An official website of the United States government
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Kathleen C. Thomas, PhD, MPH, Gabriela L. Stein, PhD, Christianna S. Williams, PhD, MPH, Mónica Pérez Jolles, PhD, Betsy L. Sleath, PhD, Maria Martinez, PhD, San Juanita García, PhD, Linda E. Guzman, Charlotte E. Williams, MPH, and Joseph P. Morrissey, PhD.
Latina/o children with mental health needs are half as likely to use services as other children. Activation is a promising strategy to eliminate disparities. Evidence on parent activation in this underserved population is sparse.
This study examines the comparative effectiveness of MePrEPA (metas, preguntar, escuchar, preguntar para aclarar/goals, questioning, listening, questioning to clarify), an activation intervention for Latina/o parents raising children with mental health needs.
Secondary analyses explore the association of parent activation and child self-reported outcomes using the child Patient Reported Outcomes Measurement Information System (cPROMIS).
We conducted a randomized controlled trial (n = 172) in a Spanish-language mental health clinic to assess the effectiveness of MePrEPA compared with a parent support group.
Inclusion criteria were Latina/o parents raising a child with mental health needs receiving services and able to attend weekly sessions. After completion of the trial, we tested feasibility of a refined version of the intervention MePrEPAv2 in a nonrandomized sample (n = 114). The primary outcome was parent-reported activation; secondary outcomes were parent mental health, observed parent activation, and child visit attendance. We added 2 additional secondary outcomes midtrial. Education activation explores parents' activation regarding their child's education (n = 134). Quality of school involvement captures the parents' report of their rapport with the child's teacher (n = 134). Six cPROMIS measures captured child-reported anger, anxiety, depression, fatigue, pain interference, and peer relationships (n = 97). We tested effectiveness of MePrEPA with a difference-in-difference approach estimating linear mixed models.
We examined heterogeneity of treatment effect. We assessed concurrent and discriminant validity of cPROMIS scores with Pearson correlations and independent sample t tests. We used linear mixed models to assess the relationship between change in parent activation and cPROMIS scores over time.
Among consented randomized controlled trial participants, 3-month completion rates ranged from 82% to 87%. The impact of MePrEPA compared with the parent support group was significant for the primary outcome parent activation (β = 5.98; CI, 1.42-10.53) as well as secondary outcomes education activation (β = 7.98; CI, 3.01-12.94) and quality of school involvement (β = 1.83; CI, 0.14-3.52). We observed no statistically significant intervention effects for secondary outcomes parent stress (β = −.21; CI, −1.99 to 1.58) or depression (β = .25; CI, −1.32 to 1.82). Tests for heterogeneity of treatment effect on outcomes did not indicate significant differences by subgroups. Chi-square and F tests indicated no association between treatment arm and secondary outcomes observed parent activation (P = .76) or child attendance (P = .26), respectively. The cPROMIS measures show good concurrent and discriminant validity in a population of Latino children receiving mental health services.
Improved parent outcomes show trends toward association with improved child-reported cPROMIS outcomes on pain interference (β = −.203; CI, −.40 to −.01; β = −.493; CI, −.96 to −.02), anger (β = −.702; CI, −1.23 to −.17), and peer relationships (β = .380; CI, −.01 to .77). However, after adjustment for multiple comparisons on cPROMIS outcomes, findings suggest only that increases in parent and education activation were associated with poorer cPROMIS outcomes on anxiety (β = .350; CI, .10-.60; β = .410; CI, .14-.68, respectively).
MePrEPA increased activation among Latina/o parents who bring children for mental health services. Moreover, MePrEPA is a psychoeducational intervention that mental health clinics can readily incorporate in their current practices. Findings support the use of cPROMIS scores to assess child outcomes, but no conclusions can be drawn about distal impacts of the MePrEPA intervention on the cPROMIS scores. Findings underscore the importance of parent partnerships in the clinical process and at school as well as managing child mental health needs at home.
These outcomes are based on parent self-reporting of perceived activation and may not reflect actual skill level. Future work should replicate these findings in a large number of sites, adding behavioral measures and distal impacts while examining MePrEPA's effects across settings, populations, and time.
Latina/o children are disproportionately affected by poverty and other factors associated with increased risk of psychiatric disorder.1 However, Latina/o children with mental health needs (defined variously as perceived need, screened disorders, and diagnosed disorders) are half as likely to use services as children in white non-Latino families.2,3 Latina/o families are more likely to report problems with getting services, lack of a usual source of care and a medical home, and dissatisfaction with the care they receive4,5. Some evidence shows that Latina/o patients experience poorer-quality patient–provider communication6 and have lower patient activation scores than the general population.7 Unmet mental health needs, in turn, are associated with poor social and economic outcomes over the lifespan.1 Latinos are the largest and fastest-growing minority population in the United States.8,9 Developing interventions to overcome these disparities is a major national health priority.10
Activation is a construct developed by Hibbard and colleagues11 that reflects self-efficacy in chronic disease self-management, knowing when and where to go for help, and getting needs met in a health care visit. Patient activation is associated with improved health management practices and outcomes in the short and long run,12-15 including mental health outcomes.16,17 Moreover, activation is independent of education and health literacy,18 making it a promising strategy to reduce health disparities. A small body of work provides evidence on factors associated with parent activation.19,20 Even though Latino culture values deference to authority and simpatia (harmonious interactions), which may hinder parents' efforts to be activated during a provider visit, there has been success fostering activation skills among Latina/o adults.7,21,22 Further, qualitative data indicate that Latina/o adults with higher levels of activation are more confident asking questions and feel that there is value in participating in the visit with their clinician and asking questions.7 Increased parent confidence managing child behavior problems and laying a foundation for success; self-efficacy identifying when and how to partner with the various professionals that work with the child in treatment, school, and other social environments; and confidence and skills to articulate concerns, treatment goals, and refinements to treatment strategies have the potential to improve timely service use, retention, and patient-centered care for Latino children with mental health needs. However, evidence on parent activation in this underserved population is sparse, as is the measured potential of parent activation to support improved child mental health outcomes for Latino children.
