Baseline Characteristics of Subjects with Chronic Obstructive Pulmonary Disease Associated to the Improvement in Activities of Daily Living after Exercise Training

Abstract

This study aimed to identify baseline variables predicting improvement in ADLs following an exercise training program in subjects with COPD. Sixty-seven patients with COPD underwent assessment of spirometry, modified Medical Research Council scale, COPD Assesment Test (CAT), Six Minute Walk Test (6MWT), London Chest Activity of Daily Living (LCADL) scale and Glittre-ADL test (TGlittre). After 24 sessions, they were reassessed for limitation in ADLs (LCADL and TGlittre). The main outcome was the achieving of minimal important difference (MID) of TGlittre, LCADL and both (ADLs). The cut-off points to discriminate the subjects who achieved the MID of TGlittre, LCADL and ADLs were established using the ROC curve. A cut-off point of 3.7 min in baseline TGlittre was able to discriminate subjects who achieved the MID of TGlittre (AUC = 0.77). Subjects with baseline TGlittre ≥3.7 min were 6.92 (95%CI 2.2–20.9) times more likely to achieve the MID of TGlittre post-exercise training. A cut-off point of 32% in LCADL was able to discriminate subjects who achieved the MID of LCADL (AUC = 0.81) and in ADLs (AUC = 0.78). Subjects with baseline LCADL ≥32% were 12.3 (95% CI 2.50 − 60.7) times more likely to achieve the MID of LCADL. In conclusion, the baseline variables that best predict the improvement of individuals after exercise training are TGlittre and LCADL, showing that subjects with significant functional impairment are more likely to clinically significantly improve their ADLs.

Keywords:

• Minimal clinically important difference
• chronic obstructive pulmonary disease
• rehabilitation
• activities of daily living
• outcome assessment

Introduction

In patients with chronic obstructive pulmonary disease (COPD), the difficulty or inability to perform activities of daily living (ADLs) compromises health-related quality of life [1], which is strongly related to poorer prognosis [2,3]. Thus, limitation in ADLs is an important outcome to be evaluated in pulmonary rehabilitation programs (PRP) [4].

It is important to measure a wide range of tasks when assessing ADLs limitation [5,6]. For more specific assessment, multiple-task tests are recommended [5]. In this context, the Glittre-ADL test (TGlittre) combines several tasks, it is valid, reliable, responsive to a PRP [7] and it has a known minimal important difference (MID) [8].

Due to the variability of the impact of COPD on functionality, no functional capacity test can perfectly reflect the actual patients’ limitations [9]. Therefore, it is recommended to include questionnaires and scales to complement the assessment, since these instruments can assess patients' perceptions of the limitation in ADLs [9,10]. The London Chest Activity of Daily Living scale (LCADL) is used to evaluate the impact of dyspnoea during ADLs in patients with COPD [11]. It is valid, reliable [11,12], responsive to a PRP [13] and presents a MID [14].

Pulmonary rehabilitation is an effective intervention to reduce symptoms and improve functional status [4]. However, patients respond differently to it and not all patients improve ADLs in a clinically significant way [15]. Some studies have already shown that more severe patients benefit most from PRP [15–21]. However, these studies have focussed on improving exercise capacity, quality of life, and level of physical activity. As far as it is known, no study has investigated whether baseline characteristics can predict the improvement in ADLs by combining the MID of specific instruments for ADLs assessment.

Identifying the baseline characteristics of patients who significantly improve ADLs may help to determine which are the patients who can clinically respond to traditional therapies and who may need tailored strategies to optimise gains. Thus, the present study aimed to: (a) compare baseline variables of pulmonary function, dyspnoea, mortality index, health and functional status between COPD patients who did and did not significantly improve their ADLs post-exercise training program; (b) investigate whether these baseline variables are related to improvement in ADLs; (c) identify a cut-off point in baseline ADLs performance that can discriminate who significantly improves ADLs; and (d) identify which baseline variables are better predictors of improvement in ADLs.

