Gastrointestinal
shedding
Key points
• SARS-CoV-2 isolation from feces and RNA detection regardless of gastrointestinal symptoms
• Faecal sampling not recommended for diagnostic screening (unless laboratory diagnosis of suspected cases with negative respiratory tract results)
• Prolonged gastrointestinal viral RNA detection up to seven weeks, well after respiratory tract clearance and symptoms resolution in some patients
We reviewed 47 studies providing data on gastrointestinal (GI) sampling in ≥629 COVID-19 patients, including stool specimens (≥486 cases), anal/rectal swabs (≥198), and others (endoscopic samples, n=14) (Supplementary Dataset) [9, 27, 39, 40, 42, 43, 48-51, 53, 58-60, 63, 64, 77, 84-86, 89, 91, 93, 100, 102, 105, 109, 117, 120, 121, 126, 131, 137, 152, 158, 168, 171, 172, 178, 179, 191, 203, 206, 229, 237, 240, 271].
SARS-CoV-2 was isolated from stool sample 15 dps from a COVID-19 patient with severe pneumonia (using Vero cells) [271] and from two patients without diarrhoea [77]. In a study involving nine mild cases, virus isolation (on Vero E6 cells) was unsuccessful in stool samples 6-12 dps from 4 patients, and no virus replication evidence was found through sgRNA assays despite detectable high VL [131]. SARS-CoV-2 nucleocapsid protein was detected in the cytoplasm of gastric, duodenal, and rectum glandular epithelial cells in one patient [58]. The gastric fluid samples of six of 13 critically ill patients were positive for SARS-CoV-2 RNA [172].
GI disease has been described for some COVID-19 patients [9, 10, 12, 14, 16, 63, 83, 110, 272] and was one of the clinical signs associated with a positive SARS-CoV-2 test [272]. Though there was evidence of SARS-CoV-2 RNA detection in the GI tract, it was not necessarily in cases with GI symptoms [48, 102]. Examining studies with an available timeline of sampling (Figure 1, Figure 2, Table, and Supplementary Table) SARS-CoV-2 RNA detection was reported between 3 and 50 dps in stool samples of 14 mild [59, 63, 120, 131] and 11 severe cases [58, 59, 172], regardless of the presence of diarrhoea. Anal swabs had detectable SARS-CoV-2 RNA between 3 and 45 dps in six mild [158, 171, 206] and seven severe cases [42, 172]. No systematic comparison between viral detection in stool and anal swabs was available.
Prolonged GI SARS-CoV-2 RNA detection after the resolution of respiratory symptoms and/or convalescence was observed in several studies [53, 58, 60, 102, 131, 137, 191], though infectious virus shedding is still an outstanding question. In a German study, stool samples remained RNA positive ≥21 dps for 6 mild cases, including a patient with potential independent intestinal tract replication, suggested by the authors by comparison with the SARS-CoV-2 URT kinetics [131]. Zheng and colleagues reported a 59% positivity rate in 842 stool samples from 96 patients and a median viral RNA detection duration of 22 (IQR 17-31) dps that was significantly longer than in sputum/saliva and serum samples [191]. In a study on 98 COVID-19 cases (18 severe), paired OP swabs and stool specimens for 74 cases yielded the following results: 41 with SARS-CoV-2 detection in stool for a mean 27.9 dps (SD 10.7) that was 11 days after the clearance in OP swabs with mean detection 16.7 dps (SD 6.7); while the remaining 33 patients with negative stool results had positive OP swabs for a mean 15.4 dps (SD 6.7). A patient had detectable SARS-CoV-2 RNA in stool 47 dps and another for 33 days after respiratory clearance. Disease severity was not associated with prolonged GI shedding in this cohort [102]. In another cohort of 42 patients (11 severe cases) with GI symptoms, 28 cases (nine severe) had a median RNA detection period of 11 (IQR 7-13) dps until first positive stool sample compared to 6.5 (IQR 3‐7.25) dps for OP swabs. More than half (n=18, 5 severe) remained stool positive for median 7 (IQR 6‐10) days after negative OP swabs [137]. A total of 39 (53%) of 73 hospitalized patients had detectable SARS-CoV-2 RNA in stool and for 17 patients (20%) it remained positive after respiratory samples turned negative [58]. Another Chinese study on recovering patients (n=55) found a median of 11 (9-16) dps until the first negative results in stool: 43 patients had a 2 (1-4) days median delay in clearance in feces compared to OP swabs, while in 12 patients both turned negative at the same time [53]. Sun and colleagues aggregated data on 165 stool samples and estimated a median/95th percentile time until loss of detection of 24.5/45.6 dps for 43 mild cases and 32.5/48.9 dps for six severe cases, both comparable to the estimates for NP swabs [240].
