Interpreting atmospheric pollen counts for use in clinical allergy: allergic symptomology

Review article

Interpreting atmospheric pollen counts for usein clinical allergy: allergic symptomologyDavid A. Frenz

Background: Allergists generally consider atmospheric pollen counts to be anestimate of the antigenic challenge confronting allergic individuals. The nature ofthis challenge depends on the particular pollen types found in the atmosphere andalso the airborne concentration of these pollen types. Both clinical experience andclinical investigations support these assumptions; however, a coherent system forrelating pollen counts and allergic symptomology does not exist.Objective: This review article will systematically review the medical and tech-

nical literature concerning the clinical significance of atmospheric pollen counts.Data sources: This review article will consider three independent bodies of

literature: 1) data contrasting human exposure patterns with rooftop pollen counts;2) data concerning dose-response relationships between atmospheric pollen countsand allergic symptomology; and 3) data concerning methods for indexing atmo-spheric pollen counts based on a pollen type’s in vivo allergenicity and terminalvelocity.Results: Three principal results emerged. First, rooftop pollen counts imperfectly

approximate human exposure to atmospheric pollen. Differences in both the con-centration and type of pollen encountered by humans can be expected to differ fromsamples obtained on rooftops.Second, allergic symptomology is positively correlated with atmospheric pollen

counts. Investigations involving Betula (birch) pollen offer quantitative dose-re-sponse models. Complex, nonlinear relationships that seem to reflect both thepriming effect and late-phase reactions exist.Last, atmospheric pollen counts can be indexed based on a contemporary appli-

cation of Thommen’s postulates. This system provides allergists with a means toestimate the clinical significance of various pollen types by combining data con-cerning in vivo allergenicity and terminal velocity.Conclusions: These conclusions should allow allergists to judge the clinical

significance of atmospheric pollen counts with greater sophistication than waspreviously possible.

Ann Allergy Asthma Immunol 2001;86:150–158.

INTRODUCTIONAllergists generally consider atmo-spheric pollen counts to be an estimateof the antigenic challenge confrontingallergic individuals. The nature of thischallenge depends on the particular pol-

len types found in the atmosphere andalso the airborne concentration of thesepollen types. Both clinical experienceand clinical investigations support theseassumptions; however, a coherent sys-tem for relating pollen counts and aller-gic symptomology does not yet exist.A systematic evaluation of the med-

ical and technical literature concerningthe relationship between atmosphericpollen counts and allergic symptomol-ogy appears long overdue. This review

article will summarize the availableevidence on the subject and presentguidelines for estimating the publichealth impact of pollen counts.

ApproachThis review article will consider threebodies of literature concerning atmo-spheric pollen counts and allergicsymptomology. First, studies concern-ing human exposure to atmosphericpollen will be presented. These datawill be contrasted with pollen countssupplied by rooftop pollen samplers todemonstrate the limits of rooftop datafor estimating human exposure.Second, investigations that related

atmospheric pollen counts to allergicsymptomology will be evaluated. Theimplications and limitations of thesestudies will be explored.Last, the properties of common pol-

len types will be considered. A pollentype’s in vivo allergenicity and termi-nal velocity will be offered as a meansto judge its clinical significance.

Human ExposureOne must consider whether rooftoppollen samples reasonably approxi-mate human exposure to atmosphericpollen before exploring the relation-ship between pollen counts and aller-gic symptomology. Several investiga-tions that used personal samplers andsamplers placed near homes will bereviewed. Pollen counts from these en-vironments will be contrasted withrooftop pollen measurements.Personal SamplersKailin1 conducted a study involving 94people in the Washington, DC, metro-politan area. Her personal samplers

Multidata LLC, St. Louis Park, Minnesota,and Medical School, University of Minnesota,Minneapolis, Minnesota.Received for publication April 8, 2000.Accepted for publication in revised form July

8, 2000.

150 ANNALS OF ALLERGY, ASTHMA, & IMMUNOLOGY

consisted of a greased microscopeslide placed within a protective hous-ing. The sampler was worn in the lapelarea (waking hours) or placed near thebed (sleeping hours) during one 24-hour period in September. These datawere then compared with atmosphericpollen counts obtained with threeDurham samplers.Kailin’s data suffer the various lia-

bilities of nonvolumetric devices andare therefore difficult to interpret.2 Shefound that pollen recovery varied be-tween patients; those with greater ex-posure to outdoor air tended to offerhigher pollen counts. In all cases, how-ever, the personal samplers offeredmuch lower pollen counts than theDurham samplers.Kailin’s data may reflect differences

in human exposure and illustrate a dis-cordance between rooftop pollen con-centrations and personal environments.A more probable explanation, how-ever, is that her instruments sampledunequal volumes of air. The Durhamsamplers probably sampled the mostair and patients with minimal exposureto outdoor air the least.Leuschner and Boehm3,4 extended

this work with a modified version ofKailin’s sampler. Pollen recovery bypersonal samplers exposed outdoorswas compared with volumetric pollencounts obtained from Burkard sporetraps (Burkard Manufacturing Co.,Rickmansworth, Hertfordshire, En-gland)5 in several cities in Switzerland.Although direct comparisons betweenthe devices are not possible, qualitativeanalyses appear permissible.Paired samples from 19 sampling

periods were analyzed.4 In some casesthe personal samplers recovered pollentypes that were absent in Burkard sam-ples, and vice versa. In addition, oncertain days, the personal samplerscollected large rafts of packed pollengrains, a phenomenon that the au-thors described as pollen “clouds” or“swarms.” On these occasions, theBurkard recovered little or no pollenof that pollen type.These data reflected differences in

