Small Geographic Area Variations in PrescriptionDrug Use

WHAT’S KNOWN ON THIS SUBJECT: Prescribing patterns in theUS pediatric population are changing but not uniformly. A detailedexamination of prescription variation is needed to betterunderstand pharmacotherapy of children and to inform futureexploration of the causes and consequences of diverse practices.

WHAT THIS STUDY ADDS: We examine pediatric pharmacotherapyand quantify payer type differences and small geographic areavariation. Substantial payer-type differences and regionalvariations were found, likely reflecting local practice cultures.Variation was greatest for medications used in situations ofdiagnostic and therapeutic uncertainty.

abstractBACKGROUND: Despite the frequency of pediatric prescribing little isknown about practice differences across small geographic regions andpayer type (Medicaid and commercial).

OBJECTIVE: The goal of this research was to quantify variation in pre-scription drug use among northern New England children.

METHODS: Northern New England, all-payer administrative data (2007–2010) permitted study of prescriptions for 949 821 children ages 0 to 17years (1.75 million person-years [PYs]; 54% Medicaid, 46% commercial).Age- and gender adjusted overall and drug group–specific prescriptionuse was quantified according to payer type (Medicaid or commercial)and within payer type across 69 hospital service areas (HSAs). We mea-sured prescription fills per PY (rate) and annual, mean percentage of thepopulation with any drug group–specific fills (prevalence).

RESULTS: Overall mean annual prescriptions per PY were 3.4 (commer-cial) and 5.5 (Medicaid). Generally, these payer type differences weresmaller than HSA-level variation within payer type. HSA-level rates ofattention-deficit/hyperactivity disorder drug use (5th–95th percentile)varied twofold in Medicaid and more than twofold in commercially in-sured children; HSA-level antidepressant use varied more than twofoldwithin each payer type. Antacid use varied threefold across HSAs and washighest in infants where commercial use paradoxically exceeded Medicaid.Prevalence of drug use varied as much as rates across HSAs.

CONCLUSIONS: Prescription use was higher among Medicaid-insured thancommercially insured children. Regional variation generally exceeded payertype differences, especially for drugs used in situations of diagnostic andtherapeutic uncertainty. Efforts should advance best pediatric prescribingdiscussions and shared decision-making. Pediatrics 2014;134:563–570

AUTHORS: Shelsey J. Weinstein, BA,a Samantha A. House,DO, MPH,a,b Chiang-Hua Chang, PhD, Jared R. Wasserman,MS,c David C. Goodman, MD, MS,a,b,c,d and Nancy E. Morden,MD, MPHa,c,d

aGeisel School of Medicine at Dartmouth, Hanover, NewHampshire; bDepartment of Pediatrics, Dartmouth HitchcockMedical Center, Lebanon, New Hampshire; cThe DartmouthInstitute for Health Policy & Clinical Practice, Lebanon, NewHampshire; and dDepartment of Community and Family MedicineDartmouth Hitchcock Medical Center, Lebanon, New Hampshire

KEY WORDSantibiotic use, gastroesophageal reflux, geography,methylphenidate, pharmacotherapy, attention-deficit/hyperactivity disorder

ABBREVIATIONSADHD—attention-deficit/hyperactivity disorderHSA—hospital service areaPY—person-year

Dr Morden, Dr Goodman, and Ms Weinstein conceptualized anddesigned the study and drafted the initial manuscript; Dr Housereviewed and revised the manuscript; and Dr Chang andMr Wasserman analyzed the data obtained for the manuscriptand critically reviewed the numbers reported. All authorsapproved the final manuscript as submitted.

www.pediatrics.org/cgi/doi/10.1542/peds.2013-4250

doi:10.1542/peds.2013-4250

Accepted for publication Jun 12, 2014

Address correspondence to Shelsey J. Weinstein, Geisel School ofMedicine at Dartmouth 1 Medical Center Drive, 228 RubinBuilding, Lebanon, NH 03756. E-mail: shelsey.j.weinstein.med@dartmouth.edu

PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).

Copyright © 2014 by the American Academy of Pediatrics

FINANCIAL DISCLOSURE: The authors have indicated they haveno financial relationships relevant to this article to disclose.

FUNDING: Funding for this research was received from theCharles H. Hood Foundation and the Robert Wood JohnsonFoundation.

POTENTIAL CONFLICT OF INTEREST: The authors have indicatedthey have no potential conflicts of interest to disclose.

