Monthly variations in aneurysmal subarachnoid hemorrhage incidence and mortality

16 Pages·2017·10.13 MB·English

RESEARCHARTICLE
Monthly variations in aneurysmal
subarachnoid hemorrhage incidence and
mortality: Correlation with weather and
pollution
Myung-HoonHan1☯,JinheeKim2☯,Kyu-SunChoi3,ChoongHyunKim1,JaeMinKim1,Jin
HwanCheong1,Hyeong-JoongYi3,SeonHeuiLee4*
1 DepartmentofNeurosurgery,HanyangUniversityGuriHospital,Gyeongchun-ro,Guri,Gyonggi-do,Korea,
2 DepartmentofNursing,CollegeofMedicine,ChosunUniversity,Gwangju,Korea,3 Departmentof
a1111111111
Neurosurgery,HanyangUniversityMedicalCenter,Wangsimni-ro,Seongdong-gu,Seoul,Korea,
a1111111111
4 DepartmentofNursingScience,CollegeofNursing,GachonUniversity,Hambangmoe-ro,Yeonsu-gu,
a1111111111
Incheon,Korea
a1111111111
a1111111111 ☯Theseauthorscontributedequallytothiswork.
*[email protected]
Abstract
OPENACCESS
Citation:HanM-H,KimJ,ChoiK-S,KimCH,Kim
JM,CheongJH,etal.(2017)Monthlyvariationsin
Backgroundandpurpose
aneurysmalsubarachnoidhemorrhageincidence
andmortality:Correlationwithweatherand
Althoughtheeffectofweatherandairpollutionontheoccurrenceofsubarachnoidhemor-
pollution.PLoSONE12(10):e0186973.https://doi.
rhage(SAH)hasbeeninvestigated,resultshaveremainedinconsistent.Thepresentstudy
org/10.1371/journal.pone.0186973
aimedtodeterminetheseasonalityofaneurysmalsubarachnoidhemorrhageoccurrence
Editor:ThanhG.Phan,MonashUniversity,
andmortality.
AUSTRALIA
Received:June6,2017
Methods
Accepted:October11,2017
Published:October26,2017 WeusedtheNationalInpatientSampledatabasetoevaluatetheeffectofmeteorological
factorsandairpollutantsonpatientswithsubarachnoidhemorrhageinKoreabetween2011
Copyright:©2017Hanetal.Thisisanopen
accessarticledistributedunderthetermsofthe and2014.MonthlyvariationsinSAHoccurrenceandmortalitywereanalyzedusinglocally
CreativeCommonsAttributionLicense,which weightedscatterplotsmoothingcurves.MultivariatePoissongeneralizedlinearregression
permitsunrestricteduse,distribution,and
modelswereusedtoevaluatepotentialindependentmeteorologicalandpollutantvariables
reproductioninanymedium,providedtheoriginal
associatedwithSAHoccurrenceandmortality.
authorandsourcearecredited.
DataAvailabilityStatement:Allrelevantdataare
withinthepaperanditsSupportingInformation
Results
files.
Intotal,21,407patientswhounderwentcliporcoiltreatmentowingtoaneurysmalSAHin
Funding:Thisstudywassupportedbyagrant
fromtheKoreaHealthTechnologyR&DProject KoreafromJanuary1,2011,toDecember31,2014,wereincluded.Thecrudeincidence
throughtheKoreaHealthIndustryDevelopment rateofSAHinKoreawas10.5per100,000peopleperyear.Anapproximately0.5%lower
Institute(KHIDI),fundedbytheMinistryofHealth
riskofSAHwasobservedper1˚Cincreaseinmeanmonthlytemperature(relativerisk,
&Welfare,RepublicofKorea(grantnumber:
0.995;95%confidenceinterval[CI],0.992–0.997;p<0.001),whileanapproximately2.3%
HC15C3401).URL:https://www.khidi.or.kr/board;
jsessionid=n3vhZ3dBmrs2TrBkQTPGfGkRJV8pqV higherriskofSAHwasobservedper1˚Cincreaseinmeanmonthlydiurnaltemperature.
