SSRN-id1359506

8/14/2019 SSRN-id1359506

1/33

NBER WORKING PAPER SERIES

CHANGES IN U.S. HOSPITALIZATION AND MORTALITY RATES FOLLOWING

SMOKING BANS

Kanaka D. Shetty

Thomas DeLeire

Chapin White

Jayanta Bhattacharya

Working Paper 14790

http://www.nber.org/papers/w14790

NATIONAL BUREAU OF ECONOMIC RESEARCH

1050 Massachusetts Avenue

Cambridge, MA 02138

March 2009

We thank seminar participants at the Research in Progress Seminar at Stanford Medical School for

their insights. We thank Dr. Alan Garber, Dr. Douglas Owens, and Dr. Mayer Brezis for their helpful

comments. Finally, we thank Dr. Catherine Su and Dr. Priya Pillutla for helpful comments on earlier

versions of this manuscript. The views in this paper are those of the authors and should not be interpreted

as those of the Congressional Budget Office. Dr. Shetty was supported by a U.S. Veterans Affairs'

Fellowship in Ambulatory Care Practice and Research. Dr. Bhattacharya thanks the U.S. National

Institute on Aging for partial funding. The authors have no relationships (financial or otherwise) with

any company making products relevant to this study. The views expressed herein are those of the author(s)

and do not necessarily reflect the views of the National Bureau of Economic Research.

2009 by Kanaka D. Shetty, Thomas DeLeire, Chapin White, and Jayanta Bhattacharya. All rights

reserved. Short sections of text, not to exceed two paragraphs, may be quoted without explicit permission

provided that full credit, including notice, is given to the source.

8/14/2019 SSRN-id1359506

2/33

Changes in U.S. Hospitalization and Mortality Rates Following Smoking Bans

Kanaka D. Shetty, Thomas DeLeire, Chapin White, and Jayanta Bhattacharya

NBER Working Paper No. 14790

March 2009

JEL No. I1,I18

ABSTRACT

U.S. state and local governments are increasingly restricting smoking in public places. This paper

analyzes nationally representative databases, including the Nationwide Inpatient Sample, to compare

short-term changes in mortality and hospitalization rates in smoking-restricted regions with control

regions. In contrast with smaller regional studies, we find that workplace bans are not associated with

statistically significant short-term declines in mortality or hospital admissions for myocardial infarction

or other diseases. An analysis simulating smaller studies using subsamples reveals that large short-term

increases in myocardial infarction incidence following a workplace ban are as common as the large

decreases reported in the published literature.

Kanaka D. Shetty

RAND Corporation

1776 Main Street

Santa Monica, CA 90401

[email protected]

Thomas DeLeire

La Follette School of Public Affairs

University of Wisconsin - Madison1225 Observatory Drive

Madison, WI 53706

[email protected]

Chapin White

Congressional Budget Office

2nd and D Streets

Washington, DC 20015

[email protected]

Jayanta Bhattacharya

117 Encina Commons

Center for Primary Careand Outcomes Research

Stanford University

Stanford, CA 94305-6019

and NBER

[email protected]

8/14/2019 SSRN-id1359506

3/33

4

1Introduction

Stateandlocalgovernmentshaveincreasinglybannedsmokinginpublicplaces

(includingworkplaces,restaurantsandbars)asameansoflimitingnonsmoker

exposureandofdiscouragingsmoking(CentersforDiseaseControlandPrevention

2007).Severalrecentstudiesinthemedicalliteratureusingasmallnumberofregions

suggestthatsmokingbansleadtoashortterm8to40%decreaseintheannual

incidenceofacutemyocardialinfarction(AMI)(Sargent,Shepardetal.2004;Bartecchi,

Alseveretal.2006;Cesaroni,Forastiereetal.2008).Despitethesefindings,itisunclear

howwelltheresultswouldtranslatetotypicalU.S.communities.Weexaminewhether

governmentalsmokingrestrictionsaffecthospitalizationandmortalityratesinalarge

sampleofU.S.communities.

WecalculatedeathandhospitalizationratesforAMIandotherdiseasesusing

MedicareProviderAnalysisandReview(MEDPAR)files,nationaldeathrecords

(otherwiseknownasthemultiplecauseofdeathfiles,hereafterMCD),and

hospitalizationdatafromtheHealthcareCostandUtilizationProjectsNationwide

InpatientSample(NIS).Wecompareratesbeforeandafterimplementationofthese

bansrelativetocommunitiesthatdidnotimplementbans.Weusethevariationin

implementationdatesacrossthecountryandfixedeffectsmodelstocontrolfor

unobservablefactorsincludingimprovingpreventionandtreatmentofcardiovascular

disease,decreasingsmokingrates,andsmokingrestrictionsenactedbyprivate

8/14/2019 SSRN-id1359506

4/33

5

businesses.Wefindthatsmokingrestrictionsareunlikelytosubstantiallyaffectshort

termmortalityandhospitalizationratesinboththeelderly,workingage,andchild

populations.Wefindsomeevidencethatsmokingbanscouldreducemortalityinthe

elderlybuttheresultsarenotstatisticallysignificant(1.4%,95%confidenceinterval:

3.0to0.2%).

Allpreviouspublishedstudiesonthehealtheffectsofsmokingbanssharea

commonmethodology:theycomparetheoutcomesinasinglecommunitythathas

passedasmokingbanwithoutcomesinasmallsetofnearbycommunitiesthathave

notpassedbans. Amajorcontributionofthispaperisthatwesimulatetheresultsfrom

allpossiblesmallscalestudiesusingsubsamplesfromthenationaldata. Wefindthat

largeshorttermincreasesinAMIincidencefollowingasmokingbanareascommonas

thelargedecreasesreportedinthepublishedliterature.

2Background

Inthissection,wediscusshowenvironmentaltobaccosmoke(alsoknownas

secondhandsmoke)isrelatedtohealthoutcomes,thehistoryandeffectsofsmoking

bansintheU.S.andinternationally,andtheimplicationofpreviousstudiesforU.S.

smokingpolicy.

