JouMf AL or EpronaoLoar VoL 88, No. 3Copyright O 1968 by The John* Hopklo* Univerrity PrnUtd in UM.A.

DIFFERENCES IN THE DISTRIBUTION OF SENSITIVITY TOHISTOPLASMIN AND ISOLATIONS OF

HISTOPLASMA CAPSULATUM1

GEORGE W. COMSTOCK, CARLOS N. VICENS,1 NORMAN L. GOODMAN*AND SUSAN COLLINS*

(Received for publication November 10,1967)

Comstock, G. W . (Training Center for Public Health Research, Box 2067,Hagerstown, Md. 21740) C. N. Vicens, N. L Goodman, and S. Collins. Differencesin the distribution of sensitivity to histoplasmin and isolations of Histoplasma capsu-latvm. Amer. J. Epid., 1968, 88:195-209.—In 1963, 7,787 junior and senior highschool students in Washington County, Maryland were tested with histoplasmin.Among 1,450 students who were lifetime residents of a single house, there wasmarked geographic variation in the prevalence of histoplasmin reactors, rangingfrom 10 to 9 0 % with no apparent correlation with any topographic or geo-graphic features except for low prevalence within the city of Hagerstown.Marked variation in the frequency of histoplasmin sensitivity within short distancesmade it unlikely that meterologic factors were involved and suggested thatinfection was usually acquired close to home. No association was found with thepresence of chickens or domestic birds on the premises. Students who lived onalkaline soils were more likely to have been infected than those who lived on acidsoils. Positive soil cultures for Histoplasma capsvlahim did not show a similarpattern of associations, being more frequent from premises with chickens and inareas with acid soils. The discrepancies between skin sensitivity and soil isolationscould result from some characteristic of the organism, such as spore size, whichvaried with soil conditions.

The geographic distribution of histo- challenge to investigators during theplasmosis has been both a puzzle and a quarter-century since it first became ap-

parent that infection with Histoplasma1 From the Traming Center for Public ^ ^ t u m w a 8 v e r y c o m m O n in some

Health Research, School of Hygiene andPublic Health, Johns Hopkins University, give helpful advice throughout the study. TheBox 2067, Hagerstown, Maryland 21740. This authors are also grateful to Dr. Philip E.investigation was supported in part by Grad- Sartwell for his suggestions and criticisms,uate Training Grant No. CD-1-Ol-l-Tl from 'From the Department of Preventive Med-the Bureau of Disease Prevention and En- i c m e and Public Health, University of Puertovironmental Control and by Research Career Rico School of Medicine, San Juan, PuertoAward No. K6-HE-21.670 from the National Rico.Heart Institute, VS. Public Health Service; 'From the Department of Botany andand by a grant from the Washington County Microbiology, University of Oklahoma, Nor-(Md.) Tuberculosis Association. man, Oklahoma.

The initial suggestion that Washington 'From the Research Section, TuberculosisCounty, Maryland might be a fruitful loca- Program, National Communicable Diseasetion for studies on histoplasmosis was made Center, UJ3. Public Health Service, Bethesda,by Dr. Carroll E. Palmer, who continued to Maryland.

195

196 COMSTOCK, VICENS, GOODMAN AND COLL1NB

parts of the United States and very rarein others (1-3). Large scale studies bythe Tuberculosis Program of the U.S.Public Health Service showed thatcounties in which the majority ofyoung adult lifetime residents reactedto histoplasmin were heavily concen-trated in the east central portion of thecountry, a triangular area with its apicesnear Omaha, Nebraska, Columbus,Ohio and Natchez, Mississippi (3-5).Elsewhere in the nation, there were fewsuch counties; most of them were scat-tered in eastern Texas and westernLouisiana, along the Appalachian moun-tains from Bristol, Tennessee to Gettys-burg, Pennsylvania, and on Maryland'sEastern Shore. But infection was notuniformly distributed within the majorareas of high and low prevalence, andsubsequent studies of counties and townshave shown considerable differences inthe frequency of reactors to histoplas-min even within these small areas(6-15). Indeed, variation appears to bethe hallmark of the geographic dis-tribution of histoplasmin sensitivity.The cause or causes of this variationare still undetermined.

Studies of the distribution of H. cap-sulatum have been handicapped by thedifficulties of isolating the organism.Although it is believed to be a normalinhabitant of the soil, it can be found inonly a small proportion of soil sam-ples even in apparently endemic areas.Nevertheless, accumulated experienceindicates H. capsulatum is most likelyto be demonstrable in soil that is moist,shaded, moderately acid and enrichedby bird or bat manure (15-19). Althoughthese factors have often been foundto be associated with human infections,particularly in epidemic situations (19-21), there have been instances in whichhuman and animal infections could

not be correlated with isolations ofH. capsulatum (11,15).

The present study was suggested bythe observation that histoplasmin reac-tor rates among U.S. Navy recruits fromcounties to the south and east of Wash-ington County, Maryland were veryhigh while those from counties to thenorth and west were low (22). Washing-ton County appeared to straddle a steepgradient of histoplasmin sensitivity, asituation expected to provide a widerange of reactor rates in its population.An additional consideration in select-ing this area was the unusually com-plete information about all residents ofthe county that had just been collectedin a private census. The combined find-ings of a histoplasmin skin test survey,the census and a preliminary soil sam-pling for H. capsulatum have providedevidence that factors associated withthe ability to isolate the organism fromthe soil are not necessarily the same asthose associated with human sensitivityto histoplasmin. This finding has im-portant implications for investigationsof the epidemiology of histoplasmosis.

