This retrospective cohort study was conducted from 18 January to 24 August 2020 at the Fifth Affiliated Hospital of Sun Yat-Sen University (SYSU5) in Zhuhai, China. We recruited 105 patients who were hospitalized in the COVID-19 medical center of SYSU5. The diagnosis of COVID-19 was based on epidemiological history, symptoms, lung imaging manifestations and the presence of SARS-CoV-2 RNA in respiratory tract samples using the China CDC-standardized real-time reverse transcriptase polymerase chain reaction (rRT-PCR). The copy numbers of SARS-CoV-2 RNA were indicated by rRT-PCR cycle threshold (CT) values, with lower CT values corresponding to higher viral copy numbers. A CT value <40 was defined as positive, as reported (12). GI infection of SARS-CoV-2 was defined as simultaneously positivity for SARS-CoV-2 RNA in a fecal sample. After excluding one patient for whom the fecal SARS-CoV-2 RNA test was not completed, 104 patients were included and divided into two groups according to whether SARS-CoV-2 GI infection was present (infection group; n = 54) or absent (non-infection group; n = 50).

This study was approved by the Ethics Committee of SYSU5 [No. ZDWY [2020] Lunzi No. (K22-1)]. All experiments in this study were performed according to the principles of the Declaration of Helsinki and Good Clinical Practice. All patients or their legal guardians provided written informed consent.

Information about sex, age, ethnicity, smoking, alcohol consumption, epidemiological history, medical history, clinical classification, and antiviral treatment was obtained from the hospital information system of SYSU5. Epidemiological history was recorded as local or overseas cases according to the possible location of infection with SARS-CoV-2. In the present study, local cases had travel history to Wuhan or contact with patients with COVID-19 or other person with fever or respiratory symptoms from Wuhan, and no overseas travel history within 14 days before illness onset. Overseas cases had overseas travel or residence history and tested positive for SARS-CoV-2 RNA in respiratory tract samples when they entered China. The identification of hypertension and diabetes was based on previous relevant medical history or diagnoses during hospitalization. Clinical classification was classified as mild (mild clinical symptoms and absent of pneumonia manifestations on imaging), moderate (obvious symptoms and present of pneumonia manifestations on imaging), severe (respiratory rate ≥30 breaths/min; oxygen saturation ≤ 93% at a resting state; arterial partial pressure of oxygen/oxygen concentration ≤ 300 mmHg) and critical (occurrence of respiratory failure requiring mechanical ventilation; presence of chock; other organ failure that requires monitoring and treatment in the ICU) according to the severity of clinical symptoms, imaging manifestations and hypoxia.

Venous blood sampling was performed immediately upon hospitalization, and the white blood cell count (WBC), neutrophil count (NEU), lymphocyte count (LYM), and procalcitonin (PCT) and C-reactive protein (CRP) levels were determined using standard laboratory methods. Consistent laboratory methods were adopted.

The primary endpoint was the conversion time from the first positive of SARS-CoV-2 RNA test to negative in respiratory tract samples. Negative conversion of SARS-CoV-2 RNA was defined as undetectable viral RNA for two consecutive respiratory tract or fecal samples (sampling interval more than 24 h). The secondary outcome was the time of hospitalization for COVID-19. Discharge standards were defined as follows: (1) body temperature remaining normal (axilla temperature ≤ 36.6°C or oral temperature ≤ 37.2°C or rectal temperature ≤ 37.8°C) for at least 3 days; (2) respiratory symptoms improved, SpO₂ >93% without assisted oxygen inhalation; (3) lung imaging showing obvious improvement in lesions, CT improvement defined as that exudation or consolidation of the lesion are absorbed, the lesion area was gradually narrowed, and there may be residual linear fibrosis, which are independently judged by two radiologist until they reached the same conclusion; (4) negative conversion of SARS-CoV-2 RNA in respiratory tract and fecal samples; (5) disease course at least 14 days from onset; and (6) discharge approved by multi-disciplinary medical team.

