Medication Adherence in the MTA: Saliva Methylphenidate Samples Versus Parent Report and Mediating Effect of Concomitant Behavioral Treatment

Copyright @ 2009 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

Medication Adherence in the MTA:Saliva Methylphenidate Samples Versus ParentReport and Mediating Effect of Concomitant

Behavioral TreatmentELIZABETH PAPPADOPULOS, PH.D., PETER S. JENSEN, M.D.,

ALANNA R. CHAIT, B.A., L. EUGENE ARNOLD, M.D., JAMES M. SWANSON, M.D.,LAURENCE L. GREENHILL, M.D., LILY HECHTMAN, M.D., SHIRLEY CHUANG, M.S.,

KAREN C. WELLS, M.D., WILLIAM PELHAM, PH.D., THOMAS COOPER, M.S.,GLENN ELLIOTT, M.D., PH.D., AND JEFFRY H. NEWCORN, M.D.

ABSTRACT

Objective: Although research supports the use of appropriately administered stimulant medication to treat children

with ADHD, poor adherence and early termination undermine the efficacy of this treatment in real-world settings.

Moreover, adherence measures often rely on parent report of medication use, and their validity and reliability are

unknown. Method: Drawing on data from 254 participants in the NIMH Collaborative Multisite Multimodal Treatment

Study of Children With Attention-Deficit/Hyperactivity Disorder, we examine the discrepancy between parents’ verbal

reports of medication adherence and physiological adherence measures determined via methylphenidate saliva

assays collected at four time points during the 14-month treatment period. In addition, we examine the impact of

physiologically documented medication adherence on parent- and teacher-reported outcomes through 14 months.

Results: Overall, nearly one fourth (24.5%) of the saliva samples indicated nonadherence. Among subjects, 63 (24.8%)

of the 254 participants were nonadherent on 50% or more of their repeated saliva assays. Only 136 (53.5%) of the

subjects were adherent at every time point at which saliva assays were taken, indicating that some degree of

nonadherence characterized nearly half of all other NIMH Collaborative Multisite Multimodal Treatment Study of Children

With Attention-Deficit/Hyperactivity DisorderYtreated children. Findings also indicated that nonadherence produced

greater deleterious effects in children in the medication-only condition compared with those receiving both medication

and behavioral treatment. Conclusions: Same-day saliva methylphenidate assays suggest that nearly half of the

parents are inaccurate informants of their child’s ADHD medication adherence and that parents may overestimate

actual (physiological) adherence. This finding suggests the need for interventions to improve accuracy of parental

report. Clinicians need to focus on adherence enhancement strategies to improve outcomes of children being treated

with medication, particularly when benefits are suboptimal. J. Am. Acad. Child Adolesc. Psychiatry, 2009;48(5):00Y00.

Key Words: medication, pharmacotherapy, methylphenidate, saliva levels. Clinical trial registration informationVThe

NIMH MTA Study. URL: http://www.clinicaltrials.gov. Unique identifier: NCT00000388.

Accepted December 18, 2008.See end of text for author affiliations.The work reported was supported by cooperative agreement grants and contracts

from the National Institute of Mental Health to the following: University ofCalifornia, Berkeley: U01MH50461 and N01MH12009; Duke University: U01MH50477 andN01MH12012; University of California, Irvine: U01MH50440and N01MH 12011; Research Foundation for Mental Hygiene (New York StatePsychiatric Institute/Columbia University): U01MH50467 and N01MH12007;Long Island-Jewish Medical Center U01 MH50453; New York University:N01MH 12004; University of Pittsburgh: U01MH50467 and N01MH12010;and McGill University N01MH12008. The Office of Special Education Programsof the U.S. Department of Education and the Office of Juvenile Justice andDelinquency Prevention of the Justice Department also participated in funding.

The opinions and assertions contained in this article are the private views ofthe authors and are not to be construed as official or as reflecting the views of theNational Institute of Mental Health, the National Institutes of Health, or theDepartment of Health and Human Services.

This article is the subject of an editorial by Dr. Philip L. Hazell inthis issue.

Correspondence to Peter Jensen, M.D., The REACH Institute, 450 SeventhAvenue, Suite 1107, New York, NY 10123; e-mail: [email protected].

0890-8567/09/4805-0000�2009 by the American Academy of Child andAdolescent Psychiatry.

DOI: 10.1097/CHI.0b013e31819c23ed

WWW.JAACAP.COM 1J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 48:5, MAY 2009

Copyeditor: Lorelie Eugenio


Although numerous studies have shown that stimulantmedication can significantly improve behavior, atten-tion, concentration,1Y6,7 visual short-term memory,8

and driving performance,9 poor treatment adherenceand early termination can undermine the effectiveness ofthis treatment in real-world settings, producing sub-optimal outcomes,10Y12 a problem recognized even twodecades ago.13 In a recent article, children with attention-deficit/hyperactivity disorder (ADHD) who participatedin long-term stimulant treatment exhibited greater im-provement in teacher-reported symptoms than thosewho discontinued stimulant medications.14 In theNIMH Collaborative Multisite Multimodal TreatmentStudy of Children With Attention-Deficit/Hyperactiv-ity Disorder (MTA), medication use mediated thedegree of baseline to 14-month, and 14- to 24-monthchanges in ADHD symptom ratings, buttressing theinitial 14-month randomized clinical trial findings 4,5,15

that the use of a carefully monitored t.i.d. scheduleof stimulant medication (with or without behavioraltherapy) produced significantly better outcomes thancommunity care (CC) or behavioral therapy alone (Beh).Moreover, analysis of parent-reported actual (versusassigned) treatment with medication suggested thatsubgroups whose parents reported stopping medicationafter 14 months showed the largest deterioration frompreviously achieved levels of symptom improvement.11

Despite the risks associated with insufficient treat-ment adherence and persistence, estimated adherencerates in children with ADHD only range from 56%to 75%,16 and more than 50% of patients withADHD discontinue treatment regardless of its efficacyor symptom severity.17,18 For example, one study ofchildren (N = 1,635; aged 3Y7 years) taking meth-ylphenidate (MPH) immediate-release revealed that54.0% of the subjects received only one prescription,and only 10.9% received five or more prescriptionsduring a 1-year period.19 Another study found thatonly 74% of children with ADHD initially assignedto stimulants took 50% or more of their pills for12 months,20 and only 52% continued to use stimulantmedication for 3 consecutive years.21

