eTable 1. The Effect of Adjustments at the Within-Patient Comparison
eTable 2. The Effect of Adjustments at the Between-Patient Comparison
eTable 3. Associations Between ADHD Medication Use and SweSAT Scores in Between-Patient Comparison
eTable 4. Associations Between ADHD Medication Use and SweSAT Scores in Within-Patient Comparison, Stratified by Performance on the Initial Test
eTable 5. Pattern of Concomitant Use of SSRIs
Customize your JAMA Network experience by selecting one or more topics from the list below.
Lu Y, Sjölander A, Cederlöf M, et al. Association Between Medication Use and Performance on Higher Education Entrance Tests in Individuals With Attention-Deficit/Hyperactivity Disorder. JAMA Psychiatry. 2017;74(8):815–822. doi:10.1001/jamapsychiatry.2017.1472
Can attention-deficit/hyperactivity disorder medications improve performance on higher education entrance tests for individuals with attention-deficit/hyperactivity disorder?
In this within-patient cohort study of 930 individuals with attention-deficit/hyperactivity disorder, test scores were significantly higher during periods the individuals were taking medication compared with nonmedicated periods.
Treating patients with attention-deficit/hyperactivity disorder medication might help to improve their academic performance.
Individuals with attention-deficit/hyperactivity disorder (ADHD) are at greater risk for academic problems. Pharmacologic treatment is effective in reducing the core symptoms of ADHD, but it is unclear whether it helps to improve academic outcomes.
To investigate the association between the use of ADHD medication and performance on higher education entrance tests in individuals with ADHD.
Design, Setting, and Participants
This cohort study observed 61 640 individuals with a diagnosis of ADHD from January 1, 2006, to December 31, 2013. Records of their pharmacologic treatment were extracted from Swedish national registers along with data from the Swedish Scholastic Aptitude Test. Using a within-patient design, test scores when patients were taking medication for ADHD were compared with scores when they were not taking such medication. Data analysis was performed from November 24, 2015, to November 4, 2016.
Periods with and without ADHD medication use.
Main Outcomes and Measures
Scores from the higher education entrance examination (score range, 1-200 points).
Among 930 individuals (493 males and 437 females; mean [SD] age, 22.2 [3.2] years) who had taken multiple entrance tests (n = 2524) and used ADHD medications intermittently, the test scores were a mean of 4.80 points higher (95% CI, 2.26-7.34; P < .001) during periods they were taking medication vs nonmedicated periods, after adjusting for age and practice effects. Similar associations between ADHD medication use and test scores were detected in sensitivity analyses.
Conclusions and Relevance
Individuals with ADHD had higher scores on the higher education entrance tests during periods they were taking ADHD medication vs nonmedicated periods. These findings suggest that ADHD medications may help ameliorate educationally relevant outcomes in individuals with ADHD.
Attention-deficit/hyperactivity disorder (ADHD) is a common psychiatric disorder among children and adolescents that can persist to adulthood.1 It affects approximately 5% to 7% of the school-aged population2-4 and slightly less than 3% of adults.5-7 On average, individuals with ADHD earn lower school grades or standardized test scores and receive less schooling compared with peers without ADHD.8-10 The core symptoms of ADHD, including inattention, impulsivity, and hyperactivity, may affect school performance; associated deficits, such as in attention span and working memory, may exacerbate academic difficulties.11,12
Clinical trials have shown that ADHD medications are efficacious at reducing the core symptoms of ADHD13 and are generally tolerated in children, adolescents, and adults,5,14 although recent Cochrane systematic reviews graded this evidence as low quality.15,16 It is, however, less clear whether such improvement in behavior translates into better academic outcomes. Results from previous studies are mixed; more important, it is difficult to evaluate the educational significance of the measured outcomes.17-20
Using information from the Swedish national registers, we examined the link between the use of ADHD medication and a nationally valued academic outcome in people who have received a diagnosis of ADHD. We chose a within-patient design, in which the same individual’s test scores were examined when he or she was taking vs not taking medication.