The present study addressed this knowledge gap by conducting a randomized controlled trial of MePrEPA, (metas, preguntar, escuchar, preguntar para aclarar/goals, questioning, listening, questioning to clarify), a psychoeducational intervention developed to teach activation skills to Latina/o parents raising children with mental health needs compared with a nondirected social support group. To expand patient-reported outcomes in this population, this study explored the acceptability of child-reported patient outcome measures using the NIH-supported Patient-Reported Outcomes Measurement Information System for children (cPROMIS) and the association of increases in parent activation on cPROMIS scores.
The cPROMIS measures have been assessed both for stability23 in the general population and for concurrent and discriminant validity in several clinical populations of children with chronic physical conditions,24 but the usefulness of cPROMIS for clinical populations of children and adults with mental health needs remains an unanswered question.25 Use of cPROMIS measures in a clinical population of children with mental health needs is promising for 3 reasons. First, children with mental health needs regularly report on their functioning and experiences as part of health services and clinical research.26,27 Moreover, evidence shows that it is feasible to collect patient-reported outcomes in clinical practices that serve adults with serious mental illness.28 Second, evidence shows that children and youth want and have the capacity to be involved in treatment decisions.27,29-32 Finally, there is consensus that, consistent with the mandate and mission of PCORI, shared decision making based on patient-centered outcomes is the foundation of patient-centered care.33
This study is grounded in the notion that parent activation is potentially an important enabler of children's service use, controlling for other predisposing factors (such as child age, gender, race), enabling conditions (such as income, health insurance coverage, language fluency), and health care need characteristics (such as condition severity, stability, comorbidities) of the child and family.
The figure below depicts the Andersen Behavioral Model of Health Care Use,34 a classic public health conceptual framework, in which predisposing, enabling, and need characteristics of parents and their child with mental health needs determine health practices and use of health services within the context of the health care system and external environment.
Andersen Behavioral Model of Health Care Use.
This work contributes to the literature by building on prior research to provide evidence about activation of parents on behalf of their children; by extending cPROMIS reporting in a low-income Latina/o population; by teaching activation in a group format, including school-based outcome measures; and by including parents and teens in a structured advisory role throughout the research process.35
In this study, stakeholders shaped the study's inception, conceptualization, and design, and participated in both structured and unstructured ways throughout its implementation. Our clinic partner, El Futuro, held a subcontract from PCORI; we met quarterly and at other times on an as-needed basis. Parents and teens were formed into regularly meeting advisory groups as well as ad hoc focus groups.
During proposal development, we spoke with Latino parents (n = 6) about their experiences seeking mental health care and other services for their children. The important themes of influences of language fluency and parent–provider trust emerged. We developed the study and intervention design with input from our clinic host. Dr. Smith and the staff at El Futuro emphasized the importance of engaging new clientele quickly to begin to address needs, and the significant loss due to clientele who drop out over time. We chose a group intervention format to align with the clinic's focus at that time; staff explained that the group could be billed as family therapy, and even if parents paid out of pocket, a group would be able to cover a counselor's time.
Two preliminary focus groups (n = 6) were held at El Futuro to develop new and nuanced hypotheses about facilitators and barriers to mental health services by Latino children and to improve the parent activation intervention. Sessions were held in Spanish and conducted by experienced facilitators. The 6 participants were mothers of children receiving mental health services at El Futuro identified by clinic staff as parents who would have insights into our questions. Participants included both new and existing clients.
A focus group protocol was developed, translated into Spanish, and institutional review board approved prior to the focus group sessions. We asked the mothers what factors supported their bringing their children to El Futuro and what made it difficult. We described our planned activation intervention and asked if we'd omitted anything important or if anything we'd planned did not seem important to them. A group of investigators reviewed an existing intervention from a prior activation project, and we adapted the intervention to incorporate the themes and perspectives noted in the focus groups.
Investigators coded the focus group transcripts for themes related to the research questions using an open-coding approach to identify both common themes and divergent perspectives. The group recommended shortening the prior activation intervention to 4 sessions, a time commitment deemed more feasible. The final activation intervention manual contained the following 4 sessions: (1) understanding my child's mental health needs, (2) working with mental health care providers, (3) working with the schools, and (4) parental stress and mental health. Based on parents' feedback, we built classes around a 60-minute structure, with flexibility to stay late and continue talking.
We held 2 focus groups (n = 6 and 4, respectively) of intervention group graduates to get input on MePrEPAv2 refinements in spring 2015. Parents liked the group format that provided an opportunity for sharing, role playing, and learning from one another. They spoke movingly about how the group facilitated their maturation as parents and explained that individual sessions would not provide this important element. They all wanted more time: more time to ask questions, more time for sharing, a way to make up missed sessions, and more time to cover parent stress management. Their examples of being activated focused on parenting and school interactions. As a result of both these focus groups and our own observation of the groups, we made the following changes to the intervention:
Early in the study, we established an ongoing mentor parent advisory group to facilitate a regular collaboration with participating parents. The goal of the Mentor Parent Group was to provide opportunities for parents to discuss research protocols, provide input into study implementation decisions, interpret findings, and suggest next steps. Soon after the Padres project received funding, the mentor parent group convened and met every 3 months to discuss various aspects of the research study. The first mentor parent who joined the group had participated months earlier in a parent focus group to inform the Padres study proposal. Thereafter, the study team facilitators and the project coordinator identified parent study participants from whom we had completed data collection, who showed leadership and meeting commitment, and who were interested in giving back to the community by providing feedback about their study group sessions and our research efforts. The goal was to include parents who had had the experience of participating in the study so that they could provide informed input on the experience of participating and interpreting data. The mentor parent group grew over time, with group size ranging from 1 to 7 parents, and with an average attendance of 4 parents per meeting.
A Teen advisory group was convened in year 3 of the project, when we added child-reported outcomes to the study design. The goal of the teen advisory group was to provide input on the process and outcomes of collecting the cPROMIS measures. For the teen advisory group, the project coordinator identified teens from whom we were collecting data, whose families were able to commit to coming to meetings, and who were comfortable discussing in a group setting. The teen group grew over time, with some fluctuation in membership and a core group who came to repeated meetings. The average attendance was 3 per meeting.
