In this prospective study, a convenience sampling strategy was used to select stable patients with COPD in a Class III general hospital in Nantong City, Jiangsu Province as the survey objects from January 15, 2023 to April 20, 2024. Inclusion criteria: Patients specifically diagnosed with COPD according to “China’s guidelines of diagnosis and treatment of chronic obstructive pulmonary disease (2020 revision)” (17), patients who participated in the respiratory center’s outpatient PR program, had no symptoms of acute exacerbation in the first 4 weeks, were in a stable condition, and were 60 to 89 years old, receiving regular drug therapy, permanent residents of Nantong (living in Nantong for more than half a year), with clear consciousness and, no language communication obstacles were included in the study. Patients with other respiratory diseases, mental illness, or serious heart, liver, or kidney disease; patients with motor dysfunction, Patients who did not complete the corresponding PR courses (less than 70%); and patients with incomplete information, such as those whose PR program could not be reviewed were excluded from the study. Patients were divided into a benefit group (achieving both: ≥30 m 6MWD and ≥4-point SGRQ) and non-benefit group. In this study, binary Logistic regression analysis was used to screen the influencing factors of pulmonary rehabilitation. The sample size calculation was based on the following parameters: with 6MWD as the primary outcome measure, the minimum clinically important difference (MCID) for 6MWD in patients with COPD was 30 m. G*Power 3.1 software was used for a priori analysis, and the test power (1-β) was set as 0.8, the significance level α = 0.05, and the effect size Cohen’s d = 0.5 (moderate effect). At least 128 samples were needed for prediction. Sixteen independent variables were included in this study. According to the rule, the sample size should be 5 to 10 times the number of independent variables. An initial sample size of 160 cases was considered for this study. Considering a 10–20% rate of invalid questionnaires (18), at least 178 cases were required for the study. The study was approved by the Ethics Committee of the Sixth People’s Hospital of Nantong (approval number: NTLYLL2023015), and all patients provided written informed consent.

According to the criteria recommended by the 2019 Asian Working Group on Sarcopenia (AWGS) (28), factors such as muscle strength, limb skeletal muscle mass index (RASM), and low physical performance should be considered in the evaluation of sarcopenia. (1) When measuring grip strength, participants were asked to hold the dynamometer as hard as they could to evaluate the handedness and non-dominant hand grip strength (kg) using dynamometer evaluation. According to 2019 AWGS standards (28), for men a grip strength < 28 kg, and for women a grip strength < 18 kg is considered as low muscle strength. (2) The formula used to calculate RASM is limb skeletal muscle mass (kg)/height (m)², and limb skeletal muscle mass (ASM) refers to the following formula:

ASM = 0.193 * weight (kg) + 0.107 * height (cm) − 4.157 * gender − 0.037 * age (years) − 2.631. Weight was measured by Omron TMHN-286 scale, height was measured by SecaTM213 altimeter, and gender was set to 1 if the patient was male and 0 if the patient was female. Several studies have demonstrated that ASM calculated by this formula is in good accordance with the dual-energy X-ray absorbent (DXA) method (28). If a female had RASM ≤ 5.7 and RASM ≤ 7 in males along with low muscle strength, it was considered as confirmed sarcopenia. (3) Gait speed was measured by timing the patients while walking 50 meters. If the gait speed was less than 1.0 m/s, the patient was considered to have low physical function. The formula may overestimate the muscle mass of patients with edema (such as right heart failure), and is only recommended as an alternative by AWGS (when DXA/BIA is unavailable).

St. George’s Respiratory Questionnaire (SGRQ) was used to evaluate the quality of life of patients from three main aspects of health including symptoms, mobility, and disease impact. The scores range from 0 to 100. A higher score indicates worse health status of patients. A reduction of 4 or more units in the SGRQ score between groups is considered clinically significant. The Cronbach’s α coefficient for the SGRQ in patients with COPD is 0.98 (29).

