This prospective, single-center, parallel-group randomized controlled pilot study was conducted in Qingdao, China, between March 2023 and September 2023. The study was performed at the Department of Rehabilitation Medicine, Qingdao Municipal Hospital, a tertiary public hospital affiliated with the University of Health and Rehabilitation Sciences. The study followed CONSORT guidelines for randomized trials. The study flow diagram is presented in Figure 1, which shows the recruitment, allocation, and follow-up processes for participants in this trial.

This study was reviewed and approved by the Ethics Committee of Qingdao Municipal Hospital (Approval No. 2023 Pro-review No. Y020), and clinical trial registration was completed (Registration No. ChiCTR2100052874).

40 stable COPD patients were randomly assigned to either the ST group (n = 20) or the Health Education (HE) group (n = 20), with “HE” referring to structured disease education sessions provided by trained healthcare professionals. Participants were recruited from six communities within Qingdao, China. The recruitment process involved conducting community outreach presentations and posting recruitment flyers at local healthcare centers and public areas within each community. These efforts aimed to raise awareness of the study and inform potential participants about the research.

Eligible participants were then screened based on the inclusion and exclusion criteria for stable COPD patients. Those who met the criteria were invited for further screening, which included a comprehensive assessment to confirm their suitability for participation. Informed consent was obtained from all participants before enrollment.

The demographic and baseline clinical characteristics of both groups are detailed in Table 1. There were no significant differences between the groups in terms of age, sex, BMI, or other baseline characteristics (P > 0.05).

Randomization was conducted by an independent researcher using sequentially numbered, opaque, sealed envelopes. Allocation concealment was maintained until interventions began. The investigator who performed randomization was not involved in assessments or data analysis. Participants and facilitators were aware of group allocation, while outcome assessors and data analysts were blinded to group assignment.

Inclusion criteria were: 1) age <80 years; 2) diagnosis according to GOLD criteria, with FEV₁/FVC <70% after inhalation of bronchodilators for inclusion in primary screening; 3) negative bronchodilator test to exclude asthma; 4) stable condition with no acute exacerbations in the 4 weeks prior to enrollment; 5) ability to undergo subgroup randomization; and 6) ability to participate in regular aerobic exercise and complete the 6-min walking test.

Exclusion criteria were: 1) unstable cardiovascular disease, such as recent hospitalization for decompensated heart failure, unstable angina, or myocardial infarction within the past 6 months; 2) vocal disorders (e.g., vocal cord nodules, chronic laryngitis); 3) severe cognitive impairment; 4) participation in aerobic training (or other disease-related systematic physical training) within the past 6 months; 5) severe hypoxic respiratory failure that could not be safely managed during pulmonary rehabilitation (e.g., resting SpO₂ < 88% on room air or requiring long-term supplemental oxygen therapy); 6) insufficient education and comprehension levels to qualify for ST (Qiao et al., 2024).

Both groups received the same structured COPD HE delivered weekly by a trained respiratory nurse, in addition to usual care. The ST group additionally attended facilitator-led ST sessions, whereas the control group received HE only. Thus, the comparison was HE + ST vs. HE alone. Participants in the ST group received a 12-week intervention, which involved two 60-min sessions per week. Based on the practices outlined in McNamara et al. (2017), which included training durations ranging from 60 min, 2 times per week, to 90 min, 1–2 times per week, we selected a 60-min session, delivered twice a week, in line with local standard practices. This duration allows for a balanced integration of singing training, breathing control exercises, and relaxation, while minimizing potential physical strain on participants (McNamara et al., 2017). Sessions were conducted by facilitators specialized in respiratory therapy and music-based interventions.

Detailed ST:

Part I: Preparation and Warm-up (about 20 min). 1.

Breathing Awareness and Positive Thinking Exercise (5 min):

Instruct the patient to sit in a comfortable position with one hand on the abdomen and one hand on the chest. Guide them to close their eyes and feel their natural breathing, then consciously perform slow and deep inhalations and exhalations with the goal of feeling the forceful sensations of inhalation and exhalation and focusing their mental attention on their body. 2.

The warm-up process consists of 10 min of respiratory muscle:

Sternocleidomastoid, obliques, intercostals stretching, and 5 min of combined chest and abdominal breathing exercises. This was followed by gentle vocal warm-ups such as lip trills (lip blowing), humming, and sustained exhalation of “sssss” and “shhhh” to activate the respiratory muscles and vocal organs.