Like observations for stool specimens, prolonged SARS-CoV-2 RNA detection was reported for anal swabs [42, 70, 158, 168, 171, 172, 206]. Presence of SARS-CoV-2 RNA in anal swabs seemed linked to disease severity in a Guangzhou cohort (two severe and 16 mild cases) [9]. Zhang and colleagues found positive anal swabs in four of 16 patients upon hospitalization and six of 16 cases at day 5 [27]. A Chinese study with discharge criteria of two consecutive negative OP swab and a negative anal swab, reported a median RNA detection duration of 12 (IQR 9-14, range 4-34) days for 24 patients [158]. SARS-CoV-2 RNA detection in anal swabs ≥17 days was observed in an asymptomatic patient [70].
Large well-documented cohort studies are needed to estimate the proportion of COVID-19 cases with continuous GI shedding and the SARS-CoV-2 VL levels over time. A 29% SARS-CoV-2 RNA positivity rate in stool samples was observed in a study aggregating data on 205 patients (unclear how many provided the analysed 153 stool samples) [77]. Few studies (n=10) provided quantitative data on SARS-CoV-2 RNA detection in stool and anal swabs for 38 COVID-19 cases [27, 59, 63, 131, 158, 168, 171, 172, 206]. There was no significant difference between the VL in stool between 22 mild and 71 severe cases [191]. SARS-CoV-2 GI VL in adults seemed to be subjectively lower (higher Ct values) than in the respiratory tract in 25 cases [59, 63, 131, 168, 171, 172, 206], and higher (lower Ct values) in four mild cases [131, 168, 171], though meaningful conclusions cannot be drawn from such small sample sizes and non-systematic observations.
SARS-CoV-2 detection in
blood
Key points
• Blood sampling not recommended for initial diagnostics
• No evidence of SARS-CoV-2 isolation from blood nor blood-borne transmission
• SARS-CoV-2 RNA detection as a sign of severe disease up to 4 weeks post symptoms onset and use as a clinical monitoring tool
We reviewed 32 studies providing blood samples (whole blood, plasma or serum) data of ≥389 COVID-19 patients (Supplementary Dataset) [9, 10, 27, 39, 42, 48, 49, 51, 53, 58, 59, 63, 64, 77, 89, 105, 109, 118, 120, 122, 131, 133, 134, 145, 155, 172, 178, 182, 191, 196, 206, 223]. A systematic comparison of SARS-CoV-2 RNA detection in different types of blood samples was lacking. No virus isolation from blood samples was reported.
Summarizing data on COVID-19 patients with known infection timeline (Figure 1, Figure 2, Table, and Supplementary Table) SARS-CoV-2 RNA was detected 3-18 dps in 14 patients: 18 blood samples of ten severe cases [9, 42, 58, 59, 172] and 4 samples of four mild cases respectively [9, 39, 206]. SARS-CoV-2 RNA presence in blood was linked with disease severity [42] and reported in further 54 severe cases [10, 27, 105, 109, 120, 172, 182, 191]. Additionally, viral RNA was detected in blood samples from 35 mild cases [10, 27, 105, 109, 178, 191, 206], one asymptomatic infant [51] and 3 samples from unspecified cases [77]. SARS-CoV-2 RNA detection in blood might be useful as a laboratory sign of deterioration in severe cases. SARS-CoV-2 RNA was detected in blood samples from 16 mild cases for 10±6 days and seven ICU patients for 15±6 days [105]. A Chinese study reported a median viral RNA detection duration in the serum of 16 (IQR 11-21) dps, and 27% serum positivity rates in 22 mild cases compared to 45% in 74 severe cases. Detection rates peaked in weeks 2-3 since symptoms onset in all patients with detectable SARS-CoV-2 RNA (17 severe and 3 mild cases) and dropped to 11% (n=5) for severe cases and 0 for mild cases in week 4. However, VL had no significant difference between severe and mild cases [191]. A patient in critical condition had lower, but detectable SARS-CoV-2 RNA in plasma (Ct 35.8-38.4, 7-12 dps) compared to NP swabs (6.7-4.4 log10 copies/1000 cells, 7-24 dps) [120].
None of nine adults diagnosed with COVID-19 using OP swabs had detectable viral RNA in blood when tested with three different kits [48]. Although SARS-CoV-2 was detected and successfully isolated from respiratory samples, all 31 serum samples from nine mild cases tested negative [131]. Finally, none of the serum samples from 14 convalescent patients (no respiratory symptoms and two consecutive negative OP swabs) were positive for SARS-CoV-2 RNA despite simultaneous detection in OP swabs and stool [53].
