DOSIMETRY RADIATION

WHOlE HG/9S.16 English only Distr.: Limited

ULTRAVIOLET RADIATION EXPOSURE DOSIMETRY OF THE EYE

/~.··.:·'~;::'·~·:·'··F::·v·::··A:

.

.!':·~:;:::::I.~j.::.:''';:$,:,:.'~)/$;},::.);:;:;:^..::·:·,i'·.'·:.':.,; .

W"'OIEHG/9S.18 Engllsnonly DIlJtr.:Urnltecl

ULTUVlOLET UDIATION EXPOSURE DOSIMETRY OF THE EYE

-. I' ,

ID World Health Organization, 1995

TillsdocumentI~notIssuedt(llhe generalpUblic, andall rightsare reserved by theWol1d Haaltn Organlutlon (WHO). The document may nolbereviewed,abstracted, Quoted,te~t'OlfLlC::$d or ttanl;llated, Inpart or in whOle, withOut the prtor wnnen p6mllsslon of WHO. Nopanof Inls dO(:l,Imant maytH:I stOtad In a retrieval system or transmittedInany fOnnorbyanymeans - eleelronlc. mecnanlcal or otMr - wll1'lout the prior written permission of WHO.

Tileview elilpreS$ee;IIn documentsbynamadaomersare solelythe respon~[bilityQfthoSe authOrs.

WHO/EHG/95.18

ULTRAVIOLET RADIATION OCULAR EXPOSURE DOS/MErRY OF THE EYE

A Report

^to

the World Health Organization and Unitecl Nations Environment Programme

prepared by:

DavidH.Sliney,Ph.D.

Manager, Laser/Optical Radiation Program

US Army CenterforHealth Promotion and Preventive Medicine Aberdeen Proving Ground. MD 21010^{K 5 4 2 2}USA

ABSTRACT

Acareful studyof ocularexposure to environmental sunlight demonstrates thatitis not at all simple to accurately determine the level of solar ultraviolet radiation exposure of the human eye. Epidemiological studies of cataractde- pend heavily upon realistic and reasonably accurate dosimetry of UVR expo- sure. Unfortunately, most pastattempts to measure or calculateUVRexposure of the eyeforsuch studieshave generally relied upon monitors to measure the ambientUVRinsunlighL falling upona horizontal surface (i.e.,the globalUVR), This approach is insufficient to properly assess the large role of ground reflec- tion, the horizon sky contribution! the degree of lid openingand the extreme lateral component ofUVRincident on the eye. Current dosimetric estimates may leadto incorrect assignment of lifetime exposures to different cohorts.

This report summarizes a series of recent oculardosirnetrystudies under- taken La assess all of these factors. An algorithm is developed for calculating ocular exposure for a given environment AddilioI'1ally, the value of different types ofeye protection is shown to vary widely depending upon frame design.

Thedosimetry studiescan beconfirmed bya biological dosimeter - the human cornea. Because the action spectrum and threshold for the human cornea are well defined, the living cornea actually serves as a biological dosimeter to call- bratethis method for calculating ocular exposure. More accurate dosimetric methods are expected to aid in the resolution of the current controversy as to theeuologic role of UVR in cataract andother oculardiseases. Concerns about the depletion ofstratosphenc ozoneand the related increase in terrestrial UVR exposure have emphasized the importance of resolving this controversy.

KEY WORDS: Ultraviolet radiation,eye,cataract,UVdosimetry

""" '''·A

~," ' t:

'1'1':",

,~~:' ~\,.: f I,~~, J··r, ''11 ;

">,~~\ '.. } ». ..:1

, i , " ,'Iol ",~. ''."

WHO/EHG/95.IB

e

I. INTRODUCTION

1.1 Purpose

The objective of thisstudywas to evaluate the geometrical, physi- ological and environmental exposure factors which play a role in de- termining theactual ultraviolet radiation (UVR) exposure ofthe lens to a person when outdoors in daylight, and to provide an algorithm for calculating exposure in a given environment.

1.2 Background

The prevalence of the blinding disease of cataract worldwide ex- ceeds 50million. Prevention or slowing the progress of lenticular opacities is an important objective in public health [1]. Despite the fact that animal experiments clearly show that ultraviolet radiation (UVR)Can produce cataract under acute laboratory exposures [1~81,

refined epidemiological studies show an increased riskofcortical cata- ractwithUVB (280-315nm)exposure [9],and the derivation ofocu- lar exposure guidelines based upon these studies [10]^l the clear role ofUVRin cataractogenesis is still debated. Some experts argue that UVRplays a major role [11] and others suggest thatUVRdoes not [I2L To a large extent, the controversy is fuelled by poor ocular dosimetry The goal of the studies reported here was to enlarge our understanding of the relative importance ofseveral factors thatdeter- mine the ocular dose fromUVRin sunlight.

1.3 Ambient UVR Exposure

The ambient out-door UVRconstantly changes during theday.

Because these substantial changes arenot directly proportional to the much less dramatic changes in visible light; we aresubjectively largely unaware of the degree of these changes. For example, the Summer- time terrestrial solarspectral irradiance at a wavelength of300nm is ten times greater thanat eitherthree hours before Or three hoursafter solarnoon. An untanned person with fair skin would receive a mild

fit

UIlravioJet RadiationExposureDosimetrr.

01

theJye

sunburnin 25 minutes at noon,but would have to lietn thesun forat least two hoursto receive the same dose' after 3:00 (standard time)

r

13-15}. Theintegrated total exposure dose ofbiologically-weighted UVR fallingOn a horizontalsurface (the globalUVR exposure). oc- curs primarily during the middayhours,and 70% occurs during the fourhourscanteredonnoon-time zenith asshownisTable1 [13].

Because short-wavelengthUVRisstronglyscatteredbyatmospheric molecules,itisquitepossible to receive a sunburnwhilelyingin the shade,ifexposedtoalargesegment oftheskyAlthough mostvisible light isnotstrongly scattered,and well over 85% ofthe light received atground levelon a dear sunnydayare directrays,more than50% at 300nmisscatteredand diffused. Duetothestrongscatteringfactor in the UV spectrumit has beenconventional to distinguish between thedirectand thediffuse componentsin sunlight. The sum of the global and diffuse components ~ the total falling ona horizontal ground surface, is termed theglobal radiation1asshown in Figure 1.

