Summary page In addition to their profound impact of quality of life, seven of the top ten leading causes of death in the United States in 2010 were chronic conditions, and 86% of health care expenditures were for patients with one or more chronic diseases. A common feature of most chronic disease care is that decision-making is not just a matter of whether to intervene, but when the optimal time to intervene is and which of the available treatments should be tried first. This task becomes even more difficult when there are multiple competing treatments directed at multiple different target outcomes. The VA is reaching a critical point in its ability to develop integrated learning systems into the care of chronic conditions. Clinically-detailed data now dates back almost 15 years and the computing power to use it clinically is now available. This proposal describes the work for the Integrated Preventive Cardiology Initiative (IPCI), which seeks to improve care for the prevention of cardio-cerebrovascular disease (CVD) with an underlying goal of making theoretical and methodological advancing models for integrated chronic disease treatment strategies. CVD is an ideal model for this goal. CVD is not just important in its own right (the leading cause of both morbidity and mortality in VA, the nation and now, worldwide, and a leading cause of ethnic and SES mortality disparities), but CVD has excellent evidence for benefit from multiple treatments which influence multiple target conditions. Further, the risk factors for the different target conditions (heart attacks, stroke, CHF, renal disease) and treatment effects on these outcomes vary substantially. Yet guidelines remain fairly simplistic, without integration of blood pressure (BP), lipid and ASA guidelines. To examine these issues, we developed a multi- faceted study with 3 Specific Aims. Aim1: Examine the degree to which longitudinal baseline patient data improves prediction of overall CVD risk—the key determinant of statin’s and BP medication’s absolute risk reduction. Aim2: Develop and validate methods for adjusting estimates of effect sizes, model calibration, and model discrimination for measurement error in EHR-derived predictor and outcome variables. Aim3: Estimate how the timing, order, and intensity of treatment impact CVD absolute risk reduction within an integrated CVD prevention framework. This 4-year study is designed to substantively improve primary CVD treatment choices, by dramatically advancing how we use existing historical clinical data and integrating the alternative treatment options by analyzing their strengths, weaknesses, and their differential impact on various CVD outcomes. In Aim 1 we will analyze 13-years of longitudinal EHR data on Veterans age 45 to 80 using data from national VA datasets, the National Death Index, CMS data and focused chart reviews. We will test a series of hypotheses trying to understand the relationships of risk factors to different CVD ri...