BMJ2012;344:e363doi:10.1136/bmj.e363(Published24January2012) Page1of10 Research RESEARCH Consumption of fried foods and risk of coronary heart disease: Spanish cohort of the European Prospective Investigation into Cancer and Nutrition study OPENACCESS Pilar Guallar-Castillón associate professor12, Fernando Rodríguez-Artalejo professor12, Esther Lopez-Garciaassistantprofessor12,LuzMLeón-Muñozpostdoctoralresearchfellow12,PilarAmiano epidemiologist32,EvaArdanazepidemiologist,headofsection42,LarraitzArriolaepidemiologist 32, AurelioBarricarteepidemiologist,headofservice42,GenevieveBucklandnutritionalepidemiologist6, María-Dolores Chirlaque specialist in preventive medicine and public health72, Miren Dorronsoro unitchief32,José-MaríaHuertaresearchassociate72,NereaLarrañagaepidemiologist32,PilarMarin registerednurse42,CarmenMartínezseniorscientist92,EstherMolinaresearchassociate92,Carmen Navarroseniorscientist,headofdepartment728,JRamónQuirósepidemiologist10,LaudinaRodríguez medical doctor10, María José Sanchez professor and director of research92, Carlos A González senior scientist, unit chief6, Conchi Moreno-Iribas epidemiologist245 1DepartmentofPreventiveMedicineandPublicHealth,SchoolofMedicine,AutonomousUniversityofMadrid,28029Madrid,Spain;2CIBERde EpidemiologíaySaludPública,Spain;3PublicHealthDivisionofGipuzkoa,InstitutoInvestigaciónIISBioDonostia,BasqueGovernment,Donostia, Spain;4PublicHealthInstituteofNavarra,Pamplona,Spain;5DepartmentofPediatrics,DepartmentofObstetricsandGynecology,andPreventive Medicine,AutonomousUniversityofBarcelona,Barcelona,Spain;6UnitofNutrition,EnvironmentandCancer,CancerEpidemiologyResearch Programme,CatalanInstituteofOncology,HospitaletdeLlobregat,Barcelona,Spain;7DepartmentofEpidemiology,MurciaRegionalHealthCouncil, Spain;8DepartmentofPublicHealthandPreventiveMedicine,UniversityofMurcia,Spain;9AndalusianSchoolofPublicHealth,Granada,Spain; 10PublicHealthandHealthPlanningDirectorate,Oviedo,Asturias,Spain Abstract 1.07(0.83to1.38),andinthefourthquarterwas1.08(0.82to1.43;P ObjectiveToassesstheassociationbetweenconsumptionoffriedfoods fortrend0.74).Theresultsdidnotvarybetweenthosewhousedolive andriskofcoronaryheartdisease. oilforfryingandthosewhousedsunfloweroil.Likewise,noassociation wasobservedbetweenfriedfoodconsumptionandallcausemortality: DesignProspectivecohortstudy. multivariatehazardratioforthehighestversusthelowestquarteroffried SettingSpanishcohortoftheEuropeanProspectiveInvestigationinto foodconsumptionwas0.93(95%confidenceinterval0.77to1.14;Pfor CancerandNutrition. trend0.98). Participants40757adultsaged29-69andfreeofcoronaryheart ConclusionInSpain,aMediterraneancountrywhereoliveorsunflower diseaseatbaseline(1992-6),followedupuntil2004. oilisusedforfrying,theconsumptionoffriedfoodswasnotassociated MainoutcomemeasuresCoronaryheartdiseaseeventsandvital withcoronaryheartdiseaseorwithallcausemortality. statusidentifiedbyrecordlinkagewithhospitaldischargeregisters, Introduction populationbasedregistersofmyocardialinfarction,andmortality registers. Fryingisoneofthemostcommonlyusedmethodsforcooking ResultsDuringamedianfollow-upof11years,606coronaryheart inWesterncountries.Whenfoodisfrieditsnutritionalcontent diseaseeventsand1135deathsfromallcausesoccurred.Compared changes—thefoodloseswaterandtakesupfat,increasingits withbeinginthefirst(lowest)quarteroffriedfoodconsumption,the energydensity.12Fryingmodifiesboththefoodsandthefrying multivariatehazardratioofcoronaryheartdiseaseinthesecondquarter mediumbecauseoilsdeteriorateduringfrying,especiallywhen was1.15(95%confidenceinterval0.91to1.45),inthethirdquarterwas reused,throughtheprocessesofoxidationandhydrogenation, Correspondence to: P Guallar-Castillón [email protected] No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe:http://www.bmj.com/subscribe BMJ2012;344:e363doi:10.1136/bmj.e363(Published24January2012) Page2of10 RESEARCH leadingtoalossofunsaturatedfatsandanincreaseintrans Participantswereaskedaboutfoodconsumedinatypicalweek fats.1Thusfriedfoodabsorbsdegradationproductsofthefrying duringtheprevious12months.Foodsconsumedatleasttwice oil.3Atthesametime,fryingmayalsoimprovepalatabilityby amonthwererecorded.Thedietaryhistoryquestionnaire makingfoodcrunchy.45 recordedinformationonupto662differentfoods,including Friedfoodshavebeenassociatedwithvariouscardiovascular thoseconsumedindividuallyandthoseconsumedaspartof203 riskfactorsincrosssectionalstudies.InSpain,thePizarrastudy recipesandregionaldishes.Ofthesefoods,212werefried.We of1226adultsshowedthatconsumptionoffoodfriedwith calculatedenergyandnutrientintakeusingtheEPICfood reusedoilswasassociatedwithahigherprevalenceofarterial compositiontable.21 hypertension.6Furthermore,acrosssectionalanalysisofthe Informationoncookingmethodswascollectedaspartofthe SpanishcohortoftheEuropeanProspectiveInvestigationinto dietaryhistoryquestionnaire.Thesamefoodmaybecookedby CancerandNutrition(EPIC)studyconcludedthatconsumption severalmethodsaspartofasinglerecipe.Forexample,inthe offriedfoodswaspositivelyassociatedwithgeneralandcentral preparationofpaella(atypicalSpanishrecipe),riceisbothfried obesity.7Similarly,intheSUN(SeguimientoUniversityof andboiled.Thus,friedfoodwasdefinedasfoodforwhich Navarra)cohortinSpain,morefrequentconsumptionoffried