A randomised controlled study of the effectiveness of breathing retraining exercises taught by a physiotherapist either by instructional DVD or in face-to-face sessions in the management of asthma in adults

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 09/104/19. The contractual start date was in November 2011. The draft report began editorial review in October 2016 and was accepted for publication in May 2017. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Mike Thomas is a member of the Health Technology Assessment (HTA) Primary Care, Community and Preventive Interventions (PCCPI) Panel. In the last 3 years he has received speaker’s honoraria for speaking at sponsored meetings or satellite symposia at conferences from the following companies, marketing respiratory and allergy products: Aerocrine, GlaxoSmithKline (GSK) and Novartis International AG. He has received honoraria for attending advisory panels with Aerocrine, AstraZeneca, Boehringer Ingelheim, GSK and Novartis. He has received sponsorship to attend international scientific meetings from GSK and AstraZeneca and has received funding for research projects from GSK. He is a member of the British Thoracic Society (BTS)/Scottish Intercollegiate Guidelines Network (SIGN) Asthma Guideline Group and the National Institute for Health and Care Excellence (NICE) Asthma Guideline Group. Paul Little is Editor-in-Chief of the Programme Grants for Applied Research journal and is a member of the National Institute for Health Research (NIHR) Journals Library Board. Lucy Yardley Is a member of the Public Health Research journal Research Funding Board and a member of the HTA Efficient Study Designs Board and reports grants from the NIHR during the conduct of the study. James Raftery is a member of the NIHR Journals Library Editorial Group and the HTA and EME Editorial Board and was previously Director of the Wessex Institute and head of the NIHR Evaluation, Trials and Studies Coordinating Centre (NETSCC). He is also a former member of the NIHR HSDR Research Led Board. Ian Pavord has received speaker’s honoraria for speaking at sponsored meetings in the last 5 years from AstraZeneca, Boehringer Ingelheim, Aerocrine, Almirall Ltd, Novartis and GSK. He has received honoraria for attending advisory panels with Almirall, AstraZeneca, Boehringer Ingelheim, GSK, MSD, Schering-Plough, Novartis, Dey Pharma, Napp Pharmaceuticals and RespiVert Ltd. He has received sponsorship to attend international scientific meetings from Boehringer Ingelheim, GSK, AstraZeneca and Napp. He is Chief Medical Advisor to Asthma UK, a member of the UK Department of Health Asthma Strategy Group, a member of the BTS/SIGN Asthma Guideline Group and joint Editor-in-Chief of Thorax. Neither Ian Pavord nor any member of his family has any shares in pharmaceutical companies. David Price reports other Board Membership (fees paid to Research in Real Life Ltd) from Aerocrine, Almirall, Amgen Inc., AstraZeneca, Boehringer Ingelheim, Chiesi Ltd, Meda, Mundipharma, Napp, Novartis and Teva Pharmaceutical Industries Ltd; consultancy (fees paid to Research in Real Life Ltd) from Almirall, Amgen, AstraZeneca, Boehringer Ingelheim, Chiesi, GSK, Meda, Mundipharma, Napp, Novartis, Pfizer Inc. and Teva; grants from the UK NHS, British Lung Foundation, Aerocrine, AKL Ltd, Almirall, AstraZeneca, Boehringer Ingelheim, Chiesi, Eli Lilly and Co., GSK, Meda, Merck & Co., Inc., Mundipharma, Napp, Novartis, Orion, Pfizer, Respiratory Effectiveness Group, Takeda, Teva and Zentiva; lectures/speaking engagement fees (paid to Research in Real Life Ltd) from Almirall, AstraZeneca, Boehringer Ingelheim, Chiesi, Cipla Ltd, GSK, Kyorin Pharmaceutical Co., Inc., Meda, Merck, Mundipharma, Novartis, Pfizer, Skyepharma, Takeda and Teva; manuscript preparation fees (paid to Research in Real Life Ltd) from Mundipharma and Teva; payment for travel/accommodation/meeting expenses (paid to Research in Real Life Ltd) from Aerocrine, Boehringer Ingelheim, Mundipharma, Napp, Novartis and Teva; funding for patient enrolment or completion of research (paid to Research in Real Life Ltd) from Almirall, Chiesi, Teva and Zentiva; and payment for the development of educational materials (paid to Research in Real Life Ltd) from GSK and Novartis, outside the submitted work. In addition, David Price has an AKL Ltd patent pending and owns shares in AKL Ltd, which produces phytopharmaceuticals. He owns 80% of Research in Real Life Ltd (which is subcontracted by Observational and Pragmatic Research Institute Pte Ltd), 75% of the social enterprise Optimum Patient Care Ltd and 75% of the Observational and Pragmatic Research Institute Pte Ltd. Ratko Djukanovic has received fees for lectures at symposia organised by Novartis and Teva and for consultation for these two companies as a member of advisory boards. He is a co-founder and current consultant, and has shares in, Synairgen, a University of Southampton spin-out company.

Copyright statement

© Queen’s Printer and Controller of HMSO 2017. This work was produced by Thomas et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

Trial design

The BREATHE (Breathing REtraining for Asthma – Trial of Home Exercises) trial was a pragmatic, observer-blinded, three-arm, parallel-group RCT comparing breathing retraining delivered using a DVD with face-to-face physiotherapy and with usual care (control) for adults with asthma and impaired health status.

Participants

We recruited 655 adult patients with a diagnosis of asthma and impaired asthma-related health status from 34 primary care sites (UK NHS general practices) in the Wessex region. Patients were recruited between November 2012 and January 2015, with follow-up ending in February 2016. Practices were recruited and supported by the UK Clinical Research Network (CRN), who also supported patient recruitment and follow-up.

Interventions

Randomised controlled trial processes

We performed a parallel-group, three-arm RCT over a 12-month period to assess the effect of the DVD programme compared with usual care and face-to-face physiotherapist-led training with a similar content on the following parameters: asthma-related health status, parameters of symptomatic and physiological asthma control and asthma-related health resource use. Consenting participants were randomly assigned to (1) receipt of the DVD intervention plus supporting written material, (2) three sessions of face-to-face physiotherapy breathing instruction, consisting of an initial ‘small-group’ (up to five patients) session of approximately 45 minutes and two subsequent individual sessions of up to 45 minutes or (3) usual care, which included recruitment and follow-up assessments but no additional intervention or care.

Outcomes

Prespecified outcome measures were between-group and within-group changes from baseline to the end of the study (12 months). The statistical analysis plan (SAP) is provided below (see Statistical analysis plan).

The primary outcome was the between-group [intention-to-treat (ITT) population] change in asthma-specific health status [AQLQ (short version)] score, adjusted for potential confounders.

Secondary outcome measures were between-group (ITT population) change in Asthma Control Questionnaire (ACQ)25 score, lung function [FEV₁, forced volume vital capacity (FVC), FEV₁-to-FVC ratio, peak expiratory flow rate (PEFR)], fraction of exhaled nitric oxide (FeNO), health status [EuroQol-5 Dimensions (EQ-5D)26], anxiety and depression [Hospital Anxiety and Depression Scale (HADS)27], hyperventilation (Nijmegen) questionnaire28 score, oral corticosteroid courses for asthma exacerbations and bronchodilator use.

Sensitivity analyses included analyses of the primary and secondary outcomes in both the ITT and per-protocol (PP) populations. We also performed a prespecified sensitivity analysis including participants with missing baseline or outcome data for the primary efficacy parameter, the AQLQ, as described below and in the SAP. In addition, health economic and process evaluation analyses were carried out and are presented in Chapters 4 and 6 respectively.

Sample size

Changes to the original protocol

There were two significant amendments to the original protocol developed in 2012. There was concern from the DMEC that the attrition rate may be higher than the 10% initially anticipated. The protocol was subsequently amended to allow recruitment to be extended to ensure that the original target of 525 complete data sets was reached.

The second noteworthy amendment was to the hypothesis tests of the statistical analysis. The decision was made to change the trial’s superiority sample size calculations from a one-sided 5% analysis to a two-sided 5% analysis. The trial was adequately powered for either option but, after extensive discussions with the funder, sponsor, Trial Steering Committee (TSC), TMG and DMEC, it was agreed that the use of a two-sided sample size calculation would be optimum.

Statistical analysis plan

Recruitment

We recruited patients from 34 primary care sites in Wessex through the PCRN (subsumed during the study into the CRN). We identified potential recruits to the study by searching general practice electronic medical records for patients with a coded diagnosis for asthma, undergoing current treatment and meeting inclusion criteria. Potential recruits were sent information about the study, the AQLQ (as impaired disease-specific health status was an inclusion criteria to enable an improvement to be demonstrated) and a prepaid return envelope.