Part II: Core ST (approximately 35 min). 1.

Breathing control exercises

Slow inhalation-long exhalation exercises: Instruct the patient to inhale deeply through the nose for 3 s, and then exhale through the mouth by slowly retracting the lips, prolonging the exhalation time to 6–10 s as much as possible. 2.

Music Theory Learning Stage

Tone name C D E F G A B and the ascending and descending notes of each note, singing name 1 2 3 4 5 6 7 and the ascending and descending notes of each note, so as to facilitate the patient’s recognition of music. 3.

Song singing (30 min)

Choose songs with soothing melodies, steady rhythms, and moderate phrase lengths. Initially choose old songs and folk songs with short phrases, and gradually introduce songs with slightly longer phrases as the patient’s ability improves.

Part III: Relaxation and organization (about 5 min).

Practice slow, deep breathing exercises while singing an extremely soothing song or engaging in group humming to gradually restore heart rate and breathing rate to resting levels. Use the Brog scale to assess the exertion level of this training session. Adaptations were made for COPD patients to avoid vocal strain, excessive respiratory load, or breath-holding.

For transparency and reproducibility, we also provide a more detailed description in the Supplementary Appendix.

A soothing song and relaxation were used to reduce physical tension after prolonged breath control and vocalization.

Assessments were performed at both the beginning and the end of the 12-week rehabilitation program. Three domains were evaluated: clinical characteristics, diaphragmatic function, and functional performance indicators.

Results include clinical characteristics, primary outcomes, and secondary outcomes.

Primary outcome: Change in diaphragmatic function (mobility and thickening fraction) measured by ultrasound during quiet and deep breathing.

Secondary outcomes: Respiratory muscle strength; Pulmonary function (FEV₁, FVC, FEV₁/FVC); Exercise capacity (6-Minute Walk Distance, 6MWD); Health status (COPD Assessment Test, CAT; St George’s Respiratory Questionnaire, SGRQ); Psychological symptoms (Hospital Anxiety and Depression Scale, HADS).

We report between-group differences in change from baseline to post-intervention with 95% CIs and, for selected outcomes with established Minimal clinically important difference (MCID), the proportion achieving prespecified MCID thresholds (defined as the smallest change perceived as important and sufficient to justify a change in management). For outcomes without well-defined MCIDs, we additionally applied a transparent, pragmatic ≥10% improvement-from-baseline threshold, following prior post hoc analyses by Kaasgaard et al., as an exploratory indicator of potentially clinically relevant change.

This study used 40 participants to assess feasibility and estimate variance for future trials. Assuming a within-group effect size of 0.8 for diaphragmatic mobility, 80% power, α = 0.05, and 15% dropout, a sample of 20 per group was considered adequate.

All analyses were performed using SPSS 26.0 and GraphPad Prism 10.1.2. Missing data were handled using last observation carried forward (LOCF). Continuous variables are presented as mean ± SD. Between-group comparisons used independent t-tests; within-group pre–post comparisons used paired t-tests. A p-value <0.05 was considered statistically significant.

Twenty patients were included in each of the ST group and the HE group. Two participants withdrew from the ST group during the intervention period (2/20, 10%), leaving 18 participants who completed the trial (90%). Five participants withdrew from the health education group due to personal reasons (5/20, 25%; four withdrew on their own and one withdrew for planned lung decongestive surgery), leaving 15 participants who completed the trial (75%). Overall, 33/40 participants completed the study (82.5%). A comparison of the baseline data showed no statistically significant differences in baseline characteristics between the two groups (P > 0.05).

COPD patients are characterized by decreased diaphragm mobility, which related to a progressive decline in lung function and exercise tolerance (Shiraishi et al., 2024). Given this pathological feature, exploring strategies to optimize respiratory muscle activity patterns has become an important direction in pulmonary rehabilitation research (Dominelli and Molgat-Seon, 2022). Although studies on ST in COPD have shown improvements in various outcomes, such as exercise capacity, spirometry, and global respiratory muscle pressures, as well as symptoms and health-related quality of life, as a result, commonly used endpoints provide limited insight into whether ST significantly impacts diaphragmatic mechanics. (Kim, Yeo, and Kim, 2023; Qiao et al., 2024; Philip et al., 2024; Fang et al., 2022; Pacheco et al., 2014). Against this background, our study used diaphragm ultrasound (diaphragmatic mobility and thickening fraction) to provide a direct, muscle-level assessment of the physiological effects of ST.