fryingwastheonlycookingmethodused.Suchfoodcouldbe foodsatbaselinewasassociatedwithahigherriskofbecoming deepfriedorpanfriedanditcouldbebattered,crumbed,or overweightorobeseduringasixyearfollow-up.8Lastly,in sautéed.Toestimatetheabsorptionoffatfromfrying,weused anothercrosssectionalstudyin2090Italianadults,consumption anabsorptioncoefficientforeachfriedfood.18Wealsorecorded offriedfoodwasassociatedwithlowerhighdensitylipoprotein thetypeofoilusedfordressing,cooking,andfrying. cholesterollevelsandalargerwaistcircumference.9 Assessment of non-dietary variables Onlyafewstudieshaveevaluatedtheeffectoffriedfoodson riskofcardiovasculardisease.Acase-controlstudyinCosta Informationondemographicvariables,educationallevel, Ricafoundnoassociationbetweenconsumptionoffriedfoods smoking,andphysicalactivitywasobtainedthroughinterview andriskofnon-fatalacutemyocardialinfarction.10Another atrecruitment.Physicalactivitywasassessedwithavalidated case-controlstudy,INTERHEART,observedapositive questionnairethatincludedactivityatwork,athome,andduring associationbetweenfriedfoodsandacutemyocardial leisuretime.22Participantswerealsoaskediftheyhaddiabetes infarction.11 mellitus,hypercholesterolaemia,hypertension,cancer,orangina, Therefore,althoughinterestintheassociationsofdietary orhadexperiencedmyocardialinfarctionorstroke.Inaddition patternsandspecificcookingtechniqueswithriskofdiseaseis theywereaskedtoreporttheirmenopausalstatusandanyuse increasing,12littleisknownaboutthehealtheffectsoffried oforalcontraceptivesorhormonereplacementtherapy.Trained foods.Weevaluatedprospectivelytheassociationbetween observersusedstandardisedprocedurestomeasureweight, consumptionoffriedfoodsandtheriskofcoronaryheartdisease height,andwaistcircumference.Thebodymassindexwas intheEPIC-Spaincohort. calculatedasweight(kg)dividedbyheightsquared(m2). Methods Ascertainment and validation of coronary heart disease ThemethodsoftheEPICprojecthavebeenreported Atbaseline,prevalentcoronaryheartdiseasewasselfreported. elsewhere.713-16Forthepresentanalysisweusedthedatafrom Duringfollow-up,weascertainedincidentcoronaryheartdisease theSpanishcohortofEPIC,whichincluded41438healthy withatelephonequestionnaire(atthreeyearsafterrecruitment) adults(15632men),aged29-69.Studyparticipantswere andthroughrecordlinkagewiththreesourcesofinformation: recruitedbetween1992and1996infiveSpanishregions.Three hospitaldischargedatabases,populationbasedmyocardial oftheseregionsarelocatedinthenorth(Asturias,Gipuzkoa, infarctionregistries(availableinMurcia,Navarra,and andNavarra)andtwointhesouth,includingtheMediterranean Gipuzkoa),andregionalmortalityregistriesandthenational shore(GranadaandMurcia).Selectionoftheseregionsaimed mortalitydatabase(managedbytheNationalStatisticalInstitute, tomaximisevariabilityindietbecausefoodandnutrientintake Madrid,Spain),whichprovidedinformationondateandcause hastraditionallybeendifferentbetweenthenorthandthesouth ofdeath. ofSpainandbetweentheinnerregionsandcoastalareas.To increasevariabilityindiet,participantswerealsorecruitedin Coronaryheartdiseasewasclassifiedaccordingtothe urbanandruralareasfromvariouseducationalandsocial InternationalClassificationofDiseases,ninthrevision(ICD-9 sectors.Studyparticipantsincludedmostlyblooddonors,civil codes410-414)and10threvision(ICD-10codesI20-I25).A servants,andthegeneralpopulation.Participationratesvaried teamoftraineddoctorsandnursesvalidatedcoronaryheart from55%to60%betweenregions.Theexposuresofinterest diseaseeventsagainsthospitalrecordsandautopsyreports. wereassessedatbaseline,andparticipantswerefollowedup Coronaryheartdiseaseeventswereclassifiedonthebasisof until31December2004.Allparticipantsgavewritteninformed symptoms,signs,biomarkers,andfindingsonelectrocardiogram consentbeforeenrolment. oratautopsy,accordingtostandardcriteria.23Adefinite coronaryheartdiseaseeventwasdefinedasonemeetingall Assessment of food consumption relevantcriteriaforafatalornon-fatalacutemyocardial Usingacomputeriseddietaryhistoryquestionnaire,17previously infarction,oranginarequiringrevascularisation(coronaryartery validatedinSpain18-20andadministeredbytrainedinterviewers, bypassgraftorpercutaneoustransluminalcoronaryangioplasty). Wedefinedprobableandpossibleacutecoronaryheartdisease weobtaineddataonusualfoodconsumptionforeachparticipant. eventsascaseswhereelectrocardiogramfindingswere Weusedhouseholdmeasures,standardunits,andacollection non-specificorthebiomarkerswereequivocalormissing.13 of35setsofphotographsofsimplefoods,mixedfoods,and drinkstoquantifytheportionofeachfood(g/day)consumed. Statistical analysis Eachinterviewondietaryhistorylastedabout40-50minutes. Ofthe41438participantsinthecohort,weexcludedthe followingfromtheanalyses:193forhavingcoronaryheart No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe:http://www.bmj.com/subscribe BMJ2012;344:e363doi:10.1136/bmj.e363(Published24January2012) Page3of10 RESEARCH diseaseatbaseline;167foranimplausiblyhighorlowdietary diagnosedhypertension.Amongwomen,thefrequencyof consumption,definedas3standarddeviationseitherwayfrom menopausalstatusdecreasedwiththeconsumptionoffriedfood thecohortmean(<788kcal/dayor>5710kcal/day);and321 (table2). forlackingdataonimportantvariablessuchasdateofcoronary Duringamedianfollow-upof11years,606definitecoronary heartdiseaseevent(n=12),smoking(n=22),diabetesmellitus heartdiseaseeventsoccurred(466myocardialinfarctionsand (n=71),hypercholesterolaemia(n=197),andhypertension 140anginasrequiringrevascularisation).Therewerealso712 (n=60).Thustheanalyseswerecarriedouton40757 definite,possible,orprobablecoronaryheartdiseaseeventsand participants. 