Unfortunately, it is often unclear from such reports20

how much of the 50% Badherence[ rate resulted fromplanned drug holidays or deliberate medication onlyon school days. For example, with 180 to 185 schooldays in a year, one who takes medication only onschool days with 100% fidelity could end up with

50% Badherence.[ Nevertheless, it is unlikely that suchplanned nonmedication days would account for allreported nonadherence.Nonadherence has been linked to a variety of factors,

ranging from proximal and practical aspects of medica-tion administration and/or immediately recognized sideeffects to more distal factors, including family socio-economic status and ethnicity.3,22,23 Although preciseestimates of actual adherence and persistence have re-mained unavailable, recent national data concerningADHD treatment suggest that such problems may beincreasing, perhaps in part because of many youthsreceiving insufficient medication treatment visits.1,19Y21

The validity and reliability of medication adherencemeasures, which often rely on parent report, vary orare unknown for stimulants24 and nonstimulants25 inyouths. The current analysis represents the first attemptto investigate the discrepancy between parent report andphysiological measures of medication compliance. Usingdata from the MTA Cooperative Group’s 14-monthrandomized controlled trial of ADHD treatments4,5

initiated in 1994, the present study evaluates the dis-crepancy between parent-reported and physiologicalmeasures of adherence by comparing saliva assay valuesof MPH adherence with parent verbal-report measuresand also examines the extent to which physiologicalversus parent-reported adherence measures predictADHD treatment outcomes through the 14-monthtreatment period of this clinical trial.

METHOD

Study Population

As described in the initial report of the MTA outcomes,4,5 579children with ADHD combined type (aged 7Y9.9 years) were ran-domly assigned to medication management consisting of titrationwith t.i.d. administration of 5 to 20 mg of immediate-release MPHfollowed by monthly visits (MedMgt); behavior treatment consistingof intensive parent, school, and child components, with therapistinvolvement gradually reduced over time (Beh); the two combined(Comb); or usual community care consisting of treatments by com-munity providers (CC). The experimental interventions (MedMgt,Beh, and Comb) were delivered by MTA staff for 14 months andwere carefully monitored for treatment fidelity. Multiple assess-ments from various sources and across domains were collected atbaseline and at 3, 9, and 14 months (see MTA Cooperative Groupfor more details4,5). The present study focuses on the 289 subjectsin the medication-only and combined treatment groups, who wereassigned to systematic, carefully titrated,26 monitored, and adjustedmedication.27

PAPPADOPULOS ET AL.

2 WWW.JAACAP.COM J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 48:5, MAY 2009


Study Design

As the original reports of the MTA outcomes indicate,4,5 subjectsin the medication management and combination groups startedwith a 28-day, double-blind, daily-switch titration of MPH hydro-chloride, using 5 randomly ordered repeats each of placebo, 5, 10,and 15 or 20 mg, with the higher dose, 15 mg versus 20 mg, beingdetermined on the basis of the child’s weight (threshold Q26 kg).Participants in the combined treatment group received medication(outlined above) and behavioral treatment; namely, parent groupand individual sessions, teacher consultation, and a classroom aide.4

At the 1-, 3-, 9-, and 14-month assessments, MPH saliva swabswere taken as a part of and at the end of the standard monthlymedication visit. Saliva samples were subsequently assayed for thepresence of MPH by gas-chromatographic analysis at the NathanKline Institute.

Medication Adherence Measures

Parents’ verbal reports of their child’s medication adherenceduring the medication maintenance phase of the study and subject-based physiological measures of adherence were available for com-parative analyses. For the sake of clarity throughout this report, thesetwo methods are termed verbal adherence and physiological adherence,respectively. Verbal adherence information (based on parental reportsof their child’s medication status) was acquired from the MTAMedical Visit clinical report form. At each monthly medicationvisit, if the parent reported that their child was taking the studymedication as prescribed, the child was then coded as being ad-herent for that visit (adherence defined as the parent’s report ofoverall adherence since the last visit and having taken medicationas prescribed on the day of the visit). If the parent indicated that thechild was not taking the study medication, then the child was codedas being nonadherent for that visit. However, because children werescheduled for monthly (up to 13) medication follow-up visits,5 allComb and MedMgt subjects were reclassified as verbally adherentor verbally nonadherent, if parents reported that the children tookmedication on 80% or more of these monthly visits. Although ourprevious analyses revealed significant mediating effects of parent’sverbally reported medication adherence5 on ADHD treatment out-comes, our earlier analyses did not examine the extent to whichparent reports correlated with physiological adherence (defined hereas MPH saliva assays), nor whether physiological adherence wasrelated to overall treatment outcomes.

Physiological Adherence

The protocol specified that four saliva assays would be conductedon each MTA medication-treated child during the course of the14-month treatment period (at follow-up visits 2, 6, 9, and 14),with no prenotification given to families of whether saliva assayswould be drawn during that particular visit. Saliva swabs weregathered from the child only after the parent had verbally reportedabout the child’s medication adherence. For each saliva record,detectable MPH was used to indicate the presence of MPH takenthat day (limit of detection for the assay used was 93ng/mL). Allassays were refrigerated upon receipt at Nathan Kline Institute andpromptly analyzed on the next available workday. Assays were codedto be physiologically adherent if MPH was detectable upon chro-matographic analysis and nonadherent if MPH was not detectable.Across all 289 subjects in the MTA Comb and MedMgt groups,31 \children did not have any saliva swabs collected and, as a result,were excluded from the analyses: 20 children who terminated early

from the maintenance phase before their first scheduled medicationvisit when saliva samples would have been collected and 11 childrenwho continued in the study but for whom no saliva measures weretaken because of scheduling problems or inadvertent omission bystudy staff. Of the remaining 258 children (123 MedMgt and 135Comb subjects), 254 had one or more readable saliva assays (32saliva vials were broken in transit or otherwise not readable). Nosignificant differences were found between the two medicationgroups in medication participation, swab frequency, or studyretention. The available 254 subjects (87.9%) provided a total of780 assays (3.1 assays/participant; range 1Y4 assays, with 748 (2.9/participant) usable assays.Subject-based physiological adherence during the maintenance

phase was computed by dividing the number of adherent assays foreach child by the total number of assays collected during themedication maintenance phase for the same child. Because of thevariable number of assays per subject, subjects were recoded asadherent if greater than 50% of the given subject’s assays haddetectable MPH levels.

