Eliciting preferences for continuing medication among adult patients and parents of children with attention‐deficit hyperactivity disorder

Abstract Background Adherence to medication for attention‐deficit hyperactivity disorder (ADHD) is less than optimal. Previous studies have primarily focused on qualitative assessment of factors that influence medication adherence. Objective This study aimed to quantify the factors that influence patient and parent preferences for continuing ADHD medication. Method A discrete‐choice experiment was conducted to investigate preferences. Adults, and parents of children, with ADHD were presented with eight hypothetical choice tasks of three options (Medication A, Medication B, No Medication) described by six attributes related to medication outcomes. Preferences were estimated using a mixed multinomial logit model. Results Overall, respondents' preferences (n = 216) for continuing medication were negative (mean [β] = −1.426, p < .001); however, a significant heterogeneity in preferences was observed amongst respondents (standard deviation = 0.805, p < .001). Improvements in education, aggressive behaviour, social behaviour and family functioning, and side effects and stigma, influenced respondents' decision to continue taking medication. The respondents were willing to continue medication if they experienced positive effects, but side effects (even moderate) were the strongest concern for not continuing medication. While side effects were the most important factor for both adult patients and parents of children with ADHD, improvement in education was relatively more important for adults and improvement in aggressive behaviour, social behaviour and family functioning was relatively more important for parents of children with ADHD. Parents were more likely to not continue a medication with severe side effects even at the highest level of improvement in education. Conclusions Side effects are the most important factor that influenced preferences for continuing medication for both adults with ADHD, as well as parents of children with ADHD. While overall the respondents preferred not to take/give medication, discrete‐choice experiment showed that the relative importance of factors that influenced continuation of medications was different for the two groups. Patient and Public Involvement Adults, and parents of children, with ADHD participated in this study by completing the online questionnaire. The questionnaire was based on findings of research in the literature, as well as earlier focus groups conducted with adults, and parents of children, with ADHD. The face validity of the questionnaire was determined by asking parents of children, and adults, with ADHD (n = 3) to complete the survey and participate in a short discussion on their understanding of the questions and their recommendations on improving the clarity of the survey.

with ADHD. While overall the respondents preferred not to take/give medication, discrete-choice experiment showed that the relative importance of factors that influenced continuation of medications was different for the two groups.
Patient and Public Involvement: Adults, and parents of children, with ADHD participated in this study by completing the online questionnaire. The questionnaire was based on findings of research in the literature, as well as earlier focus groups conducted with adults, and parents of children, with ADHD. The face validity of the questionnaire was determined by asking parents of children, and adults, with ADHD (n = 3) to complete the survey and participate in a short discussion on their understanding of the questions and their recommendations on improving the clarity of the survey.