1135deathsfromallcauses. WeusedCoxregressiontoobtainhazardratiosforcoronary Fordefinitecoronaryheartdiseaseevents,noassociationwith heartdiseaseaccordingtosexspecificquartersoffriedfood friedfoodconsumptionwasobservedinanalysesadjustingfor consumption.Thequarter(first)withlowestconsumptionwas energyintake,age,sex,andcentre(table3⇓,model1).The usedasreference.IntheCoxmodels,agewastheunderlying hazardratioofcoronaryheartdiseaseforthehighestversusthe timevariable,withentrytimedefinedastheparticipant’sage lowestquarteroffriedfoodconsumptionwas0.94(95% atrecruitmentandexittimeastheageatthecoronaryheart confidenceinterval0.72to1.23;Pfortrend0.52).Similarresults diseaseevent,death,or31December2004,whichevercame wereobtainedafteradditionaladjustmentforthemain first.Toreduceviolationsoftheproportionalhazards confounders(table3,model2):1.11(0.84to1.46;Pfortrend assumptionwealsostratifiedmodelsbyageatrecruitment(five 0.60).Finally,noassociationwasobservedbetweenfriedfood yeargroups),sex,andcentre.Webuiltthreemodels,with andincidentcoronaryheartdiseaseafteradjustingforpossible progressiveadjustmentforpotentialconfounderstodetermine mediatorssuchasbodymassindex,waistcircumference,and theirinfluenceontheresults.Model1wasadjustedforenergy hypertension(table3,model3).Comparedwiththefirst(lowest) intakeandmodel2wasadditionallyadjustedforeducational quarteroffriedfoodconsumption,themultivariatehazardratios level,smoking,physicalactivity(atwork,athome,andduring ofcoronaryheartdiseasewere1.15(0.91to1.45)inthesecond leisuretime),diabetesmellitus,hyperlipidaemia,cancer,oral quarter,1.07(0.83to1.38)inthethirdquarter,and1.08(0.82 contraceptiveuse,menopause,hormonereplacementtherapy, to1.43)inthefourthquarter(Pfortrend0.74).Lastly,afterfull ethanolintake,andconsumptionofnon-friedfoods(vegetables, adjustment,a100gincreaseintheconsumptionoffriedfood fruits,nuts,dairyproducts,meat,andfish).Model3was didnotshowanassociationwithriskofcoronaryheartdisease additionallyadjustedforpossiblemediatorsofthestudy (hazardratio1.00,0.90to1.11).Theresultsweresimilarafter associationreportedintheliterature:bodymassindex,waist censoringthefirsttwoyearsoffollow-upandafterexcluding circumference,andhypertension.Wecategorisedcontinuous thosewhoreportedachangeintheirdietduringtheprevious variablesintoquarters,exceptforage,whichwasanalysedas year.Theresultsdidnotvarybetweenthosewhousedoliveoil acontinuousvariable.ThePforlineartrendwascalculatedby forfryingandthosewhousedsunfloweroilorothervegetable modelingthequartersforfriedfoodasacontinuousvariable. oils(Pforinteraction0.22),orbetweenthesexes(Pfor Moreover,wealsocalculatedhazardratiosofcoronaryheart interaction0.19). diseaseandofallcausemortalityfora100gincreaseinfried Similarresultswerealsoobtainedinthecombinedanalysisof foodconsumptionanalysedasacontinuousvariable. definite,possible,andprobablecoronaryheartdiseaseevents. Newanalysesweredoneaftercensoringthefirsttwoyearsof Comparedwiththefirst(lowest)quarteroffriedfood follow-upandafterexcludingthosewhoreportedachangein consumption,themultivariatehazardratioofcoronaryheart theirdietduringthepreviousyear.Totestwhetherthestudy diseaseinthefullyadjustedmodel(model3)was1.08(0.88to associationvariedwithtypeofoilusedforfrying(oliveoil 1.34)forthesecondquarter,1.06(0.84to1.33)forthethird versussunfloweroilorothervegetableoils),weusedthe quarter,and1.04(0.81to1.34)forthefourthquarter(Pfor likelihoodratiotest,whichcomparedmodelswithandwithout trend0.83). interactionterms(productsofthequartersforfriedfoodand Friedfoodconsumptionalsofailedtobeassociatedwithall typeofoil).Wealsousedthesameproceduretotestwhether causemortality.Comparedwiththelowestquarteroffriedfood resultsvariedbysex. consumption,thoseinthehighestquarterhadahazardratioof StatisticalsignificancewassetatP<0.05.Theanalyseswere 0.93(0.77to1.14;Pfortrend0.98).Moreover,noassociation carriedoutwithSAS,version9.1forWindows.Thismanuscript wasobservedbetweena100gincreaseinintakeoffriedfood followstherecommendationsoftheStrengtheningtheReporting andtotalmortality(1.00,0.92to1.08;table3,model3). ofObservationalStudiesinEpidemiology(STROBE) Ofthetotalamountoffriedfoodconsumed,24%(34g/day) initiative.24 wasfish,22%(31g/day)meat,21%(30g/day)potatoes,and 11%(15g/day)eggs.Noassociationwasobservedbetween Results eachofthesefriedfoodgroupsandincidentcoronaryheart Anaverage138goffriedfoodwasconsumeddaily,including disease(table4⇓).Comparingthehighestandlowestquarters, 14gofoilusedforfrying.About7%ofthetotalamountof themultivariatehazardratioforthefullyadjustedmodelwas foodconsumedwasfried.Friedfoodconsumptionrangedfrom 1.13(0.89to1.44;Pfortrend0.26)forfriedfish,1.09(0.82to 0-817g/dayformenand0-657g/dayforwomen.Sixtytwoper 1.43;Pfortrend0.32)forfriedmeat,0.90(0.70to1.15;Pfor centofparticipantsusedoliveoilforfryingandtherestused trend0.45)forfriedpotatoes,and0.87(0.68to1.13;Pfortrend sunfloweroilorothervegetableoils. 