Other specimens: oral fluid, tears, urine, cerebrospinal
fluid, peritoneal fluid,
semen
Key points
• Oral fluid/saliva as a self-collectable alternative to respiratory sampling
• SARS-CoV-2 RNA detection in oral fluid/saliva up to 4 weeks
• Rare RNA detection and no SARS-CoV-2 isolation in conjunctival secretions
• Limited SARS-CoV-2 RNA detection in urine and no virus isolation
• SARS-CoV-2 visualisation in brain tissue and viral RNA detection in cerebrospinal fluid
• SARS-CoV-2 RNA detection in semen
Fifteen studies reported on oral fluid sampling (with varying collection methods) in >339 COVID-19 cases [21, 62, 89, 97, 105, 109, 148, 164, 178, 190, 191, 194, 208, 209, 265]. No study compared the different collection methods, e.g. self-collection, sampling by a healthcare worker, swabbing, stimulated secretion. Self-collected deep throat (posterior oropharyngeal) saliva was suggested as an alternative to sputum and yielded positive PCR results in 11 out 12 hospitalized COVID-19 patients in Hong Kong, as well as three positive and two negative virus cultures [21]. Further 23 patients with 173 saliva and endotracheal aspirate samples, studied by the same group, had median VL 5.2 log10 copies/mL (IQR 4.1–7.0) at presentation. The highest saliva VL were reported in week 1 since symptoms onset (20 patients), followed by a gradual decline, and prolonged RNA detection ≥20 days (seven patients) [109]. In a summary of the first cases in Hong Kong viral loads in saliva were reported as 5.9x106 copies/ml compared to 3.3x106copies/ml in combined NP+OP swabs [62]. An Australian study screened 522 paired saliva and NP swabs and detected SARS-CoV-2 RNA in 33 saliva samples out of 39 cases confirmed by NP swab. Viral loads were lower in saliva compared to NP swab with both positive up to 21 dps. Among 50 subjects with negative NP swab, one had a positive saliva sample [190]. Similarly, a Thai study screened 200 participants with paired saliva and NP+OP swabs and found 16 cases with matched positive saliva and swabs, two with only saliva positive and three with only NP+OP swab positive. Viral loads were comparable and a 97.5% agreement was observed between saliva and combined NP+OP swabs [265]. A pre-print US study (later published in NEJM [273]) including 44 cases reported comparable/superior sensitivity of saliva to NP swabs and higher SARS-CoV-2 saliva VL for 38 matched samples [194]. In another pre-print study saliva was collected from 31 patients (26 mild, 5 severe cases) after stimulation of the salivary gland, paired with OP swab, and tested positive for SARS-CoV-2 RNA in four (3 severe, 1 mild) out of 13 cases with positive OP swab [97]. A study in Zhejiang confirmed COVID-19 in 96 patients by testing 668 sputum and 1178 saliva samples but did not specify samples types positivity rates separately. Taken together the latter declined from 95 to 54% in weeks 1-4 since symptoms onset with a median RNA detection duration of 18 (IQR 13-29) days [191]. In 25 cases SARS-CoV-2 RNA detection in the saliva was reported for 13±5 days in mild cases and 16.5±6 days in ICU patients [105]. A mild case in Wuhan had SARS-CoV-2 RNA detectable in OP swabs, saliva (Ct=18.7), and urine sediment 54 dps, and continuous detection in OP swabs over 70 dps [208]. SARS-CoV-2 was detected in all saliva samples collected by drooling technique from 25 severe cases [164], including two patients with same-day negative respiratory sampling in NP and bronchoalveolar swabs [164, 209]. A comparison between throat wash with saline solution and NP swabs collected 48-57 dps in 11 cases found inconsistent results and higher positivity rates in throat wash [148].
