Predicting Diabetes Risk Using Glucose Data

描述(申请人提供)：在美国有超过700万人患有未确诊的糖尿病，另有7300万人患有未确诊的前驱糖尿病。虽然早期诊断和治疗可以改善这两种情况下的健康结果，但在过去3年中，只有一半符合糖尿病筛查条件的人进行了筛查。自动化、基于电子病历(EMR)的糖尿病风险评估和糖尿病筛查的系统化方法可能会提高筛查率。尽管目前EMR筛查指南的自动化具有挑战性，但数据表明，与国家筛查指南(基于年龄、性别、种族、体重指数和其他健康状况)相比，单一随机血糖值是更好的糖尿病预测指标。然而，这种方法诊断糖尿病前期和糖尿病的敏感性很差。随着时间的推移利用多个血糖值--患者的血糖病史--可能会提高随机血糖筛查策略的敏感性，以检测未诊断的糖尿病前期和糖尿病。这项提案描述了一项职业发展计划，该计划将使候选人为成为一名成功的独立调查员做好准备，并实现他的长期职业目标，即成为开发和实施EMR干预措施以改善2型糖尿病患者诊断和健康结果的国家领先者。这一拟议的研究策略将开发一种新的、基于计算机血糖病史的糖尿病风险评估工具，然后利用该工具在EMR中开发和实施临床决策支持，以促进在常规临床实践中进行糖尿病筛查。PI的近期目标是利用EMR中可用的纵向血糖值--“血糖史”--开发一种基于随机血糖(RBG)的糖尿病风险评估工具和临床决策支持。为了实现这一目标，他提出了一项职业发展计划，其中包括教学课程、参加地方和国家会议和研讨会，以及在德克萨斯州达拉斯的德克萨斯大学西南医学中心和Parkland医院的支持性研究环境中进行指导性研究。这一环境包括由NIH资助的CTSA和临床与翻译研究中心、AHRQ资助的以患者为中心的结果研究中心，以及Parkland临床创新中心，这是一个利用EMR数据进行高级医疗分析和预测建模的实体。该项目的研究目标是：1)利用EMR中提供的血糖数据描述非已知糖尿病患者的血糖病史特征，并描述血糖异常史与糖尿病筛查之间的关系；2)进行前瞻性糖尿病筛查研究，以开发和优化RBG风险工具的性能，以使用EMR血糖病史识别未诊断的糖尿病和糖尿病前期病例；以及3)开发和评估使用AIM 2的RBG风险工具启用EMR的糖尿病筛查临床决策支持的可行性。这些研究目标将作为职业发展计划和培训目标的平台，包括：1)应用医学信息学培训；2)高级量化分析；3)实施科学；4)比较有效性研究。总而言之，K23提案的研究和培训目标将为两个R01应用提供培训、经验和初步数据。一个R01将是一个更大的糖尿病筛查研究，在AIM 2的基础上增加口服葡萄糖耐量测试。另一个R01将以AIM 3的试点为基础，提出一项全功能、多点随机对照试验，以评估基于EMR的糖尿病筛查工具和临床决策支持在临床实践中识别糖尿病和糖尿病前期病例的有效性。拟议的研究和培训目标将战略性地将PI定位为开发、实施和评估循证干预措施以改善2型糖尿病预后的领导者。K23提案中概述的糖尿病风险识别和筛查的创新方法可能会对这种非常常见、严重和昂贵的疾病的临床护理、人口健康管理和国家筛查指南政策产生非常大的影响。

DESCRIPTION (provided by applicant): Over 7 million people in the US have undiagnosed diabetes, and an additional 73 million have undiagnosed prediabetes. Although early diagnosis and treatment can improve health outcomes in both conditions, only half of individuals eligible for diabetes screening have been screened in the past 3 years. Automated, electronic medical record (EMR)-based diabetes risk assessment and systematic approaches to diabetes screening may improve screening rates. Although automation of current screening guidelines within EMRs is challenging, data suggest that a single random glucose value is a better predictor of diabetes than national screening guidelines (which are based on age, sex, race, body mass index, and other health conditions). However, the sensitivity of this approach to diagnose both prediabetes and diabetes is poor. Utilization of multiple glucose values over time - a patient's glucose history - may improve the sensitivity of random glucose screening strategies to detect undiagnosed prediabetes and diabetes. This proposal describes a career development plan that will prepare the candidate to become a successful independent investigator and attain his long-term career goal of becoming a national leader in the development and implementation of EMR interventions to improve the diagnosis and health outcomes of patients with type 2 diabetes. This proposed research strategy will develop a novel, computerized glucose history-based diabetes risk assessment tool and then utilize this tool to develop and implement clinical decision support in the EMR to promote diabetes screening in routine clinical practice. The PI's immediate goal is to use the longitudinal glucose values available within the EMR - the "glucose history" - to develop a random blood glucose (RBG)-based diabetes risk assessment tool and clinical decision support. To meet this goal, he has proposed a career development plan that integrates didactic coursework, participation in local and national conferences and workshops, and a progression of mentored research studies within the supportive research environment at University of Texas Southwestern Medical Center and Parkland Hospital in Dallas, TX. This environment includes a NIH-funded CTSA and Clinical and Translational Research Center, an AHRQ-funded Center for Patient-Centered Outcomes Research, and the Parkland Center for Clinical Innovation, an entity that does advanced healthcare analytics and predictive modeling with EMR data. The research aims of this project are to: 1) characterize the glucose history of patients without known diabetes using glucose data available in the EMR and describe associations between an abnormal glucose history and diabetes screening; 2) conduct a prospective diabetes screening study to develop and optimize performance of a RBG risk tool to identify cases of undiagnosed diabetes and prediabetes using the EMR glucose history; and 3) develop and assess the feasibility of EMR-enabled diabetes screening clinical decision support using the RBG risk tool from Aim 2. These research aims will serve as the platform for the career development plan and training aims which include: 1) training in applied medical informatics; 2) advanced quantitative analyses; 3) implementation science; and 4) comparative effectiveness research. Together, the research and training aims of the K23 proposal will provide the training, experience, and preliminary data for two R01 applications. One R01 will be a larger diabetes screening study with additional oral glucose tolerance testing based on Aim 2. The other R01 will be based on the pilot in Aim 3 and propose a fully-powered, multisite randomized controlled trial to assess the efficacy of the EMR-based diabetes screening tool and clinical decision support to identify cases of diabetes and prediabetes in clinical practice. The proposed research and training aims will strategically position the PI to become a leader in the development, implementation, and evaluation of evidence-based interventions to improve outcomes in type 2 diabetes. The innovative approach to diabetes risk identification and screening outlined in this K23 proposal has the potential for very high impact on clinical care, population health management, and national screening guideline policies for this very common, serious, and costly disease.