Both the mentor parent and teen groups were scheduled at the convenience of group members. Typically, groups were held midday on Saturdays. Families found it most convenient to schedule groups 1 week before convening: Group facilitators would phone members to determine a convenient day and time, and then call a second time to confirm the chosen time. A meal was provided at each group; childcare was provided at each parent meeting.
Participants were paid $20 in thanks for their time.
Qualitative data describe mentor parents' conceptualization of their parenting role, maturation in that role, the context for seeking help on behalf of their child, and implications for the relevance of selected variables and factors to be measured in the study and variables that were not documented in the literature. For the teen advisory group, qualitative data describe the acceptability of completing cPROMIS measures and the resulting reflections that this engendered. We use a descriptive case study approach to describe the structure, process, and outcomes of the Padres mentor parent and teen groups in the context of the study.
Parents suggested focusing on improving communication between parents and children. This content area was then incorporated in the first curriculum session of the intervention. Parents explained that having a group intervention was preferable because they valued the support of other parents who were experiencing similar situations. They also suggested that the intervention be delivered at the start of their child's treatment because it would help build a sense of support and lessen their feelings of isolation early on. We asked parents about the preferred number of sessions and they agreed that a short intervention was preferable. As a result, we implemented 4 group sessions early in the clinical process, when families had just finished their initial evaluation at the clinic or when they were placed on a waiting list. Impacts: We incorporated each of these recommendations in the study.
The parents provided important insights on language. They emphasized the importance of the language we use. The mentor parents stressed that the language that health care service providers, group facilitators, and researchers use with parents should be simple, easy to understand, and without medical terminology or jargon. Use of simple language avoids raising fears among caregivers and helps increase their trust, especially since involvement with the mental health care system may be new for many. One parent described an experience that a friend had with a mental health service provider: “and the doctor told [the parent] that the child had [name of the diagnosis] and she thought the child was going to die, and she was nervous the entire session.” Impacts: This made us aware that common ways of communicating diagnoses can be frightening to many parents, and it reminded us to be transparent by avoiding overly technical terms and jargon.
Once the study was implemented, parents expressed concern that participants felt they had to drop out if they missed a session. Impacts: The project coordinator began to emphasize to each parent at each contact that they were welcome to continue participating in the groups even if they missed a session or if their child was no longer receiving services at the partner clinic. Upon implementation of this strategy, recruitment and retention improved significantly.
Parents recommended that a measure of parent activation in a school setting be included, particularly regarding the parent–teacher relationship. Impacts: As a result, we included 2 additional measures early in the study—an exploratory measure of education activation and the Parent–Teacher Involvement Scale.36 Importantly, in interim reports to the funder, the education-focused parent activation outcome measures showed an impact of the intervention before the patient activation measure. The strong impacts on these measures underscore their relevance for parents.
During a mentor parent discussion of the importance of fairly comparing parents who are participating in the groups, participants stressed the need to account for differences in the level of confidence that parents have at the beginning of the project when comparing outcomes between the groups. Impacts: Participants' framework for understanding growth in activation over time underscored the importance of our randomized controlled trial with difference-in-difference analytical approach.
The teens said they liked data collection via the iPad; data collection takes only about 5 minutes. They said that answering the questions enhanced their own activation skills and described how the PROMIS measures led them to reflect on their life and plan what they wanted to discuss with their therapist.
It was important to mentor parents that study results be shared with others after they had an opportunity to help researchers understand those findings. Mentor parents felt that the importance of the findings and their generalizability motivated dissemination, particularly to other parents and the people working with their children. Mentor parents preferred activities that were more immediate and personal over peer reviewed publications. Foremost, parents stressed the need to pursue each area using clear and concrete language, to make sure other parents would feel at ease and engaged in learning about the study findings.
The mentor parents emphasized the importance of helping other parents and community members understand the information gathered by the researchers. Mentor parents were interested in sharing personal stories with other community members and organizations about why the study's parent activation intervention worked for them. Mentor parents were also interested in conveying their collaboration and its impacts on the dissemination efforts. Impacts: Researchers incorporated parents' testimonials in study reports, manuscripts, and a video37, underscoring the collaborative research process as a key dissemination strategy.
Mentor parents preferred in-person mediums of communication that used visual prompts. Nonetheless, mentor parents were still interested in providing recommendations to researchers in such publications, as they understood this to be an important channel of communication for researchers to share study results with others. Parents recommended the inclusion of their direct quotes in research manuscripts and the use of colloquial language to better convey the information to journal audiences. Mentor parents' understanding of the importance of peer reviewed publications for researchers as a way to disseminate findings was facilitated by the ongoing parent–researcher interactions and by researchers' education efforts. Impact: Researchers partnered with a videographer to create a short video describing the collaborative research process for the parents. The video, which is 4-1/2 minutes long, is available to the public and portrays mentor parents' testimonials. The mentor parents planned their testimonials at one of their meetings, decided who would be in the video, and selected the various scenes to include. Researchers also published a manuscript that described the parents' advisory role in earlier stages of the research process, and they followed parents' recommendations, such as including parent quotes in the manuscript.35
Regarding the context of dissemination, mentor parents wanted to share the intervention with community members in informal settings. They suggested sharing study findings with school personnel because children's problems at school motivated referrals to mental health services. Mentor parents were also very interested in presenting the study's video in the waiting room at the partner clinic as a way to showcase the leadership role patients and parents can take on to foster community mental health and well-being. Parents were also very interested in presenting study findings to the clinic's funders, to encourage charitable donations and thereby increase access to services for families. Impacts: Parents' interest in disseminating study findings to community contexts, such as peers and schools, supported the research team's efforts to seek funding to replicate the study in a school setting; consequently, a school pilot study is underway. Because they also supported seeking funding to replicate the study in an ethnically diverse hospital clinic, a new study is underway that will continue to rely on the mentor parent group for guidance. Finally, a mentor parent volunteered to speak at a community-based workgroup meeting to educate the group on Latino-based collaborative research. By linking the parents and the community workgroup, researchers have sought to empower parents to lead the dissemination of study findings in their communities.