Methods

Subjects

The study included subjects with clinical diagnosis of COPD confirmed by spirometry (GOLD II-IV) [22], referred to the exercise training program of the Universidade do Estado de Santa Catarina (UDESC) and Universidade Federal de Ciências da Saúde de Porto Alegre (UFCSPA) from 2013-2018. The sample of this study is part of a study previously published by our group [14] and therefore the inclusion and exclusion criteria are the same.

This is a retrospective study and was approved by the Ethics Committee for Research involving Human Beings at UDESC, UFCSPA and Irmandade da Santa Casa de Misericórdia de Porto Alegre (Protocol Numbers: 80831117.5.0000.0118, 1.413.342 and 836.248, respectively). All participants signed the informed consent form.

Study protocol

The protocol consisted of baseline evaluation, 24 exercise training sessions and post-exercise training evaluation. At baseline, data on pulmonary function, mortality index, dyspnoea, health status and functional status were collected. Immediately after 24-sessions of exercise training, subjects were reassessed for TGlittre and LCADL.

Baseline variables

Pulmonary function was assessed according to the American Thoracic Society/European Respiratory Society (ATS/ERS) recommendations [23]. Dyspnoea was assessed by the modified Medical Research Council scale (mMRC) [24]. To assess health status, the COPD Assessment Test (CAT) [25,26] total score and its cut-off point of 18 were used in the analysis [27]. The six-minute walking test (6MWT) was performed according to the ATS/ERS recommendations [28,29]. The greatest distance of two tests with a 30-min of interval and the cut-off point of 82%pred were considered in analysis [30]. The predicted values were calculated by the second equation of Brito et al. [31]. The body mass index (B), degree of obstruction (O), perception of dyspnoea (D), and exercise capacity (E) (BODE) was calculated. The total score was used in the analysis (0-10), in which the higher the score, the higher the mortality risk [32].

Assessment of ADLs

TGlittre and LCADL were considered specific measures of ADLs. Two TGlittre were conducted in pre- and post-exercise training program, as described previously [7]. For the analysis, the shortest test was considered, in minutes and in percentage of the predicted (%pred) [33]. Values over 100%pred mean that the subject takes more time to complete the test then the expected. In addition, subjects were classified into two groups according to the achievement of MID of –0.38 min [8]. The LCADL scale was applied as an interview pre- and post-exercise training [12]. The percentage score of the total LCADL was used for the analysis [12]. Higher scores indicate greater functional limitation due to dyspnoea [11]. Subjects were classified into two groups according to the MID of –4 points [14]. For the analysis, subjects were further classified into two groups: who achieved the MID of ADLs (whether they achieved the MID of both instruments: TGlittre and LCADL) and failed to achieve the MID of ADLs (whether they failed to achieve the MID of either instrument or none of the instruments). Baseline TGlittre and LCADL values were also used as independent variables in the regression analysis.

Exercise training program

The exercise training program was conducted by physical therapists according to ATS/ERS recommendations [4], thrice a week, for 24 sessions (approximately 8 weeks), according to previous description (NCT03251781).

Statistical analysis

Data normality was checked with the Shapiro-Wilk test. Pre- and post-exercise training comparisons were done with paired t-test or Wilcoxon test. Subjects who achieved or failed to achieve the MID of TGlittre, LCADL, and ADLs were compared with independent t-test or Mann-Whitney U test. Correlations between baseline variables and change (post - pre-exercise training) in TGlittre and LCADL were checked with Pearson's or Spearman's tests. Baseline variables with p < 0.1 [34] in the correlation analysis were used in the stepwise multiple regression to identify the best predictors of change in TGlittre and LCADL. The Receiver operating characteristic (ROC) curve was conducted to determine the cut-off point in the baseline ADLs variables (TGlittre and LCADL) capable of discriminating subjects who achieved the MID of TGlittre, LCADL and ADLs. To build the ROC curve, a correlation ≥0.3 between the baseline variable and the change in it was established [35]. Satisfactory discriminatory power was considered when area under the curve (AUC) was ≥0.7 [36]. Subjects were then classified according to the cut-off point found and this classification was used as an independent variable. Binary logistic regression was performed to verify the predictors of achieving MID and adjusted for the variables that presented p < 0.1 in the raw analysis. Basal variables were used as independent variables and MID of TGlittre, LCADL and ADLs as dependent variables. For variables with values analysed in absolute and %pred values (FEV₁, FVC), the one with the lowest p values in the raw analysis was chosen for inclusion in the adjusted model. The odds ratio and 95% confidence intervals (95% CI) were calculated. The significance level was set at 5%. The analysis was performed with the SPSS Statistics 20.0, MedCalc 17.1 and GraphPad Prism 6.

