Lifetime predictions for individualized vascular disease prevention: Whom and when to treat?

Abstract

Since cardiovascular disease (CVD) is a result of a lifelong exposure to risk factors and involves numerous people, it is utterly important to identify which people are most likely to get CVD in order to prevent or delay disease. Therefore, cardiovascular risk prediction is a keystone in the prevention of CVD to help identify patients at high-risk who may benefit most from drug treatment. Although an overwhelming of 250 risk prediction models are made over the last 15 years, this thesis focus on the translations of trial results to the individual patient by predicting individualized treatment benefits for lifelong prevention of CVD instead of 10-year risk reduction. It also aimed to improve the applicability of prediction models in clinical practice. In this thesis, the diabetes lifetime prediction (DIAL) model was developed and validated to predict lifelong treatment effects in patients with T2DM. Also, a cost-effectiveness analyses was performed which showed a favorable cost per QALY when treatment decisions were made based on lifetime predictions compared to 10-year predictions. To improve the use of prediction models, methods to deal with missing patient characteristics when using these models were explored, were median imputation of the population median was sufficient for missing characteristics as long as important predictors such as age were available. In addition, it is shown that treatment of risk factors in terms of cardiovascular disease free life years gained (lifetime benefit) is most favorable when started early in life.