The modified Medical Research Council (mMRC) dyspnea index is a widely used tool in clinical practice to measure the severity of dyspnea in patients (30). The scale is a 5-point scale. A score of 0 indicates difficulty in breathing only during strenuous activity and increasing in turn, and a score of 4 indicates severe difficulty in breathing while leaving the house, or breathlessness when dressing or undressing.

The BODE index is a multidimensional tool used to assess the prognosis and severity of COPD. To evaluate disease impact, it integrates 4 key components including BMI (B), degree of airway obstruction (O), degree of dyspnea (D), and exercise capacity (E) (31). The sum of the four points is the BODE index, with lower scores indicating better conditions. A score of 0 to 2 indicates mild disease condition, 3 to 4 indicates moderate, 5 to 6 indicates severe, and a score of 7 to 10 extremely severe Table 1.

The Family Care Index (Family APGAR Index, APGAR) (32), is a member of the Family subjective evaluation tool for Family satisfaction. It is scored based on five items. Each item is scored on a scale of 0 to 2, where 2 indicates “normal,” 1 indicates “sometimes” and 0 indicates “rarely.” The total score ranges between 0 to 10 points. A score of 7 to 10 is considered good family functioning, and a score of 0 to 6 is considered a severely dysfunctional family. The scale of Cronbach’s alpha coefficient is 0.80 ~ 0.88, and is widely used in the family in China, and has demonstrated good validity in assessing family satisfaction.

Data was collected by five trained personnel during the PR program. The BMI, 6MWD test, SGRQ, mMRC, and BODE index scores were recorded before the start of the PR program. The 6MWD and SGRQ were re-evaluated and recorded in the medical record system after the PR course by the evaluators who did not participate in the PR program. The researchers conducted a thorough check and verification of data collected by two researchers to ensure its completeness, authenticity, and accuracy.

SPSS 26.0 statistical software was used for data analysis. Patient age, 6MWD results, SGRQ scores, and mMRC scores were found to conform to a normal distribution. The mean and standard deviation describe; gender, drinking, smoking, and body mass index (BMI). Other classification data descriptions included frequencies and composition ratios. χ² test, t-test, and Mann-Whitney U test were used for univariate analysis. Logistic regression analysis was used to identify factors associated with the benefits of PR in patients. The significance level was set at alpha = 0.05. Sensitivity analyses included three levels: (1) mandatory inclusion of demographic variables (Model 1); (2) increasing indicators of disease severity (Model 2); (3) Multiple imputation analysis (Model 3). All analytic variables including age, gender, 6MWD, SGRQ, mMRC scores, and smoking status. Twenty imputed datasets were generated using chained equations. The Fully Conditional Specification algorithm was implemented with predictive mean matching for continuous variables and logistic regression for categorical variables. Convergence was confirmed after 50 iterations per chain when autocorrelation function values fell below 0.1. All procedures were executed in SPSS 26.0 MVA module with random seed fixed at 202305.

The generalized estimating equation (GEE)was used to analyze the 6MWD and SGRQ scores of patients in the benefit group and the non-benefit group before rehabilitation and at the end of 3, 6, and 9 weeks of rehabilitation, and the trend profile was drawn. The GEE method can effectively incorporate information on missing data to estimate model parameters, eliminating the need for imputation or deletion of missing data.

Among 254 initially enrolled patients with COPD, 196 (77.2%) completed the PR program. 58 patients (22.8%) discontinued participation. Detailed temporal patterns of missing data (baseline, 3-week, 6-week, 9-week) and Little’s MCAR test results (χ² = 7.32, p = 0.29) are provided in Supplementary Table S1. 6 patients were readmitted due to acute exacerbation, 34 patients did not respond to the urging, 7 patients could not be admitted to the hospital due to force majeure factors, and 4 patients did not respond. 11 cases did not have corresponding 6MWD or SGRQ evaluations, as shown in Figure 1. Among the 196 completers, 107 cases (54.59%) met the benefit criteria (an increase of ≥ 30 meters in 6MWD and a reduction of ≥ 4 points in SGRQ), 121 patients only met the 6MWD improvement standard, and 107 patients only met the SGRQ improvement standard.