Ifonecould see at 300 nm and see only thisradiant energy, the'sky would appear perpetually in aheavyhaze, andshadowswouldprob- ably not benoted. Forepidemiologicalstudies ofskin cancerandfor atmospheric science, normallytheglobal UVR hasbeenused. Thisis clearlyunacceptable forstudies ofocular exposure,sincethe eyes are only exposedto diffuse, horizonUVRand^togroundreflections.

Table 1. Calculated ACGIH Effective Global UV·B and Total UV-A Exposure Dose on aHorizontalSurface fromData of Bener: Davos, Switzerland in June 1131

£Xp05UUDuration :£creet~veUV-B TotalUV~.A. TotalIrradiance

Cantered on Noon in 315.-500 nm

(tJOl,lrs)("laof8h) U/em!}(%8h» (Jlcm!) (%8h) (estlm,Jl<;m¹)

2 hours (25%) 0.023 (40%) 39,1 (32%) 140 Pl%)

4 hours ^(50%) 0.042 (72%) 77.4 (63%) 290 (64%)

6 hours (7S%) 0,053 (91%) 105 (86%) 390 (8'%)

8hours (100%) 0.058 (100%) 122 POOO/",) 4$0 (100%) l:2hours ^(1500.10) 0.060 (103%) 145 (JIg%) 540 (1200;;;)

Note; The percentage of the total possible exposure relative to an8-hourday cantered on noonisprovided within the parentheses).

'6

.. V!HO/8IG/95.18 _

Figure 1. Global, Diffuse andDirect radiation. Global radiation is the sum of the direct and diffuse components (after Slineyand Wolbarsht,1980).

Thus,the geometryofexposureas well as the spectrum plays a major role in determining theocular effects from the UVR in sun- light.

Ocularexposureis farmore affectedbygeometrical factors than

isskin exposure.Thisisevidentby the observation that although the corneaismore sensitive toUVR injury than the skin, one seldom experiences«snowblindness» (photokerattns) when out in sunlight, even when oneexperiencesa «sunburn» of the skin. Fortunately,

persons seldom lookdirectlyoverhead when the sunis veryhazard-

ous to view although mostpeoplereadily stare atthesunwhenitis

comfortabletolookatnear the horizon, and because of atmospheric

filtering of DVRand blue light, the yellow-orange sunis then not hazardous toview [16]. Whenthe solar elevation angle exceedsap~

proximatelyl(}Oabove thehorizon, strongsquintingisobserved, and this behavioral pattern effectively shieldsboth the COrnea and the retinafromdirectexposure [16]. These factors have beenpreviously calculated to reducetheexposureofthe cornea to amaximumofabout 5% of thatfalling upon the exposed topof head (the globalDVirradi- ance) [14Lbut the exact fraction remains unknown. This conclu-

sionwas baseduponthe fact thatifonemathematically weightsthe

UltravioletRadiation ExposureDosimefry

of

Ihe Eye

"".j,,,

human photokeratitisaction spectrum with the solar spectral irradi- ance, the photokeratins thresholdisachieved in less that15minutes for the midday summer sun. Experience shows that the special ge·

ometry of exposure precludes photokeratitisexceptwhen ground re- flectance exceeds approximately15%. Furthermore,ifsquinting and otherbehavioral factors are not considered;the dose to the eyelid determinedby film-badgedosimerryisapproximately20%ofthe dose fallingOna horizontalsurface[17-19]. . .

Inaddition to the protective value of squinting, facial. features, such asthe eyelid, brow ridge, and cheek,mayfurtherlimit-the UVR .levels reaching the cornea and lens. While bone structure

and

^'other

facial characteristics are genetically determined, the eyelid opening, which varies with the scenebrightness, Or luminance, are instrumen- talinattenuating the amount of light which enters the eye. At the outset of this series of studies, it became dear that a correlation be- tween the luminance and theamount of lid closurewasneeded to determine the amount ofUVRwhich reaching the eye in different conditions.

1.4 Ocular Exposure Geometry

When the sunisoverhead andUVBexposureismost severe; the upper lid and the brow ridge shield the cornea. Iftheeyeisturned away from the sun, squintingisdecreased and the scatteredUVRfrom overhead striking the corneaincreases. When the skyissufficiently overcast? squinting is greatly reduced and overhead skylight Can be incidenton the cornea-but overhead rays strike the corneaata graz- ing angleofincidence where greater Fresnel reflection (and less ab- sorption) takes place [17]. Onlywhen the incidentUV rays are par- allel to the pupillary axis (as from snow reflections) are roost (ap- proximately 98%) incidentrays absorbed. The traditional eye pro- tector of thelnuitor Eskimoisa slit in whalebone or ina seal-skin mask to provide geometrical rather than spectral protection against the intenseUVRreflections from snow [16,20]. thelack of protec- tion above and to the sides of sunglasses can be a seriousshortcom~

ing. Coroneo has shown that vety obliquetemporal :rays can be re-

'r!~_ _. ~ ~ ,

W/IO/EHG!?5.J8

e

fracted into the critical nasal equatorial (genninative) region of the lens as shown in Figure2;and this could explain the increased inci- denceofopacities originating in this nasal sector of the lens in corti- calcataract 121-24]' Clearly, quantitative data for this componentof exposureto the eye have been needed [24J.

Theproblemthenremainsto quantify theprotective value ofthe upperandlowerlidswhen they dose down during squintingandto determine the ocular UVR exposure dose in different environments from sky radiation and from groundreflection. Furthermore, the ef- feet upon ocular exposure of a brimmed hat and other head wear require quantification. Although abrim can Virtually eliminate the overheadUVRexposuredose, the lid opening may increase, and the ground reflection becomes important. Several experimental ap- proaches are possible. Uv-sensuive contact-lens dosimeters could provide important information. Likewise, studies of lid opening of different peoples along withUVR directionalfieldmeasurements could be performed for different environmental conditions. Because of the difficultyofobtaining a sufficientlysensitive contact-lensdosimeter,

mtmcr BLOCKJ!:D RAY

08L1QUE DIRE(;'tNAY

Figure 2. The Coroneo Effect. Oblique optical rays from the extreme temporal edgeof the cornea can be refractedtopass through (he pupil and arrive at the germinative layer of the lens(theequatorial zone). The path length through theabsorbing lens is approximately the same asaraydirectlythrough thelens along theopticalaxis; hence approximately 1% ofthis oblique,300-nmradia- tion will reach the equator1241.