K E Y W O R D S
ADHD, adults, discontinuation, implementation, medication, parents, preferences 1 | BACKGROUND Attention-deficit hyperactivity disorder (ADHD) is defined as a persistent pattern of inattention, hyperactivity and/or impulsivity that interferes with an affected person's daily functioning. 1 ADHD is a common neurodevelopmental disorder among children, with a worldwide prevalence of 7.2%. 2 The prevalence of ADHD in adults is between 1.2% and 7.3% 3 ; however, it has been suggested that these estimates are not accurate as only one-quarter of adults living with ADHD are diagnosed. [4][5][6] ADHD can have a significant impact on the academic, occupational and social life as well as family relationships of those affected. 7 ADHD also has a significant economic impact and is associated with higher healthcare costs. The annual cost of raising children and adolescents with ADHD in the United States (US) has been estimated as US$124.5 billion. 8 Similarly, the estimated healthcare and productivity cost associated with adults with ADHD ranged between US$87 billion and US$138 billion in the US in 2010. 9 Given the breadth of impact, management of ADHD is important in achieving better health and social outcomes for those affected.
Pharmacotherapy is an important component of the overall management and is recommended by various international guidelines. [10][11][12] These guidelines suggest that long-term use of medication is important to achieve the desired medication outcomes; however, evidence shows that patients do not adhere well to medication 13 and often discontinue medication within the first few months of therapy. 14 Adherence to medication is defined as the process by which patients take their prescribed medication and is composed of initiation, implementation and discontinuation as well as persistence. 15 Numerous factors can influence medication adherence in people with ADHD. [14][15][16][17][18][19] Lack of medication effectiveness, side effects, stigma and high costs have primarily contributed to the higher rates of nonadherence. 13,20,21 While there is some information about how these factors impact adherence, 22 there are very limited quantitative data on the relative importance (RI) of these factors. Initiation is defined as the consumption of the first dose of a prescribed medication. Implementation is defined as the continuation of medication as prescribed. Discontinuation is defined as the cessation of a prescribed medication. 15 Once initiated, people with ADHD can struggle with their decision-making and often stop and restart medication multiple times through their medication-taking journey. 23,24 Identifying adherence phase-specific factors and their relative importance is central to designing targeted interventions, as factors influencing adherence vary in their impact on different phases of medication-taking. 22,25,26 Adherence research in ADHD has used sociobehavioural models to explain the complex phenomenon of adherence. [27][28][29][30][31] However, there is no clear evidence that these models can help in designing interventions to promote adherence that are effective and sustainable. 7 Furthermore, there is limited research investigating the choice that people make between various medication attributes, such as benefits, side effects and cost, to decide on medication adherence.
Patients continuously make a conscious decision to adhere/not adhere to medication by valuing the medication and its outcomes. 32 The valuation or utility can be estimated through a discrete-choice experiment (DCE). DCEs have been widely used for eliciting preferences in health research. 33  | 1095 these factors and investigate which factors are more important than others. Understanding the relative importance of the factors is important in prioritizing interventions to improve medication adherence and its outcomes. In particular, as noted above, it is important to assess the importance of factors at each of the three phases of medication-taking to develop tailored interventions. 25 A few studies have used DCEs to assess preference for medication in people with ADHD. [35][36][37][38] These studies have shown that the benefits of medication, particularly improvement in behaviour and social situations, were the most preferred attributes for patients and their carers.
However, none of the previous studies examined medication attributes in the context of the three phases of medication adherence. Moreover, previous studies were either limited to a class of medication, such as stimulants, 38 or a population, such as parents of children, 37 children 36 or adults with ADHD. 35 Therefore, the objective of this study was to investigate preferences for continuing ADHD medication based on key factors shown to influence adherence to ADHD medication, and determine the relative importance of those factors, in parents of children and adults with ADHD.

| Study participants and recruitment
Participants were recruited by advertising the research on online platforms such as Facebook. The invitation was advertised on various Australian-based ADHD support groups on Facebook and also on their websites. By using the 'Boost' option (available on Facebook), the invitation was customized to the target audience by predefining their location (Australia), age (18-65 years) and interest (ADHD, managing a child with ADHD, ADHD awareness). Potential participants were provided with an online link to access the anonymous survey. However, eligibility screening was performed first, which ensured that only eligible participants could access the survey.

| Study instrument
A web-based survey was conducted with parents of children, and adults, with ADHD in Australia. The survey was divided into three parts.
The first part screened the respondents for their eligibility to participate.
The second part involved choice-based questions (DCE) in which eligible respondents were asked to choose their preferred hypothetical medication (or no medication) from a set of alternatives based on given attributes and attribute levels. The third part asked questions about participants' characteristics such as, age, gender and income.

| Part 1: Eligibility criteria
The inclusion criteria were (1) parent of a child with ADHD aged up to 17 years, or an adult (18-65 years) diagnosed with ADHD, irrespective of when the diagnosis was made (as a child or as an adult), and (2) taking prescribed medication or giving it to their child for ADHD. Participants were only recruited if they were based in Australia. The survey was conducted in English.