PLOSONE|https://doi.org/10.1371/journal.pone.0186973 October26,2017 1/16
SeasonalvariationinSAHincidenceandmortality
DlRtr63n4MysznphH0yvZC!-1652677014? Conclusions
menuId=MENU00776&siteId=null.
Weshoweddistinctpatternsofseasonalandmonthlyvariationintheoccurrenceandmor-
Competinginterests:Theauthorshavedeclared
talityofSAH.Ourfindingssuggestthatmeteorologicalfactorsmayplayanimportantrolein
thatnocompetinginterestsexist.
monthlyvariationsintheoccurrenceofaneurysmalSAH.
Introduction
Theinfluenceofweatherandairpollutionontheoccurrenceofaneurysmalsubarachnoid
hemorrhage(SAH)hasbeenwidelyinvestigatednumerouscountries;however,theirresults
remaininconsistent.AlthoughseveralresearchershavereportedthattheincidenceofSAHis
lowerinthesummer,[1,2]additionalstudieshaverevealednoseasonalvariabilityintheoccur-
renceofSAH.[3–6]Variousmeteorologicalfactorssuchastemperature,atmosphericpressure,
andotherweatherparametershavebeeninvestigatedtoevaluatetheirpossibleassociation
withSAHseasonalityandmortality.[2,7–13]Severalhypothesizedmechanismslinkingtem-
perature,SAHoccurrence,andmortalityhavebeensuggested.Coldertemperaturemaybe
associatedwithhigherSAHoccurrenceandmortalitybyincreasedfibrinogenlevels,increased
bloodpressureandsympatheticnerveactivity,andincreasedsystemicinfectionrates.[11]In
addition,thepossiblelinkbetweenSAHoccurrence/mortalityandairpollutionhasalsobeen
investigated.[14–17]Exposuretopollutantswassuggestedtobelinkedtodeteriorationofvas-
cularendothelialfunction,increasedsystemicinflammationandplateletactivation,and
decreasedantioxidantenzymeactivity.[18]
WeaimedtodeterminetheseasonalityofaneurysmalSAHoccurrenceandmortality.In
addition,weusedtheNationalInpatientSample(NIS)databasetoevaluatetheeffectofmeteo-
rologicalfactorsandairpollutantsonSAHoccurrenceandmortalityinKoreafrom2011to
2014.
Methods
Datacollection
ThisstudywasapprovedbytheinstitutionalreviewboardofGachonUniversity(IRBNo.
1044396-201606-HR-043-01).Owingtotheretrospectivenatureofthestudy,theneedfor
informedconsentwaswaived.
WecollectedNISdataonpatientswithSAHfromJanuary1,2011,toDecember31,2014,
availablebytheHealthInsuranceReviewandAssessmentService(HIRA;http://opendata.
hira.or.kr)inKoreaeachyear.DetailedinformationregardingtheNISandHIRAhasbeen
describedelsewhere.[19,20]Aseachpersonhasauniqueresidentregistrationnumber,dupli-
cationofpatientscanbeavoided.[21]AllpatientswithprimaryorsecondarySAHdiagnosis
codesofI60toI609,basedontheInternationalClassificationofDiseases,tenthedition(ICD-
10),weresearchedandextracted.Initially,weidentifiedatotalof71,737patientswithsponta-
neousSAH.However,ourinitialsearchcaptureddatafrompatientswithprolongedhospital
owingtosequelaeofSAHthathadoccurredpriorto2011,idiopathicornon-aneurysmal
SAH,andadmissiondiagnosiscodeerrors.Toenhancethereliabilityofdetectingmonthly
variationintruecasesofnewlydevelopedaneurysmalSAHbetween2011and2014,weonly
includedpatientswhohadundergonetreatmentprocedures(surgicalclippingorendovascular
coiling),despitethepotentialforunderestimationoftheincidencerate.Therefore,we
extracteddatafrompatientswithprocedurecodesforclipping(S4641,simple;S4642,
PLOSONE|https://doi.org/10.1371/journal.pone.0186973 October26,2017 2/16
SeasonalvariationinSAHincidenceandmortality
complex)orcoiling(M1661,assistedcoiling;M1662,other(simple)coiling)fromamongthe
initialsampleof71,737patients.Whenbothclippingandcoilingwereperformedinthesame
individual,thecasewasregardedaseitherclippingorcoilingbasedontheprocedureper-
formedattheearliestdate.Wefinallyincluded21,407patients,whowerefollowedupuntil
December31,2015,toobtaindataregardingSAHmortality.Sex,age,demographicdistribu-
tionofhospitaladmissions,datesofhospitaladmissionanddischarge,andothervariables
wereextractedforallpatients.WedefinedthedayofonsetofSAHbasedontheadmission
dateofthepatient.ThenumberofSAHsoccurringeachmonthwasalsorecorded.