e

U1travi"'etRaJiaIicmExposure Dosim"lry "I/h"Eye

the approach taken in thisseries of studies was to separate out the various geometrical and environmental factors, evaluate each and thereby develop analgorithmfor ocularexposure. The factors to be analysed wouldbe based upon:ambient UVR measurementsalong

with skyluminance andlid-opening measurements,

Because of theclearly different contributionsfrom skylightand from ground reflection, these two components of ocular exposure mustbe treated separately. ThetotalocularexposureHotnj/cm'can becalculatedby addingthese two components: the upper, skylight componentH,andthelower-hemisphere, ground-reflectioncompo·

nentHitfrom below. Thus:

H .. H_{() s}+H_~ [11_,

where, H₅isdeterminedbytheproduct ofsolid angleQ\I(insteradians) determined bythefield-of-viewabove the horizon(which varieswith the upper lid position)and the time-integrated UVsky radiance L_s [tnunitsofW/(cm²-sr) 1 over the exposure timet:

H_s

=

Q.L.t_u [2]

~

Radiance L~ is avery powerful radiometric quantity which is very usefulwhenattemptingtodescribethe radiant energyarriving froma projectedareainspace. The ocular radiant exposure from the ground H,is determined bythesolid angle n of ocularexposure over the lotverhemisphere andbythe groundtfvRradianceL (i.e., theinci- dent global UV multipliedbytheground reflectance dfvidedbyIt,the effective solidangle for diffuseLambertian reflection):

The groundreflection factors _g have been previously measured (Table 2); hence, the new studies reported here have concentrated upon measuringthehorizon sky DV radiance L_sand the solid angles

nu

and

ad

determinedby the degree ofsquinting as a function of sceneluminanceL ._\I

WHO!EIIG!95.IS

et

Table 2. Reflectance of ACGIH-Effective Solar UV-B from Terrain Surfaces [14,17l

Rcpreserit~ti"eTerrain Surfaces. OitluseReOe(;t~r'Iee ACCIH-Wcightcd Solaruv-n

Green Mountain Grassland 0.8- 1.6%

01)1.pafcheaGrassland ;2·3,7%

Wooden BoatDock ^{6.4 %}

BlacKAsphalt S ·9%

COncretePavement g-I~%

Atlantic Beach Sand(Dry) 15 - 18%

Atlantic Beach Sand (Wet) 7%

Seafoam(SI,lrf) ZS.30%

Aged."Dirty" Snow 50%

Fresh Snow 88%

1.5 Spectral Considerations

Any measurements ofUVRfor the purpose of studying health ef- fects must consider the spectral component to be measured.. For a photobiological effectitis necessary to determine an actionspectrum whichis relevant to the study. All laboratory studies designed to determine an actionspectrum have shown that wavelengths between approximately295nm and325nrn are responsible for acute lenticu- Jar opacities in animal models [2)5]. Theonlyepidemiological study that employed individual dose assessment and attempted to dlstin- guish betweenUVBand UVA - the ChesapeakeBay WatermanStudy - found an increased risk withUVB and cortical cataract. The action spectrum for corneal injury - phctokeratins - has also been deter- minedin animals andhumans and extends overa muchgreater range^I as shown in Figure 3. The action spectrum for current internation- ally accepted guidelines to limit human exposure are based upon an envelope of thephotokeratitisand erythema actionspectra[1,10,25- 27}. Theintemanonalguidelines were promulgated by the predeces- sor committee to the International Commission on Non-Ionizing Radiation Protection (ICNIRP), and this and the(lE action spec- trum for erythema are mimickedbya number

or

DV monitoring in- strumentssuitable forfieldmeasurements [15.18,28-30]. Several re- cent efforts tomeasure environmental UVR (spurred bythe ozone

e ullroviolet

Radiation Exposure Dosimelryof!heEr!

depletion issue), have employed both spectroradiometers and the broad-band meters. Although none of the broad-band meters per- fectly simulate either the elE (Commission International de l'Eclairage) or the ICNIRP standardizederythema (skin reddening) actionspectrum forUVRexposure, theycan be calibratedfor differ- ent solar conditions against these spectraasshown in Figure 4.

1000

...-1

-1"'"'....-1"'

C"I_E 100

... -

~:J.

wu

z~

-

Cl 0:~

c::

...I I.L.I

<

Za:

g

0.01

280 ^{300 320 340 360 400}

WAVELENGTH ^(nm)

Figure 3. TheActionSpectra forphotokeratlus and acutecataractfromlabora-

torystudies.

12

...":.,.'_".J._I ••~...:..:."----'-_.,.::~.,L....o..:...:.--'-,....:.:....IoIoo..ll..I-..r...IoIoo...~_..."",I....I..."",..,... "., ...A_I,,.,,,...,- ••'-N"'""NI'\.,,',.JJ.IItI~w:.I.iNII.I~IoIIll....lJ

WHO/EHG/95.18

~

v _~

'\

^{~ACGIH S(A)}

\

^{"~1/3 CIE EE(A)}

-'

^..'.

~\ <.

_~~,

~~^...

10~

10⁰⁶

250 275 300 325 350 375 400

Wavelength (nm) 10-4

Figure4.TheACGIHIlCNIRPaction spectrum for UVR hazards andthe CIE action spectrum for erythema are compared.By adjusting the normalization value ofthe CIE action spectrumbya factor of 3,the two spectra are very similar.

2. PROGRAM OF STUDIES

To explore the expectedrangeofvalues forthe dosimetrlcvari- ables describedinSection 1.3 (Equation [3]), a program of studies was conducted at theUS Army Environmental Hygiene Agency (USAEHA)during 1994 based upon earlierexperience in evaluation human exposure to solar UVRduring 1972-1994. (The USAEHA was renamedtheUSArmy(enterfor HealthPromotion andPreven- tiveMedicine (U$ACHPPM) during the fall of 1994). To thisend,

ambientUVRmeasurements alongwithskyluminanceandlid-open- ingmeasurementswere made. Further experimentsemployedaman- nequintosimulate theocular geometryin sunlightwere made with

et

ultraviolet Radiation

fx~SCJre Do~imefry of"'" ~

and without sunglasses. The lid opening of individuals was meas- ured under different outdoorconditions. An equation wasderived to estimate the actual ocularUVRexposure in sunlight as a functionof sky luminance.. ground reflection andskyconditions. Each series of studies are reported separatelyinthe following section.

