Research in Social and Administrative Pharmacy Potential normalization of an asthma mHealth intervention in community pharmacies: Applying a theory-based framework

A B S T R A C T Background: E ﬀ ective mobile health (mHealth) interventions have been developed to support patients with their medication use, however to date few are widely used in pharmacy practice. Normalization of an intervention is essential to have a population impact, which is de ﬁ ned as ‘ the process of getting a new intervention into routine practice ’ . Objective: The aim of this study was to assess the normalization potential of a complex mHealth intervention for adolescents with asthma (ADolescent Adherence Patient Tool; ADAPT) in community pharmacy practice. Methods: The Normalization Process Theory (NPT), a sociological action theory, was retrospectively applied to study the normalization potential of ADAPT. NPT explains factors that promote or hinder implementation, embedding, and integration of new interventions in clinical practice. Evaluation data (structured interviews and questionnaires) of 23 pharmacists who used the ADAPT intervention were used for this study. Results: Pharmacists understood the purpose of the ADAPT intervention and were prepared to undertake the necessary work of implementation. However changes at di ﬀ erent levels are needed to support full normalization, such as changes in the intervention itself and changes in the pharmacist's work ﬂ ow. The potential for nor- malization could also be enhanced by the use of product champions and appropriate reimbursement guidelines, to ensure uptake of the intervention by other pharmacists. Support from professional bodies for the use of mHealth could also promote normalization. Conclusions: Normalization of mHealth is a complex continuous process. The ADAPT intervention has the po- tential to be normalized in community pharmacy practice, but full normalization would require changes in both daily pharmacy practice and reimbursement models.


Introduction
Suboptimal asthma control is common, i.e., around 50% of patients have uncontrolled asthma, mostly caused by medication non-adherence. Patients with uncontrolled asthma are at increased risk of exacerbations, which can result in hospitalizations or even deaths. Furthermore, decreased quality of life has been described. 1,2 These consequences contribute to increased healthcare costs for society.
Non-adherence rates are especially high during adolescence. 3 Mobile health (mHealth) interventions have the potential to support patients with their medication use, and can be in particular interesting for adolescents. 4,5 Medication intake behaviour is complex as it is affected by multiple unintentional (practical) and intentional (perceptual) barriers. 6 Therefore, interventions with multiple components are more effective in improving medication adherence than those aimed at only one aspect of non-adherent behaviour. 7,8 Several mHealth interventions with multiple components have been developed and use of these interventions resulted in increased adherence rates, improved self-management, or improved health status. [9][10][11][12] However, hardly any of them are currently implemented in clinical practice.
Complex interventions may be hard to implement in clinical practice, as they often require change at multiple levels involving different stakeholders, e.g., the patient, the healthcare professional, the healthcare organization, and the wider environment such as the national healthcare system. [13][14][15][16][17] These contextual factors are dynamic and can change over time. It has been suggested that the fit between an intervention and its context determines the success of the implementation. 13 Implementation science focuses on strategies to promote the uptake of interventions into routine practice. The Normalization Process Theory (NPT), a sociological action theory, focusses on the work required to implement new interventions in clinical practice. Normalization is defined as 'to become part of routine practice', and it covers different stages: implementation, embedding, and integration (Table 1). 18,19 NPT was developed to address factors that promote or hinder implementation, embedding, and integration of new practices. 18 It can be used to describe how complex healthcare interventions can become normalized.
Prior research showed that the ADolescent Adherence Patient Tool (ADAPT), a pharmacy based interactive mHealth intervention, improved medication adherence in adolescents with asthma having poor adherence rates. 20 However, the population impact of an intervention depends on both the effect size and the extent to which the intervention reaches the target population. 21 Thus the actual impact of an intervention is likely to be enhanced by integration into routine clinical practice. 14,19 Most previously developed mHealth interventions were local or isolated initiatives. Hardly any intervention is widely implemented and not much attention has been paid to a sustained normalization plan. 11,12,22 Therefore, the aim was to study the normalization potential of a mHealth intervention for adolescents with asthma in the community pharmacy, using the ADolescent Adherence Patient Tool (ADAPT) as an example.

