PrevalenceofallergicsensitizationintheUnitedStates:Results from
the National Health and Nutrition
ExaminationSurvey(NHANES)2005-2006P aivi M. Salo, PhD,aSamuel J.
Arbes, Jr, DDS, MPH, PhD,bRenee Jaramillo, MStat,cAgustin
Calatroni, MA, MS,bCharles H. Weir, MS, MPH, PhD,dMichelle L.
Sever, MPH, PhD,bJane A. Hoppin, ScD,aKathryn M. Rose,
PhD,cAndrewH. Liu, MD,fPeter J. Gergen, MD, MPH,eHerman E.
Mitchell, PhD,band Darryl C. Zeldin, MDaResearch TrianglePark,
Chapel Hill, and Durham, NC, Bethesda, Md, and Denver and Aurora,
ColoBackground:Allergicsensitizationisanimportantriskfactorfor
thedevelopmentof atopic disease.The
NationalHealthandNutritionExaminationSurvey(NHANES)2005-2006providesthemostcomprehensiveinformationonIgE-mediatedsensitizationinthegeneralUSpopulation.Objective:Weinvestigatedclustering,sociodemographic,andregionalpatternsofallergicsensitizationandexaminedriskfactorsassociatedwithIgE-mediatedsensitization.Methods:Dataforthiscross-sectionalanalysiswereobtainedfromNHANES2005-2006.Participantsaged1yearorolder(n59440)weretestedforserumspecicIgEs(sIgEs)toinhalantandfoodallergens;participants6yearsorolderweretestedfor19sIgEs,andchildrenaged1to5yearsweretestedfor9sIgEs.SerumsampleswereanalyzedbyusingtheImmunoCAPSystem.Informationondemographicsandparticipantscharacteristicswascollectedbymeansofquestionnaire.Results:
Of the study population aged 6 years and older,
44.6%haddetectablesIgEs,whereas36.2%ofchildrenaged1to5yearsweresensitizedto1ormoreallergens.Allergen-specicIgEsclusteredinto7groupsthatmighthavelargelyreectedbiologicalcross-reactivity.Althoughsensitizationtoindividualallergensandallergentypesshowedregionalvariation,theoverallprevalenceofsensitizationdidnotdifferacrosscensusregions,exceptinearlychildhood.Inmultivariatemodelingyoungage,malesex,non-Hispanicblackrace/ethnicity,geographic
location (census region), and reported pet
avoidancemeasuresweremostconsistentlyassociatedwithIgE-mediatedsensitization.Conclusions:
The overall prevalence of allergic sensitization doesnot vary
across US census regions, except in early life,
althoughallergen-specic sensitization differs based on
sociodemographicand regional factors. Biological cross-reactivity
might be animportant but not the sole contributor tothe
clusteringof allergen-specic IgEs. (J Allergy Clin Immunol
2014;134:350-9.)Key words: Allergen, allergy, allergic
sensitization, serum IgEThe increased prevalence of allergic
diseases is a major publichealth concern worldwide.1In the United
States millions ofpersonsareaffectedbyIgE-mediateddiseases,
whichnotonlyaffect thequalityoflifebut
alsoplaceconsiderableeconomicburden on patients and
society.2,3Thecommonhallmarkofatopicdiseaseistheproductionofserum
specic IgE (sIgE) against allergens. Assessment of
sIgEantibodieswithin vivoskintestchallengesorin
vitroserologicanalyses conrms allergicsensitization, whereas the
patientsclinical history and physical examination remain
importantcornerstones of the diagnosis of atopic
disease.4Monitoringthe prevalence and patterns of IgE-mediated
sensitization inpopulations over time is important because allergic
sensitizationis a signicant risk factor for the development of
atopic disease.1In the United States the prevalence of allergic
sensitization in thegeneral population has been estimated in 3
National Health
andNutritionExaminationSurveys(NHANESs).5-8InNHANESII(1976-1980)andIII(1988-1994)allergytestingwasconductedby
using skin prick tests, whereas NHANES 2005-2006
measuredsIgElevelsinserum.NHANES2005-2006providesthelargestandmostrecentnationallyrepresentativedataonIgE-mediatedsensitizationinthe
USpopulation. Participants aged6yearsandolder weretestedfor
19sIgEantibodies, andthoseaged1 to 5 years were tested for a subset
