Department of Clinical Laboratory, Shengli Clinical Medical College of Fujian Medical University, Fuzhou, Fujian, China.
Department of Clinical Laboratory, Maternal and Child Health Hospital of Xianyou County, Putian, Fujian, China.
Sci Rep. 2024 Nov 25;14(1):29146. doi: 10.1038/s41598-024-80310-8.
The coronavirus disease 2019 (COVID-19) has exhibited escalating contagion and resistance to immunity, resulting in a surge in infections and severe cases. This study endeavors to formulate two nomogram predictive models aimed at discerning patients at heightened risk of severe and fatal outcomes upon hospital admission. The primary objective is to enhance clinical management protocols and mitigate the incidence of severe illness and mortality associated with COVID-19.
1600 patients diagnosed with COVID-19 and discharged from Fujian Provincial Hospital were chosen as the subjects of this study. These patients were categorized into three groups: mild group (n = 940), severe group (n = 433), and fatal group (n = 227). The patients were randomly divided into training and validation cohorts in a 7:3 ratio. COVID-19 symptoms were treated as dependent variables, and univariate regression analysis was conducted for the laboratory indicators. Risk factors with p-values greater than 0.05 in the univariate regression analysis were eliminated. The remaining risk factors were then analyzed using direct multiple regression analysis to establish an unadjusted model. Subsequently, risk factors with p-values greater than 0.05 were further removed. Clinical characteristics were added to the model as adjustment factors, and the method of multiple stepwise regression analysis was employed to derive the final fully adjusted model. The severe and fatal COVID-19 models were converted into nomograms, respectively. Receiver operating characteristic (ROC) curves were utilized to evaluate the discrimination of the nomogram models. Calibration was assessed using the Hosmer-Lemeshow test and calibration curves. Clinical benefit was evaluated by decision curve analysis.
Compared to the mild group, individuals in the severe COVID-19 group exhibited significant increases in age, neutrophil (NEU), and lactate dehydrogenase (LDH) levels, alongside notable decreases in lymphocyte (LYM) and albumin (ALB) levels. Nomogram model incorporating age, NEU, LDH, LYM, and ALB demonstrated efficacy in predicting the onset of severe COVID-19 (AUC = 0.771). Furthermore, history of cerebral infarction and cancer, LDH and ALB as risk factors for fatal COVID-19 cases compared to the severe group. The nomogram model comprising these factors was capable of early identification of COVID-19 fatalities (AUC = 0.748).
Elevated age, NEU, and LDH levels, along with decreased LYM and albumin (ALB) levels, are risk factors for severe illness in hospitalized patients with COVID-19. A history of cerebral infarction and tumors, along with elevated LDH and decreased ALB levels, are risk factors for death in critically ill patients. The nomogram model based on these factors can effectively predict the risk of severe or fatal illness from COVID-19, thereby assisting clinicians in timely interventions to reduce the rates of severe illness and mortality among hospitalized patients. However, the model faces challenges in processing longitudinal data and specific points in time, indicating that there is room for improvement.
2019 年冠状病毒病（COVID-19）的传染性不断增强，对免疫的抵抗力也不断增强，导致感染和重症病例急剧增加。本研究旨在制定两个列线图预测模型，以区分入院时发生严重和致命结局风险较高的患者。主要目的是改进临床管理方案，降低 COVID-19 相关严重疾病和死亡的发生率。
选择福建省立医院收治的 1600 例 COVID-19 出院患者作为研究对象。将这些患者分为三组：轻症组（n=940）、重症组（n=433）和死亡组（n=227）。患者按 7：3 的比例随机分为训练和验证队列。将 COVID-19 症状作为因变量，对实验室指标进行单因素回归分析。单因素回归分析中 p 值大于 0.05 的危险因素被剔除。然后对剩余的危险因素进行直接多元逐步回归分析，建立未调整模型。随后，剔除 p 值大于 0.05 的危险因素，将临床特征作为调整因素加入模型，采用多元逐步回归分析方法得出最终完全调整模型。将严重和致命 COVID-19 模型分别转化为列线图。使用受试者工作特征（ROC）曲线评估列线图模型的判别能力。采用 Hosmer-Lemeshow 检验和校准曲线评估校准。通过决策曲线分析评估临床获益。
与轻症组相比，重症 COVID-19 组患者的年龄、中性粒细胞（NEU）和乳酸脱氢酶（LDH）水平显著升高，淋巴细胞（LYM）和白蛋白（ALB）水平显著降低。纳入年龄、NEU、LDH、LYM 和 ALB 的列线图模型在预测严重 COVID-19 方面具有良好的效果（AUC=0.771）。此外，与重症组相比，脑梗死和癌症史、LDH 和 ALB 是 COVID-19 死亡病例的危险因素。包含这些因素的列线图模型能够早期识别 COVID-19 死亡病例（AUC=0.748）。
入院 COVID-19 患者中，高龄、NEU 和 LDH 水平升高，LYM 和白蛋白（ALB）水平降低是发生重症的危险因素。脑梗死和肿瘤史、LDH 升高和 ALB 降低是重症患者死亡的危险因素。基于这些因素的列线图模型可以有效预测 COVID-19 严重或致命疾病的风险，从而帮助临床医生及时进行干预，降低住院患者的严重疾病和死亡率。然而，该模型在处理纵向数据和特定时间点时存在挑战，表明仍有改进的空间。