Data Analyses

As a first step, adherent versus nonadherent subjects were com-pared on baseline demographic characteristics. In addition, thedegree of concordance between parent-reported verbal adherenceand physiological adherence was examined, first by cross-tabulatingall 748 instances of MPH assays with the parents’ verbal reportsduring the same visits and again, at the subject level, by reclassifyingthe 254 available subjects for whom data were available into verballyadherent/nonadherent and physiological adherent/nonadherentbinary variables.Given the advantages of random-effects regression (RR) tech-

niques over traditional analyses of variance for longitudinal clinicaltrials data,4,11 we used RR for subsequent analyses examining thepotential mediating impact of our subject-based binary physiologicaladherence/nonadherence variable on ADHD symptom outcomes,with tests for two between-subject factors (site and adherence) andone within-subject factor (time), plus the interaction of these factors,time � treatment (any differences over time between MedMgt andComb), adherence � time (any differences over time betweenadherent and nonadherent subgroups), and adherence � time �treatment (differential effects of medication adherence over timebetween the two treatment groups) within the RR analyses.Following previous reports,4,11 ADHD symptom ratings were

used as the primary outcome measures. These ratings were derivedfrom the SNAP-IV (an acronym for the scale developers, Swanson,Nolan, and Pelham) rating scale completed by parents and teachersof the child. The SNAP summary scores were computed by theweighted averages of the nine inattention and nine hyperactivity-impulsivity items in the SNAP-IV questionnaire, separately forparents and teachers. Following previously established procedures,11

we examined the separate parent and teacher informants’ scores in acombined analysis of parent and teacher ADHD symptom ratingsusing a mixed-effects regression model with two factors: a fixed-effectinformant (parent or teacher) factor nested within a factor definedby subject (random effect). This approach allows the power and reli-ability of both informants to be considered in the test, with allsubjects included even if some observations are missing, while alsotesting for informant differences by the interaction of informant withtreatment group. According to conventions for factorial analyses inwhich main effects and interaction effects are orthogonal, significancelevels were set at p < .05, to determine whether adherence � time

MEDICATION ADHERENCE IN THE MTA STUDY



or adherence � time � treatment effects could be identified. Ineffect, these analyses addressed the question whether any differencesmight be found because of differential medication taking based onthe physiological MPH assays.

RESULTS

Physiological Adherence

Overall, nearly one fourth (176 of 748 [23.5%]) ofthe saliva samples were physiologically nonadherent,below the detectable 3 ng/mL (indicating no MPHingestion within 8 hours). To characterize the findingsby child rather than by assay, for each child, wecalculated the proportion of positive assays versus totalnumber of assays drawn. Because these proportions wereconstructed based on anywhere from 1 to 4 salivasamples, depending on the number of swabs from eachsubject, proportions varied from 0, 0.25, 0.33, 0.50,0.67, 0.75, and 1. When findings were grouped bychildren (rather than by records), 24.8% (n = 63) of 254children’s medication regimens were classified physio-logically nonadherent (i.e., e50% of their saliva samplesbelow detectable 3 ng/mL). Moreover, across allchildren, 31 subjects (12.2%) were nonadherent on allcollected saliva samples, and just more than half ofthe subjects (n = 136 [53.5%]) were physiologicallyadherent on all of their available samples. SeeT1 Table 1 fordetails. The MedMgt and Comb groups did not differsignificantly in rates of physiological adherence.

T2 Table 2 presents data on the demographic character-istics of physiologically adherent versus nonadherentsubjects. As seen here, two factors were differentially

related to greater physiological medication adherence,ethnicity (more adherence in whites), and having amother who was working outside of the home.

Verbal (Parent-Reported) Medication Adherence

Most parents reported that their child took themedication during that day. A total of 228 (89.8%) of254 parents reported their children as taking the studymedication on all occasions, 4 children (1.2%) werereported as taking their medications three fourths of thetime, 14 children (5.6%) took their medication twothirds of the time, 5 children (2.0%) took their medi-cations half of the time, 1 child (0.4%) took medicationone third of the time, and 2 children (0.8%) none of thetime by parent report. Thus, applying the same defini-tion for nonadherence to parental reports (e50%) as forphysiological adherence, only 8 (3.1%) of the parentsreported nonadherence.

Concordance Between Verbal and Physiological

Adherence

T3Table 3 presents concordance information of verbaland physiological adherence for the 641 of 748 salivaassays for which parental reports of compliance werealso obtained. As seen in the table, concordance ofparental reports of adherence with physiological mea-sures was substantially better than chance, but actualagreement was only Bslight[ (i.e., 0’s < 0.20), using therating system of Landis and Koch,28 suggesting parentaloverreporting of actual medication taking as measuredby saliva MPH. From the perspective of the sensitivity

T AQ1ABLE 1

Proportion of Positive Assays vs. Total Assays Drawn, by Subject

No. and Percentage of Subjects forEach Proportion of Positive vs. Total Assays

Drawn (N = 254)

Subjects’ Physiological Adherence Classification* Proportion of Positive Assays n % Cumulative %

Nonadherent total (n = 63) 0.0 31 12.2 12.20.25 6 2.4 14.60.33 13 5.1 19.70.50 13 5.1 24.8

Adherent total (n = 191) 0.67 36 14.2 39.00.75 19 7.5 46.51.0 136 53.5 100.0Totals 254 100.0 100.0

Note: Each subject was classified as physiologically adherent or nonadherent, based on his/her ratio of positive saliva assays to the total numberof saliva assays drawn. By protocol, each subject should have had four assays, but actual range was one to four assays (average 2.9 assays/subject).Subjects with assay positive proportions of 0.50 or lesser were classified as Bphysiologically nonadherent,[ whereas those with proportions ofgreater than 0.50 were classified as Bphysiologically adherent.[

PAPPADOPULOS ET AL.



and specificity of a test against a gold standard, if weuse parent verbal reports as a Btest[ to detect non-adherence against the physiological assay, the test haspoor sensitivity, with only 14 of 136 physiologicalnonadherence cases (10.3%) identified by parentalreport. Negative test results (i.e., the parent reportsadherence) has a specificity of 80.3% (496/618), indi-cating that reliance on reports of adherence within asingle visit will be correct only 80% of the time, mis-classifying 1 in 5 visits as adherence that is not. Froma clinical perspective, therefore, a positive parental re-port of nonadherence merits clinical attention, whereasa negative report of nonadherence (i.e., report of ad-herence) merits clinical suspicion if clinical results arenot satisfactory.