PEDIATRICS Volume 134, Number 3, September 2014 563

ARTICLE

by guest on December 14, 2020www.aappublications.org/newsDownloaded from

Prescription drugs are a common andimportant component of children’s healthcare. In 2010, a total of 2.64 million pre-scriptions were dispensed to childrenaged ,18 years in the United States or∼4 prescriptions per child per year.1 Al-though prescription use among childrenin the United States seems to be de-creasing, this general trend masks het-erogeneous changes across diverse druggroups. From 2002 to 2010, the reporteddecrease in number of prescriptionswritten for children was largely driven bya decline in antibiotic prescribing, whileuse of other drug classes, including selectpsychotropic agents and antacids, in-creased. Drivers of these diverse utiliza-tion trends for distinct drug groups arenot well understood.2–4

Pediatric prescribing of specific druggroups has been shown to vary acrosslarge geographic areas and across se-lect populations. For example, studies ofpsychotropic use have highlighted dif-ferential increases inprescribingacrosspayers and across populations definedby socioeconomic standing,5–8 race, orethnicity.9–11 Medicaid-enrolled childrenconsistently fill more psychotropic pre-scriptions than the commercially in-sured,12 but even among the Medicaidinsured, use of certain drug groupssuch as stimulants varies substantiallyacross regions of the United States (ie,west, midwest, south).10,13 Althoughothers have described the influence ofpayer and geographic region on pediat-ric pharmacotherapy, little is knownabout regional variation in prescriptionuse at the level of health care markets.In contrast to macroscopic observa-tions, study of smaller regional areas,such as hospital service areas (HSAs),reveals local pharmacotherapy practice.Novel analysis of practice variation atthis local level, especially across diversedrug groups, can prompt discussions onevidenced-based practice that may leadto improvements at the level of healthsystems.14

In this population-based study of an all-payer claims data set for northern NewEngland, we examine differences in useof diverse drug groups across payer-types (Medicaid versus commercialinsurance), and across small geo-graphic areas within payer-type. Wecontrol for population differences ingender, age, and payer to quantifyvariation in local patterns of care forchildren.

METHODS

Data and Population

This study used pediatric all-payer ad-ministrative data sets for northern NewEngland states (New Hampshire, Ver-mont, and Maine). The data, resultingfrom state-level efforts to developcomprehensive health care claims datasets, include all claims meeting state-level data-reporting mandates forchildren ages 0 to 17 years with one ormore months of enrollment in a com-mercial or Medicaid insurance plan,2007–2010 (Maine Medicaid data werenot available July 2009–December2010). To achieve statistically stablepopulations necessary for small areavariation analyses, the 4 years of datawere combined. Categorized by state,payer, and region, these children’s in-surance enrollment months served asthe denominators for prescriptiondrug use calculations.

Main Measures

Outpatient prescription fills for eachperson-month of enrollment were usedto develop 2 complementary prescrip-tion use measures: (1) prescription fillsper 100 person-years (PYs); and (2) theannual average proportion of the pop-ulation with any use of a particularmedication type. Together, these mea-sures reveal how many fills the popula-tion received and the proportion of thepopulation over which all observed fillsweredistributed.Preliminaryanalysisofdrug use over time revealed no signifi-

cant trends. We attributed this to ourrelatively short data span (2007–2010)and did not pursue further descriptionof secular trends.

We measured prescription fills overalland for select groups of drugs including:(1) prescription antacids (proton pumpinhibitors and histamine2-receptor an-tagonists); (2) antidepressants; (3) antipsy-chotics; (4) attention-deficit/hyperactivitydisorder (ADHD) drugs; and (5) anti-biotics (Supplemental Appendix Tables 1and 2). Specific drug groups analyzedwere drawn from those appearing mostfrequently in the claims of our enroll-ment cohort; prescription antacidswereconsidered because of recent focus ontheir increasing use among pediatricprescribers.1,15,16 Asthma medications,while common, were not specificallyexamined in this descriptive study dueto the complex interactions betweenmaintenance and rescue treatmentsthat warrant a methodologically dis-tinct examination. In addition, com-monly prescribed oral contraceptiveswere not specifically studied becausethese medications are frequently ac-cessed through family planning clinicsthat do not uniformly bill insurance,the source of our data. For overall fillmeasures, we did not include claimsfor fluoride because use of this productis influenced by local water fluorida-tion. To test the validity of prescriptionfill count as a measure of prescriptionuse, payer and year-specific days’ sup-ply per prescription fill were and com-pared across payer, state, and year.Fills were found to be a stable measureof prescription drugs received (Sup-plemental Appendix Table 3).

Covariates

Each patient-month was categorizedaccording to age, gender, and payer(Medicaid or commercial). ResidentialZIP codeswere used to link each child to1 of the 3 states of study and to 1 of 69HSAs. HSAs are relatively self-contained

564 WEINSTEIN et al by guest on December 14, 2020www.aappublications.org/newsDownloaded from

geographic health care markets de-fined initially with Medicare utilizationdata but validated for the purposes ofthis pediatric health services researchthrough calculated localization indices.Localization indiceswere defined as thepercentage of health care events forchildren residing in an HSA provided byclinicians within that HSA. The meanlocalization index overall for outpatientvisits in our data was 71% (5th–95thpercentile range: 35%–84%); the meanfor emergency department visits was74% (5th–95th percentile range: 48%–89%).14 Localization indices calculatedaccording to payer type were verysimilar (Supplemental Appendix Ta-ble 4).

