Introduction: Respiratory failure is a common organ failure syndrome in hospitalized patients1. Vital sign monitoring (like respiratory rate & oximetry) is a necessary aspect of risk stratification, but it is not sufficient. In one study of hospitalized patients, 46% of the patients had no significant vital sign change in the 24 hours before an unplanned intubation2. Therefore, clinicians must also monitor for physical diagnostic signs that link the appearance of breaths to respiratory instability. Many pathognomonic patterns of high-risk labored breathing have been described. For example, when rib-dominant breaths alternate with abdomen-dominant ones, the patient is said to exhibit respiratory alternans, a sign of inspiratory muscle overload3. However, the manual assessment of such signs lacks sensitivity, inter-rater reliability, and scalability4. We sought to (a) identify technologies that can measure labored breathing and (b) assess their readiness for clinical adoption by hospitals. Methods: We selected four well-established diagnostic signs of labored breathing: (1) respiratory rate variability, (2) recruitment of accessory muscles (upper-rib elevation by the scalene and sternocleidomastoid muscles), (3) Abdominal Paradox (rib-abdomen asynchrony), and (4) respiratory alternans (rib-dominant breaths alternate with abdomen-dominant ones). We systematically searched PubMed using pre-specified keywords corresponding to these four signs. We identified 2868 abstracts. Two reviewers independently screened each abstract to ensure that it reported on technology that quantified the diagnostic sign of interest. A third reviewer resolved any disagreements. We excluded 2423 articles with an abstract review and included 445 articles for full paper review. We excluded an additional 127 articles after full paper review, and we were unable to acquire 4 articles. We included the remaining 314 articles for analysis. Results: Quantification of labored breathing has been attempted for over 50 years; the earliest study included in our analysis was published in 1975. Over 30 different hardware configurations have been tried, either alone or in combination; but none of them has been validated as a comprehensive solution to measure all the four diagnostic signs that we studied. Despite enormous improvements in sensor technologies and computing capacity, the scale of investigation has not meaningfully increased since 1975. In the first decade of kinematic measurements (1975-1984), there average annual number of studies was 2.7 and the median sample size was 19. In the decade prior to our study (2013-2022), the average annual number of studies was 11.3 and the median sample size was 20. To this day, a majority of the studies are conducted in a specialized laboratories (73% between 2013-2022) rather than clinical practice settings. Most studies aimed to measure the construct validity of a technology (19%) or to describe kinematic distributions in specific clinical scenarios (77%). Rarely did studies attempt to quantify the predictive validity for a clinical outcome (4%). We did not find any clinical trial where a kinematics-based early warning intervention was tested. Conclusions: This study describes a major bottleneck in the translation of bedside diagnostic signs of high-risk labored breathing patterns into measurable physiomarkers of respiratory instability. Despite half a century of attempted measurement, the technology readiness level for clinical adoption remains low.

中文翻译：

---

呼吸困难模式：呼吸病理生理学的一个无法测量的维度

简介：呼吸衰竭是住院患者常见的器官衰竭综合征1。生命体征监测（如呼吸频率和血氧测定）是风险分层的必要方面，但这还不够。在一项针对住院患者的研究中，46% 的患者在计划外插管前 24 小时内没有明显的生命体征变化2。因此，临床医生还必须监测将呼吸外观与呼吸不稳定联系起来的身体诊断体征。已经描述了高风险呼吸困难的许多特征模式。例如，当肋骨主导的呼吸与腹部主导的呼吸交替时，据说患者会表现出呼吸交替，这是吸气肌超负荷的迹象3。然而，对此类迹象的手动评估缺乏敏感性、评估者间的可靠性和可扩展性4。我们试图 (a) 确定可以测量呼吸困难的技术，以及 (b) 评估其是否为医院临床采用做好准备。方法：我们选择了四种行之有效的呼吸困难诊断体征：(1) 呼吸频率变异性，(2) 辅助肌募集（斜角肌和胸锁乳突肌抬高上肋骨），(3) 腹部悖论（肋骨-腹部）异步），（4）呼吸交替（肋骨主导呼吸与腹部主导呼吸交替）。我们使用与这四个标志相对应的预先指定的关键词系统地搜索 PubMed。我们确定了 2868 个摘要。两名审稿人独立筛选每个摘要，以确保其报告了量化感兴趣的诊断标志的技术。第三位审稿人解决了任何分歧。我们排除了 2423 篇带有摘要评论的文章，​​并纳入了 445 篇文章进行全文评论。经过完整的论文审查后，我们排除了另外 127 篇文章，其中 4 篇文章无法获取。我们纳入了其余 314 篇文章进行分析。结果：呼吸困难的量化尝试已超过 50 年；我们分析中包含的最早的研究发表于 1975 年。已经尝试了 30 多种不同的硬件配置，无论是单独的还是组合的；但它们都没有被验证为衡量我们研究的所有四种诊断标志的综合解决方案。尽管传感器技术和计算能力取得了巨大进步，但自 1975 年以来，研究规模并没有显着增加。在运动学测量的第一个十年（1975-1984 年），平均每年研究数量为 2.7 项，中位样本量为 19 个。在我们研究之前的十年（2013-2022）中，年均研究数量为 11.3，中位样本量为 20。迄今为止，大多数研究都是在专门实验室进行的（2013-2013 年期间为 73%）。 2022）而不是临床实践环境。大多数研究旨在衡量技术的结构有效性（19%）或描述特定临床场景中的运动学分布（77%）。很少有研究尝试量化临床结果的预测有效性（4%）。我们没有发现任何测试基于运动学的早期预警干预的临床试验。结论：这项研究描述了将高风险劳累呼吸模式的床边诊断体征转化为可测量的呼吸不稳定生理标志物的主要瓶颈。尽管进行了半个世纪的尝试测量，临床采用的技术准备水平仍然很低。