medication patterns in patients with autism: temporal, regional, and demographic influences

Medication Patterns in Patients with Autism:Temporal, Regional, and Demographic Influences

Michael G. Aman,1 Kristen S.L. Lam,1 and Mary E. Van Bourgondien2

ABSTRACT

To date, there have been relatively few surveys of psychotropic medicine use in individualswith autism. Data were analyzed from three statewide surveys that employed the same ques-tionnaire and survey methodology. The first was done in the Autism Society of North Car-olina in 1992–1993 (NC-1, n = 838; Aman et al. 1995); the second was done in the AutismSociety of Ohio in 1999 (Ohio, n = 417; Aman et al. 2003), and the third was done again in theAutism Society of North Carolina in 2001 (NC-2, n = 1538; Langworthy-Lam et al. 2002). Re-sponse rates ranged from 48%–56%. Longitudinal trends were examined by comparing theNC-1 and NC-2 data, and regional effects were assessed by comparing the NC-2 and Ohiodata. There was a very large increase in antidepressant utilization from 1993 to 2001, with sig-nificant increases also occurring for antipsychotics, psychostimulants, and alpha-agonistsand beta-blockers. Among youths with autism, the use of any psychotropic increased from30.5% in NC-1 to 45.2% in NC-2. Psychotropic medication patterns were remarkably consis-tent across North Carolina and Ohio, except that significantly more autism supplements wereused in Ohio. We also examined subject and demographic variables across studies and foundseveral robust correlates of psychotropic medication use. Greater age and handicap, and morerestrictive placements, were associated with the use of several drug classes. Knowledge ofthese patterns may help families and medical planners anticipate future needs.

116

JOURNAL OF CHILD AND ADOLESCENT PSYCHOPHARMACOLOGYVolume 15, Number 1, 2005Mary Ann Liebert, Inc.Pp. 116–126

1The Nisonger Center, Ohio State University, Columbus, Ohio.2Division TEACCH, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina.This study was supported by funding from contract N01MH80011 from the National Institute of Mental Health to

Dr. Michael Aman, by a Faculty Fellowship Award from the Center for Survey Research and Behavioral Sciences atOhio State University (Columbus, Ohio) to Dr. Michael Aman, and by funding from the Kobacher Foundation.

INTRODUCTION

OVER THE YEARS, drug surveys have beenvery common in the mental retardation

field, with over 46 studies published to date(see Rinck 1998; Singh et al. 1998b). The earlysurveys documented what were, generally,very high rates of medication use. Collectively,these acted as a catalyst for social change, withless reliance on psychotropic medicines inmore recent years (Rinck 1998; Singh et al.

1998b). A relatively small percentage of thesestudies attempted to relate demographic vari-ables to use of medicine, and these were sum-marized by Aman et al. (1995). Severity ofmental retardation has been associated withprescription patterns in both directions (bothmore and less prescription with more severemental retardation). Aman et al. (1995a) noteda large association between clinician-recordedDiagnostic and Statistical Manual of MentalDisorders, 4th edition (DSM-IV) psychiatric

13914C16.pgs 2/17/05 10:10 AM 16

diagnoses and level of mental retardation,with a much greater tendency to give diag-noses at higher functional levels. This mayhelp to explain some of the inconsistencies be-tween the severity of mental retardation andmedication prescribing, as there may be greaterpressure to assign a DSM-IV diagnosis, insome settings than others, before prescribingdrugs. Linkage with age has been inconsistent,with a group of studies showing more pre-scription with age and others showing nopattern. Generally, studies have found noassociations between drug utilization and gen-der and ethnicity in treated populations; psy-chotropic drug utilization was lower for subjectswith epilepsy, substantial visual impairment,impaired ambulation, and the presence of cere-bral palsy (Aman et al. 1995a).

Medication surveys are relatively newamong patients with autism-spectrum condi-tions, with the first report (a survey conductedby newsletter) being reported by Rimland(1988). Another investigation (a clinic-basedsurvey of 109 high-functioning patients [withIQs � 70] seen in a center for autism and re-lated conditions) reported that 55% of patientswere taking at least one psychotropic drug(Martin et al. 1999). The most commonly-prescribed agents were antidepressants (32.1%),stimulants (20.2%), antipsychotics (16.5%),and mood stabilizers (7.2%). In this article, wecompared three surveys among the membersof two states’ Autism Society (AS) member-ships (Aman et al. 1995b; Langworthy-Lam etal. 2002; Aman et al. 2003). The results of thesesurveys will be elaborated upon further in theresults and discussion sections.