Analysis

We calculated age- and gender-adjustedpopulation prescription use measuresoverall, by payer type (Medicaid orcommercial), by state-specific payer(eg, Maine Medicaid, Vermont [commer-cial]), and by HSAwithin eachpayer type.Diagnoses were not used to adjust druguse measures because studies usingclaims have shown that diagnostic as-certainmentcorrelateswith the intensityof health services use but only veryweaklywithmeaningful health outcomesin adults.14,17–19 Drug use was measuredaccording to age group (0–2, 3–4, and5–9 years), but we present only agegroup–specific data for antacid use,with the 0- to 2-year-old age group fur-ther stratified into 0- to 1-year-olds and1- to 2-year-olds. These data are selec-tively presented because antacids werethe only drug group for which a re-markable age group pattern emerged.For graphic displays, to address thebroad range of scales for distinct druggroup use, we calculated the ratio ofthe observed (HSA-adjusted) rate to theexpected (northern New England) rateof use of each drug for each HSA,stratified by payer type. To achievemeasures of treatment prevalence, we

calculated observation-time-weightedoverall, payer type–specific, and HSA-level age- and gender-adjusted annualmean proportion of the population withany use of each drug group.

Pearson’s coefficient was used to as-sess correlation between HSA-level useof specific drug groups across payers(eg, Medicaid versus commercial useof ADHD medications) and to assessthe correlation in use of specific druggroups, regardless of payer (eg, the cor-relation between antipsychotic agentsand ADHD medications). The Committeefor the Protection of Human Subjectsat Dartmouth College approved thisstudy. Analyses were conducted byusing SAS version 9.3 (SAS Institute,Inc, Cary, NC) and Excel 2010 (MicrosoftCorporation, Redmond, WA).

RESULTS

In total, 949 821 individuals aged 0 to 17years contributed 1.75 million PYs from2007 through 2010, with a mean follow-up time of 22.1 months. Overall, 53.9%of observed PYs were insured by com-mercial plans and 46.1% by Medicaid.Payer-specific PYs ranged from 216 929(Vermont, Medicaid) to 402 203 (Maine,commercial) (Table 1).

Differences Across and WithinPayer Type

Overall, age- and gender-adjusted pre-scription fill rates were higher amongMedicaid-insured children than amongcommercially insured children: 572 vs346 fills per 100 PYs, respectively.

Differences in drug use by payer typewere larger for somemedication groupsthan forothers. Antibioticswere themostcommonly used medications, and theiruse differed the least across payers.Antibiotics accounted for 18.8%of all fills.Within each year studied, on average,37.9% of Medicaid-insured children filledan antibiotic prescription (84.6 fills per100 PYs), whereas 33.3%of commercially

insured children filled a prescription(79.3 fills per 100 PYs). Drug use differ-ences by payer were more striking forpsychotropic agents. For example, forADHD medications, which comprised12.8%of allfills, 7.7%ofMedicaid-insuredchildrenfilledat least 1prescription (84.0fills per 100 PYs), whereas 4.2% of com-mercially insuredchildrenfilledat least 1prescription (34.1 fills per 100 PYs).Antidepressants comprised 4.9% of allfills; 3.8% of Medicaid-insured childrenreceived an antidepressant (29.0 fills per100 PYs), whereas 2.7% of commerciallyinsured children received an antide-pressant (16.3 fills per 100 PYs). Anti-psychotic use followed a similar trend(Table 2).

Prescription antacid use was alsohigher among Medicaid-insured chil-dren (2.9% with $ 1 fill, 11.8 fills per100 PYs) compared with the commer-cially insured (1.7% with $ 1 fill, 7.6fills per 100 PYs). Overall, use of ant-acids was highest among 0- to 2-year-olds (21.3 fills per 100 PYs overall);rates in this age group ranged from29.6 fills per 100 PYs (Vermont, Medic-aid) to 58.1 fills per 100 PYs (Maine,commercial). When use among 0- to 1-year-olds and 1- to 2-year-olds was ex-amined separately, the relationshipbetween payer type and prescribingintensity for 0- to 1-year-olds was thereverse of that seen for drug groups ingeneral. Among children aged,1 year,antacid use was higher among thecommercially insured (54.1 fills per 100PYs) compared with the Medicaid in-sured (37.5 fills per 100 PYs) (Fig 1).

Small Area Variation

In the 69 northern New England HSAs,area-specificPYsobservedranged from1457 to 141 644. HSA-level analysesdemonstrated substantial variation innearly all age- and gender-adjustedmeasures of prescription fills.