0.30)forfriedeggs.Thesefriedfoodsalsofailedtoshowan associationwithallcausemortality(table4). Participantswithahigherfriedfoodintakewereyounger,ofa higherlevelofeducation,moreoftensmokers,andlessoften Discussion sedentaryatwork(table1⇓).Moreover,theyhadalower prevalenceofdiabetesmellitusandhyperlipidaemia,ahigher ThisanalysisoftheSpanishcohortoftheEuropeanProspective energyandethanolintake,andalowerconsumptionofnon-fried InvestigationintoCancerandNutritionfoundnoassociation vegetables,fruit,driedfruitandnuts,milkproducts,and betweenconsumptionoffriedfoodandriskofcoronaryheart non-friedfish(table2⇓).Theyalsohadalowerfrequencyof diseaseorallcausemortality. No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe:http://www.bmj.com/subscribe BMJ2012;344:e363doi:10.1136/bmj.e363(Published24January2012) Page4of10 RESEARCH Comparison with other studies Othermechanismsmaybeoperatingduringthefryingprocess. Forexample,inaninvitrostudyfriedeggsincreasedthe Fewstudieshaveevaluatedthisassociation.Ourresultsare productionofangiotensinconvertingenzymeinhibitorypeptides consistentwiththoseofacase-controlstudyofnon-fatal morethanboiledeggsdid.33Ontheotherhand,fryingalso myocardialinfarctioninCostaRica,whichobtainedanodds increasesthelevelsofcholesteroloxidationproducts34and ratioof1.06(0.59to1.91;Pfortrend0.65)forthehighest versuslowestfifthoffriedfoodconsumption.10However,in reducesparaoxonaseactivity,anenzymethatinhibitsoxidation oflowdensitylipoproteincholesterol.35 theCostaRicaninvestigation,palmandpartiallyhydrogenated soyoilwereusedforfrying,whereasinthisstudymainlyolive ThisanalysisoftheEPIC-Spaincohortsuggeststhattheabove andsunfloweroils—themostcommonlyusedoilsin mentionedmechanismsmayoffseteachotherinthegeneral Mediterraneancountries—wereused. population,withtheresultthatfryingwithmainlyoliveoilor sunfloweroilisnotassociatedwithahigherriskofcoronary OurresultsdifferfromthoseoftheINTERHEARTstudy,where heartdisease.Fryingwithothertypesoffatsmaystillbe friedfoodconsumptionwasassociatedwithahigherriskof acutemyocardialinfarction.11However,somemethodological harmful. differenceslimitthecomparisonofresults.Forinstance, Limitations and strengths of the study INTERHEARTdidnotassessquantitativelytheintakeoffried food;moreover,itrecordedtheconsumptionofonlyninefried Thisstudyhassomelimitations.Firstly,dietwasmeasuredonly foodsanditdidnotreportthetypeofoilusedforfrying. atbaseline.Secondly,someresidualconfoundingcannotbe InacrosssectionalanalysisoftheMultiEthnicStudyof ruledout,sincediabetesmellitus,hypertension,and Atherosclerosis(MESA),noassociationwasfoundbetween hypercholesterolaemiawereselfreported.InSpain,however, consumptionoffriedfishandsubclinicalatherosclerosis, evidenceexistsofthediagnosticvalidityofselfreported assessedbythethicknessofintima-mediainthecommon diabetesmellitusandhypertension.3637Thirdly,somedegreeof carotid.Norwasfriedfishconsumptionassociatedwith under-reportingoffriedfoodconsumptionispossible,given significantchangesinserumlevelsoftotalcholesterol,low thatpeopleperceiveitasunhealthy;andnon-differential densitylipoproteincholesterol,highdensitylipoprotein under-reportingwouldbiasthehazardratiotowards1,hiding cholesterol,ortryglicerides.25TheCardiovascularHealthStudy, apossibleassociation.Fourthly,wedonotknowthetypeof acohortof3919olderpeople(≥65years)followedfornine fryingprocedure(deeporpan)carriedout,theextentthatoils years,alsodidnotfindanassociationofconsumptionoffried werereused,andthetimeandtemperatureusedforfrying.Itis fishorfishsandwicheswithmortalityfromcoronaryheart possiblethatthesefactorsaredifferentiallyassociatedwith diseaseorcardiacarrhythmia,orwithincidenceofnon-fatal coronaryheartdisease.Fifthly,wecannotseparatetheeffectof acutemyocardialinfarction.26 aspecificfoodfromtheeffectofthecookingmethodused.For instance,wecannotseparatetheeffectoffryingwithoil(for Possible mechanisms of the study example,oliveoil)fromthecardioprotectiveeffectofomega association 3fattyacidsinfish.Sixthly,resultsarebasedonwhatpeople atebetween1992and1996.Unfortunately,sincethennostudies Thefryingprocessiscomplexandasyetisnotwellunderstood. havebeenpublishedontheconsumptionoffriedfoodbythe Laboratoryinvestigationsshowthatfriedfoodsmayactthrough Spanishpopulation.Datafromhouseholdconsumptionsurveys, manymechanisms,suchthattheresultingeffectoncoronary however,indicatethatthemacronutrientintakehasnotchanged heartdiseaseisdifficulttoanticipate.Fryingchangesthequality materiallyfrom1991to200638;also,overalladherencetothe andcompositionoffood.Forexample,fryingusuallyaugments Mediterraneanpatternseemsfairlystableintheperiod1998to thefatcontent,andforsomefoods,suchaspotatoes,thedietary 2005,39althoughanincreaseintheconsumptionofnon-alcoholic fibreisincreasedowingtotheformationofresistantstarch.1 beveragesanddairyproductsandadecreaseinconsumptionof Fryingcanspecificallyincreasetheamountoftransfattyacids potatoesandlegumeshasbeenobservedoverthepasttwo infoods.27Theamountdependsonthefryingtechnique(deep decades.38Finally,althoughaverageconsumptionoffriedfood orpanfrying),thedegreeofthermaldegradationoftheoil,the inthehighestquarterwasfairlyhigh,ourresultscannotrule typeoffood,and,aboveall,thetypeofoil(thedegreeofoil