International Archives of
Int Arch Occup Environ Health (1990) 62:133-137
OWC Upationland
Environmental health © Springer-Verlag 1990
Assessment of changes in eye redness by a photographic method and the relation to sensory eye irritation Seren K Kjaergaardl , Ole F Pedersen l , Ebbe Taudorf2, and Lars Molhavel 'Institute of Environmental and Occupational Medicine, University of Aarhus, Bygning 180, Universitetsparken, DK-8000 Arhus C, Denmark 2 Allergy Unit Medical Department TTA, State University Hospital, Copenhagen, Denmark Received October 10 / Accepted November 2, 1989
Summary The purpose of the present study was to evaluate the relation between photographically assessed changes in eye redness and sensory eye irritation in 17 patients suffering from hayfever, provoked by increasing doses of birch pollen in the eye Pre and post-exposure diapositives were compared in a randomized and double blind design by a panel of five members By evaluating the photographs we found a high reproducibility between panel members and for repeated evaluation by the individual panel member Furthermore, we found that the photographic method was more sensitive for measurements of changes in eye redness than the observations normally performed during such conjunctival provocation tests The correlations of the objective findings with the symptoms reported were good at high pollen doses, but at lower doses of birch pollen (about half the effective dose) a decreased eye redness was observed. We found that sensory irritation (often expressed as dryness) measured simultaneously by questionnaires, in general occurred before any increased eye redness The study indicates that photographic measurement of eye redness is reproducible, sensitive, unbiased and therefore a useful tool in environmental studies Furthermore we demonstrated that the relation between eye redness and eye irritation is interestingly biphasic and may need further clarification. Key words: Sensory eye irritation physiology Air pollutants
Standardized photography of eye redness is a possibility in such measurements Photographic methods have been used in evaluation of mucosal redness in the mouth (Heyden and Mattson 1988) and by us for the measurement of changes in eye redness of subjects exposed to formaldehyde, n-Decane, and tobacco dust (Bach et al. 1988 ; Kjxergaard et al 1989 ; Kjaergaard and Pedersen 1989). The aim of this study was to evaluate the photographic method and the association between sensory eye irritation and eye redness in subjects with birth birch pollen induced hayfever.
Material and methods Seventeen patients, all suffering from seasonal allergic rhinitis and all participants in a clinical pollen provocation test to evaluate different treatments, were included The results of the clinical study have been reported elsewhere (Taudorf et al 1987) A conjunctival provocation test (CPT) was performed, as previously described (Moller et al 1984) Each CPT was preceded by an inspection of the eye, and in the case of congestion or complaint of itching, CPT was not performed If no such symptoms were present, one drop of the albumin diluent was instilled in the inferior conjunctival sack. If no reaction occurred, i e no congestion or itching, one drop of Pharmalgen birch dilution, 1 BU/ml, was instilled into the other eye after 10 min The provocation doses were increased by half logarithmic steps in the alternate eye to positive reaction or to the
Eye redness, PathoFrame
dilaposillve
Introduction In environmental and occupational studies of air pollution there is an increasing demand for objectively measured indicators of sensory exposure effects Such measurements allow objective criteria for the diagnosis of mucous membrane irritation due to air pollution, and give a better foundation for appropriate intervention. Offprint requests to: 50ren K Kjaergaard
Fig 1 Photographic equipment used
134
Fig 2 Photographs taken before (upper) and after (lower) provocation
top dose of 320000 BU/ml The criteria for positive CPT was increased redness of the conjunctiva as diagnosed by a trained observer and itching of the eye as reported by the patient All CP Ts were performed by one person only Before CPT, all medication that could influence the results was withdrawn The patients, 8 females and 9 males were aged between 18 and 46 (mean 32 8 years) The range of pollen doses for positive reaction was from 100 to 10000 BU/ml. "Changes in eye redness" were evaluated by pairs of photographs (diapositives) taken after installation of the diluent, and after each step in the provocation procedure. Diapositives were taken with a Nikon F3-camera using a 105mm micronikkor lens mounted on bellows (Fig 1) at a distance providing a 2:1 magnification of the area photographed An automatic flash and a white reflector screen were used to provide equal amount of indirect lighting without reflexes Photographs were taken under standardized conditions and included an area defined by the lids and the pupil Professional films of the same production batch were used and pictures for comparison were taken on the same film to avoid influence of differences in film processing.
