Double disadvantage: Biological and social consequences of ADHD and educational outcomes

The research literature on the consequences of ADHD for children’s lives indicates two mutually reinforcing pathways through which the disorder affects educational outcomes: direct biological effects on cognitive development and indirect effects on children’s home environment that likely affect learning.

Neurological research has shown that children with ADHD experience delayed maturation of the brain—more specifically delayed growth of the pre-frontal cortex, i.e. the part of the brain associated with the regulation of executive functions, such as complex cognition, decision-making processes, and the moderation of social behavior. In their seminal longitudinal study of 223 children diagnosed with ADHD and 223 typically developing controls, Shaw et al. [10] show substantial delay in the maturation of the pre-frontal cortex among children with ADHD. Among children with ADHD, 50 percent of cortical points had reached peak thickness at age 10.5 years, three years later than among the typically developed controls. More recent work by Sripada, Kessler and Angstadt [11] further renders probable that children with ADHD also experience delayed maturation of deeper parts of the brain’s architecture and pathways, which are associated with attention, control of impulsivity, disregard of irrelevant stimuli, and other cognitive tasks. In total, there is strong evidence that delayed neurobiological development of areas of the brain that moderate behavior is important for conductive learning-behavior among children with ADHD. The neurobiological findings dovetail with studies of educational performance among children with ADHD, which find that both inattention and externalizing behavior predicts worse educational outcomes [5,7,28–30].

Yet, delayed cognitive maturation may not be the only cause driving the relationship between ADHD and low educational achievement. Neither learning nor ADHD occur in a social vacuum. Beyond the direct neurobiological effects, ADHD also places strain on children’s life circumstances outside the educational system in ways that could deter learning. Untreated ADHD increases the risk of experiencing family instability, such as parental divorce [13–14] and, more dramatically, the risk of entering an out-of-home placement [15]. The association between family instability and lower educational achievement is well-established, although recent work has called (the generalizability of) causal claims into question [31–33]. Yet, for children with ADHD, family stability could likely worsen an already vulnerable learning situation. Thus, the effect of ADHD on educational achievement is likely not only driven directly by delayed maturation of the brain, but also by secondary effects of ADHD on children’s life and learning circumstances. Because research has shown that medical treatment of ADHD both mitigates the direct behavioral symptoms and lowers the risk of family instability and engaging in risky behavior, treatment could also have substantial impact on educational achievement.

Medical treatment of ADHD and educational outcomes

The strong association between ADHD and the above-mentioned host of negative outcomes has given rise to a number of treatment types, with pharmacological treatment with stimulants being the most common form [20,34]. Overall the short term efficacy of medical treatment on core ADHD-symptoms is well documented for the age group 6–18 years by over 200 clinical RCT’s since the 1960s [20]. Short-term medical treatment effects on academic achievement outcomes have also been studied extensively and show positive effects [35], whereas long-term studies of the effects on education are scarce and effects are generally more diverse, smaller, or less significant than short-term effects [36]. In their recent review, Baweja et al. [3] conclude that the evidence for a positive effect of medical ADHD treatment on school performance remains more substantial for acute than long-term indicators of academic performance. Baweja et al. [3] also note that evidence is strongest for the effect of stimulants to improve teacher ratings of behavioral outcomes. Methodological problems, such as sample attrition, are a probable explanation for the lack of empirical support for long-term effects, wherefore the want of research on this matter is substantial [35]. Nevertheless, as outlined above, the substantial number of both clinical and non-clinical findings on the impact of medical treatment of ADHD on domains of children’s lives connected with education do suggest that medical treatment could increase educational achievement among affected children.

Data sources

We use data from Danish administrative registers collected by Statistics Denmark. The primary data sources are the Danish Registry of Medicinal Product Statistics (RMPS) and the national registers on educational attainment (REA). Danish registers offer extensive individual-level information on all educational courses commenced in the public educational system and various academic performance measures. Register data has almost no sample drop out or attrition and validity and coverage is very high; e.g. in 2008, 96 percent of the Danish population aged 15–69 have non-missing education information [37]. From 2002, REA contains teacher evaluation grades and exam grades of students, who graduate the compulsory level of schooling (ninth grade in Denmark). From these school-leaving grades, we create our outcome measures by calculating standardized overall teacher evaluation and exam GPA.

