Features of successful interventions to improve adherence to inhaled corticosteroids in children with asthma: A narrative systematic review

Abstract Introduction Nonadherence to inhaled corticosteroids (ICSs) in children with asthma leads to significant morbidity and mortality. Few adherence interventions have been effective and little is known about what contributes to intervention effectiveness. This systematic review summarizes the efficacy and the characteristics of effective interventions. Methods Six databases were systematically searched on October 3, 2020 for randomized control trials measuring adherence to ICS in children with asthma. A narrative synthesis was conducted focusing on intervention efficacy and study reliability. Intervention content was coded based on the National Institute for Health and Care Excellence guidelines for medicines adherence (the Perceptions and Practicalities Approach, PAPA) and behavior change techniques (BCTs), to determine the effective aspects of the intervention. Results Of 240 studies identified, 25 were eligible for inclusion. Thirteen of the 25 studies were categorized as being highly reliable. Nine of the 13 interventions were effective at increasing adherence and 6 of those met the criteria for a PAPA intervention. Techniques targeting perceptions and practicalities in successful interventions included rewards, reminders, feedback and monitoring of adherence, pharmacological support, instruction on how to take their ICS/adhere, and information about triggers for symptoms and nonadherence. Conclusion Adherence interventions in children with asthma have mixed effectiveness. Effective intervention studies were more frequently of higher quality, were tailored to individuals' perceptual and practical adherence barriers, and used multiple BCTs. However, due to the small number of included studies and varying study design quality, conclusions drawn here are preliminary. Future research is needed to test a PAPA‐based intervention with a rigorous study design.


| INTRODUCTION
Asthma is the most common, chronic noncommunicable disease in children worldwide. 1 Asthma prevalence is higher in children in Europe (8.9%), compared with the rest of the world (7.2%) 2 but varies between countries. 3 Most children with asthma achieve good disease control with maintenance low-dose inhaled corticosteroids (ICS), which are effective at preventing most asthma hospitalizations and deaths. 4 However, some children remain poorly controlled despite being prescribed high-dose ICS treatment, often due to poor adherence. This contributes to suboptimal asthma control and severe attacks. 5,6 Up to half of patients attending tertiary care pediatric asthma clinics are nonadherent (defined as taking <80% of their prescribed dose). 7 The Global Initiate for Asthma (GINA) highlights that suboptimal use of asthma treatment is a patient-specific barrier that contributes to the burden of asthma. 8 Similarly, the UK National Review of Asthma Deaths reported that 67% of asthma deaths were avoidable and one of the most important avoidable factors was low ICS adherence in the month and/or year before death. 9 Many interventions have been developed to address the issue of poor ICS adherence in children. A meta-analysis in adults and children identified that interventions for improving adherence in asthma can be effective. 10 However, the meta-analysis did not examine the intervention characteristics, for example, content, channel of delivery, and context of the intervention, which form the three components of a behavior change framework (3CBC 11 ) in relation to intervention efficacy. It is important to be able to identify characteristics of effective interventions so that they may be applied in practice. The current review will address this lack of detail regarding features of successful interventions within this population.
Moreover, the reliability of the diagnosis of asthma and the adherence measurement tool have not previously been used to identify high-reliability interventions. A possible belief/behavioral pattern related to a misdiagnosis is if patients do not believe they have asthma, as adherence to ICS does not improve their symptoms, or they do not suffer any symptoms so they may become nonadherent, as they consider the treatment unnecessary. If patients who are misdiagnosed with asthma are included in asthma interventions, the results of the study may not be relevant for patients with asthma.
Similarly, if adherence is overestimated in studies using unreliable adherence measurements, then the conclusions drawn from the studies will also be inaccurate. By investigating these missing elements within the current review, the data presented in this review are likely to be more relevant to practice, as they represent a rigorous test of the intervention.
The National Institute for Health and Care Excellence (NICE: https://www.nice.org.uk/), a group within the National Health Service of England and Wales, who develop evidence-based recommendations within a committee of professionals, lay members and, in consultation with stakeholders, have developed guidelines intended to aid the design of adherence support for longterm conditions at any stage of the life span. 12 The guidelines apply the Perceptions and Practicalities Approach (PAPA 13 ; Figure S1). This approach recognizes that adherence varies within the individual, over time and across treatments. Adherence/nonadherence is best understood in terms of the interaction between an individual and a particular treatment. It is a variable behavior rather than a trait characteristic. PAPA conceptualizes adherence as including both intentional and unintentional nonadherence.
The application of the PAPA approach to adherence interventions has the following key features: first, the need for a "no-blame approach" as patients are often reluctant to admit to nonadherence, or to concerns about the treatment, as they fear that this may be interpreted by the clinician as doubting their expertise. Hence, nonadherence and the reasons for it are often hidden. The second key feature is the need to tailor support to address both perceptions (e.g., beliefs about asthma and its treatment) and practicalities (e.g., clear instructions on inhaler technique and establishing a medication routine). Both perceptions and practicalities influence the patients' motivation and ability to start and continue taking the treatment. Indeed, research in asthma has shown beliefs about ICS are often important perceptual barriers to adherence, in particular doubts about the personal need for regular inhaler use, particularly in the absence of symptoms and concerns about corticosteroids. 14,15 Although this approach has been used within an adult asthma review, 16 the current review will be the first to assess the PAPA approach in a pediatric setting.
This systematic review aims to address the above research gaps by the following: (1) specifically examining ICS adherence interventions in children with asthma; (2) using quality indicators to identify those studies that may be more informative; and (3) examining the characteristics of successful adherence interventions to identify features that may be relevant to practice.

