What are the Gastrointestinal Manifestations of COVID-19?

Diarrhea
A total of 43 studies including 10,676 COVID-19 patients (confirmed by laboratory real-time reverse transcription polymerase chain reaction [RT-PCR] testing) were included in the overall analysis.20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35 , 37 , 38 , 41 , 42 , 45 , 47 , 49, 50, 51, 52, 53, 54, 55 , 58 , 59 , 62, 63, 64, 65, 66, 67, 68, 69, 70, 71 , 73 , 76 The pooled prevalence of diarrhea symptoms across these studies was 7.7% (95% CI, 7.2%–8.2%). When analyzing by country (studies from China vs studies from other countries), the pooled prevalence of diarrhea in studies from countries other than China was much higher at 18.3% (95% CI, 16.6%–20.1%). This is in comparison to studies from China, where the prevalence was much lower at 5.8% (95% CI, 5.3%–6.4%) (Figure 1 ).
Figure 1 Forest plot of the prevalence of diarrhea in all patients.
In hospitalized patients, across 39 studies including 8,521 patients, the pooled prevalence was slightly higher at 10.4% (95% CI, 9.4%–10.7%) compared with outpatients.20, 21, 22, 23 , 25, 26, 27, 28 , 30 , 33, 34, 35 , 38 , 41 , 42 , 44, 45, 46 , 49, 50, 51, 52, 53, 54, 55 , 60 , 62, 63, 64, 65, 66, 67, 68 , 70, 71, 72, 73, 74 , 76 In 3 studies including 1701 outpatients, the pooled prevalence was 4.0% (95% CI, 3.1%–5.1%).31 , 59 , 63 As part of the sensitivity analysis, we identified 35 studies including 9717 patients that described diarrhea, and explicitly reported that it was one of the initial presenting symptoms.20, 21, 22, 23 , 26, 27, 28 , 31 , 33, 34, 35 , 38 , 41 , 42 , 44 , 45 , 47 , 48 , 50, 51, 52, 53, 54, 55 , 60 , 63, 64, 65, 66 , 68, 69, 70, 71 , 74 , 76 The pooled prevalence in these studies was 7.9% (95% CI, 7.4%–8.6%). A total of 33 studies including 8070 patients reported on hospitalized COVID-19 patients presenting with diarrhea as one of the initial symptoms of COVID-19.20, 21, 22, 23 , 26, 27, 28 , 33, 34, 35 , 38 , 41 , 42 , 44 , 45 , 48 , 50, 51, 52, 53, 54, 55 , 60 , 63, 64, 65, 66 , 68 , 70, 71, 72, 73 , 76 The pooled prevalence was 9.3% (95% CI, 8.6%–9.9%) (Supplementary Figure 3, Supplementary Figure 4, Supplementary Figure 5, Supplementary Figure 6).

Description of diarrhea
Only a handful of studies provided any details on the type and severity of diarrhea symptoms.55 , 60 , 74 In the study by Lin et al,55 23 of 95 patients (24%) reported having diarrhea (described as loose or watery stools, ranging from 2–10 bowel movements per day); however, only a small number of patients actually had diarrhea on admission (5.2%). Most patients developed diarrhea during the hospitalization, which may have been attributable to other treatments or medications. In the study by Jin et al60 of 651 hospitalized patients, 8.6% of patients had diarrhea on admission before receiving any treatments. The diarrhea symptoms were described as more than 3 loose stools per day. Stool cultures were negative (including Clostridium difficile) in all patients. There was no mention of fecal leukocytes. Median duration of symptoms was 4 days (range, 1–9 days) and most patients had self-limited diarrheal symptoms.60 One additional study on 175 hospitalized patients reported that 19.4% of patients had diarrhea, with an average of 6 episodes per day, with symptom duration ranging from 1 to 4 days.74

Diarrhea as the only presenting symptom in the absence of upper respiratory symptoms
In the 43 studies that informed our analysis on the prevalence of diarrhea, we extracted information on whether diarrhea was reported as the only presenting symptom.50 , 60 In only 2 studies, there was explicit reporting of diarrhea in the absence of upper respiratory infection symptoms. In a study by Luo et al50 of 1141 patients, 183 patients (16%) presented with GI symptoms only in the absence of respiratory symptoms. Of 1141 patients, loss of appetite (15.8%) and nausea or vomiting (11.7%) were the most common symptoms, but diarrhea was reported in 6.0% and abdominal pain in 3.9% of patients. Notably, the majority of patients (96%) had lung infiltrates on chest computed tomography. In the study by Jin et al60 of 651 hospitalized patients, 21 patients (3.2%) presented with GI symptoms only (and no respiratory symptoms of coughing or sputum production).60 GI symptoms were defined as at least 1 of the following symptoms: nausea, vomiting, and diarrhea. Conversely, in the US study of 116 patients with COVID-19, Cholankeril et al70 reported that 31.9% of patients had GI symptoms on admission (median duration, 1 day); diarrhea was reported in 10.3% (12 of 116), nausea and/or vomiting in 10.3% (12 of 116), and abdominal pain in 8.8% (10 of 116). The authors explicitly reported that none of the 116 patients had isolated GI symptoms as the only manifestation of COVID-19.