Overall, prescription fills per PY acrossHSAs varied more than twofold within

ARTICLE

PEDIATRICS Volume 134, Number 3, September 2014 565 by guest on December 14, 2020www.aappublications.org/newsDownloaded from

Medicaid-insured children, rangingfrom 3.4 to 8.1 fills per PY (5th–95thpercentile: 4.6–7.4). Among the com-mercially insured, the correspondingHSA range was 2.1 to 5.2 (5th–95thpercentile: 2.8–4.3). Among Medicaid-insured children, ADHD medicationfills per 100 PYs varied more thanthreefold, with an HSA-level range of39.7 to 130.4 (5th–95th percentile:54.7–106.9); corresponding commer-cial use varied even more, rangingfrom 12.3 to 58.3 (5th–95th percentile:20.3–50.3). Antidepressant use variedsimilarly, with more than fourfold var-iation across each payer (more thantwofold from the 5th to 95th percen-tile). Antipsychotic fills per 100 PYsranged among the Medicaid-insuredchildren from 11.8 to 52.3 (5th–95thpercentile: 16.9–47.8) and among thecommercially insured from 1.0 to 17.4

(5th–95th percentile: 3.1–10.4). Antaciduse across HSAs ranged withinMedicaid-insured children from 3.9 to22.2 fills per 100 PYs (5th–95th per-centile: 6.7–19.2); the correspondingcommercial range was 1.2 to 11.7 (5th–95th percentile: 3.3–10.1). HSA-levelvariation was narrower for antibioticuse: Medicaid-insured range, 52.1 to112.3 fills per 100 PYs and commercialrange, 50.5 to 105.6 fills per 100 PYs(Fig 2, Table 3, Supplemental AppendixTable 5).

Small Area Prescription UseCorrelations

Drug group–specific correlations inrate of prescription use betweenMedicaid-insured and commercially in-sured children were as follows: anti-biotics, R = 0.69; ADHD medications, R =0.46; antidepressants, R = 0.57; anti-

psychotics, R = 0.57; and antacids, R =0.52. Considering all use (regardless ofpayer), HSA-level correlations betweendrug groups were strong for psycho-tropic agents but weaker between dis-similar drug types. For example, thecorrelation between antidepressantuse and ADHD medication use wasstrong (R = 0.60); the same was true forantipsychotics and ADHD medications(R = 0.54). Conversely, there was nocorrelation between antidepressantuse and antacid use (R = 0.04) (Sup-plemental Appendix Table 6).

DISCUSSION

This population-level study revealedsubstantial HSA variation in prescrip-tion drug use among children residingin northern New England. Although pre-scription use was higher in Medicaid-insured children compared with thecommercially insured, substantial vari-ation was found across HSAs, withinpayer type, even after controlling forage and gender. This finding suggeststhe importance of local clinician prac-tice styles as one important deter-minant of drug use.14,20–22 The majorityof children residing in each HSA re-ceived most of their ambulatory carefrom within HSA providers. Therefore,the HSA prescription use measureslikely reflect the care pattern of eacharea’s clinicians.22 These findings may

TABLE 1 Distribution of Person Years (PYs) by Age, State, and Payer Type

State and Payer % of Total PYs No. of Unique Children by Age

All 0 to 2 y 3 to 4 y 5 to 9 y 10 to 14 y 15 to 17 y

MaineCommercial 23.0 402 203 51 745 38 660 105 490 121 743 84 566Medicaid 15.5 271 114 51 454 31 391 74 363 70 249 43 659

New HampshireCommercial 18.3 319 856 40 541 30 527 84 002 97 573 67 214Medicaid 18.3 319 863 63 231 37 637 88 907 82 910 47 179

VermontCommercial 12.7 222 456 28 050 20 247 56 040 68 546 49 574Medicaid 12.4 216 929 39 948 25 114 60 455 57 065 34 347

Total PYs 1 752 422 274 968 183 574 469 256 498 086 326 538

Person years represent 949 821 unique individuals enrolled in a Medicaid or commercial plan for$1 month, 2007 to 2010.

TABLE 2 Age- and Gender-Adjusted Prescription Fill Measures (2007–2010) by to Drug Group, State, and Payer

Drug Group Proportionof all Fills

OverallCommercial

OverallMedicaid

MECommercial

MEMedicaid

NHCommercial

NHMedicaid

VTCommercial

VTMedicaid

Rate of use: payer specific fills per 100 PYsAntibiotics 18.8% 79.3 84.6 74.6 80.4 81.2 86.7 85.2 86.8ADHD medications 12.8% 34.1 84.0 33.7 94.1 37.4 79.5 30.1 78.2Antidepressants 4.9% 16.3 29.0 16.9 31.9 17.3 28.1 14.0 26.5Antipsychotics 3.6% 5.8 29.8 6.1 33.8 6.1 30.1 5.0 24.4Antacids 2.2% 7.6 11.8 7.1 12.5 7.7 12.3 8.3 10.0

Prevalence of use: annual proportion of the population with one or more fillAntibiotics 33.3% 37.9% 32.8% 35.3% 33.8% 39.2% 33.5% 36.8%ADHD medications 4.2% 7.7% 4.3% 8.2% 4.7% 7.9% 3.3% 6.5%Antidepressants 2.7% 3.8% 2.9% 4.1% 2.8% 3.6% 2.1% 3.4%Antipsychotics 0.7% 2.6% 0.7% 3.2% 0.8% 2.5% 0.5% 2.1%Antacids 1.7% 2.9% 1.7% 3.0% 1.6% 2.8% 1.9% 2.7%

Missing Maine Medicaid data from July 2009 through December 2010. Antacids include both proton pump inhibitors and histamine2-receptor antagonists. The proportion of the populationwith $1 fill is the mean of each annual prevalence value 2007–2010. ME, Maine; NH, New Hampshire; VT, Vermont.

