Evaluation of Motivational Interviewing to Improve Psychotropic Medication Adherence in Adolescents

Abstract

Problem:

Medication adherence rates in adolescents are poor. The World Health Organization identified that those at greatest risk were nonwhite adolescents with depression. Medication nonadherence results in poorer mental health outcomes.

Objective:

The first aim of the study was to investigate if two motivational interviewing (MI) sessions would improve medication adherence in adolescents taking antidepressants and mood stabilizers. The second aim was to evaluate if attitudes toward medication correlated with adherence. The third aim was to determine if self-reported adherence scores were similar to electronic adherence data collected.

Methods:

The quasi-experimental study contained an MI intervention, including a baseline and postintervention assessment of adherence over 30 days. A total of 48 adolescents, ranging in age from 12 to 18 years, were recruited from a university mental health center to participate in the study; 41 completed the study. Four nurse practitioners and two child psychiatrists mastered the MI techniques evaluated with standardized measures. The Medication Electronic Monitoring System (MEMS) was the primary measure of medication adherence. The Drug Attitude Inventory (DAI) was used as a secondary measure. Finally, participants completed the client evaluation of MI and a satisfaction survey of MI.

Results:

At endpoint, 70.7% of the participants were taking their medications between 80% and 100% of the time, as measured over 30 days, compared with 43.9% of participants at baseline. Mean adherence scores significantly improved by 17% after two MI sessions. Mean baseline adherence scores were 63.7%, whereas mean endpoint adherence scores were 80.6% (p < 0.0001). The effect size was 0.65, demonstrating moderate effect. Participants (n = 29) who demonstrated 80% or greater adherence had DAI mean scores of 16.48, whereas those below 80% had a DAI mean of 15.5 (p = 0.73), demonstrating no significant difference on drug attitudes between the two groups at endpoint. DAI baseline mean scores were 14.2, whereas endpoint mean scores were 16.2. There was a significant difference between self-rated adherence and objective data collected from the MEMS caps as participants over-reported medication adherence by an average of 18.4% at baseline (t = 6.84, df = 40, p < 0.001). Participants reported a high degree of satisfaction with MI.

Conclusions:

MI is a promising intervention for adolescents to improve psychotropic medication adherence.

Introduction/Problem Statement

Aserious problem exists in the adolescent population with adherence to psychotropic medications. Medication adherence is defined as the patient's conformance with the provider's recommendation with respect to timing, dosage, and frequency of taking medication during the prescribed length of time (World Health Organization 2003) (WHO). Adherence rates to psychotropic medications in youth range from 34% to 89%, as reported by parents and child pill counts, electronic monitoring (Medication Electronic Monitoring System [MEMS]), saliva tests of stimulants, and therapeutic serum levels of mood stabilizers (Ghaziuddin et al. 1999; Burns et al. 2008; Hamrin et al. 2010; Dean et al. 2011; Fontanella et al. 2011; Yang et al. 2012). These rates cannot be compared or averaged as different measures and definitions of adherence were utilized. In the study evaluating parent-reported child medication adherence rates of 89% to stimulants, saliva tests of stimulants demonstrated a 53.5% rate of adherence to methylphenidate in these same youth (Pappadopulos et al. 2009). Therefore, self-report was significantly overestimated and the highest rates of adherence evaluated thus far were inaccurate. The level–dose ratio of drug metabolites was found to be an inaccurate measure of assessing antidepressant adherence (Woldu et al. 2011). Therefore, a more objective and systematic measure of adherence is needed. Several research studies have used the rate of 80% or greater as the measure of satisfactory adherence to medication (Ho et al. 2009; Nakonezny et al. 2010; Brain et al. 2013). Adherence is especially imperative in treatment with the selective serotonin reuptake inhibitors (SSRIs) or selective norepinephrine reuptake inhibitors (SNRIs) since patients could develop the uncomfortable discontinuation syndrome after 18–24 hours of missing the drug depending on the SSRI half-life. Therefore, patients taking SSRIs or SNRIs would benefit from taking medications from 80% to 100% of the time to prevent adverse effects and achieve optimal drug effect. When looking specifically at adherence to mood-stabilizing drugs, Mullins et al. (2005) and Fontanella et al. (2011) found that antidepressant discontinuation rates were greater than 70% in adult and adolescent patients when evaluating long-term medication maintenance. In addition, the Tordia team evaluated short-term use of the SSRIs for depression and found only 50% of 190 adolescents were adherent to SSRIs as measured by taking 70% of their antidepressants by pill count at 3 months (Waldu et al. 2011). Short-term adherence to medication treatment in a large sample of 5460 youth with bipolar disorder was also problematic, despite the severe consequences of being unmedicated for bipolar disorder (Bhowmik et al. 2013). Medication adherence rates tend to decline during adolescence due to increased autonomy and decreased parental involvement with medication management (Gau et al. 2006; Dean et al. 2011). The WHO (2003) identified that the greatest patient risk factors for medication nonadherence were being of adolescent age, nonwhite race, and having the diagnosis of depression. Outcomes of psychotropic medication nonadherence include longer duration of illness, more psychiatric episodes, reduced likelihood of remission, poorer functional outcomes, increased psychopathology, greater relapse rates, poorer school performance, and difficulty in interpersonal relationships with family, teachers, and friends, as well as increased suicidal behavior (Hamrin et al. 2010). Nakonezny et al. (2010) found that youth with greater than 80% adherence to fluoxetine demonstrated significantly lower severity of depressive symptoms than youth who were considered nonadherent by taking their medication less than 80% of the time at the 12-week follow-up study period. Therefore, adherence to greater than 80% of antidepressant and mood stabilizer treatment is necessary to achieve improved mental health outcomes for adolescents with depression and suicidality (Gibbons et al. 2006; TADS team 2007; Brent 2009; TORDIA 2010). In considering the length of time on medications required to achieve remission of symptoms for depression, the American Academy of Child and Adolescent Psychiatry (2013) recommends 9 months to 1 year of antidepressant medication treatment after symptom improvement to prevent relapse.

Review of the Existing Literature in Motivational Interviewing Interventions

There is a dearth of evidence-based interventions to successfully address the problem of medication nonadherence in adolescents taking psychotropic medications. Patient education has demonstrated limited usefulness in improving psychiatric medication adherence in youth and adults (Zygmunt et al. 2002; Hamrin et al. 2010). While motivational interviewing (MI) has demonstrated success in adolescents for adherence to medical treatments and substance abuse treatment, it has not been tested for efficacy in adolescents for psychotropic medication adherence. Studies addressing adherence to treatment using MI as an intervention included youth with diseases such as diabetes, asthma, substance use, and dietary adherence. These studies demonstrated improved physical and mental health behaviors, reductions in HBA1C levels, patient self-reported readiness for change, increased blood glucose monitoring, decreased substance use, improved asthma symptom scores, significant reduction of calories from fat intake, and significant decreases in cholesterol levels (Berg-Smith et al. 1999; Channon et al. 2007; Lundahl et al. 2010; Seid et al. 2012). In a randomized controlled trial using the MEMS cap as the primary measure, 50 Latino adults on SSRI medication for depression received brief MI (two to three sessions) intervention that resulted in significant improvement (30%) in medication adherence rates from 42% at baseline to 72% at completion (p < 0.01) (Interian et al. 2013). In evaluating population and session requirements for MI, Lundahl et al.'s (2010) meta-analysis of MI for substance use indicated that MI should be used in populations older than 12 years of age due to the requirement for abstract thinking. Four studies demonstrated that brief MI intervention with two sessions was efficacious in improving treatment adherence (Cannon et al. 2007; Lundahl et al. 2010; Seid et al. 2012; Interian et al. 2013).

Purpose

This study plans to contribute to the evidence-based literature to determine if MI can be used by child and adolescent psychiatric prescribers as a method to improve mental health outcomes in adolescents through improved treatment adherence to psychotropic medications. The first aim is to evaluate if psychotropic adherence to selective SSRIs/SNRIs and mood stabilizers improves in adolescents from baseline levels of adherence through the intervention of brief MI to postintervention levels of adherence. A second aim is to evaluate if adolescent attitudes positively correlate with actual medication adherence. A third aim is to evaluate if patient-reported adherence levels match the results obtained from objective MEMS. The last aim is to evaluate patients’ experiences and satisfaction with MI treatment as a part of their medication appointments.