outthatafriedfoodintakehigherthanthatobservedinthis unsaturationincreasestheformationoftransfattyacids).28 studyisassociatedwithcoronaryheartdisease. However,somefoodssuchascertaintypesoffishshowminimal Thisanalysisalsohasimportantstrengths,suchastheuseofa variationsinthecontentoftransfattyacidsafterfrying, regardlessofwhetheroliveoilorsunfloweroilisused.29Onthe validateddietaryhistorywithquantitativeinformationonthe consumptionofalargenumberoffriedfoods,thelargesample otherhand,theuseofsolidfatsforfryingcanalsoincreasethe size,longfollow-up,andadjustmentfornumerouspotential contentoftransfattyacids. confounders. Theassociationbetweenconsumptionoffriedfoodsand endothelialdamageiscontroversial.Inacrosssectionalanalysis Conclusions oftheMultiEthnicStudyofAtherosclerosis(MESA), consumptionoffriedfishwasinverselyassociatedwiththe InaMediterraneancountrywhereoliveandsunfloweroilsare levelofsolubleintercellularadhesionmolecule-1,amarkerof themostcommonlyusedfatsforfrying,andwherelarge endothelialactivation.Theauthorsdidnotexpectthisfinding, amountsoffriedfoodsareconsumedbothatandawayfrom sinceitwasnotobservedforconsumptionofnon-friedfish.30 home,noassociationwasobservedbetweenfriedfood Incontrast,mealsrichinoilspreviouslyusedfordeepfrying consumptionandtheriskofcoronaryheartdiseaseordeath. infastfoodrestaurantshavebeenassociatedwithendothelial OurresultsaredirectlyapplicableonlytoMediterranean damage.31Inanotherstudy,however,foodsrichinoils countrieswithfryingmethodssimilartothoseinSpain.Firstly, previouslyusedforfryingandhighinoxidationproducts oil(mainlyoliveandsunflower)ratherthansolidfatisusedfor increasedpostprandialtriglyceridaemiabutdidnotaffect fryinginSpain.Itiswellestablishedthatoliveoilislessprone endothelialfunction.32 tooxidationthanotheredibleoilsorfats.4041Secondly, consumptionoffriedfoodsinSpainisnotaproxyforfastfood No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe:http://www.bmj.com/subscribe BMJ2012;344:e363doi:10.1136/bmj.e363(Published24January2012) Page5of10 RESEARCH intake.Fastfoodsaregenerallypreparedbydeepfryingwith 11 IqbalR,AnandS,OunpuuS,IslamS,ZhangX,RangarajanS,etal.Dietarypatternsand theriskofacutemyocardialinfarctionin52countries:resultsoftheINTERHEARTstudy. oilsusedseveraltimes,andareconsumedmostlyawayfrom Circulation2008;118:1929-37. home.InSpain,friedfoodsareconsumedbothatandaway 12 AstrupA,DyerbergJ,ElwoodP,HermansenK,HuFB,JakobsenMU,etal.Theroleof reducingintakesofsaturatedfatinthepreventionofcardiovasculardisease:wheredoes fromhome,andbothdeepfryingandpanfryingareused. theevidencestandin2010?AmJClinNutr2011;93:684-8. Moreover,wecanassumethatoilisnotreusedmanytimesfor 13 DaneshJ,SaracciR,BerglundG,FeskensE,OvervadK,PanicoS,etal.EPIC-Heart: foodsconsumedathome;however,thecardiovasculareffects thecardiovascularcomponentofaprospectivestudyofnutritional,lifestyleandbiological factorsin520,000middle-agedparticipantsfrom10Europeancountries.EurJEpidemiol offoodfriedwithoverlyreusedoilsmeritfurtherresearch. 2007;22:129-41. Finally,consumptionoffriedsnackshighinsaltisfairlylow 14 RiboliE,KaaksR.TheEPICproject:rationaleandstudydesign.EuropeanProspective InvestigationintoCancerandNutrition.IntJEpidemiol1997;26(suppl1):S6-14. inSpain,whereasinothercountriessuchastheUnitedStates 15 BinghamS,RiboliE.Dietandcancer—theEuropeanProspectiveInvestigationintoCancer theyprovideanimportantpercentageofenergyintake.42 andNutrition.NatRevCancer2004;4:206-15. 16 RiboliE,HuntKJ,SlimaniN,FerrariP,NoratT,FaheyM,etal.EuropeanProspective InvestigationintoCancerandNutrition(EPIC):studypopulationsanddatacollection. Contributors:CM-IiscoordinatoroftheEuropeanProspective PublicHealthNutr2002;5:1113-24. 17 BurkeBS.Thedietaryhistoryasatoolinresearch.JAmDietAssoc1947;23:104-11. InvestigationofCancerandNutrition(EPIC)-HEART-Spain.CAGis 18 EPICgroupofSpain.Relativevalidityandreproducibilityofadiethistoryquestionnaire coordinatoroftheEPIC-Spain.CM-IandCAGcontributedequallyto inSpain.II.Nutrients.EPICGroupofSpain.EuropeanProspectiveInvestigationinto CancerandNutrition.IntJEpidemiol1997;26(suppl1):S100-9. thestudy.PG-CandFR-Aconceivedthestudyanddraftedthe 19 EPICgroupofSpain.Relativevalidityandreproducibilityofadiethistoryquestionnaire manuscript.Theyaretheguarantors.PG-C,FR-A,EL-GandLML-M inSpain.I.Foods.EPICGroupofSpain.EuropeanProspectiveInvestigationintoCancer analysedthedata.PA,EA,LA,AB,GB,M-DC,MD,J-MH,NL,PM, andNutrition.IntJEpidemiol1997;26(suppl1):S91-9. 20 EPICgroupofSpain.Relativevalidityandreproducibilityofadiethistoryquestionnaire CM,EM,CN,JRQ,LR,andMJSprovidedtheoriginaldataand inSpain.III.Biochemicalmarkers.EPICGroupofSpain.EuropeanProspective informationontherespectivepopulationsandhelpeddesignandanalyse InvestigationintoCancerandNutrition.IntJEpidemiol1997;26(suppl1):S110-7. 21 SlimaniN,TorrentM,FarriollN.EuropeanProspectiveInvestigationintoCancerand thestudy.AB,MD,CN,JRQ,MJS,andCAGweretheprincipal Nutrition(EPIC):foodcompositiontables—Spain.InternationalAgencyforResearchon investigatorsforEPIC.Allauthorsinterpretedtheresultsandcontributed Cancer,1991. 22 HaftenbergerM,SchuitAJ,TormoMJ,BoeingH,WarehamN,Bueno-de-MesquitaHB, towritingthemanuscriptandtakeresponsibilityfortheintegrityofthe etal.Physicalactivityofsubjectsaged50-64yearsinvolvedintheEuropeanProspective dataandtheaccuracyofthedataanalysis. InvestigationintoCancerandNutrition(EPIC).PublicHealthNutr2002;5:1163-76. 