Integrating Multiple Electronic Health Records Systems to Improve Lung Cancer Outcomes

筛查和治疗的最新进展增加了肺癌（LC）幸存者的数量（~ 571，340 截至2019年的LC幸存者）。然而，研究表明，这些幸存者有很高的风险发展， 第二原发性肺癌（SPLC），在生存5年后，中位10年SPLC风险为8.36%， 初步诊断。此外，SPLC幸存者的死亡率明显高于那些仍然存在的人。 单次LC。存在许多未解决的挑战：（1）虽然先前的研究确定了几个风险因素， SPLC，这些大多是在初步诊断时测量和固定的，结果集中在幸存者身上 谁曾经吸烟。然而，SPLC风险可能受到各种因素动态变化的影响 (e.g.,我们的初步数据显示，SPLC的风险在幸存者中仍然很高 从不抽烟的人(2)然而，目前主要用于SPLC的流行病学数据没有提供详细的数据 在初步诊断后测量，（3）也没有评估从不吸烟的风险因素或预测 幸存者(4)此外，有限的试验证据存在，以解决重要的临床问题，是否和如何 在LC幸存者中，IPLC诊断后继续CT筛查，这需要长期随访， 在临床试验中往往不可行。(5)重要的是，对LC幸存者进行详细筛查的数据通常缺乏 在大多数基于人口的数据。我们计划通过利用电子健康来应对这些多重挑战 电子病历（EHR）和新的分析方法来生成证据，为临床决策提供信息。我们的长期 我们的目标是通过利用大量EHR数据结合新的统计数据， 整合初次诊断后测量的患者数据的方法。我们的具体目标是：（目标1）使用 新颖的三向链接以建立用于LC的集成共享数据库（即，Oncoshare-Lung）使用来自 基于社区和学术的医疗保健系统（人口种族多样， 亚洲从不吸烟者）与加州癌症登记处（CCR）联系;（AIM 2）提供一组临床 通过开发一种新的统计框架来预测有效管理LC幸存者的决策工具 通过捕获IPLC诊断后测量的数据来动态SPLC风险;以及（AIM 3）评估可行性 和一种新的因果推理方法的效用，以评估LC幸存者中SPLC的有效筛查策略 使用EHR。我们将应用一种新的因果推理方法，明确地模拟目标试验（假设 随机试验，以回答感兴趣的问题），以估计长期持续CT筛查的影响， 根据不同的资格标准，定期LC幸存者。我们预计，这项研究的完成将填补 通过提供：（1）临床决策工具来评估个人的动态SPLC风险，以识别 高风险幸存者进行量身定制的监测，（2）新的分析管道，以评估有效的筛选标准， SPLC，以及（3）一个精心策划的数据库，用于LC结局和监测的高影响力转化研究 在一个种族多元化的人口，提供了一个独特的机会，研究在SPLC的关键问题。

Recent advances in screening and treatment have increased the number of lung cancer (LC) survivors (~571,340 LC survivors as of 2019). However, studies have shown that these survivors have a high risk for developing second primary lung cancer (SPLC), with the median 10-year SPLC risk of 8.36% after surviving 5 years from the initial diagnosis. Further, survivors with SPLC have significantly higher mortality vs. those who remain with single primary LC. Many unaddressed challenges exist: (1) While prior studies identified several risk factors for SPLC, these are mostly measured and fixed at the time of initial diagnosis, with findings focused on survivors who have ever smoked. However, SPLC risk is likely to be influenced by dynamic changes of various factors (e.g., smoking cessation), and our preliminary data show that SPLC risk remains just as high among survivors who never smoked. (2) Nevertheless, current epidemiologic data mainly used for SPLC do not offer detailed data measured after initial diagnosis, (3) nor have risk factors or predictions been evaluated for never-smoking survivors. (4) Further, limited trial evidence exists to address the important clinical question of whether and how to continue CT screening after IPLC diagnosis among LC survivors, which requires a long-term follow-up that is often not feasible in clinical trials. (5) Importantly, data on detailed screening for LC survivors are typically lacking in most population-based data. We plan to address these multiple challenges by leveraging electronic health records (EHRs) and novel analytical approaches to generate evidence to inform clinical decisions. Our long-term goal is to improve LC outcomes by focusing on SPLC utilizing large EHR data combined with novel statistical methods that integrate patient data measured after initial diagnosis. Our Specific Aims are: (AIM 1) to use a novel 3-way linkage to establish an integrated shared database for LC (i.e., Oncoshare-Lung) using EHRs from community-based and academic healthcare systems (with an ethnically diverse population with a high proportion of Asian never smokers) linked to the California Cancer Registry (CCR) ; (AIM 2) to provide a set of clinical decision tools for efficiently managing LC survivors by developing a novel statistical framework for predicting dynamic SPLC risks by capturing data measured after IPLC diagnosis; and (AIM 3) to evaluate the feasibility and utility of a novel causal inference method to assess efficient screening strategies for SPLC in LC survivors using EHRs. We will apply a new causal inference method that explicitly emulates the target trials (hypothetical randomized trials to answer the question of interest) in estimating the effects of continuing CT screening in long- term LC survivors under varying eligibility criteria. We expect that the completion of this research will fill the critical gaps in SPLC by providing: (1) clinical decision tools to assess individuals’ dynamic SPLC risks to identify high-risk survivors for tailored surveillance, (2) new analytic pipelines to evaluate efficient screening criteria for SPLC, and (3) a well-curated database for high-impact translational research for LC outcomes and surveillance in an ethnically diverse population that provides a unique opportunity to examine critical questions in SPLC.