Theoretical/Conceptual Framework Intervention Description

The self-determination theory (SDT) (Deci and Ryan 1985) is a theory highlighting the importance of autonomy, competence, and relationships as three fundamental needs that motivate people to initiate behaviors. Prescribers support patient self-efficacy through change theory stages, including the precontemplation stage (lack of awareness of the problem); the contemplation stage (identifying the problem, but not yet ready to commit to change); the preparation stage (readying for the change); the action stage (identifying goals and implementing the necessary changes); and the maintenance stage (continuing behaviors to maintain change) (Prochaska and Velicer 1997). MI (Miller and Rollinick 2012) is a direct, nonconfrontational communication approach that fits well with SDT and elicits intrinsic motivation toward improving one's health behavior. MI is a collaborative goal-oriented style of communication with particular attention to the language of change. It is designed to strengthen personal motivation and commitment to a specific goal by eliciting and exploring the person's own reasons for change within an atmosphere of acceptance and compassion (Miller and Rollnick 2012). The key elements of MI are to improve the clinician's working alliance with a client, to teach clinicians to manage resistance and express empathy toward clients, and to help clients build motivation to change while addressing ambivalence to change (Miller and Rollnick 2012). Clinicians use MI to help patients become discerning enough to evaluate the advantages and disadvantages of healthcare changes. MI explores the reasons for change by facilitating self-reflection questions about the problem, asking open-ended questions, and encouraging reflective and supporting statements demonstrating change talk. MI teaches prescribers to create a positive alliance with the patient, which has been found to be an important factor in increasing medication adherence (Verbeek-Heide and Mathot 2006; Pescosolido et al. 2007). MI was chosen for this study due to its prior success with adherence in youth with medical problems and for developmental reasons. Since adolescents will soon be entering young adulthood, they will be expected to manage their own psychiatric care. Improving self-efficacy and personal ownership for medication treatment is a developmental task. Adolescents receiving MI for substance abuse disorders endorsed a high degree of satisfaction with MI treatment (D'Amico et al. 2010, 2012a, 2012b), allowing clients to identify their personal motivators for medication treatment and increase autonomy in treatment decisions. There is no known harm or side effect for patients using MI identified in the research. Participant benefits could include improved medication adherence, increased involvement in treatment planning, development of intrinsic motivation for treatment adherence, increased collaboration with the prescriber, and improved mental health. Self-management of medication in adolescence could improve mood symptoms and prepare youth for young adulthood where self-management will be expected of them.

Methods

The study design is a preliminary efficacy and proof-of-concept evaluation of the effectiveness of MI to improve antidepressant adherence in adolescents. This is a quasi-experimental study containing an intervention and a pre and postintervention assessment. This is the first study to evaluate the effects of MI on psychotropic medication adherence in adolescents.

Procedures

Recruitment of 48 adolescent participants taking SSRIs, SNRIs, and mood stabilizers occurred. Evaluation of participants’ practice of taking medication for 30 days and their attitudes toward medication before and after two MI sessions was completed. The diagnosis were obtained from the patient's electronic medical record and included the patient's primary and secondary psychiatric diagnosis as defined by the treating psychiatric clinician based on Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM V) (American Psychiatric Association 2013) criteria. Participants had other psychiatric comorbidities and take additional psychotropic medications; however, only the SSRI/SNRI and mood stabilizer adherence was monitored by the MEMS cap system. SSRIs/SNRIs and mood stabilizers were selected as the drug chosen to monitor because of the potential for serious negative outcomes of nonadherence, including suicide as well as adverse side effects such as discontinuation syndrome. Participants were instructed not to put any other medications in their MEMS medication bottle, not to remove the medication from their bottle and place it in another container, and to close the cap after each use. The study evaluated if two MI sessions delivered within standardized medication treatment appointments delivered by the prescriber improve psychotropic medication adherence compared with medication adherence at baseline. While participant recruitment took place over a 2-month period, four nurse practitioners and two child psychiatrist prescribers agreed to receive 24 hours of MI training and implement it twice with patients from their caseload who were recruited for the study. Twenty-four hours of MI training were selected as Berg-Smith et al. (1999) identified that 18 hours of MI training were insufficient for the providers to learn the MI skills. The MI training occurred over a 3-day period. A hired certified MINT trainer evaluated each prescriber's MI skills at baseline before training with the video assessment of simulated encounters-revised (VASE-R), and then the trainer evaluated each prescriber's MI skills after the 24-hour training using the VASE-R to rate the prescriber's ability in using the MI principles with a patient. The prescribers were required to demonstrate baseline proficiency of MI skills with a test patient at 1 month of completion of the MI training according to the Motivational Interviewing Treatment Integrity Manual (MITI-4) standards. All prescribers passed the MITI-4 audiotaped examination at baseline. MI skills were evaluated again by audiotape at midpoint during the study to ensure treatment fidelity by attaining a baseline proficiency score of MI skills using the MITI-4. All prescribers received a passing proficiency score or greater at the midpoint evaluation. After the recruitment, consent, and assent, demographic and Drug Attitude Inventory (DAI) information was collected on each patient in a private room within the clinic by the principal investigator. Once all 48 patients completed demographic and DAI forms, all participants received the baseline MEMS cap and bottle for their antidepressant/mood stabilizer for the duration of 4 months during the study period. Once the adolescents had been monitored by the MEMS cap for 30 days, they were eligible to receive their first MI intervention session within their monthly medication clinic appointment. At the end of the 4-month monitoring period, the patients returned the MEMS cap to the principal investigator and completed postintervention ratings using the DAI form, Client Evaluation of Motivational Interviewing Scale (CEMI) form, and patient satisfaction survey. The last 30 days of the patients’ MEMS cap data were then utilized for their endpoint measure of adherence. Adolescents received the MI treatment alone; parents were not included as some adolescents attend appointments independently at times and the consistency of the intervention needed to be the same for all participants. Adolescents only were assessed for intrinsic motivation of the MI intervention on the adolescent toward medication adherence as measured by the DAI and MEMS caps. One hundred percent adherence meant taking the medication over the 30-day period exactly as the medication was prescribed. For example, if a patient missed 2 days of medication, 2 days was divided by 30 days for a percentage of medication missed. Excessive bottle opening was not counted in adherence evaluation. During the first month, patient adherence to medication was monitored electronically using the MEMS cap and a baseline adherence score was obtained. At the start of the second month, each patient had one MI session with a trained MI prescriber, and at the start of the third month of monitoring, each patient had a second MI session with the MI prescriber. The data obtained by the MEMS during the fourth month were the data used to assess patient endpoint compliance. Excessive bottle opening was not counted in overall adherence. Participants received $20 for their participation in the study since additional time at the clinic was required to fill out forms. While MI was individualized for each patient, each prescriber would ask the participants what their goals were for their mental health treatment as well as how medication treatment fit in with their mental health goals. They were asked how many of their medication pills they thought they took over the previous month at each visit and how motivated they were to take their medications on a scale of 0–10. Once the participants identified a number for level of motivation, they were asked why not a lower number to endorse competency. Participants were asked to identify the pros and cons of taking medications, as well as barriers to taking medications. Once barriers were identified, the prescriber asked the participant for suggestions to improve the number of medication doses they were taking. If the participant had difficulty providing ideas to improve adherence, the prescriber asked the participant if they were open to hearing adherence ideas. The participant was then asked if any of the ideas seemed useful to them. Once a plan was made to improve medication adherence, the participant was asked to identify a numerical value of commitment to the plan based on a Likert scale of 0–10. On the MI follow-up visit, the prescriber would ascertain adherence and problem solve with the patient on ways to modify or continue the plan. Prescribers would acknowledge verbally patient-reported success with the plan. Clinicians were required to make two reflective comments for each question asked and summarize the discussion with the participant. T tests were used to analyze pre and postdata from the same population. Pearson's correlation r coefficient was used to compare medication rates with DAI scores. Cohen's D was used to calculate effect size.

Participant characteristics

Adolescents between 12 and 18 years of age with disorders, including major depression, generalized anxiety, bipolar disorder, and post traumatic stress disorder (PTSD), who were receiving SSRIs, SNRIs, SARIs, mood stabilizers, and aminoketone drugs for at least 1 month, were recruited from an outpatient university-based child and adolescent mental health clinic. These diagnoses were obtained from the patient's medical record once consent was obtained. Two months were allotted for recruitment with rolling admission into the study until 48 participants were enrolled.

