A Commercial Insurance Claims Analysis of Correlates of Behavioral Therapy Use Among Children With ADHD

Multiple variables were associated with increased behavioral therapy uptake, including age, geographic region, gender, prescriber type, severity of presentation, and psychiatric comorbidity. Children prescribed medication were 2.5 times less likely to use behavioral therapy than those not prescribed medication. Medication was associated with a fourfold reduction in future behavioral therapy uptake, even among preschoolers. This association held after the analysis adjusted for multiple covariates affecting behavioral therapy uptake. Results parallel what was observed in an RCT, in which despite robust medication effects, use of behavioral therapy before medication produced better outcomes at lower costs than the reverse sequence, partly because of poor uptake of behavioral therapy by participants who were initially prescribed medication (23, 24). Because over 6% of U.S. school-age youths are treated with CNS stimulants (1) and because medications are the most common initial ADHD treatment (30), the potential impact of these findings is appreciable. To the extent that combined treatments enhance long-term functioning over unimodal treatments (31, 32), the negative association of use of one treatment with use of the other could partly explain why medication alone produces limited long-term benefits even when adherence is strong (33, 34).

The rate of use of any behavioral therapy approached 50%, comparable to rates in past studies (3, 13, 35). However, it appears that the rate of receipt of potentially efficacious behavioral therapy was far lower than 50%. The number of therapy sessions attended affects outcomes (36, 37). In any 4-month span, barely 50% of youths attended at least four sessions, which is half the duration of the shortest evidence-based programs (38). Under 2% of therapy codes were for parent-only sessions, one of the most well-supported formats for ADHD treatment (6). Individual therapy accounted for most sessions, and many forms of this modality are not evidence based for ADHD (5, 10). Hence, both the dose and format of received therapy services may often be suboptimal.

Systems, parent, and provider factors have been identified as barriers (4, 22, 27, 39). Offering services in easily accessible settings and using engagement interventions to increase motivation for care increase uptake of therapy (40–43). In a prior RCT, engagement techniques were employed while free and accessible therapy was offered in the same setting as medication. Yet, use of medication was still significantly associated with lower behavioral therapy uptake (23). Therefore, a multimethod approach targeting patients, parents, and providers may be needed to meaningfully increase rates of use of behavioral therapy.

What drives this negative association of medication use with behavioral therapy uptake could not be ascertained in this analysis. One possibility is that medication alleviates any need for additional treatment. We adjusted for severity, and this relationship persisted, although our severity threshold was coarsely limited to use of emergency or inpatient services. Medication effects on use of behavioral therapy were also robust in a federally funded study in which only children manifesting objectively verified impairment after 8 weeks of systematic medication treatment were randomly assigned to receive additional treatment (23). Availability of therapy providers is limited, and they are often not located in the same setting as medication prescribers. However, low uptake occurs even when access is readily available, such as in integrated primary care practices (16, 41, 44). Combined treatments may be too burdensome for families, but previous studies have not found that use of behavioral therapy inhibits use of medication (23, 44, 45).

We theorize that the immediate and visible reduction in ADHD symptoms produced by medication (2, 46) reduces parental and possibly provider motivation to seek more intensive treatments because time commitments are a barrier to use of behavioral therapy (47). Symptom changes are easily observed, but acute symptomatic improvement and sustained functional improvements are poorly correlated because impairment often persists when symptoms remit (48–50). This disconnect makes it challenging to predict long-term outcomes based on initial response. Therefore, increased dissemination efforts focusing on ADHD as a chronic condition (5), emphasizing functional remission over acute symptomatic improvement, and using impairment versus symptoms as the metric for titrating treatment may improve behavioral therapy uptake.

Children under 6 (versus those 6–12), those with more severe clinical courses, and those with comorbid psychiatric disorders were more likely to receive behavioral therapy. These findings are consistent with prior work (3, 13, 22, 44) and treatment guidelines in which behavioral therapy is the recommended first-line treatment for young children (5, 46, 51, 52). For school-age youths, medication is recommended as a core component of ADHD treatment (5, 46). Combining behavioral therapy with medication can optimize long-term functioning at lower medication dosages; therefore, it would be reasonable to consider behavioral therapy as an adjunctive treatment for medicated youths of all ages (9, 11). However, the sequence of medication followed by behavioral therapy is less efficacious and cost-effective than the sequence that starts with behavioral therapy (23, 24). Yet medication first is the most common pattern according to these results and other naturalistic studies (30, 45). Therefore, it does not seem advisable to refer patients for behavioral therapy only when impairment persists after medication treatment. Delaying behavioral therapy until after medication use may be one reason why it has been challenging to find evidence of improved long-term outcomes for youths with ADHD despite large increases in health care spending for this disorder over the past 2 decades (34, 53).

The primary limitation of this study was the exclusion of children with public health insurance. Past work has found higher rates of use of behavioral therapy among youths covered by Medicaid (3, 13, 22, 35, 45, 54). It is possible that the impact of medication use on behavioral therapy use may be lower in this population. Therefore, replication of results in a nationwide pool of publicly insured youths is advised. However, even when publicly insured youths are included, rates of use of behavioral therapy remain well below rates of medication use regardless of patient age (4, 13, 25, 30). Because of limitations of the data set, we could not measure the impact of race or ethnicity, verify family relationships, or determine whether patients exiting and returning to insurance coverage were counted as the same or distinct cases. Our analyses would miss unbilled services, such as those provided through schools. However, past studies that included school-based supports found similar predictors of use of behavioral therapy (3, 13, 22). We were not able to verify the diagnosis of ADHD but we required that an ADHD diagnostic code be present for at least two separate visits. Unlike other behavioral health conditions, ADHD is not thought to be underreported (55). Rates of comorbid behavior disorders were low for an ADHD sample (56) but consistent with those in past studies of claims data (57), which was likely attributable to providers preferentially billing for ADHD because it is the only disruptive behavior disorder with FDA-approved medications.

In a sample of over 800,000 privately insured youths with ADHD, ADHD medication use was a robust predictor of behavioral therapy uptake. Children prescribed ADHD medication were 2.5 times less likely to have used behavioral therapy and were four times less likely to use behavioral therapy in the future, compared with youths not prescribed ADHD medication. Effects were strong for both younger and older youths. These results suggest that the common sequence of medication initiated in primary care followed by referral for behavioral therapy may not be the ideal pathway, because behavioral therapies were even less likely to be used once medication had been prescribed. Future work should examine in naturalistic settings whether use of behavioral therapy with or before medication treatment is associated with increased treatment use and subsequent improvement in long-term outcomes.

The authors acknowledge Sara Mills, M.S., for editorial assistance. They also acknowledge support from the Penn State Center for Applied Studies in Health Economics.