Browse the corpus

Since the start of the COVID-19 pandemic, the question of potential reinfection has been ever present. Although there has been much debate about potential reliance on herd immunity through natural infection, human coronaviruses are well adapted to subvert immunity1 and reinfection occurs for seasonal coronaviruses (229E, OC43, NL63, and HKU1) that cause the common cold due to ephemeral immunity that is poorly protective between infections.2 Furthermore, detailed mapping of immune parameters in cohorts such as health-care workers emphasises the heterogeneity of immune responsiveness to SARS-CoV-2, from those with high neutralising antibody titres and broad T-cell repertoires, to the minority with barely detectable immunity.3 These very low levels of immunity after infection would be hard to equate with protection from reinfection. Furthermore, among the longitudinal studies that have investigated waning antibody levels against SARS-CoV-2, responses have been found to last for 6 months or longer; although, depending on which components of the antibody repertoire are assayed, a substantial minority serorevert to negativity.4, 5

n from reinfection. Furthermore, among the longitudinal studies that have investigated waning antibody levels against SARS-CoV-2, responses have been found to last for 6 months or longer; although, depending on which components of the antibody repertoire are assayed, a substantial minority serorevert to negativity.4, 5 Despite the substantial advances in all aspects of COVID-19 analysis and data collection over the past year, calculation of the risk of reinfection has been difficult and there are two key reasons for this. The most obvious reason for difficulty is that most individuals around the world who became infected during the first wave of the pandemic did not access a PCR or antibody test and were not admitted to or treated in hospital, and so are not included in many COVID-19 datasets. The second reason is that scientific journals require specific evidence for formal reporting of reinfection, leading to probable under-reporting. For instance, peer reviewers and editors have required evidence from individuals who tested positive by PCR, then recovered and became negative by PCR, and then subsequently tested positive by a second PCR test, with distinct sequenced viral isolates on each occasion.6 Outside of a research cohort setting, such evidence gathering is rarely achievable and potential confounders exist to reinfection analysis. For example, a minority of individuals can harbour a reservoir of persistent SARS-CoV-2 in the gut,7 such that distinguishing between true reinfection as opposed to recurrence of the original infection is challenging. A study of health-care workers in Sergipe, Brazil, indicated a relatively high rate of reinfections correlated with the lowest antibody responses,8 but in most cases the researchers could not confirm de-novo reinfection. From that study, the investigators estimated risk of reinfection to be approximately 7%.8

llenging. A study of health-care workers in Sergipe, Brazil, indicated a relatively high rate of reinfections correlated with the lowest antibody responses,8 but in most cases the researchers could not confirm de-novo reinfection. From that study, the investigators estimated risk of reinfection to be approximately 7%.8 In The Lancet, Christian Hansen and colleagues report their population study of a Danish cohort investigating the risk of becoming positive for SARS-CoV-2 by PCR for the second time, presumed to indicate reinfection.9 The study makes use of data from Denmark's national PCR-testing strategy whereby approximately 4 million people took 10·6 million PCR tests. Because the data in the system were person-identifiable, the authors were able to determine that 3·27% of those who were uninfected during the first surge had a positive test during the second surge, compared with 0·65% among those who had previously recorded a positive test. Thus, they determined from that, in general, past infection confers 80·5% protection against reinfection, which decreases to 47·1% in those aged 65 years and older. Hansen and colleagues acknowledge the many limitations of their analysis being restricted to only PCR data, including the possibility that people might change their behaviour after a positive PCR test. This confounder is addressed by noting that the findings are similar in a sensitivity analysis of nurses, doctors, social workers, and health-care assistants who were tested regularly due to their profession.

to only PCR data, including the possibility that people might change their behaviour after a positive PCR test. This confounder is addressed by noting that the findings are similar in a sensitivity analysis of nurses, doctors, social workers, and health-care assistants who were tested regularly due to their profession. Set against the more formal reinfection case reports that are based on differential virus sequence data and make reinfection appear an extremely rare event, many will find the data reported by Hansen and colleagues about protection through natural infection relatively alarming. Only 80·5% protection from reinfection in general, decreasing to 47·1% in people aged 65 years and older are more concerning figures than offered by previous studies. Until now, one of the largest datasets has come from Qatar during a period of high disease burden and reported an estimated reinfection risk of 0·2%.10 However, a key difference between the studies is that the Danish study is based on a universally accessible national testing programme for both symptomatic and non-symptomatic individuals, whereas the Qatar data are derived from a programme of PCR testing in the context of symptomatic disease. PCR-positive cases within the Danish dataset are thus likely to encompass a far higher proportion of asymptomatic cases presumed to elicit more marginal levels of protective immunity.

d non-symptomatic individuals, whereas the Qatar data are derived from a programme of PCR testing in the context of symptomatic disease. PCR-positive cases within the Danish dataset are thus likely to encompass a far higher proportion of asymptomatic cases presumed to elicit more marginal levels of protective immunity. The quality, quantity, and durability of protective immunity elicited by natural infection with SARS-CoV-2 are poor relative to the much higher levels of virus-neutralising antibodies and T cells induced by the vaccines currently being administered globally.11, 12 Emergence of variants of SARS-CoV-2 with variable escape from natural and vaccine-induced immunity complicates matters further. Precise correlates of protection against SARS-CoV-2 are not known, but emerging variants of concern might shift immunity below a protective margin, prompting the need for updated vaccines.13 Interestingly, vaccine responses even after single dose are substantially enhanced in individuals with a history of infection with SARS-CoV-2.14 These data are all confirmation, if it were needed, that for SARS-CoV-2 the hope of protective immunity through natural infections might not be within our reach, and a global vaccination programme with high efficacy vaccines is the enduring solution. © 2021 Henning Bagger/Getty Images2021