Major guidance documents (e.g., GOLD and ATS/ERS statements) define the key outcomes of Pulmonary Rehabilitation as improvements in dyspnoea, exercise capacity, and health-related quality of life—all of which are strongly influenced by diaphragmatic dysfunction (Rochester et al., 2023). Hyperinflation and diaphragm flattening limit tidal-volume expansion, increase neuromechanical dissociation, and lead to exertional dyspnoea and premature exercise termination, further impairing these PR outcomes (O'Donnell et al., 2020). Thus, improved diaphragm mechanics are not merely physiological observations but have clear translational relevance to PR goals. In this pilot study, ST was associated with a clearer between-group improvement in dynamic diaphragmatic excursion and thickening than health education, whereas static mobility changes were smaller.

This improvement aligns with the expectations of previous research on ST, as singing requires precise regulation of deep and prolonged respiratory cycles that provide continuous and rhythmic stimulation to the diaphragm. In COPD, lung hyperinflation flattens and shortens the diaphragm, placing it at a mechanical disadvantage on its length–tension curve and increasing the inspiratory neural drive required to generate a given tidal volume (Zhang et al., 2022).

Previous studies highlighting diaphragmatic mobility as a key functional measure in COPD rehabilitation, and diaphragmatic mobility influences dyspnea and exercise tolerance in patients with COPD (Shiraishi et al., 2020; Shiraishi et al., 2021; Paulin et al., 2007). In conjunction with our findings, the study indicates that ST may enhance diaphragmatic function, which in turn could improve various clinical symptoms in COPD patients, such as dyspnea, exercise capacity, and overall respiratory efficiency.

This study shows that compared to the HE group, the ST group showed more significant improvements in MIP and MEP. Our MIP results are consistent with those of Kaasgaard et al. (2022b), where one of the main findings was that the Singing for Lung Health group, in those who achieved SGRQ-MID (Minimal Important Difference in Quality of Life), showed significant improvements in MIP, whereas the Physical Exercise Training group did not show similar changes (Kaasgaard et al., 2022a). Contacting with the findings of Kaasgaard et al., our results suggest that singing training may improve diaphragmatic function and respiratory muscle strength, which in turn enhances health-related quality of life. This training method offers a highly promising supplementary approach to pulmonary rehabilitation for COPD patients, particularly when traditional physical training methods fail to meet patients’ needs, and holds significant clinical relevance. Moreover, physiological models of breathing control have suggested that singing can optimize diaphragmatic recruitment, improving both diaphragmatic function and overall respiratory muscle coordination, which may explain why ST improves inspiratory pressures and related PR outcomes (Lewis et al., 2021).

Some studies have shown that ST can increase lung capacity by improving breathing patterns (Bonilha et al., 2009; Lewis et al., 2016). The study showed that the 12-week ST program resulted in a significant improvement in the pulmonary function index FEV1/FVC, with a change of +3.7 (from 53.4 ± 12.9 to 57.1 ± 14.99) in the ST group. In contrast, the HE group showed a minimal change of +0.9 (from 47.2 ± 11.6 to 48.1 ± 11.3). The within-group comparison revealed a statistically significant improvement in FEV1/FVC in the ST group (p < 0.05), but no significant change was observed in the HE group (P > 0.05). Furthermore, the between-group comparison did not reach statistical significance (P > 0.05). Theoretically, singing requires continuous adjustment of breathing and intonation and precise control of airflow, which cannot be separated from the diaphragm’s regulation of intra-abdominal pressure and airflow dynamics. ST encourages the use of diaphragmatic breathing to enhance diaphragm engagement, which improves alveolar ventilation efficiency and reduces dead space ventilation, thereby enhancing overall ventilation efficiency (Lu et al., 2020). The subglottal pressure during singing may be similar to the positive expiratory pressure of lip-contracted breathing, which could create resistance to expiratory airflow and potentially help prevent bronchial collapse in early COPD (Philip et al., 2021). Bonilha’s study (Bonilha et al., 2009)demonstrated a significant improvement in lung function indices (Inspiratory capacity) during singing, but there was no significant difference in post-training measurements. This suggests that there is a transient improvement in hyperinflation during singing in patients with COPD, and may explain why patients with non-significant improvement in static lung function are effective in terms of motor function and quality of life.