Study design, setting, and participants
The NPT is retrospectively applied, per construct and per component, to the ADAPT intervention. 20,23 Evaluation data from the ADAPT study (which were previously collected) were used for this research. 24 The ADAPT study was a cluster randomised controlled trial to evaluate the effectiveness of the newly developed ADAPT intervention in Dutch community pharmacies. This study was approved by the Medical Review Ethics Committee of the University Medical Centre Utrecht (NL50997.041.14) and by the Institutional Review Board of the Utrecht Pharmacy Practice network for Education and Research (UPPER). In total, 66 pharmacists (independent pharmacies, of which some were part of pharmacy chains) individually decided to participate in the ADAPT study. Of these, 23 had access to the ADAPT intervention. At the end of the ADAPT study, a research assistant conducted structured interviews with those pharmacists to evaluate their experiences with the intervention. These structured interviews (i.e., questionnaires) contained open-ended questions on their experiences with the ADAPT intervention and their perceptions on implementation and integration in clinical practice. The pharmacists also completed a brief questionnaire where they used a 5-point Likert scale (totally agree to totally disagree) for statements related to their experiences and opinions about the ADAPT intervention. 24 The NPT was retrospectively applied to the evaluation data and the NPT Toolkit, consisting of 16 questions, 25 was used to evaluate the implementation, embedding, and integration of ADAPT in daily pharmacy practice.

Normalization Process Theory (NPT)
NPT is a sociological action theory, proposing first that complex interventions become routinely embedded and integrated in contexts as the result of people working, individually and collectively, to implement them. Action is regarded as more important than people's attitudes or intentions when implementing an intervention in healthcare. Second, the work of implementation is operationalized through four constructs of social action ( Table 2); (1) coherence: does it makes sense?, (2) cognitive participation: do I want to take part?, (3) collective action: what is the impact on work?, and (4) reflexive monitoring: is it worth it? These four constructs represent different stages and different kinds of work that people do as they work around a set of new practices, such as the use of a new intervention. Lastly, NPT proposes that the integration of a complex intervention requires continuous investment by people Table 1 Terminology and definitions used in the Normalization Process Theory (NPT).

Term Definition
Normalization To become part of routine practice, i.e., to take it for granted Embedding The process through which a practice (e.g., use of a new intervention) become routinely incorporated in everyday work of individuals and groups Implementation The social organization of bringing a practice (e.g., use of a new intervention) into action, thus actually using the intervention Integration The process by which a practice (e.g., use of a new intervention) is reproduced and sustained among social matrices of an organization or institution Table 2 The four constructs of the Normalization Process Theory (NPT) with the four corresponding components.

Construct Components Explanation
Coherence Sense-making

Internalization
Understanding the value, benefits, and importance Individual specification Understanding specific individual tasks and responsibilities Differentiation Understanding the distinctiveness Communal specification Working together with others to build a shared understanding of the aim, objective, and expected benefits Cognitive participation Effort

Initiation
Key participants drive implementation forward Enrolment Organizing or reorganizing of participants (and others) to collectively contribute Legitimation Ensuring that other participants believe it is right for them to be involved, and can make a valid contribution Activation Defining the actions and procedures needed to sustain using mHealth and stay involved Collective action Commitment Interactional workability Impact on interactions, particularly the interactions between healthcare professionals and patients (consultations) Relational integration Impact on relations between groups of professionals Skill set workability Fit between new intervention and existing skill sets Contextual integration Fit with overall organizational context; goals, morale, leadership and resources Reflexive monitoring Appraisal

Systematization Determining how effective and useful it is for participants and for others Communal appraisal
Working together (in formal collaboratives, or in informal groups) to evaluate the worth Individual appraisal Working experientially as individuals to appraise its effects on them and the contexts in which they are set Reconfiguration Attempting to redefine procedures or modify the intervention itself that carry forward in space and time. This means that continuous investing in sense-making, effort, commitment, and appraisal are necessary for the normalization of a complex intervention.

