Treatment guided by fractional exhaled nitric oxide in addition to standard care in 6- to 15-year-olds with asthma: the RAACENO RCT

Article history

The research reported in this issue of the journal was funded by the EME programme as project number 15/18/14. The contractual start date was in February 2017. The final report began editorial review in April 2021 and was accepted for publication in September 2021. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The EME editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Copyright statement

Copyright © 2022 Turner et al. This work was produced by Turner et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This is an Open Access publication distributed under the terms of the Creative Commons Attribution CC BY 4.0 licence, which permits unrestricted use, distribution, reproduction and adaption in any medium and for any purpose provided that it is properly attributed. See: https://creativecommons.org/licenses/by/4.0/. For attribution the title, original author(s), the publication source – NIHR Journals Library, and the DOI of the publication must be cited.

2022 Turner et al.

Childhood asthma

Asthma is the most common chronic condition in childhood and affects 1.1 million children in the UK. 2 Asthma is characterised by coughing, wheezing and shortness of breath. Symptoms can be brought on by many triggers, including an upper respiratory tract infection, exercise or exposure to inhaled allergens (e.g. cat dander and pollen) or chemicals (e.g. cigarette smoke). Physiological testing reveals that children with asthma have obstructed lung function compared with their non-asthmatic peers, although this obstruction is reversible. Pathologically, asthma is characterised by epithelial shedding, eosinophilic inflammation, proliferation of mucus-producing goblet cells, smooth muscle hypertrophy and basement membrane thickening. 3 Despite the high prevalence of asthma in the population of all countries worldwide, there is no universally accepted definition of asthma, nor is there a diagnostic test of sufficient sensitivity and specificity. 4

Asthma causation

Asthma is a typical complex condition in which genetic and environmental factors interact and result in disease. Twin studies suggest that up to 70% of causation can be attributed to heritable factors. 5 Many environmental factors are implicated in childhood asthma causation, including exposure to second-hand smoke, pollution and moulds, and dietary factors. 6 Asthma symptoms can develop at any stage of life, and thereafter follow a pattern of remission and relapse. Asthma is a major risk factor for chronic obstructive pulmonary disease in the seventh decade of life, a condition that is characterised by irreversible airway obstruction and is a leading cause of death in the Western world. 7 Asthma is known to be a syndrome in which a number of different mechanisms lead to common symptoms and pathology. 8 In adults, childhood-onset asthma is genetically distinct from other forms of asthma and is associated with allergic conditions, such as eczema, hay fever and food allergies. 9–12 Two multifaceted interventions10,11 that promoted breastfeeding and affected early-life exposures to smoking, ingested allergens and inhaled allergens were found to reduce asthma incidence in at-risk individuals, but many other interventions aimed at preventing asthma have not been successful. At the time of writing, there is no accepted intervention that prevents asthma development de novo. The current management approach to asthma is to control symptoms with inhaled and ingested medications that relieve acute symptoms and prevent chronic symptoms and adverse events (AEs), such as lung function decline.

Asthma control and exacerbations

Asthma control can be defined as ‘the extent to which the various manifestations of asthma are reduced or removed by treatment’. 13 A key goal of asthma treatment is to achieve asthma control. 14 There are several questionnaires that are used to objectively measure asthma control, and the Asthma Control Test (ACT) is currently the most commonly used test for older children and adults. 15 The Children’s Asthma Control Test (CACT) has been validated for use in children aged between 4 and 11 years. 16 Patients (or parents/carers) complete the ACT or CACT by answering questions about nocturnal and exertional asthma symptoms and how frequently they take asthma-reliever treatment. The maximum score on the ACT is 25, while the maximum score on the CACT is 27. A score of > 19 from the ACT or CACT indicates that an individual’s asthma is controlled; this means that, despite subtle differences in methodology, the definitions of ‘uncontrolled asthma’ within the ACT and CACT are equivalent. A second key goal of asthma management is to prevent asthma exacerbations. Persistent, poorly controlled asthma symptoms place a child at increased risk of an asthma exacerbation. 14,17

Asthma exacerbations (synonymous with asthma attacks) are a worsening of symptoms and may lead to coughing, wheezing and breathing difficulties. During an asthma exacerbation, the airways become swollen and inflamed, and the muscles around the airways contract. Asthma exacerbations are usually treated with oral corticosteroids (OCs) at home but can result in hospitalisation. They are associated with morbidity and occasionally mortality. The incidence of death from an asthma exacerbation in the UK is at least double that of other European countries. 18 In the UK, a child is admitted to hospital with an asthma exacerbation every 20 minutes. 2 The annual cost to the NHS that is attributable to asthma is £1.1B, with one-third of this incurred from unscheduled care2 and, on a pro rata basis, at least £150M as a result of asthma exacerbations in children. Parental absence from work is an additional cost to the economy that is attributable to asthma exacerbations; one parent will typically be absent from work for 4 days when a child has an asthma exacerbation. 19 Another important impact of asthma is that a child’s education suffers as a result of missed days of school. 20

Pharmacological management of childhood asthma

Medications that are used to manage asthma are considered to be relievers [short-acting beta-agonists (SABAs) taken on an ad hoc basis to relieve symptoms] or preventers (taken on a daily basis independent of symptoms). There is an extensive evidence base that shows that asthma symptoms can be reduced by preventer treatments (also known as maintenance therapy), such as inhaled corticosteroids (ICs), inhaled long-acting beta-agonists (LABAs) and oral leukotriene receptor antagonists (LTRAs). 14,17,21 Current UK guidelines for the management of asthma recommend that decision-making for asthma preventer treatment should be driven by very recent symptoms, for example symptoms over the last 2–4 weeks. 17,21 The initial asthma preventer treatment recommended by all guidelines is ICs, but there is uncertainty about what the best treatment option is when symptoms are not controlled by IC-preventer treatment alone. 14,17,21 The British Thoracic Society (BTS) and Scottish Intercollegiate Guidelines Network (SIGN) guideline17 suggests adding either LABA or LTRA, whereas the National Institute for Health and Care Excellence (NICE) guideline21 suggests adding LTRA and the Global Initiative for Asthma guideline suggests increasing IC dose before adding LABA or LTRA. 14 Disappointingly, 70% of children in the UK experience asthma symptoms despite treatment, and in the USA 50% of children have uncontrolled asthma symptoms. 22,23 The reasons for children’s symptoms being uncontrolled despite receiving preventer treatment are described elsewhere and include misdiagnosis, illness behaviour, poor inhaler technique and non-adherence to preventer treatment. 17

Adherence to asthma preventer treatment is important to achieving good asthma control and reducing the risk of exacerbations,24 but adherence to preventers is typically < 50% in children with asthma. 24 Interventions to improve adherence can be effective in improving asthma outcomes. Examples of interventions that address poor adherence include electronic logging devices that objectively record inhaler use and let the patient (or parent) know that the treatment is due, and direct observation of therapy. 25,26

There is a widely recognised need for objective tests to help guide asthma treatment decision-making in children to prevent asthma exacerbations and improve day-to-day control. Spirometry, which measures how fast children can breathe out and for how long, is recommended by some, but not all, asthma guidelines; however, its use in asthma management has not been formally evaluated. Other tests have been evaluated for use in guiding the management of childhood asthma, but investigations such as peak expiratory flow, bronchial hyperresponsiveness and exhaled breath analysis (e.g. exhaled breath temperature, exhaled breath condensate, exhaled air volatile organic compounds) have so far proven unsuccessful or not feasible in routine practice. 27–29 We all produce a gas called nitric oxide in our breath, and the concentration of nitric oxide in exhaled breath [called fractional exhaled nitric oxide (FeNO)] is elevated in children with asthma compared with children without asthma. FeNO is considered to be a surrogate for the allergic airway inflammation (also called airway eosinophilia) that is characteristic of childhood asthma.

The potential role of fractional exhaled nitric oxide in monitoring childhood asthma

The evidence that FeNO may be useful in guiding asthma treatment comes from studies that show that treatment with ICs and LTRAs leads to reduced FeNO, and that FeNO rises before an asthma exacerbation and in association with poor asthma control. 30–35 There are limitations to the application of FeNO in clinical practice, including the fact that some children with high FeNO but little or no eosinophilia fail to respond to treatment with ICs despite being known to be compliant. 36,37 A small clinical trial found no added benefit of sputum eosinophils to guide asthma treatment in children in addition to symptoms. 38

Fractional exhaled nitric oxide fulfils many of the criteria of a biomarker for allergic airway inflammation and is approved by the Federal Drug Authority for the management of asthma. 39 The measurements can be made in children as young as 5 years of age using hand-held and portable apparatus. Measurements of FeNO are reproducible, are available within a minute of testing and have been costed at £22.90 per test (2005 prices). 40

There have been eight published clinical trials that have used FeNO to guide asthma treatment in children and the results have been mixed. 41 A Cochrane review of the evidence concluded that there was evidence that FeNO-guided treatment was associated with reduced exacerbations, but not with improved asthma control. 41 These studies had different inclusion criteria for participants, had different primary outcomes and applied different treatment algorithms to (often) different changes in FeNO values. 42 Not all of the clinical trials found that the intervention reduced exacerbations or improved control, and there was evidence of an overall increase in IC treatment associated with FeNO-guided management. 43

The original data from seven of these eight trials have been obtained and combined to facilitate understanding of how FeNO might be used to guide asthma treatment in children. To date, three relevant papers have been published from this data set. 44–46 The first paper46 described the variability in repeated measurements of FeNO over time and suggested that concentrations in children with controlled asthma symptoms, whose treatment is classed as ‘stable’, may change by ≥ 50% over a 3-month period. These findings are consistent with earlier results from children without asthma. 35 The second paper44 described how percentage change in FeNO (but not absolute change in FeNO) over a 3-month period was weakly associated with loss of future asthma control (but not a future asthma exacerbation). The third paper45 reported that FeNO-guided treatment was associated with improved asthma outcomes in some subgroups. Overall, children who were not treated with LTRAs had better outcomes if their treatment was guided by FeNO than children treated with LTRAs. In addition, there was evidence that treatment guided by FeNO led to better outcomes in children who were not obese and who were allergic than in children who were obese or who were not allergic. 45

Current guidelines14,17,21 do not recommend that FeNO should be routinely used to guide asthma treatment in children, but the Cochrane review suggests that FeNO-guided asthma management may be useful in reducing asthma exacerbations in children. 41

The Reducing Asthma Attacks in Children using Exhaled Nitric Oxide (RAACENO) trial was designed to compare the efficacy of asthma treatment guided by symptoms and FeNO with asthma treatment guided by symptoms alone for the risk of asthma exacerbation (full details in Chapter 2). The RAACENO trial differed from previous FeNO trials in the following aspects:47

1. The change in FeNO was expressed as a percentage where previous values were available, based on recent evidence. 44 This recognised that some individuals have higher FeNO values than others.

2. A threshold of ≥ 50% change from the previous value was used to define a clinically meaningful change, based on recent evidence. 46

3. Within the FeNO group there were different treatment algorithms between those with symptoms and a meaningful change in FeNO and those experiencing symptoms and no meaningful change in FeNO. This strategy has proved effective in pregnant women but has not previously been tested in children. 48

Hypothesis

Our hypothesis is that the proportion of children with one or more asthma exacerbations over 12 months will be reduced when asthma treatment guided by FeNO plus symptoms is compared with treatment guided only by symptoms.

In Chapter 2, we describe the trial design and methods. Chapters 3 and 4 include the baseline characteristics and clinical results. In Chapters 5 and 6 we describe the health economic and qualitative process evaluations, respectively. Finally, in Chapter 7, we discuss the results of the trial and consider implications for practice and recommendations for research.

Trial design

The study protocol has been published in an open access journal47 and is a standalone document.

RAACENO was a pragmatic, multicentre, parallel, randomised controlled trial (RCT) that aimed to compare the efficacy of asthma treatment guided by symptoms and FeNO with that of asthma treatment guided by symptoms alone for the risk of asthma exacerbation. The aim was to recruit 502 children with asthma who were treated with ICs and who had had an asthma exacerbation treated with OCs in the past year, with at least 50 participants being recruited in primary care.

The trial design is summarised in Figure 1. Participants were recruited to the trial and were followed up for 12 months post randomisation. Clinical assessments (including assessment of symptoms and measurement of FeNO) took place at recruitment and then at 3, 6, 9 and 12 months post randomisation. FeNO was measured in all participants at every assessment, but guided treatment decisions in the experimental group only (asthma treatment guided by symptoms and FeNO). In the control group (asthma treatment guided by symptoms alone), FeNO was also measured, but the results were not used for any treatment decisions. At each visit, a treatment recommendation was made in relation to recent asthma symptoms, FeNO measurements, treatment adherence and current asthma treatment.

FIGURE 1.

Trial design. FEV₁, forced expiratory volume in 1 second.

The primary outcome (one or more asthma exacerbations treated with OCs) was collected at each assessment and determined up to 12 months (or to the time at which the participant was lost to follow-up or withdrew from the trial).

Adherence to IC treatment was logged electronically where possible using a smart inhaler; some IC inhalers did not have a compatible smart inhaler device.