This study is a randomized controlled trial of MePrEPA, a psychoeducational intervention developed to teach activation skills to Latina/o parents raising children with mental health needs compared with a nondirected social support group (Figure 2, ClinicalTrials.gov ID: NCT02329431 Appendix 1). We also explored 3 secondary analyses. Figure 1, the Padres project timeline, illustrates the sequence of project events. First, based on the advice of our mentor parent group, we added outcomes capturing education activation and the quality of school involvement to the protocol in April 2014 (Figure 3). We collected these outcomes from the subset of 134 trial participants who enrolled after this addition was made to the protocol. Second, in June 2015, in accordance with the original protocol, we refined the intervention based on parent and clinic feedback to create MePrEPAv2, which we implemented in a nonrandomized feasibility trial (n = 120, Figure 4). The goal of this stage was to test using the effectiveness of the refined intervention MePrEPAv2 compared with the historical peer support group. Third, in July 2015, upon receipt of supplementary funding, we began to collect child-reported outcomes (Figure 5). We collected these outcomes from the subset of 97 MePrEPAv2 families who enrolled after this addition was made to the trial. These analyses explored the acceptability of child-reported patient outcome measures using the NIH-supported cPROMIS and the association of increases in parent activation on cPROMIS scores.
CONSORT Diagram of Randomized Controlled Trial Participants.
Project Timeline of Events.
Education-Focused Parent Activation Outcomes Sample.
MePrEPAv2 Sample.
cPROMIS Sample.
The study took place in a Spanish-language mental health clinic in a medium-size urban city in North Carolina. Participants were Latina/o parents or primary caregivers (hereafter referred to as parents) raising a child with mental health needs who was receiving or newly accessing mental health services. The randomized controlled trial took place during the 22-month study window of November 2013 through August 2015. Inclusion criteria were having a child 22 years old or younger (to be inclusive of all children through high school), able to attend weekly sessions for 4 weeks, and able to give informed consent. Exclusion criteria were focal child not living with the potential participant and evidence of urgent parent mental health needs (eg, active suicidal ideation). One parent was enrolled per child. All participants self-identified as Latina/o as confirmed by clinic staff. In July 2015, we began to collect child-reported outcomes in study children aged 8 or older.
The project coordinator enrolled participants at the point of scheduling a therapy visit. Enrollment took place in person at the clinic, at the close of an initial diagnostic visit that confirmed that the child would be appropriately served by the clinic. Upon obtaining informed consent from the parent and assent from the child, a parent was randomized via computer algorithm to treatment or control using a block design stratified by child Medicaid coverage status. We anticipated that Medicaid would support expression of activation skills. The coordinator remained blind to the treatment group by giving the treatment arms color names that did not convey content; the structure of participant involvement was identical. All interactions took place in Spanish. We conducted a baseline interview shortly before the scheduled session start. We scheduled sessions at the convenience of families; most were held in the evening, and childcare was provided. We audio-recorded the first parent–provider visit conversation scheduled after parent completion of our group sessions. The coordinator collected interview data at baseline and at 1- and 3-month follow-ups, augmented with data from the child's chart. Interviews were scheduled at a time convenient to the parent, typically while the child was in a therapy session or directly before or after. Participants were paid $20 at each data collection point. We collected child-reported outcomes via iPad at each child visit to the clinic during the 3-month period that their parents were involved in study data collection. Child PROMIS data collection began in July 2015 and included all children whose parents were currently participating in the study regardless of what stage of the study the parents were in at that time. The project coordinator provided the child with an iPad survey while they were in the clinic waiting room.
Child participants were paid $20 at baseline. Study participants were enrolled through June 2016 and data collection was completed in July of that year, regardless of what stage of the study participants were in at that time. This ensured that all parent participants received a group intervention. We conducted the study in compliance with the Office of Human Research Ethics.
MePrEPA is a psycho-educational group intervention developed to teach activation skills to Latina/o parents raising children with mental health needs (Appendix 2).38 The intervention consists of four 60-minute facilitated sessions that address understanding and managing child mental health needs, working with health providers (part 1: partnering with providers; part 2: practicing activation skills), and working with the school system. Each session included direct instruction, discussion, and role play.
The comparison group was a nondirected social support group designed to address what we believed was the main threat to validity, namely, that the effects observed might be most parsimoniously attributed to peer learning and support and not to the MePrEPA educational content. So, for this group, we created a parent support group, with a facilitator and 4 sessions. Here, the facilitator only laid the ground rules for confidentiality and encouraged discussion but did not contribute to the content of the dialogue.
The groups were facilitated by doctoral clinical psychology graduate students who were bilingual and bicultural with experience working in community-based settings. We chose these criteria to reflect typical master's-level clinical staff qualifications at community-based mental health clinics. All sessions were audio-recorded. A PhD psychologist monitored the audiotapes for fidelity and met with facilitators to resolve questions and maintain consistency to the curriculum and study design as needed.
In the nonrandomized feasibility of MePrEPAv2, we rearranged the elements of the curriculum to minimize missed content, addressing (1) the child's mental health needs, (2) working with mental health care providers, (3) working with the schools, and (4) managing parent stress and mental health. In each session, we introduced and practiced MePrEPA activation skills. We offered the group 2 evenings a week on a rolling basis, so that participants could repeat the curriculum or make up any missed sessions. The goal of the refinements of MePrEPAv2 was to minimize scheduling burden and facilitate parent attendance.