We used the National Patient Register21 to identify 61 640 individuals with ADHD (code 314 in the International Classification of Diseases, Ninth Revision, and F90 in the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision) born between 1976 and 1996. We observed these individuals from January 1, 2006, to December 31, 2013. The Migration and Cause of Death Registers were linked, using the unique personal identification number, to refine individual follow-up periods. The Integrated Database for Labor Market Research was linked to obtain information on parents’ highest educational level, and the Swedish Military Service Conscription Register was linked to obtain information on the individual’s IQ. The study was approved by the regional ethics review board in Stockholm, Sweden. By Swedish law, there is no requirement for informed consent in register-based research.
The outcome was the score on the Swedish Scholastic Aptitude Test (SweSAT; henceforth referred to as the test score). This is an optional standardized test22 that has been an instrument, together with the grade point average from upper secondary schools in Sweden, for higher education selection in Sweden. Applicants to institutions of higher education are ranked by grade point average and/or the test score. There is no restriction on the number of times that one may retake the SweSAT; universities automatically consider an applicant’s best test score. Therefore, it is common for applicants to take the test multiple times. The SweSAT is usually administered twice a year during April and October. Because the level of difficulty varied between tests, raw test scores were normed so that scores from different test occasions were comparable.23 Since 2011, the test scores are normed to a scale between 0.00 and 2.00 (in increments of 0.05). We analyzed the normed test scores on its original scale multiplied by 100, yielding test scores ranging between 0 and 200.
A total of 16 test occasions were included during the follow-up period. We extracted test records for the identified individuals with ADHD and excluded records of incomplete tests from the analyses. Age at testing was restricted to ages between 17 (because students start taking the tests during the second year of upper secondary school) and 30 years. For comparison, we extracted the test scores for population controls who were not diagnosed with ADHD and were matched with each case on sex, birth year, and residential area at the time of the first diagnosis at a ratio of 10:1.
The exposure was ADHD medication. We extracted medication records of stimulants (including methylphenidate hydrochloride, amphetamine sulfate, and dextroamphetamine sulfate) and nonstimulants (atomoxetine hydrochloride) from the Prescribed Drug Register. In line with previous definitions,24-26 we defined the medicated periods as having 2 dispensing records fewer than 6 months (183 days) apart. The nonmedicated periods are thus any period other than those defined as medicated periods during the follow-up. We assigned the test dates to the predefined medicated and nonmedicated periods to determine the medication status at the testing.
In agreement with previous studies,27,28 we observed significant linear and quadratic effects of age and practice (as measured by the number of previous tests taken) in association with the test scores, meaning that test scores improve at a declining rate as people get older and as they take more tests. This observation forms the basis of the adjustment in the main analyses using a within-patient design (the effects of covariates are shown in eTable 1 in the Supplement). Other covariates, including sex, test year, IQ (measured in stanine [standard nine] scores), and parents’ highest educational level, were considered at the cohort level (the estimates are shown in eTable 2 in the Supplement).
Data analysis was performed from November 24, 2015, to November 4, 2016. We first compared the basic characteristics between individuals with ADHD who had taken the SweSAT and their matched population controls, as well as between individuals with ADHD who took medication and did not take medication during the study follow-up period.
In the main analyses using the within-patient design, the eligible study participants were individuals with ADHD who had taken repeated tests and used ADHD medication during the follow-up period. We tested the association between medication use and the test scores using a conditional generalized estimating equation, with each patient as a separate cluster.29,30 The analyses were performed with and without the adjustment of time-varying confounders, including age and the number of previous tests taken. Because each individual serves as his or her own control in this design, all the time-invariant confounders were implicitly adjusted for. We also tested the presence of carryover effects to ensure that the estimates were not biased (eAppendix in the Supplement).
For comparison, we also conducted analyses at the cohort level (ie, without controlling for confounding by indication).31,32 In these analyses, the test scores from all patients during medicated periods were compared with those during nonmedicated periods (referred to as “between-patient comparison”; in the eAppendix in the Supplement). All analyses were performed in R, version 3.2.3 (The R Project for Statistical Computing), using the R package drgee for the generalized estimating equation and conditional generalized estimating equation models.33P < .05 (2-sided) was considered significant.
