Management of Asthma in School age Children On Therapy (MASCOT): a randomised, double-blind, placebo-controlled, parallel study of efficacy and safety

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 05/503/04. The contractual start date was in April 2009. The draft report began editorial review in October 2011 and was accepted for publication in May 2012. 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

none

Copyright statement

© Queen's Printer and Controller of HMSO 2013. This work was produced by Lenney 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.

Background

Respiratory disease has recently been declared a target for improved management by the Department of Health. Although a major focus of this initiative is adult chest disease, especially chronic obstructive pulmonary disease, it has been recognised that asthma is an important unresolved burden, especially in childhood where much asthma originates. The longitudinal studies from Aberdeen1 have shown that 60% of adults with asthma suffered their first symptoms in early childhood, continuing with these through later childhood and into adult life.

Rationale

Asthma remains a common medical condition seen in children in primary care and a frequent cause of medical paediatric hospital admission. It affects one in eight children nationwide, of whom many are prescribed low-dose inhaled corticosteroids (ICSs). 1 When treatment with low-dose ICSs fails to control asthma symptoms or adequately prevent exacerbations, national guidelines2 suggest ensuring compliance, including giving appropriate information about the disease to children and their families, maximising inhaler technique and treating comorbidities such as rhinitis. Once these measures have been established and if asthma remains uncontrolled, step 3 of the national guidelines recommends increasing the treatment. The evidence base at this step of the guidelines is much more limited in children than it is in adults. The reasons for this are that few studies have been undertaken in children and most that have taken place have used inappropriate adult-based outcomes such as lung function measurements, which suffer from assay insensitivity and fail to capture the episodic nature of much of childhood asthma. Pharmaceutical company studies have generally been conducted only as part of their requirements to obtain a licence to market their product. These studies have generally been of short duration and have not added to clinicians' understanding of how and where to use the medications. 3,4 In addition, they have not necessarily selected a representative population because of their tight entry criteria and their intensive study requirements. Such requirements mean that ‘real-life’ compliance and hence outcomes do not occur. In the independent national Dutch study,5 which enrolled patients uncontrolled on low-dose ICSs, three treatment groups were employed: ICSs alone, ICSs in double the dose and ICSs plus a long-acting beta2-agonist. There was essentially no difference in outcome measures between the three treatment groups, probably because the primary outcome measure was lung function (forced expiratory volume in 1 second or FEV₁). Comparing this study with a similar adult study,6 which also used lung function as the primary outcome measure, the mean FEV₁ on entry into the Dutch paediatric study was approximately 89% of that expected for the children's heights and the mean FEV₁ on entry into the adult study was 74% of that expected. It is therefore not surprising that the paediatric study was unable to show any differences between the treatment groups.

There is little scientific evidence on how to treat children with asthma who are not well controlled on low-dose ICS therapy, apart from the limited value of increasing the ICS dose. 7 There is no clinical study evidence showing that, when control is poor in children with asthma, the dose of the inhaled steroid should be increased. We therefore decided not to introduce into this study a treatment limb with a higher ICS dosage. There is anecdotal information, however, from many studies undertaken within the pharmaceutical industry that, when children enter a study that is controlled and double blind in nature, up to 30% of them improve, their symptoms reduce and their lung function increases (Bisgaard H, Professor of Respiratory Paediatrics at the University Hospital in Copenhagen, personal correspondence – during development of paediatric asthma studies). It is therefore surprising that approximately one-third of children receiving ICSs are prescribed high-dose inhaled steroid therapy (≥ 800 μg beclometasone dipropionate or equivalent) or are commenced on ‘add-on’ therapies such as long-acting beta-2 agonists or leukotriene receptor antagonists in addition to low-dose ICSs. Concerns about the safety of high-dose ICSs have been raised in relation to growth impairment,8 hypoglycaemia9 and suppression of the adrenal cortex,10 resulting in warnings on prescribing from the Medicines and Healthcare products Regulatory Agency (MHRA) in the UK. 11 Asthma is a very common condition and the worth of these regimens has not been proven by appropriately devised paediatric studies. The UK guidelines on asthma management2 have been developed in a ‘stepwise’ manner, the amount of medication increasing at each step if symptoms are not controlled; however, as stated above, it may be that childhood asthma differs from that in adults. It seems that children with relatively poorly controlled asthma who exhibit frequent symptoms do not necessarily show abnormal lung function between their periods of symptoms. It is for this reason that our study focused on outcome measures such as exacerbations and quality of life, although spirometric values at the first (T0) and last (T48) visits were measured and recorded.

A study is needed that is simple, is pragmatic (but placebo controlled and double blind), has outcomes that will be of practical benefit to children and will provide evidence for the use of add-on medications in the most cost-effective and efficient way. Since the MASCOT study was commenced the Best Add-on Therapy Giving Effective Response (BADGER) trial has been completed and reported. 12 This study, however, required reversibility or hyper-responsiveness as an entry criterion, which excluded many patients, was short term in nature and focused primarily on symptomatic control as measured by the Childhood Asthma Control Test as opposed to exacerbations.

Potential risks and benefits

The medications used in this study are subject to marketing authorisations and prescribed in accordance with their licensed indications. The management of any symptoms or exacerbations was in accordance with usual clinical practice, and a research worker, either the local principal investigator or research nurse, was available throughout the study to discuss specific issues with individuals concerned. Patients were free to withdraw from the study at any time with no detriment to their future care. All ethical aspects of the study were discussed when informed written consent was obtained. Appropriate patient and family information leaflets (Appendix 1) were developed and were discussed at the screening consultation. Patients and their families were provided with copies of the information sheets and their signed consent/assent forms.

All of the medications have been shown to be efficacious for children with chronic asthma when used appropriately as preventative therapy. The ultimate aims of preventative asthma treatment are the prevention of chronic symptoms and the maintenance of near-normal lung function and normal activity levels with prevention of recurrent acute episodes in order to maximise quality of life. The potential benefit for participants of taking these medications as part of the trial is that they will improve control of their asthma, reducing symptoms and exacerbations and meeting the goals above.

Objective

Trial design

At the time that consent was obtained, the children commenced the run-in part of the study. This was an open run-in for 4 weeks commencing at T–4 (time minus 4 weeks) and continuing until T0 (time zero). If they continued with sufficient symptoms they were then eligible to enter the double-blind randomised study lasting 48 weeks (i.e. to T48).

Study procedure

Data collection

The data were recorded on to standardised CRFs designed collaboratively by the Trial Management Group (TMG). These were returned to the Medicines for Children Research Network Clinical Trials Unit (MCRN CTU) and the data entered onto a validated study database (MACROĸ™ v3.0, InferMed, London, UK) by trained staff. Confirmation of patient existence was by receipt of a fully signed consent form. Each CRF was checked for adherence to the trial protocol and for missing and/or erroneous values. Discrepancies were raised and queried with study sites to enable the correct data to be obtained or reasons, where possible, for missing data/errors.

Outcomes

Sample size calculation

Assuming that the number of asthma exacerbations requiring treatment with oral corticosteroids was a Poisson random variable, the sample size for the trial was estimated using the following formula as described by Friede and Schmidli:14

where n_c is the number of patients in the control arm, λ_c is the control group rate, t is the length of follow-up and θ* is the RR. The equation does not allow for overdispersion, which would lead to an inflated sample size.

Data from the UK General Practice Research Database for 1032 children who had at least one course of oral steroids in the previous 12 months were used to estimate a mean rate of 1.5 per year with variance 1.02 and dispersion parameter 0.68. 15 This was the best available estimate of exacerbation rate and dispersion parameter at the time of designing the MASCOT trial. However, as the MASCOT randomised population would have had their inhaler technique corrected, and other population differences may alter the exacerbation rate, an initial ‘target’ sample size was estimated with the intention of undertaking an internal blinded pilot after the first 75 children had been randomised and completed their 24-week follow-up assessment to check parameter assumptions and adjust the sample size if required. Recruitment rates and percentage of children randomised after the 4-week run-in were also to be closely monitored.

Bonferroni's adjustment was used in the sample size calculation (1.7% two-sided significance level) to allow for the three primary treatment comparisons of interest. To have 80% power to detect, as significant, at least a 30% reduction in exacerbation rate (from 1.5 per year to 1.05 per year, equivalent to a RR of 0.7) and allowing for a loss to follow up of 10%, 147 patients per group were required. Our preliminary target number to be randomised was 150 children per treatment group (450 in total). This number would also give > 99% power to detect, as significant, a difference of 0.5 points between treatment groups on the PAQLQ(S), with assumed standard deviation (SD) 0.71.

Randomisation and blinding

Randomisation code lists were generated (by an individual at the MCRN CTU who was not involved with the MASCOT trial) with the software package Stata (Release 9, StataCorp LP, College Station, TX, USA) using block randomisation with variable block length, stratified by secondary care centre, with allocation to the three treatment arms in the ratio of 1 : 1 : 1. The pharmacy at each secondary care centre held the randomisation list for that centre with treatment allocations labelled A, B or C. After determining a patient's treatment allocation from the list, the pharmacist selected an appropriate treatment pack and removed a serrated label showing A, B or C before dispensing to the patient. Study drugs were identical in appearance and identically packaged, with all patients, clinicians and trial personnel blinded to treatment allocation throughout. Patients could be unblinded at any time if clinically required using emergency contact details for a 24-hour pharmacy. All patients were routinely unblinded when they completed the 48-week study or when they withdrew prematurely. Routine unblinding was undertaken by a member of the MCRN CTU, who was not involved with the MASCOT study, who then contacted the child's GP. The MASCOT trial clinician and research nurse were not aware of the unblinding information to minimise the potential of unblinding other patients still in the study.