A prospective cohort study was conducted with 196 patients aged 60 to 87 years (mean age 73.18 ± 6.59 years). Univariate analysis identified several baseline factors including 6MWD, chronic pain, sarcopenia, economic situation, and APGAR to be statistically significant (p < 0.05), Table 2.

The study investigated whether the benefits of PR could be predicted by various baseline factors. The significant variables from the univariate analysis were included in a binary logistic regression analysis. The specific assignment of each variable is shown in Table 3. Logistic regression analysis identified five significant predictors of PR response (Table 4). Chronic Pain (OR = 0.43, 95% CI 0.22 ~ 0.83; p = 0.011), indicating that patients with chronic pain demonstrated a 57% reduction in the probability of benefit (1-OR) compared to those without pain; Sarcopenia (OR = 0.50, 95% CI 0.27 ~ 0.95; p = 0.035), suggesting that sarcopenic patients exhibited half the probability of benefit attainment; Economic Advantage (OR = 1.96, 95% CI 1.03 ~ 3.71; p = 0.039), showing that economically stable participants had nearly double the likelihood of benefit; Family Support (APGAR: OR = 2.11, 95% CI 1.08 ~ 4.11; p = 0.029), indicating that high family support doubled the probability of benefit; Baseline 6MWD (OR = 0.98, 95% CI 0.96 ~ 0.99; p < 0.001), with each 10-meter increment in baseline walking distance conferring an 18% reduction in benefit odds (1–0.98¹⁰).

Effect sizes remained stable (< 5% change) for chronic pain (OR = 0.45 vs. 0.43) and sarcopenia (OR = 0.52 vs. 0.50) after adjustment for age, sex, and BODE index, as shown in Table 5. In addition, to validate the potential impact of baseline 6MWD on the efficacy evaluation, baseline 6MWD was mandatorily included as a covariate in the GEE analysis. After adjusting for baseline 6MWD, the time trend in the benefit group (β = 4.1 vs. original 4.3) and the independent effect of baseline walk distance (β = −0.12/m, p < 0.001) remained significant. For every 1-m increase in baseline, the subsequent improvement in 6MWD decreased by 0.12 m, suggesting a persistent influence of baseline functional status on the rehabilitation trajectory, as shown in Table 6.

This study utilized GEE to analyze repeated measures data. The model specifications were as follows: (1) an exchangeable correlation structure was selected as the optimal working correlation matrix based on the Quasi-Likelihood Independence Criterion (QIC = 328.7 vs. 335.2 for unstructured and 341.5 for autoregressive); (2) an identity link function with Gaussian distribution was applied for the continuous dependent variables (6MWD and SGRQ).

6MWD changes (Figure 2) GEE showed significant time effect (Wald χ² = 67.3, p < 0.001), between-group difference (χ² = 14.8, p < 0.001) and time-by-group interaction (χ² = 8.9, p = 0.003). Changes in SGRQ (Figure 3) shows the group main effect (χ² = 11.2, p = 0.001), time effect (χ² = 53.7, p < 0.001) and interaction effect (χ² = 5.9, p = 0.015), and the specific improvements are shown in Table 7. The changes in Δ6MWD and ΔSGRQ during pulmonary rehabilitation in the benefit group and the non-benefit group are shown in Figure 4.

As a comprehensive intervention measure, PR includes exercise training, health education, and strategies to promote behavior change, etc. Evidence indicates that (30) these measures will have a positive impact on reducing the rate of readmission and mortality of patients with COPD. However, some studies have shown that at the end of the PR, nearly one-third or one-half of the patients do not show significant improvements in terms of sports ability and/or quality of life (5, 7). This research shows that at the end of a nine-week PR course, the benefit rate for patients with COPD was 53.76%, which is consistent with the findings from previous research results. PR is a challenging intervention for patients because it involves the feasibility assessment of PR, the design of clinical and community rehabilitation nursing PR programs, the adherence of patients to rehabilitation behavior, and the support of family members. Therefore, the influencing factors of PR benefits can be considered in the preliminary evaluation, and if there are risk factors, intervention should be performed before rehabilitation which will be conducive to improving the clinical rehabilitation benefit rate.

