**4. Discussion**

We reported a case series of five clinically and laboratory confirmed COVID-19 patients suffering a second episode of COVID-19 after a gap of over 70 days free of symptoms. The recurrent cases had largely mild symptoms in both the first and second episodes of illness (only one case had moderate symptoms during their second episode). Most of the samples had high CT values (>30) indicating a low viral load. Whole genome sequencing is essential for determining true reinfections; however, our attempt failed due to low viral loads in collected NP swabs. Therefore, it is still unclear whether these were due to true reinfections or possible persistent low-level of infection sustained by viable viral particles or extra viral RNA. Nevertheless, our small case series indicates that symptomatic SARS-CoV-2 recurrence can possibly occur among individuals who recovered from their initial illness.

Our institute data showed that approximately 1% of all staff to date had the second episode of COVID-19. The proportion of recurrent RNA positivity after recovering from COVID-19 in our study was lower than that reported by Camilla et al. (2020) in a metaanalysis, where the proportion was reported between 7 and 23% [17]. Several factors might explain the low percentage. Firstly, we followed a strict case definition for a patient to be included in our investigation with a combination of clinical symptoms, PCR positivity, at least a one-month gap between a negative test report and the reappearance of COVID-19 symptoms followed by the presence of viral RNA in a second PCR test. Secondly, we could not actively monitor and follow-up all first-time positive cases in person because of resource constraints, and the follow-up system depended on the passive reporting of symptoms by the staff to the clinic. As most employees worked from home, they were probably reluctant to report symptoms to the staff clinic in order to avoid social stigma [27]. Thus, we might have missed many mild to moderate symptomatic recurrences of COVID-19 among our study population. Thirdly, there is a high probability of asymptomatic reinfection [28] which was not captured in our surveillance system. Hence, the actual number of recurrences of SARS-CoV-2 RNA positivity in our population may have been underestimated. Consensus criteria for deciding true SARS-CoV-2 reinfection should be set as new findings accumulate. Finally, our cohort did not include a collection of blood samples to describe the immunological profile that could allow us to evaluate the immune status of the cases.

The protective role and longevity of antibody responses to the virus remain unanswered. Some recent reports sugges<sup>t</sup> that neutralizing antibodies against the virus remain relatively stable for several months after infection. In line with another case series [29], it is not clear whether the virus can persist in some COVID-19 patients for a longer period and transmit to close contacts and later reappear as an apparent second illness with COVID-like symptoms. Therefore, we should still advise our staff to maintain social distance, wear masks and self-isolate as much as possible to prevent reinfection. Bangladesh has received 9 million doses of the COVID-19 vaccine, Covishield (AstraZeneca) from the Serum Institute of India and vaccinated a total of 3,682,152 individuals as of 7 March 2021 (Health Bulletin of DGHS, 7 March 2021). Given the global spread of new variants, it is ye<sup>t</sup> to be explored whether mass vaccination against existing strains can reduce the incidence rate to sufficient levels required to halt the pandemic.

Two of our patients showed concomitant influenza virus infection during their second COVID-19 episode. Both SARS-CoV-2 and influenza virus have common signs and symptoms, so it is likely that many patients with respiratory tract symptoms can be co-infected with influenza (H3). The influenza positive cases were clinically indistinguishable from the other three cases. The role of influenza in altering the clinical course of COVID-19 is unclear, thus further studies will be required to explore the pros/cons of co-infection.

Most of our cases reported mild symptoms. The immunological memory against SARS-CoV-2 is not well established and it is uncertain whether the first infection can prevent re-infection [11]. Limited data sugges<sup>t</sup> that SARS-CoV-2-specific CD4+/CD8+ T cells can persist for up to 3–5 months and IgG to the spike protein up to eight months in COVID-19 patients [13,14]. If true, all of our initial cases may have developed immunological memory which may have partially protected them during their second illness.

Though our study was based on systematically collected data and prospectively investigated by an experienced group of clinicians, epidemiologists, and virologists, it had several limitations. Firstly, we tried to compare whole genome sequence data obtained from the first and second episodes from individual cases. However, only two complete sequences (with 99% coverage) were recovered from the 10 isolates (Supplementary Figure S1). Therefore, we could not determine whether the virus detected was identical or a different variant. Secondly, we could not conclude whether these cases were true reinfections or just relapsed cases. It is possible that the virus in the first episode diminished in the upper respiratory tract, became PCR negative, but was sustained in other parts of the body such as the pharynx, trachea, lungs, heart, kidneys, intestines, brain, genitals, or in body fluids such as the cerebrospinal fluid, semen and breast milk [30]. A systematic review showed that the virus could be sustained in other samples, such as the stool, for a longer period of time after the nasopharyngeal swab was negative [31]. Natural infection should, in theory, offer a degree of protection against future reinfection. However, in reality, this is not always achieved, given the differences in innate immunity and host genetics. Influenza vaccines that include previous variants may not protect against a new variant, so reinfection is likely. A lack of acquired immunity may partially explain the observed COVID-19 recurrence, but further immunogenetic studies examining innate, acquired and concomitant immunity are required.

Active case finding and isolation, quarantine, regular hand washing, wearing face masks, cough etiquette, avoiding public meetings, and maintaining physical distancing are the steps of choice in the absence of necessary vaccination coverage. The efficacy of such initiatives, however, is largely dependent on public willingness to cooperate and their perceptions of risk towards COVID-19 [32]. All our COVID-19 positive cases were well aware of the coronavirus disease: how it spreads and what preventive measures should ideally be maintained to minimize the risk. Our findings support a previous study conducted in ten countries, including Asia, which found that people who had personal and direct experience with COVID-19 had significantly higher risk perceptions [33]. Although our whole genome sequencing was inconclusive, all five cases believed their second episode of the illness was reinfection because their sufferings were identical to (or even worse than) the first. Epidemiologically, they had the potential exposures before the second episode, and clinically their signs and symptoms were also compatible with COVID-19 infection.