23 LuepkerRV,AppleFS,ChristensonRH,CrowRS,FortmannSP,GoffD,etal.Case Funding:ThisstudywasfundedbyresearchgrantsfromFIS(PI04-0257, definitionsforacutecoronaryheartdiseaseinepidemiologyandclinicalresearchstudies: PI06-0366,PI04-2342,PI04-1822,PI04-1821,PI04-2188,and astatementfromtheAHACouncilonEpidemiologyandPrevention;AHAStatistics Committee;WorldHeartFederationCouncilonEpidemiologyandPrevention;theEuropean PI08-0166);RETIC(RD06/0020)oftheInstitutodeSaludCarlosIII; SocietyofCardiologyWorkingGrouponEpidemiologyandPrevention;CentersforDisease andfundsfromtheSpanishregionalgovernmentsofAndalusia,Asturias, ControlandPrevention;andtheNationalHeart,Lung,andBloodInstitute.Circulation 2003;108:2543-9. BasqueCountry,Murcia,andNavarra,andtheCatalanInstituteof 24 VonElmE,AltmanDG,EggerM,PocockSJ,GotzschePC,VandenbrouckeJP. Oncology.EL-GhasaRamónyCajalcontractfromtheMinistryof StrengtheningtheReportingofObservationalStudiesinEpidemiology(STROBE) Education.Authorsofthisreport,andnotthefundingagencies,are statement:guidelinesforreportingobservationalstudies.BMJ2007;335:806-8. 25 HeK,LiuK,DaviglusML,Mayer-DavisE,JennyNS,JiangR,etal.Intakesoflong-chain responsibleforitscontent.Thefunderdidnothaveanyroleinthestudy n-3polyunsaturatedfattyacidsandfishinrelationtomeasurementsofsubclinical design;collection,analysis,andinterpretationofthedata;thewriting atherosclerosis.AmJClinNutr2008;88:1111-8. 26 MozaffarianD,LemaitreRN,KullerLH,BurkeGL,TracyRP,SiscovickDS.Cardiac ofthereport;andthedecisiontosubmitthearticleforpublication. benefitsoffishconsumptionmaydependonthetypeoffishmealconsumed:the Competinginterests:AllauthorshavecompletedtheICMJEuniform CardiovascularHealthStudy.Circulation2003;107:1372-7. 27 LitinL,SacksF.Trans-fatty-acidcontentofcommonfoods.NEnglJMed disclosureformatwww.icmje.org/coi_disclosure.pdf(availableon 1993;329:1969-70. requestfromthecorrespondingauthor)anddeclare:nosupportfrom 28 BoskouG,SaltaFN,ChiouA,TroullidouE,AndrikopoulosNK.Contentof trans,trans-2,4-decadienalindeep-friedandpan-friedpotatoes.EurJLipidSciTechnol anyorganisationforthesubmittedwork;nofinancialrelationshipswith 2006;108:109-15. anyorganisationsthatmighthaveaninterestinthesubmittedworkin 29 AnsorenaD,GuembeA,MendizabalT,AstiasaranI.Effectoffishandoilnatureonfrying processandnutritionalproductquality.JFoodSci2010;75:H62-7. thepreviousthreeyears;andnootherrelationshipsoractivitiesthat 30 HeK,LiuK,DaviglusML,JennyNS,Mayer-DavisE,JiangR,etal.Associationsofdietary couldappeartohaveinfluencedthesubmittedwork. long-chainn-3polyunsaturatedfattyacidsandfishwithbiomarkersofinflammationand endothelialactivation(fromtheMulti-EthnicStudyofAtherosclerosis[MESA]).AmJ Ethicalapproval:Thisstudywasapprovedbytheethicscommitteesof Cardiol2009;103:1238-43. theInternationalAgencyforResearchonCancer(Lyon,France)and 31 WilliamsMJ,SutherlandWH,McCormickMP,deJongSA,WalkerRJ,WilkinsGT. Impairedendothelialfunctionfollowingamealrichinusedcookingfat.JAmCollCardiol BellvitgeHospital(Barcelona,Spain). 1999;33:1050-5. Datasharing:Noadditionaldataavailable. 32 WilliamsMJ,SutherlandWH,McCormickMP,YeomanD,deJongSA,WalkerRJ.Normal endothelialfunctionaftermealsrichinoliveorsaffloweroilpreviouslyusedfordeepfrying. NutrMetabCardiovascDis2001;11:147-52. 1 FillionL,HenryCJ.Nutrientlossesandgainsduringfrying:areview.IntJFoodSciNutr 33 MajumderK,WuJ.AngiotensinIconvertingenzymeinhibitorypeptidesfromsimulated 1998;49:157-68. invitrogastrointestinaldigestionofcookedeggs.JAgricFoodChem2009;57:471-7. 2 PokornJ,PanekJ,TrojakovaL.Effectoffoodcomponentchangesduringfryingonthe 34 EcharteM,AnsorenaD,AstiasaranI.Fattyacidmodificationsandcholesteroloxidation nutritionvalueoffriedfood.ForumNutr2003;56:348-50. inporkloinduringfryingatdifferenttemperatures.JFoodProt2001;64:1062-6. 3 Sanchez-MunizFJ,CuestaC,Garrido-PolonioMC.Evaluationofasunfloweroilusedfor 35 SutherlandWH,WalkerRJ,deJongSA,vanRijAM,PhillipsV,WalkerHL.Reduced fryingbydifferentanalyticalindexesandcolumnandgaschromatography.Z postprandialserumparaoxonaseactivityafteramealrichinusedcookingfat.Arterioscler Ernahrungswiss1994;33:16-23. ThrombVascBiol1999;19:1340-7. 4 SamraRA.Fatsandsatiety.In:MontmayeurJP,leCoutreJ,eds.Fatdetection:taste, 36 HuertaJM,TormoMJ,Egea-CaparrosJM,Ortola-DevesaJB,NavarroC.Accuracyof texture,andpostingestiveeffects.CRCPress,2010. self-reporteddiabetes,hypertensionandhyperlipidemiaintheadultSpanishpopulation. 5 DrewnowskiA,Almiron-RoigE.Humanperceptionandpreferencesforfat-richfoods.In: DINOstudyfindings.RevEspCardiol2009;62:143-52. MontmayeurJP,leCoutreJ,eds.Fatdetection:taste,texture,andpostingestiveeffects. 37 TormoMJ,NavarroC,ChirlaqueMD,BarberX.Validationofselfdiagnosisofhighblood CRCPress,2010. pressureinasampleoftheSpanishEPICcohort:overallagreementandpredictivevalues. 