Sample size

Using the baseline values of the LCADL and TGlittre of the first 10 subjects who achieved and the first 10 who failed to achieve the MID of LCADL and TGlittre, and considering a two-tailed α of 0.05 and a β of 80%, a sample of 10 and 20 subjects per group was estimated, respectively. In addition, aiming to reach an AUC ≥0.7 for the cut-off points [36], the estimated total sample size was 60. The sample size was calculated with the G*Power 3.0.10 and MedCalc 17.1 softwares.

Results

From 89 eligible subjects, 22 subjects were excluded for the following reasons: death (n = 3); started smoking again (n = 1); exercise training program withdrawal (n = 8); severe COPD exacerbation during the exercise training (n = 7), cataract surgery (n = 2); and diverticulitis surgery (n = 1). Thus, 67 subjects (40 men) completed the study. The baseline characteristics of the total sample are described in Table 1.

Table 1. Baseline characteristics of the total sample and comparison between groups that achieved and failed to achieve the MID of TGlittre, LCADL and ADLs (TGlittre and LCADL) after the pulmonary rehabilitation program.

		MID of TGlittre		MID of LCADL		MID of ADLs
	Total sample (n = 67)	Achieved (n = 37)	Failed to achieve (n = 30)	Achieved (n = 27)	Failed to achieve (n = 40)	Achieved (n = 15)	Failed to achieve (n = 52)
Age, years	64.4 (4.7)	64.7 (1.6)	64 (1.3)	61.4 (1.5)	66.4 (1.4) ‡	61.5 (2.5)	65.2 (1.1)
Body mass, kg	70.5 (1.9)	71.9 (2.2)	69 (3.1)	69.8 (2.4)	71 (2.7)	74.3 (2.6)	69.4 (2.3)
Height, m	1.64 (0.09)	1.65 (0.09)	1.63 (0.09)	1.61 (0.08)	1.66 (0.09)	1.61 (0.09)	1.65 (0.09)
BMI, kg/m²	26 (0.6)	26.4 (0.08)	25.5 (1)	26.7 (0.9)	25.5 (0.9)	28.7 (1.2)	25.2 (0.7)
Smoking history, pack years	57 (4.7)	58.1 (5.78)	55.5 (7.7)	45.6 (5.4)	64.7 (6.7)	44.5 (6.3)	60.6 (5.7)
FEV1, L	1.12 (0.47)	1.07 (0.45)	1.18 (0.50)	1.05 (0.52)	1.17 (0.44)	0.92 (0.49)	1.18 (0.46) §
FEV1, %pred	39 (1.7)	36 (2.2)	41 (2.7)	38 (2.9)	39 (2.1)	33 (3.5)	40 (1.9) §
FVC, L	2.42 (0.81)	2.44 (0.87)	2.39 (0.74)	2.20 (0.86)	2.57 (0.75) ‡	2.00 (0.83)	2.54 (0.77) §
FVC, %pred	64 (2.2)	63 (3.2)	66 (3)	61 (4)	66 (2.5)	54 (5.3)	67 (2.3) §
FEV1/FVC	0.46 (0.08)	0.44 (0.08)	0.48 (0.09)	0.47 (0.07)	0.45 (0.09)	0.45 (0.07)	0.46 (0.09)
mMRC*	2 (2)	2 (4)	2 (2)	2 (3)	1 (2)	3 (3)	2 (2) §
CAT	17.5 (0.9)	16.8 (1.4)	18.3 (1.3)	20.3 (1.3)	15.6 (1.2) ‡	21.6 (1.5)	16.4 (1.1) §
BODE*	4 (8)	4 (3)	3 (3)	4 (3)	3 (2)	5 (3)	3 (2) §
6MWT, m	436 (11.6)	423 (17.5)	453 (14.3)	433 (22)	438 (12.9)	405 (34.5)	445 (11.2)
6MWT, %pred	79.4 (1.8)	76.3 (2.7)	83.2 (2.2)	78.8 (3.3)	79.8 (2)	72.7 (5)	81.3 (1.8) §
TGlittre, min	4.5 (0.2)	5.1 (0.3)	3.64 (0.2) †	4.4 (0.3)	4,5 (0.3)	5.1 (0.5)	4.3 (0.2)
TGlittre, %pred	145 (6.6)	167 (9.8)	120 (5.7) †	145 (11.4)	146 (8.1)	169 (17.9)	139 (6.6)
LCADL, %total	35.8 (1.6)	35 (2.4)	36.7 (2.1)	43.2 (2.6)	30.8 (26.4) ‡	45.4 (3.9)	33 (1.5) §

