Climatic and meteorological exposure and mental and behavioral health: A systematic review and meta-analysis

Abstract

As climate change exerts wide ranging health impacts, there is a surge of interest in the associations between climatic factors and mental and behavioral disorders (MBDs). Existing quantitative syntheses focus mainly on heat and high temperature exposure, neglecting the effects of other climatic factors and their synergies. The objective of this study is to conduct a systematic review and meta-analysis of the evidence of associations between climatic exposure and combined mental and behavioral health conditions and specific mental disorders (e.g., schizophrenia, dementia). A systematic search was conducted April 11-16, 2022 using Web of Science, Medline, ProQuest, EMBASE, PsycINFO, CINAHL, and Environment Complete. Screening and eligibility screening followed inclusion criteria based on population, exposure, comparator, and outcome guidelines. Risk of bias assessment was performed, a narrative synthesis was first presented for all studies, and random-effect meta-analyses were performed when at least three studies were available for a specific exposure-outcome pair. Certainty of evidence was evaluated following the Grading of Recommendations Assessment, Development and Evaluation (GRADE) tool. The search process yielded 7696 initial results, from which we identified 88 studies to include in the review set. Climatic factors reported included air temperature, solar radiation/sunshine, barometric pressure, precipitation, relative humidity, wind direction/speed, and thermal comfort index. Outcomes included MBD incidences (e.g., schizophrenia, mood disorders, neurotic disorders), mental health-related mortality, and self-reported psychological states. Meta-analysis showed that heatwaves (pooled RR = 1.05, 95 % CI = 1.02-1.08) and extreme high temperatures (99th percentile: pooled RR = 1.18, 95 % CI = 1.08-1.29) were associated with higher risk of MBD. Cold extremes, however, were not associated with MBD risk. The findings further identified an association between increases in a thermal index (i.e., apparent temperature) and elevated risk of MBD (pooled RR = 1.06, 95 % CI = 1.03-1.12); specifically, a 99th percentile high temperature was associated with increased schizophrenia risk (pooled RR = 1.07, 95 % CI = 1.01-1.12). Risk of bias assessment showed most studies to have low or moderately low risks, while a few studies were rated probably high in confounding, selection bias, outcome measurement, and reporting bias. GRADE evaluation revealed moderate certainty of evidence on thermal comfort index and MBD, but low certainty related to air temperature or sunshine duration. These findings call attention to the heterogeneity of exposure measures and the utility of thermal indices that consider the synergistic effects of meteorological factors. Methodological concerns such as the linearity assumption and cumulative effects are discussed.

Full text

CORRECTED PROOF
Review
Climatic and meteorological exposure and mental and behavioral health: A
systematic review and meta-analysis
DongyingLia, ,YueZhanga,XiaoyuLia,KaiZhangb,YiLuc,RobertD. Browna
a DepartmentofLandscapeArchitectureandUrbanPlanning,TexasA&MUniversity,CollegeStation,TX77843,USA
b DepartmentofEnvironmentalHealthSciences,SchoolofPublicHealth,UniversityatAlbany,StateUniversityofNewYork,Albany,NY12222,USA
c DepartmentofArchitectureandCivilEngineering,CityUniversityofHongKong,HongKong
A R T I C L E I N F O
Editor:SCOTTSHERIDAN
Keywords:
Climate
Mentalhealth
Schizophrenia
Mooddisorders
Neuroticdisorders
meta-analysis
A B S T R A C T
Asclimatechangeexertswideranginghealthimpacts,thereisasurgeofinterestintheassociationsbetweencli-
maticfactorsandmentalandbehavioraldisorders(MBDs).Existingquantitativesynthesesfocusmainlyonheat
andhightemperatureexposure,neglectingtheeffectsofotherclimaticfactorsandtheirsynergies.Theobjective
ofthisstudyistoconductasystematicreviewandmeta-analysisoftheevidenceofassociationsbetweenclimatic
exposureandcombinedmentalandbehavioralhealthconditionsandspecificmentaldisorders(e.g.,schizophre-
nia,dementia).
Asystematic searchwasconductedApril11
–
16,2022 usingWebofScience,Medline, ProQuest,EMBASE,
PsycINFO,CINAHL,andEnvironmentComplete.Screeningandeligibilityscreeningfollowedinclusioncriteria
basedonpopulation,exposure,comparator,andoutcomeguidelines.Riskofbiasassessmentwasperformed,a
narrativesynthesiswasfirstpresentedforallstudies,andrandom-effectmeta-analyseswereperformedwhenat
leastthreestudieswereavailableforaspecificexposure-outcomepair.Certaintyofevidencewasevaluatedfol-
lowingtheGradingofRecommendationsAssessment,DevelopmentandEvaluation(GRADE)tool.
Thesearchprocessyielded7696initialresults,fromwhichweidentified88studiestoincludeinthereview
set.Climaticfactorsreportedincludedairtemperature,solarradiation/sunshine,barometricpressure,precipita-
tion,relativehumidity,winddirection/speed,andthermalcomfortindex.OutcomesincludedMBDincidences
(e.g.,schizophrenia,mooddisorders,neuroticdisorders),mentalhealth-relatedmortality,andself-reportedpsy-
chologicalstates.Meta-analysisshowedthatheatwaves(pooledRR=1.05,95%CI=1.02
–
1.08)andextreme
hightemperatures(99thpercentile:pooledRR=1.18,95%CI=1.08
–
1.29)wereassociatedwithhigherrisk
ofMBD.Coldextremes,however,werenotassociatedwithMBDrisk.Thefindingsfurtheridentifiedanassocia-
tion between increases in a thermal index (i.e., apparent temperature) and elevated risk of MBD (pooled
RR=1.06,95%CI=1.03
–
1.12);specifically,a99thpercentilehightemperaturewasassociatedwithin-
creasedschizophreniarisk(pooledRR=1.07,95%CI=1.01
–
1.12).
Riskofbiasassessmentshowedmoststudiestohavelowormoderatelylowrisks,whileafewstudieswere
ratedprobablyhighinconfounding,selectionbias,outcomemeasurement,andreportingbias.GRADEevalua-
tionrevealedmoderatecertaintyofevidenceonthermalcomfortindexandMBD,butlowcertaintyrelatedtoair
temperatureorsunshineduration.Thesefindingscallattentiontotheheterogeneityofexposuremeasuresand
theutilityofthermalindicesthatconsiderthesynergisticeffectsofmeteorologicalfactors.Methodologicalcon-
cernssuchasthelinearityassumptionandcumulativeeffectsarediscussed.
1. Introduction
Climate extremes have adverse effects on health. With climate
change, climate extremes are expected to increase in intensity, fre-
quency,andhealthimpact(Petkovaetal.,2013;Guoetal.,2018).Such
extremeeventsarenotisolatedincidences,butarereflectiveoflarger-
scale, often persistent changes in the thermodynamic environment
acrosscitiesandcountryside(Trenberthetal.,2015).Infact,climate
changeisalreadyaffectingeveryinhabitedregionworldwide,causing
changesinecosystemservices(Shawetal.,2011)andambientfactors
towhichhumansareroutinelyexposedineverydaylife,suchastem-
perature,relativehumidity,andterrestrialradiation.Forexample,ob-
⁎
Correspondingauthor.
E-mailaddress:dli@arch.tamu.edu(D.Li).
https://doi.org/10.1016/j.scitotenv.2023.164435
Received21 January 2023;Received in revised form22 May 2023;Accepted22 May 2023
0048-9697/© 20XX
Edit Preview PDF
This is the pre-proof version of the published manuscript generated by the publisher &
recompiled by the authors. Some content may be slightly different from the final published
version.

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
servedglobalsurfacetemperatureincreasedby>1.2°Cfrom1850to
2020, with the steepest increase occurring after 1990 (Masson-
Delmotteetal.,2021).Atthemicroclimatescale,thisincreaseinsur-
facetemperatureisdirectlylinkedtorisingmeanradianttemperature
andhumanbodyheatfluxintheenvironment(GálandKántor,2020).
Strongevidencehas associated climate extremeswithcardiovascular
disease,respiratorydisease,infectiousdisease,andotherphysicaldis-
eases(Bragaetal.,2002;Medina-Ramónetal.,2006;Wuetal.,2016;
Obradovichet al.,2018a).Moreover,climatechangemay exacerbate
health disparities by affecting minority groups and socially-
disadvantagedgroupsdisproportionately(Voelkeletal.,2018;Lietal.,
2022a).Thus,acomprehensiveunderstandingoftherelationshipsbe-
tweenmeteorologicalfactorsandhealthisaresearchprioritytotheur-
gentsocietal goals of climate changeadaptationanddisparityreduc-
tion.
Intherecentdecade,epidemiologicalstudieshaverevealedcritical
links between meteorological conditions, especially temperature ex-
tremes,andmentalillness(Hayesetal.,2018;Berryetal.,2010).Re-
searchhasidentifiedimpactsofthreetypesofclimate-relatedevents
(i.e., acute, subacute, and long-lasting changes) on mental health
(PalinkasandWong,2020).Forexample,high andlowtemperatures
mightbefactorsthatcontributetomentaldisorders(Obradovichetal.,
2018b; Shiloh et al., 2005; Williams et al., 2012; Yoo et al., 2021;
Zhangetal.,2020),andheateventsthatoccurduetoincrementaltem-
peraturechangesinsummercouldposeasalientrisktohumanmental
conditions().Todate,severalsystematicreviewshaveassessedtheas-
sociationsbetweenextremeweathereventsandhumanmentalhealth.
Ratajet al.(Ratajetal.,2016)investigatedthe prevalence ofmental
disordersindevelopingcountries(exceptAfricanregions)duringex-
tremeweatherevents(i.e.,stormsandflooding)with17articlespub-
lishedby2014.Rotheretal.(Rotheretal.,2021)filledtheresearchgap
inAfricatosomeextentandexaminedexistingfindingsontheimpact
offloodingonchildandadolescentmentalhealthbasedonasystematic
reviewthat identified only two articles meeting criteria.Focusingon
Europeancountries,Cruzetal.(Cruzetal.,2020)andWeilnhammeret
al.(Weilnhammeretal.,2021)examinedthefindingsfrom21and35
articlespublishedupto2019
–
2020,respectively,andsynthesizedthe
mentalhealthimpactsarisingfromextremeweathereventsinvolving
heat/coldwaves,droughts,wildfires,andfloods.Furthermore,system-
aticreviewsfocusingonextremeheathavesummarizedtheepidemio-
logicalevidenceoftheimpactofhightemperatureandheatwaveson
mentalhealth-relatedmorbidityandmortality(Thompsonetal.,2018;
Liuetal.,2021).Thompsonand colleagues summarized thefindings
from34articlespublishedthrough2017andpresentedstronglinksbe-
tweenhightemperaturesandsuiciderisk(Thompsonetal.,2018).No-
tably,Liuetal.(Liuetal.,2021)conductedasystematicreviewof53
articlespublishedbetween1990and2020andameta-analysisof41by
poolingtheeffectsizesofhighambienttemperaturesandreporteda
0.9%increase in mentalhealth-related morbidity for every1 °C in-
creaseintemperature.
Therecent surgeofpublishedresearchin climatedeterminantsof
mentalhealthwarrantsacomprehensivequantitativesynthesis.Specif-
ically,severalmajorresearchgapsremaintobeaddressedwithasys-
tematicreview.Existingreviewsoftenfocusonhotweatherorextreme
heatconditionsasthesoleexposureofinterest.Althoughtheirfindings
providesolidgroundforarelationshipbetweenambienttemperatures
andmentalillness,wehavealimitedunderstandingoftheinfluencesof
multiple,orthecombinedeffects,ofmeteorologicalvariablesonhu-
man-environmentheatfluxandhealthoutcomes.Accordingtobiome-
teorology,humanheatandcoldregulationdependsonabalancedheat
budget(Jietal.,2022;BrownandGillespie,1986).Environmentalfac-
torsthatinfluencehumanheat/coldstressinclude:radiation,tempera-
ture,vaporpressure,anddiffusionconductancetoheatandvapor.As
such, widely accepted thermophysiological comfort and heat/cold
stressmodels consider fourmeteorologicaldeterminantsairtempera-
ture,humidity,windspeed,andsolarradiation(Jietal.,2022),along
withindividualcharacteristicssuchasBMI,physicalactivity,andcloth-
ing insulation (Zhao et al., 2021; Höppe, 1999a). Research has re-
viewedthat,controllingforairtemperature,otherclimatefactorssuch
ashumidity and radiationaffectshumanthermalsensationandcom-
fort.Therelationshipsamongtemperature,humidity,andradiationare
non-linear(Lietal.,2018)andtheirrelativeimportanceintheircontri-
butiontothermalsensationvaryacrossseasons(e.g.,airtemperature
contributesstronglyinsummerwhileradiationcontributesstronglyin
winter)(Liuetal.,2016).Assuch,acomprehensivereviewthatcriti-
callyassesses the fullrangeofmeteorologicalfactorsassociatedwith
mentalhealthiswarranted.
Itisalsoworthnotingthatpreviousempiricalarticlesandreviews
outliningtheimportanceofclimateandweatherconditionsoftendif-
ferentiatemorbidity/hospital admissions and mortality, but consider
mentaldisorderas one outcomecategory without differentiating the
specificdiagnoses/subdiagnoses(Williamsetal.,2012;Charlsonetal.,
2021).Suchanapproachmaynotcapturecomplexitiesrelatingtothe
variouscausesandsymptomologyofeachdifferenttypesofMBD(e.g.,
schizophrenia,depression,andpost-traumaticstressdisorder).Forex-
ample,inarecentreviewandmeta-analysis,Liuetal.presenteddiffer-
enteffectsizesbetweenhightemperaturesandspecificMBDs,suggest-
ingvariationsinthepresence,magnitude,andpossiblyevendirection
andmechanismoftherelationships(Liuetal., 2021). In addition to
temperature, some mental disorders (e.g., bipolar disorders, depres-
sion,andschizophrenia)maybestronglyrelatedtooneoracombina-
tionofmultipleclimaticconditionsbutnotothers,suchassunlightin-
tensity, humidity, and wind conditions (Gu et al., 2019; Kim et al.,
2021;Leeetal.,2007;Molin et al., 1996). As such,estimatedeffect
sizesmaydifferacrossdifferenttypesofmentalhealthconditions,re-
quiringamorenuancedreviewthatconsiderseachexposure-outcome
pair.Inaddition,asthecontemporarydefinitionofhealthemphasizes
notonlytheabsenceofdiseasebutalsopositivemoodstatesandhappi-
ness (World Health Organization, 2002), evidence on mental health
outcomesotherthanmorbidityandmortalityshouldbeincluded.
Furthermore,climatechangeandclimaticfactorsaffecteachregion
differently.MorerecentstudieswereconductedintheGlobalSouth,ex-
panding our understanding of the global mental health impact of
weathervariability;ithasbeensuggestedthattherelativeriskofmen-
talhealth-relatedmortalitymightvaryindifferentclimatezones,e.g.,
higherintropicalzonesthaninsubtropicalorcontinentalzones(Liuet
al.,2021).Givensuchpotentialgeographicalvariability,itiscriticalto
conductareviewthatidentifiesthegeographicalandclimatezonesthat
yetremainunderrepresented.
Toourknowledge,noreviewhasprovidedaquantitativesynthesis
regardingtherelationshipsofthefullsetofclimateandmeteorological
factorswithmentalhealthandMBD.Consideringthegrowingbodyof
evidenceonthistopic,thecurrentstudyaimstodepictacompletepic-
tureoftheserelationships,modelpooledeffectsizes,discussresearch
gapsandrisk of bias, andpropose future directions. Thespecific re-
searchquestionsexaminedinclude:
1. What are the geographical, temporal, and to pical trends in the
literatureonclimatefactorsandmentalhealth?
2. Whatarethedirectionandmagnitudeoftherelationshipbetween
each climate/meteorological factor and mental and behavioral
disorders?
3. How strong and consistent are the relationships when each
mentalandbehavioraldisorderisassessed?
4. Whataretheknowledgegapsintheexistingliterature,andwhat
arethemethodologicalconcernsrelatingtostudybias?
2

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
2. Methods
2.1. PECOframeworkandliteraturesearch
TheprotocolwasregisteredattheInternationalProspectiveRegis-
ter of Systematic Reviews database (PROSPERO, registered ID:
CRD42022321928).Wedevelopedthesystematicreviewprotocolfol-
lowingthePreferredReportingItemsforSystematicReviewsandMeta-
Analysis(PRISMA)guidelines(Moheretal.,2009),whichsetstandards
forafour-stepprocesstoconductrigoroussystematicreviews:identifi-
cation,screening,eligibility,andinclusion.
ThestudysearchwasconductedinApril11
–
16,2022,insevenelec-
tronic databases: Web of Science, Medline, ProQuest, EMBASE,
PsycINFO,CINAHL,andEnvironmentComplete.Thesedatabaseswere
selectedsoastobestcovertheliteraturefromabroadarrayoffields
suchasclimatescienceandmeteorology,environmentalscience,psy-
chologyandpsychiatry,public health, and medicine.Wedid not re-
strictthesearchperiod,sothesearchidentifiedarticlesfromtheincep-
tionofeachdatabaseuntilthesearchdate.Weadoptedthepopulation,
exposure,comparator, and outcome (PECO) framework to guidethe
searchandeligibilitycriteria.
2.1.1. Population
Weincluded studies reporting the general human population ex-
posedtovaryinglevelsofclimate/meteorologicalconditions.Animal
studieswereexcluded.Inordertoidentifyacompletesetofstudies,we
didnot restrict the geographicalareasordemographicandsocioeco-
nomiccharacteristics.
2.1.2. Exposure
Climaticandmeteorologicalindicatorsthatarerelevanttohuman
healthandwell-beingwereconsideredasexposure.Toensurecompa-
rability,studiesthatreportedobjectivelymeasuredclimate/meteorol-
ogy/weatherconditions were included,while those onlyconsidering
perceivedambientconditionswereexcluded.Toensurethereviewen-
compassedrelevantexposuretypes,weusedexistingbiometeorological
modelsofhuman-environmentenergyexchangesuchasthePETmodel
(Höppe, 1999b) and the COMFA outdoor thermal comfort model
(BrownandGillespie,1986).Theyestimatehumanheatfluxinanenvi-
ronmentbasedonambientfactorssuchas airtemperature,humidity,
windspeed,andsolarradiation.Inaddition,otherclimateconditions
suchasbarometricpressure, precipitation (e.g.,rain,snow), and fog
werealsoconsidered.Adetailedtablelistingtheexposuredomainsis
presentedinSupplementarymaterialS1.
2.1.3. Comparators
Acomparablepopulationexposedtodifferentlevelsofclimatecon-
ditionsneedstobepresented inordertoestimate therisksofmental
healthdisordersorthevaluesofreportedpsychologicalconditions.As
such,onlystudiesinvolvingacontrolpopulationorreportingvarying
exposureofthesamepopulationwereincluded.
2.1.4. Outcome
Mental health-related outcomes in this study include mortality,
morbidity,andself-reportedmentalhealthandemotionalstates.Mor-
talityandmorbidityrelatedtermswereselectedbasedontheInterna-
tionalClassificationofDiseasesTenthRevision(ICD-10)mental,behav-
ioral, and neurodevelopmental disorders (F00
–
99) classification, in-
cludingsubcategoryterms.ThesetermsencompasstheentireMBDdo-
main,includingdementiaand organic disorders(F00
–
09),psychoac-
tivesubstanceuse(F10
–
19),schizophrenia(F20
–
29),mooddisorders
(F30
–
39), neurotic disorders (F40
–
49), behavioral syndromes
(F50
–
59), personality disorder (F60
–
69), intellectual disabilities
(F70
–
79),developmentaldisorders(F80
–
89),childhoodbehavioraldis-
orders(F90
–
98),andothermentaldisorders(F99).Equivalentdefini-
tionsusingICD-9or the Diagnostic andStatisticalManualof Mental
Disorders, Fifth Edition (DSM-5) were included and cross-walked to
ICD-10classes.Inaddition,weincludedgeneraltermssuchas
“
mental
health
”
andtermsthatdescribesubjectivepsychologicalstate,suchas
“
happiness.
”
and
“
sadness
”
.A detailed table listing the outcome do-
mainsispresentedinSupplementarymaterialS1.
Thesearchsyntaxwasdeveloped based on the PECOframework.
Weusedwildcardstoaccountforvaryingformsofthekeywords.Inad-
dition,forwardandbackwardsearchesusingeligiblearticlesandprevi-
ously conducted systematic reviews were also performed. Example
searchsyntaxusedfortheWebofScienceisprovidedinSupplementary
materialS2.
2.2. Studyselection
AfterimportingtheretrievedarticlesintoEndnote20andremoving
duplicates,weperformedthescreeningandeligibilitystepsbyexamin-
ingthetitles,abstracts,andfulltextsforconcurrencewiththeeligibil-
itycriteriadefinedbasedonthePECOframework.Theselectiondeci-
sionforeachrecordwasmadeindependentlybytworesearchers,with
anydisagreementresolvedthroughdiscussionandthenbyconsultinga
thirdresearcherwhennecessary.Studieswereincludedinthereviewif
theymetthefollowingcriteria.
1) Studyreportedoriginalempiricalresearchandwaspublishedina
peer-reviewedjournal
2) StudywaswritteninEnglish
3) Study reported an observational study on a human population
(seeSection2.1.1)
4) Study included objective climate measurements as the exposure
(seeSection2.1.2)
5) Study included mental health or behavior as the outcome (see
Section2.1.3)
6) Studiesinvolvedacomparisonpopulationorvaryingexposureof
thesamepopulation(seeSection2.1.4)
7) Studywasquantitativeandreportedatleastoneeffectestimateof
aclimate-MBDpair
Studieswereexcludedifthey:1)werenotpeer-reviewedjournalar-
ticlesorwerenon-empiricalsuchasarevieworcommentarypiece;2)
were not written in English; 3) reported an experimental or quasi-
experimentalstudy,4)targetednon-humanbeingsasresearchsubjects
(e.g.,animalstudies);5)comprisedacasestudy/casereportofasingle
subject;6)didnothavearelevantexposureoroutcomevariable;7)ex-
aminedclimateimpactsunderaspecificprogram,training,oroccupa-
tionalenvironment(e.g.,militarytraining);7)investigatedartificially
designed/controlledambientconditions(e.g.,hospitalsorclassrooms
withdifferent levelsoflighting);8)took theseasonorself-reportsof
perceivedclimateasexposure without objective measuresofclimate
factors;9)reportedanon-time-seriesstudywithmeteorologicalfactors
compared at a spatial scale that was too coarse (cross-continent or
cross-country);10)reportedall-causemortalityormorbidityduetocli-
mateexposurebutnotmentalhealth-relatedoutcomes;11)werequali-
tativeor did notconductanyestimateoftheexposure-outcomerela-
tionship; and 12) examined reproductive health or linked-life pairs
(e.g.,amother-childdyad).
The search strategy and selection procedure guided by PRISMA
(Moheretal.,2009)ispresentedinFig.1.
2.3. Datacollectionandnarrativesynthesis
Informationextractionwasconductedbythreeresearchersindepen-
dentlyandthencross-checked.Adescriptive informationspreadsheet
andameta-analysissheetweredevelopedinMicrosoftExceltoextract
and tabulate information from the included studies. The following
3