Statistical methods

Health economics

The MASCOT economic analysis focused on determining the differences in the patient pathways between the three groups in terms of their costs and benefits. The intention was therefore that the analysis would take a number of different forms. It was intended to compute incremental cost-effectiveness ratios (ICERs) for (B) inhaled fluticasone propionate 100 μg and salmeterol 50 μg twice daily (combination inhaler) plus placebo tablet once daily and (C) inhaled fluticasone propionate 100 μg twice daily plus montelukast 5-mg tablet once daily compared with the base case of (A) inhaled fluticasone propionate 100 μg twice daily plus placebo tablet once daily, and for (B) compared with (C).

Protocol amendments

All protocol amendments are summarised in Appendix 5.

Early trial closure

Recruitment rates were closely monitored throughout the study, which highlighted within 8 months of the first patient being registered that additional strategies and additional centres would be required to improve recruitment and achieve the target of 450 patients randomised. Site visits were undertaken from October 2009 and a number of new strategies were developed through discussion with sites (see Appendix 3). A funding extension application was submitted to the NIHR HTA programme in February 2010 requesting an extension of recruitment time together with funding for new sites. One of the differences noted about recruitment centres that were more successful than others was how well integration had occurred between staff within the newly formed Medicines for Children Research Network (MCRN), Primary Care Research Network (PCRN) and Comprehensive Local Research Network (CLRN). If the study was to succeed it would also have been necessary to open other recruitment centres and these had been carefully selected based on the knowledge we had accrued through discussions and particularly the ability to liaise and work effectively between secondary and primary care. It was recognised that some of the initial recruitment sites would have to be closed. Funding was not granted, however, and the study was closed down prematurely on 24 June 2010. For those patients who were already randomised, follow-up was to continue to T48 or the end of January 2011, whichever was the earliest, as January 2011 was the date that the medications expired. Data cleaning and site close-down visits took place between February and July 2011, with final analyses undertaken during August and September 2011.

Participant flow and recruitment

The first patient was registered on 27 January 2009, the first patient was randomised on 19 May 2009, the last patient was registered on 25 June 2010 and the last patient was randomised on 24 June 2010. Table 3 shows all 13 recruiting centres, the date each site was initiated, the original target recruitment figure, the number of participants registered, the number of participants randomised and the dates of the first and last randomisation. All centres registered at least one patient and 12 centres randomised at least one patient.

A total of 166 patients were registered at T–4 and 63 (38%) of these were eligible and consented to be randomised at T0. The percentage randomised at each centre ranged between 20% and 58% for those that randomised at least one patient.

Figure 1 provides a CONSORT flow diagram showing the numbers of participants recruited and randomly assigned to the three trial arms.

FIGURE 1.

CONSORT flow diagram. SABA, short-acting beta-2 agonist. All patients who did not reach the minimum of 43 weeks because of early closure of the trial have primary outcome data up to at least T24.

Baseline comparability of randomised groups

A summary of the baseline characteristics for all randomised participants is given in Table 4. Overall there was a preponderance of males with the major comorbid condition being eczema, neither of which was unexpected. The male-to-female ratio in children with asthma is approximately 2 : 1 and there is a strong association between eczema and asthma in children. The age of the children ranged from 6.50 to 14.67 years (average 10.39 years).

The distribution of characteristics is similar across treatment groups but a higher percentage of males were randomised to fluticasone alone and a higher percentage of children with respiratory and dermatological abnormalities were randomised to fluticasone plus salmeterol. Because of the small number of children recruited this probably occurred by chance but could suggest that patients randomised to fluticasone plus salmeterol had more severe asthma.

No ineligible patients were randomised.

Unblinding of randomised treatments

No patient required unblinding from randomised treatment other than at the end of the study. The process was described in Chapter 2, Randomisation and blinding.

Protocol deviations

Protocol deviations were classified as major or minor according to a preplanned classification system outlined in the SAP (see Appendix 2). There were 18 minor protocol deviations reported for 17 patients (three randomised to fluticasone, five randomised to fluticasone plus salmeterol, two randomised to fluticasone plus montelukast and seven who were not randomised) (Table 5). There were no major protocol deviations. Eleven of the protocol deviations were due to visits that occurred outside the predefined time interval, one of which led to missing data. One protocol deviation occurred because two different parents completed the quality of life questionnaire at two different visits. Six protocol deviations were related to the spirometry test undertaken at T0 or T48.

Efficacy outcomes

Secondary outcomes

Quality of life of children as measured by the Paediatric Asthma Quality of Life Questionnaire with Standardised Activities

A total of 37 children completed the quality of life questionnaire at baseline and at 48 weeks' follow-up. The results show an improvement in mean scores for activity limitations, symptoms and emotional function across all treatment groups at 48 weeks, with the largest improvement in mean score for fluticasone plus montelukast across all domains (Table 9). However, the pair-wise treatment comparison results from ANCOVA show that none of the mean differences adjusted for baseline score is statistically significant. The 24-week analysis results based on 50 patients show an improvement in mean scores for activity limitations, symptoms and emotional function across all treatment groups (Table 10). The largest improvement was seen for fluticasone plus salmeterol across all domains but none of the mean differences adjusted for baseline score is statistically significant.

TABLE 9 - Quality of life of children measured using the PAQLQ(S) over 48 weeks

A, futicasone; B, futicasone plus salmeterol; C, futicasone plus montelukast.

One patient with missing question on the activity limitations domain was excluded from the domain score calculation but included in the total score as the missing data made up < 10% of the total.

Scores for PAQLQ(S) range from 1 to 7; lower scores are worse.

Domain	Fluticasone			Fluticasone plus salmeterol			Fluticasone plus montelukast			Fluticasone vs futicasone plus salmeterol	Fluticasone vs futicasone plus montelukast	Fluticasone plus salmeterol vs futicasone plus montelukast
	Baseline mean (SD), range	T48 mean (SD), range	Change mean (SD), range	Baseline mean (SD), range	T48 mean (SD), range	Change mean (SD), range	Baseline mean (SD), range	T48 mean (SD), range	Change mean (SD), range	Adjusted mean difference (95% CI), p-value	Adjusted mean difference (95% CI), p-value	Adjusted mean difference (95% CI), p-value
Activity limitations (nA = 10, nB = 14,anC = 12)	4.76 (0.91), 3.60 to 6.00	6.20 (1.07), 3.40 to 7.00	1.44 (1.00), −0.20 to 3.40	4.39 (1.30), 2.60 to 6.60	5.60 (1.47), 3.00 to 7.00	1.21 (2.12), −2.80 to 4.40	4.25 (1.47), 1.80 to 6.80	6.30 (0.85), 4.40 to 7.00	2.05 (1.61), −0.60 to 4.20	0.59 (−0.43 to 1.61), p = 0.25	−0.12 (−1.18 to 0.94), p = 0.82	−0.70 (−1.67 to 0.26), p = 0.16
Symptoms (nA = 10, nB = 15, nC = 12)	4.90 (0.92), 3.70 to 6.30	6.28 (0.84), 4.40 to 7.00	1.38 (0.93), 0.00 to 3.10	4.35 (1.51), 1.30 to 6.80	5.30 (1.77), 1.50 to 7.00	0.95 (2.17), −2.20 to 5.70	4.32 (1.63), 1.50 to 6.70	6.23 (0.92), 4.70 to 7.00	1.91 (1.79), −0.80 to 4.40	0.90 (−0.22 to 2.02), p = 0.11	−0.03 (−1.20 to 1.15), p = 0.96	−0.93 (−1.98 to 0.12), p = 0.08
Emotional function (nA = 10, nB = 15, nC = 12)	5.31 (0.82), 3.50 to 6.25	6.40 (0.86), 4.25 to 7.00	1.09 (0.62), 0.13 to 2.13	4.33 (1.68), 1.75 to 7.00	5.58 (1.62), 2.13 to 7.00	1.25 (1.72), −1.50 to 4.38	4.79 (1.54), 2.00 to 6.75	6.42 (0.83), 4.88 to 7.00	1.63 (1.66), −0.63 to 4.13	0.50 (−0.48 to 1.49), p = 0.31	−0.19 (−1.19 to 0.82), p = 0.71	−0.69 (−1.60 to 0.22), p = 0.13
Total (nA = 10, nB = 15, nC = 12)	5.01 (0.64), 3.91 to 5.78	6.30 (0.88), 4.13 to 7.00	1.29 (0.61), 0.22 to 2.52	4.34 (1.39), 2.04 to 6.83	5.44 (1.56), 2.26 to 7.00	1.10 (1.86), −2.09 to 4.96	4.47 (1.50), 1.74 to 6.65	6.31 (0.85), 4.78 to 7.00	1.84 (1.66), −0.70 to 4.26	0.73 (−0.29 to 1.74), p = 0.15	−0.12 (−1.17 to 0.94), p = 0.82	−0.84 (−1.78 to 0.10), p = 0.08

TABLE 10 - Quality of life of children measured using the PAQLQ(S) over 24 weeks

A, futicasone; B, futicasone plus salmeterol; C, futicasone plus montelukast.