ADolescent Adherence Patient Tool (ADAPT)
The ADAPT intervention consisted of a smartphone application (app) for patients (iOS or Android), which was connected to a desktop management system of the patient's own community pharmacist. In the Netherlands, patients are usually registered in one single pharmacy and generally collect all their prescription medication in that specific pharmacy. The ADAPT intervention was an interactive mHealth intervention with several components to support different aspects of medication adherence and self-management: a weekly questionnaire to monitor symptoms, a medication reminder, short educational and motivational movies, a peer chat, and the opportunity to contact the pharmacist. During the ADAPT study, pharmacists and patients voluntarily participated and they had free access to the intervention. During the study, patient information was encrypted, using a code consisting of a pharmacy and patient number, to ensure privacy. All (personal) app data were encrypted using 128-bits Advanced Encryption Standard (AES) and were securely saved using Hypertext Transfer Protocol with a Secure Sockets Layer (HTTPS with SSL certificate). More details on the ADAPT intervention and study design have been described elsewhere. 20,23 Pharmacists were asked to support patients with their medication use by contacting them via chat messages, sending additional movies, or adjusting the frequency of the symptom monitor, if needed.
Pharmacists received e-mail notifications when patients sent a chat message or when the weekly symptom monitor indicated poor symptom control. All use of the app was recorded automatically. The ADAPT intervention was evaluated in a six months cluster randomised controlled trial with 234 patients, and improved adherence in adolescents with asthma demonstrating poor adherence rates. 20

Data analysis
The structured interviews with pharmacists were audiotaped and the recordings were transcribed verbatim. Summaries of responses per question were made, and a combination of analytical techniques (searching and finding answers to the questions) and tactics (connecting similar answers) were used to obtain a comprehensive data overview. NPT was mapped onto this data, i.e, retrospective thought experiment. The results were discussed with experts involved with the development of the NPT, resulting in several hypothesis results. Questionnaire data were divided in three groups per statement: agree (fully agree and agree), neutral, and disagree (disagree and totally disagree). Descriptive statistics were calculated using IBM SPSS Statistics for Windows, version 24.0.

Results
The pharmacist descriptives are shown in Table 3, and the results of the brief questionnaire are shown in Table 4. The results of the application of NPT to ADAPT are described below and summarized in Fig. 1.

Coherence
The concept of coherence refers to the extent to which users can make sense of the intervention. There are four subsidiary constructs: differentiation (the extent to which the intervention can be differentiated from similar interventions), communal specification (shared understanding of the intended benefits), individual specification (individual understanding of the intended benefits and the work required to realise these benefits), and internalization (understanding the value, benefits, and importance of the intervention). 25

Differentiation
Pharmacists were easily able to differentiate ADAPT from alternative methods to improve adherence to asthma medication, as, for most of them, this was their first experience with mHealth in the pharmacy. The ADAPT intervention consisted of a unique combination of interactive components to improve adherence. Pharmacists were aware that the desktop management system enabled the use of multiple components, such as the pharmacist chat, that facilitated contact between patients and pharmacist. These electronic consults (e-consults) were new for patients and pharmacists (Table 4).

Communal specification
Pharmacists who participated in the ADAPT study were aware of the problem of sub-optimal adherence to asthma medication in adolescents, agreed that this was an important problem, and understood that ADAPT aimed to improve adherence. Moreover, almost all pharmacists (22/23) thought that the pharmacy is the right place for mHealth interventions, like ADAPT, which indicates that they collectively perceive medication adherence as their responsibility. At the same time it was not clear whether and how pharmacists worked within their pharmacies with others to come to a shared understanding of using ADAPT in their specific context. "The pharmacy is the right place for mHealth interventions like ADAPT, because medication adherence and medication counselling belong to the core business of pharmacists." Male pharmacist, age 50 years.