| Search strategy
Design framework. Any interventions that focused on adherence to ICS with at least one outcome measure of adherence and used a randomized control trial (RCT) design were included. The comparison group was either usual treatment or a basic education arm. Articles were included where the full text was written in English and where the population of interest was patients aged 0-18 years old with a diagnosis of asthma. Although many preschool children with wheeze do not respond to ICS, 42 studies often recruit younger children and therefore this age-range was included to avoid missing relevant articles. If they do not have the treatable trait of airway eosinophilia likely to respond to ICS, 43 this will be highlighted in the section regarding reliability of the criteria for asthma diagnosis. Studies were excluded if they did not meet the above criteria, if they were an RCT comparing two medications only, or where the majority of participants were not children (e.g., the mean age of participants was over 18 years old or only adults were recruited).

| Data extraction and synthesis
Following full text review, CP and TJ independently extracted details of the following: study characteristics (setting, number of participants, diagnosis criteria, intervention and control descriptions, and the outcome of interest); effectiveness (a statistically significant [p < .05] improvement in adherence in the intervention group compared with the control group); behavior change techniques (BCTs); target of the BCTs; and relationship to PAPA. Where there were differing opinions or uncertainty, a third opinion was sought from a senior colleague (RH).

Intervention content
Intervention content were coded for PAPA as follows: Specific components within the interventions for changing adherence (BCTs) were also coded independently using the BCT taxonomy V1 app. 44 Any differences in the selected BCTs were discussed until consensus was reached (Table 1).

| Risk of bias
Risk of bias (RoB) was assessed independently using the Cochrane Risk of Bias Handbook 45 by CP, AC, and HF using the Covidence platform (www.covidence.org) to record coding decisions and consensus discussions. The RoB score was based on the adherence outcome. Each study was scored across five domains: selection bias; performance and detection bias, attrition bias and reporting bias, and was scored as either low, high, or unclear risk for each study. Authors were contacted for clarity when information relating to the domains seemed unclear.

| Study reliability
To ascertain which interventions were truly effective, study reliability was considered. Although other validated tools have been used to assess quality such as the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) tool, 46 the authors felt that there were several crossovers between RoB, the reliability scores, the 3CBC approach, and, in particular, the indirectness section of the GRADE tool. Based on the aim of this review, the reliability measurements would be more useful when considered with RoB. Both diagnosis and adherence measures can range from being subjective to objective; therefore, considering the reliability of the approaches used is key for determining study reliability. Through multidisciplinary team discussions (including with respiratory physicians, pharmacists, and a chartered psychologist), a coding hierarchy that considered the reliability of the asthma diagnosis and adherence measurement used was created and applied to the specific studies within this review (Table S2).
Based on the RoB, the reliability of the asthma diagnosis, and the objectivity of the adherence measurement, the most reliable and least biased studies were used to ascertain what components constituted an effective intervention. Previous literature suggests that optimizing the content, channel of delivery, and context of the intervention is important for intervention effectiveness, 11 and thus the 3CBC 11 was also applied to this review.
Studies were summarized by a narrative synthesis. Meta-analysis was not conducted due to the wide study heterogeneity in terms of setting, asthma diagnosis criteria, and outcome measures used. The study protocol is published on PROSPERO (https://www.crd.york.ac. uk/prospero/#searchadvanced) (ref: CRD42016029213).