One patient with missing question on the activity limitations domain was excluded from the domain score calculation but included in the total score as the missing data made up < 10% of the total.

Scores for PAQLQ(S) range from 1 to 7; lower scores are worse.

Domain	Fluticasone			Fluticasone plus salmeterol			Fluticasone plus montelukast			Fluticasone vs futicasone plus salmeterol	Fluticasone vs futicasone plus montelukast	Fluticasone plus salmeterol vs futicasone plus montelukast
	Baseline mean (SD), range	T24 mean (SD), range	Change mean (SD), range	Baseline mean (SD), range	T24 mean (SD), range	Change mean (SD), range	Baseline mean (SD), range	T24 mean (SD), range	Change mean (SD), range	Adjusted mean difference (95% CI), p-value	Adjusted mean difference (95% CI), p-value	Adjusted mean difference (95% CI), p-value
Activity limitations (nA = 17, nB = 15,anC = 17)	4.52 (1.18), 2.20 to 6.00	5.60 (1.50), 2.20 to 7.00	1.08 (1.02), −0.40 to 3.60	4.40 (1.26), 2.60 to 6.60	6.00 (0.91), 4.40 to 7.00	1.60 (1.39), −0.80 to 4.20	4.21 (1.32), 1.80 to 6.80	5.75 (1.48), 2.40 to 7.00	1.54 (1.33), −0.60 to 3.80	−0.47 (−1.28 to 0.35), p = 0.26	−0.33 (−1.12 to 0.47), p = 0.41	0.14 (−0.68 to 0.96), p = 0.73
Symptoms (nA = 17, nB = 16, nC = 17)	4.64 (1.15), 2.60 to 6.60	5.49 (1.31), 2.70 to 7.00	0.85 (0.97), −1.00 to 2.90	4.45 (1.51), 1.30 to 6.80	5.91 (1.18), 2.70 to 7.00	1.46 (1.42), −1.10 to 3.60	4.24 (1.42), 1.50 to 6.70	5.46 (1.67), 1.50 to 7.00	1.22 (1.42), −1.40 to 3.20	−0.52 (−1.34 to 0.29), p = 0.20	−0.20 (−1.01 to 0.61), p = 0.63	0.33 (−0.49 to 1.14), p = 0.43
Emotional function (nA = 17, nB = 16, nC = 17)	5.21 (1.13), 3.38 to 6.88	5.86 (1.38), 1.50 to 7.00	0.65 (1.11), −1.88 to 2.63	4.45 (1.70), 1.75 to 7.00	6.22 (1.04), 3.75 to 7.00	1.77 (1.65), 0 to 4.88	4.79 (1.32), 2.00 to 6.75	5.93 (1.66), 1.38 to 7.00	1.14 (1.41), −1.50 to 3.63	−0.73 (−1.61 to 0.15), p = 0.10	−0.28 (−1.13 to 0.57), p = 0.51	0.45 (−0.41 to 1.31), p = 0.30
Total (nA = 17, nB = 16, nC = 17)	4.81 (1.01), 2.96 to 6.52	5.64 (1.29), 2.17 to 7.00	0.83 (0.90), −1.09 to 2.96	4.43 (1.39), 2.04 to 6.83	6.05 (1.01), 3.43 to 7.00	1.62 (1.38), −0.52 to 3.74	4.43 (1.29), 1.74 to 6.65	5.69 (1.59), 1.65 to 7.00	1.26 (1.30), −0.83 to 3.13	−0.63 (−1.42 to 0.15), p = 0.11	−0.27 (−1.05 to 0.50), p = 0.48	0.36 (−0.42 to 1.14), p = 0.36

Quality of life of caregivers as measured by the Paediatric Asthma Caregiver's Quality of Life Questionnaire

A total of 38 caregivers completed the quality of life questionnaire at baseline and at 48 weeks. The results show an improvement in mean scores for activity limitations and emotional function across all treatment groups at 48 weeks (Table 11). In support of the PAQLQ(S) analysis, the largest improvement in mean score at 48 weeks is seen for fluticasone plus montelukast across all domains. The pair-wise treatment comparison results from ANCOVA, however, show that none of the mean differences adjusted for baseline score is statistically significant. Similar results were obtained using data up to 24 weeks (Table 12).

TABLE 11 - Quality of life of caregivers measured using the PACQLQ over 48 weeks

A, futicasone; B, futicasone plus salmeterol; C, futicasone plus montelukast.

Scores for PAQLQ(S) range from 1 to 7; lower scores are worse.

Domain	Fluticasone			Fluticasone plus salmeterol			Fluticasone plus montelukast			Fluticasone vs futicasone plus salmeterol	Fluticasone vs futicasone plus montelukast	Fluticasone plus salmeterol vs futicasone plus montelukast
	Baseline mean (SD), range	T48 mean (SD), range	Change mean (SD), range	Baseline mean (SD), range	T48 mean (SD), range	Change mean (SD), range	Baseline mean (SD), range	T48 mean (SD), range	Change mean (SD), range	Adjusted mean difference (95% CI), p-value	Adjusted mean difference (95% CI), p-value	Adjusted mean difference (95% CI), p-value
Activity limitations (nA = 11, nB = 15, nC = 12)	4.75 (1.34), 2.50 to 7.00	6.07 (1.16), 3.50 to 7.00	1.32 (1.41), −1.25 to 3.50	5.08 (1.43), 1.25 to 6.75	6.30 (0.99), 4.25 to 7.00	1.22 (1.30), −1.00 to 3.75	3.85 (1.53), 1.75 to 6.50	5.81 (1.41), 3.50 to 7.00	1.96 (1.21), 0.50 to 4.75	−0.09 (−0.92 to 0.75), p = 0.83	−0.13 (−1.03 to 0.77), p = 0.77	−0.04 (−0.90 to 0.83), p = 0.93
Emotional function (nA = 11, nB = 15, nC = 12)	4.44 (1.09), 2.22 to 6.22	5.80 (1.56), 1.67 to 7.00	1.36 (1.05), −0.56 to 3.00	4.35 (1.35), 1.00 to 6.00	5.73 (1.27), 2.33 to 7.00	1.38 (1.45), −0.56 to 4.44	4.05 (1.49), 2.00 to 6.00	5.67 (1.27), 3.78 to 7.00	1.62 (1.48), −1.11 to 4.11	0.02 (−0.95 to 0.99), p = 0.97	−0.07 (−1.10 to 0.96), p = 0.89	−0.08 (−1.04 to 0.87), p = 0.86
Total (nA = 11, nB = 15, nC = 12)	4.54 (1.08), 2.46 to 6.46	5.88 (1.42), 2.23 to 7.00	1.34 (1.07), −0.23 to 3.15	4.57 (1.28), 1.08 to 6.15	5.91 (1.13), 3.08 to 7.00	1.34 (1.25), −0.38 to 4.00	3.99 (1.43), 2.00 to 6.15	5.71 (1.25), 3.77 to 7.00	1.72 (1.27), −0.38 to 4.08	−0.01 (−0.87 to 0.85), p = 0.99	−0.12 (−1.04 to 0.80), p = 0.79	−0.12 (−0.97 to 0.74), p = 0.78

TABLE 12 - Quality of life of caregivers measured using the PACQLQ over 24 weeks

A, futicasone; B, futicasone plus salmeterol; C, futicasone plus montelukast.

Scores for PAQLQ(S) range from 1 to 7; lower scores are worse.

Domain	Fluticasone			Fluticasone plus salmeterol			Fluticasone plus montelukast			Fluticasone vs fluticasone plus salmeterol	Fluticasone vs fluticasone plus montelukast	Fluticasone plus salmeterol vs fluticasone plus montelukast
	Baseline mean (SD), range	T24 mean (SD), range	Change mean (SD), range	Baseline mean (SD), range	T24 mean (SD), range	Change mean (SD), range	Baseline mean (SD), range	T24 mean (SD), range	Change mean (SD), range	Adjusted mean difference (95% CI), p-value	Adjusted mean difference (95% CI), p-value	Adjusted mean difference (95% CI), p-value
Activity limitations (nA = 17, nB = 14, nC = 16)	4.49 (1.49), 2.00 to 7.00	6.00 (1.31), 3.25 to 7.00	1.51 (1.29), −1.25 to 3.50	5.36 (0.99), 3.25 to 6.75	6.04 (1.24), 3.25 to 7.00	0.68 (1.33), −2.00 to 3.25	4.09 (1.64), 1.75 to 7.00	5.92 (1.07), 4.00 to 7.00	1.83 (1.55), −2.25 to 4.25	0.30 (−0.53 to 1.12), p = 0.47	−0.07 (−0.85 to 0.70), p = 0.85	−0.37 (−1.23 to 0.49), p = 0.39
Emotional function (nA = 17, nB = 16, nC = 16)	4.22 (1.36), 1.67 to 6.78	5.66 (1.42), 2.33 to 7.00	1.44 (1.27), −0.56 to 4.44	4.35 (1.31), 1.00 to 6.00	5.42 (1.82), 1.78 to 7.00	1.07 (1.22), −1.78 to 2.89	4.32 (1.43), 2.00 to 6.33	5.63 (1.18), 3.00 to 7.00	1.31 (1.57), −3.33 to 4.11	0.32 (−0.57 to 1.20), p = 0.47	0.09 (−0.79 to 0.98), p = 0.83	−0.23 (−1.12 to 0.67), p = 0.62
Total (nA = 17, nB = 16, nC = 16)	4.30 (1.33), 2.00 to 6.85	5.76 (1.36), 2.62 to 7.00	1.46 (1.18), −0.77 to 4.15	4.54 (1.24), 1.08 to 6.15	5.57 (1.66), 2.28 to 7.00	1.03 (1.21), −1.85 to 2.85	4.25 (1.41), 2.00 to 6.54	5.72 (1.08), 3.54 to 7.00	1.47 (1.51), −3.00 to 4.08	0.32 (−0.51 to 1.15), p = 0.44	0.02 (−0.81 to 0.85), p = 0.96	−0.30 (−1.15 to 0.54), p = 0.47