566 WEINSTEIN et al by guest on December 14, 2020www.aappublications.org/newsDownloaded from

reflect uncertainty regarding the indi-cations and benefits of some drugsprescribed to children and a lack ofconsensus around best pediatric pre-scribing practice.

ThehigherdruguseweobservedamongMedicaid-insured children confirmsother studies on payer type–specific

pediatric prescription use. The HSA-level variation we report enhances thegeographic resolution of payer type–specific patterns and expands the scopeof studied medications beyond psycho-tropic agents.10,12,13 Overall prescriptionuse among Medicaid-enrolled childrenwas 62% higher than use among the

commercially insured. Across HSAs,within payer-type variation was notable.For example, for Medicaid enrollees,HSA-level ADHD medication fill ratesvaried more than twofold; antacid usevaried approximately threefold (5th–95th percentile).

As with the variation seen amongMedicaid enrollees, the overall com-mercial drug use rates mask HSA-levelvariations more than threefold forantipsychotics (5th–95th percentile)and antacids (5th–95th percentile).These findings suggest that the likeli-hood of receiving a prescriptiondepends as much on where the childlives as on his or her socioeconomicstatus. The correlations observed be-tween drug group–specific use ratesand within drug groups across payerssimilarly support the role of prescriberpractice style as a key determinant ofchildren’s pharmacotherapy.

The importantbut, in relative terms, lowlevel of variation in antibiotic usewithinpayer type hints at potential sources of

FIGURE 1Antacid prescription fills per 100 PYs according to age group and state-specific payer type. The graphicportrays paradoxically higher rates of fills among commercially insured infants (compared withMedicaid-insured infants).

FIGURE 2Prescription fill rates in HSAs plotted as the ratio of observed use (area-adjusted rate) to expected use (overall northern New England rate), stratified by payer.Red dots are small HSAs (14 of 69); minimum number of children per area is 346 commercial and 525 Medicaid. Rate range is in legend.

ARTICLE

PEDIATRICS Volume 134, Number 3, September 2014 567 by guest on December 14, 2020www.aappublications.org/newsDownloaded from

the variation in pediatric prescribingobserved in this study. Useof antibioticsdiffers from use of psychotropic agentsin several ways. Antibiotics are morecommon overall, and differences in usemust therefore be substantial to swaystatistical variation measures. In ad-dition, antibiotics have long beena popular subject of treatment guide-lines, clinical quality measures, andnational reports on overuse.1,4 Suchconsensus-promoting efforts have be-gun to emerge only recently for psy-chotropic agents.23–26 In addition,although substantial subjective judg-ment is involved in the decision toprescribe a course of antibiotics foracute illness, moremay be required forthe decision to prescribe psychotropicagents usually intended to treatchronic conditions and symptoms. Al-though there is a clear role for somepsychotropic agents in the treatment ofspecific pediatric conditions,4,6 di-agnostic and therapeutic uncertaintycan be high for many clinical sit-uations.4,27,28 For patients and families,cultural beliefs and social stigma sur-rounding medication use may alsovary. These factors likely contribute tothe relatively high level of variationobserved in psychotropic prescribing.

The use of antacids seems subject toa distinct set of prescribing determi-nants. Little evidence supports the useof thesedrugs in children, especially forthe very young.16 Despite a growingbody of literature documenting over-use in the pediatric population, rates ofantacid use continue to increase.29,30

The fact that proton pump inhibitor andhistamine2-receptor antagonist useamong infants (the age group with thegreatest use) was twice as high incommercially insured children com-pared with Medicaid-insured childrenraises questions about the determi-nants of infant use of these drugs. Whywould the usual and now expectedpattern of diffusely higher prescriptionuse in Medicaid-insured children bereversed for infant use of antacids? Wecan speculate that parents of com-mercially insured children advocatefor treatment of common, and in mostcases physiologic, reflux more thanparents of Medicaid-insured children,but this theory cannot be inferred fromour data. Based on the epidemiology ofgastroesophageal pathology in this agegroup, the majority of this drug use islikely unwarranted.30 Although deter-minants of this practice remain un-clear, this pattern highlights the need

for dissemination and uptake of prac-tice guidelines as well as clinician andparent education on the value anddisadvantages of antacid use in thetreatment of pediatric reflux.

