Prenatal Blood Pressure Patterns to Predict Pregnancy-Related Hypertension and Later Life Cardiovascular Risk

摘要 妊娠高血压疾病(PRH)、子痫前期(PE)和妊娠高血压(GH)， 导致多达10%的妊娠并发症，是孕产妇和围产期发病率的主要原因 美国的死亡率。这些疾病也与老年高血压(HTN)的更高风险有关。 以及女性的心血管疾病(CVD)。2013年，美国妇产科学院 认识到有必要及早识别有患PRH障碍风险的妇女，因为现有的预测 模型不能证明在低风险女性中的临床实用性。2017年，美国预防服务局 特别工作组的一份报告列举了研究方面的差距，包括需要“进一步开发和验证风险工具” 使用严格的方法进行预测，包括适当的校准统计数据和经过验证的模型 使用常规护理中提供的参数(例如，临床病史和临床测试)。建议进行的研究 利用统计方法解决这些差距，这些方法旨在识别具有相似特征的潜在类别的个人 血压(BP)的模式会随着时间的推移而变化。这项先进的统计技术将女性分类为 BP轨迹组，可以确定那些在“低风险”女性中PE和GH风险较高的人。我们选择了 这种方法是因为它已经被证明对预测非妊娠患者未来的心血管疾病是非常有效的 成年人。这项研究将利用BP轨迹组和临床风险因素来评估和验证模型 预测该指数和后续妊娠期间的PE和GH，以及HTN和CVD的后期风险。 我们建议对2009-2018年(~330,000)分娩的妊娠进行一项回顾性队列研究，同时 女性晚年HTN和CVD结局的前瞻性随访。这个大型的、基于社区的、高度 来自Kaiser Permanente North California(KPNC)集成医疗保健提供系统的多样化样本 利用自2008年以来建立的电子健康记录(EHR)链接所有临床数据来源。这项研究 将开发PE和GH的预测模型，这些模型在大多数情况下显示出对早期的高临床实用价值 低风险妇女的风险分层，以及晚年新发HTN和CVD的预测。具体目标 目标1：确定与PE和GH风险相关的前20周妊娠血压轨迹组，以及 评估和验证BP轨迹模型识别PE和GH风险的预测能力； 目的2：评价和验证前20周妊娠血压轨迹组的预测能力。 整个妊娠BP轨迹组，每个组识别有或没有PE和GH的妇女谁有风险 新发病的HTN和CVD最长为产后12年；目标3：在没有PE或GH的妇女中(目标1-2)， 评估和验证整个妊娠血压轨迹组预测随后的PE和GH的能力 怀孕了。临床翻译是为患有EHR临床的低风险女性开发一种自动化算法 改善PE和GH以及后期HTN和CVD的早期风险分层的数据。该方法可以 直接临床监测，使用可用的治疗方法，并测试新的治疗方法以进行早期预防。

ABSTRACT Pregnancy-related hypertensive (PRH) disorders, preeclampsia (PE) and gestational hypertension (GH), complicate up to 10% of all pregnancies and are a leading cause of maternal and perinatal morbidity and mortality in the U.S. These disorders also have been linked with higher risk of later life hypertension (HTN) and cardiovascular disease (CVD) in women. In 2013, the American College of Obstetrics and Gynecology identified the need for early identification of women at risk for PRH disorders because available predictive models could not demonstrate clinical utility among low risk women. In 2017, the U.S. Preventive Services Task Force report cited research gaps including the need “to further develop and validate tools for risk prediction using rigorous methodology, including appropriate calibration statistics and validated models that use parameters available in routine care (e.g., clinical history and clinical testing).” The proposed study addresses these gaps utilizing statistical methods designed to identify latent classes of individuals with similar patterns of blood pressure (BP) change over time. This advanced statistical technique classifies women into BP trajectory groups that may identify those at higher risk for PE and GH among “low risk” women. We chose this method because it has already proven to be highly effective for prediction of future CVD in non-pregnant adults. This study will utilize BP trajectory groups and clinical risk factors to evaluate and validate models for prediction of PE and GH during the index and subsequent pregnancies, as well as later risk of HTN and CVD. We propose a retrospective cohort study of pregnancies delivered in 2009-2018 (~330,000) along with prospective follow up for later life HTN and CVD outcomes in women. This large, community-based, highly diverse sample from the Kaiser Permanente Northern California (KPNC) integrated healthcare delivery system leverages the established electronic health record (EHR) since 2008 linking all clinical data sources. The study will develop prediction models for PE and GH that show high clinical utility across most settings for the early risk stratification of low risk women, and prediction of new onset HTN and CVD in later life. The specific aims are: Aim 1: To identify the first 20 wks' gestation BP trajectory groups associated with risk of PE and GH, and evaluate and validate the BP trajectory model's predictive ability to identify women at risk for PE and GH; Aim 2: To evaluate and validate the predictive ability of first 20 wks' gestation BP trajectory groups, and the entire pregnancy BP trajectory groups to each identify women with or without PE and GH who are at risk for new onset HTN and CVD up to 12 years post-delivery; Aim 3: Among women without PE or GH (Aims 1-2), to evaluate and validate the entire pregnancy BP trajectory groups ability to predict PE and GH in a subsequent pregnancy. The clinical translation is to develop an automated algorithm for low risk women with EHR clinical data to improve early risk stratification for PE and GH, and later life HTN and CVD. The method may direct clinical monitoring, use of available therapies, and testing of new therapies for early prevention.