Random Regression Analyses of Adherence as a

Mediator of Outcomes

For our analyses using mixed effects regression, weexamined ADHD outcome as a function of the com-posited parent-teacher SNAP reports, examining site,time, physiological adherence (adherent versus non-adherent), treatment (Comb versus MedMgt), site �treatment, time � treatment, treatment � adherence,time � adherence, and time � treatment � adherence.Effects were as follows: site (F5,468 = 0.92, ns), time(F1,248 = 754.8, p < .0001), treatment (F1,468 = 3.52, p =.06), adherence (F1,468 = 0.23, ns), site � treatment(F5,468 = .35, ns), time � treatment (F1,468 = 0.46, ns)treatment� adherence (F1,468 = 5.73, p = .017), time�adherence (F1,468 = 12.2, p < .0005), and time �

TABLE 2Demographic Characteristics of Physiological Adherencea Groups

VariableNonadherent(n = 63)

Adherent(n = 191)

Group Comparison

Statistic p

Age in years, mean (SD) 7.89 (0.81) 7.79 (0.79) t252 = 0.85 .39Sex, male, n (%) 51 (80.95) 150 (78.53) #2 = 0.17 .68Ethnicity, n (%) #2 = 12.6 .05White 33 (52.38) 124 (64.92)Black 15 (23.81) 27 (14.14)Nonblack Hispanic 8 (12.70) 12 (6.28)Black Hispanic 2 (3.17) 3 (1.57)Asian 1 (1.59) 0 (0.00)Mixed 3 (4.76) 23 (12.04)Other 1 (1.59) 2 (1.05)

Grade, n (%) #2 = 5.88 .121 3 (4.76) 32 (16.75)2 26 (41.27) 72 (37.70)3 25 (39.68) 66 (34.55)4 9 (14.29) 21 (10.99)

Previous stimulant medication, n (%) 22 (34.92) 60 (31.41) #2 = 0.27 .61Welfare, n (%) 11 (17.46) 35 (18.32) #2 = 0.01 .92High school graduate, n (%)Mother 59 (93.65) 177 (93.16) #2 = 0.02 .89Father 45 (88.46) 140 (90.91) #2 = 0.03 .61

Employed, n (%)Mother 21 (33.87) 103 (53.93) #2 = 7.54 .006Father 45 (86.54) 129 (83.77) #2 = 0.23 .63

Income, $, n (%) #2 = 0.16 .920Y20,000 11 (18.03) 37 (19.68)20,000Y50,000 26 (42.62) 75 (38.89)950,000 24 (39.34) 76 (40.43)

Married, n (%) 43 (68.25) 132 (69.47) #2 = 0.03 .86

aEach subject was classified as physiologically adherent or nonadherent, based on his or her ratio of positive saliva assays to the total number ofsaliva assays drawn. By protocol, each subject could have up to four assays, but actual range was one to four assays (average 2.9 assays/subject).Subjects with assay positive proportions of 0.50 or lesser were classified as Bphysiologically nonadherent,[ whereas those with proportions greaterthan 0.50 were classified as Bphysiologically adherent.[




treatment� adherence (F1,468 = 0.50, ns). Actual meansof parent-teacher composited SNAP scores by physio-logical adherence status are shown inT4 Table 4 andplotted inF1 Figure 1.These findings demonstrated that ADHD symptoms

decreased significantly over time and was mediated byphysiological adherence (adherent subjects showedgreater improvement over time). The significant effectof treatment � adherence suggests that the main effectof adherence on ADHD ratings was different in theComb versus MedMgt groups, as seen in Figure 1, butthe lack of a triple interaction (time � treatment �adherence) indicates that the MedMgt and Combgroups did not show a significant differential effect ofadherence on the rates of improvement over time.

DISCUSSION

This article uniquely examines prospective medica-tion adherence in the MTA study by contrastingparental reports of adherence with MPH saliva contentvalues collected at the same visit. Before considering theimplications of our findings, several caveats are in order.First, findings must be seen in light of the possibilitythat MPH saliva assays may not have yielded detectablelevels (i.e., 93ng/mL) if the child had last taken MPH 8or more hours earlier. So if the parent reported at a late-day assessment that the child had taken medication asdirected that day but was unaware that the noon-daydose had been missed at school, a discrepancy betweenthe late-day saliva assay (negative) and the parent’sreport would have resulted inadvertently. In addition,our conclusions should be tempered by the uncertaintyabout how long saliva values are detectable (Q3 ng/mL)on chromatographic analysis and whether that consti-tutes an appropriate threshold for adherence.

Caution is also warranted, to the extent that ourresults apply only to the MTA sample (combined typeADHD, age 7Y10 years) and medication regimen,principally based on immediate-release MPH t.i.d.Short-acting formulations have been mostly replacedby newer longer-acting profiles, so the adherence/compliance rates for newer formulations may differfrom those reported here.From an optimistic perspective, one might view our

medication nonadherence based on saliva samples asrelatively low, with only one fourth (176 records[23.5%]) of all the saliva samples indicating nonadher-ence. Pessimistically, however, 24.8% of the subjects

TABLE 4Composite Parent-Teacher SNAP Scale Scores by Physiological

Adherence Groupa

Monthb

Nonadherentb (n = 63) Adherentb (n = 191)

Comb MedMgt Comb MedMgt

Mean SD Mean SD Mean SD Mean SD

0 1.88 0.49 2.11 0.36 2.07 0.45 2.05 0.423 0.95 0.57 1.63 0.56 1.07 0.44 1.04 0.599 1.01 0.57 1.30 0.51 0.97 0.47 0.90 0.4814 0.93 0.58 1.20 0.58 0.90 0.46 0.88 0.52

Note: Comb = combined treatment (medication management +behavioral therapy); MedMgt = medication management only;SNAP = Swanson, Nolan, and Pelham.

aEach subject was classified as physiologically adherent ornonadherent, based on his/her ratio of positive saliva assays to thetotal number of saliva assays drawn. Subjects with assay positiveproportions of 0.50 or lesser were classified as Bphysiologically non-adherent,[ whereas those with proportions of greater than 0.50 wereclassified as Bphysiologically adherent.[

bSignificant differences found for time (i.e., month) (F1,248 =754.8, p < .0001), treatment (i.e., MedMgt vs. Comb) � adher-ence (F1,468 = 5.73, p = .017), and time � adherence (F1,468 = 12.2,p < .0005).