Those with comorbid psychiatric conditions and diagnosed medical problems were not excluded. Adolescents were on medication with both single daily dosing and twice daily dosing. The study only measured one antidepressant or mood-stabilizing drug for the 4-month duration of the study. The adolescents were recruited from an outpatient university psychiatric clinic servicing 1080 adolescent patients in an urban setting. This outpatient university psychiatric setting provides medication management, evidence-based psychotherapies, and psychological testing for children and adolescents with any psychiatric problem. There were 12 psychiatric prescribers within the child and adolescent outpatient team, and 6 agreed and signed consent forms to participate in the study. The entire outpatient adolescent psychiatric population consists of 51% girls and 49% boys. Five hundred sixty-four (52%) patients had TennCare, and 455 (42%) patients had commercial insurance, and 19 patients were private pay (6%). Coding of ethnicity did not allow for analysis of data of the entire clinic population. Flyers and a staffed information table advertising the study were available in the waiting area to promote interest in participation. Interested patients who contacted the principal investigator (PI) were provided with information about the purpose, procedures, confidentiality, voluntary participation, and risks and benefits of involvement and assessed for eligibility criteria. Participants and guardians met with the PI to provide written parental consent and adolescent written assent.

Inclusion criteria

Criteria considered were adolescents 12–18 years of age, of either gender and of any ethnicity, and currently receiving SSRI/SNRI or mood-stabilizing treatment for 1 month minimum at one university outpatient clinic. Patients needed to currently be receiving medication management from prescribers at the university outpatient child and adolescent clinic. There was no maximum length of time that one could be on medications so that a large enough sample could be recruited. The adolescents needed to be able to read, write, and speak to complete the DAI and CEMI.

Exclusion criteria

Adolescents with verbal communication deficits were not included in the study as MI requires a dialog between the clinician and adolescent. Adolescents with severe self-care deficits were excluded as the intervention of MI involves planning and processing skills, as well as independence in activities of daily living such as the ability to distribute their own medication. Patients 17 and younger could not participate without parental written consent. Illiteracy and current psychosis were also exclusion criteria.

Study measures

Medication Electronic Monitoring System

MEMS was selected as the primary measure to determine rates of medication adherence since the MEMS has been shown to be a more accurate and objective measure of adherence compared with clinician report, patient report, or parent report (Nakonezny et al. 2010; Interian et al. 2013). The MEMS caps measure how many times the patient opens the pill bottle during the time of evaluation of medication monitoring period and records the date, times, and frequency of the opening of the pill bottle. Blood level concentration of SSRIs is not a reliable adherence measure due to dosing issues, interindividual blood concentration rates, drug absorption rates, and drug metabolism (Woldu et al. 2011; Bosman et al. 2014). Nakonezny et al. (2010) compared medication adherence with four different adherence methods in 31 adolescents with major depression taking fluoxetine and found electronic monitoring to be the most accurate measure for medication adherence compared with clinician report, pill count, and parent/patient diary. De Blesser et al. (2010) evaluated the MEMS and found that the 24-month battery life MEMS produced no errors of missing registrations or over-registrations of 1025 openings during the testing. The MEMS was used as an outcome measure in this study and not as an intervention. Participants were not given feedback about the number of pills actually taken as delivered in intervention studies. In a study where the MEMS cap was used as an intervention to increase adherence to highly active antiretroviral treatment (HAART), results were not significant that the MEMS feedback increased medication adherence (Davies et al. 2010).

Drug Attitude Inventory

The secondary measure used in this study was the DAI, which is a 30-item yes/no format comprising seven components, including positive and negative attitudes toward medication, an illness model, external and internal loci of control, relapse prevention, and a measure of harm/toxicity. Total scale scores range from −30 to +30, with negative scores being associated with nonadherence and positive scores being associated with adherence. The DAI scale has been validated by Hogan et al. (1983) in 150 outpatient schizophrenic adults. The DAI discriminated 88% of the time between adherent and nonadherent patients. Townsend et al. (2009) also validated the DAI in a study with 122 adolescents with bipolar disorder, major depression, and attention-deficit/hyperactivity disorder (ADHD). Internal consistency was 0.889, demonstrating a correlation between attitudes and self-reported adherence.

The Client Evaluation of MI scale

The third rating scale, CEMI, measured the client's perceptions of MI interventions after receiving two sessions. The CEMI is a 16-item scale measuring the participant's evaluation of the technical and relational aspects of the implementation of MI used by the clinician. The CEMI questions assess clinicians’ use of MI concepts, including collaboration, evocation, respecting autonomy, expressing empathy, rolling with resistance, developing discrepancy, supporting self-efficacy, open questions, affirmations, reflections, and summaries that the clinician uses with the patient. Participants were asked to rate each item with the instruction, “During your most recent counseling session, how much did your clinician demonstrate each behavior?” using a four-point Likert-type scale (1 = never to 4 = a great deal). The CEMI demonstrated an internal consistency of 0.90 for technical factors and 0.88 for relational factors used in MI (Madson et al. 2013). Scores range from 0, indicating not MI consistent, to 32, demonstrating the highest level of MI consistency. The participants met with the PI to complete the CEMI and the second DAI after the final MEMS cap monitoring at the clinic.

Motivational Interviewing Treatment Integrity Manual

Prescriber fidelity to MI skills was measured by the MITI-4.1. The fourth edition of this scale, the MITI-4.1, was utilized in this study (Moyers et al. 2014 “Motivational Interviewing Treatment Integrity,” Coding Manual 4.1.; unpublished manual). There are two components of the MITI, the global scores and the behavior counts. The global score takes into consideration how the MI clinician creates a partnership that exudes empathy. The behavior counts focus on the behaviors of the clinician in the sessions. When looking at the scores from MITI-4.1, there are two passing scores: fair and good. To be classified in the fair competency range, the clinician should foster collaboration, make demonstrated efforts to understand the client's point of view, consistently reflect the client's change talk, and avoid emphasis on the status quo. When evaluating the behavior counts, the clinician should use more reflections than questions. The ratio goal is one question for every reflection for fair competency and two reflections for every question for good competency. Reflections should consist of more complex rather than simple reflections. For fair competency, at least 40% of the reflection should be complex, and for good competency, at least 50% should be complex. Furthermore, for the conversation to meet standards for MI, there should be no clinician confrontations. For the purposes of this study, clinicians needed to demonstrate basic beginning proficiency (fair) before providing MI to clients in the study and once again at the study midpoint.

Video Assessment of Simulated Encounters-Revised

The VASE-R is a video-based examination method used for evaluating MI skillfulness that involves videotaped vignettes simulating real-world clinical encounters (Rosengren et al. 2009). After enrollment, demographic data were collected from each study participant, including age, race, sex, type of payee, name and date of diagnoses, name and number of all medications, and date the medication was started. Demographic information included one open-ended question about any concerns that adolescents may have about being on psychiatric medications. The DAI was administered at the same time as the collection of demographic information. Use of the MEMS caps began after all participants were enrolled to prevent variations in time of data collection. After completion of the first 30-day MEMS cap evaluation, participants will individually receive two MI sessions from their treating prescriber as a part of their standard medication clinic appointment 1 month apart. Each adolescent received the MI treatment without a parent in the room to ensure that the treatment is completed in the same format as some adolescents come to some return appointments without a parent. The PI reviewed the patient charts to ensure that two MI interventions are documented in each patient's chart before the MEMS caps are redistributed to all participants as the endpoint measure. Participant's rates of adherence were monitored for 1 month after the intervention using the MEMS caps. The CEMI and repeat DAI measure were given at the clinic by the PI after the second 30-day MEMS monitoring is complete. Finally, a three-question satisfaction survey was given to each participant: How would you rate the quality of the MI treatment? How satisfied are you with the MI treatment? Has the MI you received helped you deal more effectively with taking your medication? Likert scale scoring ranged from a score of 1, indicating a response of poor, unsatisfied, or MI made things worse, to 4, indicating excellent, very satisfied, and MI helped a great deal. The duration of the study was 6 months from enrollment to completion of the study for participants. All study procedures were approved by the Vanderbilt University human subject review committee.