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Fig.1. PRIMAflowchartofthestudyselectionprocess.
study characteristics were extracted in the descriptive information
sheet:authors, citationdetails, country,population, KöppenClimate
Zoneofstudyarea,studytype,samplesize,spatialresolution,temporal
resolution,climate/meteorologicalmeasures,mentalhealthmeasures,
statisticalanalysis,effectsize,lagperiod,andmoderator/mediator.The
climatic/meteorological measures category included check boxes for
thermalcomfortindex,airtemperature,winddirection/speed,solarra-
diation,relativehumidity, barometric pressure,precipitation, among
others.Wealsotooknotesofwhetheraworkingdefinitionofaclimate/
weatherdisasterwasused,detaileddescriptionsoftheexposuremea-
sures,anddatasources.TheMBDmeasuresincludedtypeofdata(e.g.,
mortality/morbidityrecord,medicationdispensation,self/caregiverre-
port),detaileddescriptionsofoutcome measures/scales, check boxes
foreachMBDtype,timepointsofmeasurement,anddatasources.The
check boxes included categories corresponding to ICD-10 F01
–
F09,
F10
–
F19,F20
–
F29,F30
–
F39,F40
–
F48,F50
–
F59,F60
–
F69,F70
–
F79,
F80
–
F89,F90
–
F98,andF99andequivalentICD-9andDSMcategories.
WerecodedandtabulatedalldatacollectedusingMicrosoftExcel
PivotTableandRpackages.Descriptiveplotswereproducedfordata
synthesis.Amapshowinggeolocationandclimatezonewasproduced
inArcGISPro.Studiesweregroupedbasedonthespatiotemporalreso-
lutions,population characteristics, and exposure andoutcomes mea-
suresandchartssuchascircularbarchartwereproducedusingRpack-
ages.(SeeFig.2.)
4

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
2.4. Riskofbiasassessment
Astheincludedstudiesrangedintypefromecologicaltime-seriesto
individual-level cross-sectional, case-control, and cohort studies, we
developedariskofbiasassessment(RoB)rubricbyintegratingrelevant
itemsfromassessmenttoolscommonlyusedforassessingstudiesonen-
vironmentalexposureandhealthoutcomeatindividualoraggregated
populationlevels.ThesetoolsincludedtheNationalHeart,Lung,and
BloodInstitutesStudyQualityAssessmentTools(NationalHeart,Lung,
andBloodInstitute,2014a;NationalHeart,Lung,andBloodInstitute,
2014b),theJBI'sCriticalAppraisalTools(Moolaetal.,2017),andthe
Riskofabiasassessmentinstrumentforsystematicreviewsinforming
WorldHealthOrganizationglobalairqualityguidelines(WorldHealth
Organization,2020).Specifically,sixdomainswithtwelveitemswere
considered: selection bias, exposure assessment, outcome measure-
ment, confounding, missing data/attrition bias, and reporting bias.
Eachdomainincludedonetothreesub-items,whichwereratedona
four-pointscale:lowrisk,probablylowrisk,probablyhigh risk, and
highrisk. For example, for confounding, we assessed whether time-
invariantfactors(e.g.,age,sex, socioeconomic conditions) andtime-
variant factors (e.g., air pollution, noise, seasonality) potentially re-
lated to the outcomes were considered, measured using valid ap-
proaches,andstatisticallyadjustedforinthemodels.Studiesthatdid
notconsiderrelevantcontrolvariables,didnot adjust for these vari-
ablesinthe statistical modelsestimating the exposure-outcomerela-
tionshipswereratedasprobablyhigh/highinriskofbiasfortherespec-
tiverubricitemsunderconfounding.Forselectionbias,weconsidered
whetherstudy population wasclearlydefinedandrecruitment/inclu-
sioncriteriaconsistentandspecified.Individual-basedstudiesthatused
convenientsamplesoronlinepanels,andecologicalstudiesthatused
datafromasinglehospitaloragencywithoutdescribingthepopulation
servedandcoveragewereratedprobablyhigh/highinoutcomemea-
sure.Fortheoutcomemeasurement,weassessedwhetherthehealth
outcomeassessorswereblindedtotheexposurestatusofparticipants
andwhetherthemeasureswereclearlydefined,valid,reliable,andim-
plementedconsistentlyacrossstudypopulations.Studiesthatusedun-
validatedscales ofoutcomevariableswereconsideredhigh inriskof
bias.Studiesthat only reportedsignificantresults and omittedother
findingsfromplannedanalysiswereevaluatedasprobablyhighinre-
portingbias.Thedetailsoftheassessmentinstrumentarepresentedin
SupplementarymaterialS3.Thegraphicsweregeneratedusingaweb-
basedvisualizationtoolnamedRoB2(Sterneetal.,2019).
Eacharticlewas independently assessedbytwo researchers; con-
flicts were resolved by discussion and then consulting a third re-
searcher.Theitemscoreswerethenaveragedandroundedtogenerate
domainscores.Wedidnotassignoverallriskofbiasscoresduetothe
arbitrarinessofassigningweightstodifferentdomainsofrisk.Instead,
wepresentthe by-item and-domainRoB scores, whichretaintrans-
parency and better reflect the methodological strengths and weak-
nessesofeacharticle.Accordingly,wealsodidnotexcludeanystudies
basedonRoBresultsbutinsteaddiscussedbiasesandprovidedrecom-
mendationsforfuturestudies.
2.5. Meta-analysis
Meta-analysiswasperformedtoestimatethepooledeffectsizesof
theassociationsbetweenclimateexposuresandmentalhealth.Forthe
meta-analysisdataextractionsheet,wecollectedinformationoneach
5

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
exposure-outcomepairincludingexposurevariable,outcomevariable,
statisticalmodel,typeofeffectsizestatistic,effectsizeestimate,stan-
darderror,95%confidence interval lower bound andupperbound,
anddetailedstepsofdataprocessing.Thenaturallogarithmsoftherel-
ativerisk (RR) estimatesandthecorrespondingstandarderrorswere
calculatedandusedinthemeta-analysis.
Whenmultipleestimatesonthesameexposure-outcomepairwere
reportedinonearticle,weextractedonlyoneeffectsizefollowingthe
approaches adopted in previous environmental epidemiology-related
meta-analyses(ChenandHoek,2020;Khreisetal.,2017).Studiesre-
portingoddsratio(OR)wereconvertedintorelativerisk(RR)usingEq.
1whendatawereavailable,andbyassumingequalitywhenrisksinthe
controlgroupwereextremelylow(Harreretal.,2021).Whenanarticle
reportedtwoormoreestimatesforsubgroups(e.g.,byregionorpopu-
lationgroup),wecalculatedthepooledeffectsizeusingafixed-effects
meta-analysis model. When multiple outcomes that belonged to the
sameICD-10subclassweretestedseparately,weusedEq.2tocalculate
thecombinedeffectsize(Harreretal.,2021).Whenmultipleestimates
existedduetomodelspecification,function,estimator,orlagduration,
wefollowedthedecisionruleinsequentialordertoselectthemodelre-
sults:1)selecttheauthors'favoredmodel(mainmodelormodelfollow-
ing main conceptual framework/highlighted in the Abstract, High-
lights,Results,orMainfindingssectionoftheDiscussion),2)selectthe
fullyadjustedmodelratherthanthecrudemodel,andfinally3)select
themodelwith the smallestlagvalue. Pooled effectswerenot com-
putedforself-reportedhealth or psychologicalconditionsbecause of
thesmallnumberofstudiesandthehighheterogeneityofpsychometric
scalesused.
(1)
wherecrepresentstheeventsinthecontrolgroup, isthetotal
numberof participantsinthecontrolgroup, andORistheestimated
oddsratio,RRistherelativeriskcalculated.
(2)
where and arethetwooutcomesand and arethevari-
ance, isthemergedeffectsize.
Heterogeneityofeffectestimatesacrossstudieswasassessedusing
,Cochran'sQ,and statistics. isthevarianceofthedistributionof
thetrueeffectsizeacrossstudiesandiseasytointerpretbasedonthe
originalmeasurementmetric,but andCochran'sQbothheavilyre-
flectstatisticalpowerwhile isnotassensitivetothenumberofstud-
ies;assuch,presentingbothQand helpsindepictingacompletepic-
tureofbetween-studyheterogeneity.Theruleofthumbemployedfor
interpreting wasasfollows:upto25%aslowheterogeneity,50%
for moderate heterogeneity, and 75 % for substantial heterogeneity
(HigginsandThompson,2002).Duetothesmallnumberofstudieson
eachexposure-outcomepairandhighoverallheterogeneity,wedidnot
removeoutliersinthemeta-analysisorperformsensitivityanalysesbut
flaggedontheforestplotthosestudiesreportingeitherextremelysmall
effects(whentheupperboundofthe95%CIislowerthanthelower
boundofthepooledeffect)orextremelylargeeffects(whenthelower
boundoftheCIishigherthantheupperboundofthepooledeffect).
Subgroupanalysisbasedongeographyorpopulationwasnotpossible
duetosmallnumberofexposure-outcomepairs.TheRpackagemetafor
(Viechtbauer, 2010) was used to perform the random-effects meta-
analysismodeling,identifyoutliers,andproduceforestplots.
2.6. Evaluationofcertaintyofevidence
WeutilizedtheGradingofRecommendationsAssessment,Develop-
mentandEvaluation(GRADE)frameworktoassesstheoverallquality
ofevidence (Guyattetal.,2008;Balshemet al., 2011).Abaselineof
moderatecertaintywasinitiallyapplied,andsubsequentlydowngraded
orupgraded based oneach of theGRADE domains. Factorsthat de-
creasedthecertaintyofevidenceincludedD1)Riskofbiasacrossstud-
ies,D2)Indirectness,D3)Inconsistency,D4)Imprecision,andD5)Pub-
licationbias.FactorsthatincreasedcertaintyincludedU1)Largemag-
nitudeofeffects,U2)Consistentdose-responsegradient,andU3)Con-
foundingminimizeseffect.Thedetailedcriteria fordowngradingand
upgradingareasfollows.
Downgrading.D1)The certainty of evidencewas downgraded by
onelevelifatleastonestudythathadanon-negligibleweightinthe
pooledeffectsizeestimateshowedatleastthreehighorprobablyhigh
biasratingsintheRoBevaluations.Inourcase,becausethenumberof
studies in each exposure-outcome pair was small, downgrading was
performedwhenany study exhibitedatleast three highor probably
highbias ratings. D2) The certainty of evidence was downgraded if
studiesdidnot adhere tothe population, exposure,comparator, and
outcomespecifiedfortheresearchquestion,forexamplenotmeasuring
outcomesdirectlybutusingsurrogateoutcomemeasures.Here,thisap-
pliedwhenastudydefinedMBDcasesbasednotonclinicaldiagnosis,
butonoutcomesfromapreliminaryscreeningtoolorevaluationsbyre-
searcher(s)whosecredentialswerenotreported.D3)Thecertaintyof
evidencewasdowngradedwhenverylargeheterogeneityorvariability
in results was detected. Specifically, as observational studies are
demonstratedtohavemoderatetolargeheterogeneityacrossgeogra-
phies and populations (Chen and Hoek, 2020; Schwingshackl et al.,
2021),wedowngradedbyoneortwolevelsifthepredictioninterval
wasmorethantwicetheconfidenceintervalandstudiespresentedthe
associationasbidirectional(bothpositiveandnegative).D4)Thecer-
taintyofevidencewasdowngradedifthenumberofparticipantsand
person/population-time were small (n < 500, determined based on
previousstudiesdiscussingsamplesizeinepidemiology(Vergouweet
al.,2005; Rigby and Vail,1998)).D5)Thecertaintyofevidencewas
downgradedifpublicationbiaswasidentified basedonvisualassess-
mentoffunnelplots.However,funnelplotsandEgger'stesthadlimited
valueduetothefactthatthenumberofstudieswassmallerthanthe
empiricalthresholdof10.
Upgrading. U1) The certainty of evidence was upgraded by one
level if the pooled effect size was large. Defining the threshold for
“
large
”
 was challenging, especially given that RRs were mostly low
comparedtothethresholdvaluesused in other medical studies.U2)
Thecertainty of evidence was upgraded by one level if studies pre-
sentedabiologicaldose-responsegradient(e.g.,higherRRassociated
withhigherexposure). U3). The certaintyofevidence was upgraded
whenpossibleresidualconfounderswouldreducethedemonstratedef-
fect.
3. Results
3.1. Literaturesearchandselectionresults
Theinitial search yielded7696 articles, to which59 were added
fromforward/backwardsearchesandotherliterature.Afterduplicate
removal,abstract/titlescreening,andfull-textreview,thefinalreview
setcomprised88articlespublished1972
–
2022(Table1).Apost-2016
surgeinpublicationswasevident.Thevastmajorityofincludedstudies
(n=84,95.5%)examinedallagesoradults(e.g.,18+);onlythree
(3.4%)focusedonolderadults(e.g.,50+),andonlyone(1.1%)on
children.
6

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Table1
Studycharacteristics(N=88).
Citation Population Sample
size
Study
periods
Typeof
study
Spatialunitof
analysis
Temporal
unitof
analysis
Microclimate
exposure
Typeof
mental
health
outcome
Mentalhealth
outcome
Statistical
analysis
Included
inmeta-
analysis
Goldstein,
1972
(Goldstein,
1972)
Collegestudents:
Acohortof
studentsin
anintroductory
psychologycourse
atacommunity
collegeinU.S.
22
persons
11days Cohort City/town
level
Daily Air
temperature,
Humidity,
Barometric
pressure
Self-reported
affect/
mental
health
F30
–
F39 Correlation No
Persinger,
1975
(Persinger,
1975)
Collegestudents:
Acohortof
university
studentsfroma
classatLaurentian
University,
Canada
10
persons
90days(9
January
–
8
April1974)
Cohort City/town
level
Daily Air
temperature,
Windspeed,
Sunshine,
Relative
humidity,
Barometric
pressure,
Self-reported
affect/
mental
health
F30
–
F39 Correlation,
(Generalized)
linearmodel
No
Howarth&
Hoffman,
1984
(Howarth
and
Hoffman,
1984)
Collegestudents:
Acohortof
university
students
participatingin
thestudyfor
course
requirementin
Canada
24
persons
11days Cohort City/town
level
Daily Wind
direction/
speed,
Sunshine,
Humidity,
Barometric
pressure,
Precipitation
Self-reported
affect/
mental
health
F30
–
F39 Correlation,
(Generalized)
linearmodel
No
Mawsonand
Smith,1981
(Mawson
andSmith,
1981)
General
population:
Mentaldiseases-
relatedhospital
admissionswithin
theareaofthe
GreaterLondon
Councilrecorded
bytheStatistics
andResearch
Departmentofthe
DHSS
2126
records
1975 Timeseries City/town
level
Daily Barometric
pressure,
Relative
humidity
Medical
records/
clinical
diagnosis
F30
–
F39 Correlation No
Barnston,
1988
(Barnston,
1988)
Collegestudents:
Undergraduate
psychologyclasses
attheUniversity
ofIllinoisat
Urbana,U.S.
62
persons
September
16
–
October
27,1974
Cohort City/town
level
Daily Air
temperature,
Solar
radiation/
Sunshine,
Barometric
pressure,
Precipitation,
Relative
humidity,
Wind
direction/
speed
Self-reported
affect/
mental
health
General
psychological
health
Correlation No
Carneyetal.,
1988
(Carneyet
al.,1988)
General
population:
Mentaldisorders-
relatedhospital
admissions
University
Departmentof
Psychiatryatthe
RegionalHospital,
Galway.
104
records
1980
–
1984 Timeseries Neighborhood/
Hospital
catchment
level
Monthly Air
temperature,
Solar
radiation/
Sunshine
Medical
records/
clinical
diagnosis
F30
–
F39 Correlation No
Molinetal.,
1996(Molin
etal.,1996)
Acohortof
patientsdiagnosed
withdepression
126
persons
1991
–
1994 Cohort City/town
level
Weekly Air
temperature,
Solar
radiation/
Sunshine,
Barometric
pressure,
Precipitation
Symptom
worsening/
Newepisode
(Generalized)
linearmodel
No
(continuedonnextpage)
7

