[Sample] Durability of protection from mRNA COVID-19 vaccines

Background

The rapid deployment of the BNT162b2 and mRNA-1273 vaccines in late 2020 was followed by an equally rapid question: does protection last? Early trial readouts reported efficacy at roughly two months of follow-up ¹², which is a short window for a durability question. The subsequent literature splits into three strands: extended trial follow-up, population-scale observational waning estimates, and laboratory studies of immunological memory ⁸.

Findings

Initial efficacy

Both pivotal trials were large, blinded, placebo-controlled RCTs. BNT162b2 showed 95.0% efficacy against symptomatic disease ¹; mRNA-1273 showed 94.1% ². Both estimates carry narrow confidence intervals and held across age subgroups, but both measured over roughly two months.

Waning of protection

Extended follow-up of the BNT162b2 trial showed efficacy declining from ~96% (months 2–4) to ~84% (months 4–6) ³. An Israeli cohort measured antibody titres falling several-fold by six months after the second dose, with faster decline in older adults ⁴. A systematic review and meta-regression of 18 studies found effectiveness against infection waning substantially within six months, while effectiveness against severe disease remained high — on the order of 90% — over the same period ⁷.

Correlates of protection

A modelling synthesis across seven vaccines found neutralising antibody titres to be highly predictive of protection against symptomatic infection, providing a mechanistic reading of the waning data: as titres fall, protection against infection falls with them ⁵. The correlate is strongest for infection, weaker for severe disease, which is consistent with the observation that severe-disease protection holds up better ⁸⁷.

Immunological memory

Despite declining circulating antibodies, longitudinal sampling shows durable memory B cells and stable antibody responses months after vaccination, with cross-reactivity against variants of concern ⁶. This separates two questions the waning literature sometimes conflates: the durability of circulating neutralising activity versus the durability of memory that can be recalled on exposure ⁸.

Critical appraisal

The RCT efficacy estimates are the strongest evidence but cover short windows ¹². The waning estimates are largely observational and vulnerable to testing bias and confounding by risk profile ⁴⁷. Correlate analyses are inferential — titre predicts protection in aggregate, not for an individual ⁵. Finally, most of this evidence predates Omicron, so absolute effectiveness numbers should not be carried forward to later variants ⁸.

Gaps

The literature to this window does not answer: how long memory responses persist beyond a year; whether correlates for severe disease can be established with the same confidence as correlates for infection; and how variant-specific updates reset the durability clock. A standard-depth update would target these three questions directly.

Conclusion

Two-dose mRNA vaccination gives strong but time-limited protection against infection, with more durable protection against severe disease, and neutralising antibody titre is the best available predictor of the trajectory. Confidence in this conclusion is moderate: the pattern is consistent across designs, but the evidence base is young and largely pre-Omicron.