Several staffmembers at USACHPPM assisted in these studies.

Ms. Dawn Deaver and Mr.jackieDavis designed an experimental ap- paratus to measurefield-of-view and then performed the lid opening measurement'] under differentlightconditions. Dr. D. H. Slineyper- formedmost of the field UVR measurements in different locations.

Mr. Rodney Wood and Mr. Stephen Wengraitis performed numerous calibrations andcharacterization measurements with thedifferent UVR radiometers andspectroradiometers, Mr. CharlesSpiceraided in field measurements, and as a meteorologist, helped describe cloudcondi- tions for thesky radiance measurements. Mr. Daniel Berger and Mr.

Adrian Morys of Solar Light Company provided UV measurement data from Philadelphia,PA.

3. ENVIRONMENTAL ULTRAVIOLET MEASUREMENTS

Ambient UVR measurements of the horizon sky were obtained using several types of direct-readingDV instrumentsandDVmoni- tors. Measurementsof the near-horizon DV radiance were performed by two general techniques: with hand-portable, narrow field-of-view (FOV) UV survey instruments and with modified, fixed global DV monitors. Each approach was required to better understandtheef- fects ofspecific cloudconditions.

3.1 Materials and Methods for Field Measurements in DiHerent Environments

Field measurements were performed in varied geographicalloca- lionswithportable UV survey Instruments havmg spectralresponses

74

, ,..1':':.1.",: :" ••AI,,__ ••.:.J..I...:.L.J.I._.:.,_•.(...::.••• _ . - - . . _ ·~••~•..I_.,•.l,....L.L..Ll---'--'-.r...L."_,_".L.l_. _ _ •••• ,•• _., •• ,•• ,•••,_,yl~-.."..,,,.,11"","·I'I"I.,NI"".'1111'JII_I'.YA·~ 1o\.""',,I'~.~_."",.,_ _ •• _ ,.,' ,'",.,.J..JJ_,~_ _ , _ _,_.I,'_ .••.,l I.I ~

.WHO/EHG/95.J8

-

designed^tofollow the ACGIHIICNIRPDVhazard function 125-26).

TheACGIHlICNIRPaction spectrumisan envelope spectrum forboth skin and eye effects, but is closest to the action spectrum for photokeratitis [25-27]. The only action spectrum for cataract are based upon acute cataract in animal studies [2,3,5]., but these action spectra arevery similar to the ACGIHlICNIRP curve for terrestrial solarwavelengths of significance(i.e., 295-325 nm). Any standard- izedaction spectrum chosen will not focus only onthe narrow,300- 325-nm band, but both theelE erythema and the ACGIHlICNIRP action spectra greatly emphasize the 295-325-nm band and broad- band, portable field instruments will tend^to haveverysimilar spec- tral response curves. Figure4illustrates that the two action spectra arequitesimilar andexplainswhyWester was able to obtain thesame environmental UVR readings by a conversion factor of 4.0. Hence several broad-band meters could be calibrated against either action spectrum for a typical solar spectrum near zenith and for one when the sun was lower in the sky (for best results).

Ahighly portable International Light Model1400Radiometer with a Model 1400-$EL240actinic detector head. However, this highly portable radiometer had a sensitivity limited to approximately 0_1 'tlWlcrn2 which was insufficient to measure reflected UVR from the ground and some winter-time exposure conditions. For field meas- urements requiring greater sensitivity, International Light Model1700 radiometers were usedwith Model SED240 Detectors. Narrow FOV cones with solid angles of acceptance of 0.5 and0.1 steradian (sr) were constructed to mountoneach radiometer detector.

All of the UVR monitors were carefully characterized in the labo- ratory against several types of simulated solar spectra and also cali- brated against simultaneous solar UVR spectral measurements weighted against the ACGIHllCNIRP action spectra. Since none of the field portable monitors perfectly fit the appropriate action spec- trum,it was necessary to measure the solar spectrum at several ze- nith angles (e.g., 20'\40°and 50^Ufrom nadir). Figure5shows the spectral irradiance measured with an Oriel Model 7244double-

" UIIrtIvioIet RaJia/ion Exf!!!.5Ure Dosimelly

of

/heEye

monochromatorwitha UVsolar-blind detector andOrielModel 7070 DetectorSystem duringJune at Edgewood, Maryland. Also shownin Figure 4 for comparison, is (heACGIHlICNIRP actionspectrum.

Notice the enormouschangein the spectralirradiance while at the same time thehazard action spectrum in changing over several or- ders ofmagnitude inthe opposite direction. Conesto provideFOVs of 0.1, O.Sand1.0 steradian were used in ordertoaverage overdiffer- e:ntskyconditions. Near-horizon measurementswere made in at least four different azimuths foreach solar elevation angle, and notations of cloudconditions andthe solar elevation anglewere noted for each measurement Mostmeasurementsweremadewitha FOVof0.5 sr

with the measurement axis at approximately IOU. A FOV of0.5 sr

corresponds (0aright circular cone with afullangle of 23°. These measurements were made in many locations (Arizona, California, Florida, Louisiana,and Maryland) in order to understand the impact ofdesert, snow,water, andcoastal conditions.

ttf"L...L----I... " ' - - ....

~ XlO ]20 i<\(I 360 3flO .j1X

Figure 5. The ACGIHlICNIRP action spectrumfOT UV hazardsisplotted in comparison with a typical spectral irradiance of summertime (Z= 20^{1' )} solar UVR. Because of thestrongchanges in the300~32.5nm band, greatefforts must be taken to minimize measurement errors.

16

WHO/fHG/95.IB

e

3.2 Findings

Measured effective (ACGIHllCNIRP) sky radiance values were generally of the orderof one mW/(m^2·sr). typicalvalues areshown in Table 3. Itisvery difficult to summarize thefindings; however, SOme general observations can be made. Clouds strongly affect the measurements. Generally, thepresence ofatmospheric haze andclouds is to redistribute the sky radiation to larger zenith angles, i.e., to low elevation angles near the horizon. Itis well known that the horizon skyisslightly brighter in the visible [35L and this apparently is al- ways true for the UVB band as well. Therefore, the eye is directly exposed to higher values of UVB on slightly overcast or hazy days thanOnvery dear days. However, theabsolute UVB readings become less for all but the lightest overcast. The meteorologist attempting to describe the cloud conditions for this study found it very difficult to draw general conclusions. Clouds on the opposite side of the sky from the open sun had generally twice the UVB radiance of the adja- . centbluesky; however, theshadierunderside orcumulus clouds near the sun had only half the localized radiance as the blue sky in the general area of the sun.

