Back

Long COVID Recovery And Exercise Adherence: 32-Month Study

Rolo-Duarte, A.; Prada, D.; Carvalho, A. S. M.; Borges, A.; Bettencourt, P. J. G.

2025-11-13 rehabilitation medicine and physical therapy
10.1101/2025.11.11.25340028 medRxiv
Show abstract

ObjectiveTo evaluate symptom progression in COVID-19 survivors, adherence to prescribed exercise therapy, and its association with pre-infection physical activity at 21 days (T0), 6 months (T1), and 32 months (T2) post-discharge. DesignRetrospective longitudinal study in a hospital-based rehabilitation unit in Portugal. The cohort included 276 patients (mean age 56.6 {+/-} 13.5 years) with confirmed SARS-CoV-2 infection. ResultsAdherence was higher among patients reporting prior physical activity (48.8%; p = .003). Symptom prevalence declined over time: dyspnea (T0 = 22.4%, T2 = 7.3%), fatigue (T0 = 32.4%, T2 = 14.5%), and pain (T0 = 17.6%, T2 = 4.8%). Asymptomatic cases increased from 27.4% (T0) to 54.5% (T2). Early adherence, particularly by day 15, was associated with continued participation at day 21, and adherence at day 21 correlated with reduced dyspnea at follow-up (p = .02). Importantly, patients who remained symptomatic at day 21 took significantly longer to recover (t = -6.386; p < .001), indicating this time point as a prognostic marker of delayed resolution of exercise-modifiable symptoms. ConclusionEarly initiation of individualized, structured exercise proved safe, adaptable, and associated with reduced symptom burden, especially dyspnea. Persistence of symptoms at day 21 highlights the prognostic value of early follow-up and underscores the decisive role of timely rehabilitation engagement. Structured home-based and tele-rehabilitation programs supported adherence and accessibility, reinforcing exercise as a cornerstone of long COVID management and potentially applicable to other post-respiratory rehabilitation contexts.

Matching journals

The top 4 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.