Early Changes in eDiary COPD Symptoms Predict Clinically Relevant Treatment Response at 12 Weeks: Analysis from the CRYSTAL Study

ABSTRACT

Early detection of treatment response is important in the long-term treatment and management of patients with chronic obstructive pulmonary disease (COPD). This analysis evaluated whether early improvement in symptoms, recorded in the first 7 or 14 days via an electronic diary (eDiary) compared with baseline, can predict clinically meaningful treatment responders at 12 weeks.

CRYSTAL was a 12-week, randomized, open-label study that demonstrated the increased effectiveness of indacaterol/glycopyrronium (IND/GLY) or glycopyrronium (GLY), after a direct switch from on-going baseline therapies, in patients with symptomatic COPD and moderate airflow obstruction. The co-primary endpoints were trough forced expiratory volume in 1 second (FEV₁) and transition dyspnea index (TDI) at Week 12. Patients' symptom status was recorded daily in an eDiary. Of 4,389 patients randomized, 3,936 and 3,855 reported symptoms on Days 7 and 14, respectively. Patients who reported an early decrease in symptoms on Day 7 or 14 were more likely to achieve the minimal clinically important difference of ≥100 mL in trough FEV₁ or ≥ 1 point in TDI at Week 12. Using stepwise multivariate regression models we identified as best predictors of FEV₁ responders the decrease in wheeze on Day 7, and nighttime symptoms and wheeze on Day 14; best predictors of TDI responders were decrease in nighttime symptoms and wheeze on Day 7, and nighttime symptoms, sputum and wheeze on Day 14. Early symptom improvement at Day 7 or 14, especially wheeze and nighttime symptoms, may identify patients with clinically important improvement in lung function and dyspnea at Week 12.

KEYWORDS:

• Chronic obstructive pulmonary disease
• electronic diary
• eDiary
• forced expiratory volume in 1 second
• transition dyspnea index
• early bronchodilator response
• CRYSTAL

View correction statement:

Correction

Introduction

Chronic obstructive pulmonary disease (COPD) is a complex, heterogeneous, and progressive lung disease characterized by persistent respiratory symptoms and airflow limitation; it is associated with deterioration in quality of life, exacerbations, hospital admissions and death (1). Respiratory symptoms characteristic of COPD include dyspnea, cough, wheezing, and sputum production and these have significant adverse effects on patient functioning and quality of life (1).

Prevention and improvement of symptoms are critical in the management of COPD (2), which can be a challenge as many symptoms are minimized by patients (3) and there is significant disparity between physicians' and patients' perception of symptoms and severity of disease (4). Spirometric measures, though useful for diagnosis and evaluating changes in lung function, do not capture variability in symptom severity and patients' overall status (5). Health-status measures such as the Saint George's Respiratory Questionnaire (SGRQ) (6, 7), Clinical COPD Questionnaire (CCQ) (8), and COPD Assessment Test (CAT) (9, 10) are routinely used symptom assessment tools in clinical trials; however, these symptom questionnaires are completed retrospectively by patients at varied intervals and can be associated with poor recall and sensitivity of response (3). Directly recording patient-reported outcome (PRO) measures on a daily basis may provide greater precision in the assessment of day-to-day symptoms and severity than existing COPD-specific health status questionnaires (11). Daily symptom reporting would also minimize the bias associated with answering a lengthy questionnaire less frequently (weekly or monthly) as this relies on patient recollection. Hence, PRO instruments such as an electronic diary (eDiary) can be used as effectiveness endpoints in clinical trials to help elucidate the impact of an intervention on one or more aspects of patients' health status (12, 13). A daily eDiary is an effective PRO tool that can record daily scores and reduce follow-up bias. The same eDiary was used in the SHINE study and was validated for its precision in reporting symptoms (13).