Data from RMPS contain detailed information on all legal purchases of prescription drugs in Denmark (date, type of drug, dosage, etc.). The prescription data are linked to the individual identification number of the patient through the Danish health insurance system [38]. By using register data we avoid bias from refusal of participation, which is often a problem for survey data studies of ADHD [39–40] as well as recall bias from self-reported information on medical treatment patterns and academic performance. The RMPS is found to have a high completeness and validity, which is mainly due to reimbursement-driven record-keeping with automated bar-code-based data [40–41].

Operationalization of treatment patterns

In order to determine the effect of medical treatment of ADHD, we use longitudinal data from RMPS and define three types of pharmacological treatment patterns of use in the analysis: Discontinued Pharmacological Treatment (DPT), Ambiguous Pharmacological Treatment (APT), and Continuous Pharmacological Treatment (CPT). The difference between these three types of treatment allows us to estimate the effect of treatment by comparing the outcome across the three different treatment patterns with the DPT acting as a control group and the CPT as a treatment group and the APT as a partial compliance group. The definition of discontinuation (DPT) follows strict restriction rules: we define DPT as having purchased medication for maximum three months within the data window, which just allows for initial medicine trial and dose titration. Likewise, the definition of continuous treatment (CPT) is restricted to those, who have a regular and stable use of medication with purchases being no more than three months apart. The patterns of the remaining group, the APTs, are then per definition ambiguous as they cannot be convincingly placed in either the DPT or CPT category. These are children, who repeatedly enter in and out of medical treatment. All three categories are mutually exclusive. In all analyses, CPT is reference group, whereby we investigate inverse treatment effects of DPT and APT (discontinuation and ambiguous treatment) compared to CPT (continuous treatment).

Sample criteria

We sample all children diagnosed with and treated for ADHD from the total population of children, who completed Danish compulsory schooling from 2002 to 2011. To identify medically treated children (irrespective of their subsequent treatment pattern), we select children, who redeem at least one prescription for any type of ADHD medication (methylphenidate, atomoxetine, or modafinil) between age twelve and prior to school-leaving exams (treatment sample). Note that amphetamine salts were not introduced in Denmark for treatment of ADHD until 2013 due to official national restrictions and that these types of drugs are therefore not included in our sample criteria. We exclude children, who begin treatment before age twelve because scholastic impairment is likely less traceable among children at this age and because they more likely suffer from hyperactivity than attention problems (the latter of which especially predict lower academic achievement). All analyses were also conducted for the total sample yielding very similar results. We define an additional sample of children, who start treatment after school-leaving exams and use this as a placebo sample (we elaborate on this in section 4 on methods).

We restrict both samples further using the following criteria. First, children who either discontinue medical treatment within the first three months of the RMPS data window or begin treatment during the last three months of the data window are excluded from analyses, as the treatment patterns of these children cannot be convincingly determined. Second, the sample is restricted to cases with valid information on main outcome variables; teacher evaluation and exam GPA. In our data, 92 percent of all children have valid information for GPA, compared to 70 percent of children diagnosed with and treated for ADHD. For students in public schools, the school-leaving examinations are obligatory, but exemptions can be applied for (reasons for exemption could be severe functional impairments) [42]. Exams are not obligatory in private schools, though a school-leaving diploma is required for almost all upper secondary educational tracks.

Third, we exclude children, who commence treatment around the time point of exams to minimize the risk that we include children, whose (expected) poor exam performance causes them to commence treatment to boost future academic performance [43]. These restrictions result in a treatment sample consisting of 2659 individuals from the birth cohorts 1984–1996, and, correspondingly, a placebo sample of 3785 individuals from the birth cohorts 1983–1994.

Variables

To account for any effect of the duration of medical ADHD treatment, severity of symptoms, or potential correlations between age and treatment response [36], we control for individual age at treatment start. Age at graduation is included to control for grade retention. We further control for the medical dosage using the number of Defined Daily Doses (DDD) [44] for the individuals’ first and third purchase of ADHD medication. The severity of ADHD symptoms (which is unobserved) may both influence dosage and whether an individual child continues treatment [24–25]. Controlling for dosage titration may break the correlation between severity of the condition and treatment continuation. Hence, if controlling for medication dosage alters the estimated effect of treatment continuation this is evidence that severity and continuation are correlated. To control for comorbidity issues [45], we also include information on use of psychoactive medication for conditions other than ADHD. Data on any comorbid diagnoses are not, however, available.