Individual specification
To ensure individual specification, a half-day training about 'asthma and medication use by adolescents' was organized at the start of the ADAPT study, which was rated as useful by two thirds of the pharmacists (Table 4). In addition, pharmacists received instructions in the pharmacy how to use the ADAPT intervention and an intervention guide was designed to explain the (use of the) intervention. Most pharmacists (18/23) stated that they understood their specific tasks and responsibilities, such as using the intervention when receiving e-mail notifications. Some pharmacists (7/23) reported requiring additional information to implement the intervention in clinical practice, and suggested an electronic support manual integrated into the intervention.
"The e-mail notifications were easy to deal with and they could easily be found in the system." Female pharmacists, age 29 years

Internalization
The aim of the ADAPT intervention was to increase adherence among adolescents with asthma. Pharmacists voluntarily participated in the ADAPT study, and were thereby self-selecting and not surprisingly, already convinced of the importance of improving adherence. However, only 7 out of 23 pharmacists were familiar with electronic health (eHealth) interventions in the pharmacy, including mHealth. Thus, future implementation strategies should include an emphasis on the importance of adherence and the benefits of the ADAPT intervention as a starting point to ensure sense-making among pharmacists as well as providing information about the potential benefits of mHealth.

Cognitive participation
The second stage of normalization is about commitment, i.e., "do I want to participate", including the role of key participants, reorganization of tasks, and defining actions to stay involved. The four constructs which make up cognitive participation are initiation, enrolment, legitimation, and activation. 25

Initiation
Initiation refers to the extent to which key participants drive the work of implementation forward. In this case, pharmacists were expected to support patients by chat messages, sending additional movies, and adjusting the frequency of the symptom questionnaire if needed. In the current study, initiation was supported by the research team sending a monthly digital newsletter aiming to motivate and remind the participating pharmacists to be actively involved in use of the intervention. Data showed that most pharmacists contacted patients through the pharmacist chat, and monitored patients' symptoms. However, not many additional movies were sent 30 .

Enrolment
Enrolment refers to the extent to which participants have to (re-) organize themselves and others, in order to implement the intervention. Pharmacists need to change their daily routine in order to create time to use a new intervention. As the number of enrolled adolescents per pharmacy was relatively low, the total time devoted to the use of the ADAPT intervention remained limited (Table 4). This supported enrolment in the context of this research. However, with more widespread use of the intervention, the time required would be a significant barrier unless reimbursement systems were changed to reflect this additional work. 24 Twenty-one of the 23 participating pharmacists stated they would continue to use the ADAPT intervention in clinical practice only if their time would be reimbursed.
"Use of the ADAPT intervention costed very little time, approximately 5-10 minutes per week" Male pharmacist, age 40 years

Legitimation
Legitimation refers to participants believing that promoting the implementation is a legitimate part of their role. Almost all pharmacists in this study thought that the pharmacy was the right place for mHealth interventions like ADAPT (Table 4). Moreover, around half of the pharmacists thought that they could eventually make a valid contribution, because the ADAPT intervention resulted in improved medication use (11/23) and the intervention improved their insights into patients' asthma symptoms and medication use (13/23).

Activation
Activation is about defining actions and procedures to remain involved with the intervention. In order to support pharmacists, an intervention guide was developed to explain (the use of) the intervention. Moreover, in the desktop application a decision tree was added to support proper use of the intervention, i.e., define the appropriate actions. Additionally, pharmacists received e-mail notifications when patients needed help and a monthly digital newsletter reminded pharmacists to stay involved. For most pharmacists, the use of the intervention was clear and they used the intervention for the whole study period (Table 4). Unfortunately no (quantitative) data was available on the use of the intervention guide, and the usefulness of the decision tree.
"The ADAPT intervention was easy to work with, it was very clear to me." Female pharmacist, age 29 years

Collective action
The next stage of the normalization process is collective action, which refers to the impact on work and workflows of getting the intervention routinely embedded in clinical practice. The four constructs of collective action are interactive workability, relational integration, skill set workability, and contextual integration. In the ADAPT study, most pharmacists (17/23) used the ADAPT intervention during the complete study period.