MATEBIALS AND METHODS

To conduct the histoplasmin sur-vey, a team of nurses thoroughly exper-ienced in intracutaneous testing wasassigned to Washington County by theTuberculosis Program of the U.S. PublicHealth Service, which also providedantigens and supplies for the survey.The Washington County Health De-partment assigned nurses to assist withthe testing, and the Washington CountyTuberculosis Association provided educa-tional materials and most of the clericalhelp at the survey sites. The surveywas done in November, 1963. All of the10,227 junior and senior high schoolstudents in Washington County enrolled

DIFFERENCES IN DISTRIBUTION OF SENSITIVITY TO HKTOPIiASMIN 197

in 19 public and parochial schools wereoffered skin tests with tuberculin(5 TU PPD-S) and histoplasmin(1:100 H-42). Coccidioidin tests (1:100Cutter #99096) were also given to stu-dents in Hancock, Clear Spring andWilliamsport. The prevalence of histo-plasmin sensitivity was expected tobe low in the western area around Han-cock, intermediate farther to the east atClear Spring, and high in Williamsporton the south central border of the county.This expectation was subsequently con-firmed; consequently, testing with coc-cidioidin in only these three areas wasthought to be sufficient to indicate thefrequency and nature of cross-reactionsbetween histoplasmin and coccidioidinin the county.

Tuberculin was administered intracu-taneously on the volar surface of theright forearm, histoplasmin similarlyon the left forearm. Coccidioidin wasalso given on the left, half of the testsbeing given above the site of injectionof histoplasmin and half below. Glasssyringes and platinum needles wereused. The needles were sterilized byflaming between each test. Syringes wereused for only one antigen.

At approximately 72 hours after test-ing, the test sites were carefully exam-ined. The transverse diameter of anyinduration was measured and recordedin millimeters. Eighty-six per cent ofthe students had their tests read by thesame nurse. When more than one readerwas needed to complete the day's work-load, a second nurse helped with thereadings, but insofar as possible sawonly those children who had lived out-side the county at some time. For chil-dren with two tests on the left forearm,the readers were not allowed to knowwhich reaction was due to coccidioidinand which was due to histoplasmin.

Tests were read for 7,787 students,76 per cent of the total enrollment. Lessthan 2 per cent were non-white, and allbut 2 per cent were between the ages of12 and 19.

All survey records were matchedagainst lists from a special census donein the summer of 1963 by the JohnsHopkins School of Hygiene and PublicHealth, National Cancer Institute andWashington County Health Department.Information from the census includedthe type of house and water supply,and the presence of domestic animals onthe premises. Also included was the datewhen families moved into their presenthomes. If both these data and the resi-dence history taken during the surveyshowed that not more than one year hadbeen spent in another dwelling, the stu-dent was classified as a lifetime one-house resident.

Detailed maps available from thecensus and previous work by the Na-tional Cancer Institute allowed very ac-curate location of homes throughout thecounty. The type of soil on which thehouses of the subjects were located wasdetermined from soil maps publishedby the U.S. Department of Agriculture,based on examinations by soil scientistsof every field and parcel of land in thecounty (23).

In 1965, it was decided to begin stud-ies involving soil cultures for H. capsu-latum. As an initial step, locations whichseemed likely to yield positive cul-tures were selected. These were 14elementary schools in areas of highprevalence of histoplasmin sensitivityand all homes, 86 in number, in whichtwo or more lifetime residents had beentested and in which all were positivehistoplasmin reactors. From each site,soil was collected from numerous areas.Samples were taken from the superficial

198 COMSTOCK, VICENS, GOODMAN AND COLLINS

TABLE 1

Correlation of sizes of reactions to coceidioidin and histoplasmin among studentsin selected schools in Washington County, Maryland

Mm induration to oocoidi-oidin

0-2-4-6-8-

10-12-14-18-

TotelaNumberPeroent

0-

801

311

1

1

80711.8

2-

1

10.1

4-

20

203.2

Mm Induration

6-

83

11

884.7

8-

1234311

13311.2

10-

10447

3

30717.J

to histopl&imin

12-

1474313

18713.3

14-

68

3

708.0

16-

18

1

101.0

18-

3

1

40.3

10-

3

1

30.3

22-

1

10.1

Tot»l

Number

1,138122348

1

1

1,182

Per cent

98.11.01.90.30.8

0.1

0.1

100.0

layers of the soil. Whenever possible,the selected areas were moist, shadedand likely to be contaminated withbird droppings. Between 10 and 15pounds of soil were collected from eachsite, and treated as a single sampleafter mixing.

Each sample was examined for H.capsulatum in two ways. After beingallowed to ran in cages in which thefloor was covered with soil, mice weresacrificed and examined for H. capsu-latum. The soil was also examined byintraperitoneal injection into mice, afterthe soil had been exposed to high hu-midity in a closed vessel (24). Char-acteristics of sites were not disclosedto the laboratory group until the exam-inations had been completed.

RESULTS

Association of histoplasmin sensitiv-ity with personal and household char-acteristics. Of the 1,182 students testedwith coceidioidin, only 6 had reactionsto this antigen which were larger thantheir reactions to histoplasmin, as shownin table 1. Even if clerical error could

be excluded, it does not appear that in-fection with Coccidioides immitis is asignificant problem in WashingtonCounty. More important, there was noindication of the type of sensitivityfound by Palmer, Edwards and All-father to be characteristic of the areaaround east Texas, with small reactionsto both histoplasmin and coceidioidinsuggesting that the primary source ofsensitivity to these antigens was infec-tion with some other organism (25).