A total of 15,203 invitation letters were mailed to potential participants and 1481 responses were received (a response rate of 9.7%). In total, 680 (45.9%) respondents were deemed ineligible, with lack of impairment according to AQLQ score being the most common reason for ineligibility; 655 participants (81.8% of the eligible respondents) were randomised, 261 (39.8% of randomised participants) to the DVD-delivered breathing retraining group, 262 (40.0%) to the usual-care group and 132 (20.2%) to the physiotherapist breathing retraining group (Table 2). The numbers of patients randomised from the different practices are shown in Appendix 1. The CONSORT diagram showing the flow of participants through the trial is provided in Figure 1.

FIGURE 1.

Consolidated Standards of Reporting Trials diagram.

Because of errors in completion of the primary outcome (AQLQ) at baseline for 45 out of 655 participants (7%), it was not possible to assign an AQLQ score to these patients. However, one or more AQLQ returns was achieved for all randomised participants.

Baseline characteristics and demographics

Baseline demographic and clinical features of the participants in the study are shown in Table 3. The demographic and clinical profiles of the recruited population were typical of adult patients with mild-to-moderate asthma in the community and were similar between treatment arms. More women than men consented to participate in the trial and the median age of participants was 57 years, with approximately one-third of participants aged < 50 years. The median (IQR) age at diagnosis of asthma was 29 (11–45) years. In total, 8% of participants were current smokers and 33% were ex-smokers. Approximately one-third of participants had a family history of asthma. The most common self-reported asthma triggers were dust (77%), pollen (64%), smoke (64%), exercise (63%), stress (46%), cats (41%), dogs (24%) and food (19%).

Participants had mildly impaired lung function, with a mean FEV₁% predicted of 91% and a mean FEV₁-to-FVC ratio of 0.8. The median (IQR) FeNO, a measure of eosinophilic airway inflammation, was 22 (14–34) parts per billion, which is at the top end of the normal range; over one-quarter of participants had a raised reading, indicative of persisting active inflammation despite inhaled steroid treatment. In terms of the BTS treatment step, our patient sample was typical of adult asthma patients treated in the community, with 7% at step 1 (needing short-acting beta-agonists only), 27% at step 2, 42% at step 3 and 12% at step 4 or 5. Self-reported atopic comorbidity was common (Table 4), with approximately two-thirds of participants reporting allergic problems including allergic rhinitis (69%) and eczema (31%), with most requiring pharmacological treatment. Over one-third of participants reported psychological problems (anxiety and depression), again with most requiring medication. Diagnosed COPD was present in only 2% of participants and had to be mild to meet entry criteria.

Baseline questionnaire data (Table 5) showed moderate impairment in asthma-specific QoL, with a mean (SD) AQLQ score of 4.3 (0.9), and also impaired symptomatic asthma control, with a mean (SD) ACQ score of 1.5 (0.9). The baseline psychological assessment with the HADS questionnaire showed a median (IQR) anxiety score of 6 (4–9) and depression score of 3 (1–5). A score of ≤ 7 on this tool is considered ‘normal’ and so > 255 patients had scores suggestive of significant anxiety, in keeping with population-based survey data. Across the five domains of the EQ-5D generic QoL questionnaire, 42% of participants reported problems with pain or discomfort, 31% with activities, 27% with anxiety or depression, 23% with mobility and 5% with self-care.

In summary, we successfully recruited a population of primary care asthma patients with mild-to-moderate asthma and with evidence of incomplete asthma control and QoL impairment, our target group. Their demographic profile was similar between randomisation arms and appears to be typical of the demographic and disease control profile of UK primary care adult asthma populations reported in recent surveys of asthma control.

Missing data, withdrawals and study retention

Primary outcome measure: disease-specific health status measured using the Asthma Quality of Life Questionnaire

The primary efficacy analysis was a comparison of between-group changes in AQLQ scores in the ITT population, with adjustments for prespecified covariates. Secondary analyses included an unadjusted comparison of between-group changes in AQLQ scores in the ITT population, adjusted and unadjusted comparisons of between-group changes in AQLQ scores in the PP population and analyses of the subdomains measured by the AQLQ instrument. Prespecified sensitivity analyses are also reported. We also assessed the time course of changes in AQLQ score from 3-month and 6-month AQLQ postal questionnaire data.

Table 6 presents the baseline and 12-month mean (SD) AQLQ scores and the unadjusted mean changes in AQLQ scores (with 95% CIs) in the DVD, physiotherapy and usual-care arms in the ITT and PP populations. Table 7 presents the primary efficacy analysis, the adjusted mean difference in 12-month AQLQ scores in the DVD, physiotherapy and usual-care treatment arms in the ITT and PP populations, with comparisons between the DVD arm and the control arm and between the physiotherapy arm and the control arm (superiority analysis) and between the DVD arm and the physiotherapy arm (equivalence analysis). The total AQLQ score (the primary efficacy measure) and the scores on the four subdomains of the AQLQ (which may be analysed and compared individually, measuring symptoms, activity, emotions and environment) are presented. The time course of AQLQ score changes in the ITT and PP populations is shown in Table 8.

Number needed to treat to achieve a clinically important improvement in the primary outcome measure, Asthma Quality of Life Questionnaire score

The number needed to treat (NNT) was calculated using the formula recommended by Guyatt et al. 32 (Juniper and Guyatt produced the AQLQ24 and ACQ25 tools). This analysis is based on an individual patient assessment of the proportions in each treatment arm showing a clinically significant improvement (≥ 0.5), the proportions showing unchanged scores (–0.49 to 0.49) and the proportions with a clinically significant deterioration (≤ –0.5) (see Appendix 5).

We found that, in the ITT population, 62% of participants in the DVD group reported a clinically significant improvement compared with 64% in the physiotherapy group and 56% in the control group (Table 9). The corresponding figures for deterioration were 5% in the DVD group, 4% in the physiotherapy group and 9% in the control group. The proportions in the PP population were slightly higher (see Table 9), with 74% of the DVD group, 76% of the physiotherapy group and 62% of the control group showing an improvement in QoL and 6% of the DVD group, 5% of the physiotherapy group and 10% of the control group showing a deterioration in QoL.

In between-group comparisons, these figures equated to a NNT for one patient to have a clinically relevant improvement in QoL in the ITT population of eight for the DVD arm compared with the usual-care arm (Table 10), seven for the physiotherapy arm compared with the usual-care arm (Table 11) and 41 for the physiotherapy arm compared with the DVD arm (Table 12). In the PP population the corresponding NNTs were eight, seven and 56 (Tables 13–15).

Prespecified sensitivity analysis on the primary outcome

As per the prespecified SAP, we carried out a sensitivity analysis on the primary outcome data, the change in AQLQ scores between baseline and 12 months adjusted for the prespecified covariates. In this analysis we included all randomised patients regardless of whether baseline or follow-up AQLQ data were present and we used the following rules (based on recommendations from Juniper) to assign values to missing data:

1. For missing baseline AQLQ scores, the following rules were adopted. Values for each AQLQ subdomain (symptoms, emotions, activities, environment) were calculated provided at least two-thirds of the items were scored, otherwise the domain value was set to missing. If any domain score was missing, the overall AQLQ score was set to missing and step 2 was followed.

2. Following step 1, any remaining missing baseline AQLQ scores were replaced by their cohort mean (as specified in the SAP).

3. For missing 12-month AQLQ scores, the method of LOCF was applied. If the 12-month score was missing, the 6-month score was used. If both the 12-month and 6-month scores were missing, the 3-month score was used to replace the 12-month score. If the 3-month, 6-month and 12-month scores were missing, that is, no follow-up information was available, then it was assumed that the patient returned to his or her baseline AQLQ score. Hence, the baseline and 12-month scores were the same.

Table 16 shows the baseline and 12-month unadjusted scores and the unadjusted mean differences in the three treatment arms in the ITT and PP populations, and Table 17 shows the adjusted mean difference in 12-month AQLQ scores in the three treatment arms in the ITT and PP populations. This sensitivity analysis provides a similar message to the primary analyses: we see that there are significant improvements in both of the active arms (DVD and physiotherapy) above usual care in both the ITT and PP populations, although predictably of slightly lower magnitude (as we assume ‘no change’ in cases with missing data) than in the main analysis. We continue to show equivalence between the two active arms.