The primary outcome examined for the trial was parent activation. Secondary outcomes examined were education activation, school involvement, parent stress and depression; observed parent activation; and child attendance. The Patient Activation Measure (PAM) captured parent activation on behalf of their child.39 The PAM is an adult self-report 13-item scale with 4-level Likert responses and scores ranging from 0 to 100. It is valid with excellent reliability.11,16,40 The PAM has been translated into Spanish and has been used successfully in Latina/o patient and general populations (mean, 40).7 The PAM has also been used to measure activation of parents on behalf of their children (mean, 70).41 A change of 4 points in the PAM is associated with improved health behaviors in the general population.14,42
Two secondary outcomes assessed the parent's report of activation regarding their child's education. Education activation was an exploratory measure derived from the PAM at the request of our parent advisory group35 to capture parent activation supporting the child's education. It was measured with a 10-item PAM-like measure that captured parent self-efficacy in managing their child's homework tasks, knowing where to go for help at school, and managing school conversations, scored on a scale from 0 to 100. The Quality subscale of the Parent-Teacher Involvement Questionnaire captured the quality of conversation and rapport between the parent and the child's teacher.43,44 This subscale is scored on a scale from 0 to 24 and has good validity and reliability in a sample of children with ADHD.45 We collected education-focused parent activation outcomes on a reduced sample because they were a post hoc addition recommended by our parent advisory group.
Two secondary measures captured parent mental health. We measured parent stress with the 17-item Parental Stress Scale.46,47 The Parental Stress Scale is scored on a scale from 0 to 75, has been translated into Spanish, and has been shown to have excellent validity and reliability (for women: mean, 22).46,47 We measured parent depression with the 8-item Patient Health Questionnaire (PHQ-8).48 The PHQ-8 is scored on a scale from 0 to 27 and has excellent validity and reliability.48-50 The parent PHQ-9 has been translated into Spanish and used successfully in Latina/o populations.51 A change of 5 points in the PHQ-8 is associated with a shift in level of depression.50
The parent activation (α = .892), education activation (α = .894), quality of parent–teacher relationship (α = .864), stress (α = .830), and depression (α = .895) measures each demonstrated good internal consistency reliability.
We collected qualitative data on parent–provider communication after completion of the 4- week MePrEPA and parent support groups, in an effort to capture observed activation, a secondary outcome, coding activation in 4 levels (low, medium low, medium high, high).
We collected child attendance, also a secondary outcome, at clinic visits during a 4-month window of time: during the 3-month period parents were participating in the study and 1 additional month following. We measured child clinic visit attendance by number of visits attended, rescheduled, and missed.
We also collected data at baseline on predisposing, enabling, and need characteristics of the child and family, consistent with the Andersen Behavioral Model.34 We collected time-varying measures at 1- and 3-month follow-up interviews as well. Published Spanish versions of the study measures were used whenever available; otherwise, items were translated and reviewed by 2 bilingual Latina/o team members.52
Children provided self-reports of the cPROMIS mental (anger, anxiety, depression), physical (fatigue, pain interference in accomplishing daily activities and resulting negative feelings), and social functioning (peer relationships) measures.23,24,53 We also collected 3 sets of parent-reported assessments of their child designed in the PROMIS system as parent proxy measures of these outcomes. We used the Child Behavior Checklist (CBCL) to capture parent assessment of internalizing and externalizing behavior problems.54 Internalizing problems have raw scores from 0 to 72; externalizing problems have raw scores from 0 to 48. Standardized T scores are used to identify children in the clinical, population 98th percentile range. Internal reliability and validity are good to excellent. The CBCL has been translated into Spanish and used successfully in Latina/o populations.55,56 We measured school suspension or expulsion from the National Longitudinal Transition Study-2.57 Parents also reported on parent proxy measures for the 6 cPROMIS items collected from children.53
Quantitative analyses used intent-to-treat principles; participants were included in the analyses when they completed the baseline interview.58 All analyses were prespecified unless stated otherwise. We conducted analyses SAS software version 9.4. We used 2-tailed Fisher exact tests and t tests with unequal variance to examine the balance between the MePrEPA and support groups at baseline on family and child characteristics in the Andersen Behavioral Model.34
We show unadjusted means of outcome measures at each data collection time point. We calculated internal consistency reliability of the outcome measures using Cronbach's alpha. We used a chi-square test to examine the association between the ordinal measure of observed activation and treatment arm. We used a 2-tailed Fisher exact test to examine the association between child visit attendance and treatment arm.
We tested our main hypothesis about the effectiveness of the MePrEPA intervention compared with the parent support group with a difference-in-difference approach estimating linear mixed models containing time (1 month and 3 months, reference baseline), randomization group (intervention, reference support group), and a time × group interaction, controlling for child Medicaid coverage status per our randomization design and child experience with therapy (novice, not)—a post hoc design decision due to group imbalance. The coefficient for the time × group interaction measured the change in outcome between baseline and 1 or 3 months for the intervention compared with the support group. We report detailed findings for the randomized controlled trial where group membership distinguishes the MePrEPA intervention and peer support groups. We also report findings to assess the impact of the nonrandomized MePrEPAv2 arm; in this model, group membership distinguishes the MePrEPA intervention, the MePrEPAv2 intervention, and the peer support group.
Ignorability of missingness in the data (ie, missing completely at random or missing at random) is examined by comparing the proportion of missing outcome measures by treatment arm and Medicaid status with 2-tailed Fisher exact tests and by comparing mean values of baseline outcome measures by presence or absence of outcome measures at baseline and 1- and 3-month follow-ups with t tests. We addressed nonignorable missing values for outcome measures via multiple imputation. The multiple imputation model used the following predictors: child covered by Medicaid, number of children in the household, parent age, parent education, family monthly income, parent work hours per week, acculturation, any family members in the household (beside child) with emotional or developmental challenges, any family members in the household (including child) with chronic health conditions, and child taking psychotherapeutic medication. Because our modeling approach accommodates unbalanced data, we used no techniques to handle ignorable missing data. We present the randomized controlled trial findings both without and with imputed data for transparency. Because a small number of children not covered by Medicaid might have private or other insurance, we conducted sensitivity analyses using child insurance coverage status (some/none) in place of Medicaid coverage.
