医学
逻辑回归
回顾性队列研究
流血
卡帕
急诊医学
威尔科克森符号秩检验
干预(咨询)
重症监护医学
内科学
外科
曼惠特尼U检验
语言学
精神科
哲学
作者
Mohammad Ali,Haseeb Ahmed,Sagar S. Patil,Bhawna Halwan
标识
DOI:10.14309/00000434-201410002-02225
摘要
Introduction: Accurately triaging acute upper GI bleed (UGIB) patients who are most likely to benefit from early intervention may help improve outcomes and optimize allocation of resources related to management. Our research aims to study the potential impact of using a clinical decision support system (CDSS) and Glasgow-Blatchford scales (GBS) versus clinical judgment on outcomes in patients with acute UGIB. We also hope to evaluate if any correlation exists between early intervention and length of stay (LOS). Methods: A retrospective chart review was done of 140 patients admitted to the hospital from January 2012 to December 2012 with a diagnosis of UGIB. Twenty-six clinical/demographic variables were recorded and used to calculate GBS and CDSS percentiles, stratifying high-risk patients who could potentially benefit from emergent EGD (eEGD). eEGD is defined as requiring endoscopy within 24 hours of admission. The CDSS was constructed employing the random forest (RF) mathematical model to correctly predict the need for eEGD. Applying Cohen’s Kappa and logistic regression, we tested for an association between CDSS, GBS, and clinical decision for eEGD. Wilcoxon rank-sum test enabled comparison of LOS between patients who underwent eEGD and those who did not. Results: Of the 140 patients identified, 126 underwent EGDs. GBS scores for all these patients categorized them as “high risk” (score >0), warranting intervention. CDSS recommended eEGD for 42 patients (33.3%). Based on the clinician’s judgment, 46 patients (36.5%) had eEGD. Although the proportions are similar, there was poor agreement between the 2 groups, kappa=-0.0814, 95% confidence interval (CI) -0.2518, 0.0890. The median LOS (IQR) of patients who underwent eEGD based on clinician’s judgment was 3 (2,6) days compared to a median LOS (IQR) of 5 (3,7) days for patients without eEGD (p=0.0012). There were no statistical differences in the LOS between patients recommended and those not recommended to undergo eEGD by CDSS (4 days; p=0.1898). Within the group recommended to undergo eEGD based on CDSS, there was no difference in the LOS whether or not eEGD was performed (p=0.4453). Conclusion: There was no change in outcomes when comparing CDSS with clinical judgment regarding recommendations for eEGD. However, patients who had eEGD based on clinical judgment had a significantly shorter LOS than those who did not. A larger sample sized prospective study along with correlation to EGD findings would be the next steps for the validation of the CDSS model.Table 1: eEGD Recommendations
科研通智能强力驱动
Strongly Powered by AbleSci AI