All three of the surveys by the authors em-ployed the same questionnaire. The Aman etal. (1995b) survey was conducted late in 1992and early in 1993 among the AS membershipof North Carolina. The Langworthy-Lam et al.(2002) study, conducted in 2001, also surveyedthe AS membership of North Carolina. TheAman et al. (2003) survey was conductedamong the AS membership of Ohio in 1999.Thus, one of these surveys was done beforethe widespread introduction of the atypicalantipsychotics and the selective serotonin re-uptake inhibitors (SSRIs) (Aman et al. 1995b),whereas two were done well after these agents

were on the market (Langworthy-Lam et al.2002; Aman et al. 2003). The state of NorthCarolina has a reputation for providing excep-tional service for individuals with autism, andNorth Carolina is one of only two states with alegislative mandate and state appropriationsfor statewide services specifically designed forchildren and adults with autism (NC State Bill383, Chapter 1007, 1971). Ohio is more typicalof the 50 states, and services must be obtainedfrom county boards for mental retardation anddevelopmental disabilities and from educationboards on a case-by-case basis.

The three surveys enabled triangulation ontwo important questions. The first concernedthe effects of time. The introduction of newerantipsychotics and antidepressants (as well asover-the-counter compounds, such as mela-tonin and St. John’s wort) may have had at leasttwo possible consequences. Their introductionmay have: (a) resulted in greater overall use ofpsychotropic agents or (b) encouraged physi-cians to use safer and/or more effective agentsto replace older ones but at the same rate. The hy-pothesis addressed in this study is that therewould be no change in drug utilization overtime (Aman et al. 1995b; Langworthy-Lam et al.2002) but that physicians’ choice of agentswould have changed over time owing to a pref-erence for newer agents. The second hypothesiswas that current medication utilization wouldbe significantly higher in Ohio (Aman et al.2003) than in North Carolina (Langworthy-Lamet al. 2002) because of speculation that patientswould have better access to psychoeducationalservices in North Carolina, which might allevi-ate some of the target symptoms presented bythese children. Finally, the three previous publi-cations also looked at demographic correlatesof drug utilization. Therefore, these correlateswere systematically analyzed for consistencyover time and across regions.

METHODS

Survey form

The same survey form was used for all threeinvestigations, with the exception that a ques-tion on ethnicity was added for the second and

MEDICATION IN PATIENTS WITH AUTISM 117

13914C16.pgs 2/17/05 10:10 AM Page 117

third investigations. The questionnaires had: (a)a demographic section, including type of schooland work placement; (b) a part on medicationsused, associated side-effects, and consumer sat-isfaction; (c) presence of epilepsy, severity ofepilepsy if present, type of epilepsy, and diag-nostic tests performed; and (d) parents’ educa-tional level. The printed survey form was 31⁄2pages long; copies are available on request fromthe senior author. The following operationaldefinition was used to classify medications as“anti-epileptic drugs” (AEDs) or “mood stabi-lizers:” If prescribed to manage epilepsy,AED/mood stabilizers like divalproex sodium,carbamazepine, and gabapentin were classed asAEDs. If no epilepsy was reported, such agentswere classed as mood stabilizers (as was lithiumcarbonate at all times).

Procedures for mailings

Mailing labels for each state’s AS member-ship were provided free of charge by the Execu-tive Committees of the AS of both states. Up tothree mailings were sent to the family of eachrespondent. If the family failed to respond tothe first mailing, a second mailing was sent, andfailure to respond to this led to a third (and last)mailing. The cover letters indicated the purposeof the survey and the importance of obtaining arepresentative sample, the fact that all re-sponses were confidential, and our intention toshare the findings with respondents throughthe state’s AS newsletter. In an effort to boostthe response rate, the cover letters indicatedthat there would be a random drawing for a$200 reward among the respondents (done forthe last two surveys only).

Respondents

Three sets of subjects responded, describedhere as the NC-1, NC-2, and Ohio samples.Thorough descriptions of the survey proce-dures and samples are provided in Aman et al.(1995b), Langworthy-Lam et al. (2002), andAman et al. (2003), respectively.

North Carolina 1. Of a target sample of 1595families, there were 859 responses (53%).Many features of the sample, including age,

severity of autism, level of mental retardation,and parents’ education are reported in Table 1.There were 681 males (82.0%) and 149 females(18.0%). In all, 19.2% of subjects had epilepsy.Ethnic group was not queried in this survey.

North Carolina 2. Of 3228 members of AS ofNorth Carolina, 1538 families (48%) responded.Demographic features are reported in Table 1.There were 1268 males (82.4%) and 270 fe-males (17.6%). Table 1 indicates functional lev-els were rather high for young people withautism (nearly 45% were reported to have anormal IQ). However, of those youths who at-tended school, 27.1% attended regular classes,62.4% attended special classes, and 10.5% at-tended special schools for children with devel-opmental disabilities. This suggests that manyrespondents were inaccurate about functionallevel and tended to overestimate the IQs ofthese individuals. Ethnic composition was asfollows: Caucasian, n = 1101 (71.6%); African-American, n = 283 (18.4%); Hispanic, n = 25(1.6%); Asian, n = 22 (1.4%); other or not re-ported, n = 107 (7.0%).

