. . .
and artificial food additives.
Becauseofsmallandnonrepresentativesamples,pre-vious studies have been limited in their applicability tothegeneralpopulation.Furthermore,studies
havere-lied on parent- and teacher-reported cases, which over-estimate true prevalence. Accurate estimates of diseaseburden in a large study population will provide infor-mation for determining health care resource allocationforADHDpreventionprograms.Inaddition,researchonthe influence of child race/ethnicity on ADHD trends issparse, and identifying potential disparities in ADHDprevalenceisanimportantstepineliminatinghealthin-equalities. The Kaiser Permanente Southern California(KPSC)systemshavedetailedmedicalrecordsandtreat-ment information for a large number of children in thehealth plan. Using these integrated patient medical rec-ords, we investigated recent trends in ADHD diagnosisfrom 2001 to 2010 by child race/ethnicity, age, sex, andmedianhouseholdincome.Specifically,wetestedthefol-lowing hypotheses: (1) that the diagnosis of ADHD hasincreasedmorerapidlyamongchildrenofracesotherthanwhite during the study period, (2) that the diagnosis of ADHD has increased among children of families withhigher socioeconomic status and children across all agegroups from 5 to 11 years, and (3) that the sex gap inADHD diagnosis has narrowed over time.
DATA SOURCE AND PATIENT SELECTION
For this analysis, demographic and clinical information from2001 to 2010 was obtained from the KPSC medical records,which include information about membership, inpatient andoutpatientphysicianencounters,andpharmacyuse.TheKPSCmembership records contain information about race/ ethnicity, sex, and date of birth. Information on supplementaldrugbenefits,Medicaidstatus,andchangestomembershipanddemographics are also available.The study cohort is composed of member children aged 5to11yearswhoreceivedcareattheKPSCfromJanuary1,2001,through December 31, 2010, regardless of membership statusatthetimethecarewasgiven.Childreninthisagegroupwereselected to maintain uniformity in diagnostic criteria (
Diag-nostic and Statistical Manual of Mental Disorders [Fourth Edi-tion]
]). Children with diagnoses of autism spectrumdisorder (n=15400) were excluded from the final study co-hort (n=842830).Potential confounders and mediators included race/ ethnicity (categorized as non-Hispanic white [white], non-Hispanic black [black], Hispanic, Asian/Pacific Islander, andother/multiple [other] racial/ethnic groups), sex (male or fe-male),age(5-7,8-9,and10-11years),andarea-based(USCen-sustract)medianfamilyhouseholdincome(
$29999,$30000-$49000, $50000-$69999, $70000-$89999, and
$90000).Child race/ethnicity was determined on the basis of maternaland paternal race/ethnicity. The other/multiple (other) race/ ethnicity category includes non-Hispanic children with mul-tiple recorded races.PrimarydiagnosisofADHDwasdeterminedusing
Interna-tional Classification of Diseases, Ninth Revision, Clinical Modi- fication
diagnosis code 314.x from child hospitalization, out-patient office visit, and emergency department visit across allthe KPSC facilities. The outpatient, inpatient, and emergencydepartmentsystemsareintegrated.Duringthechildvisitinanyof these facilities, the practitioner has access to the child’s di-agnosis, but often the diagnosis of ADHD is made in an out-patient setting. The following criteria were used to diagnoseandcodeADHDwithintheKPSC:(1)aChildBehaviorCheck-listmustbefilledoutbyparentsandteacherstodescribechildbehavioral and emotional problems, and (2) a clinical inter-view must be performed by a qualified mental health profes-sional. In a preliminary study conducted for this project, 96%of children with ADHD were found to have had their condi-tions diagnosed by child and adolescent psychiatrists, devel-opmentalandbehavioralpediatricians,childpsychologists,andneurologists.Tofurtherincreasethespecificityofthecasedefi-nition, we used data for receipt of drugs specific to ADHD ex-tractedfrompharmacyrecords.Thesedrugsincludedamphet-amine aspartate, amphetamine sulfate, dextroamphetamineaspartate,dextroamphetaminesulfate,andmethylphenidatehy-drochloride.
First, we compared the distribution of child age, sex, race/ ethnicity,andmedianhouseholdincomebyADHDstatususingthe
tests. Second, the annual rates of ADHD per 100 chil-drenwereestimatedusingPoissonregression.Forthis,theyearlycount of ADHD was the outcome variable and year of diagno-sis was the independent variable, adjusting for potential con-founding factors (
). Dummy variables were con-structed for the various categories of covariates. The meanpredictedprobabilitieswereestimatedfromtheadjustedmod-els. Third, increases of relative risk (RR) from 2001 to 2010were quantified using regression analysis. The population-
Table 1. Distribution of Child Characteristics Basedon ADHD Status
CharacteristicChildren, %With No ADHD(n = 803630)With ADHD(n = 39200)
Child age, y5-7 23.9 16.48-9 17.2 26.610-11 59.0 57.0Child sexFemale 50.2 25.4Male 49.9 74.6Child race/ethnicityNon-Hispanic white 18.3 34.3Non-Hispanic black 8.1 10.2Hispanic 40.8 31.8Asian/Pacific Islander 5.6 2.6Other
2.1 2.5Unknown 25.2 18.6Household income, median, $
30000 5.7 4.030000-49 999 27.6 22.350000-69 999 29.1 28.170000-89 999 19.8 22.8
90000 17.5 22.3Abbreviation: ADHD, attention-deficit/hyperactivity disorder.
Differences between children with and without ADHD by childcharacteristics were statistically significant at
Other race/ethnicity includes non-Hispanic children with multiplerecorded races/ethnicities.
Median household income based on US Census tract information.
JAMA PEDIATR PUBLISHED ONLINE JANUARY 21, 2013 WWW.JAMAPEDS.COM