Mortalityassessment
The21,407selectedstudypatientswerefollowedupuntilDecember31,2015,usingdatafrom
theNISdatabase.SincetheHIRAdoesnotprovidemortalityinformation,[19]themortality
datewasdefinedasthedateonwhichthelastdischargecodewasrecorded,whensatisfying
thefollowingcriteria:(1)visitedtheoutpatientdepartment≦2times(incaseofdeathcertifi-
cate)withnomedicationprescribed,andnofurtheradmissionwithin3monthspost-dis-
charge,and(2)didnotusemedicalfacilitiesforanypurposeinKoreauntilDecember31,
2015,after3monthspost-discharge.AllKoreansarebeneficiariesoftheKoreanNational
HealthInsuranceSystem,thusHIRAdatabaseenablingtotracerecordsofnation-wideinpa-
tientandoutpatientdata.[22]Therefore,wecouldidentifiedallrecordsofSAHpatientsvisit-
inganymedicalfacilitiesincludingprimaryhospitals,privateclinics,publichealthcenters,
andpharmaciesinSouthKorea.[23]IntreatedSAHpatients,allpatients(evenpatientswith
modifiedRankinscale0atdischarge)weretobefollowedupbyaphysicianandprescribed
medicationinanoutpatientdepartment(atthehospitalwherepatientwastreatedoranyother
hospitalsinKorea),especiallyduringtheearlyperiodafterdischarge.Mostpatientsinavege-
tativestatearetransferredtoarehabilitationhospitalornursingfacility.Eveniffewvegetative
patientsmaytransfertotheirhome,theguardiansmayneedtousemedicalfacilitiestoadmin-
isterprescribedmedications.Whenthevegetativepatientisfounddeadbytheguardianat
home,thedeadbodymustbecarriedtotheemergencyroomforcadaverexamination,and
subsequently,thedischargecodeshouldberecordedintheemergencyroom.Therefore,we
assumedthedateofdeathtobethedayofdischargeaftercliporcoiltreatmentwhenthese
conditionsaremet.WepresentthedetailedSAHadmissionandmortalitydatabasedoneach
month,classifiedbysexandagegroup(S1Data)Wealsoperformedsensitivityanalysisto
evaluatethesuitabilityofourcriteriawithacut-offvalueof3months.Nosignificantdiffer-
encesinmortalitywereobservedwheneachmonth(2–5)wassetasthecut-offvalue(S1
Table).
Meteorologicalandairpollutiondata
Meteorologicaldataincludingmonthlymeasuresofmeantemperature,averagediurnaltem-
peraturerange,andinsolationfrom2011to2014inKoreawereobtainedfromtheMeteoro-
logicalAdministrationofSouthKorea(http://www.kma.go.kr/eng/).Thesedatawere
collectedandanalyzedfrom73observationstationslocatedthroughoutSouthKorea.[24]
Airpollutantdataincludedaveragemonthlymeasuresofparticulatematterwithanaero-
dynamicdiameter<10μm(PM ),nitrogendioxide(NO ),andsulfurdioxide(SO )in
10 2 2
Koreaforthedurationofthestudyperiod.ThesedatawerecollectedfromAirKorea,which
wasestablishedin2002bytheMinistryofEnvironmentofSouthKorea(www.airkorea.or.
kr/eng/).Thecomprehensiveairpollutiondatawerecollectedandanalyzedfrom251sites
in79cities.[25]
PLOSONE|https://doi.org/10.1371/journal.pone.0186973 October26,2017 3/16
SeasonalvariationinSAHincidenceandmortality
Season
Seasonsweredefinedaswinter(DecemberthroughFebruary),spring(MarchthroughMay),
summer(JunethroughAugust),andautumn(SeptemberthroughNovember).
