Modeling the role of modifiable risk and protective factors in opioid use disorder and non-fatal opioid overdose among AI/AN using EMR

摘要/摘要 随着阿片类药物流行继续肆虐美国，美国印第安人和阿拉斯加原住民(AI/ANS) 继续受到不成比例的影响，AI/ANS中与阿片类药物相关的死亡人数是黑人的三倍 还有西班牙裔白人。AI/AN的高阿片类药物过量比率不仅在AI/ANS中更高，他们还 坚持：大都市和非大都市的AI/ANS的阿片类药物过量比率都比任何 其他种族群体在近16年期间(2008年至2015年)。人工智能/人工智能不是同质的种群， 全国各地都有重大而重要的变化。例如，华盛顿有26.7个人工智能/年 年，每10万名阿片类药物过量中就有6.7人死亡，而南达科他州每10万人中只有6.7人死亡 2018年。然而，虽然人工智能/AN社区与 一般人群(例如，药物滥用家族史、某些特征或精神疾病)，他们在这方面有所不同 此外，由于创伤和虐待的比率很高，以及文化差异。由于样本量较小，历史上 被排除在研究之外，以及分散的人群，阿片使用障碍(OUD)和阿片类药物的真实程度- 人工智能/AN社区的相关过量可能被低估了。这项拟议的研究针对的是弱点。 在先前的研究中，通过一)对人工智能/ANS的单一关注和二)使用一个大型的健壮数据库：CENER 公司的健康实况®。Health Fact®拥有524,959名独特的AI/AN患者，包含电子健康 包括诊断、程序、药物和实验室以及人口统计数据在内的记录-允许 患者级别的分析和不断变化的趋势。这项研究将确定OUD和阿片类药物的AI/AN比率- 国家一级和美国九个人口普查地区(新英格兰、大西洋中部、东部)的相关过量 中北部、中西部、中西部、南大西洋、中南部东部、中南部西部、山区、太平洋)、 以及确定风险和保护因素。具体地说，我们的目标是估计UD的费率和调整后的赔率 和非致命性阿片类药物过量在国家一级对AI/ANS(目标1)，并创建预测模型以确定 可改变的风险和保护因素在OUD和非致命性阿片类药物过量中的作用(目标2)。这是第一次 研究，据我们所知，使用大型强大的电子病历数据库(EMR)来分析和 患者一级人工智能/AN社区中与阿片类药物相关的死亡。我们的纵向数据将显示出变化 随着时间的推移的趋势，以及各种人口统计信息将使风险和防护分析成为可能 各种因素。这些数据将允许对AI/ANS中的OUD比率和阿片类药物相关死亡进行细微的比较 普通民众。最终，我们的分析将被用来开发关于各种因素作用的预测模型 可修改和不可修改的风险因素，可用于确定哪里最需要 财政资源和政策。作为国家毒瘾多样性研究所的补充资料，该项目将 为该领域贡献一支多元化且训练有素的药物使用研究人员队伍，帮助美国国立卫生研究院实现其 多元化的健康相关科学队伍。

Summary/Abstract As the opioid epidemic continues to ravage the United States, American Indians and Alaska Natives (AI/ANs) continue to be disproportionally affected with opioid-related deaths three times higher in AI/ANs than in Blacks and Hispanic whites. High AI/AN opioid overdose rates have not only been higher in AI/ANs, they have also been persistent: both metropolitan and non-metropolitan AI/ANs had the highest opioid overdose rate than any other racial group over a nearly 16-year period (2008 to 2015). AI/ANs are not a homogenous population and there are significant and important variations across the country. For example, Washington saw 26.7 AI/AN deaths per 100,000 of opioid overdoses, whereas South Dakota only saw 6.7 AI/AN deaths per 100,0000 in 2018. However, while the AI/AN community shares some risk and protective factors for substance use with the general population (e.g., family history of drug abuse, certain traits or psychiatric conditions), they differ in this as well due to high rates of trauma and abuse and cultural differences. Due to small samples sizes, a history of exclusion from research, and a dispersed population, the true extent of opioid use disorder (OUD) and opioid- related overdoses in the AI/AN community is likely underestimated. This proposed study addresses weaknesses in prior research through i) a singular focus on AI/ANs and ii) use of a large robust database: Cerner Corporation’s Health Facts®. With 524,959 unique AI/AN patients, Health Facts® contains electronic health records including diagnosis, procedures, medications, and labs, as well as demographic data—allowing for patient-level analysis and changing trends overtime. This study will determine AI/AN rates of OUD and opioid- related overdoses at the national level and by the nine U.S. Census regions (New England, Mid Atlantic, East North Central, West North Central, South Atlantic, East South Central, West South Central, Mountain, Pacific), as well as identifying risk and protective factors. Specifically, we aim to estimate rates and adjusted odds of OUD and non-fatal opioid overdose for AI/ANs at the national level (Aim 1), and create a predictive model to determine the role of modifiable risk and protective factors in OUD and non-fatal opioid overdoses (Aim 2). This is the first study, to our knowledge, to use large robust electronic medical records database (EMR) to analyze OUD and opioid-related fatalities in the AI/AN community at the patient level. Our longitudinal data will show changing trends over time, and a variety of demographic information will enable an analysis of both risk and protective factors. These data will allow for a nuanced comparison of OUD rates and opioid-related fatalities in AI/ANs to the general population. Ultimately, our analysis will be used to develop predictive models on the role of various modifiable and non-modifiable risk factors, which may be used to determine where there is the greatest need for fiscal resources and policies. As a National Institute of Drug Addiction Diversity Supplement, this project will contribute a diverse and highly trained substance use researcher to the field, helping NIH achieve its goal of a diversified health-related sciences workforce.