8
The Research Landscape for Primary Care and Children’s Behavioral Health
One of the panels looked at aspects of the research agenda—in particular, at interventions that can be used in primary care and at issues involved in parenting. As the panelists observed, meta-analyses can survey a wide range of studies and determine which programs exhibit the most evidence of effectiveness, and continued research could both improve existing programs and point the way toward more effective programs.
RESEARCH ON INTEGRATED SERVICES
As noted throughout the workshop, behavioral health problems among children and adolescents are common in primary care. These patients also are heavy users of primary care—for example, children who keep coming back with stomachaches and headaches.
Effective treatment for behavioral health problems is essential for reducing suffering and dysfunction, as well as premature death, said Joan Asarnow, professor of psychiatry and biobehavioral sciences at the UCLA David Geffen School of Medicine. Suicide, which is generally considered a complication of untreated, undertreated, or ineffectively treated behavioral health problems, is a leading cause of death for adolescents, the second-leading cause of death for adolescents and young adults, and the third-leading cause of death among 10- to 14-year-olds. Rates of self-inflicted injury are particularly high in girls, compared to boys, and hit a peak in adolescence. We need to intervene in childhood and adolescence. If we intervene in adulthood, we can help some people, but we are going to miss kids at an earlier stage in their lives when we might have been able to catch
them early and prevent later years of suffering and dysfunction.” Results of a systematic review and meta-analysis were presented.
In a systematic meta-analysis, Asarnow and her colleagues searched for randomized controlled trials in English, peer-reviewed journals published between January 1960 and June 2014 that compared integrated behavioral health services in primary care versus treatment as usual (Asarnow et al., 2015). Integrated care was defined as behavioral health care provided through primary care services. The sample encompassed largely adolescents and children, though some transitional ages up to age 21 were included. The search identified 31 studies with a total of more than 13,000 participants. Nineteen of the trials looked at mental health treatments, four looked at substance use treatments, nine considered preventive interventions, three looked at mental health prevention, and six covered substance use prevention.
This meta-analysis found that integrating behavioral health care with primary medical care makes a significant difference, though the effect is small overall and the trials exhibited significant heterogeneity. Treatment trials have a small to medium effect, while prevention trials have a weaker effect.
Even among the prevention trials, some had a significant effect. For example, the Pbert trial on smoking cessation, which used the “5A model”—ask, advise, assess, assist, and arrange, delivered by the primary care provider, followed by one visit and four telephone calls by peer counselors—had a strong effect (Pbert et al., 2008). Mental health trials had a significant effect, with a weaker effect for substance use trials.
With regard to the models used, collaborative care had a larger effect than studies using other models. Four of the five collaborative care trials had significant effects in their meta-analysis, and the one that did not had a very strong comparison group. Trials that enhanced the primary care provider as a resource were more effective than colocated care interventions, which resonated with Asarnow. The bottom line, said Asarnow, is “integrated primary medical and behavioral health care provides at least part of the solution for addressing the behavioral health needs of children.” The effects are small to medium, so there is room for improvement. “But the probability is 66 percent that a randomly selected kid would have a better outcome after receiving integrated care than a randomly selected kid after usual care. This is good news.”
The large variation in studies calls for looking at the most promising models for integration. As an example of a promising model, Asarnow cited the Youth Partners in Care program, in which children were screened for depression symptoms, with referrals to a care manager. The care manager contacted and briefed the primary care provider on the patient’s needs and how to approach the patient. Patients received a booklet called “Stress and
Your Mood,” which talked about stress, the kinds of problems related to stress, depression, and ways of helping with depression. The care manager helped the patients and families pick the kinds of treatments they wanted. The care manager then briefed the primary care provider so he or she could come up with a shared treatment plan and consider whether specialty mental health consultation was required, which was something that physicians often felt they needed. “We don’t want kids to die because somebody has missed something,” said Asarnow. “Working with depression, the thing you realize is that it is a potentially fatal disease.”
Other models include colocated care, technology-enhanced care, behavioral health consultation, and coordinated care. “We need to understand what models are best,” said Asarnow, while recognizing that effectiveness may vary by setting.
This and other studies have uncovered several challenges in the treatment of behavioral health issues in primary care, Asarnow reported. Primary care providers often feel ill prepared, requiring training, consultation, the use of resource materials, or other possible solutions. Resources for collaborative care are often inadequate, requiring better referral networks and information systems to support linkage. Finally, quality-of-care problems are common, particularly inadequate follow-up. Rigorous evaluation and a continuous quality improvement process are needed to improve care in practice settings, said Asarnow, and tracking outcomes is probably the most critical thing to do to make evidence-based decisions in clinical care.