Statisticalmethods
ThecrudeSAHincidenceratewasestimatedastheaveragenumberofSAHsoccurringinthe
studyperiodper100,000peopleperyear.Discretevariablesareexpressedaspercentages;con-
tinuousvariables,themean±standarddeviationormedianwithinterquartilerange(IQR).
DescriptivestatisticswereusedtodeterminethemedianandIQRofmonthlymeteorologi-
calandairpollutionmeasures.Therelationshipsamongmeteorologicalandairpollutionfac-
torswereevaluatedusingthePearsoncorrelationtest.
Weusedalocallyweightedscatterplotsmoothing(LOWESS)curvewitha95%confidence
interval(CI)tographicallyrepresentthetotalmonthlyvariationintheoccurrence/mortality
ofaneurysmalSAH,whichwasthenstratifiedinaccordancewitheachyear.Student’st-test
wasusedtoidentifyseasonalvariationsinSAHoccurrenceandmortality.Linearregression
linesfortheLOWESScurvewereusedtoidentifytheassociationamongtemperature,diurnal
temperaturerange,andSAHoccurrence.SAHoccurrencewasnaturallog-transformedto
normalizedistributions.
UnivariateandmultivariatePoissongeneralizedlinearregressionmodelswereusedtoeval-
uatepotentialindependentmeteorologicalandpollutantvariablesassociatedwithSAHoccur-
renceandmortality,usingalog-linkagefunctionoffsetbythelogofthepopulationineach
yearfrom2011to2014.ThepopulationofKoreabetween2011and2014wasinvestigated
basedontheKoreanpopulationcensus(http://rcps.egov.go.kr).
WeusedtheggplotfunctioninRtovisualizethemapofSouthKoreausing“ggplot2”and
“Kormaps”packages(e.g.ggplot(kormap1,aes(x=long,y=lat,group=group,fill=id))+
geom_polygon(colour="black")+...).WecreatedamapofSouthKoreainExceltodisplay
geographicdistributionofaveragetemperature,diurnaltemperaturerange,andPM concen-
10
trationsusingagradientspectrumoftwocolors.ThescatterplotwithLOWESScurveandlin-
earlinewereproducedby“moonBook”,“ggthemes”,and“ggplot2”packages(e.g.ggplot
(data=RRR,aes(x=factor(Month),y=Total,group=1))+stat_smooth(colour=
"#2E9FDF",fill="gray75",size=1.1)+geom_point(fill="#2E9FDF",size=2,shape=21)+...
andggplot(data=RRR,aes(x=Temperature,y=log_SAH,group=1))+stat_smooth
(method=lm,size=1.1,aes(colour="Linear"),se=FALSE)+stat_smooth(size=1.1,aes
(colour="LOWESS"),se=FALSE)+...).AllstatisticalanalyseswereperformedusingRver-
sion3.3.2.
Results
DemographicdistributionofSAHoccurrenceandregionalheterogeneity
ofthemeteorologicalfactorsandpollutantsinKorea
Duringthestudyperiod,theaveragepopulationofKoreawas51,037,983.8(Table1andFig1).
Intotal,21,407patientsunderwentsurgicalclippingorendovascularcoilingowingtoaneu-
rysmalSAHbetween2011and2014.ThecrudeincidencerateofSAHinKoreawas10.5per
100,000peopleperyear.TheincidenceratewashighestinGwangju(23.4)andlowestinJeon-
nam(1.4)(Fig1A).OnepossibleexplanationforthisdifferenceisthattheRegionalEmergency
MedicalCenterisinGwangju,whichislocatedinsideJeonnam.Inaddition,Jeonnaminhabi-
tantspreferenceforthehospitalinGwangjualsoaffectedthedifference,despiteanemergency
situation.Fig1B,1Cand1Dshowregionaldifferencesofaveragetemperature,diurnal
PLOSONE|https://doi.org/10.1371/journal.pone.0186973 October26,2017 4/16
SeasonalvariationinSAHincidenceandmortality
Table1. Characteristicsoftheaveragepopulationandpatientswithspontaneousaneurysmalsub-
arachnoidhemorrhageinKoreabetween2011and2014.