Statistical analyses

Demographics and other characteristics were screened for associations by adherence group using chi-square tests to evaluate for differences in categorical data, with Fisher's exact test when cell size was less than five. Differences in means by adherence group were evaluated with T tests with folded F testing to determine equality of variance in samples, with the Satterthwaite method used where unequal variances were present. Paired t-tests were used to analyze pre and postintervention data on the medication adherence (MEMS) and drug attitudes (DAI). To further evaluate change associated with the intervention, adherence was evaluated dichotomously with 80%–100% adherence considered adherent versus <80% adherence considered nonadherent, and McNemar's test (Q _M) was used to evaluate change in adherence from baseline to postintervention. Cohen's D was used to calculate effect size of the intervention. Comparison of correlation coefficients was used to evaluate the associations between variables. A two-sided p-value of 0.05 for significance was used in all types of analyses, which were conducted using SAS (SAS Institute, Cary, NC), version 9.4, of the SAS System for Windows.

Results

Participants

A total of 48 adolescents between 12 and 18 years of age self-selected to participate in the study from an outpatient university-based child and adolescent mental health clinic serving 1080 adolescent patients.

Only 41 adolescents were included in the final analysis of the study as seven youth withdrew early from the study due to needing a higher level of psychiatric care (3), moving out of state (1), or discontinuing medication treatment appointments (3) (Table 1).

Table 1.

Characteristics of Sample by Baseline and Postintervention Adherence

		Baseline adherence		Postintervention adherence
Characteristic	Total (n = 41)	≥80%	<80%	≥80%	<80%
Gender, n (%)
Female	24 (58.5)	9 (50)	15 (65.2)	16 (66.7)	8 (33.3)
Male	17 (41.5)	9 (50)	8 (47.1)	13 (76.5)	4 (23.5)
Age (years), mean (SD)	15.6 (1.5)	15.7 (1.6)	15.5 (1.4)	15.7 (1.5)	15.2 (1.5)
Range (years)	12–18	12–18	13–18	12–18	13.17
Insurance, n (%)
State	17 (41.5)	7 (41.2)	10 (58.8)	12 (70.6)	5 (29.4)
Private	24 (58.5)	11 (45.8)	13 (54.2)	17 (70.8)	7 (29.2)
Race, n (%)
Nonwhite	9 (21.9)	2 (22.2)	7 (77.8)	4 (44.4)	5 (55.6)
White	32 (78.1)	16 (50)	16 (50)	25 (78.1)	7 (21.9)
Diagnosis, n (%)
Depression	30 (73.2)	10 (33.3)	20 (66.7)^a	20 (66.7)	10 (33.3)
Mood disorder NOS	2 (4.9)	1 (50)	1 (50)	2 (100)	0
Anxiety	20 (48.8)	8 (40)	12 (60)	16 (80)	4 (20)
ADHD	20 (48.8)	11 (55)	9 (45)	16 (80)	4 (20)
OCD	3 (7.3)	2 (66.7)	1 (33.3)	2 (66.7)	1 (33.3)
PTSD	8 (19.5)	3 (37.5)	5 (62.5)	6 (75)	2 (25)
AS	5 (12.2)	4 (80)	1 (20)	5 (100)	0
BPD	5 (12.2)	4 (80)	1 (20)	3 (60)	2 (40)
Eating disorder	1 (2.4)	1 (100)	0	1 (100)	0
Substance abuse	1 (2.4)	1 (100)	0	1 (100)	0
Separation anxiety	1 (2.4)	1 (100)	0	1 (100)	0
Social phobia	2 (4.9)	1 (50)	1 (50)	2 (100)	0
ODD	1 (2.4)	0	1 (100)	1 (100)	0
Panic disorder	2 (4.9)	1 (50)	1 (50)	2 (100)	0
Number of psychiatric
diagnoses, mean (SD)	2.5 (0.7)	2.7 (0.7)	2.3 (0.8)	2.7 (0.7)	1.9 (0.7)^b
1 Diagnosis, n (%)	4 (9.8)	1 (25)	3 (75)	1 (25)	3 (75)^ad
2	16 (39.0)	5 (31.3)	11 (68.8)	9 (56.3)	7 (43.8)
3	19 (46.3)	11 (57.9)	8 (42.1)	17 (89.5)	2 (10.5)
4	2 (4.9)	1 (50)	1 (50)	2 (100)	0
Psychotropic medication, n (%)
Mood Rx only	11 (26.8)	3 (27.3)	8 (72.7)	7 (63.6)	4 (36.4)
SSRI	31 (75.6)	13 (41.9)	18 (58.1)	22 (71.0)	9 (29.0)
SNRI	5 (12.2)	1 (20)	4 (80)	5 (100)	0
Bupropion	6 (14.6)	4 (66.7)	2 (33.3)	4 (66.7)	2 (33.3)
Mood stabilizer	6 (14.6)	5 (83.3)	1 (16.7)^a	4 (66.7)	2 (33.3)
Antipsychotic	9 (22.0)	6 (66.7)	3 (33.3)	7 (77.8)	2 (22.2)
Stimulant	15 (36.6)	7 (46.7)	8 (53.3)	12 (80)	3 (20)
Alpha agonist	5 (12.2)	3 (60)	2 (40)	3 (60)	2 (40)
Benzodiazepine	1 (2.4)	1 (100)	0	1 (100)	0
Buspirone	6 (14.6)	4 (66.7)	2 (33.3)	6 (100)	0
Trazodone	7 (17.1)	5 (71.4)	2 (28.6)	7 (100)	0
Medical medication	18 (43.9)	9 (50)	9 (50)	12 (66.7)	6 (33.3)
Number of psychotropic
medications, mean (SD)	2.3 (1.2)	2.9 (1.2)	1.8 (1.0)^b	2.6 (1.2)	1.7 (0.9)^a
Range	1–5	1–5	1–4	1–5	1–3
1 Medication, n (%)	14 (34.2)	2 (14.3)	12 (85.7)	7 (50)	7 (50)^e
2	10 (24.4)	5 (50)	5 (50)	8 (80)	2 (20)
3	10 (24.4)	6 (60)	4 (40)	7 (70)	3 (30)
4 or more	7 (17.1)	5 (71.4)	2 (28.6)^ad	7 (100)	0
Mood Rx medication dosing
Once daily, n (%)	38 (92.7)	16 (42.1)	22 (57.9)	27 (71.1)	11 (28.9)
Twice a day	3 (7.3)	2 (66.7)	1 (33.3)	2 (66.7)	1 (33.3)
Months treated with medication (all), mean (SD)	47.0 (40.4)	50.3 (43.3)	44.4 (38.7)	55.0 (42.6)	27.5 (26.9)^a
Range in months	1–156	8–156	1–120	9–156	1–72
<6 months treated	4 (9.8)	0	4 (100) .12	0	4 (100)^b
6–18 months	10 (24.4)	5 (50)	5 (50) 0.7	8 (80)	2 (20)
>18 months	27 (65.9)	13 (48.2)	14 (51.9) 0.4	21 (77.8)	6 (22.2)
Months treated with antidepressant, mean (SD)	28.7 (33.1)	37.3 (40.6)	21.9 (24.8)	32.8 (35.3)	18.9 (26.0)
Range in months	1–144	3–144	1–84	1–144	1–72
Other medication, n (%)
None	23 (56.1)	9 (39.1)	14 (60.9)	17 (73.9)	6 (26.1)
1 other medication	12 (29.3)	6 (50)	6 (50)	8 (66.7)	4 (33.3)
2	3 (7.3)	1 (33.3)	2 (66.7)	1 (33.3)	2 (66.7)
3	1 (2.4)	1 (100)	0	1 (100)	0
4	1 (2.4)	0	1 (100)	1 (100)	0
5	1 (2.4)	1 (100)	0	1 (100)	0
CEMI-R score, mean (SD)	25.0 (4.0)	25.1 (4.3)	24.9 (3.8)	25.1 (4.4)	24.6 (2.8)
Range	12–30	12–30	17–30	12–30	19–27
CEMI-T score, mean (SD)	25.8 (4.8)	27.3 (4.1)	24.5 (5.0)	25.4 (5.3)	26.7 (3.2)
Range	11–32	15–32	11–32	11–32	21–32
Satisfaction score, mean (SD)	10.3 (1.4)	10.3 (1.4)	10.2 (1.5)	10.3 (1.5)	10.2 (1.3)
Range	7–12	8–12	7–12	7–12	8–12

Statistical Significance testing using t-test, χ ², or Fisher's exact test unless otherwise noted: ^a p ≤ 0.05, ^b p ≤ 0.01; significant Cochran–Armitage test for trend ^d p ≤ 0.01, ^e p ≤ 0.05.