The baseline photograph and the photograph of the provoked eye was shown simultaneously on a screen The photographs were compared twice with a day or more between (i e two evaluation series) in a randomized double blind design by a five person panel. An example of such a pair is shown in Fig 2 Furthermore, in the pollen study, "adjacent" pairs of photographs were compared, i e. the baseline with the second, the second with the third and so on. All comparisons were randomized, both with respect to the order of pairs and the position on screen, i e upper or lower photo. Members of the panel made individual judgements of differences in redness of the eye on a 0-2 point scale 0 was given if no difference was observed If the upper photo had a small increase in redness it was given the score 1 If a strong and convincing increase was seen it was given score 2 The lower photo was scored similarly if this showed more redness than the upper one After breaking the code and identification of the baseline photo, this score was transformed to a score between 1 and 5 in the following way: 2 given to baseline (previous) photo becomes equivalent to 1. 1 given to baseline (previous) photo becomes equivalent to 2.
135 0 becomes equivalent to 3. 1 given to later photo becomes equivalent to 4. 2 given to later photo becomes equivalent to 5. By this procedure each pair of photographs was scored ten times By the mean (i e the subject score) and the variance estimates (m and S2) of the ten scores each subject was classified either as "most red before", "most red after" or "no difference" The following classification criteria were used: Most red before No difference Most red after
if m < 2 4 and S2 < 1 14 if 3 6-m-2 4 or S2 > 1 14 if m> 3 6 and S2 < 1 14
The classification limits are based on the 95 % confidence limits in a normal distribution with 3 as the mean value Furthermore we set criteria for the variance (s 2), which by high values indicate very different scores among the panel members These cases were interpreted as no difference However, to get a more graded score we also used the subject score (m). "Sensory irritation in the eyes" was measured each time a photograph was taken A questionnaire with a 100-mm-long visual analogue rating scale (Huskisson 1983) with the endpoints "no irritation" and "very strong irritation" was used Questions about the type of sensations (itching, pain, dryness, stinging) and about watering eyes were included. The results were evaluated statistically using correlation analysis to evaluate associations between panel members All statistical analyses were performed using SPSSpc (Norusis 1986 ; 1988) A probability less than or equal to 0 05 was considered significant.
Results Reproducibility and sensitivity of the photographicmethod The correlation coefficients between the five different panel members within each of the two series are given in Table 1 Similar correlation coefficients for the individual subjects between the two series (not shown) varied between 0 59 and 0 73 and were not significantly different from those shown in Table 1 The within panel member correlation coefficients for the two series varied between 0 70 and 0 78 The differences in the scores on the five point scale between randomly selected panel members and within panel members showed no observable shift in either of the two possible directions, indicating that only random errors in the judgements existed. Furthermore, differences did not exceed an absolute value of 2 points except for a few cases which could be attributed to obvious mistakes during scoring The mean difference between the two series (series 2 minus series 1) was 0 023 equivalent to about 0 6% of maximal difference The coefficient of variation (CV) for the ten scores given had an average of 19% for all the comparisons performed in this study. The sensitivity of the photographic method was measured as the likelihood of detecting increased redness if the simple clinical observation (the CPT-criterium) indicated this The sensitivity of the photographic method is 0.72 when the classification based on the 95% confidence limit and variance estimate as described above (i.e less red, no differences, more red) is used If instead we defined more red by subject score values larger than 3 and less red by subject score values smaller than
Table 1 Correlation coefficients between panel members in the two series of evaluation (see text) Letters A-E designate panel members, numbers 1-2 designate first and second evaluation series Between panel member correlations in each of the two series 2
1
A A B C D E
B
C
D
1 0 67 1 0 74 0 62 1 0 67 0 61 0 73 1 0 71 0 61 0 75 0 80
E
A
B
1
1 0 63 1 0 68 0 63 0 63 0 68 0 70 0 65
C
D
E
1 0 78 0 81
1 0 73
1
INCREASED VALUE
IRRITATION CHANGE FROM BASEUNE -e REDNESS CHANGE FROM BASEUNE
0 DECREASED VALUE
20 40 60 80 100 BIRCH POLLEN DOSE STEP IN PERCENT OF MAXIMAL NEEDED STEPS
Fig 3 Standardized mean values of changes of eye irritation and redness at different provocation steps defined as percent of maximal step (cf Table 2) "Redness" is given as the mean of the subject scores (cf text) "Irritation" is measured as the change from baseline measured in mm on a visual analogue rating scale Both redness and irritation scores are transformed using the z-transformation, i e (Zi = (yi-y)/SD) Table 2 Self-reported sensory eye irritation measured in mm on a 100-mm-long visual analogue scale and compared with the clinical criteria used by the observer, i e reporting distinct itching in the eye Irritation Visual analogue rating scale (Values are in mm)
Iching No Yes a
b
Mean
SD
No of observations
12 9a 34 4b
18 2 21 4
28 21
p c 0 02 for Ho: mean = 0, Student's t-test < 0 001 for Ho: mean = 0, Student's t-test
3, we found the sensitivity to be 0 88 If on the contrary we consider the photographic scores to be true, the sensitivity of the CPT-criterium is only 0 60 for the classification (less red, no difference, more red) and 0 47 for the classification based on the subject score larger or smaller than 3 This means that by the simple clinical observation we have only a 60 % chance of detecting an increased redness as measured by the photographic method.
