Personalized Medication Adherence Management in Asthma and Chronic Obstructive Pulmonary Disease: A Review of Effective Interventions and Development of a Practical Adherence Toolkit

BACKGROUND: The management of medication nonadherence of patients with asthma or chronic obstructive pulmonary disease (COPD) remains challenging. Given the multitude of underlying causes, a personalized approach is required. The Test of Adherence to Inhalers (TAI) can identify reasons for nonadherence, but it does not provide guidance regarding how to act effectively after results. OBJECTIVE: To develop a practical, evidence-based decision support toolkit for health care professionals managing adult patients with asthma and/or COPD, by matching TAI-identified


INTRODUCTION
With adherence ranging from 22% to 78%, asthma and chronic obstructive pulmonary disease (COPD) medication adherence is the lowest for all medication groups. 1,2 Besides being related to poor symptom control and an increased number of exacerbations, poor adherence is associated with increased mortality, decreased quality of life, and increased direct and indirect costs. 1,[3][4][5][6][7] Medication adherence is defined by the World Health Organization (WHO) as "the degree to which use of medication by the patient corresponds with the prescribed regimen." 8 The WHO distinguishes three types of nonadherence: erratic, intelligent, and unwitting. Erratic (also called sporadic) nonadherence reflects forgetfulness, intelligent (or deliberate) nonadherence reflects a reasoned choice for not taking the medicine (eg, owing to fear of side effects), and unwitting (or unconscious) nonadherence reflects the failure to understand fully either the specifics of the regimen or the necessity for adherence. 8 There is no typical nonadherent patient, because nonadherence is a complex and multifactorial problem. 9 Bourbeau and Bartlett 10 highlighted that medication, dosing regimen, patient factors, health care provider, and caregivers could all have a role. Most of these factors are modifiable and therefore can make the difference between adherence and nonadherence. Because there is so much diversity, there is no simple one-size-fits-all approach.
To manage nonadherence, the first step is to recognize and classify it. Yet for many years, no respiratory specific tool could easily identify the three WHO-defined phenotypes. In 2016, the Test of Adherence to Inhalers (TAI), 11 a questionnaire consisting of 10 patient-reported items and two health care professional reported items, was launched and validated. The TAI is increasingly translated in other languages and seems to be the first tool that can truly guide personalized adherence interventions. However, once completed, there is no guidance regarding how to act on each answer provided by the patient. To support health care professionals, there is the need for a toolkit that can efficiently match TAI-identified adherence barriers with effective interventions.
The objectives of this study were to identify effective medication adherence-enhancing interventions for adult patients with asthma or COPD and to use this information to develop a practical, evidence-based decision support toolkit for health care professionals to manage nonadherence.

Study design
We searched the PubMed and Embase databases for articles published between 2003 and May 2020. We chose 2003 because the landmark WHO report about medication adherence, 8 which can be seen as the starting signal for the emergence of articles on the topic of adherence, was published that year. The current study is reported according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses reporting guidelines. 12

Eligibility criteria
For inclusion, studies needed to meet the following criteria: (1) full text (ie, no abstracts), (2) English language, (3) randomized controlled trial (RCT), 4) adults with asthma and/or COPD, (5) medication adherence as the outcome (primary or secondary), (6) showing improvement in medication adherence, and 7) intervention led by health care professionals. No limits were set for sex or study setting (eg, clinical or community setting).