Time from randomisation to first exacerbation requiring treatment with a short course of oral corticosteroids

In total, 63 randomised patients could be included in the analysis of time to first exacerbation (Table 13) [median length of follow-up 47 weeks (range 0 to 57 weeks, IQR 33 to 49 weeks) for the 49 censored patients]. Overall, there were 14 (22.2%) patients who had at least one exacerbation event, with most events in the fluticasone plus salmeterol group [seven events (30.4%)] and least events in the fluticasone plus montelukast group [three events (14.3%)]. No statistically significant difference in time to first exacerbation could be detected between the three treatment groups with the log-rank test (Figure 2, χ² = 1.90, 2df, p = 0.39). Because of the limited data, wide CIs for pair-wise HRs (Table 14) make it impossible to draw any meaningful conclusions.

TABLE 13 - Time to first exacerbation

Treatment	No. of patients	No. of events (%)	Median exacerbation-free time	24-week exacerbation-free probability (95% CI)
Fluticasone	19	4 (21.1)	Not estimable	0.79 (0.52 to 0.91)
Fluticasone plus salmeterol	23	7 (30.4)		0.80 (0.55 to 0.92)
Fluticasone plus montelukast	21	3 (14.3)		0.83 (0.57 to 0.94)
Total	63	14 (22.2)

FIGURE 2.

Kaplan–Meier survival curves for time to first exacerbation by treatment.

TABLE 14 - Hazard ratios for time to first exacerbation

Comparison	HR (95% CI)
Fluticasone vs fluticasone plus salmeterol	0.63 (0.19 to 2.08)
Fluticasone vs fluticasone plus montelukast	1.52 (0.34 to 6.70)
Fluticasone plus salmeterol vs fluticasone plus montelukast	2.37 (0.68 to 8.20)

Number of school days missed due to respiratory problems

A total of 37 children with data available up to 48 weeks were included in the ITT analysis. Fewer children missed at least 1 day of school over 48 weeks on fluticasone plus montelukast (18.2%) than on fluticasone (63.6%) and fluticasone plus salmeterol (60%) (Table 15); however, this pattern is not supported by the 24-week data, based on 54 patients, which show that a similar percentage of patients missed at least 1 day across treatment groups (Table 16). The RRs for number of school days missed, estimated from Poisson regression, are not statistically significant (Table 17) and wide CIs prevent useful conclusions for the 48-week and 24-week analyses.

TABLE 15 - Number of school days missed over 48 weeks (ITT analysis)

Patients included if total exposure time is within the range of 43–53 weeks.

	Fluticasone (n = 11)	Fluticasone plus salmeterol (n = 15)	Fluticasone plus montelukast (n = 11)	Total (n = 37)
Children with at least 1 school day missed, n (%)	7 (63.6)	9 (60.0)	2 (18.2)	18 (48.6)
Total no. of school days missed over 48 weeks,a n (%)
0	4 (36.4)	6 (40.0)	9 (81.8)	19 (51.4)
1	2 (18.2)	1 (6.7)	0 (0.0)	3 (8.1)
2	1 (9.1)	3 (20.0)	0 (0.0)	4 (10.8)
3	1 (9.1)	1 (6.7)	0 (0.0)	2 (5.4)
4	1 (9.1)	0 (0.0)	0 (0.0)	1 (2.7)
5	1 (9.1)	0 (0)	1 (9.1)	2 (5.4)
6	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
7	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
8	1 (9.1)	0 (0.0)	0 (0.0)	1 (2.7)
9	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
10	0 (0.0)	3 (20.0)	0 (0.0)	3 (8.1)
11	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
12	0 (0.0)	0 (0.0)	1 (9.1)	1 (2.7)
13	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
14	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
15	0 (0.0)	1 (6.7)	0 (0.0)	1 (2.7)
Median (IQR)	1 (0 to 4)	2 (0 to 10)	0 (0 to 0)	0 (0 to 3)
Range	0 to 8	0 to 15	0 to 12	0 to 15
Total exposure time (weeks)
Mean (SD)	48.04 (1.06)	48.09 (1.96)	48.38 (2.46)	48.16 (1.87)
Range	46.14 to 50.00	43.00 to 50.00	43.00 to 53.00	43.00 to 53.00

TABLE 16 - Number of school days missed over 24 weeks (ITT analysis)

Patients included if total exposure time is within the range of 19–29 weeks.

	Fluticasone (n = 18)	Fluticasone plus salmeterol (n = 17)	Fluticasone plus montelukast (n = 19)	Total (n = 54)
Children with at least 1 school day missed, n (%)	6 (33.3)	8 (47.1)	6 (31.6)	20 (37.0)
Total no. of school days missed over 24 weeks,a n (%)
0	12 (66.7)	9 (52.9)	13 (68.4)	34 (63.0)
1	0 (0.0)	3 (17.6)	0 (0.0)	3 (5.6)
2	2 (11.1)	2 (11.8)	3 (15.8)	7 (13.0)
3	2 (11.1)	2 (11.8)	0 (0.0)	4 (7.4)
4	1 (5.6)	0 (0.0)	0 (0.0)	1 (1.9)
5	0 (0.0)	1 (5.9)	1 (5.3)	2 (3.7)
6	1 (5.6)	0 (0.0)	0 (0.0)	1 (1.9)
7	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
8	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
9	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
10	0 (0.0)	0 (0.0)	2 (10.5)	2 (3.7)
Median (IQR)	0 (0 to 2)	0 (0 to 2)	0 (0 to 2)	0 (0 to 2)
Range	0 to 6	0 to 5	0 to 10	0 to 10
Total exposure time (weeks)
Mean (SD)	24.24 (1.03)	24.78 (1.43)	24.45 (1.27)	24.48 (1.25)
Range	22.00 to 26.00	22.29 to 29.00	22.00 to 28.14	22.00 to 29.00

TABLE 17 - Relative treatment effects from Poisson regression model adjusted for overdispersion: number of school days missed

A, futicasone; B, futicasone plus salmeterol; C, futicasone plus montelukast.

The scale parameter was estimated as the square root of Pearson's chi-squared statistic/df.

p-value = Wald chi-square p-value.

Analysis set	No. of patients	Time point (weeks)	Scale parametera	Fluticasone vs fluticasone plus salmeterol			Fluticasone vs fluticasone plus montelukast			Fluticasone plus salmeterol vs futicasone plus montelukast
				RR	95% CI	p-valueb	RR	95% CI	p-valueb	RR	95% CI	p-valueb
ITT	nA = 11, nB = 15, nC = 11	48	2.58	0.60	0.17 to 2.05	0.41	1.42	0.29 to 7.07	0.67	2.39	0.59 to 9.71	0.22
	nA = 18, nB = 17, nC = 19	24	2.00	1.07	0.30 to 3.84	0.91	0.69	0.22 to 2.12	0.51	0.64	0.20 to 2.05	0.45
Per protocol	nA = 9, nB = 11, nC = 8	48	2.72	0.69	0.14 to 3.33	0.65	0.93	0.15 to 5.66	0.94	1.35	0.27 to 6.68	0.72
	nA = 15, nB = 13, nC = 16	24	2.00	1.97	0.41 to 9.46	0.40	0.74	0.24 to 2.31	0.61	0.38	0.08 to 1.68	0.20

Number of hospital admissions due to respiratory problems

Tables 18 and 19 show the number of hospital admissions over 48 weeks and 24 weeks respectively. Two hospital admissions occurred in two patients on fluticasone plus salmeterol over 48 weeks (see Table 18). Analysis of the number of admissions with Poisson regression was not possible because of sparse data and no conclusions can be drawn (Table 20). For the patients with data up to 24 weeks, four (7.4%) patients required at least one hospital admission, with a similar frequency across treatment groups (see Table 19). The CIs for the 24-week RRs estimated from Poisson regression (see Table 20) are very wide and include relative effects in both directions.

TABLE 18 - Number of hospital admissions over 48 weeks (ITT analysis)

Patients included if total exposure time is within the range of 43–53 weeks.