8/14/2019 SSRN-id1359506

5/33

6

2.1Environmentaltobaccosmokeandhealthoutcomes

Inarecentreview,theU.S.SurgeonGeneralreportsthatnumerous

epidemiologicandlaboratorystudieshavelinkedenvironmentaltobaccosmoke(ETS)

exposuretoincreasedratesofcardiovasculardisease,respiratoryillnessandlung

cancer(GlantzandParmley1991;He,Vupputurietal.1999;BarnoyaandGlantz2005;

U.S.DepartmentofHealthandHumanServices.2006).Laboratorystudiessupportthe

notionthatsmallquantitiesofinhaledcigarettesmokecaninducesimilarbiochemical

responsesinnonsmokersasinchronicsmokers.Sucheffectscouldpredisposenon

smokerstogreatlyelevatedriskofAMIandstroke.Epidemiologicstudiestypically

comparedoutcomesnonsmokingspousesofsmokersandnonsmokers.Although

somemetaanalysesdisputedthesefindings,mostagreedthatchronicETSexposure

increasedriskofAMIby20to30%.Althoughnoamountofsecondhandsmokeislikely

tobebeneficial,theabovestudiesdonotaddressthepotentialeffectsofintermittent

exposuretosecondhandsmoke,asmightbecausedbyexposuretocigarettesmokein

publicplaces.

SecondhandsmokeinpublicplaceshasbeenmoststronglylinkedtoAMI

amongallpotentialadversehealthoutcomes(OngandGlantz2004;Sargent,Shepardet

al.2004;Bartecchi,Alseveretal.2006).Thereissomebiologicaljustificationforthisin

themedicalliterature(U.S.DepartmentofHealthandHumanServices.2006).Thefull

8/14/2019 SSRN-id1359506

6/33

7

effectsofeliminatingtobaccosmokemaytakeyearstooccurbecausesomeaspectsof

coronaryarterydisease(suchasnarrowingofthecoronaryarteries)developslowly

overtime.However,aheartattackoccursduetosuddenclotformationindiseased

arteries;inlaboratorysettings,exposuretoevensmallquantitiesoftobaccosmokecan

inducebiochemicalstatesthatpredisposetoheartattacks.Therefore,asmokingban

couldplausiblyreduceAMIincidenceandmortalityasearlyasthefirstyearafteraban

ifiteliminatesevenrelativelyminorexposure.Asaresultmanypriorstudiesexamined

AMIratesinsingleregionsinthe6to18monthsfollowingasmokingban.Inaddition

toincreasedriskofAMI,thoseexposedtosecondhandtobaccosmokemaysuffer

higherratesofasthma,chronicobstructivepulmonarydisease,infections,cancerand

otherdiseases.

2.2PublicbansonsmokinginU.S.publicplaces

AsevidenceonETSaccumulated,manyU.S.employersbeganrestricting

smokingintheworkplace;theproportionofcoveredworkersincreasedfrom25%to

70%between1986and1993(Farkas,Gilpinetal.1999;Farrelly,Evansetal.1999;

CentersforDiseaseControlandPrevention2000).Followingtheseprivaterestrictions,

severalcommunitiesinCaliforniabannedsmokinginworkplaces,restaurants,andbars

intheearly1990s.Manyotherstatesandmunicipalitiesfollowed(American

Nonsmokers RightsFoundation2007).Inadditiontotheselocalpolicies,several

8/14/2019 SSRN-id1359506

7/33

8

prominentpoliticiansadvocatedanationalpolicybanningsmokinginpublicplaces

(OnTheIssues.org2007).

Althoughthesebanshaveprovenpopular,thescientificliteraturetodatehas

onlyexaminedtheimpositionofasmokingbaninafewspecificU.S.regionsand

severalEuropeancountries.AMIratesdecreasedapproximately40%inHelena,

Montanaand27%inPueblo,Colorado(relativetosurroundingcommunities)following

theimpositionofbroadrestrictions(Sargent,Shepardetal.2004;Bartecchi,Alseveret

al.2006).AlargerstudycomparedratesofAMIandacutestrokeadmissionsinNew

YorkStatebeforeandaftercomprehensivesmokingbans,whichwerelargely

implementedinMarchandJuly2003(Juster,Loomisetal.2007).Thisstudyestimated

thatthelawsreducedAMIadmissionsby8%,althoughitdidnotcomparethese

changesinAMIandacutestrokeadmissionstochangesthatmayhaveoccurredin

nearbystatesthatdidnotimplementsmokingbansoverthisperiod.

Examiningadifferentmechanism,AdamsandCotti(2008)findincreasedratesof

vehiculardeathsfollowingtheenactmentofsmokingbans. Theyattributethisincrease

tosmokersdrivingoutoftheirnativeareatofindaplacetosmokeinpublic. Another

possibilityisthatthesebansleadsmokerstosmokemoreinvehicles,whichcouldbea

distractionwhiledriving.AdamsandCottiusenationaldataandhencemeasuresthe

averageeffectofabanacrossallcommunitiesintheU.S.;unlikethepreviouslycited

studies,theydonotmeasuretheeffectinindividualcommunities.