TABLE 3Record-Based Concordance Physiological (by Saliva Assay)

Adherence vs. Parent-Reported Adherencea

Verbal Adherence on this Visit?

Physiological adherence this assay? No Yes TotalNo 14 (60.9%) 122 (19.7%) 136 (21.2%)Yes 9 (39.1%) 496 (80.3%) 505 (78.8%)Total 23 (100%) (3.6%) 618 (100%) (96.4%) 641b (100.0%) (100.0%)

aAn individual record was classified as physiologically adherent on a given visit, if the saliva assay showed a detectable level (93 ng/mL);otherwise, the record was coded as physiologically nonadherent. During the same visit, same case records were coded as verbally adherent if theparent reported that the child had taken the medication that same day.

b0 = 0.12, #2 = 22.4, p < .001.

PAPPADOPULOS ET AL.



were nonadherent in 50% or more of their collectedsamples, although almost 90% of the parents reportedthat their child was adherent at every visit. Only 3.1% ofthe parent reports indicated adherence at or below the50% threshold versus 23.5% of the physiologicalsamples below the adherence threshold.

Interestingly, our results of verbal adherence (89.8%across all visits) are higher than generally seen in lon-gitudinal studies,16Y21 likely because of the intensivefollow-up procedures put in place to facilitate adherenceduring 14 months. Despite our high reported adherencerates, our physiological adherence findings are consistentwith studies1,16,18,29 that have linked poor adherenceto adverse outcomes. However, unlike some previousstudies, we found no associations between adherenceand demographic factors such as age, sex,17,22 familyincome, and single-parent status1,16,18,29,30 In contrast,we did find somewhat greater physiological nonadher-ence as a function of minority ethnic status, as well asamong children whose mothers were not employed. Themeaning of this finding is unclear and may result fromthe different types of subjects who participate in long-term treatment studies versus those drawn from morenaturalistic service use databases and short-term studies.

The current literature on ethnic disparities in ad-herence is somewhat inconsistent: some studies indicatethat minority youths are at increased risk for premature

treatment discontinuation and lower use of psychiatricservices and medication,23,31 but other studies suggestno ethnic disparities in mental health prescription rates,referrals, and counseling once a diagnosis is made.32,33

Moreover, although some research suggests that mi-nority parents, compared with their white counterparts,harbor more negative attitudes about ADHD medica-tion treatment,32,34 a recent examination of other MTAvariables showed no significant ethnic differences inacceptance of or compliance with either medication orbehavioral treatments.35 The discrepancy between ourcurrent saliva findings and our previously reported35

adherence rates may be because our earlier findingsconcerning ethnicity were based on the definition ofmedication adherence as B80% or greater attendance[ atthe 14 months of medication-monitoring visits.To the extent that our saliva assays were a fair

representation of the child’s having taken medication onthe day that parents reported compliance, concordancein Table 3 between parental reports of compliance andphysiological measures was only fair, suggesting thatparents overreport their child’s actual medication-takingbehavior. Given recent research on factors related tononadherent behavior,36 the present discrepancy be-tween parental reports and saliva levels may haveresulted from both intentional (purposeful actions)and unintentional (patterned behaviors, such as forget-ting) variables. Parents may misreport their child’smedication-taking behavior to satisfy their physicians’expectations and to project a favorable impression.Relatedly, one study of youths’ adherence to humanimmunodeficiency virus therapy found that manyparticipants viewed their health care providers aspowerful authorities and admitted to giving theminaccurate adherence information to avoid blame andcriticism.37 Recent literature has similarly demonstratedthe impact of the therapeutic relation on adherencereports and behaviors, suggesting that a more positiveparentYtherapist relationship predicts greater treatmentacceptability and persistence.38 Further research on thedevelopment of a collaborative doctorYpatient relation-ship and methods for increasing parents’ comfort incandid disclosure might lead to improvements in actualmedication adherence or at least more accurateinformation about the degree of actual adherence.Stigma associated with medication therapy may also

contribute to the discrepancy between self-report andbiological measures of adherence. Previous research

Fig. 1 Parent-teacher composite SNAP values for 14 months. Four groupsare plotted by physiological adherence (adherent versus nonadherent) andby treatment group (Comb versus MedMgt). Significant differences foundfor time (F1,248 = 754.8, p < .0001), treatment � adherence (F1,468 = 5.73,p = .017), and time � adherence (F1,468 = 12.2, p < .0005). See text forfull details. ADH = physiologically adherent; NADH = physiologicallynonadherent; Comb = combined treatment (medication management +behavioral therapy), MedMgt = medication management only.




indicates that inaccurate information about treatment,embarrassment, and anticipated disapproval from familyand peers can lead patients or parents to skip medicationdoses or prematurely discontinue treatment.37,39 More-over, parents of children with ADHD are reported toconsider medication treatments less socially acceptablethan psychosocial or combination interventions.40

Stigma reduction programs may be needed to nullifystereotypes about ADHD care and encourage treatmentadherence among youths and their parents.To the extent that unintentional barriers to adher-

ence, such as lifestyle factors, busyness, or simplyforgetting, might play a role in adherence, parents ofchildren with ADHD and long-term pharmacotherapyregimens may benefit from medication recall tools, suchas pill boxes or treatment tracking systems, as well assimplified (single daily dose) medication regimens.41