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Table1(continued)
Citation Population Sample
size
Study
periods
Typeof
study
Spatialunitof
analysis
Temporal
unitof
analysis
Microclimate
exposure
Typeof
mental
health
outcome
Mentalhealth
outcome
Statistical
analysis
Included
inmeta-
analysis
Salibetal.,
1999(Salib
andSharp,
1999)
General
population:
Admissiontoa
largepsychiatric
hospital(Winwick
Hospital)inNorth
Cheshire
2070
records
1993 Timeseries Countylevel Daily Air
temperature,
Solar
radiation/
Sunshine,
Relative
humidity,
Precipitation
Medical
records/
clinical
diagnosis
F00
–
F09 Correlation No
Leeetal.,
2002(Leeet
al.,2002)
General
population:
Admissiontothe
psychiatricunitof
thetwohospitals
affiliatedwiththe
KoreaUniversity
MedicalCenter
152
persons
1996
–
1999 Cohort Neighborhood/
Hospital
catchment
level
Monthly Air
temperature,
Solar
radiation/
Sunshine,
Barometric
pressure,
Precipitation,
Relative
humidity
Medical
records/
clinical
diagnosis
F30
–
F39 Correlation No
Saliband
Sharp,2002
(Saliband
Sharp,2002)
General
population:
Admissiontoa
largepsychiatric
hospital(Winwick
Hospital)inNorth
Cheshire
1084
records
1993 Timeseries City/town
level
Daily Air
temperature,
Solar
radiation/
Sunshine,
Relative
humidity,
Precipitation
Medical
records/
clinical
diagnosis
F10
–
19,F20
–
F29,F30
–
F39,F40
–
F49
Correlation No
Cornalietal.,
2004
(Cornaliet
al.,2004)
Dementia
patients:
Patientsdiagnosed
withdementia
admittedtothe
Alzheimer
Rehabilitation
Unit,Richiedei
MedicalCenter,
Palazzolos/O
–
Brescia,Italy.
25
persons
June14
–
21,
2002
Case-control Neighborhood/
Hospital
catchment
level
Daily Air
temperature
Symptom
worsening/
Newepisode
&Drug
dispensation
Dementia (Generalized)
linearmodel
No
Bulbenaetal.,
2005
(Bulbenaet
al.,2005)
General
population:
Psychiatric
emergenciesatdel
MarHospital,
Barcelona
368
records
2002 Other
ecological
City/town
level
Daily Air
temperature,
Solar
radiation/
Sunshine,
Barometric
pressure,
Precipitation,
Relative
humidity,
Wind
direction/
speed
Medical
records/
clinical
diagnosis
F40
–
F49 (Generalized)
linearmodel
No
Kelleretal.,
2005(Keller
etal.,2005)
General
population:
Acohortof
populationliving
inAnnArbor,
Michigan,U.S.
605
persons
April5
–
June
15,2001,
April16
–
July27,
2003,
January
–
December
2002
Cross-
sectional
City/town
level
Daily Air
temperature,
Barometric
pressure
Self-reported
affect/
mental
health
F30
–
F39 (Generalized)
linearmodel
No
(continuedonnextpage)
8

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Table1(continued)
Citation Population Sample
size
Study
periods
Typeof
study
Spatialunitof
analysis
Temporal
unitof
analysis
Microclimate
exposure
Typeof
mental
health
outcome
Mentalhealth
outcome
Statistical
analysis
Included
inmeta-
analysis
Shilohetal.,
2005(Shiloh
etal.,2005)
General
populationaged
18+:
Hospital
admissionsand
psychiatric
diagnoses
recordedbythe
IsraeliNational
Psychiatric
Registry(INPR),
Departmentof
Informationand
Evaluation,
MentalHealth
Services(Ministry
ofHealth,
Jerusalem,Israel)
33,614
records
1981
–
1991 Timeseries City/town
level
Daily Thermal
comfort
index,Air
temperature,
Solar
radiation/
Sunshine,
Barometric
pressure,
Relative
humidity
Medical
records/
clinical
diagnosis
F20
–
F29 Correlation No
Hartigetal.,
2007(Hartig
etal.,2007)
General
population:
Dispensationof
selectiveserotonin
reuptake
inhibitors(SSRIs)
recordedbythe
ApoteketABin
Sweden
NR 1991
–
1998 Timeseries Nationallevel Monthly Air
temperature
Drug
dispensation
Depression Auto
regressive
integrated
moving
average
No
Leeetal.,
2007(Leeet
al.,2007)
General
population:
Hospitalization
recordinthe
TaiwanNational
HealthInsurance
ResearchDatabase
15,060
records
1999
–
2003 Timeseries Nationallevel Monthly Air
temperature,
Solar
radiation/
Sunshine,
Precipitation
Medical
records/
clinical
diagnosis
F30
–
F39,F99 Auto
regressive
integrated
moving
average
No
Christensenet
al.,2008
(Christensen
etal.,2008)
Bipolarpatients:
Bipolarpatients
admittedtothe
departmentsof
psychiatryin
threeCopenhagen
University
Hospitals,aged18
and75yearsold
56
persons
1990
–
1993 Cohort Neighborhood/
Hospital
catchment
level
Multi-
month
Air
temperature,
Solar
radiation/
Sunshine,
Barometric
pressure,
Precipitation,
Relative
humidity,
Wind
direction/
speed
Medical
records/
clinical
diagnosis
F30
–
F39 Correlation No
Denissenetal.,
2008
(Denissenet
al.,2008)
General
population:
Acohortof
populationsigning
uptoanonline
diarystudyin
Germany
1233
persons
July2005
–
February
2007
Cohort Nationallevel Daily Air
temperature,
Windspeed,
Sunlight,
Barometric
pressure,
Precipitation
Self-reported
affect/
mental
health
F30
–
F39 (Generalized)
linearmodel
No
Hansenetal.,
2008
(Hansenet
al.,2008)
General
population:
Hospitalizations
formentaland
behavioral
disordersordeath
recordsassociated
withmentaland
behavioral
disordersin
Adelaide,South
Australiacollected
bytheSouth
Australian
171,
614
records
1993
–
2006 Timeseries City/town
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F99 (Generalized)
linearmodel
Yes
(continuedonnextpage)
9

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Table1(continued)
Citation Population Sample
size
Study
periods
Typeof
study
Spatialunitof
analysis
Temporal
unitof
analysis
Microclimate
exposure
Typeof
mental
health
outcome
Mentalhealth
outcome
Statistical
analysis
Included
inmeta-
analysis
Bulbenaetal.,
2009
(Bulbenaet
al.,2009)
General
population:
Psychiatric
emergenciesat
HospitaldelMar
hospitaland
InstitutMunicipal
Psiquiatria
hospital,
Barcelona
872
records
2003
summer
days
Timeseries Neighborhood/
Hospital
catchment
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F10
–
F19,
F20
–
F29,
F30
–
F39,
F40
–
F49,
F60
–
F69
(Generalized)
linearmodel
Yes
Huibersetal.,
2010
(Huiberset
al.,2010)
General
population(aged
18
–
65):
Reportedpresence
ofmental
disordersin
southern
Netherlands
14,478
persons
2005
–
2007 Timeseries Nationallevel Daily Air
temperature,
Solar
radiation/
Sunshine,
Precipitation
Medical
records/
clinical
diagnosis
F30
–
F39 (Generalized)
linearmodel
Yes
Khalajetal.,
2010(Khalaj
etal.,2010)
General
population:
Hospital
admissionsinfive
regions(Sydney
East;Sydney
West;Gosford
Yong;Newcastle
andIllawarra)of
NewSouthWales
1,497,
655
records
springand
summer
days
(September
–
February)of
1998
–
2006
Timeseries Regionallevel Daily Thermal
comfort
index,Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F99 (Generalized)
linearmodel
Yes
Raduaetal.,
2010(Radua
etal.,2010)
General
population:
Admissionstothe
AcuteUnitatthe
Departmentof
Psychiatryin
Bellvitge
University
Hospital,
Barcelona,Spain
421
persons
1997
–
2004 Timeseries City/town
level
Daily Thermal
comfortindex
Medical
records/
clinical
diagnosis
F30
–
F39 Autoregressive
Integrated
Moving
Average
No
Klimstraetal.,
2011
(Klimstraet
al.,2011)
Adolescentsand
theirmothers:
Acohortof
adolescentsand
theirmothers
enrolledinan
ongoing
longitudinal
projectinThe
Netherlands,
entitledResearch
onAdolescent
Developmentand
Relationships
(RADAR).
823
persons
6weeks Cohort Nationallevel Daily Air
temperature,
Sunlight,
Precipitation
Self-reported
affect/
mental
health
F30
–
F39 Correlation No
Sungetal.,
2011(Sung
etal.,2011)
General
population:
Medicalrecordsof
psychiatric
hospital
admissionsin
Psychiatric
InpatientMedical
Claim(PIMC)
datasetofthe
NationalHealth
Insurance
ResearchDatabase
inTaiwan
41,023
records
1996
–
2007 Timeseries Nationallevel Daily Air
temperature
Medical
records/
clinical
diagnosis
F20
–
F29 (Generalized)
linearmodel
Yes
(continuedonnextpage)
10

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Table1(continued)
Citation Population Sample
size
Study
periods
Typeof
study
Spatialunitof
analysis
Temporal
unitof
analysis
Microclimate
exposure
Typeof
mental
health
outcome
Mentalhealth
outcome
Statistical
analysis
Included
inmeta-
analysis
Yackersonet
al.,2011
(Yackerson
etal.,2011)
General
population:
Mentaldiseases-
relatedemergency
visitsand
hospitalizationin
theregional
MentalHealth
Center(MHC)of
Ben-Gurion
University,Israel
4325
persons
2001
–
2003 Timeseries Neighborhood/
Hospital
catchment
level
Weekly Air
temperature,
Windspeed/
direction,
Relative
humidity
Medical
records/
clinical
diagnosis
F20
–
F29 Correlation No
Gasparriniet
al.,2012
(Gasparrini
etal.,2012)
General
population:
Deathrecords
associatedwith
heatrecordedby
theOfficefor
NationalStatistics
inEnglandand
Wales
92,439
records
summers
(June
–
September)
of1993
–
2006
Timeseries Regionallevel Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F09,
F10
–
F19,
F20
–
F29
(Generalized)
linearmodel
Yes
Vidaetal.,
2012(Vida
etal.,2012)
General
population:
Emergency
departmentvisits
formental
disordersinthree
geographicareas
ofQuébec,aged
15yearsandover
347,
552
records
1995
–
2007 Other
ecological
Climatezone Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F09,
F10
–
F19,
F20
–
F29,
F30
–
F39
(Generalized)
linearmodel
Yes
Williamsetal.,
2012
(Williamset
al.,2012)
General
population:
Mortality,hospital
admissions,and
emergency
departmentvisits
collectedbythe
SouthAustralian
Departmentof
Health
NR 1993
–
2009 Timeseries City/town
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F99 GEE Yes
Alexander,
2013
(Alexander,
2013)
General
population:
Psychiatric
disease-related
callstothepublic
emergencyservice
ofthecityof
BuenosAires,
Argentina
80,724
records
1999
–
2004 Timeseries City/town
level
Monthly Thermal
comfort
index,Air
temperature,
Precipitation,
Relative
humidity
Medical
records/
clinical
diagnosis
General
mentalhealth
Correlation No
McWillimaset
al.,2013
(McWilliams
etal.,2013)
General
population:
Admissionsto
psychiatric
hospitalsinthe
Republicof
Ireland
48,347
records
1971
–
2002 Timeseries Regionallevel Daily Air
temperature,
Solar
radiation/
Sunshine,
Barometric
pressure,
Precipitation,
Wind
direction/
speed
Medical
records/
clinical
diagnosis
F20
–
F29 Autoregressive
Integrated
Moving
Average;
(Generalized)
linearmodel
No
(continuedonnextpage)
11

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Table1(continued)
Citation Population Sample
size
Study
periods
Typeof
study
Spatialunitof
analysis
Temporal
unitof
analysis
Microclimate
exposure
Typeof
mental
health
outcome
Mentalhealth
outcome
Statistical
analysis
Included
inmeta-
analysis
Sungetal.,
2013(Sung
etal.,2013)
General
population:
Medicalrecordsof
psychiatric
hospital
admissionsin
Psychiatric
InpatientMedical
Claim(PIMC)
datasetofthe
NationalHealth
Insurance
ResearchDatabase
inTaiwan
9071
records
1996
–
2007 Timeseries Nationallevel Daily Air
temperature,
Precipitation
Medical
records/
clinical
diagnosis
F30
–
F39 (Generalized)
linearmodel
Yes
Tsutsui,2013
(Tsutsui,
2013)
Collegestudents:
Acohortof
university
studentsrecruited
oncampus
groundsaswellas
throughawebsite
atOsaka
University,Japan
75
persons
516days(1
November
2006
–
31
March2008)
Cohort City/town
level
Daily Air
temperature,
Windspeed,
Sunshine,
Humidity,
Precipitation
Self-reported
affect/
mental
health
General
psychological
health,F30
–
F39,F40
–
F49,F50
–
F59
Correlation,
(Generalized)
linearmodel
No
Vaneckovaet
al.,2013
(Vaneckova
and
Bambrick,
2013)
General
population:
Admissionstoall
privateandpublic
hospitalslocated
intheSydney
930,
322
records
1991
–
2009 Case-
crossover
City/town
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F20
–
F29,
F40
–
F49,
F70
–
F79
(Generalized)
linearmodel
Yes
Wilsonetal.,
2013
(Wilsonet
al.,2013)
General
population:
Medicalrecordsof
hospital
admissionsor
deathrecordsin
theNewSouth
WalesDepartment
ofHealth
NR 1997
–
2007
and1997
–
2010
Case-
crossover
City/town
level
Daily Thermal
comfort
index,Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F99 DLNM Yes
Henriquez-
Sanchezet
al.,2014
(Henriquez-
Sanchezet
al.,2014)
CollegeStudents:
Acohortof
Spanishuniversity
graduates
participatingin
theSUNproject
fromSeguimiento
Universityof
Navarra
13,938
persons
1999
–
2009 Cohort Regionallevel Multi-
year
Air
temperature,
Solar
radiation/
Sunshine,
Precipitation
Medical
records/
clinical
diagnosis
F30
–
F39 Cox
(Generalized)
linearmodel
Yes
McWillimaset
al.,2014
(McWilliams
etal.,2014)
General
population:
Admissionsto
psychiatric
hospitalsinthe
Republicof
Ireland
34,465
records
1971
–
2002 Timeseries Regionallevel Daily Air
temperature,
Solar
radiation/
Sunshine,
Barometric
pressure,
Precipitation,
Wind
direction/
speed
Medical
records/
clinical
diagnosis
F30
–
F39 Autoregressive
Integrated
Moving
Average;
(Generalized)
linearmodel
No
Obrienetal.,
2014
(Obrienet
al.,2014)
Populationaged
15+:
Acohortof
populationfrom
theHousehold,
Incomeand
LabourDynamics
inAustralia
(HILDA)Survey
5012
persons
2007
–
2008 Cross-
sectional
(onewave
ofa
longitudinal
dataset)
Nationallevel Monthly Precipitation Self-reported
affect/
mental
health
General
psychological
health
(Generalized)
linearmodel
No
(continuedonnextpage)
12

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Table1(continued)
Citation Population Sample
size
Study
periods
Typeof
study
Spatialunitof
analysis
Temporal
unitof
analysis
Microclimate
exposure
Typeof
mental
health
outcome
Mentalhealth
outcome
Statistical
analysis
Included
inmeta-
analysis
Simonsetal.,
2014
(Simonset
al.,2014)
General
population:
Acohortof
population
admittedto3
medicalcenters
(Radboud
University
NijmegenMedical
Centrein
Nijmegen,
University
MedicalCentrein
Utrecht,and
JeroenBosch
Hospital,'s-
Hertogenbosch)in
TheNetherlands
3198
persons
2008
–
2012 Cohort Neighborhood/
Hospital
catchment
level
Monthly Solar
radiation/
Sunshine
Medical
records/
clinical
diagnosis
F00
–
F09 (Generalized)
linearmodel
Yes
Wangetal.,
2014(Wang
etal.,2014)
General
population:
Emergencyroom
visitsformental
illnessinNational
AmbulatoryCare
ReportingSystem
inTorontothat
capturedover
97%oftheER
visitsinthe
provinceof
Ontarioandhad
goodreabstraction
accuracy
271,
746
records
2002
–
2010 Timeseries City/town
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F99 DLNM Yes
Kimetal.,
2015(Kim
etal.,2015)
General
population:
Deathrecord
providedby
StatisticsKorea
50,055
records
1992
–
2009 Timeseries City/town
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F99,
F10
–
F19,
F20
–
F29
(Generalized)
linearmodel
Yes
Beecheretal.,
2016
(Beecheret
al.,2016)
Mentalhealth
distresspatients:
Acohortof
university
students
participatingin
mentalhealth
treatmentat
BrighamYoung
University,U.S.
16,452
persons
2008
–
2014 Other City/town
level
Daily Sunshine Self-reported
affect/
mental
health
General
mentalhealth
(Generalized)
linearmodel
No
Dingetal.,
2016(Ding
etal.,2016)
Populationaged
45+
Acohortof
residentsaged45
andoverof
thesoutheast
Australianstateof
NewSouthWales,
Australia.
53,144
persons
2006
–
2008 Cohort State/Province
level
Daily Air
temperature,
Relative
humidity
Self-reported
affect/
mental
health
General
psychological
health
(Generalized)
linearmodel
No
Noelkeetal.,
2016
(Noelkeet
al.,2016)
General
populationaged
18+:participants
reportingpresence
ofmental
disordersinthe
GallupG1K
dataset
1,854,
746
persons
2008
–
2013 Other Nationallevel Daily Air
temperature
Self-reported
affect/
mental
health
General
psychological
health
(Generalized)
linearmodel
No
(continuedonnextpage)
13

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Table1(continued)
Citation Population Sample
size
Study
periods
Typeof
study
Spatialunitof
analysis
Temporal
unitof
analysis
Microclimate
exposure
Typeof
mental
health
outcome
Mentalhealth
outcome
Statistical
analysis
Included
inmeta-
analysis
O'Hareetal.,
2016
(O'Hareet
al.,2016)
Populationaged
50+:
Acohortof
populationfrom
thefirstwaveof
TheIrish
Longitudinal
StudyonAgeing
(TILDA)
8027
persons
late2009to
early2011
Cross-
sectional
(onewave
ofa
longitudinal
dataset)
Nationallevel Monthly Air
temperature,
Solar
radiation/
Sunshine,
Precipitation
Medical
records/
clinical
diagnosis
F30
–
F39 (Generalized)
linearmodel
No
Trang,
Rocklöv,
Giang,
Kullgren,et
al.,2016
(Trangetal.,
2016a)
General
population:
Mentaldisorders-
relatedhospital
admissionsina
databasefrom
HanoiMental
Hospital(oneof
mentalhospitals
inHanoiCity)in
northernVietnam
21,443
records
2008
–
2012 Timeseries City/town
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F09,
F70
–
F79
(Generalized)
linearmodel
Yes
Shiue,Perkins,
&Bearman,
2016(Shiue
etal.,2016)
General
population:
Mentalbehavioral
disorders-related
hospital
admissionsin
Germanhospitals
NR 2009
–
2011 Timeseries Regionallevel Daily Thermal
comfortindex
Medical
records/
clinical
diagnosis
F01
–
F09,
D10
–
F19,
F20
–
F29,
F30
–
F39,
F40
–
F49,
F50
–
F51,F98
Polynomial
regression
model
No
Trang,
Rocklöv,
Giang,&
Nilsson,
2016(Trang
etal.,
2016b)
Populationwith
medicalrecordsof
mentaldisorders-
relatedhospital
admissionsina
databasefrom
HanoiMental
Hospital(oneof
themental
hospitalsinHanoi
City)innorthern
Vietnam
23,525
records
2008
–
2012 Timeseries City/town
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F09,
F10
–
19,F20
–
F29,F30
–
F39,F40
–
F49,F50
–
F59,F70
–
F79,F99
(Generalized)
linearmodel
Yes
Yi-Fanetal.,
2016(Yi-
Fanetal.,
2016)
General
population:
Participantsofthe
International
SocialSurvey
Program(ISSP)
reporting
psychological
conditionsfrom
29counties
5420
records
NR Cross-
sectional
Nationallevel Daily Air
temperature,
Relative
humidity,
Wind
direction/
speed
Self-reported
affect/
mental
health
General
psychological
health
(Generalized)
linearmodel
No
Bullocketal.,
2017
(Bullocket
al.,2017)
Bipolarpatients:
Patientsdiagnosed
withbipolar
disorderfroma
publichealth
serviceinregional
Victoria,Australia
11
persons
Anaverage
of130
consecutive
days(range:
14
–
231days)
Case-
crossover
Regionallevel Daily Air
temperature,
Solar
radiation/
Sunshine,
Barometric
pressure,
Relative
humidity
Symptom
worsening/
Newepisode
Mood (Generalized)
linearmodel
No
Linaresetal.,
2017
(Linareset
al.,2017)
General
population:
Mentaldisorders-
relatedemergency
admissionsto
municipal
hospitalsin
Madrid
1175
records
2001
–
2009 Timeseries City/town
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F09 (Generalized)
linearmodel
Yes
(continuedonnextpage)
14

