Anti-diabetic drug utilization of pregnant diabetic women in us managed care

PAGE 1

RESEARCHARTICLEOpenAccessAnti-diabeticdrugutilizationofpregnantdiabetic womeninusmanagedcareCaitlinAKnox1*,JosephACDelaney2andAlmutGWinterstein1,3AbstractBackground: Withtheincreasingprevalenceoftype2diabetesinyoungadulthood,treatmentofdiabetesin pregnancyfacesnewchallenges.Anti-diabeticdrugutilizationpatternsofpregnantwomenwithpre-existing diabetesarepoorlydescribed.Weaimtodescribeanti-diabetic(AD)agentutilizationamongdiabeticpregnant women. Methods: WeutilizedIMSLifeLink,includingadministrativeclaimsdataofpatientsinUSmanagedcareplans,to establisharetrospectivecohortofwomen,age18 – 46years(N=96,740)withbilledproceduresforalivebirth,and a12montheligibilityperiodbeforeand3monthafterdelivery.Diabetesmellituswasidentifiedfrom 2in-or outpatientclaimswithdiagnoses(ICD-9-CM250.XX)beforepregnancy.WeestimatedtheprevalenceofADdrugs before,duringandafterpregnancy,andseculartrendsacrossthestudyperiod(1999 – 2009),usinglinearregression. Asensitivityanalysiswasconductedtoidentifytheextentofmisclassificationoftrimesters. Results: Almostsixpercent(n=5,581)ofthelivebirthcohorthaddiabetesmellitus.Throughoutthestudy,48% (1999)and78%(2009)(p<0.0001)ofdiabeticwomenreceivedADdrugsduringpregnancy.ThemostcommonAD drugsduringpregnancywereinsulin,metformin,sulfonylureas,thiazolidinediones(TZD),andcombinationAD.The annualprevalenceofinsulinuseincreasedbyonly1%from39%(1999)to40%(2009)(p=0.589)duringpregnancy, whileuseofsulfonylureasandmetforminincreasedfrom2.5%and4.2%(1999)to17.3%and15.3%(2009) (p<0.0001),respectively.Insulinandsulfonylureausesteadilyincreasedinprevalencefromthe1stto3rdtrimester (16.5%and3.3%to33.0%and7.5%),whilemetforminandTZDusedecreased(11.4%and1.6%to3.8%and0.2%). Conclusions: ADuseduringpregnancydemonstratestheneedforadditionalinvestigationregardingsafetyand efficacyofADdrugsonmaternaloutcomes. Keywords: Pharmacoepidemiology,Drugutilization,Pregnancy,ManagedcareBackgroundCurrentestimatesprojectthatby2025oneinthree adultsintheUnitedStates(US)willhavediabetesmellitus [1].In2010,approximately11%ofUSwomenaged20 yearsorolderwereeitherdiagnosedorhadundiagnosed diabetes[1].Thisreflectsanincreaseindiabetesprevalenceof2%inthisagegroupoverthelastfiveyears,with acorresponding1.9millionnewcasesofdiabetesdiagnosedin2010[1].Thisgrowthisalmostexclusivelyattributabletotype2diabetesmellitus,whichtraditionallyhas haditsonsetinlaterstagesofadulthood[1,2]. Thegrowingprevalenceoftype2diabetesinyoung adultsisparticularlyimportant,asmoreyoungwomen willbediagnosedduringreproductiveyears[2].Poorly controlleddiabetesbothbe foreandduringthefirst trimesterofpregnancycancausemajorbirthdefects, spontaneousabortions,andstillbirths[2].Despitethis well-establishedfact,morethan60%ofwomenwith pre-existingdiabeteshave difficultymanagingtheir glycemiccontrolduringpregn ancy[3-5].Researchers andprovidersagreethatglycemiccontrolisoneofthe mostimportantmodifiableriskfactorsinminimizing birthdefectsofinfantsborntowomenwithpre-existing diabetes[6-10].However,littleexperienceandevidence regardingthesafetyandeffectivenessoforalagentsduring pregnancyexists. *Correspondence: cknox@ufl.edu1DepartmentofPharmaceuticalOutcomes&Policy,CollegeofPharmacy, Gainesville,FL,USA Fulllistofauthorinformationisavailableattheendofthearticle 2014Knoxetal.;licenseeBioMedCentralLtd.ThisisanopenaccessarticledistributedunderthetermsoftheCreative CommonsAttributionLicense(http://creativecommons.org/licenses/by/2.0),whichpermitsunrestricteduse,distribution,and reproductioninanymedium,providedtheoriginalworkisproperlycited.Knox etal.BMCPregnancyandChildbirth 2014, 14 :28 http://www.biomedcentral.com/1471-2393/14/28