The distribution of all the tested stu-dents by size of reaction to histoplasminwas clearly bimodal, as is shown infigure 1. Slightly less than half of thestudents had no induration to histoplas-min. Among those with some induration,the mean reaction size was 10.3 mm, avalue almost identical to that foundamong Navy recruits from states in theso-called endemic area (25). Althoughit would be justifiable with a distribu-tion such as this to define a positive re-actor as a person with 2 or more mm ofinduration, it was decided to abide bythe usual, if somewhat arbitrary, defini-

DIFFERENCES IN DISTRIBUTION OF SENSITIVITY TO HISTOPLASMIN 199

tion of a positive reactor as a personwith 5 or more mm of induration.

There were appreciable differencesin the prevalence of histoplasmin sensi-tivity according to the residence historytaken at the time of the survey. This isshown in table 2. Students who had livedless than one year outside of WashingtonCounty had the highest frequency ofpositive reactions, somewhat more thanhalf of them being classified as positive.Those who had moved into the countybut had not lived outside of Marylandhad a lower frequency of positive re-actions; those coming from otherstates ranked lowest in this respect,only one-third having positive reactionsto histoplasmin. Lifetime residents ofthe county who had moved from place toplace had a different frequency of sensi-tivity than those who had lived in onlyone area. Students coming from a high-prevalence area to a low-prevalencearea were more likely to be positivereactors than the residents of the low-prevalence area. The reverse was true forstudents moving from a low- to a high-prevalence area.

To obtain the best possible index tofactors possibly related to the geographicdistribution of histoplasmin sensitiv-ity, the subsequent analysis is restrictedto children who were classified as life-time residents of one house. The geo-graphic distribution of histoplasmin sen-sitivity based on this group of 1,450students is shown in figure 2. The basicunits for this figure are socio-economicareas in the city of Hagerstown andelection districts in the rest of the county.The fundamental pattern is not changedby the use of enumeration districts orelementary school districts. Only 10 percent of the children in the westernmostdistrict were infected. Approximately

48.2

0 5 10 15 20INDURATION (mm)

FiouBB 1. Percentage distribution of aiiesof histoplaamin reactions among 7,787 stu-dents tested in Washington County, Md.,1963.

TABLE 2

Prevalence of histoplasmin reactors by residencehistory at time of survey

Rexidence other

Totals*Leas than 1 year else-

whereOther counties of

MarylandOther states or coun-

tries

Total

7,7876,925

370

1,489

5 + mm tohistopl&smin

Number

3,9343,271

162

501

Percent

50.555.2

43.8

33.6

* Includes 3 with no residence history.

one-third of those in the two northeast-ern districts were reactors, but there wasa sharp division in these areas. Chil-dren living in the eastern portion ofeach district had almost no sensitivity;however, it was high among childrenfrom the western half of these districts.The city of Hagerstown is the rectangu-lar area of moderate to low prevalencetoward the upper right portion of thecounty. Throughout most of the city,prevalence ranged from 33 to 45 per

200 COMSTOCK, VICEN8, GOODMAN AND COLLINS

PREVALENCE OF HISTOPLASMIN SENSTTMTY AMONG LIFETIME ONE-HOUSE RESIDENTSATTENDING JUNIOR AND SENIOR HIGH SCHOOLS OF WASHINGTON COUNTY, MARYLAND

FMUBB 2. Map of Washington County showing prevalence of histoplaamin sensitivityby geographic subdistricta.

cent, except for the northernmost areaand a keystone-shaped segment of theelection district to the north of thecity, which is a residential suburb,where the prevalence was only about25 per cent. Except for the fact thatthese two low-prevalence areas are resi-dential and wooded, there are no obviousphysical differences from the farmlandto the north with scattered houses andsome trailer homes along the main high-way. In this rural part of the electiondistrict directly north of the city, 96per cent of the children were reactors. Inaddition, it can be seen that there areother bands of very high prevalenceacross each arm of the L-shaped county.

No features of climate, topography oragriculture can be related to these ex-treme variations in the frequency ofhistoplasmin sensitivity. WashingtonCounty is located at the narrowest por-tion of western Maryland where at onepoint only a few miles of the countyseparate Pennsylvania from WestVirginia. The southern boundary of thecounty is formed by the Potomac river,and all of the county lies within itswatershed, with most of the creeks me-

andering southward to the Potomac.The eastern edge of the county is formedby the western slopes of the Blue RidgeMountains, containing the highest pointof the county (elevation, 2,145 feet). Attheir foot is the Hagerstown Valley,part of the Great Limestone Valley ofeastern states. It is approximately 20miles wide at this point, with an eleva-tion of 300 to 600 feet. The valleyfloor is generally level to rolling, withlocal ridges and frequent outcroppingsof limestone. The western half of thecounty, as measured along the straightnorthern border, extends into the east-ern slopes of the Appalachian Moun-tains, and consists of a series of narrowvalleys and ridges, ranging up to 2,000feet high. Most of the county is welldrained, and there are no swamps ormarshes.

Sixty per cent of the acreage in Wash-ington County is suited for regularcultivation, 15 per cent is limited, and12 per cent is fit only for pasture. Theremaining 13 per cent is mostly roughand rocky, and some of it is severelyeroded and mountainous (23). Dairyfarming and orchards comprise the ma-

DIFFERENCES IN DISTRIBUTION OF SENSITIVITY TO HISTOPIiASMIN 201

jor portions of the county's agriculture,the former located mostly in the Hagers-town Valley, and the latter concentratedin the westernmost district and thenorthern half of the Blue Ridge slopesin the east. Poultry raising, whilecommon, is not a major component oflocal agriculture (26).