Ohio. Of 747 member families of AS Ohio havingmembers with autism, 417 responses (56%)were obtained. There were 340 males (81.5%)and 74 females (17.7%), with missing data forthree subjects (0.7%). Once again, a sizable pro-portion reported average or near-average IQ(see Table 1), but the figures for educationalplacement again suggested a higher rate ofhandicap. Of those youths who attended school,109 (32.2%) attended regular classes, 176 (52.1%)attended special classes, and 53 (15.7%) at-tended special schools. Ethnicity was as follows:Caucasian, n = 391 (93.8%); African-American, n= 5 (1.2%); Asian, n = 5 (1.2%); Hispanic, n = 4(1.0%); and Other or missing, n = 12 (2.9%).

RESULTS

Temporal effects on medication use(NC-1 versus NC-2)

The medication utilization rate for the firstand second North Carolina samples are com-pared in Table 2, using Z-tests for propor-

118 AMAN ET AL.

13914C16.pgs 2/17/05 10:10 AM Page 118

tions. It is clear that the rate of prescribingincreased over time (NC-1 versus NC-2) forall frequently prescribed classes, includingthe antipsychotics (p < 0.01), antidepressants(p < 0.001), psychostimulants (p < 0.001), andalpha-agonists/beta-blockers (p < 0.001). Therate of use actually declined slightly for hyp-notics and anxiolytics (p < 0.05). As wouldbe expected, measuring the broader categoryof “any psychotropic” through “all classes”combined (p < 0.001) also increased acrosstime. The prevalence of “any psychotropic”rose from 30.5% in 1993 to 45.2% in 2001, a48% increase. Medications showing no in-crease across time included the mood stabiliz-ers, opiate blockers, antiparkinsonians, andautism supplements (vitamin B6, dimethyl-glycine, and so forth).

Among the antipsychotics within the NC-2sample, the most commonly prescribed med-ications were risperidone (62% of antipsy-chotic users), olanzapine (17%), and thioridazine(7%). Within the earlier NC-1 sample, themost common antipsychotics were thiori-dazine (48% of antipsychotic prescriptions),haloperidol (30%), and chlorpromazine (10%).The most frequently prescribed antidepres-sants in the NC-2 sample were fluoxetine(31% of antidepressant prescriptions), sertra-

line (15%), and paroxetine (15%). In the NC-1sample, the most common antidepressantswere clomipramine (37%), imipramine (22%),and fluoxetine (16%) (which had only re-cently been released on the market). Othersubstantial changes from NC-1 to NC-2, bro-ken out by drug class, and in rank order,were as follows: (a) Hypnotics/anxiolytics:Chloral hydrate (19%), buspirone (17%), andhydroxyzine (17%) were most common inNC-1, whereas melatonin (40%), buspirone(26%), and lorazepam (13%) were prevalentin NC-2; (b) alpha agonists/ beta-blockers:Clonidine (68%) and propranolol (32%) wereprevalent in NC-1, whereas clonidine (67%)and guanfacine (26%) were most used in NC-2; and (c) stimulants: Immediate-release methyl-phenidate (76%), dextroamphetamine (16%),and pemoline (7%) were most frequent inNC-1, whereas methylphenidate (36%), mixedamphetamine salts (Adderall; 36%), andmethylphenidate SR (16%) were the mostcommon formulations in NC-2. The onlysubstantial change in the use of AEDs was ashift from phenytoin (14%) as the third mostused agent in NC-1 to gabapentin (11%) inNC-2. Carbamazepine and sodium valproatewere the most commonly prescribed AEDs inboth surveys.


TABLE 1. SUBJECT AND FAMILY CHARACTERISTICS

Variable NC1 NC2 Ohio

Age Range (years) 1–82 3–56 2–46Mean Age (years) 15.96 15.62 13.24Severity of Autism (%)

Mild 32.4 44.5 35.5Moderate 34.1 34.1 44.8Severe/Profound 18.7 15.2 13.9Unknown 14.7 6.3 5.8

Level of Mental Retardation (%)Normal IQ 22.1 44.9 52.3Mild MR 19.8 17.6 11.3Moderate MR 22.2 16.8 12.7Severe/Profound 18.9 10.8 6.5Unknown 17.1 10.0 17.3

Parents’ Education F/M (%)High School or Less 31/30 23/21 17/17Tech. School/Some College 25/30 23/27 22/25College Degree 44/40 49/50 59/57Not Reported —/— 5/2 2/1

F, fathers; M, mothers.

13914C16.pgs 2/17/05 10:10 AM Page 119

Regional effects (NC-2 versus Ohio)

There was only one regional difference inagents used, this being the rate of “autism sup-plements,” which was 5.7% for NC-2 and10.3% for Ohio (p < 0.001) (see Table 2). Themost commonly used supplements (usually inmegadose) were vitamin B6, dimethylglycine,Super Nu-Thera, and DMAE (dimethylamino-ethanol). In fact, there was remarkable similar-ity in the rate of use for all major drug classes(antipsychotics, antidepressants, mood stabi-lizers, stimulants, hypnotics/anxiolytics, andalpha agonists/ beta-blockers), with a meandifference of 1.4% across states. The utilizationrate of “any psychotropic” in Ohio was 45.6,not appreciably different from the rate of 45.2in NC-2.