The research design also included an evaluation of health-care costs, which considered primary and secondary care contacts and asthma treatment. This information was collected at each assessment, supported by a patient-held diary. The inclusion of a health economic evaluation was in response to a recent Health Technology Assessment publication which noted that ‘little or no empirical evidence exists’ for economic evaluation of FeNO monitoring. 49 Since RAACENO was funded, a health economic evaluation of FeNO monitoring in children has been published. 50 It found no difference in cost or quality-adjusted life-years (QALYs) between the group whose asthma treatment was guided by FeNO plus symptoms and the group whose treatment was guided by symptoms alone.

A qualitative process evaluation was also incorporated into the study design to explore participant experiences and determine the acceptability of the intervention. A sample of up to 20 families (parent and child pairs) across both groups of the trial, as well as between 15 and 20 trial staff, were interviewed until data saturation was reached.

The University of Aberdeen and NHS Grampian co-sponsored the trial. The trial was approved by North of Scotland Research Ethics Committee A (16/NS/0106).

Participants

Participants were children aged 6–15 years with asthma. We aimed to recruit 502 children, with 452 recruited in secondary care sites across the UK and 50 recruited in primary care centres in the east of England. The following inclusion criteria were used to identify eligible participants.

Identification

Potential participants were recruited from both primary and secondary care sites across the UK. Recruitment strategies differed between centres depending on local NHS organisational factors and are described in the following sections.

Secondary care

The usual clinical team identified eligible children. The initial approach was in person (at a clinic appointment) by a member of the usual-care team [which may have also included embedded research nurses (RNs)] or by a letter/e-mail from the managing clinician or asthma specialist nurse. For those approached in clinic, the parent and age-appropriate child patient information leaflets (PILs) were given to the family by a member of the usual clinical team. At this time, the parent was asked whether or not they would be happy for a member of the study team to speak to them in the clinic or to contact them by telephone to answer any questions that they may have about the study. For those approached by post or e-mail, the parent and child PIL and letters of invitation were sent to the family by the managing clinician or asthma specialist nurse. A member of the usual care team (which included embedded RNs) contacted the parent by telephone around 2 weeks after the initial approach to answer any questions they had about the study.

In Scotland, the NHS Research Scotland Primary Care Network also identified potential participants in primary care for three secondary care sites. The initial approach was by letter from the primary care practice, which was sent with the short PIL, a reply slip and a pre-paid envelope. Those interested in taking part in the research were asked to return the reply slip or contact the research team based at the secondary care site. Interested families were then sent the parent and age-appropriate child PILs and arrangements were made for those who wished to take part to be seen at the secondary care site for a recruitment appointment.

Primary care

Primary care recruitment was conducted through a number of primary care practices in the east of England. Practices participated as independent research sites, supported by local Clinical Research Network (CRN) RNs. Staff at the primary care practice identified eligible children from electronic records. The initial approach was by a letter from a general practitioner (GP) at the practice, which was sent with the short PIL, a reply slip and a prepaid envelope. Those interested in taking part in the research were asked to return the reply slip or contact the research team. A member of the primary care team or CRN member of staff contacted the parent by telephone around 2 weeks after the initial approach to answer any questions that they had about the study. Interested families were then sent the parent and age-appropriate child PILs, and arrangements were made for those who wished to take part to be seen at the primary care practice for a recruitment appointment.

Recruitment

For families that were interested in participating, a baseline appointment was then arranged in either primary or secondary care. At this baseline appointment, consent was received in accordance with good clinical practice. Fully informed, written consent was received from parent(s)/carer(s) and (where appropriate) from the child. If the child did not provide written consent, they were asked to give verbal assent. At the baseline visit, the participant’s eligibility was confirmed by a medically qualified doctor. Baseline data were also collected (described in more detail in Baseline).

Some recruitment sites were able to offer one or more of the optional mechanistic components (listed below). In these sites, consent for each relevant component was received from participants who wished to take part in one or more of the components:

• bronchodilator-derived response

• skin-prick test

• saliva collection for deoxyribonucleic acid (DNA) analysis

• sputum collection for eosinophil count.

Children who turned 16 years of age during follow-up and who wished to re-consent to their continued participation in the study could do so at their next follow-up visit. They were given a PIL and were asked to sign a consent form. If they wished to continue with the study, but did not wish to complete a new consent form, the study team confirmed verbally that they were happy to continue taking part.

Randomisation/treatment allocation

Eligible and consenting participants were randomised to one of the two groups (1 : 1) [treatment decisions based on FeNO plus symptoms (intervention group) or treatment decisions based on symptoms alone (standard-care group)]. A member of the research team at the site randomised participants using a 24-hour web-based randomisation application (bespoke in-house system), which was hosted by the Centre for Healthcare Randomised Trials, University of Aberdeen. Random allocation used a minimisation algorithm, including a random element (20%). Stratification was by centre (primary care sites were considered as one centre for the purpose of minimisation), age (< 11 years, ≥ 11 years), sex and asthma severity [as evidenced by BTS/SIGN treatment step (BTS step 2, BTS step 3, BTS step 4)]. 51

The principal investigator (PI), or the individual at a site with delegated authority, accessed the web-based randomisation system. Minimisation characteristics (centre, age, sex and asthma severity) were entered into the web-based system, which returned the allocation. Randomisation was completed in the clinic at the baseline visit after consent had been given and baseline data collected. Participants were informed of their allocated treatment group following randomisation. The study teams at each site were also made aware of the allocated treatment group. Each participant’s GP was informed that the child was taking part in RAACENO and which treatment group they had been randomised to.

Treatment decisions

In this study, treatment decisions in the intervention group were guided by FeNO measurement and symptoms. Symptoms were assessed using the ACT or, if the participant was aged < 12 years, the CACT. In the standard-care group, treatment decisions were guided by symptoms alone (assessed using ACT/CACT), as per the BTS/SIGN national guideline. 51

Treatment decisions in both groups were protocolised. The relevant data for the treatment decision (current medication, any recent change to medication, adherence to IC medication, symptoms and, in the intervention group, the FeNO result) were entered into the study website. For participants in the standard-care group, the RN and clinician were blinded to the FeNO result until after the visit was complete and the treatment decision had been made. The treatment algorithm was incorporated into the study website and, therefore, this was applied without subjectivity. The treatment algorithm could recommend no change to current treatment, a reduction in treatment, an increase in treatment or, for participants who received the highest treatment via a dry powder inhaler, a change to receive the same treatment but by metered dose inhaler (MDI) via a spacer. Where the child was already receiving the maximum treatment for their age, the algorithm recommended referral to the asthma specialist. At each visit, the clinical team could follow the recommendation of the treatment algorithm or decide to make an alternative treatment decision in consultation with the family.

Where an alternative treatment decision was made, this was documented within the study case report forms (CRFs) and not considered to be a protocol deviation.

At each follow-up, the algorithm was applied from the treatment that they were currently taking.

For participants recruited in secondary care, the asthma team could either prescribe any change in treatment or write to the GP to request a change in treatment, depending on local prescribing practice. For participants recruited in primary care, the GP could prescribe any change in treatment.

Experimental intervention

In the intervention group, asthma treatment decisions were guided by FeNO and symptoms. Figure 2 summarises the experimental intervention and Table 1 summarises the treatment steps. The experimental intervention and subsequent adjustment of treatment steps were applied at baseline and at each of the follow-up visits.

FIGURE 2.

Simplified schematic of the experimental intervention.

When developing the experimental intervention, the following considerations were made:

• FeNO-guided treatment escalated either anti-inflammatory medication (algorithm 1, elevated FeNO) or early intervention with bronchodilators (algorithm 2, FeNO not elevated). The response to increasing ICs or the addition of LABA or LTRA is known to be heterogeneous in children,52 and this algorithm uses FeNO to stratify treatment with early escalation of IC treatment or early addition of LABA and LTRA ‘add-on’ therapies, an approach that has been proven in a FeNO trial in pregnant mothers. 53

• Inhaled corticosteroid was increased only once in controlled children with elevated FeNO, and only if their IC dose had not been increased since their asthma exacerbation and they were adherent.

• In accordance with the BTS/SIGN 2016 guideline,51 poor adherence was also considered (and discussed) before escalating treatment. Elevated FeNO is a result of airway inflammation but does not identify the cause of airway inflammation, and poor adherence is an important mechanism for persisting airway inflammation; therefore, adherence was part of the algorithm. Adherence was a continuous measure of a complex outcome, and a single cut-off value to prompt a discussion about adherence was arbitrary but needed. Poor adherence was defined as < 70%; an audit of children attending the asthma clinic in Aberdeen found that 48% had collected ≥ 70% of their IC treatment from their GP over 12 months (data from unpublished NHS Grampian audit by Dr Steve Turner).

• Interpretation of FeNO on the first visit relied on population norms, rather like lung function or height, but at subsequent visits it was interpreted as a percentage change from the previous measurement.

• Asthma control was defined as ACT score of ≥ 20. We know from two previous FeNO intervention trials54,55 and one observational study of children attending secondary care asthma clinics56 that 75% of children with asthma have an ACT score of ≥ 20 at any one time.

More details of the treatment steps are provided in Appendix 1 and the detailed decision tree is given in Appendix 2.

Control intervention (standard care)

In the standard-care group, asthma treatment was guided by symptoms alone.

Figure 3 summarises the control intervention and Table 2 summarises the treatment steps, which are in accordance with the national guideline that was in place when RAACENO started. 51 The control intervention and subsequent adjustment of treatment steps were applied at recruitment and at each of the follow-up visits (3, 6, 9 and 12 months).

FIGURE 3.

Simplified schematic of the control intervention.

More details of the treatment steps are provided in Appendix 1 and the detailed decision tree is shown in Appendix 3.

Measurement of fractional exhaled nitric oxide

FeNO was measured prior to spirometry.

Although FeNO was measured in children in each of the groups at each of the follow-up visits that were attended, the results of FeNO were not used in treatment decisions in the standard-care group of the trial. In this group, the FeNO result was recorded after the assessment was completed and the child had left the assessment room. Participants in this group were blinded to the result and it was not available to the patient’s GP or paediatrician during the trial period.

In the intervention group (FeNO plus symptoms), it was not feasible to blind the participant and parent to FeNO; for example, in the scenario in which a child’s asthma is controlled and treatment stepped down, it will be clear that FeNO has fallen.

Code break/emergency unblinding procedures

There was no requirement for emergency unblinding procedures. This was because knowledge of whether a participant was in the standard-care group or intervention group would not affect any management decisions being taken if an AE occurred.

Study exit at 12-month follow-up for both the intervention and the control group

The treatment algorithm was applied to participants in both groups of the trial at the 12-month assessment. For participants who were recruited in secondary care, the letter that was sent to the GP confirmed that ongoing management of the child would revert to current standard practice.

Data collection

Baseline and safety data were collected during face-to-face assessments at baseline and then outcome and safety data were collected at the 3-, 6-, 9- and 12-month follow-ups. The schedule for data collection is outlined in Table 3.

Baseline

At baseline, the web-based CRF recorded the participant’s details [date of birth, sex, ethnicity, information about their asthma (including current treatment, recent history), co-morbidities, family history of atopic disease, second-hand smoke exposure and other information about environmental exposure]. Spirometry was undertaken using standard methodology (described in Methodology for spirometry) to assess forced expiratory volume in 1 second (FEV₁) inhaler technique, and height and weight were measured.

At the baseline assessment, children and their parent/carer completed asthma questionnaires: for children aged ≤ 11 years the CACT and the Paediatric Asthma Quality of Life Questionnaire (PAQLQ) were used, and for children aged ≥ 12 years the ACT and the PAQLQ were used.

If a smart inhaler device was available for the recommended IC inhaler, this was given to the family along with a charging wire.

At the end of the baseline assessment, a diary card was given to families to record any asthma exacerbations, asthma-related health care and other related resource use. Families were asked to bring it back to follow-up appointments as an aide-memoire to recollect outcome data.

Follow-up visits

Participants were followed up in clinic at 3, 6, 9 and 12 months. The web-based CRF captured current asthma treatment, height, spirometry data (as described for baseline) and FeNO (again as described for baseline). In addition, at the 12-month assessment weight was also measured. Patient-reported adherence to their IC medication was recorded (never, occasionally, about half of the time, most of the time, all of the time). Patients were classed as adherent to IC medication if they reported taking their inhaler ‘most of the time’ or ‘all of the time’. Where children had been provided with a smart inhaler at their previous visit and brought it back to the follow-up visit, adherence data (daily adherence, presented as a percentage) were downloaded from the smart inhaler. Using smart inhaler data, adherence was defined as ≥ 70% of the doses taken. Where there was a mismatch between patient-reported adherence and data from the smart inhaler, the research team explored potential explanations and, based on the discussion, recorded whether or not they thought that the child used the inhaler most/all of the time (adherence ≥ 70%). Similarly, if smart inhaler data were not available, the research team discussed the patient-reported adherence and recorded whether or not they thought that the child used the inhaler most/all of the time (adherence ≥ 70%).

Inhaler technique was also checked at each follow-up visit. Technique was categorised as satisfactory, satisfactory after training (which was provided during the visit) or not satisfactory. Technique was not incorporated into the treatment algorithm.

Information about any asthma exacerbations was recorded at each follow-up, including date of onset, treatment and associated resource use. Asthma exacerbations were not incorporated into the treatment algorithm. Other resource use associated with the child’s asthma was also recorded.