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Table1(continued)
Citation Population Sample
size
Study
periods
Typeof
study
Spatialunitof
analysis
Temporal
unitof
analysis
Microclimate
exposure
Typeof
mental
health
outcome
Mentalhealth
outcome
Statistical
analysis
Included
inmeta-
analysis
Pengetal.,
2017(Peng
etal.,2017)
General
population:
Participatedin
HealthInsurance
Systemof
Shanghaiwith
medicalrecordsof
hospital
admissionsfor
mentaldisorders
inShanghai,
China
93,971
records
2008
–
2015 Timeseries City/town
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F99 DLNM Yes
Sarranetal.,
2017
(Sarranet
al.,2017)
Patientswith
seasonalaffective
disorder
Acohortof
patientsdiagnosed
withseasonal
affectivedisorder
attheUniversity
Centerof
Psychiatryatthe
University
MedicalCenter
Groningen,
Netherlands
291
persons
2003
–
2009 Cohort Neighborhood/
Hospital
catchment
level
Weekly Air
temperature,
Solar
radiation/
Sunshine,
Barometric
pressure,
Relative
humidity
Symptom
worsening/
Newepisode
Depression (Generalized)
linearmodel
No
Basuetal.,
2018(Basu
etal.,2018)
General
population:
Medicalrecordsof
mentaldisorders-
relatedemergency
roomvisitsby
CaliforniaOffice
ofStatewide
HealthPlanning
andDevelopment
219,
942
records
2005
–
2013 Timeseries State/Province
level
Daily Thermal
comfortindex
Medical
records/
clinical
diagnosis
F20
–
F29,
F40
–
F49
(Generalized)
linearmodel
Yes
Chanetal.,
2018(Chan
etal.,2018)
General
population:
Mentaldisorders-
relatedhospital
admissions
collectedbythe
HospitalAuthority
ofHongKong
contained>99%
ofcasesduesto
mentaldisorders
inHongKong
44,600
records
2002
–
2011 Timeseries City/town
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F09,
F10
–
F19,
F20
–
F29,
F30
–
F39,
F40
–
F49,F99
DLNM Yes
M. Leeetal.,
2018(Leeet
al.,2018a)
General
population:
Acorhortof
population
participatingthe
JapaneseHealth
DiaryStudyin
Japan
4548
persons
Octoberof
2013
Cohort Nationallevel Daily Air
temperature,
Relative
humidity
Medical
records/
clinical
diagnosis
F40
–
F49 (Generalized)
linearmodel
No
S. Leeetal.,
2018(Leeet
al.,2018b)
General
population:
Emergency
admissions
collectedbythe
KoreanNational
HealthInsurance
Corporation
containing
medical
informationfor
almost100%of
theKorean
population
166,
578
records
2003
–
2013 Timeseries City/town
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F09,
F20
–
F29,
F30
–
F39,
F40
–
F49
DLNM Yes
(continuedonnextpage)
15

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Table1(continued)
Citation Population Sample
size
Study
periods
Typeof
study
Spatialunitof
analysis
Temporal
unitof
analysis
Microclimate
exposure
Typeof
mental
health
outcome
Mentalhealth
outcome
Statistical
analysis
Included
inmeta-
analysis
Obradovichet
al.,2018
(Obradovich
etal.,
2018b)
General
populationaged
18+:
Participants
completingthe
BehavioralRisk
Factor
Surveillance
System(BRFSS)
undertheCenters
forDisease
Controland
Prevention(CDC)
1,961,
743
persons
2002
–
2012 Other Nationallevel Monthly Air
temperature,
Precipitation
Self-reported
affect/
mental
health
General
mentalhealth
(Generalized)
linearmodel
No
Sherbakovet
al.,2018
(Sherbakov
etal.,2018)
General
population:
Hospitalizationsin
Californiainthe
Officeof
StatewideHealth
Planningand
Development
(OSHPD)Patient
DischargeData
(PDD)
130,
065
records
May
–
Octoberof
1999
–
2009
Timeseries Climatezone Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F99 DLNM Yes
Tapaketal.,
2018(Tapak
etal.,2018)
General
population:
hospitalizationin
Farshchian
hospital(theonly
psychiatric
hospitalacross
Hamadan
Province)in
westernIran
20,406
persons
2005
–
2017 Timeseries Neighborhood/
Hospital
catchment
level
Daily Solar
radiation/
Sunshine,
Barometric
pressure,
Precipitation
Medical
records/
clinical
diagnosis
F20
–
F29,
F30
–
F39
(Generalized)
linearmodel
Yes
Wangetal.,
2018(Wang
etal.,2018)
General
population:
Admissionstothe
AnhuiMental
HealthCenterin
China
17,744
records
warmseason
(May
–
October)in
2005
–
2014
Timeseries City/town
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F20
–
F29 DLNM Yes
Xuetal.,2019
(Xuetal.,
2019)
General
population:
Emergency
departmentvisits
recordedbythe
Queensland
Health
NR 2013
–
2015 Timeseries Neighborhood/
Hospital
catchment
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F99 DLNM Yes
Almendraet
al.,2019
(Almendra
etal.,2019)
General
population:
Mentaldisorders-
relatedhospital
admissionsinthe
DiagnosisRelated
Groupsgeneral
databaseprovided
bythePortuguese
HealthSystem
Central
Administration
30,139
records
2008
–
2014 Timeseries City/town
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F09,
F10
–
F19,
F20
–
F29,
F30
–
F39,
F40
–
F49,
F50
–
F59,
F60
–
F69
DLNM Yes
Guetal.,2019
(Guetal.,
2019)
General
population:
Mentaldisorders-
relatedhospital
admissionatthe
biggestpsychiatric
hospitalin
Ningbo,China
10,132
records
2012
–
2016 Timeseries City/town
level
Daily Solar
radiation/
Sunshine
Medical
records/
clinical
diagnosis
F20
–
F29 DLNM Yes
(continuedonnextpage)
16

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Table1(continued)
Citation Population Sample
size
Study
periods
Typeof
study
Spatialunitof
analysis
Temporal
unitof
analysis
Microclimate
exposure
Typeof
mental
health
outcome
Mentalhealth
outcome
Statistical
analysis
Included
inmeta-
analysis
Hoetal.,2019
(Hoand
Wong,2019)
General
population:
Deathrecordsina
mortalitydataset
providing
informationofall
decedentsinHong
Kong
133,
359
records
2007
–
2014 Timeseries City/town
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F99 (Generalized)
linearmodel
Yes
Liuetal.,
2019(Liuet
al.,2019)
General
population:
Mentaldiseases-
relatedhospital
admissionsatthe
MentalHealth
Centerof
Shandong
provinceinChina
19,569
persons
14days(14
June14
–
17,
June28
–
30,
July4
–
7,
andJuly29
–
31) in2010
Case-
crossover
City/town
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F99 (Generalized)
linearmodel
No
Minetal.,
2019(Min
etal.,2019)
General
population:
Mentaldisorders-
relatedhospital
admissionsinthe
hospitalmedical
recordsystemsof
Yanchengcity,
China
8438
records
2014
–
2017 Timeseries City/town
level
Daily Thermal
comfortindex
Medical
records/
clinical
diagnosis
F00
–
F99 DLNM Yes
Mullins&
White,2019
(Mullinsand
White,
2019)
General
population:
Mentaldisorders-
relatedemergency
departmentvisits
and
hospitalizations
collectedbythe
California'sOffice
ofStatewide
HealthPlanning
andDevelopment
&participants
reportingmental
healthby
interviewsfrom
theBehavioral
RiskFactor
Surveillance
System(BRFSS)
undertheCenters
forDisease
Controland
Prevention(CDC),
aged18yearsand
older
5,996,
037
records
2005
–
2016 Timeseries Countylevel Daily Air
temperature
Medical
records/
clinical
diagnosis&
Self-reported
affect/
mental
health
F00
–
F99 (Generalized)
linearmodel
No
Panetal.,
2019(Panet
al.,2019)
General
population:
Mentaldiseases-
relatedhospital
admissions
collectedbythe
AnhuiMental
HealthCenterin
Hefei,China
30,022
records
2005
–
2014 Timeseries City/town
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F20
–
F29 DLNM No
(continuedonnextpage)
17

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Table1(continued)
Citation Population Sample
size
Study
periods
Typeof
study
Spatialunitof
analysis
Temporal
unitof
analysis
Microclimate
exposure
Typeof
mental
health
outcome
Mentalhealth
outcome
Statistical
analysis
Included
inmeta-
analysis
Xuetal.,2019
(Xuetal.,
2018)
Childrenaged6
–
11:
Acohortof
childrenaged6to
11yearsold,
participatingin
theLongitudinal
Studyof
Australian
Children
6875
records
2008
–
2014 Cohort Nationallevel Yearly Air
temperature,
Precipitation
Self-reported
affect/
mental
health
F00
–
F99 (Generalized)
linearmodel
No
Yietal.,2019
(Yietal.,
2019)
General
population:
Emergency
admissionsinthe
AnhuiMental
HealthCenterin
China
36,607
records
2005
–
2014 Timeseries City/town
level
Daily Thermal
comfortindex
Medical
records/
clinical
diagnosis
F20
–
F29 DLNM Yes
daSilvaetal.,
2020(da
Silvaetal.,
2020)
General
population:
Hospitalizations
formentaland
behavioral
disordersinthe
publicSingle
SystemofHealth
(SUS)inCuritiba,
Brazil
5397
records
2010
–
2016 Timeseries City/town
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F99 DLNM Yes
Lietal.,2020
(Lietal.,
2020)
General
populationaged
18+:
Participantsofthe
BehavioralRisk
Factor
Surveillance
System(BRFSS)
undertheCenters
forDisease
Controland
Prevention(CDC)
3,060,
158
persons
1993
–
2010 Other
ecological
Countylevel Daily Air
temperature
Self-reported
affect/
mental
health
General
mentalhealth
(Generalized)
linearmodel
No
Liuetal.,
2020(Liuet
al.,2020)
General
population:
Deathrecords
collectedbythe
HongKongCensus
andStatistics
Department
19,534
records
2006
–
2016 Timeseries Nationallevel Daily Air
temperature,
Relative
humidity
Medical
records/
clinical
diagnosis
F00
–
F99 DLNM Yes
Niuetal.,
2020(Niuet
al.,2020)
General
population:
Mentaldisorders-
relatedemergency
admissionsin30
hospitalsin
Beijingrecorded
byBeijing
MunicipalHealth
Commission
Information
Centerthat
coveredall
admissions
16,606
records
2016
–
2018 Timeseries City/town
level
Daily Thermal
comfortindex
Medical
records/
clinical
diagnosis
F10
–
F19,
F20
–
F29,
F30
–
F39,
F40
–
F49
DLNM Yes
(continuedonnextpage)
18

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Table1(continued)
Citation Population Sample
size
Study
periods
Typeof
study
Spatialunitof
analysis
Temporal
unitof
analysis
Microclimate
exposure
Typeof
mental
health
outcome
Mentalhealth
outcome
Statistical
analysis
Included
inmeta-
analysis
Ohetal.,2020
(Ohetal.,
2020)
General
population:
Emergency
departmentvisits
forpanicattacks
intheNational
Emergency
Department
Information
System(NEDIS)in
Seoul,South
Korea
1926
persons
2008
–
2014 case-
crossover
City/town
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F40
–
F49 (Generalized)
linearmodel
Yes
Zhangetal.,
2020(Zhang
etal.,2020)
General
population:
Admissionstothe
publiclyfunded
andauthoritative
psychiatric
specialisthospitals
formental
disordersinthree
Chinesecities
(Shenzhen,
Zhaoqing,and
Huizhou)
1,133,
220
records
2013
–
2018 Case-
crossover
City/town
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F09,
F20
–
F29,
F30
–
F39,
F40
–
F49
DLNM Yes
Bundoetal.,
2021(Bundo
etal.,2021)
General
population:
Mentaldisorders-
relatedhospital
admissionsinthe
University
Psychiatric
HospitalinBern,
Switzerland
88,996
records
1973
–
2017 Case-
crossover
Neighborhood/
Hospital
catchment
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F09,
F10
–
F19,
F20
–
F29,
F30
–
F39,
F40
–
F49,
F60
–
F69,
F70
–
F79,
F80
–
F89
Distributedlag
linearmodel
Yes
Burdett,
Davillas,&
Etheridge,
2021
(Burdettet
al.,2021)
General
population:
Acohortof
populationfrom
theUKHousehold
Longitudinal
Study(UKHLS)on
mentalhealth
NR April
–
Julyin
2020
Cohort Nationallevel Daily Air
temperature,
Sunshine,
Precipitation
Self-reported
affect/
mental
health
General
mentalhealth
(Generalized)
linearmodel
No
Eun-hyeetal.,
2021(Eun-
hyeetal.,
2021)
General
population:
Mentaldisorders-
relatedemergency
roomvisitsinthe
Statewide
Planningand
Research
Cooperative
Systemoperated
bytheNewYork
StateDepartment
ofHealth.
92,627
records
2009
–
2015 Timeseries State/Province
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F99 DLNM Yes
Jahan&
Wraith,
2021(Jahan
andWraith,
2021)
General
population:
Mentaldisorders-
relatedhospital
admissionsinthe
Queensland
HospitalAdmitted
PatientData
Collectionthat
collects
informationfrom
allpublicand
privatehospitals
inQueensland,
Australia
132,
088
records
1996
–
2015 Timeseries Regionallevel Daily Air
temperature,
Solar
radiation/
Sunshine,
Barometric
pressure,
Precipitation,
Relative
humidity
Medical
records/
clinical
diagnosis
F20
–
F29 DLNM No
(continuedonnextpage)
19

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Table1(continued)
Citation Population Sample
size
Study
periods
Typeof
study
Spatialunitof
analysis
Temporal
unitof
analysis
Microclimate
exposure
Typeof
mental
health
outcome
Mentalhealth
outcome
Statistical
analysis
Included
inmeta-
analysis
Kimetal.,
2021(Kim
etal.,2021)
General
population:
Acohortthat
underwenthealth
examinations
fromtheKorean
NationalHealth
InsuranceService
25,589
persons
2002
–
2013 Case-control Nationallevel Multi-
day
Air
temperature,
Solar
radiation/
Sunshine,
Barometric
pressure,
Precipitation,
Relative
humidity
Medical
records/
clinical
diagnosis
F30
–
F39 (Generalized)
linearmodel
Yes
Middletonet
al.,2021
(Middleton
etal.,2021)
General
population:
Mentalhealth-
relatedclinic
visitsatfive
communityclinics
inNunatsiavut
228,
104
records
2012
–
2018 Timeseries Neighborhood/
Hospital
catchment
level
Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F99 (Generalized)
linearmodel
No
Sonetal.,
2021(Son
andShin,
2021)
General
population:
Mentaldisorders
fromHealth
InsuranceReview
andAssessment
Service-National
PatientSample
providedbythe
NationalHealth
Insuranceof
Korea
531,
342
records
2016 Timeseries Nationallevel Daily Solar
radiation/
Sunshine
Medical
records/
clinical
diagnosis
F30
–
F39 (Generalized)
linearmodel
Yes
Tangetal.,
2021(Tang
etal.,2021)
General
population:
Medicalrecordsof
hospitalizationin
thelargest
psychiatric
hospital(Anhui
MentalHealth
Center)inAnhui
Province,China
53,288
records
2005
–
2019 Timeseries Neighborhood/
Hospital
catchment
level
Daily Thermal
comfortindex
Medical
records/
clinical
diagnosis
F20
–
F29 (Generalized)
linearmodel
No
Yooetal.,
2021(Yooet
al.,2021)
General
population:
International
SocialSurvey
Progental
disorders-related
emergencyroom
visitsinthe
Statewide
Planningand
Research
Cooperative
Systemoperated
byNewYorkState
Departmentof
Health
2,893,
794
records
2009
–
2016 Timeseries Regionallevel Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F99 DLNM Yes
Zapataetal.,
2021
(Zapata,
2021)
General
population:
Participantsofthe
employment
surveyofEcuador
conductedbythe
NationalInstitute
ofStatisticsand
Census(INEC)
54,541
persons
NR Cross-
sectional
Nationallevel Monthly Thermal
comfort
index,Air
temperature,
Precipitation,
Relative
humidity
Self-reported
affect/
mental
health
General
psychological
health
(Generalized)
linearmodel
No
(continuedonnextpage)
20

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Table1(continued)
Citation Population Sample
size
Study
periods
Typeof
study
Spatialunitof
analysis
Temporal
unitof
analysis
Microclimate
exposure
Typeof
mental
health
outcome
Mentalhealth
outcome
Statistical
analysis
Included
inmeta-
analysis
Gongetal.,
2022(Gong
etal.,2022)
General
population:
Mentaldisorders-
relatedemergency
roomvisitsinthe
NationalHealth
Service(NHS)
DigitalinEngland
NR 1998
–
2009 Timeseries Regionallevel Daily Air
temperature
Medical
records/
clinical
diagnosis
F00
–
F09 Distributedlag
linearmodel
Yes
Note.NR:notreported;record:thenumberofmedicalrecords(e.g.,mentaldisorder-relatedadmissions,hospitalization,oremergencyvisit);person:thenumberof
individuals.GEE:generalizedestimatingequation;DLNM:distributedlagnon-linearmodel.
3.2. Narrativesynthesisofstudycharacteristics
3.2.1. Geographicalandclimatezonedistribution
The included studies represented 23 countries/regions, predomi-
nantly in Asia (n = 29, 33.0 %) and Europe (n = 27, 30.7 %), followed
by North America (n = 17, 19.3 %). Outside of a few Asian countries
and Australia, the rest of the Global South (e.g., Latin America, Africa)
was underrepresented (n = 3, 3.4 %). In terms of climate zone, most
studies focused on temperate and dry climates (n = 74, 84.1 %); only a
few from Ecuador (Zapata, 2021), Brazil (da Silva et al., 2020), Taiwan
(Sung et al., 2011; Sung et al., 2013), northern Australia (Xu et al.,
2019; Jahan and Wraith, 2021) and Vietnam (Trang et al., 2016a) fea-
tured tropical climates, and none polar climates (Fig. 2).
3.2.2. Typesofclimaticandmeteorologicalexposuresreported
Frequently-investigated climatic and meteorological factors in-
cluded air/ambient temperature, relative humidity, precipitation, sun-
light/solar radiation, and barometric pressure (Fig. 3, left). At least two
factors were evaluated in 74 studies (84.1 %), and just one in the re-
mainder (15.9 %). Most studies (n = 74, 84.1 %) included air tempera-
ture and temperature-based metrics (e.g., heatwave occurrence),
chiefly mean daily temperature, mean of daily maximum or minimum
temperature, and daily temperature range. Heat and cold were also
commonly considered, typically defined with either a) an absolute
threshold based on local climate norms (Hansen et al., 2008; Bulbena et
al.,2009;Trangetal.,2016a;Liuetal.,2019)orb)aparticularrelative
percentile(e.g.,95%,99%)ofthetemperaturedistribution(Liuetal.,
2020;Sherbakovetal.,2018;Khalajetal.,2010).Likewise,heatwaves
orextremeheateventsweredefinedasmaximumtemperaturesexceed-
inganabsoluteorrelativethresholdforconsecutivedays.
Inadditiontotemperature,30studies(34.1%)examinedtheeffect
ofsolarradiationintermsofsunlightradiation(MJ/m2),sunshinedu-
ration (h), day length (h), cloud cover (Okta) (National Weather
Service,n.d.),andglobalradiation(MJ/m2);25studies(28.4%)evalu-
atedprecipitation-relatedfactors,includingrainfall,snowfall,andthe
numberofrainydays;23studies(26.1%)examinedhumidity,mea-
suredasrelativehumidityordewpointtemperature;19(21.6%)in-
vestigatedbarometricpressure;andnine(10.2%)studiedtheimpactof
windspeed/velocityanddirection.Several studiesalsoaccountedfor
horizontalvisibility(Yi-Fanetal.,2016;Sarranetal.,2017;Tapaket
al.,2018),mist(Sarranetal.,2017;Tapaketal.,2018),andthenum-
berofdustyandfoggydays(Carneyetal.,1988;Noelkeetal.,2016).
Inmoststudies,differentclimaticfactorswereconsideredassepa-
rateexplanatoryvariables;however, some applied biometeorological
modelstosynthesizehumanthermalcomfortorstressscores(n=13,
14.8%).Ninearticles(10.2%)usedapparenttemperatureastheheat
indicator(Khalajetal.,2010;Wilsonetal.,2013;Basuetal.,2018;Min
et al., 2019; Yi et al., 2019; Niu et al., 2020; Zapata, 2021; Oudin
Åström et al., 2015); this thermal comfort index integrates ambient
temperature,relativehumidity, and wind velocity (Steadman,1984).
21

Fig.2. Geographicandclimatezonedistributionofstudiesincludedinthereview(n=88).
Fig.3. Typesofmeteorologicalexposuresandmentalhealthoutcomes(n=88).