Clinical and demographic factorsand medication use

All three of the statewide surveys includedanalyses for associations between subject vari-ables and medication use. The subject vari-ables examined included the following: Age

(0–6, 7–13, 14–20, and 21+ years); degree of au-tism (mild, moderate, severe); level of mentalretardation (not present, mild, moderate, se-vere/profound); type of housing (living withfamily, partially sheltered, other); educationalsetting (regular class, special class in regularschool, special school); and work placement(sheltered workshop, supported/regular em-ployment). These subject variables were com-pared with medication use (for each medicineclass) by means of cross tabs and chi-squareanalyses.

The results of these comparisons for all ofthe studies (p � 0.01) are presented in Table 3.The letter “D” signifies that the relationshipwas direct (i.e., greater drug use with greaterage, more severe condition, or more restrictiveplacement). The letter “N” indicates that therelationship was negative, and “C” indicates acurvilinear relationship (in all cases, highermedication use in middle childhood than inearly childhood or adulthood). The samplesizes (838 and 1538) were substantially largerfor the two North Carolina surveys than forthe Ohio survey (n = 417). Thus, there wasgreater power in the North Carolina analyses

120 AMAN ET AL.

TABLE 2. PERCENTAGE OF SUBJECTS TAKING MEDICATION CLASSES ON DATE OF THE SURVEYS: TEMPORAL AND GEOGRAPHIC DIFFERENCES

Z-scores across Z-scores acrossType of agent NC-1 (1993) NC-2 (2001) Ohio (1999) time region

Antipsychotics 12.2% 16.5% 14.9% 2.80** 0.93Antidepressants 6.1% 21.4% 21.6% 9.75*** 0.03Mood Stabilizers 3.9% 4.9% 4.5% 1.00 0.36Stimulants 6.6% 13.8% 11.3% 5.35*** 1.40Hypnotics/Anxiolytics 6.3% 7.3% 8.7% 2.08* 1.94Alpha-agonists and Beta-blockers 4.4% 9.6% 12.5% 4.36*** 1.78Opiate Blockers 0.4% 0.9% 0.0% 0.16 1.93Antiparkinsonians 1.3% 2.9% 2.6% 0.46 0.32Autism Supplements 5.0% 5.7% 10.3% 0.70 3.44***Any Psychotropic 30.5% 45.2% 45.6% 6.99*** 0.05Antiepileptic Drugs (AEDs) 13.2% 12.5% 11.5% 0.65 0.50Psychotropic Drugs + 38.9% 50.2% 51.6% 5.26*** 0.38

AnticonvulsantsPsychotropic Drugs + Autism 33.8% 48.6% 51.6% 6.97*** 1.19

SupplementsPsychotropic Drugs + AEDs + Autism 42.0% 53.5% 55.4% 5.21*** 0.85

SupplementsMiscellaneous Other 12.1% 30.5% 31.2% 10.05*** 0.45Any Psychotropic or Other 53.3% 64.4% 65.0% 4.74*** 0.53

*p < 0.05; **p < 0.01; ***p < 0.001.

13914C16.pgs 2/17/05 10:10 AM Page 120

and more significant findings noted. The oddsof obtaining two successive p � 0.01 findingsin a row by chance alone are �1 in 10,000, andthe odds of obtaining three successive p � 0.01are �1 in 1,000,000.

The significant associations with antipsy-chotic medicine were remarkably consistent

across studies. Greater age, more severe au-tism and mental retardation, residence outsidethe family home, and more restrictive educa-tional placement were all associated with agreater likelihood of receiving antipsychoticdrugs. More restrictive work placement wasassociated with a greater use of antipsychotics


TABLE 3. DEMOGRAPHIC VARIABLES RELATED TO MEDICATION USE (P � 0.01)

Medication and variables NC-1 NC-2 Ohio

AntipsychoticsAge D D DSeverity of Autism D D DLevel of MR D D DType of Housing D D DEducational Placement D D DWork Placement D — —

AntidepressantsAge D D DType of Housing D D —Father’s Education — D —Ethnicitya — N —

Mood StabilizersAge D D DSeverity of Autism — — DLevel of MR — — DType of Housing D D D

StimulantsAge C C —Severity of Autism — N —Level of MR — N —

Medication and VariablesType of Housing — N —Educational Placement — N —

Sedatives/AnxiolyticsAge — D —Level of MR — D —Type of Housing D D DWork Placement D — —

Antiepileptic DrugsAge C D —Severity of Autism D D —Level of MR D D DType of Housing D D —Educational Placement — D —

Autism SupplementsAge N N —Mother’s Education D D —Father’s Education — D —

“D” refers to a direct relationship, “N” refers to a negative relationship, and“C” denotes a curvilinear relationship. A dash indicates no correlation. “MR”refers to mental retardation. Direction of relationship was as follows: D,greater age, greater severity of autism, more severe mental retardation, morerestrictive housing, more restrictive educational placement, member of ethnicminority, greater education for father or mother.

aNot included in NC–1 survey.