8/14/2019 SSRN-id1359506

8/33

9

2.3Internationalsmokingbans

TworelativelylargestudiesoftheeffectofsmokingbansonAMIincidencein

RomeandthePiedmontregionofItalyconcludedthatsmokingbansreducedAMI

incidenceby7to11%inyoungerpopulations(BaroneAdesi,Vizzinietal.2006;

Cesaroni,Forastiereetal.2008).Theauthorsdemonstratedthattheprevalenceof

smoking,cigaretteconsumptionpersmoker,andnonsmokerexposuredroppedin

Italyafterthenationalban(BaroneAdesi,Vizzinietal.2006;Gallus,Zuccaroetal.2007;

Cesaroni,Forastiereetal.2008).1

AstudyoftheScottishpublicsmokingbanfoundlargereductionsinAMIrates

aswell(Pell,Hawetal.2008).TheScottishgovernmentimplementedacomprehensive

baninMarch2006.Theauthorsmeasuredadmissionsto9hospitals(whichtreatover3

millionpeople)aswellasdeathrecordsforthegeneralpopulation.Theymeasured

actualexposuretocigarettesmokeamongnonsmokers(formersmokersandnever

smokers)andcurrentsmokers.Inpriorstudiesofthegeneralpopulation,serum

cotininelevelsdeclinedfrom0.43to0.25nanograms/dLinnonsmokersandfrom167

to103nanograms/dLincurrentsmokers.Theauthorsthennotedstatisticallysignificant

declinesinadmissionsforacutecoronarysyndromesinallgroups 14%insmokers,

19%informersmokersand21%inthosewhoneversmoked.ThedeclineintheAMI

1 The authors used cotinine levels, obtained from blood or salivary samples, to measure cigarette smoke exposure in

non-smokers reliably.

8/14/2019 SSRN-id1359506

9/33

10

ratewasnoticeablylargerthanthe4%declineinneighboringEnglandduringthesame

timeperiod(whichlackedacomprehensiveban).

2.5ImplicationsofpriorworkforU.S.smokingpolicy

Theaforementionedstudieslinkingsmokingbanstoimpressivepublichealth

gainssuggestthatwidespreadpublicsmokingbanswoulddemonstrablyimproveU.S.

publichealth.However,theinternationalexperiencealsomaynottranslatebecause

nonsmokersexposuretosecondhandsmokeandsmokingprevalenceinItalyandin

Scotlandweremuchhigherandprivatesmokingrestrictionswereweakerthaninthe

U.S.

RestrictingtheanalysistoU.S.studiesdoesnoteliminatequestionsabout

generalizability.PriorU.S.studiesweresmallinscale,havingexaminedonlyafew

regions;itispossiblethatthoseregionsarenotrepresentativeoftypicalU.S.

communities.Althoughdifferenceindifferenceanalysescancontrolforunobserved

factors,asimplepairwisecomparisonusinganatypicalpairofcommunitieswillyield

resultsthatmaynotberepresentative.Bycontrast,asimulationstudyfoundthat

extendingsmokingrestrictionsfrom70%to100%ofU.S.workplaceswouldprevent

roughly1,500myocardialinfarctionsinthefirstyear(OngandGlantz2004).Although

thismaybeaclinicallyrelevantimprovement,itrepresentsamuchsmallerreduction

8/14/2019 SSRN-id1359506

10/33

11

(

8/14/2019 SSRN-id1359506

11/33

12

similarityimpliesexposuretosecondhandsmokepresentslargeheathrisksatlow

levelsandnoadditionalhealthrisksathigherlevels,whichseemsunlikely.2

WeaddresstheseissuesbyanalyzingtheimpactofU.S.publicsmoking

restrictionsonhealthoutcomesinalarge,heterogeneousgroupofU.S.communities.By

analyzingmuchlargerpopulationsinadiversegroupofU.S.hospitalsandcounties

andbyaccountingforunderlyingseculartrendsandregionspecificcharacteristics,we

mitigatethepossibilityofselectionbias.

3.Data

Toconstructthedatasetsusedforouranalysis,wemergedataonthetimingand

locationofsmokingbanstothreelargenationwidedatasourcesonhealthoutcomes.

3.1Dataonsmokingbans

WeuseordinancedatafromtheAmericanNonsmokersRightsFoundationto

identifystates,counties,andmunicipalitiesthatimplementedrestrictionsonsmoking

between1990and2004.WeadapttheclassificationschemefromtheAmerican

NonsmokersRightsFoundationtoidentifythosebansthatrestrictsmokinginall

workplacesexceptbarsandrestaurantsasworkplacebans.Althoughnotincludedin

2Similar risk reductions were found in European and U.S. studies though baseline smokingprevalence, secondhand smoke exposure, and efficacy varied, which argues against a plateauin risk.

8/14/2019 SSRN-id1359506

12/33

13

theformerlist,Californiaadoptedanearlycompletebanonworkplacesmokingin

1995.WeclassifiedCaliforniaassmokingrestrictedstartingin1995.Wealsocreatea

datasetofbansofanysite workplaces,barsorrestaurants.Weclassifyeach3digitzip

code,city,andcountyintheU.S.bysmokingbanstatusanddateofimplementation.

3.2Datasourcesforhealthoutcomes

WeanalyzehealthoutcomesusingtheMultipleCauseofDeath(MCD)database

(19892004),Medicareclaims(19972004),andtheNationwideInpatientSurvey(NIS),

collected19932004bytheHealthcareCostandUtilizationProject,whichissponsored

bytheAgencyforHealthcareResearchandQuality(AgencyforHealthcareResearch

andQuality2006).TheMCDdatabaseidentifiestheunderlyingcauseforeachdeathin

theU.S.OursourceforMedicareclaimsaretheMedicareProviderAnalysisand

Review(MEDPAR)files,whichincludeallfeeforserviceMedicarebeneficiariesinthe

U.S.TheNISisanationallyrepresentative20%sampleofalldischargesfromU.S.

communityhospitals(whichincludesallnonfederalacutecarehospitals).Excluded

hospitalsincludeVeteransAffairshospitalsandlongtermrehabilitationhospitals.

Weidentifiedmortalityandhospitalizations duetoAMIandallcausedeaths

andhospitalizationsbecausebothmightplausiblyimproveintheshortrun.In

addition,broaddiseasemeasureslikeAMIandallcauseeventsarelesslikelymiscoded

inadministrativedata,reducingmeasurementerror(Petersen,Wrightetal.1999;

8/14/2019 SSRN-id1359506

13/33

14

TirschwellandLongstreth2002).Wealsoconsidercasesofasthmaandchronic

obstructivepulmonarydisease;whilethesearechronicdiseases,secondhandsmoke

possiblytriggersacuteexacerbations(U.S.DepartmentofHealthandHumanServices.