Characteristics AveragepopulationofKoreabetween2011and2014
Overall,mean 51,037,983.8
Men,mean(%) 25,542,156.5(50.1)
Women,mean(%) 25,495,827.3(50.0)
<65years,mean(%) 44,924,827.5(88.0)
(cid:21)65years,mean(%) 6,113,156.3(12.0)
Patientsreceivingtreatmentduetospontaneousaneurysmal
SAHinKorea
Overall,n 21,407
Crudeincidencerateper100,000person- 10.5
year
Sex
Men,n(%) 7,679(35.9)
Women,n(%) 13,728(64.1)
Age,mean±SD(median) 56.3±12.9(55)
<65years,n(%) 15,470(72.3)
(cid:21)65years,n(%) 5,937(27.7)
Treatment
Clip,n(%) 11,485(53.7)
Coil,n(%) 9,922(46.3)
Lengthofstay,median(IQR) 20(12–29)
Charlsonco-morbidityIndex,mean±SD 2.3±1.5(2)
(median)
Inhospitalmortality,n(%) 1,926(9.0)
Mortalityrate,% 9.0
Medicalcharges,median(IQR),₩ 13,833,740(10,564,160to18,790,240)
NumberofSAHsperyear
2011,n(%) 5,471(25.6)
2012,n(%) 5,255(24.5)
2013,n(%) 5,307(24.8)
2014,n(%) 5,374(25.1)
SAH,subarachnoidhemorrhage;SD,standarddeviation;IQR,interquartilerange
https://doi.org/10.1371/journal.pone.0186973.t001
temperaturerange,andPM inSouthKoreabetween2011and2014basedonsummerand
10
winterseasons.Theaveragetemperaturevariedbyabout3˚C~8˚Cand4˚C~6˚Cvariation
ofdiurnaltemperaturerangethroughoutnationfrom2011to2014insummerandwinter.
PM rangedfromapproximately13μg/m3to27μg/m3.Wealsopresentdetailedregional
10
measuresofthemeteorologicalfactorsandairpollutantsinSouthKoreaeachyearclassified
bysummerandwinterseasons(S2Data).
Characteristicsofthestudypopulation
TheaverageageofpatientsatSAHoccurrencewas56.3years,and64.1%ofpatientswere
women.Surgicalclippingwasperformedin11,485(53.7%)patients.Themedianlengthof
admissionwas20days,andthemortalityratewas9.0%.Patientcharacteristicsarefurther
detailedinTable1.
PLOSONE|https://doi.org/10.1371/journal.pone.0186973 October26,2017 5/16
SeasonalvariationinSAHincidenceandmortality
Fig1. Demographicdistributionofpatientswhounderwentproceduresforaneurysmalsubarachnoidhemorrhageandregionalheterogeneity
ofthemeteorologicalfactorsandpollutantsinSouthKoreabetween2011and2014:(A)demographicdistributionofsubarachnoidhemorrhage
occurrence(B)regionalaveragetemperaturebasedonsummerandwinterseasons;(C)regionaldiurnaltemperaturerangebasedonsummerandwinter
seasons;(D)regionalparticulatematterwithanaerodynamicdiameter<10μmconcentrationsbasedonsummerandwinterseasons.
https://doi.org/10.1371/journal.pone.0186973.g001
Monthlymeasuresofmeteorologicalfactorsandpollutants
Table2showsthemonthlydistributionofmeteorologicalparametersandairpollutantswith
medianandIQRvaluesfrom2011to2014.ThepairwisePearsoncorrelationcoefficients
amongmeteorologicalfactorsandpollutantswereestimated(S2Table).Temperatureexhib-
itedasomewhatsignificantcorrelationwithairpollutionvariables,withthestrongestnegative
PLOSONE|https://doi.org/10.1371/journal.pone.0186973 October26,2017 6/16
SeasonalvariationinSAHincidenceandmortality
Table2. MonthlysubarachnoidhemorrhageoccurrenceandaveragemonthlymeteorologicalfactorsandairpollutantsinKoreabetween2011
and2014.