Search strategy
The search was performed with the use of key words and Medical Subject Headings of the National Library of Medicine. Search terms included "asthma," "chronic obstructive pulmonary disease/lung disease/COPD," "pulmonary/lung emphysema," "chronic bronchitis," "intervention," "strategy," "medication," "drug," "medicine," "medication adherence," "(non [-])adherence", "(non [-])compliance", "(non[-])concordance," and "(non [-])persistence." (The full PubMed and Embase search strategies are presented in Table E1 in this article's Online Repository at www.jaci-inpractice.org). The reference lists of all included articles, a previously published Cochrane review, 13 and the WHO report were manually searched to identify any additional relevant articles. 8 Data extraction and quality assessment The following data were extracted for each study: (1) first author, country, and year of publication; (2) sample size; (3) duration of the intervention and follow-up; (4) age of the study population; (5) setting; (6) intervention; (7) control; (8) method for measuring adherence; and (9) outcomes. The Cochrane Risk of Bias Tool was used as for quality assessment of the included studies. 15 The tool consists of six domains (selection bias, performance bias, detection bias, attrition bias, reporting bias, and other bias). For all included articles, the risk of bias was judged as low, unclear, or high risk (see Figure E1 and Table E2 in this article's Online Repository at www. jaci-inpractice.org). Finally, we assessed the overall level of evidence (high, medium, or low) based on the power of the study (powered on adherence or not) as well as the effect on adherence (between or within groups). Study data were first extracted and assessed by one researcher (J.E.P. Aarts) to maintain consistency throughout coding. Subsequently, data in included studies were validated by an independent second reviewer (S.J. van de Hei).

Development of adherence toolkit
All data were summarized and used to design a practical toolkit to support health care professionals selecting an effective adherenceenhancing intervention matching specific responses to the TAI (see Table E3 in this article's Online Repository at www.jaciinpractice.org).

Study selection
The initial search yielded 369 studies, 105 of which were discarded as duplicates. After titles and abstracts were screened, 131 articles remained for full text review. Of these, 103 articles were excluded. Finally, 28 studies were included, with 12 additional studies identified through manual searching of reference lists, WHO reports, and Cochrane review (total n ¼ 40). Two studies were identified as eligible from the reference lists, but were not included. 16,17 One study evaluated the same intervention in two separate studies; of those, we included one (the largest study with a longer follow-up). 16 Another study evaluated the same intervention: one study had a control group (which we included), and in the second study, the intensity of interventions was different between groups. 17 Figure 1 provides s Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram of the process ( Figure E1 and Table E2).
Interventions could be classified into seven categories ( Figure 2): (1) reminders, 19,20,50,53 (2) educational interventions, [21][22][23][24][25][26][27][28][29][30][31]55 (3) multiple component intervention (eg, pharmacy care and self-management), [32][33][34][35][36][37][38][39][40][41][42]45,52,53,55,56 (4) motivational strategies, 43,45,47 (5) shared decision-making, 46 (6) simplifying the medication regimen, 47 and (7) feedback on medication use. 48,49,57 Reminders Four studies examined the impact of reminders on medication adherence. [18][19][20]50 Three studies were conducted on patients with asthma, and one was on patients with COPD and/or asthma. Charles et al 18 evaluated the effectiveness of a metered dose inhaler with an audiovisual reminder function in asthma patients. The audiovisual reminder device, which is attached to the inhaler, has the ability to emit both audio and visual reminders at predesigned times. In total, 110 patients with asthma were randomized to either the intervention or control group for 24 weeks. The proportion of medication taken in the last 12 weeks of the study was higher in the intervention group (93%) compared with the control group (74%), with a significant difference of 18% (P < .0001). Furthermore, the proportion of patients who were taking greater than 50%, greater than 80%, or greater than 90% of the medication was significantly higher in the reminder group. The second study, by Foster et al, 19 examined the effectiveness of two interventions (inhaler reminders and feedback [IRF] and/or personalized adherence discussions [PADs] about barrier[s], goal setting, and goal achievement strategies) compared with usual care. The 143 patients were divided into four groups: IRF, PAD, IRF and PAD and usual care for 6 months. At 6 months, medication adherence was significantly higher in the IRF groups compared with the non-IRF groups (73% vs 43%; P < .0001). Strandbygaard et al 20 examined the effectiveness of daily text reminders for medication adherence. In total, 26 adult patients were randomized to an intervention period of 8 weeks. After 12 weeks, the absolute difference in mean adherence rate between groups was 17.8% (P ¼ .019). The 6-month study of Gregoriano et al 50 randomized 169 asthma or COPD patients to either acoustic smartphone reminders and support phone calls or usual care. The reminder was always used, whereas support phone calls were employed only when medication was used incorrectly for more than 2 consecutive days. The intervention group had significantly more days, in which 80% to 100% of patients were adherent (pressurized metered dose inhaler: 82% vs 60%, P < .001; dry powder inhaler: 90% vs 80%, P ¼ .01) and had better timing adherence (ie, correct dosing intervals).

