Performance of an automated electronic acute lung injury screening system in intensive care unit patients
Performance of an automated electronic acute lung injury screening system in intensive care unit patients
复制标题
DOI：
10.1097/ccm.0b013e3181feb4a0
复制
发表时间：
2011-01-01
影响因子：
8.8
通讯作者：
Fuchs, Barry D.
中科院分区：
文献类型：
--
作者：
Koenig, Helen C.;Finkel, Barbara B.;Fuchs, Barry D.
目的：肺保护性通气可降低急性肺损伤患者的死亡率，但对急性肺损伤的认识不足限制了其应用。我们最近验证了一个自动化的电子急性肺损伤监测系统的患者在一个单一的重症监护病房的重大创伤。在这项研究中，我们评估了该系统的性能作为一个前瞻性的急性肺损伤筛查工具，在一个不同的人口重症监护病房patients.Design：患者进行了前瞻性筛选急性肺损伤超过21 wks的自动化系统和经验丰富的研究协调员谁手动筛选受试者急性呼吸窘迫综合征临床试验网络（ARDSNet）试验。通过将其结果与手动筛选过程进行比较，对自动化系统的性能进行了评估。不一致的结果由两名医生评审员盲法判定。此外，还使用一系列假设进行了敏感性分析，以更好地估计系统的性能。设置：宾夕法尼亚大学医院、学术医疗中心和ARDSNet中心（1994-2006）.患者：内科和外科重症监护室的插管患者。干预措施：无。测量和主要结果：在筛选的1270例患者中，84例被确定为急性肺损伤（发生率为6.6%）。自动筛查系统的灵敏度为97.6%（95%置信区间，96.8-98.4%），特异性为97.6%（95%置信区间，96.8-98.4%）。手动筛选算法的灵敏度为57.1%（95%置信区间，54.5-59.8%），特异性为99.7%（95%置信区间，99.4-100%）。灵敏度分析表明，在不同假设条件下，自动化系统的灵敏度范围为75.0-97.6%。在所有的假设下，自动化系统表现出更高的灵敏度比和可比的特异性手动screeningmethod.Conclusions：一个自动化的电子系统识别急性肺损伤患者具有高灵敏度和特异性，在不同的重症监护病房的一个大型学术医疗中心。需要进一步的研究来评估这种系统可以启动的自动提示对急性肺损伤患者使用肺保护性通气的影响。(Crit Care Med 2011; 39：98-104）
Objective: Lung protective ventilation reduces mortality in patients with acute lung injury, but underrecognition of acute lung injury has limited its use. We recently validated an automated electronic acute lung injury surveillance system in patients with major trauma in a single intensive care unit. In this study, we assessed the system's performance as a prospective acute lung injury screening tool in a diverse population of intensive care unit patients.Design: Patients were screened prospectively for acute lung injury over 21 wks by the automated system and by an experienced research coordinator who manually screened subjects for enrollment in Acute Respiratory Distress Syndrome Clinical Trials Network (ARDSNet) trials. Performance of the automated system was assessed by comparing its results with the manual screening process. Discordant results were adjudicated blindly by two physician reviewers. In addition, a sensitivity analysis using a range of assumptions was conducted to better estimate the system's performance.Setting: The Hospital of the University of Pennsylvania, an academic medical center and ARDSNet center (1994-2006).Patients: Intubated patients in medical and surgical intensive care units.Interventions: None.Measurements and Main Results: Of 1270 patients screened, 84 were identified with acute lung injury (incidence of 6.6%). The automated screening system had a sensitivity of 97.6% (95% confidence interval, 96.8-98.4%) and a specificity of 97.6% (95% confidence interval, 96.8-98.4%). The manual screening algorithm had a sensitivity of 57.1% (95% confidence interval, 54.5-59.8%) and a specificity of 99.7% (95% confidence interval, 99.4-100%). Sensitivity analysis demonstrated a range for sensitivity of 75.0-97.6% of the automated system under varying assumptions. Under all assumptions, the automated system demonstrated higher sensitivity than and comparable specificity to the manual screening method.Conclusions: An automated electronic system identified patients with acute lung injury with high sensitivity and specificity in diverse intensive care units of a large academic medical center. Further studies are needed to evaluate the effect of automated prompts that such a system can initiate on the use of lung protective ventilation in patients with acute lung injury. (Crit Care Med 2011; 39:98-104)