To account for the event that treatment patterns (DPT and APT versus CPT) vary across child and family characteristics, we include a rich set of control variables. Individual level information includes gender, immigrant status, birth weight, and gestational age. Studies show that birth characteristics predict ADHD symptoms [46–47] and correlate with academic performance measures such as GPA [48]. We also include a dummy indicator for whether or not the child was graded using a new Danish grading scale introduced in 2008 [49] to account for any impact of this change on our results. (Grades from before 2008 are converted to the grade scale for 2008 using the official grade conversion table. The dummy variable thus captures any potential differences in grading across 2007/2008 that are not fully accounted for by the conversion of grades. The estimate of the dummy variable is insignificant in all analyses).

We measure parental characteristics one year prior to the birth of the child to avoid post treatment confounding. We control for parental income (including dummies for negative income generated by negative income from self-employment), unemployment, educational level (ISCED code dummies), parent age at the birth of their child, and whether either parent has an ADHD diagnosis.

Table 1 presents the descriptive statistics of all background variables. We show data for the population of the full population sample of 2002–2011 graduates and for our restricted treatment and placebo sample. Of the most notable differences, the proportion of males and non-immigrant Danes are higher in both the treatment and placebo sample compared to the population. The gender difference is most pronounced in the treatment sample, however. The overrepresentation of males and ethnic Danes is consistent with clinical accounts of diagnostic patterns for ADHD. Parents of children in the ADHD samples are slightly younger, more likely to have an ADHD diagnosis, have higher unemployment rates, lower income, and a greater proportion of parents have lower educational attainment levels. For several characteristics, there are notable differences between the treatment and placebo sample. However, as we show in the results section, these differences do not seem to alter how medical treatment affects the outcome. We offer explanations as to why below.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Descriptive statistics for all background variables.

Population sample of compulsory school graduates from years 2002–2011 and the ADHD sample.

https://doi.org/10.1371/journal.pone.0207905.t001

Table 2 presents descriptive statistics on outcome measures. Teacher evaluation GPA reflects students’ academic performance in class as evaluated by the teacher, whereas exam GPA reflects assessments made by the students’ teacher as well as an external examiner during the school-leaving exams (oral and written). Whereas teacher evaluation GPA thus reflects students’ behavior as well as their academic performance, exam GPA can be interpreted as an estimate of student academic performance. Because exams include an external assessment as well as written exams, student behavior that is unrelated to their actual achievement less likely influences these assessments. Grades are assessed using the Danish 7 point grading scale, which corresponds to the European Credit Transfer and Accumulation System (ECTS)-scale [49]. All empirical models use GPAs standardized within the population sample. Both the main ADHD sample and the placebo sample have substantially and significantly lower GPAs than found for the full population.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Descriptive statistics for unstandardized and standardized dependent variables.

Population sample of compulsory school graduates from years 2002–2011 and the ADHD sample.

https://doi.org/10.1371/journal.pone.0207905.t002

Analytical setup

To measure the effect of medication on students’ school-leaving GPA we estimate the following regression: (1) where GPA_i is student GPA, X_i is a vector of individual characteristics, β is a vector of corresponding regression coefficients, DPT and APT are binary indicators of discontinued treatment and ambiguous treatment. e_i is the error term. The regression coefficient γ₁ measures the average difference in outcomes between those continuing medication (CPT) and those discontinuing medication (DPT) from the onset of diagnosis and until completing compulsory school, where GPA is measured.

The interpretation of γ₂ is less clear-cut, but captures the average difference in outcome between a continued treatment pattern and the average ambiguous treatment pattern (APT). The effect thus both depends on the effectiveness of the medication (generating a difference between those with an ambiguous and a continuous medical consumption) and it depends on the distribution of the ambiguous treatment patterns in the data. If APT patterns on average are close to full compliance, children in this group should have the same outcome as the CPTs. If, however, they, on average, are close to a full discontinuation of treatment, children in this group should, on average, have the same outcome as the DPTs. At any rate, a negative estimate of γ₂ indicates that medication is effective.

Although random nonresponse is supported by empirical evidence, drop out from medical treatment in observed data may be nonrandom for a number of other reasons. Take-up rates of prescriptions have a social gradient due to economic constraints for poorer families to finance medication consistently, and at-risk families may be less inclined to follow medical advice or to comply with medical treatment despite negative side effects. Studies find that this applies for health care in general [50–51] and for mental health care and ADHD treatment in specific [52–53]. Such social gradient patterns may also help explain the existence of APT patterns in data. However, for differences in background characteristics to be important confounding factors, background characteristics should both correlate with the allocation into treatment patterns and simultaneously have a direct effect on the GPA outcomes [54].