Education
Eleven studies evaluated the impact of education on medication adherence in asthma; one study focused on COPD. [21][22][23][24][25][26][27][28][29][30][31]51 Educational interventions were frequently described, but often in insufficient detail and combined with other types of interventions (ie, multiple component intervention). Therefore, we chose to describe the effects of successful education in general rather than at the study level. Education was given by telephone calls, 21,22,30,31 face-to-face sessions, 23,25,26,28,31,51 written information (book, brochure, or infographic), [23][24][25]29 audiotape, 29 and video. 24,27 Some studies combined these formats. Educational subjects were pathophysiology, triggers and symptoms, the role of medication, importance of adherence, side effects, and rescue medication instructions. Details regarding exact content were frequently missing. The duration of education ranged from 30 minutes to 24 months.

Multiple component intervention
Multiple component interventions were examined in 16 studies (asthma: n ¼ 11; COPD: n ¼ 5). 19,[32][33][34][35][36][37][38][39][40]42,46,[52][53][54][55][56] There were two subtypes of multiple component interventions: (1) pharmacy care interventions, which were led by pharmacists and focus on medication (eg, education and counseling, review and correct inhaler technique including physical demonstration, motivational interviewing on drug-related problems, attitudes regarding medication and adherence, and shared decisionmaking); and (2) self-management interventions, which were led by different types of health care professionals educators and researchers, and focus on self-management (eg, education and counseling, review and correct inhaler technique, motivational J ALLERGY CLIN IMMUNOL PRACT VOLUME 9, NUMBER 11 interviewing (exercise and smoking), and self-management of stress, triggers, and attacks). These interventions consisted of multiple components, which makes it difficult to determine which component had most impact on medication adherence.

Motivational strategies
Three studies evaluated the impact of motivational strategies on medication adherence in asthma patients. [43][44][45] Gamble et al 43 investigated psycho-education (based on motivational interviewing) covering self-motivation and resolving ambivalence toward taking medication. In total, 20 adults were randomized to either the intervention (eight visits in 12 weeks) or the control group (usual care). At 6 months, medication adherence improved (intervention: 37.6% to 61.9% vs control: 31.7% to 28.8%; P < .01). In the second study, 27 adults were randomized to either motivational interviewing and education or education only. The study examined the effectiveness of motivational interviewing on attitudes about medication adherence. 45 Motivational interviewing consisted of one 30-to 60-minute session and covered an assessment of the patient's readiness to use medication, personalized feedback about current pulmonary function, a discussion about the patient's thoughts about received feedback, and specific intervention strategies matched to the patient's readiness to change. One week after the intervention, the level of readiness to adhere over time was stable or increased in the intervention group and the attitude toward taking medication over time was higher. One study used a text message program that sent predefined tailored text messages to counteract illness and medication beliefs supporting nonadherence. In that study, Petrie et al 44 randomized 147 patients to either intervention or control (no tailored text messages). At 6, 12, and 18 weeks, and at 6 and 9 months, medication adherence was significantly higher in the intervention group compared with the control group, with a relative average increase in adherence of 10% (P < .001).

Shared decision-making
One study examined the effectiveness of shared and clinician decision-making on medication adherence and clinical outcomes in patients with asthma. 46 In shared decision-making, the beliefs and preferences of the patient are considered when making a treatment decision, whereas in clinician decision-making those beliefs and preferences are not taken into account. Patients (n ¼ 612) were randomized into three groups: shared decisionmaking, clinical decision-making, and usual care. After 12 months, medication adherence was significantly higher in the shared decision-making group compared with the clinical decision-making group (67% vs 59%; P ¼ .03) and the control group (67% vs 46%, P ¼ .0001).

Simplifying medication regimen
A study 47 evaluated the impact of medication administration once daily instead of twice daily on medication adherence in patients with asthma. For 12 weeks, 1233 patients were randomized to receive mometasone furoate 400 mg once daily or mometasone furoate 200 mg twice daily. The mean adherence rate was significantly higher in the once-daily group compared with the twice-daily group (93.3% vs 89.5%; P < .001).