At each follow-up appointment, children and their parent/carer completed the ACT or CACT, depending on the age of the child. At the 12-month appointment, they also completed the PAQLQ.

At the end of the 3-, 6- and 9-month follow-up visits, a diary card (described above) was given to families. In addition, if a smart inhaler was previously not available but was now available, this was given to the family, along with a charging wire.

Those who did not attend for follow-up at 3, 6, 9 or 12 months were contacted by telephone and another appointment was offered. There was a 6-week visit-window before and after each assessment date for that assessment to take place.

Telephone follow-up with families was offered to those who could not attend a follow-up appointment and, on the instruction of the sponsor, all follow-up from 20 March 2020 (around the time of the first COVID-19 lockdown in the UK) was carried out by telephone.

Where the assessment was carried out by telephone, it was not possible to measure FeNO or download adherence data from the smart inhaler; family-reported adherence was recorded. In these cases, for both groups the treatment algorithm was applied without FeNO measurements or smart inhaler data, and in the FeNO group the symptom-only algorithm was applied. In these cases, the treatment algorithm recommendation would be discussed with the PI, particularly if the child had an ACT/CACT test that suggested that symptoms were poorly controlled (score of < 20). The local team was responsible for informing the GP and family if any changes to treatment were needed.

Those who did not attend for follow-up, and for whom telephone follow-up could not be arranged, were sent the CACT or ACT by post and were asked about any asthma exacerbations since their last visit. Those who did not attend the 12-month visit were also sent the PAQLQ.

At the end of the 12-month follow-up period, if primary outcome data had not been obtained from the participant for the full follow-up period, data on asthma exacerbations was sought from primary and secondary care medical records by the local or central trial office team.

Methodology for fractional exhaled nitric oxide measurement

Fractional exhaled nitric oxide was measured using the NIOX VERO^® device (Circassia, Oxford, UK) and standard methodology. The NIOX VERO^® was switched on and a clean mouthpiece was applied. When the machine indicated that it was ready, the participant breathed in slowly through the mouthpiece and then breathed out slowly (as directed by the visual incentive) for 10 seconds (6 seconds if aged ≤ 10 years). Only one technically acceptable measurement was required. Children could make multiple attempts until a technically acceptable measurement was achieved. If children aged > 10 years could not complete the 10-second measurement, they were asked to try the 6-second measurement. Wherever possible, the same apparatus was used for the same individual and the measurement took place at the same time (± 2 hours) at each assessment. The apparatus took 1 minute to produce the result and an acceptable result was indicated by a beep.

Fractional exhaled nitric oxide was measured before spirometry for all participants because FeNO is known to fall slightly after spirometry; therefore, measuring FeNO before spirometry more accurately reflects the value. 39,57 Furthermore, by consistently measuring FeNO before spirometry throughout the trial, intra-subject comparison of FeNO results would be valid.

Methodology for spirometry

The standard methodology was used. 58 Apparatus was calibrated in accordance with manufacturer recommendations. Apparatus varied between centres, but we recommended that the same apparatus was used for each individual throughout their follow-up. Incentive spirometry was used where appropriate. The participant was seated and a nose clip was used. After a full breath in, the participant exhaled as quickly as possible and for as long as possible. As many measurements were taken as were required to produce three technically acceptable measurements. FEV₁ was the spirometric outcome.

Mechanistic components

Data processing

Research nurses at each centre entered data into the study website. Given that the treatment decisions were protocolised as part of the study website, the data were added (ideally) in real time to allow the treatment algorithm to be applied and any treatment step-up/step-down decisions to be communicated with participants and their families during the visit.

Staff in the trial office worked closely with local RNs to ensure that the data were as complete and as accurate as possible. Follow-up questionnaires to participants unable to attend follow-up appointments were sent from and returned to the trial office in Aberdeen (or, if sites preferred, sent from and returned to the site team). Data from questionnaires returned to the trial office were entered into the study website by trial office staff.

As part of the trial’s monitoring plan, the trial office carried out accuracy checks on a sample of questionnaires entered by each site.

Based on the free-text information that research teams added to record why they were not complying with the recommendation of the treatment algorithm, we coded reasons for algorithm non-compliance into seven overarching themes (knowledge, beliefs, behaviour, emotion, environment, social and technical) and 28 subthemes. The reason could be coded into up to four subthemes. We differentiated between knowledge and beliefs on the basis of whether underlying knowledge was suggested in the free text (e.g. it was known that there had been a recent change in treatment) or the decision-making was based on a belief (e.g. that no change in treatment was needed). Where free-text information was provided about who had made the decision, this information was categorised (clinical team, family, joint decision).

Primary outcome

The primary outcome was one or more asthma exacerbations treated with OCs in the 12 months after randomisation (yes/no). This outcome was identified in a meta-analysis43 published before a Cochrane analysis,41 which reached the same conclusions. The outcome is approved as an exacerbation outcome by the American Thoracic Society and European Respiratory Society task force on asthma exacerbations. 13 The decision to prescribe OCs is usually made by clinicians independent of the research team and working in accordance with the national guideline. 17 Some children may have had ‘rescue’ packs of OCs at home and use of such rescue packs for an asthma exacerbation met the criteria for definition as an asthma exacerbation. Asthma exacerbations were determined from the child/parent or carer and supplemented with information from primary and secondary care medical records where appropriate.

Secondary outcomes

There were a number of secondary outcomes:

• time to first exacerbation

• number of exacerbations during follow-up, based on prescribed OCs

• need for unscheduled health-care assessment during follow-up (yes/no)

• number of unscheduled health-care assessments

• asthma control during follow-up (i.e. age-appropriate ACT score of ≥ 20), as used in our observational study and other FeNO studies16,35,54,55

• spirometry during the 12-month follow-up (i.e. %FEV₁, a standard objective index of asthma severity)

• FeNO at the 12-month follow-up

• dose of IC at the 12-month follow-up (i.e. daily dose of budesonide equivalent averaged over 3 months)

• PAQLQ score at 12-month follow-up60

• qualitative outcomes from interviews

• health economic evaluation (derived from participant-reported data, supplemented with information from primary and secondary care medical records where appropriate). 60

Adverse events

The NIOX VERO^® is CE (Conformité Européenne) certified and known to be safe for use in this age group. In RAACENO, we recorded only AEs and serious adverse events (SAEs) that were related to the use of the NIOX VERO^® or other study assessments.

An asthma exacerbation (defined as an increase in asthma symptoms requiring treatment with OCs) was the primary outcome and was not classified as an AE or SAE.

An AE was defined as any untoward medical event affecting a clinical trial participant. Each initial AE was considered for severity, causality or expectedness and was reclassified as a SAE, related or not related to the use of the NIOX VERO^® or other study assessments, based on prevailing circumstances.

A SAE was defined as any AE that:

• resulted in death

• was life-threatening (i.e. the subject was at risk of death at the time of the event; it does not refer to an event which hypothetically might have caused death if it were more severe)

• required hospitalisation or prolongation of existing hospitalisation

• resulted in persistent or significant disability or incapacity

• was a congenital anomaly or birth defect

• was otherwise considered medically significant by the investigator.

Hospitalisations for treatment planned prior to randomisation and hospitalisation for elective treatment of a pre-existing condition were not considered to be AEs or SAEs. Emergency hospital admissions that did not relate to the use of the NIOX VERO^® or other study assessments were not considered to be SAEs.

In this trial, the following events were potentially expected:

• feeling faint following any of the study interventions

• itching (following skin-prick testing)

• coughing/wheezing (following spirometry or sputum collection).

All AEs and SAEs meeting the criteria for recording within RAACENO were recorded from the time that a participant consented to join the trial to the last trial visit.

Participant withdrawal

Participants remained in the trial unless they chose to withdraw consent or they were unable to continue for a clinical reason. All changes in status (with the exception of complete withdrawal of consent) meant that the participant was followed up for all trial outcomes wherever possible. All data collected up to the point of complete withdrawal were retained and used in the analysis. If participants withdrew from the intervention or standard care, they were encouraged to remain in the trial and be followed up as per trial schedule. Participants who wished to withdraw from study follow-up were encouraged to allow routine follow-up data from hospital or GP records to be used for trial purposes. Patients who did not attend for follow-up assessment but for whom any outcome data were available were included in the analysis. Participants who had their asthma diagnosis overturned while participating in the trial were no longer managed through the intervention or control treatment algorithm; data collected up to this point were included in the analysis.

Sample size

We knew from an audit of GP records of children attending the asthma clinic at the Royal Aberdeen Children’s Hospital that 55% of children had received one or more course of OCs in the past year. An exacerbation incidence of 55% is consistent with results from a study of children in secondary care asthma clinics across Scotland; however, we anticipated a lower exacerbation rate of 44% in the control group given that outcomes in children with asthma participating in clinical trials are often better than in observational studies. 54,56 Our meta-analysis found a relative 33% reduction in the proportion of participants with one or more exacerbation receiving FeNO-guided treatment. 43 Assuming an exacerbation proportion of 44% for the symptom-guided treatment group and using a 33% relative reduction as the target difference implies 29.5% exacerbation in the intervention group [equating to an odds ratio (OR) of 0.52]. We had 90% power with 5% significance (two sided) to detect this absolute 14.5% reduction if we recruited 238 children per group. Allowing for 5% incomplete follow-up, we planned to recruit 502 children (i.e. 251 per group).

Statistical analysis of clinical outcomes

All analyses were prespecified in the statistical analysis plan, which was approved by both the TSC and the Data Monitoring Committee (DMC) in advance of analysis. The statistical analysis plan is a stand-alone document. 47 There was no interim analysis or stopping rules. A 5% two-sided significance level was used to denote statistical significance, and estimates are presented alongside their 95% confidence intervals (CIs). No adjustments were made for multiple testing. Analysis was in accordance with the intention-to-treat (ITT) principle and was undertaken in Stata^® version 15 (StataCorp LP, College Station, TX, USA).

We describe categorical variables with number and percentage in each category. We describe normally distributed continuous variables with mean and standard deviation (SD) and skewed continuous variables with median and interquartile range. We report the number of missing data for each variable.

Primary outcome

A logistic regression model was used to determine whether or not the intervention was associated with the primary outcome, that is any exacerbation within 12 months of randomisation. The primary analysis was adjusted for the factors used in the minimisation process to allocate participants to groups: centre, age (< 11 years, ≥ 11 years), sex and asthma severity (BTS step 2, 3 or 4). An unadjusted analysis was also carried out. Odds ratios and their 95% CIs were used as the main measure of effect with an absolute measure (difference in percentages) also reported. This analysis was on those with available primary outcome data (complete-case analysis).

Sensitivity analysis (primary outcome only)

The primary analysis was by ITT, but we also conducted a post hoc per-protocol analysis according to whether or not the treatment algorithm was followed (including individuals in both groups of the trial who complied with the treatment algorithm). For this analysis, a definition of compliance is required. A participant was considered to be compliant with the trial if the treatment decision was based on the algorithm recommendation at three or four out of the four scheduled visits (from baseline, 3, 6 and 9 months). A treatment decision could be made only if a participant attended a visit (or, during the COVID-19 pandemic, this was carried out by telephone). Our definition of compliance meant that children who missed two or more visits could not be classified as compliant in the trial.

A complier-average causal effect (CACE) analysis was also undertaken for the primary outcome. For the CACE analysis, the treatment effect calculated for the primary outcome was divided by the proportion of compliers in the FeNO group (the same definition of compliance was used as described above for the per protocol analysis). The Stata command ‘bpbounds’ was used to obtain the 95% CI. In this analysis, all participants in the standard-care group were assumed to be compliant, irrespective of whether or not the algorithm was followed.

We planned to assess the influence of any missing data on the robustness of the findings and to conduct sensitivity analyses using pattern mixture models or multiple imputation, if appropriate. 61,62

Prespecified subgroup analyses

The analysis for the primary outcome was repeated for eight subgroups. The primary outcome model included a randomised group-by-subgroup interaction term, with the estimates of treatment effect estimated in each group separately, with an interaction p-value reported. Results of the subgroup analyses were presented graphically. The subgroups are as follows:

• baseline treatment (includes LTRA, does not include LTRA)

• age at baseline (< 11 years, ≥ 11 years)

• sex (male, female)

• severity (BTS step 2, BTS step 3, BTS step 4)

• wheeze only with colds (yes, no)

• obesity at baseline (obese, non-obese)

• current eczema in last 12 months (yes, no)

• skin-prick testing (positive for any of cat, house dust mites, hen’s egg, Timothy/mixed grass; yes, no).

Secondary outcomes

The total number of asthma exacerbations, based on prescribed OCs, during the 12-month follow-up period was analysed using negative binomial regression, adjusting for the minimisation factors and presented using an incidence rate ratio (IRR) and its 95% CI. Time to first exacerbation was compared between groups using Cox proportional hazards regression and reported using a hazard ratio (HR) and 95% CI. Four secondary outcomes were collected at baseline and at each of the four study visits: ACT (combined ACT/CACT), FeNO, FEV₁ and dose of IC. These outcomes were analysed using linear mixed-effects models with unstructured covariance to account for the correlation between repeated measures, using Stata’s ‘xtmixed’ command. Centre and participant were included as random effects. Design covariates, treatment and nominal time were included as fixed effects, and a treatment-by-time interaction was included. The primary effect of interest was the between-treatment group marginal difference over all time points. The benefit of this approach is the ability to include all individuals who have had at least one clinical assessment. Unscheduled health-care contacts (yes or no) were compared between treatment groups using logistic regression and the total number of unscheduled health-care contacts were compared using negative binomial regression. Comparison of quality of life (using the total PAQLQ score and its subscales) at the final assessment (12 months) was assessed using linear regression.