	Fluticasone (n = 11)	Fluticasone plus salmeterol (n = 15)	Fluticasone plus montelukast (n = 11)	Total (n = 37)
Children with at least one hospital admission, n (%)	0 (0.0)	2 (13.3)	0 (0.0)	2 (5.4)
Total no. of hospital admissions over 48 weeks,a n (%)
0	11 (100.0)	13 (86.7)	11 (100.0)	35 (94.6)
1	0 (0.0)	2 (13.3)	0 (0.0)	2 (5.4)
Total exposure time (weeks)
Mean (SD)	48.04 (1.06)	8.09 (1.96)	48.38 (2.46)	48.16 (1.87)
Range	46.14 to 50.00	43.00 to 50.00	43.00 to 53.00	43.00 to 53.00

TABLE 19 - Number of hospital admissions over 24 weeks (ITT analysis)

Patients included if total exposure time is within the range of 19 and 29 weeks.

	Fluticasone (n = 18)	Fluticasone plus salmeterol (n = 17)	Fluticasone plus montelukast (n = 19)	Total (n = 54)
Children with at least one school day missed, n (%)	2 (11.1)	1 (5.9)	1 (5.3)	4 (7.4)
Total no. of hospital admissions over 48 weeks,a n (%)
0	16 (88.9)	16 (94.1)	18 (94.7)	50 (92.6)
1	1 (5.6)	1 (5.9)	1 (5.3)	3 (5.6)
2	1 (5.6)	0 (0.0)	0 (0.0)	1 (1.9)
Total exposure time (weeks)
Mean (SD)	24.24 (1.03)	24.78 (1.43)	24.45 (1.27)	24.48 (1.25)
Range	22.00 to 26.00	22.29 to 29.00	22.00 to 28.14	22.00 to 29.00

TABLE 20 - Relative treatment effects from Poisson regression model adjusted for overdispersion: number of hospital admissions

A, fluticasone; B, fluticasone plus salmeterol; C, fluticasone plus montelukast.

The scale parameter was estimated as the square root of Pearson's chi-squared statistic/df.

p-value = Wald chi-square p-value.

Unable to calculate RRs in the 48-week ITT and per-protocol analyses as there were only two patients (from the fluticasone plus salmeterol group) who had any hospital admissions throughout the duration of the trial.

Analysis set	No. of patients	Time point	Scale parametera	Fluticasone vs fluticasone plus salmeterol			Fluticasone vs fluticasone plus montelukast			Fluticasone plus salmeterol vs fluticasone plus montelukast
				RR	95% CI	p-valueb	RR	95% CI	p-valueb	RR	95% CI	p-valueb
ITT	nA = 11, nB = 15, nC = 11	48 weeksc	–	–	–	–	–	–	–	–	–	–
	nA = 18, nB = 17, nC = 19	24 weeks	1.08	2.90	0.25 to 33.11	0.39	3.19	0.28 to 36.51	0.35	1.10	0.06 to 21.79	0.95
Per protocol	nA = 9, nB = 11, nC = 8	48 weeksc	–	–	–	–	–	–	–	–	–	–
	nA = 15, nB = 13, nC = 16	24 weeks	1.14	1.77	0.12 to 27.21	0.68	2.15	0.14 to 33.06	0.58	1.22	0.05 to 28.50	0.90

Amount of rescue beta-2 agonist therapy prescribed for asthma symptoms

A total of 23 out of 37 (62.2%) children were prescribed at least one beta-2 agonist over 48 weeks of follow-up (Table 21), with a median total dose per patient of 24,000 μg. In total, 28 out of 53 (52.8%) children were prescribed at least one beta-2 agonist over 24 weeks of follow-up (Table 22), with a median total dose per patient of 20,000 μg. There was no significant difference in the mean amount prescribed between the three groups from ANOVA at 48 weeks (F-value = 0.96, 2df, p = 0.39) or 24 weeks (F-value = 1.2, 2df, p = 0.31) (Table 23).

TABLE 21 - Number of beta-2 agonists prescribed over 48 weeks

For one patient included in the ITT analysis set, the number of beta-2 agonists prescribed between the T8 and T24 visits was missing.

Note that one patient on fluticasone plus salmeterol had 28 beta-2 agonists prescribed between T0 and T8.

	Fluticasone (n = 11)	Fluticasone plus salmeterol (n = 14a)	Fluticasone plus montelukast (n = 12)	Total (n = 37)
Children with at least one beta-2 agonist prescribed, n (%)	6 (54.5)	10 (71.4)	7 (58.3)	23 (62.2)
Total no. of beta-2 agonists prescribed over 48 weeks,a n (%)
0	5 (45.5)	4 (28.6)	5 (41.7)	14 (37.8)
1	0 (0.0)	3 (21.4)	1 (8.3)	4 (10.8)
2	0 (0.0)	1 (7.1)	1 (8.3)	2 (5.4)
3	2 (18.2)	1 (7.1)	3 (25.0)	6 (16.2)
4	2 (18.2)	0 (0.0)	0 (0.0)	2 (5.4)
5	1 (9.1)	1 (7.1)	0 (0.0)	2 (5.4)
6	0 (0.0)	2 (14.3)	1 (8.3)	3 (8.1)
7	1 (9.1)	0 (0.0)	1 (8.3)	2 (5.4)
8	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
9	0 (0.0)	1 (7.1)	0 (0.0)	1 (2.7)
40	0 (0.0)	1 (7.1)	0 (0.0)	1 (2.7)
Median (IQR)	3 (0 to 4)	1.5 (0 to 6)	1.5 (0 to 3)	2 (0 to 4)
Range	0 to 7	0 to 40	0 to 7	0 to 40
Amount of beta-2 agonists prescribed over 48 weeksa (µg)
Median (IQR)	36,000 (0 to 80,000)	22,000 (0 to 120,000)	22,000 (0 to 60,000)	24,000 (0 to 80,000)
Range	0 to 140,000	0 to 800,000	0 to 140,000	0 to 800,000

TABLE 22 - Number of beta-2 agonists prescribed over 24 weeks

One patient who is included in the ITT analysis set had missing number of beta-2 agonists prescribed between the T8 and T24 visits.

Note that one patient on fluticasone plus salmeterol had 28 beta-2 agonists prescribed between T0 and T8.

	Fluticasone (n = 18)	Fluticasone plus salmeterol (n = 16a)	Fluticasone plus montelukast (n = 19)	Total (n = 53)
Children with at least one beta-2 agonist prescribed, n (%)	9 (50.0)	10 (62.5)	9 (47.4)	28 (52.8)
Total no. of beta-2 agonists prescribed over 24 weeks,a n (%)
0	9 (50.0)	6 (37.5)	10 (52.6)	25 (47.2)
1	2 (11.1)	4 (25.0)	4 (21.1)	10 (18.9)
2	4 (22.2)	4 (25.0)	4 (21.1)	12 (22.6)
3	1 (5.6)	0 (0.0)	1 (5.3)	2 (3.8)
4	2 (11.1)	1 (6.3)	0 (0.0)	3 (5.7)
31	0 (0.0)	1 (6.3)	0 (0.0)	1 (1.9)
Median (IQR)	0.5 (0 to 2)	1 (0 to 2)	0 (0 to 2)	1 (0 to 2)
Range	0 to 4	0 to 31	0 to 3	0 to 31
Amount of beta-2 agonists prescribed over 24 weeksa (µg)
Median (IQR)	6000 (0 to 40,000)	20,000 (0 to 40,000)	0 (0 to 24,000)	20,000 (0 to 40,000)
Range	0 to 80,000	0 to 620,000	0 to 60,000	0 to 620,000

TABLE 23 - Difference in mean amount of beta-2 agonists prescribed (ANOVA)

Used Bonferroni's method to calculate 95% CIs as there were an unequal numbers of cases in each group.

Comparison of means	Fluticasone vs fluticasone plus salmeterol, mean difference (µg) (95% CIa)	Fluticasone vs fluticasone plus montelukast, mean difference (µg) (95% CIa)	Fluticasone plus salmeterol vs fluticasone plus montelukast, mean difference (µg) (95% CIa)
48 weeks	−59,481 (−195,859 to 76,897)	6758 (−134,532 to 148,048)	66,238 (−66,920 to 199,396)
24 weeks	−34,861 (−107,549 to 37,827)	7942 (−61,642 to 77,525)	42,803 (−28,980 to 114,585)

Time from randomisation to treatment withdrawal (because of lack of efficacy or side effects)

Although 13 patients were withdrawn from treatment or the study, only one (on fluticasone) withdrew from treatment because of poor asthma control (patient also recorded intercurrent illness as reason) during follow-up (see Table 34). The planned analysis of time to treatment withdrawal was therefore not possible.

Lung function at 48 weeks assessed by spirometry

A total of 30 children had lung function assessed by spirometry at baseline and at 48 weeks and could be included in ANCOVA (Table 24). There were no statistically significant adjusted mean differences between the three treatment groups, with wide CIs that include clinically significant differences in both directions.

TABLE 24 - Lung function over 48 weeks assessed by spirome

A, futicasone; B, futicasone plus salmeterol; C, futicasone plus montelukast.