To evaluate the extent that omitted variables bias our results we estimate Eq (1) both with and without observed covariates. If the differences between CPT, APT, and DPT are unaffected by including observed covariates, this indicates either that observed background characteristics are uncorrelated with our treatment indicators or with our outcome of interest. If so, this may also suggest that the observed differences in outcomes across CPT, APT, and DPT are unaffected by unobserved covariates as well, which in turn makes a causal interpretation of the treatment estimates more likely.

During the last 20 years, the number of medical products available has increased, while existing medical products have been enhanced [55–56]. Such developments may increase the probability that individuals continue medical treatment for ADHD once they start. However, the observed changing rates of DPT, APT, and CPT across time may also reflect a changing composition of the students in our data (students diagnosed with ADHD who commence treatment) across births cohorts. This could lead to selection bias. To control for this, we use a two-stage Heckman selection estimator to correct for sample selection into our data across birth cohorts [57]. As exclusion restrictions, we use children’s residential region at age three and their birth cohort in the selection equation only [58]. The propensity to prescribe medication differs across Danish regions [15,59], wherefore region of residence at a young age likely affects the propensity to enter medical ADHD treatment. As medical treatment has become increasingly more widespread, birth cohort may also affect the propensity to enter medical treatment. Birth cohort does not affect GPA because grades are by construction independent across cohorts in accordance with the grading legislation of the Danish compulsory schools [60]. We control for region of residence at graduation in the main equation to allow for the effect of unobservables across regions.

To further test whether we can view treatment patterns as an exogenous indicator of nonresponse to medical treatment, we use the availability of a placebo treatment group: students, who commence medical treatment after graduation (thus after receiving their grades) and for whom we should expect no treatment effect on their GPA (see Johansen et al. [61] for a similar design). Estimating Eq (1) on the placebo sample should yield no effect unless nonresponders (DTP) are selected on unobservables compared to responders (CPT) and ambiguous responders (APT). Children in the placebo and the treatment sample do on average differ in the timing of their treatment, and this may imply additional underlying differences in their background characteristics, as Table 1 shows. However, as we show in the result section the differences across DPT, APT and CPT are statistically insignificant indicating that estimated effects are unaffected by the inclusion of covariates in the treatment sample. Hence, it is unlikely that the differences in estimated effects with and without controls are important for comparing the differences between DPT, APT and CPT across the treatment and placebo sample. Moreover, from our data we find that students in the treatment sample and placebo sample are equally negatively affected by having an ADHD diagnosis in terms of their academic performance. This speaks in favor of comparability of the treatment and the placebo sample.

Limitations

This study finds that medicine exerts a profound influence on the educational outcomes of children diagnosed with ADHD. Methodologically, our analysis rests on the stable unit treatment value assumption, implying that potential individual outcomes are unrelated to the treatment status, outcomes or characteristics of other individuals. However, children with ADHD are part of social contexts—school classes, families, etc.—and the individual effects of treatment presented should ideally be construed in its context. Therefore, additional research should study the impact of ADHD treatment on peers and families. A few recent studies suggest that ADHD and medical treatment hereof affects academic outcomes of non-ADHD siblings [8] and classroom peers [62]. Further, our effects are found in a context where only severe or moderate cases of ADHD are recommended to initiate pharmacological treatment. Hence, further studies are needed to assess the effects of treatment on long-term academic outcomes in contexts where less severe cases are also treated.

Perspectives

More than four decades of sociological research has voiced concerns about medicalization, social control, and the social construction of problem behaviors [16–18]. Recent research shows that increasing medical treatment rates of ADHD in populations with high baseline-treatment levels likely has adverse consequences for children [63]. Yet, it remains a well-documented fact that children with behavioral and mental health problems, such as ADHD, have poorer later life outcomes than their peers. In addition, research from countries with low baseline-levels of ADHD treatment consistently shows that in this context, medical treatment lowers problem behavior [59]. Our study adds to this research and our results indicate that medical treatment can improve the educational trajectories of children diagnosed with moderate to severe symptoms ADHD. Furthermore, in our theory section we outline two distinct but interrelated pathways through which ADHD likely affects educational outcomes—a direct, neurobiological impact on cognition and attention, as well as an indirect pathway changing learning environments. Further knowledge on the importance of each of these pathways in shaping children’s educational outcome could provide valuable insight into how to best address the educational challenges faced by children with behavioral problems, and should be the subject of future research.