2006).Wealsoidentifyhipfracturehospitalizationsbecausethesecanactasanegative

controlbecausetheirincidencewouldbeunlikelytochangequicklyafterasmoking

ban(Hoidrup,Prescottetal.2000). Wedontexpecttheincidenceofhipfracturetobe

affectedbysmokingbans. Ifweweretofindanassociationbetweensmokingbansand

theincidenceofhipfracture,wewouldquestionthevalidityoftheempiricalstrategy.

Weassembleeachdatasetinasimilarfashion:wefirstclassifydeathsand

hospitaladmissionsaccordingtotheirprimarydiagnoses(AMI,asthma,etc.);wethen

sumalloutcomesineachregion(hospitalcatchmentarea,county,orzipcode);finally,

wemergeinformationonsmokingordinancesforthatregion.

Workplacesmokingbanscouldhavedifferentialeffectsbyage.Theelderlyand

childrenmaybemorevulnerabletothediseasesexacerbatedbyETS,andcouldstandto

gainmorebenefitthanatypicalworkingadult.Ontheotherhand,childrenandthe

elderlyareprimarilyexposedtoworkplaceETSascustomers,whichwouldreduce

theirbenefitfromasmokingban.Toaccountforthesedifferences,wefurtherstratify

outcomesintothreeagegroupsperregion:children(017years),workingageadults

(1864years),andtheelderly(65+years).

8/14/2019 SSRN-id1359506

14/33

15

Fromeachdataset,weexcludedeathsandhospitalizations whereweareunable

todeterminewhetherthepersonlivedinanareawhereasmokingrestrictionwas

implemented(3540%oftheNISandMCDdata).Inaddition,weexcludethefollowing

fromouranalysisofNIShospitals:transferpatients,hospitalsincludedinasingle

surveyyear,hospitalsthatmergedduring19932004,hospitalsdevotedtoacute

rehabilitation(becausefewAMIsareadmitted)andsmallhospitals(

8/14/2019 SSRN-id1359506

15/33

16

example,manyregionsexperiencedunobservedincreasesinprivatesmoking

restrictions,reductionsinsmokingprevalence,orimprovedmedicaltreatmentthat

couldhavecausedchangesinoutcomes.Tomitigatethesepotentialconfounding

factors,wecomparetrendsinregionswheresmokingbanswereimplementedtothose

incontrolregionswheresmokingrestrictionswerenotimposed.Inparticular,we

estimateregionfixedeffectsmodels. Foreachoutcome(e.g,AMIdeathor

hospitalizations), weusethefollowingregressionmodel:

(1) ititstiit SmokingBanOutcome +++=

Here itOutcome representsthenumberofdeathsorhospitaladmissionsinregion

i(1N)andtimet(1T), itOutcome isanindicatorforeachyear,and it istheerror

term.Weinclude i ,anindicatorforeachregion(county,3digitzipcode,orhospital),

tocontrolforidiosyncraticdifferencesbetweenregions. s isthecoefficientofinterest,

representingthebreakinthetimetrendinducedbyasmokingban,aftercontrollingfor

seculartrends.Inpresentingfinalresults,wepresentthemeanpercentagechangein

outcomes %)100(

swhere is = =

N

i

T

t itOutcome

TN 1 11

.Weuseblockbootstrap

clusteredattheregionalleveltocalculateallstandarderrors.Ofnote,theeffectsof

privaterestrictionsenactedpriortoagovernmentbanareincludedinprebantrends

andareexcludedfromthefinalestimate.Asaresult,thisstrategyplausiblyidentifies

theshorttermeffectsofgovernmentrestrictionsalone.Duetodatalimitations,wealter

8/14/2019 SSRN-id1359506

16/33

17

ourapproachinanalyzinghospitaladmissionsinNIShospitals;wecomparechangesin

admissionsinthefirstyearfollowingasmokingrestrictiontochangesincontrol

hospitalswithoutsmokingrestrictions.3

4.2Sensitivityanalyses.

Inourmainregressionmodel(1)wedonotincludecovariatesotherthantime

andregionspecificindicators.However,itispossiblethatfactorsmightchangeover

timedifferentiallybetweenregions.Ifso,thefixedeffectsmodelwouldyieldbiased

estimates.Wethereforetestmodelspecificationsofthefollowingform:

(2) ititstiit SmokingBanOutcome ++++= X

Thetermsandresultsarethesameasin(1)exceptfortheadditionofX,avectorof

countylevelcharacteristicstakenfromthe2005AreaResourceFile.Thevariables

includepopulationsize,numberofphysiciansandhospitalbedspercounty,household

income,andpercentofpopulationinlaborforce.Thesedatawerelinkedbycountyand

year,whereavailable,tocountiesfromtheMCDaswellashospitalsintheNISdata.

Wedonotincludethesevariables(whichcouldpotentiallyaddexplanatorypower)in

3In our analysis of the NIS data, our primary unit of analysis is the hospitalscatchment area, which includes the hospitals home city but whose full extent is unobservable.We use the number of admissions to a hospital in a particular month as a measure of theadmission rate within its catchment area. Catchment areas tend to remain stable over time, sothe approximation error is likely to be small. It is possible that we misclassified someadmissions. However, many bans were enforced in the county or state of origin, which wouldinclude the entire catchment area. In addition, we only considered patients with seriousillnesses who tend to be taken to the nearest hospital, which further reduces bias.

8/14/2019 SSRN-id1359506

17/33

18

allmodelspecificationsbecausedataforX isoftenmissingforseveralyears,which

substantiallylimitstheirsamplesizes.Weperformsensitivityanalysesbycomparing

theresults

using

model

specification

(2)

to

the

results

using

model

specification

(1)

for

thesamesetofincludedregions.