We reviewed six studies reporting conjunctival swab sampling in 137 COVID-19 cases [45, 105, 111, 120, 130, 172]. SARS-CoV-2 RNA was detected, but virus not isolated, in the tears and conjunctival secretions of one mildly symptomatic patient with conjunctivitis [45], while samples and cultures from 46 patients without ocular symptoms were negative [45, 111]. SARS-CoV-2 RNA detection in tears was also reported for one critically ill patient [172] and in 5 cases with unspecified disease severity [105]. Two severe cases out of 12 COVID-19 patients with ocular symptoms had positive conjunctival swabs [130]. No SARS-CoV-2 was detected in conjunctival swabs of 4 cases, incl. a severe one with conjunctivitis [120]
We reviewed 31 studies featuring urine samples from ≥369 patients [9, 39, 40, 43, 48, 49, 51, 53, 58, 59, 63, 64, 77, 89, 105, 109, 120, 122, 129, 131, 133, 137, 172, 178, 191, 197, 206, 208, 223, 237, 274]. None reported SARS-CoV-2 isolation from urine. A letter published shortly after the cut-off date of this review (not included in Supplementary dataset) described successful isolation of SARS-CoV-2 (on Vero E6 cells) at 12 dps in a severe case [275]. SARS-CoV-2 RNA was only detected in the urine of four patients (three patients with a positive sample upon OP swab turning negative) [53], at 7 dps in one woman with positive OP swab [206], in one critically ill patient with suspected systemic COVID-19 infection [172], in a case with urine abnormalities that later developed severe COVID-19 [274], and in a neonate with mild infection 6-17 dps [178]. Urine sediments were positive for SARS-CoV-2 RNA in five COVID-19 cases [129, 208]
Cerebrospinal fluid (CSF) sampling for a total of 14 patients was reported in eight studies [128, 138, 185, 189, 222, 230, 252, 257]. SARS-CoV-2 presence in the brain was evidenced post-mortem in a severe case: viral particles were observed in the frontal lobe and tissues samples, but not CSF, had detectable viral RNA [189]. SARS-CoV-2 RNA was detected (Ct>36 for N target only in a N/N2-based Japanese assay) in the CSF of a patient with meningitis 9 dps [138]. However, viral RNA was not detectable in CSF of six patients with Guillain‐Barré syndrome [185, 252], nor of 2 patients with mild respiratory symptoms and suspected viral meningoencephalitis [230], nor of the two patients with intracranial haemorrhage and positive NP swabs [222], nor of a child with Kawasaki disease [257]. CSF was not tested in a patient with suspected COVID-19 related acute necrotizing encephalopathy [128], nor in another two with Guillain‐Barré syndrome [174, 244].
SARS-CoV-2 was not detected in peritoneal fluid samples collected during an appendectomy in a patient without respiratory symptoms, but positive NP swab [242].
We reviewed 5 studies reporting on semen testing in 102 COVID-19 cases [181, 197, 235, 267, 268]. SARS-CoV-2 RNA was detected in semen from six (four with acute COVID-19 and two recovering) out of 38 patients [235]. SARS-CoV-2 was not detected in semen collected during the recovery of 64 cases (9 convalescent severe, 54 mild, 1 asymptomatic) [181, 197, 267, 268], nor in the testis samples from a deceased severe case [181].
Pregnancy and infancy, vaginal sampling
We reviewed 30 studies including 400 pregnant women and their infants [19, 47, 54, 57, 64, 66, 81, 84, 92, 106, 110, 115, 118, 127, 146, 159, 162, 163, 173, 179, 201, 204, 210, 213, 223, 225, 226, 231, 234, 255]. No confirmed mother-to-child transmission was reported for these babies delivered mainly via Caesarean section. A single infant, delivered via C-section, isolated, and formula-fed, had a positive pharyngeal swab 36 hours after delivery. Cord blood and placenta tested negative, but it was uncertain whether it was a vertical transmission or nosocomial infection [84]. Similarly, another infant delivered via C-section and quarantined, had 5 NP swabs collected until 16 days of age negative for SARS-CoV-2 RNA, but serology suggestive of in utero infection. Amniotic fluid or placenta were not tested and vertical transmission could not be confirmed/excluded [115]. Buonsenso and colleagues reported on a single case of SARS-CoV-2 RNA detectable in placenta, umbilical cord blood and 3 of 5 breast milk samples collected in the first 5 days since delivery via Caesarean section. The infant was fed with expressed breast milk that tested negative on days 14-17 and his NP swabs were negative on days 1, 3, 18 as well as the OP and anal swabs on day 18 [225]. An infant and her mother were diagnosed with COVID-19 using NP swabs on day 7 post-delivery via C-section. The infant had negative NP swabs after 14 days and horizontal transmission was considered most likely [238]. Altogether SARS-CoV-2 RNA was not detected in 41 amniotic fluid [19, 54, 64, 92, 162, 179, 201, 223], nor in 41 umbilical cord blood [19, 54, 64, 84, 92, 162, 201, 225], nor in 10 placental tissue samples [54, 57, 64, 84, 92, 223, 225]. It is unclear whether SARS-CoV-2 could be transmitted through breastmilk, so far it was not detected in samples from 39 mothers [19, 51, 64, 84, 91, 92, 162, 178, 201, 225, 226, 237, 238].