Noserious attempt to measure the horizontal radiance ofmoun- tains and tree lineswere made. This would be approximately the ter- rain reflectance factor multiplied by the global irradiance. TheIL 1400 Radiometer was not sufficiently sensitive to measure this. How- ever, where mountains obscure the horizon sky and the mountains are more than several km away, the UV radiance wouldincrease due to atmospheric Scatter. This had to be left to future studies.

Terrain reflectance influenced thehorizon-skyUVBradiance only when snow was present. The high albedo of snow increased the ho- rizon-sky measurements by at least 20%.

_ UltravioletRaQlOtion

Ex~re

^Dosimetry

^{01 the fl!}

Table 3. Measured ACGIH Effective UV-B from the Sky witha 40°ConeField

of View(FOV) "

Sky Conditions and Zenith Directly Opposite Bod:w,..

Locatiol't,Elevafion Re~l(.Iir'lg at Sun Sun Sky (llWJcmⁱ) (JIW/cm¹) (uWJem^J) (uWJcm^J)

ClearSky (dry) Sm 0,036 0.5. Z1:lI/00 0.08 0,10 ClearSky(humid), 5 m 0.10 1,.5, Z =SOD 0.10 0.09

Ground Fog.j m 0.010 0,07,Zc 7S^D 0.015 0.01

Hazy, Humid. 5 m 0.005 0.50. Z~7if O.OS O.::W

Cloudy Bright, 700 m 0.20 0.16. Z::zz45° 0,10 0,02-

H3Z)',Beach,0,3 m 0.20 0.22, Z =75~ 0.20 0.22

Hazy. Beach.0.3J'Ii 0.14 1.30.Zc:40^D 0.20 0.16

Clear, Mtn Top 2750 m 0.20 0.60. Z·25° 0,30 0,03

Clr.MlI"lVillage.2S00m 0,14 U. Z=45" , 0.08 0.03 NOTE: Z .. 70°referstothe Zenith Angleof70ⁿ(i.e.,ElevatIon Angleo(2if).

3.3 Ambient UVR Fixed Monitoring

Apair of Solarlight Company Model501 DV Biorneters (Robertson-Berger meters) were used for fixed monitoring in Edgewood, Maryland. Oneofthedetectorswas modifiedwithahood which blocked all UVR arrivingfrom zenith angles lessthanapproxi- mately 70° (Le.^televationangles greaterthan approximately 20^{0) I}as shown in Figure 6.' This allowed aconstantcomparison ofthehori- zonUVR(towhich theeye wouldbedirectly exposed) withthetotal globalUVR that is routinelymeasuredbyotherscientificorganiza- tions. ABiosphericalInstrumentsModelGUVSll DVmonitorwas also usedforSOmemonitoringatEdgewoodfor intercomparisonwith theSolarLight instruments.

',18

," ~,.,',I"JJ,'IA'.V,,'••}""}"I"",,} J..u" •...:."._A.l._,"""'~ I

WHO/EHG/95.18

e

3,,4 Findings

Figure 7showsa typical dailyplot ofeffectiveUV irradianceasa function of time fortotal globalUVand for horizon (0(1 to 20^li• Note that thehorizon DYE doesnot vary asdramaticallyasglobalDVE.

a.Model 501 UVaiQrn~UH b. UV Biotl'lolar with 2t}(iegr~&

aeceptance anglo

Figure6. TwoSolar Light Model 501UV Biometerswereused for monitoring

globalandhorizon-skyUVB. Azenith hood limitedthe incident UVR inone detector to horizonelevationanglesof0(1to approximately20^1'from the hori- zon.

HORIZON AND GLOBAL MEASUREMENTS

uvFlIrr_dIAfl(loll

I.

f:::C:-::---:---: iA::'~:C::~ :-- ' j

:~:k~':>;c5~j::'-:--:~'I:::.:.:;:.~:].

'6.S 7,$ 8.5 $.! 10.5 11.5 ,:a.5 13.5 14,5 15.5 Time ofDay

~Glllbal +^Ho,I;fQn

Figure7. The effectiveUV irradiance (Solar LightModel SOl UV^Biometcrs) withspectral weighting close¹⁰the ACG U·I1ICN IRPactionspectrum, is shown asa functionoflimefortotalglobal DVand forhorizon(O'l to20'1), Note that the horizon UVBdoesnotvaryasdramaticallyas globalUVB.

4ft

UltravioletRadiation

~5ure

^Dosimelry

^of

^/heEl!

4. LID OPENING

Of all the variables that have been studied, there has been little previous attention to lid opening. However, lid openingdearly af- fected the final UVRdose to the lensgreatly, but this variable posed the greatest challenge to quantify. Since lid opening and squinting varydepend uponwherea person is looking and the brightness (lu- minance) of the viewed scene, behavioral aswell as environmental factors playarole. The degree of squinting in bright environments apparently had not been previously investigated. Since squinting depended upon the visible componentinthe solar spectrum and not upon theUVRcomponent, this natural protective mechanism of the eyecould be ineffective in environments where UVR was high rela- tive to the visible component.

The objective of this series of studies was to determine if scene luminance was theonlydeterminant oflidopening and to develop an algorithm for the degree oflid opening. It was also hoped that the sample size would be sufficient to determine if other factors} such as age} sex, ethnic origin Or iris colour affected the results. Subjects were volunteers whoworked at USACHPPM and were curious about theirOwn lid opening response.

4. 1 Materials and Methods

To measure the position of the upperand lower lid in a bright out- door environment, it was recognized that this could be best deter- mined by measuring the upper and lower extent of an individuals visual field. A ruled surveyor's pole was positionedvertically in the opensunshine and individuals were asked to standat a fixed distance of 1.8 m from the pole. Apulley fixture was mounted on the top of the pole to accept a movable red ball which could be lowered from the top toanyvertical position (Figure8). Each subject was asked^to fixate on a ruled point which corresponded to his or her eye level above ground. The person was then asked to note when he or she 20

WHO/EHG/95.IB

e

couldjust detect the presence of the ball asitwas lowered from the top of the pole, andthat ball positionwas recorded. Alsorecorded

wastheaverage sceneluminancein the directionoftheperson'sgaze.