Secondary outcome measures

Asthma exacerbations

Asthma exacerbations (attacks) were assessed [using the European Respiratory Society (ERS)/American Thoracic Society (ATS) Task Force on asthma outcomes recommendations for defining an exacerbation33] according to prescribed short courses of oral corticosteroids for worsening of asthma by the usual physician, with data obtained from the manual and electronic reviews of the patient medical record at 12 months. In keeping with the mild asthma population recruited, only 12% of the ITT population (13% of the PP population) had one or more asthma attack over the 12-month period (Table 22). The percentage of patients in the ITT population randomisation treatment arms (DVD, physiotherapy, control) having one or more asthma attack was 9%, 11% and 15%, respectively, and in the PP population was 10%, 12% and 16% respectively (Table 23). There was no statistically significant difference in the crude exacerbation rate between the DVD arm and the physiotherapy arm (p = 0.6 in the ITT population, p = 0.8 in the PP population; Table 24) or between the physiotherapy arm and the control arm (p = 0.4 for both populations; Table 25). The DVD group narrowly failed to reach a statistically significant reduction in exacerbations compared with the control group, with a p-value of 0.06 in the ITT population and 0.12 in the PP population) (Table 26).

TABLE 22 - Oral corticosteroid courses at 12 months in the DVD, physiotherapy and usual-care treatment arms in the ITT and PP populations

Oral corticosteroid courses	ITT population, n (%)				PP population, n (%)
	DVD (N = 261)	Physiotherapy (N = 132)	Usual care (N = 262)	Total (N = 655)	DVD (N = 215)	Physiotherapy (N = 110)	Usual care (N = 231)	Total (N = 556)
None	237 (90.8)	117 (88.6)	223 (85.1)	577 (88.1)	193 (89.8)	97 (88.2)	195 (84.4)	485 (87.2)
1	17 (6.5)	10 (7.6)	26 (9.9)	53 (8.1)	16 (7.4)	9 (8.2)	26 (11.3)	51 (9.2)
2	2 (0.8)	4 (3.0)	10 (3.8)	16 (2.4)	2 (0.9)	3 (2.7)	8 (3.5)	13 (2.3)
3	2 (0.8)	0 (0.0)	1 (0.4)	3 (0.5)	2 (0.9)	0 (0.0)	1 (0.4)	3 (0.5)
≥ 4	3 (1.1)	1 (0.8)	2 (0.8)	6 (0.9)	2 (0.9)	1 (0.9)	1 (0.4)	4 (0.7)
Total	261 (100.0)	132 (100.0)	262 (100.0)	655 (100.0)	215 (100.0)	110 (100.0)	231 (100.0)	556 (100.0)

TABLE 23 - Oral corticosteroid courses at 12 months in the DVD, physiotherapy and usual-care treatment arms in the ITT and PP populations: one or more courses

Oral corticosteroid courses	ITT population, n (%)				PP population, n (%)
	DVD (N = 261)	Physiotherapy (N = 132)	Usual care (N = 262)	Total (N = 655)	DVD (N = 215)	Physiotherapy (N = 110)	Usual care (N = 231)	Total (N = 556)
None	237 (90.8)	117 (88.6)	223 (85.1)	577 (88.1)	193 (89.8)	97 (88.2)	195 (84.4)	485 (87.2)
≥ 1	24 (9.2)	15 (11.4)	39 (14.9)	78 (11.9)	22 (10.2)	13 (11.8)	36 (15.6)	71 (12.8)
Total	261 (100.0)	132 (100.0)	262 (100.0)	655 (100.0)	215 (100.0)	110 (100.0)	231 (100.0)	556 (100.0)

TABLE 24 - Comparison of oral corticosteroid courses at 12 months in the ITT and PP populations: DVD vs. physiotherapy treatment arms

Oral corticosteroid courses	ITT population, n			PP population, n
	DVD (N = 261)	Physiotherapy (N = 132)	p-value	DVD (N = 215)	Physiotherapy (N = 110)	p-value
None	237	117	0.62	193	97	0.81
≥ 1	24	15		22	13
Total	261	132		215	110

TABLE 25 - Comparison of oral corticosteroid courses at 12 months in the ITT and PP populations: physiotherapy vs. usual care

Oral corticosteroid courses	ITT population, n			PP population, n
	Physiotherapy (N = 132)	Usual care (N = 262)	p-value	Physiotherapy (N = 110)	Usual care (N = 231)	p-value
None	117	223	0.42	97	195	0.45
≥ 1	15	39		13	36
Total	132	262		110	231

TABLE 26 - Comparison of oral corticosteroid courses at 12 months in the ITT and PP populations: DVD vs. usual care

Oral corticosteroid courses	ITT population, n			PP population, n
	DVD (N = 261)	Usual care (N = 262)	p-value	DVD (N = 215)	Usual care (N = 231)	p-value
None	237	223	0.06	193	195	0.12
≥ 1	24	39		22	36
Total	261	132		215	231

A negative binomial regression model was constructed to assess the between-group differences in asthma exacerbation frequency in the ITT (Tables 27–29 and Figure 2) and PP (Figure 3 and see Appendix 6) populations, adjusting for differences in exacerbation frequency at baseline and for covariates (age, sex, BTS treatment step and smoking status). In the adjusted analysis for the ITT population, the risk of exacerbations was lower in the DVD arm than in the usual-care arm [incidence rate ratio (IRR) 0.68), but the 95% CI crossed the line of unity (95% CI 0.34 to 1.38) and so this finding was not statistically significant and could have occurred through chance. The analysis of the physiotherapy arm compared with the usual-care arm produced a similar result, with a (lower magnitude) non-significant reduction in the risk of an asthma attack in the physiotherapy arm (IRR 0.85, 95% CI 0.36 to 1.98). There was also no significant difference in the risk of an exacerbation in the DVD group compared with the physiotherapy group, but the DVD group was favoured (IRR 0.81, 95% CI 0.33 to 1.99). The unadjusted negative binomial regression model produced very similar results, as did the adjusted and unadjusted models in the PP population. Our study was not powered to show an effect on exacerbations and, as the annual exacerbation rate is modest in patients with mild and moderate asthma treated in the community, a larger sample would be needed to show a statistically significant reduction in exacerbations. This will be considered further in Chapter 7.

TABLE 27 - Group differences in asthma exacerbations in the 12 months post randomisation in the ITT population

Oral corticosteroid courses	Treatment arm, n			Total, n
	DVD	Physiotherapy	Usual care
0	237	117	223	577
1	17	10	26	53
2	2	4	10	16
3	2	0	1	3
4	2	1	1	4
6	1	0	0	1
10	0	0	1	1
Total	261	132	262	655

TABLE 28 - Unadjusted analysis of the negative binomial regression for asthma exacerbations in the 12 months post randomisation in the ITT population

Comparison	Unadjusted IRR	95% CI	p-value
Physiotherapy vs. usual care	0.69	0.34 to 1.41	0.31
DVD vs. usual care	0.65	0.37 to 1.16	0.15
DVD vs. physiotherapy	0.94	0.45 to 1.96	0.87

TABLE 29 - Adjusted analysis of the negative binomial regression for asthma exacerbations in the 12 months post randomisation in the ITT population

As there were only 78 cases with one or more oral corticosteroid courses, only the covariates age, sex, BTS treatment step and smoking status were included in the model.

Comparison	Adjusted IRRa	95% CI	p-value
Physiotherapy vs. usual care	0.85	0.36 to 1.98	0.89
DVD vs. usual care	0.68	0.34 to 1.38	0.42
DVD vs. physiotherapy	0.81	0.33 to 1.99	0.84

FIGURE 2.

Frequency of one or more exacerbations in the 12 months post randomisation by treatment arm in the ITT population. OC, oral corticosteroid.

FIGURE 3.

Frequency of one or more exacerbations in the 12 months post randomisation by treatment arm in the PP population. OC, oral corticosteroid.

Bronchodilator use

In the 12 months following the baseline assessment, 177 of 261 (67.8%) participants randomised to the DVD arm required one or more prescriptions for rescue bronchodilators compared with 104 of 132 (78.8%) in the physiotherapy arm and 206 of 262 (78.6%) in the control arm. The number of inhalers prescribed over the 12 months post randomisation by treatment arm in the ITT population is shown in Table 30 and the proportions of participants having each number of inhalers are shown graphically in Figure 4. A negative binomial regression model was constructed for the ITT population to estimate the difference in rescue bronchodilator use between arms. In the unadjusted analysis (Table 31) there was a non-significant trend for lower bronchodilator use in the DVD arm compared with the usual-care arm (IRR 0.83, 95% CI 0.68 to 1.03; p = 0.09) and for the DVD arm compared with the physiotherapy arm (IRR 0.81, 95% CI 0.63 to 1.04; p = 0.10), with little difference between physiotherapy arm and the usual-care arm (IRR 1.03, 95% CI 0.81 to 1.33; p = 0.80). In the regression model adjusted for covariates (age, sex, BTS treatment step, baseline smoking status, baseline HADS scores and baseline Nijmegen questionnaire score; Table 32), a similar pattern was seen, with the adjusted IRR being 0.91 (95% CI 0.72 to 1.15; p = 0.61) for the DVD arm compared with the control arm, 0.88 (95% CI 0.66 to 1.15; p = 0.50) for the DVD arm compared with the physiotherapy arm and 1.03 (95% CI 0.79 to 1.37; p = 0.93) for the physiotherapy arm compared with the usual-care arm. The same analyses were performed on the PP population, which also found no significant differences in rescue bronchodilator use between randomisation treatment arms in the adjusted or the unadjusted analysis (Figure 5 and see Appendix 5).