6 SoriguerF,Rojo-MartinezG,DobarganesMC,GarciaAlmeidaJM,EstevaI,BeltranM, EPICGroupofSpain.JEpidemiolCommunityHealth2000;54:221-6. etal.Hypertensionisrelatedtothedegradationofdietaryfryingoils.AmJClinNutr 38 Varela-MoreirasG,AvilaJM,CuadradoC,delPozoS,RuizE,MoreirasO.Evaluation 2003;78:1092-7. offoodconsumptionanddietarypatternsinSpainbytheFoodConsumptionSurvey: 7 Guallar-CastillonP,Rodriguez-ArtalejoF,FornesNS,BanegasJR,EtxezarretaPA, updatedinformation.EurJClinNutr 2010;64(suppl3):S37-43. ArdanazE,etal.IntakeoffriedfoodsisassociatedwithobesityinthecohortofSpanish 39 Bach-FaigA,Fuentes-BolC,RamosD,CarrascoJL,RomanB,BertomeuIF,etal.The adultsfromtheEuropeanProspectiveInvestigationintoCancerandNutrition.AmJClin MediterraneandietinSpain:adherencetrendsduringthepasttwodecadesusingthe Nutr2007;86:198-205. MediterraneanAdequacyIndex.PublicHealthNutr2011;14:622-8. 8 Sayon-OreaC,Bes-RastrolloM,Basterra-GortariFJ,BeunzaJJ,Guallar-CastillonP,de 40 Gallina-ToschiT,CerreteniL,BendiniA,Bonoli-CarbogninM,LerckerG.Oxidativestability laFuente-ArrillagaC,etal.ConsumptionoffriedfoodsandweightgaininaMediterranean andphenoliccontentofvirginoliveoil:ananalyticalapproachbytraditionalandhigh cohort:theSUNproject.NutrMetabCardiovascDis2011:publishedonline6Aug. resolutiontechniques.JSepSci2005;28:859-70. 9 DonfrancescoC,LoNC,BrignoliO,RiccardiG,CiccarelliP,DimaF,etal.Italiannetwork 41 CasalS,MalheiroR,SendasA,OliveiraBPP,PereiraJA.Oliveoilstabilityunder forobesityandcardiovasculardiseasesurveillance:apilotproject.BMCFamPract deep-fryingconditions.FoodChemToxicol2010;48:2972-9. 2008;9:53. 42 NielsenSJ,Siega-RizAM,PopkinBM.TrendsinenergyintakeinUSbetween1977and 10 KabagambeEK,BaylinA,SilesX,CamposH.Individualsaturatedfattyacidsandnonfatal 1996:similarshiftsseenacrossagegroups.ObesRes2002;10:370-8. acutemyocardialinfarctioninCostaRica.EurJClinNutr2003;57:1447-57. Accepted:14November2011 No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe:http://www.bmj.com/subscribe BMJ2012;344:e363doi:10.1136/bmj.e363(Published24January2012) Page6of10 RESEARCH Whatisalreadyknownonthistopic Theconsumptionoffriedfoodhasbeenassociatedwithsomecardiovascularriskfactors,suchashypertension,obesity,andlowlevels ofhighdensitylipoproteincholesterol Theassociationbetweenfriedfoodconsumptionandcoronaryheartdiseasehasbeenevaluatedinsomecase-controlstudies,with heterogeneousresults Whatthisstudyadds InSpain,aMediterraneancountrythatusesoliveorsunfloweroilforfrying,theconsumptionoffriedfoodswasnotassociatedwithrisk ofcoronaryheartdiseaseorallcausemortality Fryingwithothertypesoffats,reusingoilsseveraltimes,orconsumingfriedsnackshighinsaltmaystillbeharmful Citethisas:BMJ2012;344:e363 anymedium,providedtheoriginalworkisproperlycited,theuseisnoncommercialand isotherwiseincompliancewiththelicense.See:http://creativecommons.org/licenses/by- Thisisanopen-accessarticledistributedunderthetermsoftheCreativeCommons nc/2.0/andhttp://creativecommons.org/licenses/by-nc/2.0/legalcode. AttributionNon-commercialLicense,whichpermitsuse,distribution,andreproductionin No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe:http://www.bmj.com/subscribe BMJ2012;344:e363doi:10.1136/bmj.e363(Published24January2012) Page7of10 RESEARCH Tables Table1|Consumptionoffriedfoods,sociodemographicvariables,andlifestylesofstudyparticipantsatbaseline,byquartersoffriedfood consumption.Valuesaremeansunlessstatedotherwise Quartersoffriedfoodconsumption Characteristics 1(lowest) 2 3 4(highest) Consumptionoffriedfoods Totalamountoffriedfoodconsumed(g/day) 47.0 105.7 158.4 249.6 Friedfish(g/day) 12.6 28.8 40.6 53.6 Friedmeat(g/day) 8.3 21.7 36.2 59.9 Friedpotatoes(g/day) 9.3 21.6 33.3 57.7 Friedeggs(g/day) 4.0 10.7 17.1 28.0 Sociodemographicvariables Age(years) 50.4 49.5 48.9 48.3 Educationallevel(%): Noformaleducation 41.9 36.4 31.7 27.6 Primaryeducation 31.6 35.9 40.8 46.6 Secondaryeducation 12.6 14.2 16.0 16.0 Universityeducation 13.3 12.8 10.8 9.0 Lifestyle Smoking(%): Neversmoker 58.3 59.3 59.5 59.7 Currentsmoker 22.5 22.5 23.7 24.2 Formersmoker 19.2 18.2 16.8 16.1 Physicalactivityatwork(%): Sedentary 22.8 22.1 20.4 18.8 Standing 62.1 64.2 63.7 63.4 Manuallabour 6.2 8.5 10.3 12.3 Heavymanuallabour 2.2 1.9 2.1 2.5 Doesnotwork 3.8 3.3 3.6 3.1 Physicalactivityathome(METsh/week) 68.6 69.8 70.0 68.1 Physicalactivityinleisuretime(METsh/week) 28.7 28.2 28.6 28.4 MET=metabolicequivalents.Rangeofquartersoffriedfoodconsumptioning/dayformen:firstquarter(0-112),secondquarter(113-176),thirdquarter(177-246), andfourthquarter(247-817),andforwomen:firstquarter(0-59),secondquarter(60-103),thirdquarter(104-154),andfourthquarter(155-657). No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe:http://www.bmj.com/subscribe BMJ2012;344:e363doi:10.1136/bmj.e363(Published24January2012) Page8of10 RESEARCH Table2|Morbidity,menopausalstatus,useofhormonetreatment,consumptionofnon-friedfood,andpotentialmediatorsofstudy associationamongstudyparticipantsatbaseline,byquartersoffriedfoodconsumption.Valuesaremeansunlessstatedotherwise Quartersoffriedfoodconsumption Characteristics 1(lowest) 2 3 4(highest) Diseasesdiagnosed: Diabetesmellitus(%) 7.5 5.1 3.9 3.2 Hyperlipidaemia(%) 24.8 20.5 18.5 16.7 Cancer(%) 1.0 0.9 0.8 0.8 Menopause(%)* 42.2 36.9 33.0 29.3 Hormonetreatment(%)*: Oralcontraceptives 24.7 26.4 26.6 27.4 Hormonesubstitutiontherapy 6.4 6.2 5.5 5.6 Othernutrientsandfoodsconsumed: Totalenergy(kcal/day) 1920.1 2133.0 2294.8 2566.6 Ethanol(g/day) 11.5 14.5 17.0 20.4 Non-friedvegetables(g/day) 255.0 241.1 232.3 237.4 Fruit(g/day) 336.6 322.0 315.3 311.2 Driedfruitandnuts(g/day) 5.0 4.6 4.2 3.9 Milkproducts(g/day) 