《用于评估可疑急性主动脉综合征的诊断工具》

  • 来源专题:心血管疾病防治
  • 编译者: 张燕舞
  • 发布时间:2018-05-22
  • 在评估怀疑患有急性主动脉综合征(AAS)的患者时,误诊和过度怀疑是个问题。该项前瞻性多中心主动脉夹层检测风险评分(ADD-RS)加D二聚体在疑似急性主动脉夹层(ADvISED)研究中使用ADD-RS和D-二聚体的组合作为超过1800名此类患者的诊断工具。 ADD-RS≤1加负向D-二聚体的组合有效地排除了AAS,在约300名患者中仅缺少一个病例,并且将使大约60%的具有低AAS概率的患者免于不必要的结局性血管成像。 虽然这种初步的经验似乎很有希望,但在推荐将这种组合作为诊断工具的常规使用之前,需要在更广泛的患者群体中进行额外的验证。

    BACKGROUND Acute aortic syndromes (AASs) are rare and severe cardiovascular emergencies with unspecific symptoms. For AASs, both misdiagnosis and overtesting are key concerns, and standardized diagnostic strategies may help physicians to balance these risks. D-dimer (DD) is highly sensitive for AAS but is inadequate as a stand-alone test. Integration of pretest probability assessment with DD testing is feasible, but the safety and efficiency of such a diagnostic strategy are currently unknown.

    METHODS In a multicenter prospective observational study involving 6 hospitals in 4 countries from 2014 to 2016, consecutive outpatients were eligible if they had≥1 of the following: chest/abdominal/back pain, syncope, perfusion deficit, and if AAS was in the differential diagnosis. The tool for pretest probability assessment was the aortic dissection detection risk score (ADD-RS, 0-3) per current guidelines. DD was considered negative (DD-) if<500 ng/mL. Final case adjudication was based on conclusive diagnostic imaging, autopsy, surgery, or 14-day follow-up. Outcomes were the failure rate and efficiency of a diagnostic strategy for ruling out AAS in patients with ADD-RS=0/DD- or ADD-RS≤1/DD-.

    RESULTS A total of 1850 patients were analyzed. Of these, 438 patients (24%) had ADD-RS=0, 1071 patients (58%) had ADD-RS=1, and 341 patients (18%) had ADD-RS>1. Two hundred forty-one patients (13%) had AAS: 125 had type A aortic dissection, 53 had type B aortic dissection, 35 had intramural aortic hematoma, 18 had aortic rupture, and 10 had penetrating aortic ulcer. A positive DD test result had an overall sensitivity of 96.7% (95% confidence interval [CI], 93.6-98.6) and a specificity of 64% (95% CI, 61.6-66.4) for the diagnosis of AAS; 8 patients with AAS had DD-. In 294 patients with ADD-RS=0/DD-, 1 case of AAS was observed. This yielded a failure rate of 0.3% (95% CI, 0.1-1.9) and an efficiency of 15.9% (95% CI, 14.3-17.6) for the ADD-RS=0/DD- strategy. In 924 patients with ADD-RS≤1/DD-, 3 cases of AAS were observed. This yielded a failure rate of 0.3% (95% CI, 0.1-1) and an efficiency of 49.9% (95% CI, 47.7-52.2) for the ADD-RS≤1/DD- strategy.

    CONCLUSIONS Integration of ADD-RS (either ADD-RS=0 or ADD-RS≤1) with DD may be considered to standardize diagnostic rule out of AAS.

    CLINICAL TRIAL REGISTRATION URL: https://www.clinicaltrials.gov. Unique identifier: NCT02086136.

相关报告
  • 《筛查急性冠脉综合征后的抑郁症》

    • 来源专题:心血管疾病防治
    • 编译者:张燕舞
    • 发布时间:2018-05-22
    • 急性冠脉综合征后,患者抑郁症风险增加。 在对6项前瞻性观察研究(超过1700名患者)的系统回顾中,急性冠状动脉综合征患者单相性重度抑郁症筛查的诊断准确性(表1)与令人满意,并与一般人群中观察到的相当。建议在急性冠状动脉综合征患者中筛查抑郁症,并在筛查中实施服务以确保随访的诊断和治疗。 Background Patients who have had an acute coronary syndrome (ACS) event have an increased risk for depression. Purpose To evaluate the diagnostic accuracy of depression screening instruments and to compare safety and effectiveness of depression treatments in adults within 3 months of an ACS event. Data Sources MEDLINE, EMBASE, PsycINFO, CINAHL, and Cochrane Database of Systematic Reviews from January 2003 to August 2017, and a manual search of citations from key primary and review articles. Study Selection English-language studies of post-ACS patients that evaluated the diagnostic accuracy of depression screening tools or compared the safety and effectiveness of a broad range of pharmacologic and nonpharmacologic depression treatments. Data Extraction 2 investigators independently screened each article for inclusion; abstracted the data; and rated the quality, applicability, and strength of evidence. Data Synthesis Evidence from 6 of the 10 included studies showed that a range of depression screening instruments produces acceptable levels of diagnostic sensitivity, specificity, and negative predictive values (70% to 100%) but low positive predictive values (below 50%). The Beck Depression Inventory-II was the most studied tool. A large study found that a combination of cognitive behavioral therapy (CBT) and antidepressant medication improved depression symptoms, mental health-related function, and overall life satisfaction more than usual care. Limitation Few studies, no evaluation of the influence of screening on clinical outcomes, and no studies addressing several clinical interventions of interest. Conclusion Depression screening instruments produce diagnostic accuracy metrics that are similar in post-ACS patients and other clinical populations. Depression interventions have an uncertain effect on cardiovascular outcomes, but CBT combined with antidepressant medication produces modest improvement in psychosocial outcomes. Primary Funding Source Agency for Healthcare Research and Quality (PROSPERO: CRD42016047032).
  • 《人工智能和机器学习可以成功诊断多囊卵巢综合征》

