ICMR- National Institute of Cholera and Enteric Diseases, Beliaghata, Kolkata, India.
PLoS One. 2022 Sep 9;17(9):e0273006. doi: 10.1371/journal.pone.0273006. eCollection 2022.
To develop an accurate lab score based on in-hospital patients' potent clinical and biological parameters for predicting COVID-19 patient severity during hospital admission.
To conduct this retrospective analysis, a derivation cohort was constructed by including all the available biological and clinical parameters of 355 COVID positive patients (recovered = 285, deceased = 70), collected in November 2020-September 2021. For identifying potent biomarkers and clinical parameters to determine hospital admitted patient severity or mortality, the receiver operating characteristics (ROC) curve and Fischer's test analysis was performed. Relative risk regression was estimated to develop laboratory scores for each clinical and routine biological parameter. Lab score was further validated by ROC curve analysis of the validation cohort which was built with 50 COVID positive hospital patients, admitted during October 2021-January 2022.
Sensitivity vs. 1-specificity ROC curve (>0.7 Area Under the Curve, 95% CI) and univariate analysis (p<0.0001) of the derivation cohort identified five routine biomarkers (neutrophil, lymphocytes, neutrophil: lymphocytes, WBC count, ferritin) and three clinical parameters (patient age, pre-existing comorbidities, admitted with pneumonia) for the novel lab score development. Depending on the relative risk (p values and 95% CI) these clinical parameters were scored and attributed to both the derivation cohort (n = 355) and the validation cohort (n = 50). ROC curve analysis estimated the Area Under the Curve (AUC) of the derivation and validation cohort which was 0.914 (0.883-0.945, 95% CI) and 0.873 (0.778-0.969, 95% CI) respectively.
The development of proper lab scores, based on patients' clinical parameters and routine biomarkers, would help physicians to predict patient risk at the time of their hospital admission and may improve hospital-admitted COVID-19 patients' survivability.
基于住院患者的有效临床和生物学参数，开发一种准确的实验室评分，以预测住院期间 COVID-19 患者的严重程度。
为了进行这项回顾性分析，我们构建了一个推导队列，其中包含了 2020 年 11 月至 2021 年 9 月期间收集的 355 例 COVID 阳性患者（康复=285，死亡=70）的所有可用生物学和临床参数。为了确定有潜力的生物标志物和临床参数来确定住院患者的严重程度或死亡率，我们进行了接收者操作特征（ROC）曲线和 Fischer 检验分析。相对风险回归用于为每个临床和常规生物学参数开发实验室评分。实验室评分还通过 2021 年 10 月至 2022 年 1 月期间住院的 50 例 COVID 阳性患者组成的验证队列的 ROC 曲线分析进行验证。
推导队列的灵敏度与 1 特异性 ROC 曲线（>0.7 曲线下面积，95%置信区间）和单变量分析（p<0.0001）确定了五种常规生物标志物（中性粒细胞、淋巴细胞、中性粒细胞：淋巴细胞、白细胞计数、铁蛋白）和三个临床参数（患者年龄、预先存在的合并症、入院时患有肺炎）用于开发新的实验室评分。根据相对风险（p 值和 95%置信区间），对这些临床参数进行评分，并分别分配给推导队列（n=355）和验证队列（n=50）。ROC 曲线分析估计了推导和验证队列的曲线下面积（AUC），分别为 0.914（0.883-0.945，95%置信区间）和 0.873（0.778-0.969，95%置信区间）。
基于患者的临床参数和常规生物标志物，开发适当的实验室评分将有助于医生在患者入院时预测其风险，并可能提高住院 COVID-19 患者的生存率。