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Othersusedthebodydiscomfortindex(Shilohetal.,2005)orphysio-
logicallyequivalenttemperature(PET)(Shiueetal.,2016).Finally,one
studydevelopedaspecificheatindexbygeneratingacumulativetem-
peraturemeasurereflectiveofhourlyextreme-heatexposurewithina
day(Tangetal.,2021
3.2.3. Typesofmentalhealthoutcomesexamined
Themajorityofstudiesconducted before 1990 usedself-reported
psychologicalconditions,whilemostrecentstudiesutilizedhospitalad-
mission/utilizationrecordsorclinicaldiagnosisasoutcomevariables
[e.g.17,23,65,76](n=65,73.9%).Inaddition,afewstudiesused
drugdispensation (Cornali et al.,2004; Hartig etal., 2007) (n= 2,
2.3%),symptomworseningordevelopmentofanewepisode(Molinet
al.,1996;Cornalietal.,2004;Bullocketal.,2017;Sarranetal.,2017)
(n=4,4.5%).Ofthosedrawinguponhospitalrecords(Fig.3,right),
83(94.3%)examinedmentalandbehavioralmorbidity(Obradovichet
al.,2018b;Shilohetal.,2005;Carneyetal.,1988;Obrienetal.,2014)
andsix(6.8%)usedmentalhealthrelatedcause-specificmortality(Liu
etal., 2020; Gasparrini etal.,2012;Kimetal.,2015;HoandWong,
2019;OudinÅströmetal.,2015;Rocklövetal.,2014).Mostextracted
MBD status from hospital records based on the ICD-9 or ICD-10
[e.g.,23,38](n=55,62.5%),althoughsomeusedclinicaldiagnosis
byatrainedpsychiatristbasedontheDiagnosticandStatisticalManual
ofMentalDisorders(DSM-IV)(Leeetal.,2002;Huibersetal.,2010).
AmongstudiesutilizingICDstandards,thetopfivemostexamined
outcomecategorieswereschizophrenia(n=25,45.5%),mooddisor-
ders (n = 20, 36.4 %), organic mental disorders such as dementia
(n=15,27.3%),neuroticdisorderssuchasanxietyanddepression
(n=13,23.6%),andpsychoactivesubstanceuse(n=10,18.2%).A
fewstudiesinvestigatedintellectualdisabilities(n=5,9.1%),behav-
ioralsyndromes(n= 3,5.5%),personalitydisorder(n=2,3.6%),
developmentaldisorders(n=2,3.6%),andchildhoodbehavioraldis-
orders(n=2,3.6%).Finally,31studies(56.4%)examinedMBDout-
comesthatincludedmultiplecategories.
3.2.4. Studytype,spatiotemporalscale,andstatisticalanalysis
The reviewed studies most commonly used a time series design
(n = 51, 57.9.2 %), followed by cohort (n = 17, 19.3 %), case-
crossover/case-control(n=9,10.2%),cross-sectionaldataorasingle
wavefromlongitudinaldata(n=5,5.6%),andotherdesigns(n=6,
6.8%).Time-seriesstudiesmostlyreportedalargenumberofrecords
(median=34,000),whilecohortstudiesshowedsmallersamplesizes
(median=480).Fig.4illustratesthespatialandtemporalresolutions
ofthestudies. Concerning spatial resolution, most studieswere con-
ductedatmacro-ormeso-scaleduetotheresolutionofthedatasource;
forexample,dataonhospitalizationandemergencyroomvisitswere
oftenreportedataggregatedlevels(Bulbenaetal.,2005;Bulbenaetal.,
2009;daSilvaetal.,2020)rangingfromahospitalcatchmentorneigh-
borhood(n=13,14.8 %)toanentire nation(n=19,21.6%).Re-
gardingtemporalresolution,moststudies(n=71,80.7%)examined
daily climatic factors, with a few considering monthly patterns
(10.2%).
Whenestimatingexposure-response associations, the mostwidely
usedstatisticaltechnique,especiallyinearlierstudies,waslinearmod-
eling (including generalized linear and linear mixed modeling)
(n=49,55.7%).Poissonornegativebinomiallinkfunctionswerefre-
Fig.4. Spatiotemporalresolutionofthestudies(n=88).
22

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
quentlyemployed to account for MBDincidencecountorprevalence
data;linearmixedmodelswereappliedforthenesteddatastructureof
repeated measurements from individual participants (Bullock et al.,
2017;Sarranetal.,2017). Recentstudiesfavoredthetechniquenon-
lineardistributedlagmodeling(DLNM,n=22,25.0%),whichcansi-
multaneouslyfitnon-linearexposure-responserelationships and non-
lineardelayed effects using a bi-dimensional matrix(i.e.,cross-basis)
(GasparriniandArmstrong,2010;Gasparrini,2011).Includedstudies
oftencomparedmultiplelagvaluesoftheeffects,includingcumulative
lageffects,inthemainanalysesorsensitivityanalyses.Themaximum
lagvalueswereoftendeterminedbasedontheliterature,modelfitsta-
tistics,orRR-lagplots.Asidefromlag0,themostfrequentlyusedlag
valuewas0
–
7(n=15,68.2%),followedby0
–
3(n=11,50%),0
–
2
(n=8,36.4%),and0
–
21(n=6,27.3%).Manystudiesconsidered
sevendaysasthedelayperiodforshort-termexposuretoambientcon-
ditions(Almendraetal.,2019),and21daysformedium-longexposure
(Eun-hyeetal.,2021).However,findingsrelatedtoappropriatelagval-
uesvaried.Table2providesdetailsaboutthecurvefunctions,lagval-
ues, and single-day versus cumulative effect estimates. In addition,
othermethods included Pearsonand Spearman rankcorrelation, au-
toregressiveintegratedmovingaverage(ARIMA),Coxregression,and
generalizedestimationequation(GEE)methods.
3.3. Meta-analysisoftherelationshipbetweenclimatecharacteristicsand
mentalhealth
3.3.1. ClimaticfactorsandMBD
Ofthe76studiesinthesystematicreviewset,42wereincludedin
themeta-analysis, whichprovidedestimatedreportedcomparable ef-
fectsizesthatcouldbeconvertedtorelativerisk.Asdifferentstudies
examineddifferenttypesofMBD,andsomeonlyreportedcasesofMBD
withoutdifferentiatingsubclasses(e.g.,schizophrenia,mooddisorder),
wefirstestimatedrandom-effectmodelsbycombiningallsubclassesof
MBDandconsideringMBDriskasasingleoutcomevariable(Fig.5).
Whenmultipleoutcomesormodelswerereported,weadheredtothe
rulesdetailedinSection2.5toselectonlyoneeffectsize.Asthemajor-
ityofthereviewedstudiesreportedmorbidityandonlyafewinvesti-
gatedmortalityrisk,wereportthefindingsrelatedtomorbidityfirst,
followedbymortality.Pooledeffectswerecreatedforeachexposure,
suchasthermalindex, hot/cold air temperature,andsunshine dura-
tion.Whenlessthanthreestudieswereavailableforacertainexposure-
outcomepair,wepresentedtheeffectsizesfromtheindividualstudies
forthesakeoftransparencyandcompletenessbutdidnotincludethem
in meta-analysis models. The exposure variables omitted from the
meta-analysiswerewinddirection/speed,barometricpressure,humid-
ity,andprecipitation.
3.3.1.1. Thermal comfort  index and MBD. Apparent temperature is a
summaryindexthatconsiderspositivecontributionstothehumanen-
ergy budget from air temperature, relative humidity, and radiation,
alongwiththenegativecontributionofwindspeed.Drawingonthree
studiesthatexaminedapparenttemperature,ourmeta-analysisresults
suggested a heightened risk of MBD during energy budget overload
butnot energy loss conditions. Namely, when apparent temperature
was at the 90th percentile (heat overload), MBD risk was elevated
(pooledRR= 1.08, 95%CI = 1.03,1.12) compared tomedianor
minimum risk temperatures. Heterogeneity across the studies was
small (I2 = 5.3 %). Meanwhile, apparent temperature in the 10th
percentile (heat loss) was not associated with MBD (pooled
RR = 1.02, 95 % CI = 0.99, 1.05) and heterogeneity was high
(I2= 80.4%).
3.3.1.2. AirtemperatureandMBD. Themeta-analysisresultssuggested
thatheatconditionsexceedingcertainlocalthresholds(e.g.,consecu-
tivedailytemperaturesof35°Corexceedingthe97.5thor99thlocal
Table2
Studiesusingdistributedlagmodelsandselectionoflagandcumulativeef-
fects(N=22).
Author,
year
Linearity Statistical
analysis
Immediate
or
cumulative
effect
Lagtime/
exposurepriorto
outcome
Temporal
unit
Wilsonet
al.,2013
Non-
linear
Distributed
lagnon-
linear
model
Immediate
&
cumulative
0,1,2,3,0
–
2 Daily
Wanget
al.,2014
Non-
linear
Distributed
lagnon-
linear
model
Immediate
&
cumulative
0,1,2,3,4,5,6,
7,8,9,10,11,
12,13,14,15,
16,17,18,19,
20,21,22,23,
24,25,26,27,
28,29,30,0
–
7
Daily
Penget
al.,2017
Non-
linear
Distributed
lagnon-
linear
model
Cumulative 0,0
–
1,0
–
2,0
–
3,
0
–
4,0
–
5,0
–
6,0
–
7,0
–
14,0
–
21
Daily
Chanet
al.,2018
Non-
linear
Distributed
lagnon-
linear
model
Cumulative 0
–
2,0
–
8 Daily
Leeetal.,
2018
Non-
linear
Distributed
lagnon-
linear
model
Cumulative 0
–
7 Daily
Sherbakov
etal.,
2018
Non-
linear
Distributed
lagnon-
linear
model
Cumulative 0
–
3 Daily
Wanget
al.,2018
Non-
linear
Distributed
lagnon-
linear
model
Cumulative 0,0
–
1,0
–
2,0
–
3,
0
–
4,0
–
5,0
–
6
Daily
Almendra
etal.,
2019
Non-
linear
Distributed
lagnon-
linear
model
Cumulative 0,0
–
1,0
–
2,0
–
3,
0
–
4,0
–
5,0
–
6,0
–
7
Daily
Guetal.,
2019
Non-
linear
Distributed
lagnon-
linear
model
Cumulative 0
–
7,0
–
14,0
–
21 Daily
Minetal.,
2019
Non-
linear
Distributed
lagnon-
linear
model
Immediate
&
cumulative
0,1,2,3,4,5,6,
7,8,9,10,11,
12,13,14,15,
16,17,18,19,
20,21,0
–
1,0
–
2,
0
–
3,0
–
4,0
–
5,0
–
6,0
–
7,0
–
8,0
–
9,
0
–
10,0
–
11,0
–
12,0
–
13,0
–
14,
0
–
15,0
–
16,0
–
17,0
–
18,0
–
19,
0
–
20,0
–
21
Daily
Panetal.,
2019
Non-
linear
Distributed
lagnon-
linear
model
Immediate 0,1,2,3,4,5,6,
7,8,9,10,11,
12,13,14,15,
16,17,18,19,
20,21
Daily
Xuetal.,
2019
Non-
linear
Distributed
lagnon-
linear
model
Immediate 0,1,2 Daily
Yietal.,
2019
Non-
linear
Distributed
lagnon-
linear
model
Immediate
&
cumulative
0,1,2,3,4,5,6,
7,8,9,10,11,
12,13,14,0
–
1,
0
–
2,0
–
3,0
–
4,0
–
5,0
–
6,0
–
7,0
–
8,
0
–
9,0
–
10,0
–
11,
0
–
12,0
–
1,0
–
14
Daily,0
–
14
cumulative
days
(continuedonnextpage)
23

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Table2(continued)
Author,
year
Linearity Statistical
analysis
Immediate
or
cumulative
effect
Lagtime/
exposurepriorto
outcome
Temporal
unit
daSilvaet
al.,2020
Non-
linear
Distributed
lagnon-
linear
model
Cumulative 0,0
–
6,0
–
7 Daily
Liuetal.,
2020
Non-
linear
Distributed
lagnon-
linear
model
Cumulative 0,1
–
5,6
–
21,0
–
21
Daily
Niuetal.,
2020
Non-
linear
Distributed
lagnon-
linear
model
Immediate
&
cumulative
0,1,2,3,4,5,6,
7,0
–
1,0
–
2,0
–
3,
0
–
4,0
–
5,0
–
6,0
–
7
Daily,0
–
7
cumulative
days
Zhanget
al.,2020
Non-
linear
Distributed
lagnon-
linear
model
Immediate
&
cumulative
0,1,2,3,4,5,6,
7,8,9,10,0
–
1,
0
–
3,0
–
5,0
–
7,0
–
9
Daily
Bundoet
al.,2021
Linear Distributed
laglinear
model
Cumulative 0
–
3,0
–
7 Daily
Eun-hyeet
al.,2021
Non-
linear
Distributed
lagnon-
linear
model
Cumulative 0,0
–
3,0
–
7,0
–
14,0
–
21
Daily
Jahan&
Wraith,
2021
Non-
linear
Distributed
lagnon-
linear
model
Cumulative 0
–
3,0
–
7,0
–
10,
0
–
15,0
–
21,
−
0-
28,0
–
30
Daily
Yooetal.,
2021
Non-
linear
Distributed
lagnon-
linear
model
Cumulative 0
–
7 Daily
percentile)were consistently associated with increased risk of MBD:
heatwave,pooledRR=1.05,95%CI=1.02,1.08;97.5thpercentile
airtemperature,pooledRR=1.18,95%CI=1.07,1.30;99th per-
centileairtemperature,pooledRR=1.18,95%CI=1.08,1.29.In
contrast,therelationshipofcoldweatherandMBDwaslessconsistent
acrossmeasures;temperaturesinthe1stpercentile(pooledRR=0.97,
95%CI=0.86,1.09)and2.5thpercentile(pooledRR=1.14,95%
CI=0.94,1.35)were not associated with MBD.Poolingoffindings
fromstudiesthatassumedalog-linearmonotonalrelationshipbetween
1°CincreaseintemperatureandMBDriskalsoyieldedinsignificantre-
sults(pooledRR=1.01,95%CI=0.99,1.03),althoughresultswith
a linearity assumption needed to be interpreted with caution (see
Section 4.3). Furthermore, Four studies among them considered the
year-roundtemperaturerange(Kimetal.,2021;Huibersetal.,2010;
Trangetal.,2016b;Oh et al., 2020), (e.g., May to September in the
Northern Hemisphere or October to March in the Southern Hemi-
sphere)(Williamsetal.,2012;Vidaetal.,2012;Bundoet al.,2021),
whiletheotherfiveuseddatayear-round.Finally,whileheterogeneity
waslowforheatwave
–
MBD(I2=17.2%),itwashigh(I2>80%)
forallothermodels,likelyduetothevastdifferencesinpopulations,
climateconditions,andhightemperatureexposurevariables.
Studiesoncause-specificmortalityrelated toMBDmostlyfocused
onairtemperaturemetrics.Poolingtheeffectsofthethreestudiesthat
assumedlog-linearity,wefounda1°Cincreaseintemperaturetobeas-
sociated with 3 % increased risk of MBD-related mortality (pooled
RR = 1.03, 95 % CI = 1.02, 1.04), and low heterogeneity
(I2=0.0%).
3.3.1.3 Solar radiation/sunshine, barometric pressure, precipita-
tion,andhumidity,andMBD.
Amongthevariousexposuremeasuresrelatedtosunshineandsolar
radiation,onlysunshinedurationbyhourwasusedbythreestudiesand
therefore amenable to risk estimation using a random effects meta-
analysismodel.This estimation didnot show anassociation (pooled
RR= 0.98, 95% CI = 0.93, 1.03). Regardingbarometric pressure,
onlytwostudieswereavailable.Likewise,forprecipitationandrelative
humidity,avarietyofvariableswereexamined,eachinasinglestudy;
theseincludedrainvolume(mmorpercentilethreshold),rain/snow/
foggyday,andrelativehumidity(%).Thus,meta-analysescouldnotbe
performedfortheseexposures.
3.3.2. Outcome-specificanalysis
Tounderstandthedirectionandmagnitudeoftheassociationsbe-
tweenclimatefactorsandspecificsubcategoriesofMBD,wetalliedef-
fect sizes according to the specific MBD subclass reported and per-
formedameta-analysisoneachsubclass.Aftersortingtheclimatefac-
tor-mentaldisorderpairs, only threeoutcomecategories were repre-
sented in at least three studies with the same exposure-outcome
pairs
—
schizophrenia (F20
–
F29, n = 4), mood disorders (F30
–
F39,
n=7),andneuroticdisorders(F40
–
F49,n=6)
—
andthusabletobe
pooledusingrandom-effectmodels.Otheroutcomessuchasdementia
andbehavioralsyndromeswerenotincludedinthemeta-analysisdue
tofeaturinginlessthanthreestudies.Wepresentforestplotsshowing
thepooledeffects of the threeanalyzableexposure-outcome pairs in
Fig.6,whilethecompletesetofindividualeffectsizesarepresentedin
SupplementarymaterialFigs.S4-1toS4-6.
3.3.2.1. Schizophrenia. Four studies fitted models to compare 99th
percentile and median/minimum air temperatures for relation to
schizophrenia.Basedonthesedata,wefound a 99th percentile high
temperature to be associated with higher risk of schizophrenia
(pooled RR = 1.07, 95 % CI = 1.01, 1.12), with low to moderate
heterogeneity (I2 = 40.4 %).
3.3.2.2. Mood disorders. Pooled results from four studies estimating
theeffect of a 1 °C increase in temperature yielded insignificant re-
sults(pooled RR = 1.00, 95 % CI = 0.97, 1.03), while the pooled
associationof 99thpercentiletemperatureandelevatedrisk ofmood
disorderapproachedstatisticalsignificance(pooledRR=1.23,95%
CI = 1.00, 1.51). These models featured high heterogeneity scores
(I2 > 80 %) due to geographical and population differences, along
with variation in outcome classification: three studies defined their
outcomevariableusingICD-9or10,wherethefourthusedtheDSM-
IV.Moreover, Kim et al. (2015) only considered ICD codes F31
–
33,
whilethe other twoconsidered F31
–
40.
3.3.2.3. Neuroticdisorders. Pooledresultsrelatedtoneuroticdisorders
(e.g., depression) were generally not significant. The association of
1°Cincreaseintemperaturewithneuroticdisordersdidapproachsig-
nificance (pooled RR = 1.02, 95 % CI = 1.00, 1.04), but that of
99th percentile temperature did not (pooled RR = 1.11, 95 %
CI = 0.92, 1.33). Heterogeneity scores were high for these models
(I2> 75 %) for reasons similar to those described above.
3.4. Riskofbiasandcertaintyofevidence
Fig.7showstheresultsoftheriskofbiasassessmentbystudyand
assessmentcategory.Theindividualscoresforeachitemevaluatedare
givenin Supplementary material S5Fig.S5
–
1.Across all studies,the
majorbiasconcernssamplingselection, outcome measurements,and
confounding factors (Supplementary material S5 Fig. S5
–
2). Studies
thatutilizedrecordsfromafewhospitalswithoutexplainingthesam-
plingmethod or theirrepresentativenessofthepopulationwerecon-
sideredtohaveelevatedbias.Afewearlyindividual-levelstudiesuti-
lized convenient samples from students, hospitals, or online panels
(Barnston,1988),whichmightbesubjecttovolunteerbiasandhence
wereconsideredprobablyhighbias.Regardingoutcomemeasurement,
studiesthatextractedoutcomesfromhospitalrecordsusingstandard
classificationsystems,suchasICD-9or10orDSM-IV,wereconsidered
24