The results of this study indicate a negative correlation between the baseline 6MWD and benefits from PR. An odd ratio (OR) of 0.977 indicates that the lower the baseline 6MWD, the benefits from PR are more likely. That is, for every 10-meter reduction in the baseline 6MWD, the probability of achieving a PR response increases by 23% (1/0.977^10 = 1.23). This is consistent with the stratified analysis results of Costi et al. (33): Compared with the baseline 6MWD > 350 m group, the baseline 201–300 m group (OR = 1.9) and ≤200 m group (OR = 1.5) had a significantly higher benefit probability, suggesting the presence of a “low start-high gain” effect. Patients with frailty follow-up after 6 months found a significant correlation between exercise capacity and the likelihood of benefiting from PR in patients with COPD (34). PR primarily targets improving the physical activity and respiratory functions in patients with COPD. Those with poor baseline sports ability have more room for improvement and are more likely to benefit from targeted exercise interventions. Due to the ceiling effect, patients with COPD with good exercise capacity may reach a plateau state and be unable to further enhance their exercise capacity through self-directed efforts in the short term. That is why targeting patients with poor physical function for PR training is beneficial. Clinical and community nursing staff play a crucial role in supporting patients undergoing PR, especially for those with lower baseline physical function. Key strategies (35) for nursing staff to enhance patient outcomes and promote a positive attitude toward PR include: educating the patients about the benefits of PR, training them, and encouraging them thus developing a positive mental attitude of patient toward PR. This approach can improve patient outcomes, enhance the success rate of PR programs, and empower individuals to actively manage their chronic respiratory condition and improve their overall quality of life. We verified the robustness of group differences and dynamic changes by adjusting for baseline 6MWD. Although the baseline value had a predictive effect on the outcome (β = − 0.12), the effect size of the core interaction term (group × time) was only slightly attenuated (4.1 vs. 4.3), indicating that the rehabilitation advantage in the benefit group was not driven by differences in baseline functioning. This finding further supports the use of baseline 6MWD as a patient stratification indicator rather than a confounder.

34.79% of patients participating in this study were reported to experience chronic pain issues. A study reported that the prevalence of chronic pain in COPD ranged from 21 to 82% (33), and studies report systemic inflammatory state associated with COPD contributed to the development and persistence of pain (36). Patients with COPD mainly experience pain caused by multiple factors such as pulmonary pain, chest discomfort, musculoskeletal problems, anxiety, depression, and drug side effects (37, 38). The lower rehabilitation benefit rate observed in patients with chronic pain undergoing PR can be due to poor physical adaptability and the influence of long-term disease makes it difficult to adapt to the intensity and frequency of PR training. Other major factors lowering the rehabilitation benefit rate are psychological factors such as anxiety and depression (38). Therefore, when carrying out PR for patients with chronic pain and COPD, it is necessary to comprehensively consider their physical and psychological conditions, formulate personalized rehabilitation programs, and provide professional guidance and support to improve their benefit rates.