ADHD, attention-deficit/hyperactivity disorder; AS, autism spectrum disorder; BPD, bipolar disorder; CEMI, Client Evaluation of Motivational Interviewing Scale; NOS, not otherwise specified; OCD, obsessive compulsive disorder; ODD, oppositional defiant disorder; PTSD, post traumatic stress disorder; SSRI, selective serotonin reuptake inhibitor.

Participants included 17 males (41.5%) and 24 females (58.5%) and the average age of participants was 15.6 years (SD = 1.5) (Table 1). Twenty-four adolescents had private medical insurance (58%) and 17 (41%) participants had state medical insurance. Racial backgrounds consisted of 32 (78.1%) Caucasian youth, 3 (7.3%) black youth, 3 (7.3%) Hispanic youth, and 3 (7.3%) youth of mixed race. Due to the small size of each subgroup, race was classified as nonwhite (9 [21.9%]) and white (32 [78.1%]) for analyses. The majority of the adolescents had a primary diagnosis of major depressive disorder (n = 30), of whom 17 were dually diagnosed with having major depression and generalized anxiety disorder. Four adolescents had a primary diagnosis of having bipolar disorder, three had a primary diagnosis of having generalized anxiety disorder, two had a primary diagnosis of having mood disorder not otherwise specified (NOS), and two had a primary diagnosis of having posttraumatic stress disorder; comorbid diagnoses included posttraumatic stress disorder, obsessive-compulsive disorder, ADHD, and autism spectrum disorder, social phobia, panic disorder, substance abuse, eating disorder, and one oppositional defiant disorder. The average number of diagnoses per patient was 2.5 (SD = 0.7). Thirty-one (76%) adolescents reported lifetime past medication nonadherence. The average number of medications per patient was 2.3 (SD = 1.2). Twenty-seven (65.9%) study participants received medication treatment for more than 18 months before the study. Ten patients had been on medications between 6 to 18 months, and four patients had been on medication less than 6 months.

Characteristics by baseline adherence

At baseline, there were no significant differences by adherence status (≥80% adherence vs. <80% adherence) based on gender, age, race, or insurance status (Table 1). Trends by race suggested that baseline adherence was lower in black and Hispanic participants and higher in Caucasians and multiracial participants; however, subgroup size was very small (n = 3 in each nonwhite group) and Fisher's exact and chi-square tests were not significant. When grouped as nonwhite versus white participants, a nonsignificant trend toward lower adherence was seen in the nonwhite group (p ≤ 0.25, Fisher's exact test).

At baseline, individuals with a diagnosis of having depression were 5.3 times more likely to be nonadherent (66.7% vs. 33.3%, odds ratio = 5.3, 95% confidence interval: 1.2–24.6, χ ² = 5.07, df = 1, p = 0.02). Individuals with a secondary diagnosis of ADHD, autism spectrum disorder, or a primary diagnosis of having bipolar disorder were more likely to be adherent at baseline; however, relationships did not attain statistical significance (Table 1). A test of trend from one to four psychiatric diagnoses by adherence showed more adherence with three or more diagnoses and neared significance (Cochran–Armitage trend test, Z = −1.56, p = 0.12). Type of psychotropic medication did not exhibit any associations with baseline adherence, although trends suggest that individuals treated with mood stabilizers (n = 6) were more adherent (p ≤ 0.07, Fisher's exact test). Number of psychotropic medications prescribed did show differences by baseline adherence, with those meeting 80% or more adherence being prescribed an average of 2.9 medications (SD = 1.2) compared with those nonadherent at an average of 1.8 (SD = 1.0) (t = −3.07, df = 39, p = 0.0038). In addition, evaluation of number of medications (range 1–4 or more) by adherence was significant (p = 0.0324, Fisher's exact test) and a trend test was also significant (Cochran–Armitage trend test, Z = −2.75, p = 0.0059). Dosing (once daily vs. twice daily) was not significant by baseline adherence. Medication treatment length was greater for those who were adherent. Those with less than 6 months of treatment were nonadherent at baseline using the greater than 80% criteria, with average adherence of 32% (SD = 21.6, range 12%–57%), and those with longer treatment were approximately equal in adherence at baseline. Furthermore, there were no significant associations between baseline adherence and having medical medications prescribed.

Self-report of adherence versus MEMS adherence at baseline

Adolescents significantly overestimated their baseline adherence by 18.4% percent (82.1% self-reported mean adherence) compared with objective MEMS cap data (63.8% mean adherence) during initial adherence evaluation (t = 6.8, df = 40, p < 0.0001). Self-reported adherence was an overestimate for all except for 7 youth. Only five participants (12.2%) were accurate in their self-report of adherence versus MEMS data at baseline, of these, three took their medication 100% of the time, one adolescent took the medication 97% of the time, and one adolescent took medication 0% of the time. Mean self-reported adherence at baseline was significantly lower in those nonadherent versus adherent (71.8 [SD = 22] vs. 95.3 [SD = 5.3]) using the MEMS cap measure of adherence (t = −4.92, df = 24.2, p < 0.0001). Of those self-reporting adherence of 80% or greater, 37.9% were found to be nonadherent using the MEMS cap (p < 0.0003, Fisher's exact test) with an average of 28% (SD = 16.8) over-reporting (t = −4.92, df = 25.2, p < 0.0001; Table 2).

Table 2.

Baseline Adherence by Medication Electronic Monitoring System Cap and Self-Report Measures

		Baseline adherence
Characteristic	Total (n = 41)	≥80%	<80%
MEMS, baseline percent adherence
Mean (SD)	63.8 (28.4)	89.2 (7.7)	43.8 (21.7)^c
Range	0–100	80–100	0–77
≥80% adherence, n (%)	18 (43.9)	18 (100)	—
<80%	23 (56.1)	—	23 (100)
Self-report, baseline percent adherence
Mean (SD)	82.1 (20.4)	95.3 (5.3)	71.8 (22.0)
Range	0–100	83–100	0–100
≥80% adherence, n (%)	29 (70.7)	18 (62.1)	11 (37.9)^c
<80%	12 (29.3)	0	12 (100)
Percent difference, self-report vs. MEMS
Mean % (SD)	−18.4 (17.2)	−6.1 (6.7)	−28 (16.8)^c
Range	−51 to 3	−20 to 2	−51 to 3
Over-report ≥15%, n (%)	21 (51.2)	3 (14.3)	18 (85.7)^c
Over-report <15%	20 (48.8)	15 (75)	5 (25)

p < 0.001.

MEMS, Medication Electronic Monitoring System.

Postintervention characteristics

A similar trend remained in race with nonwhite participants having lower adherence, although the percent adherence doubled from baseline (22.2%) to postintervention in the nonwhite subgroup (44.4%; p = 0.0929, Fisher's exact test). This could indicate that nonwhite participants are very responsive to the MI intervention. McNemar's test showed that change in adherence was significant in the white subgroup (10 individuals changed from nonadherent to adherent, and 1 moved from adherent to nonadherent; Q _M = 7.36, df = 1, p = 0.0067), with no significant change in the nonwhite subgroup (three individuals moved from nonadherent to adherent pre to post, while one moved from adherent to nonadherent; Q _M = 1.0, df = 1, p = 0.3173).

The direct relationship between the greater number of psychiatric diagnoses and higher adherence remained at postintervention, with the mean number of psychiatric diagnoses being 2.7 (SD = 0.7) in those with greater than 80% adherence and 1.9 (SD = 0.7) in those considered nonadherent (t = −3.4, df = 39, p = 0.0016; Table 1). As the number of diagnoses increased, adherence increased (p < 0.0175, Fisher's exact test) and the trend test also remained significant (Cochran–Armitage trend test, Z = −3.0608, p = 0.0022; Table 1). Similar findings were identified with number of psychotropic medications. Finally, a significant difference was found with months of treatment, with those adherent having a significantly higher average length of psychotropic medication treatment (55 months [SD = 42.6] vs. 27.5 [SD = 26.9]; t = −2.1, df = 39, p = 0.05). Individuals with less than 6 months of psychotropic medication treatment had an average postintervention adherence of 56.3% (SD = 24.7, range 20%–73%), the average change in adherence from baseline to postintervention was 24.3% (SD = 22.2) with a range of 8% to 57%. For those on medications longer than 6 months, the average change was less, 16.1% (SD = 21.1) with a range of −16% to 64%. There was no significant difference in the mean percent change in adherence from baseline to postintervention (t = −0.73, df = 39, p = 0.47).