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Fig.5. Forestplotofclimate/meteorologicalfactorsandriskofmentalandbehavioraldisorders(*denotesoutliers;REModel).
tohavelowriskofbias,aswerethoseinvolvingclinicaldiagnosisbya
trained psychiatrist or self-report based on validated diagnostic/
screeninginstruments.Studiesthatusedmeasuresnotvalidatedforthe
specificconstructtobeexaminedweregradedashavingprobablyhigh
bias.Inaddition,iftheoutcomewasclinicallydiagnosed,weconsid-
eredwhether the assessors wereblindedtoexposureconditions;out-
comesmeasuredby self-reported scale wereuniformlyconsidered to
haveelevated bias. Forconfounding factors, themajority of popula-
tion-andindividual-levelstudiesincludedtime-invariantfactorssuch
asage,sex,socioeconomicconditions,andpre-existing health condi-
tions,time-variantfactors such asseason, day ofweek, time of day,
andother exposurefactorssuchasair pollutionandnoiselevel.Sev-
25

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Fig.5.(continued)
eral studies that referenced morbidity records used auxiliary place-
baseddatasetsthatprovidedpopulation-leveldemographicconditions.
Studiesthatonlyadjustedfor temporalfactors(season,day of week,
diurnal)but didnotincludestudypopulation/individual- orenviron-
ment-relatedconfoundingfactorsweregradedashavinghigherriskof
bias,while thosethatdidnotconsideror adjustforanyconfounding
factorsintheirexposure-response estimates were consideredtohave
probablyhigh risk of bias.Finally, selective reportingof results was
observedinsomestudieswhenmultipleexposure,outcome,orstatisti-
cal models were mentioned in the planned analyses, but only those
withsignificantresultswerepresentedanddiscussed.
UsingtheGRADEcriteria,weassessedthecertaintyofevidencefor
eachcategorywherepooledeffectswereproduced.Weassignedabase-
lineofmoderatecertaintybasedonthestudydesign,followedbydeci-
sionstodowngradeorupgradeforeachoftheGRADEdomains.Effect
estimatesforwhichmorethanhalfofthecontributingstudiesexhibited
26

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Fig.6. Forestplotofclimate/meteorologicalfactorsandschizophrenia,mooddisorders,andneuroticdisorders.(*denotesoutliers;REModel).
greaterthanlowriskofbiasinatleastonedomainweredowngraded
byonelevel,suchastheassociationsbetweenMBDandheatwaveand
sunshineduration.Allstudiesmetthespecifieddesiredpopulation,ex-
posure,comparator,and outcome criteriaand were notdowngraded
basedonthosecategories.Forexample,nostudyusedasoutcomeasur-
rogateofaclinicalendpoint.Whenweidentifiedinconsistencies,such
asapredictioninterval more than twicetheconfidenceinterval and
studiesdisagreeinginthedirectionoftheassociation(i.e.,positiveand
negativeassociationforthesame exposure-outcome pair), wedown-
gradedtheevidence;acaseinpointistheassociationofhighairtem-
peratureandMBD.Forthenumberofparticipantsandspecificallyper-
son-time,allbutonestudyusedatime-seriesdesignorfeaturedalarge
cohort.Theonlystudythatdidnotreportthenumberofparticipants
usedhospitaladmissionsrecordsduringa16-yearspanandwas thus
consideredtohaveadequatesamplesize.Therelativerisksyieldedby
therandomeffectsmeta-analysismodelsweresmall,andthereforeno
evidence was upgraded for large magnitude of effect. As meta-
regressionwasnotpossibleduetothesmallnumberofstudies, there
was insufficient evidence for a dose-response gradient. Confounders
mayeitherhavepositiveornegativeimpactsontheassociations,and
thereforeno upgradingwasperformedforthe criteriaregardingcon-
founding.
Ultimately,basedontheoutcomesofthestructuredGRADEprocess,
thereisamoderatelevelofevidencesupportingthatMBDiselevated
whenthe thermal index(apparent temperature) increases.Similarly,
theassociationbetweenriskofschizophreniaandhighairtemperature
hasmoderatecertainty.Otheridentifiedassociationssuchasbetween
heatwaveor high temperature and MBDstill require more attention
duetohavingonlylowcertaintyofevidence.Table3presentsthede-
tailsoftheevaluationoutcome.
4. Discussion
4.1. Mainfindings
Inthisstudy,weperformedasystematicreviewandmeta-analysis
toexaminetheassociationsbetweenclimaticandmeteorologicalfac-
torsandmentalhealth.Theresultsrevealedunevengeographicaland
climatezonedistributions and uncoveredunderrepresentedclimates.
Temporally, earlier studies focused on psychological states using
smallersamples,recentstudiesemphasizedpsychiatricmorbiditywith
longitudinal/time-seriessurveillancedata.Extremeheatbasedonheat
thresholdsandhighthermalindexintegratingtemperature,humidity,
andwindvelocity were identifiedas risk factorsforMBD, while ex-
treme cold was not. Evidence was limited to generate medium-high
confidencepooledeffectssizesforhumidity,wind,andsolarradiation.
Regarding different subclasses of MBD, schizophrenia risk increased
whentemperatureroseabovethe99thpercentile,andmoodandneu-
27

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Fig.7. Riskofbiasassessmentresults.
roticdisorderriskswereapproachingsignificancewithtemperaturein-
crease.
Ourmeta-analysisrevealedthat,ingeneral,heatextremesandtem-
peraturesexceedinglocalthresholdswererelatedtomentalandbehav-
ioraldisorders;specifically,theseassociationswereidentifiedamong
studiescomparingheatwavewithnon-heatwavedaysandalsoamong
studiesestimatingMBDriskin99thor97.5thpercentiletemperatures
relativetomedianorminimumtemperatures.Theseresultsareconsis-
tent with findings from prior systematic reviews and meta-analyses,
whichindicatedhightemperaturesandheatwavestoberiskfactorsfor
mentaldisorders(Thompsonetal., 2018;Liuetal.,2021).Neverthe-
less,thecertaintyofevidenceisconsideredloworverylowformostex-
posure-outcomepairsduetoriskofbiasissuesandhighheterogeneity.
Thompsonetal.(Thompsonetal.,2018)reportedasimilarfindingbe-
tweenheatandmentalhealthoutcomes,althoughameta-analysiswas
notconducted due toheterogeneity. Liu et al. (Liuet al., 2021)ob-
tainedpooledeffectsizes(RR)forvariousdefinitionsofheatwave,and
reported pooled RRs between 1.048 and 1.753. Our effect estimate
28

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Table3
Evaluationofcertaintyofevidence.
Ratingfactors D1 D2 D3 D4 D5 U1 U2 U3 Finalrating
Apparenttemperature(heat)
–
MBD
0 0 0 0 0 0 0 0 Moderate
(+++)
Apparenttemperature(cold)
–
MBD
0 0
−
1 0 0 0 0 0 Low(++)
Heatwave
–
MBD
−
1 0 0 0 0 0 0 0 Low(++)
Airtemperature(high
temperature)
–
MBD
0 0
−
1 0 0 0 0 0 Low(++)
Airtemperature(low
temperature)
–
MBD
0 0
−
1 0 0 0 0 0 Low(++)
Airtemperature(linear
increase)
–
MBD
−
1 0
−
1 0 0 0 0 0 Verylow
(+)
Sunshineduration
–
MBD
−
1 0
−
1 0 0 0 0 0 Verylow
(+)
Airtemperature(high
temperature)
–
Schizophrenia
0 0 0 0 0 0 0 0 Moderate
(+++)
Airtemperature(high
temperature)
–
mood
disorders
0 0
−
1 0 0 0 0 0 Low(++)
Airtemperature(linear
increase)
–
mooddisorders
−
1 0
−
1 0 0 0 0 0 Verylow
(+)
Airtemperature(high
temperature)
–
neurotic
disorders
0 0
−
1 0 0 0 0 0 Low(++)
Airtemperature(linear
increase)
–
neurotic
disorders
−
1 0
−
1 0 0 0 0 0 Verylow
(+)
overlapwith the lower endof the RRrange, for estimatesincluding
heatwave(RR=1.05,95%CI=1.02
–
1.08),97.5thpercentiletem-
perature(RR=1.18,95%CI=1.07
–
1.30),and99thpercentiletem-
perature(RR=1.18,95%CI=1.08
–
1.29).Thedifferenceontheup-
perendoftheRRswaslargelyattributabletothefactthatthepresent
studydid notconsideroutcomessuchas suicideideationorattempts
whileLiuetal.did.Coldextremes,ontheotherhand,werenotsignifi-
cantlyassociatedwithMBDrisk.Nopreviousstudiesareavailableon
thisrelationshiptocomparetheeffectestimate.
Beyondtemperaturemetricsalone,wealsoidentifiedathermalin-
dex,which combines temperature,humidity, wind speed, andradia-
tion,tobefactorforelevatedrisksofMBD.Temperatureisthetopcli-
maticfactorcontributingtothermalstress,butnottheonlyone;humid-
ityandradiation variations canhave as largeeffectsas temperature
change.Whilepopularmetricsusing99thor97.5thpercentiletempera-
turesweresignificantbutlow-certaintycorrelatesofMBD,thermalin-
dexatthe90thpercentileyieldedresultswithmoderatecertainty.This
indicatestheimportanceofconsideringmultiplesourcesofheatstrain
inthe course of human-environment heatexchange.Studiesonsport
andoccupationalmedicineusingestablishedthermalmodels such as
physiological equivalent temperature (PET), the Comfort Formula
(COMFA),andwet-bulbglobetemperature(WBGT)havedemonstrated
thesemodelstohavestrongexplanatorypowerassummariesofther-
malstress(Budd,2008).Meanwhile,wefoundnosignificantrelation-
shipswhenconsideringsunshinedurationalone,likelyduetothesmall
numberofstudiesthatusedthismetric.Otherclimateexposures,such
ashumidity,windspeed,barometricpressure,andsolarradiation,were
measuredheterogeneously and couldnot be pooled,thus the results
wereinconclusive.
Whenitcomestospecificmentaldisorders,ingeneral,studiesusing
homogeneousclimate/meteorological metrics were lacking. Previous
systematic reviews that provided narrative summaries of patterns
mostlynotedthathightemperaturesmightbeatriggerofschizophre-
nia,bipolardisorder,anddementia(Thompsonetal.,2018;Monteset
al., 2021; Bongioanni et al., 2021). To our knowledge, our study is
amongthefirstto conduct a meta-analysisonvariousmental health
outcomes,especiallyforeachtypeofMBDbasedonICDclassifications
and other self-reported psychological states. The only significant
pooledeffectidentifiedwasanassociationof99thpercentiletempera-
turewithschizophrenia,whichalsoshowedmoderatecertaintyofevi-
dence.Therelationshipsoftemperaturewithmoodandneuroticdisor-
dersapproached but did not achievestatistical significance, and the
confidenceinthebodyofevidencewasassessedaslowatbest.
Theheterogeneityof effect estimates was high on most exposure-
outcomepairs,likelyduetodifferencesinclimateconditions,popula-
tioncharacteristics,andheatadaptationandmitigationcapacities.It
haspreviouslybeennoted that general populationstudiesmay yield
heterogeneityvaluesintherangeof75%andabove,higherthanthose
reportedinrandomizedcontrolledtrialsorothersmallerstudies(Chen
andHoek,2020).
4.2. Mechanismsoftheobservedassociationsbetweenclimateconditions
andmentalhealth
Themechanismsunderlyingtheobservedassociationsbetweencli-
mate/meteorological factors and mental health may be multifaceted
andcomplex.First,the associations between heatandmentalhealth
may be explained by dysregulation of the serotonin system (5-
hydroxytryptamine,5-HT),whichis involved in moodandcognition
modulation.Studieshaveshownthatheatandhumiditymaycausewa-
terdeprivationanddehydration,whichleadtoadecreaseinserotonin;
suchdeficiencyisrelatedtoincreasedmooddisorders,depression,anx-
ietydisorders,schizophrenia,andotherMBDs(Linetal.,2014).Ithas
alsobeensuggestedthat,asserotoninplaysanessentialroleinthermal
regulation,acuteambienttemperaturechangesmaycausesystemdys-
regulation,therebyaggravatingmentalhealthsymptoms(Craneetal.,
2015).Ontheotherhand,manyantipsychoticandpsychotropicmed-
icationshavethermoregulatorysideeffects,andintakeofsuchmedica-
tionsmay berelatedtocompromisedthermoregulationandperspira-
tion(Zammitetal.,2021).Second,recentresearchhasindicatedthat
heatstresscaninduceneurodegeneration;inparticular,heatstrokecan
cause excitotoxicity, necrosis, and apoptotic death of neuronal cells
(Gongetal.,2022;Zammitetal.,2021;Kourtisetal.,2012).Clinical
studiessuggestthatneuronalcellandneuralnetworkalterationscon-
tributeto the pathologyofmentaldisorders(Quachetal., 2016). Fi-
nally,inadditiontobiologicalmechanisms,theremaybeasocioeco-
logicalexplanationofthelink:extendedperiodsofheatmayinfluence
individualand family daily routines andsocial networks,presenting
challengesfor commute and childcare routines, and alsoincur emo-
tionalstrainrelatedtofamilymemberswithpre-existinghealthcondi-
tions;alloftheseareexternalstressorstomentalhealth.Adversemicro-
climateconditionsalsolimitopportunitiestoengageinoutdooractivi-
tiesthatofferrestorativeeffects(e.g.,visitingaparkorabeach)(Hartig
etal.,2007;Lietal.,2019).Finally,disastersituationssuchasheatand
coldstormsoftenalsocomewithdisruptionoftransportationanden-
ergy/water infrastructures (Li et al., 2022b), exacerbating the psy-
chosocialstressexperiencedbyindividuals.
Onepotential pathway that was notsubstantiated in the current
studyishumansunlightexposureandthecircadianclock.Specifically,
researchon seasonalaffectivedisorderanddepressionhas outlineda
circadiantimingmechanismviathehypothalamicsuprachiasmaticnu-
cleusthatisdependentonambientconditions,withthelight-darkcycle
beingconsidered themostrelevantsynchronizerforthebody's circa-
diantiming(e.g.,sleep-wakecycle,locomotoractivity,hormones);dis-
turbanceofthattimingisrelatedtodepressionanddistress(Mendoza,
2019).
4.3. Futuredirectionsandlimitations
Whiletherehasbeenarecentsurgeofresearchinterest,giventhe
broadarrayofclimaticfactors,thevarioustypesofmentalhealthcon-
ditions,andtheinsufficientevidenceofassociationdeterminedinthis
studyformostexposure-responsepairs,moreresearchiswarrantedre-
29