PAGE 2

Whiletype1diabetesmanagementrequiresinsulinand thusleaveslittlechoiceduringpregnancy,type2diabetes maybemanagedwithlife-stylemodifications,oralantidiabeticagents,and/orinsulin.Amongoralagents,several newmolecularentitieshavebeenaddedwithinthelastten yearswithlimiteddataonpregnancyoutcomes.Giventhe limitedresearchthatisavailableonanti-diabeticagentuse duringpregnancy,weaimedtodescribeanti-diabetic agentutilizationbefore,duringandafterpregnancyand determineseculartrendsamongclassesofanti-diabetic drugsacrossthe10-yearstudyperiod(1999 – 2009)in womenwithpre-existingdiabetes.MethodsWeutilizedtheIMSLifeLinkDatabase,whichconsists ofcommercialhealthplaninformationfrommorethan 100managedcareplansthroughouttheUS.Themajority ofthepayertypewithinthedatabaseiscommerciallyinsured.TheIMSLifeLinkdatabasealsoincludesMedicaid, Medicare,self-insuredandunknownpayertypes.The databaserecordsaregenerallyrepresentativeofthecommerciallyinsuredpopulationintermsofgenderandage. TheIMSLifeLinkdatabaseiscomprisedofeligibility anddemographicinformation,aswellas,inpatientand outpatientclaimsdatawithdetailondiagnosisandprocedures,andprescriptiondrugclaims.Thisdatabase containedarandomsampleof6millionwomenaged18 to46yearswithnoprescriptiondrugclaimforcontraceptives.Tobeincludedinourstudycohortwerequiredwomentohaveabilledmedicalprocedurescode forlivebirth(Table1),and12monthscontinuousinsurancecoveragebeforeand3monthsafterdelivery. Womenwererequiredtohaveatleastoneprescription drugclaimbeforepregnancy,toconfirmprescriptiondrug coverage.Atotalof96,740womenmettheinclusion criteriaforthecohort. Toidentifypatientswithpre-existingdiabetes,werequiredtwoin-oroutpatientclaimswithdiagnosisof Table1Delivery-relatedprocedure(CPT-4)codesusedtoidentifylivebirthsCodeDescription 01960Anesthesiaforvaginaldeliveryonly 01961Anesthesiaforcesareandeliveryonly 01962Anesthesiaforurgenthysterectomyfollowingdelivery 01963Anesthesiaforcesareanhysterectomyw/oanylaboranalgesia/anesthesiacare 01967Neuraxiallaboranalgesia/anesthesia,plannedvaginaldelivery 01968Anesthesiaforcesareandeliveryfollowingneuraxiallaboranalgesia/anesthesia 01969Anesthesiaforcesareanhysterectomyfollowingneuraxiallaboranalgesia/anesthesia 59050Fetalmonitoringinlabor,physicianw/writtenreport 59051Fetalmonitoringinlabor,physicianw/writtenreport;interpretationonly 59400Routineobstetriccare,antepartumcare,vaginaldelivery,&postpartumcare 59409Vaginaldeliveryonly(w/woepisiotomy&/orforceps) 59410Vaginaldeliveryonly(w/woepisiotomy&/orforceps);w/postpartumcare 59412Externalcephalicversion,w/wotocolysis 59414Delivery,placenta(separateprocedure) 59430Postpartumcareonly(separateprocedure) 59510Routineobstetriccarew/antepartumcare,cesareandelivery,&postpartumcare 59514Cesareandeliveryonly 59515Cesareandeliveryonly;w/postpartumcare 59525Subtotal/totalhysterectomyaftercesareandelivery 59610Routineobstetriccare,vaginaldelivery,w/antepartum,postpartumcare,previousc-section 59612Vaginaldeliveryonly,previouscesareandelivery 59614Vaginaldeliveryonly,previouscesareandelivery;w/postpartumcare 59618Routineobstetriccare,ante/postpartum,cesareandeliveryafterfailedvaginaldelivery,previouscesareandelivery 59620Cesareandelivery,afterfailedvaginaldelivery,previouscesareandelivery 59622Cesareandelivery,afterfailedvaginaldelivery,previouscesareandelivery;w/postpartumcare 99436Attendanceatdelivery,atrequestofdeliveringphysician,&stabilizationofnewborn 99440Newbornresuscitation Knox etal.BMCPregnancyandChildbirth 2014, 14 :28 Page2of8 http://www.biomedcentral.com/1471-2393/14/28