13914C16.pgs 2/17/05 10:10 AM Page 121

in the NC-1 survey only. With regard to antide-pressants, greater use was associated withgreater age in all surveys, and more restrictivehousing in the two North Carolina surveys. Inthe NC-2 survey only, higher paternal educa-tion, ethnic majority status, and female gender(not tabulated) were associated with greateruse (this was the only significant gender find-ing in any of the surveys).

Henceforth, all details of the associationsare not enumerated, but the broad strokesof what occurred are provided; readers arereferred to Table 3 for precise details. Mood-stabilizer use (as defined in the Methodssection above) was associated with greaterage, more restrictive housing (consistent acrossall surveys), and more severe handicap (Ohioonly). The use of stimulants was linked tomiddle childhood (both North Carolina sur-veys) and was negatively correlated withgreater handicap (autism or mental retarda-tion) or restrictive placements.

Use of sedative or anxiolytics was linkedwith more restrictive housing (all surveys), aswell as greater age, more severe retardation,and restrictive work arrangements. Anticon-vulsant drug use was tied to age (peaking latein the adolescent and early in the adulthoodrange), more severe handicap, and more re-strictive placements. Finally, the use of autismsupplements was tied to younger age andhigher parental education.

DISCUSSION

Temporal effects on medication use

It is clear that the hypothesis that use of med-ication would hold constant over time (i.e.,1992–1993 to c.1999–2001) was proven quitewrong. Almost all drug classes, with the excep-tion of mood stabilizers, opiate blockers, anti-convulsants, and autism supplements weremore frequently prescribed in the NC-2 than in

the NC-1 survey. Similar increases have beenobserved in other populations as well, such as:(a) child and adolescent Medicaid recipients inTexas (Patel et al. 2002), (b) children served byMedicaid in two states and children in a group-model health-maintenance organization (Zito etal. 2003), and (c) Ontario Drug Benefit recipi-ents (Dewa et al. 2002). All of these investiga-tors noted major increases in drug utilization,which was often attributed to the introductionof SSRIs and the atypical antipsychotics. Thepresent surveys also showed increases in theuse of antipsychotics and, especially, in the useof antidepressants, which increased 3.5-fold.However, there were also sizable increases inthe use of psychostimulants and alpha-ago-nists/beta-blockers (both of which more thandoubled). These data do not permit judgmentson appropriateness of use (Walkup 2003).

Although robust, the 35% increase in the useof antipsychotics was the result, in part, ofboth an expanded use and a transition fromclassical to atypical antipsychotics. In 1993 (NC-1), almost all use of antipsychotics (�93%)1

was confined to the classical agents, whereas inthe latter surveys (NC-2 and Ohio) most pre-scriptions (83.7% and 85.5%, respectively) werefor atypical agents. Clearly, physicians havebegun to switch antipsychotic agents for theirpatients from the older to the newer agents, atrend noted with other populations as well(Zito et al. 2003; Patel et al. 2002; Dewa et al.2002). As the atypical antipsychotics have lesstendency to cause extrapyramidal side-effects(including tardive dyskinesia), this may wellbe a good development. However, there are in-creasing concerns about weight gain, diabetesmellitus, and (in some cases) hyperprolactine-mia, so one should not be nonchalant aboutthis transition. The recent report of risperi-done’s effectiveness in children with autismand highly irritable behavior (Research Unitsin Pediatric Psychopharmacology 2002) in-dicated clinical improvement well beyondanything previously seen with classical anti-

122 AMAN ET AL.

1We no longer have the breakdown for all antipsychotics from the NC-1 study. However, we do know that thiori-dazine, haloperidol, and chlorpromazine collectively accounted for 93% of prescriptions. As clozapine was the onlyatypical agent available at the time of NC-1, and given the precautions necessary to protect against agranulocytosis, itis highly unlikely that even 3% of those taking antipsychotics were receiving atypical agents.

13914C16.pgs 2/17/05 10:10 AM Page 122

psychotics (e.g., Anderson et al. 1984; Ander-son et al. 1989). This suggests greater benefitwith new antipsychotics for autism. There isa sizable (although fairly unsophisticated) liter-ature suggesting that the atypical antipsychoticsimprove a variety of target symptoms, in-cluding self-injury, aggression, destructiveness,stereotypic behavior, and obsessions (Aman andMadrid 1999; McDougle et al. 2000). Thus, someof the increase in antipsychotic use may be theresult of not only an improved tolerability pro-file but an improvement in behavioral symp-toms as well.