Diarrhea as the initial presenting symptom preceding other COVID-19 symptoms
Of the studies included in our review, based on our selection framework, we identified only 1 study that reported on timing of diarrhea in relation to other COVID-19–related symptoms. In a study by Ai et al76 of 102 hospitalized patients, 15 patients reported diarrhea symptoms on hospital admission, and diarrhea was the first symptom in 2 patients. In a study by Wang et al77 of 138 consecutive hospitalized patients, not included in our pooled analysis, a total of 14 patients presented with diarrhea and nausea 1–2 days before the development of fever and dyspnea.

Nausea/vomiting
A total of 26 studies including 5955 patients with COVID-19 (confirmed by laboratory RT-PCR testing), were included in the overall analysis for nausea and/or vomiting.20 , 22 , 23 , 25 , 27 , 29 , 34 , 37 , 41 , 45, 46, 47 , 50 , 51 , 54 , 55 , 59 , 63 , 65 , 67 , 68 , 70, 71, 72, 73 , 76 The pooled prevalence of nausea/vomiting was 7.8% (95% CI, 7.1%–8.5%). A subgroup analysis of 1901 patients from 7 studies (including patients from Germany, Singapore, United States, Australia, and The Netherlands) demonstrated a higher pooled prevalence of 14.9% (95% CI, 13.3%–16.6%).37 , 46 , 47 , 59 , 63 , 68 , 70 This is in comparison to the prevalence of symptoms in studies from China, which was 5.2% (95% CI, 4.4%–5.9%) (Figure 2 and Supplementary Figure 7).
Figure 2 Forest plot of the prevalence of nausea/vomiting in all patients.

Abdominal pain
A total of 15 studies including 4031 COVID-19 patients (confirmed by laboratory RT-PCR testing) were included in the overall analysis for abdominal pain.21 , 23 , 27 , 37 , 50 , 54 , 55 , 59 , 63 , 69, 70, 71, 72, 73 , 76 The pooled prevalence of abdominal pain was 3.6% (95% CI, 3.0%–4.3%). A subgroup analysis of 1584 patients from the United States, Australia, South Korea, and The Netherlands, demonstrated a slightly higher pooled prevalence of 5.3% (95% CI, 4.2%–6.6)% compared with studies from China 2.7% (95% CI, 2.0%–3.4%), which included 10 studies of 2447 patients.37 , 59 , 63 , 69 , 70 The symptoms were variably described as stomachache, epigastric pain, and abdominal discomfort, without further details regarding the quality or nature of pain (Figure 3 and Supplementary Figure 8).
Figure 3 Forest plot of the prevalence of abdominal pain in all patients.

Stool shedding
Our study selection criteria prioritized including studies with diarrhea as a GI manifestation and avoiding overlap in populations and, therefore, did not include a comprehensive set of studies reporting on stool shedding. A recently published systematic review by Cheung et al10 found a 48.1% (95% CI, 38.3%–59.7%) pooled prevalence of stool samples positive for virus RNA in 12 studies. Stool RNA was positive in 70.3% of samples taken from patients after respiratory specimens were no longer positive for the virus.
From the 57 studies included in our analysis, 4 studies reported on presence of viral RNA in stool.24 , 32 , 57 , 68 Of these, 3 studies were published after the systematic review by Cheung et al.10 First, Dreher et al68 conducted a retrospective cohort study in Germany, stratifying patients by presence of acute respiratory distress syndrome. In this study, 8 of 50 patients had diarrhea, and stool PCR was positive in 15 of 50 patients. In a US study by Kujawski et al,57 stool PCR was positive in 7 of 10 patients. Finally, in a case series from Germany by Wolfel et al,32 the authors not only examined stool RNA but also tried to isolate virus from laboratory specimens. In this study, 2 of 9 patients had diarrhea as an initial symptom and stool PCR remained positive for up to 11 days, but notably, the authors were unable to isolate infectious virus, despite a high stool viral RNA load, even though the virus was successfully isolated from respiratory specimens. The authors concluded that stool is not a primary source of spread of infection.32 Conversely, in a letter published by Wang et al,78 the authors collected 1070 specimens from 205 hospitalized patients with COVID-19 and 44 of 153 stool specimens (29%) were positive for viral RNA. Four specimens with high copy numbers were cultured and electron microscopy was performed to detect live virus, which was observed in the stool from 2 patients who did not have diarrhea. The authors concluded that although this does not confer infectivity, it raised the possibility of fecal–oral transmission.78 The small sample size of the reports that assessed the presence of live virus in stool combined with the conflicting findings limit our certainty in the evidence and thus the question of fecal–oral transmission remains unsettled.