Effect of a Third COVID-19 Vaccine Dose on the Incidence of Long COVID Among Adults Who Completed a Primary Vaccine Series: a Target Trial Emulation in a Community-Based Cohort

Yanhan Shen; Zach Shahn; McKaylee M Robertson; Kelly Gebo; Denis Nash

doi:10.1101/2025.11.03.25339444

Effect of a Third COVID-19 Vaccine Dose on the Incidence of Long COVID Among Adults Who Completed a Primary Vaccine Series: a Target Trial Emulation in a Community-Based Cohort

medRxiv [Preprint]. 2025 Nov 6:2025.11.03.25339444. doi: 10.1101/2025.11.03.25339444.

Authors

Yanhan Shen^{1

2}, Zach Shahn^{1

2}, McKaylee M Robertson^{1

2}, Kelly Gebo³, Denis Nash^{1

2}

Affiliations

¹ Institute for Implementation Science in Population Health (ISPH), City University of New York (CUNY); New York, NY, USA.
² Department of Epidemiology and Biostatistics, Graduate School of Public Health and Health Policy, City University of New York (CUNY); New York, NY, USA.
³ Department of Medicine, Johns Hopkins University; Baltimore, MD, USA.

Abstract

Background: Evidence on whether a third COVID-19 vaccine dose lowers long COVID risk is mixed. We estimated the effect of receiving ≥1 third dose versus completing only a primary series on 6- and 12-month long COVID incidence using a target-trial emulation in a U.S. community cohort.

Methods: We analyzed the CHASING COVID Cohort, a prospective, community-based study of U.S. adults. Eligible participants were ≥18 years, had completed a two-dose primary series, had no prior long COVID, and had no SARS-CoV-2 infection in the 3 months before time zero. Strategies compared were: receive a third dose at time zero vs. not receive a third dose during follow-up. Long COVID was defined as ≥1 new symptom at or beyond 3 months post-infection with concurrent activity limitation, both absent in the prior year. Follow-up was 6 and 12 months. We used a per-protocol analog: participants were artificially censored upon deviating from their assigned strategy or lost-to-follow-up, with inverse-probability weights to address selection due to censoring and time-varying confounding. We fit weighted pooled logistic models to estimate weighted incidence, differences, and ratios at each horizon.

Results: Across 16 sequential trials (18,930 person-trials; 4,044 unique individuals), 3,321 person-trials received a third dose at time zero and 15,609 did not. At 6 months, weighted long COVID incidence was 0.9% (95% CI, 0.5%, 1.3%) with a third dose vs. 1.0% (0.8%, 1.1%) without (risk difference (RD), -0.1%; 95% CI, -0.5%, 0.4%; risk ratio (RR), 0.93; 95% CI, 0.54, 1.44). At 12 months, incidence was 4.9% (4.1%, 5.9%) with a third dose vs. 4.5% (4.1%, 4.8%) without (RD, 0.4%; 95% CI, -0.5%, 1.4%; RR, 1.09; 95% CI, 0.90, 1.33).

Conclusion: In this community-based target-trial emulation, receiving a third COVID-19 vaccine dose did not meaningfully reduce 6- or 12-month long COVID incidence compared with completing only a primary series.

Keywords: COVID-19 vaccine; Intention to Treat; Per Protocol; Sequential Target Trial Emulation; long COVID; longitudinal.

Publication types

Preprint

Grants and funding

UH3 AI133675/AI/NIAID NIH HHS/United States