Initiallya woodedsceneandskyscenewith a greengrassyforeground were used, but because of concerns regarding the great lackof uni~

fonnityinsceneluminance, theprimary measurement sitewas moved to the sideofawindowlesswhite wall of a warehouse, sothat avery uniformluminance wouldexist.

Figure 8. Outdoorfield ofviewmeasurements

8

Uhraviolet RadiationEXe!2.wre Dasimelrr.af/heEr.e

4.1.1 InitialStudy MClhods

Since the objective of the lid opening studieswasto model con- ditions that simulated sunlight exposure for

an

average person while outside duringa lifetime,every attemptwasmade 'initially to model human behaviour in sunlight as accurately as possible.

While walking out doors, mostpeople do notlookstraightahead, but down at anangle which averages to approximately 15de- grees below thehorizontal." To account forthisdownward gaze, subjects were initially instructed to look at an object a distance , away forcing them to look15degrees below thehorizontal. With this line of sightfixed,the uppervertical acceptance angle' was measured. In addition, a second field-of-view measurement was taken while thesubjects looked horizontally. The majority of tests showedthat the acceptance angle for eachsubjectasmeas- ured from the line of sight remained the same regardless of this viewingangle. However, itisinterestingtonote thatoften the acceptance angle increased byas muchaslO~15 degrees when the line-of-sight was below thehorizontal. Thismight have been expectedsince the luminance of theground is usually lower than thatof thesky,causing lessofa squint.

The first series ofmeasurements were takeninnatural settings which often includedamixture of forest, opensky, buildings, and green grassy fields. Itwasfound thatsuch these attempts toward realistic environments seriously complicated the lumi- nance measurements. Later tests were thereforeperformed in more idealsettings,where a definite luminance value could be obtained using asimplehand-held MinoltaLuminance Meter whichrespondsto aone-degreeacceptanceangle.

4.1.2 Final Study Methods

These latter test settings included a windowless.. painted white wall of the side ofa large building, a wide openuniformskywith

22

WHO/fHG/95.18

e

adistanthorizon; andabackdrop oftalltrees. Becausethelower limitofthevertical extent ofthevisual fieldwas almostalways - BO° to -90(1and was notmeasured exceptfor the verybrightest condition. Luminancevalues rangingfrom 170 cd/m' whenthe subject viewed awoodedarea ofdense treesona heavily over- cast morning; to an upperlimit of 15,000 cd/m'which was achievedbyusingthe white wall ona bright sunnyafternoon.

Thus luminancevalues spanned threeordersof magnitude. For mostofthesestudies photographsofthesubject'seye weremade for the different sunlight conditions, but aside from confirming the visualfieldmeasurements, these have: been set aside forfur- ther study todetermineiffacialgeometryplays a rolein the indi- vidualvariationinourmeasurements.

Toassure that thescene luminancewastheprimarydeterminant for thelid opening, subjects werealso tested wearingtwo differ- ent pairs of sunglasses. Bothpairreduced the luminancebya [actorof approximatelyfour (3.8 and4.3)~however,one pairof lenses was chosen with an orange tint^toblockmorestronglyin the short-wavelength(blue-green)portion ofthespectrum,while the other was chosentobe spectrally neutral (grey) lens.

To test the reproducibilityandreliability of the measurements;

one series of measurementswere repeated forthe same lumi- nance condition two weeks followingthe firstmeasurementand thereadingswere indistinguishable.

4.2Findings

Althoughit hadbeen hypothesized that individuals with dark irises would squint less than blue-eyed mdividuals,itwas found that there was a wide distribution of data and only a general trendin that direction, asshowninFigure9. Theupperfield-of-view angle ofthe groupof subjectstested typically varied widelywith individualsand

8

Ultraviolet Radiation Exe!!.sure Dosimetryofthe Eye

spanned a range ofapproximately 25 degrees. Asshownby the solid line in Figure 10the FOV angles varied verynearly with a linear function of luminance.