To explore sample heterogeneity, we conducted separate analyses for 3 subgroups that test study hypotheses about differential effects of the intervention on patient activation. We hypothesized that the impact of the intervention would be greater for children covered by Medicaid because they had insurance to cover the cost of desired care, for parents who reported low activation at baseline, and for children who were novices to mental health treatment because these parents had more to learn about the system of care and thus more to gain. These subgroup models tested effectiveness using a difference-in-difference-in-difference approach, where a 3-way time × treatment group × subgroup level interaction tests whether the treatment effect (time × treatment group) differed between the 2 levels of each subgroup.
The parent–provider conversations were transcribed in Spanish. All transcriptions were reviewed by 2 bilingual, bicultural team members and discrepancies reconciled. Team members coded the level of parent activation evidenced in each transcript on a 4-level scale—high, medium high, medium low, and low—taking into account behaviors suggested by the PAM and MePrEPA, such as asking questions and speaking with the child's teacher. We calculated a kappa statistic for each level and overall to assess coder reliability.
For the cPROMIS analysis, we present mean cPROMIS scores at each point in time; we assessed change over time using paired t tests. We examined concurrent validity by estimating Pearson correlations of cPROMIS scores with parent proxy PROMIS scores and raw Child Behavior Checklist internalizing and externalizing scores. We examined discriminant validity by dividing children into 2 groups based on each of 3 measures: experience of suspension or expulsion and meeting clinical thresholds for internalizing and for externalizing behaviors as reported by the parents on the CBCL.59 For each time point, we assessed differences in means using independent sample t tests. Finally, we used linear mixed models to assess the relationship between change in parent activation and cPROMIS scores over time. Since we conducted analyses for each of 6 cPROMIS scores, we adjusted P values for multiple comparisons.60
Among consented randomized controlled trial participants (n = 181), 172 (95%) completed baseline interviews, 151 (83%) 1-month interviews, and 148 (82%) 3-month interviews (Figure 2). For the education-focused parent activation outcomes sample (n = 134), 89 (66%) completed baseline interviews, 97 (72%) 1-month interviews, and 116 (87%) 3-month interviews (Figure 5). In the MePrEPAv2 sample (n = 120), 95%, 73%, and 65% completed baseline, 1-, and 3-month follow-ups, respectively (Figure 4). In the cPROMIS sample (n = 97), completion rates were 89%, 81%, and 61%, respectively (Figure 5). Missing data were ignorable except for the reduced sample of education-focused parent activation outcomes: education activation and school involvement.
For those outcomes, the mean value of baseline patient activation was lower (68 vs 76; P < .001) when baseline education activation and parent involvement scores were missing. The intervention and support groups were balanced on all measures except child novice to therapy (Fisher exact test; P = .014; Table 1). Child age range was 3-9 years, and all of the children attended some kind of school setting. There were 21% and 30% of children who had 2 or more mental health diagnoses in the MePrEPA and support groups, respectively. The most common diagnoses were adjustment, mood, depressive, and attention deficit disorders. There were 22% and 32% of children taking psychotherapeutic medications. The most common medications were selective serotonin reuptake inhibitors and stimulants. Among consented participants of the nonrandomized feasibility trial of the refined intervention MePrEPAv2 (n = 120), 114 (95%) completed baseline interviews, 100 (88%) 1-month interviews, and 78 (68%) 3-month interviews. Nearly every participant in the randomized controlled trial (92%) attended a session; 43% attended all sessions; and the mean number attended per person was 2.9. Attendance at any session was lower (76%) in MePrEPAv2. The mean number of sessions attended was 2.6 overall and did not differ significantly between the trial MePrEPA, trial parent support, and nontrial refined MePrEPAv2 groups.
Baseline Sample Characteristics by Treatment Group.
Table 2 shows randomized trial unadjusted outcome measures by group, time, and single-point-in-time differences between groups. There were no differences in unadjusted outcome measures between groups at baseline. Positive unadjusted outcomes increased over time in both groups for every measure. At 1- and 3-month follow-ups, the unadjusted education-focused parent activation outcomes were higher for the intervention group (t = 2.10-3.28; df = 94.3-113.2; P < .05).
Self-reported Outcome Means at Baseline and 1-Month and 3-Month Follow-up.
Table 3 shows the difference in the change in outcome measures for the intervention group compared with the support group at follow-up times, adjusting for Medicaid and child novice to therapy status. The regression coefficients represent the adjusted mean difference-in-difference. For example, using the Table 2 unadjusted differences as estimates of adjusted differences, the difference in PAM score experienced by the intervention group between baseline and 1-month follow-up was 18 (= 88 − 70) and by the support group was 11 (= 84 − 73); the mean difference between these 2 differences was 7 (= 18 − 11), which is the reported coefficient in Table 3. Each coefficient represents the mean difference-in-difference in the respective scale (PAM, education activation, school involvement, parent stress, and parent depression).
Linear Mixed Models for Treatment Differences in Change in Outcome Measures.
The MePrEPA intervention group experienced significantly greater improvement from baseline to both 1- and 3-month follow-ups compared with the support group for parent activation (1-month β = 7.04; CI, 2.51-11.56; P = .003; 3-month β = 5.98; CI, 1.42-10.53; P = .011), education activation (1-month β = 9.88; CI, 4.82-14.93; P < .001; 3-month β = 7.98; CI, 3.01-12.94; P = .002), and school involvement (1-month β = 2.15; CI, .45-3.86; P = .014; 3-month β = 1.83; CI, .14-3.52; P = .035). We observed no statistically significant intervention effects in parent stress (β = −.21; CI, −1.99 to 1.58) or depression (β = .25; CI, −1.32 to 1.82) change scores between groups over time. Sensitivity analysis controlling for any insurance coverage (some/none) in place of Medicaid (covered by Medicaid yes/no) did not change findings. Figure 6 illustrates the impact of the intervention for parent activation: The slope of scores over time was greater for the intervention than for the support group. Table 4 shows MePrEPA impacts after we used multiple imputation to impute missing baseline values of education activation and parent involvement.
Adjusted Parent Activation Scores by Time and Treatment Group.
Linear Mixed Models for Treatment Differences in Change in Education Outcome Measures After Multiple Imputation for Baseline Value.
After imputation of missing values, education activation effects are only slightly smaller while parent involvement effects are slightly larger. These findings show the added value of the MePrEPA curriculum over and above a parent support group.
