Dear Editor,

An outbreak of COVID-19 has spread worldwide outside Wuhan, China [1]. It is urgent to evaluate the 15-day in-hospital mortality rate for this disease, and its associated risk factors [2].

For this single-center, ambispective cohort study of 548 patients [3], data at admission were retrospectively obtained. All the patients were followed up prospectively until day 15 after hospitalization. The primary endpoint of this study was death within 15 days of hospitalization. The risk factors for in-hospital mortality from COVID-19 were analyzed by means of multivariable Cox proportional hazards regression models.

Within 15 days of hospitalization, 78 patients died, resulting in mortality rates of around 14.2%. 127 patients were critically ill, i.e. needed ventilation support during the 15-day hospitalization. The risk of death within 15 days of admission was found to be higher in elderly male patients with a smoking history and underlying conditions such as chronic obstructive pulmonary disease, diabetes and hypertension (Supplementary Table 1). Compared with the survivors, the non-survivors more commonly presented the symptoms of dyspnea (88.5%), chest tightness (66.0%) and confusion (15.4%) (Supplementary Table 1). The non-survivors also presented with lower lymphocyte counts and higher levels of ferritin, lactate dehydrogenase (LDH) and C-reactive protein compared with the survivors (Supplementary Table 1). The majority of deaths were due to respiratory failure (88.5%), followed by cardiac arrest (5.1%) and shock (2.6%). Common complications among the death included acute respiratory distress syndrome (ARDS) (97.4%), hyperglycemia (67.9%), myocardial damage (62.8%), liver damage (24.4%), and renal damage (38.5%). Systemic glucocorticoid treatment was more common in non-survivors. Instead, more survivors than non-survivors received lopinavir/ritonavir and arbidol (33.4% vs. 9%, and 76.4% vs. 53.8%, respectively) (Supplementary Table 1).

Multivariable Cox proportional hazards regression showed that higher hazard ratios (HRs) for 15-day in-hospital death were associated with age older than 65 years, low oxygen saturation, lymphopenia, high LDH, and multiorgan dysfunction (Table 1). The most influential risk factor was severe lymphopenia (< 0.5 × 109/L), which was associated with a higher hazard for mortality (HR = 4.410) compared with a lymphocyte count > 0.5 × 109/L. The presence of multi-organ dysfunction increased the hazard (HR = 1.273, per organ increase). Treatment with lopinavir/ritonavirt (HR = 0.350) was associated with a lower hazard [4] (Table 1). Low-dose glucocorticoid treatment (< 1 mg/kg) or no glucocorticoid use was associated with a lower hazard compared with high-dose treatment (≥ 1 mg/kg) (Table 1). The risk factors for mortality in the critically ill patients are shown in Supplemental Table 2. However, in the critically ill cohort, the HR estimates were similar with and without systemic glucocorticoid therapy. This suggests glucocorticoid therapy requires strict control of indications [5]. The type of respiratory support (noninvasive vs. invasive ventilation) had no effect on mortality in critically ill patients (supplementary Table 2).

Table 1 Multivariable COX proportional risk regression modeling of risk factors for 15-day mortality in 548 inpatients
Full size table

In conclusion, 14.2% patients of this cohort reached the primary endpoint of death within 15-day hospitalization. Aging, hypoxia, lymphopenia, high LDH level and multiple organ dysfunction were associated with increased 15-day in-hospital mortality from COVID-19. Lopinavir/ritonavir was associated with decreased mortality. Among the critically ill patients, the hazard ratio was not reduced in those receiving invasive ventilation versus non-invasive ventilation.