Recently emerging coronavirus disease-2019 (COVID-19) has created a major public health challenge globally. A highly contagious severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) strain is identified as the primary cause of COVID-19 symptoms. As of 24 May, 2022, the COVID-19 pandemic has been lingering in most countries, accounting for 161,523,786,368 confirmed cases and claiming 6,279,667 lives, according to the World Health Organization (WHO) data (1).

Currently, quantitative reverse transcriptase-PCR (qRT-PCR)-based screening and detection of SARS-CoV-2 infection is one of the standard laboratory diagnostic methods. Recurrent positive results in qRT-PCR tests have been commonly observed in COVID-19 patients. Not only that, patients could be also alternatively tested as negative and positive or even re-positive after confirmed double negative tests, as commonly noticed in many patients who got discharged from hospitals after recovery (2–5).

Recently, the Omicron variant has rapidly replaced the Delta as the most dominant one in many geographical regions owing to its highest infectivity, transmissibility, and immune evasion capabilities (6). Most COVID-19 patients infected with the Omicron were either asymptomatic or presented mild symptoms and were released from quarantine within a few days. It is still puzzling whether these patients are contagious when they get re-tested as positive for SARS-CoV-2. For emerging infectious diseases like COVID-19, it is critical to understand the way to gauge the viral shedding kinetics and identify different epidemiological parameters to predict the cross-infection capacity of COVID-19 patients who tested re-positive. Here, we aimed to retrospectively examine the clinical features of SARS-CoV-2 RNA re-emergence among non-critical cases and develop a risk tool to assess re-positivity at the early stage as well as to conduct tests for negative-positive patients.

Early prediction of positive re-testing in non-critical COVID-19 patients would be beneficial in making better management and screening policies. Here, we summarized the clinical features and viral shedding kinetics of patients with recurring SARS-CoV-2 positivity. Moreover, we identified that the milder clinical types, incomplete vaccination, myalgia, headache, reduction in admission time (might be associated with symptoms, Supplementary Table 1), and long duration of the first negative conversion were correlated with higher odds of positive re-test for COVID-19. We also established a clinical risk nomogram model as well as a web-based calculator with sufficient sensitivity and specificity to predict the probability of COVID-19 recurrence within 30 days in non-critical cases who had confirmed negative result(s). These values are generally available in any isolation sites. DCA and CICA results further supported the clinical benefit of these models.

According to the Shanghai Municipal Health Commission, as of 4th May, 2022, approximately 90% (547, 056) of 601, 942 infected cases were asymptomatic carriers, (7) which were consistent with our results (86.29% for asymptomatic). It is shown that the median duration of the virial shedding for symptomatic Omicron infected subjects is 6 (range 4–8) days (8). Notably, our cohort of mostly asymptomatic subjects showed similar virial shedding kinetics, suggesting that it might be independent of symptom presentation. Consistently, our results indicate that more severe COVID-19 patients might be at higher risk for re-positive tests (9). Interestingly, we also found that patients with a long duration of first negative conversion might be at higher risk of COVID-19 recurrence. Therefore, we hypothesized that the long duration of first negative conversion and less severe COVID-19 symptoms, especially asymptomatic ones, might elicit a weaker immune response and minimum viral clearance.

Our study presented a total of 4,649 (20.09%) individuals who had at least one negative test which turned into a positive at later stages. Initially, 889 (3.84%) patients were released from quarantine after having two consecutive negative tests (tested at 24 h intervals), but they turned positive later. Several possibilities for the positive re-tests for COVID-19 have been proposed (10–12). In the present scenario, it’s been challenging to differentiate between COVID-19 recurrence due to the delayed shedding of non-viable viral remnants and truly contagious viral particles. The patients who meet the criteria for release from quarantine still need to be monitored for symptomatic manifestations via RT-PCR test.