The 1,450 lifetime residents of onehouse came from 1,121 households. Therewere 851 children who were the onlymembers of their household to be tested.The remaining 599 came from 270 house-holds, and as could be predicted fromthe geographic distribution of reactors,there was a marked tendency for allchildren in a household to react similarlyto histoplasmin.

At all ages, boys were slightly morelikely than girls to be positive reactors,the overall prevalence for boys being56.7 per cent, and for girls 54.5 per cent.Prevalence increased markedly withage, ranging from 46.4 per cent amongthose born in 1951 to 65.1 per centamong those born in 1946. The proba-bility of a resident becoming infected,as estimated from the age-prevalencecurves, varied markedly with geographicarea. Where the overall prevalence wasmore than 83 per cent, the estimatedrate of acquiring new infections wasnearly 20 per cent per year. In areaswhere only 18 to 33 per cent of the stu-dents were infected, the estimated in-fection rate was between 1 and 2 percent per year. In the westernmost area,there was no increase in prevalencewith age.

The use of crude prevalence ratiosdid not seem appropriate for the analysisof the relationship of histoplasmin sen-sitivity to various housing factors. Be-cause 36 per cent of the residents of thecity of Hagerstown and 63 per cent ofthe non-city residents were reactors, it

TABLE 3

Adjusted* prevalence of histoplasmin reactionsamong lifetime residents of one house, by year

house was built

Yeai built

Before 19001900-19441945-1954

ToUltoted

410632361

5+ Tnm tohiitopluniin

Number

276324168

Adjmtedper cent*

58.156.447.8

* See text for method of adjustment.

is obvious that anything associated withrural living would automatically show apositive correlation with prevalence ofhistoplasmin sensitivity. To minimizethis built-in association, the followingprocedure was adopted. The geographicdistricts were stratified with respect tofrequency of histoplasmin sensitivity inthe same subgroups used for figure 2,namely 0-17, 18-33, 34-50, 51-67, 68-83and 84-100 per cent. It was reasonedthat any factor that might be causallyassociated with histoplasmin sensitivityshould be manifest within each stratum,and that a stratum-adjusted rate wouldbe the best summary of the strength ofthat association. Stratum-adjusted ratesare used for tables 3 through 7. In theseand subsequent tables, the totals are notalways the same because persons forwhom pertinent information was notavailable have been omitted.

The prevalence of histoplasmin sen-sitivity among students according to theage of the house in which they lived isshown in table 3. There is a modesttrend toward increased likelihood ofbeing a reactor to histoplasmin forresidents of old houses. This findingmight signify that dusts containingHistoplasma capsulatum are more com-mon in or around old houses, possibly as

202 COMSTOCK, VICENS, GOODMAN AND COLLINS

TABLE 4

Adjusted prevalence of hisloplasmin reactionsamong lifetime residents of one house, by

material used for outside walls

Materialfor walls

StoneWood & atoneAsbestos shingleBrickWoodConcrete

Totaltetted

6572

326387361118

5-f- mm to

Number

6144

17621018963

Adjustedpercent

60.268.254.964.053.849.6

TABLE 5

Adjusted prevalence of hisloplasmin reactionsamong lifetime residents of one house, by type

of basement floor

Basementfloor

ConcreteOther

Totaltested

862463

5+ nun tohistoplasndn

Number

436264

Adjustedprevalence

52.665.5

TABLH 6

Adjusted prevalence of hisloplasmin reactionsamong lifetime residents of one house, by source

of drinking water

Source ofwater

SpringDeep wellCity systemCisternShallow well

Totaltested

10627372720074

5 + mm tohiitnplsuntfn

Number

7816831913245

Adjustedprevalence

62.853.251.851.549.6

the result of accumulated exposures tobirds or bate over the years, or that oldhouses were more likely to have beenbuilt in areas of the county where con-ditions are favorable for the growth of

or infection by the organism. Relation-ship of material used for house walls tohistoplasmin sensitivity is shown intable 4. Prevalence was highest amongresidents of stone houses, and lowestamong residents of houses with concretewalls. This finding may reflect the samefactors as age of house, for with theexception of a number of log buildingsstill in use, many of the older homesare constructed of stone while concretewalls are found only in more recentlybuilt homes. Because of findings asso-ciated with two epidemics of histoplas-mosis, it was thought that conditionsfavorable to infection with H. capsvla-tum might be found in basements with-out concrete floors (21). However, asshown in table 5, the evidence did notfavor this hypothesis.

H. capsidatum is reported to be ableto grow in water (27, 28), and bird drop-pings are thought to improve the likeli-hood of its growth (18). Because manycisterns are filled with rainwater fromroofs, subject to contamination withspores and bird droppings, it seemedpossible that infections with histoplas-mosis might be more common amongstudents whose drinking water camefrom cisterns. This turned out not to beso, as can be seen in table 6. Studentswhose drinking water came from springsshowed a higher than average preva-lence of positive reactions, which mayreflect the same factors as those asso-ciated with old houses, for the earlyhomes were often located near springs.Another possibility is that the soil nearsprings may be damp, and dampness isbelieved to favor the growth of H. cap-sulatum (6, 17,29-31).