The current study provides new insights into thepotential causes of ADHD medication Bnonresponse[and suggests that clinicians should assess medicationadherence in patients with a poor response beforealtering dosage or changing medications. The physicianshould consider asking about adherence in a fashion thatmakes full and candid disclosure more likely such assaying, BMany parents have concerns about themedication, or find it hard to remember doses,sometimes during the week, or sometimes on weekends.Or some parents cut the dose because they are worriedabout side effects, or because of something they readabout medicine that troubled them. What’s been thebiggest problem for you, in either remembering to giveit, or cutting the dose?[Generally, our findings seem to suggest that

adherence may be more important for patients whoseonly treatment is medication (MedMgt) than for thosewho have additional behavioral therapy (Comb).Possibly, in the Comb group, the added benefit ofbehavioral therapy allowed a good outcome even whenmedication adherence was low, whereas the MedMgtoutcomes were more fully dependent on medicationadherence.Because our current findings, as well as previous

reports, have shown that medication adherence isassociated with children’s outcomes and because parentreports may often be incomplete, it is worth theclinicians’ time to focus on strategies for enhancingadherence at treatment outset and throughout treatmentduration. Such approaches might include encouraging

parents to adopt a regular time and routine formedication, encouraging parents to monitor andreinforce children’s pill taking carefully, using long-acting formulations to reduce the number of times thefamily needs to remember medication, enlisting coop-eration of both child and parent, using a medicationdiary or log (perhaps as simple as a calendar), putting thefamily at ease so they candidly report adherence, andgiving the child a chance privately to talk aboutadherence and his/her feelings about medication.Research on these adherence strategies, as well as studiesof medication adherence across different psychiatricdisorders or using different methods (such as drug urinedipstick tests), are sorely needed. In particular, studiesare needed to identify adherence strategies that actuallywork, including whether they can improve children’soutcomes. Such strategies are especially important forchildren who seem unresponsive to medication andwhere adherence is of potential concern.

The NIMH Collaborative Multisite Multimodal Treatment Studyof Children With Attention-Deficit/Hyperactivity Disorder (MTA)was a National Institute of Mental Health (NIMH) cooperativeagreement randomized clinical trial involving six clinical sites.Collaborators from the National Institute of Mental Health: Peter S.Jensen, M.D. (currently at Columbia University), L. Eugene Arnold,M.D., M.Ed. (currently at Ohio State University), Joanne B. Severe,M.S. (Clinical Trials Operations and Biostatistics Unit, Division ofServices and Intervention Research), Benedetto Vitiello, M.D.(Child and Adolescent Treatment and Preventive InterventionsResearch Branch), Kimberly Hoagwood, Ph.D. (currently atColumbia); previous contributors from NIMH to the early phase:John Richters, Ph.D. (currently at National Institute of NursingResearch); Donald Vereen, M.D. (currently at National Institute onDrug Abuse). Principal investigators and coinvestigators from theclinical sites are as follows: University of California, Berkeley/SanFrancisco: Stephen P. Hinshaw, Ph.D. (Berkeley), Glen R. Elliott,Ph.D., M.D. (San Francisco); Duke University: C. Keith Conners,Ph.D., Karen C. Wells, Ph.D., John March, M.D., M.P.H., JefferyEpstein, Ph.D.; University of California, Irvine/Los Angeles: JamesSwanson, Ph.D. (Irvine), Dennis P. Cantwell, M.D., (deceased, LosAngeles), Timothy Wigal, Ph.D. (Irvine); Long Island JewishMedical Center/Montreal Children’s Hospital: Howard B. Abikoff,Ph.D. (currently at New York University School of Medicine), LilyHechtman, M.D. (McGill University); New York State Psychi-atric Institute/Columbia University/Mount Sinai Medical Center:Laurence L. Greenhill, M.D. (Columbia), Jeffrey H. Newcorn,M.D. (Mount Sinai School of Medicine); University of Pittsburgh:William E. Pelham, Ph.D. (currently at State University of NewYork, Buffalo), Betsy Hoza, Ph.D. (currently at University ofVermont), Brooke Molina, Ph.D. Original statistical and trial designconsultant: Helena C. Kraemer, Ph.D. (Stanford University). Follow-up phase statistical collaborators: Robert D. Gibbons, Ph.D.(University of Illinois, Chicago), Sue Marcus, Ph.D (Mt. SinaiCollege of Medicine), Kwan Hur, Ph.D. (University of Illinois,

PAPPADOPULOS ET AL.



Chicago). Collaborator from the Office of Special EducationPrograms/U.S. Department of Education: Thomas Hanley, Ed.D.Collaborator from Office of Juvenile Justice and DelinquencyPrevention/Department of Justice: Karen Stern, Ph.D.

Disclosure: During the course of the MTA, since 1992, Dr.Pappadopulos has been an employee of Pfizer. Dr. Jensen has receivedtraining and/or research funds from the Lowenstein Foundation, theKlingenstein Third Generation Foundation, the Kellogg Foundation,Annie E. Casey Foundation, Casey Family Programs, Novartis,Magellan, and Value Options; participates in speakers’ bureaus forCMED, UCB Pharma, CME Outfitters, and the NeuroscienceEducation Institute, McNeil; and consults to Best Practice, McNeil,Novartis, and Shire. Dr. Arnold has had research funding from AutismSpeaks, NIMH, Neuropharm, Novartis, Noven, Shire, Sigma Tau, andTargacept; has consulted to Abbott, Neuropharm, Novartis, Noven,Organon, Shire, and Sigma Tau; and is/was on the speakers’ bureaus ofMcNeil, Novartis, and Shire. Dr. Swanson has received research supportfrom Alza, Richwood, Shire, Celgene, Novartis, Celltech, Gliatech,Cephalon, Watson, CIBA, Janssen, and McNeil; has been on theadvisory boards of Alza, Richwood, Shire, Celgene, Novartis, Celltech,UCB, Gliatech, Cephalon, McNeil, and Lilly; has been on the speakers’bureaus of Alza, Shire, Novartis, Celltech, UCB, Cephalon, CIBA,Janssen, and McNeil; and has consulted to Alza, Richwood, Shire,Celgene, Novartis, Celltech, UCB, Gliatech, Cephalon, Watson, CIBA,Janssen, McNeil, and Lilly. Dr. Greenhill has received research fundingor has been on the speakers’ bureaus of Lilly, Alza, Shire, Cephalon,McNeil, Celltech, Novartis, Sanofi Aventis, Otsuka, and Janssen. Dr.Hechtman has received research funding from the National Institute ofMental Health, Eli Lilly, GlaxoSmithKline, Janssen-Ortho, PurduePharma, and Shire; has been on the speakers’ bureaus of NationalInstitute of Mental Health, Eli Lilly, Purdue Pharma, Janssen-Ortho,and Shire; and has been on the advisory boards of Eli Lilly, Janssen-Ortho, Purdue Pharma, and Shire. Dr. Wells receives speaking fees fromthe REACH Institute. Dr. Pelham has received research funding fromAlza, Shire, Noven, Lilly, and Cephalon; has served on the advisoryboards of or has consulted to Alza/McNeil Richwood/Shire, Noven, Lilly,Cephalon, Novartis, Celgene, and Abbott; and has been on the speakers’bureaus of Shire and McNeil. Dr. Elliott has received research fundingfrom Cephalon, McNeil, Shire, Sigma Tau, and Novartis; has consultedto Cephalon and McNeil; and has been on the speakers’ bureaus ofJanssen, Lilly, and McNeil, but has no current competing interests.Dr. Newcorn has received research funding from or has been on thespeakers’ bureaus of Lilly, Alza, Shire, Celgene, McNeil, Celltech/UCB,Novartis, Sanofi Aventis, Janssen, and Bristol-Myers Squibb. The otherauthors report no conflicts of interest.