of the antibodies (9 sIgEs).NHANES 2005-2006 not only tested a
greater number ofallergens across awider agerangethantheprior
studies,
butalsoprovidedquantitativeinformationontheextentofallergicsensitization.This
articleprovides acomprehensivereport onclustering,sociodemographic,
and regional patterns of allergic sensitizationin the US
population. Patterns of sensitization in NHANES2005-2006 were
compared with NHANESIII data. We alsoidentiedfactors
independentlyassociatedwithIgE-mediatedsensitization in the general
population.Fromathe Division of Intramural Research, National
Institute of Environmental HealthSciences, National Institutes of
Health, Research Triangle Park;bRho Federal SystemsDivision, Chapel
Hill;cSocial &ScienticSystems, Durham;dtheDepartment
ofEnvironmentalSciencesandEngineering,GillingsSchoolofGlobalPublicHealth,UniversityofNorthCarolina,
Chapel Hill;etheDivisionofAllergy,
Immunology,andTransplantation,NationalInstituteofAllergyandInfectiousDiseases,NationalInstitutesofHealth,Bethesda;andfNationalJewishHealth,Denver,andUniversityof
Colorado School of Medicine, Aurora.Supportedinpart bytheIntramural
ResearchProgramof theNational InstitutesofHealth, National
Institute of Environmental Health Sciences.Disclosureofpotential
conict ofinterest: S. J. Arbeshasreceivedagrant fromtheNational
InstituteofAllergyandInfectiousDisease/DivisionofAllergy,
Immuno-therapy,andTransplantation/NationalInstitutesofHealth.A.H.LiuisontheDataSafetyMonitoringCommittee
for GlaxoSmithKline, has consultant arrangementswith DBV, and has
received payment for lectures from Merck. The rest of the
authorsdeclare that they have no relevant conicts of interest.The
ndings and conclusions in this article are those of the author or
authors and do notnecessarily represent the views of the Research
Data Center, the NCHS, or the Centersfor Disease Control and
Prevention.Received for publication May 9, 2013; revised December
6, 2013; accepted for publica-tion December 11, 2013.Available
online February 9, 2014.Correspondingauthor: Darryl C. Zeldin, MD,
NIEHS/NIH, 111T.W. Alexander Dr,Rm A214, Research Triangle Park, NC
27709. E-mail:
[email protected]://dx.doi.org/10.1016/j.jaci.2013.12.1071350Abbreviations
usedGM: Geometric meanNCHS: National Center for Health
StatisticsNHANES: National Health and Nutrition Examination
SurveySES: Socioeconomic statussIgE: Serum specic
IgEMETHODSDataData were obtained fromNHANES 2005-2006, which is
designed to assessthe health and nutritional status of the
civilian, noninstitutionalizedUS population. NHANES 2005-2006,
which includes 10,348 subjects,oversampled persons of low income,
adolescents aged 12 to 19 years, persons60yearsofageandolder,
AfricanAmericans, andMexicanAmericanstoensure adequate samples for
subgroup analyses. All data analysis withrestricted and not
publicly available variables (census region, level ofurbanization)
was conductedat the National Center for HealthStatistics(NCHS)
Atlanta Research Data Center to protect
participantcondentiality.Thesurveyprotocol was
approvedbytheNCHSEthics ReviewBoard,9and written informed consent
was obtained from all participants.Adetailed
descriptionofthesurveydesignandmethodsisavailableonlineat
http://www.cdc.gov/nchs/nhanes/nhanes2005-2006/nhanes05_06.htm.10Toexaminesensitizationpatternsovertime,weusedNHANESIIIdataforcomparisons.
Study procedures and methods for the NHANES III data havebeen
previously
described.7Assessmentofallergen-specicIgEsandatopyParticipantsaged1yearandolderwereeligibleforserumIgEtesting.Blood
samples were analyzed with the Pharmacia Diagnostics ImmunoCAP1000
System (Kalamazoo, Mich), now known as Thermo
ScienticImmunoCAPSpecicIgE.