PAGE 3

diabetesmellitus(InternationalClassificationofDiseases, NinthRevision,ClinicalModification(ICD-9-CM)250. XX)withinthe3monthsprecedingconception[11,12]. ICD-9-CMcodesidentifythespecifictypeofdiabetesmellitusbythefifthdigitofthecode,butpreviousresearch suggeststhatcurrentcodingpracticesdonotprovide sufficientaccuracytoidentifythetypeofdiabetesfrom thefifthdigitoftheICD-9-CMcode[13,14].Therefore, wedidnotseparatewomenaccordingtotheirdiabetes mellitustype,butdecidedtopresentasubgroupanalysisinwomenwithhighpropensitytohavetype2diabetes[15].Fortheanalysis,weassumedthepresenceof type2diabeteswhen100%ofdiabetesmellitusdiagnosesintheclaimsindicatedtype2diabetes(ICD-9-CM 250.0or250.2)[15].The “ otherdiabetes ” subgroup includedallwomenwithtype1diabetesmellitusclaims andthosewithmixedorunspecificclaims(i.e.,missing the5thdigit). Wedefinedthedeliverydateasthedateofthefirst CurrentProceduralTerminology(CPT-4)claimforlive birthforeachwoman.Weincludedwomenwithlive birthsonlybecauseweneededtoobtainanestimatedconceptiondateforeachwoman.Inclusionofterminated pregnancieswouldnothaveallowedthedeterminationof trimestersbecausethedateofterminationrelativetoconceptionwouldbeunknown.Sincewedidnothaveaccess totheLastMenstrualPeriod(LMP)date,itwasnecessary forustocalculatetheconceptiondateofthepregnancy. Thiswasdonebysubtractingninemonths(270days)from thedeliverydate[16,17].Eachtrimesterofpregnancywas assigneda91-dayinterval[16,17].Pre-pregnancyinterval wereidentifiedasthethreemonthsprecedingtheimputed conceptiondatetoclassifythepresenceofpre-existing diabetesandcapturepre-pregnancydrugutilization.We separatelydefinedanafterpregnancyintervalasthethreemonthtimeperiodafterthedeliverydate,resultinginfive distinct3-monthperiods,whichwereusedtodefinedrug useprevalence. Becausewereliedonanimputedconceptiondate,there wasapotentialformisclassificationofthepre-pregnancy periodandthefollowingtrimesters.Theriskofmisclassificationofthepregnancyperiodswasparticularlyhighin womenwithdiabetes,becausetheyareathigherriskfor pre-termbirths(i.e.shortergestationperiods).Inorderto evaluatethiseffectweconductedasensitivityanalysis, whereweutilizedthefirsthealthcareencounterwitha pregnancycode(ICD-9-CM:V22,V23,V72.40,V72.42 andCPT-4:81025)toestimateconceptiondate[18].Using thefirstpregnancyclaimastheconceptiondate,themean lengthofpregnancyforthecohortwas6.55monthswith atstandarddeviationof1.64months.Fivepercentofthe diabetescohorthadanICD-9-CMclaimforearlydelivery (644.2,644.20,and644.21).Therefore,weconductedanothersensitivityanalysiswhereweadjustedthegestation lengthto245daysforwomenwithanearlydeliveryclaim [19].Wesawthatthedrugclassutilizationinthesensitivityanalysisdidnotdiffersignificantlywiththeoriginally imputedpregnancyperiodswecalculatedusing270days subtractedfromthedeliverydate. Amongthetenanti-diabeticdrugclassesidentified withintheIMSpregnancycohort,wefocusedouranalysis onthefivemostcommonlyutilizeddrugclassesbefore, duringandafterpregnancy:insulin,biguanides,sulfonylureas,thiazolidinediones,andoralanti-diabeticcombinations.Theremaininganti-diabeticdrugclassesallshowed prevalenceratesoflessthan1%(alpha-glucosidaseinhibitors,amylinanalogs,dipeptidylpeptidase-4,GLP-1 receptoragonistsandmeglitinideanalogues).Allinsulin productswerecollapsedintoonecategoryreflecting AmericanHospitalFormularyService(AHFS)drugclass 68.20.08.Wecalculatedrespectivedrugclassprevalence astheproportionofdiabeticwomenwithadrugclaimof aparticularclassduringeachofthedesignatedperiods. Weestimatedtheprevalencealongwiththe95%confidenceintervalsofanti-diabeticdrugclassutilizationbefore,during(foreachtrimester)andafterpregnancy.We investigatedtheseculartrendofannualdrugutilization prevalenceusinglinearregression.AllanalyseswereconductedwithSAS9.2,Cary,NC.TheUniversityofFlorida InstitutionalReviewandPrivacyBoardsapprovedthis study.ResultsOverthecourseoftheten-yearstudyperiod(1999 – 2009), weidentified5,581(5.9%)womenwithpre-existingdiabetesamongallwomenwithaprocedurecodeforlive birth.Atotalof4,043womenhadonlyICD-9-CMcodes consistentwithtype2diabetesmellitus.Diabeticwomen wereonaverageslightlyolder,hadmorephysicianvisits andmoreprescriptiondrugclaimsbeforepregnancy (Table2).Asexpected,diabeticwomenhadahigher prevalenceofadditionalco-morbidconditionsthannondiabeticwomenhad,buthadsimilarfrequenciesofsmoking,alcoholanddrugabuse.Diabeticwomenalsohada higherprevalenceofcesareansectiondeliverycompared tonon-diabeticwomen,with44.8%versus27.2%respectively.Therewasahigherprevalenceofdiabeticsinthe EastregionoftheUS,andalowerprevalenceintheSouth andWest.Withinthesub-groupanalysis,type2diabetic womenhadsimilarbaselinecharacteristics,buttheaveragenumberofADprescriptionsbeforepregnancywas lowerinthetype2diabetesonlygroupwhencompared toalldiabeticwomen. Theannualmeanageforthediabeticpregnancycohort hadastatisticallysignificantincreasefrom29.85yearsto 32.90yearsfrom1999to2009(beta=0.34,p<0.0001). Theprevalenceofanti-diabetictreatment(insulinor oralanti-diabeticdrugs)amongdiabeticwomenbeforeKnox etal.BMCPregnancyandChildbirth 2014, 14 :28 Page3of8 http://www.biomedcentral.com/1471-2393/14/28