304.7 298.1 296.9 292.0 Non-friedmeat(g/day) 93.9 98.0 97.9 96.2 Non-friedfish(g/day) 37.2 29.6 26.0 23.8 Mediatorsofstudyassociation: Bodymassindex(kg/m2) 28.5 28.3 28.1 28.1 Waistcircumference(cm) 92.8 92.0 91.4 91.4 Hypertension(%) 23.7 20.9 18.4 17.4 Seetable1forrange(g/day)ofquartersoffriedfoodconsumptionformenandforwomen. *Calculatedforwomen. No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe:http://www.bmj.com/subscribe BMJ2012;344:e363doi:10.1136/bmj.e363(Published24January2012) Page9of10 RESEARCH Table3|Associationbetweenfriedfoodconsumptionandincidenceofdefinitecoronaryheartdiseaseandallcausemortalityduring11 yearsoffollow-upinSpanishcohortofEuropeanProspectiveInvestigationintoCancerandNutrition Quartersoffriedfoodconsumption For100gincrease Variables 1(lowest) 2 3 4(highest) Pfortrend offriedfood Coronaryheartdisease: TotalNo 10188 10190 10190 10189 Personyears 109578 110394 110943 111762 Noofevents 154 163 150 139 Model1*:hazardratios 1 1.05(0.84to1.33) 0.97(0.76to1.24) 0.94(0.72to1.23) 0.52 0.95(0.86to1.06) (95%CI) Model2†:hazardratios 1 1.15(0.91to1.46) 1.08(0.84to1.39) 1.11(0.84to1.46) 0.60 1.01(0.91to1.12) (95%CI) Model3‡:hazardratios 1 1.15(0.91to1.45) 1.07(0.83to1.38) 1.08(0.82to1.43) 0.74 1.00(0.90to1.11) (95%CI) Allcausemortality: TotalNo 10188 10190 10190 10189 Personyears 110135 111019 111560 112342 Noofdeaths 328 276 273 258 Model1*:hazardratios 1 0.90(0.76to1.07) 0.91(0.77to1.09) 0.89(0.73to1.08) 0.26 0.98(0.91to1.06) (95%CI) Model2†:hazardratios 1 0.94(0.79to1.11) 0.95(0.80to1.14) 0.94(0.77to1.15) 0.60 1.00(0.92to1.08) (95%CI) Model3‡:hazardratios 1 0.94(0.79to1.11) 0.95(0.79to1.13) 0.93(0.77to1.14) 0.98 1.00(0.92to1.08) (95%CI) Quartersaresexspecific.Seetable1forrange(g/day)ofquartersoffriedfoodconsumptionformenandforwomen. *Adjustedforage,sex,centre(Asturias,Gipuzkoa,Granada,Murcia,orNavarra),andenergyintake(quarters). †Adjustedforvariablesinmodel1andforethanolconsumption(quartersamongconsumers),educationallevel(noformaleducation,primary,vocational,secondary, universityeducation),smoking(never,current,former),physicalactivityatwork(sedentary,standing,manuallabour,heavymanuallabour,nowork),physical activityathome(metabolicequivalentsh/weekinquarters),physicalactivityinleisuretime(metabolicequivalentsh/weekinquarters),diabetesmellitus, hyperlipidaemia,cancer,oralcontraceptives,menopause,hormonereplacementtherapy,andconsumptionoffruit(quarters),nuts(quarters),anddairyproducts (quarters),andnon-friedfoods(quarters):vegetables,meat,andfish. ‡Adjustedforvariablesinmodel2andforbodymassindex(<25and≥25,<30and≥30,<35and≥35),waistcircumference(quarters),andhypertension. No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe:http://www.bmj.com/subscribe BMJ2012;344:e363doi:10.1136/bmj.e363(Published24January2012) Page10of10 RESEARCH Table4|Associationbetweenconsumptionofmaintypesoffriedfoodandincidenceofdefinitecoronaryheartdiseaseandallcause mortalityduring11yearsoffollow-upinSpanishcohortofEuropeanProspectiveInvestigationintoCancerandNutrition Quartersoffriedfoodconsumption For100gincrease Variables 1(lowest) 2 3 4(highest) Pfortrend offriedfood Coronaryheartdisease Friedfish: Noofevents 144 136 149 177 — — Hazardratio(95%CI)* 1 0.99(0.78to1.26) 1.06(0.83to1.35) 1.13(0.89to1.44) 0.26 1.03(0.82to1.29) Friedmeat: Noofevents 137 158 174 137 — — Hazardratio(95%CI)* 1 1.17(0.93to1.49) 1.40(1.09to1.79) 1.09(0.82to1.43) 0.32 1.05(0.82to1.35) Friedpotatoes: Noofevents 188 151 138 129 — — Hazardratio(95%CI)* 1 0.92(0.74to1.15) 0.95(0.75to1.19) 0.90(0.70to1.15) 0.45 0.91(0.71to1.17) Friedeggs: Noofevents 184 151 145 126 — — Hazardratio(95%CI)* 1 0.99(0.80to1.24) 0.95(0.76to1.20) 0.87(0.68to1.13) 0.30 0.84(0.51to1.40) Allcausemortality Friedfish: Noofevents 297 276 271 291 — — Hazardratio(95%CI)* 1 0.99(0.84to1.17) 0.97(0.81to1.15) 0.92(0.77to1.10) 0.34 0.84(0.69to1.01) Friedmeat: Noofevents 346 246 264 279 — — Hazardratio(95%CI)* 1 0.80(0.67to0.95) 0.90(0.75to1.08) 0.94(0.78to1.14) 0.72 0.96(0.79to1.16) Friedpotatoes: Noofevents 353 273 249 260 — — Hazardratio(95%CI)* 1 0.90(0.77to1.06) 0.92(0.78to1.09) 0.99(0.83to1.18) 0.87 1.10(0.91to1.32) Friedeggs: Noofevents 350 265 245 275 — — Hazardratio(95%CI)* 1 0.96(0.81to1.13) 0.90(0.75to1.06) 1.05(0.87to1.25) 0.90 1.11(0.76to1.64) *Adjustedthesameasmodel3intable3. Rangeofquartersoffriedfoodconsumptioning/day:fishconsumptionformen:firstquarter(0-15),secondquarter(16-35),thirdquarter(36-61),andfourthquarter (62-439),andforwomen:firstquarter(0-8),secondquarter(9-23),thirdquarter(24-40),andfourthquarter(41-346);meatconsumptionformen:firstquarter (0-12),secondquarter(13-34),thirdquarter(35-65),andfourthquarter(66-384),andforwomen:firstquarter(0-2),secondquarter(3-15),thirdquarter(16-36), andfourthquarter(37-347);potatoconsumptionformen:firstquarter(0-12),secondquarter(13-33),thirdquarter(34-59),fourthquarter(60-560),andforwomen: firstquarter(0-3),secondquarter(4-17),thirdquarter(18-34),andfourthquarter(35-282);eggconsumptionformen:firstquarter(0-6),secondquarter(7-15), thirdquarter(16-30),fourthquarter(31-176),andforwomen:firstquarter(0-0),secondquarter(1-8),thirdquarter(9-17),andfourthquarter(18-123). No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe:http://www.bmj.com/subscribe