Antidepressant adherence

Mean baseline adherence for the entire sample was 63.8% (SD = 28.4) and significantly improved at postintervention to a mean adherence rate of 80.7% (t = −5.15, df = 40, p < 0.001, Table 3), with an effect size of 0.65 demonstrating moderate effect. Adherence to antidepressants was poor at baseline with only 18 of the 41 (43.9%) adolescents adherent when defined as taking medication 80%–100% of the time. Postintervention adherence significantly improved to 70.7% with 29 adolescents meeting 80% or greater adherence using the MEMS cap (Q _M = 8.07, df = 1, p = 0.0045; Table 2). Results indicate that 13 youth, nonadherent at baseline, were adherent postintervention, while only two moved from adherent to nonadherent. Adolescents who were less than 80% adherent at baseline had significantly greater mean change in adherence (using the MEMS cap) from baseline to endpoint (29.3%) compared with a mean change of 1.2% in those with 80% or greater adherence at baseline (t = 6.22, df = 28.663, p < 0.0001).

Table 3.

Postintervention Outcomes: Adherence and Drug Attitude Inventory Scores

Measure	Baseline	Postintervention	Significance
Mean adherence % (SD)	63.8 (28.4)	80.7 (23.4)	^a t = −5.15, df = 40, p < 0.001
>80% adherent, n (%)	18 (43.9)	29 (70.7)	^b(Q _M) = 8.07, df = 1, p = 0.0045
DAI score (SD)	14.2 (8.4)	16.2 (8.2)	^a t = −1.99, df = 40, p = 0.0539
	Mean difference (baseline to postintervention)		95% confidence interval
Adherence, %	16.93		10.28 to 23.57
Range	−16 to 64
DAI score	1.98		−0.03 to 3.99
Range	−14 to 18

Paired t-test.

McNemar's test.

DAI, Drug Attitude Inventory.

Drug Attitude Inventory

The correlation between postadherence scores and the DAI scores for the entire sample was r = 0.16 (p = 0.31). In this sample, 17 participants demonstrated increased scores on the DAI (range 2–18), 12 participants’ scores decreased on the DAI (range −2 to −14), and 12 participants’ DAI scores remained the same from baseline to endpoint. There was a nearly significant change of 2.0 points from baseline (14.2, SD = 8.4) to endpoint (16.2, SD = 8.2) on DAI scores as a result of MI (t = 1.99, df = 40, p = 0.054, Table 3). The 18 participants who demonstrated greater than 80% adherence had a significantly higher baseline DAI mean score of 17.1 (SD = 5.5) than the baseline mean scores on the DAI (12.0, SD = 9.7) of the 23 participants with less than 80% baseline adherence (t = −2.16, df = 35.919, p = 0.0377, Satterthwaite method). Of the 29 participants who had greater than 80% adherence at endpoint, their DAI mean score was 16.5 (SD = 8.5), which was not significantly different than the 12 participants with less than 80% adherence at endpoint (DAI mean score of 15.5, SD = 7.5, t = −0.35, df = 39, p = 0.7312). This finding demonstrated a significant improvement in attitude in the nonadherent group. The correlation between DAI at baseline and change in adherence was −0.042 (p = 0.006), suggesting that those with high baseline attitudes toward medication had less change in adherence and those with low drug attitudes at baseline had greater change in adherence.

The top participant concerns about medications, as ascertained by open-ended questions, included side effects such as sedation, weight gain, diet restrictions, and fears of developing suicidal thoughts (n = 9); disliking having to try different medications to find the right one for symptom improvement (n = 5); fears of medication changing one's personality (n = 3); fears that the medication would not alleviate depression and anxiety symptoms (n = 3); and feeling forced to take medication by parents and other adults (n = 2). In addition, participants endorsed both shame of being on medication (n = 2) and fears of being on the medication for the rest of their life (n = 2).

Treatment fidelity

Six prescribers received MI training and passed the initial skills examination before intervention with patients and were found proficient at the midpoint evaluation with the MITI-4.1.

Adolescents rated their prescriber's performance of MI on both relationship measures and technical performance of MI skills. The CEMI-R mean score for relationship scores was 25.0 (SD = 4.0) and the CEMI-Technical score mean was 25.8 (SD = 4.8) for technical scores (Table 1). A three-question satisfaction survey found that an overall mean score was 10.3 of a possible 12, suggesting high levels of satisfaction with the MI intervention by participants.

Discussion

This was the first study to evaluate the effectiveness of MI treatment to improve medication adherence in adolescents on psychotropic medications. The intervention of two MI sessions by the patient's prescribing clinician significantly improved adolescent adherence mean scores by 16.9% (t = 5.15, df = 40, p < 0.0001) and demonstrated a moderate effect size of 0.65. This is a notable finding considering three patients were at 100% adherence at baseline and had no room to improve. MI improved adolescents’ self-management of medication, preparing youth for young adulthood where they will be responsible for all of their mental health treatment. Twenty-four hours of evidence-based MI training for prescribers were sufficient to obtain both adequate proficiency in skills, as assessed by the MITI-4.1, and CEMI scores, as well as improve medication treatment adherence (MEMS). This is a promising result for prescribers, as demonstrated by the six prescribers who were able to master MI communication skills within 24 hours of training and apply it with patients in their caseload to achieve adherence improvements.

Findings also suggest that as the number of psychiatric diagnoses and number of psychotropic medications taken increase, adherence time on medication was also greater. This may be related to greater motivation to take medications if symptoms are more severe, although this study was not designed to evaluate this relationship. Further study of the relationship between severity of illness and medication adherence should be pursued in the future.

This study also highlighted the fact that self-reported medication adherence is flawed. Adolescents initially overestimated their initial adherence rates by 18.4% percent compared with objective MEMS cap data at baseline. In a meta-analysis of several studies comparing MEMS with self-rated adherence in mostly HIV populations and single studies of youth on psychiatric medications, patient-reported adherence was overestimated by 6%–13% (Nakonezny et al. 2010; Shi et al. 2010; Yang et al. 2012). These findings are concerning due to the fact that prescribers make decisions to either increase doses of medication or change medications based on the patient and parent report of how frequently medications are taken and the impact of the medication on the patient's symptoms as well as side effects. During this study, 5 of 41 participants (12.2%) reported wanting better symptom relief from their medications when their actual medication adherence rates (MEMS) were less than 80%, classifying these participants as nonadherent. This raises the possibility that the medication was not being taken often enough to be effective. Only 60% of those who reported adherence greater than 80% of the time were actually adherent patients at that level. Self-reported adherence should be interpreted with caution in clinical settings. While the MEMS cap is a more accurate measure of medication adherence, a few limitations of using the MEMS cap in a clinical setting include the high cost of the MEMS cap, coordination of refills with the pharmacy, and transfer of prescriptions from a pharmacy bottle to an MEMS cap bottle. A few of the participants in the study used a daily pill box to assist in the organization of their medication, and an MEMS bottle and cap could disrupt their current pill box organization strategy.

Scoring on the DAI ranges from −30, indicating an extremely negative attitude, to +30, indicating an extremely positive attitude. The difference in DAI score showed that adherent patients (taking medication greater than 80% of the time) demonstrated significantly more positive attitudes toward medication than nonadherent patients at baseline. This is consistent with De las Cuevas et al. (2014) finding that positive attitudes towards medications on the DAI was consistent with actual medication adherence in 160 psychiatric outpatients taking antidepressant medication. The lack of difference in DAI scores at endpoint between those who were adherent greater than 80% and those who were adherent less than 80% was due to nonadherent participants demonstrating an increase in their DAI scores, which were more similar in drug attitude scores to those who were adherent at the endpoint. This suggests that the MI intervention may have had some influence on overall drug attitude in those who were not adherent despite not significantly changing in the overall group. Baby et al. (2010) evaluated medication adherence in 75 adults with schizophrenia where adherence was considered as taking medications greater than 75% of the time. The adherent group DAI score was 11.5 and the nonadherent group DAI score was 8.55, showing no significant difference between the adherent and nonadherent groups on drug attitudes similar to the finding in this study. There were no significant changes in the seven drug attitude subscores from baseline to endpoint, including subjective negative experiences with medication, attitudes toward health and illness, physician knowledge, locus of control, prevention, and views about medications causing harm. Free choice is a very important concept in transtheoretical change theory as goals of MI are to help individuals develop intrinsic reasons for personal health self-care. However, adolescents are still minors and their parents are still involved in their healthcare treatment, which has implications for total autonomy resting with the adolescent alone for medication. In this sample, 75.61% of adolescents positively endorsed the questions that they take medication of their own free choice at baseline and 78% positively endorsed that they take their medication of their own free choice at endpoint on the DAI. Prescribers want adolescents to have input and investment into their medication treatment plan as this could improve mental health outcomes. Chowdhury et al. (2012) conducted a 6-week medication educational training for 18 adolescents using the DAI as the primary measure pre and postintervention and found that attitudes decreased toward medication by −9.2% after the educational intervention. MI is well suited for changing behaviors as the focus is on identifying an area of change the client wants to make and devising a plan to modify behavior. It is encouraging to report that behaviors toward medication can be changed regardless of attitudes toward medications.

