We investigated association between epidemiological and clinical characteristics of coronavirus disease 2019 (COVID-19) patients and clinical outcomes in Korea. This nationwide retrospective cohort study included 5621 discharged patients with COVID-19, extracted from the Korea Disease Control and Prevention Agency (KDCA) database. We compared clinical data between survivors (n = 5387) and non-survivors (n = 234). We used logistic regression analysis and Cox proportional hazards model to explore risk factors of death and fatal adverse outcomes. Increased odds ratio (OR) of mortality occurred with age (≥ 60 years) [OR 11.685, 95% confidence interval (CI) 4.655-34.150, p < 0.001], isolation period, dyspnoea, altered mentality, diabetes, malignancy, dementia, and intensive care unit (ICU) admission. The multivariable regression equation including all potential variables predicted mortality (AUC = 0.979, 95% CI 0.964-0.993). Cox proportional hazards model showed increasing hazard ratio (HR) of mortality with dementia (HR 6.376, 95% CI 3.736-10.802, p < 0.001), ICU admission (HR 4.233, 95% CI 2.661-6.734, p < 0.001), age ≥ 60 years (HR 3.530, 95% CI 1.664-7.485, p = 0.001), malignancy (HR 3.054, 95% CI 1.494-6.245, p = 0.002), and dyspnoea (HR 1.823, 95% CI 1.125-2.954, p = 0.015). Presence of dementia, ICU admission, age ≥ 60 years, malignancy, and dyspnoea could help clinicians identify COVID-19 patients with poor prognosis.