To investigate whether the results were sensitive to how the medicated and nonmedicated periods had been defined, we also assigned the medication status for each test date according to the number of days since the last dispensing. We reasoned that an individual is more likely to be taking medication if the test date is closer to the last dispensing, and conversely, less likely to be taking medication if there is a large gap in between the last dispensing and the test. The following cutoff values were tested: 6 months (183 days) to define both medicated and nonmedicated status, 3 months (91 days) to define both medicated and nonmedicated status, medicated status if the test date was no more than 3 months since the last dispensing, and unmedicated status if the last dispensing and the test date were more than 6 months apart (those with the gap between 3 and 6 months were set as missing).
Stimulant ADHD medication might have different efficacy from nonstimulant medication.34 To test whether different types of drugs might influence the test scores differently, we identified 2 groups: those who used only stimulants and those who used nonstimulants or a combination of medications. The association of medication use and the test scores was assessed within these 2 groups.
Attention-deficit/hyperactivity disorder frequently co-occurs with learning disabilities (LDs), but it has been shown that attention problems in patients with ADHD are not limited to its association with LDs.35 However, it is not clear whether the association between ADHD medication and academic outcomes might differ in patients with or without LDs. To answer this question, we performed a sensitivity analysis restricting our sample to the subset of individuals with ADHD but without coexisting LDs. We identified people with LDs either from the National Patient Register (code 315 in International Statistical Classification of Diseases, Ninth Revision, and code F81 in International Statistical Classification of Diseases and Related Health Problems, Tenth Revision) or from the test records (there is a special test category for individuals with dyslexia).
To evaluate whether the association with test scores was specific to ADHD medications, we examined the effect of selective serotonin reuptake inhibitors (SSRIs) on the test scores within these individuals with ADHD, both before and after the adjustment for ADHD medication use. The dispensing records of SSRIs (including fluoxetine hydrochloride, citalopram hydrobromide, paroxetine hydrochloride, sertraline hydrochloride, fluvoxamine maleate, and escitalopram oxalate) were extracted from the Prescribed Drug Register, and the medicated and nonmedicated periods of SSRIs were defined in the same manner as with ADHD medications.
A total of 6225 of all 61 640 individuals with an ADHD diagnosis (10.1%) had taken the test (3718 [6.0%] of all individuals with ADHD had taken the test during study follow-up). The corresponding number in the matched population controls was 196 135 of 616 400 (31.8%) (115 262 [18.7%] during follow-up).
Individuals with ADHD who had taken the test took significantly fewer tests compared with the matched population controls (mean [SD], 1.58 [1.09] vs 1.74 [1.20] ; P < .001), and they took the test at a significantly later age than did the matched population controls (mean [SD], 22.10 [3.30] vs 20.55 [2.44] years; P < .001) (Table 1). These individuals, however, did not differ from matched controls in mean (SD) test scores (91.40 [46.16] vs 90.51 [43.02]; P = .20) despite scoring slightly lower in mean (SD) IQ stanine scores (5.62 [1.82] vs 5.86 [1.74]; P = .002). No difference was observed in parents’ highest educational levels between test takers with ADHD and matched controls. Of the 3718 individuals with ADHD who had taken the test, 2745 (73.8%) received ADHD medication (referred to as the medicated group), while the remaining 973 individuals (26.2%) were not medicated during the entire follow-up period (referred to as the never-medicated group). The mean (SD) value of the test scores was more than 10 points higher among the medicated group than among the never-medicated group (94.12 [46.13] vs 83.46 [45.32]; P < .001; Table 2).