Participantsaged6yearsandolderweretestedforallergen-specicIgEantibodiesto15inhalant(ie,
indoorandoutdoor)allergens (Alternaria alternata, Aspergillus
fumigatus, Bermuda grass[Cynodon dactylon], birch [Betula
verrucosa], cat dander, cockroach[Blatella germanica], dog dander,
dust mite [Dermatophagoides
farinaeandDermatophagoidespteronyssinus],mouseurineproteins,oak[Quercusalba],
ragweed [Ambrosia elatior], rat urine proteins, Russian thistle
[Salsolakali], and rye grass [Lolium perenne]) and 4 food allergens
(egg white, cowsmilk, peanut [Arachis hypogaea], and shrimp
[Pandalus borealis]).
Becauseofsmallerquantitiesofserum,theIgEpanelforchildrenlessthan6yearsof
age was limited to 9 allergens (Alternaria alternata, cat dander,
cockroach[Blatella germanica], dog dander, dust mite [D farinae and
D pteronyssinus],egg white, cows milk, and peanut [Arachis
hypogaea]). Of the participants6yearsofageandolder(n58086),
89.9%(n57268)haddataforall19specicIgEs,
andofthechildrenaged1to5years, 63.2%(n5856)had complete data for
the partial IgE
panel.Thethresholdforapositivetestresult,whichwasconsideredindicativeofsensitization,wasthelevelofdetection(>_0.35kUA/L).Subjectswithatleast
1 positive sIgE result were considered atopic. A detailed
description ofthe laboratory procedures is presented
elsewhere.11,12StatisticalanalysesDescriptiveanalysesandpredictivemodelingwereperformedwithSAS(version9.2;
SASInstitute, Cary, NC) andSUDAAN(version10.0;
RTIInternational,ResearchTrianglePark,NC).Clusteranalysiswasconductedwith
the Rsystemfor statistical computing (version 2.10.1). To account
for thecomplex survey design, survey nonresponse, and
poststratication, samplingweights and design variables were applied
to all analyses (NHANES2005-2006, NHANES III) to obtain unbiased
national prevalence and varianceestimates.Participantsaged6years
andolder andthoseaged1to5years wereanalyzed separately because of
the differences in IgE panels. Differences inthe prevalence of
positive allergen-specic IgE test results by generalpopulation
characteristics were tested with x2statistics. Among
atopicsubjects, weusedFstatisticstotest
differencesingeometricmean(GM)concentrations of sIgEs across
populationcharacteristiccategories. Whenevaluating racial/ethnic
differences, we focused on the 3 main groups:non-Hispanicwhite,
non-Hispanicblack, andMexicanAmericansubjects,excluding the group
others because of racial/ethnic heterogeneity. Toexamine clustering
of sIgEs, we used different statistical methods,
includinghierarchic clustering, factor analysis, and
multidimensional scaling.We used logistic regression analysis to
identify predictors of specic IgEpositivity (ie, sensitization) in
the general population. For the predictormodeling,
weusedsociodemographicandother characteristics
associatedwithskintest
positivityinNHANESIII.7Wealsoincludedreportedpetavoidance measures
because avoidance, removal, or both of pets because ofallergyor
asthmawas stronglyassociatedwiththeoverall
prevalenceofsensitization (data not shown). In the full model we
includedage, sex,race/ethnicity, poverty/incomeratio, education,
serumcotininelevel, bodymass index, year of home construction,
census region, level of urbanization,number of household members,
household crowding, presence of cats and/ordogs in the home, and
reported pet avoidance and used backward
eliminationformodelselection.
AlloftheremainingpredictorsinthenalmodelhadPvalues of .05 or less.