REFERENCES

1. Cox DJ, Merkel RL, Penberthy JK, Kovatchev B, Hankin CS. Impact ofmethylphenidate delivery profiles on driving performance of adolescentswith attention-deficit/hyperactivity disorder: a pilot study. J Am AcadChild Adolesc Psychiatry. 2004;43:269Y275.

2. Greenhill LL, Findling RL, Swanson JM, ADHD Study Group. Adouble-blind, placebo-controlled study of modified-release methylphe-nidate in children with attention-deficit/hyperactivity disorder. Pediatrics.2002;109:E39.

3. Wolraich ML, Greenhill LL, Pelham W et al. Randomized, controlledtrial of OROS methylphenidate once a day in children with attention-deficit/hyperactivity disorder. Pediatrics. 2001;108:883Y892.

4. MTA Cooperative Group. A 14-month randomized clinical trial oftreatment strategies for attention-deficit/hyperactivity disorder. Multi-

modal Treatment Study of Children with ADHD. Arch Gen Psychiatry.1999;56:1073Y1086.

5. Moderators and mediators of treatment response for children withattention-deficit/hyperactivity disorder: the Multimodal TreatmentStudy of Children With Attention-Deficit/Hyperactivity Disorder.Arch Gen Psychiatry. 1999;56:1088Y1096.

6. Abikoff H, Hechtman L, Klein RG et al. Symptomatic improvement inchildren with ADHD treated with long-term methylphenidate andmultimodal psychosocial treatment. J Am Acad Child Adolesc Psychiatry.2004;43:802Y811.

7. Tourette’s Syndrome Study Group. Treatment of ADHD in childrenwith tics: a randomized controlled trial. Neurology. 2002;58:527Y536.

8. Mehta MA, Goodyer IM, Sahakian BJ. Methylphenidate improvesworking memory and set-shifting in AD/HD: relationships to baselinememory capacity. J Child Psychol Psychiatry. 2004;45:293Y305.

9. Bedard AC, Ickowicz A, Logan GD, Hogg-Johnson S, Schachar R,Tannock R. Selective inhibition in children with attention-deficit hyper-activity disorder off and on stimulant medication. J Abnorm Child Psychol.2003;31:315Y327.

10. McGough JJ, Biederman J, Wigal SB et al. Long-term tolerability andeffectiveness of once-daily mixed amphetamine salts (Adderall XR) inchildren with ADHD. J Am Acad Child Adolesc Psychiatry. 2005;44:530Y538.

11. MTA Cooperative Group. National Institute of Mental Health Multi-modal Treatment Study of ADHD follow-up: changes in effectivenessand growth after the end of treatment. Pediatrics. 2004;113:762Y769.

12. Hoagwood KE, Jensen PS, Arnold LE et al. Reliability of the services forchildren and adolescents-parent interview. J Am Acad Child AdolescPsychiatry. 2004;43:1345Y1354.

13. Sleator EK. Measurement of compliance. Psychopharmacol Bull. 1985;21:1089Y1093.

14. Charach A, Ickowicz A, Schachar R. Stimulant treatment over five years:adherence, effectiveness, and adverse effects. J Am Acad Child AdolescPsychiatry. 2004;43:559Y567.

15. MTA Cooperative Group. National Institute of Mental Health Multi-modal Treatment Study of ADHD follow-up: 24-month outcomesof treatment strategies for attention-deficit/hyperactivity disorder. Pedi-atrics. 2004;113:754Y761.

16. Hack S, Chow B. Pediatric psychotropic medication compliance: aliterature review and research-based suggestions for improving treatmentcompliance. J Child Adolesc Psychopharmacol. 2001;11:59Y67.

17. Firestone P. Factors associated with children’s adherence to stimulantmedication. Am J Orthopsychiatry. 1982;52:447Y457.

18. Perwien A, Hall J, Swensen A, Swindle R. Stimulant treatment patternsand compliance in children and adults with newly treated attention-deficit/hyperactivity disorder. J Manag Care Pharm. 2004;10:122Y129.

19. Olfson M, Gameroff MJ, Marcus SC, Jensen PS. National trends in thetreatment of attention deficit hyperactivity disorder. Am J Psychiatry.2003;160:1071Y1077.

20. Corkum P, Rimer P, Schachar R. Parental knowledge of attention-deficit hyperactivity disorder and opinions of treatment options: impacton enrollment and adherence to a 12-month treatment trial. Can JPsychiatry. 1999;44:1043Y1048.

21. Thiruchelvam D, Charach A, Schachar RJ. Moderators and mediatorsof long-term adherence to stimulant treatment in children with ADHD.J Am Acad Child Adolesc Psychiatry. 2001;40:922Y928.

22. Swanson J. Compliance with stimulants for attention-deficit/hyperactivitydisorder: issues and approaches for improvement. CNS Drugs. 2003;17:117Y131.

23. Marcus SC, Wan GJ, Kemner JE, Olfson M. Continuity of methyl-phenidate treatment for attention-deficit/hyperactivity disorder. ArchPediatr Adolesc Med. 2005;159:572Y578.

24. Berg J, Dunbar-Jacob J, Rohay JM. Compliance with inhaled medica-tions: the relationship between diary and electronic monitor. Ann BehavMed. 1998;20:36Y38.

25. Strunk RC, Bender B, Young DA et al. Predictors of protocol adherencein a pediatric asthma clinical trial. J Allergy Clin Immunol. 2002;110:596Y602.

26. Greenhill LL, Swanson JM, Vitiello B et al. Impairment and deportmentresponses to different methylphenidate doses in children with ADHD:




the MTA Titration Trial. J Am Acad Child Adolesc Psychiatry. 2001;40:180Y187.