Paulin (Paulin et al., 2007) demonstrated that diaphragmatic mobility was positively correlated with 6MWT scores, suggesting that dyspnea, exercise capacity, and quality of life are interconnected and important indicators in COPD patients. Overall, the synergistic improvement of dyspnea, enhancement of exercise capacity, and improvement in quality of life have become the core goals of COPD rehabilitation therapy (Zhang et al., 2024). In the Cochrane systematic review of conventional pulmonary rehabilitation in stable COPD, the pooled between-group improvement in 6-min walk distance (6MWD) compared with usual care was 44 m (95% CI 33–55 m). Real-world data from the UK National Asthma and COPD Audit Programmed (NACAP) further show that approximately two-thirds to 70% of patients undergoing center-based PR achieve at least the 30 m MCID for the 6MWT. In our trial, the mean change in 6MWD in the ST group was 52 m, and 55.6% of patients (10/18) exceeded the 30 m threshold, indicating a moderate effect that is broadly comparable in magnitude to the average 6MWD gains reported with conventional PR, although the proportion of patients reaching the MCID was somewhat lower than that observed in large PR cohorts.

Specifically, the results of this study showed that the 6MWT scores in the ST group were significantly better than those in the health education group, indicating that ST effectively improved patients’ exercise endurance.

Consistent with Lewis et al. (2021) on the physiological link between singing and respiratory control and pulmonary mechanics (Lewis et al., 2021). ST, which involves coordinated deep inhalation and slow exhalation, improved the breathing pattern in COPD patients, reduced dynamic hyperinflation, thereby improving lung ventilation efficiency and alleviating dyspnea. This aligns with previous research suggesting that ST can coordinate deep inhalation and slow exhalation to improve breathing patterns in COPD patients, reduce dynamic hyperinflation, and enhance lung ventilation efficiency, thereby reducing dyspnea.

The larger improvement in 6MWD observed in our study compared with Kaasgaard et al. (2022a) may reflect several complementary factors related to intervention duration, delivery context, and task individualization (Kaasgaard et al., 2022b). While both programmes were delivered twice weekly, the intervention period in our study was longer (12 weeks vs. 10 weeks in Kaasgaard et al., 2022a), resulting in a greater cumulative exposure to singing-based respiratory training. This extended time course may be particularly relevant for consolidating changes in breathing pattern, respiratory muscle coordination, and tolerance to exertion.In addition, the smaller group size in our study allowed greater flexibility for individual adjustment within the group format. For participants who experienced difficulty keeping up with the full musical or respiratory demands, the facilitator could adapt tasks in real time—for example, assigning simpler phrases within the same song, encouraging humming or softer phonation during more demanding passages, and providing posture or alignment cues to reduce unnecessary accessory muscle loading. Such individualised adjustments may help maintain participation, reduce symptom-related interruption, and support gradual physiological adaptation, particularly in participants with lower baseline capacity.Finally, differences in song selection and task grading may also have contributed. In our intervention, songs were selected and adapted to emphasise moderate tempo, predictable meter (e.g., 3/4 or 4/4), manageable phrase length with planned breath points, and transposition to a comfortable pitch range. Within the same song, phrase-level task grading enabled participants with higher capacity to engage in longer phrases or sections requiring greater breath support, while participants with lower capacity focused on simplified phrases, joining selectively and re-entering at planned breath points. This symptom-guided, graded approach may have facilitated more effective engagement of diaphragmatic mechanics while preserving a cohesive and supportive group environment, potentially contributing to the observed gains in exercise capacity.

COPD not only severely impairs patients’ ability to perform activities of daily living but also significantly impacts their mental health. Psychological issues, such as anxiety and depression, frequently experienced by patients, further exacerbate disease symptoms and negatively affect quality of life (Li et al., 2025). In the present trial, ST was associated with greater improvements in health-related quality of life and psychological status than HE at the between-group level. From a clinical perspective, established MCID thresholds of four points for SGRQ and 1.5 points for both HADS subscales provide a useful benchmark for interpreting these changes. Using these thresholds, 15 of 18 patients (83.3%) in the ST group and five of 15 (33.3%) in the HE group achieved an MCID-level reduction in SGRQ total score, while 13 of 18 (72.2%) versus one of 15 (6.7%) achieved an MCID-level reduction in HADS-A and nine of 18 (50.0%) versus two of 15 (13.3%) achieved an MCID-level reduction in HADS-D. Taken together, these results indicate that a substantial proportion of patients receiving ST experienced clinically meaningful improvements in both disease-specific quality of life and mood, whereas such improvements were much less common in the HE group.

