7 years, 47.5 % male), free of clinically evident CVD at baseline, were studied. From baseline (2000-2002), 1606 deaths (including 387 CVD deaths) were observed over a median follow-up of 17.7 years. Multivariable Cox regression analysis was performed to assess the association of plasma FGF21 levels with mortality. RESULTS: FGF21 levels at baseline were associated with all-cause mortality, even after adjustment for traditional risk factors, including demographic, socioeconomic and cardiovascular risk factors (adjusted hazard ratio 1.08 [95% confidence interval 1.01, 1.16] per 1 SD increase in ln-transformed levels; 1.27 for the highest vs, lowest quartile). Baseline FGF21 levels were significantly associated with both CVD and non-CVD mortality in unadjusted models. However, the association with non-CVD mortality, but not CVD mortality, remained statistically significant after adjusting for covariates. Similar results were obtained in FGF21 quartile analyses and also when using competing risk regression or matched case-control cohort in sensitivity analyses. CONCLUSIONS: In subjects without clinically-evident CVD at baseline, over 17.7 years follow-up there is a modest association of baseline FGF21 levels with all-cause mortality. The finding that this is driven primarily by a significant association with non-CVD mortality over almost two decades merits further investigation.

Using machine learning to predict cardiovascular risk using self-reported questionnaires: Findings from the 45 and Up Study.

BACKGROUND: Machine learning has been shown to outperform traditional statistical methods for risk prediction model development. We aimed to develop machine learning-based risk prediction models for cardiovascular mortality and hospitalisation for ischemic heart disease (IHD) using self-reported questionnaire data. METHODS: The 45 and Up Study was a retrospective population-based study in New South Wales, Australia (2005-2009). Self-reported healthcare survey data on 187,268 participants without a history of cardiovascular disease was linked to hospitalisation and mortality data. We compared different machine learning algorithms, including traditional classification methods (support vector machine (SVM), neural network, random forest and logistic regression) and survival methods (fast survival SVM, Cox regression and random survival forest). RESULTS: A total of 3687 participants experienced cardiovascular mortality and 12,841 participants had IHD-related hospitalisation over a median follow-up of 10.4 years and 11.6 years respectively. The best model for cardiovascular mortality was a Cox survival regression with L1 penalty at a re-sampled case/non-case ratio of 0.3 achieved by under-sampling of the non-cases. This model had the Uno's and Harrel's concordance indexes of 0.898 and 0.900 respectively. The best model for IHD hospitalisation was a Cox survival regression with L1 penalty at a re-sampled case/non-case ratio of 1.0 with Uno's and Harrel's concordance indexes of 0.711 and 0.718 respectively. CONCLUSION: Machine learning-based risk prediction models developed using self-reported questionnaire data had good prediction performance. These models may have the potential to be used in initial screening tests to identify high-risk individuals before undergoing costly investigation.