CV Benefits and Safety of Glucose-Lowering Therapies in Adults with Diabetes

描述（由申请人提供）：心血管（CV）事件是2型糖尿病成人最常见的死亡原因。虽然常用的血压（BP）和胆固醇药物的获益和安全性已得到充分证实，但对涉及多种降糖药物治疗的常用血糖控制策略的CV获益和CV风险知之甚少。为了解决这一重要的临床和公共卫生问题，该项目检查了130万名接受治疗的2型糖尿病成年人在13年期间的详细临床数据，以量化降糖药物的特定组合对心肌梗死，中风，CV死亡率，总死亡率和充血性心力衰竭（CHF）住院的影响。我们检验了四种比较有效性假设，以确定主要CV事件是否与以下因素存在差异性相关：（a）使用二甲双胍与磺酰脲类（SU）、噻唑烷二酮（TZD）、DPP-4抑制剂或GLP-1受体激动剂的特异性双药组合;（B）当添加基础胰岛素时，使用二甲双胍与SU、TZD、DPP-4抑制剂或GLP-1受体激动剂的特异性双药组合;（c）在已经使用基础胰岛素的受试者中添加餐时胰岛素;和（d）用人胰岛素相对于类似物胰岛素的治疗。分析采用现代比较有效性统计方法，包括新的用户设计和具有逆概率加权（IPW）和目标最小损失估计（TMLE）的边际结构建模（MSM），在某些条件下，允许适当调整早期暴露和临床结局之间因果途径上的时间依赖性混杂因素，以及适当调整信息删失导致的选择偏倚。研究结果有可能通过确定哪些常用的降糖方案最大限度地提高CV获益并最大限度地降低CV风险，从而大幅改善数百万美国2型糖尿病患者的CV结局。

DESCRIPTION (provided by applicant): Cardiovascular (CV) events are the most common cause of death in adults with type 2 diabetes. While the benefits and safety of commonly used blood pressure (BP) and cholesterol medications are well established, much less is known about the CV benefits and CV risks of commonly used glucose-control strategies involving treatment with multiple glucose-lowering agents. To address this important clinical and public health concern, this project examines detailed clinical data from 1.3 million adults with treated type 2 diabetes over a 13-year period to quantify the impact of specific combinations of glucose-lowering agents on myocardial infarction, stroke, CV mortality, total mortality, and congestive heart failure (CHF) hospitalizations. We test four comparative effectiveness hypotheses to determine whether major CV events are differentially related to: (a) use of specific two-agent combinations of metformin with sulfonylurea (SU), thiazolidinedione (TZD), DPP-4 inhibitors, or GLP-1 receptor agonists; (b) use of specific two-agent combinations of metformin with SU, TZD, DPP-4 inhibitors, or GLP-1 receptors agonists when basal insulin is added; (c) addition of prandial insulin in subjects already using basal insulin; and (d) treatment with human insulin versus analog insulin. Analysis employs modern comparative effectiveness statistical approaches, including new user designs and marginal structural modeling (MSM) with inverse probability weighting (IPW) and targeted minimal loss-based estimation (TMLE) which, under certain conditions, allow proper adjustment for time-dependent confounders on the causal pathway between early exposures and clinical outcomes, as well as proper adjustment for selection bias due to informative censoring. The study results have the potential to substantially improve CV outcomes in millions of Americans with type 2 diabetes by identifying which commonly used glucose-lowering regimens maximize CV benefits and minimize CV risks.