	Fluticasone			Fluticasone plus salmeterol			Fluticasone plus montelukast			Fluticasone vs futicasone plus salmeterol	Fluticasone vs futicasone plus montelukast	Fluticasone plus salmeterol vs fluticasone plus montelukast
	Baseline mean (SD), range	T48 mean (SD), range	Change mean (SD), range	Baseline mean (SD), range	T48 mean (SD), range	Change mean (SD), range	Baseline mean (SD), range	T48 mean (SD), range	Change mean (SD), range	Adjusted mean difference (95% CI), p-value	Adjusted mean difference (95% CI), p-value	Adjusted mean difference (95% CI), p-value
FEV1% predicted (nA = 8, nB = 13, nC = 9)	86.13 (22.56), 48.00 to 116.00	86.25 (17.77), 49.00 to 106.00	0.12 (12.74), −16.00 to 19.00	74.46 (19.70), 47.00 to 103.00	90.00 (13.91), 64.00 to 107.00	15.54 (19.77), −17.00 to 56.00	84.56 (16.94), 56.00 to 104.00	89.11 (13.63), 73.00 to 120.00	4.55 (15.83), −14.00 to 30.00	−8.58 (−20.71 to 3.56), p = 0.16	−3.51 (−16.24 to 9.22), p = 0.58	5.07 (−6.57 to 16.71), p = 0.38
FVC% predicted (nA = 8, nB = 13, nC = 9)	87.50 (19.93), 57.00 to 112.00	92.00 (12.22), 74.00 to 111.00	4.50 (10.88), −7.00 to 23.00	84.08 (18.29), 49.00 to 111.00	95.15 (16.35), 67.00 to 128.00	11.07 (24.59), −12.00 to 79.00	91.89 (20.19), 53.00 to 128.00	91.67 (13.52), 68.00 to 113.00	−0.22 (14.59), −15.00 to 28.00	−4.17 (−16.79 to 8.45), p = 0.50	1.64 (−12.03 to 15.30), p = 0.81	5.81 (−6.53 to 18.15), p = 0.34

Safety outcomes

Adverse events were assessed at each visit between randomisation and the final visit. Adverse events were classified according to the Medical Dictionary for Regulatory Activities (MedRA) (www.meddramsso.com) System Organ Class by the MCRN CTU Data Manager, with all classifications approved by the chief investigator after discussion with the trial working group.

The number (and percentage) of patients experiencing each adverse event is shown for each treatment in Table 25; this is categorised further by severity in Table 26. For each patient who experienced more than one episode of the same adverse event, the maximum severity is shown. No formal statistical testing was undertaken on these data.

In total, 53 out of 63 (84%) randomised patients experienced at least one adverse event between randomisation and last follow-up, with a similar percentage of patients across treatment groups [fluticasone 17 (89%); fluticasone plus salmeterol 18 (78%); fluticasone plus montelukast 18 (86%)]. The most common events were respiratory and nervous system disorders, infections and infestations, and general and gastrointestinal disorders. The percentage of patients reporting each adverse event was similar across treatment groups (see Table 25) except for nervous system disorders. Fewer patients reported nervous system disorders on fluticasone plus salmeterol [one patient (4.3%)] than on fluticasone plus montelukast [seven patients (33.3%)] and fluticasone [five patients (26.3%)], with a much greater number of events on fluticasone plus montelukast (37 events compared with 13 on fluticasone and two on fluticasone plus salmeterol) because of one patient experiencing 20 events. All adverse events were graded as mild or moderate (see Table 26) with a similar distribution of severity across treatment groups except for the nervous system disorders described previously.

Health economics

The sample size is extremely small in each group; therefore, although every endeavour has been made to fully analyse and report the economic results and to take necessary statistical safeguards, the results should be interpreted with extreme caution.

It can be seen from Table 28 that there was some loss of patients completing the health economics diary over each time period but completion of the health economics booklet was very similar across the groups. This included patients whose 48-week visit took place within 48 weeks because of the early closure of the trial. We removed patients from the 48-week visit analysis when the final visit took place inside 43 weeks from baseline.

Table 29 indicates the total events of care (resource use) for each of the three treatment groups during the study period.

Table 30 shows the main cost drivers in the treatment of childhood asthma. The main cost driver, not surprisingly, is the cost of the intervention itself, followed by the costs of GP visits, prescribed inhalers and prescribed medications. With larger patient numbers these results could change.

Given that the cost of the trial intervention is the greatest cost driver it is not surprising that the fluticasone plus salmeterol group and the fluticasone plus montelukast group have the greatest overall costs. That said, even though the intervention cost is higher in the fluticasone plus montelukast group, once the overall costs have been factored in it is the fluticasone plus salmeterol group that has the greatest overall cost, although the difference is small. This shows the importance of calculating the full economic cost of an intervention, not just the cost of the intervention itself.

Outcome

The QALY gains shown in Table 31 are based upon independent bootstrapped samples of n = 2000.

TABLE 31 - Quality-adjusted life-year gains

	Mean	SD
Incremental QALYs
Fluticasone	0.09	0.04
Fluticasone plus salmeterol	0.12	0.14
Fluticasone plus montelukast	0.13	0.12
Intervention + treatment costs
Fluticasone	144.75	82.13
Fluticasone plus salmeterol	458.80	95.24
Fluticasone plus montelukast	447.99	128.46

Bootstrapped t-tests and Wilcoxon rank-sum tests did not reveal any statistically significant differences in mean QALYs between baseline and 48 weeks when comparing any of the treatment groups in two-way tests. Figure 3 provides a graphical representation of quality of life over time.

FIGURE 3.

Quality-adjusted life-years over time in the trial. A, fluticasone; B, fluticasone plus salmeterol; C, fluticasone plus montelukast.

Cost-effectiveness analysis

In terms of economics the variables of interest between the groups are the differences in costs and effects. Table 32 shows the mean differences in incremental costs and QALY gains between the three groups.

TABLE 32 - Mean differences in costs and QALYs by group using bootstrapped data

t-tests of the mean difference in QALYs between the two groups.

Model	Difference in costs (£)			Difference in QALYs
	Mean	SD	p-valuea	Mean	SD	p-valuea
Fluticasone plus salmeterol vs futicasone	314.05	34.90	0.48	0.03	0.04	0.46
Fluticasone plus montelukast vs futicasone	303.24	44.59	0.35	0.04	0.04	0.43
Fluticasone plus montelukast vs futicasone plus salmeterol	−10.81	44.50	0.19	0.02	0.05	0.89

The measure of interest from these data is incremental cost-effectiveness, captured by the ICER, which shows the difference in the sum of the costs over effects for each individual patient between each of the treatment arms. Table 31 shows that there is very little difference in QALYs before and after the interventions. Similarly, cost differences are not great between the groups, with fluticasone plus salmeterol and fluticasone plus montelukast incurring approximately three times the cost per person as fluticasone. The difference in outcomes was greater for fluticasone plus salmeterol and fluticasone plus montelukast even though the cost was greater; hence, it can be seen that a positive ICER is generated for fluticasone plus salmeterol compared with fluticasone and fluticasone plus montelukast compared with fluticasone (Table 33).

TABLE 33 - Incremental cost-effectiveness ratios

	ICER (£/QALY)
Fluticasone plus salmeterol vs fluticasone	12,054
Fluticasone plus montelukast vs fluticasone	6827
Fluticasone plus montelukast vs fluticasone plus salmeterol	–588

It can be seen from the ICER results that the treatment of choice is inhaled fluticasone propionate 100 μg twice daily plus montelukast 5-mg tablet once daily, as this produces 1 additional QALY for £6827, whereas it would cost £12,054 to produce an additional QALY from inhaled fluticasone propionate 100 μg and salmeterol 50 μg twice daily (combination inhaler) plus placebo tablet once daily. Both are by comparison with inhaled fluticasone propionate 100 μg. To further aid interpretation of the results the CEACs are presented for fluticasone plus salmeterol and fluticasone plus montelukast compared with fluticasone, and fluticasone plus montelukast compared with fluticasone plus salmeterol (Figure 4).

FIGURE 4.

Cost-effectiveness acceptability curves. A, futicasone; B, futicasone plus salmeterol; C, futicasone plus montelukast.

Figure 4 shows the probabilities that fluticasone plus salmeterol and fluticasone plus montelukast are under the £30,000 per QALY cost-effectiveness threshold as required by NICE. The probability of cost-effectiveness at this level is around 80% for fluticasone plus montelukast and 60% for fluticasone plus salmeterol. Moreover, the comparison between fluticasone plus montelukast and fluticasone plus salmeterol shows that fluticasone plus montelukast dominates fluticasone plus salmeterol; however, the CEAC for fluticasone plus montelukast compared with fluticasone plus salmeterol declines after the threshold ICER because of increasing uncertainty, reflecting the lack of statistical power in the incremental QALYs. In fact, there is very little evidence supporting fluticasone plus salmeterol or fluticasone plus montelukast over the other.

Figure 5 illustrates the relative spread of incremental costs and QALYs for fluticasone plus salmeterol and fluticasone plus montelukast compared with fluticasone on a cost-effectiveness plane. This is useful because it shows that, although fluticasone plus montelukast has a lower average incremental cost and higher average incremental QALY gain, it also has greater uncertainty, which is why fluticasone plus salmeterol and fluticasone plus montelukast cannot easily be compared directly.

FIGURE 5.

