The study, conducted between March and September 2023, recruited convalescent individuals across Fuzhou and assessed them at baseline, Month 3, and Month 6. Serum IgG concentrations targeting SARS-CoV-2 were quantified using a chemiluminescence immunoassay, a platform widely used in clinical serology because of its high sensitivity and throughput. Rather than treating each measurement as an isolated snapshot, the investigators applied linear mixed-effects models to the longitudinal data, a statistical framework that allows each participant to serve, in effect, as their own control while accounting for repeated measurements nested within individuals. Because antibody titers are typically right-skewed, the team used a Yeo–Johnson transformation, a flexible power transformation that can accommodate both positive and negative values and stabilizes variance across a wide dynamic range, improving the validity of the mixed-model estimates.

One of the methodological challenges inherent to any longitudinal cohort is attrition: not every participant returns for every scheduled blood draw, and simply discarding incomplete records can bias results if dropouts differ systematically from those who complete follow-up. The Fuzhou team addressed this with multiple imputation, generating plausible values for missing follow-up IgG measurements based on the observed data structure and incorporating the uncertainty of those imputations into the final estimates. This approach preserves statistical power and reduces the risk that the apparent shape of antibody waning is an artifact of who happened to show up for their third visit. It is the kind of unglamorous but essential statistical hygiene that determines whether a cohort study’s conclusions can be trusted.

The central finding is that IgG trajectories diverged sharply by age group. Participants aged 0 to 18 years entered the study with higher baseline IgG levels than older participants, yet they experienced a more pronounced overall decline across the follow-up period. In other words, children and adolescents mounted robust antibody responses after infection but shed them relatively quickly. At the opposite end of the age spectrum, adults aged 65 and older showed no significant change in IgG from baseline to Month 3, followed by a subsequent decrease. This pattern suggests a relatively stable early phase of antibody persistence in older adults, with waning emerging later in the observation window, a tempo that is nearly the mirror image of what the youngest participants displayed.

The immunological logic behind such age-dependent divergence is a matter of active investigation. Younger immune systems generally generate vigorous germinal center reactions and produce high titers of neutralizing antibody after infection or vaccination, but those titers contract rapidly once the acute response subsides, leaving behind a smaller pool of long-lived plasma cells. Older adults, by contrast, often start from a lower peak but may exhibit slower contraction of the antibody compartment, although their overall humoral reserve is diminished and their protection against reinfection can be more fragile. The Fuzhou data cannot disentangle these mechanisms directly, since the study measured circulating IgG concentrations rather than memory B cell dynamics, but the trajectories are consistent with the broader literature on immune system aging, or immunosenescence, and with prior observations that antibody decay curves after SARS-CoV-2 infection are not uniform across populations.

Beyond age, the study identified several factors independently associated with IgG levels. Sex, notably, was not one of them; men and women showed no meaningful difference in antibody concentrations once other variables were accounted for. Symptom status, however, mattered. Asymptomatic individuals had lower IgG levels than those who had experienced symptomatic infection with recovery within one week, a finding that aligns with the well-established relationship between antigen load and the magnitude of the adaptive response. A symptomatic infection typically involves higher viral burden and more sustained antigenic stimulation, driving a stronger and more durable antibody output, whereas an asymptomatic or minimally symptomatic infection may fail to fully engage the germinal center machinery that produces high-affinity, long-lived antibody-secreting cells.

Vaccination history and smoking also left their fingerprints on the antibody curves. Participants with recent vaccination had higher IgG levels than those without, an expected but important confirmation that vaccine-boosted humoral immunity remains measurable in convalescent populations even months after infection. This hybrid immunity, the combination of vaccine-primed and infection-elicited responses, is now recognized as producing the most durable antibody profiles against SARS-CoV-2. Smoking history, on the other hand, was associated with lower IgG levels, adding to a growing body of evidence that cigarette smoke alters respiratory mucosal immunity and systemic inflammatory states in ways that can blunt or reshape antibody responses. The authors also observed inter-district heterogeneity in IgG levels across Fuzhou, meaning that where participants lived was associated with measurable differences in antibody concentrations. The study did not determine the cause of this geographic variation, and the authors note that it warrants further investigation; plausible explanations could include differences in local vaccination uptake, prior exposure intensity, or demographic composition, but the data available do not settle the question.

Technically, the study’s design choices deserve attention from readers who follow seroepidemiology. The chemiluminescence immunoassay used provides quantitative IgG values rather than simple positive-or-negative calls, which is what makes trajectory modeling possible in the first place. The Yeo–Johnson transformation is a thoughtful choice over the more common log transformation because it does not require strictly positive values, giving the modelers more flexibility when handling the full distribution of titers. Linear mixed-effects models with random intercepts and slopes allow the estimation of both average population-level decline and individual-level variation around that average, and the inclusion of fixed effects for age group, sex, symptom status, vaccination, smoking, and district permits simultaneous adjustment for potential confounders. Multiple imputation, when properly specified, is preferable to complete-case analysis in cohorts with uneven follow-up. Together, these methods produce estimates that are more robust than the simple before-and-after comparisons that dominated early pandemic serology.

The public health implications of the findings are straightforward even if the mechanisms remain open. If antibody persistence genuinely differs by age, then blanket assumptions about population-level immunity, the kind that feed into booster scheduling and reinfection risk models, may misrepresent both the youngest and oldest segments of the population. Children with rapidly waning antibodies may become increasingly susceptible to reinfection between waves, while older adults, despite a stable early antibody phase, face the steepest consequences when that protection eventually erodes. The authors frame their results as supporting age-stratified longitudinal surveillance, and the recommendation is well taken: understanding who loses antibody protection and when is a prerequisite for rational booster policy, particularly as new variants continue to emerge and the antigenic distance from prior infection grows.

The study also carries the usual caveats that apply to observational cohort research. It was conducted in a single Chinese city over a six-month window in 2023, a period when the circulating variant landscape and vaccination coverage differed from earlier phases of the pandemic, so the absolute IgG levels and decay rates may not generalize directly to other populations or time points. The six-month follow-up captures only the early-to-intermediate phase of antibody waning, and longer observation will be needed to determine whether the age-related patterns persist, accelerate, or plateau. The inter-district differences, meanwhile, remain an intriguing but unexplained signal. Still, with 2,482 participants tracked across three time points and a modeling framework built to handle the messiness of real-world longitudinal data, the study provides a valuable reference point for the ongoing effort to map the durability of anti-SARS-CoV-2 immunity, and it reinforces a conclusion that has been accumulating across the field since 2020: antibody memory after this virus is not a single curve but a family of curves, shaped by age, exposure history, and the biology of the individual carrying them.

Subject of Research: Longitudinal dynamics of SARS-CoV-2 IgG antibodies in convalescent individuals across age groups

Article Title: SARS-CoV-2 IgG antibody and associated factors in a large longitudinal cohort of convalescent individuals

Article References: Wang, X., Liang, Z., Peng, H., Yang, X., Zhang, C., Chen, Z., & Zhang, X. (2026). SARS-CoV-2 IgG antibody and associated factors in a large longitudinal cohort of convalescent individuals. Virology Journal. https://doi.org/10.1186/s12985-026-03328-6

Image Credits: AI Generated

DOI: 10.1186/s12985-026-03328-6

Keywords: SARS-CoV-2, IgG antibodies, antibody dynamics, longitudinal cohort, humoral immunity, age-specific dynamics, seroepidemiology, vaccination, smoking, asymptomatic infection, immunosenescence, Virology Journal