Meteorologicalfactors Pollutants
Month SAH Temperature Diurnaltemperaturerange Insolation(hour), PM (μg/m3), NO (ppb), SO (ppb),
10 2 2
occurrence, (˚C), (˚C), median(IQR) median(IQR) median(IQR) median(IQR)
n(%) median(IQR) median(IQR)
Jan 1954(9.1) -1.7(-3.8to0.1) 10.0(9.6to10.8) 196.3(183.0to 57.3(44.9to 26.6(23.4to 7.5(6.3to
214.4) 61.1) 28.7) 7.6)
Feb 1735(8.1) 1.3(-0.4to2.4) 10.2(10.0to11.0) 173.0(150.4to 51.4(48.3to 24.6(23.6to 6.4(5.9to
213.5) 64.1) 28.7) 6.7)
Mar 1960(9.2) 6.2(4.9to7.4) 11.3(10.1to12.8) 226.6(187.9to 58.5(51.5to 22.0(20.9to 5.5(5.3to
246.5) 61.8) 23.3) 5.8)
Apr 1815(8.5) 11.2(10.5to 12.2(12.1to12.2) 213.2(212.5to 53.6(51.5to 22.2(20.2to 4.9(4.8to
12.9) 217.2) 56.4) 36.2) 5.2)
May 1779(8.3) 18.1(17.4to 11.8(10.9to12.7) 236.2(191.0to 62.3(56.8to 20.3(19.4to 5.2(4.8to
18.4) 278.5) 71.1) 20.6) 5.3)
Jun 1659(7.7) 22.0(21.9to 9.2(9.0to9.5) 179.6(173.1to 44.2(39.9to 16.7(15.1to 4.6(4.3to
22.5) 188.0) 46.7) 17.9) 5.0)
Jul 1612(7.5) 25.3(25.1to 7.4(6.9to8.2) 157.5(123.4to 33.9(31.8to 14.4(13.3to 4.2(4.1to
26.1) 177.1) 37.7) 14.8) 7.1)
Aug 1628(7.6) 25.8(24.2to 7.3(6.9to8.6) 142.0(114.1to 30.1(26.8to 14.2(12.7to 3.8(3.6to
27.1) 223.7) 36.0) 14.6) 4.3)
Sep 1642(7.7) 21.0(20.4to 9.5(9.1to10.0) 184.9(180.2to 32.6(32.3to 16.7(16.2to 4.0(3.8to
21.2) 198.1) 32.9) 16.8) 4.7)
Oct 1915(8.9) 14.6(13.9to 11.7(11.5to12.1) 224.9(206.8to 38.1(35.5to 21.7(20.1to 4.1(3.9to
15.3) 233.3) 43.7) 23.4) 4.2)
Nov 1919(9.0) 8.0(6.7to10.5) 10.0(9.3to10.4) 177.8(140.7to 44.2(43.8to 23.4(22.5to 4.9(4.8to
190.2) 46.3) 25.2) 5.2)
Dec 1789(8.4) 0.1(-1.4to1.3) 9.2(9.0to9.3) 179.3(173.4to 43.5(41.0to 24.5(23.4to 5.9(5.7to
188.5) 49.4) 25.9) 6.2)
SAH,subarachnoidhemorrhage;IQR,interquartilerange;PM ,particulatematterlessthan10mminaerodynamicdiameter;NO ,nitrogendioxide;SO ,
10 2 2
sulfurdioxide
https://doi.org/10.1371/journal.pone.0186973.t002
correlationobservedforNO2(r=-0.758).Diurnaltemperaturerangewaspositivelycorre-
latedwithinsolation(r=0.787).
MonthlyvariationsofSAHoccurrenceandmortality
SAHoccurrencewashighestinMarch(n=1,960)andlowestinJuly(n=1,612)(Table2).The
overallmonthlyvariationinSAHexhibitedaV-shapedpattern,withatendencytodecrease
fromJanuarytosummerandatendencytoincreasefromsummertoDecember,withfour
peaksinJanuary,March,October,andNovember(Fig2A).
SAHoccurrencewaslowerinthesummerthaninthewintereachyear(Fig2C).Thispat-
ternwasmaintainedwhenpatientsweredividedintosexandagegroups(S1AandS1BFig).A
significantdifferenceinSAHoccurrencebetweensummerandwinterwasobservedinwomen
(p=0.038),notinmen(p=0.104).