The cost-effectiveness plane.

Withdrawals

In total, 13 (20.6%) patients withdrew from the study before the 48-week visit or before early study closure. The numbers of withdrawals and reasons for withdrawal are similar across the three treatment groups (Table 34). Six of the 13 patients had indicated a withdrawal from treatment but no further follow-up data were available. There were no crossovers to another treatment arm.

Compliance

The number of doses missed (inhaler and tablets) since the previous visit was captured on the follow-up CRF. The expected number of tablet doses for each visit was calculated as the number of days since the last visit (i.e. one per day) and the expected number of inhaler doses was calculated as twice the number of days since the last visit (i.e. two per day).

For each visit, each patient had the proportion of doses missed (number of doses missed/number of expected doses) calculated separately for inhalers and tablets. These proportions were averaged over all patients per treatment group separately for inhalers and tablets and are displayed as summary measures in Table 35. The proportion of treatments missed is similar across the groups and time points, except for fluticasone plus montelukast at the 48-week visit.

Outcome of non-randomised patients

The GPs of all 103 children who were registered at T–4 but who were not randomised at T0 were contacted towards the end of the study and asked to complete a questionnaire about each of their patients, as described in the protocol. As the exposure time since date of registration varied, data are summarised according to the time intervals for which the data were collected. The data presented for ‘up to 1 year’ are likely to be an underestimate of events because only the cumulative data over the period from registration until completion of the form were collected. Some patients with data reported over periods > 1 year may have had a number of their events during the first year from registration.

The questionnaire was kept very simple and short to try and maximise response. Nine (14.5%) children had at least one exacerbation that required a course of oral corticosteroids (Table 36). A total of 50 (80.6%) required at least one prescription of beta2 agonist after registration (Table 37). Unfortunately it was not possible to summarise the amount of beta2 agonist prescribed as the reporting of type of inhaler and amount prescribed was inconsistent and incomplete. In total, 18 (29.0%) patients were prescribed at least one additional treatment for asthma control (Table 38), with the majority (61%) of these being combination therapy (ICS plus long-acting beta2 agonist). Four (6.5%) children required an A&E visit (Table 39) and one (1.6%) child needed a hospital admission for their asthma (Table 40).

Main findings

The MASCOT study faced a number of challenges from the outset. Issues relating to study set-up, the pharmaceutical companies and the lack of a NHS facility for packaging and distribution are dealt with in the section on strengths and weaknesses. The most significant issue was the difficulty in recruitment of children with asthma into the study, particularly from primary care. A number of novel strategies were developed to improve recruitment; however, although recruitment did increase towards the end of the study, the percentage of children eligible for randomisation after the run-in fell and an application for additional funding for the study was rejected.

The strategies are clearly outlined in Appendix 3 and should be considered in any study in the future that attempts to recruit children from both primary and secondary care. Had these strategies been considered from the outset of the study the recruitment issue may have been less of a problem. Unfortunately, although all centres agreed that they would institute the strategies appropriate to them from January 2010 onwards, it often takes time before the rewards of such strategies take hold. Had the study been able to progress into and through the 2010 autumn peak seen in children with asthma there was a feeling among the MASCOT staff that we would have shown an upturn in both registration and randomisation of suitable patients. We still have no doubt that such patients exist in the UK.

Because of the early study closure and resulting small sample size there was insufficient power to detect any statistically significant differences between the three groups for the efficacy analyses. At 48 weeks the RR of exacerbations requiring treatment with oral corticosteroids was 0.91 (98.3% CI 0.07 to 12.05, p = 0.93) for fluticasone compared with fluticasone plus salmeterol; 1.10 (98.3% CI 0.06 to 18.6, p = 0.94) for fluticasone compared with fluticasone plus montelukast; and 1.21 (98.3% CI 0.09 to 15.97, p = 0.86) for fluticasone plus salmeterol compared with fluticasone plus montelukast. The CIs are extremely wide and include clinically important RRs that could favour any of the treatments. The results for the 24-week analysis were similarly inconclusive as were the results for the time to first exacerbation comparing fluticasone with fluticasone plus salmeterol (HR 0.63, 95% CI 0.19 to 2.08); time to first exacerbation comparing fluticasone with fluticasone plus montelukast (HR 1.52, 95% CI 0.34 to 6.7); and time to first exacerbation comparing fluticasone plus salmeterol with fluticasone plus montelukast (HR 2.37, 95% CI 0.68 to 8.2).

Although there were no statistically significant differences in mean quality of life scores adjusted for baseline values for any of the pair-wise treatment comparisons, the mean quality of life score had improved at 48 weeks and at 24 weeks for all treatment groups across all domains, both for the child and for the caregiver. Fewer children missed at least 1 day of school over 48 weeks on fluticasone plus montelukast (18.2%) than on fluticasone (63.6%) and fluticasone plus salmeterol (60%), whereas more children on fluticasone plus salmeterol (71.4%) required at least one beta-2 agonist than children on fluticasone (54.5%) or fluticasone plus montelukast (58.3%) over 48 weeks. These patterns were not supported by the 24-week data and wide CIs for pair-wise comparisons of relative treatment effects make conclusions difficult to draw. Only a few children required a hospital admission during the study, with relative treatment effects difficult to estimate.

The majority of patients (84%) experienced at least one mild or moderate adverse event during follow-up. The most common events reported were respiratory disorders, nervous system disorders, infections and infestations, general disorders and gastrointestinal disorders. The percentage of patients reporting each adverse event type was similar across treatment groups except for nervous system disorders, which were reported by fewer patients on fluticasone plus salmeterol [one patient (4.3%)] than patients on fluticasone plus montelukast [eight patients (38.1%)] and fluticasone [five patients (26.3%)], with a much greater number of events on fluticasone plus montelukast because of one patient experiencing 20 events.

Health economic analyses were extremely limited by the small sample size and very small non-significant differences in quality of life outcomes between the three treatment groups, which mean that very little confidence can be placed in the ICER and CEAC results. Nevertheless, interpreting the results with great caution indicates that fluticasone plus montelukast could be the economic treatment of choice. A key component of this treatment is montelukast, which lost patent status in August 2012. Historically, prices of drugs that lose patent status have fallen between 30% and 70%, particularly within 4–5 years. 23,24 This will further improve the cost-effectiveness profile of fluticasone plus montelukast relative to fluticasone plus salmeterol; however, this would be based solely on intervention cost.

The limited data available for children who were registered at T–4 but who were not randomised at T0 suggest that their regimens over the subsequent 12 months were not entirely successful in spite of the control achieved over the 4-week run-in, with 14.5% having an exacerbation, 80.6% requiring at least one short-acting beta-2 agonist prescription and 29% requiring a prescription for at least one further asthma treatment. These results also require careful interpretation as the data represent only 60.2% of the registered non-randomised patients.

Strengths and weaknesses

MASCOT was planned as the largest long-term paediatric asthma study ever. Although UK guidelines on asthma management2 have advocated that, when control is poor on low-dose ICSs alone, combination therapy is the first-line step-up treatment, the evidence base for this remains poor. Before the development of the MCRN in 2005, clinical studies in children were often inadequate in both number and design. MASCOT, it was hoped, would address this by being designed by a paediatric CTU in Liverpool, supported by key respiratory paediatric doctors, nurses and patients. As in all studies the key to success is appropriate study design, good working relations between all concerned and successful recruitment.

MASCOT was in the first wave of MCRN studies funded by the HTA programme. Funding for MASCOT was approved in January 2006 pending the agreement of GSK and MSD to supply the medications, including the necessary placebos. Reaching this agreement took a considerable length of time. It was then discovered that there was no NHS facility large enough to package and supply the medicines for all of the participating research sites. An application was made to the HTA for additional funding for a commercial company to undertake this procedure. The HTA supported the application.

The protocol indicated that metered dose inhalers (MDIs) would be used for the study. Unfortunately, we were then informed that GSK was closing down its factory which was the only worldwide manufacturer for the supply of their MDIs for research purposes. With the approval of all of the principal investigators the study protocol was amended to allow the GSK medications to be in the form of dry powder through Accuhalers. When the montelukast tablets from MSD arrived from America it was noted that their expiration date was only 5 months later. These challenges delayed the study opening and have been well documented in a recent publication in Thorax. 25

The study finally opened fully in all centres in May 2009, some 3 years and 4 months after the initial funding had been approved. When the study was being developed in 2005 there was clinical belief that it would not be difficult to recruit up to 90 children over a 12-month period from each of the major recruitment centres. The reality was very different. Prescribing habits in both primary and secondary care changed between these dates, with increasing numbers of children with asthma being commenced on long-acting beta-2 agonists or montelukast, rendering them ineligible for the MASCOT study. The chief investigator and the trial co-ordinator visited all of the research centres from October 2009 to January 2010 to try to discover ways to improve recruitment. A number of novel ideas were developed and these were put in place by January/February 2010.