Both adjusted and unadjusted analyses were performed for the secondary outcomes. Unadjusted analyses were controlled for the baseline score where possible. All adjusted analyses were controlled for the minimisation factors and, where appropriate, the baseline score.

Exploratory analysis

We carried out a prespecified exploratory analysis, using both randomised groups of the study pooled together. We assessed whether or not changes in FeNO over a 3-month period affected exacerbation (yes/no) in (1) the same 3-month period or (2) the subsequent 3-month period. For this analysis, we used a multilevel model that allowed data from more than one time period for each trial participant to be included.

Post hoc analysis

A post hoc analysis was performed to further examine the relationship between baseline FeNO and future asthma exacerbation. This analysis was restricted to the standard-care group only and included asthma exacerbations for the total follow-up period.

Additional post hoc analyses investigated the impact of COVID-19 (where face-to-face visits were replaced with telephone follow-up) and the impact on algorithm treatment recommendations if children had been allocated to the alternative treatment group.

Health economic evaluation

An economic evaluation compared treatment guided by FeNO plus symptoms with treatment guided by symptoms alone in terms of asthma-related NHS costs, number of asthma exacerbations and expected QALYs over a 12-month follow-up period. The analysis relied on resource use data collected at 3, 6, 9 and 12 months post randomisation, and estimated incidence rates for exacerbations by treatment group applied in a decision analytic modelling framework. The mean difference in costs was estimated for major cost categories of medication use, exacerbation-related costs and background asthma-related health service costs. The model-based cost-effectiveness analysis results were expressed in terms of both the incremental cost per exacerbation avoided and the incremental cost per QALY gained for FeNO plus symptom-guided treatment compared with standard symptom-guided treatment. A further secondary analysis also quantified direct travel costs incurred by participants and their parents in attending any unscheduled asthma-related health-care appointments, and indirect costs associated with time lost from productive activities owing to asthma and associated health-care contacts. Full details of the methods and results of the economic analysis are provided in Chapter 5.

Qualitative assessment

A qualitative process evaluation63,64 was carried out as part of the trial. This was to explore experiences and ascertain acceptability of the intervention and to solicit in-depth feedback on the process of taking part in the trial from both participant and trial staff perspectives. Interviews with families (child and parent pairs) in both the intervention and the control group, and with a range of trial staff representing a number of roles and across different sites, were planned. The full qualitative process evaluation assessment including methodology and rationale is provided in Chapter 6.

Patient and public involvement

A mother of a child with asthma who is also an asthma nurse was an independent voting member of the TSC.

Early versions of the PILs and consent forms were reviewed by the Young Persons’ Advocacy Group (part of the NHS Scottish Children’s Research Network), which provided valuable feedback for the development of these documents. The Young Persons’ Advocacy Group and patient and public involvement (PPI) member of the TSC were asked to comment on the Plain English summary.

We anticipate that the PPI member of the TSC will comment on the results letter to be sent to trial participants and GPs and we also anticipate liaising with the Young Persons’ Advocacy Group for this activity. It is anticipated that the publication of the trial results will be co-ordinated with press releases from Asthma UK and British Lung Foundation Partnership (London, UK).

Protocol amendments

There were six protocol amendments, which are summarised in Appendix 4, Table 46. Most were minor clarifications within the protocol. Of note, we refined the definition of the primary outcome to also include asthma exacerbations treated for < 3 days (e.g. with one dose of dexamethasone, which became common practice after the study started). We also revised details with respect to the qualitative process evaluation and the mechanistic components of the study.

Trial oversight

A TSC, with independent members including a PPI representative, oversaw the conduct and progress of the trial. The TSC met approximately every 6 months. An independent DMC oversaw the safety, rights and well-being of subjects within the trial. The DMC met approximately every 6 months and reviewed accruing data. The DMC did not have any statistical stopping rules.

Breaches

There was one breach recorded within the study: two siblings were recruited into the study at the same time. These siblings remained in the study and are included in the ITT analysis. The breach was assessed by the chief investigator and the sponsor as non-serious.

Recruitment

Participants were recruited to the trial between June 2017 and August 2019, at which point the sample size was achieved (with follow-up completed in August 2020). During this 26-month period, 52 UK sites were opened for recruitment: 35 secondary care sites and 17 primary care sites. Ten primary care sites, all of which opened for recruitment in 2019, did not recruit any patients to the study.

In total, 515 participants were recruited from 42 sites, comprising 499 participants recruited from secondary care and 16 from primary care sites (Table 4). A detailed summary of recruitment by site is given in Appendix 5, Table 47.

The initial protocol included a 24-month recruitment period. There were delays in opening sites as a result of several factors, including the complexity of contract arrangements and staff capacity. Once recruitment started, the accrual of participants persistently rose 6–8 weeks behind the projected recruitment numbers (Figure 4) and recruitment was completed 2 months after the planned end date (without additional funding).

FIGURE 4.

Cumulative recruitment (vs. target recruitment).

Permission was obtained to over-recruit beyond our original target of 502 participants, given that some families had already been approached about the study and had indicated a willingness to be recruited.

Six participants were recruited but did not meet the inclusion criteria and were excluded after randomisation (two in the intervention group and four in the standard-care group). These were children aged < 12 years who were on a dose of ICs higher than the protocol allowed for children of that age. Permission was also obtained to over-recruit to replace these six post-randomisation exclusions.

Baseline characteristics

Baseline characteristics are presented for the 509 included participants (after exclusion of the six post-randomisation exclusions); the two randomised groups were well balanced in terms of demographic and disease characteristics at baseline (Tables 5–9).

The mean age of the participants was 10.1 years and 60.5% were male. The majority of children (61.8%) were a healthy weight, while 21.0% were overweight, 9.1% were obese and 8.1% were thin. The majority (75.6%) of participants identified themselves as being of white ethnicity. The median age that participants’ mothers left education was 18 years. A total of 18.5% of participants were exposed to smoking at home.

The mean age at asthma diagnosis was 3.8 years, and 13.6% of participants were born early (before 36 weeks’ gestation). The median number of admissions to hospital because of asthma in the previous year was one. With regard to previous hospital admissions, 12.6% of participants had never been admitted to hospital for their asthma, 41.5% had been admitted between one and five times during their lifetime and 46.0% had been admitted more than five times. The median days of absence from school in the previous year was 10 days. The median number of courses of steroid tablets for an asthma exacerbation in the previous year was three courses.

At baseline, the median daily dose of IC was 400 µg of budesonide equivalent. In total, 75.8% of participants were prescribed a LABA (usually combined with their IC or in a separate inhaler) and 59.3% were prescribed a LTRA. A total of 52.8% used their rescue inhaler occasionally, while 17.9% used it once per day, 27.9% used it more than once per day and 1.4% reported never using their rescue inhaler. Inhaler technique was satisfactory at the first time of asking in 85.2% of participants and was satisfactory after training in 14.6%; in only one participant (0.2% of participants) was it not considered satisfactory at either stage. Using recognised cut-off points from the ACT/CACT, asthma was poorly controlled in 49.7% of participants and was partly or fully controlled in 50.3% of participants.

The median baseline FeNO was 21 parts per billion (ppb). Lung function was measured at baseline in 455 participants and the mean per cent predicted FEV₁ in children was 90%. It was not possible to measure lung function in the remaining patients owing to a lack of equipment or trained personnel to make these assessments.

The proportions of children reported to have ever had eczema, rhinitis and food allergy were 57.6%, 59.7% and 27.4% respectively.

A total of 62% of participants had a first-degree family member (parent, sibling or half-sibling) with asthma, 51.8% had a first-degree family member with eczema and 37.0% had a first-degree family member with rhinitis.

The median score for the PAQLQ was 5.6 points. The median scores on the individual domains were as follows: symptoms domain 5.5 points, limitations domain 5.4 points and emotional domain 6.0 points.

Bronchodilator response was determined in 160 participants and the mean change was + 10%. Skin-prick testing was carried out in 140 participants, of whom 102 (73%) had at least one positive response.

Primary outcome: asthma exacerbations

Prespecified subgroup analyses

The interaction terms for the subgroup by treatment group interactions were also not statistically significant. Table 11 and Figure 6 detail the results of the analyses for the prespecified subgroups for the primary outcome: any exacerbation within 12 months of randomisation. Results are presented using adjusted ORs and 95% CIs. There was no evidence at the 5% level of statistical significance for any subgroup differences by baseline treatment, age, sex, wheezing status with cold, obesity, eczema in the last year, asthma severity or allergy skin-prick test.

TABLE 11 - Subgroup analyses of the primary outcome

Adjusted for age, sex, asthma severity and centre.

Subgroup factors	N, n (%)		Adjusted OR (95% CI)	p-value	p-value for interaction
	Intervention group	Standard-care group
All participants	255, 123 (48.2)	251, 129 (51.4)	0.88 (0.61 to 1.27)	0.486
Treatment at baseline
Includes LTRA	150, 83 (55.3)	151, 86 (57.0)	0.95 (0.56 to 1.60)	0.842	0.154
Does not include LTRA	105, 40 (38.1)	100, 43 (43.0)	0.81 (0.46 to 1.44)	0.476
Age at baseline
< 11 years	146, 72 (49.3)	139, 70 (50.3)	0.95 (0.62 to 1.46)	0.823	0.630
≥ 11 years	109, 51 (46.8)	112, 59 (52.7)	0.79 (0.42 to 1.49)	0.469
Sex
Male	156, 78 (50.0)	150, 82 (54.7)	0.86 (0.57 to 1.29)	0.457	0.949
Female	99, 45 (45.4)	101, 47 (46.5)	0.86 (0.49 to 1.52)	0.607
Wheeze only with cold
Yes	100, 41 (41.0)	88, 44 (50.0)	0.67 (0.38 to 1.20)	0.179	0.214
No	155, 82 (52.9)	162, 84 (51.9)	1.03 (0.69 to 1.56)	0.872
Obesity
Obese	24, 16 (66.7)	22, 11 (50.0)	1.91 (0.53 to 6.86)	0.324	0.246
Non-obese	229, 106 (46.3)	227, 117 (51.5)	0.81 (0.56 to 1.18)	0.278
Eczema in the last year
Yes	91, 46 (50.65)	90, 44 (48.9)	1.03 (0.59 to 1.81)	0.913	0.655
No	164, 77 (47.0)	161, 85 (52.8)	0.79 (0.50 to 1.25)	0.315
Asthma severitya
BTS step 2	39, 14 (35.9)	37, 9 (24.3)	1.41 (0.45 to 4.37)	0.553	0.559
BTS step 3	109, 50 (45.9)	108, 58 (53.7)	0.74 (0.39 to 1.43)	0.376
BTS step 4	107, 59 (55.1)	106, 62 (58.5)	0.86 (0.53 to 1.40)	0.552
Allergy skin-prick test
Positive	54, 22 (40.7)	48, 19 (39.6)	1.02 (0.51 to 2.04)	0.959	0.244
Negative	16, 7 (43.8)	22, 12 (54.6)	0.56 (0.16 to 2.01)	0.378
Not carried out	185, 94 (50.8)	181, 98 (54.1)	0.89 (0.55 to 1.42)	0.618

FIGURE 6.

Forest plot showing subgroup analyses of the primary outcome.

Secondary outcomes

Time to first exacerbation

Given that more than half of the participants in the intervention group did not have an exacerbation, it was not possible to estimate the median time to first event, although the lower quartile (25th percentile) value was 137 days. The median time to first exacerbation for participants in the standard-care group was 344 days (p25, 112 days) (Figure 7 and Table 12). In the Cox regression analysis, the adjusted HR for the time to first exacerbation was 0.92 (95% CI 0.71 to 1.18) for participants in the intervention group compared with participants in the standard-care group.

FIGURE 7.

Kaplan–Meier plot for time to first exacerbation.

TABLE 12 - Secondary exacerbation outcomes

NR, not reached.

Adjusted for age, sex, asthma severity and centre.

Outcome	Intervention group (N = 255)	Standard-care group (N = 251)	Effect size (95% CI); p-value
Time to first exacerbation (from randomisation)
Number with at least one exacerbation	123	129
Median time to first exacerbation (days)	NR	344	Unadjusted HR: 0.92 (0.72 to 1.17); 0.489
25th percentile (time to first exacerbation) (days)	137	112		Adjusted HR:a 0.92 (0.71 to 1.18); 0.497
75th percentile (time to first exacerbation) (days)	NR	NR
Number of exacerbations
Total number of exacerbations	254	255
Mean number of exacerbations per participant	0.99	1.01
Median number of exacerbations	0	1	Unadjusted IRR: 0.98 (0.77 to 1.26); 0.893
Number of exacerbations, p25, p75	0, 2	0, 2		Adjusted IRR:a 0.99 (0.82 to 1.18); 0.873

Number of exacerbations during follow-up

There were a total of 254 exacerbations among those allocated to the intervention group and 255 among those allocated to the standard-care group (Figure 8; see Table 12). The median (p25, p75) number of exacerbations was 0 (0, 2) in the intervention group and 1 (0, 2) in the standard-care group. The adjusted IRR for the number of exacerbations for those in the intervention group relative to the standard-care group was 0.99 (95% CI 0.82 to 1.18).