Mean ± SD.

SD: standard deviation; MID: minimal important difference; TGlitre: Glittre-ADL test; LCADL: London Chest Activity of Daily Living scale; n: number of patients; kg: kilogram; m: metres; BMI: body mass index; FEV₁: forced expiratory volume in one second; L: litres; %pred: percentage of predicted; FVC: forced vital capacity; mMRC: modified Medical Research Council scale; CAT: COPD Assessment Test; BODE: Body Mass Index, Airflow Obstruction, Dyspnoea, and Exercise Capacity; 6MWT: 6-minute walk test; min: minutes; %total: percentage of total LCADL score.

* Results presented in median [interquartile range].

† p<.05 vs. achieved the MID of TGlittre.

‡ p<.05 vs. achieved the MID of LCADL.

§ p< .05 vs. Achieved the MID of ADLs.

Subjects reduced the time in TGlittre post-exercise training compared to pre- (mean difference = –0.73, 95%CI = –0.51 to –0.94 min, p < 0.001; and –25.9 95%CI = –11.5 to –40.2%pred, p = 0.001), as well as the LCADL score (mean difference = –3.46, 95%CI = –1.18 to –5.72 points; p = 0.01). Thirty-seven subjects (55%) achieved the MID of TGlittre, 27 (40%) achieved the MID of LCADL and 15 (22%) MID of ADLs.

Comparison between subjects who achieved and failed to achieve the MID of TGlittre, LCADL, and ADLs

There were no differences in anthropometric characteristics between subjects who achieved and failed to achieve the MID of TGlittre and ADLs, while subjects who achieved the MID of LCADL were younger than those who did not (Table 1). Subjects who achieved the MID of TGlittre took longer time to perform baseline TGlittre than those who failed to achieve it (p < 0.001). Also, subjects who achieved the MID of LCADL had lower FVC (p = 0.038) and higher CAT (p = 0.02) and LCADL (p < 0.001) scores than those who failed to achieve it. Subjects who achieved the MID of ADLs (in both TGlittre and LCADL) had lower FEV₁ (p = 0.033) and FVC (p = 0.01), as well as shorter distance walked in the 6MWT%pred (p = 0.046) than subjects who failed to achieve it. In addition, they had higher scores in the BODE index and mMRC (p = 0.02), as well as lower LCADL scores (p = 0.001).

Correlations and predicting model of change in TGlittre and LCADL

The change in TGlittre correlated with the baseline distance in the 6MWT in metres (r = 0.31; p = 0.01) and as a %pred (r = 0.33; p = 0.006), as well as the time to perform the TGlittre in minutes (Figure 1(A)) and as a %pred (r = −0.61; p < 0.001). The change in LCADL correlated with baseline FVC in litres (r = 0.33; p = 0.006), CAT score (r = –0.28; p = 0.02) and LCADL (Figure 1(B)).The remaining baseline variables in the predicting model for the change in TGlittre (Constant: 3.46, 95%CI = 1.57 to 5.36; R²: 0.57; p < 0.001) were: TGlittre as a %pred (β: –0.02; 95%CI = –0.02 to –0.01) and distance in the 6MWT as a %pred (β: –0.02; 95%CI = –0.04 to –0.01). For the LCADL change (constant: 12.5, 95%CI = 6.97 to 18.1; R²: 0.34; p < 0.001), only the baseline LCADL remained in the predictor model (β: –0.44; 95%CI = –0.59 to –0.29).

