Chronic morbidities place longstanding burdens on our health as we age. Although protein biomarkers are critical for the early detection of such diseases, current studies are limited by low sample sizes, variability in proteomics methods and fluctuations in inflammatory protein expression. Here, we present a novel framework for protein-by-proxy analysis of incident disease. We show that DNA methylation proxies for nine inflammatory and seven neurology plasma proteins (generated in up to 875 individuals in the Lothian Birth Cohort 1936) predict the incidence of seven leading causes of morbidity in the Generation Scotland cohort (n=9,537), ascertained via electronic health data linkage over a follow-up period of up to 14 years. After correction for multiple testing and adjustment for common disease risk factors, these included proxy associations between CCL11 and depression (Hazard Ratio: HR = 1.45, P = 1.8 x 10-4), VEGFA and ischaemic heart disease (HR = 1.16, P = 0.02) and associations between incident diabetes and FGF-21 (HR = 1.39, P = 9.7 x 10-7), NEP (HR = 1.32, P = 2.8 x 10-6) and N-CDase (HR = 1.16, P = 0.02). Several of the protein-proxy associations with disease pinpoint proteins that are already therapeutic targets for the diseases in question. These results provide new opportunities to identify circulating biomarkers for disease detection and candidate pathways for drug targeting.