Early detection of efficacy is^{Table 1}important in deciding treatment choices for COPD patients; however, acute bronchodilator responsiveness cannot accurately predict the long-term effects of long-acting bronchodilators (14, 15). The objective of this analysis of the CRYSTAL study was to evaluate whether early changes (reported at 7 or 14 days) in symptoms recorded via eDiary can predict clinically meaningful treatment responses at 12 weeks in lung function (trough forced expiratory volume in one second [FEV₁]) and dyspnea (transition dyspnea index [TDI]).

Methods

Study design

This was a post-hoc analysis of the CRYSTAL study (ClinicalTrials.gov number: NCT01985334), a 12-week, prospective, open-label, pragmatic study that was designed to evaluate the efficacy of a direct switch to indacaterol/glycopyrronium (IND/GLY) or glycopyrronium (GLY) from prior treatment in patients with moderate COPD (16). Patients were enrolled from hospital and primary care settings and randomized to treatment based on their previous treatment and modified Medical Research Council (mMRC) score. The two co-primary endpoints for all study groups were improvements in trough FEV₁ and TDI at Week 12 (Figure S1). The study was approved by the ethics committee or institutional review board at each site and was done in accordance with the Declaration of Helsinki, International Conference on Harmonization Good Clinical Practice, and applicable local regulations.

Patients

Patients aged ≥40 years were included if they had a clinical diagnosis of moderate COPD with a history of ≤1 moderate-to-severe exacerbation in the previous year, smoking history of ≥10 pack-years, mMRC score ≥1, moderate airflow limitation with FEV₁ ≥50% and <80% predicted, and FEV₁/forced vital capacity (FVC) ratio <0.7. Enrolled patients received baseline treatment for at least 3 months before screening. At randomization, patients were either switched to GLY from short-acting β₂-agonist (SABA) and/or short-acting muscarinic antagonist (SAMA) or long-acting β₂-agonist (LABA) or long-acting muscarinic antagonist (LAMA) (and mMRC score ≥1); or were switched to IND/GLY from LABA + inhaled corticosteroid (ICS) or from a LABA or LAMA (and mMRC score ≥2).

Assessment of symptoms using eDiary

In this analysis, the data from the intention-to-treat (ITT) population were used with particular focus on the seven questions related to respiratory symptoms from the 12-question eDiary. Symptoms were scored on a 0–3 (defined in table 1) rating scale. The questionnaire was completed each evening during the study. Patients reaching the minimal clinically important difference (MCID) of ≥100 mL change from baseline for trough FEV₁ were considered FEV₁ responders and MCID of ≥1 unit for TDI were considered TDI responders. The respiratory symptom scores assessed by the eDiary at Days 0, 7, and 14 were evaluated. A 1-point reduction in symptom score (night time respiratory symptoms, activity impairment, cough, wheeze, sputum production, sputum color and breathlessness) on Day 7 or 14 compared with Day 0 reported by any individual patient was considered as an early symptom improvement. The eDiary provides the option to the study participants to rate the individual symptoms using the scores 0, 1, 2, and 3. Since we decided to evaluate the impact of a change in symptoms in an individual patient level, the minimal change that could be evaluated was 1 point. Information on rescue medication use and treatment compliance were collected via the first two questions (EQ1 and EQ2); the subsequent ten questions collected information about the major (EQ3 to EQ9 were used in this analysis) and minor (EQ10 to EQ12) respiratory symptoms (Table 1).

Table 1. Evaluated questions (EQ3 to EQ9) of an electronic diary symptom questionnaire.