136 Table 3 Qualitative characterisation of the sensory irritation as reported by hayfever patients exposed to increasing doses of birch pollen extract Qualitative description or irritation
No irritation Dryness Itching Pain Other irritation
Sensory irritation measured on the visual analogue rating scale (values are in mm i e equivalent to percent of full scale) Mean
SD
No.
00 11 3 32 2 72 0 11 0
00 11 17 6 00 96
14 3 27 2 3
1.5 1.2
,"'
0.9
o CHANGEFROM BASEUNE
0.6
CHANGEFROM PREVIOUS
0.3
0
DIFFERENCES
0 "I-
-0.3 -0.6 0
20
40
60
80
100
Fig 4 Eye redness measured either as the change from baseline photography or from the photograph taken at the previous provocation step Mean changes are given for each provocation step defined as percentage of the maximal number of steps (cf Table 2). Means are based on subjects scores for each pair of photographs
Associations between eye redness and sensory irritation In the present study we found the associations shown in Fig 3 All values are standardised using the z-transformation (Zi = (yi-y)/SD) giving comparable values on a scale without dimensions The figure shows that eye redness first decreased and then increased until finally all subjects became distinctively positive according to the criteria set Contrary to this, sensory irritation slowly but significantly increased from the beginning and later, when redness increased, a steeper increase was seen. Both decreased and increased redness was significant using Mann-Whitney's test (Siegel 1956) Comparisons of mean sensory eye irritation using visual analogue rating scales to the dichotomized (i e yes/no) self-reported criterium of itching showed significantly irritated eyes even before the distinct itching was reported to the observer in the CPT-trial (Table 2) As seen from Table 3 we found that low levels of sensory irritation measured on the visual linear rating scale are associated with the qualitative description of the irritation as dryness, and the higher levels as itching or pain.
Discussion In environmental and occupational studies there is a need for objective methods to evaluate small changes in mucous membrane since it is often difficult to keep exposures unknown This increase reduces the risk of psychological effects, such as mass hysteria, which is a risk in indoor air studies (Colligan 1981) Therefore we used the photographic method The method has reasonable reproducibility as observed by the correlations between and within panel members and the insignificant differences between the mean score for the first and the second series The photographic method seems to be in agreement with the simple inspection (CPT-criterium) performed by the observer in the clinical trial, and to be even more sensitive than this simple inspection, where the observer compares new observations with pictures in his memory only This may become more evident if there is a long time between the first and the final inspection, as could be the case in studies of long-term expo-
sures However, it seems as if the photographic method can detect smaller changes than the naked eye. Taking pictures is fast, easy, and non-invasive, and can be done repeatedly over longer periods of time. Furthermore comparisons can be made randomized and double blind The comparisons may be time consuming and could be avoided by computerized video methods described in the literature (Kovalcheck et al 1982 ; Tsusima et al 1982) These methods focus on single vessels under high magnification, and are developed to detect changes, occurring within short periods of time. A promising method may be computer aided photographic analyses as described by Heyden and Mattsson (1988) This method combines the ease of our method with the ease and effectiveness of computers for complicated analysis However, it is still very expensive and not fully developed The method, which has been used to evaluate changes in the mucosa of the mouth after exposure to nicotine or tobacco smoke, includes a possibility of correcting for differences due to the photographic process, i e artifactual changes in color intensity during the film processing, which is a potential error in our method if different films are used. A higher sensitivity of the method might result in detection of increased redness before or simultaneous with the clinical finding of increased redness This was found in most of our cases Furthermore, decreased redness as found by the photographic method is unexpected to the clinical observer and therefore will be missed Comparing the photographs with the preceding photographs we find an even earlier detection of eye redness than by using the baseline photograph (Fig 4) This is a further validation of the photographic method. This study surprisingly showed initially decreased redness, and a corresponding slight sensory eye irritation We cannot exclude an effect of the trial itself, since a stress effect of provocation due to the expectancy of a hayfever attack, is a possibility This could cause an increased blood adrenalin concentration, and thereby cause peripheral vasoconstriction This still needs to be investigated However, low level sensory irritation may be correlated directly to the decreased redness This could be a reflex mediated response, which is later followed by a