TABLE 30 - Distribution of rescue inhaler use in the 12 months post randomisation in the ITT population

Number of rescue inhalers	Treatment arm, n			Total, n
	DVD	Physiotherapy	Usual care
0	84	28	56	168
1	36	26	61	123
2	39	16	32	87
3	20	18	26	64
4	23	14	24	61
5	17	4	12	33
6	12	6	11	29
7	6	4	5	15
8	6	3	11	20
9	5	2	3	10
10	5	2	6	13
11	2	1	2	5
12	1	3	5	9
13	1	2	0	3
14	2	2	0	4
16	1	0	1	2
17	0	0	1	1
18	0	0	3	3
22	0	0	2	2
26	0	0	1	1
28	1	0	0	1
40	0	1	0	1
Total	261	132	262	655

FIGURE 4.

Distribution of rescue bronchodilator prescriptions in the 12 months post randomisation by treatment arm in the ITT population.

TABLE 31 - Unadjusted analysis of the negative binomial regression for rescue inhaler use in the 12 months after baseline in the ITT population

Comparison	Unadjusted IRR	95% CI	p-value
Physiotherapy vs. usual care	1.03	0.81 to 1.33	0.80
DVD vs. usual care	0.83	0.68 to 1.03	0.09
DVD vs. physiotherapy	0.81	0.63 to 1.04	0.10

TABLE 32 - Adjusted analysis of the negative binomial regression for rescue inhaler use in the 12 months after baseline in the ITT population

Adjusted for age, sex, BTS treatment step, baseline smoking status, baseline HADS scores and baseline Nijmegen questionnaire score.

Comparison	Adjusted IRRa	95% CI	p-value
Physiotherapy vs. usual care	1.03	0.79 to 1.37	0.93
DVD vs. usual care	0.91	0.72 to 1.15	0.61
DVD vs. physiotherapy	0.88	0.66 to 1.15	0.50

FIGURE 5.

Distribution of rescue bronchodilator prescriptions in the 12 months post randomisation by treatment arm in the PP population.

Adverse events

Adverse events were collected for all randomised patients by patients and study centre principal investigators (PIs) on adverse event report forms and by manual and electronic review of medical records at 12 months. Adverse events were categorised by PIs according to the classification scheme provided in Appendix 9.

Introduction

The health economic analysis evaluated the superiority of each intervention compared with usual care and made a non-inferiority comparison between the DVD arm and the face-to face physiotherapy arm. The ITT population was employed, with imputation of missing values. The time frame was that of the trial. Adjustment was made for baseline values and for repeated measurement. The perspective was that of the NHS, but with exploration of non-NHS costs.

Data collected

Resource use data were collected using three forms. The medication review form was used to record data on all asthma-related NHS consultations and prescriptions over the period of the trial, collected from patient records held by each general practice. It was also used to record outpatient visits, accident and emergency (A&E) visits, out-of-hours consultations and hospital admissions related to asthma. A concomitant medications form was used to record data on six conditions that might be linked to differences in outcomes. A third form was used to record data from patients relating to over-the-counter medications, non-NHS consultations and any other privately accessed items related to asthma or its palliation.

The results presented here relate only to the medication review data. The data on concomitant medications proved unusable for several reasons, principally because of lack of detail provided on relevant medications and dates. The data on patient-reported costs also proved of little use as very few items were recorded. This we interpret as indicating that few such costs were incurred; hence, this justified taking a NHS perspective as opposed to a societal perspective.

Intervention costs

Intervention costs (shown in Appendix 12) were £83.45 per patient in the face-to-face physiotherapy breathing retraining group and £2.85 in the DVD breathing retraining group. The physiotherapy group cost was based on the duration of the planned three consultations with the practice nurse, adjusted for attendance. The cost of the DVD intervention was that of printing the discs and documentation. The development cost of these materials was treated as a research cost and was not included here.

Resource use and costs

The unit costs for each of the resource groups (GP consultations, outpatient visits, A&E attendances, out-of-hours consultations, inpatient admissions) and sources are shown in Appendix 12.

EuroQol-5 Dimensions and Asthma Quality of Life Questionnaire

Asthma Quality of Life Questionnaire data collection was discussed earlier (see Chapter 3). EQ-5D data were collected at the initial visit at baseline and at 3 and 6 months (both by post) and 12 months (assessment visit).

Analyses

Cost–utility analyses were based on all patients randomised. Utility scores were derived based on the UK tariff. 34 QALYs were calculated based on the area under the curve approach. The time frame of the analysis was that of the trial (12 months).

Bootstraps based on 1000 samples with replacements were used to estimate the cost per AQLQ score improvement and cost per QALY gained, with CIs presented to illustrate the uncertainty. Scatterplots are presented to illustrate the uncertainty in the cost–utility estimates.

Results

The medical review form was used to extract patient resource use data in the 12 months following randomisation. Three patients died and two patients left their practice. We assumed that no additional costs were incurred for these patients after the date of their last visit.

The mean NHS cost per patient by treatment arm was aggregated from the costs of asthma-related prescriptions, consultations and hospital admissions. The unadjusted mean cost was highest in the usual-care arm (£356; Table 43), with a similar cost in the physiotherapy arm (£335) and the lowest cost in the DVD arm (£296). The main cost items for each group were asthma-related medications and GP consultations, with lower costs reflecting low levels of use of other services; although there were few hospital admissions (usual-care arm, n = 8; physiotherapy arm, n = 0; DVD arm, n = 4), they were by far the most costly item. The numbers of patients using each service are also shown in Table 43.

The mean costs in each arm changed only slightly when bootstrap methods were used: £377 in the usual-care arm, £333 in the physiotherapy arm and £293 in the DVD arm (Table 44). The differences were not statistically significant at the 5% level.

Given that the intervention costs were higher in the physiotherapy arm (£83.5) and the DVD arm (£2.85) than in the usual-care arm, the inclusion of NHS costs offset these higher costs, leading to lower overall mean costs in both intervention arms than in the usual-care arm.

EuroQol-5 Dimensions data were collected at baseline (n = 653/655), 3 months (n = 519/655), 6 months (n = 472/655) and 12 months (n = 437/655) (see Table 45). For the whole sample, completion fell from an estimated 99% at baseline to 67% at 12 months. Some participants missed some but not all time points, but data for at least two time point were available for 87% of participants. Linear interpolation was used for those with two or three time points, with imputation for those with fewer time points.

The mean EQ-5D QoL scores varied between 0.751 and 0.826, with small differences at baseline and over time.

The mean EQ-5D QoL scores based on complete cases are shown in Table 46.

Cost-effectiveness analysis

The primary analysis of AQLQ score differences was based on the population with baseline measurement of AQLQ scores (n = 244, 120 and 246 in the DVD, physiotherapy and usual-care arms respectively; see Chapter 3). The cost-effectiveness analyses were based on the same population (n = 610 in total).

Bootstrapped mean costs per person in this population were £380 (95% CI £310 to £459), £296 (95% CI £228 to £374) and £334 (95% CI £299 to £269) for the DVD, physiotherapy and usual-care arms respectively. This gave a difference in costs of –£83 (95% CI –£187 to £12) for the DVD arm and –£45 (95% CI –£134 to £33) for the physiotherapy arm compared with the usual-care arm (Table 47).

The estimated differences in AQLQ scores, based on bootstrap methods with adjustments for prespecified covariates, were similar to those in the primary analyses, with differences of 0.23 (95% CI 0.06 to 0.40) for the physiotherapy arm compared with the usual-care arm and 0.26 (95% CI 0.11 to 0.41) for the DVD arm compared with the usual-care arm (see Table 47).