The present claims-based study hasa number of limitations. First, becausediagnoses appearing in claims havebeen shown to be a poor substitute fortrue health status (in adults) we havenot included these diagnoses in ouranalyses. This lack of association be-tween claims diagnoses and healthstatus has not been conclusively dem-onstrated for children. Such measure-ments in children are complicated bythe relative dearth ofmeaningful healthindicators and the very low rate ofmortality, arguably the best indicator ofhealth, in thispopulation. In theabsenceof research quantifying the associationbetween health and claims-based dis-ease measurement, we aligned ourapproach in this study with the bestavailable evidence of health care de-livery patterns in the United States,which is derived from adult studies.Future studies should test this ap-proach through examination of healthservices use in reliably identified,disease-specific pediatric cohorts. Webelieve thismethod isbest for this studyof prescription drug use.14,17–19 None-theless, meaningful differences inhealth states that are not accountedfor with our age- and gender-adjusted,payer type–stratified analysis may re-sult in an overestimate of unwarrantedvariation.

Second, some may worry that a fewHSAs with a small number of childrenare driving observed variation. TheHSAs for which we report prescriptionuse measures have a minimum of 346infants and children and 691 PYs. Fig-ure 2 demonstrates that relativelysmall-population HSAs span the rangeof use for drugs studied. Therefore,overall rates are not unduly influencedby a few small-population HSAs.

TABLE 3 Age- and Gender-Adjusted, Drug Group–Specific, Hospital Service Area Variation in Ratesof Use (Prescription Fills Per 100 Person Years) by Payer Type

Drug Group Minimum Maximum Mean Median Ratio of5th/95thPercentile

Ratio ofMaximum/Minimum

Coefficient ofVariation

AntibioticsMedicaid 52.1 112.3 86.3 85.2 1.65 2.16 0.14Commercial 50.5 105.6 76.4 76.8 1.59 2.09 0.15

ADHD medicationsMedicaid 39.7 130.4 79.0 79.2 1.95 3.28 0.22Commercial 12.3 58.3 34.9 34.5 2.48 4.74 0.26

AntidepressantsMedicaid 11.1 44.7 26.4 24.5 2.40 4.01 0.27Commercial 8.1 37.5 17.3 17.7 2.46 4.63 0.29

AntipsychoticsMedicaid 11.8 52.3 27.0 24.6 2.82 4.41 0.34Commercial 1.0 17.4 6.3 5.9 3.37 17.98 0.43

AntacidsMedicaid 3.9 22.2 11.7 11.5 2.88 5.73 0.32Commercial 1.2 11.7 6.9 6.9 3.09 9.47 0.30

Antacids include both proton pump inhibitors and histamine2 receptor antagonists. For commercial HSAs, n values ranged from 346to 50 487 individuals (691–103 175 PYs). For Medicaid HSAs, n values ranged from 525 to 30 336 individuals (767–56 354 PYs).

568 WEINSTEIN et al by guest on December 14, 2020www.aappublications.org/newsDownloaded from

Third, populations may differ acrossstates and areas. Medicaid eligibilitycan vary acrossMaine, NewHampshire,and Vermont, resulting in differences instate-level average socioeconomic sta-tus of the Medicaid-insured population.Commercial enrollment eligibility andcostsmaybeequally varied, resulting inunmeasured population differencesacross the regions studied. Similarly,northern New England is relatively ho-mogeneous in terms of race and eth-nicity. Although this makeup may limitthe generalizability of our findings tomore diverse populations, it alsostrengthens the study’s internal validitybecause race/ethnicity cannot reason-ably be suggested as a source of thevariation observed in this study. If ra-cial and cultural differences increasehealth care variation, it is likely nationalvariation in pediatric prescribing is

even greater than that we observedin this relatively homogeneous pop-ulation.

Fourth, Maine Medicaid data were notavailable for the latter months of 2009(July–December) and all of 2010. Regionalpopulations were sufficient despite thisabsence, but observed prescription usewill not reflect any changes occurringin this later time frame among MaineMedicaid-insured children.

Lastly, our measure of prescription usecounted prescription-fill events. Theseevents have been shown to be a goodproxy of medications consumed (inadults) and represent our best avail-able measure at this regional level.31,32

CONCLUSIONS

The high levels of variation in use ofpsychotropic agents and antacids we

observed suggest that practice stylessubstantially influence the pharmaco-therapy experience of children. Thesefindings should prompt discussion aboutthe definitions and determinants of pe-diatric prescribing quality. In turn, theseconversations should inform effortsaimed at developing a consistent ap-proach to pediatric prescribing. Childrenand their families should be included ineducation efforts aimed at assuringsound prescribing practice, and shareddecision-makingshouldbeahighpriority,especially when treatment decisions in-volve relatively high levels of diagnosticand therapeutic uncertainty with trade-offs between benefits and risks.

ACKNOWLEDGMENTThe authors thank Kristen K. Bronner ofthe Dartmouth Atlas Team for her helpwith data presentation.