Differentiation by Eye Color

~ :~; ; i~

~~~===~~=il ~ .. ..^, ...

". t·

::>

~

-10

'000 10000 100000

Luminance of Outdoor Environment(od/m~)

-30 1.-...L...--O-.L...J...L...L..L..J..L....--L----L-.L.l..Ji....L.i..l..L.----l...I...i...i..i..i-i.LJ

100

Figure 9. The measure of the upper lid openingwasbased uponmeasurement ofupperlimitofField-of-View(FOV). Measuresatdllferentdays

or

the same individual were highly reproducible,but values varied greatly fromindividual toindividual. The colour oftheiriseswasnotthemajordeterminant.and blue- eyed individualsdid not allsquint morethanbrown-eyedtndividuals, although there was ageneral trendillthat direction,

A mathematical function thatdescribed the elevation angle of the upper lid position angle (EA) relativetothe horizon as a function of the ambient luminanceLofthe scene in cd/m'was:

24

WHO/EHG/95.IB

e

The dashed lines represent one standarddeviation above and below themean,encompassing77% ofthe data. Figure 11provides a semi- logarithmic plot of the data covering three orders of magnitude of luminance. Theangle EA rangedfrom 34° for 100cd/m?and15⁰for about 15,000cd/m', Figure 11illustrates the impactoflidopening underdifferent skylightconditions. Sincethe total dose-rate entering

Typical~Dl1Qefor Fi0ldofView Angle

figure 10. Theupperlimitoflidopening varies greatly for different skylight and terrain luminance conditions.

Vertical Visual Fields Under Daylight Conditions With and Without Sunglasses

Figure 11. Eachindividuals visualfield was measured wearing two different pairsorsunglassesandwithoutsunglasses. It is dear that thevisual fidd is increasedwhen the eyesare shieldedbylenseswhich reduce the effective lumi- nance of the scene.

_ UlfravioletRodiotion ExE!!!.wre DosimefryoftheEye

the eye is the product of the effective (ACGIHlICNIRP-spectrally weighted)skyradiance andthesolid'angleofthe,sky'[groundreflec- tionistreatedseparately], the solid angle calculated from-theabove formulais^criticaL Normally the human eye (whenone is standing) has atypicaldownwardgaze angleof-lOOto -15⁰ [31). The result is that theeye receives little or no direct skyradiationon verybright dayseven whenthehorizonis visible. Because ofhorizon-sky shield- ing bytree-linesandbuildings,ground reflection will play thepri- maT)' roleindeterminingocular exposureinmostenvironments.

Onlyinsignificant. variations in EAwerefound forthe testswere thesubjectswore twopairs of sunglasses. Notsurprisingly, thesub- jeershada much smaller FOV·whennot wearing sunglasses. Figure 11 showstheFOVangles plottedagainsttheeffective luminancewhen viewing a^standardwhite wall with eachpairof sunglasses and with- out sunglasses. Thewhole rangeof visual field angles increasedby about 10degreeswhenthesesunglasseswereworn.Again, the actual transmittedluminanceseenbythe observerdetermined theangleEA.

Several possible factors that mightcause thelarge individual vari- ationin theangle EAwere examined. Iriscolor and skin color did notappearto bestrongpredictors. Facialbone structureappeared ^[0

be a strong lactor,but was difficult to quantify. Figure 9 shows a significant^overlapbetween^thedarkerandlighter colourediris_group~

ings. The samplesizewasnotsufftcient.to determineifskintypewas

a strong predictor. facialstructure caused some people to have a natural^{limit upon} their vertical visual field. Those with «deep set»

eyes tended tohave asmalleracceptanceangle regardless ofthebright- ness, presumablydue toblockagebythebrow ridgeand cheeks. The lower extremeofthedata was comprisedmostly ofpeople with deep- seteyes. Two subjects with thelargestlidopenings in thebrightest environments were noted tohave miotic pupils - approximately 1 mm diameter comparedto nearly2 mmfor most other subjects. Un- fortunately, there was no readymeansavailable to measure pupil^di- ameterwith sufficientaccuracyto determine

If

the constriction ofthe

pupil compensated for a lackof squintingin some individuals. ^This remains afuture area for additional study.

26

WHO/EHG/95.J8

e

4.3 Discussion

43.1 LidOpening Algorithm

Since the solid angle determined by the field measurements re- ported here depend upon skyluminanceLv'Equation[3]canbe further refined as:

H .. K(O.0034 - L )-L-t [s}

~ v s

where: Kis a factor that is approximated by the average linear angleof the horizontal FOV

The ground component of the ocular exposure dose was gener- ally much simpler to calculate. As noted previously, it is the product of the globalUVRhorizontal surface irradianceE~l and the ground reflectance - divided bythe effective solid angle ^1t into whichdiffuselyreflected radiant energy is reflected accord- ing to Lambert's La'W:

Although Equation [2]assumes a cosine dependence which is true exactly for totally diffuse reflection, thisisreasonable forall ground surfaces other that water. The measured coefficientfor diffuse reflection measuredbySliney[17}was only0.04:whereas, the effective «reflectance» measured by Rosenthal using polysulfone film dosimeters [18-19] was approximately 0.2 (2.0%). Thusanystudyofwatermustnecessarily deriveavariant of Equation [4]\or employ contact-lens dosimeters or estimate this with film-badge dosimetry with a squinting factor. Typical measured values of- are about0.01forgrass,0.1forpavements (both masonry and asphalt), and0.8forsnow Actual measured values of- reportedbySliney [18] are given in Table 2.

_ Uhraviolet Radiation

Ex~sure

^{Dosimetryof/he}

^Ire

Combining theabove equations, the following summary equa- tion of [1]results:

Thisequationis probably underestimating the solid angle.. since the lid openingswere somewhatgreater when taken withthe eye lixated downward at -15\}. The function Kiscurrently being determinedbyanalysing visual field dataandgenerallyisabout 0.1.. Althoughthe function notedas Eqn. [5] is thekeydetermi- nant, thege,ometrical expansionof the field-of-view solid-angle is not quite linearwiththe vertical angleEA.

4.3.2 Implications forEyeProtection

The human eye receives approximately 10to 25%of the UVR dose when wearing lenses opaque totheUVRcompared to the eye withoutalens. Furthermore, the lateral raysentering the eye from thesideof thesunglass may heparticularly dangerous

to the lens as a result ofthe Coroneo effect [21], even ifthe lensesvirtually blockallofthe UVRol concern. Therefore,un- lessoneemploysa goggle geometry withside-shields,etc.,UVR

transmission factors inlensesmuchless than 2^{r 5}%becomemis-

leading.

The strong dependence ofreflectance with angleof incidence is termed FresnelsLawof Reflection. This law notonlyexplains the survivability of thecornea in an overheadbath ofUVR,but also the glare experienced over water. When theSun is over- head, a body of water reflects theUVR upward, but onlyap- proximately 2~3% is reflected[14,17}. When thesunislow in thesky,much ofthe incidentlightisreflected, but now the UVR and blue light have been filtered outof the direct rays by the atmosphereandare therefore have been thought to be relatively 28

WHO/EHG/95,J8

e

harmless [17]. Nevertheless, thestrongreflections from water at these low sun angles create discomfort glare and UVR exposure of the cornea is further reduced because ofsquinting. The meas- urement ofhorizon DVB show thatseveral conditions exist when the discomfort glare may not be adequate to produce a strong squint. These instances occur during overcast sky conditions when the lids are open) but the overcast produces more UVB scattered to large zenith angles (low elevation angles). TheUVB dose appears to be higher duringthese conditions, particularlyif out over the. water, where the water reflects thishorizon UV-BOr