PAGE 4

pregnancyincreasedfrom25.9%(CI:25.74,26.06)to 36.1%(CI:35.94,36.26)from1999to2009(beta=1.3%, p<0.0001)(Figure1).Duringpregnancy,weobserveda statisticallysignificantriseinanti-diabeticdrugtreatmentfrom48.3%in1999to77.9%in2009(beta=3.4%, p<0.0001).Thegrowthintheprevalenceoftreatmentwas similaracrosstrimesterswiththelargestgrowthinthe thirdtrimester,whichsawa23%increaseintreatmentfrom 36.8%(CI:36.58,37.01)in199 9to59.4%(CI:59.18,59.62) in2009(beta=2.4%,p<0.0001).Theprevalenceofdiabetic womenwhoweretreatedwithananti-diabeticdrugafter pregnancyapproximatelydoubledfrom18.3%(CI:18.11, 18.49)to39.7%(CI:39.51,39.89)from1999to2009ata rateof1.8%(CI:1.14,2.53)peryear(p<0.0001)(Figure1). Table2BaselinecharacteristicsofthestudycohortsVariableCategoryDM*DM*NotDM* N=5,581N=5,581N=91,159 T2DM*OtherDM* N=4,043N=1,538 Age:Years(SD) 32.3(5.1)32.5(5.1)32.1(5.0)30.1(5.3) Eligibility:Months(SD)56.8(25.2)57.4(25.2)55.1(24.9)51.3(24.4) Physicianofficevisits3monthsbeforePregnancy(SD)21.3(24.5)20.8(25.2)22.5(22.5)14.6(17.8) NumberofPrescriptionDrugClaims3monthsbeforePregnancy(SD)Anti-Diabetic3.0(7.5)1.4(4.5)7.2(11.2)0.1(0.9) Other14.6(18.8)13.9(18.5)16.4(19.4)9.7(12.4) DeliveryRouteVaginal55.2%58.8%47.7%72.8% CesareanSection44.8%41.2%52.3%27.2% ComorbidConditionsPCOS12.5%13.1%10.9%5.5% Hypertension18.2%17.0%21.4%4.3% Infertility14.7%15.6%12.2%10.6% IVFClaim0.1%0.1%0.1%0.0% Obesity13.9%13.4%15.2%3.8% Smoking6.0%6.0%5.9%4.6% Alcohol0.8%0.8%1.0%0.9% DrugAbuse0.7%0.7%0.6%0.7% RegionEast32.6%33.6%26.6%20.6% Midwest43.1%43.1%40.6%43.8% South12.1%11.1%19.7%17.9% West12.3%12.2%13.2%17.7% ConceptionYear(SD)2005(2.4)2005(2.3)2005(2.6)2005(2.6)*DM=pre-existingdiabetesmellitus,T2DM=type2diabetesmellitus. Figure1 Annualprevalenceofanti-diabetictreatmentbefore,duringandafterpregnancy. Knox etal.BMCPregnancyandChildbirth 2014, 14 :28 Page4of8 http://www.biomedcentral.com/1471-2393/14/28