the composition of pollen counts ob-tained in personal environments and

rooftop samples. One could cautiouslyspeculate that differences in the air-borne pollen concentrations also ex-isted.Various volumetric personal sam-

plers have been developed for indus-trial hygiene applications.6 This tech-nology has occasionally crossed overinto clinical allergy.7Gautrin et al8 employed volumetric

personal samplers to evaluate pollenexposure among municipal park work-ers who mowed lawns. Pollen countsobtained from employees’ shirt collarswere compared with pollen recoveryby a device placed 25 meters away anda rooftop pollen sampler that repre-sented the general community.Paired samples were obtained on 8

days. The median pollen count was 11,4, and 1 pollen grains per meter of air(p/m3), from shirt, park, and commu-nity, respectively. In five cases, thehighest pollen counts occurred on theshirt and in four cases, shirt � park �community.Like the previous investigations,

the particle collection efficienciesof the three devices were probablyunequal. These data, however, likelyargue for a discordance betweenrooftop samples and personal envi-ronments. Pollen counts were highestnear pollen sources (shirt) and lowestat a distance from source regions(community).Home EnvironmentsO’Rourke and Lebowitz9 compared pol-len counts collected on rooftops to thoseobtained near homes. Four clusters (2 to4 � 3 to 5 km) in Tucson, AZ, werestudied. A Burkard spore trap (5 mabove grade) placed in each cluster pro-vided regional pollen counts. RotorodSamplers (Sampling Technologies Inc.,St. Louis Park, MN)10,11 (1.5 m abovegrade) were placed in the yards of 51homes; pollen was collected over 72-hour periods.The rooftop devices recovered pol-

len in every sample, and in 50 cases,yielded higher pollen concentrationsthan samplers placed near homes; a10-fold difference was noted overall.In 21 cases, no measurable pollen was

recovered near homes. An equal orgreater number of pollen types wascollected on the rooftops.These results reflect intrinsic differ-

ences in particle recovery between thedevices12,13 and the effect of long sam-pling periods.14 Despite these con-founding factors, however, it appearsthat real differences existed betweenthe rooftop pollen samples and thoseobtained near homes. Rooftop sampleswere higher than pollen counts ob-tained near homes, generally by an or-der of magnitude. In addition, theformer offered a greater diversity ofpollen types compared with the latter.Differences are perhaps best ex-

plained by the filtering effect of urbanbuildings. Houses and other structuresprobably interfered with pollen trans-port near the ground, leading to lowerpollen counts and less taxonomic di-versity compared with the less ob-structed rooftop samples.

DiscussionA recent review article demonstratedthat spatial variability in atmosphericpollen concentrations exists within acity.15 One could therefore anticipatesimilar variation in human exposure topollen. The investigations consideredin this section evaluated the evidenceconcerning this claim.Although the available data are not

overwhelming, these studies seem todemonstrate that pollen counts in per-sonal environments differ from rooftoppollen measurements in both the amountand type of pollen recovered. This sug-gests that rooftop pollen measurementsimperfectly approximate human expo-sure to atmospheric pollen.Unfortunately, these studies do not

permit formulation of general guide-lines concerning human exposure.Rooftop samplers may offer pollencounts that are both greater or less thanpollen counts obtained near theground; diversity in the pollen typesrecovered can be similarly variable.Any system that relates pollen countsto allergic symptomology must there-fore accept these potential limitations.

VOLUME 86, FEBRUARY, 2001 151

Dose-Response DataVarious publications have related al-lergic symptomology to the associatedpollen challenge. These investigationswill be grouped by type in increasingorder of utility.Burge’s SystemThe most popular system for translat-ing pollen counts into allergic sympto-mology appears to have been derivedfrom Burge’s work.16 She analyzedpollen counts from 51 pollen measur-ing stations to calculate ranges in pol-len abundance for three broad catego-ries of pollen types. Although she didnot assign clinical significance to thesestatistical categories, her system hasbeen widely construed by others tohave such meaning.17,18 Table 1 pre-sents Burge’s ranges and their apparentclinical consequences. It is not clearhow the clinical inferences were devel-oped and on what evidence they rest.Another obvious liability of this sys-

tem is that certain pollen types arenonallergenic or poorly allergenic. Ahigh Pinus (pine) pollen count, for ex-ample, has little clinical significancebut would be interpreted quite differ-ently under this system.Controlled-Exposure ExperimentsSome investigators have exposed hu-mans to pollen in special chambers19–27or via inhalation devices28,29 and thenrecorded both subjective and objectiveallergic symptomology. In most cases,pollen concentrations greatly exceededenvironmental conditions. For exam-ple, in Day’s work,27 the target Ambro-

sia (ragweed) pollen concentration was3,500 p/m3. Although these pollen con-ditions have been documented in out-door air, the atmospheric pollen con-centration is generally substantiallylower.30,31 Data from these experimentsare thus difficult to translate into typi-cal human exposure patterns.Nonstatistical AnalysesVarious investigators have comparedatmospheric pollen counts with symp-tom and/or medicine scores.32–42 Datawere presented graphically and oftenillustrated striking temporal associa-tions between parameters. No statisti-cal analyses were performed, however,which precludes operationalizing theresults.Correlational AnalysesBrown and Ipsen43 compared symptomscores from patients with allergic rhi-nitis and/or asthma (n � 317) withmeteorological variables, gravimetricpollen and spore counts, and pollutiondata obtained during a 3-month studyperiod (August through October). Am-brosia pollen counts demonstrated thestrongest correlation with the clinicaldata (r � 0.670, P � .01). Remainingnonpollen variables were poorly corre-lated with allergic symptomology.Van Metre et al44 replicated this