The increased use of antidepressants was themost notable temporal change seen in thisstudy, and the SSRIs were at the forefront of thistrend. Whereas clomipramine (a serotonergictricyclic agent) was the leading antidepressantin 1992–1993 and the SSRIs assumed preemi-nence in the subsequent surveys (NC-2 andOhio), the finding strongly suggests that thesemedicines were often being used for manage-ment of perseverative behavior (compulsions,self-injury, stereotypic behavior). Such behav-iors can interfere with learning, are very stig-matizing, and can lead to behavioral upheavalif interfered with. There is a very sizable (andagain largely unsophisticated) literature sug-gesting that these agents may ameliorate suchperseverative behavior (Aman et al. 1999; Gor-don et al. 1993).

The other two major increases occurredwithin the alpha-agonist/beta-blocker classesand psychostimulants, and are hard to ex-plain. Among the beta-blockers, the formeruse of propranolol was largely displaced byguanfacine. Both clonidine and guanfacinehave been advocated for the treatment of hy-peractivity and tics (Newcorn et al. 2003), al-though in the authors’ clinical studies theyhave often been used for treating sleep-onsetinsomnia. Zito et al. (2003) also commentedon the marked rise in alpha-agonists, and it ispossible that the use of these agents in autismis merely reflecting what is happening inmore general populations of children. A newdelivery system for methylphenidate (the os-motic pump system; Concerta, Alza, Moun-tain View, CA) was released about the time ofNC-2, but it is unlikely that there was enoughtime for this to have much impact on stimu-

lant use. The increment may have been the re-sult of enhanced awareness by families and/orphysicians about the extent of ADHD symp-toms in children with autism.

Finally, it is worth commenting on two cate-gories of medicine that did not increase overtime, namely mood stabilizers and AEDs.Agents such as carbamazepine were classifiedas “mood stabilizers” (rather than AEDs)by ruling out epilepsy in these individuals.Several newer AEDs, such as gabapentin, lam-otrigine, and topiramate appear to have mood-stabilizing properties (Davanzo and McCracken2003). Although these provide physicians witha wider array of choices, these do not appearto have had much impact yet within the popu-lation with autism. The utilization of AEDs(for treatment of epilepsy) remained steadyover the three surveys. This seems predictable,as the manifestations of epilepsy are reason-ably objective, and one would expect themto be consistent from year to year in patientswith autism.

Regional effects

The second hypothesis, that prescribingwould be higher in Ohio than in North Carolina,was not borne out by the data. The only classof agents to show regional differences was theautism supplements, which included vitaminB6, magnesium, and dimethylaminoethanol(DMAE). The autism supplements have re-ceived insufficient evidence of benefit in clinicalstudies (Singh et al. 1998a), and their use issometimes encouraged by advocacy groups andparents. It is reasonable to assume that the infor-mation traded in regard to these supplementsvaries from place to place and reflects local con-sumer interests at the time.

However, the real lesson from the regionalcomparisons (NC-2 and Ohio) was the remark-able similarity across regions, with a mean dif-ference (irrespective of direction) of 1.4% overall categories of psychotropic medicine. Thissuggests that there was considerable consis-tency among physicians in the ways that theytreated patients with autism, and it may because for some optimism that these agents arebeing employed systematically.


13914C16.pgs 2/17/05 10:10 AM Page 123

Clinical and demographic factors

The most obvious correlate of medicationuse, across several medicines, was the subject’sage. Of course, greater age could be associatedwith more obvious psychiatric symptoms asthe individual grows older and, hence, lead toDSM-IV-disorder-specific treatment. However,greater age is also associated with greaterphysical size, more responsibilities, and, forsome in these surveys, passage through puberty(the “terrible teens”). Martin et al. (1999) alsofound that age was directly associated withthe prevalence of “any psychotropic” drug,and, specifically, for antipsychotics, antide-pressants, anxiolytics, and mood stabilizers. Inaddition to age, greater severity of autism andmental retardation and more restrictive place-ments (educational, housing, and work) werevariably associated with more antipsychotics,antidepressants, and mood stabilizers. To statethe obvious, all of these subject variables prob-ably reflect a greater degree of functionaland/or behavioral handicap. It has long beenknown that mental retardation is associatedwith a much higher rate of psychiatric andemotional problems than found in the generalpopulation (Stark et al. 1987). It should comeas no great surprise that greater level of handi-cap tends, it appears, to be associated withhigher rates of behavioral and psychiatricproblems.

For AEDs, the association with age is consis-tent with adolescence and early adulthoodbeing a period of risk. Even the direct “linear”relationship for NC-2 was somewhat consis-tent with this, as the inflated use in the oldestgroup tended to occur among younger indi-viduals within that age bracket (prevalence ofepilepsy 28.2% for age 29 years and younger,18.1% for age 30 years and higher). The com-mon factor linking severity of mental retarda-tion, severity of autism, and higher use ofAEDs is probably the presence of greater CNSdysfunction. For example, the rates of AEDs inthe NC-2 survey were 4.7%, 11.6%, 22.7%, and34.8% for individuals with normal IQ, andmild, moderate, and severe mental retarda-tion, respectively.