REFERENCES

1. Chai G, Governale L, McMahon AW, TrinidadJP, Staffa J, Murphy D. Trends of outpatientprescription drug utilization in US children,2002–2010. Pediatrics. 2012;130(1):23–31

2. Comer JS, Olfson M, Mojtabai R. Nationaltrends in child and adolescent psychotropicpolypharmacy in office-based practice,1996–2007. J Am Acad Child Adolesc Psy-chiatry. 2010;49(10):1001–1010

3. Alexander GC, Gallagher SA, Mascola A,Moloney RM, Stafford RS. Increasing off-label use of antipsychotic medications inthe United States, 1995–2008. Pharmaco-epidemiol Drug Saf. 2011;20(2):177–184

4. Finkelstein JA, Stille C, Nordin J, et al. Re-duction in antibiotic use among US children,1996–2000. Pediatrics. 2003;112(3 pt 1):620–627

5. Crystal S, Olfson M, Huang C, Pincus H,Gerhard T. Broadened use of atypical anti-psychotics: safety, effectiveness, and policychallenges. Health Aff (Millwood). 2009;28(5):w770–w781

6. Cooper WO, Hickson GB, Fuchs C, ArbogastPG, Ray WA. New users of antipsychoticmedications among children enrolled inTennCare. Arch Pediatr Adolesc Med. 2004;158(8):753–759

7. Ghandour RM, Kogan MD, Blumberg SJ,Jones JR, Perrin JM. Mental health con-

ditions among school-aged children: geo-graphic and sociodemographic patterns inprevalence and treatment. J Dev BehavPediatr. 2012;33(1):42–54

8. Zito JM, Safer DJ, DosReis S, et al. Psy-chotropic practice patterns for youth: a 10-year perspective. Arch Pediatr AdolescMed. 2003;157(1):17–25

9. Zito JM, Safer DJ, Valluri S, Gardner JF,Korelitz JJ, Mattison DR. Psychotherapeuticmedication prevalence in Medicaid-insuredpreschoolers. J Child Adolesc Psycho-pharmacol. 2007;17(2):195–203

10. Zuvekas SH, Vitiello B. Stimulant medicationuse in children: a 12-year perspective. Am JPsychiatry. 2012;169(2):160–166

11. Olfson M, Blanco C, Liu L, Moreno C, Laje G.National trends in the outpatient treatmentof children and adolescents with antipsy-chotic drugs. Arch Gen Psychiatry. 2006;63(6):679–685

12. Zito JM, Safer DJ, dosReis S, Gardner JF,Boles M, Lynch F. Trends in the prescribingof psychotropic medications to pre-schoolers. JAMA. 2000;283(8):1025–1030

13. Cox ER, Motheral BR, Henderson RR, MagerD. Geographic variation in the prevalenceof stimulant medication use among chil-dren 5 to 14 years old: results from

a commercially insured US sample. Pedi-atrics. 2003;111(2):237–243

14. Goodman DC, Morden NE, Ralston SL, ChangCH, Parker D, Weinstein SJ. The DartmouthAtlas of Children’s Health Care in NorthernNew England. Lebanon, NH: The Trustees ofDartmouth College, 2013

15. Cox ER, Halloran DR, Homan SM, Welliver S,Mager DE. Trends in the prevalence ofchronic medication use in children: 2002–2005. Pediatrics. 2008;122(5). Available at:www.pediatrics.org/cgi/content/full/122/5/e1053

16. Ward RM, Kearns GL. Proton pump inhib-itors in pediatrics: mechanism of action,pharmacokinetics, pharmacogenetics, andpharmacodynamics. Paediatr Drugs. 2013;15(2):119–131

17. Wennberg JE, Staiger DO, Sharp SM, et al.Observational intensity bias associatedwith illness adjustment: cross sectionalanalysis of insurance claims. BMJ. 2013;346:f549

18. Song Y, Skinner J, Bynum J, Sutherland J,Wennberg JE, Fisher ES. Regional variationsin diagnostic practices. N Engl J Med. 2010;363(1):45–53

19. Welch HG, Sharp SM, Gottlieb DJ, SkinnerJS, Wennberg JE. Geographic variation in

ARTICLE

PEDIATRICS Volume 134, Number 3, September 2014 569 by guest on December 14, 2020www.aappublications.org/newsDownloaded from

diagnosis frequency and risk of deathamong Medicare beneficiaries. JAMA. 2011;305(11):1113–1118

20. Alak A, Seabrook JA, Rieder MJ. Variations inthe management of pneumonia in pediatricemergency departments: compliance withthe guidelines. CJEM. 2010;12(6):514–519

21. Richer LP, Laycock K, Millar K, et al; Pedi-atric Emergency Research Canada Emer-gency Department Migraine Group. Treatmentof children with migraine in emergencydepartments: national practice variationstudy. Pediatrics. 2010;126(1). Available at:www.pediatrics.org/cgi/content/full/126/1/e150

22. Goodman DC. Unwarranted variation inpediatric medical care. Pediatr Clin NorthAm. 2009;56(4):745–755

23. Subcommittee on Attention-Deficit/HyperactivityDisorder; Steering Committee on QualityImprovement and Management, WolraichM, Brown L, Brown RT, et al ADHD: clinicalpractice guideline for the diagnosis, eval-uation, and treatment of attention-deficit/hyperactivity disorder in children andadolescents. Pediatrics. 2011;128(5):1007–1022