| Part 2: Choice-based survey
The choice-based survey was conducted using a DCE. 36 DCEs provide a systematic method of quantifying trade-offs and the importance that a person assigns to a set of treatment attributes to decide whether they prefer a particular treatment, such as, medication. Each attribute (e.g., medication side effects) has various attribute levels (e.g., mild, moderate, severe) that describe the dimensions of that attribute. Each participant was provided with eight hypothetical choice tasks. In each choice task, participants were asked to choose their preferred medication (Medication A or Medication B) or a 'no medication' option (I will not continue either Medication A or Medication B). Medications A and B differed in terms of attribute levels. Those who chose the 'no medication' option were provided with a supplementary question which asked that if they had to choose between Medication A or B, which one would they prefer to continue. The supplementary question was asked to encourage people to make trade-offs between different treatment attributes.
The underlying theory of DCEs, the random utility theory, assumes that individuals are rational in making their decision and choose the alternative that maximizes their utility function. The utility function of each alternative is defined by its attribute levels and individuals' valuation of each parameter estimate (β). Therefore, the first step in designing the survey was to select a set of attributes and attribute levels to describe each alternative that was relevant to the study population.
A review of the literature was conducted to identify the factors that influence adherence to ADHD medication at the implementation and discontinuation phases of adherence. 39 This was followed by focus group discussions (FGs) with people with ADHD and parents of children with ADHD to further identify factors influencing adherence. 21,40 Identified factors were discussed amongst authors for their potential inclusion in the DCE. Six attributes that were consistently identified as important in the literature and the FGs were selected: (1) improvement in education and learning; (2) improvement in aggressive behaviour; (3) improvement in social behaviour; (4) improvement in family functioning; (5) severity of side effects; and (6) presence of social stigma. The first four attributes had three levels (no improvement, somewhat improved, considerably improved), the fifth attribute had four levels (none, mild, moderate, severe) and the last attribute had two levels (present or not present) ( Table 1). The attribute levels were chosen based on the literature 39  | 1097 used in ADHD, but to assess the factors that influence a decision of whether or not to continue with the medication. Given the large number of possible choice sets (n = 419,904), a full-factorial design was not feasible. Therefore, a fractional-factorial design was used, and a subset of 24 possible choice tasks was selected by generating an attribute level-balanced D-efficient design (Appendix S1).
Efficient designs have the potential to draw a subset that can yield more information, produce smaller standard errors and increase the reliability of the parameter estimates. 41 An important consideration in using efficient designs is the use of prior parameter estimates. However, it is also suggested that if the priors are not correct or close to actual behaviours, the efficient design can easily become inefficient. 42 Given that no priors were available from the literature, we set very small prior values, almost equal to zero, with the expected signs as predicted in Table 1.
Further, we ensured that no dominant alternatives were present in our survey. 43 Eight choice tasks, consistent with previous health research, 44,45 were deemed appropriate for each participant, which means that we used three blocks for our design. An example of a choice task is illustrated in Figure 1. provided detailed information about the choice survey and its objectives, as well as a detailed description of attributes and levels. Furthermore, a practice choice task was also given to the participants before the actual choice tasks (Figure 1). Since priors were not known, it was not possible to estimate the required sample size. Rose and Bliemer 42 suggest that the most practical approach in such a situation is to maximize the efforts to increase sample size within the available time and budget.

| Part 3: Participants' characteristics
The last section of the survey asked demographic questions such as age, gender, education, employment and other questions related to participants' past and current ADHD treatment.

| Data analysis
Data were analysed using NLOGIT version 6.0 to estimate a mixed multinomial model (MMNL). 46 McFadden's pseudo R 2 was used to The model also quantified the relative importance (RI) of each attribute. The RI of attributes is useful to describe how much difference each attribute could make in the overall utility of an alternative. The RI of attribute k was obtained by computing its impact on utility, expressed as the range X k between the highest part-worth and lowest part-worth utility, and then dividing by Σ k X k.
The RI of an attribute was calculated by dividing the utility range of that attribute by the total utility. 47 The utility range of an attribute was obtained by subtracting the lowest part-worth utility of that attribute from its highest part-worth utility, whereas the total utility was obtained by adding the utility range across the attributes. RI was presented as percentages. Subgroup analysis was conducted, and the RI was calculated for both parents of

| Relative importance of attributes
Overall, both parents and adults considered side effects as the most important attribute in their decision-making about the continuation of medication. The RI of side effects for parents and adults was 30.77% and 39.79%, respectively. The next important factor was improvement in education (RI = 25.37% for parents; 28.78% for adults), while stigma was the least important factor in both groups (RI = 2.41% for parents and 2.01% for adults).

| Simulated probabilities
The simulation findings showed that both parents and adults were more likely to continue taking medication if they were presented with mild levels of side effects compared to moderate and severe levels. At any given level of side effect (mild, moderate or severe), the probability of continuing medication was higher when there was considerable improvement in education compared to somewhat improvement in education. The probability of continuing medication was similar between parents and adults when the side effects were mild and moderate; however, a notable difference in preference was observed between the two groups when side effects were severe. Parents were less likely to continue giving medication to their child compared to adults when either the improvement in education was somewhat improved (probability of continuing medication 0.93% for parents vs. 2.22% for adults) or considerably improved (1.12% for parents vs. 2.42% for adults; Figure 3).