PAGE 5

Amongthevarioustherapeuticclasses,metforminand sulfonylureasshowedthegreatestincreaseoverthestudy periodfrom4.2%(CI:4.11,4.29)and2.5%(CI:2.44,2.56) in1999to15.3%(CI:15.21,15.39)and17.3%(CI:17.26, 17.36)in2009,respectively(p<0.0001foreachdrugclass) (Figure2).Thesulfonylureadrugclassincludedglimepride,glipizide,andglyburideat28%,47%and25%respectively.Thiazolidinedionesandcombinationdrugs remainedconstantat1.6%(CI:1.56,1.64)and0.5%(CI: 0.47,0.53)overthecours eofthestudy.Asexpected rosiglitazonesawadecreasefrom50%to23.1%within thethiazolidinesdionedrugclass(beta= 3.4,p=0.12). Insulinuseduringpregnancyincreased1%overthe10-year periodfrom39%(CI:38.81,39.19)in1999to40%(CI: 39.81,40.19)in2009(p-value=0.589(Figure2). Whenpooledacrosstheentirestudyperiod,the prevalenceofmetforminusebeforepregnancyinalldiabeticwomenwas13.7%(CI:13.11,14.28),whichdecreasedto12.3%(CI:12.21,12.39)duringpregnancy andthendecreasedfurtherafterpregnancyto7.7%(CI: 7.62,7.78)(Figure3).Insulinutilizationbehavedasexpected,withaloweraverageprevalencebeforepregnancyof10.7%(CI:10.52,10.88),whichmorethan tripledto35.3%(CI:35.11,35.48)duringpregnancyand thendroppedagainto11.8%(CI:11.67,11.93)after pregnancy.Sulfonylureautilizationfollowedasurprisinglysimilartrend,asthepre-pregnancybaselineprevalenceof3.1%(CI:3.08,3.12)grewduringpregnancyto 10.4%(CI:10.34,10.46)andthendecreasedto2.5%(CI: 2.43,2.57)afterpregnancy. Insub-groupanalysisofthetype2diabeticwomenonly, thepatternofanti-diabeticdrugutilizationwasverysimilartoalldiabeticwomen,exceptanexpectedlowerprevalenceofinsulinusethroughouteach3-monthperiod.We notedanevenlowerreboundinutilizationofmetformin afterpregnancy,withonly5.6%(CI:5.51,5.69)oftype2 diabeticwomenwhousedmetforminpost-delivery,but morethan12.4%(CI:12.30,12.50)beforepregnancy.DiscussionWeconductedadescriptiveanalysisonanti-diabeticdrug utilizationbefore,duringandafterpregnancyindiabetic womeninordertoguidefutureresearchondrugsafety andeffectiveness.Ourstudyhadseveralkeyfindings. First,overallanti-diabeticdrugutilizationduringpregnancydoubledoverourten-yearstudyperiod.Second, despitethisincrease,wefoundasignificantproportion ofdiabeticwomenwithnodrugtreatmentbefore,duringandafterpregnancy.Third,amongtheanti-diabetic drugclasseswefoundinterestingADutilizationpattern indicatingchangesinthet reatmentregimenduring pregnancy.Forexample,theprevalenceofmetformin utilizationdecreasedaspregnancyprogressedfromthe 1sttothe3rdtrimester,butwasre-establishedintothe treatmentregimenafterpregnancy,althoughatlessthan halftheprevalenceofthebeforeperiod.Insulinandsulfonylureademonstratedareversedpattern,astheoverall prevalenceofutilizationinbothdrugclassesincreased overthecourseofpregnancyandthenreturnedtoalow prevalenceafterpregnancysimilartoutilizationbefore. Currentrecommendationsfortheinitialtherapyof diabetes(bothtype1andtype2)emphasizepharmacologicalinterventions[20].AccordingtotheAmerican DiabetesAssociation(ADA)andtheEuropeanAssociationfortheStudyofDiabetes(EASD),theinitialmanagementoftype2diabetesshouldusecombinationtherapy ofmetforminandlifestylechangeswithaugmentationof therapywithadditionaloralanti-diabeticdrugtomaintain glycemiccontrol[20].Orinitiateinsulinatdiagnosisfor individualswhopresentwithseverehyperglycemicsymptoms[20].Therecommendationreflectsachangeintreatmentapproachesadecadeago,whichsuggestedthatearly type2diabetesmightbemanagedwithdietandlifestyle modificationsalone.Thisevolutionintreatmentparadigm mayexplaintheincreaseinoverallanti-diabeticdrug utilizationthatweobservedamongdiabeticwomenbefore,during,andafterpregnancyoverourstudyperiod. Figure2 Annualprevalenceofanti-diabeticdrugutilizationduringpregnancybydrugclass. Knox etal.BMCPregnancyandChildbirth 2014, 14 :28 Page5of8 http://www.biomedcentral.com/1471-2393/14/28