Feedback on medication use
Three studies evaluated feedback on medication use (asthma: n ¼ 2; COPD: n ¼ 1). 48,49,57 Nides et al 57 used the Nebulizer Chronolog, a microprocessor device that records the exact date and time of each inhalation, to monitor metered-dose inhaler use electronically for 4 months in 251 COPD patients. The intervention group received detailed feedback at the end of weeks 1 and 7 to discuss inhaler use, whereas the control group did not receive feedback. After 4 months, adherence was higher in the intervention group compared with the control group (mean puffs per day: 1.95 vs 1.63; P ¼ .003). The second study, by Onyirimba et al, 48 randomized 30 asthma patients to the intervention group (direct clinician-to-patient feedback discussion) or the control group (usual care) for 3 weeks. Betweentreatment adherence rates were comparable at week 1 but were significantly higher in the intervention group compared with the control group starting at week 2 (81% vs. 47%; P ¼ .003). A significant group difference (favoring the intervention group) existed in adherence rates over the course of the study period of 10 weeks (P < .0001). Finally, Sulaiman et al 49 examined the effectiveness of feedback on inhaler technique and medication adherence. The feedback consisted of visual feedback training based on records on the electronic monitor. In total, 218 patients were randomized to the intervention (with feedback) or control group (without feedback) for a study period of 3 months. Mean adherence in the last month was significantly higher in the intervention group compared with the control group (73% vs 63%; P ¼ .02).

Test of Adherence to Inhalers Toolkit
We integrated all effective adherence-enhancing strategies as identified in this review into a toolkit that provides recommendations for each TAI answer (Figure 3). The toolkit can guide health care professionals to effective interventions based on the main behavioral phenotype level (sporadic, deliberate, and unconscious nonadherence) and the individual question level (TAI questions 1-12) (see Table E3 in this article's Online Repository at www.jaci-inpractice.org for a description of the development of the TAI Toolkit). The Toolkit consists of a wheel that can be used digitally or printed on paper. The wheel is accompanied by a user guide with further elaboration on the practical application of the interventions and the strength of underlying evidence (see Table E4 in this article's Online Repository at www.jaciinpractice.org). The first prototype of the Toolkit was assessed by a panel of eight health care professionals (physicians, nurses, and pharmacists) on usability (System Usability Scale), 58,59 feasibility, and practical implications. The median System Usability Scale score was 71.4 (range, 57.5-80.0) (see Figures E2  and E3 and Table E5 in this article's Online Repository at www.jaci-inpractice.org). Their feedback was integrated into the final version (Figure 3).

Main findings
In this review, we identified multiple medication adherenceenhancing interventions for adults with asthma and/or COPD. The studies included seven different types of interventions; the most commonly reported ones were educational and multiple component interventions. Other effective intervention types were reminders, motivational strategies, shared decision-making, simplifying the medication regimen, and feedback on medication use. The effective adherence-enhancing interventions were integrated into a practical, evidence-based toolkit.