We also examined the comparative effectiveness of the MePrEPA intervention compared with the support group on single-time-point outcomes: observed activation and child clinic visit attendance. Table 5 shows observed activation and child clinic visits by treatment group. A chi-square test indicated no association between observed activation and treatment arm (P = .808). From our audio recordings, primary coding of observed levels of activation as well as actions tightly tied to our curriculum, such as asking questions, disagreeing, or mentioning child functioning in various realms, turned out to be difficult to code reliably (P > .1 for all items). A 2-tailed Fisher exact test indicated no association between child attendance and treatment arm (P = .257). Given these findings and the limited variation in these outcomes, we conducted no further modeling.
Observed Activation and Child Clinic Visits by Treatment Group.
None of the tests for heterogeneity of treatment effect on outcomes with a 3-way interaction of treatment × time × subgroup status indicated significant differences (Table 6).
Parent Activation Change Scores by Treatment Group and Subgroup.
Table 7 shows the difference in the change in outcome measures for the MePrEPA and MePrEPAv2 intervention groups compared with the support group at follow-up times, adjusting for Medicaid and child novice to therapy status. The MePrEPAv2 intervention group experienced significantly greater improvement from baseline to 1-month follow-ups compared with the support group, but the effect was not statistically significant at 3 months. As with the MePrEPA group, we observed no differences in parent stress and depression change scores between groups over time.
Linear Mixed Models for Treatment Differences in Change in Outcome Measures for MePrEPA and MePrEPAv2.
Children found it acceptable to complete the child Patient Reported Outcomes Measurement Information System (cPROMIS) measures at each clinic visit. Responses took less than 5 minutes, and there was minimal missing data. Baseline mean cPROMIS T scores (normalized at 50) ranged from 45 for peer relationships to 57 for depression (Table 8). Measures of change between baseline and 1- and 3-month follow-up are not significant. Over time, children's reports showed greater dispersion than parent reports.
PROMIS Child Measures, Means at Each Time Point and Change.
We explored concurrent validity of the cPROMIS scores relative to parent proxy measures in the Patient Reported Outcomes Measurement Information System (PROMIS) system. Findings indicate parent proxy scores for each child-reported measure—anger, anxiety, depression, fatigue, pain, and peer relationships—were significantly associated with the relevant child-reported PROMIS outcomes (P < .05; Table 9). Parent-reported Child Behavior Checklist internalizing scores were significantly associated with cPROMIS anger, depression, fatigue, and pain scores (P < .05; Table 10). Parent-reported Child Behavior Checklist externalizing scores were not associated with cPROMIS reports.
Descriptive Statistics and Child-Proxy Correlations for Paired Assessments (T-Scores), N 224 Pairs.
Cross-sectional Correlations Between Child PROMIS Measures and CBCL Raw Scores.
We explored discriminant validity of the cPROMIS scores relative to school suspension or expulsion and to the Child Behavior Checklist. Examination of discriminant validity indicates that suspension or expulsion was significantly related to child-reported anger, anxiety, and depression at 3 months, as well as the Child Behavior Checklist externalizing score at baseline and 1 month (P < .05; Table 11). Parent-reported Child Behavior Checklist internalizing and externalizing scores were not associated with cPROMIS scores after adjustment for multiple comparisons (Table 12).
Mean Child PROMIS Scores at Each Parent Time Point by Ever Suspended or Expelled.
Child PROMIS Measures According to Whether CBCL in Clinical Range.
We also explored the association between parent activation and improved child-reported cPROMIS outcomes. Table 13 shows trends that an increase in parent activation was associated with less pain interference with function and feelings (β = −.203; CI, −.40 to −.01), and an increase in parent school involvement was associated with less anger (β = −.702; CI, −1.23 to −.17) and pain interference (β = −.493; CI, −.96 to −.02) and better peer relationships (β = .380; CI, −.01 to .77). However, after adjusting for multiple comparisons, findings suggest that increases in parent and education activation were associated only with more anxiety (β = .350; CI, .10-.60; β = .410; CI, .14-.68).
Linear Mixed Models for Treatment Differences in Change in Outcome Measures.
Activation among Latina/o parents who bring children for mental health services can be improved with MePrEPA, a psychoeducational intervention that mental health clinics can readily incorporate in their current practices. Use of MePrEPA has the potential to increase patient-centered care among these families.
These findings are consistent with our hypothesis that MePrEPA enhances parent activation. The 33% gain in parent activation experienced by parents in the MePrEPA intervention is consistent with the activation literature but much larger than the 5% increase reported in other studies of Spanish-speaking patients7, the 13% increase among adults with mental illness,61 and the 6% to 7% increase associated with improved health behaviors.14,42 Parents' motivation to support their child's success may facilitate the uptake of these new skills. To our knowledge, the 28% gain in education-focused parent activation outcomes is a novel finding. The fact that the intervention has a positive impact on targeted activation outcomes, but not on secondary outcomes of stress and depression, suggests good specificity. To our knowledge, null findings on the impact of MePrEPA on child clinic visit attendance and observed parent activation are new and reflect the limited scope of evidence on parent activation to date. The PAM captures perception of self-efficacy in a broad set of activities. Child visit attendance or the parent–provider conversations coded in our observed parent activation measure may be too narrow to pick up meaningful differences. For example, our parent advisors have explained that they do not have control over work shift schedules, and unanticipated changes may interfere with scheduled child appointments. However, they feel this situation does not reflect on their perceived activation.
It is noteworthy that positive outcomes increased in the nondirected social support group, which is the control group in this trial. We compared the MePrEPA intervention against an active support group to address the principal rival hypothesis that parents build activation through peer learning and support. As a result, findings provide strong evidence that the measured effect is not due simply to participation in a support group.
