Background: We aimed to analyze clinical characteristics and find potential factors to predict poor prognosis in patients with coronavirus disease 2019 (COVID-19). Methods: We analyzed the clinical characteristics and laboratory tests of COVID-19 patients and detected SARS-CoV-2 RNA in urine sediments collected from 53 COVID-19 patients enrolled in Renmin Hospital of Wuhan University from 31 January 2020 to 18 February 2020 with qRT-PCR analysis. Then, we classified those patients based on clinical conditions (severe or non-severe syndrome) and urinary SARS-CoV-2 RNA (URNA− or URNA+). Results: We found that COVID-19 patients with severe syndrome (severe patients) showed significantly higher positive rate (11 of 23, 47.8%) of urinary SARS-CoV-2 RNA than non-severe patients (4 of 30, 13.3%, p = 0.006). URNA+ patients or severe URNA+ subgroup exhibited higher prevalence of inflammation and immune discord, cardiovascular diseases, liver damage and renal dysfunction, and higher risk of death than URNA− patients. To understand the potential mechanisms underlying the viral urine shedding, we performed renal histopathological analysis on postmortems of patients with COVID-19 and found severe renal vascular endothelium lesion characterized by an increase of the expression of thrombomodulin and von Willebrand factor, markers to assess the endothelium dysfunction. We proposed a theoretical and mathematic model to depict the potential factors that determine the urine shedding of SARS-CoV-2. Conclusions: This study indicated that urinary SARS-CoV-2 RNA detected in urine specimens can be used to predict the progression and prognosis of COVID-19 severity.