when snowison the ground.

The otherevidence just reviewed would suggest thatifdarklenses were placed over our eyesl our natural aversion to bright light~

which leads tosquinting that normally greatly lowers theUVR and retinal exposure to the eye - would be «disabled.. Thismay appear tobean unusual way to consider the comfort thatshaded lenses bring about. But, just consider thatsome ofour «discom- fort» derived from notwearing sunglasses ora brimmed hatstems from muscle fatigue associated with squinting. Wemust, there- fore,askwhether sunglasses may not actually lead to a higher UVR exposure condition rather than provide reduction. Ofpar- ticular interest recently, hasbeen theobservation ofCoroneo that extreme peripheral rays falling on the edge of the COrnea can be focused in the nasal sector of the lens behind the pupil. Since mostcortical cataracts begin in thissector,ithasbeen theorized that these rays^I unattenuated bysunglasses Or a brimmed hat may play an etiologic role in cataract. Ifthis were so, measure- ment of this extreme peripheral UVR must also be made under different environmental conditions.

e

Ultraviolet Radiation

Ex~re

^{Do5imetryoftheEye}

5. MANNEQUIN STUDIES

5.

J

Materials and Methods

Because of behavioral patterns, lid openingvariations, and head and eyemovements, itisobvious thatUV contact-lensdostmeters are needed to makethebest determinationofactual ocular'exposure. In the absence of a practical contactlensdosimeter.a detector from the

IL~1700 UVRirradiance meter was mounted in the position of the eyein a painted,styroloam,anthropomorphicmodelofthehead(Fig^M ure 12) to simulateocular exposure from a fixed position. Through theuse of variable slits to simulate exposure through the palpebral fissure (lid opening),UVRexposure could be measured in different locations to take into accountthe effect ofshadingofthe skybybuild- ings and trees, plus variable ground reflectance factors. SinceUVR

ocularexposure under daylight conditions with differing times of the day, sky and geographical conditions can varybya factor oflOO~these measurements by necessity must takeintoaccountlid openingsimu- lation, The typical lid openings for different environmental condi- tions were estimatedfrom the lid openingmeasurements. The sensi- tivity of the instrument (approximately0.01l1Wlcm1)limitedthe size of the slit to approximately1mm.

Mannequin Dosimetry Sludies

Figure 12. AUVR irradiance meter is mounted in the position of the eye inan anthropomorphic modelofthehead. Variable slnscanbe placed over the detector to simulate exposure through the palpebral fissure (lid opening). Themannequin can be moved todifferent locations to take into account {he effect ofshading ofthe skybybuildings and trees,plus variable ground reflectance (actors.

30

5.2 Findings

Mannequin Studies

Thecurrentstudies improve our knowledge of ocular exposure with thesquinting factor considered. However, eye and headmove- ments were Iactoredout and the assumption was madeinEquation [4]ofa fixed vertical fixation angle. To carry these studies further, it is probably necessary to employ contact-lens dosimeters,

Theuseof the mannequin dosimetry system haspermitted usto evaluate the effectiveness of different sunglassesintermsofUVat- tenuationat the eye. AlthoughUVBblocking by the lenswas impor- tant,we confirmed that many frame designs couldpermitasmuch as 20%of theUVirradiance to reach the corneal plane comparedwith no lenses. Consideringthe relationship oflid opening to luminance, the reduced luminance perceived through the sunglass lens leads to anincreasedlid opening which can actually increase theUVdose to

thelens. Sunglasses with wrap-around frame designs, goggles, and Inuit whale-bone slit goggles were all generally effective in reducing actual UV-B exposure to less than 1% (thelimitof the IL 1700instru- ment.

5.3 Discussion

Specific,off-axis lateral UVR shouldbe measuredbya special probe in order to examine the significant dose attributed ^to the Coroneo effect. This is a future project. Also, the pupillary opening of those with large lidopenings should be measuredtodetermine ifhigh miosis compensates for the increasedlid opening inSOmesubjects. Evenif miosis does compensate for retinal exposure, the geometry oflen- ticular and retinal exposure would still be affected differently from those having average lid-closure responses to brightlight environ- ments.

fit

Uhraviolet RadiationEXE!!!.wfe

Do~imelry ^of

^/he

E~

6. GENERAL DISCUSSION

The Cornea asBiological Dosimeter

By a careful microscopic examination ofthe cornea, Pitts and colleagues

! 51

^l in their several studies, determined the threshold for photokeratitis as the beginning ofa very slight cornealhaze. This subtle change was described as less than a clinical case of photokeratitis,and such changes are typical after walking over sand foranumber of hours indaylight. Thus, the COrnea itselfcould be used as a biological dosimeter if a corneal specular microscope or similarinstrumentwereused to count the number of vacuoles in the corneal epithelium. Although it would be difficult to use this ap- proach as an absolutemeasure,it could be applied to the determine- tionofthe relative doses obtainedwhen working overdifferent ground surfaces or for different outdoor tasks. Pitts has stated that he could determine bycorneal microscopy whether an indoor worker was working in an office withunfilteredfluorescentlamps compared to an office usingonly incandescent sources - because of the difference inUVemissions [32].

FurtherStudies withContact Lens Dosimeurs

Eye and head movements can further reduce exposure. Ifone observes human behaviour inbrightsunlight, oneisstruck by the Iact that most people squint or avoid looking into the sun sector of thesky. Thesebehavioraland physiological factors are notatalltaken into accountbysimpleUVRmeasurements; hence, thereisan obvi- ous need for an in-depth studyto properly determine corneal and lenticular exposureto ambientUVR. Indeed, the results of the previ- ous epidemiological studies of cataract can be thrown into question, because of inadequate dosimcrry The failure of some epidemiologi- cal studies to demonstrate an increased risk from UVROr sunlight exposure could,well have resulted from wrongly assigning different exposure levelstodifferent population groups baseduponanassump- 32

,.

..

WHO!EHG!95.J8

e

nonthatoverheadUVR exposurepredicts corneal exposure. Asshown

above, this assumption isgenerally wrong. The use of globalUVR measurements which havevaluein studies of UVR and skin cancer are inappropriate foreye studies. Under the same overhead UVR measurement, the critically important horizon-sky contribution to

'ocular dose can varybyafactorof ten or more depending'uponter-

mill'and weather' conditions." 'Sydenham has recently reported on progress indevelopment of a polysulfone,scleral contact lens [33);

andthis type ofcontactlens may have thesensitivity required for

definitiveenvironmental studies.

DISCUSSION

TheAppropriateA,tion Spectrum

As noted initially, the action spectrum chosen for usein thisse- ries of studies wasbased upon the assumption that UVB radiant en- ergy was the primary potentialagent ofconcern, since bothanimal

and human studies supported thishypothesis. However,itis worthy to note that many laboratory biochemical studies of the effects of

UVR on lensproteins suggest thatUVA (315-400 urn)mayalsoplay an etiological role [34]. Ifthat were the case,muchofthework here is still relevant, although the atmospheric scatter functions forUVA are^notas strong as^for DVB and the terrain reflectance values are greater.