137
normal inflammatory response where stronger irritation is correlated with increasing redness The response observed here could be an equivalent to the so called "white response" seen in skin, where a mechanical challenge like a scratch or pressing induces vasoconstriction seen as a white line or area (Ganong 1977) Increased redness and sensory irritation is the normal reaction to allergen challenges in the eye and is most probably induced by the release of substances which can interact with the histamine responsive receptors found in conjunctival beds (Kirkegaard et al 1982) Potential substances are the prostaglandines, leukotrienes and histamine (Bisgaard and Thorsen 1986 ; Wandal 1988 ; Hammarstr 6 m 1983). The initially decreased redness may then be related to trigeminal reflexes, systematic release of vasoconstricters in the blood, or some unknown locally released substances Leukotrienes may also be potential causes of such reactions since animal studies indicate microcirculatory reduction in blood flow after application of leukotrienes (Bisgaard and Thorsen 1986) This has, however, not been shown in humans. The findings regarding dryness associated with quantitatively low values of irritation, and pain associated with high level irritation, are in agreement with the findings forming the base for the McGill pain questionnaire (Melzack 1983). The photographic method presented here has previously been used in three environmental studies investigating effects of formaldehyde, n-decane, and tobacco dust (Bach et al 1988 ; Kjargaard et al 1989 ; Kja Trgaard and Pedersen 1989) The results from these studies demonstrate significantly decreased redness during the day or study period and a simultaneous slight increase in mean sensory irritation as measured by visual analogue scales. In these three studies we also found indications of regression towards the mean, i e that high baseline redness is associated with decreasing redness during the study period and vice versa Further, we found that tobacco smoking, but not sex and age, may be a possible confounder of change in redness These findings, however, are preliminary and need to be supported by future studies. Only in the tobacco dust study did we find a slight increase in redness associated with exposure to increasing amounts of dust in the air (Kjargaard and Pedersen 1989). In conclusion, the photographic method is easy to use in environmental studies It is more sensitive than simple inspection, reproducible, and reduces the risk of bias.
Acknowledgements This work was supported by the Danish Health Insurance Foundation.
References Bach B, M 01lhave L, Pedersen OF (1988) Human reactions to formaldehyde vapors Institute of Environmental and Occupational Medicine, Aarhus (In Danish with English summary) Bisgaard H, Thorsen S (1986) Leukotrienes in allergic and inflammatory conditions Ugeskr Laeg 148: 3435-3439 Colligan MJ (1981) The psychological effects of indoor air pollution Bull NY Acad Med 57:1014-1025 Draize JH, Woddard G, Calvery HO (1944) Methods for the study of irritation and toxicity of substances applied topically to the skin and mucous membranes J Pharmacol Exp Ther 82:377390 Ganong WF (1977) Review of medical physiology Lung medical publications, Los Angeles Hammarstrom S (1983) Leukotrienes Ann Rev Biochem 52:355377 Heyden G, Mattsson U (1988) Computer aided photograph analyses in oral medicine a pilot study Swed Dent J 12:93-99 Huskisson EC (1983) Visual analoque scales, In: Pain measurement and assessment Melzack R (ed) Raven Press, New York, pp 33-37 Kirkegaard J, Secher C, Mygind N (1982) Effect of the H 1 antihistamine chlorpheniramine maleate on histamine induced symptoms in the human conjunctiva Allergy 37:203-208 Kjargaard S, Pedersen OF (1989) Dust exposure, eye redness, eye cytology and mucous membrane irritation in a tobacco industry Int Arch Occup Environ Health 61:519-525 Kjwargaard S, Pedersen OF, Molhave L (1989) Human exposure to indoor pollutants: n-decane Environ Int 15: 473-482 Kovalcheck SW, Chen P, Wadleigh FR, Zweifach BW (1982) Characterization of conjunctival microvessels/network in diabetic and control individuals a computer based method. Microvasc R 23 :262 Melzack R (1983) The