MortalitywashighestinMarch(n=199)andlowestinDecember(n=124)(Fig2B).
VariationsinmortalityexhibitedapatternsimilartothoseforSAHoccurrencefromJan-
uaryuntilearlyautumn.However,mortalitytendedtodecreaseattheendoftheyear,
despiterelativelyhigherSAHoccurrence.Relativelyirregularpatternsofmortalitywere
observedeachyearfrom2011to2014(Fig2D).Monthlyvariationsinmortalityaccording
tosexandagearedepictedinFigCandDoftheS1Fig.
PLOSONE|https://doi.org/10.1371/journal.pone.0186973 October26,2017 7/16
SeasonalvariationinSAHincidenceandmortality
Fig2.LOWESSregressionlinewith95%confidenceintervalshowingthepatternofmonthlySAHoccurrenceand
mortality(overallandstratifiedbyyear).
https://doi.org/10.1371/journal.pone.0186973.g002
Combinedeffectofmonthlymeantemperatureanddiurnaltemperature
rangeonSAHoccurrence
ThelowerincidenceofSAHinthesummermaybesomewhatexplainedbytheeffectoftem-
peratureonSAH.Weobservedanapproximate0.5%decreaseinSAHoccurrenceper1˚C
increaseinambienttemperature,withabout40%explanatorypowerinthelog-transformed
linearregressionanalysis(β=-0.005;p<0.001;R2=0.396).Acontinuousdownwardslope
wasalsonotedfortemperaturesaboveapproximately10–15˚C(Fig3A).
Thisstatisticalsignificancewasmaintainedwhenwedividedpatientsintosexandage
groups(Fig3B).Theresultsoftheunivariateandmultivariatelinearregressionanalysesof
PLOSONE|https://doi.org/10.1371/journal.pone.0186973 October26,2017 8/16
SeasonalvariationinSAHincidenceandmortality
Fig3.LinearandLOWESSregressioncurveofnaturallog-transformedsubarachnoidhemorrhageoccurrence
basedontemperatureanddiurnaltemperaturerange,overall,andstratifiedaccordingtogenderandage.
https://doi.org/10.1371/journal.pone.0186973.g003
log-transformedSAHoccurrencebasedonmeteorologicalandairpollutionparametersare
presentedinS3Table.WeobservedthattemperaturewasassociatedwithSAHoccurrence
(RR,0.995;95%CI,0.992–0.997;p<0.001per1˚Cincrement)inthemultivariatePoisson
regressionafteradjustingforallmeteorologicalfactorsandpollutants(Table3).Thissignifi-
cantrelationshipwasalsoobservedforwomen,youngerpatients(<65years),andolder
patients(≧65years)inthemultivariateanalysis(S4Table).
HigherSAHoccurrencewasalsoobservedwithhigherdiurnaltemperaturechangesduring
thechangeofseasonsinMarch,October,andNovember.Diurnaltemperaturerangeexhibited
apositiveassociationwithSAHoccurrence,withapproximately26%explanatorypower(β=
0.025;p<0.001;R2=0.255).Anupwardslopewasalsoobservedbetween8–11˚Cinthe
LOWESScurve(Fig3C).Thisassociationpersistedwhenpatientsweredividedintosexand
agegroups(Fig3D).MultivariatePoissonregressionrevealedanapproximately2.3%
increasedriskofSAHoccurrenceper1˚Cincreaseinaveragemonthlydiurnaltemperature
range(RR,1.023;95%CI,1.003–1.044;p=0.025)(Table3).WithregardtoSAHmortality,
diurnaltemperaturerangeandinsolationshowedsignificantassociationintheunivariateanal-
ysis(RR,1.042;95%CI,1.013–1.071;p=0.004;RR,1.001;95%CI,1.000–1.003;p=0.032,
respectively).However,therewerenoassociationsbetweenmeteorologicalandairpollution
factorsandSAHmortalityinthemultivariateanalysis.