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
gardingeachexposure-response pair and alsothejoint effects of cli-
maticexposures.Here,wediscusspotentialfuturedirectionsforfilling
critical knowledge gaps and addressing the risk of bias and other
methodologicalconcerns.
4.3.1. Standardizingmeteorologicalmetricsandthresholds
Thewidevarietyofmeteorologicalmetricsusedinincludedstudies
reducestheabilitytoperformmeta-analysisandpromoteshighhetero-
geneity.Forexample,measuresusedtorepresentsunshineandsolarra-
diationincludeddailysunshineavailability,duration,solarradiation,
daylength,cloudcover,anddirectandglobalradiation.Likewise,for
airtemperature,studiesestimatedtheeffectsof1°Cincreaseindaily
mean or maximum temperature, 1 °C increase in daily temperature
range,variouspercentilethresholds(e.g.,99,97.5,95,90,75),andvar-
iousdefinitionsofaheatwave.Werecommendthatfuturestudiesse-
lectmetricsthatprovideadequategranularityandreflectbiometeoro-
logicalmechanisms. Regarding temperature, for instance,dailymean
ormaximum temperature mayfail to capturethe critical impactsof
nighttimetemperatureonmental health, especially duringmulti-day
heatevents.Theliteraturealsoindicatesthathighnighttimetempera-
ture,especiallyifpersistent,isrelatedtosleepingdisturbances(Haskell
etal.,1981;LibertandBach,2005),whichcouldtriggerepisodesof
mentaldisorders(LiewandAung,2021).Italsoremainsnecessaryto
considerthejointeffectofmultiplemeteorologicalconditions.Asther-
malstressisrelatedtomost recognizedbiologicalneurodegeneration
pathwaysinvolvedinmentalhealth,futurestudiesmayconsiderusing
establishedthermalindicesthatconsidermetabolic,conductive,con-
vective,andradiativeheatflux,eitherasthemainexposurevariableor
inthesensitivityanalysis.
4.3.2. Addressingscalemismatchesandattentiontomicroclimateexposure
Although the majority of studies included in this review utilized
ground-measuredmeteorologicalconditions,spatialmismatchesoften
existbetweenthemeasurementstakenatweatherstationsandthemi-
cro-environmentsexperiencedbyindividualparticipants.Studiesthat
considerclimateratherthanmicroclimatemaybesubjecttotheuncer-
taingeographiccontextproblem(UGCoP):namelythatarealmeasures
donotmatchtheenvironmentalexposurethatexertsimpactsonhealth
andbehavior,aproblemthathasbeendiscussedingeographyandspa-
tialepidemiology(Kwan,2012;Jiaetal.,2020).Nevertheless,studies
onhuman exposure to heat, wind,andsolarradiationreviewedhere
stillmostly usedareal-basedclimatemeasuresderivedfrom sparsely-
locatedweatherstationswhichmaybetoocoarsetoestablishexposure-
responserelationships.Infact,microclimateconditionscanbemodi-
fiedbyurbanheatislandeffects,site-levelstructuralelementsandveg-
etation,waterbodies,andthealbedoofmaterials.Allofthesefactors
directlyimpacthumanphysiology,thermalsensation,andbehavioral
adaptationatagivenlocation.Futurestudiesarewarrantedthatpre-
ciselymeasurethemicroclimatestowhichindividualsareexposedin
residentialneighborhoodsandthirdplaces,suchasthroughtechnolo-
gieslikedownscalingand/ormicroclimateloggernetworks(Georgeet
al.,2015).
4.3.3. Studyingdisease-specificexposuresandpathways
Morestudiesareneededonspecificmentaldisorders(e.g.,schizo-
phrenia,depression,anxiety,dementia).Inthesetofstudiesreviewed
here,manyutilizedcombineddiseaseprevalencefrommultipleMBDs
thatmayreflectdifferentriskpathways,leadingtoresultsthatarelev-
eledoutandchallengingtointerpret.Forstudiesthatconsideredspe-
cificmentaloutcomes,ICDandDSMcodesweretypicallyused;how-
ever,ICDclassificationsdependongoodclinicaljudgmentandcanbe
lessaccurate,whiletheDSMislesscommonlyusedoutsidetheU.S.In
addition,admission/dischargerecordswereoftenaggregated without
distinctionofclinicalstages.Astheimpactsofriskfactorsandinterven-
tionsmaydifferoverthediseasecourse,itiscriticaltoconsiderdisease
trajectoryand climateat differentstages suchas in prodromal, first
episode,persistent,andremittedcases.Additionally,morestudiesthat
addressneurotransmittersandothermechanismswouldhelpadvance
knowledgeandpolicy.
4.3.4. Expandingoutcomesfrommorbiditytowell-being
Publichealthpoliciesacrosstheworld,suchastheHealthyPeople
2030frameworkoftheU.S.,havebeenadvocatingfornotonlydisease
prevention/treatmentbut also forpromoting people to achievetheir
fullpotential.Althoughthisreviewsetouttoframementalhealthina
broaddefinitionthatencompasseshappinessandwell-being,thenum-
berofstudies employing such constructswas limited. Therefore,the
quantitativesynthesismostlyaddressedMBD;obtainingpooledeffects
foraffectandotherpositiveemotionswasnotpossibleduetothesmall
numberofstudiesandhighheterogeneity.Futurestudiesarewarranted
thatusevalidinstrumentstoexaminewhetherclimate-relatedrisksim-
pactsubjective well-being. In addition toidentifying modifiablerisk
factors,thereisvaluein adoptingasalutogenicview thatdetermines
theambientconditionsthatprovideforaqualityexperienceandsup-
portsubjectivewell-being.
4.3.5. Modelingtechniquethataccountsfornon-linearanddelayed
dependencies
Thetwomainfactorsthatpertainedtostatisticalmodelselectionfor
estimatingclimateandmentalhealthrelationshipswerenon-linearity
andlagged andcumulativeeffects.Recentevidencesuggeststhat the
relationshipsbetweencontinuousclimaticand meteorological factors
andmentalhealthoutcomesarenotlinearormonotonicallyincreas-
ing/decreasing(Almendraetal.,2019).Infact,humanphysiologysug-
geststhatforclimaticconditionswithinareasonablerange,theenergy
budgetismaintained,meaningthatextremevaluesatbothendsmay
havenegativeimpacts(Eun-hyeetal.,2021).Toaccountforthenonlin-
earity,studies can use heatandcoldthresholdsoronlyrecordsfrom
warmseasons.Additionally,recent studies that considerednonlinear
exposure-responserelationships haveoften fittedmodels with spline
functions(Yooetal.,2021;Leeetal.,2018b;Bundoetal.,2021).Ofall
studiesthatconsiderednon-linearexposure-responsecurvesusingdis-
tributedlagmodels,onlyonefoundalinearrelationship;theremainder
presentedcurvilinear functions for both the predictor and lags. This
raisesaquestionastothevalidityofthefindingsofstudiesthatassume
linearity/log-linearitywithout appropriate theoreticalgroundsorsta-
tistical tests. In the absence of specified geographic/seasonal condi-
tions,interpretationsofMBDriskasincreasingwithevery1°Cincrease
in air temperature can be misleading. Future meta-analysis articles
shouldalsousecautionwhenassuminglinear/log-linearormonotonic
relationships.Standardizationandmethodologicalcomparisonsacross
modelspecificationsandsingle-dayandcumulativelagvaluesmaybe
informative.Asmorestudiesaccumulatethatemployatwo-stagetime-
seriesdesign to estimate location-specific exposure-response relation-
ships,advancedmultivariatemeta-analysismethodsshouldbeconsid-
eredforfuturereviews(GasparriniandArmstrong,2013).
Limitationsofthisreviewalsoneedtobenoted.First, one major
limitationis the small numberof studies foreach exposure-outcome
pair.Oftenonlythreestudieswereavailable,renderingitimpossibleto
performsensitivityanalysis,excludeoutliersandstudiesbasedonspe-
cificconsiderations,andexaminetheimpactsofsuchexclusionsonthe
meta-analysisresults.Second,ourreviewdidnotincludeexperimental
andquasi-experimentalstudies,andhencethefindings,dependingon
theconfidenceofevidence,contributetobutcannotdirectlyestablish
causality. Third, we did not include outcomes related to suicide
ideationand attempts. Althoughpreviously suicide has been consid-
eredasymptomassociatedwithotherpsychiatricdisorders,theDSM-5
positionedsuicidalbehavioraldisorderasanindependentconditionfor
furtherstudy(DSM-IV-TR,2000).Fourth,weexcludedstudiesconsid-
eringmaternalexposureand child mental health conditions, because
30

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
theremaybeadditionalbiophysiologicalmechanismsinvolved.Never-
theless,we recognize theimportanceofstudyingreproductivehealth
andlinked lives in climate exposure. Fifth, althoughtheGRADEtool
providesa greatframework for evaluating the certainty of evidence
(Balshemetal.,2011;Schünemannetal.,2019a),ithaslimitationsin
assessing evidence from nonrandomized/observational research
(Schünemannetal.,2019b).Finally,inlightofourresearchquestions,
weconducted outcome-specific analysis, but could not perform sub-
groupanalysisbasedonage,gender,orgeographyduetothehighhet-
erogeneityandsmall number ofstudies available for eachexposure-
outcomepair.Narrativereviewevidence,suchasSharpeandDavison
(Sharpe and Davison, 2021), highlighted the links between climate-
relateddisastersandmentaldisordersspecificallyforlow-andmiddle-
incomecountries.Astheevidencebasecontinuestogrow,futurestud-
iesshould focus onhealthdisparitiesacrosspopulationgroups,espe-
ciallyvulnerablegroupssuchasolderadults,low-incomegroups,and
sociallyisolatedpopulations,whichmayofferscientificinsightsandin-
formtargetedpolicyinterventions.
5. Conclusion
Astheimpactofclimatechangeonmentalhealthgainsincreasing
recognition,acomprehensiveinvestigationoftherelationshipbetween
variousclimaticandmeteorologicalfactorsandmentalhealthcanre-
vealgapsinknowledgeandinformpolicyrelatedtopublichealthand
climateadaptation.Accordingly,utilizingthePRISMAframework,this
meta-analysisaimedtoidentifyfactorsassociatedwithincreasedriskof
mental and behavioral disorders. Our random-effects meta-analysis
modelsrevealedthathigherthermalindexvalues,heatwaves,andex-
tremetemperaturessurpassing certain thresholdswerelinked to ele-
vatedrisksofMBD.Furthermore,whenconsideringspecificsubtypesof
MBD,hightemperatureswerefoundtobeassociatedwithanincreased
riskofschizophrenia.Notably,theresultsunderscoredtheheterogene-
ityof exposuremeasuresandascarcityof evidenceregardingtheef-
fectsof climatefactorsotherthanairtemperature, suchashumidity,
wind,solarradiation,andbarometricpressure.Thefindingsalsounder-
scoredtheimportanceofinvestigatingthesynergisticeffectsofmulti-
pleclimatefactorsusingthermophysiologicalmodels,non-linearexpo-
sure-outcomerelationships,andcumulativeandlagged effects of cli-
mateexposure.
Funding
ThisworkwassupportedbytheNationalAcademiesofSciencesEn-
gineering and Medicine Gulf Research Program [grant numbers
#2000012329and2000013443]; and theNationalInstitute of Envi-
ronmentalHealthSciences[grantnumber#P42ES027704-01],andthe
HoustonMethodistResearchInstitute.
CRediTauthorshipcontributionstatement
DongyingLi1:Conceptualization,methodology,studyscreeningand
eligibility,biasassessment,synthesis&analysis,visualization,writing
–
originaldraft,writing
–
review&editing,andprojectadministration.
YueZhang:Conceptualization, methodology, studyscreening and
eligibility,biasassessment,synthesis&analysis,visualization,writing
–
originaldraft,andwriting
–
review&editing.
XiaoyuLi:Conceptualization,methodology,studyscreeningandeli-
gibility,biasassessment,synthesis&analysis,visualization,writing
–
originaldraft,andwriting
–
review&editing.
KaiZhang:Conceptualization,methodology,studyscreeningandel-
igibility,biasassessment,synthesis&analysis,visualization,writing
–
originaldraft,andwriting
–
review&editing.
1Guarantorofthereviewprotocol.
YiLu: Conceptualization, methodology, study screening andeligi-
bility, bias assessment, synthesis & analysis, visualization, writing
–
originaldraft,andwriting
–
review&editing.
Robert Brown: Conceptualization, methodology, study screening
and eligibility, bias assessment, synthesis & analysis, visualization,
writing
–
originaldraft,andwriting
–
review&editing.
Uncitedreference
Declarationofcompetinginterest
Theauthorsdeclarethefollowingfinancialinterests/personalrela-
tionshipswhichmaybeconsideredaspotentialcompetinginterests:
DongyingLireportsfinancialsupportthatwasprovidedbytheNa-
tionalAcademiesofSciencesEngineeringandMedicineGulfResearch
Program.Dongying Lireportsthatfinancialsupportwas providedby
theNational Institute ofEnvironmentalHealthSciences.DongyingLi
reportsthatfinancialsupportwasprovidedbytheHoustonMethodist
Hospital.
Dataavailability
Dataaresharedinthesupplementarydocuments
AppendixA. Supplementarydata
Supplementarydata tothisarticlecanbefoundonlineathttps://
doi.org/10.1016/j.scitotenv.2023.164435.
References
Alexander, P .,2013.Associationofmonthlyfrequenciesofdiversediseasesinthecallsto
thepublicemergencyserviceofthecityofBuenosAiresduring1999
–
2004with
meteorologicalvariablesandseasons.Int.J.Biometeorol.57(1),83
–
90.
Almendra, R .,etal.,2019.Short-termimpactsofairtemperatureonhospitalizationsfor
mentaldisordersinLisbon.Sci.TotalEnviron.647,127
–
133.
Balshem, H .,etal.,2011.GRADEguidelines:3.Ratingthequalityofevidence.J.Clin.
Epidemiol.64(4),401
–
406.
Barnston, A . G .,1988.Theeffectofweatheronmood,productivity,andfrequencyof
emotionalcrisisinatemperatecontinentalclimate.Int.J.Biometeorol.32(2),
134
–
143.
Basu, R .,etal.,2018.Examiningtheassociationbetweenapparenttemperatureand
mentalhealth-relatedemergencyroomvisitsinCalifornia.Am.J.Epidemiol.187(4),
726
–
735.
Beecher, M . E .,etal.,2016.Sunshineonmyshoulders:Weather,pollution,andemotional
distress.J.Affect.Disord.205,234
–
238.
Berry, H . L .,Bowen, K .,Kjellstrom, T .,2010.Climatechangeandmentalhealth:acausal
pathwaysframework.Int.J.PublicHealth55(2),123
–
132.
Bongioanni, P .,etal.,2021.Climatechangeandneurodegenerativediseases.Environ.
Res.201,111511.
Braga, A . L .,Zanobetti, A .,Schwartz, J .,2002.Theeffectofweatheronrespiratoryand
cardiovasculardeathsin12UScities.Environ.HealthPerspect.110(9),859
–
863.
Brown, R . D .,Gillespie, T . J .,1986.Estimatingoutdoorthermalcomfortusingacylindrical
radiationthermometerandanenergybudgetmodel.Int.J.Biometeorol.30(1),
43
–
52.
Budd, G . M .,2008.Wet-bulbglobetemperature(WBGT)
—
itshistoryanditslimitations.J.
Sci.Med.Sport11(1),20
–
32.
Bulbena, A .,etal.,2005.Panicanxiety,undertheweather?Int.J.Biometeorol.49(4),
238
–
243.
Bulbena, A .,etal.,2009.Impactofthesummer2003heatwaveontheactivityoftwo
psychiatricemergencydepartments.ActasEspanolasDePsiquiatria37(3),158
–
165.
Bullock, B .,Murray, G .,Meyer, D .,2017.Highsandlows,upsanddowns:meteorology
andmoodinbipolardisorder.PLoSOne12(3),e0173431.
Bundo, M .,etal.,2021.AmbienttemperatureandmentalhealthhospitalizationsinBern,
Switzerland:A45-yeartime-seriesstudy.PLoSOne16(10),e0258302.
Burdett, A .,Davillas, A .,Etheridge, B .,2021.Weather,mentalhealth,andmobilityduring
thefirstwaveoftheCOVID-19pandemic.HealthEcon.30(9),2296
–
2306.
Carney, P .,Fitzgerald, C .,Monaghan, C .,1988.Influenceofclimateontheprevalenceof
mania.Br.J.Psychiatry152(6),820
–
823.
Chan, E . Y .,etal.,2018.Associationbetweenambienttemperaturesandmentaldisorder
hospitalizationsinasubtropicalcity:atime-seriesstudyofHongKongspecial
administrativeregion.Int.J.Environ.Res.PublicHealth15(4),754.
Charlson, F .,etal.,2021.Climatechangeandmentalhealth:ascopingreview.Int.J.
Environ.Res.PublicHealth18(9),4486.
Chen, J .,Hoek, G .,2020.Long-termexposuretoPMandall-causeandcause-specific
mortality:asystematicreviewandmeta-analysis.Environ.Int.143,105974.
31