FIGURE 8.

Distribution of total number of exacerbations during follow-up.

Percentage forced expiratory volume in 1 second

The mean (SD) percentage predicted FEV₁ at the baseline assessment was 90% (18%) (n = 234) for participants in the intervention group and 89% (18%) (n = 219) for participants in the standard-care group (Figure 9 and Table 14). At the 12-month assessment, the corresponding values were 92% (16%) (n = 156) and 91% (17%) (n = 137). In the repeated-measures model, the overall adjusted mean difference in percentage predicted FEV₁ across all four follow-up time points was 0.24% (95% CI –1.68% to 2.16%).

FIGURE 9.

Mean (error bars denote SD) percentage predicted FEV₁ at each time point.

TABLE 14 - Lung function: percentage predicted FEV1

Adjusted for age, sex, asthma severity and centre.

Time point	Intervention group	Standard-care group	Overall mean difference (95% CI); p-value
Baseline
Total (n)	234	219
Mean (SD)	89.8 (17.8)	89.3 (18.3)
3 months
Total (n)	210	188
Mean (SD)	93.3 (18.4)	91.9 (16.8)
6 months
Total (n)	193	183
Mean (SD)	93.0 (16.3)	93.4 (16.8)
9 months
Total (n)	179	162
Mean (SD)	93.2 (16.5)	92.7 (16.3)
12 months
Total (n)	156	137
Mean (SD)	92.2 (16.0)	91.0 (17.4)
Unadjusted			0.37 (–1.60 to 2.34); 0.713
Adjusteda			0.24 (–1.68 to 2.16); 0.807

Fractional exhaled nitric oxide

The median (p25, p75) FeNO at the baseline assessment was 20 ppb (11, 45 ppb) (n = 255) for participants in the intervention group and 23 ppb (10, 51 ppb) (n = 254) for participants in the standard-care group. At the 12-month assessment, the corresponding values were 19 ppb (10, 47 ppb) (n = 163) and 22 ppb (11, 39 ppb) (n = 147) (Figure 10 and Table 15). Across the four follow-up time points, the overall mean group difference in FeNO was 2.75 ppb (95% CI –0.97 to 6.47 ppb).

FIGURE 10.

Median (error bars denote p25 and p75) FeNO at each time point (ppb).

TABLE 15 - FeNO results (parts per billion) as measured in participants at different time points

Adjusted for age, sex, asthma severity and centre.

Time point	Intervention group	Standard-care group	Overall mean difference (95% CI); p-value
Baseline
Total (n)	255	254
Median (p25, p75)	20.0 (11, 45)	22.5 (10, 51)
3 months
Total (n)	219	209
Median (p25, p75)	17.0 (10, 34)	15.0 (9, 33)
6 months
Total (n)	207	202
Median (p25, p75)	17.0 (9, 40)	17.0 (9, 30)
9 months
Total (n)	190	181
Median (p25, p75)	19.0 (10, 44)	17.0 (11, 37)
12 months
Total (n)	163	147
Median (p25, p75)	19.0 (10, 47)	22.0 (11, 39)
Unadjusted			2.59 (–1.16 to 6.35); 0.176
Adjusteda			2.75 (–0.97 to 6.47); 0.147

Daily dose of inhaled corticosteroid

At baseline, the median (p25, p75) daily dose of IC was 400 µg (400 µg, 1000 µg) (n = 255) of budesonide equivalent for participants in the intervention group and 400 µg (400 µg, 1000 µg) (n = 251) of budesonide equivalent for participants in the standard-care group. At the 12-month follow-up, the corresponding figures were 500 µg (400 µg, 1000 µg) (n = 224) of budesonide equivalent and 400 µg (400 µg, 1000 µg) (n = 208) of budesonide equivalent (Figure 11 and Table 16). The overall adjusted mean group difference in IC was –10.88 µg (95% CI –70.76 µg to 49.01 µg).

FIGURE 11.

Daily IC dose (budesonide equivalent) (µg) (median, error bars denote p25, p75 at each time point).

TABLE 16 - Inhaled corticosteroids (IC budesonide equivalent) (µg)

Adjusted for age, sex, asthma severity and centre.

Time point	Intervention group	Standard-care group	Overall mean difference (95% CI); p-value
Baseline
Total (n)	255	251
Median (p25, p75)	400.0 (400, 1000)	400.0 (400, 1000)
3 months
Total (n)	228	216
Median (p25, p75)	800.0 (400, 1000)	500.0 (400, 1000)
6 months
Total (n)	211	210
Median (p25, p75)	500.0 (400, 1000)	400.0 (400, 1000)
9 months
Total (n)	209	198
Median (p25, p75)	500.0 (400, 1000)	400.0 (400, 1000)
12 months
Total (n)	224	208
Median (p25, p75)	500.0 (400, 1000)	400.0 (400, 1000)
Unadjusted			–10.62 (–71.12 to 49.88); 0.731
Adjusteda			–10.88 (–70.76 to 49.01); 0.722

Asthma control

The ACT (score range of 0–25, definition of partial or full control is a score of > 19) and CACT (score range of 0–27, definition of partial or full control is a score of > 19) were combined and presented as a single measure. The median (p25, p75) ACT/CACT score was 19 (14, 22) (n = 255) for participants allocated to the intervention group and 18 (14, 21) (n = 251) for participants allocated to the standard-care group (Figure 12 and Table 17). At the 12-month follow-up, the corresponding scores were 22 (18, 24) (n = 222) and 21 (18, 24) (n = 208). The overall adjusted mean group difference in ACT/CACT across the 12-month follow-up was 0.05 (95% CI –0.47 to 0.58).

FIGURE 12.

Asthma control (ACT/CACT) (median, error bars denote p25, p75) at each time point.

TABLE 17 - Asthma control

Adjusted for age, sex, asthma severity and centre.

Time point	Intervention group	Standard-care group	Overall mean difference (95% CI); p-value
Baseline
Total (n)	255	251
Median (p25, p75)	19.0 (14, 22)	18.0 (14, 21)
3 months
Total (n)	231	224
Median (p25, p75)	20.0 (16, 24)	21.0 (17, 23)
6 months
Total (n)	214	215
Median (p25, p75)	21.0 (17, 24)	20.0 (17, 23)
9 months
Total (n)	213	204
Median (p25, p75)	22.0 (18, 24)	22.0 (18, 24)
12 months
Total (n)	222	208
Median (p25, p75)	22.0 (18, 24)	21.0 (18, 24)
Unadjusted			0.04 (–0.50 to 0.57); 0.895
Adjusteda			0.05 (–0.47 to 0.58); 0.840

Quality of life

Quality of life was measured using the PAQLQ. The overall PAQLQ scores (and scores for the individual domains of the PAQLQ) range from 1 (severely affected) to 7 (unaffected). At baseline, the median (p25, p75) overall PAQLQ score was 5.7 (4.3, 6.6) (n = 254) in the intervention group and 5.5 (4.2, 6.4) (n = 254) in the standard-care group (Figure 13). At the 12-month follow-up the corresponding figures were 6.4 (5.5, 6.9) (n = 215) and 6.2 (5.4, 6.8) (n = 199) (Table 18). The adjusted mean difference in total PAQLQ scores at the 12-month follow-up was 0.098 (95% CI –0.088 to 0.283) between the groups.

FIGURE 13.

Quality of life (PAQLQ) (median, error bars denote p25, p75) at baseline and 12 months. (a) Overall score, (b) symptom domain score, (c) activity limitation domain score and (d) emotional function domain score.

TABLE 18 - Quality of life (PAQLQ)

Adjusted for age, sex, asthma severity and centre.

Outcome	Intervention group (N = 215)	Standard-care group (N = 199)	Mean difference (95% CI); p-value
PAQLQ score at 12 months: overall quality of life
Median (p25, p75)	6.39 (5.52, 6.86)	6.22 (5.35, 6.78)
Unadjusted			0.092 (–0.095 to 0.279); 0.333
Adjusteda			0.098 (–0.088 to 0.283); 0.302
PAQLQ score at 12 months: symptoms domain
Median (p25, p75)	6.33 (5.33, 6.89)	6.11 (5.22, 6.67)
Unadjusted			0.162 (–0.050 to 0.374); 0.135
Adjusteda			0.169 (–0.042 to 0.379); 0.115
PAQLQ score at 12 months: limitations domain
Median (p25, p75)	6.20 (5.20, 6.80)	6.20 (5.20, 6.80)
Unadjusted			–0.012 (–0.216 to 0.192); 0.908
Adjusteda			–0.007 (–0.209 to 0.196); 0.947
PAQLQ score at 12 months: emotional domain
Median (p25, p75)	6.63 (5.75, 7.00)	6.63 (5.75, 7.00)
Unadjusted			0.074 (–0.116 to 0.264); 0.445
Adjusteda			0.079 (–0.111 to 0.268); 0.413

There were no differences between treatment groups during follow-up with respect to the symptoms, limitations and emotional domains of the quality-of-life score (see Table 18).

Algorithm recommendation at each clinic attendance

At baseline, the majority of recommendations from the treatment algorithm in both groups was a step up in treatment (68.2% in the intervention group and 59.8% in the standard-care group; Table 19). A step down in treatment was possible at baseline in the intervention group; 22.4% of recommendations were for a step down in treatment at that time point. Such a step down was not possible at baseline in the standard-care group. At subsequent time points, the proportion of step-up recommendations from the treatment algorithm was lower than that at baseline. At baseline and 3, 6 and 9 months, the proportion of recommendations involving no change in treatment was higher in the standard-care group than in the intervention group (3 months: 31.5% vs. 7.1%; 6 months: 20.6% vs. 13.7%; and 9 months: 26.1% vs. 19.0%, respectively).

Compliance with the recommendation of the treatment algorithm

At each time point (baseline and the 3-, 6- and 9-month follow-ups), between 20.0% and 24.0% of treatment recommendations made by the algorithm were not followed (Table 20). Overall non-compliance with the algorithm took place in 200 out of 887 (23%) assessments of participants in the intervention group and 177 out of 874 (20%) assessments of participants receiving standard care (see Table 20). The proportion of assessments in which the algorithm recommendation was not followed was similar in both treatment groups (see Table 20). Where the treatment recommendation was ‘remain the same’, compliance with the treatment recommendation was much more common than if the recommendation was to step up or step down treatment.

Research teams added free text that recorded why they were not complying with the recommendation of the treatment algorithm, which was coded into overarching themes and subthemes (see Chapter 2).

The majority of reasons for non-compliance with the treatment algorithm were based on beliefs, most frequently that no step up or step down in treatment was required (Table 21). The second most frequently cited reason for non-compliance was knowledge, including knowledge of previous side effects of LTRA (when the treatment algorithm recommended a step up in treatment to include LTRA). Other frequently cited reasons included non-adherence to current treatment (within the behaviour theme) and the cold season (within the environment theme). The pattern of reasons was similar across all time points (Table 22). The decision-maker in the majority of the decisions was the clinical team (Table 23).

Safety

Safety reporting within the study was limited to events related to the use of the NIOX VERO^® or other study assessments. There were no SAEs or deaths in the trial. In total, 28 participants experienced at least one adverse reaction during the trial: 12 in the intervention group and 15 in the standard-care group (Table 24).

Exploratory analysis: relationship between change in fractional exhaled nitric oxide and asthma exacerbation

Table 25 explores whether or not the change in FeNO from the previous FeNO measurement to the current follow-up was associated with an exacerbation in the same 3-month follow-up period. The adjusted OR for asthma exacerbation was 1.02 (95% CI 0.97 to 1.08) for a 10-unit (ppb) change in FeNO. Similarly, it was investigated whether or not the change in FeNO in a given 3-month period was associated with an asthma exacerbation during the subsequent 3-month follow-up period. The adjusted OR for asthma exacerbation was 1.05 (95% CI 0.99 to 1.12) for a 10-unit (ppb) change in FeNO.

Post hoc analysis

Introduction

This chapter reports the methods and results of the economic analysis that was carried out as part of the RAACENO trial and includes a discussion of the modelling. The economic evaluation compares treatment guided by FeNO and symptoms (intervention) with treatment guided by symptoms alone (standard care) in terms of asthma-related NHS costs, the number of asthma exacerbations and total QALYs over a 12-month follow-up period. Costs falling directly on participants and indirect costs associated with time lost from productive activities are also considered in a separate analysis. The methods for measuring and valuing the resource use are summarised and the results are presented by a participant’s treatment allocation. These analyses are then incorporated into a simple Markov model to assess cost-effectiveness.