4.3Simulatingsmallsampleresults

Wecompleteouranalysisbyusingsubsamplesofthenationaldatatosimulatea

completesetofpairwisecomparisonstudiesofthesortavailableinthepublished

literature,includingBartecchietal.(2005)andSargentetal.(2004).Thesestudies

comparedatreatmentunitwhereabanwaspassedagainstacontrolunitwithnoban

onthebasisofthechangeinanoutcomevariable(heartattackadmissionrates,for

instance)inashortperiod(618months)afterthebanwaspassedinthetreatment

region.Wesimulatetherangeofsucheffectsbyfirstcalculatingthepercentchangein

admissionsineachhospitallocatedinaregionwithworkplacesmokingrestrictions

betweentheyearbeforeandtheyearfollowingaworkplacesmokingban;wealso

calculatethesamestatisticsforallcontrolhospitalsfromthesametimeperiod.Wethen

subtractthechangeinoutcomesineachcontemporaneouscontrolhospitalfromeach

interventionhospital.Theresultingdatasetconsistsoftheuniverseofpossible(19,406)

pairwisecomparisonsofonesmokingrestrictedhospitalwithonecontrolhospital.We

conductasimilaranalysisofheartattackmortality,exceptinthiscase,weuseMCD

8/14/2019 SSRN-id1359506

18/33

19

mortalitydataandourunitofanalysisisthecounty.Thefinaldatasetconsistsof23,938

pairwisecomparisonsandrepresentstheuniverseofsuchcomparisonsintheMCD

data.

5.Results

5.1Smokingrestrictionsovertime

ThepercentageofU.S.regionsthatimposedsmokingrestrictionsincreased

dramaticallybetween1988and2004;theprevalenceofsmokingrestrictionsrose

sharplyin1995and20032004andmoregraduallyinotheryears(seeFigure1).

Notably,thisnationaltrendfollowstheactionsofprivateemployers.By1993,most

workplacesisolatedcigarettesmoke,whichwoulddiminishtheimpactoflegislative

bans(Farrelly,

Evans

et

al.

1999;

Centers

for

Disease

Control

and

Prevention

2000).

5.2Mainestimates

Workplacesmokingrestrictionsareunrelatedtochangesinallcausemortalityor

mortalityduetootherAMIinallagegroups.Restrictionsonsmokingofanysortare

associatedwithreducedallcausemortalityamongtheelderly(1.4%,95%CI: 3.0to

0.2%)buttheresultisonlysignificantatthe10%level(p=0.06)(seeTable2).

WefindnostatisticallysignificantreductioninadmissionsduetoAMIamong

workingageadults(4.2%,95%CI: 10.2to1.7%,p=0.165)oramongtheelderly(2.0%,

8/14/2019 SSRN-id1359506

19/33

20

95%CI: 3.7to7.7%,p=0.48)followingtheenactmentofaworkplacesmoking

restriction(seeTable3).Wesimilarlyfindnoevidenceofreductioninadmissionsfor

otherdiseasesinanyagegroup,thoughsmokingrestrictionsofallsortsareassociated

withstatisticallyinsignificantincreasesinasthma(11.4%, 95%CI: 2.4to25.3%,p=0.11)

andtotaladmissions(3.7,95%CI: 2.1to9.5%,p=0.21)amongchildren.Amongboth

theelderlyandworkingageadults,wefindnostatisticallysignificanteffectofsmoking

bansonhipfractureadmissions,ournegativecontrol.Thissupportsthehypothesisthat

unobservedcharacteristicsofregionsdonotconfoundourresults.

5.3Sensitivityanalyses

Solelyusingregionspecificindicatorstocapturetrendspotentiallyomits

importantconfoundingvariablesbutcomprehensivedataisnotavailableforallregions

andyears.Wetestforbiasbyaddingvariablesfordemographicsandmedicalresource

availabilitytotheregressionmodel.(Weobtainthesedatafromthe2005AreaResource

File,whichcontainsinformationforallcountiesandsomeyears.)Weexaminethe

importanceoftheseomittedvariablesbycomparingresultswithandwithout

additionalvariablesforthesametimeperiods.

Table4showstheeffectofanysmokingbanontotaldeathratesinallagegroups

usingdifferentsetsofcontrolvariables.Theestimatesdonotchangedramaticallywith

inclusionofdifferentvariables,butdochangeasthestudysamplechanges. Thesample

8/14/2019 SSRN-id1359506

20/33

21

isreducedfrom24,884countyquartersincolumn(1)to7,308countyquartersin

column(4).Regressionestimatesusingthesamesamplesbutdifferentregression

modelsaresimilar. Forexample,basedonalimitedsampleof15,512countyquarters

across9yearsincolumn(3),theestimatedassociationbetweenanysmokingbanand

totaldeathrateswas+0.330%(withastandarderrorof0.794%)withoutadditional

controlsand+0.328%(withastandarderrorof0.804%)withcontrolsforhospitalbeds

percapitaandphysicianspercapita.WeshowsimilarresultsinTables5and6.These

resultssuggestthatordinarydemographicchangesareaccountedforusingthe

differenceindifferenceidentificationstrategy.

Inalloftheresultsreportedinthetables,wecalculatestandarderrorsthatallow

forclusteringatthearealevel(county,hospitalarea,orzipcode). Ifinaddition,we

weretocorrectformultiplecomparisonsusingHochbergsmethod(Hochberg1988),

ourstandarderrorsincreasefurther.

5.5Pairwisecomparisonssimulatingsmallsampleresults

Figures2and3plotallpossiblepairwisecomparisonsofchangesinAMI

incidenceafteraworkplacesmokingbantochangesinrandomlyselectedcontrol

regions.Figure2showsthatthemeanmeasuredeffectofworkplacesmokingbanson

heartattackadmissionsisclosetozero,but10%orgreaterdeclinesand10%ofgreater

increasesinAMIadmissionsarecommon.Figure3showssimilarresultsfora

8/14/2019 SSRN-id1359506

21/33

22

comparisonofAMImortalityinsmokingrestrictedcountiesfromtheyearaftera

workplacebanwithratesincountieswithoutaban. Theresultsofthissimulation

analysisshowsthatresultsfrompriorsmallsamplestudies,whichfoundverylarge

decreasedinAMIadmissionsandmortalityfollowingtheenactmentofsmokingbans,

arefeasible. However,resultswiththeoppositesignandofsimilarmagnitudearealso

feasibleandshouldbeequallycommon.