The incremental cost per AQLQ score improvement was –£400 (95% CI –£1545 to £106) for the physiotherapy arm compared with the usual-care arm and –£340 (95% CI –£986 to £52) for the DVD arm compared with the usual-care arm (see Table 47). Both interventions dominated usual care. Given the non-inferiority of the DVD intervention compared with the physiotherapy intervention, cost-minimisation analysis favoured the DVD intervention over the physiotherapy intervention. This was the base case for the cost-effectiveness analysis.

Cost–utility analysis

Incremental differences between the arms in terms of costs and QALYs and the incremental cost-effectiveness ratios (ICERs) are shown in Table 48. There were small baseline adjusted QALY differences between the arms, with the face-to-face physiotherapy arm having 0.007 more QALYs than the usual-care arm and the DVD arm having 0.02 more QALYs than the usual-care arm. The DVD arm showed a gain of 0.014 QALYs compared with the face-to-face physiotherapy arm. The gains were not statistically significant.

In the adjusted analysis (bottom part of Table 48), the ICERs had negative values for both the physiotherapy arm and the DVD arm compared with the usual-care arm. The physiotherapy arm cost £877 less per QALY than the usual-care arm and the DVD arm cost £3057 less per QALY than the usual-care arm. Both interventions dominated usual care. The CIs for these ICERs were wide and not statistically significant. Similar results were found in the unadjusted analysis.

The probability of dominance was investigated for each of the comparisons. This can be interpreted as the proportion of the simulated results (points on a scatterplot) that fall into the lower right-hand (south-east) quadrant. Scatterplots are shown of the joint distribution of the incremental mean cost and mean QALYs for the DVD arm compared with the usual-care arm (Figure 6) and for the physiotherapy arm compared with the usual-care arm (Figure 7).

FIGURE 6.

Scatterplot of the joint distribution of the incremental mean cost from a NHS perspective and mean QALYs adjusted for baseline QoL over 12 months based on the imputed data: 95% CI ellipse DVD arm compared with the usual-care arm.

FIGURE 7.

Scatterplot of the joint distribution of incremental mean cost from NHS perspective and mean QALYs adjusted baseline QoL over 12 months based on the imputed data: 95% CI ellipse physiotherapy arm compared with the usual-care arm.

The probability of usual care being dominated by the DVD intervention (lower costs and higher QALYs) was 82%. The probability of usual care being dominated compared with the face-to-face physiotherapy intervention (lower costs and higher QALYs) was 51%.

Discussion and conclusions

This well-conducted randomised trial provides evidence of the likely cost-effectiveness of both face-to-face physiotherapy and breathing retraining delivered using a DVD and booklet. Limitations include the time frame used, the reliance of the cost-effectiveness estimates on a single trial and the restriction of the costing perspective to the NHS. Although longer-term follow-up would be desirable, it seems likely that patients who have been able to improve their QoL using breathing retraining would continue to apply the techniques that they have learned. If so, the benefits recorded here may have been understated. Although cost-effectiveness should ideally be based on the totality of evidence, such as a meta-analytical estimate of effect size, we note similar results for the one similar trial of breathing exercises in asthma. 17 Our adoption of a NHS, as opposed to a societal, perspective was based on the returns from a questionnaire, which indicated that few costs fell outside the NHS.

The cost/AQLQ analysis favoured both the DVD intervention and the face-to-face physiotherapy intervention compared with usual care, with the DVD intervention achieving equivalent outcomes to the face-to-face physiotherapy intervention at a lower cost. The cost/QALY analysis showed similar results for the DVD and face-to-face physiotherapy comparisons with usual care, albeit based on small non-significant differences in outcomes. Both interventions dominated usual care. The DVD intervention had similar outcomes to the face-to-face physiotherapy intervention but at lower costs. The probabilities of the interventions being dominant were 85% for the DVD intervention compared with usual care and 51% for the face-to-face physiotherapy intervention compared with usual care. Thus, both the cost-effectiveness analysis and the cost–utility analysis provided congruent results. Both interventions achieved outcome gains at lower total costs than usual care. The lower cost in the DVD group meant that the DVD intervention was preferable to face-to-face physiotherapy. The low cost of the DVD (£2.85) meant that, if effective, it was highly likely to be cost-effective. Further, if made available on the internet, its cost would fall to close to zero.

In conclusion:

• the QALY differences between the treatment arms were in the same direction as the differences in the primary outcome, but were smaller

• the increased cost of each intervention was offset by reductions in total costs so that both interventions dominated usual care

• the DVD/booklet training was preferable to face-to-face physiotherapy, as the outcomes were within the equivalence margin and the DVD intervention had a lower cost.

Face-to-face physiotherapy

Although breathing retraining for asthma is provided by some physiotherapists within the NHS and privately, in clinical practice there may be considerable variation in the content and frequency of face-to-face sessions. For this research we needed to standardise the intervention as much as possible to ensure that each participant received similar treatment and potentially to allow for comparability across different therapists. Our initial plan was to recruit two to three part-time physiotherapists to provide the face-to-face intervention during the life of the BREATHE trial.

The first step was to decide on the frequency of sessions. In clinical practice, physiotherapists have the option to tailor their treatments to the needs of individual patients, which results in large variations in treatment session numbers from patient to patient. However, previous related research trials by the chief investigator17,19 had demonstrated that three face-to-face sessions were sufficient to have a positive effect on the primary outcome, and discussions with local physiotherapists also suggested that three was the average number of sessions provided within the NHS. It was therefore decided that all participants should be offered three sessions with the physiotherapist, with intervention adherence being predefined as attending any two out of these three sessions.

The content for each session was derived from discussions with physiotherapists, from AB’s own experience and from discussion with other members of the research team. From these discussions, AB produced two documents: a complete description of the protocol for face-to-face physiotherapy [Face-To-Face Physiotherapy Protocol (Anne Burton, University of Southampton, 2012)] and a reduced version of the protocol to be given to intervention physiotherapists during training workshops [Training Manual for Physiotherapists (Anne Burton, University of Southampton, 2012)].

Training workshops were held by AB at various locations, at which the physiotherapists received information about the trial and the training manual. Details of the training provided at these workshops can be found in the manual. Although five physiotherapists attended these workshops, ultimately, only one physiotherapist (Ruth DeVos) provided all of the intervention sessions throughout the life of the trial. Fidelity of intervention delivery was determined in two ways: first, by the physiotherapist completing specifically designed checklists (see the Face-To-Face Physiotherapy Protocol for details) and, second, through the use of direct observation, whereby AB went out to the general practices during both the pilot trial and the main trial to observe the physiotherapist delivering a session of the intervention for approximately 5% of the participants (6 out of the 132 recruited to this arm). There was 100% adherence to the intervention protocol by the physiotherapist (physiotherapist adherence to the protocol had been predefined as conforming to 90% of the checklist).

Booklet development (Breathing Freely)

Previous experience from within the research team had indicated that accompanying materials (such as a written booklet) containing behavioural content to motivate and reassure patients were very useful to support user self-management. 35 Our booklet, Breathing Freely: Your Guide to Breathing Retraining for Asthma, was developed by a team of health psychologists, physiotherapists, physicians and patient representatives (Emily Arden-Close, Sarah Kirby, Emma Teasdale, Anne Bruton, Mike Thomas, Mark Stafford-Watson, Denise Gibson, Lucy Yardley, University of Southampton, 2013).

A draft booklet was first created by three of the authors (EAC, LY), based on advice from physiotherapists (AB, DG) and a GP with expertise in asthma (MT) and a similar booklet that LY had created that had successfully supported the exercise-based self-management of symptoms in patients with dizziness. 35 The first part of the booklet (pages 1–8) was designed to build motivation (in terms of social cognitive theory and positive outcome expectancies) and confidence (self-efficacy) to undertake breathing retraining. To convince users that breathing retraining would be effective, the cognitive–behavioural rationale for breathing retraining was first explained in terms of the vicious cycle of breathing problems, anxiety and hyperventilation and how this could be reversed by breathing retraining. Positive expectancies for treatment benefit were promoted by explaining how breathing retraining can improve not only breathing but also psychological well-being, and can allow users to do more without becoming breathless. The second part of the booklet (pages 10–15) contained detailed instructions (with illustrations) on how to progressively master breathing retraining, with reassurance and advice addressing common difficulties and barriers. The final part of the booklet (pages 16–21) encouraged users to use their new breathing skills to engage in more daily activities that they might have previously found difficult because of asthma (e.g. exercise, stressful situations). Implementation of breathing retraining was encouraged by providing success stories from other patients and charts to plan when to carry out the breathing exercises and physical activity and log progress.