Figure 1. (A) Correlation between Glittre-ADL test (TGlittre) pre-exercise training program (ETP) and change in TGlittre (post-ETP—pre-ETP); (B) correlation between London Chest Activity of Daily Living scale (LCADL) pre-ETP and change in LCADL (post-ETP—pre-ETP).

Cut-off point to identify subjects who achieved the MID of TGlittre, LCADL, and ADLs

The ROC curve indicated cut-off points in the baseline TGlittre of 3.7 min (Figure 2(A)) and 125% pred (Figure 2(B)) to discriminate subjects who achieved the MID of TGlittre. Seven of the 37 subjects with baseline TGlittre <3.7-min achieved the MID of TGlittre (Figure 3(A)). The analysis demonstrated a cut-off point of 32% in baseline LCADL that discriminates subjects who achieved the MID of LCADL (Figure 2(C)) and subjects who achieved the MID of ADLs (Figure 2(D). Two subjects with baseline LCADL score <32% achieved the MID of LCADL (Figure 3(B)) and one subjects achieved the MID of ADLs (Figure 3(C)).

Figure 2. Receiver operating characteristic curve for the: (a) Glittre-ADL test (TGlittre) in minutes cut-off point to identify the subjects with COPD who achieved the minimal important difference (MID) of TGlittre, sensitivity= 76%, specificity= 73%, area under the ROC curve (AUC)= 0.77 (95%CI: 0.65 to 0.89), p < 0.001; (B) TGlittre in percentage of predicted cut-off point to identify the subjects with COPD who achieved the MID of TGlittre, sensitivity= 62%, specificity= 70%, AUC= 0.67 (95%CI: 0.54 to 0.80), p = 0.010; (C) London Chest Activity of Daily Living scale (LCADL) cut-off point to identify the subjects with COPD who achieved the MID of LCADL, sensitivity= 85%, specificity= 70%, AUC= 0.81 (95%CI: 0.71 to 0.92), p < 0.001; [D] LCADL cut-off point to identify the subjects with COPD who achieved the MID of Activities of Daily Living (ADLs – MID of TGlittre and LCADL), sensitivity= 93%, specificity= 60%, AUC= 0.78 (95%CI: 0.66 to 0.90), p < 0.001.

Figure 3. Comparison of pre- (open circles) and post- (closed circles) exercise training program: (A) Glittre-ADL test (TGlittre) in minutes of total sample, subjects who achieved the minimal important difference (AMID) and who failed to achieve the minimal important difference (NAMID) of TGlittre; (B) London Chest Activity of Daily Living scale (LCADL) in %total of total sample, subjects who achieved the minimal important difference (AMID) and who failed to achieve the minimal important difference (NAMID) of LCADL; and (C) LCADL of total sample, subjects who achieved the minimal important difference (AMID) and who failed to achieve the minimal important difference (NAMID) of LCADL. The horizontal line highlights the cut-off point of TGlittre and LCADL.

Association between achieving the MID of TGlittre, LCADL and baseline characteristics

In the raw analysis, it was observed that subjects with TGlittre <3.7 min were more likely to achieve MID. There was also a tendency for an association between lower FEV₁/FVC and achieving the MID of TGlittre. In the adjusted analysis, only the baseline TGlittre was a predicting variable. Separately, female subjects with CAT score <18 and baseline LCADL <32% were more likely to achieve the MID of LCADL. In the adjusted analysis, only the baseline LCADL was able to predict the achievement of the MID of LCADL (Table 2). No logistic regression analysis was performed for the MID of ADLs due to the low prevalence of this outcome in the sample (n = 15; 22.4%).

Table 2. Binary logistic regression: predicting factors of achieving of MID of TGlittre and MID of LCADL.

