Back

Natural History of Self-reported Symptoms Following SARS-CoV-2 Infection: A Target Trial Emulation in a Prospective Community-Based Cohort

Shen, Y.; Shahn, Z.; Robertson, M.; Gebo, K.; Nash, D.

2025-11-04 infectious diseases
10.1101/2025.11.01.25339316 medRxiv
Show abstract

BackgroundThe natural history of symptoms after SARS-CoV-2 infection remains uncertain because many studies are non-representative, ignore background symptom prevalence, lack longitudinal tracking, and omit appropriate controls. Using a prospective, community-based cohort with repeated symptom measures, we estimated post-infection risks of long COVID symptoms versus contemporaneous uninfected controls. MethodsWe analyzed the CHASING COVID Cohort, a U.S. longitudinal study with surveys and serology (March 2020-December 2023). Infection status (January 2021-December 2022) was determined from self-reported PCR/antigen results, serology, or CSTE probable criteria. We emulated 24 monthly target trials comparing individuals newly infected at time zero with those remaining uninfected. Outcomes were new-onset long-COVID symptoms not reported pre-infection, assessed overall and within three clusters (neurological, autonomic, exercise intolerance) at 4-8 and 9-12 months post-infection. Inverse probability of treatment and censoring weights adjusted for confounding and informative loss to follow-up. ResultsThe analysis included 1,055 infected and 52,310 uninfected person-trials. At 4-8 months, the adjusted risk of any long-COVID symptom was 22.6% (95% CI, 20.5-24.8) among infected versus 11.3% (11.1-11.5) among uninfected (adjusted risk difference [aRD], 11.3% [9.2-13.5]; adjusted risk ratio [aRR], 2.01 [1.81-2.20]). At 9-12 months, risks were 19.2% (17.0-21.3) vs 12.4% (12.2-12.7) (aRD, 6.7% [4.6-8.9]; aRR, 1.54 [1.37-1.72]). Across all three clusters, infected participants had consistently higher risks at both intervals. ConclusionsSARS-CoV-2 infection was associated with elevated risk of new-onset long-COVID symptoms persisting up to 12 months. Using a national community-based cohort, contemporaneous uninfected controls, and target-trial emulation clarifies the burden attributable to infection and supports ongoing surveillance and targeted prevention and care.

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.