Interpretation
We identified 40 RCTs evaluating interventions that could enhance medication adherence; of those, seven different types of interventions could be distinguished that allow further personalization of adherence management. However, their content, relationship to clinical outcomes, and methodologic quality were all variable. Power calculations and randomization methods were often not reported. Furthermore, it was difficult to compare studies because of the heterogeneity in settings and adherence measures applied. These findings regarding effectiveness and shortcomings were also identified in most previous reviews on adherence interventions in adults and children with asthma/ COPD. 13,60-62 Indeed, a Cochrane review also pointed out inconsistent effects on clinical outcomes of adherence-enhancing interventions. 13 Potential reasons for this discrepancy may be the short follow-up of interventions (median, 6 months), the lack of selection of patients with room for clinical improvement, and the inaccurate adherence measures that were used. Especially in asthma, owing to its variable nature and the adaptation of medication use to symptom frequency, it has been difficult to show the direct relation of adherence to outcomes. 63 Because we aimed to inform the Toolkit with effective interventions regarding adherence, studies that showed no improvement in medication adherence were excluded from this review. During study selection, eight studies were excluded for this reason. This is a small part of all studies in our literature search, which could be a consequence of publication bias.
Another explanation could be the small number of patients included in some of these studies.
Although it was difficult to compare studies directly, a general observation was that medication adherence in these RCTs was often much higher than in real-world settings. 7 For example, the study by Price et al 47 indicated an adherence rate of 89.5% in the control group at 12 weeks' follow-up. This is higher than adherence rates found in many observational studies using pharmacy refills. 64 Several factors could explain this, such as information bias (blinding was often not possible) and selection bias (motivated patients are willing to take part in studies). Moreover, not every patient has to be fully adherent (100%) for optimal clinical outcomes. For some patients, lower adherence may result in appropriate disease control. Further optimization of adherence is required only in case of disease deterioration.
Regarding the use of toolkits to manage adherence, we could identify no toolkit specifically designed to personalize nonadherence management in adults with asthma and/or COPD. In other fields such as diabetes, however, toolkits have been developed and studies evaluating those seem to be effective. 65 Regarding the TAI specifically, most previous studies using the TAI were observational. 11, 66 We chose the TAI as the adherence measure tool for this study because it is the only respiratoryspecific tool that identifies reasons for nonadherence, which can be linked to the three WHO-defined phenotypes. The TAI is a validated tool with a Cronbach a of 0.860 and a test-retest reliability of 0.883. Furthermore, the TAI scores correlated FIGURE 3. The Test of Adherence to Inhalers (TAI) toolkit. A full description of how to use the TAI toolkit can be found in Table E4. HCP, health care professional; SDM, shared decision-making.
with electronically monitored adherence (r ¼ 0.293; p ¼ .01). 11 The first evidence regarding the use of the TAI to guide interventions seems promising, highlighting the practical value of an evidence-based toolkit. 67

Strengths and limitations
This review provides an overview of all effective interventions on medication adherence in adults with asthma and COPD and summarize them into a practical, evidence-based toolkit, thus fulfilling a long-standing need as identified by the WHO. 8 Several tools exist to identify nonadherence or reasons for nonadherence in patients with asthma or COPD, 11,68-71 yet none of those provide advice regarding interventions a health care provider could actually apply to enhance medication adherence. To our knowledge, this is the first evidence-based toolkit that can provide tailored interventions to nonadherent patients with asthma and COPD. The Toolkit allows health care professionals to manage nonadherence in an efficient, personalized, yet protocol-based manner.
Some limitations should also be noted. Our review of effective interventions was limited to RCTs. As such, other useful interventions, not yet tested in RCTs, or not possible to test in RCTs, may have been excluded. Examples of such potentially effective interventions are caregiver support and removing financial barriers. 72,73 Another limitation was the partial description of most educational and multiple component interventions. This made it difficult to describe the exact content of those interventions and which parts were effective.

Implications and recommendations
Assessment and management of medication adherence are advised in many national and international asthma and COPD guidelines. 74,75 However, health care professionals generally have little time to identify nonadherence, discover the reasons for it, and provide the right interventions all in one consultation. The TAI Toolkit may help health care professionals to select the right evidence-based intervention efficiently for the right patient. The Toolkit could create more awareness about the topic of nonadherence, and could potentially be time-saving; yet optimal implementation of the TAI Toolkit in practice, including its usability, feasibility, and validity, needs to be assessed in future studies. Moreover, practical barriers, such as integration into current clinical workflows and consultations, need to be evaluated. The Toolkit is designed to be a dynamic tool that can be periodically updated when novel evidence emerges. Also, specific training and/or tools may be required to deliver each recommended intervention properly. Finally, further personalization of the TAI Toolkit will be required for populations with poorer healthy literacy, because this has been shown to be an important driver of suboptimal medication adherence. 76 This may include the use of intuitive graphical information. 77 Regarding novel evidence on adherence management, we recommend future studies to employ objective and uniform adherence measures, to allow better comparison of study outcomes. Also, because little is known about the long-term effects of adherence interventions, longer follow-up is recommended.

CONCLUSION
This review provides an overview of interventions that can enhance medication adherence in adult patients with asthma and/or COPD. Effective interventions were integrated into the practical TAI Toolkit, which can help health care professionals personalize adherence management by efficiently selecting the right adherence enhancing invention for the right patient.