This study included young adults aged 19 to 39 years experiencing symptoms of long COVID. The criteria for long COVID symptoms were based on the commonly reported symptoms in the literature¹⁷, including persistent fatigue and dyspnea, with symptoms relieving at 6 months post-infection¹⁸. Inclusion criteria were (1) 4 weeks to 6 months since diagnosis of COVID-19; (2) experiencing persistent fatigue and/or dyspnea; (3) no history of hospitalization due to COVID-19; (4) no history of COVID-19-specific treatment or medication. Participants underwent a screening process, including health and medical history questionnaires, and were excluded if they had cardiovascular, musculoskeletal, metabolic, inflammatory diseases, or any medical/physical limitations.

Initially, 39 participants were recruited through community online boards and flyers posted at local community. After a thorough explanation of the study’s purpose and procedures, those who voluntarily agreed to participate signed an informed consent. Nine participants were excluded based on the inclusion criteria and lack of interest or scheduling conflicts, resulting in a total of 30 participants. These participants were randomly assigned to either a combined exercise (EX) group (n=15) or a control (CON) group (n=15). The EX group engaged in a structured exercise program 3 times/wk for 8 weeks, while the CON group maintained their usual lifestyles. The overall study design is presented in Fig. 1. All procedures were approved by the Institutional Review Board (IRB) of University of Seoul (No. IRB-2022-11-004).

All dependent variables were measured by the same researcher before and after the intervention using consistent standard operating procedures. All measurements were taken under the following conditions; participants refrained from vigorous physical activity, smoking, alcohol consumption, and food intake prior to the measurements. Post-intervention measurements for the EX group were taken 48 hours after the last exercise session to ensure enough recovery.

Anthropometrics were measured using extensometer and weight scales. Body composition measures, including body mass index, body fat percentage (%), and skeletal muscle mass (kg), were assessed using a bioelectrical impedance analysis device (BWA 2.0, Inbody). Blood pressure (BP) was measured after 10-minute rest in a supine position using an automated oscillometric device (Mobile-O-Graph NG, IEM) on the left upper arm. BP measurements were taken in duplicate, with a 3-minute interval, and the mean value was used for analysis. Additional measurements were taken when the difference between the measurements was greater than 10 mm Hg, and the mean of the two closest readings was used.

Long COVID symptoms and severity were assessed using self-reported questionnaires. The types of symptoms were surveyed based on the major symptom list provided by WHO, while the severity of fatigue and dyspnea was evaluated using specific questionnaires. Fatigue was measured using the validated Korean version of the Fatigue Severity Scale (FSS), which evaluates the severity of fatigue over nine items on a scale from 1 to 7¹⁹. The FSS scores are presented as mean values, with a score of 4 or above considered clinically significant fatigue²⁰. Dyspnea was assessed using the Korean version of the modified Medical Research Council (mMRC) scale recommended by the American Thoracic Society for measuring dyspnea. The mMRC scale ranges from 0 to 4, classified as “no symptoms” to “severe dyspnea”²¹.

CRF was evaluated using a maximal cardiopulmonary exercise test (CPET) on a treadmill (TM 55 Treadmill, Quinton Cardiology Systems) following the Bruce protocol. Criteria for maximal exercise included achieving two of the following: a rate of perceived exertion of 18 or above, a respiratory exchange ratio (RER) of 1.15 or above, or an exercise heart rate reaching 90% of the predicted maximum. Continuous 12-lead electrocardiogram monitoring and BP measurements at 2-minute intervals during the test stages were used to monitor exercise responses. Gas exchange measurements were taken every 15 seconds with a gas analyzer (TrueOne 2400, Parvomedics) using a breath-by-breath method. Maximal oxygen consumption (VO_2peak) was determined by the highest 30-second average of oxygen consumption obtained during the test. The percentage of predicted VO_2peak was calculated based on age, sex, height, weight, and exercise protocol. The anaerobic threshold (AT) was identified as the point where the carbon dioxide production exceeds oxygen consumption, presented as a percentage of VO_2peak at that point.