PLOSONE|https://doi.org/10.1371/journal.pone.0186973 October26,2017 9/16
SeasonalvariationinSAHincidenceandmortality
Table3. Relativeriskofsubarachnoidhemorrhageoccurrenceandmortalitybasedonmeteorologicalfactorsandairpollutants.
SAHoccurrence
UnivariatePoissonregression MultivariatePoissonregression
Variable RR(95%CI) p RR(95%CI) p
Meteorologicalfactors(per1unitincrease)
Temperature 0.995(0.993to0.996) <0.001 0.995(0.992to0.997) <0.001
Diurnaltemperaturerange 1.025(1.016to1.033) <0.001 1.023(1.003to1.044) 0.025
Insolation 1.001(1.000to1.001) 0.001 1.000(0.999to1.000) 0.350
Pollutants(per1unitincrease)
PM 1.003(1.002to1.004) <0.001 1.000(0.998to1.002) 0.811
10
NO 1.008(1.006to1.011) <0.001 0.999(0.994to1.004) 0.672
2
SO 1.025(1.012to1.037) <0.001 0.995(0.975to1.015) 0.616
2
SAHmortality
Variable RR(95%CI) p RR(95%CI) p
Meteorologicalfactors(per1unitincrease)
Temperature 0.999(0.994to1.004) 0.678 1.000(0.991to1.008) 0.909
Diurnaltemperaturerange 1.042(1.013to1.071) 0.004 1.063(0.994to1.137) 0.073
Insolation 1.001(1.000to1.003) 0.032 0.999(0.997to1.002) 0.578
Pollutants
(per1unitincrease)
PM 1.003(0.999to1.007) 0.128 1.000(0.993to1.007) 0.944
10
NO 1.004(0.995to1.012) 0.372 0.995(0.979to1.011) 0.552
2
SO 0.991(0.952to1.033) 0.678 0.991(0.927to1.061) 0.802
2
SAH,subarachnoidhemorrhage;RR,relativerisk;CI,confidenceinterval;PM ,particulatematterlessthan10mminaerodynamicdiameter;NO ,
10 2
nitrogendioxide;SO ,sulfurdioxide
2
https://doi.org/10.1371/journal.pone.0186973.t003
MonthlyambienttemperaturewassignificantlyassociatedwithSAHoccurrenceamong
womenandinbothagegroups,whilediurnaltemperaturerangeexhibitedapositiveassocia-
tionwithSAHoccurrenceinmen(RR,1.049;95%CI,1.014–1.085;p=0.006),afteradjusting
forallmeteorologicalfactorsandpollutants(S4Table).
Discussion
Inthepresentstudy,wefoundthatbothmeanmonthlyanddiurnaltemperaturerangeaffect
SAHoccurrence.Lowermeantemperaturesweresignificantlycorrelatedwithhigheraneurys-
malSAHoccurrence.Inaddition,higherdiurnaltemperaturerangewaspositivelyassociated
withSAH,thisassociationbeingsignificantlymoreprominentinmenthaninwomen.In-hos-
pitalmortalitywashighestinMarchandtendedtodecreaseattheendoftheyear,despitean
increaseintheincidenceofSAHatthistime.Mortalityexhibitednoassociationwitheither
meteorologicalfactorsandpollutantsinthemultivariateanalysis.
PreviousstudieshavesupportedourfindingthattheincidenceofSAHislowerinthesum-
merthaninotherseasons.[26–28]Arecentmeta-analysisalsoreportedthataneurysmalSAH
occurrencewaslowinsummerthaninwintermonths,peakinginJanuary.[29]Previousstud-
ieshavereportedinconsistentfindingsregardingtheassociationbetweenambienttempera-
tureandSAHoccurrence.Inthepresentstudy,wealsoobservedlowlinearcorrelation
betweenSAHoccurrenceandlowerambienttemperature(belowaround10–15˚C),despitean
overallsignificantnegativecorrelationbetweenSAHoccurrenceandtemperature.Further-
more,weobservedasignificantcorrelationbetweenSAHandmeanmonthlydiurnaltempera-
turerange.Gilletal.reportedthatcoldertemperaturesandtemperaturefluctuationsfrom
PLOSONE|https://doi.org/10.1371/journal.pone.0186973 October26,2017 10/16