We also evaluated potential 2-way interactions betweenage, sex, and
race/ethnicity, but no strong evidence for effect modication
wasfound.Sensitivityanalyseswereconductedtodeterminewhetherthemodelingresults
were inuenced by using a different cut point for a positive serum
IgEtest result. We also investigated whether similar risk factors
were associatedwith increased sIgE levels among atopic subjects. We
considered specic IgElevels to be increased if any of the sIgE
concentrations exceeded 17.5 kUA/Lamong subjects aged 6 years and
older and 3.5 kUA/L among the 1- to 5-year-old subjects. The
cutoffs were not based on clinical relevance; however,
theydistinguishedthosewiththehighest sIgElevels(_ 0.78). Of those
aged 6 years and older,27.1%were sensitized to plant-related
(grass-, tree-, weed-,and/or peanut-specic) allergens; 20.3% to
dust mites (D farinaeand/or Dpteronyssinus); 15.7% to pets (dog
and/or cat); 11.7% tocockroachand/or shrimp;10.4% tomolds
(Alternariaalternataand/orAspergillusfumigatus);6.6%toeggwhiteand/orcowsmilk;
and1.8%torodents(mouseand/orrat). Thepercentageof the population
with positive cluster responses decreased withthe increasing number
of the clusters. Among those aged 6 yearsand older, 19.8%had
positive responses originating from1cluster, 10.8% from 2 clusters,
7.4% from 3 clusters, 3.9% from4 clusters, 1.8%from5 clusters,
0.6%from6 clusters, and0.2% from 7
clusters.DemographiccharacteristicsofallergicsensitizationTheprevalenceofallergicsensitizationdifferedsignicantlybyageforeachofthetestedallergensexceptrat.
Forinhalantallergens, the prevalence of sensitization generally
peaked in thesecond or third decades of life and decreased toward
lateadulthood (Fig 3). In contrast, sensitization to milk and egg
wasmostprevalentamongsubjectsaged1to5yearsbutdecreasedsharply after
age 6 years throughout childhood.Overall, the prevalence of
allergic sensitization was higher inmale than in female subjects
(Fig 4),butthe difference was notalways present at each age
category (see Fig E1 in this articlesOnline Repository at
www.jacionline.org). Among the youngestagegroup(1-5yearsofage),
thegreatestsexdifferenceswereseen for pet allergens, especially for
cat.The prevalence of allergic sensitization to all types of
allergensvaried signicantly by race/ethnicity (Fig 4). Among those
aged6yearsandolder,
non-Hispanicblacksubjectshadthehighestprevalence of sensitization
to all tested allergens, except forRussianthistle andegg.
Amongsubjects aged1to5years,allergic sensitization tended to be
less prevalent in non-Hispanic white subjects than in other
race/ethnic groups,although differences did not always reach
statistical signicancefor individual sIgEs. Fig E2 in this articles
Online Repository atwww.jacionline.org shows race/ethnic variation
by age.Socioeconomic patterns of allergic sensitization were
lessconsistent (Fig 4). However, sensitizations to German
cockroachand shrimp were associated with indicators of lower
socio-economic status (SES), whereas
sensitizationtodogandcatallergens was more prevalent among higher
SESgroups.
AlthoughsensitizationtooutdoorallergenswasmoreprevalentinhigherSES
groups, most differences were not statistically signicant.FIG1.
Prevalence of positive serumIgE test results and distributions of
allergen-specic antibodyconcentrations in theUS population.A andB,
Subjectsaged 6 years and older. C andD, Subjectsaged 1to5years.
Boxplotsdisplayminimumandmaximumvaluesand25th, 50th,
and75thpercentiles. Themeasuring range was 0.35 to 1000 kUA/L.
LLOD, Lower limit of detection; ULOD, upper limit of detection.J
ALLERGY CLIN IMMUNOLAUGUST 2014352SALO ET
ALDifferencesinallergicsensitizationacrosslevelsofurbanizationAmong
those aged 6 years and older, sensitization to allergenswas more
prevalent in metropolitan areas than in less urbanizedareas (Fig
4). In central metropolitan areas (population >_1million)
50.0%of thepopulationwas sensitizedtoat least 1allergen, whereas
innonmetropolitanareas the prevalence ofsensitization was less than
40.0%(see Table E1). The mostconsistent
associationswereobservedforoutdoorallergens; inlarge central
metropolitan areas, 37.8%were sensitized tooutdoor allergens,
whereas less than one fourth (22.9%to24.9%) of the nonmetropolitan
population was sensitized tooutdoor allergens. Among subjects aged
1 to 5 years, sensitizationto all types of allergens was more
prevalent in large metropolitanareas than in other areas, but none
of the differences werestatistically signicant (see Table
E3).DifferencesinallergicsensitizationacrosscensusregionsAlthough
the overall prevalence of sensitization did not differsignicantly
by census region, there were regional differences forindividual
allergens and allergen types (Fig 5 and see Tables E1and E3).
Sensitization to indoor allergens was most prevalent inthe South
(34.5%among those aged 6 years and older and26.5% among those aged
1 to 5 years), particularly to dust miteand cockroach allergens.