Finally, the curvilinear relationship betweenage and psychostimulants appears to reflect

the fact that ADHD is most prevalent andproblematic (or at least most visible) in middlechildhood (American Psychiatric Association1994). The data from NC-2 also indicatedan inverse relationship between severity ofautism and mental retardation and more re-strictive placements on the one hand and pre-scription of psychostimulants on the other.Elsewhere, Aman (1982; 1996) and Aman andLangworthy (2000) argued that the stimulantsappear to work less well (or have more hetero-geneous effects) in children with more severemental retardation and in youngsters with au-tism than in the general population. The NC-2data seem to support this view, in that fewerstimulants were employed as autism and func-tional handicap became more severe.

Correspondence with researchin mental retardation

Rinck (1998) reviewed the literature on psy-chotropic drug prevalence in people withmental retardation. Since 1987, the rates inthe community have fallen in the range of19%–45% for psychotropic medications withoutanticonvulsants and 27%–50% for psychotropicagents and anticonvulsants combined. Withininstitutions over the same period, the rates havebeen in the range of 19%–35% for psychotropicdrugs without anticonvulsants and 37%–44%for both psychotropic and anticonvulsants com-bined. Most of these surveys involved adoles-cents and, especially, adults, so the rates appearto be lower than those observed here. However,there are no recent comparable surveys amongindividuals with mental retardation, and therates of prescription for mental retardation maybe climbing as well.

CONCLUSIONS

It is obvious that the use of psychotropicmedications for autism grew substantiallyover the 8-year period surveyed. As with otherclinical populations, the dollar cost associatedwith this increase is substantial. For example,within the Ontario Drug Benefit population,the number of claimants increased approxi-mately 25% between 1992 and 1998 (Dewa

124 AMAN ET AL.

13914C16.pgs 2/17/05 10:10 AM Page 124

et al. 2002), while expenditures for antipsy-chotics in a Texas Medicaid population grewby approximately 325% (Patel et al. 2002). Thishas obvious implications for families caringfor a member with autism and for medicalplanners.

The findings reveal a number of subject anddemographic variables that reliably correlatewith greater or lesser use of psychotropic med-ications. These can be helpful for families andplanners to anticipate future medication needs.This knowledge is clearly helpful, but thecrudeness of these correlates must be acknowl-edged. Hopefully, investigators in the futurewill be able to examine the utility of more so-phisticated indices, such as the actual charac-terization of autism (as revealed in structureddiagnostic instruments like the Autism Diag-nostic Interview) (Lord et al. 1994), receiptof psychoeducational intervention (such asapplied behavior analysis and parent manage-ment training), and rating scale profiles. Also,through refined cohort analyses, researchersmay be better able to identify what types ofautistic individuals will need medication and/or other services in the future.

REFERENCES

Aman MG: Stimulant drug effects in developmen-tal disorders and hyperactivity: Toward a resolu-tion of disparate findings. J Autism Dev Disord12:385–398, 1982.

Aman MG: Stimulant drugs in the developmentaldisabilities revisited. J Dev Phys Disab 8:347–365,1996.

Aman MG, Arnold LE, Armstrong SC: Review ofserotonergic agents and perseverative behaviorin patients with developmental disabilities. MentRetard Dev Disab Res Rev 5:279–289, 1999.

Aman MG, Lam KL, Collier-Crespin A: Prevalenceand patterns of use of psychoactive medicinesamong individuals with autism in the Autism So-ciety of Ohio. J Autism Dev Disord 33:527–534,2003.

Aman MG, Langworthy KS: Pharmacotherapy forhyperactivity in children with autism and otherpervasive developmental disorders. J AutismDev Disord 30:451–459, 2000.

Aman MG, Madrid A: Atypical antipsychotics inpersons with developmental disabilities. MentRetard Dev Disab Res Rev 5:253–263, 1999.

Aman MG, Sarphare G, Burrow W: Psychoactivedrugs in group homes: Prevalence and relation to

demographic/psychiatric variables. Am J MentRetard 99:500–509, 1995a.

Aman MG, Van Bourgondien ME, Wolford PC,Sarphare G: Psychotropic and anticonvulsantdrugs in subjects with autism: Prevalence andpatterns of use. J Am Acad Child Adolesc Psychi-atry 34:1672–1681, 1995b.

American Psychiatric Association: Diagnostic andStatistical Manual of Mental Disorders, 4th ed.(DSM-IV). Washington (DC): American Psychi-atric Association, 1994.

Anderson LT, Campbell M, Adams P, Small AM,Perry R, Shell J: The effects of haloperidol on dis-crimination learning and behavioral symptomsin autistic children. J Autism Dev Disord 19:227–239, 1989.

Anderson LT, Campbell M, Grega DM, Perry R,Small AM, Green WH: Haloperidol in the treat-ment of infantile autism: Effects on learning andbehavioral symptoms. Am J Psychiatry 141:1195–1202, 1984.

