Regular Exercise Is Associated With Low Fatigue Levels and Good Functional Outcomes After COVID-19 A Prospective Observational Study

The study addresses the persistent symptom burden and functional outcomes among individuals recovered from SARS-CoV-2 infection, amid highly variable prevalence estimates of post-COVID (long COVID) due to differing definitions. The CDC defines post-COVID conditions as health problems persisting 4+ weeks after infection. Prior work links comorbidities and acute disease severity to post-COVID manifestations, but determinants of long-term recovery, symptom resolution, and trajectories remain unclear. This prospective observational study aimed to characterize the distribution of persistent symptoms and their impact on function and quality of life after COVID-19, and to identify patient-level risk factors for these ongoing symptoms. Data from the initial study visit are reported. Research questions: What is the distribution of persistent symptoms and their impact on function and quality of life among patients recovered from SARS-CoV-2? Do these patients have identifiable risk factors for persistent symptoms?

Multiple reports indicate substantial post-infection symptom burden, with fatigue frequently among the most prevalent long COVID symptoms. Persistent symptoms can occur irrespective of hospitalization or ICU-level care, though definitions of post-COVID vary. Studies have found both comorbidity profiles and severity of acute illness to relate to post-COVID outcomes; one prospective study identified hypothyroidism and hypoxia as risk factors for long COVID sequelae. Neurologic manifestations are a major concern. Comparisons of hospitalized vs nonhospitalized cohorts show that a sizable fraction have symptoms at 30, 60, or ≥90 days post-onset, and hospital discharge cohorts have high readmission or mortality within 60 days. Calls exist to expand rehabilitation referrals post-hospitalization. Prior work also documents impacts on physical and cognitive function, and quality of life, and emphasizes the importance of standardized outcome measures in rehabilitation medicine.

Design: Prospective observational study within a large health system. Population and recruitment: Adults (≥18 years) with documented positive SARS-CoV-2 PCR or antigen test identified via electronic medical records at Inova Health System starting October 14, 2020. Eligible participants were ≥28 days post-diagnosis and able to consent. Exclusion: receipt of neutralizing monoclonal antibody treatment for COVID-19. IRB approval obtained; informed consent written (in-person) or verbal (virtual). Data collection approach: Conducted in-person at clinic or remotely (phone/video), using a standardized protocol. The International Classification of Functioning, Disability and Health (ICF) framework guided selection of outcomes across impairments, activity, and participation. Structured interviews captured social/medical history, details of COVID-19 illness, current exercise habits, and changes in health behaviors post-COVID (exercise, alcohol, smoking, diet, weight, sleep). Symptom assessment: Participants reported any new symptoms present at the initial visit that were not present pre-COVID; responses were free-text without prompting lists. Investigators categorized symptoms into common groups, reviewed by a study physician. Physical assessments (in-person): Physical exam, vital signs, 2-minute walk distance (2MWD) test, and handgrip strength. Absolute 2MWD and grip values were normalized to age- and sex-adjusted population norms and expressed as percent difference: (actual − norm)/norm × 100%. Patient-reported outcomes (electronic self-administered): FACIT-F (fatigue and well-being domains), PHQ-9 (depression), Post-COVID Cognitive Questionnaire, Edmonton Symptom Assessment System (ESAS), GAD-7 (anxiety), Stigma Questionnaire, NCCN Distress Thermometer (DT), EQ-5D-5L (index and health score). Cognitive assessments (in-person or online): Subtests from D-KEFS (Color Word Interference: color naming, word reading, inhibition, inhibition/switching) and WAIS-IV (Arithmetic, Coding, Symbol Search, Matrix Reasoning). Scaled scores (mean 10, SD 3) used with age corrections. Statistical analysis: Descriptive statistics as n (%) or mean (SD). Group comparisons via chi-square (categorical) or Kruskal–Wallis (continuous). Generalized linear models (continuous outcomes) and logistic regression (categorical outcomes) evaluated independent associations using bidirectional stepwise selection (entry α=0.15; stay α=0.05). Two-sided P<0.05 deemed significant. False discovery rates were estimated for multiple comparisons. SAS 9.4 used.