finding during a clinical trial dealingwith various immunotherapy sched-ules. Volumetric Ambrosia pollencounts were correlated with dailysymptom-and-medication scores fortwo treatment groups and a placebo

group (r � 0.664 to 0.850, P � .001for all groups).Lebowitz et al45,46 investigated the

relationship between allergic sympto-mology and environmental conditionsusing time series analysis. Unlike tra-ditional correlational analyses, thesemethods can be used to explore timelags between potential stimuli and hu-man responses.Their first study obtained weekly

symptom data from 80 patients ran-domly selected from a large population(n � 3,800) over a 143-week period.45Rhinitis symptoms were correlatedwith Ambrosia and Chenopodiaceae(pigweed/goosefoot) pollen countslagged by 1 week (r � 0.325 and0.234, respectively; P � .05 in bothcases).A follow-up investigation obtained

daily symptom data from a pool of204 patients over a 2-year period.46Rhinitis symptoms were associatedwith pollen counts lagged by 2 days(P � .05) during the spring and earlysummer.In a derivative study, the relation-

ship among daily symptom data, me-teorology variables, and volumetricpollen counts was evaluated using pathanalysis (this type of analysis allowsone to determine the independent con-tribution of each variable in a complexsystem in which many variables areinterrelated).47 Rhinitis symptomswere positively associated with pollenconditions during the early spring andfall (P � .01).

Table 1. Burge’s System for Categorizing Atmospheric Pollen Counts (Pollen Grains per Cubic Meter of Air) and the Apparent ClinicalConsequence of Each Category. Data are provided for discussion only; see main text for the various liabilities of this system. Adapted fromReferences 16–18.

Category Tree Grass Weed Allergy symptoms

Absent 0 0 0 No symptomsLow 1–15 1–5 1–10 Only individuals extremely sensitive to these pollen types will

experience symptomsModerate 16–90 6–20 11–50 Many individuals sensitive to these pollen types will experience

symptomsHigh 91–1500 21–200 51–500 Most individuals with any sensitivity to these pollen types will

experience symptomsVery high �1500 �200 �500 Almost all individuals with any sensitivity to these pollen types will

experience symptoms; extremely sensitive people could havesevere symptoms


Dose-Response ModelingTaudorf and Moseholm48 studied dailysymptom and medicine scores in Bet-ula-sensitive patients (n � 15) overtwo consecutive pollen seasons. Ananalysis involving volumetric Betulapollen counts yielded three principalconclusions.First, the relationship between the

clinical measures and pollen countswas found to be nonlinear (Fig 1a; see

Appendix 1). This finding agrees withdata from an earlier clinical trial49 (Fig1b) and appears to be the experience ofother investigators.50,51

Second, the clinical consequences ofa given pollen load increased as theseason progressed. This likely repre-sented Connell’s priming effect whererepeated pollen challenges increasednasal sensitivity and thus allergicsymptomology despite a constant ordeclining antigenic challenge.28,52Last, development of significant

symptom scores lagged peaks in atmo-spheric pollen conditions by severaldays; the effect decayed exponentiallywith a characteristic half-life of 1 to 2days. Brostrom and Moller53 uncov-ered a similar phenomenon in theiranalysis of data from two clinical tri-als.54,55 This finding was thought to bedistinct from the priming effect andmay represent late phase reactions.DiscussionMany studies published throughout thelast century demonstrated a positiveassociation between allergic symptom-ology and atmospheric pollen condi-tions. Formal statistical analyses con-firmed this association and quantifiedthe strength of the correlation betweenthese phenomena.A nonlinear relationship appears to

exist between allergic symptomologyand atmospheric pollen counts. Asdemonstrated by Figure 1, human suf-fering reaches significant levels atmodest pollen counts; substantial pol-len challenges are required to achieveincremental changes in symptomologythereafter. These findings thus supportthe general approach, but not the nu-meric data, offered in Table 1.Several investigators demonstrated

that allergic symptomology can lagpollen events and also appears tochange over the course of the pollenseason. These observations may repre-sent late phase reactions and the prim-ing effect, respectively. These findingssuggest that Table 1 may be too rigidand that a dynamic system may bemore appropriate.

Allergen IndexingDozens of pollen types are recoveredin atmospheric samples throughoutthe course of a typical pollen season.This section will present some guide-

lines for judging their clinical signif-icance.

Thommen’s PostulatesThommen outlined five postulates thatcharacterize the features of clinicallyimportant pollen types: 1) the pollenmust contain an excitant of hayfever;2) the pollen must be anemophilous(wind-pollinated); 3) the pollen mustbe produced in sufficiently large quan-tities to lead to human exposure; 4) thepollen must be sufficiently buoyant tobe carried considerable distances; and5) the plant producing the pollen mustbe widely and abundantly distribut-ed.56,57 Although this framework wasfirst presented 70 years ago, it with-stands contemporary scrutiny and canscarcely be improved.Thommen’s first and fourth postulates

are the most important tenets of his sys-tem because they can be used to quantifythe significance of atmospheric pollencounts. These postulates will be exploredin greater detail below.Thommen’s second postulate is obvi-

ously true, but should be slightly refinedto recognize the clinical importance ofamphiphilous (wind-and-insect-pollinated)genera.58 Acer (maple) and Salix (wil-low) pollen, for example, are recoveredin appreciable quantities in atmosphericsamples and cause pollinosis.59 Evenpollen from entomophilous (insect-polli-nated) genera can occasionally be aero-biologically and clinically significant.60Thommen’s third postulate, although

true, would be difficult to translate intoclinical practice. Paleoecologists haveestimated the pollen production ofcertain genera: one Alnus (alder) tree,for example, produces approximately3.6 � 1011 pollen grains during a 50-year period.61 Data are not availablefor all genera, however, which pre-cludes developing an indexing methodbased on this postulate.Thommen’s fifth postulate also seems

to be true, but would be exceedinglydifficult to translate into clinical practice.Most vegetation maps indicate the range,not geographic abundance of certainplants.62