6.Conclusions

WefindnoevidencethatlegislatedU.S.smokingbanswereassociatedwith

shorttermreductionsinhospitaladmissionsforacutemyocardialinfarctionorother

diseasesintheelderly,childrenorworkingageadults.Wefindsomeevidencethat

smokingbansareassociatedwithareducedallcausemortalityrateamongtheelderly

(1.4%)butonlyatthe10%significancelevel.

Wealsoshowthatthereiswideyeartoyearvariationinmyocardialinfarction

deathandadmissionrateseveninlargeregionssuchascountiesandhospital

catchmentareas.Comparisonsofsmallsamples(whichrepresentsubsamplesofour

dataandaresimilartothesamplesusedinthepreviouspublishedliterature)might

haveledtoatypicalfindings.Itisalsopossiblethatcomparisonsshowingincreasesin

cardiovasculareventsafterasmokingbanwerenotsubmittedforpublicationbecause

theresultswereconsideredimplausible.Hence,thetruedistributionfromsingle

8/14/2019 SSRN-id1359506

22/33

23

regionswouldincludebothincreasesanddecreasesineventsandameanclosetozero,

whilethepublishedrecordwouldshowonlydecreasesinevents.Thus,publicationbias

couldplausiblyexplainwhydramaticshorttermpublichealthimprovementswere

seeninpriorstudiesofsmokingbans.

Ourstudyfocusesonlyonthehealtheffectsofsmokingbans. Futureresearch

shouldestimatenonhealthrelatedbenefitsofthesebanstononsmokers.Priortoa

smokingban,nonsmokersatriskforrespiratorysymptomsorcardiovascularevents

mighthaveavoidedbusinesseswithhighETSlevels.Afteraban,nonsmokerscould

gaincomfortableaccesstothesebusinesses,butbasedonourfindingsinthisstudy,this

benefitwouldnotalsoresultinreducedhospitalizationordeathrates.Ourstudy

designplausiblyidentifiesonlyshorttermbenefitsofsmokingbans(ashasthestudy

designsusedbypreviousstudies).Wecannotanalyzewhethersmokingbansimprove

longtermtrendsforchroniccardiovasculardiseaseorlungcancer.Inaddition,

smokingbansmayinducesmokerstoquitordiscouragenonsmokersfromstarting

smoking.Thesepotentiallongtermbenefitswillnotbeapparentinstudyofshortterm

outcomesandwouldbenefitfromfurtherstudy.

8/14/2019 SSRN-id1359506

23/33

24

REFERENCE

Adams,S.andC.Cotti.(2008).DrunkDrivingAfterthePassageofSmokingBansin

Bars,JournalofPublicEconomics92(56):12881305.Adda, J. and F. Cornaglia. (2006, 18 December 2007). "The Effect of Bans and Taxes on Passive

Smoking." CeMMAP working paper series, from

http://www.ucl.ac.uk/~uctpjea/passive.pdf.

Agency for Healthcare Research and Quality. (2006). "INTRODUCTION TO THE HCUPNATIONWIDE INPATIENT SAMPLE (NIS), 2004." Retrieved 18 December 2007,

from http://www.hcup-us.ahrq.gov/db/nation/nis/NIS_Introduction_2004.pdf.

American Nonsmokers' Rights Foundation. (2007). "Percent of U.S. State/Commonwealth

Populations Covered by 100% Smokefree Air Laws." from http://www.no-smoke.org/pdf/EffectivePopulationList.pdf.

Barnoya, J. and S. A. Glantz (2005). "Cardiovascular effects of secondhand smoke: nearly as

large as smoking." Circulation 111(20): 2684-98.Barone-Adesi, F., L. Vizzini, et al. (2006). "Short-term effects of Italian smoking regulation on

rates of hospital admission for acute myocardial infarction." Eur Heart J 27(20): 2468-72.Bartecchi, C., R. N. Alsever, et al. (2006). "Reduction in the incidence of acute myocardial

infarction associated with a citywide smoking ordinance." Circulation 114(14): 1490-6.

Centers for Disease Control and Prevention (2000). "State-specific prevalence of current

cigarette smoking among adults and the proportion of adults who work in a smoke-freeenvironment--United States, 1999." MMWR Morb Mortal Wkly Rep 49(43): 978-82.

Centers for Disease Control and Prevention (2007). "Reduced secondhand smoke exposure after

implementation of a comprehensive statewide smoking ban--New York, June 26, 2003-

June 30, 2004." MMWR Morb Mortal Wkly Rep 56(28): 705-8.Cesaroni, G., F. Forastiere, et al. (2008). "Effect of the Italian smoking ban on population rates

of acute coronary events." Circulation 117(9): 1183-8.

Evans, W., M. Farrelly, et al. (1999). "Do Workplace Smoking bans Reduce Smoking?"American Economic Review 89(4): 728-747.

Farkas, A. J., E. A. Gilpin, et al. (1999). "The effects of household and workplace smoking

restrictions on quitting behaviours." Tob Control 8(3): 261-5.Farrelly, M. C., W. N. Evans, et al. (1999). "The impact of workplace smoking bans: results from

a national survey." Tob Control 8(3): 272-7.

Gallus, S., P. Zuccaro, et al. (2007). "Smoking in Italy 2005-2006: effects of a comprehensive

National Tobacco Regulation." Prev Med 45(2-3): 198-201.Glantz, S. A. and W. W. Parmley (1991). "Passive smoking and heart disease. Epidemiology,

physiology, and biochemistry." Circulation 83(1): 1-12.

