The study was approved by the Research Ethics Committee of the Kowloon West Cluster (reference No. KW/EX-20–039; 144–27) of the Hospital Authority of Hong Kong. We provide methods for virus nucleoprotein (N) gene copy number quantification (6), virus culture, and sgRNA detection of RT-PCR–confirmed coronavirus disease (COVID-19) patients (Appendix). Virus sgRNA was tested in specimens that had >5 log₁₀ N gene copies/mL.

A total of 68 specimens from 35 patients were studied (Table 1; Appendix); patients with prolonged virus shedding (10 who remained virus RNA positive for >30 days) and patients readmitted because RT-PCR positivity was detected after discharge (n = 6) were oversampled (i.e., selected to make up a larger share of the survey sample than is performed for the patient population). Patient age ranged from 17 to 75 years (median 38 years); 23 were male and 12 female (Table 1). Specimens submitted for virus culture were nasopharyngeal aspirates and throat swab specimens (n = 46), nasopharyngeal aspirates (n = 2), nasopharyngeal swab specimens and throat swab specimens (n = 4), nasopharyngeal swab specimens (n = 3), sputum (n = 11), and saliva (n = 2). The duration after onset of illness to specimen collection ranged from 1 to 67 days.

Figure 1

Figure 1. Severe acute respiratory syndrome coronavirus 2 RNA load, virus culture, and days after onset of illness in respiratory specimens and duration of illness for patients with mild coronavirus disease, Hong...

Virus was isolated from 16 specimens for 16 patients. The median age of the culture-positive patients was 39 years and of the culture-negative patients was 38 years. SARS-CoV-2 N gene copy number in the specimens overall ranged from 9.5 log₁₀ copies/mL to undetectable (limit of detection 10 copies/mL) (Figure 1). The median viral load in culture-positive samples was 7.5 log₁₀ copies/mL and in culture-negative samples was 3.8 log₁₀ copies/mL (p = 0.00001) (Table 1).

Virus was isolated from 12 of 17 specimens with viral loads ≥7.0 log₁₀ copies/mL, 3 of 11 specimens with viral loads 6.0–6.99 log₁₀ copies/mL, 1 of 7 specimens with viral loads 5.0–5.99 log₁₀ copies/mL, and 0 of 33 specimens viral loads <5 log₁₀ copies/mL.

Figure 2

Figure 2. Severe acute respiratory syndrome coronavirus 2 viral RNA load, virus culture, and subgenomic virus RNA (sgRNA) in relation to days after onset of illness for patients with mild coronavirus disease,...

The sgRNA provides evidence of replicative intermediates of the virus, rather than residual viral RNA. Detection of virus sgRNA was attempted for 33 of the 35 the clinical specimens that had viral loads >5.0 log₁₀ virus genome copies/mL; 2 specimens had insufficient specimen for this testing. Of 33 specimens tested for sgRNA and by virus culture, both tests showed positive results for 12 (36.4%) specimens, both tests showed negative results for 12 (36.4%), sgRNA showed positive results and culture was negative for 7 (21.2%) specimens, and culture was positive and sgRNA showed negative results for 2 (6.1%) samples (Cohen κ 0.467, p = 0.005 against κ 0) indicating a moderate agreement between virus culture and sgRNA detection. Virus sgRNA was detectable in 18 (81.8%) of 22 specimens collected <8 days after symptom onset and in 1 (9.1%) of 11 specimens collected >9 days after onset of disease (p = 0.0003 by χ² test with Yates correction) (Figure 2). We also provide culture and sgRNA results stratified by specimen type (Table 2).

We conducted a subset analysis for 42 specimens collected from patients who did not receive antiviral drugs or specimens that were collected before antiviral therapy (Table 1). This sample included all 16 specimens that were culture positive and 18 of 19 specimens that were sgRNA positive. The main conclusions remained unchanged. Median viral RNA load in culture-positive specimens was 7.54 log₁₀ genome copies/mL and in culture-negative specimens was 4.0 log₁₀ genome copies/mL (p<0.00001 by Mann-Whitney 2-tailed U test). Of the 16 culture positive specimens, 15 (94%) had viral RNA load >6 log₁₀ copies/mL (p<0.01 by Fisher exact test). All of them were collected within the first 8 days of illness (p = 0.01 by Fisher exact test ) (Table 1). However, the duration of illness in this subset of specimens was limited to 31 days. Five specimens with viral load >6 log₁₀ virus N gene copies/mL collected >50 days after onset of illness were negative by virus culture and virus sgRNA, but all of these patients had received antiviral therapy.

For a cohort of patients with predominantly mild COVID-19, our findings suggest that virus isolation and sgRNA detection were positive within the first 8 days after onset of illness and mainly for specimens with >6 log₁₀ virus N gene copies/mL of clinical specimen. We did not carry out serologic testing in parallel with viral culture, but data on a larger cohort of patients in Hong Kong showed that most patients had detectable virus neutralizing antibodies after day 9 of illness (7). Two other studies of virus culture for mildly ill or moderately ill patients showed virus culture was only successful within the first 9 days after onset of illness (8,9). Patients who are severely ill and immunocompromised might shed infectious virus for much longer periods (J.J.A. van Kampen, Erasmus University Medical Center, pers. comm., 2020 Jun 9), and this shedding might also be prolonged by corticosteroid therapy.

The World Health Organization has recently amended its guidelines for releasing COVID-19 patients from isolation (i.e., 10 days after symptom onset and >3 additional days without symptoms), but these guidelines do not distinguish between mild and severely ill patients (10). Our findings suggest that patients with mild or moderate illness might be less contagious 8 days after symptom onset. Mildly ill patients who have clinically recovered and are not immunocompromised might be discharged from containment >9 days after symptom onset, as long as they are not being discharged into settings that contain other highly vulnerable persons (e.g., old age care homes).