Using our ‘person-based approach’ to developing accessible, persuasive and helpful intervention materials,36 feedback on the draft version was sought through the use of semistructured think-aloud interviews lasting approximately 1 hour with 29 people with asthma. The interviews included two components: open-ended questions exploring attitudes to breathing retraining exercises in the context of health beliefs and lifestyle and think-aloud methods to elicit spontaneous reactions to the booklet. Interview questions covered participants’ experiences of managing their asthma. This included questions about breathing difficulties, previous experience of breathing techniques for asthma management and the perceived relevance of breathing training exercises to asthma control. Participants then read the draft booklet and provided feedback about each page. Further open-ended questions explored impressions of the booklet, attitudes to breathing exercises, perceived attitudes of relevant others (family, health-care professionals), barriers to carrying out breathing exercises at home and possible ways of overcoming these barriers.

After each interview the feedback on each page was tabulated into recommended changes and positive aspects of the booklet. After the first few interviews had been conducted, the recommended changes were collated and a MoSCoW (must, should, could, would) analysis was conducted (breaking changes down into things we must, could, should and were not going to do), based on the nature of the comments and how many people mentioned each comment. The booklet was then revised before further interviews were conducted. This cycle continued until no further major changes were suggested. Further details of this process and the changes made in response to feedback can be found in two publications. 36,37

DVD development

Our aim was to produce an intervention that replicated the experience of face-to-face physiotherapy breathing retraining in an accessible self-guided format. We acquired the media in a high-quality video file format to ‘future-proof’ our content as much as possible and facilitate distribution using different technologies. The available technologies for distribution at the time of the original funding application were videotape, DVD and the internet. Although we knew that distributing video via the internet was possible, we chose the DVD as the distribution format for this project because, at the time, DVD players were becoming more popular and fewer people in the UK had access to the internet. In 2009 (when the funding application was written) the broadband network and hardware required to view internet-distributed video was not established or as widespread as it is now. Since then, distributing video via the internet has become ubiquitous. Reformatting our original video files for distribution via the internet will be relatively straightforward.

To enable people to identify with the patients seen on screen on the DVD, it was decided to involve genuine individuals with asthma (rather than actors) to take on the roles of patients. A local specialist respiratory physiotherapist (Denise Gibson) was asked to play the role of the breathing retraining physiotherapist. Dr Mark Porter (who has considerable experience of fronting medical programmes on both radio and television) kindly agreed to be the ‘presenter’ for the DVD.

The DVD production team consisted of AB, MT and DG, the psychology team developing the booklet, patient representatives, a commercial company (Zemedia) and staff at Solent University Studios, Southampton. AB produced a draft script for the DVD, based on a combination of the written Face-To-Face Protocol, the developing Breathing Freely booklet and discussions with physiotherapists. This was then shared with our psychology team members, patient representatives and others in the research team. Input from TMG patient representatives was of great importance in these revisions, and these representatives were involved in and commented on all revision steps, providing verbal and written comment on patient perspectives of the intervention. In addition, the iterative person-based think-aloud process for the development of digital behavioural interventions that was used in the development process for this intervention (described in detail in Chapter 6) used a representative sample of 29 adults with asthma purposively sampled for diversity of age, sex, education and symptom profile. These patients looked at draft versions of the intervention and provided feedback and comment using a formal qualitative methodology, and this was used to modify the intervention based on their reactions and comments. Several revisions to the script were subsequently made in response to comments and feedback from within the team. A professional media production team (Zemedia) had been identified by AB in 2009 during the initial funding application process and its main representative (Tim O’Riordan) had been regularly attending project meetings from 2011. Zemedia were employed to translate our script into a filmable story, to film the scenes and to create and produce the final DVD. Script refinement and preparation for filming continued over the first year of the research programme, during which period there were several team meetings to agree on patient roles and locations and to design ‘story boards’ for individual scenes. Filming took place over 3 days in May and June 2012, in Solent University Studios and on location in patients’ homes, with Steve Bowles of Zemedia and AB directing. Clips from the filming were shared within the research team for comment in terms of what to include/exclude, what voiceovers were appropriate and general formatting of the DVD. It was not felt appropriate to seek patient feedback on the developing DVD itself at that stage as it would not have been possible to refilm scenes because of time and financial constraints. Tim O’Riordan was responsible for the editing and post-production work that resulted in the final version of the DVD used in the trial. Feedback on the DVD as used by participants was sought as part of the research trial and is described elsewhere in this report.

Patient and public involvement (PPI) was central to this project, with two PPI representatives on the TMG attending TMG meetings and commenting on all stages of the process, and with the patient charity Asthma UK being a partner in the process. With the partnership of the charity and the active involvement of patients in all stages of the trial, in particular in the development and optimisation of the DVD intervention and booklet, we feel that there was complete PPI involvement overall. We did not experience any difficulty in obtaining PPI or in acting on it and would recommend that further trials in this clinical area integrate PPI into their processes as much as possible and involve patient charities if possible.

Qualitative process evaluation

The qualitative process evaluation aimed to explore the experiences of participants in the intervention arms (DVD and face-to-face physiotherapy) of the BREATHE trial, to understand what patients perceive as the strengths and weaknesses of the different modes of delivery of breathing retraining.

Quantitative process evaluation

The objective of the quantitative process evaluation was to quantify patient-reported expectations of treatment, experiences of treatment and engagement. This process evaluation included five main sets of analyses:

1. Sensitivity analyses – of engagement with and amount of practice of the BREATHE exercises reported at 3 months’ follow-up. The rationale for these analyses was to assess treatment efficacy in participants who met ‘threshold engagement’ levels; that is, efficacy is expected to be greater in those who have looked at the DVD/booklet and carried out breathing retraining at least once, and to investigate whether participants who practised the exercises more regularly derived more benefit.

2. Between-group differences – to explore whether there was a difference between the two intervention groups on measures of expectancy, experience, treatment engagement and practical barriers.

3. Predictors of the amount of practice and continuous engagement – to assess whether expectancy, experience and practical barriers are associated with the amount of practice (at 3 months) and continuous engagement (at 6 and 12 months).

4. Analyses were carried out to model the theory of planned behaviour and explore whether intentions, the amount of practice, engagement and continued engagement can be predicted by the model.

5. The factor structure of the Problematic Experiences of Therapy Scale (PETS)43 was checked using exploratory factor analysis as it has not previously been used in an asthma population.

Conclusions

Study findings

Strengths and weaknesses of the study

This study has many strengths and a few weaknesses. This is by far the largest study of breathing retraining in asthma to have been reported and we know of no other studies of comparable size currently under way or planned. To our knowledge, this is the first study of breathing retraining in asthma that has compared a face-to-face programme with a self-guided programme and the first to have a rigorous, prospective health economic evaluation and process evaluations embedded within it.

In terms of study design, this study followed two previous smaller studies from the study team17,19 that together represent an evaluation strategy in keeping with the recommendations of the MRC framework for the evaluation of complex interventions,38 with this project being a Phase IV study. The study design was a ‘pragmatic’ RCT, with a focus on confirming the clinical effectiveness and cost-effectiveness of breathing retraining in a broad and representative primary care population. We therefore had wide entry criteria and study procedures that were designed to be as easy to comply with as possible and to interfere with normal care (other than the provision of the interventions) as little as possible. This allowed primary care sites to host the study without too much disruption to usual care and, we believe, encouraged patient participation. We were able to recruit and retain 655 primary care patients into a 12-month study that required informed consent, study visits and questionnaire completion. The study was designed and supervised by a multidisciplinary team, with strong PPI input at all stages provided by our very active PPI representatives and by the patient charity Asthma UK, which was a full partner in the project from its inception. The multidisciplinary team included physiotherapists, primary and secondary care clinicians, scientists, primary and secondary care respiratory nurses, health psychologists, behaviour change experts specialising in developing and evaluating self-guided behaviour change interventions, statisticians and health economists. The involvement of researchers with a general practice background facilitated a study design that was feasible and acceptable in a busy ‘routine care’ setting and acceptable to patients with mild-to-moderate asthma. The use of an internal pilot allowed hindrances to recruitment and study procedures to be identified and corrected early in the study, with subsequent smooth running of the main trial. Regular steering group meetings were well attended and active, and the team was fortunate to have the support of very interested and active TSC and DMEC members. There was a regular exchange of information with the funder, the National Institute for Health Research (NIHR) HTA programme, as well as the provision of support. Other NIHR structures also provided excellent support to the study. The SCTU undertook trial management most effectively, with several changes in key personnel not impairing the continuity and effectiveness of the team. The friendly, accessible and proactive involvement of the SCTU team was a key strength in delivering a successful study. Invaluable support was provided by the NIHR CRN team involved in the study, with strong support from the research nurse team with regard to the recruitment of practices and patients. There was good partnership with the NHS organisations (clinical commissioning groups and trusts) involved in the study, who also provided support. The general practices hosting the study were invariably supportive and appreciated the pragmatic design of the study and the involvement of practice asthma nurses in the study. Very good feedback was received from the recruiting sites and the GPs, practice nurses and practice managers involved, who were positive about the study and enjoyed their involvement in it. A key strength underlying these effective partnerships was good and open communication throughout the study, allowing ‘buy-in’ and effective team working. As a result of these good relationships, coupled with a sound study design, we were able to recruit beyond the original recruitment targets and to achieve high retention rates.