Figure 1. Relationship between Betula pollencounts, symptom scores and the percent of pa-tients reporting symptoms. Symptom scoresrange from 0 to 3 (�0.5 � Few symptoms;0.6–1.0 � Some symptoms; 1.1–1.5 � Manysymptoms; 1.6–2.0 � High symptoms; �2 �Very high symptoms). Adapted from References48 and 49. See also Appendix 1.


AllergenicityThommen recognized that certain pol-len types such as Pinus rarely causepollinosis, although others such as Am-brosia are strongly allergenic. Theformer apparently lacked an excitantpossessed by the latter. From a con-temporary perspective, this excitantrepresents a protein allergen or groupof allergens.Traditionally, the allergenicity of

various pollen types has been estab-lished by in vivo methods such as skintesting. These data can be used tojudge the clinical importance of pollenrecovered in atmospheric samples.Many skin test surveys have been

conducted, but most must be inter-preted with caution because the patientpopulation was not broadly representa-tive and/or the skin test extracts werenot standardized. With these hazardsacknowledged, some skin test data willbe briefly reviewed.Gergen et al63 sought to establish the

skin-test sensitivity of a large, randomsample of the US population. Extractsof four pollen types were employed:Quercus (oak), Lolium (darnel or ryegrass), Cynodon dactylon (Bermudagrass) and Ambrosia. Positive skintests occurred in 4.7%� 0.36, 10.2%�0.56, 4.4% � 0.42, and 10.1% � 0.57of people tested, respectively.Although this study used a large

sample population, its results are lim-ited by the narrow panel of allergensused. Several smaller studies are morehelpful because they tested patientswith larger panels of pollen allergens.Lewis and Imber64–66 and Chap-

man67,68 conducted similar skin testsurveys on large groups of atopic pa-tients in Missouri. Data from thosestudies were reanalyzed for this reviewarticle and are presented in Table 2(see Appendix 2). A pattern similar toGergen et al’s findings emerged. Alarger fraction of the patient populationwas sensitive to weed and grass aller-gens than to tree allergens.Terminal VelocityThommen recognized that pollen mustbe buoyant to be of clinical signifi-cance. He attributed buoyancy to a pol-

len grain’s size, morphology, andweight (or specific gravity).These variables have some explana-

tory power but are neatly subsumed bya pollen grain’s terminal velocity, orits rate of fall per unit time (generallyexpressed in centimeters per second).A pollen grain’s terminal velocity canbe calculated from Stoke’s law69 (whichconsiders a particle’s size and density)and experimentally measured by variousmeans. The latter approach is generallypreferred for pollen grains because theterminal velocities of nonspherical parti-cles often depart significantly from val-ues offered by Stoke’s law.Various investigators have mea-

sured the terminal velocities of manycommon pollen types.69–71 Results fre-quently vary between studies becauseof different experimental approaches.Durham’s data will be used for thisreview article because his effort in-volved 38 species from 31 genera, thusallowing comparisons among mostmajor pollen types.70In Durham’s investigation, Ambro-

sia trifida (giant ragweed) and Zeamays (corn) offered the lowest andhighest terminal velocities, respec-tively (0.823 and 18.288 cm/sec). Con-sidered as groups, the mean terminalvelocity of weed pollen (1.311 �0.512 cm/sec; n � 17) was less thantree pollen (2.026� 0.786 cm/sec; n�15) and grass pollen (2.128 � 0.621cm/sec; n � 6, Zea and Secale cereale(rye) excluded).Pollen grains with lower terminal

velocities travel farther from theirsource than grains with higher terminal

velocities.15 Data adapted from Sugi-ta’s work demonstrate that the relation-ship between terminal velocity and dis-tance traveled is logarithmic (Fig 2;see Appendix 3).72

DiscussionData concerning allergenicity and ter-minal velocity can be combined to in-dex the clinical importance of pollenrecovered in atmospheric samples.This new indexing system is pre-

sented in Figure 3 for nine pollentypes. All data were scaled to valuesfor Ambrosia because it is highly aller-genic and has a low terminal velocity.Skin test data from Table 2 were trans-formed in a linear manner and appearon the y-axis. Terminal velocity datawere transformed in a logarithmicmanner (per an adaptation of Equation3) and appear on the x-axis.The right upper quadrant represents

pollen types of greatest allergenic sig-nificance: pollen grains have a low ter-minal velocity and are highly aller-genic. The left lower quadrantrepresents pollen types of least aller-genic significance: pollen grains havea high terminal velocity and are poorlyallergenic. Remaining quadrants are ofintermediate clinical significance.