The Ct value for viral load is applied to evaluate the course of infection and virulence, with every ∼3.3 cycle increase in the Ct value reflecting a 10-fold reduction in starting material (13). Recently, any qRT-PCR values less than a cutoff Ct < 35 indicate infectious status. However, the range of Ct values (Ct: 24–35) in predicting viral infectivity varies greatly across studies (14, 15). It is found that the mean Ct for the contagious patient is very close to that of a non-infectious individual (23.99 and 24.02, respectively) (16). It has also been revealed that the average Ct for Omicron infection is higher than that of the Delta variant (17). In our research, 80.20% of patients featured a higher Ct value in the current Omicron pandemic. It might need more observations on the Ct value to confirm if a cutoff value of 35 could be universally utilized to identify individuals who are positive and/or at a higher risk of transmission.

The guidelines for isolation, admission, and discharge for SARS-CoV-2 infected patients are not universal. As per the WHO clinical management guidelines, a COVID-19 patient should be declared clinically recovered after having at least two sequential negative PCR tests (at intervals of > 24 h) at the time of discharge. The proportion of recurring COVID-19 patients varies between 2.4 and 69.2% and could occur any time between 1 and 38 days of hospital discharge (2–5). The CDC considers there should be no infectious viral particle/remnant produced in 88 and 95% of patient samples, respectively, collected at 10 and 15 days after symptom onset. Therefore, adults with asymptomatic, mildly symptomatic, moderate, and severe COVID-19 patients could be released from isolation in 5–10 days post-infection (18).

However, these observations are mainly based on the initial epidemiology of COVID-19. Omicron cases demonstrated a lot of variabilities compared with other strains, such as release from quarantine as early as < 5–10 days. Notably, the detection of viral RNA in patients does not confirm the infectious status of the subject. Hence, challenges and uncertainties still exist in the public health policies for global epidemic preparedness based on existing evidence. Although a single negative test would not be a confirmatory result for the actual viral shedding, it may suggest a connection between individualized immunity and recurrent infection. It is crucial to conduct epidemiological surveys on individuals, especially those with recurrent COVID-19 diagnosis and high viral loads, to evaluate their contagiousness and monitor health conditions.

At least seven major limitations for the findings are as follows. First, the detection frequency was varying and inconsistent after discharging from the hospital, which might underestimate the proportion of re-testing positive after discharge. Second, the data for the nomogram model were entirely obtained from the Fangcang Shelter Hospital in Shanghai, China. These patients might have had a milder illness or presented asymptomatic nature with fewer comorbid and immunodeficient conditions, which could introduce potential bias and limit the universal application of the risk model in other geographical regions. Third, other potential values (such as treatment) might affect viral shedding and were not available due to incomplete clinical data. Fourth, given that false negative results of COVID-19 PCR were an objective limitation, and PCR tests could not reflect the true viral load, virus culturing is required to determine the infectivity. Fifth, there could be bias in self-reported symptoms and comorbidities. Sixth, the prediction tool incorporates the clinical features and Immunization, despite more accessible and suitable during Omicron predominance, lack of the weight of serology, imaging, or other potential factors. Finally, although the validation cohort was conducted by external validation, the strength of the verification with retrospective cohort may not be as good as the prospective cohort.

We established a risk model based on the clinical features of COVID-19 to predict the probability of re-testing positive within 30 days of negative test in less severely affected patients who were confirmed negative for SARS-CoV-2 infection in the current outbreak in Shanghai. Our model demonstrated satisfactory performance and might have clinical utilization in the early identification and management of concerned individuals.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Ethics Committee of the Huashan Hospital Affiliated with Fudan University. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

MC and AL designed the study. HH had full access to all of the data and took responsibility for the accuracy of data in the study. AL, LZ, and AC took responsibility for the accuracy of the data analysis and interpretation. AL drafted of the manuscript. CW, WH, SM, and DZ revised the manuscript for important intellectual content. All authors contributed to the article and approved the submitted version.