Another strength of the study is that we used a variety of clinically relevant outcome measures, including a panel of validated patient-reported outcome measures in questionnaire format, physiological lung function measures (that included quality-assured spirometry measured by trained and accredited respiratory nurses) and airway inflammation measures. In addition, patients provided consent to access their routine primary care medical records, which allowed us to collect consultation, prescribing and other health resource use data. This combination of outcome measures allowed us to assess the effects of the intervention on QoL, the key patient-focused outcome measure that reflects patients’ experience of their disease and the amount of disturbance that it causes in their lives, and relate this to other physiological and psychological measures. This multidimensional assessment allows some insight into the mechanisms of effect of this complex intervention and further helps to dispel the myths propagated by some proponents of breathing retraining (such as some of those advocating the Butekyo method) that breathing exercises can ‘cure’ asthma and replace the need for anti-inflammatory and bronchodilator medication. 66–68 In keeping with our previous studies, we found no significant effect of the interventions on parameters measuring the pathophysiology of asthma, such as lung function and airway inflammation, despite a significant lessening of the impact of asthma on people’s lives.

A key strength of the study was the rigorous development process used for transferring the physiotherapy-based programme to a self-guided intervention, with printed support material. The development process that we have described is based on a well-established methodology using an iterative process and extensive and effective patient input. Members of the intervention development team were very experienced, having previously developed a number of effective self-guided interventions in other clinical areas, and applied a tried and tested procedure to develop and refine the materials. The involvement of experienced qualitative researchers allowed patient perspectives to be efficiently and accurately gathered and assessed and to be effectively incorporated into the intervention. A professional media production firm was used to produce the high-quality and patient-friendly self-guided intervention. The self-guided intervention proved to be accessible and well received by patients of all educational levels. The same team designed and performed the process evaluation (including both the qualitative evaluation and the quantitative evaluation), which has contributed to the assessment of the intervention and will inform subsequent implementation work. We were also fortunate to identify experienced local physiotherapists with expertise in breathing retraining who were able to assist with the DVD development and provide the face-to-face physiotherapy intervention.

Similarly, a strength of the study was the involvement of health economists at all stages, from protocol design to writing of the report. The health economic evaluation was seamlessly integrated into the study design and execution. High-quality health economic information is imperative in the current medicopolitical environment if a new intervention is going to be accepted and utilised in clinical care and promoted by commissioners. We believe that the powerful and persuasive health economic arguments, combined with the simplicity and ease of provision of the self-guided intervention, will allow rapid uptake and implementation throughout the NHS, to the benefit of patients.

A further strength of the study was the full involvement of the statistical team at all stages of the project, with input from statisticians on the TSC and the DMEC throughout the study. The choice of primary outcome, primary and secondary statistical analyses and regression models was made with the help of consensus processes after long and constructive debate. The use of the full ITT population in the primary effectiveness analysis, with a number of prespecified sensitivity analyses on the PP population, using different ways of handling missing data and using different regression models, allows confidence to be had in the key messages of the study. The statistically significant superiority of usual care and the equivalence of the self-guided and physiotherapy-based programmes (using prespecified equivalence margins) were maintained in all sensitivity analyses, adding to the confidence that this was a correct finding.

In terms of the limitations of the study, the CONSORT diagram shows that only 10% of the total primary care asthma population invited by post to participate in the study responded to the invitation. It is possible that these respondents represent an atypical sample of patients and that the response to the intervention could be different in the total population. The demographic and asthma severity profile of respondents were, however, typical of those of UK asthma populations. We also feel that our recruitment rate was much higher than that achieved by most asthma studies and it is well documented that the evidence base for asthma guidelines comes from studies with tight inclusion and exclusion criteria that result in < 2% of all community-treated patients being eligible. 69 One cannot force patients to enter trials and we feel that changes in the design of the study and the processes used would not improve on the recruitment rate that we achieved. However, there is now the need for evaluated implementation studies to confirm that the benefits seen in our ‘real-world’ but trial-consenting population can be translated to the wider adult asthma population, as the intervention is made available. Such assessments will need to use routinely collected data or minimally intrusive outcome measures if they are to capture outcomes in the wider population.

Approximately 40% of our responders had unimpaired asthma-related QoL and so were ineligible for the study. This is in keeping with previous population-based studies on QoL in asthma, with the majority being found to have impaired health. We feel that those with unimpaired QoL are probably already coping well with their illness and so are not in need of further help from an intervention such as breathing retraining; however, this applies only to a minority of adult asthma sufferers in the UK and so there is potentially a wide use for this intervention in routine practice.

Our study sample size was powered on the primary outcome measure, the AQLQ. For a number of the secondary outcome measures, including asthma attacks, GP consultations and rescue medication prescriptions, we observed numerical, but non-significant or marginally significant, improvements in outcome in the breathing retraining arms compared with the usual-care arm. We cannot say whether these improvements were chance findings or whether lack of statistical power prevented a significant result being seen. This is particularly true of asthma attacks as these occur only in about 1 in 10 patients with mild-to-moderate asthma treated in the community over a single year. Larger studies would be needed to investigate this fully. As all of these factors (consultations, asthma attacks and rescue medication prescriptions) can be assessed through the analysis of routinely collected data, this could be investigated in implementation studies using analysis of anonymised routine data.

Participants in our study were aware of which study arm they had been allocated to. Although this is a limitation, it is clearly not possible to ‘blind’ a participant in such a study to group allocation. The study was, however, ‘observer blinded’, with the research nurses collecting data and the statistical team blinded as to study arm allocation until the statistical report was finalised. The control arm in this study consisted of usual care and so those in the active arms had a greater provision of care. Although it is possible that participants randomised to the usual-care arm were disappointed not to be receiving an active intervention, they were assured that at the completion of the study they would be offered the intervention if it was found to be effective. It is noteworthy that there was a large improvement in QoL in the control arm during the study, probably reflecting the beneficial effects of trial involvement and a ‘regression to the mean’ effect. Those in the DVD arm received audio-visual and printed support materials but there was no provision of additional care, whereas those in the physiotherapy arm underwent three face-to-face visits with a respiratory physiotherapist and also received the printed support materials. Additional professional care can improve health outcomes in a non-specific placebo-like way. In previous studies earlier in the MRC complex intervention framework process, we included additional professional contact in the control arm17,19 and observed similar improvements in the breathing retraining arm compared with the control arm. As this was a Phase IV pragmatic study, we felt that usual care was the appropriate comparator as it provided a better estimate for the health economic evaluation and provides better information for pragmatic implementation. In this situation the new intervention is provided in addition to current usual care. In our study we did see large improvements in the usual-care arm, although less than in the active arms, reflecting the well-described phenomenon whereby patients benefit simply from taking part in research trials.

Our study included only adults. The intervention materials were designed for adults and would not be transferable to younger age groups without modification. There is anecdotal and very limited study evidence that breathing retraining is feasible and effective in adolescents and in younger children. 70 Asthma has a high prevalence in younger people and there is considerable public and parental interest in non-drug interventions for asthma in children. There is therefore still a need for studies to clarify the clinical effectiveness and cost-effectiveness of breathing retraining interventions in younger people. As we excluded children and adolescents, the groups in whom asthma is most prevalent, the average age of our trial population was higher than that in population-based asthma demographic data. It is possible that there was some under-representation of younger adults, who are generally harder to recruit to time-consuming clinical trials, but the demography in our trial was similar to that in other pragmatic community-based adult asthma clinical trials. For example, in the study by Price et al. ,63 the mean age of subjects was 46 years, 56% were aged ≥ 46 years and the overall demographic profile was similar to ours.

We were able to provide the intervention only to English-language speakers in this study. There is therefore a need to transfer the intervention into other languages and to frame it in culturally appropriate forms for ethnic minorities. Consideration of adaptations for people with limited literacy skills or learning disabilities is also needed.