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Christensen, E . M .,etal.,2008.Climaticfactorsandbipolaraffectivedisorder.NordicJ.
Psychiatry62(1),55
–
58.
Cornali, C .,etal.,2004.Effectofhighclimatetemperatureonthebehavioraland
psychologicalsymptomsofdementia.J.Am.Med.Dir.Assoc.5(3),161
–
166.
Crane, J . D .,etal.,2015.Inhibitingperipheralserotoninsynthesisreducesobesityand
metabolicdysfunctionbypromotingbrownadiposetissuethermogenesis.Nat.Med.
21(2),166
–
172.
Cruz, J .,etal.,2020.Effectofextremeweathereventsonmentalhealth:anarrative
synthesisandmeta-analysisfortheUK.Int.J.Environ.Res.PublicHealth17(22),
8581.
Denissen, J . J .,etal.,2008.Theeffectsofweatherondailymood:amultilevelapproach.
Emotion8(5),662
–
667.
Ding, N .,Berry, H . L .,Bennett, C . M .,2016.Theimportanceofhumidityintherelationship
betweenheatandpopulationmentalhealth:evidencefromAustralia.PLoSOne11
(10),15.
DSM-IV-TR,2000.DiagnosticandStatisticalManualofMentalDisorders.American
PsychiatricAssociation.
Eun-hye, Y .,etal.,2021.Effectofextremetemperaturesondailyemergencyroomvisits
formentaldisorders.Environ.Sci.Pollut.Res.28(29),39243
–
39256.
Gál, C . V .,Kántor, N .,2020.Modelingmeanradianttemperatureinoutdoorspaces,A
comparativenumericalsimulationandvalidationstudy.UrbanClim.32,100571.
Gasparrini, A .,2011.Distributedlaglinearandnon-linearmodelsinR:thepackagedlnm.
J.Stat.Softw.43(8),1.
Gasparrini, A .,Armstrong, B .,2010.Timeseriesanalysisonthehealtheffectsof
temperature:advancementsandlimitations.Environ.Res.110(6),633
–
638.
Gasparrini, A .,Armstrong, B .,2013.Reducingandmeta-analysingestimatesfrom
distributedlagnon-linearmodels.BMCMed.Res.Methodol.13(1),1
–
10.
Gasparrini, A .,etal.,2012.Theeffectofhightemperaturesoncause-specificmortalityin
EnglandandWales.Occup.Environ.Med.69(1),56
–
61.
George, A . D .,Thompson,III, F . R .,Faaborg, J .,2015.UsingLiDARandremote
microclimateloggerstodownscalenear-surfaceairtemperaturesforsite-levelstudies.
RemoteSens.Lett.6(12),924
–
932.
Goldstein, K . M .,1972.Weather,mood,andinternal-externalcontrol.Percept.Mot.Skills
35(3),786.
Gong, J .,Part, C .,Hajat, S .,2022.Currentandfutureburdensofheat-relateddementia
hospitaladmissionsinEngland.Environ.Int.159,7.
Gu, S . H .,etal.,2019.Exposure-lag-responseassociationbetweensunlightand
schizophreniainNingbo,China.Environ.Pollut.247,285
–
292.
Guo, Y .,etal.,2018.Quantifyingexcessdeathsrelatedtoheatwavesunderclimatechange
scenarios:Amulticountrytimeseriesmodellingstudy.PLoSMed.15(7).
Guyatt, G . H .,etal.,2008.Whatis
“
qualityofevidence
”
andwhyisitimportantto
clinicians?Bmj336(7651),995
–
998.
Hansen, A .,etal.,2008.Theeffectofheatwavesonmentalhealthinatemperate
Australiancity.Environ.HealthPerspect.116(10),1369
–
1375.
Harrer, M .,etal.,2021.DoingMeta-analysisWithR:AHands-onGuide.CRCPress.
Hartig, T .,Catalano, R .,Ong, M .,2007.Coldsummerweather,constrainedrestoration,
andtheuseofantidepressantsinSweden.J.Environ.Psychol.27(2),107
–
116.
Haskell, E .,etal.,1981.Metabolismandthermoregulationduringstagesofsleepin
humansexposedtoheatandcold.J.Appl.Physiol.51(4),948
–
954.
Hayes, K .,etal.,2018.Climatechangeandmentalhealth:Risks,impactsandpriority
actions.Int.J.Ment.Heal.Syst.12(1),1
–
12.
Henriquez-Sanchez, P .,etal.,2014.Geographicalandclimaticfactorsanddepressionrisk
intheSUNproject.Eur.J.Pub.Health24(4),626
–
631.
Higgins, J . P .,Thompson, S . G .,2002.Quantifyingheterogeneityinameta-analysis.Stat.
Med.21(11),1539
–
1558.
Ho, H . C .,Wong, M . S .,2019.Urbanenvironmentalinfluencesonthe
temperature
–
mortalityrelationshipassociatedmentaldisordersand
cardiorespiratorydiseasesduringnormalsummerdaysinasubtropicalcity.Environ.
Sci.Pollut.Res.26(23),24272
–
24285.
Höppe, P .,1999a.Thephysiologicalequivalenttemperature
–
auniversalindexforthe
biometeorologicalassessmentofthethermalenvironment.Int.J.Biometeorol.43,
71
–
75.
Höppe, P .,1999b.Thephysiologicalequivalenttemperature
–
auniversalindexforthe
biometeorologicalassessmentofthethermalenvironment.Int.J.Biometeorol.43(2),
71
–
75.
Howarth, E .,Hoffman, M . S .,1984.Amultidimensionalapproachtotherelationship
betweenmoodandweather.Br.J.Psychol.75(FEB),15
–
23.
Huibers, M . J . H .,etal.,2010.Doestheweathermakeussad?Meteorologicaldeterminants
ofmoodanddepressioninthegeneralpopulation.PsychiatryRes.180(2
–
3),
143
–
146.
Jahan, S .,Wraith, D .,2021.Immediateanddelayedeffectsofclimaticfactorsonhospital
admissionsforschizophreniainQueenslandAustralia:atimeseriesanalysis.Environ.
Res.197,10.
Ji, Y .,Song, J .,Shen, P .,2022.Areviewofstudiesandmodellingofsolarradiationon
humanthermalcomfortinoutdoorenvironment.Build.Environ.214,108891.
Jia, P .,etal.,2020.Spatiallifecourseepidemiologyreportingstandards(ISLE-ReSt)
statement.HealthPlace61,102243.
Keller, M . C .,etal.,2005.Awarmheartandaclearhead-Thecontingenteffectsof
weatheronmoodandcognition.Psychol.Sci.16(9),724
–
731.
Khalaj, B .,etal.,2010.ThehealthimpactsofheatwavesinfiveregionsofNewSouth
Wales,Australia:acase-onlyanalysis.Int.Arch.Occup.Environ.Health83(7),
833
–
842.
Khreis, H .,etal.,2017.Exposuretotraffic-relatedairpollutionandriskofdevelopmentof
childhoodasthma:asystematicreviewandmeta-analysis.Environ.Int.100,1
–
31.
Kim, C . T .,etal.,2015.Heat-attributabledeathsbetween1992and2009inSeoul,South
Korea.PLoSOne10(2),e0118577.
Kim, S . Y .,etal.,2021.Short-andlong-termexposuretoairpollutionandlackofsunlight
areassociatedwithanincreasedriskofdepression:Anestedcase-controlstudyusing
meteorologicaldataandnationalsamplecohortdata.Sci.TotalEnviron.757,9.
Klimstra, T . A .,etal.,2011.Comerainorcomeshine:individualdifferencesinhow
weatheraffectsmood.Emotion11(6),1495
–
1499.
Kourtis, N .,Nikoletopoulou, V .,Tavernarakis, N .,2012.Smallheat-shockproteinsprotect
fromheat-stroke-associatedneurodegeneration.Nature490(7419),213
–
218.
Kwan, M .- P .,2012.Theuncertaingeographiccontextproblem.Ann.Assoc.Am.Geogr.
102(5),958
–
968.
Lee, H .- C .,Tsai, S .- Y .,Lin, H .- C .,2007.Seasonalvariationsinbipolardisorderadmissions
andtheassociationwithclimate:apopulation-basedstudy.J.Affect.Disord.97
(1
–
3),61
–
69.
Lee, H . J .,etal.,2002.Effectsofseasonandclimateonthefirstmanicepisodeofbipolar
affectivedisorderinKorea.PsychiatryRes.113(1
–
2),151
–
159.
Lee, M .,etal.,2018a.Weatherandhealthsymptoms.Int.J.Environ.Res.PublicHealth
15(8),15.
Lee, S .,etal.,2018b.Mentaldisease-relatedemergencyadmissionsattributabletohot
temperatures.Sci.TotalEnviron.616,688
–
694.
Li, B .,etal.,2018.Amodifiedmethodofevaluatingtheimpactofairhumidityonhuman
acceptableairtemperaturesinhot-humidenvironments.EnergyBuild.158,393
–
405.
Li, D .,etal.,2019.Subtypesofparkuseandself-reportedpsychologicalbenefitsamong
olderadults:Amultilevellatentclassanalysisapproach.Landsc.UrbanPlan.190,
103605.
Li, D .,etal.,2022a.Modelingtherelationshipsbetweenhistoricalredlining,urbanheat,
andheat-relatedemergencydepartmentvisits:anexaminationof11Texascities.
Environ.Plann.B.Plann.Des.49(3),933
–
952.
Li, M .,Ferreira, S .,Smith, T . A .,2020.Temperatureandself-reportedmentalhealthinthe
UnitedStates.PLoSOne15(3),e0230316.
Li, X .,etal.,2022b.Amelioratingcoldstressinahotclimate:EffectofWinterStormUri
onresidentsofsubsidizedhousingneighborhoods.Build.Environ.209,108646.
Libert, J .- P .,Bach, V .,2005.Thermoregulationandsleepinthehuman.In:The
PhysiologicNatureofSleep.ImperialCollegePress,London,UK,pp.407
–
431.
Liew, S . C .,Aung, T .,2021.Sleepdeprivationanditsassociationwithdiseases-areview.
SleepMed.77,192
–
204.
Lin, S .- H .,Lee, L .- T .,Yang, Y . K .,2014.Serotoninandmentaldisorders:aconcisereview
onmolecularneuroimagingevidence.Clin.Psychopharmacol.Neurosci.12(3),196.
Linares, C .,etal.,2017.Short-termassociationbetweenenvironmentalfactorsand
hospitaladmissionsduetodementiainMadrid.Environ.Res.152,214
–
220.
Liu, J .,etal.,2020.Cause-specificmortalityattributabletocoldandhotambient
temperaturesinHongKong:atime-seriesstudy,2006
–
2016.Sustain.CitiesSoc.57,
102131.
Liu, J .,etal.,2021.Isthereanassociationbetweenhotweatherandpoormentalhealth
outcomes?Asystematicreviewandmeta-analysis.Environ.Int.153,106533.
Liu, W .,Zhang, Y .,Deng, Q .,2016.Theeffectsofurbanmicroclimateonoutdoorthermal
sensationandneutraltemperatureinhot-summerandcold-winterclimate.Energy
Build.128,190
–
197.
Liu, X .,etal.,2019.Influenceofheatwavesondailyhospitalvisitsformentalillnessin
Jinan,China-acase-crossoverstudy.Int.J.Environ.Res.PublicHealth16(1),87.
Masson-Delmotte, V .,etal.,2021.Climatechange2021:thephysicalsciencebasis.In:
ContributionofWorkingGroupItotheSixthAssessmentReportofthe
IntergovernmentalPanelonClimateChange.p.2.
Mawson, D .,Smith, A .,1981.RelativehumidityandmanicadmissionsintheLondonarea.
Br.J.Psychiatry138,134
–
138.
McWilliams, S .,Kinsella, A .,O
’
Callaghan, E .,2013.Theeffectsofdailyweathervariables
onpsychosisadmissionstopsychiatrichospitals.Int.J.Biometeorol.57(4),497
–
508.
McWilliams, S .,Kinsella, A .,O
’
Callaghan, E .,2014.Dailyweathervariablesandaffective
disorderadmissionstopsychiatrichospitals.Int.J.Biometeorol.58(10),2045
–
2057.
Medina-Ramón, M .,etal.,2006.Extremetemperaturesandmortality:assessingeffect
modificationbypersonalcharacteristicsandspecificcauseofdeathinamulti-city
case-onlyanalysis.Environ.HealthPerspect.114(9),1331
–
1336.
Mendoza, J .,2019.Circadianinsightsintothebiologyofdepression:Symptoms,
treatmentsandanimalmodels.Behav.BrainRes.376,112186.
Middleton, J .,etal.,2021.TemperatureandplaceassociationswithInuitmentalhealthin
thecontextofclimatechange.Environ.Res.198,11.
Min, M .,etal.,2019.Effectofapparenttemperatureondailyemergencyadmissionsfor
mentalandbehavioraldisordersinYancheng,China:atime-seriesstudy.Environ.
Health18(1),1
–
12.
Moher, D .,etal.,2009.Reprint
—
preferredreportingitemsforsystematicreviewsand
meta-analyses:thePRISMAstatement.Phys.Ther.89(9),873
–
880.
Molin, J .,etal.,1996.Theinfluenceofclimateondevelopmentofwinterdepression.J.
Affect.Disord.37(2
–
3),151
–
155.
Montes, J .,Serrano, C .,Pascual-Sanchez, A .,2021.Theinfluenceofweatheronthecourse
ofbipolardisorder:asystematicreview.Eur.J.Psychiatry35(4),261
–
273.
Moola, S .,etal.,2017.JoannaBriggsInstituteReviewer
’
sManual.TheJoannaBriggs
Institute,p.5.
Mullins, J . T .,White, C .,2019.Temperatureandmentalhealth:Evidencefromthe
spectrumofmentalhealthoutcomes.J.HealthEcon.68,102240.
NationalHeart,Lung,andBloodInstitute,2014a.QualityAssessmentofCase-Control
Studies.
NationalHeart,Lung,andBloodInstitute,2014b.QualityAssessmentToolfor
ObservationalCohortandCross-sectionalStudies.
NationalWeatherService.Glossary.Availablefrom:https://w1.weather.gov/glossary/
index.php?letter=o.
Niu, Y .,etal.,2020.Short-termeffectofapparenttemperatureondailyemergencyvisits
formentalandbehavioraldisordersinBeijing,China:atime-seriesstudy.Sci.Total
Environ.733,139040.
32

CORRECTED PROOF
D. Li et al. Science of the Total Environment xxx (xxxx) 164435
Noelke, C .,etal.,2016.Increasingambienttemperaturereducesemotionalwell-being.
Environ.Res.151,124
–
129.
Obradovich, N .,etal.,2018a.Empiricalevidenceofmentalhealthrisksposedbyclimate
change.Proc.Natl.Acad.Sci.115(43),10953
–
10958.
Obradovich, N .,etal.,2018b.Empiricalevidenceofmentalhealthrisksposedbyclimate
change.Proc.Natl.Acad.Sci.U.S.A.115(43),10953
–
10958.
Obrien, L . V .,etal.,2014.Droughtasamentalhealthexposure.Environ.Res.131,
181
–
187.
Oh, S .,etal.,2020.Emergencydepartmentvisitsforpanicattacksandambient
temperature:atime-stratifiedcase-crossoveranalysis.Depress.Anxiety37(11),
1099
–
1107.
O
’
Hare, C .,etal.,2016.Seasonalandmeteorologicalassociationswithdepressive
symptomsinolderadults:Ageo-epidemiologicalstudy.J.Affect.Disord.191,
172
–
179.
OudinÅström, D .,etal.,2015.Theeffectofheatwavesonmortalityinsusceptiblegroups:
acohortstudyofamediterraneanandanorthernEuropeanCity.Environ.Health14
(1),30.
Palinkas, L . A .,Wong, M .,2020.Globalclimatechangeandmentalhealth.Curr.Opin.
Psychol.32,12
–
16.
Pan, R . B .,etal.,2019.Impactsofheatandcoldonhospitalizationsforschizophreniain
Hefei,China:anassessmentofdiseaseburden.Sci.TotalEnviron.694,8.
Peng, Z . X .,etal.,2017.Effectsofambienttemperatureondailyhospitaladmissionsfor
mentaldisordersinShanghai,China:atime-seriesanalysis.Sci.TotalEnviron.590,
281
–
286.
Persinger, M . A .,1975.Lagresponsesinmoodreportstochangesintheweathermatrix.
Int.J.Biometeorol.19(2),108
–
114.
Petkova, E . P .,etal.,2013.Projectedheat-relatedmortalityintheUSurbannortheast.Int.
J.Environ.Res.PublicHealth10(12),6734
–
6747.
Quach, T . T .,etal.,2016.Neuronalnetworksinmentaldiseasesandneuropathicpain:
beyondbrainderivedneurotrophicfactorandcollapsinresponsemediatorproteins.
WorldJ.Psychiatry6(1),18.
Radua, J .,Pertusa, A .,Cardoner, N .,2010.Climaticrelationshipswithspecificclinical
subtypesofdepression.PsychiatryRes.175(3),217
–
220.
Rataj, E .,Kunzweiler, K .,Garthus-Niegel, S .,2016.Extremeweathereventsindeveloping
countriesandrelatedinjuriesandmentalhealthdisorders-asystematicreview.BMC
PublicHealth16(1),1
–
12.
Rigby, A . S .,Vail, A .,1998.Statisticalmethodsinepidemiology.II:Acommonsense
approachtosamplesizeestimation.Disabil.Rehabil.20(11),405
–
410.
Rocklöv, J .,etal.,2014.Susceptibilitytomortalityrelatedtotemperatureandheatand
coldwavedurationinthepopulationofStockholmCounty,Sweden.Glob.Health
Action7(1),22737.
Rother, H .- A .,etal.,2021.ImpactofextremeweathereventsonSub-SaharanAfrican
childandadolescentmentalhealth:theimplicationsofasystematicreviewofsparse
researchfindings.J.Clim.Chang.Health,5,100087.
Salib, E .,Sharp, N .,1999.Doestheweatherinfluencedementiaadmissions?Int.J.
Geriatr.Psychiatry14(11),925
–
935.
Salib, E .,Sharp, N .,2002.Relativehumidityandaffectivedisorders.Int.J.Psychiatry
Clin.Pract.6(3),147
–
153.
Sarran, C .,etal.,2017.Meteorologicalanalysisofsymptomdataforpeoplewithseasonal
affectivedisorder.PsychiatryRes.257,501
–
505.
Schünemann, H . J .,etal.,2019a.Completing
‘
Summaryoffindings
’
tablesandgradingthe
certaintyoftheevidence.In:CochraneHandbookforSystematicReviewsof
Interventions.pp.375
–
402.
Schünemann, H . J .,etal.,2019b.GRADEguidelines:18.HowROBINS-Iandothertoolsto
assessriskofbiasinnonrandomizedstudiesshouldbeusedtoratethecertaintyofa
bodyofevidence.J.Clin.Epidemiol.111,105
–
114.
Schwingshackl, L .,etal.,2021.Evaluatingagreementbetweenbodiesofevidencefrom
randomisedcontrolledtrialsandcohortstudiesinnutritionresearch:meta-
epidemiologicalstudy.BMJ374.
Sharpe, I .,Davison, C . M .,2021.Climatechange,climate-relateddisastersandmental
disorderinlow-andmiddle-incomecountries:ascopingreview.BMJOpen11(10),
e051908.
Shaw, M . R .,etal.,2011.TheimpactofclimatechangeonCalifornia
’
secosystemservices.
Clim.Chang.109(1),465
–
484.
Sherbakov, T .,etal.,2018.Ambienttemperatureandaddedheatwaveeffectson
hospitalizationsinCaliforniafrom1999to2009.Environ.Res.160,83
–
90.
Shiloh, R .,etal.,2005.Effectsofclimateonadmissionratesofschizophreniapatientsto
psychiatrichospitals.Eur.Psychiatry20(1),61
–
64.
Shiue, I .,Perkins, D . R .,Bearman, N .,2016.Physicallyequivalenttemperatureandmental
andbehaviouraldisordersinGermanyin2009
–
2011.J.Ment.Health25(2),
148
–
153.
daSilva, I .,etal.,2020.Riskassessmentoftemperatureandairpollutantson
hospitalizationsformentalandbehavioraldisordersinCuritiba,Brazil.Environ.
Health19(1),1
–
11.
Simons, K . S .,etal.,2014.Effectofpreadmissionsunlightexposureonintensivecareunit-
acquireddelirium:Amulticenterstudy.J.Crit.Care29(2),283
–
286.
Son, J .,Shin, J .,2021.Bimodaleffectsofsunlightonmajordepressivedisorder.Compr.
Psychiatry108,6.
Steadman, R . G .,1984.Auniversalscaleofapparenttemperature.J.Appl.Meteorol.
Climatol.23(12),1674
–
1687.
Sterne, J . A .,etal.,2019.RoB2:arevisedtoolforassessingriskofbiasinrandomised
trials.Bmj366.
Sung, T . I .,etal.,2011.Relationshipbetweenmeandailyambienttemperaturerangeand
hospitaladmissionsforschizophrenia:Resultsfromanationalcohortofpsychiatric
inpatients.Sci.TotalEnviron.410,41
–
46.
Sung, T . I .,Chen, M . J .,Su, H . J .,2013.Apositiverelationshipbetweenambient
temperatureandbipolardisorderidentifiedusinganationalcohortofpsychiatric
inpatients.Soc.PsychiatryPsychiatr.Epidemiol.48(2),295
–
302.
Tang, C .,etal.,2021.Effectsofdifferentheatexposurepatterns(accumulatedand
transient)andschizophreniahospitalizations:atime-seriesanalysisonhourly
temperaturebasis.Environ.Sci.Pollut.Res.11.
Tapak, L .,etal.,2018.Investigatingtheeffectofclimaticparametersonmentaldisorder
admissions.Int.J.Biometeorol.62(12),2109
–
2118.
Thompson, R .,etal.,2018.Associationsbetweenhighambienttemperaturesandheat
waveswithmentalhealthoutcomes:asystematicreview.PublicHealth161,
171
–
191.
Trang, P . M .,etal.,2016a.Heatwavesandhospitaladmissionsformentaldisordersin
northernVietnam.PLoSOne11(5),e0155609.
Trang, P . M .,etal.,2016b.Seasonalityofhospitaladmissionsformentaldisordersin
Hanoi,Vietnam.Glob.HealthAct.9(1),32116.
Trenberth, K . E .,Fasullo, J . T .,Shepherd, T . G .,2015.Attributionofclimateextremeevents.
Nat.Clim.Chang.5(8),725
–
730.
Tsutsui, Y .,2013.WeatherandIndividualHappiness.WeatherClim.Soc.5(1),70
–
82.
Vaneckova, P .,Bambrick, H .,2013.Cause-SpecificHospitalAdmissionsonHotDaysin
Sydney,Australia.PLoSOne8(2),e55459.
Vergouwe, Y .,etal.,2005.Substantialeffectivesamplesizeswererequiredforexternal
validationstudiesofpredictivelogisticregressionmodels.J.Clin.Epidemiol.58(5),
475
–
483.
Vida, S .,etal.,2012.Relationshipbetweenambienttemperatureandhumidityandvisits
tomentalhealthemergencydepartmentsinQuebec.Psychiatr.Serv.63(11),
1150
–
1153.
Viechtbauer, W .,2010.Conductingmeta-analysesin{R}withthe{metafor}package.J.
Stat.Softw.36(3),1
–
48.
Voelkel, J .,etal.,2018.Assessingvulnerabilitytourbanheat:astudyofdisproportionate
heatexposureandaccesstorefugebysocio-demographicstatusinPortland,Oregon.
Int.J.Environ.Res.PublicHealth15(4),640.
Wang, S .,etal.,2018.Effectofincreasingtemperatureondailyhospitaladmissionsfor
schizophreniainHefei,China:atime-seriesanalysis.PublicHealth159,70
–
77.
Wang, X .,etal.,2014.Acuteimpactsofextremetemperatureexposureonemergency
roomadmissionsrelatedtomentalandbehaviordisordersinToronto,Canada.J.
Affect.Disord.155,154
–
161.
Weilnhammer, V .,etal.,2021.Extremeweathereventsineuropeandtheirhealth
consequences
–
Asystematicreview.Int.J.Hyg.Environ.Health233,113688.
Williams, S .,etal.,2012.HeatandhealthinAdelaide,SouthAustralia:assessmentofheat
thresholdsandtemperaturerelationships.Sci.TotalEnviron.414,126
–
133.
Wilson, L . A .,etal.,2013.TheimpactofheatonmortalityandmorbidityintheGreater
MetropolitanSydneyRegion:acasecrossoveranalysis.Environ.Health12(1),1
–
14.
WorldHealthOrganization,2002.PreventionandPromotioninMentalHealth.World
HealthOrganization.
WorldHealthOrganization,2020.RiskofBiasAssessmentInstrumentforSystematic
ReviewsInformingWHOGlobalAirQualityGuidelines.WorldHealthOrganization.
RegionalOfficeforEurope.
Wu, X .,etal.,2016.Impactofclimatechangeonhumaninfectiousdiseases:Empirical
evidenceandhumanadaptation.Environ.Int.86,14
–
23.
Xu, Y .,Wheeler, S . A .,Zuo, A .,2018.Willboys
’
mentalhealthfareworseunderahotter
climateinAustralia?Popul.Environ.40(2),158
–
181.
Xu, Z .,etal.,2019.Assessingheatwaveimpactsoncause-specificemergencydepartment
visitsinurbanandruralcommunitiesofQueensland,Australia.Environ.Res.168,
414
–
419.
Yackerson, N . S .,etal.,2011.Theinfluenceofseveralchangesinatmosphericstatesover
semi-aridareasontheincidenceofmentalhealthdisorders.Int.J.Biometeorol.55
(3),403
–
410.
Yi, W .,etal.,2019.Examiningtheassociationbetweenapparenttemperatureand
admissionsforschizophreniainHefei,China,2005
–
2014:atime-seriesanalysis.Sci.
TotalEnviron.672,1
–
6.
Yi-Fan, P .,etal.,2016.Analyzingpersonalhappinessfromglobalsurveyandweather
data:ageospatialapproach.PLoSOne11(4),e0153638.
Yoo, E . H .,etal.,2021.Associationbetweenextremetemperaturesandemergencyroom
visitsrelatedtomentaldisorders:Amulti-regiontime-seriesstudyinNewYork,USA.
Sci.TotalEnviron.792,11.
Zammit, C .,etal.,2021.Neurologicaldisordersvis-à-visclimatechange.EarlyHum.Dev.
155,105217.
Zapata, O .,2021.Happinessinthetropics:climatevariablesandsubjectivewellbeing.
Environ.Dev.Econ.22.
Zhang, S . Y .,etal.,2020.Theeffectoftemperatureoncause-specificmentaldisordersin
threesubtropicalcities:acase-crossoverstudyinChina.Environ.Int.143,7.
Zhao, Q .,Lian, Z .,Lai, D .,2021.Thermalcomfortmodelsandtheirdevelopments:A
review.EnergyBuiltEnviron.2(1),21
–
33.
33