For most outdoor allergens, theprevalence of sensitizationwas
highest inthe West (36.0%),especiallyfor Russianthistle (17.9%)
andgrasses (21.4%to25.0%). Among those aged 6 years and older,
sensitizationto food allergens was most prevalent in the South. Of
thefood-specicIgEs, onlypeanut inthethose6yearsandoldershowed
regional variation (see Tables E1 and
E3).Concentrationsofallergen-specicIgEThe GMs of the sIgE
concentrations ranged from 0.74
kUA/L(egg)to3.89kUA/L(ryegrass)amongpersonswithapositivetestresulttothespecicallergen(seeTablesE2andE4).sIgElevels
tended to decrease with increasing age group after
peakinginchildhood andearlyadulthood. Themajorityof theGMsdidnot
differ signicantly by sex. Among those aged 6
yearsandolder,non-HispanicblacksubjectsdidnotalwayshavethehighestlevelsofsIgE,
buttheGMsofdustmite, cockroach-,mold-, and shrimp-specic IgEs were
signicantly higher innon-Hispanic black subjects than in other
groups. Amongsubjectsaged1to5years,
onlycockroach-specicIgElevelswere lower in non-Hispanic white
subjects than in other groups.FIG2. Clusteringof
allergen-specicIgEs. Thedendrogram(right)
useshierarchicclusteringtogroupsimilarsIgEsbasedonthelog-transformedcorrelationmatrix(left).Hierarchicclusteranalysisidentied7IgEclusters:
(1) plants(grass-, tree-, weed-, andpeanut-specicIgEs); (2) dust
mites(DfarinaeandDpteronyssinus); (3) pets(dogandcat); (4)
cockroachandshrimp; (5) molds(AlternariaalternataandAspergillus
fumigatus); (6) foods (egg white and cows milk); and (7) rodents
(mouse and rat).J ALLERGY CLIN IMMUNOLVOLUME 134, NUMBER 2SALO ET
AL353Across all age groups, lower SES was consistently associated
withhigher cockroach-specic IgE levels. Although the level
ofurbanizationwasnotassociatedwithsIgElevelsinearlyyearsof life,
except for peanut, levels of several sIgEs (ie, pet-,cockroach-,
weed-, tree-, mold-, shrimp-, andpeanut-specicIgEs) in the older
age groups varied signicantly across urbaniza-tionlevels.
Variationacross census regions was apparent forindoor allergen and
outdoor allergenspecic IgE concen-trations but not for food-specic
IgE levels.PredictorsofallergicsensitizationTableI showsparticipant
characteristicsthat wereindepen-dentlyassociatedwith1or
morepositivetest responses.
TheoddsofspecicIgEpositivityincreasedthroughout
childhoodandearlyadulthoodandwereconsistentlyhigheramongmalesubjects
(except in early years of life), among non-Hispanic
blacksubjects,and among those who reported avoidance, removal,
orbothofpets. AlthoughtheoddsofspecicIgEpositivitywerehighest in
the South in early childhood, no signicant regional dif-ferences
were found in older age groups. Age, sex, race/ethnicity,and census
region were also consistently associated withincreased sIgE levels
among atopic subjects (data not
shown).ComparisonsbetweenNHANES2005-2006andNHANESIIIIn NHANES
2005-2006 grass, dust mite, and ragweedallergens were the most
common sensitizers among subjectsaged 6 years or older, which is
consistent with NHANES III.7FigsE3 to E5 in this articles Online
Repository at www.jacionline.orgshowsimilar patterns of allergic
sensitization by age, sex,and race/ethnicity in both surveys. In
contrast, patterns ofsensitizationacross census regions
showedgreater variabilityin NHANES 2005-2006; regional variation
was observedforall
typesofallergens(seeTablesE1andE3)inNHANES2005-2006, whereas only
sensitization to outdoor allergens
variedbycensusregioninNHANESIII.7Theoverall
prevalenceofsensitizationdidnot varyacross census regions
inNHANES2005-2006, as it did in NHANES
III.DISCUSSIONNHANES2005-2006providesthemostdetailedinformationonIgE-mediatedsensitizationintheUSpopulation,
includingquantitative information on the extent of allergic
sensitization. Itis the rst nationwide study to examine allergic
sensitization fromearly years of life (children aged 1-5 years) to
old age (aged >_75years). Almost half of the US population aged
6 years and olderand more than one third of children aged 1 to 5
years had positivesIgEresponsestoat least 1allergen.
