Study Selection and Data Extraction
Independent screening of titles and abstracts was performed by independent reviewers (P.D., S.S., J.F.) to identify potential studies for inclusion. A second reviewer (O.A.) subsequently reviewed the full-text articles and identified articles for inclusion. Any disagreements about inclusion were resolved through discussion. We incorporated any studies (prospective or retrospective) that reported on patient characteristics and symptoms of interest. For studies published in Chinese, we used Google translate to assess for potential inclusion and for data extraction.
Due to concerns about inclusion of the same patients in different publications, we used a hierarchical model of data extraction to minimize double counting of patients across similar institutions with coinciding dates of study inclusion. We aimed to identify and include data from the largest possible cohort from each location or hospital.12 Data extraction was performed using a 2-step process. The initial data extraction focused on data elements for study and patient characteristics. Subsequently, we identified studies for full data extraction based on study location (unique hospitals) and total number of patients. Additionally, when a study from a specific hospital did not provide all of the necessary information for the diarrhea symptoms, the next largest study from the same hospital (when available) was selected for inclusion in our analysis.
Data extraction was performed using a standardized Microsoft Excel data extraction form. Data extraction was performed in pairs; one study author independently extracted data while a second reviewer checked for accuracy of the data extraction (S.S., O.A., S.M.S., P.D., J.D.F., J.K.L., Y.F.Y., H.B.E.). Because the reporting of the data in the primary studies was suboptimal, a third reviewer (O.A.) additionally verified the extracted data to confirm the numbers and to resolve any disagreements. Studies with discrepancies in the data were excluded.
The following data elements were extracted:1. Study: author, year, location (hospital name, city, province or state), dates of inclusion, and date of last follow-up
2. Patient characteristics: number of patients, age (mean, median, interquartile interval or range), number of females, severity of illness, inclusion criteria (hospitalized or outpatients), GI comorbidities (pre-existing conditions, such as chronic liver disease, hepatitis, and inflammatory bowel disease)
3. Outcomes: diarrhea, nausea, vomiting, abdominal pain, and liver function test (LFT) abnormalities
4. Additional information: severity, characteristics, duration, timing (before or concurrent with respiratory symptoms), relationship with clinical outcomes (need for ventilator, survival, discharge, and continued hospitalization), and viral stool shedding.