Evening Questions	Possible Answers
EQ1: Number of puffs of rescue medication during the past 24 hours:	0–20 puffs	Not evaluated in this analysis
EQ2: Did you take the study medication as prescribed by your study doctor today?	Yes/No
EQ3: How would you rate your respiratory symptoms last night?	0 = No waking due to symptoms
	1 = Woke up once due to symptoms
	2 = Woke up more than once due to symptoms
	3 = Woke up frequently or could not sleep due to symptoms
EQ4: Did your respiratory symptoms stop you performing your usual daily activities today?	0 = Not at all
	1 = A little
	2 = Quite a lot
	3 = Completely
EQ5: How was your cough during the past 24 hours?	0 = None	Evaluated in this analysis Major symptoms
	1 = Mild
	2 = Moderate
	3 = Severe
EQ6: How was your wheeze during the past 24 hours?	0 = None
	1 = Mild
	2 = Moderate
	3 = Severe
EQ7: How much sputum did you produce the past 24 hours?	0 = None
	1 = Less than 5 mL (1 teaspoon)
	2 = Between 5 and 25 mL (1–5 teaspoons)
	3 = More than 25 mL (5 teaspoons)
EQ8: What color was the sputum you produced during the past 24 hours?	0 = None
	1 = White-grey
	2 = Yellow
	3 = Green
EQ9: During what activities did you first feel breathless in the last 24 hours?	0 = Never or only when running
	1 = When walking uphill or upstairs
	2 = When walking on flat ground
	3 = At rest
EQ10: Did you have a sore throat during the past 24 hours?	0 = No
	1 = Yes, mild
	2 = Yes, moderate	Not evaluated in this analysis Minor symptoms
	3 = Yes, severe
EQ11: Did you have a cold (nasal discharge and/or nasal congestion) during the past 24 hours?	0 = No
	1 = Yes, mild
	2 = Yes, moderate
	3 = Yes, severe
EQ12: Did you have a fever during the past 24 hours?	0 = No
	1 = Yes, mild
	2 = Yes, moderate
	3 = Yes, severe

eDiary, electronic diary.

Statistical analysis

The data were summarized by treatment responses in lung function and TDI. The changes in eDiary symptom scores from Day 0 to Day 7 or 14 in responders and non-responders were analyzed as continuous variables and assessed using the Wilcoxon signed-rank test. Changes in symptom scores between Day 0 and Day 7 or 14 were also dichotomized according to whether the symptom decreased (an “early decrease”), or did not decrease (either no change or increase in symptom score). A logistic regression model was used to investigate the association between the patients with an MCID (response variable) with respect to the early decrease in respiratory symptoms, controlling for the effects of age, sex, baseline through FEV₁, baseline dyspnea index and current smoking status. The odds ratio for responders vs. non-responders for trough FEV₁ and TDI were also presented graphically. A logistic regression analysis with backward stepwise elimination was performed to select the most important symptom changes (at Days 7 and 14 separately) that could predict trough FEV₁ or TDI responders at Week 12. The changes in symptoms were included in the model as quantitative variables and the odds ratios of responders (ORs) were estimated per unit decrease in the score.

Results

Of 4,389 randomized patients, 4,324 were in the ITT population, and 3,936 and 3,855 reported symptoms via eDiary on Days 7 and 14, respectively. Patients' demographics and baseline characteristics of the ITT population are shown in Table 2.

Table 2. Patients' demographics and baseline characteristics.

Characteristics	Value (N = 4324)
Age, years	64.6 ± 8.30
Men, n (%)	2908 (67.3)
Body mass index, kg/m^2	27.6 ± 5.0
Current smokers, n (%)	2292 (53.0)
Time since COPD diagnosis, years	6.1 ± 5.5
Post-bronchodilator FEV1, L	1.85 ± 0.49
Post-bronchodilator FEV1, % predicted	64.7 ± 8.70
Post-bronchodilator FEV1/FVC	0.57 ± 0.08
Dyspnea—mMRC grade, n (%)
1	2591 (59.9)
≥2	1693 (39.1)
Number of exacerbations in the previous 12 months, n (%)
1	867 (20.1)
≥2	16 (0.4)
Baseline Treatment, n (%)
SABA, SAMA, SABA+SAMA	496 (11.5)
LABA + ICS (fixed dose or free combination)	1059 (24.5)
LABA (only monotherapy)	1193 (27.6)
LAMA (only monotherapy)	1512 (35.0)
Others/NA/Unknown^*	184 (4.3)

Data presented as mean ± SD unless stated otherwise.