A final attribute available from thecensus records was the presence or ab-sence of certain domestic animals on theproperty. Their relationship to the

DIFFERENCES IN DISTRIBUTION OF SENSITIVITY TO HLSTOPLASMTN 203

TABLB 7

Adjusted prevalence of histoplasmin reactions among lifetime residents of one house, by presenceor absence of domestic animals on premises

Anjnrl

HorsesChickens or birdsCattleCateDogs

Animal present

Total tested

79410241674889

5+ mm to hiBtoplasmiB

'Samba

63262180415514

Adjustedprevalence

62.660.259.556.756.2

Animal absent

Total tested

1,3501,0181,188

754539

5+ mm to hiitoplasmln

Number

715515598362263

Adjustedprevalence

54.052.453.051.961.4

frequency of reactors is shown in table 7.In all instances, the presence of theseanimals—horses, chickens or birds, cat-tle, cats and dogs—was associated witha slightly increased prevalence of histo-plasmin sensitivity, whether consideredsingly or in combination. However, inno instance was this association marked,and horses were as closely associatedwith histoplasmin sensitivity as chickensor birds. This finding lends little sup-port to the widely held theory that birddroppings are highly associated withhistoplasmosis (18).

Association of histoplasmin sensitivitywith soil characteristics. Because it isgenerally agreed that H. capsulatumis an inhabitant of the soil, it seemedreasonable to assume that ite presencemight be correlated with certain soilcharacteristics. And if this were so,children whose homes were located onsoils conducive to the presence of theorganism should be more likely to beinfected than children whose homeswere located on unfavorable soils. Thus,a clue to soils that might be best suitedto the growth of H. capsulatum couldbe provided by the prevalence of histo-plasmin sensitivity among children liv-ing in areas with different soil types.

Because the locations of all houses

had been accurately recorded for the1963 census, the type of soil on whichthey were situated could be determinedfrom detailed soil maps published bythe U.S. Department of Agriculture, ex-cept for houses in Hagerstown and itsbuilt-up suburbs (23). The analysis wasrestricted to homes of children who hadbeen lifetime residents of one house.These homes were located on 75 differentsoil types but many soils were repre-sented by very small numbers of chil-dren, and only 15 soil types were repre-sented by 20 or more children each. Theprevalence of histoplasmin reactorsamong children residing on these 15soils ranged from 88.0 per cent to 16.7per cent, with no pattern discernible toinspection.

There was less but still considerablevariation in the prevalence of reactorsliving on the great soil groups. Theseresults are shown in table 8. The highestprevalence was found among residentson gray-brown podzolic soils, and thelowest among those living on red-yellowpodzolic soils. One difference in thesetwo soil groups is the degree of acidity,the former being neutral or slightly acid,the latter tending to be strongly acid.

The prevalence of reactors accordingto the parent material of the soils is

204 COMSTOCK, VICEN8, GOODMAN AND COLLINS

TABLE 8

Prevalence of hittoplasmin reactions amonglifetime residents of one house in rural

areas, by great soil group

Great >oil

Gray-brown pod-zolic

AlluvialGray-brown & red-

yellow pod^olioLdthosolsRed-yellow podzolic

Totaltested

566

3648

124122

5+1Tmi t o

hiitoplumin

Number

407

2429

6651

Percent

71.9

66.760.4

63.241.8

TABLE 9

Prevalence of histoplasmin reactions amonglifetime residents of one house in rural areas,

by parent material for soil

Parent material

Micaceous schista& phyllites

LimestonesLimestone with

sandstone or shaleQuartziteSandstonesShalesCrystalline rock ma-

terialsMetabasaltShales and sand-

stones

Totaltoted

27

323

2134832

12225

6474

5+ mm tohistoplasnin

Number

24

240

15829195912

2629

Per cent

88.9

74.3

74.260.469.448.448.0

40.739.2

shown in table 9. Soils derived fromlimestones tended to have a higher pro-portion of reactors than almost all ofthe others.

A similar conclusion emerged whenthe prevalence of reactors was studiedaccording to the agricultural capabilityof the soils. There was no correlationwith suitability for different types ofcrops or other agricultural uses. Of the

soil capability units represented by 20or more tested children, the units withthe highest and lowest proportion of re-actors were both steep and eroded. Forthe high-prevalence unit, it was notedthat "rockiness consists of limestoneoutcroppings", while the low-prevalenceunit was not associated with limestone.Only 15 children lived on soils consideredto be limited because of wetness and poordrainage, but it should be noted that 11or 73 per cent of them were reactors,while reactors comprised only 60 percent of those living on soils limited byerosion.

In keeping with the foregoing findings,there was a marked correlation of theprevalence of reactors with pH of thesoil (table 10). Only a third of the life-time residents on strongly acid soilswere positive reactors, while amongthose on alkaline soils almost 90 percent reacted to histoplasmin.

No discernible correlation of preva-lence of reactors could be noted with thesuitability of the soil for forestry, irri-gation or sewage disposal. A similar lackof correlation was noted with variousengineering characteristics of the soilssuch as weight bearing capacity, dis-persion, and shrink-swell potential.

TABLE 10

Prevalence of histoplasmin reactions amonglifetime residents of one house in rural areas,

by pH of soil type on which housewas located

Median pH

7.0 and over6.5-6.96.0-6.45.5-5.9

Under 5.5

Totaltested

2777

234156181

5+ nun tohlstoplumin

Number

2457

18010266

Percent

88.974.065.665.536.5

DIFFERENCES IN DISTRIBUTION OP SENSITIVITY TO HISTOPLASMIN 205

Soil cultures for Histoplasma cap-sulatum. None of the soil specimens fromschools was positive. From the 86 speci-mens collected around houses in whichall the tested lifetime residents werepositive histoplasmin reactors, there were11 which yielded H. capsulatum. Therewas a marked association of a historyon the 1963 census regarding the pres-ence or absence of chickens and domesticbirds with the proportion of positiveisolations (table 11). H. capsulatumwas isolated from 26 per cent of thesites with chickens or birds, and fromonly 2 per cent of those without birds.

