Real-life assessment of the multidimensional nature of dyspnoea in COPD outpatients

Strengths and weaknesses

Patients were recruited by a population of respiratory physicians representative of all types of medical practice. These physicians received the necessary training to explain the MDP (using a standard script) and the other questionnaires to their patients, but the extent of this training was purposely kept compatible with daily clinical practice. The number of patient exclusion criteria was kept to a minimum.

We believe that the main strength of the study resides in its pragmatic, real-life nature. In addition, de novo psychometric validation of the MDP (see supplement S3) was conducted in order to verify that its previously described properties held true in our study population [11, 13, 14].

We acknowledge that the study also has several weaknesses. First, a significant amount of missing data was observed for certain descriptive criteria and certain outcomes. For example, only 97 of the 276 MDP questionnaires were fully completed, which might seem problematic; however, the median amount of missing answers was only one out of 12 questions, and it should be kept in mind that in this study the patients had a large number of questionnaires to complete. Secondly, the large number of participating physicians is bound to have induced a certain degree of heterogeneity, but this can also be considered to be a strength. Thirdly, the consecutive nature of inclusions was not strictly controlled. Finally, we could not evaluate the responsiveness of the MDP, because only a small number of patients chose the exact same type of dyspnoea-triggering event to focus on for the “before” and “after” measures.

On this point, it is important to remember that the MDP “is intended to refer to a specific event or time” [14], as opposed to, for example, the Dyspnoea-12 scale, which refers to an indeterminate recent period [8]. We asked our patients to focus on “their worst experience within the last 15 days”. This slightly deviates from the MDP doctrine because it introduces some recall subjectivity regarding “the event”, but this was a constraint of our “real-life” preoccupation (it would have been somewhat artificial to designate a unique dyspnoea-triggering activity for all the patients, clinically too vague to “lump” the last 2 weeks together, and too restrictive to focus on the day before the visit). Insofar as previous data have shown that the test–retest reliability of the MDP can be weak for recall periods of 4–6 weeks [13], we took the precaution to limit our recall period to 2 weeks. Reassuringly from a methodological standpoint, the test–retest reliability of the MDP applied with these choices was good in our population (supplement S3, table S5), and we found test–retest intraclass correlation coefficients similar to those obtained by Parshall et al. [13] over a very short recall period (during the emergency department visit) and way above those reported by the same authors over a 4–6-week interval.

Contribution to the available MDP dataset

The MDP is a recent tool for the evaluation of dyspnoea. It has been developed over several years, and, as only to be expected, has been somewhat unstable during this development [14]. The consolidated version of the questionnaire was only made easily and freely available in 2015 [14]. Consequently, only a limited volume of “MDP dyspnoea” data is currently available.

The team that developed the questionnaire has published an extensive review [14]. Very briefly, the use of an MDP precursor in a study of experimental dyspnoea demonstrated that, for a given level of sensory intensity, the unpleasantness of respiratory sensations varied with their nature (“air hunger” sensation was more unpleasant than “work/effort” sensation) [10]. A study conducted in 151 adult patients admitted to three urban emergency departments in the southwestern USA and presenting with respiratory symptoms of various origins showed that the MDP exhibited good reliability, validity and responsiveness to clinical improvement [11]. This study is important because it establishes the validity and the reliability of the instrument for the first time according to a generic approach (unselected population) and under pragmatic conditions.

Our study differs from the 2012 study by Meek et al. [11] in that we studied stable patients suffering from a single disease (COPD), while Meek et al. [11] studied acutely ill patients suffering from various respiratory diseases. Nevertheless, our results on validity and reliability are similar to those reported by Meek et al. [11]. Parshall et al. [13] completed the analysis performed by Meek et al. [11] by evaluating test–retest reliability in the same cohort of patients. As aforementioned, we observed similar performances in this regard in our patients. It is of note that, in the studies by Parshall et al. [13] and Meek et al. [11], sensory and affective scores were higher than those obtained in our study, possibly due to the acute nature of the dyspnoeic episode evaluated. A small pilot study in COPD patients demonstrated that activities of daily living elicit similar sensory and affective responses, as evaluated by the MDP, to experimental dyspnogenic stimuli [12].

