Results

After searching all of the prior mentioned databases, we came up with a total of 661 retrieved results. Of these, 282 were duplicates and were thus removed. After the initial screening of the remaining 379 articles, 67 articles were eligible to enter the full-text screening process. Finally, following the careful implementation of the inclusion/exclusion criteria mentioned above, we encountered 16 eligible articles to enter the study. The schematic view of the study selection process is provided in Figure 1 .
Cohorts accounted for the most observed type of articles in our study, with 14 included papers. one cross-sectional and one case-control study were also eligible to enter this systematic review. All the data reported in the articles are gathered from three distinct countries of China (14 articles), South Korea (1 article), and Brunei (1 article) (Figure 2 ). Two studies from South Korea provided the whole country’s data using official South Korean governmental statistics (South Korea Center for Disease Control and Prevention). Therefore, we excluded the study with the smaller population due to their high possibility of population overlap (Figure 1 ).
The details of the included studies are discussed in Table 1 . The included articles yield a total of 10,475 studied individuals. Of these participants, a sum of 489 (4.67%) patients were considered to be re-positive cases (Figure 3 ).
Considering the articles reporting gender prevalence, the overall population sums up to 1275 patients, including 695 (54.51%) females and 580 (45.49%) males. Among the studies reporting gender prevalence of both re-positive and non-re-positive subgroups, total females and total males are 405/723 (56.02%) and 318/723 (43.98%), respectively. In the re-positive subgroup, the female and male numbers are 50/88 (56.82%) and 38/88 (43.18%), sequentially. In the non-re-positive subgroup, females and males account for 355/635 (55.91%) and 280/635 (44.09%) of the population (Figure 4 ).
To understand the mean age of different groups, we combined the studies that published the age data for total, re-positive, and non-re-positive groups. The mean age for the total population, the re-positive subgroup, and the non-re-positive subgroup were 47.5 ± 16.9, 50.0 ± 18.0, and 46.6 ± 17.1, respectively.
Other than the details presented in Table 1 , PCR conversion time (days between symptom onset to first negative PCR test), hospital stays (days) during the first hospitalization, days between last negative test and first re-positive PCR, second-time clinically symptomatic patients among re-positive patients, second-time clinical symptoms, comorbidities, days to second negative PCR, second-time chest CT changes among re-positive patients, number of re-positives infecting close contacts, number of re-positive patients with positive IgM and IgG, number of non-re-positive patients with positive IgM and IgG, and mean/median IgM and IgG levels for the included articles are demonstrated in Supplementary Table 1 .
Combining all studies which reported symptom status among re-positive cases, the rate of symptom presentation upon second positive test turned out to be 20.0%. Among those, observed symptoms included cough (68.2%), fever (31.8%), throat pain (9.1%), fatigue (22.7%), chest discomfort (4.5%), chest expectoration (4.5%), itchy throat (4.5%) and constipation (4.5%).
The most common comorbidities in re-positive patients according to available data were hypertension (15.9%), diabetes (3.2%), cerebrovascular disease (3.2%), depression (3.2%), tuberculosis in mediastinal lymph nodes (1.6%), hepatopathy (1.6%), chronic lymphocytic leukemia (1.6%), and pregnancy (1.6%).
Two studies reported the average number of days from second positive to second negative, being 2.73 ± 2.03 (25) and 1.83 ± 0.22 (26) (combined mean = 2.32 ± 1.56). The only re-positive patient in Qiao et al.’s study (27) stayed in the hospital for approximately 17 days after her second admission.
When comparing CT scans obtained upon first discharge and second positive PCR of a total of 46 patients, thirteen (28.3%) showed no improvement of lesions since initial discharge, while in the other 33 patients (71.7%), CT lesions were resolved to some extent.
Out of 64 patients whose antibody data were available, 55 (85.9%) had positive IgG and 35 (54.7%) had positive IgM titers, based on each study’s definition of “positive”. Some of the studies had also gathered antibody information of their non-re-positive patients. When combining available data, 93.9% of non-re-positive patients had positive IgG profile, showing a significant difference with that of re-positive patients (85.9%, p = 0.03), while 53.5% were positive for IgM, which is not statistically different to that of re-positive patients (54.7%, p = 0.86).