This is the first study to determine the minimal important difference of the modified version of the PFSDQ in patients with severe COPD using different approaches based on state-of-the-art analytic techniques. Our estimates for the minimal important difference of the different PFSDQ-M components ranged from −3 to −6 points with anchor-based methods (using CRDQ dyspnea and total score and 6MWD as anchors) and from −3 to −5 points using different distribution-based methods (Cohen’s effect size, empirical rule effect size and the standard error of measurement method).
Anchor and distribution based method are conceptually very different . Regardless of the apparent differences between methods, there is evidence that diverse methods yield apparently similar findings [33, 34].
Despite the low correlation found between changes in the target instrument and changes of the anchors, the consistency of the result using different techniques to establish the MID and our large sample size [23, 35] support the validity of our findings. As a consequence of the low correlation coefficient (< 0.3)  between the instruments we could not use the CRDQ fatigue domain as an anchor, neither the 6MWD for the PFSDQ-M dyspnea and fatigue component.
Estimates of the magnitude of clinically meaningful change in physical performance measures can contribute to the needs of clinical and research outcomes  of interventions. Functional status assessment is an important aspect in evaluating the patient’s ability to perform daily activities. The PFSDQ-M is a reliable and valid questionnaire [3, 22]. In addition, it has been shown to be responsive to changes following pulmonary rehabilitation  and based on our data changes are independent of baseline functional status.
We suggest that when a group of patients decrease their PFSDQ-M score around 5 points this would be indicative of a clinically significant improvement of functional status. Estimations of the MID were similar in subgroups of patients with different baseline functional status. A change of −5 points corresponds approximately to a 12% change from the baseline scores and 5% of the range. This % of the range is in line with that observed in other questionnaires (e.g. SGRQ; MID of 4 points corresponds with 4% of the range) . Other studies using other interventions may further fine tune the minimal important difference of the PFSDQ-M.
The PFSDQ-M is an instrument used in pulmonary rehabilitation that provides information about patient’s symptoms and their functional performance . The meaningful change criteria of the PFSDQ-M appear achievable, because it is comparable with the magnitudes of improvement reported in research studies [37, 38]. In addition, previous studies showed significant improvement in functional status after 3-month rehabilitation program [39, 40], with decrease in PFSDQ-M scores that are in line with our results.
The estimates of change should be considered as preliminary evidence and will require further confirmation using similar as well as additional techniques. Patient ratings that could reflect perceived changes in exercise capacity and functional performance have also been used to identify minimal clinically important differences in previous studies [24, 41, 42]. However this technique may be biased by poor recollection and particularly changes in expectations when patients have to judge on themselves over a large period of time.
Limitations of the study
Although we introduced original and new findings, the study has limitations which need to be addressed. First, this is a retrospective study over a long time period (eleven years). Although over this time period some changes in staffing were inevitable, the composition of the multidisciplinary team and the approach of the exercise training and the assessments have been consistent. The occupational therapist who administered the PFSDQ-M questionnaires was the same person throughout the study period. Second, the majority of included patients suffered from severe COPD, which prevents from generalizing our findings to all GOLD stages. Third, the Dutch version of the PFSDQ-M has not been validated. Although it is generally a translation of the original English version, the items brush hair and walk on bumpy terrain were replaced by putting on socks and shopping to optimize content validity. This adaptation was made with permission of the developers of the PFSDQ-M questionnaire.
This study is the first to determine the MID of the modified version of the PFSDQ and provides a framework to judge impact of rehabilitation interventions on this patient reported outcome. Furthermore the established MID is a potential tool for clinical rehabilitation programs to assess efficacy of the program. It is however important to mention that the MID is not designed to evaluate training response in individual patients, due to the test-retest variability of the PFSDQ-M.