In the within-patient analysis, we compared the test scores from 930 individuals (493 males and 437 females; mean [SD] age, 22.2 [3.2] years) who took repeated tests (2524 tests) during their own medicated vs nonmedicated periods. The use of ADHD medications was associated with an increase of 13.13 points (95% CI, 9.73-16.53; P < .001) in the test scores. However, the effect was reduced to a 4.80-point increase (95% CI, 2.26-7.34; P < .001; Table 3) after accounting for age and practice effects. That is, among people who had taken multiple tests and used ADHD medications intermittently, the test scores were a mean of 4.80 points higher (on the scale of 0-200, or 0.048 on the original scale) when the patient was taking medication. The estimated improvement in the test scores appeared to be larger, although not significantly different, in males than in females (mean difference, 5.69; 95% CI, 2.14-9.23 for males vs 3.60; 95% CI, 0.06-7.14 for females; P = .31 for interaction of medication status and sex; Table 3).
In the between-patient comparison, we did not find a significant association between medication use and test scores (eTable 3 in the Supplement).
Within patients, medication was also linked with higher test scores when we used alternative definitions of medication status at the test date. The estimates of mean difference in the test scores, ranging between 3.56 (95% CI, 1.28-5.84) and 4.02 (95% CI, 1.34-6.69), were highly consistent even when we changed the number of days as different cutoff values (Table 4).
Among the 930 individuals who took repeated tests and received ADHD medication, 665 (71.5%) used stimulants only. The association between stimulant medication and the test scores (difference, 3.81; 95% CI, 0.94-6.69) was weaker than the association with nonstimulant medications or a combination of medications (difference, 6.93; 95% CI, 1.81-12.05), although the estimates were not significantly different (P = .23 for interaction of medication status and type of users; Table 5). We identified fewer than 100 individuals with ADHD and coexisting LDs; this subset was too small to warrant separate analyses. Instead, we examined the association among individuals with ADHD who did not have comorbid LDs; the estimated mean test score difference was 5.05 points (95% CI, 2.47-7.62) in this subset.
In contrast, we found only a small, nonsignificant improvement in test scores associated with SSRI use regardless of adjustments for the use of ADHD medication. Within individuals with ADHD who had taken an SSRI, there was a 2.37-point difference (95% CI, –0.88 to 5.63; Table 5) in the mean test scores during periods when participants took SSRIs vs periods when they did not take SSRIs after adjusting for the use of ADHD medication.
In this study, we examined the performance of individuals with ADHD on the higher education entrance examinations during medicated and nonmedicated periods. Among those with a diagnosis of ADHD, the use of ADHD medication was linked with higher scores in the SweSAT; this result survived several sensitivity analyses. The size of the effect was a 0.048-point improvement on the original scale of the test scores, equivalent to 0.11 SD. Comparing the magnitude of effects on test performance, the benefit from taking ADHD medication was comparable to having 1 previous test experience (0.12 SD). The effect might be stronger for individuals with average or high performance on the initial test (eAppendix and eTable 4 in the Supplement). Our results corroborate the emerging evidence from randomized clinical trials36-38 and observational studies39-41 that ADHD medication is associated with a small improvement in results of standardized tests or grade point average18 and further extend the evidence to the age range of adolescence and adulthood, which has been particularly understudied.42
This effect size, albeit small, approximates to an increment of 0.05 in the normed test scores. Such improvement might translate to a higher rank among test applicants, potentially enhancing the chances of receiving higher education. The current results, therefore, might have an implication on an individual’s educational attainment. In addition, educational level has been shown to be a strong indicator of occupational outcomes in adult patients with ADHD43; thus, the resultant better education might lead to other long-term implications. However, the small effect size suggested that other treatment programs are needed to help support individuals with ADHD in educational settings.
In the within-patient design, confounders that are constant within individuals (eg, genetic makeup for disease severity, intelligence, and conscientiousness) or family and school environment that stimulate learning (eg, receiving special educational services) are implicitly adjusted for. We adjusted for age and practice effects to address the confounding of increased test scores on repeated tests owing to the natural curve of improvement in ADHD symptoms44 and improvement in test taking ability. In addition, we did not find evidence for exposure-outcome carryover effects (eAppendix in the Supplement). This finding suggests that the association between the use of ADHD medication and the test scores after controlling for age and practice effects is relatively robust. However, unmeasured confounders that vary within individuals might still be present (eg, periods of time when individuals decide to take medication might represent a time when they make other health and life decisions that may enhance test performance): for such a potential confounder, we tested the association of the test scores with the use of SSRIs, the next commonly used medication in individuals with ADHD. No statistically significant association was detected between SSRI use and test scores, indicating a limited effect of such unmeasured factors (eAppendix and eTable 5 in the Supplement). However, as in all observational studies, we are unable to rule out all time-varying confounders.