PAGE 6

Interestingly,accordingtotheNationalHealthInterview Survey,forthepasttenyearstheprevalenceofdiabetic adultswhodonotuseinsulinororalanti-diabeticdrugs tocontroltheirdiabeteshasslightlygrownby1%tomore than15%ofUSadultswithdiabetes[2,21].Thefactthat weobservedevenhigherproportionsofnon-treatedpatientsinamanagedcarepopulationwithcomprehensive drugbenefitsandincreasedmedicalattentionduetopregnancyissurprisingandraisesquestionsaboutcurrent treatmentapproaches.Whilethemajorityofpregnant womanmayhavehadonsetoftype2diabetesinrecent yearsandthuslimitedneedforaggressiveglycemiccontrol,westillexpectedthatcurrenttreatmentguidelines andtheemphasisontightglucosecontrolduringpregnancywouldhaveresultedinmorecomprehensivedrug therapy.Unfortunately,becauselaboratoryvaluesarenot availableinclaimsdata,itisunclearwhetherthesewomen wereabletoachieveormaintainnormoglycemiathroughoutpregnancy. Ingeneral,theuseofinsulintotreattype2diabetes mellitusduringpregnancyisacceptedandrecommended assafeandeffectiveinachievingnormalbloodglucose levels[22,23].Thereisalsosufficientevidencetosupport thatmetforminisnon-inferiortoinsulinwithregardsto neonatalsafety,butcomparativedataregardingefficacy duringpregnancyarelacking[24,25].Furthermore,the majorityofthemetforminsafetystudieshavebeencompletedinwomenwithpolycysticovariansyndromeor gestationaldiabetes,notinwomenwithpre-existing diabetes,andhavenotevaluatedpregnancyoutcomes (pre-termlabor,preeclampsia,cesareansectiondelivery) assafetyendpoints[24,26]. Thedistinctincreaseofsulfonylureasuseduringpregnancyobservedinthiscohortwasnoteworthyinthis context,becausethisdrugclassiscurrentlynotrecommendedforthisindication,includinganFDAcontraindicationforuseinthelast4weeksofpregnancy.This notwithstanding,severalstudieshavefoundnoharmful effectsandreportgoodglycemiccontrolwiththe useofsulfonylureasduringpregnancy[27-30].Again, themajorityofstudiesontheuseofthisdrugclass inpregnancyaddresstheuseingestationaldiabetic womenandlackevidenceonvariousaspectsofsafety andefficacy. Weconductedthisstudyusingadministrativeclaims data.TheIMSLifeLinkdatabaseallowedustoinvestigate theprevalenceofdiabetesinpregnancyinarelativelylarge cohortofmorethan96,000pregnantwomen,sampled fromover100commercialhealthplansacrosstheUS, withfullyadjudicatedmedicalandpharmaceuticalclaims. Thestudyperiod,whichspannedtenyears,allowedfor theexaminationoftimetrendsindrugutilizationofthese womenbefore,duringandafterpregnancy.Focusona managedcaresettingallowedgoodobservationofprescribingpractices,becauseaccesstocareisnotamajor limitation. Alladministrativedatahav elimitationsandthisstudy isnoexception.Despiteourlargesamplesize,focuson womeninprivateinsuranceisnotrepresentativeofdiabetestreatmentpatternintheUS.Inaddition,theselectionofhealthplansinIMSmaynotberepresentativeas suggestedbythegeographicdistributionofdiabetes prevalencethatdoesnotfollownationallyreporteddata. However,comparisonsacrosstimeandacrosstrimestersareexpectedtobevalidwithinthestudycohort.By restrictingthepregnancycohorttoonlywomenwitha prescriptiondrugclaimbeforepregnancy,weareomittingwomenwithoutdrugcoverageandtheoretically omittingwomenwhoarenotreceivingdrugtreatment orwhoarenon-adherenttotheirmedicationregimen. Wepotentiallyunderestimateddruguseprevalencein instanceswherepharmacyclaimswerenotsubmitted forreimbursementbecausepatientspaidcashfortheir prescriptions.However,sinceprescriptioncopaysshould haveprovidedcheaperalternativestopatientsduringthe studyperiod,weexpectminimalmisclassification.We limitedthisdescriptiveanalysistopregnantwomenwith livebirths.Therefore,thedrugutilizationpatternsthat wereseenwithinthisdescriptiveanalysismaybedifferent frompre-existingdiabeticwomenwithunsuccessfulpregnancies(i.e.miscarriage,stillbirth,ortermination). Finally,weaimedtofocusourstudyonpatientswith type2diabetes,buthadtoacceptlimitationsinthe granularityofICD9-CMcodesusedforbilling.Studies havebeenlargelyunsuccessfulinvalidatingalgorithms todistinguishbetweendiabetestypes[13-15,31].In ordertoestimatetheextentof potentialmisclassificationweprovideresultsofasensitivityanalysisusinga conservativeapproachthatdidnotallowanycodesfor type1orunspecifieddiabetesinoursub-cohortoftype 2diabeticpatients.Thisapproachwaslikelynotsensitiveandmayhaveexcludedpatientsdifferentially. Figure3 Pooledannualprevalenceofanti-diabeticdrug utilization,bydrugclass:before,during,andafterpregnancy. Knox etal.BMCPregnancyandChildbirth 2014, 14 :28 Page6of8 http://www.biomedcentral.com/1471-2393/14/28

