Risk Factors for Mortality in 244 Older Adults With COVID-19 in Wuhan, China

A Retrospective Study

Haiying Sun, MD; Ruoqi Ning, MD; Yu Tao, MD; Chong Yu, MD; Xiaoyan Deng, MD; Caili Zhao, MM; Silu Meng, MM; Fangxu Tang, MD; Dong Xu, MD, PhD


J Am Geriatr Soc. 2020;68(6):E19-E23. 

In This Article


A total of 244 older patients with a definite clinical outcome recorded by March 5, 2020, were enrolled in the study, including 123 who were discharged and 121 who died (Table 1). The median age of the discharged patients was 67 years, while that of the deceased group was 72 years (P < .001). Among all patients, 58.5% in the discharged group and 32.2% in the deceased group were female (P < .001). Most patients exhibited fever (86.5%) and respiratory symptoms (88.5%), and the main respiratory symptom was dry cough (73.4%). Approximately one-third of patients experienced digestive symptoms, such as diarrhea. Among all patients, 21% had diabetes, while 14.4% had coronary heart disease. These two common chronic comorbidities of older patients were not associated with the outcome, while the proportion of hypertension was significantly different between the two groups (P = .042). While 16.7% of deceased patients had a history of respiratory problems, only 3.3% of the discharged patients had a history of respiratory problems (P < .001). Approximately 13.1% of the patients were admitted with consciousness disorders. Approximately one-third of the patients were admitted with oxygen saturation less than 90%. The median SpO2 was 97% and 90% for discharged and deceased patients, respectively (P < .001). The respiratory rate and heart rate of the deceased patients were higher than values of those who were discharged (P < .001). Vital signs on admission, including consciousness disorders, SpO2 less than 90%, respiratory rate greater than 20 breaths/min, and heart rate greater than 100 beats/min, were often associated with poor outcomes.

Most laboratory variables, including white blood cell (WBC) count, lymphocyte (LYM) count, amino-terminal pro-brain natriuretic peptide, alanine aminotransferase, aspartate aminotransferase, high-sensitivity cardiac troponin I (hs-TnI), D-dimer and creatinine levels, estimated glomerular filtration rate, high-sensitivity C-reactive protein, procalcitonin, serum ferritin, and interleukin (IL)-6 levels, were significantly different between the two groups. There was no statistical difference in erythrocyte sedimentation rate between the two groups (Table 1). The comparison of outcomes, clinical characteristics, and laboratory variables between older male and female patients revealed several differences (Supplementary Table S1). To eliminate bias possibly caused by sex, the clinical characteristics and laboratory variables between the two groups for male and female patients were compared separately, and the results were similar to those of all patients (Supplementary Table S2).

Statistically significant variables identified in the univariate analysis were entered into a multivariable logistic regression analysis (Table 2). Due to many missing values (>30%), serum ferritin and IL-6 levels were excluded. Age and LYM count were significantly different (odds ratio [OR] = 1.122; 95% confidence interval [CI] = 1.007–1.249; P = .037; and OR = 0.009; 95% CI = 0.001–0.138; P = .001, respectively), indicating that older age and lower LYM count on admission were independently associated with increased risk for death. In addition, WBC count demonstrated a P value of .052 (OR = 1.28; 95% CI = 1.00–1.64). The results of logistic regression and the independent risk factors (ie, age and LYM count) were used to generate ROC curves (Figure 1). The area under the ROC curve in the logistic regression model was 0.913, and that for age and LYM count were 0.653 and 0.823, respectively, indicating that LYM count and age were the most important risk factors for death.

Figure 1.

Receiver operating characteristic (ROC) curves for univariate and multivariate logistic regression models of risk factors. The purple curve is the ROC curve for the logistic regression model of age alone. The red curve is the ROC curve of the logistic regression model of lymphocyte (LYM) count alone. The blue curve is the ROC curve for the multivariable logistic regression model.