InNHANES2005-2006allergen-specic IgEs clustered into 7 biologically
relevantgroups. Across the United States, sensitization to
individualallergens and allergen types showed regional variation,
whereastheoverall prevalenceofsensitizationdidnot, except
inearlyyears of life. Inmultivariate models age, sex,
race/ethnicity,census region, and reported pet avoidance measures
wereconsistently associated with IgE-mediated sensitization,
althoughsex differences were less pronounced in early life.The
NHANES 2005-2006 data offered a unique opportunity toexplore
clustering of sIgEs in the general US population.Although
IgE-mediated allergy is known to aggregate in clustersof
allergens,15,16fewpopulation-basedstudies
haveexaminedclusteringpatterns. Structural
similarityofallergenicepitopes,the major determinant of
cross-reactivity,17might be a keycontributor to the clustering.
However, sIgE patterns might alsoreect a subjects propensity to
coexposure/sensitization thatcan occur without sequence homology
and/or commonepitopes (eg, egg and
milksIgEs).17,18Asexpected,thehighestcorrelations were observed for
dust mite, grass-, andtree-specic allergens, which share groups of
cross-reactiveallergens.17The clustering of cockroach and
shrimp-specicFIG3. Prevalenceof positiveserumIgEtest
resultsbyageamongtheUSpopulation. Solidsymbolsrepresent
statisticallysignicant differences, whereasprevalencesthat arenot
statisticallydifferent aremarked with open symbols. Red symbols
indicate the subgroup of allergens measured in 1- to
5-year-oldchildren.J ALLERGY CLIN IMMUNOLAUGUST 2014354SALO ET
ALIgEs might reect indirect sensitization to a
cross-reactiveallergen, tropomyosin. In a recent study the
correlation betweencockroach- and shrimp-specic allergens among
inner-citychildrenwasassociatedwithresidentialexposuretocockroachallergen.19Althoughpeanut
is traditionallyconsideredafoodallergen,
itwasnotsurprisingthatpeanut-specicIgEgroupedtogether with
plant-specic IgEs because peanut allergens showextensive IgE
cross-reactivity between homologous pollenallergens.20,21Studies
have shown that peanut-sensitized subjectsare often also sensitized
to pollens.21,22Consistent with
previousreports,21,22sensitizationtoplant-relatedallergens was
highlyprevalent among peanut-sensitized NHANES
2005-2006participants. The vast majority of those with positive
testresponsestopeanut
weresensitizedtograss-specic(88.7%),tree-specic(88.6%),
orweed-specic(90.6%)allergens(datanot shown). However, a large
proportion of serologic cross-reactivity might not be clinically
relevant. For example, a
recentstudyshowedthatchildrensensitizedtobothpeanutandbirchwerelesslikelytoreport
symptomstopeanut thanthosewhowere sensitized to peanut but not to
birch pollen.21InNHANES2005-2006thepatternsof
sensitizationacrosssociodemographic factors were similar toprevious
NHANESreports.7,8IgE-mediated sensitization was strongly
associatedwith age. The prevalence and levels of inhalant
allergen-specicIgEs peaked in childhood and early adulthood,
whereassensitization to foods (ie, milk and egg) was most prevalent
beforeage 6 years.13The decreasing trends in the prevalence
ofsensitizationandsIgElevelsinolderagegroupsmight
reectchangesintheimmunesystemthat accompanyaging.