The TMG requested a meeting with the HTA to discuss recruitment challenges, the development of novel additional recruitment strategies and the possibility of opening new recruitment sites. The TMG considered closing sites that were failing to recruit successfully and the chief investigator and trial co-ordinator had communicated with other sites where there were good working relationships between the research networks and where the prospective principal investigator believed that it would be practically possible to recruit ≥ 30 patients within the allocated time. The HTA met with the TMG but wanted to see a doubling of patients recruited and randomised over the subsequent 3 months before considering the possibility of opening any new potential sites. Over those 3 months patient recruitment into the study did double but there was little increase in patients with sufficient symptoms to be randomised into the double-blind arm of the study. The HTA therefore closed the study to recruitment in June 2010. Patients who had been randomised up to that date were allowed to continue in the study either for a full 48 weeks or until the end of January 2011, whichever occurred first.

As indicated in the results section, the small number of children recruited and randomised was insufficient to show clinically or statistically significant differences in asthma exacerbations between the three treatment groups. Interestingly, however, quality of life measurements improved from baseline in all three groups for the duration of the study. This phenomenon is well reported in clinical trials and is thought to be the result of participating in the study itself rather than a specific effect of the individual medications. Differences were seen between the results obtained at 24 weeks and those at 48 weeks but these differences were not consistent and were probably once again related to the small number of patients randomised.

Because of previous safety concerns regarding long-acting beta-2 agonists, particularly in adult male US asthma sufferers, it was encouraging that we saw no safety issues related to any of the treatment groups in this study, but our patient numbers were too small to draw any conclusions. Other recently published papers26,27 have, to a certain degree, also allayed previous concerns.

Our health economic data suggest a possible benefit of fluticasone plus montelukast over fluticasone therapy alone but caution is needed because of the small number of children participating in the MASCOT study.

Implications for health care

Unfortunately, the early cessation of this study leaves the question of how best to treat children uncontrolled on ICSs unanswered. Since commencing the MASCOT study an American asthma study, the BADGER trial,12 has been published which concluded that long-acting beta-2 agonist step-up was more likely to provide a better response than either fluticasone alone or fluticasone plus montelukast. This is in keeping with the present national UK guidelines for asthma therapy. 2 Further studies that have compared add-on long-acting beta-2 agonist with doubling the dose of fluticasone in children28,29 have favoured long-acting beta-2 agonist step-up. However, these studies have major limitations, such as short duration, required reversibility to be recruited and inappropriate outcome measures for children, which make it difficult to generalise the results. Now that MASCOT has been prematurely terminated, the question may never be adequately addressed.

Recommendations for research

We believe that there are two major priorities arising from this work. The first is to examine whether or not patients treated in primary care for a chronic condition can be approached by someone from outside their current care team to consider study participation.

Second, it is important to look at alternative study designs to answer the key research question of what is the most appropriate treatment for children uncontrolled on low-dose ICSs. This might be carried out using routine data included in primary care databases, possibly supplemented with patient-reported outcome data.

Setting the study in the context of existing research

Compared with the adult literature, the results of studies of asthma in children have been inconsistent, and there are far fewer studies available. The earlier studies, in particular, produced negative results5 and there was little information available for the Cochrane systematic review. 30 The MASCOT study, therefore, had the potential to be the definitive paper to state what is the appropriate treatment at step 3 of the national and international guidelines for asthma when low-dose ICSs are not successfully controlling symptoms and exacerbations. Failure to recruit sufficient numbers has prevented this study goal.

Contribution of authors

Warren Lenney was the chief investigator, liaised with all the of the clinical centres involved, worked closely with the trial co-ordinators and headed up the team that drew up the interim and final reports.

David Price was a member of the TMG and contributed to the design and conduct of the study, provided expertise in primary care and prepared the report for publication.

Andrew McKay (Trial Statistician) was a member of the TMG, performed the statistical analyses and report preparation during the study and prepared the report for publication.

Catrin Tudur Smith (Senior Lecturer in Biostatistics) was a member of the TMG, led the statistical team, contributed to the design of the study, its conduct and analysis and prepared the report for publication.

Paula Williamson (Director of the CTU and Professor of Medical Statistics) was a member of the TMG, contributed to the design and conduct of the study and reviewed a draft of the report.

Marilyn James (Professor of the Economics of Health and Social Policy) led the health economics team and contributed to the design and analysis of the study and prepared the report for publication.

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.

Publications

1. Lenney W, Perry S, Price D. Clinical trials and tribulations: the MASCOT study. Thorax 2011;66:457–8.

1 Introduction

The Statistical Analysis Plan (SAP) provides a detailed and comprehensive description of the main, pre-planned analyses for the study ‘MASCOT: Management of asthma in school-age Children On Therapy: a prospective, multi-centre, randomised, double blind, controlled, trial comparing inhaled fluticasone/salmeterol (combination inhaler) + placebo tablet, and inhaled fluticasone + montelukast tablet with inhaled fluticasone propionate + placebo tablet for improved asthma control’. This study is carried out in accordance with the World Medical Association Declaration of Helsinki (1964) and the Tokyo (1975), Venice (1983), Hong Kong (1989) and South Africa (1996) amendments and will be conducted in compliance with the protocol, MCRN CTU Standard Operating Procedures and EU Directive 2001/20/EC, transposed into UK law as the UK Statutory Instrument 2004 No 1031: Medicines for Human Use (Clinical Trials) Regulations 2004.

This statistical analysis plan details the intended analyses and should be clear and detailed enough to be followed by any statistician. This will prevent the introduction of bias or data dredging.

These planned analyses will be performed by the trial statistician under the supervision of the lead statistician. The analysis results will be described in a statistical analysis report, to be used as the basis of the primary research publications according to the study publication plan.

All analyses are to be performed with Standard Statistical Software (SAS). The final analysis datasets, programs and outputs will be archived following good clinical practice guidelines (ICH E9). The testing and validation of the statistical analysis programs will be performed following the relevant Standard Operation Procedure.

Due to the early trial closure and lack of data some analyses described in protocol v8.1 are not planned now.

2 Design

3 Description of Study Population

8 Reporting Protocol Deviations

The table (given in Appendix A) lists potential deviations of important protocol specifications, including eligibility criteria, treatment regimens and study assessments. Protocol deviations are classified according to this system prior to unblinding of treatment. The number (and percentage) of patients with major and minor protocol deviations will be summarised by treatment group with details of type of deviation provided. No formal statistical testing will be undertaken.

9 Setting Results in Context of Previous Research

Once the trial has been completed the results of the trial will be set in context of the existing evidence base.

References

Protocol Deviations Table

Note:

1. Impact refers to the impact of the potential protocol deviation on the risk of introducing bias in the defined en-points of the trial. This is generally graded as:

   1. Major (in which case patients who experience this protocol deviation would generally be excluded from the ‘per protocol’ analysis set)

   2. Minor (in which case patients who experience this protocol deviation would generally be included in the ‘per protocol’ analysis set)

2. Justification refers to the protocol-specific justification for the assessment of the impact of each potential protocol deviation.

Resource components sources and their unit costs

Unit cost for prescribed medicines

Unit cost for over-the-counter medicines

Version 2.0 (19 March 2008)

Amendments and clarifications prior to full ethical approval (v1.0 22 January 2008 to v2.0 19 March 2008).

Version 3.0 (16 May 2008)

Amendments and clarifications (v2.0 19 March 2008 to v3.0 16 May 2008).

Version 4.0 (24 July 2008)

Amendments and clarifications (v3.0 16 May 2008 to v4.0 24 July 2008)

Version 5.0 (30 January 2009)

Amendments and clarifications (v4.0 24 July 2008 to v5.0 30 January 2009)

Version 6.0 (20 May 2009)

Amendments and clarifications (v5.0 30 January 2009 to v6.0 20 May 2009)

Version 6.1 (18 January 2010)

Minor amendments and clarifications (v6.0 20 May 2009 to v6.1 18 January 2010)

Version 7.0 (26 March 2010)

Amendments and clarifications (v6.1 18 January 2010 to v7.0 26 March 2010)

Version 8.0

Amendments and clarifications (v7.0 26 March 2010 to v8.0 21 March 2011)

Version 8.1

Amendments and clarifications (v8.0 21 Mach 2011 to v8.1 13 June 2011)

Primary outcome intention-to-treat analysis set inclusion/exclusions with reasons: T48

Primary outcome intention-to-treat analysis set inclusion/exclusions with reasons: T24

Primary outcome per-protocol analysis set inclusion/exclusions with reasons: T48

Primary outcome per-protocol analysis set inclusion/exclusions with reasons: T24

PAQLQ(S) activity limitations domain score analysis set inclusion/exclusions with reasons: T48

PAQLQ(S) activity limitations domain score analysis set inclusion/exclusions with reasons: T24

PAQLQ(S) activity limitations domain score analysis set inclusion/exclusions with reasons: T24

PAQLQ(S) symptoms domain score analysis set inclusion/exclusions with reasons: T48

PAQLQ(S) symptoms domain score analysis set inclusion/exclusions with reasons: T24

PAQLQ(S) total score analysis set inclusion/exclusions with reasons: T48

PAQLQ(S) total score analysis set inclusion/exclusions with reasons: T24

PACQLQ activity limitations domain score analysis set inclusion/exclusions with reasons: T48

PACQLQ activity limitations domain score analysis set inclusion/exclusions with reasons: T24

PACQLQ emotional functioning domain score analysis set inclusion/exclusions with reasons: T48

PACQLQ emotional functioning domain score analysis set inclusion/exclusions with reasons: T24