Because of the manner in which thesampling sites were selected, there wasrelatively little variation in the greatsoil types represented. However, thevariation in pH was sufficient to makeanalysis worthwhile. The results areshown in table 12, both according to themedian pH of the soil type reportedby the U.S. Department of Agriculture(23), and by the pH found on examina-tion of the sample collected. Becausethe sampling extended into the built-upsuburbs of Hagerstown, it was notpossible to estimate the soil type for 12of the sampled sites. However, in con-trast to the association of alkaline pHwith histoplasmin sensitivity, there wasno evidence of any association of medianpH with the frequency of isolations.

TABLE 11

Isolations of Histoplaama capsulatum fromselectedsitesinWashington County, by presenceor absence of chickens or other domestic

birds on premises

TABLE 12

Isolations of Histoplasma capsulatum fromselected sites in Washington County, by pH ofsoil type on which house was located and by

pH of sample collected

Chickens ordomestic birds

PresentAbsent

Sitescultured

3843

Isolations

Number

101

Percent

26.32.3

7.0 and over8.5-8.08.0-6.45.5-S.8

Under 8. S

Totals

pH of soil type

Sitescultured

418831110

74

Twnla rinn*

Num-ber

13331

g

Percent

35.018.88.1

18.310.0

12.2

pH of sample

Sitescultured

3833ISIS8

88

Isolations

Num-ber

03361

11

Percent

0.08.1

30.033.313.5

12.8

Furthermore, when pH was determinedby actual measurement of the samples,it was found that isolations were mostfrequent from soils of moderate acidity,a result in keeping with those of others(16,32).

DISCUSSION

A number of fortunate circumstancescombined to strengthen the presentstudy. First was the ability to limit theanalysis to a large number of childrenwho were lifetime residente of singlehouses which could be accurately locatedon detailed maps. Basing residence his-tory on questions asked in different waysat two different times increased thecertainty that the sample represented ahighly stable population. The studyvariables were collected entirely inde-pendently. Skin teste were measuredwithout knowledge of residence or otherhistorical information. Because most ofthe tested children attended consolidatedschools representing large geographicareas, the reader of the histoplasmintests could identify the area of residenceonly within very broad limits. Informa-tion about housing, soil types, and ex-posure to animals was collected inde-

206 COMSTOCK, VICENS, GOODMAN AND COLLINS

pendently of the skin test survey. Soilspecimens were identified only by codenumbers, so that soil isolations wereperformed without knowledge of thecharacteristics of the sites from whichan individual specimen had come.

A major strength was the wide rangeof frequencies of histoplasmin reac-tors among junior and senior high schoolchildren in Washington County. Thedistribution of sensitivity within thecounty bore no relation to any obvioustopographic feature. Areas of very highand low prevalence were found alongthe Potomac river valley; high and lowprevalence areas were also found alongthe Blue Ridge mountains. The lowprevalence of sensitivity within the cityof Hagerstown and one of its built-upsuburbs to the north afforded the onlyclearcut association with any discerniblegeographic feature. The other built-upsuburbs east and southwest of the cityshowed moderate to high prevalence ofsensitivity among the lifetime residents.There was no correlation with the majortype of farming practiced.

The marked variation in sensitivitywithin short distances in itself indicatesthat climatic or meteorologic factors donot directly affect the probability ofhuman infections except in the sense ofimposing broad limitations upon thegrowth or infectivity of the organism.Man's disturbance of the soil, indicatedas the cause of human infections inurban epidemics (12), could not beassociated with the observed patternsof histoplasmin sensitivity in Washing-ton County. The only major disturb-ances of soil during the lifetime of thetested children involved construction ofportions of three highways, an airportrunway, a large industrial plant, a shop-ping center and two high schools. Theairport runway is located in the north-

west corner of the very high-prevalencearea north of Hagerstown, and withgenerally westerly winds could havebeen related to the sources of these in-fections. All of the other sites are eitherlocated within low-prevalence areas, orare so situated that prevailing windconditions should have exposed highand low prevalence areas equally. Homeconstruction with its attendant soil dis-turbance has been most marked in thesuburbs to the north, east and south-west of Hagerstown. The considerablevariation in prevalence in these areashas already been noted. Most strikingis the keystone-shaped suburb to thenorth of the city, which has remaineda low-prevalence area in spite of con-siderable construction in a wooded sec-tion, in marked contrast to the veryhigh prevalence around it. It is alsonoteworthy that the prevalence of his-toplasmin sensitivity among boys isonly very slightly greater than amonggirls, in spite of the fact that the farmboys are heavily exposed to dust duringplowing and harvesting.

A further inference can be drawn fromthe striking variation in frequency ofhistoplasmin sensitivity relative to thelocation of the children's homes. Evenchildren who have lived in one house alltheir lives lead highly mobile lives. Ifexposures away from home were impor-tant determinants of the likelihood ofbecoming infected, such marked varia-tions could not be demonstrated byplace of residence. Something very closeto home, applying equally to boys andgirls, must be the major source of in-fections in this part of the country, andpossibly elsewhere too.

Although studies in WashingtonCounty based on isolations of H. capsu-latum are limited, confidence in theresults is increased because they are

DIFFERENCES IN DISTRIBUTION OF SENSITIVITY TO HISTOPLASMIN 207

consistent with those of others (16, 18,32). The presence of chickens or otherdomestic birds on the premises greatlyincreased the likelihood that H. capsuia-tum could be demonstrated in the soil.Isolations were also more frequent whenthe soil samples had a moderately lowpH.