PAGE 7

ConclusionPre-existingdiabetesisanincreasingcomorbidityin pregnantwomanintheUS.Whiletheoveralluseof anti-diabeticmedicationsduringpregnancyincreased,a largerthanexpectedproportionofpregnantwomendid nothaveADdrugclaimsthroughoutthestudyperiod. Contrarytocurrentrecommendations,metforminuse decreasedaspregnancyprogressedwhilesulfonylurea useincreased.Thehighrateoforalanti-diabeticdrug useduringpregnancyemphasizestheneedforconclusive evidenceregardingsafetyandefficacyintermsofglucose controlaswellasmaternaloutcomes.Furtherresearchis neededinordertoevaluatethesafetyoforalanti-diabetic agentuseinpregnantwomenwithpre-existingdiabetesin termsofpregnancyandneonataloutcomes.Withinthis study,itwewereunabletoassesstheimpactofglycemic controlandco-morbidconditionsonthechoiceofantidiabeticagentsthroughoutpregnancy.Therefore,itwill alsobeimportantforfutureresearchtofocusonthedeterminatesofmedicationchoicesduringpregnancyincludingthepresenceoron-setofco-morbidconditions andchangesinglycemiccontrol.Additionally,thelower prevalenceofanti-diabeticdrugutilizationpost-delivery indicatesapossibleneedforfurtherinvestigation.Competinginterests Theauthorsdeclarethattheyhavenocompetinginterests. Authors ’ contributions Conceivedanddesignedtheresearch:CK,JD,AW.Analyzedthedata:CK. Wrotethepaper:CK.Interpretationofdata:CK,JD,AW.Criticallyrevised manuscript:CK,JD,AW.Allauthorshaveapprovedthemanuscriptas submitted. Presentationofwork Thesub-groupanalysisofthetype2diabetesdatafromthispaperwas presentedatthe28thInternationalConferenceonPharmacoepidemiology andTherapeuticRiskManagement,inAugust2012. Acknowledgements Wereceivednofundinginsupportofthisresearchandthispaperisnot underreviewelsewhere.TheAmericanFoundationPharmaceutical EducationFellowshipandtheMaryKOwensFellowshipsupportedCK.None oftheseentitieshadinfluenceonthedesignandconductofthestudy, collection,management,analysis,andinterpretationofthedata,andthe preparationorapprovalofthemanuscript. Authordetails1DepartmentofPharmaceuticalOutcomes&Policy,CollegeofPharmacy, Gainesville,FL,USA.2DepartmentofEpidemiology,UniversityofWashington, Seattle,WA,USA.3CollegesofMedicineandPublicHealthandHealth Professions,UniversityofFlorida,Gainesville,FL,USA. Received:5April2013Accepted:23December2013 Published:17January2014 References1. DiagnosedandundiagnoseddiabetesintheUnitedStates,allages,2010. [http://www.cdc.gov/diabetes/pubs/estimates11.htm#1] 2.CowieCC,RustKF,Byrd-HoltDD,EberhardtMS,FlegalKM,EngelgauMM, SaydahSH,WilliamsDE,GeissLS,GreggEW: Prevalenceofdiabetesand impairedfastingglucoseinadultsintheU.S.population:NationalHealth AndNutritionExaminationSurvey1999 – 2002. DiabetesCare 2006, 29 (6):1263 – 1268. 3.HolingEV,BeyerCS,BrownZA,ConnellFA: Whydon'twomenwith diabetesplantheirpregnancies? DiabetesCare 1998, 21 (6):889 – 895. 4.CoustanDR: Pre-conceptionplanning.Therelationship'sthething. DiabetesCare 1998, 21 (6):887 – 888. 5.JanzNK,HermanWH,BeckerMP,Charron-ProchownikD,ShaynaVL,Lesnick TG,JacoberSJ,FachnieJD,KrugerDF,SanfieldJA: Diabetesandpregnancy. Factorsassociatedwithseekingpre-conceptioncare. DiabetesCare 1995, 18 (2):157 – 165. 6.KitzmillerJL,GavinLA,GinGD,Jovanovic-PetersonL,MainEK,ZigrangWD: Preconceptioncareofdiabetes.Glycemiccontrolpreventscongenital anomalies. JAMA 1991, 265 (6):731 – 736. 7.BellR,GlinianaiaSV,TennantPW,BilousRW,RankinJ: Peri-conception hyperglycaemiaandnephropathyareassociatedwithriskofcongenital anomalyinwomenwithpre-existingdiabetes:apopulation-based cohortstudy. Diabetologia 2012, 55 (4):936 – 947. 8.Schaefer-GrafUM,BuchananTA,XiangA,SongsterG,MontoroM,KjosSL: Patternsofcongenitalanomaliesandrelationshiptoinitialmaternal fastingglucoselevelsinpregnanciescomplicatedbytype2and gestationaldiabetes. AmJObstetGynecol 2000, 182 (2):313 – 320. 9.MillerE,HareJW,ClohertyJP,DunnPJ,GleasonRE,SoeldnerJS,Kitzmiller JL: ElevatedmaternalhemoglobinA1cinearlypregnancyandmajor congenitalanomaliesininfantsofdiabeticmothers. NEnglJMed 1981, 304 (22):1331 – 1334. 10.DunneF,BrydonP,SmithK,GeeH: PregnancyinwomenwithType2 diabetes:12yearsoutcomedata1990 – 2002. DiabetMed 2003, 20 (9):734 – 738. 11.ChenG,KhanN,WalkerR,QuanH: ValidatingICDcodingalgorithmsfor diabetesmellitusfromadministrativedata. DiabetesResClinPract 2010, 89 (2):189 –195. 12.HebertPL,GeissLS,TierneyEF,EngelgauMM,YawnBP,McBeanAM: IdentifyingpersonswithdiabetesusingMedicareclaimsdata. AmJMed Qual 1999, 14 (6):270 – 277. 13.WeiWQ,LeibsonCL,RansomJE,KhoAN,CaraballoPJ,ChaiHS,YawnBP, PachecoJA,ChuteCG: Impactofdatafragmentationacrosshealthcare centersontheaccuracyofahigh-throughputclinicalphenotyping algorithmforspecifyingsubjectswithtype2diabetesmellitus. JAmMed InformAssoc 2012, 19 (2):219 – 224. 14.O'MalleyKJ,CookKF,PriceMD,WildesKR,HurdleJF,AshtonCM: Measuring diagnoses:ICDcodeaccuracy. HealthServRes 2005, 40 (5Pt2):1620 – 1639. 15.ZgiborJC,OrchardTJ,SaulM,PiattG,RuppertK,StewartA,SiminerioLM: Developingandvalidatingadiabetesdatabaseinalargehealthsystem. DiabetesResClinPract 2007, 75 (3):313 – 319. 16.TohS,MitchellAA,WerlerMM,Hernndez-DazS: Sensitivityand specificityofcomputerizedalgorithmstoclassifygestationalperiodsin theabsenceofinformationondateofconception. AmJEpidemiol 2008, 167 (6):633 – 640. 17.AndradeSE,RaebelMA,MorseAN,DavisRL,ChanKA,FinkelsteinJA, FortmanKK,McPhillipsH,RoblinD,SmithDH, etal : Useofprescription medicationswithapotentialforfetalharmamongpregnantwomen. PharmacoepidemiolDrugSaf 2006, 15 (8):546 – 554. 18.HardyJR,LeadererBP,HolfordTR,HallGC,BrackenMB: Safetyof medicationsprescribedbeforeandduringearlypregnancyinacohort of81,975mothersfromtheUKgeneralpracticeresearchdatabase. PharmacoepidemiolDrugSaf 2006, 15 (8):555 – 564. 19.MargulisAV,SetoguchiS,MittlemanMA,GlynnRJ,DormuthCR,HernndezDazS: Algorithmstoestimatethebeginningofpregnancyin administrativedatabases. PharmacoepidemiolDrugSaf 2012, 22 (1):16 – 24. 20.AssociationAD: Standardsofmedicalcareindiabetes – 2011. Diabetes Care 2011, 34 (Suppl1):S11 – 61. 21. 2003NationalDiabetesFactSheet. [http://www.cdc.gov/diabetes/pubs/ general.htm#treatment] 22.AssociationAD: Standardsofmedicalcareindiabetes – 2009. Diabetes Care 2009, 32(Suppl1):S13 – 61. 23.BulletinsACoP: ACOGpracticebulletin.Clinicalmanagementguidelines forobstetrician-gynecologists.Number60,march2005.Pregestational diabetesmellitus. ObstetGynecol 2005, 105 (3):675 – 685. 24.RowanJA,HagueWM,GaoW,BattinMR,MooreMP,InvestigatorsMT: Metforminversusinsulinforthetreatmentofgestationaldiabetes. NEnglJMed 2008, 358 (19):2003 – 2015. 25.GroupGD: Managementofdiabetesfrompreconceptiontothe postnatalperiod:summaryofNICEguidance. BMJ 2008, 336 (7646):714 – 717.Knox etal.BMCPregnancyandChildbirth 2014, 14 :28 Page7of8 http://www.biomedcentral.com/1471-2393/14/28