^* LAMA + ICS, LABA + LAMA, ICS monotherapy, LABA + LAMA + ICS (triple), systemic corticosteroids, methylxanthines, roflumilast and others.

FEV₁, forced expiratory volume in 1 second; FVC, forced vital capacity; ICS, inhaled corticosteroid; IND/GLY, indacaterol/glycopyrronium; LABA, long-acting β₂-agonist; LAMA, long-acting muscarinic antagonist; mMRC, modified Medical Research Council; NA, not available; SABA, short-acting β₂-agonist; SAMA, short-acting muscarinic antagonist.

Symptom predictors for FEV₁ responders

FEV₁ responders, compared with non-responders, demonstrated statistically significant differences in the change from Day 0 in: nighttime respiratory symptoms at Day 14 (P = 0.0013); wheeze at Days 7 (P = 0.0005) and 14 (P = 0.0039); and sputum at Days 7 (P = 0.0345) and 14 (P = 0.0146) (Table 3).

Table 3. Mean changes in symptoms on Days 7 and 14 in patients with and without a MCID in trough FEV₁ (≥100 mL; responders and non-responders) at Week 12.

	Responders^*(n = 1680)	Nonresponders (n = 2421)	P-value
Night-time respiratory symptoms
Day7−Day0	−0.042 ± 0.61	−0.022 ± 0.61	0.1664
Day14−Day0	−0.084 ± 0.63	−0.026 ± 0.62	0.0013
Activity impairment
Day7−Day0	−0.056 ± 0.62	−0.016 ± 0.57	0.0613
Day14−Day0	−0.045 ± 0.59	−0.010 ± 0.59	0.1161
Cough
Day7−Day0	−0.041 ± 0.67	−0.041 ± 0.66	0.9606
Day14−Day0	−0.096 ± 0.67	−0.083 ± 0.65	0.5513
Wheeze
Day7−Day0	−0.135 ± 0.62	−0.066 ± 0.59	0.0005
Day14−Day0	−0.129 ± 0.62	−0.070 ± 0.62	0.0039
Sputum
Day7−Day0	−0.055 ± 0.59	−0.022 ± 0.56	0.0345
Day14−Day0	−0.086 ± 0.59	−0.045 ± 0.60	0.0146
Sputum color
Day7−Day0	−0.011 ± 0.49	−0.016 ± 0.50	0.7901
Day14−Day0	−0.034 ± 0.50	−0.027 ± 0.53	0.4990
Breathlessness
Day7−Day0	−0.064 ± 0.64	−0.043 ± 0.59	0.3600
Day14−Day0	−0.077 ± 0.63	−0.041 ± 0.62	0.1066

Data presented as mean ± SD.

P-values indicate statistically significant difference.

^* Responders: patients achieved MCID in trough FEV₁ (≥100 mL increase) at Week 12.

FEV₁, forced expiratory volume in 1 second; MCID, minimal clinically important difference.

Improvements in^{Figure 1}nighttime respiratory symptoms on Day 14 and in activity on Day 7 were associated with trough FEV₁ response at Week 12; improvements in wheeze, sputum and breathlessness on Days 7 and 14 were associated with trough FEV₁ response at Week 12 (Figure 1).

Figure 1. FEV₁ responders at Week 12 as predicted by the improvement in individual symptoms on Days 7 and 14. The odds ratio and 95% confidence interval (CI) presented are adjusted for age, sex, baseline, baseline dyspnoea index and current smoking status. FEV₁, forced expiratory volume in 1 second.