Sarcopenia is defined as a progressive and systemic skeletal muscle disease, characterized by loss of muscle mass and function over time (28). It is estimated that about 5 to 13% of the “healthy” older adults population may experience sarcopenia (39). In patients with COPD, the prevalence of muscle disease ranges from 8.38 to 52.1% (40). This study found that 39.28% of the patients had sarcopenia, which may be related to the older age of the patients included in this study. Sarcopenia in patients with COPD may be caused by multiple factors such as hypoxia, malnutrition, inflammatory response, reduced muscle activity, and drug side effects (39). Due to the reduction of muscle mass, patients with sarcopenia may have poor exercise tolerance and difficulty in adapting to the intensity and frequency of PR training, which will lead to fatigue and discomfort during training and affect the rehabilitation effect, thus being the main reason for observing low rehabilitation benefits in such patients (41). Further, patients with sarcopenia may lack muscle strength and face difficulty in controlling their position, affecting the correctness of pose during PR training thus limiting the training effects. Therefore, given the sarcopenia patients with COPD, it is important to pay special attention to formulating individualized rehabilitation plans, gradually increasing the intensity of training, and providing professional guidance and support, to improve its benefit rate. This study used an equation method to assess skeletal muscle mass, which, despite rigorous validity validation, may underestimate muscle mass heterogeneity. Integration of standardized tools such as BIA/DXA should be prioritized in future multicenter studies. The clinical predictive value of the 50-meter walk test needs to be further validated in a larger sample.

A multi-center cross-sectional survey in China suggests that family support and better economic situations contribute to improved rehabilitation outcomes for patients (42). However, the improvement in rehabilitation outcomes is the subjective feeling of patients in this study, and there is no unified standard.

Patients with better economic and family environments often have enhanced lifestyle support (43). Economic stability ensures access to rehabilitation resources (such as protein supplements and Sports bracelet), Economic security reduces cost anxiety and enhances participation in sports (44). These patients also obtain better psychological support, including psychological counseling and psychotherapy, which can help them overcome psychological obstacles in the process of rehabilitation and improve the rehabilitation outcomes (45). In the PR plan, therefore it is essential to recognize the role of family and encourage family members to actively participate in the patient’s rehabilitation process, to improve the effectiveness of rehabilitation. However, the family of patients with poor economic backgrounds should also try their best to provide support and resources, to ensure effective rehabilitation outcomes.

The results of this study demonstrated that after 9 weeks of rehabilitation, both the 6MWD and SGRQ scores showed improved trends in both groups. The statement is generally consistent with Bishp’s study (46), however, it fails to address the evolving trend observed in both benefit and non-benefit groups. The exercise capacity of the benefit group improved more than that of the non-benefit group from 6 to 9 weeks of rehabilitation, and the SGRQ scores of the benefit group showed greater improvement than those of the non-benefit group at any time point. The possible reasons are as follows: Through systematic rehabilitation training, patients experienced improvements in lung function, muscle strength, and endurance after 6 weeks, resulting in significant improvement in exercise capacity. (Inferences about lung-function improvement in our study were based on optimization of the response to exercise ventilation rather than on direct spirometry). The improvement in the SGRQ scores of the benefit group at any time point may be attributed to the positive effects of rehabilitation training on the quality of life and psychological state of patients. Thus, rehabilitation training can not only improve the physical condition of patients but also enhance their self-confidence and psychological state. It helps in reducing anxiety and depression, leading to an overall improvement in their quality of life and mental health.

Compared with other studies, the Hafner study (47) focused on the physiological characteristics of patients, suggesting that a more muscular body composition and higher ability to deliver oxygen from the blood to the muscles may be beneficial for PR outcomes. Ragaselvi (6) and Crisafulli (7) reported Osteoporosis was independently associated with poorer recovery outcomes, but the current study did not confirm this finding. BMI had no impact on the patient’s recovery benefits, which is consistent with most research findings (47, 48). It is important to note that this study did not systematically assess sedentary behavior or levels of daily physical activity. Existing evidence suggests that sedentary time is significantly associated with decreased exercise tolerance in patients with COPD, which may affect PR effect by reducing muscle metabolic fitness. “Although we indirectly reflected functional status as measured by baseline 6MWD, future studies are needed to integrate objective monitoring tools such as accelerometers to quantify the dynamic impact of sedentary behavior on PR response. The reliance on self-reported exercise intensity in tele-rehabilitation lacks validation against accelerometer data. This may introduce measurement bias, as COPD patients typically overestimate daily step counts (49). Future studies must integrate wearable sensors (e.g., ActiGraph GT9X) to objectively quantify sedentary behavior and its dynamic impact on PR response.