Diabetes Mellitus, Comorbid Conditions and Mortality

描述（由申请人提供）： 背景和意义。糖尿病是一种不断发展的、使人衰弱的慢性疾病，导致发病率和死亡率。很少有研究调查糖尿病和共病及其与18岁及以上成年人9年以上死亡率的关系。 目的。对于1997-2006年期间连续参加团体健康的成年糖尿病患者，目标是：1）估计1997-2006年糖尿病和合并冠心病（CAD）、充血性心力衰竭（CHF）、高血压和抑郁症的患病率和发病率，分别描述1997年糖尿病患病率与1998-2006年糖尿病发病率的时间模式; 2）估计1998-2006年四种共病的组合对死亡率的贡献，相对于仅患有糖尿病的登记者; 3）在1997年和2006年存活的糖尿病患病率登记者中，比较1998-2006年治疗失败的血糖负担是否在仅患有糖尿病的登记者中低于患有共病的登记者;和4）通过估计在1997年患有糖尿病和CHF和/或CAD的登记者中，HbA 1c和胆固醇水平较高是否与1998-2006年较低的死亡率相关，来测试心血管风险的“反向流行病学”。 人口1997-2006年连续入组的团体健康成人入组者，在此期间患有流行或偶发的1型或2型糖尿病。排除了1997年死亡的登记者，包括了1998-2006年活着或死亡的登记者。符合条件的入选者分为两个相互排斥的队列：1）糖尿病患病率（时间序列所有年份的糖尿病入选者）; 2）糖尿病发病率（1998-2006年糖尿病发作的入选者）。 研究设计.所有目标都将采用纵向队列设计，以1997年为9年（1998-2006年）时间序列中的基线年。 数据源集团健康自动化数据库。 措施死亡率、四种共病、调整诊断组、糖尿病并发症、糖尿病治疗类型、个人和计划特征、吸烟、肥胖（BMI）、血压、HbA 1c和胆固醇水平、提供者和信息的利用率、主治医生的连续性。分析。对于目的1，将为两个队列生成描述其他4种共病的患病率和发作的描述性统计量和时间序列图。将构建时间序列中共病状况的类型和时间的措施。在目标2中，考克斯比例风险回归模型将估计合并症与死亡率之间的关联。在目标3中，多变量回归模型将估计合并症与血糖负荷之间的关系。目标4将使用考克斯比例风险模型检验反向流行病学。

DESCRIPTION (provided by applicant): Background and Significance. Diabetes is a growing, debilitating chronic disease that contributes to morbidity and mortality. Few studies have examined diabetes and comorbid conditions and their relationships to mortality over 9 years among adults aged 18 and over. Aims. For the population of adult enrollees with diabetes mellitus continuously enrolled in Group Health between 1997-2006, aims are to: 1) estimate the prevalence and incidence of diabetes and combinations of comorbid coronary artery disease (CAD), congestive heart failure (CHF), hypertension and depression in 1997-2006, describing temporal patterns separately for enrollees with prevalence diabetes in 1997 versus incidence diabetes in 1998-2006; 2) estimate the contribution of combinations of the four comorbid conditions to mortality in 1998-2006, relative to enrollees with diabetes alone; 3) among prevalence enrollees with diabetes in 1997 and alive through 2006, compare whether glycemic burden from treatment failure in 1998-2006 is lower in enrollees with diabetes alone than in enrollees with comorbid conditions; and 4) test for `reverse epidemiology' of cardiovascular risk by estimating whether greater HbA1c and cholesterol levels are related to lower mortality in 1998-2006 among enrollees with diabetes and CHF and/or CAD in 1997. Population. Group Health adult enrollees who are enrolled continuously from 1997-2006 and who had prevalent or incident type 1 or 2 diabetes during that time interval. Enrollees who died in 1997 are excluded, and enrollees who were alive or died in 1998-2006 are included. Eligible enrollees are divided into two mutually-exclusive cohorts: 1) prevalence diabetes (enrollees with diabetes in all years of the time series); and 2) incidence diabetes (enrollees with diabetes onset in 1998-2006). Study Design. A longitudinal cohort design will be used in all aims, with 1997 a baseline year in the 9-year (1998-2006) time series. Data Sources. Group Health automated data bases. Measures. Mortality, four comorbid conditions, Adjusted Diagnosis Group, diabetes complications, diabetes treatment type, personal and plan characteristics, smoking, obesity (BMI), blood pressure, HbA1c and cholesterol levels, utilization of providers and information, continuity of primary physician. Analyses. For Aim 1 descriptive statistics and time series graphs will be produced for the two cohorts describing the prevalence and onset of other four comorbid conditions. Measures will be constructed of the type and timing of comorbid conditions in the time series. In Aim 2 Cox proportional hazards regression models will estimate the association between the combinations of comorbid conditions and mortality. In Aim 3 multivariate regression models will estimate the relationship between combinations of comorbid conditions and glycemic burden. Aim 4 will test for reverse epidemiology using the Cox proportional hazards model.