The cPROMIS measures show good concurrent and discriminant validity with respect to the parent proxy measures and parent-reported internalizing behaviors in a population of Latino children receiving mental health services. These findings support the use of cPROMIS scores to assess distal impacts of our MePrEPA intervention. The mixed positive and null findings on the association of change in parent- and child-reported outcomes suggest a need for future research. On the one hand, trends toward improved outcomes on anger, peer relationships, and pain interference suggest the importance of establishing parent partnerships in the clinical process and at school as well as managing child mental health needs at home to support improvement in child mental health outcomes. On the other hand, we found increased parent activation and education activation were associated with increased anxiety. To the extent that parents are more actively involved in the child's school and other realms of functioning, it is understandable that this might lead to increased child anxiety in the short run.
We conducted this study in a Spanish-language culturally adept mental health clinic highly valued by Latino families.62 While results are promising, future research should demonstrate that the positive effects we obtained are reproducible in other ethnically diverse mental health settings. Since we cannot describe families who refused participation, the findings reported here reflect the impact of MePrEPA among families who volunteered to participate. Required participation could have a different outcome. On the other hand, advice from parents and teens contributed importantly to the project's recruitment and retention successes. In addition, their insights on measurement issues enriched the study data and improved researchers' ability to document valued study outcomes. Their thoughtful input broadens the generalizability of the study for Latino families. Moreover, our engagement findings can inform future research efforts that seek active Latina/o parent collaboration and the timely incorporation of parent and teen voices in each phase of the research process.
Findings did not indicate differing intervention effects by subgroups. In this study sample, different kinds of parents benefited from the intervention. Future work should explore intervention effects for a wider range of parents and children.
There are 3 limitations of this trial important to consider. First, we enrolled families whose children were receiving services in a mental health clinic. In this busy setting, if parents were not ready to talk with us, we waited to find a better time to talk. It is likely that we missed the opportunity to offer participation to some eligible individuals. We did not track the proportion of eligible participants who were unresolved soft refusals or those we missed. There may be a small group of families who would not choose to participate in an activation class. The findings reported here reflect the impact of MePrEPA among families who volunteered to participate.
Second, our outcomes are based on parent self-report and may not reflect actual skill level. Parents may overreport due to enthusiasm and social desirability, or underreport due to lack of familiarity with the concepts under study. With a randomized design, any bias associated with self-reports should be the same in both intervention and support groups and thus unlikely to account for the findings reported. Nonetheless, future research should develop behavioral measures of activation to determine whether self-reported measures are correlated with actual performance. Third, our time frame for follow-up was only 3 months from study initiation. Future research should examine benefits over time to assess the possibility of both decay and discovery of new benefits that were not expressed within this study's window of data collection.
There are both benefits and new questions to address from this trial. We were inclusive with our recruitment criteria rather than exclusive. Our goal was to assess impacts in a typical clinic population to increase generalizability. By doing this, we sacrificed control over a homogeneous sample, but we conducted exploratory analyses to understand which groups benefited the most from the intervention. These findings provide guidance on how to target the intervention to those who will benefit the most.
Future research should explore the role of child functioning at school and seek to identify the mechanisms of impact of the intervention. While our manual specified didactic topics at each session of the intervention, providing structured time for questions and role playing opened the door to a wide range of topics. Future research should explore if those families less experienced with therapy learned from their more experienced peers. Also, while our facilitators were PhD students, we believe this intervention could be facilitated by less-specialized individuals, such as those with a Bachelor's degree or lay health promotoras for whom there is sizable indication of success.63,64 We conducted our trial in a single setting held in high regard by the Latina/o community.62
Future research should explore effectiveness in a variety of settings. The 33% gain in parent activation experienced by parents was much larger than measured among other study populations.7,14,42,61 Future research should seek to elucidate facilitators and barriers to the uptake of activation skills among parents and those who are not. Our measure of education activation was exploratory. Future work should examine the validity of this measure.
Practices may find cPROMIS an efficient way to gather child-reported outcomes on a regular basis. In the comparison analyses, the lack of correlation between cPROMIS scores and parent-reported externalizing behaviors raises questions about children's and parents' impressions about motivations underlying behaviors. Future work should examine the impact of children's outcome reporting on their insights into the therapeutic process and associated child and parent outcomes.
Our teen advisory group has been explicit in the benefits they see from completing these self-report measures:
“A veces … hay preguntas que te hacen como reaccionar y decir ‘¡Necesito mejorar esto.’” / “There are also … some questions that make you reflect and say ‘I need to improve this.’”
Bivariate analyses suggest that child outcomes do not change significantly over time, and we found that increased parent activation was associated with increased anxiety. Future work should explore child outcomes over a longer time frame. Changes may increase anxiety in the short run but prove useful in the long run. Stratified analyses should also be explored to identify whether some children gain from their parents' learned activation skills.
Activation among Latina/o parents who bring children for mental health services can be improved with MePrEPA, a psychoeducational intervention that mental health clinics can readily incorporate into their current practices. Future work should develop behavioral measures of parent activation to determine actual performance effects. Distal impacts on child service use should also be explored.
Implementation and dissemination can be facilitated by considering curriculum refinements (content, methods of parent participation) and structural refinements (shortened, one-on-one vs group, video format). Future work should also examine replications in a large number of Latina/o clinic sites for more robust external validity tests, as well as adapt MePrEPA for non-Latino settings and determine whether effects generalize across settings and populations.
Research reported in this report was [partially] funded through a Patient-Centered Outcomes Research Institute® (PCORI®) Award AD-12-11-4900). Further information available at: https://www.pcori.org/research-results/2013/helping-latino-parents-learn-skills-manage-their-childrens-mental-health-care
Thomas KC, Stein GL, Williams CS, et al. (2018). Helping Latino Parents Learn Skills to Manage Their Children's Mental Health Care. Patient-Centered Outcomes Research Institute (PCORI). https://doi.org/10.25302/8.2018.AD.12114900
The [views, statements, opinions] presented in this report are solely the responsibility of the author(s) and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute® (PCORI®), its Board of Governors or Methodology Committee.
This book is distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License which permits noncommercial use and distribution provided the original author(s) and source are credited. (See https://creativecommons.org/licenses/by-nc-nd/4.0/
Your browsing activity is empty.
Activity recording is turned off.
See more...