A striking finding of this study is thathuman infections, as indicated by reac-tions to the histoplasmin skin test, didnot follow the pattern observed for soilisolations. Positive histoplasmin reac-tions were only slightly more common onpremises where chickens or other domes-tic birds were kept, and were morelikely to be found among children wholived on alkaline than on acid soils.Sufficient numbers of children wereexamined to make it highly unlikely thatthese observations resulted from chance.It thus appears that environmental fac-tors favoring sensitization of humans byH. cap8idatum must be different fromthe factors favoring isolation of theorganism from soil samples with currenttechnics.

To reconcile the discrepancies be-tween the observed distribution of H.capsidatum and the distribution of his-toplasmin sensitivity requires specula-tion, which to some extent has beenput forth by others (33). Let us assumethat conditions favoring the growth ofH. capsidatum also reduce the likeli-hood that a given dose of organisms willcause an infection among humans livingin the area, and that conditions makinggrowth more difficult but not impossibleincrease the likelihood of the same num-bers of organisms causing human in-fection. Let us further assume thatlaboratory isolation of the organism ispossible from sites where growth con-ditions are favorable, but that our pres-ent technics do not allow successful iso-

lations under conditions where growthis difficult. Among the various charac-teristics of the organism which might berelated both to the ease of isolation andto its ability to infect humans is theformation of spores of varying sizes.

It is generally accepted that inhaledparticles larger than 5 microns in diam-eter rarely reach the pulmonary al-veoli, being trapped in the respiratory-passages along the way. Particles lessthan 0.1 micron reach the alveoli readilybut usually pass out again with the ex-haled air. The optimal size for alveolarretention appears to be on the order of1 micron, depending presumably ondensity of the inhaled particle as wellas on its size (34-35). The spores ofH. capsulatum are rarely larger than16 microns in diameter. Most are lessthan 5 microns but considerable varia-tion in the distribution of spore sizeshas been reported (33, 37, 38). In oneseries of 14 strains cultured on artificialmedia for 1 month, the proportion ofspores approximating 1.6 microns indiameter varied from 9 to 83 per cent(33). The infectivity of these strains,if the spores were to become airborne,would presumably show similar varia-tion.

The numbers of spores produced in agiven situation could also affect infec-tivity. Even with a growth situation thatproduced a very high proportion ofspores too large to reach the alveoli, thenumbers of small spores could reachlevels high enough to result in a highrisk of infection if vast numbers ofspores were formed. This sort of situa-tion could fit the epidemics of histoplas-mosis where it is often relatively easyto isolate the organism and exposedhumans are very likely to become in-fected.

This line of speculation is partially

208 COMSTOCK, VICENS, GOODMAN AND COLLINS

substantiated by the few available facts.Experience with dusts and other or-ganisms makes it reasonable to believethat small spores are most likely tocause airborne infections. Unfortu-nately, almost nothing is known aboutthe types of spores found under naturalconditions, largely because only thelarge tuberculated spores are readilyrecognizable microscopically, and be-cause the effect of culturing technics onspore formation and size is not known.It does appear from one study thatmarked variation in spore size mayoccur within the endemic area of theUnited States (33). But another studythat also showed variation in sporesizes between strains is not consistentwith the hypothesis set forth in thispaper. Emmons found that strains iso-lated from south Georgia where humaninfections are rare produced largenumbers of small spores, while strainsfrom Loudoun County, Virginia, anarea with a high frequency of humaninfections, had a higher proportion oflarge spores (37).

However, the major purpose of thisspeculation is not to suggest that sporesize is the important determinant of thepathogenesis and epidemiology of histo-plasmosis. By emphasizing again thatconditions favoring the isolation of theorganism from the soil are not neces-sarily the same as those favoring humaninfection, it is hoped that investigationsinto the basic biology of H. capsulatumwill be encouraged. It is likely thatmuch more will need to be learnedabout the fundamental growth habite ofthe organism before the epidemiology ofhistoplasmosis can be fully understood.

REFERENCES

1. Meleney, H. E. HistoplasmosiB (reticulo-endothelial cytomycosis): A review.Amer. J. Trop. Med., 1940, SO: 603-616.

2. Smith, C. E. Coccidioidomycosis. Med.Clin. N. America, 1943, £7: 790-S07.

3. Palmer, C. E. Nontuberculous pulmonarycalcification and sensitivity to histoplas-min. Pub. Health Rep., 1945, 60: 513-520.

4. Manoa, N. E., Ferebee, S. H. and Kersch-baum, W. F. Geographic variation in theprevalence of histoplasmin sensitivity.Dis. Chest, 1956, £9: 649-668.

5. Edwards, P. Q. and Palmer, C. E. Na-tionwide histoplasmin sensitivity andhistoplasmal infection. Pub. HealthRep., 1983, 78: 241-259.

6. Zeidberg, L. D., Dillon, A. and Gass, R.S. Some factors in the epidemiology ofhistoplasmin sensitivity in WilliamsonCounty, Tennessee. Amer. J. Pub.Health, 1951, 4?.-80-89.

7. Sachs, D., Smith, R. T., Fleming, D. S.and Furcolow, M. L. The prevalence ofpositive reactions to tuberculin and his-toplasmin in a rural Minnesota county.Amer. J. Hyg., 1955, 6t: 43-53.

8. Grayston, J. T., Heeren, R. H. andFurcolow, M. L. The geographic dis-tribution of histoplasmin reactors amongschool-age children within a rural Iowacounty. Amer. J. Hyg., 1955, 6£: 201-213.

9. Furcolow, M. L. Recent studies on theepidemiology of histoplasmosis. Ann.N.Y. Acad. Sci., 1958, 76: 127-164.

10. Wilcox, K. R. Jr., Waisbren, B. A. andMartin, J. The Walworth, Wisconsinepidemic of histoplasmosifl. Ann. Intern.Med., 1958, Jfi: 388-418.