Based on the results of the multivariate stepwise regression analysis, we have shown that among all symptoms the best predictors of FEV₁ response were: improvements in wheeze on Day 7 (OR 1.21, 95% CI 1.08–1.33; P = 0.0006) and Day 14 (OR 1.13, 95% CI 1.01–1.26; P = 0.0324); and nighttime respiratory symptoms on Day 14 (OR 1.13 95% CI 1.01–1.26; P = 0.0339).

Symptom predictors of TDI responders

TDI responders, compared with non-responders, presented with statistically significant differences in the change from Day 0 in: nighttime respiratory symptoms at Days 7 (P < 0.0001) and 14 (P < 0.0001); activity impairment at Days 7 (P = 0.0005) and 14 (P < 0.0001); cough at Day 14 (P = 0.0229); wheeze at Days 7 (P < 0.0001) and 14 (P < 0.0001); sputum at Day 14 (P = 0.0011); sputum color at Day 14 (P = 0.00110); and breathlessness at Days 7 (P = 0.0354) and 14 (P = 0.0239; Table 4). Improvements in most symptoms on Days 7 and 14 were associated with TDI response at Week 12 (Figure 2).

Table 4. Mean changes in symptoms on Days 7 and 14 in patients with and without a MCID in TDI total score (≥1 unit; responders and non-responders) at Week 12.

	Responders^*(n = 2008)	Nonresponders (n = 1994)	P-value
Night-time respiratory symptoms
Day7−Day0	−0.070 ± 0.62	0.008 ± 0.59	<0.0001
Day14−Day0	−0.098 ± 0.63	−0.001 ± 0.60	<0.0001
Activity impairment
Day7−Day0	−0.065 ± 0.60	0.002 ± 0.59	0.0005
Day14−Day0	−0.062 ± 0.61	0.018 ± 0.58	<0.0001
Cough
Day7−Day0	−0.060 ± 0.66	−0.026 ± 0.66	0.1093
Day14−Day0	−0.110 ± 0.66	−0.062 ± 0.66	0.0229
Wheeze
Day7−Day0	−0.143 ± 0.60	−0.044 ± 0.59	<0.0001
Day14−Day0	−0.147 ± 0.64	−0.040 ± 0.59	<0.0001
Sputum
Day7−Day0	−0.045 ± 0.58	−0.021 ± 0.55	0.2269
Day14−Day0	−0.089 ± 0.60	−0.027 ± 0.58	0.0011
Sputum color
Day7−Day0	−0.028 ± 0.51	−0.002 ± 0.48	0.1373
Day14−Day0	−0.057 ± 0.51	0.000 ± 0.52	0.0011
Breathlessness
Day7−Day0	−0.071 ± 0.61	−0.035 ± 0.62	0.0354
-Day14−Day0	−0.074 ± 0.64	−0.033 ± 0.62	0.0239

Data presented as mean ± SD.

P-values indicate statistically significant difference.

^* Responders: patients achieved MCID in TDI total score (≥1 unit increase) at Week 12.

MCID, minimal clinically important difference; TDI, transition dyspnea index.

Figure 2. TDI responders at Week 12 as predicted by the improvement in individual symptoms on Days 7 and 14. The odds ratio and 95% confidence interval (CI) presented are adjusted for age, sex, baseline trough forced expiratory volume in 1 second (FEV₁) and current smoking status. TDI, transition dyspnea index.

Based on the results of the multivariate stepwise regression analysis, we have shown that among all symptoms the best predictors of TDI responders were: improvements in nighttime respiratory symptoms on Day 7 (OR 1.17, 95% CI 1.05–1.30; P = 0.0062) and Day 14 (OR 1.20, 95% CI 1.07–1.34; P = 0.0013); wheeze on Day 7 (OR 1.27, 95% CI 1.14–1.43; P <0.0001) and Day 14 (OR 1.25, 95% CI 1.12–1.40; P < 0.0001); and sputum color on Day 14 (OR 1.18, 95% CI 1.03–1.34; P = 0.0143).