Overall, adolescents rated prescribers as demonstrating a high level of consistency in implementing MI skills, as measured by participant report on the CEMI. CEMI scores demonstrated that adolescents in this study found the prescribers to have high scores in relational skills of empathy and partnership (25) and technical skills of chasing change talk, avoiding sustain talk using, reflective comments, and promoting patient autonomy (25.8) of 32. Adolescents reported a very high level of satisfaction and improved adherence as a result of their MI sessions. The four participants who did not rate self-improvement in adherence as a result of the MI intervention were those who demonstrated strong levels of adherence to medications at baseline by taking medications greater than 80% of the time on the MEMS. MI is a recommended intervention for those patients who endorse ambivalence toward medications or express barriers to taking psychotropic medications.

In applying SDT to the MI intervention, patients were asked to discuss the pros and cons incurred by taking psychotropic medications. MI allowed for exploration of the patient's fears, concerns, barriers, and benefits of medication treatment and allowed prescribers to listen, support, and provide requested guidance and medication changes regarding real and feared side effects and develop a collaborative relationship around developing a plan for improved efficacy and symptom reduction. In addition, MI provided an opportunity for the adolescent to safely discuss the feelings about illness adjustment, shame, stigma, and lack of control over treatment. Adolescents did not report significant side effects as a result of taking SSRIs, SNRIs, or mood stabilizers, the drugs evaluated in this study. However, adolescents did report more physical costs related to the antipsychotics and stimulant categories of medications that they were taking for comorbid psychiatric diagnoses. MI works on the premise of creating self-efficacy for one's healthcare, which was very relevant to these adolescents’ concerns, by helping them establish their own goals for their mental healthcare and interventions to achieve their goals.

Limitations

The quasi-experimental nature of this study, including lack of randomization or a control group, limits the generalizability and inferences of the findings. The MEMS cap monitored only SSRI/SNRI and mood stabilizer medication, so this study did not capture variability in adherence by type of medications per patient. Awareness of being monitored by the MEMS cap while taking medication could possibly have had the effect of increasing medication adherence. The observation period did not include follow-up to assess the long-term effects of MI on improving medication adherence. While the study took 6 months from baseline to endpoint, seven participants were lost due to discontinuation of treatment at the university clinic and their data were not able to be included in the analysis. Of the seven patients who dropped out of the study, their baseline DAI scores were 16.9 compared with DAI scores of the sample who completed the study, which was 14.2 at baseline. Their self-reported drug adherence was 74% compared with 82% of the total sample who completed the study. Consistent MEMS data were not available; however, from the available information, the dropouts were not significantly different from those completing the intervention. Bias could have resulted due to self-selection of patients who are more adherent, demonstrating greater interest in participation; however, initial medication adherence rates were low among some in this cohort. While the MEMS cap recorded the number of times the pill bottle was opened, it could not track if the medication was actually swallowed. Finally, this study captured two of the World Health Organization high-risk populations for nonadherence, including adolescents with depression; however, this study did not represent minority ethnicities as the majority of the population in this study were Caucasian.

Suggestions for future research

Future studies should only enroll patients for whom adherence (less than 80% adherence) or expressed ambivalence about taking medications is problematic. Using prescribers who are certified in MI and are experienced in working with adolescents could enhance the quality control of the intervention. Methodology of using a randomized controlled sample would assist in obtaining generalizability for the larger population. Future studies could also evaluate if psychiatric symptoms improved as a result of increased adherence. The lack of measures of symptom improvement should be included so as not to confound the relationship between MI and adherence. In addition, evaluation of baseline adherence using the MEMS cap for 2 months rather than 1 month before the MI intervention could reduce the potential effect of being monitored by the MEMS cap as a factor in initial adherence monitoring. A 2-month time period would allow participants to get used to the cap as a part of their normal routine rather than the cap being a novelty. Another suggestion would be to evaluate the long-term outcomes of MI at 1 year to determine if the intervention had a long-term benefit for adherence. Additionally, since some participants will discontinue the study for various reasons, a larger study group should be obtained at the outset. Last, as researchers find that parental involvement increases youths’ adherence, this study design did not include parental involvement as it sought to evaluate the adolescents’ intrinsic motivation for changing their medication adherence. Further research should evaluate the effect of inclusion of a parent in the MI intervention on adherence outcomes. Evaluation of the severity of depression should be evaluated as De la Cuevas et al. (2014) found patients with more severe depression were less adherent to medication.

Conclusions

This was the first study to evaluate the effectiveness of MI to improve psychotropic medication adherence in adolescents. Medication adherence in adolescents was poor at baseline as only 43.9% of adolescents in this study took their antidepressants and mood stabilizer medication between 80% and 100% of the time as measured by the MEMS. MI proved to be a successful intervention to improve medication adherence for adolescents taking antidepressants and mood stabilizers with 70.7% of participants taking their medication greater than 80% of the time at endpoint. The majority of youth endorsed positive drug attitudes causing low variability and less movement for change in attitudes toward medications and adherence. Patients who were categorized as adherent demonstrated higher drug attitude scores than those at baseline; however, drug attitude scores increased among those who were not greater than 80% adherent at endpoint to a similar level as those who were adherent, so there was no statistically significant difference between the two groups. Participants significantly overestimated the amount of medication they took compared with objective MEMS cap data obtained. The use of MEMS caps may be useful to corroborate patient-reported adherence in patients who do not seem to demonstrate symptom improvement.

Footnotes

Acknowledgments

The authors would like to thank the following individuals for their support and contributions to this study: Lesa Abney, Meg Benningfield, Leah Bowen, Madeline Briere, Molly Butler, Cheryl Cobb, Ronald and Leanne Crabbe, Kaitlyn Fredricks, Jeff Gordon, Stephan Heckers, Teena McGuiness, and Cristina Warren.

Disclosures

Vanya Hamrin and coauthors have had no financial, institutional, or commercial conflicts of interest. The authors have no affiliations with any pharmaceutical companies. There was no financial support required for this article.

References

American Academy of Child and Adolescent Psychiatry: Retrieved from: www.uptodate.com/../depression-treatment-options-for-adolescents, 2013.

American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, 5th ed. Washington, DC: American Psychiatric Association; 2013.

Baby

, Gupta

, Sagar

: Attitudes and subjective reasons of medication compliance and noncompliance among outpatients with schizophrenia in India. J Epidemiol, 7, 2008.

Berg-Smith

, Stevens

, Brown

, Van Horn

, Gernhofer

, Peters

, Greenberry

, Snetselaar

, Ahrens

, Smith

: A brief motivational intervention to improve dietary adherence in adolescents. Health Educ Res, 14:399–410, 1999.

Bhowmik

, Aparasu

, Rajan

, Sherer

, Ochoa-Perez

, Chen

: The utilization of psychopharmacological treatment and medication adherence among Medicaid enrolled children and adolescents with bipolar depression. J Affect Disord, 150:424–429, 2013.

Bosman

, terHorst

, Hospes

: Adherence of antidepressant during pregnancy: MEMS compared with three other methods. Ther Adv Psychopharmacol, 4:61–69, 2014.

Brain

, Allerby

, Sameby

, Quinlan

, Joas

, Karliampi

, Lindstrom

, Eberhard

, Burns

, Waern

: Drug attitude and other predictors of medication adherence in schizophrenia: 12 months of electronic monitoring (MEMSR) in the Swedish Coast-study. Euro Neuropsychopharmacol, 23:1754–1762, 2013.

Brent

: Youth depression and suicide: Selective serotonin reuptake inhibitors treat the former and prevent the latter. Can J Psychiatry, 54:76–77, 2009.

Burns

, Cortell

, Wanger

: Treatment compliance in adolescents after attempted suicide: A two-year follow up study. J Acad Child Adolesc Psychiatry, 47:948–957, 2008.

10.