These findings are broadly in line with, but in some respects more pronounced than, previous studies of singing-based interventions and pulmonary rehabilitation in COPD(Philip et al., 2024; Lewis et al., 2016). Trials of “Singing for Lung Health” and group ST have generally reported modest improvements in SGRQ or other health status instruments and small-to-moderate reductions in anxiety and depressive symptoms (Lewis et al., 2016), with mean changes that often approach or slightly exceed the MCID but with a more limited proportion of participants achieving MCID-level improvements. In contrast, the proportion of patients in our ST group reaching MCID thresholds across SGRQ total and domains and both HADS subscales was relatively high, and the between-group differences versus HE were sizeable. On the other hand, our results fall within the range of effects reported for center-based pulmonary rehabilitation, where SGRQ improvements of approximately 8–15 points and MCID responder rates of 40%–80% have been described, although psychological benefits are more variable. Taken together, these comparisons suggest that ST may achieve quality-of-life and mood benefits of a magnitude comparable to those seen in structured rehabilitation programs in a subset of patients, while offering a different and potentially more acceptable format.

Several studies of inspiratory muscle training (IMT) in COPD have shown clear improvements in inspiratory muscle strength and related physiological outcomes, such as increased MIP and exercise capacity, but their benefits on subjective outcomes such as dyspnea, health-related quality of life and psychological status have been inconsistent. Isolated IMT interventions often fail to produce marked improvements in anxiety or depressive symptoms, likely because these outcomes are influenced by psychosocial factors beyond muscle training alone. These discrepancies with our findings may reflect differences in the underlying mechanisms; while IMT targets respiratory muscle function, ST integrates respiratory muscle engagement with pleasurable activity, emotional regulation and structured social interaction. Singing engages controlled breathing patterns that stimulate respiratory muscles and may reduce hyperinflation, while the group-based, music-oriented format provides social support and reduces isolation, which are important determinants of psychological wellbeing (Shi et al., 2024; Leonardi et al., 2022; Mori and Zatorre, 2024).

Traditional pulmonary rehabilitation programs are effective but remain markedly underutilized and difficult to access in many settings, with low participation rates and suboptimal long-term adherence (Chen, Zhang, and Curtis, 2022). There is thus a need to explore alternative or adjunctive rehabilitation strategies that are engaging and acceptable to patients. Incorporating ST into rehabilitation programs may help improve patient participation and adherence, thereby supporting long-term respiratory health (McNaughton et al., 2017). Singing training interventions show significant heterogeneity in content, delivery format (e.g., in-person vs. online), course structure, and program names, with variations across different studies and regions (Kim, Yeo, and Kim, 2023). Some studies even adjust the course content based on participant feedback and local conditions, indicating that singing training programs are greatly influenced by their delivery format, facilitator background, and regional cultural characteristics. Due to differences in study designs, it is difficult to directly compare studies, and there is a significant risk of bias (Lewis et al., 2016). Therefore, exploring an optimal, simplified singing training protocol is particularly important.

In recent years, various formats have emerged, including Singing for Lung Health, community choirs, Sing Your Lungs Out, Sing Strong, and SINFONIA (Kim, Yeo, and Kim, 2023). Compared to other singing programs, Singing for Lung Health has more potential as the best protocol due to its structured design and involvement of more researchers. Singing for Lung Health facilitators must undergo professional training, and the program is well-structured. Kaasgaard’s analysis of the Singing for Lung Health program versus Physical Exercise Training with a sample size of 270 participants (Kaasgaard et al., 2022b) is rare in singing training and provides valuable insights for future research. Our study design also references its structured approach.

We agree with Philip et al. (2024), who emphasized that future research should focus on variables such as content, dosage, and duration of the singing interventions (Philip et al., 2024). As mentioned, singing training programs are flexible and vary according to their format and regional specifics. We believe that future research on singing training programs should focus on universal factors within the programs, such as comparing the effects of different rhythms, tones, volume levels, and sentence lengths on singing training, or examining how different intervention durations and cycles impact outcomes. Researchers should also provide more detailed descriptions of the content of the singing training programs and any adjustments made to the songs to better suit COPD patients.

We suggest that future research should focus on identifying universal factors in singing training programs, such as optimal rhythm, tone, volume, and duration, and explore how these factors can be standardized to improve the effectiveness of interventions for COPD patients.