Discussion

The findings of this analysis of the CRYSTAL study showed that improvements in symptoms as reported in the eDiary on Days 7 and 14 may predict lung function and TDI responders at Week 12 in patients with moderate airflow obstruction. Early improvements in wheeze and nighttime respiratory symptoms in the first 7 or 14 days of treatment were the best predictors of patients being FEV₁ responders; similarly, early improvements in nighttime symptoms, wheeze and sputum were the best predictors of TDI responders at Week 12. This is the first analysis, to our knowledge, that has shown the predictive value of early symptom improvements in an eDiary for the mid-term response to long-acting bronchodilators.

As per GOLD recommendations, improving airflow obstruction and symptom control are key goals in COPD management (1). Both lung function and dyspnea are important and routinely used COPD measures in the clinical assessment of disease progression (17, 18). FEV₁—a spirometric measure—is a key parameter to determine severity of airflow obstruction and it is weakly correlated with symptoms (19). Symptoms are directly reported by patients and are recognized as important PROs (13). Breathlessness is a key symptom of COPD and is known to have serious impact on health status and the ability to perform everyday activities (20). Early indicators of treatment response in lung function and in dyspnea may help to optimize the treatment choices in COPD management (2, 19).

An important observation of the present analysis is that improvements in wheeze, nighttime respiratory symptoms and sputum were noted by Day 7 or 14 in patients who were FEV₁ responders at Week 12. Similarly, significant improvements from baseline in nighttime respiratory symptoms, cough, wheeze, sputum, sputum color and breathlessness were observed by Days 7 or 14 of treatment in TDI responders versus non-responders. These improvements in symptoms, observed as early as Day 7 or 14 of treatment, support previous reports showing early treatment benefit with long-acting bronchodilators, such as GLY or IND/GLY, on lung function and health status within days or weeks of the start of treatment (15). In this analysis, we were also able to show that the patients who will have clinically meaningful improvements in lung function and/or dyspnea in the mid-term (at Week 12) are more likely to also have an early response to COPD symptoms. An interesting association is the improvement in cough and nighttime symptoms and later FEV₁ and TDI response. A plausible speculation would be that this may be related to the introduction of a LAMA in patients without the previous use of such an agent, due to the beneficial effects of LAMA on sputum production (21) and sleep quality (22). This would be an interesting concept to explore in future studies.

Recently, the use of eDiary has been increasingly recognized in clinical trials, especially in progressive diseases such as asthma and COPD, where treatment outcomes are relevant to the individual patient's experience and reported outcomes. In these diseases, it has been found to be a valid, reliable and responsive tool that can also alert the patient and/or physician of an exacerbation (11, 13). Moreover, responses to the simple questions contained in the eDiary regarding nighttime symptoms, wheeze etc. after only one or two weeks of starting or changing therapy can predict long-term treatment response.

Electronic PROs (ePROs) are gaining regulatory and scientific support as a preferred method of data capture in clinical trials. The results from the 26-week SHINE study (Phase III clinical trial on IND/GLY) show that the same eDiary (an 11-point rating scale) as that used in the present study was a useful tool for identifying relevant symptoms, particularly dyspnea which was correlated with activity limitation (13). The only difference was in the frequency of reporting; in the SHINE study, the eDiary was used twice daily to record morning and nighttime symptoms, whereas, in the CRYSTAL study, patients only recorded their symptoms once daily at nighttime.

A post-hoc analysis from the SHINE study has shown that there is an association between TDI and eDiary scores; a decrease in ≥1 point in TDI score was found to be related to ≤ 0.6 point decrease in eDiary score and both considered to be clinically meaningful changes (13). Furthermore, the eDiary was found to be more sensitive to changes in symptoms than the SGRQ, giving a much better differentiation between improved and non-improved patients (13). This was also observed in our analysis where the eDiary showed robust sensitivity in identifying changes in symptoms in the population that were mildly symptomatic (mMRC = 1). In the present analysis we have not evaluated the overall eDiary score, but its individual components and, therefore, we are reporting the association between individual symptom/question changes of at least 1 point and the clinically meaningful changes in lung function and dyspnea at 12 weeks. The relatively small numbers of patients with ≥1 point improvement and the small range of the eDiary (0–3) do not allow us for additional analyses (e.g. by 2-point improvement). Nevertheless, low rates of symptom changes were observed at both Day 7 and Day 14 of treatment, which could be attributed to the sensitivity of the scale of the eDiary used in this analysis, which was a 3-point rating scale that required a pronounced change in symptoms to bring about just a single point shift in the score.