Muscle strength was assessed using an isokinetic muscle strength test with a Biodex System 4 Pro (Biodex Medical Systems), evaluating knee extension and flexion movements of the dominant leg. Participants were fastened to the equipment with straps around the chest and thigh to minimize movement outside the knee joint. The axis of the knee joint was aligned with the dynamometer’s axis, and the range of motion and leg weight were calibrated. Participants performed five repetitions at a velocity of 60°/sec to measure muscle strength and power, and 10 repetitions at 180°/sec to assess muscle endurance with a 2-minute rest between each test. To familiarize participants with the testing procedure and equipment, two practice attempts at 70% effort were allowed before the actual test. Muscle strength was presented as peak torque and relative peak torque (peak torque/body weight [BW]), muscle power as average power, and muscle endurance as total work performed.

The exercise intervention consisted of supervised sessions 3 times/wk for 8 weeks, with each session lasting approximately 60 minutes. The exercise protocol included aerobic, resistance, and respiratory muscle training which started and ended with a 5-minute warm-up and cool-down, respectively. This exercise protocol was designed based on effective exercise prescription for long COVID from previous research¹¹.

Aerobic training was performed on a treadmill at an intensity of 50%–70% of heart rate reserve (HHR) for 30 minutes. The intensity was set using the Karvonen equation based on the maximal heart rate measured during the CPET, and progressively increased from 50%–60% HRR in weeks 1–4 to 60%–70% HHR in weeks 5–8. Heart rate was monitored using a wireless heart rate monitor (Fitbit Charge 2, Fitbit). Resistance training targeted major muscle groups through four exercises (upper extremity, chest press, and seated row; lower extremity, leg extension, and leg curl) using weight machines. Each exercise was performed for two sets of 12 repetitions at 40%–50% of one repetition maximum (1RM), with a 2-minute rest interval between sets. The 1RM for each exercise was assessed at the first exercise session and re-evaluated at week 5 to adjust the exercise intensity. Respiratory muscle training involved inspiratory muscle training using a hand-held inspiratory resistance device (IM1910, GH INNOTEK). Participants inhaled against given resistance to strengthen the inspiratory muscles, primarily the diaphragm. Participants performed three sets of 10 repetitions at 40%–50% of their maximal inspiratory pressure (MIP), with each repetition consisting of 5 seconds of resistance breathing followed by 5 seconds of relaxed exhalation. The intensity was set at 40% of MIP for weeks 1–4 and increased to 50% for weeks 5–8. MIP was estimated using a predictive equation based on the participant’s age and sex²². Data of participants in the EX group who attended at least 80% of the total exercise sessions (20 out of 24 sessions) were analyzed. The CON group was instructed not to engage in additional physical activities and to maintain their usual lifestyle throughout the 8-week period.

Participants were randomly assigned in a 1:1 ratio to either a combined EX group or a CON group, using a simple randomization allocation method based on a random number table. The sample size for this study was determined based on previous research investigating the effects of an 8-week exercise intervention on long COVID symptoms, CRF, and muscle strength in adults with long COVID¹³. The results showed an approximate 2.1 mL/kg/min increase in VO_2peak, with an effect size (pη²) of 0.14. Using G*Power software with an effect size (f) of 0.403, a significance level of 0.05, and a power of 95%, the calculated required sample size was 24 participants. Considering potential dropouts, the final sample size was set at 30 participants (15 participants in each group).

Data were presented as frequencies, percentages, or means± standard deviations. The normality of data was tested using independent t-tests and Shapiro-Wilk tests, respectively. For normally distributed data, a two-way analysis of variance with repeated measures was used to examine changes and mean difference with a 95% confidence interval (CI) in dependent variables by intervention (EX vs. CON) and time (baseline vs. follow-up). Bonferroni post-hoc tests were conducted to examine changes within groups over time and differences between groups at each time point. For non-normally distributed variables, changes (Δ) were analyzed using the Kruskal-Wallis test (for number of long COVID symptoms and mMRC scores). Correlation analysis was performed to explore the relationship between improvements in CRF, muscle strength, and the alleviation of long COVID symptoms. All statistical analyses were conducted using SPSS-PC version 27.0 (IBM Corp.), with significance set at p<0.05.