Studieshavedemonstratedage-associatedalterations inmanyaspectsof
immune function (eg, T- and B-cell functions),
althoughimmunosenescence might vary by atopic
phenotype.23,24Sexdifferences inprevalencearecommoninpatients
withatopic disorders.25Although sex-specic differences in
immuneresponses emerge in early childhood, the clinical expression
anddiagnosticmarkersofatopicdiseaseareinuencedbygenetic,hormonal,
and environmental factors throughout
life.26,27InNHANES2005-2006male subjects were more likely to have
posi-tive sIgE test results, as well as increased sIgE levels, than
femalesubjects, except among subjects aged 1 to 5 years. Some
studieshavesuggestedthatdifferencesintheproductionofIgEmightbe
associated with sex-related genetic differences in IgEcontrol.28The
NHANES 2005-2006 data demonstrated racial/ethnicdifferences
frominfancy. Amongsubjects aged1to5years,FIG 4. Prevalenceof
positive serum IgE testresultsby demographic characteristics in the
US population.A, Subjectsaged6yearsandolder. B,
Subjectsaged1to5years. Solidsymbolsrepresentstatisticallysignicant
differences, whereas prevalences that arenot statisticallydifferent
aremarkedwithopensymbols.J ALLERGY CLIN IMMUNOLVOLUME 134, NUMBER
2SALO ET AL355sensitization to allergens, particularly
sensitization to foodallergens that often precedes the progression
of allergicdiseases,29was signicantly lower in non-Hispanic white
subjectsthanother race/ethnicgroups. Amongthoseaged6yearsandolder,
non-Hispanic black subjects were most likely to havepositive test
responses and to have increased sIgE levels.Although several
studies have reported racial and ethnicdisparities intheprevalence
of allergicsensitization, thereislimitedinformationonhowgenetic,
socioeconomic, cultural,environmental, or other factors contribute
to the observeddisparities.30Arecent studysuggestedthat theracial
disparitymight primarilyresult fromenvironmental rather
thangeneticfactors.31Although many large epidemiologic studies have
reportedsubstantial
geographicvariationintheprevalenceofsensitiza-tion,7,32,33the
overall prevalence of allergic sensitization did
notvaryregionallyinNHANES2005-2006,exceptinearlylife.Inpopulation-basedstudiesatopyisoftenassessedbyusingtestpanelslimitedtoinhalant
allergens,32,33whereasinNHANES2005-2006thepanel
consistedofavarietyofindoor, outdoor,andfoodallergens. Theuseof
alarger test panel might haveinuenced the sensitivity of the
assessment of allergic sensitiza-tion. Even though the majority of
outdoor allergens showedsignicant regional variability in
sensitization patterns, the panelincluded several allergens (eg,
pets, rodents, ragweed, molds, andmost foods) that did not show
variation by region. In
NHANES2005-2006chancesofcapturingmonosensitizedsubjectsmighthave
been increased because egg and milk were the most commonsensitizers
among monosensitized participants (data not shown).In contrast, in
NHANES III the regional variation in theoverall prevalenceof
sensitizationmight havereectedhighersensitizationrates tooutdoor
allergens intheWesternUnitedStates. The majority of the tested
allergens in NHANESIIIwereoutdoorallergens,ofwhichmany
hadsignicantlyhighersensitizationratesintheWest(eg,grassesandRussianthistle).However,
the overall prevalence of sensitization might
varygeographicallyinearlylife.Among1-to5-year-oldNHANES2005-2006
participants, those living in the South were most
likelytohavepositivesIgEresponses.
Differencesinindoorallergensensitizationratesmight
contributetotheregional variationinthis age group because
sensitization to dust mite and
cockroachallergenswassignicantlyhigherintheSouthcomparedwiththe
other census regions. Althoughregional differences werealso
observed in sIgE levels, caution is warranted
wheninterpretingtheresults of theyoungest
agegroupbecauseofhighlyvariable sIgEconcentrations anda small
samplesize.Amongatopicsubjects aged6years andolder,
grass-specicIgEs reached particularly high levels in the West.Pet
avoidance, reported avoidance, and/or removal of pets fromthe home
because of allergies or asthma were strongly
associatedwithpositivesIgEresponsesandincreasedspecicIgElevels.Althoughtherelationshipbetweenthedevelopment
of atopicdisorders and exposure to pets and pet allergens
remainscontroversial,34-36pet allergens are triggers for many
atopicsubjects and caninuence disease
severityamongsensitizedpatients.37-39Subjects or families with
allergies often avoidowning pets, either because pets trigger
symptoms or as aprecaution. Studies report that higher parental
awareness of riskfactorsfor
allergicdiseasecaninuencethelikelihoodof
petavoidance.40Becausethisstudyiscross-sectional, participantsmight
have modied their behavior because of their atopic status.Thus the
interpretation of pet-related associations is complicated.FIG 5.