A separate study using the Exacerbations of Chronic Pulmonary Disease Tool (EXACT)-Respiratory Symptoms (E–RS) eDiary tool was found to be effective in quantitatively analyzing breathlessness, cough and sputum, and chest symptoms (11). Comparative analyses between different tools will clarify the differences in sensitivity between them and the exact utility of each of these tools for symptom reporting. Finally, the use of eDiaries was also found to be useful in other disease areas, such as angina and diabetes (23, 24), suggesting a potential role for such approaches in telehealth. The daily reporting of respiratory symptoms via an eDiary provides real-time data on treatment response, which may reduce the possible bias associated with other periodic symptom assessment tools that have a greater reliance on patient recall. An interesting observation of the current analysis is that early changes in eDiary questions seem to be related to changes in TDI and, since the eDiary consists of simple questions, this may provide a potential application in clinical practice that needs to be further explored. Our analysis has potential limitations. The results are limited to patients with moderate airflow limitation, as this was the population included in the CRYSTAL study. Secondly, we have not explored whether the patient responses would be different in specific subgroups of patients, according to baseline characteristics; however, we have included, in the statistical models of the analysis, all the potential clinically relevant confounding factors. Finally, the eDiary contains a list of simple and easy-to-understand questions for reporting symptoms and observation of daily symptom variability would help the treating physicians to understand treatment response and patients' overall health status. While the CRYSTAL study data provide close to real-life validation of the eDiary to predict treatment response based on change in daily symptoms, multivariate regression analysis of predictors of FEV₁ response found the symptoms evaluated to only be modest predictors.

Conclusions

This analysis of eDiary data from the open-label CRYSTAL study showed that early symptom improvement at Day 7 or 14 may identify patients with clinically important improvement in lung function and dyspnea at Week 12. Patients showing early improvement in symptoms such as wheeze, night-time respiratory symptoms, and sputum are likely to respond better to long-acting bronchodilator treatment. These results highlight the importance of the evaluation of symptomatic responses in the days soon after a treatment change as these may provide important insights and improve disease management in the longer term. Further prospective studies are needed to identify how similar outcomes may be applied in clinical practice.

Declaration of interest statement

Konstantinos Kostikas (KK) is an employee and shareholder of Novartis Pharma AG. KK has previously received honoraria for speeches and consulting services from AstraZeneca, Chiesi, ELPEN and Takeda and received honoraria for speeches from Boehringer Ingelheim outside the submitted work. Maryam Aalamian-Mattheis (MA-M) is an employee of Novartis Pharma AG. Veronica Anna Pagano (VAP) and Xavier Nunez (XN) were statisticians for this subgroup analysis of CRYSTAL study and work at TFS. Robert Fogel (RF) is an employee and shareholder of Novartis Pharmaceutical Corporation. Francesco Patalano (FP) and Andreas Clemens (AC) are employees and shareholders of Novartis Pharma AG.

Acknowledgments

All authors have contributed substantially in data interpretation, development and revision of manuscript draft, who provided their consent and approval for publishing this manuscript. Writing and editorial assistance for this manuscript was provided by Claire Field (Novartis Pharmaceuticals & Novartis Global Service Center, Dublin) and Santanu Bhadra (Novartis Healthcare Pvt. Ltd., India).

Additional information

Funding

The study was supported and funded by Novartis.