Cardiovascular disease in Type 1 diabetes: quantifying risk and addressing limitations in the analysis of longitudinal cohort studies

[摘要] Cardiovascular disease (CVD) has historically been increased in type 1 diabetes compared to thegeneral population, but no contemporary estimates of risk are available in the United States.Additionally, the reasons for this increased risk are not fully understood, as the hyperglycemiathat characterizes type 1 diabetes is itself an inconsistent predictor of CVD incidence. Thus, theobjective of this dissertation is to quantify the contemporary incidence and excess risk of CVD inyoung adults <45 years old with type 1 diabetes and to utilize novel statistical methods toaddress limitations in the analyses of longitudinal cohort studies, in an effort to better understandthe risk factor patterns that lead to CVD in this population.Data are from the Pittsburgh Epidemiology of Diabetes Complications study, aprospective cohort study of childhood-onset type 1 diabetes diagnosed at Children’s Hospital ofPittsburgh between 1950 and 1980. CVD data from the background Allegheny County,Pennsylvania population were used to calculate age- and sex-matched standardized mortality(SMR) and incidence rate ratios (IRR). Using tree-structured survival analysis (TSSA), formalsubgroup analysis was performed to identify groups at varying levels of risk for CVD, based onthreshold effects of continuous risk factors. Joint models were used to simultaneously model thelongitudinal trajectory of HbA1c and time to CVD incidence.CVD risk was shown remain significantly increased in this type 1 diabetes cohort. TSSAidentified a range of risk groups, which were defined by combinations of diabetes duration, non-HDL cholesterol, albumin excretion rate, and white blood cell count. The longitudinal trajectoryof HbA1c was associated with CVD risk, similarly across all manifestations of CVD, includingcoronary artery disease, stroke, and lower extremity arterial disease, which is a new finding inthis cohort. This work has important impacts on public health, as it confirms that individualswith type 1 diabetes continue to be at increased risk for CVD and demonstrates that novelstatistical methods should be utilized as a complement to traditional methods to increaseunderstanding of disease etiology.

[发布机构] the University of Pittsburgh