CONCLUSIONSThis review article explored the rela-tionship between atmospheric pollencounts and allergic symptomology. Onthe basis of the literature explored,three principal conclusions emerged.First, rooftop pollen counts imper-

fectly approximate human exposure to

Table 2. Skin Test Frequency (Percentage of Patients Responding 2–4�) and TerminalVelocity Data (cm/s) for Nine Pollen Types. Adapted from References 64–68 and 70. Seealso Appendix 2.

Pollen type Lewis Chapman Mean TV

Betula 5.3 2.1 3.7 1.676Fraxinus 4.1 5.0 4.6 1.524Juglans 7.0 6.2 6.6 2.835Quercus 6.8 5.2 6.0 2.195Ulmus 7.5 6.2 6.9 2.042Phleum 16.8 19.2 18.0 2.804Amaranthus 8.6 9.0 8.8 1.859Ambrosia 22.4 25.6 24.0 0.823Plantago 11.3 11.6 11.5 1.494


atmospheric pollen. Differences inboth the concentration and type of pol-len encountered by humans can be ex-pected to differ from samples obtainedon rooftops.Second, allergic symptomology is

positively correlated with atmosphericpollen counts. Investigations involvingBetula pollen offer quantitative dose-response models. Complex, nonlinearrelationships that seem to reflect boththe priming effect and late phase reac-tions exist.

Last, atmospheric pollen counts canbe indexed based on a contemporary ap-plication of Thommen’s postulates. Thissystem provides allergists with a meansto estimate the clinical significance ofvarious pollen types by combining dataconcerning in vivo allergenicity and ter-minal velocity.Taken together, these conclusions

should allow allergists to judge the clin-ical significance of atmospheric pollencounts with greater sophistication thanwas previously possible.

APPENDIX 1The purpose of this appendix is to sup-ply equations for the curves depicted inFigure 1.The top graph was derived from Table

4 of Taudorf and Moseholm.48 Data fortheir most sensitive group of patients(group H) was reanalyzed for this reviewarticle. The midpoint of each symptomcategory was determined (0.25, 0.75,1.25, 1.75, and 2.5). The midpoint in therange of pollen counts supplied for eachsymptom category was then calculatedfor the first four categories (2.5, 10.0,45.0, and 337.5 p/m3). The pollen countassociated with the fifth category(3223.9 p/m3) was determined by regres-sion analysis because a range was notsupplied.The data depicted are described by the

following equation:

b � 0.968�101.409s r � .998

where b is the Betula pollen count, s isthe symptom score and r is the equa-tion’s correlation coefficient.The bottom graph was derived from

Table 2 of Viander and Koivikko.49 Datafor their control population was re-gressed against the Betula pollen count.

b � 3.945�100.015p r � .914

where p is the percentage of patientsreporting symptoms.

APPENDIX 2The purpose of this appendix to ex-plain the derivation of the skin test datathat are presented in Table 2.Lewis and Imber64–66 presented skin

test data obtained from a large popula-

tion of presumably atopic patients seek-ing care from two allergists over a 10-year period. Data for a sub-population(PS1) tested via the intracutaneousmethod were reanalyzed for this reviewarticle.For the tree study, patients (n� 1324)

were screened with an extract containinga mixture of various tree antigens(“mixed trees”).65 Those respondingstrongly (2 to 4�) were then tested witha panel of 12 individual tree extracts.Table 2 reflects the percentage of thetotal population tested that reacted (2 to4�) to each tree extract.Similar efforts were conducted for

grass and weed pollen.64,66 Unlike thework with tree pollen, patients were notscreened with grass or weed mixturesbefore skin testing with individual grassand weed extracts. Table 2 reflects thepercentage of each population tested(n � 1,341 to 1,407) that reacted (2 to4�) to the various extracts.Chapman67,68 conducted a similar

study involving patients referred to himby other physicians for an allergy con-sultation. Data for a sub-population (n�629) tested via the intracutaneousmethod were reanalyzed for this reviewarticle. Table 2 reflects the percentage ofthis population that reacted (2 to 4�) toeach extract.

APPENDIX 3The purpose of this appendix is to ex-plain the derivation of Figure 2.Sugita’s work was reviewed at length

previously.15 His mathematical investi-gations yielded the source areas for lakesof various radii for certain pollen types.A striking relationship between a pollentype’s terminal velocity and the sourcearea was evident: pollen types with lowterminal velocities offered very largesource areas compared with pollen typeswith high terminal velocities.The source radius contributing 50% of

pollen to the center of a lake (radius50 m) was obtained for four pollen typesrepresenting a spectrum of terminal ve-locities.72 Terminal velocity was thenplotted against source radius. The result-ing curve in Figure 2 is described by the

Figure 2. Relationship between terminal velocityand source area for four pollen types representing aspectrum of terminal velocities. Adapted from Ref-erence 72. See also Appendix 3.

Figure 3. Allergen indexing. The nine pollentypes presented in Table 2 were indexed to val-ues for Ambrosia (1.00, 1.00). The right upperquadrant represents pollen types of greatest clin-ical significance. The left lower quadrant repre-sents pollen types of least clinical significance.Remaining quadrants are of intermediate clinicalsignificance.


following equation

t � �3.246 log R � 14.495 r � .930

where t is the pollen type’s terminalvelocity (cm/sec) and R is its corre-sponding source radius (m).Notice the dramatic relationship be-

tween terminal velocity and source ra-dius. Initially, the latter increases onlyminimally despite a significant reductionin terminal velocity. Between the thirdand fourth points, however, source ra-dius increases from 1,637 to 15,759 mdespite a comparatively small decreasein terminal velocity (2.9 to 1.6 cm/sec).