He, J., S. Vupputuri, et al. (1999). "Passive smoking and the risk of coronary heart disease--a

meta-analysis of epidemiologic studies." N Engl J Med 340(12): 920-6.Hochberg, Y. (1988). "A sharper Bonferroni procedure for multiple tests of significance."

Biometrika 75(4): 800-2.Hoidrup, S., E. Prescott, et al. (2000). "Tobacco smoking and risk of hip fracture in men and

women." Int J Epidemiol 29(2): 253-9.

Juster, H. R., B. R. Loomis, et al. (2007). "Declines in hospital admissions for acute myocardial

infarction in New York state after implementation of a comprehensive smoking ban." AmJ Public Health 97(11): 2035-9.

8/14/2019 SSRN-id1359506

24/33

25

Longo, D. R., R. C. Brownson, et al. (1996). "Hospital smoking bans and employee smokingbehavior: Results of a national survey." Jama 275(16): 1252-7.

Metzger, K. B., F. Mostashari, et al. (2005). "Use of pharmacy data to evaluate smoking

regulations' impact on sales of nicotine replacement therapies in New York City." Am JPublic Health 95(6): 1050-5.

Ong, M. K. and S. A. Glantz (2004). "Cardiovascular health and economic effects of smoke-freeworkplaces." Am J Med 117(1): 32-8.OnTheIssues.org. (2007). "2007 Democratic primary debate at Dartmouth College." Retrieved

18 December 2007, from

http://www.ontheissues.org/Archive/2007_Dem_primary_Dartmouth_Drugs.htm .

Pell, J. P., S. Haw, et al. (2008). "Smoke-free legislation and hospitalizations for acute coronarysyndrome." N Engl J Med 359(5): 482-91.

Petersen, L. A., S. Wright, et al. (1999). "Positive predictive value of the diagnosis of acute

myocardial infarction in an administrative database." J Gen Intern Med14(9): 555-8.Sargent, R. P., R. M. Shepard, et al. (2004). "Reduced incidence of admissions for myocardial

infarction associated with public smoking ban: before and after study." Bmj 328(7446):

977-80.Thun, M., J. Henley, et al. (1999). "Epidemiologic studies of fatal and nonfatal cardiovascular

disease and ETS exposure from spousal smoking." Environ Health Perspect 107 Suppl 6:

841-6.Tirschwell, D. L. and W. T. Longstreth, Jr. (2002). "Validating administrative data in stroke

research." Stroke 33(10): 2465-70.

U.S. Department of Health and Human Services. (2006). The Health Consequences of

Involuntary Exposure to Tobacco Smoke: A Report of the Surgeon General. USDHS.Atlanta, GA.

8/14/2019 SSRN-id1359506

25/33

26

Figure1

0

10

20

30

40

Percentageofregionswithbans

1988 1992 1996 2000 2004Year

Any smoking restriction

Workplace restriction

Regions are 3 digit US postal (zip) codesSource: Authors' analysis of data from American Non-smoker's Rights Foundation and California Department of Industrial Relations

Figure 1. Prevalence of smoking bans over time

Regionsare3digitU.S.postal(zip)codes

Source:AuthorsanalysisofdatafromAmericanNonsmokersRightsFoundationand

CaliforniaDepartment

of

Industrial

Relations

8/14/2019 SSRN-id1359506

26/33

27

Figure2.

0

2

4

6

8

10

Percent

-200 -150 -100 -50 0 50 100 150 200Relative difference in acute myocardial infarction incidence (%)

Percent change

Normal density curve

Mean change

Source: HCUP. Hospitals with smoking restrictions: 77. Control hospitals: 631 Pairwise comparisons: 19406

Figure 2. Changes in AMI admissions relative to all control hospitals

Source:AuthorsanalysisofdatafromNationwideInpatientSample.Hospitalswith

smokingrestrictions:52.Controlhospitals:336.Pairwisecomparisons:8,529.Relative

differencemaybelessthan 100%ifalargepercentageincreaseinacontrolhospitalis

subtractedfromalargedecreaseinasmokingrestrictedhospital.

8/14/2019 SSRN-id1359506

27/33

28

Figure3.

0

2

4

6

8

1

0

Percent

-75 -50 -25 0 25 50 75Relative difference in acute myocardial infarction incidence (%)

Percent change

Normal density curve

Mean change

Source: MCD. Counties with smoking restrictions: 89. Control counties: 462 Pairwise comparisons: 23938

Figure 3. Changes in AMI deaths relative to all control counties

Source:AuthorsanalysisofdatafromMultiplecauseofdeathdatabase.Countieswith

smokingrestrictions:89.Controlcounties:462.Pairwisecomparisons:23,905.Relative

differencemaybelessthan 100%ifalargepercentageincreaseinacontrolcountyis

subtractedfromalargedecreaseinasmokingrestrictedcounty.

8/14/2019 SSRN-id1359506

28/33

29

Table1.Datasourcecharacteristics

Datasource

NationwideInpatientSample:

Hospitals 874

States

27

Years 19932004

Alladmissions 21,820,484

Admissionsbydisease

Acutemyocardialinfarction 217023

CombinedasthmaandCOPD 433674

MultipleCauseofDeath:

Counties 468

States 50

Years 19892004

Alldeaths 24,610,532

Acutemyocardialinfarction 2,042,812

Medicarepatients:

Regions(3digitzipcodes) 868

States 51

Years 19972004

IncludedMedicarepopulation(personyears) 275,303,008

Alldeaths 13,106,175

Alladmissions

Admissionsby

disease

72,542,544

Acutemyocardialinfarction 2,382,386

CombinedasthmaandCOPD 2,984,382

Hipfracture 3,381,690

Chronicobstructivepulmonarydisease

Source:AuthorsanalysisofdatafromNationwideInpatientSample,MultipleCauseof

Deathfiles,and100%MedicareProviderAnalysisandReviewfiles.