| DISCUSSION
The findings of this study showed negative preferences for continuing medication in parents of children and adults with ADHD. That is, they preferred not to continue with the medication. However, significant  Another important contribution of this study is that it highlights the difference in the impact of factors, particularly aggressive behaviour and social behaviour, on decision-making between parents and adults.
Our findings suggest that improvement in social behaviour and aggressive behaviour is more important to parents compared to adults with ADHD. We offer several possible explanations for the difference in the impact of factors. First, parents are generally more concerned about the aggressive behaviour of their child whose inability to recognize risk can put his/her and others' lives in danger. 22 Parents are also concerned that their child may hurt themselves or others unintentionally or may get expelled from school due to aggressive behaviour. These concerns may lead to psychological distress and lower sense of parenting competence, hence why they consider improvement in aggressive behaviour important. Furthermore, adults may be better at controlling aggressiveness compared to children. 53 Second, parents may be more concerned about the future of their children in terms of how they will cope with the wider social challenges of life such as educational, occupational, marital and interpersonal communication issues. Parents may believe that over time, these challenges may lead to negative social outcomes and social rejection of people with ADHD; hence, improvements in social behaviour may become more important to parents. Third, ADHD in adults, compared to children, is more likely to be associated with false beliefs, negative attitudes and lack of knowledge/understanding of ADHD among the public and health professionals, which might lead to self-as well as public stigmatization of an individual with ADHD. 54 These explanations, therefore, support the notion that the motivation to continue medication could be different for parents and adults.
Attributes were selected through a rigorous three-step process that included a comprehensive review of the literature, 39

| Limitations
The findings of this study should be interpreted within the context of some limitations. One of the main challenges in our research was in the recruitment of the right respondents to answer our survey: adults and parents. Therefore, one of the limitations of this study is our sample size. This was overcome by pooling both data sets and using interactions to identify differences in each group. However, the complexity of our model in terms of interactions and random parameters was limited by the sample size to ensure that the model was correctly estimated. However, the statistical significance of several attributes in the model suggests that lack of power was not a concern. Given the online nature of data collection, it was not possible to determine the response rate. Second, the part worth Furthermore, this study included participants who had experiences of taking medication to further reduce the hypothetical bias; however, it may limit the generalizability to those who do not have experience of taking ADHD medication. Finally, the sample may not be representative of the population with ADHD. For example, our sample included more females and more educated people compared to the national sample. 55 In the case of parents, the skewed sample may not significantly affect the interpretation of findings as evidence suggests that mothers take more responsibility of their child with ADHD compared to fathers. 56 However, in the case of adults with ADHD, it may limit the generalizability of the findings. Moreover, we relied on a self-reported diagnosis of ADHD, which may also contribute to sampling bias.

| Implications of findings for practice and research
Our study provides an important contribution to the literature about patient/parent preferences for ADHD and the RI that they place on various medication outcomes. Guidelines on the management of ADHD recommend that patients' and/or parents' preferences should be considered when designing treatment plans during clinical consultations. 57,58 There is a growing consensus that treatments that are more KHAN ET AL. acceptable are likely to result in higher adherence, which will in turn produce improved medication outcomes. 36 We believe that our findings provide evidence that each person has their own preferences for medication, and therefore, those preferences must be incorporated into clinical decision-making for designing treatment plans that meet an individual's needs. Identification of individual preferences would also allow clinicians to focus on issues that are more important to the patients, provide tailored education and address their concerns, which may contribute to improved medication adherence.

| CONCLUSIONS
Preferences for ADHD medication vary significantly amongst adults, and parents of children, with ADHD. Side effects were the most important factor that influenced decision-making about ADHD medication, followed by improvement in education, social behaviour, aggressive behaviour, family functioning and stigma. There was a willingness to continue medication for the benefits of medication despite the medication having side effects.
However, the extent to which a person balances the two attributes may vary between individuals. Therefore, the "one size fits all" approach to improve adherence is unlikely to be successful.

ACKNOWLEDGEMENT
This study received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.