ACKNOWLEDGMENTSThe author thanks Walter H. Lewis,PhD, Department of Biology, Wash-ington University, St. Louis, Missouri,for discussion concerning References64 to 66. The author also gratefullyacknowledges Michael D. Lebowitz,PhD, College of Public Health, Uni-versity of Arizona, Tucson, Arizona,for discussion concerning References45 to 47. Susan C. Gebhard and DennisE. Gebhard, both of Multidata LLC, St.Louis Park, Minnesota, reviewed adraft of this manuscript and providedhelpful suggestions.

REFERENCES1. Kailin EW. Variations in ragweed pol-len exposure of individuals in a met-ropolitan area. Med Ann DC 1964;33:1–5.

2. Ogden EC, Raynor GS, Hayes JV, et al.Manual for Sampling Airborne Pollen.New York, NY: Hafner Press, 1974:40–42, 50–51.

3. Leuschner RM, Boehm G. IndividualPollen Collector for use of hay-fever pa-tients in comparison with the Burkardtrap. Grana 1977;16:183–186.

4. Leuschner RM, BoehmG. Investigationswith the Individual Pollen Collector andBurkard trap with reference to hay feverpatients. Clin Allergy 1979;9:175–184.

5. Hirst JM. An automatic volumetric sporetrap. Ann Appl Biol 1952;39:257–265.

6. Fletcher RA. A review of personal/portable monitors and samplers for air-borne particles. J Air Pollut Control As-soc 1984;34:1014–1016.

7. Fiorina A, Scordomaglia A, MincariniM, et al. Aerobiologic particle samplingby a new personal collector (Partrap

FA52) in comparison to the Hirst(Burkard) sampler. Allergy 1997;52:1026–1030.

8. Gautrin D, Vandenplas O, DeWitte DJ,et al. Allergenic exposure, IgE-mediatedsensitization, and related symptoms inlawn cutters. J Allergy Clin Immunol1994;93:437–445.

9. O’Rourke MK, Lebowitz MD. A com-parison of regional atmospheric pollenwith pollen collected at and near homes.Grana 1984;23:55–64.

10. Grinnell SW, Perkins WA, VaughanLM. Sampling apparatus and method.Patent no. 2,973,642. Washington, DC:United States Patent Office, 1961.

11. Magill PL, Lumpkins ED, Arveson JS.A system for appraising airborne popu-lations of pollens and spores. Am IndHyg Assoc J 1968;29:293–298.

12. Frenz DA. Comparing pollen and sporecounts collected with the Rotorod Sam-pler and Burkard spore trap. Ann AllergyAsthma Immunol 1999;83:341–349.

13. Frenz DA. The effect of windspeed onpollen and spore counts collected withthe Rotorod Sampler and Burkard sporetrap. Ann Allergy Asthma Immunol2000;85:392–394.

14. Frenz DA, Boire AA. Pollen recovery inatmospheric samples collected with theRotorod Sampler over multiple-day pe-riods such as weekends. Ann AllergyAsthma Immunol 1999;83:217–221.

15. Frenz DA. Interpreting atmospheric pol-len counts for use in clinical allergy:spatial variability. Ann Allergy AsthmaImmunol 2000;84:481–491.

16. Burge HA. Monitoring for airborne al-lergens. Ann Allergy 1992;69:9–21.

17. National Allergy Bureau. Pollen andMold Report. Milwaukee, WI: AmericanAcademy of Allergy, Asthma and Im-munology 1997:3.

18. National Allergy Bureau. Reading thecharts [www.aaaai.org/nab/reading.stm].Milwaukee, WI: American Academy ofAllergy, Asthma, and Immunology.

19. Mock DC. A controlled evaluation oftwo antihistamine compounds using na-sal engorgement as an index of response.J Allergy 1963;34:432–438.

20. Davies HJ. Exposure of hay fever sub-jects to an indoor environmental grasspollen challenge system. Clin Allergy1985;15:419–427.

21. Horak F, Jager S. Die WienerProvokations-Kammer: eine neue meth-ode des allergenexpositionstests. WienKlin Wochenschr 1987;99:509–510.

22. Day JH, Buckeridge DL, Clark RH, et al.

A randomized, double-blind, placebo-controlled antigen delivery study of theonset of action of aerosolized triamcino-lone acetonide nasal spray in subjectswith ragweed-induced allergic rhinitis. JAllergy Clin Immunol 1996;97:1050–1057.

23. Donovan JP, Buckeridge DL, BriscoeMP, et al. Efficacy of immunotherapy toragweed antigen tested by controlled an-tigen exposure. Ann Allergy Asthma Im-munol 1996;77:74–80.

24. Day JH, Briscoe MP, Clark RH, et al.Onset of action and efficacy of terfena-dine, astemizole, cetirizine, and lorata-dine for the relief of symptoms of aller-gic rhinitis. Ann Allergy AsthmaImmunol 1997;79:163–172.

25. Day JH, Briscoe MP, Welsh A, et al.Onset of action, efficacy, and safety of asingle dose of fexofenadine hydrochlo-ride for ragweed allergy using an envi-ronmental exposure unit. Ann AllergyAsthma Immunol 1997;79:533–540.

26. Day JH, Briscoe M, Widlitz MD. Ceti-rizine, loratadine, or placebo in subjectswith seasonal allergic rhinitis: effects af-ter controlled ragweed pollen challengein an environmental exposure unit. J Al-lergy Clin Immunol 1998;101:638–645.

27. Day JH, Briscoe MP. Environmental ex-posure unit: a system to test anti-allergictreatment. Ann Allergy Asthma Immu-nol 1999;83:83–93.