Our study therefore extends the available MDP dataset, in the wider perspective of the description of the affective dimension of dyspnoea in COPD [23–26]. One important point is that our data lend support to the validity of the MDP as an actual multidimensional instrument. This is clearly shown by the psychometric validation (see supplementary material), but is also apparent in the clinical validation. In this respect, the “immediate perception domain” (S) component and the “emotional response domain” (A2) component of the MDP exhibited a different pattern of correlation with the various PROs (table 3). Reassuringly, S was more strongly correlated with “physical” PROs than with A2 (e.g. SF-12 PCS or DIRECT score) and A2 was more strongly correlated with “mental” PROs than with S (e.g. SF-12 MCS).

Our results also show that the affective component of the MDP increases with the severity of dyspnoea as assessed one-dimensionally (figure 2), but that there is a disproportionate difference in A2 between the mMRC grade 3 and mMRC grade 4 categories (figure 3), which was not observed for S. The multivariate analysis performed on our dataset (table 4) identified three factors as being significantly linked to the affective dimension of dyspnoea. Among these factors, the existence of clinical depression does not come as a surprise and is somewhat reminiscent of the interferences that have been described between dyspnoea and affective state in COPD patients [27]. The relationship between A2 and the type of physical activity reported by the patients is an interesting finding, inasmuch as exercise training and more generally pulmonary rehabilitation have been shown to improve dyspnoea-related anxiety [25] and more generally the affective dimension of dyspnoea [28]. In fact, in a study conducted in 48 COPD patients undergoing a rehabilitation programme, Wadell et al. [28] observed that the affective dimension of dyspnoea was more responsive to the programme than its sensory dimension. It has also been shown that COPD patients exhibiting the highest baseline dyspnoea-related fear derive greater benefits from rehabilitation than other patients [29].

At first sight, the strongly significant relationship between A2 and FEV₁ appears to be less expected and could be considered to be surprising, especially as FEV₁ has the reputation of being poorly correlated with dyspnoea, but this reputation may not be entirely warranted [30]. However, in a study conducted among 98 COPD patients in whom affective descriptors of dyspnoea were correlated with indicators of physical disability, Williams et al. [31] reported that “affective descriptors or the emotional response to the sensation of breathlessness have a significant relationship with severity of COPD impairments”, as assessed by the BODE index (body mass index, airflow obstruction, dyspnoea, exercise capacity). Williams et al. [31] interpreted this observation as indicating that affective descriptors of dyspnoea reflect the degree of threat imposed by the sensation. Our data suggest that “frustration” plays an important role in the affective dimension of dyspnoea (figure 2). Frustration is likely to become greater with the progressive shrinking of daily activities associated with progression of obstruction, which could also contribute to explaining our observations and those of Williams et al. [31].

Perspectives

Cluster analysis in COPD is a complex issue, and our intention in the present study was only to test whether or not the affective dimension of dyspnoea could contribute to COPD phenotyping. Figure 5 and table 5 suggest that this could be the case. Being able to identify an affective dimension of dyspnoea, or, more importantly, identify a population of COPD patients who are highly likely to exhibit a strong affective dimension, might have important therapeutic implications. COPD-related dyspnoea, in the wider sense of the term, is often only partially responsive to the usual symptomatic approaches. Likewise, a great many COPD patients experience “residual” exertional dyspnoea despite “maximal” pharmacological and nonpharmacological (rehabilitation) treatment, not to mention refractory dyspnoea that persists at rest despite treatment. It would therefore be desirable to widen the therapeutic arsenal against dyspnoea. Targeting the affective dimension of dyspnoea is most likely a promising option in this perspective.

Certain interventions have already demonstrated specific effects in this domain. This is the case for distractive auditory stimuli, which have been shown to reduce the unpleasantness of dyspnoea more than its intensity, during a 6-min walking test in COPD patients [32]. Perhaps more interesting from a practical point of view, cognitive behavioural strategies may have a particular impact on dyspnoea-related anxiety [33]. Recent data have shown that self-management programmes may prove particularly efficient in this regard [34]. This type of intervention is not easy to implement and requires particular skills and professional competences. It is therefore important to select patients most likely to benefit from these interventions based on a “personalised” medicine approach. The results of our clustering analysis suggest that this could be a realistic objective (figure 5 and table 5). It is of note that “simple” pulmonary rehabilitation resources are also scarce, and, according to the results of this study, selecting patients who have an important affective dimension to their dyspnoea and limited physical activity could be a means to optimise these resources.

In conclusion, our study provides the first large-scale demonstration that the MDP can be used in a clinical setting, and it demonstrates that it is possible to identify patients with a prominent affective dimension of dyspnoea. Further studies are needed to evaluate the clinical usefulness of this instrument to assess the efficacy of therapeutic interventions, and its value to tailor novel therapies to individual patient profiles.