Davanzo P, McCracken J: Mood stabilizers: Lithiumand anticonvulsants. In: Pediatric Psychopharma-cology: Principles and Practice. Edited by MartinA, Scahill L, Charney DS, Leckman JF. New York:Oxford University Press, 2003, pp. 309–327.

Dewa CS, Goering P: Lessons learned from trendsin psychotropic drug expenditures in a Canadianprovince. Psychiatr Serv 52:1245–1247, 2001.

Dewa CS, Remington G, Herrmann N, Fearnley J,Goering P: How much are atypical antipsychoticagents being used, and do they reach the popula-tions who need them? A Canadian experience.Clin Ther 24:1466–1476, 2002.

Gordon C, State R, Nelson J, Hamburger S, RapoportJ: A double-blind comparison of clomipramine, de-sipramine, and placebo in the treatment of autisticdisorder. Arch Gen Psychiatry 50:441–447, 1993.

Langworthy-Lam KL, Aman MG, Van Bourgon-dien ME: Prevalence and patterns of use of psy-choactive medicines in individuals with autismin the Autism Society of North Carolina. J ChildAdolesc Psychopharmacol 12:311–321, 2002.

Lord C, Rutter M, LeCouteur A: Autism DiagnosticInterview—Revised: A revised version of a diag-nostic interview for caregivers of individualswith pervasive developmental disorders. J Au-tism Dev Disord 24:659–685, 1994.

Martin A, Scahill L, Klin A, Volkmar F: Higherfunctioning pervasive developmental disorders:Rates and patterns of psychotropic drug use. JAm Acad Child Adolesc Psychiatry 38:923–931,1999.

McDougle CJ, Scahill L, McCracken JT, Aman MG,Tierney E, Arnold LE, Bell-Bradshaw F, FreemanBJ, Gates-Ulanet P, Martin A, McGough JJ, PoseyDJ, Ritz L, Swiezy NB, Vitiello B, Volkmar FR,Walson P: Research Units on Pediatric Psycho-pharmacology (RUPP) Autism Network: Back-ground and rationale for an initial controlled


13914C16.pgs 2/17/05 10:10 AM Page 125

study of risperidone. In: Child and AdolescentPsychiatric Clinics of North America. Edited byMartin A, Scahill L. Philadelphia (Pennsylvania):WB Saunders Co., 2000, vol. 9, pp. 201–224.

Newcorn JH, Schultz KP, Halpern JM: Adrenergicagonists: Clonidine and guanfacine. In: PediatricPsychopharmacology: Principles and Practice.Edited by Martin A, Scahill L, Charney DS, Leck-man JF. New York: Oxford University Press, 2003,pp. 264–273.

Patel NC, Sanchez RJ, Johnsrud MT, Crismon ML:Trends in antipsychotic use in a Texas Medicaidpopulation of children and adolescents: 1996–2000. J Child Adolesc Psychopharmacol 12:221–229, 2002.

Research Units on Pediatric PsychopharmacologyAutism Network: A double-blind, placebo-con-trolled trial of risperidone in children with autis-tic disorder. NEJM 347:314–321, 2002.

Rimland B: Controversies in the treatment of autis-tic children: Vitamin and drug therapy. J ChildNeurol 3:S68–S72, 1988.

Rinck C: Epidemiology and psychoactive medica-tion. In: Psychotropic Medications and Develop-mental Disabilities: The International ConsensusHandbook. Edited by Reiss S, Aman MG. Colum-bus (Ohio): Ohio State University Nisonger Cen-ter, 1998, pp. 31–44.

Singh NN, Ellis CR, Mulick JA, Poling A: Vitamin,mineral, and dietary treatments. In: PsychotropicMedications and Developmental Disabilities: TheInternational Consensus Handbook. Edited by

Reiss S, Aman MG. Columbus (Ohio): Ohio StateUniversity Nisonger Center, 1998a, pp. 311–320.

Singh, NN, Ellis CR, Wechsler HA: Psychopharma-coepidemiology in mental retardation. J ChildAdolesc Psychopharmacol 7:255–266, 1998b.

Stark JA, Menolascino FJ, Albarrelli MH, Gray VC(eds.): Mental Retardation and Mental Health:Classification, Diagnosis, Treatment, Services. NewYork: Springer-Verlag, 1987.

Walkup JT: Increasing use of psychotropic medica-tions in children and adolescents: What does itmean? J Child Adolesc Psychopharmacol 13:1–3,2003.

Zito SM, Safer DJ, dosReis S, Gardner JF, MagderL, Soeken K, Boles M, Lynch F, Riddle MA: Psy-chotropic practice patterns for youth. A 10-yearperspective. Arch Ped Adolesc Med 157:17–25,2003.

Address reprint requests to:Michael Aman, Ph.D.The Nisonger Center

Room 175McCampbell Hall

Ohio State University1581 Dodd Drive

Columbus, OH 43210-1257

E-mail: [email protected]

126 AMAN ET AL.

13914C16.pgs 2/17/05 10:10 AM Page 126

medication patterns in patients with autism: temporal, regional, and demographic influences

Documents