24. Zuckerbrot RA, Cheung AH, Jensen PS, SteinRE, Laraque D; GLAD-PC Steering Group.Guidelines for Adolescent Depression inPrimary Care (GLAD-PC): I. Identification,assessment, and initial management. Pe-diatrics. 2007;120(5). Available at: www.pe-diatrics.org/cgi/content/full/120/5/e1299

25. Mahajan R, Bernal MP, Panzer R, et al; Au-tism Speaks Autism Treatment NetworkPsychopharmacology Committee. Clinicalpractice pathways for evaluation andmedication choice for attention-deficit/hyperactivity disorder symptoms in au-tism spectrum disorders. Pediatrics. 2012;130(suppl 2):S125–S138

26. McKinney C, Renk K. Atypical antipsychoticmedications in the management of disrup-tive behaviors in children: safety guidelinesand recommendations. Clin Psychol Rev.2011;31(3):465–471

27. US Food and Drug Administration. FDA2proposes new warnings about suicidalthinking, behavior in young adults whotake antidepressant medications. Availableat: www.fda.gov/newsevents/newsroom/pressannouncements/2007/ucm108905.htm.Accessed July 8, 2014

28. Greenhill LL, Vitiello B, Abikoff H, et al. De-veloping methodologies for monitoringlong-term safety of psychotropic medi-cations in children: report on the NIMHconference, September 25, 2000. J Am AcadChild Adolesc Psychiatry. 2003;42(6):651–655

29. Nelson SP, Kothari S, Wu EQ, Beaulieu N,McHale JM, Dabbous OH. Pediatric gastro-esophageal reflux disease and acid-relatedconditions: trends in incidence of diagnosisand acid suppression therapy. J Med Econ.2009;12(4):348–355

30. Barron JJ, Tan H, Spalding J, Bakst AW,Singer J. Proton pump inhibitor utilizationpatterns in infants. J Pediatr GastroenterolNutr. 2007;45(4):421–427

31. Lau HS, de Boer A, Beuning KS, Porsius A.Validation of pharmacy records in drugexposure assessment. J Clin Epidemiol.1997;50(5):619–625

32. Tamblyn R, Lavoie G, Petrella L, Monette J.The use of prescription claims databasesin pharmacoepidemiological research: theaccuracy and comprehensiveness of theprescription claims database in Québec.J Clin Epidemiol. 1995;48(8):999–1009

570 WEINSTEIN et al by guest on December 14, 2020www.aappublications.org/newsDownloaded from

DOI: 10.1542/peds.2013-4250 originally published online August 11, 2014; 2014;134;563Pediatrics 

David C. Goodman and Nancy E. MordenShelsey J. Weinstein, Samantha A. House, Chiang-Hua Chang, Jared R. Wasserman,

Small Geographic Area Variations in Prescription Drug Use

ServicesUpdated Information &

http://pediatrics.aappublications.org/content/134/3/563including high resolution figures, can be found at:

Referenceshttp://pediatrics.aappublications.org/content/134/3/563#BIBLThis article cites 30 articles, 10 of which you can access for free at:

Subspecialty Collections

http://www.aappublications.org/cgi/collection/pharmacology_subPharmacologyactivity_disorder_adhd_subhttp://www.aappublications.org/cgi/collection/attention-deficit:hyperAttention-Deficit/Hyperactivity Disorder (ADHD)following collection(s): This article, along with others on similar topics, appears in the

Permissions & Licensing

http://www.aappublications.org/site/misc/Permissions.xhtmlin its entirety can be found online at: Information about reproducing this article in parts (figures, tables) or

Reprintshttp://www.aappublications.org/site/misc/reprints.xhtmlInformation about ordering reprints can be found online:

by guest on December 14, 2020www.aappublications.org/newsDownloaded from

DOI: 10.1542/peds.2013-4250 originally published online August 11, 2014; 2014;134;563Pediatrics 

David C. Goodman and Nancy E. MordenShelsey J. Weinstein, Samantha A. House, Chiang-Hua Chang, Jared R. Wasserman,

Small Geographic Area Variations in Prescription Drug Use

http://pediatrics.aappublications.org/content/134/3/563located on the World Wide Web at:

The online version of this article, along with updated information and services, is

http://pediatrics.aappublications.org/content/suppl/2014/08/06/peds.2013-4250.DCSupplementalData Supplement at:

by the American Academy of Pediatrics. All rights reserved. Print ISSN: 1073-0397. the American Academy of Pediatrics, 345 Park Avenue, Itasca, Illinois, 60143. Copyright © 2014has been published continuously since 1948. Pediatrics is owned, published, and trademarked by Pediatrics is the official journal of the American Academy of Pediatrics. A monthly publication, it

by guest on December 14, 2020www.aappublications.org/newsDownloaded from