Objectives of the economic evaluation

The primary economic objective, as stated in the trial protocol, was to undertake an economic evaluation to assess the asthma-related health-care costs, direct and indirect costs to participants, and quality-of-life effects (QALYs) of the intervention compared with routine care over the 12-month trial follow-up period to identify an optimal treatment regime in terms of cost-effectiveness. The analysis relied on medication and asthma-related health-care resource use data collected at 3, 6, 9 and 12 months post randomisation, and estimated incidence rates of exacerbations by treatment allocation group applied in a decision-analytic modelling framework. Cost-effectiveness analysis results are presented as the cost per exacerbation avoided and the incremental cost per QALY gained. The perspective is that of the UK NHS. The health economics analysis plan is a standalone document. 47

Methods

Modelling of cost-effectiveness

The cost-effectiveness analysis utilised a simple decision model (Figure 14) to determine the incremental cost per exacerbation avoided and the incremental cost per QALY gained with the intervention compared with standard care. The model utilised a 1-year time horizon, in line with the trial follow-up period, but we also assessed the impact of extending this to 5 years. The model took the form of a simple Markov model with a 2-weekly cycle length in line with the exacerbation definition utilised in the trial, which was informed through clinical consultation. The model was informed by the trial resource use data (translated into background costs, treatment costs per participant per year and the cost per exacerbation), the exacerbation rate by treatment group and published evidence on the health-related quality-of-life impact of asthma exacerbations. 80,81 Furthermore, the cost of an outpatient appointment with a consultant (see Table 29) was applied at 3-month intervals for both treatment groups in the model. This is consistent with the trial protocol to reflect the resource use required to achieve the effectiveness outcomes utilised.

FIGURE 14.

Schematic of the decision model.

The utility decrements associated with asthma exacerbations were applied to the expected number of asthma exacerbations for each cycle of the model. The utility decrements for asthma exacerbations reported in the literature for those requiring inpatient hospitalisation differ from those for patients managed without inpatient admission; therefore, a weighted average of QALY decrements was applied based on the proportion of exacerbations requiring hospitalisation from the trial. No death events occurred during the trial; therefore, it was assumed that the intervention has no effect on survival and that any difference in QALYs between groups is wholly driven by differences in the number of asthma exacerbations over the follow-up period. This approach is in line with a previous National Institute for Health and Care Research-funded economic evaluation, in which additional asthma-related mortality was not considered in the analysis. 49

The model was used to estimate the expected costs, numbers of exacerbations and QALYs over the observed time horizon of the trial, assuming a constant incidence rate for exacerbations by treatment group. Given the time horizon of the model, no discount rate was applied to QALYs or costs. Furthermore, it was assumed that there is no additional risk of death over the UK age-specific risk for participants in the cohort because no death events occurred in the trial. All results are presented using probabilistic analysis, in terms of both the incremental cost per exacerbation avoided and the incremental cost per QALY gained. These were calculated using the modelled expected differences in costs, exacerbations and QALYs between treatment groups. Input parameters are assigned probability distributions to reflect uncertainty in their mean estimates. Probability distributions were determined by the nature of the data that they describe. Utility decrement and total cost parameters were assigned gamma distributions. The incremental costs of the intervention and rate parameters were applied normal distributions and log-normal distributions, respectively. The variance for each cost input parameter was derived from the analysis of the trial data described in the following section. Where the standard error was not known, a value of 10% was assumed. A Monte Carlo simulation of 10,000 draws was used to characterise the uncertainty surrounding the model-based estimates of cost-effectiveness, with mean costs, QALYs and exacerbations obtained by taking the average over 10,000 simulations. Results are displayed graphically in the form of the cost-effectiveness plane and cost-effectiveness acceptability curve. 82,83

Results

Cost-effectiveness analysis results

Table 39 presents the probabilistic results of the base-case cost-effectiveness analysis. Although the expected cost and QALYs are slightly higher in the intervention group, the differences are very small (close to zero). Considering the uncertainty surrounding the expected differences, Figure 15 shows the joint difference in cost and effect from 10,000 probabilistic simulations plotted on the incremental cost-effectiveness plane. It can be noted that the probabilistic points are scattered quite equally around the origin of the plane, representing zero difference in costs (y-axis) and zero difference in effects (x-axis). Thus, the probability of the intervention offering a cost-effective approach compared with standard symptom-guided treatment never rises higher than 49%, irrespective of the monetary value placed on a QALY (Figure 16). The potential viability of the intervention is negatively influenced by the cost of FeNO testing. Where the throughput of FeNO testing was assumed to be between 60 and 300 tests per year, the probability of the intervention being cost-effective decreases to 34% (see Appendix 10, Table 57). Should the pooled cost of asthma exacerbations be utilised in the model, the probability of the intervention being cost-effective reduces to 41% (see Appendix 10, Table 57).

TABLE 39 - Cost-effectiveness results (per participant)

‘Standard care’ is treatment guided by symptoms only and the ‘intervention’ is treatment guided by FeNO and symptoms.

Strategya	Cost (£)	Δ Cost (£)	Number of exacerbations (n)	Δ Exacerbations (n)	QALYs	Δ QALYs	Incremental cost per exacerbation avoided (£)	Incremental cost per QALY gained (£)	Probability cost-effective at
									£13,000	£20,000	£30,000
Standard care	2007	–	1.03	–	0.95	–	–	–	0.52	0.52	0.53
Intervention	2014	7	1.02	–0.01	0.95	0.00	–709	165,239	0.49	0.49	0.49

FIGURE 15.

Incremental cost-effectiveness scatterplot (intervention vs. standard care).

FIGURE 16.

Cost-effectiveness acceptability curve (intervention vs. standard care).

Discussion

The results of the economic analysis indicate no statistically significant difference in any category of cost over the follow-up period of the trial, which is consistent with the lack of effect on asthma exacerbations. Correspondingly, the model-based cost-effectiveness analysis found no meaningful difference in expected costs or QALYs between the treatment groups, with the simulated differences close to zero. Considering the uncertainty around the estimated differences, the probability of FeNO plus symptom-guided treatment (intervention) being cost-effective compared with standard care remained below 50%, regardless of willingness to pay per QALY gained. Furthermore, while there was no statistically significant difference in treatment, background or exacerbation-related costs, FeNO testing would result in a small increase in monitoring costs per participant, without a demonstrable benefit. Therefore, our cost-effectiveness findings based on data from RAACENO do not support the widespread adoption of FeNO testing for guiding treatment in children.

A strength of this economic analysis is that it is based on data from a large multicentre pragmatic RCT of participants from England and Scotland. Therefore, it can be assumed that the results have high internal validity and are generalisable across the NHS. The approach of RAACENO was detailed with respect to the collection of health-care utilisation data; for the observed exacerbations, a large number of complete resource use data were collected. This analysis considered only the short-term health-care costs of exacerbations treated with OCs. However, there is evidence that treatment with OCs, such as prednisolone, carries long-term health risks, such as growth impairment, pubertal disorders, hirsutism, diabetes, hypertension and many more. 86–89

A limitation of the economic analysis is the lack of health-related quality-of-life data for children experiencing exacerbations. Furthermore, previous evidence has suggested that hospitalisations may result in greater quality-of-life loss in children than in adults. 82 However, the exacerbation rate and the proportion of exacerbations requiring hospitalisation were not found to be different between the treatment groups. Therefore, it is unlikely that the uncertainties surrounding these point estimates of utility decrements would have an impact on the cost-effectiveness results. Another potential weakness relates to the fact that no direct data were collected on the severity or duration of exacerbations in RAACENO, and these factors may affect the extent of QALY losses associated with exacerbations. 82,90 Nevertheless, it may be assumed that there was no meaningful difference in the severity of exacerbations between treatment groups in RAACENO, given that there was no meaningful or statistically significant difference in the resource use required to manage them. Another limitation to be considered is that the frequency of monitoring appointments within the RAACENO trial was greater than in NHS clinical practice: four appointments per year compared with two appointments per year. Therefore, there is a risk that we have captured protocol-induced costs and effects in this analysis. However, given that all results were not statistically significant, we are confident that it would not affect our conclusions. Finally, we did not incorporate asthma-related mortality risk into our analysis. Although asthma exacerbations can be potentially life-threating, no death events occurred during trial follow-up. Furthermore, given the short model time horizon of 1 year, it was determined that such an inclusion would have a negligible, if any, impact on the results.

A previous Health Technology Assessment report,49 which included a systematic review and economic modelling study to assess the clinical effectiveness and cost-effectiveness of using FeNO for the diagnosis and management of asthma in children, indicated the need for further cost-effectiveness evidence using data with a follow-up of at least 1 year. This highlighted the importance of the RAACENO trial. Based on a review of efficacy data from eight earlier studies, the report49 found that asthma management using FeNO and symptoms could be expected to produce a modest reduction in exacerbations in children over a lifetime horizon, compared with treatment guided by symptoms alone. Correspondingly, the health economic modelling suggested that FeNO testing would be expected to produce a small QALY gain compared with standard management for an increased cost, which is mostly due to an increase in the use of ICs. The RAACENO trial data did not provide support for a meaningful reduction in exacerbations or the use of preventative treatments, and consequently our economic evaluation results are not consistent with those of this previous Health Technology Assessment report. 49 Owing to the flat findings at 12 months, we did not extend the time horizon of the health economic model for RAACENO beyond the time horizon of the trial.

Another more recent economic analysis,50 which was carried out alongside a RCT of 268 children aged between 4 and 18 years in the Netherlands, also compared FeNO plus standard care with standard care alone over a period of 12 months. The standard care was based on the BTS guidelines for a step-up or step-down treatment approach based on the child’s symptoms. 17,55,91 Similar to RAACENO, this study found no statistically significant differences in QALYs or total health-care costs between the standard-care group and the FeNO plus standard-care group. 50

In conclusion, the economic evaluation based on RAACENO participant-level data does not provide evidence to support the cost-effectiveness of treatment guided by FeNO plus symptoms compared with treatment guided by symptoms alone in children with asthma who are at risk of an asthma exacerbation. No statistically significant or meaningful differences in health service costs, direct participant costs or indirect costs to society were identified.

Introduction

The UK Medical Research Council outlines several recommendations for the evaluation of complex interventions, which include the use of both quantitative methods, to assess effectiveness and cost-effectiveness, and qualitative methods, to understand processes of change. 92 The document offers guidance on how these should be applied before, during and after clinical trials. Qualitative research is usually conducted prior to the commencement of a trial to assess the feasibility and acceptability of the intervention and protocol. This pre-trial approach provides an opportunity to develop solutions designed to address potential barriers to and facilitators of various aspects of the trial before its implementation. Qualitative research can also be used in a process evaluation during and after a trial to understand change processes through exploring the experiences of those delivering and participating in a study, as well as the acceptability of the intervention. 64 Process evaluations are usually aligned with a ‘theory of change’ or ‘programme theory’, or use a logic model. 64

Prior to the RAACENO trial, no qualitative work was undertaken by the study team to establish the feasibility and acceptability of the intervention. Although the FeNO aspect of the intervention has a strong evidence base,43 the introduction of algorithms and adherence monitors into childhood asthma management was novel; however, some studies had indicated their feasibility and acceptability. 53–55 Given that the combination of the intervention ingredients was new in this population and condition, we undertook qualitative research during the trial with those delivering and participating in our study. Our ‘theory of change’ was underpinned by the hypothesis that a reduction in asthma exacerbations in children could be achieved by using FeNO as a biomarker to inform treatment strategy, as per the main trial protocol. 47

The qualitative process evaluation described in this chapter was conducted concurrently with the RAACENO trial to explore key aspects of staff and patient experiences, as well as the acceptability of the intervention ingredients together and in context. We explicitly focused on perspectives around the use of (1) FeNO, together with (2) algorithm-generated treatment recommendations plus clinical decision-making for medication prescription, and (3) the smart inhalers utilised to monitor medication adherence. We interviewed trial staff to explore health-care decision-making based on the intervention components and beliefs about their benefits for asthma management among children. We interviewed trial participants (patients and their parents) to elicit their experiences and views of the intervention components in determining and monitoring their treatment, including perceptions of effectiveness.

Aim and objectives

The overarching aim of this component of the project, outlined within the grant application and published protocol,47 was to undertake a qualitative process evaluation to explore staff and family experiences and the acceptability of the intervention.

The specific objectives were to:

• garner views on the RAACENO intervention from a range of trial staff representing a number of roles and different sites to understand intervention delivery from provider perspectives and to access any additional observations made around acceptability from their perspectives and their impressions of their patients’ views

• solicit in-depth feedback on the process of taking part in this trial from the perspectives of the parents and children who participated, as well as their views on the RAACENO intervention

• synthesise our findings and report them as a process evaluation alongside the main trial outcomes.

Methods

Findings

Data

Discussion

Conclusions

In this qualitative process evaluation, trial staff representing a number of roles and across different sites were interviewed to understand their experiences of delivering RAACENO from provider perspectives and to explore the acceptability of the intervention from their points of view. A number of families (mother and child pairs) were also interviewed about their experiences and views. Overall, experiences within both groups were positive. Key was that the RAACENO trial impacted positively on staff–family relationships and communication around asthma management and treatment among children, and that the use of technology and individual data within clinical appointments was considered useful: closer monitoring and the educational impacts were especially highlighted. We also ascertained that the intervention was broadly acceptable, with caveats around clinicians using the algorithm recommendation as a guide (rather than being dictated by it) and wariness around extreme step ups/downs in the light of contextual factors not taken into account by the algorithm.