Statistical analyses were performed using SPSS Statistics version 25.0. Basic characteristics of the study subjects are presented as the median (range) for continuous variables and as numbers with percentages for categorical variables. Independent t-tests or the Mann–Whitney U tests were performed for comparison of means of continuous variables depending on the normality of the distribution of groups. Paired-samples t-test was adopted to compare means of paired samples. For venous blood results, values beyond the defined lower limit of detection were set to the lower limit. The Pearson χ2 test was used to assess differences in categorical variables between groups. The log-rank test and Kaplan-Meier curve analyses were performed to compare the duration of SARS-CoV-2 shedding and hospitalization between groups. A Cox proportional hazard model was applied to estimate the HRs and corresponding 95% CIs. Subgroup analysis including sex, age, epidemiological history, clinical classification and antiviral treatment was carried out to adjust for potential imbalance in baseline characteristics. For all comparative analyses, P < 0.05 was considered to indicate statistical significance.

This study included a total of 104 patients with COVID-19, with 54 (51.9%) in the infection group and 50 (48.1%) in the non-infection group. The characteristics of the study population are listed in Table 1. A significant difference in the sex was found between the groups (P = 0.002). There were 34 male patients (63.0%) in the infection group and 16 male patients (32%) in the non-infection group. Compared with patients without SARS-CoV-2 GI infection, a significantly higher likelihood of GI infection was found for overseas cases (8 (14.8%) vs. 1 (2.0%); P = 0.020). Among these overseas cases with GI infection, 5 patients were from the UK, 2 patients from Hong Kong, and 1 patient from Macau. In contrast, age, ethnicity, smoking, drinking, medical history, and clinical classification were not significantly different between the groups. We also observed no differences in the WBC, NEU, LYM, CRP and PCT levels between the two groups (Table 1).

In analyses of the full study population, patients with SARS-CoV-2 GI infection exhibited a longer duration of SARS-CoV-2 RNA positivity in respiratory tract samples than patients without SARS-CoV-2 GI infection (median 12.0 days [95% CI: 10.0–13.2] vs. 9.0 days [95% CI: 7.5–10.5]; HR 0.575 [95% CI: 0.386–0.857]; P = 0.003) (Figure 1A). We also found that patients with SARS-CoV-2 GI infection had a longer time of hospitalization than the patients without SARS-CoV-2 GI infection (median 28.0 days [95% CI: 23.2–32.8] vs. 15.0 days [95% CI: 13.6–16.4]; HR 0.149 [95% CI: 0.087–0.252]; P < 0.001) (Figure 2A). To adjust for potential imbalance in the baseline characteristics of the two groups, which is inevitable in a retrospective study, subgroup analyses of sex, age, epidemiological history, clinical classification and antiviral treatment were performed. Combined SARS-CoV-2 GI and respiratory infection had an adverse effect on the time to negative conversion of viral RNA in respiratory tract samples as well as hospitalization time in all strata (Figures 1B, 2B). Nonetheless, the effect was not statistically significant in some subgroups because of the small sample size.

Among the GI infection group, we specially evaluated the viral load of 16 patients who remained persistently positive fecal SARS-CoV-2 RNA over nearly 2 weeks or more. Among the 16 patients, SARS-CoV-2 RNA in feces was detected 1–21 days (median: 6 days; mean: 8.44 days) after detection of viral RNA in nasopharyngeal swabs. The mean duration of positive fecal SARS-CoV-2 RNA was significantly longer than that of positive nasopharyngeal SARS-CoV-2 RNA (25.81 days vs. 9.88 days; difference in mean 15.94 [95% CI: 11.93–19.95]; P < 0.001) (Table 2).

We monitored CT value of fecal sample from 16 patients during their hospitalization. As shown in Figure 3, CT values of the fecal sample ranged from 11 to 40. The mean of the initial and the final CT values is 23.59 and 36.22, respectively (difference in mean −12.62 [95% CI: −18.67 to −6.579]; P < 0.001). Except for Patient 11, 13, and 16, the CT values of the other 13 patients increased slowly over time during their hospitalization. The CT values of Patient 1 and 6 increased <1 over 25 and 29 days.

Among 3 patients (Patient 5, Patient 11, and Patient 15) undergoing upper GI endoscopy, SARS-CoV-2 nucleocapsid was detected in the cytoplasm of gastric glandular epithelial cells in 2 patients (Patient 11 and 15), who remained positive for fecal SARS-CoV-2 RNA after negative conversion of viral RNA from pharyngeal swabs (Table 2).