Association between patient-reported outcomes and exercise test outcomes in patients with COPD before and after pulmonary rehabilitation

Background Over the years, the scope of outcomes assessment in chronic obstructive pulmonary disease (COPD) has broadened, allowing for the evaluation of various patient-reported outcomes (PROs). As it still remains unclear whether and to what extent PROs mirror the exercise performance of patients with COPD, the current study aimed to assess the association between different exercise test outcomes and PROs, before and after pulmonary rehabilitation (PR). Methods Correlations between PROs used to describe health-related quality of life (HRQoL), mood status, level of care dependency and dyspnea in patients with COPD and commonly used laboratory- and field-based exercise test outcomes were evaluated in 518 individuals with COPD attending PR. Results Overall, correlations between PROs and exercise test outcomes at baseline were statistically significant. The correlation between modified Medical Research Council (mMRC) dyspnea score and 6-min walking distance (6MWD) was strongest (ρ:-0.65; p<0.001). HRQoL related PROs showed weak correlations with exercise outcomes at baseline. Moderate correlations were found between St George’s Respiratory Questionnaire total score and 6MWD (r:-0.53; p<0.001) and maximal workload achieved during cardiopulmonary exercise testing (ρ:-0.48; p<0.001); and between Clinical COPD Questionnaire (CCQ) total score and 6MWD (r:-0.48; p<0.001) and maximal workload (ρ:-0.43; p<0.001). When significant, correlations between changes in exercise test outcomes and changes in PROs after PR were generally very weak or weak. The highest correlation was found between changes in CCQ total score and changes in 6MWD (ρ: − 0.36; p<0.001). Conclusions PROs and exercise test outcomes, although significantly correlated with each other, assess different disease features in patients with COPD. Individual PROs need to be supported by additional functional measurements whenever possible, in order to get a more detailed insight in the effectiveness of a PR program. Trial registration Netherlands Trial Register (NL3263/NTR3416). Registered 2 May 2012.


(Continued from previous page)
Conclusions: PROs and exercise test outcomes, although significantly correlated with each other, assess different disease features in patients with COPD. Individual PROs need to be supported by additional functional measurements whenever possible, in order to get a more detailed insight in the effectiveness of a PR program.
Keywords: COPD, Patient-reported outcome measures, Exercise test, Pulmonary rehabilitation, Quality of life Background Patients with chronic obstructive pulmonary disease (COPD), a highly-prevalent chronic lung disease, frequently suffer from symptoms of dyspnea, exercise intolerance, an impaired mood status and a reduced health status [1][2][3]. These features are typically weakly related to the degree of lung function impairment [4]. Therefore, the use of additional assessments such as exercise tests and patient-reported outcomes (PROs) has been advocated [3,5,6]. Appraisal of these extra-pulmonary features is necessary to better understand the patients' daily needs or problems, to identify possible treatable traits for integrated COPD care programs, and to evaluate its efficacy [7].
Several laboratory-and field-based exercise tests can be performed to measure exercise performance, which is typically affected in patients with COPD [3,8], due to a downward spiral of dyspnea, disability and physical inactivity [9]. Important aspects from the patient's perspective like health-related quality of life (HRQoL), dyspnea, anxiety, depression, and the level of care dependency, all of which have a direct impact on daily life [10], are measured using PROs.
Punekar and colleagues systematically reviewed the strength of the available evidence supporting correlations between the outcomes of different exercise tests and PROs most commonly used to assess HRQoL and dyspnea [11]. They concluded that only a limited amount of studies have focused on the correlations between exercise test outcomes and PROs in patients with COPD. The available evidence indicates a very weak to moderate negative correlation between 6-min walking distance (6MWD) and HRQoL, measured with the St. George's Respiratory Questionnaire (SGRQ). The relationship between PROs for dyspnea and 6MWD showed contrasting results, with both moderate to strong positive and negative correlations being reported [11]. So, it still remains unclear whether and to what extent PROs mirror the exercise performance of patients with COPD. It seems reasonable to hypothesize that other exercise test outcomes than 6MWD may be stronger correlated with different PROs. For example, disease-specific questionnaires like the Clinical COPD Questionnaire (CCQ) and the COPD Assessment Test (CAT) focus more on functional impairments and symptoms related to COPD and may therefore be more closely associated with exercise test outcomes in patients with COPD.
Pulmonary rehabilitation (PR) reduces dyspnea, increases exercise capacity, and improves HRQoL in individuals with COPD [6]. Exercise training is a major component of PR and therefore exercise test outcomes are consistently used to assess the individual patient's response to PR [12][13][14][15][16][17]. Nevertheless, improvements in exercise performance after PR do not necessarily lead to a concurrent decrease in symptoms in patients with COPD and vice versa [18]. Therefore, the question remains whether changes in exercise test outcomes after PR translate into changes in disease-specific PROs.
In this observational study, we aimed to assess the association between different exercise test outcomes and PROs most commonly used to describe HRQoL, anxiety, depression and disease-specific symptoms, such as dyspnea, in patients with COPD before and after PR. A priori, we hypothesized that the correlation between PROs for dyspnea and HRQoL and exercise test outcomes would be statistically significant, but that there would be no strong or very strong association. Furthermore, it was expected that improvements in exercise test outcomes after PR showed weak correlations with changes in PROs in patients with COPD.