Baseline variables (RC)	Odds ratio (95%CI)	Adjusted odds ratio^a (95%CI)
MID of TGlittre
FEV1/FVC^b	0.03 (0.00 to 1.17)	0.01 (0.00 to 5.21)
TGlittre (< 3.7 min)	7.42 (2.50 to 22.0)	6.92 (2.29 to 20.9)
MID of LCADL
Sex (M)	5.10 (1.77 to 14.7)	4.97 (0.70 to 35.1)
FVC, L^b	0.55 (0.29 to 1.06)	1.22 (0.40 to 3.71)
CAT (<18)	2.97 (1.07 to 8.26)	1.01 (0.22 to 4.81)
LCADL (<32%)	13.4 (3.81 to 47.2)	12.3 (2.50 to 60.7)

^aContinuous variable.

^b Adjusted by variables that presented p < 0.1 in isolated analysis.

RC: reference category; 95%CI: 95% confidence interval; MID: minimal important difference; TGlittre: Glittre-ADL test; LCADL: London Chest Activity of Daily Living scale; FEV1: forced expiratory volume in one second; FVC: forced vital capacity; L: litres; CAT: COPD Assessment Test.

Discussion

Based on the current results, subjects who achieved the MID of TGlittre were those with worse baseline performance in TGlittre and subjects with TGlittre ≥3.7 min were 6.92 times more likely to achieve the MID of TGlittre. Subjects who achieved the MID of LCADL were those who were younger, with worse lung function, worse health status and more limitation on baseline LCADL; and subjects with LCADL ≥32% were 12.3 times more likely to achieve the MID of LCADL. Additionally, subjects who achieved the MID for both instruments (TGlittre and LCADL) had worse lung function, dyspnoea, health status, functional capacity, as well as increased mortality index and limitation of ADLs evaluated by LCADL. The cut-off point of 32% in the baseline LCADL was also able to discriminate subjects who achieved the MID of ADLs.

Previous studies had shown that more compromised patients benefit more from PRP [15–21], corroborating our results regarding the improvement in ADLs. For all baseline characteristics that differed, the mean difference between the patients who achieved and failed to achieve the MID of ADLs was above the MID of the mMRC, CAT, TGlittre and LCADL [8,14,19,37]. Although there was no statistical difference in the 6MWT distance in metres between groups, there was a difference in the 6MWT %pred, and patients who failed to achieve the MID of ADLs performed ≥80%pred, while patients who achieved the MID performed <80%pred. This shows that patients who improve their ADLs are more symptomatic and have greater functional and health impairment in a clinically relevant manner. It can be inferred that, although more limited patients may have lower exercise tolerance and difficulty in intensity progression, short training periods may be sufficient to promote improvements [16]. While patients who are closer to their predicted maximum may need longer intervention periods.

A contradictory result of our study was the older age in the group which failed to achieve the MID of LCADL. It is expected that patients become more dependent on their ADLs with increasing age [38]. Nevertheless, they had a better baseline LCADL score when compared to those who achieved the MID.

In our study, a cut-off point of 3.7 min in baseline TGlittre was able to discriminate subjects who improved on the test after exercise training, just as a cut-off point of 32% in baseline LCADL was able to discriminate subjects who improved the scale score. Additionally, only the cut-off point for LCADL was able to discriminate who improves ADLs considering the concomitant achievement of MID from an objective instrument of functional capacity assessment and an instrument of perceived/reported limitation. It is important to point out that 48 (71.6%) subjects responded “0” for at least one item of the LCADL at baseline, and therefore had lower potential for improvement in scale (68.8% of those subjects failed to achieve the MDI). Furthermore, from those 48 participants, 31 (64.6%) presented functional limitation characterised by the time spent to perform the TGlittre ≥3.5 min at baseline. Yet, no associations between responding “0” for at least one item of LCADL and presenting functional limitation assessed by the TGlittre was found (data not shown).