| Search results
The literature search retrieved 255 articles. An additional nine were identified from other sources. Twenty-two duplicate articles were removed before abstract screening. Based on abstract screening, 202 papers were excluded and a further 13 papers were excluded based on the full text. Main reasons for exclusion were as follows: study design not an RCT, no usual care control group, medication adherence not included as a usable outcome, and trial compared medications or was conducted in adults. Twenty-five studies were included in the narrative synthesis  ; see full PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) diagram ( Figure 1). 3.2 | Narrative synthesis

| Study reliability
Although half of the interventions were reported as effective at increasing adherence, the study reliability varied widely ( Based on the coding hierarchy that considers the reliability of the asthma diagnosis (Table S2), seven studies used reliable means to diagnose asthma in their participants. 17,21,22,24,35,38,40 Three studies used less reliable methods 25,28,29 and a further seven used unreliable diagnostic methods. 20,26,27,36,37,39,41 In one study, the method of diagnosis of asthma was unclear. 19 Adherence measurement varied with studies using objective and subjective measures. Based on our coding hierarchy of objectivity of adherence measurements (Table S2), most studies used more objective measurements 19,21,22,29,35,36,40,41 or both objective and subjective measures. [26][27][28] Six used subjective measurements of adherence only 17,20,24,25,38,39 and for one study, the method of adherence measurement was unclear. 37 Based on the RoB, reliability of asthma diagnosis and objectivity of the adherence measurement within each study, the reliability of the evidence can be summarized (Table 3).

RoB across studies
The main bias identified was performance bias. Overall, RoB was low for most studies in terms of selection bias (random sequence T A B L E 1 (Continued) NAEPP, National asthma education and prevention program; NHLBI, National Heart, Lung, and Blood Institute; PACE, Physician Asthma Care Education; PEAK, pharmaceutical care evaluation of asthma in kids; PEF, peak expiratory flow; RTMM, real-time medication management. generation), detection bias (blinding of outcome assessment), and reporting bias (selective reporting bias). Section bias (allocation concealment) was often low or unclear and was generally poorly reported. Attrition bias (incomplete outcome data) was frequently unclear or high risk (Figure 2).

| Reliability of the evidence
The most reliable studies (n = 13/25) (i.e., moderate or high reliability based on asthma diagnosis and adherence measurement criteria) and low/moderate RoB are discussed in more detail below (n = 13/25).
Nine of the 13 highly reliable interventions were effective at increasing adherence 19,21,22,24,29,32,35,38,40 and four were ineffective. 17,26,28,30 The following section compares the nine effective interventions with the four ineffective interventions within this highreliability group (n = 13/25). Of those studies that reported effectiveness for increasing adherence, only one study was not considered to be in the high-reliability group.

Components of effective interventions
This section will summarize the findings of this systematic review based on the 3CBC framework, 11 to critically appraise the effectiveness of the components within the most reliable intervention study evidence.
Context. The nine effective high-reliability intervention studies (n = 9/13) were conducted in Brazil, 21 Greece, 32 New Zealand, 22,24 China, 34 United States, 38 United Kingdom, 29,35 and the Netherlands. 40 The ineffective high-reliability intervention studies (n = 4/13) were conducted in United States 17,26,30 and Taiwan. 28 Effective interventions took place in an emergency care setting, 22,38 primary care, 21,29 hospital outpatients, 19,32,35,40 and in the community. 24 The ineffective interventions took place in emergency care, 17,30 in hospital outpatients, 28 and in the community. 26 There are no data regarding whether or not the interventions used a no-blame approach 11 but four of the high-reliability effective interventions were clearly tailored to the patient, 19,21,24,32 compared with only one of the ineffective interventions. 28 Channel of delivery. Seven of the high-reliability effective intervention studies used technology to deliver the intervention (n = 7/9) including using electronic monitoring devices (EMDs 19,22,32,35,40 ), the telephone, 21 and an SMS-based system. 40 Three of the ineffective interventions used technology to deliver the intervention (n = 3/4) via a website and monthly telephone calls, 26 SMS text reminder and tips (not personalized), 30 and via the internet alone. 28 19,24,35,38,40 community health workers, 24 and researchers. 40 In one effective intervention (n = 1/9), the only channel was a letter sent from the patients' GP 29 to the parents of the child with asthma. The ineffective interventions used limited contact with any primary care provider (multiple roles), 17 pharmacist, 31 nurse, 26,28 and physician. 28  19,21,22,29,30,32,34,35,40 ; feedback and monitoring 19,22,35,38,40 ; pharmacological support (this often involved providing free medications in countries where medications were not free and providing a longer-term supply when the medications were free) 19 29 Four further studies specified that the interventions targeted the child specifically 22 and these were often with older children. 21,36,37 For extracted examples of common BCTs and the interventions they were used in, see  10 Similarly, we found that only nearly half of the included interventions (11/25) were effective at significantly increasing adherence. 19,21,22,24,25,29,32,34,35,38,40 We then explored the crucial factors for an effective intervention to increase adherence.
Of the 13 high-reliability interventions studies, nine were effective. 19 These issues are important, because necessity and concern beliefs may be the drivers of adherence as they influence motivation to adhere to treatment. 53,54 The most common BCTs used in effective interventions were prompts/cues (e.g., reminders); feedback and monitoring; pharmacological support and instruction of how to perform a beha-  33 Baren et al. 17 Chatkin et al. 21,a Teach et al. 38,a Chan et al. 22,a Julious et al. 29,a Kenyon et al. 30 Koumpagioti et al. 32,a Moderate risk Canino et al. 20 Gustafson et al. 26 Morton et al. 35,a van Es et al. 39 Jan et al. 28 Vasbinder et al. 40,a Bresolini et al. 18 Garrett et al. 24,a Kosse et al. 31 Burgess et al. 19,a Lv et al. 34,a High risk Stergachis et al. 37 Hederos et al. 27 Guendelman et al. 25,a Mosnaim et al. 36 Wiecha et al. 41 Davis et al.