PAGE 8

26.GlueckCJ,WangP,GoldenbergN,Sieve-SmithL: Pregnancyoutcomes amongwomenwithpolycysticovarysyndrometreatedwithmetformin. HumReprod 2002, 17 (11):2858 – 2864. 27.LangerO,ConwayDL,BerkusMD,XenakisEM,GonzalesO: Acomparison ofglyburideandinsulininwomenwithgestationaldiabetesmellitus. NEnglJMed 2000, 343 (16):1134 – 1138. 28.JacobsonGF,RamosGA,ChingJY,KirbyRS,FerraraA,FieldDR: Comparisonofglyburideandinsulinforthemanagementofgestational diabetesinalargemanagedcareorganization. AmJObstetGynecol 2005, 193 (1):118 – 124. 29.KorenG: Glyburideandfetalsafety;transplacentalpharmacokinetic considerations. ReprodToxicol 2001, 15 (3):227 – 229. 30.CarpenterMW,CoustanDR: Criteriaforscreeningtestsforgestational diabetes. AmJObstetGynecol 1982, 144 (7):768 – 773. 31.KhoAN,HayesMG,Rasmussen-TorvikL,PachecoJA,ThompsonWK, ArmstrongLL,DennyJC,PeissigPL,MillerAW,WeiWQ, etal : Useofdiverse electronicmedicalrecordsystemstoidentifygeneticriskfortype2 diabeteswithinagenome-wideassociationstudy. JAmMedInformAssoc 2012, 19 (2):212 – 218.doi:10.1186/1471-2393-14-28 Citethisarticleas: Knox etal. : Anti-diabeticdrugutilizationofpregnant diabeticwomeninusmanagedcare. BMCPregnancyandChildbirth 2014 14 :28. Submit your next manuscript to BioMed Central and take full advantage of: € Convenient online submission € Thorough peer review € No space constraints or color “gure charges € Immediate publication on acceptance € Inclusion in PubMed, CAS, Scopus and Google Scholar € Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Knox etal.BMCPregnancyandChildbirth 2014, 14 :28 Page8of8 http://www.biomedcentral.com/1471-2393/14/28