Compared with many studies that used retrospective parental reports on medication use,40 we inferred medication use during the test periods from medication prescription records. Thus, our results were not affected by any recall bias. The results, however, rely on whether the medication uses were correctly assigned. Using an alternative definition in which we traced the test date back to the closest drug dispensing and counted the number of days in between the 2 dates, we demonstrated that the observed association was not sensitive to the definition of medication exposure. However, there is still uncertainty about medication status on the test date. Previous studies have shown that nonadherence to an ADHD treatment regimen was the norm rather than the exception.19 In the cases in which treatment nonadherence occurred during our defined medication periods, our estimation was likely to be conservative rather than an overestimate of the true medication effect.
Because higher education entrance examinations comparable to the SweSAT exist in most countries, our results of a positive association between medication use and these tests in individuals with ADHD may be generalizeable to countries with a similar prevalence of ADHD medication use. There is a likelihood of some self-selection in our study. First, the proportion of individuals who had taken the SweSAT was nearly 3 times higher in the population controls than in the individuals with ADHD, probably owing to a higher percentage of school dropouts45 and of individuals who had chosen other educational tracks, such as vocational education and training, in the ADHD group. Further selection arose from the within-individual design, which was restricted to the individuals who had taken repeated tests. Second, IQs among the individuals with ADHD who had taken the tests were comparable to IQs among the matched controls, suggesting that our results might be specific to the individuals with normal cognitive function. This finding might contribute to the observation of similar mean test scores among individuals with ADHD and the matched controls, even though we might have expected those with ADHD diagnoses to have lower scores. We were unable to test whether the identified link between ADHD medication and improved test performance also applies to individuals with more impaired cognitive function or other severe functional impairment. Finally, by probing the link, we explicitly do not assume causality.
Prescriptions for ADHD medications dispensed in the United States increased almost 50% from 2002 to 2010.46 Evidence on links between medication and outcomes is an essential part of dispensing decisions. Like all medications, ADHD medications have unwanted effects,16,47 including concerns regarding nonserious adverse events, such as decreased appetite and sleep problems, and the possibility of serious adverse events,15,16,48 which need to be considered. However, our study suggests that ADHD medications may help improve performance on an important achievement test and, thus, enhance the educational attainment of people with ADHD.
For the people with a diagnosis of ADHD, the use of ADHD medication is associated with better performance on higher education entrance tests. This evidence should be considered together with the current list of risks and benefits of ADHD medication to guide clinical decisions.
Corresponding Author: Yi Lu, PhD, Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Nobels väg 12A, 17177 Stockholm, Sweden (email@example.com).
Accepted for Publication: April 29, 2017.
Published Online: June 28, 2017. doi:10.1001/jamapsychiatry.2017.1472
Author Contributions: Dr Lu had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Lu, Sjölander, Larsson, Lichtenstein.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Lu, Sjölander.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Lu, Sjölander.
Obtained funding: D’Onofrio, Almqvist, Larsson, Lichtenstein.
Administrative, technical, or material support: Lu, Lichtenstein.
Study supervision: Larsson, Lichtenstein.
Conflict of Interest Disclosures: Dr Larsson reported serving as a speaker for Eli Lilly and Shire and receiving a research grant from Shire. Dr Lichtenstein reported serving as a speaker for Medice. No other disclosures were reported.
Funding/Support: This study was supported by framework grant 340-2013-5867 from the Swedish Research Council through the Swedish Initiative for Research on Microdata in the Social and Medical Sciences, grant 2013-2280 from the Swedish Research Council, grant 1R01MH102221 from the National Institute of Mental Health, and the Australian National Health and Medical Research Council early career fellowship (Dr Lu).
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.