It is unclear whether shedding occurs in the female reproductive systems posing risks for vaginal delivery and sexual intercourse. In a study of 35 mild cases with age range 37-88 years, viral RNA was not detected in exfoliated cervical cells nor vaginal fluid [226]. Vaginal fluid and vaginal swabs in further 22 women also tested negative [92, 115, 133, 179, 201, 223, 237] incl. 10 severe cases [133].
Children and adolescents
We reviewed 64 studies describing COVID-19 in 1510 children and adolescents [9, 17, 31, 32, 45, 49, 51, 56, 68, 74, 83, 84, 86, 87, 89, 91, 93-95, 100, 101, 106, 108, 112, 114, 115, 121, 126, 140, 143-145, 155, 156, 160, 166, 168, 177, 178, 183, 184, 193, 199, 203, 205, 212, 214, 225, 227, 229, 232, 233, 236-238, 243, 246, 247, 249-251, 253, 257, 264]. Some further studies included cases <18 years old grouped with adults and/or provided incomplete stratification by age (marked as NA in Children No column of Supplementary dataset). In a study of 36 mild cases with ages 1-16 years it took in mean 10 (range 7-22) days until first negative results in NP swabs [114]. An asymptomatic 6-month infant maintained detectable SARS-CoV-2 RNA in NP swabs until 17 days of hospitalization and had a positive stool sample on day 9 [51]. In an Italian study on 168 paediatric cases (4 asymptomatic, 131 mild, 33 severe) nationwide diagnosed via NP/OP swabs 67% had at least 1 parent testing positive [233]. Verdoni and colleagues reported an increase of Kawasaki-like disease cases in March-April 2020 with two of ten children with positive NP swabs and eight with positive serology [257]. A case-report documented an episode of detectable SARS-CoV-2 RNA in blood and transient fever in an otherwise asymptomatic 6-month infant [51]. Prolonged GI SARS-CoV-2 RNA detection after respiratory samples clearance was observed in stool samples up to 35 dps for 20 children aged 0.15-10 years [49, 89, 91, 100, 121, 126, 237] and in rectal swabs ≥3 weeks in 15 cases aged 0.15-17 years [86, 91, 93, 168, 203]. In a cohort of 46 children and adolescents, four had positive rectal swabs within 2-12 days after recovery and discharge with negative OP swabs [229]. An asymptomatic girl had SARS-CoV-2 RNA detectable in anal swabs for 42 days, but not in NP swabs [203].
Higher SARS-CoV-2 VL in stool than OP swabs were reported for >20 days in an infant with mild COVID-19 [121]. Eight children had higher average VL in anal swabs compared to NP swabs [86]. Han and colleagues reported of a neonate with mild COVID-19 who had SARS-CoV-2 RNA detectable in respiratory swabs (NP swab VL 1.2x1010 copies/ml and OP swab VL 1.3x108 copies/ml to undetectable 4-17 dps), stool (VL 1.7x106 copies/ml to 4.1x107 copies/ml 6-18 dps), saliva (6-9 dps), plasma (5-10 dps) and urine (6-17 dps) [178].
Immunocompromised
individuals
We reviewed 32 studies including 317 immunocompromised individuals [16, 50, 80, 88, 95, 96, 103, 110, 136, 147, 151, 152, 154, 155, 165, 169, 188, 191, 193, 207, 218, 219, 221, 233, 245, 249, 251, 253, 257-261], however specifics for them were rarely outlined in cohorts. Data on SARS-CoV-2 shedding patterns in immunocompromised individuals are still limited and quantitative data are lacking. SARS-CoV-2 RNA was detected in NP swabs 57 and 63 dps in a kidney transplant recipient following clinical recovery and hospital discharge 35 dps [258]. In a report on two lung transplant recipients, an asymptomatic adolescent had a positive NP swab 26 days after diagnosis, while a mildly symptomatic adult’s NP swab was negative for SARS-CoV-2 RNA after 2 weeks [251]. In a study of 90 transplant recipients with COVID-19, seven (three mild, four severe) cases had an initial negative NP swab, but dps were not reported [207]. Evidence of endothelial cell infection and endothelitis was observed in three severe COVID-19 cases, incl. a renal transplant recipient [188].
A person living with HIV (PLHIV) and HCV had pneumonia resolution and negative NP swabs 10 dps but delayed antibody response 42 dps [152]. Further three PLHIV had a similar mild course with negative NP and OP swabs after 1 week [218] and another two PLHIV after 2 weeks from diagnosis [80, 165].