Prevalence of positive serum IgE test results by census region in
the US population. A, Subjects aged6 years and older. B, Subjects
aged 1 to 5 years.J ALLERGY CLIN IMMUNOLAUGUST 2014356SALO ET
ALAlthoughassociationsareadjustedforpetavoidancemeasures,the data
might not have captured avoidance measures throughoutlife. The
inverse association between sensitization and thepresence of a cat
or cats might be partially explained by reversecausation;
sensitization precedes and causes the absence of
cats.Atopicsubjects might
havebeenmorelikelytoremovecatsthandogsfromthehome.
InNHANES2005-2006only34.0%of those who removed a cat from the home
in the past 12
monthsweresensitizedtocatallergen,whereasalmosthalf(48.4%)ofthose
whoremovedadogweresensitizedtodogallergen(datanot shown). Studies
suggest that selectiveavoidance is oftenmore common to cats than
dogs.40,41Although several studies have reported an inverse
associationbetween smoking and atopy,7,32,42the association of
smoking onallergic sensitization has been inconsistent in the
literature.43Theinverseassociationmight either reect
selectiveavoidanceofsmoking or the immunosuppressive effects of
tobacco smoke.42,44Some studies report that those with asthma and
allergies
inchildhoodarelesslikelytoinitiatesmokingthantheirhealthypeers.44Allergic
sensitization was more prevalent in large
centralmetropolitanareascomparedwithother areas.
Supportingtheexisting literature,8,45differences were most
consistent foroutdoor allergens. Level of urbanization was not an
independentpredictor in logistic regression models, perhaps because
of lesservariation of other types of sIgEs. Microclimatic
differencesassociated with urbanization, most notably a CO2
concentrationand temperature increase, have been associated with
fastergrowth, earlier owering, and greater aboveground biomass
andpollen production of plants (eg, ragweed) in urban
areascomparedwithrural locations.46Numerousstudiesreport
thatplant-related aeroallergen exposures are inuenced by
increasingCO2concentrations, climatechange, orboth,
leadingtolongerpollen seasons, greater pollen loads, and changes in
allergenicityand spatial distributions of allergens.47,48Data also
demonstratecomplex interactions between urban air pollution,
aeroallergens,and allergic diseases; air pollution might not only
contribute to thecauseof someatopicdisorders but
alsomodifytheeffect ofaeroallergens on atopic diseases.49,50Recent
nationally represen-tative studies report that respiratory
allergies are positivelyassociated with air pollution in the United
States.51,52It isalso possible that differences across levels of
urbanizationreectotherenvironmentaleffects;manystudieshavereportedprotective
effects of growing up in a farming environment fromthe subsequent
development of atopic disorders.53,54The cross-sectional design of
NHANES 2005-2006 precludesthe assessment of temporal relationships.
Furthermore, thepresence of allergen-specic IgE antibodies does not
necessarilycorrespond to clinical allergy (ie, symptoms), but
multiple
studieshavedemonstratedthatthelikelihoodofclinicalallergycorre-lateswithresultsfromallergytesting.55-57Themethodologicaldifferences
between NHANES 2005-2006 and previousNHANESs makeit difcult
tocompare thepopulation-basedprevalence of allergic sensitization
over time. Because theperformance characteristics of serum-specic
IgE and skin pricktests vary,58,59results from these 2 tests are
not interchangeable.For example, cockroach has substantially lower
sensitivitywhen performance characteristics of ImmunoCAP are
comparedwiththoseofskinpricktests.58Althoughthesensitivityoftheassayhasimprovedoverthepast
decade, wewerenot abletoevaluate sensitization at less than the
cutoff of 0.35 kUA/L, whichmight be suboptimal.60Because of missing
sIgE data, thesmall samplesizeof theyoungest
participantslikelycompro-mised statistical power in subgroup
analyses. Despite theselimitations, NHANES 2005-2006 extends the
knowledge
onTABLEI.Prevalencesandoddsratiosforindependentpre-dictorsofspecicIgEpositivityintheNHANES2005-2006populationforsubjectsaged6yearsandolderandsubjectsaged1to5yearsPredictorPrevalenceofatopy,*%(SE)Adjustedoddsratio,yOR(95%CI)WaldFtest,PvalueySubjects>_6yAgegroup
_80 28.6(3.29) 1.0(reference)Sex
