A person’s antibody level has long been treated as a personal shield: measure it, compare it against a threshold, and predict whether infection will be blocked. A new study from the Dominican Republic challenges that comfortable assumption, showing that the amount of antibody needed to protect against SARS-CoV-2 infection is not a fixed property of the immune system at all. Instead, it shifts with the intensity of transmission in the surrounding community, a finding with far-reaching consequences for how vaccine effectiveness is measured and how immune correlates of protection are interpreted.
The research, led by Eric J. Nilles of Harvard Medical School and Brigham and Women’s Hospital together with colleagues from the Dominican Ministry of Health and Social Assistance, Yale School of Medicine, Baylor College of Medicine, the University of Queensland, and the London School of Hygiene & Tropical Medicine, was published in Nature Communications. The team drew on prospective surveillance of febrile illness among 1,828 participants followed during 2021 and 2022, a period that encompassed both quiet stretches and waves of SARS-CoV-2 circulation in the country. Because participants were monitored as they went about their lives, the study could capture infections as they occurred rather than reconstructing them retrospectively.
The central observation is deceptively simple. During periods of high community transmission, participants who became infected with PCR-confirmed SARS-CoV-2 carried higher antibody titers than participants who became infected during periods of low transmission. In other words, when the virus was circulating widely, it took more pre-existing antibody to be counted among the protected, and the antibody gap between those who were infected and those who remained uninfected narrowed. The protection curve, plotted as infection risk against antibody level, shifted rightward as transmission increased.
To quantify this shift, the researchers employed scaled logistic models, a statistical framework that relates the probability of infection to an immune measurement while allowing the relationship itself to change with external conditions such as transmission intensity. The models confirmed that the antibody threshold associated with any given reduction in infection risk rose during high-transmission periods. This was not an artifact of waning immunity or variant changes alone; the pattern persisted when the analysis accounted for the immune profiles of the participants themselves.
The most compelling evidence came from simulation. The team built discrete-time models in which individuals faced repeated opportunities for exposure, with the number of infectious contacts rising as community transmission climbed. Crucially, the models held per-exposure susceptibility constant: antibody levels reduced the chance of infection on any single exposure exactly as before. Yet when the frequency of exposures increased, the simulated protection curves shifted rightward, reproducing the empirical pattern seen in the Dominican cohort. Repeated exposures alone were sufficient to make it appear as though immunity had weakened, even though nothing about the immune response had changed.
The mathematics behind this phenomenon is intuitive once spelled out. Protection at the level of a population is a compound event: a person avoids infection only if every exposure fails. If each exposure is blocked with probability p, then the chance of avoiding infection across n exposures is p raised to the nth power. As n grows, even a high per-exposure protection probability yields a lower overall probability of escaping infection. An individual with a given antibody titer therefore faces greater cumulative risk during a wave, not because their antibodies work less well per contact, but because they are drawing from a larger number of infectious contacts.
This exposure-dependence has immediate implications for vaccine effectiveness studies. Observational estimates of effectiveness compare infection rates among vaccinated and unvaccinated people, typically during a defined period. If that period coincides with a transmission surge, both groups experience more exposures, and the measured effectiveness can fall even if the vaccine’s per-exposure benefit is unchanged. Conversely, effectiveness estimated during quiet periods may look artificially robust. The Dominican data suggest that some of the temporal variation reported in vaccine effectiveness studies around the world may reflect exposure frequency rather than genuine changes in immune performance.
The findings also complicate immunobridging, the practice of inferring clinical protection from antibody measurements alone, which became central to accelerating vaccine authorization during the COVID-19 pandemic. If the antibody level corresponding to a given level of protection depends on transmission context, then a threshold calibrated in one setting or season may not transfer to another. Regulatory and scientific bodies that rely on correlates of protection to predict vaccine performance will need to consider the epidemiological backdrop against which those correlates were established, and against which they are applied.
There are broader conceptual lessons as well. Immune correlates of protection are usually conceived as properties of the host-pathogen interaction: a particular neutralization titer, a particular binding antibody concentration. This study reframes them as joint properties of the host, the pathogen, and the transmission environment. Two individuals with identical antibody levels may experience very different degrees of protection depending on how many infectious contacts they encounter, which in turn depends on community incidence, behavior, and contact networks. Protection, in this view, is not stored entirely within the body.
The work emerged from a collaboration supported by the US Centers for Disease Control and Prevention, with field operations conducted in partnership with the Dominican Republic Ministry of Health and Social Assistance and Pedro HenrÃquez Ureña National University. The authors note that the findings and conclusions are their own and do not necessarily represent the official position of the CDC. As immune monitoring becomes an ever-larger part of pandemic preparedness, the Dominican Republic cohort offers a caution and a tool: measure antibodies, yes, but always alongside the transmission context in which those antibodies must do their work. A threshold that protects in one season may quietly fail in the next, not because immunity faded, but because the virus came calling more often.
Subject of Research: Exposure-dependent immune correlates of protection against SARS-CoV-2 infection
Article Title: Immune correlates of protection are exposure-dependent: evidence from the Dominican Republic
Article References: Nilles, E. J., Then Paulino, C., Vazquez, M., Duke, W., Jarolim, P., Sanchez, I. M., Murray, K. O., Lau, C. L., Skewes Ramm, R., & Kucharski, A. (2026). Immune correlates of protection are exposure-dependent: evidence from the Dominican Republic. Nature Communications. https://doi.org/10.1038/s41467-026-78427-7
Image Credits: AI Generated
DOI: 10.1038/s41467-026-78427-7
Keywords: SARS-CoV-2, correlates of protection, antibodies, vaccine effectiveness, transmission, epidemiology, Dominican Republic, immunobridging, mathematical modeling, cohort study, Nature Communications, public health
News Source: Kristina Jarvis. (October 18, 2026). Antibody Protection Against COVID-19 Weakens as Community Transmission Rises, Study Finds. Scienmag.
