Objective:

To estimate how often adults with likely long COVID improve or achieve complete symptom resolution over time.

Approach:

• Study design: Prospective observational analysis of the RECOVER-Adult cohort at 83 U.S. sites.
• Participants: The primary analysis included 1,019 adults meeting criteria for likely long COVID 12 months after infection; additional analyses examined participants meeting criteria at 18 or 24 months.
• Outcomes: Participants completed symptom assessments every 3 months. The primary outcome was an LCRI score below 11 at two consecutive visits; secondary outcomes included a 50% or greater score reduction with good-or-better quality of life, and an LCRI score of 0.

Key Findings:

• Among participants with likely long COVID at 12 months, estimated sustained improvement was 36% by 24 months and 50% by 36 months; median time to improvement was 36 months.
• A 50% or greater symptom-score reduction with good-or-better quality of life was estimated in 41% by 24 months and 56% by 36 months.
• Complete symptom resolution was estimated in 15% by 24 months and 27% by 36 months.
• At 36 months, adjusted improvement probabilities ranged from 60% for smell- or taste-predominant symptoms to 19% for multisystem symptoms.
• Adjusted improvement probability was 51% among adults aged 18–45, 40% among those aged 46–64, and 44% among those aged 65 or older. It was 48% for participants infected during the Omicron period versus 36% for those infected before Omicron.
• Improvement probabilities were similar for female and male participants. Analyses of the 18- and 24-month cohorts showed similar patterns, though estimates were slightly lower in the 24-month cohort.

Interpretation:

The study authors described the findings as indicating potential for natural recovery while noting the need for placebo-controlled therapeutic intervention trials.

Limitations:

• Participants were selected for meeting likely long COVID criteria at specific time points, limiting generalizability to all people with prior COVID-19.
• Symptom classification could not establish that all reported symptoms were caused by infection.
• The observational design cannot establish causal relationships, and treatments during follow-up were not evaluated.
• Missing follow-up assessments required statistical imputation.
• The primary outcome represented improvement below the likely long COVID threshold, not complete symptom resolution.
• The 12-, 18-, and 24-month cohorts overlapped but were not identical and could not be directly compared.

Sources:

• Annals of Internal Medicine