Interactional workability
Interactional workability refers to the extent of which the intervention improved interactions between pharmacists and adolescents. In total, 17 of the 23 pharmacists agreed that the intervention facilitated such contact. The e-consults were an addition to current consultations and they contributed to co-operative interactions, such as shared-decision making. 26 "I used the chat function quite often. The patient completed the questionnaire to monitor symptoms which was nice, and sometimes I needed to contact the patient based on the symptom score. The direct contact thought the chat was new, because normally I would call them afterwards." Male pharmacist, age 31 years

Relational integration
Relational integration refers to the impact of the intervention on accountability, responsibility, and trust between the users. The ADAPT intervention provided a role for pharmacists. As for most pharmacists (19/23) the intervention assisted with the medication guidance of patients.

Skill set workability
Skill set workability is the extent to which existing skills of professionals fit with a new intervention. The use of the ADAPT intervention was allocated to pharmacists as their responsibility is to ensure right medication use of patients. In two pharmacies, a pharmacy technician specialized in asthma care was appointed to use the ADAPT intervention (Table 3). Using mHealth in clinical practice is new. Therefore, it is important that healthcare providers develop the right skills to support normalization. These skills can be acquired through trainings and workshops. 27 For most pharmacists the use of the ADAPT intervention was clear and most pharmacists thought the training was useful (Table 4).
"Clear instructions. I liked ADAPT." Female pharmacist, age 25 years

Contextual integration
Contextual integration refers to the fit with the overall organizational context. The ADAPT intervention contributed to integrated care and it delivered tools to pharmacists for medication counselling and for providing extra care, which does fit with the ongoing expanding role of pharmacists. 28 All pharmacists thought that an integration of the ADAPT stand-alone desktop application in the pharmacy computer system would support further implementation (Table 4).

Reflexive monitoring
The last stage of normalization is reflexive monitoring, which is the appraisal work that people do covering the effectiveness of the intervention and the redefinition of procedures. The four constructs which contribute to reflexive monitoring are systematization, communal appraisal, individual appraisal and reconfiguration.

Systematization
Systematization refers to the effectiveness and usefulness of an intervention. The ADAPT intervention was evaluated in a cluster randomised controlled trial, showing that the intervention effectively improved adherence in adolescents with asthma having poor adherence rates. 20 The pharmacist chat was the most effective component, which is in line with previous studies. 29,30 Communal and individual appraisal Communal and individual appraisal refers to working together, or individually, to evaluate the worth of the intervention and to appraise its effects on them and on the contexts. Almost all pharmacists (95.6%; 22/23) were satisfied with the ADAPT intervention. They were in general positive about the effect of the intervention for patients and themselves (Table 4). For some pharmacists, the ADAPT intervention did not meet their expectations (43.5%; 10/23). Their reasons included a low number of participating patients per pharmacy, reluctance of patients, time constraints, and non-intuitiveness of the intervention.
"I really liked participating in the ADAPT study and the training at the start was also very useful." Female pharmacists, age 29 years

Reconfiguration
Reconfiguration is about attempts to redefine procedures, modify practices, and change the intervention itself. The ADAPT intervention was especially effective in improving medication adherence in those at greatest risk. 20 Therefore, it might be useful if pharmacists could select non-adherent patients and provide them access to the intervention, i.e., tailor the intervention. Additionally, an integration of the ADAPT intervention in the pharmacy computer system and reimbursement guidelines will support the normalization of the intervention (Table 4).