11. Aronson, D. L. and Edwards, P. Q. Anurban focus of histoplasmin sensitivity.Amer. Rev. Tuberc, 1959, 79: 83-86.

12. Tosh, F. E., Doto, I. L., D'Alessio, D. J.,Medeiros, A. A., Hedricks, S. L. andChin, T. L. The second of two epidemicsof histoplasmosis resulting from work onthe same starling roost. Amer. Rev. Resp.Dis., 1966, 94: 406-413.

13. Edwards, L. B., Peeples, W. J. and Berger,A. G. Prevalence of sensitivity to tu-berculin and histoplasmin among highschool students in Montgomery County,Maryland. Pediatrics, 1958, SI: 389-396.

14. Chase, H. V. and Campbell, C. C. Histo-plasmin «Vin test survey of elementaryschool children in Frederick County,Md. JAM A., 1962,186: 335-338.

15. Menges, R. W., Furcolow, M. L., Selby,

DIFFERENCES IN DISTBTBUTION OF SENSITIVITY TO HIBTOPLARMTN 209

L. A., Habennan, R. T. and Smith, C.D. Ecologic studies of histoplasmosis.Amer. J. Epid., 1967, 86: 108-119.

16. Zeidberg, L. D., Ajello, L., Dillon, A. andRunyon, L. C. Isolation of Histoplasmacapsulatum from soil. Amer. J. Pub.Health, 1962, Jfi: 930-936.

17. Zeidberg, L. D. and Ajello, L. Environ-mental factors influencing the occurrenceof Histoplasma capsulatum and Micro-sporum gypseum in eoil. J. Bact., 1954,68: 156-159.

18. Ajello, L. Relationship of Histoplasmacapsulatum to avian habitats. Pub.Health Rep., 1964, 79: 266-270.

19. Furcolow, M. L. Environmental aspects ofhistoplasmosis. Arch. Environ. Health,1966,10: 4-10.

20. Lehan, P. H. and Furcolow, M. L. Epi-demic histoplasmosis. J. Chronic Dis.,1957, 6: 489-503.

21. Furcolow, M. L. and Grayston, J. T. Oc-currence of histoplasmosis in epidemics.Etiologic studies. Amer. Rev. Tuberc.,1953, 68: 307-320.

22. Palmer, C. E. and Edwards, L. B. Personalcommunication, 1963.

23. Matthews, E. D. Soil survey of Washing-ton County, Maryland. UJ3. Dept. ofAgriculture, Soil Conservation Service incooperation with Maryland AgriculturalExperiment Station, Series 1959, No. 17,UJ3. Government Printing Office, Wash-ington, D.C., 1962.

24. Goodman, N. L. Unpublished data, 1964.25. Palmer, C. E., Edwards, P. Q. and All-

father, W. E. Characteristics of skin re-actions to coccidioidin and histoplasmin,with evidence of an unidentified sourceof sensitization. Amer. J. Hyg., 1957, 66:198-213.

26. Washington County Planning and ZoningCommission. Economic Survey, Wash-ington County, Maryland. Master PlanReport Number Two. Hagerstown,Maryland, 1960.

27. Furcolow, M. L. and Horn, H. W. Air andwater in the natural history of Histo-plasma capsulatum. Proc. Conf. on His-

toplasmosis 1962. Public Health Mono-graph No. 39, 1956, pp. 282-288.

28. Gordon, M. A., Ajello, L., Georg, L. K.and Zeidberg, L. D. Microsporum gyp-seum and Histoplasma capsulatumspores in soil and water. Science, 1962,116: 208.

29. Larsh, H. W., Hinton, A. and Cozad, G.C. Natural reservoir of Histoplasmacapsulatum. Amer. J. Hyg., 1956, 03:18-27.

30. Gooe, R. D. Growth and survival of His-toplasma in soil. Canad. J. Microbiol.,1965, .M: 979-985.

31. Goodman, N. L. Environmental studies ofHistoplasma capsulatum. DissertationAbstracts, Ann Arbor, Michigan, 1965.

32. Zeidberg, L. D , Ajello, L. and Webster,R. H. Physical and chemical factors inrelation to Histoplasma capsulatum insoil. Science, 1955, lti: 33-34.

33. Cozad, G. C. and Furcolow, M. L. Labora-tory studies of Histoplasma capsulatum.II. Size of spores. J. Infect. Dis., 1963,9S: 77-84.

34. Harrington, D. and Davenport, S. J. Re-view of literature on effects of breathingdusts with special reference to silicosis.U.S. Department of the Interior, Bureauof Mines, Bulletin 400, 1937, pp. 70-72.

35. Wells, W. F. Airborne contagion and airhygiene. An ecological study of dropletinfections. Published for the Common-wealth Fund, Harvard University Press,Cambridge, Mass., 1965, pp. 118-120.

36. Brown, J. H., Cook, K. M., Ney, F. G. andHatch, T. Influence of particle size uponthe retention of particulate matter inthe human lung. Amer. J. Pub. Health,1950, 40.-46<W58;48O.

37. Emmons, C. W., Morlan, H. B. and Hill,E. L. Histoplasmosia in rats and skunksin Georgia. Pub. Health Rep., 1949, 64:1423-1430.

38. Helmbright, A. L. and Larsh, H. W. Thesize of the spores of Histoplasma capsu-latum. Proc. Confer, on Histoplasmosis1952. Public Health Monograph No. 39,1956, pp. 81-83.