Co-location improves access enormously, said Asarnow, partly because it lessens the stigma and burden of going to a separate location to see a mental health provider or care manager. Also, many health-related behaviors or disorders are episodic, which requires a good monitoring system to detect a disorder like depression or a suicidal episode.
Next steps include getting effective integrated care models into routine practice in real-world settings. Rigorous scientific evaluation can inform practice, with a continuous quality-improvement loop. Also, costs are critical, said Asarnow. “If we don’t understand the costs of our services and integrating care, we probably won’t have it implemented.”
RESEARCH ON PARENTAL INVOLVEMENT
“Involvement of parents is critical for children’s health, possibly even more so for behavioral health,” said John Landsverk, a research scientist at the Oregon Social Learning Center. Whether dealing with children or adolescents, parents are almost always involved. In particular, Landsverk works with very high-risk youth, which often means that issues with competencies in parenting are involved.
The elements of “what it takes” for effective parenting have been well
researched over the past 35 years, Landsverk said. Parenting skills have been measured and changed in multiple studies, although elements differ across contexts, such as child developmental level, poverty, settings, and the demands of specific stressful situations. But there are common features that produce positive outcomes and can be taught, including
- Nurturance and reinforcement
- Emotion regulation
- Supervision, control, and discipline
- Supporting behaviors that promote effective adaptation to developmentally relevant demands (both academic and social)
- Discouraging behaviors that hinder positive adaptation, such as aggression, self-harm, association with deviant peers, and drug use
The range of positive outcomes that effective parenting can have is impressive, Landsverk said, including
- Sustained attention, improved executive function, and regular sleep
- Increased language and higher vocabulary
- Social skills and school readiness
- Less externalizing behavior
- Safer home environments
- Less abuse and neglect
- Less involvement in juvenile justice
- Less incarceration and hospitalization
- Higher grade point average and better mathematics and reading achievement
- Reduced peer aggression and association with delinquent peers
- Fewer mental health symptoms
- Less drug and alcohol use
Structured reviews have been proposed as a method for assisting the translational process, moving from discovery and testing to dissemination and implementation (Glasgow et al., 2012). Questions to be asked in a structured review include What studies met the criteria for inclusion and exclusion, and what were their salient characteristics? How were the studies carried out? What challenges were encountered in the studies? Was technology used in the intervention and/or evaluation?
Searching through the literature from 1995 to 2014, Landsverk and his colleagues identified two categories of studies: one with a full integration of behavioral health screening and services on a primary care platform, and another characterized by referrals for behavioral health from primary care
settings. The study covered family medicine and adolescent medicine, not just pediatrics, from ages 0 to 18.
The review resulted in several interesting findings, Landsverk reported. Major evidence-based treatments are being tested in primary care settings, though often in an abbreviated form and adapted to a particular setting, but they are producing promising results. Few examples exist of models in full primary care settings that use screens, behavioral health treatments, and primary care personnel. Also, no evidence was seen of cost measurement in randomized controlled trials, and there was little focus on implementation other than feasibility.
Some multisite studies and cluster-randomized designs could have been used for greater implementation research done on top of effectiveness trials, though there was some focus on variation at the site level. Also, Landsverk pointed to the potential benefits of hybrid designs (with both effectiveness and implementation aims) and anticipatory implementation measurement in efficacy/effectiveness designs.
The number and variety of parenting programs constitute both good news and bad news, said Landsverk. Some sort of decision-support tool could help primary care practices and associations to choose among interventions, perhaps based on severity level. Many kinds of disorders will need to be referred out, especially for care. Also, as Asarnow also pointed out, costs need to be considered from the start; otherwise, some interventions will be difficult to use. Many parenting interventions are done in group settings, and technology could reduce the costs of such interventions enormously, Landsverk said.
Next steps in the structure review include adding implementation and dissemination to the search terms and looking at parent as well as child outcomes. Addition issues include cost measurement and considerations for both preimplementation and implementation studies, what kinds of implementation studies are feasible, and what partnership will be needed to carry out more informative studies.
Asarnow, J. R., M. Rozenman, J. Wiblin, and L. Zeltzer. 2015. Integrated Medical-Behavioral Care vs Usual Primary Care for Child and Adolescent Behavioral Health: A meta-analysis. JAMA Pediatrics 169(10):929-937.
Glasgow, R. E., C. Vinson, D. Chambers, M. J. Khoury, R. M. Kaplan, and C. Hunter. 2012. National Institutes of Health approaches to dissemination and implementation science: Current and future directions. American Journal of Public Health 102(7):1274-1281.
Pbert, L., A. J. Flint, K. E. Fletcher, M. H. Young, S. Druker, and J. R. DiFranza. 2008. Effect of a pediatric practice-based smoking prevention and cessation intervention for adolescents: A randomized, controlled trial. Pediatrics 121(4):738-747.
This page intentionally left blank.