Source:AuthorsanalysisofdatafromNationwideInpatientSample,nationaldeath

statistics,and100%MedicareProviderAnalysisandReviewfiles.

8/14/2019 SSRN-id1359506

29/33

30

Table2.Mortalityandsmokingrestrictions*

Disease %changeinmortality(95%CI) Pvalue

Workplacesmokingrestrictions:

Alldeaths

(0

17

years

old)

4(10.2

to

2.2)

0.204

AMI (1864yearsold) 4.4(11.6to2.7) 0.224

Alldeaths(1864yearsold) 1.1(2.9to0.8) 0.247

Alldeaths(age65+) 0.8(1.2to2.8) 0.413

AMI(allages) 1.5(4.8to1.8) 0.374

Alldeaths(allages) 0.3(1.6to0.9) 0.624

Anysmokingrestrictions:

Alldeaths(017yearsold) 2(6.5to2.6) 0.400

AMI(1864yearsold) 3.5(10to3.1) 0.299

Alldeaths(1864yearsold) 0.4(2.1to1.2) 0.601

Alldeaths(age65+) 1.4(3.0to0.2) 0.062

AMI(allages) 1.1(4.1to1.9) 0.470

Alldeaths(allages) 0.5(0.6to1.6) 0.396

*AuthorsanalysisofdatafromMultipleCauseofDeathfiles,years19932004exceptin

age65+,whicharefrom100%MedicareProviderAnalysisandReviewfiles,19972004.

AMIindicatesdeathsfromacutemyocardialinfarction.

8/14/2019 SSRN-id1359506

30/33

31

Table3.Hospitaladmissionsandsmokingrestrictions

Disease %changeinadmissions

(95%CI)

Pvalue

Workplacesmokingrestrictions:

Alladmissions

(age

017)

3.7

(2.1

to

9.5)

0.211

Asthma(age017) 11.4(2.4to25.3) 0.106

AMI(age1864) 4.2(10.2to1.7) 0.165

Alladmissions(age1864) 2.2(0.4to4.8) 0.101

Asthma(age1864) 3.4(6.2to13.1) 0.485

COPD(age1864) 1.7(8.6to12) 0.746

Hipfracture(age1864) 5.1(15.6to5.4) 0.340

AMI(age65+) 2(3.7to7.7) 0.477

Alladmissions(age65+) 1.8(3.1to6.7) 0.477

Asthma(age65+) 0.8(13.5to15.1) 0.909

COPD(age65+) 3.4(3.6to10.5) 0.343

Hipfracture(age65+) 0.1(2.4to2.6) 0.946

Anysmokingrestrictions:


Asthma(age017) 13.7(2to29.3) 0.087

AMI(age1864) 4.7(10.3to01) 0.104


Asthma(age1864) 3.5(5to11.9) 0.422

COPD(age1864) 2.1(11.8to7.7) 0.676

Hipfracture

(age

18

64)

4.7

(14.6

to

5.1)

0.348

AMI(age65+) 5.1(0.4to10.6) 0.068

Alladmissions(age65+) 1.7(2.0to5.4) 0.364

Asthma(age65+) 6.8(4.78to18.4) 0.246

COPD(age65+) 2.7(2.4to7.8) 0.303

Hipfracture(age65+) 1.1(1.4to3.6) 0.385

Source:AuthorsanalysisofdatafromNationwideInpatientSample(199320004)except

age65+,whicharefrom100%MedicareProviderAnalysisandReviewfiles,years1997

2004.

Acutemyocardialinfarction/Ischemicheartdisease


8/14/2019 SSRN-id1359506

31/33

32

Table4.Comparisonofregressionmodels:

Allcausemortalityfollowingsmokingbanofanysort(MCDcounties)Variablesincluded (1) (2) (3) (4)

Anybans 0.495 0.324 0.328 1.335

(0.583) (0.573) (0.794) (0.547)

Hospitalbeds/person 0.0031

(0.0013)

Countypopulation 0.00005 0.00006 0.00005 0.00009

(0.00002) (0.00002) (0.00002) (0.00003)

Physicians/person 0.0003 0.0002 0.0007

(0.0007) (0.0008) (0.0010)

Percentpopulationinlaborforce 4.5

(6.6)

#Yearsinsample 15 13 9 5

#observations 24884 21580 15512 7308

Resultsfrom

original

model

0.495 0.322 0.330 1.343

(0.583) (0.573) (0.804) (0.546)

*Eachcoefficientrepresentsthe%changeinoutcomesduetoa1unitchangeinthe

explanatoryvariable. Standarderrorsareclusteredatthearealevelandarereported

inparentheses.

8/14/2019 SSRN-id1359506

32/33

33


Myocardialinfarctionadmissionratesinworkingageadults(NIShospitals)

followingworkplacebansVariablesincluded (1) (2) (3) (4)

Workplacebans 4.2 4.1 13.3 9.2

(3.1) (3.0) (5.8) (5.9)


(0.0029)


(0.00003) (0.00003) (0.00002) (0.00008)


(0.0025) (0.0028) (0.0077)


(324.9)


#observations

7476 7476 4932 1502

Resultsfromoriginalmodelon

thissample

4.2 4.2 13.3 9.2

(3.1) (3.1) (5.8) (5.7)



inparentheses.

8/14/2019 SSRN-id1359506

33/33


Allcauseadmissionratesinworkingageadults(NIShospitals)

Variablesincluded (1) (2) (3) (4)

Workplacebans 2.20 2.12 0.35 1.52

(1.34) (1.33) (1.73) (2.10)


(0.000998)


(0.00002) (0.00002) (0.00002) (0.00004)


(0.000993) (0.00109) (0.00299)


61.8)


#observations

7476 7476 4932 1502

Resultsfromoriginalmodel 2.20 2.20 0.26 1.40

(1.34) (1.34) (1.73) (2.08)



inparentheses.

Date post:	30-May-2018
Category:	Documents
Upload:	bill-hannegan
View:	218 times
Download:	0 times

SSRN-id1359506

Documents