Skip to main content

Prolonged persistence of SARS-CoV-2 RNA in body fluids

Our take —

This study followed individuals over time and found that patients who had more severe disease also shed viral RNA for a longer duration (median: >30 days) than patients who had mild disease (median: 15-24 days). This trend was true across all sample types collected (e.g., throat swabs, nasopharyngeal swabs, sputum samples, and fecal specimen). However, this study was relatively small and had few severe cases. Therefore, the results should be interpreted with caution (i.e., the results may be due to this specific sample of people and may not be generalizable to all populations/situations). Lastly, when interpreting these results, it should be noted that infectiousness is not always correlated with the shedding of detectable SARS-CoV-2 RNA.

Study design

Prospective Cohort

Study population and setting

This cohort consisted of 43 patients with mild COVID-19 symptoms and 6 patients with severe COVID-19 symptoms who had been admitted to two hospitals in Guangdong Province, China. Researchers collected samples from patients every three days, for up to four weeks after symptom onset or until discharge. In total, there were 490 samples from throat swabs (n=88), nasopharyngeal swabs (n=175), sputum samples (n=62), and fecal specimens (n=165). Researchers tested the samples for SARS-CoV-2 RNA by RT-PCR and modeled the time to loss of detectable viral RNA using parametric regression models.

Summary of Main Findings

The median time from symptom onset to undetectable viral RNA was greater than 14 days for all sample types and was longer in patients with severe disease as compared to patients with mild disease. Comparing severe to mild disease, the median times were 33.9 versus 15.6 days based on throat swabs, 33.5 versus 22.7 days based on nasopharyngeal swabs, 30.9 versus 20.0 days based on sputum samples, and 32.5 versus 24.5 days based on fecal specimens.

Study Strengths

This study followed patients prospectively, collecting both multiple samples and multiple sample types from each patient over a period of up to four weeks. Researchers performed sensitivity analyses showing that the type of parametric model used did not impact the conclusions.


The researchers had far fewer samples than the needed numbers to detect a statistical difference in the outcome between sample types. Additionally, this study included just six patients with severe disease and no patients with asymptomatic infections. Finally, the researchers did not provide evidence on whether the various PCR positive samples could cause disease in others (i.e., infectiousness).

Value added

This study contributes to the growing literature on the duration of detectable viral RNA following symptom onset and expands our knowledge on this topic by demonstrating its relationship with disease severity and that the trend is consistent across multiple sample types.

This review was posted on: 27 September 2020