| Strengths and limitations
Due to the heterogeneity of the adherence outcomes, limited availability of raw data and a small number of eligible studies, a meta-PEARCE ET AL. | 843 analysis was not possible within this review. This systematic review focuses on adherence as an outcome as opposed to clinical health outcomes as unlike within the adult literature, few studies in pediatric asthma include both adherence and clinical outcomes. Focusing on adherence therefore allowed a greater number of studies to be synthesized. Ideally, intervention studies should have an objective reliable clinical outcome as well as an adherence outcome to account for potential patient manipulation of the adherence measurement and for those patients that may have low adherence despite good control (likely overmedicated). However, unlike in some other conditions, adherence to ICS has been shown to be highly correlated with objective clinical outcomes 55 and, therefore, the use of adherence as a primary focus for this review is a reasonable proxy.
Most of the interventions had a moderate RoB, which was increased by the high level of performance bias that is common in behavioral interventions. This is due to the lack of ability to blind patients and personnel to the purpose of the study; however, many of the studies tried to counteract that using deception (where ethically permitted). This included objective EMDs also for control groups and additional measurements to distract from the adherence data collection. The studies often had low selection bias (for random sequence generation), detection bias, and reporting bias. However, attrition bias and allocation concealment was frequently unclear with modern recommended reporting guidelines such as CONSORT 56 not being followed. We recommend using objective methods of measuring adherence and also more than one method of measurement, and also for the diagnosis of asthma, alongside blinding to increase the reliability of future intervention findings.
T A B L E 4 PAPA categorization and reliability One further limitation is not excluding interventions where the diagnosis of asthma reported was not rigorous, for example, where primary-care medical records were used to identify those with asthma despite no record of prescribing ICS or where a physician diagnosis was given without objective measurement of asthma. 57 Future intervention studies should ensure the children recruited have a reliable diagnosis of asthma and objective measurements of adherence so the true effectiveness of the interventions can be determined. 58 Therefore, this review considered the reliability of the evidence for both the diagnosis of asthma, the measurement of adherence and the RoB of the studies.

| CONCLUSIONS
Adherence interventions in children with asthma have mixed effectiveness. Effective intervention studies were more frequently of higher quality, targeted both perceptual and practical adherence barriers in a tailored manner, and used a combination of BCTs.
However, due to the small number of included studies and varying study design quality, conclusions drawn here are preliminary.
None of the studies have explicitly addressed ICS necessity and concern beliefs. This remains a potential area of investigation as a method for enhancing adherence. Future interventions could consider a closer use of the NICE guidelines including addressing patients' beliefs and the channel by which the intervention is delivered, the increased use of EMDs, with feedback delivered in a no-blame collaborative consultation. Future research is needed to test a PAPAbased intervention with a rigorous study design as outlined in this review.

ACKNOWLEDGMENTS
We would first like to thank all our colleagues at the Asthma UK