Summary of the main findings

There is evidence that using FeNO to guide asthma treatment in patients already receiving maintenance therapy may reduce the risk of asthma exacerbations in children, but not other outcomes. This study was designed to determine whether or not intervention with an asthma treatment algorithm, which considered FeNO, current symptoms, current treatment and adherence to treatment among children, was more effective in reducing asthma exacerbations than ‘standard care’, which differed by not including FeNO. We recruited children whose characteristics were typical for those seen in a secondary care asthma clinic in the UK, with a high burden of exacerbation and symptoms and for whom an objective measurement, such as FeNO, to guide asthma treatment decisions would be welcome. The desired sample size was achieved, the participant characteristics were consistent with children who had more troublesome asthma and the primary outcome had the expected incidence. There were no SAEs or deaths. The primary outcome was determined in 99% of participants. The intervention was not associated with reduced odds for the primary outcome, that is one or more exacerbation, or for secondary outcomes between participants in the two treatment groups. In the adjusted model, the OR for the primary outcome was 0.88 (95% CI 0.61 to 1.27) for participants allocated to the intervention group compared with the standard-care group. The CI for the main result includes the possibility of a 33% reduction in the odds of having an exacerbation (as specified in the sample size calculation). Intentional non-adherence to the algorithm-generated treatment recommendations occurred and the analysis found that the primary outcome did not differ between randomised groups where the algorithm recommendations were followed. Interviews with staff, trial participants and their parents gave insights into the acceptability of algorithm-generated treatment recommendations and the use of electric logging devices, which objectively measure treatment adherence. The health economic evaluation gave insights into the educational and financial implications for children with asthma and their families. However, we conclude that, for children attending secondary care asthma clinics already receiving maintenance therapy, the addition of FeNO to standard care does not improve any of the outcomes captured in the RAACENO study.

Relevance to the existing literature

Strengths and limitations

Our study has a number of strengths. First, RAACENO recruited the desired number of participants. Second, the primary outcome was determined for 99% of participants (exceeding the expected 95%) and the expected and actual incidence of the primary outcome among participants randomised to receive standard care were 50% and 51.4%, respectively, which collectively means that RAACENO was adequately powered to conclude that the intervention did not achieve the predicted reduction in the primary outcome. Third, we recruited the population of children with troublesome asthma as intended. For example, participants typically had three exacerbations in the year prior to randomisation and the majority required add-on preventer therapy; any intervention to reduce morbidity would be welcome for this population compared with a population with mild asthma. A fourth strength is that the computer-delivered treatment algorithm ensured that standardised care was recommended to all participants in all trial centres. A fifth strength is that RAACENO was designed on an evidence base built from data collected in previous trials in which FeNO was used to guide asthma treatment in children; for example, we used percentage change in FeNO (and not absolute change) to guide treatment,45 we applied a more clinically meaningful and liberal threshold than in previous trials when defining a change in FeNO,46 and the analysis explored whether or not the intervention may have been effective in previously described subgroups. 45 Further strengths are the design of the study, which would enable easy roll-out into clinical practice and an objective marker of adherence to ICs, and that health economic and qualitative evaluations were included.

There are a number of limitations to our study, which should be considered. First, it is possible that the assumptions that were made in the algorithm were wrong. For example:

1. A well-recognised cut-off value from a validated symptom score was used to define uncontrolled asthma,15 but control is a continuum and an algorithm that included more than two categories of control (e.g. controlled, partly controlled and uncontrolled) may have yielded different results. However, this would have affected both groups of the trial and, therefore, would have been unlikely to affect the outcome.

2. The change in FeNO, which triggered a change in treatment, was based on the best evidence available,35,46 but this threshold might not have been correct. There are many factors that affect FeNO independent of asthma43 and a lower threshold would have lower sensitivity for detecting risk of future exacerbations. A higher threshold may have lacked specificity and cases at increased risk for future exacerbation would not have been detected. An alternative FeNO strategy would be to have censored all values below 20 ppb as normal;39 however, the median FeNO value at baseline was 21 ppb and such a strategy would probably have made treatment options between the two treatment groups more similar. RAACENO is one of a number of trials to have applied different FeNO criteria to changes in asthma treatment and it seems increasingly likely that there is no one-size-fits-all method to interpreting changes in FeNO.

3. Adherence to IC medication may have been over-reported. Participants with unrecognised poor adherence will have raised FeNO. 105 In the case of participants in FeNO-guided treatment group, this would lead to treatment escalation, but this would not be effective in reducing exacerbations owing to non-adherence, thus reducing the efficacy of the intervention. However, our algorithm permitted only one step up in treatment for asymptomatic individuals with raised FeNO (whose raised FeNO may reflect poor adherence), and the risk of participants being non-adherent to IC treatment was reduced by using the smart inhaler in many participants.

4. The definition of adherence that was applied, < 70% by objective measurement and neither ‘all of the time’ or ‘most of the time’ by report, may have been too liberal. Other RCTs that have evaluated the added benefit of FeNO to standard care applied a cut-off point of 50% for satisfactory adherence (more generous than RAACENO),54 excluded participants who were poorly adherent106,112,122 and reported a median adherence of 97% (range 55–124), as evidenced by a counter on the inhaler. 104 Our study was designed such that the intervention, if effective, could be rolled out throughout the NHS. For this reason, it did not incorporate elements used in other trials that may have reduced non-adherence, for example, a run-in period to identify non-adherent participants, who would subsequently be excluded. The threshold was applied to both treatment groups and, therefore, is unlikely to have affected the outcome in one group more than the other.

5. The treatment options given to participants in the two treatment groups may not have been sufficiently different to create a difference in outcomes. When the algorithm decisions were modelled after artificially allocating participants in the standard treatment group to the FeNO-guided treatment group, we found that treatment advice would have differed in 40% of participants. Two previous FeNO trials, which found no reduction in exacerbations between study groups, reported that, for 62%55 and 64%111 of assessments, the treatment advice would have been different if participants had been allocated to the other study group. A third FeNO trial,109 which reported reduced risk for exacerbation associated with FeNO-guided treatment, found that the ratio of discordant to concordant decisions, had participants been allocated to the other study group, was 1 : 1.05.

6. Our pragmatic study design used an ‘open-label’ design, given that participants, researchers and the clinical team knew which group participants were randomised to. Although the study could have had a double-blind design, this would have made the study less pragmatic and ‘real world’, and would have added to the complexity, cost and, potentially, ethics considerations. Therefore, we believe that the open-label design is a relative limitation to RAACENO, but one that is unlikely to have meaningfully affected the outcome.

Our initial application to the Efficacy and Mechanism Evaluation programme included a mechanistic component that was designed to explore the complex mechanistic relationship between sputum eosinophilia, FeNO, asthma control and exacerbation. Obtaining sputum for analysis on two occasions proved extremely challenging. Samples need to be processed immediately, and most centres did not have a local laboratory that was able to do this. Data from a study that used sputum eosinophilia to guide asthma treatment36 were shared with the RAACENO CI pre publication, and analysis demonstrated that there was a positive intra-subject correlation between percentage change in FeNO and sputum eosinophilia. Given the challenge in obtaining sputum samples and the presence of data that answered the research question, collection of sputum samples for measurement of eosinophils was stopped with the approval of the TSC and funder.

Generalisability

The RAACENO trial participants were representative of children attending asthma clinics in the UK. For example, the observational Paediatric Asthma Gene Environment Study (PAGES)56 recruited children from hospital asthma clinics across Scotland in 2008–9. The characteristics of the PAGES participants, who were aged 6–16 years, and the RAACENO participants are compared in Table 45. The RAACENO participants were selected for having had a recent asthma exacerbation and receiving higher IC treatment, and had lower %FEV₁ and bronchodilator-derived response, but were very similar to the PAGES participants on other outcomes. Every assessment included a structured history, enquiring about treatment adherence, checking inhaler technique and in many cases taking spirometry measurements; all of these aspects are recommend by UK national guidelines for the management of asthma. 17 The RAACENO results are, therefore, generalisable to the population of children managed in secondary care asthma clinics across the UK and, therefore, the subset of children across the UK who have the greatest burden of morbidity from asthma.

Public and patient involvement

The TSC included a lay member. The study documentation was created with the input of the Young Persons’ Group (a PPI group). The summary of the results that will be provided to participants and their families will be reviewed by the Young Persons’ Group.

Conclusions

In children with asthma being cared for in secondary care asthma clinics, the addition of FeNO to symptom-guided treatment did not reduce the OR for an asthma exacerbation.

Implications for practice

The RAACENO findings do not support the routine use of FeNO measurements as part of asthma management in a secondary care setting.

Recommendations for research

The role of FeNO in the diagnosis of asthma is increasingly recognised, and this work should continue. There may be some settings in which the addition of FeNO to standard care may improve patient care and future research might explore scenarios that include primary care and suspected poor adherence to IC treatment. In the RAACENO trial and other previous trials in children, FeNO was used in addition to symptoms, but in adults FeNO has been used to guide treatment independent of symptoms, either as a single biomarker123 or in combination with other biomarkers. 108 Future research in children could explore whether FeNO could be used to guide asthma treatment independent of symptoms.

RAACENO identified how some clinicians and parents/carers can be reticent to accommodate algorithm-recommended treatment and our qualitative interviews identified some of the reasons for this. Future trials could consider individualising algorithms to consider the time of year, the individual’s past exacerbation history and their future plans (e.g. holiday or sporting event). In addition, in the absence of a single ‘gold-standard’ test for monitoring asthma, two or more objective tests could be used to help to support stepping down treatment decisions, as has been previously described,103 and an algorithm based on these tests could give treatment decisions independently of symptoms.

The intervention delivered in the RAACENO trial was the addition of FeNO to an already complex intervention, which included regular standardised assessments, spirometry and an objective index of adherence. Future research could seek to understand which (if any) single component of a management intervention is associated with improved asthma outcomes. Such trials could consider whether or not different facets of intervention are more effective in different phenotypes/endotypes of asthma.

A further issue identified by trials, such as RAACENO, in which algorithms are used to generate treatment decisions is that patients and clinicians can be reluctant to follow algorithm recommendations. Future research could explore if/how algorithm recommendations could be made more acceptable.

Compared with many non-communicable diseases in children and young people (e.g. diabetes, epilepsy and inflammatory bowel disease), the management of asthma has not changed substantially over the last 25 years. Anticipating no novel therapies to cure or manage asthma, future research could evaluate the benefit of a standardised approach to treatment changes, for example using a standardised web-based algorithm to recommend a treatment regime and incorporating electronic logging devices to routine care, in addition to asthma action plans and 3- to 4-monthly assessments. Spirometry is used in many asthma clinics but its clinical utility is not well understood; future research could evaluate the benefit of adding spirometry to symptom-guided asthma treatment.

Staff at recruitment sites

Secondary care sites:

• NHS Grampian (Royal Aberdeen Children’s Hospital) – Steve Turner (PI), Richard Brooker, Margaret Connon, Patricia Flanagan, Susan MacFarlane, Marie McWIlliam, Stephen Main, Mustafa Osman and Guy Wilson.

• NHS Tayside (Ninewells Hospital) – Jonathan McCormick (PI), Richard Brooker, Susan MacFarlane, Anne MacLeod, Stephanie Menzies, Debbie Rice and Susie Roberts.

• NHS Greater Glasgow & Clyde (Queen Elizabeth University Hospital) – James Paton (PI), Elizabeth Adair, Elise Buchan, Paul Burns, Katharine Duffy, Sheryl Goddard, Kathleen McLaughlin, Barry Milligan, Leanne Milne, Lynne Murray, Ashleigh Neil, Natalie Rodden, Katrina Watts and Liz Waxman.

• University Hospitals of Leicester NHS Trust (Leicester Royal Infirmary) – Erol Gaillard (PI), Anisa Ahmed, Victor Jerome Ambrose, Clare Foxon, Teresa McNally, Sarita Makam, Deepa Patel and Simal Patel.

• Barnsley Hospital NHS Foundation Trust (Barnsley Hospital) – Diarmuid Kerrin (PI), Rikki Crawley, Abigail Crew, Ezzedin Gouta, Sian Jones, Nicola Lancaster, Jamie Matthews, Linda Schofield, Janet Shackleton, Zena Thomas and Gemma Wray.

• Sheffield Children’s NHS Foundation Trust (Sheffield Children’s Hospital) – Sonal Kansra (PI), Sara Ali, Alice Carey, Anita Critchlow, Mihaela Diaconu, Rachel Harrison, Jane Kirkby, Hemant Kulkarni and Janet Shackleton.

• University Hospitals of North Midlands Trust (Royal Stoke University Hospital) – Will Carroll (PI), Emma Addison, Martin Booth, Sadie Clayton, Angelene Cope, Francis Gilchrist, Sheng-Ang Ho, Ruth Jones, Gillian Lockett, Holly Minchin, Emily Obhrai, Helen Parker, Jane Peach, Rachel Pringle, Victoria Riches, Eric Roe, Jayne Swift and Joanne Tomlinson.

• The Royal Wolverhampton NHS Trust – Sally Edwards (PI), Lyndsay Bibb, Charlotte Busby, Rebecca Denyer, Ajay Gupta and Katherine Jones.

• NHS Ayrshire & Arran (University Hospital Crosshouse) – Tim Adams (PI) and Claire Bell.