Study design and participants
The current study is a retrospective analysis of the 'COPD, Health status and Comorbidities' (Chance) study, Netherlands Trial Register NTR3416 [19]. The Medical Ethical Committee of the Maastricht University Medical Centre+ (MEC 11-3-070) approved this trial, which conformed to the 'Declaration of Helsinki' as amended most recently by the 64th WMA General Assembly, Fortaleza, Brazil, October 2013 [20]. The Medical Research Involving Human Subjects Act (WMO) does not apply for the secondary analysis of the Chance study. Therefore, an additional official approval of this secondary analysis by the Medical Ethical Committee is not required (MEC letter 2019-0987).
Patients with mild to very severe COPD were recruited before the start of a comprehensive PR program at CIRO in Horn, The Netherlands [21]. Patients between the age of 40 and 85 years with a diagnosis of COPD according to GOLD guidelines [22] were eligible. The protocol and part of the results of the Chance-study have been published before [1,4,10,15,19,[23][24][25][26]. All patients gave written informed consent prior to inclusion in the study.

PR program
PR took place inpatient (8 weeks, 5 sessions per week; total of 40 sessions) or outpatient (8 weeks, 3 sessions per week, followed by 8 weeks, 2 sessions per week; total of 40 sessions), in line with the 2013 American Thoracic Society & European Respiratory Society Statement [4]. Extensive pre-and post-PR assessments were performed, as described before [19].
To evaluate HRQoL, three disease-specific PROs, the CAT (range 0-40 points) [29], the CCQ (range 0-6 points) [30] and the COPD-specific version of the SGRQ (range 0-100 points) [31] were assessed in all participants. Mood status was measured with the Hospital Anxiety and Depression scale (HADS; range 0-21 points) [32]. Higher scores are equivalent to a decreased HRQoL and/or increase in symptoms of anxiety or depression, respectively. The mMRC dyspnea scale was used to establish functional impairment due to dyspnea [33]. The level of care dependency was determined at baseline with the Care Dependency Scale (CDS; range 15-75 points) with a lower score representing a higher level of care dependency [34].

Statistical analyses
Analyses were performed using SPSS software (statistical package for the social sciences) for Windows (version 25.0). Results are presented as mean and standard deviation (SD), median and interquartile range (IQR), and/or proportions, as appropriate. Continuous variables were tested for normality. Differences at baseline between completers and non-completers were analyzed using independent samples T-tests or Mann-Whitney U tests.
Correlations between PROs and exercise test outcomes were analyzed using Scatter plots and Pearson's or Spearman's correlations, as appropriate. The strength of correlations has been classified according to British Medical Journal guidelines, which regard significant correlation coefficients of 0-0.19 as very weak, 0.2-0.39 as weak, 0.4-0.59 as moderate, 0.6-0.79 as strong, and 0.8-1 as very strong [39]. A priori, the level of significance was set at ≤0.01.

Results
A total of 518 patients (55.6% male, age 64.1 ± 9.1 years) volunteered to participate and attended the pre-PR assessment. The mean baseline 6MWD was 424 ± 124 m and 25.1% of the patients had a 6MWD below 350 m [40] and in 74.7% of the patients, quadriceps muscle strength was less than 80% of the predicted value [41]. The PROs showed a high degree of dyspnea (80.7% with mMRC dyspnea grade of two or higher) [22], anxiety (34.8% with ≥10 points) [32], depression (33.4% with ≥10 points) [32], care dependency (28.5% with CDS total score of ≤68 points) [25], and an impaired HRQoL (81.9% with a SGRQ total score of ≥44 points; 75.0% with a CAT total score of ≥18 points; 76.7% CCQ total score of ≥1.9 points) [22]. Baseline characteristics, exercise test outcomes and PROs at baseline are presented in Table 1.
HRQoL PROs showed weak correlations with exercise outcomes at baseline. Moderate correlations were only found between SGRQ-C and 6MWD (r: −0.53; p<0.001) and CPET maximum workload (ρ: −0.48; p<0.001) and between CCQ and 6MWD (r: −0.48; p<0.001) and CPET maximum workload (ρ: −0.43; p<0.001). See Fig. 2 for a scatter plot illustrating the relationship between HRQoL PROs and 6MWD. CDS score was significantly correlated with all exercise test outcomes, with correlations ranging from 0.24 (CWRT cycle endurance time) to 0.50 (6MWD). Both HADS-D and HADS-A showed nonsignificant or very weak to weak correlations with all exercise test outcomes.