    • 来源专题:新一代信息技术
    • 编译者:袁晨
    • 发布时间:2023-11-29
    • 美国国立卫生研究院的研究回顾了25年的数据,发现人工智能/机器学习可以检测到常见的激素紊乱。 根据美国国立卫生研究院的一项新研究,人工智能(AI)和机器学习(ML)可以有效地检测和诊断多囊卵巢综合征(PCOS),这是女性中最常见的激素失调,通常在15至45岁之间。研究人员系统地回顾了已发表的使用AI/ML分析数据来诊断和分类PCOS的科学研究,发现基于AI/ML的程序能够成功检测PCOS。 美国国家环境健康科学研究所(NIEHS)的高级研究员和内分泌学家Janet Hall医学博士说:“鉴于社区中PCOS诊断不足和误诊的巨大负担及其潜在的严重后果,我们希望确定AI/ML在识别可能有PCOS风险的患者中的应用。”NIEHS是美国国立卫生研究院的一部分,也是该研究的合著者。“人工智能和机器学习在检测多囊卵巢综合征方面的效果比我们想象的还要令人印象深刻。” 多囊卵巢综合征发生时,卵巢不能正常工作,并在许多情况下,伴随着睾丸激素水平升高。这种疾病会导致月经不规律、痤疮、面部毛发过多或脱发。患有多囊卵巢综合征的女性患2型糖尿病、睡眠、心理、心血管和其他生殖疾病(如子宫癌和不孕症)的风险通常会增加。 该研究的资深作者、NIEHS的助理研究医师和内分泌学家Skand Shekhar博士说:“鉴于多囊卵巢综合征与其他疾病的重叠,诊断多囊卵巢综合征可能具有挑战性。”“这些数据反映了将AI/ML纳入电子健康记录和其他临床环境以改善多囊卵巢综合征女性的诊断和护理的未开发潜力。” 研究作者建议将基于人群的大型研究与电子健康数据集结合起来,并分析常见的实验室测试,以确定能够促进多囊卵巢综合征诊断的敏感诊断生物标志物。 诊断基于多年来发展的广泛接受的标准化标准,但通常包括临床特征(如痤疮,毛发生长过多和月经不规律),伴有实验室(如高血睾酮)和放射学表现(如卵巢超声显示多个小囊肿和卵巢体积增大)。然而,由于多囊卵巢综合征的一些特征可能与其他疾病如肥胖、糖尿病和心脏代谢疾病共同发生,因此它经常被忽视。 人工智能指的是使用基于计算机的系统或工具来模仿人类的智能,并帮助做出决策或预测。ML是AI的一个分支,专注于从以前的事件中学习,并将这些知识应用于未来的决策。人工智能可以处理大量不同的数据,例如来自电子健康记录的数据,使其成为诊断多囊卵巢综合征等难以诊断的疾病的理想辅助工具。 研究人员对过去25年(1997-2022年)发表的关于该主题的所有同行评审研究进行了系统回顾,这些研究使用AI/ML检测多囊卵巢综合征。在一位经验丰富的NIH图书管理员的帮助下,研究人员确定了可能符合条件的研究。他们总共筛选了135项研究,其中31项被纳入了论文。所有研究都是观察性的,并评估了AI/ML技术在患者诊断中的应用。大约一半的研究包括超声图像。研究参与者的平均年龄为29岁。 在采用标准化诊断标准诊断PCOS的10项研究中,检测准确率在80-90%之间。 Shekhar说:“在一系列诊断和分类模式中,人工智能/机器学习在检测多囊卵巢综合征方面表现非常出色,这是我们研究中最重要的收获。” 作者指出,基于人工智能/机器学习的项目有可能显著提高我们早期识别多囊卵巢综合征女性的能力,节省相关成本,减轻多囊卵巢综合征患者和卫生系统的负担。 具有强大验证和测试实践的后续研究将允许人工智能/机器学习在慢性健康状况中的顺利整合。