The cut-off point found for TGlittre in the present study was close to the 3.5-min cut-off point, which is used to discriminate the functional capacity of COPD subjects [39]. Similarly, the cut-off point found for LCADL is a value within a range that discriminates COPD prognosis [40] and the one that discriminates patients' functional status [41]. This is an interesting finding, as clinicians can rely on these cut-off points to identify the most functional impaired subjects and those with the least potential for improvement to a conventional exercise training. Thus, discriminatory ability is added to TGlittre and LCADL as tools for assessing functional status in PRP. The use of the 125%pred cut-off point for baseline TGlittre is not recommended, as it did not present satisfactory discriminatory capacity and there was no statistical significance for this analysis.

The baseline values of the instruments themselves were the best predictors of their MID achieving from all variables. This was expected since the bigger the interval between the actual patient performance and the maximum possible performance/score, the greater the chance of variation after an intervention, increasing the responsiveness. Therefore, the cut-off points showed in the present study can be considered not only to identify who are the subjects with the greatest potential for improvement in their ADLs, but also to identify in which subjects the instrument may not be responsive. For example, if the subject has a baseline TGlittre time <3.7 min, a more “demanding” test may be required, while for subjects with a baseline LCADL score <32%, a scale that includes more complex ADLs or evaluates another limiting symptom besides dyspnoea may be more responsive. On the other hand, our results may also suggest the need to consider different MID values for the same instrument, taking into account the patient's initial performance. Since the less limited patients tend to show lower improvement after a PRP, the MID for LCADL and TGlittre might be lower, while in more severe, higher values of the MID’s may be more appropriate. Although MID is defined as the smallest change needed to reflect clinical improvement [35], its calculation usually does not consider the different patient profiles.

Interestingly, in the raw analysis, females were 5.1 times more likely to achieve the MID of LCADL than males. Brazilian men often report score 0 for many items in the LCADL household activities domain [12,40,41], which represents activities that the subject does not perform because they consider it irrelevant or never needed to do it [12]. This seems to overestimate the scale results for men [40]. The percentage score of the LCADL scale is an alternative to the total score seeking to minimise this bias [12]. However, since the number of items the patient can improve after PRP is limited with the many "score 0" (activities the patient does not perform), the scale may become less sensitive to detect improvement after an intervention, even by means of the score as a percentage of the total.

A possible study limitation is that for the adjusted regression analysis of the MID of LCADL, the sample may have been insufficient. However, based on the criterion of 10 individuals for each independent variable, our sample was very close to the ideal.

Up to our knowledge, this was the first study to show that worse lung function, higher dyspnoea, poorer health status, higher risk of death, and greater limitation in baseline ADLs are associated with higher ADLs gains post-exercise training in subjects with COPD.

Conclusions

Subjects who significantly improve ADLs are those with worse lung function, higher dyspnoea, greater impact on health status, higher mortality index, and greater limitation on baseline ADLs. Greater baseline functional impairment increases the chance of achieving the MID of ADLs limitation assessment instruments. The cut-off points of 3.7 min for baseline TGlittre and 32% for baseline LCADL can help to identify subjects with better response to exercise training in terms of ADLs limitation.

Acknowledgements

The authors thank those who supported the data acquisition, to the Irmandade da Santa Casa de Misericórdia de Porto Alegre, to all the pulmonologists who referred their patients to the study protocol, to the Coordenação de Aperfeiçoamento de Pessoal de Ensino Superior (CAPES/Brazil) and Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq/Brazil—Chamada Universal—MCTI/CNPq N° 14/2014) and to the Fundação de Amparo à Pesquisa e Inovação do Estado de Santa Catarina (FAPESC/Brazil—PAP UDESC—Chamada Publica N° 01/2016—Termo de Outorga: 2017TR645).

Disclosure statement

The authors report no conflict of interest.

Additional information

Funding

This research was supported in part by the Coordenação de Aperfeiçoamento de Pessoal de Ensino Superior—CAPES/Brazil and the Conselho Nacional de Desenvolvimento Científico e Tecnológico—CNPq/Brazil (Chamada Universal—MCTI/CNPq N° 14/2014) and the Fundação de Amparo à Pesquisa e Inovação do Estado de Santa Catarina—FAPESC/Brazil (PAP UDESC—Chamada Publica N° 01/2016—Termo de Outorga: 2017TR645).