Correlations between changes in exercise test outcomes and changes in PROs after PR
Four hundred nineteen patients completed the PR program. Completers and non-completers were comparable with respect to baseline characteristics (Table 1). Only the amount of current smokers was significantly higher in the noncompleter group (p<0.001). All PROs and exercise test outcomes changed significantly after PR (Table S1). When significant, correlations between changes in exercise test outcomes and changes in PROs were generally very weak or weak. The highest correlation, being classified as weak, was found between ΔCCQ and Δ6MWD (ρ: −0.36; p<0.001; Fig.  2). Changes in other HRQoL PROs demonstrated similar association with changes in exercise test outcomes (Table 3). Changes in quadriceps peak muscle strength were not correlated with changes in any of the PROs.

Discussion
This study demonstrates that PROs and exercise test outcomes are associated to some extent in patients with mild to very severe COPD, but, in general, these correlations are weak to moderate. A strong relationship was merely found between the severity of dyspnea (mMRC) and distance covered in the 6MWT at baseline. In the current study, dyspnea tended to indicate at least moderate negative correlations with exercise test outcomes at baseline, suggesting that exercise performance  Fig. 1 Association between the mMRC score and 6MWD at baseline (a) and between changes in mMRC score and changes in 6MWD after PR (b) decreases as dyspnea scores increase. However, these associations attenuated considerably or even became nonsignificant once the changes in dyspnea were correlated with changes in exercise test outcomes following PR, indicating that an improvement in exercise performance after PR does not necessarily imply that self-reported breathlessness decreases concurrently, like shown before [18]. As a side remark, it is important to note that correlations between changes in parameters are always lower than cross-sectional correlations. After all, the measurement error is included twice (pre vs. post) in the analysis, which always results in a weaker signal [44]. While the mMRC-scale is a unidimensional method to quantify only dyspnea, there are several multidimensional A B C D E F Fig. 2 Left: Association between the SGRQ-C score (a), CAT score (b), CCQ score (c), and the 6MWD. Right: association between the change in SGRQ-C score (d), CAT score (e), CCQ score (f), and the change in 6MWD after PR. On the X-axis the MCID of 30 m for the 6MWT [42] is marked, on the Y-axis the MCIDs for SGRQ-C (−4.0), CAT (−2.0) and CCQ score (−0.4) are marked [10,43]. MCID, minimal clinically important difference disease-specific PROs, which assess not only dyspnea but also other symptoms and perceived HRQoL in COPD [1]. Of these HRQoL PROs (CAT, CCQ, SGRQ), their association with exercise test outcomes was weak to moderate, indicating that no single exercise test accurately reflects HRQoL (or the other way around), proving that HRQoL is indeed a multi-dimensional concept that includes domains related to physical, mental, emotional, and social functioning. Overall, these results support the findings by Punekar et al. [11] who showed that generally there was a very weak to moderate negative correlation between the 6MWD and the SGRQ.
While guidelines on the diagnosis and treatment of COPD have intensively stated that the assessment of disease severity is substantially improved by using functional criteria [22], such as exercise capacity, the current study demonstrates that the variance in PROs can only be partially explained by attributes related to exercise performance. So, despite the fact that PROs for HRQoL, dyspnea, anxiety, depression and the level of care dependency are crucial when evaluating the disease severity and effectiveness of a treatment in COPD, it is justified to conclude that these PROs assess features not measured by exercise tests. Consequently, if we solely use a few outcome measures (for example, walking distance or HRQoL) to evaluate performance after PR, the clinical complexity and multidimensional aspect of PR in patients with COPD appears to be ignored [18].
In our study, the 6MWD showed the strongest relationship with important clinical PROs, underlining the fact that the 6MWT indeed seems to play a key role in evaluating functional exercise capacity [14]. Since the 6MWT is self-paced, test outcomes are likely to be affected by a patient's mental and emotional status [3].

Limitations
Patients were solely recruited in a specialized PR centre, resulting in a selected group of COPD patients. This should be considered when applying results to other COPD samples. Furthermore, by quantifying the associated exercise limitation, a mMRC-score of 4 reflects the most disabled COPD patients who are not always able to perform a symptom-limited CPET, as a result of their dyspnea. In the current study, patients unable to perform a CPET and, concurrently, a CWRT were automatically excluded from the correlation analysis, since they did not present any values for both exercise tests, possibly affecting the correlation coefficients.

Conclusions
In conclusion, we have found that patient-reported outcomes and exercise test outcomes, although significantly correlated with each other, assess different disease features in patients with COPD. Therefore, it can be stated that relevant features from the patient's perspective like HRQoL, anxiety, depression, and the level of care dependency are not an accurate reflection of a patient's exercise capacity. The only exception to this seems to be dyspnea, the only PRO that tended to imply at least moderate association with exercise test outcomes. We would like to highlight the complexity of evaluating the effectiveness of a personalized PR program, in which we note that changes in PROs and changes in exercise test outcomes correlate poorly. Indeed, improvements in exercise capacity obtained after PR do not necessarily result in alterations in PROs in patients with COPD. Individual PROs need to be supported by additional functional measurements whenever possible, in order to get a more detailed insight in the effectiveness of a PR program.
Additional file 1: Table S1. Changes in PROs and exercise test outcomes after PR.