• Royal Devon & Exeter NHS Foundation Trust – Patrick Oades (PI), Cassie Brady, Emma Chamberlain, Beth Enderby, Caroline Harrill, Nicola Jones, Megan Knight and Suzanne Wilkins.

• Plymouth Hospitals NHS Trust (Derriford Hospital) – Denise Ullmann (PI), Julie Alderton, Jane Hallam, Cho Hsint, Jemma Inches, Jennifer Merrison-Hort, Kathiresan Natesan, Alice Phillips, Mala Raman, Sarah-Jane Sharman, Alison Stolton and Emma Storr.

• Birmingham Women’s and Children’s NHS Foundation Trust (Birmingham Women’s and Children’s Hospital) – Priti Kenia (PI), Rehana Bi, Amy Burrill, Benjamin Davies, Susan Frost, Anas Ghivalla, Brinderjit Heyeer, Juliet Hopkins, Kiran Hunjan, Ciaran McArdle, Monica Mitchell, Prasad Nagakumar, George Preece, Satish Rao, Fiona Robinson and Ian Surplice.

• Nottingham University Hospitals NHS Trust (Nottingham Children’s Hospital) – Jayesh Bhatt (PI), Carol Bertenshaw, Lindsay Crate, Rebecca Devaney, Emma Douglas, Debra Forster, Matthew Hurley, Laura Looby, Helen Navarra, Andrew Prayle, Alan Smyth, Stefanie Stafford, Rachel Wiffen and Caroline Youle.

• Bradford Teaching Hospitals NHS Foundation Trust (Bradford Royal Infirmary) – Anil Shenoy (PI), Louise Akeroyd, Trudy Booth, Liz Ingram, Neelam Majid, Rachel Swingler, Joanna Urso, Rachel Wane, Aimee Watson and Dawn Woodward.

• University Hospital Southampton NHS Foundation Trust – Graham Roberts (PI), Cherry Alviani, Rebecca Cartwright, Amber Cook, Jillian Doherty, Lisa Fairhead, Matt Harvey, Anne Magee, Jane Martin and Cilla Snape.

• NHS Lothian (Royal Hospital for Sick Children) – Kenneth Macleod (PI), Julie Baggott, Sarah Blacklock, Naomi Matos, Debbie Miller, Maxine Ramsay, Stefan Unger and Julie Westwood.

• Derby Teaching Hospitals NHS Foundation Trust (Royal Derby Hospital) – Donna Traces (PI), Jacey Allen, Claire Baldwin, Melanie Hayman, Nigel Ruggins, Rebecca Salloway, Coral Smith and Samantha West.

• Walsall Healthcare NHS Trust (Walsall Manor Hospital) – Hesham Abdalla (PI), Joanne Butler, Sarah Freeth, Najma Iqbal, Sharon Kempson, Katharine Lewney, Joanne Mason, Priscilla Mhembere, Marie Phipps and Lisa Richardson.

• University Hospitals Bristol (Bristol Children’s Hospital) – Simon Langton Hewer (PI), Karen Coy, Rachel Helyer, Myriam Kanchanatheera, Gurnie Kaur, Deborah Marriage, Huw Thomas and Lisa Tucker.

• Leeds Teaching Hospitals NHS Trust (Leeds General Infirmary) – Emma Guy (PI), Alex Adams, Kate Baker, Nicola Balatoni, Louise Blackburn, Lai Wah Jenny Chung-Toole, Janet Clark, Lucy Fletcher, Saikiran Gopalakaje, Neil Hall and Heather Rostron.

• King’s College Hospital Foundation Trust (King’s College Hospital) – Atul Gupta (PI), Rania Abusamra, Adedamola Adebayo, Cara Bossley, Ivan Caro, Holly Dickson, Katie Leigh-Ellis, Kathleen McLaughlin, Eniola Nsirim, Sormeh Salehian and Katie Tupper.

• Barts Health NHS Trust – Chinedu Nwokoro (PI), Rasheda Choudhury, Bessie Cipriano, Emmily Clark, Mumtaz Idris, Hafiza Khatun, Ramiya Kirupananthan, Ivone Lancoma-Malcolm, Hina Mir, Tara Murray, Victoria Oyenuga, Elise Pabriaga, Caroline Pao, John Rae, Prita Rughani, Kevin Samuels, Gemma Sedgwick, Theresa Simangan, Surendan Thavagnanam and Natasha Thorn.

• Newcastle Upon Tyne Hospitals NHS Foundation Trust (Royal Victoria Infirmary) – Matthew Thomas (PI), Karen Allen, Kathryn Bell, Malcolm Brodlie, Stephen Crulley, Charlotte Reigan, Evelyn Thomson, Jennifer Townshend and Jessica Williams.

• Alder Hey Children’s NHS Foundation Trust (Alder Hey Children’s Hospital) – Ian Sinha (PI), Lynsey Brown, Debra Evans, Debbie Large, Phillip Lawrence, Nicola Rimmer, Joanne Shakeshaft, Elinor Thomason and Vicky Wheeler.

• Manchester University NHS Foundation Trust (Wythenshawe Hospital) – Naveen Rao (PI), Sandra Bothwell, Anthony Brothwood, Claire Holliday, Kate Ransom, Karen Rhodes and Louise Woodhead.

• St George’s University Hospital NHS Foundation Trust (St Georges University of London) – Richard Chavasse (PI from September 2018), Fran Mabesa, Young Ng, Jennifer Pearce, Anu Shankar (PI to September 2018) and Elia Vitale.

• Shrewsbury & Telford Hospital NHS Trust (The Royal Shrewsbury Hospital) – Kumar Sethuraman (PI), Lynette Charles, James Jones, Helen Moore, Martyn Rees and Rebecca Wilcox.

• Brighton & Sussex University Hospitals NHS Trust (Royal Alexandra Children’s Hospital) – Akshat Kapur (PI), Krishne Chetty, Kevin Donnelly, Paul Frattaroli, Sonia Sobowiec Kouman, Christine Laycock, Jason Lenton, Carolynn Lorimer, Kate Moscovici, Somnath Mukhopadhyay, Catherine Olden, Rebecca Ramsay, Vivien Richmond, Tom Ruffles, Paul Seddon, Emma Tagliavini, Cathy Warde and Edwina Wooler.

• Countess of Chester Hospital NHS Foundation Trust (Countess of Chester Hospital) – Ravi Jayaram (PI), Caroline Burchett, Sarah De-Beger, Usha Downie, Stuart Eccles and Caroline Lonsdale.

• Harrogate and District Foundation Trust (Harrogate District Hospital) – Alexandra Hardisty (PI), Sarah-Jane Foxton, Christine Morgan, Tracy Nixon, Rosalind Parkinson and Louise Wills.

• The Dudley Group NHS Foundation Trust (Russells Hall Hospital) – Sohail Nassir (PI), Daljit Kaur, Katharine Lewney, Rebecca Marks and Abigail Twiss.

• NHS Lanarkshire (Wishaw General Hospital) – Carol Dryden (PI), Angela Brown, Donna Corrigan, Karen Leitch, Jackie Quigley and Thin Thin Saing.

• The Pennine Acute Hospitals NHS Trust (Royal Oldham Hospital) – Prakash Kamath (PI), Sunil Bagewadi, Bernadette Holt, Rachel Newport, Grainne O’Connor, Jennifer Philbin, Chloe Rishton and Zainab Sarwar Yasin.

• Lancashire Teaching Hospitals – Karnam Sugumar (PI), Julie Abernethy, Andy Lancaster, Claire Lodge and Natalie Mattos-Harris.

• Southport & Ormskirk NHS Trust (Ormskirk Hospital) – Sharryn Gardner (PI), Irene Gardner, Zena Haslam, Sudhaker Kandasamy, Emily McDonald, Moira Morrison and Laura Thomas.

• Wirral University Hospital NHS Foundation Trust (Arrowe Park Hospital) – Sudeshna Bhowmik (PI), Sharon Hughes, David Lacy, Lucy Lewis and Joanne Mullen.

Primary care sites:

• Barrack Lane Medical Centre – Debasish Banerjee (PI), Karen Drury and Beverly Kennell.

• Castle Partnership – Joanne Walsh (PI), Jason Ladbrooke and Belinda Sewell.

• Hanscombe House Surgery – Funlayo Subar (PI).

• Ixworth Surgery – Vijay Chandraraj (PI), Deirdre Siddaway and Joshua Williams.

• Lakenham Surgery – Emore Echuko (PI).

• Parsonage Surgery – Jagjit Takhar (PI).

• Roundwell Medical Centre – Keeva Rogers (PI), Chaminda Dooldeniya (PI), Caiman Hardiman, Alison Mortlock and Denise Steward.

Primary care clinical RNs:

• CRN Eastern – Donna Clements, Phillipa Lincoln, Ruth Osborne and Suzanne Skinner.

Contributions of authors

Steve Turner (https://orcid.org/0000-0001-8393-5060) (Chief Investigator) contributed to the conception and design of the trial, the conduct of the trial, the recruitment and follow-up of participants, the interpretation of results and writing/editing the report.

Seonaidh Cotton (https://orcid.org/0000-0002-7883-0608) (Trial Manager) contributed to the design and conduct of the trial, the development of the treatment algorithm, the interpretation of results and writing/editing the report, and was responsible for the day-to-day management of the trial.

Jessica Wood (https://orcid.org/0000-0002-4233-811X) (Trial Manager) was responsible for the day-to-day management of the trial, and contributed to the interpretation of results and writing/editing the report.

Victoria Bell (https://orcid.org/0000-0001-9518-9296) (Assistant Trial Manager) was responsible for aspects of the day-to-day management of the trial and contributed to the interpretation of results and writing/editing the report.

Edwin-Amalraj Raja (https://orcid.org/0000-0003-0853-9724) (Statistician) contributed to the statistical analysis, the interpretation of results and writing/editing the report.

Neil W Scott (https://orcid.org/0000-0002-8067-9660) (Statistician) was responsible for the statistical analysis and contributed to the interpretation of results and writing/editing the report.

Heather Morgan (https://orcid.org/0000-0002-6118-8911) (Qualitative Researcher) was responsible for the day-to-day management of the qualitative process evaluation and contributed to the conception and design of the qualitative process evaluation, the interpretation of results and the writing/editing of the relevant parts of the report.

Louisa Lawrie (https://orcid.org/0000-0002-9867-2184) (Qualitative Researcher) was responsible for the day-to-day management of the study, including interviewing participants, and contributed to the design of the qualitative process evaluation, the interpretation of results and the writing/editing of the relevant parts of the report.

David Emele (https://orcid.org/0000-0003-1060-9963) (Senior Programmer) contributed to the design of the trial, was responsible for the development and computerisation of the treatment algorithm, and edited the report.

Charlotte Kennedy (https://orcid.org/0000-0002-1974-6318) (Health Economist) was responsible for the health economic analysis and contributed to the interpretation of results and writing the report.

Graham Scotland (https://orcid.org/0000-0001-5539-8819) (Health Economist) contributed to the design of the trial, oversaw the health economic analysis and contributed to the interpretation of results and writing/editing the report.

Shona Fielding (https://orcid.org/0000-0001-5363-8188) (Lecturer in Statistics) provided statistical expertise during the design and conduct of the trial and contributed to the statistical analysis and editing the report.

Graeme MacLennan (https://orcid.org/0000-0002-1039-5646) (CHaRT Director) contributed to the statistical analysis, the interpretation of results and writing/editing the report.

John Norrie (https://orcid.org/0000-0001-9823-9252) (Director of Edinburgh Clinical Trials Unit) contributed to the conception and design of the trial and writing/editing the report.

Mark Forrest (https://orcid.org/0000-0002-2395-8823) (Senior IT Development Manager) contributed to the development and computerisation of the treatment algorithm and writing/editing the report.

Erol Gaillard (https://orcid.org/0000-0003-0996-8133) (Clinical Senior Lecturer) contributed to the conception and design of the trial, the conduct of the trial, the recruitment and follow-up of participants, the interpretation of results and writing/editing the report.

Johan de Jongeste (https://orcid.org/0000-0003-1321-5943) (Professor, Paediatrics) contributed to the conception and the design of the trial, the interpretation of results and writing/editing the report.

Marielle Pijnenburg (https://orcid.org/0000-0003-0902-0860) (Paediatric Pulmonologist) contributed to the conception and design of the trial, the interpretation of results and writing/editing the report.

Mike Thomas (https://orcid.org/0000-0001-5939-1155) (Professor of Primary Care Research) contributed to the conception and design of the trial, the interpretation of results and writing/editing the report.

David Price (https://orcid.org/0000-0002-9728-9992) (Professor of Primary Care Respiratory Medicine) contributed to the conception and design of the trial, the interpretation of results and writing/editing the report.

Publication

Turner S, Cotton S, Wood J, Bell V, Raja E-A, Scott N, et al. Reducing asthma attacks in children using exhaled nitric oxide (RAACENO) as a biomarker to inform treatment strategy: a multicentre, parallel, randomised, controlled, phase 3 trial [published online ahead of print January 28 2022]. Lancet Respir Med 2022.

Data-sharing statement

All data will be made available and can be obtained from the corresponding author.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research funded by the National Institute for Health and Care 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, the HTA programme or the Department of Health and Social Care. 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, the HTA programme or the Department of Health and Social Care.