Channon

, Suws-Thomas

, Rollnick

, Hood

, Cannings-John

, Rogers

, Gregory

: A multicenter randomized controlled trail of motivational interviewing in teenagers with diabetes. Diabetes Care, 30:1390–1395, 2007.

11.

D'Amico

, Hunter

, Miles

JNV

, Ewing

, Osilla

: A randomized controlled trial of a group motivational interviewing intervention for adolescents with a first time alcohol or drug offense. J Subst Abuse Treat, 45:400–408, 2013.

12.

D'Amico

, Osilla

, Hunter

: Developing a group motivational interviewing intervention for adolescents at-risk for developing an alcohol or drug use disorder. Alcohol Treat Q, 28:417–436, 2010.

13.

D'Amico

, Osilla

, Miles

JNV

, Ewing

, Sullivan

, Katz

, et al.: Assessing motivational interviewing integrity for group interventions with at-risk adolescents. Psychol Addict Behav, 26:994–1000, 2012a.

14.

D'Amico

, Tucker

, Miles

JNV

, Zhou

, Shih

, Green

HDJ

: Preventing alcohol use with a voluntary after school program for middle school students: Results from a cluster randomized controlled trial of CHOICE. Prev Sci, 13:415–425, 2012b.

15.

Davies

, Asghar

, Cooper

: Does feedback of medication execution using MEMS caps aid adherence to HAART?: The MEMRI study (MEMS as Realistic Intervention). J Int AIDS Soc, 13(Suppl. 4):120, 2010.

16.

Dean

, Wragg

, Draper

, McDermott

: Predictors of medication adherence in children receiving medication. J Pediatr Child Health, 47:350–355, 2011.

17.

De Blesser

, De Geest

, Vandenbroeck

, Vanhaecke

, Dobbels

: How accurate are electronic monitoring devices? A laboratory study testing two devices to measure medication adherence. Sensors, 10:1652–1660, 2010.

18.

Deci

, Ryan

: Intrinsic Motivation and Self-Determination in Human Behavior. New York, Plenum, 1985.

19.

De las Cuevas

, Penate

, Sanz

: Risk factors for non-adherence to antidepressant treatment in patients with mood disorders. Euro J Clin Pharmacol, 70:89–98, 2014.

20.

Fontanella

, Bridge

, Marcus

, Campo

: Factors associated with antidepressant adherence for medicaid enrolled children and adolescents. Ann Psychopharmacol, 45:898–908, 2011.

21.

Gau

, Shen

, Chou

, Tang

, Chiu

, Gau

: Determinants to adherence to methylphenidate and impact of poor adherence on maternal and family measures. J Child Adolesc Psychopharmacol, 16:286–297, 2006.

22.

Ghaziuddin

, King

, Hovey

, Zaccagnini

, Ghaziuddin

: Medication non-compliance in adolescents with psychiatric disorders. Child Psychiatry Hum Dev, 130:103–110, 1999.

23.

Gibbons

, Hur

, Bhaumik

, Mann

: The relationship between antidepressant prescription rates and rate of early adolescent suicide. Am J Psychiatry, 163:1898–1904, 2006.

24.

Hamrin

, McCarthy

, Tyson

: Pediatric psychotropic medication initiation and adherence: A literature review based on social exchange theory. J Child Adolesc Psychiatr Nurs, 23:151–172, 2010.

25.

, Bryason

, Rumsfeld

: Medication adherence: Its importance in cardiovascular outcomes. Circulation, 119:3028–3035, 2009.

26.

Hogan

, Awad

, Eastwood

: A self-report scale predictive of drug compliance in schizophrenics: Reliability and discriminative validity. Psychol Med, 13:177–183, 1983.

27.

Interian

, Lewis

, Fenandez

, Gara

, Escobar

: A randomized-controlled trial of an intervention to improve antidepressant adherence among Latinos with depression. Depress Anxiety, 00:1–9, 2013.

28.

Lundahl

, Kunz

, Brownell

, Tollefson

, Burke

: A meta-analysis of motivational interviewing: Twenty-five years of empirical studies. Res Soc Work Pract, 20:137–160, 2010.

29.

Madson

, Mohn

, Zuckoff

, Schumacher

, Kogan

, Hutchison

, Magee

, Stein

: Measuring client perceptions of motivational interviewing: Factor analysis of the client evaluation of motivational interviewing scale. J Subst Abuse Treat, 44:330–335, 2013.

30.

Miller

, Rollnick

: Motivational Interviewing. New York, Guilford Press, 2013.

31.

Mullins

, Shaya

, Meng

: Persistence, switching, and discontinuation rates among patients receiving sertraline, paroxetine, and citalopram. Pharmacotherapy, 25:660–667, 2005.

32.

Nakonezkny

, Hughes

, Emslie

: A comparison of various methods of measuring antidepressant mediation adherence among children and adolescents with major depressive disorder in a 12-week open trial of fluoxetine. J Child and Adolesc Psychopharmacol, 20:431–439, 2010.

33.

Nakonezny

, Byerly

, Rush

: Electronic monitoring of antipsychotic medication adherence in outpatients with schizophrenia or schizoaffective disorder: an empirical evaluation of its reliability and predictive validity. Psychiatry Res, 157:259–63, 2008.

34.

Pappadopulos

, Jensen

, Chait

, Arnold

, Swanson

, Greenhill

, et al.: Medication adherence in the MTA. Saliva methylphenidate samples versus parent report and mediating effect of concomitant behavioral treatment. J Am Acad of Child Adoles Psychiatry, 41:514–521, 2009.

35.

Pescosolido

, Perry

, Martin

, McLeod

, Jensen

: Stigmatizing attitudes and beliefs about treatment and psychiatric medications for children with mental illness. Psychiatr Serv, 58:613–618, 2007.

36.

Prochaska

, Velicer

: The trans-theoretical model of health behavior change. Am J Health Promot, 12:38–48, 1997.

37.

Rosengren

, Hartzler

, Baer

, Wells

, Dunn

: The video assessment of simulated encounters-revised (VASE-R): Reliability and validity of a revised measure of motivational interviewing skills. Drug Alcohol Depend, 97:130–138, 2008.

38.

Seid

, D'Amico

, Varni

, Munafo

, Britto

, Kercsmar

, Drotar

, King

, Darvie

: The in vivo adherence intervention for at risk adolescents with asthma: Report of a randomized pilot trial. J Pediatr Psychol, 37:390–403, 2012.

39.

Shi

, Lui

, Fonseca

, Walker

, Kalsekar

, Pawaskar

: Correlation between adherence rates measured by MEMS and self-reported questionnaires: A meta-analysis. Health Qual Life Outcomes. 1186–1195, 2010.

40.

The Health Foundation: Evidence Scan: Training Professionals in Motivational Interviewing, 2011. www.health.org.uk/sites

41.

The TADS Team: (2007). The treatment for adolescent depression study (TADS). Arch Gen Psychiatry, 64:1132–1144, 2007.

42.

The TORDIA TEAM: (2010). Emslie GJ, Mayes T, Porta G, Vitiello B, Clark, G, … Brent D: Treatment resistant depression in adolescents (TORDIA): Week 24 outcomes. Am J Psychiatry, 16:782–791, 2010.

43.

Townsend

, Floersch

, Findling

: Adolescent attitudes toward psychiatric medications. J Child Psychol Psychiatry, 50:1523–1531, 2009.

44.

Verbeek-Heida

, Mathot

: Better safe than sorry-why patients prefer to stop using selective serotonin reuptake inhibitors (SSRI) antidepressants but are afraid to do so: Results of a qualitative study. Chronic Illn, 2:133–142, 2006.

45.

Woldu

, Porta

, Goldstein

, Sakotsky

, Perel

, Emslie

, Mayes

, Clark

, Ryan

, et al.: Pharmacokinetically and clinician determined adherence to an antidepressant regime and clinical outcome in the Tordia Trial. J Am Acad Child Adolesc Psychiatry, 50:490–496, 2011.

46.

World Health Organization: Adherence to Long-Term Therapies: Evidence for Action. Geneva: World Health Organization, 2003.

47.

Yang

, MoonYoon

, Lee

, Joe

, Jung

, Kim

: Adherence with electronic monitoring and symptoms in children with attention deficit hyperactivity disorder. Psychiatry Investig, 9:263–268, 2012.

48.

Zygmunt

, Olfson

, Boyer

, Mechanic

: Interventions to improve medication adherence in schizophrenia. Am J Psychiatry, 159:1653–1664, 2002.