Our study provided part of the self-guided intervention in the form of a DVD, although potentially this could be provided in other formats, including for the internet and smartphone-based platforms. The DVD format was chosen, first, to control access to the programme in the trial and, second, as this format was very commonly used at the time of the study. We had the ability to provide inexpensive DVD players to anyone randomised to this arm of the study who did not already have access to a DVD player, but no one needed this. However, with advances in digital technology and the widespread use of streaming and downloads over the internet, and the wide ownership and use of smartphones, DVDs are now used less frequently. We cannot guarantee that the intervention would provide identical outcomes if provided as an internet-based platform and so this needs to be investigated in further studies. We are currently transferring the content of the DVD to an internet-streamed breathing retraining programme and piloting its use.

Comparison with the results of other studies

The literature on breathing retraining for asthma provided through health professional-delivered face-to-face programmes is reasonably extensive and convincing, such that the UK BTS/SIGN asthma guideline71 and the WHO GINA guideline72 both recommend breathing retraining as an option for patients uncontrolled on standard therapy, on the basis of a meta-analysis of the improvement in asthma in RCTs. Our finding that physiotherapy was superior to usual care was therefore in keeping with this evidence base. Our study extends this evidence as the pragmatic design allows greater external validity and confidence in generalising the findings than those of previous studies, which have often used selected populations and atypical clinical settings as opposed to UK general practice. The lack of improvement in lung function and airway inflammation is in accordance with the majority of the current literature, although there are some small studies (some with methodical problems) that have reported improvements in pathophysiological parameters.

Very few previous studies have investigated audio-visual programmes for breathing retraining that do not involve contact with a health professional and none have compared face-to-face breathing retraining with such programmes. The equivalence that we have demonstrated between face-to-face physiotherapy and the DVD programme is therefore a novel finding and of considerable significance in terms of implementing the intervention. Despite the evidence of effectiveness for breathing retraining and the guideline recommendations, the vast majority of patients who could benefit do not currently have access to such training and many who do have to pay a private practitioner (often unregulated) for it.

Other novel aspects of our study include its size, duration and range of outcome measures and the rigorous health economic evaluation and process evaluations. These evaluations suggest that the widespread provision of the self-guided breathing retraining programme is likely to be well accepted and to reduce NHS costs for asthma patients.

Implications for services and future research

We report a Phase IV study of breathing retraining for adults with asthma who are uncontrolled on their current treatment, which has shown improved outcomes and reduced costs of a breathing retraining programme delivered by DVD. On the basis of this evidence, we feel that this intervention is potentially of benefit to the majority of adults with asthma in the community and can be delivered to them as a low-cost, and logistically viable, self-guided programme that has the potential to reduce NHS costs as well as benefit patients. We therefore feel that implementation studies are now needed, to optimise delivery and to assess the effects of providing the intervention in a wide and community-based programme. This has implications for service delivery.

Specific research issues that should be addressed include assessing the effectiveness of the programme delivered though internet-based and smartphone-based platforms. We are currently in the process of carrying out the technical work necessary for this. Comparisons of the acceptability and effectiveness of different delivery methods are needed. We would hope to make these comparisons and are in discussion with our charity partner, Asthma UK, on how to structure and deliver the intervention most effectively.

Given the effectiveness of the intervention in an adult population, we feel that there is a need for research to be carried out in a paediatric population, among whom there is great interest in such interventions. It is not justifiable to automatically extrapolate the results of asthma trials in older age groups to children, although this has frequently been carried out, often inappropriately. In addition, the framing of the intervention and the language and directions provided will need to be adapted for younger people. As mentioned earlier, there is also a need to frame and translate the intervention for specific groups, such as ethnic minorities, those with health literacy problems and those with learning disabilities.

Funding

The authors acknowledge funding from the UK NIHR HTA Board and the NIHR CRN and support from the NIHR CRN.

Sponsor

University of Southampton.

Data Monitoring and Ethics Committee members

Professor Robert McKinley, Dr Rachel Garrod and Dr Ly-Mee Yu.

Trial Steering Committee members

Professor Sally Singh, Dr Christopher Cates, Professor Christopher Griffiths and Mr John Price.

BREATHE trial general practices

• Adelaide Surgery.

• Alma Medical Practice.

• Barton-on-Sea Surgery.

• Bosmere Medical Practice.

• Brook Lane Surgery.

• Clanfield Surgery.

• Cowplain Family Practice.

• Derby Road Surgery.

• Forest End Surgery.

• Forton Medical Centre.

• Friarsgate Surgery.

• Gosport Medical Centre.

• Heyward Road Surgery.

• Highcliffe Medical Centre.

• Highlands Practice.

• Homewell Practice.

• Kirklands Surgery.

• Lordshill Health Centre.

• New Forest Medical Centre.

• Nightingale Surgery.

• North Baddesley Surgery.

• Old Fire Station Surgery.

• Osborne Practice.

• Park and St Francis Surgery.

• Portsdown Group Practice.

• Ramillies Surgery.

• Regents Park Surgery.

• Ringwood Medical Centre.

• Stoke Road Medical Centre.

• Sunnyside Medical Centre.

• Three Swans Surgery.

• Waterbrook Medical Practice.

• Waterside Medical Practice.

• Woolston Lodge Surgery.

Contributions of authors

Mike Thomas (Professor of Primary Care Research) was the chief investigator of the BREATHE trial. He was responsible for the study conception, provided overall leadership and was the primary care lead.

Anne Bruton (Professor of Respiratory Rehabilitation, NIHR Senior Research Fellow) was a co-applicant and lead physiotherapist for the BREATHE trial. She was responsible for the study conception, provided overall leadership and was the physiotherapy lead.

Paul Little (Professor of Primary Care Research) was a co-applicant and TMG member and was involved in the design of the study.

Stephen Holgate (MRC Clinical Professor of Immunopharmacology and Honorary Consultant Physician) was a co-applicant and TMG member and was involved in the design of the study.

Amanda Lee (Chair in Medical Statistics) was a statistician on the study.

Lucy Yardley (Professor of Health Psychology) was the study lead for the digital intervention development and behaviour change and process evaluation.

Steve George (Emeritus Fellow) was a co-applicant and TMG member and was involved in the design of the study.

James Raftery (Professor of Health Technology Assessment and previously Chairperson of NETSCC and Director of the Wessex Institute) was a TMG member and was the health economic lead for the study.

Jennifer Versnel (Patient Representative and previously Executive Director of Research and Policy at Asthma UK) was a TMG member, was responsible for the study conception and was the PPI co-lead.

David Price (Chair of Primary Care Respiratory Medicine) was a co-applicant and TMG member and was involved in the design of the study.

Ian Pavord (Professor of Respiratory Medicine) was a co-applicant and TMG member and was involved in the design of the study.

Ratko Djukanovic (Professor of Medicine, Director of the Southampton NIHR Respiratory Biomedical Research Unit, Director of the NIHR Southampton Centre for Biomedical Research) was a co-applicant and TMG member and was involved in the design of the study.

Michael Moore (Professor of Primary Health Care Research) was a co-applicant and TMG member and was involved in the design of the study.

Sarah Kirby (Associate Professor) was involved in the process evaluation and qualitative components of the study.

Guiqing Yao (Associate Professor in Health Economics) was involved in the health economic component of the study.

Shihua Zhu (Senior Research Fellow in Health Economics) was involved in the health economic component of the study.

Emily Arden-Close (Lecturer in Health Psychology) was involved in the digital intervention development and the qualitative components of the study.

Manimekalai Thiruvothiyur (Research Assistant) was a statistician on the study.

Frances Webley (BREATHE Trial Manager, SCTU) was responsible for the delivery and co-ordination of the RCT.

Mark Stafford-Watson (Patient Representative) was a TMG member and was the PPI co-lead.

Elizabeth Dixon (BREATHE Trial Manager, SCTU) was responsible for the delivery and co-ordination of the RCT.

Lynda Taylor (BREATHE Trial Co-ordinator, SCTU) was responsible for the delivery and co-ordination of the RCT.

Contribution of other

Denise Gibson (Consultant Respiratory Physiotherapist) was responsible for the physiotherapy intervention design and development.

Publications

Arden-Close E, Teasdale E, Tonkin-Crine S, Pitre N, Stafford-Watson M, Gibson D, et al. Patients’ perceptions of the potential of breathing training for asthma: a qualitative study. Prim Care Respir J 2013;22:449–53. http://dx.doi.org/10.4104/pcrj.2013.00092

Bruton A, Kirby S, Arden-Close E, Taylor L, Webley F, George S, et al. The BREATHE study: Breathing REtraining for Asthma – Trial of Home Exercises. A protocol summary of a randomised controlled trial. Prim Care Respir J 2013;22:PS1–7. http://dx.doi.org/10.4104/pcrj.2013.00047

Data sharing statement

Data included in this report can be obtained by contacting the corresponding author.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health.