27. Vitiello B, Severe JB, Greenhill LL et al. Methylphenidate dosage forchildren with ADHD over time under controlled conditions: lessonsfrom the MTA. J Am Acad Child Adolesc Psychiatry. 2001;40:188Y196.

28. Landis RJ, Koch GG. The measurement of observer agreement forcategorical data. Biometrics. 1977;33:159Y174.

29. Gau SS, Shen HY, Chou MC, Tang CS, Chiu YN, Gau CS. Deter-minants of adherence to methylphenidate and the impact of pooradherence on maternal and family measures. J Child Adolesc Psychophar-macol. 2006;16:286Y297.

30. Rostain dAL, Power TJ, Atkins MS. Assessing parents’ willingness topursue treatment for children with attention-deficit hyperactivity dis-order. J Am Acad Child Adolesc Psychiatry. 1993;32:175Y181.

31. Leslie LK,Weckerly J, Landsverk J, Hough RL, Hurlburt MS,Wood PA.Racial/ethnic differences in the use of psychotropic medication in high-risk children and adolescents. J Am Acad Child Adolesc Psychiatry. 2003;42:1433Y1442.

32. Zito JM, Safer DJ, dosReis S, Magder LS, Gardner JF, Zarin DA.Psychotherapeutic medication patterns for youths with attention-deficit/hyperactivity disorder. Arch Pediatr Adolesc Med. 1999;153:1257Y1263.

33. Kelleher KJ, Moore CD, Childs GE, Angelilli ML, Comer DM. Patientrace and ethnicity in primary care management of child behavior prob-lems: a report from PROS and ASPN. Med Care. 1999;37:1092Y1104.

34. Dosreis S, Zito JM, Safer DJ, Soeken KL, Mitchell JW Jr, Ellwood LC.Parental perceptions and satisfaction with stimulant medication for

attention-deficit hyperactivity disorder. J Dev Behav Pediatr. 2003;24:155Y162.

35. Arnold LE, Elliot M, Sachs L, Bird H, Kraemer HC, Wells KC et al.Effects of ethnicity on treatment attendance, stimulant response/dose,and 14-month outcome in ADHD. J Consult Clin Psychol. 2003;71:713Y727.

36. Horne R, Clatworthy J, Polmear A, Weinman J. Anglo-ScandinavianCardiac Outcomes Trial. Do hypertensive patients’ beliefs about theirillness and treatment influence medication adherence and quality oflife? J Hum Hypertens. 2001;(Suppl 1):S65YS68.

37. Rao D, Kekwaletswe TC, Hosek S, Martinez J, Rodriguez F. Stigma andsocial barriers to medication adherence with urban youth living withHIV. AIDS Care. 2007;19:28Y33.

38. Kazdin AE, Marciano PL, Whitley MK. The therapeutic alliance incognitive-behavioral treatment of children referred for oppositional, ag-gressive, and antisocial behavior. J Consult Clin Psychol. 2005;73:726Y730.

39. Richardson J, Barkan S, Cohen M et al. Experience and covariates ofdepressive symptoms among a cohort of HIV infected women. Soc WorkHealth Care. 2001;32:93Y111.

40. Gifford AL, Laurent DD, Gonzales VM, Chesney MA, Lorig KR. Pilotrandomized trial of education to improve self-management skills ofmen with symptomatic HIV/AIDS. J Acquir Immune Defic Syndr HumRetrovirol. 1998;18:136Y144.

41. Lehane E, McCarthy G. An examination of the intentional and un-intentional aspects of medication non-adherence in patients diagnosedwith hypertension. J Clin Nurs. 2007;16:698Y706.

PAPPADOPULOS ET AL.



AUTHOR QUERY

AUTHOR PLEASE ANSWER QUERY

AQ1 = Please provide caption for Table 1.

END OF AUTHOR QUERY

peter

Sticky Note

Number and Proportion of Subjects with Positive Assays

peter

Sticky Note

Unmarked set by peter

peter

Sticky Note

Completed set by peter

peter

Sticky Note

Accepted set by peter

peter

Sticky Note

MigrationConfirmed set by peter


Author(s) Name

Title of Article

*Article # *Publication Mo/Yr ______

*Fields may be left blank if order is placed before article number and publication

month are assigned.

Name

Address Dept/Rm

City State Zip Country

Telephone

Author Reprints

Journal of the American Academy of Child & Adolescent Psychiatry Order

Payment

Ship to

Quantity of Reprints ____ $

Covers (Optional) $

Shipping Cost $

Reprint Color Cost $

Tax $

Total $

Reprint Pricing100 copies = $375.00

200 copies = $441.00

300 copies = $510.00

400 copies = $585.00

500 copies = $654.00

Covers$108.00 for first 100 copies

$18.00 each add’l 100 copies

Reprint Color ($70.00/100 reprints)

Shipping$5.00 per 100 for orders shipping within the U.S.

$20.00 per 100 for orders shipping outside the U.S.

Tax

U.S. and Canadianresidents add theappropriate tax or submit a tax exempt form.

You may have included color figures in your article. The costs to publish those will be invoiced separately. If your article contains color figures, use Rapid Ordering www.lww.com/periodicals/author-reprints.

� MC � VISA � Discover � American Express

Account # / / Exp. Date

Name

Address Dept/Rm

City State Zip Country

Telephone

Signature

Use this form to order reprints. Publication fees, including color separation charges and page charges will be billed separately, if applicable.

Payment must be received before reprints can be shipped. Payment is accepted in the form of a check or credit card; purchase orders are accepted for orders billed to a U.S. address.

Prices are subject to change without notice.

Quantities over 500 copies: contact our Pharma Solutions Department at410.528.4077

Outside the U.S. call 4420.7981.0700

MAIL your order to: Lippincott Williams & WilkinsAuthor Reprints Dept.351 W. Camden St. Baltimore, MD 21201

FAX:410.528.4434

For questions regarding reprints or publication fees,E-MAIL:[email protected]

OR PHONE:1.800.341.2258

For Rapid Ordering go to: www.lww.com/periodicals/author-reprints

For Rapid Ordering go to: www.lww.com/periodicals/author-reprints

Documents

Medication Adherence in the MTA: Saliva Methylphenidate Samples Versus Parent Report and Mediating Effect of Concomitant Behavioral Treatment