28. Connell JT. Quantitative intranasal pol-len challenges: I. Apparatus design andtechnique. J Allergy 1967;39:358–367.

29. Connell JT. Quantitative intranasal pol-len challenge[s]: II. Effect of daily pollenchallenge, environmental pollen expo-sure, and placebo challenge on the nasalmembrane. J Allergy 1968;41:123–139.

30. Frenz DA, Palmer MA, Hokanson JM,Scamehorn RT. Seasonal characteristicsof ragweed pollen dispersal in the UnitedStates. Ann Allergy Asthma Immunol1995;75:417–422.

31. Frenz DA. Volumetric ragweed pollendata for eight cities in the continentalUnited States. Ann Allergy Asthma Im-munol 1999;82:41–46.

32. Rackemann FM, Smith LB. Studies inasthma: XII. The plant pollens of NewEngland; their relative importance. J Al-lergy 1931;2:364–371.

33. Sack SS, Golan HG. The relationshipbetween clinical symptoms and pollencount[s] in pollinosis. J Allergy 1942;13:296–299.

34. Gurney CW, Cryst S. Aeroallergen stud-ies with the molecular filter membrane. J


Allergy 1955;26:533–541.35. Connell JT, Sherman WB. Hay fever

symptoms related to immunologicalfindings. Ann Allergy 1967;25:239–250.

36. Norman PS. A rational approach to de-sensitization. J Allergy 1969;44:129–145.

37. Norman PS, Lichtenstein LM, Tignall J.The clinical and immunologic specificityof immunotherapy. J Allergy Clin Im-munol 1978;61:370–377.

38. Østerballe O. Immunotherapy in hay fe-ver with two major allergens 19, 25 andpartially purified extract of timothy grasspollen. Allergy 1980;35:473–489.

39. Peterson BN, Sandberg I. Diagnostics inallergic diseases by correlating pollen/fungal spore counts with patient scoresof symptoms. Grana 1981;20:219–224.

40. Frostad AB, Grimmer Ø, Sandvik L, etal. Clinical effects of hyposensitizationusing a purified preparation from timo-thy pollen as compared to crude aqueousextracts from timothy pollen and a four-grass pollen mixture respectively. ClinAllergy 1983;13:337–357.

41. Ickovic MR, Thibaudon M, Peltre G,David B. Tree and grass pollen allergensensitivity related to atmospheric pollencounts in 1986 and 1987. Rev PalaeobotPalynol 1990:64:45–51.

42. Brown HM. The relationship of aerobio-logical data to seasonal allergicsymptoms: a review of 27 years’ experi-ence. Immunol Allergy Pract 1992;14:318–329.

43. Brown EB, Ipsen J. Changes in severityof symptoms of asthma and allergic rhi-nitis due to air pollutants. J Allergy 1968;41:254–268.

44. VanMetre TE Jr, Adkinson NF, AmodioFJ, et al. A comparison of immunother-apy schedules for injection treatment ofragweed pollen hay fever. J Allergy ClinImmunol 1982;69:181–193.

45. Robertson G, Lebowitz MD. Analysis ofrelationships between symptoms and en-vironmental factors over time. EnvironRes 1984;33:130–143.

46. Lebowitz MD, Collins L, Holberg CJ.Time series analyses of respiratory re-sponses to indoor and outdoor environ-mental phenomena. Environ Res 1987;43:332–341.

47. Holberg CJ, O’Rourke MK, Lebowitz

MD. Multivariate analysis of ambientenvironmental factors and respiratory ef-fects. Int J Epidemiol 1987;16:399–410.

48. Taudorf E, Moseholm L. Pollen count,symptom and medicine score in birchpollinosis. Int Arch Allergy Appl Immu-nol 1988;86:225–233.

49. Viander M, Koivikko A. The seasonalsymptoms of hyposensitized and un-treated hay fever patients in relation tobirch pollen counts: correlations with na-sal sensitivity, prick tests and RAST.Clin Allergy 1978;8:387–396.

50. Wilkinson P, Taudorf E. Clinical trial ofhyposensitisation in hay fever: twomethods of relating symptoms scores todaily pollen counts. Allergy 1984;39:463–468.

51. Taudorf E, Laursen LC, Djurup R, et al.Oral administration of grass pollen tohay fever subjects. Allergy 1985;40:321–335.

52. Connell JT. Quantitative intranasal pol-len challenges: III. The priming effect inallergic rhinitis. J Allergy 1969;43:33–44.

53. Brostrom G, Moller C. A new method torelate symptom scores with pollencounts. Grana 1989;28:123–128.

54. Moller C, Dreborg S. Cross-reactivitybetween deciduous trees duringimmunotherapy: I. In vivo results. ClinAllergy 1986;16:135–143.

55. Moller C, Dreborg S, Lanner Å, Bjork-sten B. Oral immunotherapy of childrenwith rhinoconjunctivitis due to birch pol-len allergy. Allergy 1986;41:271–279.

56. Thommen AA. The etiology of hayfever: studies in hay fever. NY StateJ Med 1930;30:437–442.

57. Thommen AA. Which plants cause hayfever? In: Coca AF, Walzer M, Thom-men AA, eds. Asthma and Hay Fever inTheory and Practice. Springfield, IL:Charles C. Thomas Publisher, 1931:546–554.