Sample: N=218 enrolled; 34% (n=74) in-person. Mean age 51 ± 16 years; 45% male; 64% employed; 53% college educated; mean 194 ± 89 days since diagnosis. During acute illness: 95% symptomatic; 57% received treatment; 46% hospitalized; 34% required supplemental oxygen; 6% mechanical ventilation. Self-rated health (1–10): Pre-COVID 7.9; post-COVID 6.5. Post-COVID symptoms: Mean 1.76 symptoms; most common ≥15% were fatigue, memory loss, and shortness of breath. Among those reporting fatigue, severity: 41% mild, 43% moderate, 16% severe. Objective physical function: Grip strength significantly below norms (mean −14% vs norm; P<0.0001). 2MWD not significantly different from norms overall. No difference in 2MWD or grip between hospitalized vs nonhospitalized patients. PROs vs norms: FACIT-F fatigue scale, PHQ-9, and EQ-5D index significantly impaired vs population norms (P<0.05). Hospitalized patients had lower FACIT-F functional well-being and EQ-5D index/health scores than nonhospitalized (P<0.05). Cognition vs norms: D-KEFS color naming and WAIS-IV arithmetic scores lower than age/sex norms (P<0.05). Hospitalized patients had lower WAIS-IV symbol search (processing speed) than nonhospitalized (P=0.014). Processing speed composite tended to be lower in hospitalized (P=0.06). Symptom burden by hospitalization: Hospitalized reported more symptoms at visit (2.2 vs 1.3; P=0.015). Proportions with any symptom did not differ significantly (66.7% vs 55.3%). Severe fatigue subgroup (FACIT-F fatigue <29.7): Of 98 who completed FACIT-F, 29 (30%) had severe fatigue. Severe fatigue associated with younger age (mean 44 vs 55 years; P=0.003), lower proportion male (32% vs 54%; P=0.05), higher rates of anxiety (28% vs 9%; P=0.015), chronic kidney disease (17% vs 4%; P=0.033), and depression (28% vs 7%; P=0.0067), and less regular exercise (38% vs 63%; P=0.022). No differences in hospitalization rates. Worse post-COVID diet, weight, and sleep in severe fatigue group (all P<0.05). Severe fatigue group had markedly worse PROs across nearly all instruments (P<0.05), but no differences in 2MWD or grip strength. WAIS-IV matrix reasoning trended lower (P=0.05). They reported more symptoms (3.34 vs 1.96; P=0.0186). Correlations: 2MWD and grip r=+0.40 (P=0.0057). Cognitive scores correlated most with ESAS and Stigma scores (|r| up to 0.53), and with FACIT-F functional well-being (up to 0.39). Multivariable predictors: - FACIT-F fatigue scale (higher=less fatigue): male β=+4.5 (P=0.033); regular exercise β=+7.0 (P=0.0014); CKD β=−7.7 (P=0.051); depression β=−7.1 (P=0.024); each additional symptom β=−1.8 (P<0.0001). - Severe fatigue: per year age OR=0.94 (0.91–0.98; P=0.0014); anxiety OR=4.86 (1.21–19.5; P=0.026); each additional symptom OR=1.26 (1.04–1.53; P=0.018). No fatigue: regular exercise OR=3.63 (1.17–11.28; P=0.026); each additional symptom OR=0.65 (0.47–0.90; P=0.0086). - PHQ-9: age β=−0.07 per year (P=0.013); regular exercise β=−3.6 (P=0.0004); CKD β=+4.1 (P=0.024); depression history β=+2.9 (P=0.050); each additional symptom β=+0.9 (P<0.0001). - EQ-5D index: age β=−0.003 (P=0.0026); regular exercise β=+0.13 (P<0.0001); depression β=−0.10 (P=0.018); number of symptoms β=−0.013 (P=0.052). - Cognition: D-KEFS inhibition associated with regular exercise β=+1.9 (P=0.0015); D-KEFS inhibition/switching with college degree β=+1.3 (P=0.019); WAIS-IV symbol search lower with prior hospitalization β=−2.1 (P=0.003) and higher with college degree β=+2.6 (P=0.0003). Grip strength higher with regular exercise β=+15.1 (P=0.021).

Post-COVID patients demonstrated significantly worse fatigue, depressive symptoms, and health-related quality of life compared with population norms. Grip strength was below normative values and associated with cognitive performance measures, reinforcing its role as a marker of global function. Hospitalized patients exhibited greater depressive symptoms and lower quality of life compared with norms and performed worse on a processing speed measure than nonhospitalized peers; however, overall, hospitalization status was not an independent predictor of most PRO or cognitive outcomes after adjusting for confounders, indicating that acute disease severity alone does not fully drive long COVID burdens. Fatigue was the most prevalent symptom in both hospitalized and nonhospitalized groups and severe fatigue was linked to worse PROs across domains, more symptom burden, and adverse post-COVID health behaviors, independent of hospitalization history. Regular exercise emerged as a consistent, independent correlate of better outcomes, including less fatigue and depression, higher quality of life, better executive functioning, and stronger grip strength, suggesting a potentially modifiable factor for prevention and mitigation of long COVID impacts.

In this prospective observational cohort assessed a median ~6 months post-infection, prior SARS-CoV-2 infection was associated with reduced grip strength, higher fatigue and depressive symptoms, and lower quality of life versus norms. Regular exercise was consistently associated with more favorable patient-reported, cognitive, and physical outcomes, whereas greater symptom burden and comorbid conditions (e.g., CKD, depression, anxiety) were linked to worse outcomes. Hospitalized individuals tended to be older with more comorbidities and demonstrated more depression and lower QoL compared with norms. Encouraging regular exercise, optimizing management of comorbidities, and targeting fatigue and depressive symptoms may help mitigate long COVID effects. Ongoing longitudinal follow-up at 6 months and 1 year is planned to characterize recovery trajectories and inform interventions.

- Potential selection bias: volunteers from a single health system may over-represent individuals with greater symptom burden. - Pre-COVID health status based on retrospective self-report, subject to recall bias. - Wide window between infection and enrollment (months variability) and evolving variants over March 2020–July 2022 may introduce heterogeneity. - Cross-sectional analysis of initial visit only; longitudinal outcomes not yet reported. - Not all participants attended in-person; objective measures (grip strength, 2MWD) and some PROs/cognitive tests had missing data, which could reduce power and introduce bias. - Exclusion of patients treated with monoclonal antibodies may limit generalizability to all COVID-19 populations.