Discussion
This study describes all factors related to the normalization of a complex asthma mHealth intervention in the community pharmacy setting. The findings tentatively suggest that the ADAPT intervention has the potential to become normalized in clinical practice as long as there is adequate financial reimbursement for the additional work required by pharmacists and sufficient investment in training and motivating pharmacists to use it. These factors require change at the health service level, and lack of such change may inhibit the normalization of mHealth in clinical practice.
The context of an intervention is important when trying to implement it in clinical practice. 13,19 In the current study, a mHealth intervention for adolescents with asthma in the community pharmacy was used as an example. Thus, the normalization potential may differ for other contexts or patient groups. Moreover, it should be taken into account that trials, like the ADAPT study, are not best suited to evaluate the normalization potential of complex interventions in real life. Trials are closed systems with strict requirements for the population and the intervention use, while interventions should eventually be integrated in a dynamic real-world environment. More (or other) barriers may emerge when implementing a new intervention in the real-world environment. For example not all pharmacists will have a positive attitude towards the intervention, or be motivated to normalize the intervention. A product champion (initiation) is therefore important to support pharmacists in using the intervention. Ideally the intervention should be continuously evaluated during an implementation phase after the efficacy of the intervention has been shown. 31 If the ADAPT mHealth intervention is normalized, it has the potential to support medication adherence of non-adherent patients, 20 and it can facilitate integration of care among different healthcare providers (including pharmacists, physicians, and nurses). For example the physician's role might be affected if the healthcare providing role of pharmacists increases. 32 Further research should therefore focus on the inclusion of other healthcare practices. Ultimately mHealth might be added to multidisciplinary treatment guidelines to support normalization.
Different research fields focus on the implementation of new interventions in healthcare, [13][14][15][16][17] and many models have been developed. 16,33,34 For example, cognitive science suggests that increased knowledge increases implementation, behavioural science suggests that implementation is influenced by feedback and incentives, marketing science suggests a clear and attractive intervention, social science suggests a change in social norms, and organizational science suggests a change on system levels. 15 In this study a sociological model was used, because sociology (i.e., the study of human social relationships, institutions, and society) is important when focussing on the implementation of new interventions in a complex and dynamic everyday healthcare setting. 35 It is now time to start implementing mHealth, 36 and the Normalization Process Theory is an action theory proposing that the implementation of an intervention in healthcare is the product of action, not necessarily people's attitudes or intentions. NPT also highlights all relevant aspects related to normalization (implementation, embedding, and integration), which makes it a complementary theory.

Limitations
In the current study, the normalization potential of a mHealth intervention was studied by retrospectively applying NPT and this approach has some limitations. Data were collected using a structured questionnaire that was not informed by NPT, and therefore did not address all the relevant elements of NPT. For other implementation studies, it is recommended to use the NPT at the start of the study when developing the questionnaires, to ensure sufficient arguments to confirm or reject a hypothesis. Secondly, the pharmacists in the ADAPT study voluntarily participated, and were probably positively biased towards the intervention, and therefore more likely to demonstrate cognitive participation. Despite these limitations, this study highlights aspects which are important for normalization and might need extra attention when trying to get a new intervention in routine pharmacy practice. Further research should focus on how to get other pharmacists (with a neutral or negative attitude) involved in using mHealth. The first steps to do this are described in the current study; ensure sensemaking and cognitive participation. Support from the pharmacist community (and upper management in pharmacy chains) to use mHealth may also be important to attain greater implementation.

Conclusions
Normalization of a complex mHealth intervention, like ADAPT, is a complex process, which involves changes at different levels and requires continuous investment of pharmacists. The ADAPT intervention may have the potential to become normalized, as sense-making, effort, commitment, and appraisal were predominantly positive in this sample. However changes in pharmacy practice appear to be needed to integrate mHealth into daily routine. These changes apply to the intervention, work flow, and appointing a product champion. Moreover, reimbursement and the support of professional bodies are likely to promote implementation and normalization.

Conflicts of interest
The authors have no conflicts of interest to declare. This work was supported by the Lung Foundation Netherlands [grant number 9.2.17.213FE]. For the ADAPT study, funding was received from the Netherlands Organisation for Health Research and Development (ZonMw) and from Umenz Benelux BV. The funders had no role in the design and conduct of the study, interpretation of the data, nor in the decision to submit the manuscript for publication.