58. Hyde HA. Atmospheric pollen andspores in relation to allergy. Clin Allergy1972;2:153–179.

59. Lewis WH, Vinay P, Zenger VE. Air-borne and Allergenic Pollen of NorthAmerica. Baltimore, MD: Johns Hop-kins University Press, 1983:12–15,85–87.

60. Lewis WH, Vinay P. North Americanpollinosis due to insect-pollinated plants.

Ann Allergy 1979;42:309–318.61. Erdtman G. Handbook of Palynology.

New York, NY: Hafner PublishingCompany, 1969:118.

62. Lewis WH, Vinay P, Zenger VE. Air-borne and Allergenic Pollen of NorthAmerica. Baltimore, MD: Johns Hop-kins University Press, 1983:177–236.

63. Gergen PJ, Turkeltaub PC, Kovar MG.The prevalence of allergic skin test reac-tivity to eight common aeroallergens inthe U.S. population: results from the sec-ond National Health and Nutrition Ex-amination Survey. J Allergy Clin Immu-nol 1987;80:669–679.

64. LewisWH, ImberWE. Allergy epidemi-ology in the St. Louis, Missouri, area: II.Grasses. Ann Allergy 1975;35:42–50.

65. LewisWH, ImberWE. Allergy epidemi-ology in the St. Louis, Missouri, area: III.Trees. Ann Allergy 1975;35:113–119.

66. LewisWH, ImberWE. Allergy epidemi-ology in the St. Louis, Missouri, area:IV. Weeds. Ann Allergy 1975;35:180–187.

67. Chapman JA, Williams S. Aeroallergensof the southeast Missouri area: a reportof skin test frequencies and air samplingdata. Ann Allergy 1984;52:411–418.

68. Chapman JA. Aeroallergens of south-eastern Missouri, USA. Grana 1986;25:235–246.

69. Gregory PH. The Microbiology of theAtmosphere, second ed. NewYork: JohnWiley and Sons, 1973:15–29.

70. Durham OC. The volumetric incidenceof atmospheric allergens: III. Rate of fallof pollen grains in still air. J Allergy1946;17:70–78.

71. DiGiovanni F, Kevan PG, Nasr ME. Thevariability in settling velocities of somepollen and spores. Grana 1995;34:39–44.

72. Sugita S. A model of pollen source areafor an entire lake surface. Quat Res1993;39:239–244.

.

Requests for reprints should be addressed to:David A. FrenzVice-President/Director of ResearchMultidata LLC4838 Park Glen RoadSt. Louis Park, MN 55416-5702E-mail: [email protected]


CME Examination (5 Questions Total)No 2001–0021–5, Frenz DA. 2001;86:000–000.CME TEST QUESTIONS1. Taudorf and Moseholm demonstratedthat the relationship between pollencounts and symptom scores is:a. Linear.b. Nonlinear.c. Linear for Betula pollen but nonlin-ear for Poaceae pollen.

d. Linear for Betula pollen but nonlin-ear for Ambrosia pollen.

e. Random.2. Taudorf and Moseholm demonstratedthat a given Betula pollen load causedgreater allergic symptomology as thepollen season progressed. This likelyrepresented:a. The contribution of other pollen typessuch as Populus.

b. Connell’s priming effect.c. Late phase reactions.d. Comorbid food allergies.e. Comorbid psychiatric illness.

3. Taudorf and Moseholm demon-strated that the development of sig-nificant symptom scores laggedpeaks in atmospheric pollen condi-tions by several days. This likelyrepresented:a. The contribution of other pollen typessuch as Populus.

b. Connell’s priming effect.c. Late phase reactions.

d. Comorbid food allergies.e. Comorbid psychiatric illness.

4. Thommen:a. Outlined five postulates that char-acterize the features of clinicallyimportant pollen types.

b. Was a nineteenth century pulmo-nologist.

c. Conducted the first experimentsconcerning immunotherapy.

d. Elucidated the tertiary structure ofIgE.

e. Deduced the intracellular signalingpathways for several cytokines.

5. In this review article, pollen typeswere indexed by combining data con-cerning their:a. Mass and morphology.b. Morphology and specific gravity.c. Mass and terminal velocity.d. In vivo allergenicity and terminalvelocity.

e. In vivo allergenicity andmorphology.

Annotated answers1. The correct answer is b. Figure 1 dem-onstrates a nonlinear relationship be-tween pollen counts and symptomsscores. Please refer to References 48–51.

2. The correct answer is b. Connell’s nowclassic series of investigations (Refer-

ences 28, 29, and 52) explored variousfeatures of the pathophysiology of aller-gic rhinitis. Taudorf and Moseholm in-voked his “priming effect of the endorgan” to explain their clinical observa-tions.

3. The correct answer is c. Several authors,including Taudorf and Moseholm, havedemonstrated that allergic symptomssometimes lag peaks in atmospheric pol-len conditions. Although one could ar-gue that this represents Connell’s prim-ing effect, the temporal characteristicsseem to favor late phase reactions.Please refer to References 48 and 53–55.

4. The correct answer is a. Thommen firstoutlined his postulates in Reference 56.He presented an amplified and refinedtreatment in an early textbook on aller-gic disease that he co-authored withCoca and Walzer (Reference 57).

5. The correct answer is d. This reviewarticle argued for an evidential ap-proach for judging the clinical signif-icance of atmospheric pollen counts.Modified versions of two of Thom-men’s postulates—in vivo allergenic-ity and terminal velocity—were usedto index a pollen type’s potential tocause allergic disease. Please refer toTable 2 and Figures 2 and 3.


Documents

Interpreting atmospheric pollen counts for use in clinical allergy: allergic symptomology