McGill pain questionnaire In: Pain measurements and assessment Melzack R (ed) Raven Press, New York, pp 41-47 Miller C, Bjorksten B, Nilsson G, Dreborg S (1984) The precision of the conjunctival provocation test Allergy 39: 37-41 Norusis M (1986) SPSS/PC+ advanced statistics SPSS Inc, Chicago Norusis M (1988) SPSS/PC basic statistics SPSS Inc, Chicaco Siegel S (1956) Nonparametric statistics for the behavioral sciences McGraw-Hill, Kogakuska, LTD, Tokyo Taudorf E, Laursen LC, Lanner A, Bj6 rksten B, Dreborg S, 50borg M, Weeke B (1987) Oral immunotherapy in birch pollen hay fever J Allergy Clin Immunol 80: 153-161 Tsusima N, Satok T, Koyama T, Kikuchi Y, Nitta J (1982) Pharmacological effects on microcirculation (MC) of human conjunctiva Microvasc R 24:231 Wandall JH (1988) Neutrophilic granulocyte function Dan Med Bull 35: 237-252
Int Arch Occup Environ Health (1990) 62:133-137
OWC Upationland
Environmental health © Springer-Verlag 1990
Assessment of changes in eye redness by a photographic method and the relation to sensory eye irritation Seren K Kjaergaardl , Ole F Pedersen l , Ebbe Taudorf2, and Lars Molhavel 'Institute of Environmental and Occupational Medicine, University of Aarhus, Bygning 180, Universitetsparken, DK-8000 Arhus C, Denmark 2 Allergy Unit Medical Department TTA, State University Hospital, Copenhagen, Denmark Received October 10 / Accepted November 2, 1989
Summary The purpose of the present study was to evaluate the relation between photographically assessed changes in eye redness and sensory eye irritation in 17 patients suffering from hayfever, provoked by increasing doses of birch pollen in the eye Pre and post-exposure diapositives were compared in a randomized and double blind design by a panel of five members By evaluating the photographs we found a high reproducibility between panel members and for repeated evaluation by the individual panel member Furthermore, we found that the photographic method was more sensitive for measurements of changes in eye redness than the observations normally performed during such conjunctival provocation tests The correlations of the objective findings with the symptoms reported were good at high pollen doses, but at lower doses of birch pollen (about half the effective dose) a decreased eye redness was observed. We found that sensory irritation (often expressed as dryness) measured simultaneously by questionnaires, in general occurred before any increased eye redness The study indicates that photographic measurement of eye redness is reproducible, sensitive, unbiased and therefore a useful tool in environmental studies Furthermore we demonstrated that the relation between eye redness and eye irritation is interestingly biphasic and may need further clarification. Key words: Sensory eye irritation physiology Air pollutants
Standardized photography of eye redness is a possibility in such measurements Photographic methods have been used in evaluation of mucosal redness in the mouth (Heyden and Mattson 1988) and by us for the measurement of changes in eye redness of subjects exposed to formaldehyde, n-Decane, and tobacco dust (Bach et al. 1988 ; Kjxergaard et al 1989 ; Kjaergaard and Pedersen 1989). The aim of this study was to evaluate the photographic method and the association between sensory eye irritation and eye redness in subjects with birth birch pollen induced hayfever.
Material and methods Seventeen patients, all suffering from seasonal allergic rhinitis and all participants in a clinical pollen provocation test to evaluate different treatments, were included The results of the clinical study have been reported elsewhere (Taudorf et al 1987) A conjunctival provocation test (CPT) was performed, as previously described (Moller et al 1984) Each CPT was preceded by an inspection of the eye, and in the case of congestion or complaint of itching, CPT was not performed If no such symptoms were present, one drop of the albumin diluent was instilled in the inferior conjunctival sack. If no reaction occurred, i e no congestion or itching, one drop of Pharmalgen birch dilution, 1 BU/ml, was instilled into the other eye after 10 min The provocation doses were increased by half logarithmic steps in the alternate eye to positive reaction or to the
Eye redness, PathoFrame
dilaposillve
Introduction In environmental and occupational studies of air pollution there is an increasing demand for objectively measured indicators of sensory exposure effects Such measurements allow objective criteria for the diagnosis of mucous membrane irritation due to air pollution, and give a better foundation for appropriate intervention. Offprint requests to: 50ren K Kjaergaard
Fig 1 Photographic equipment used
134
Fig 2 Photographs taken before (upper) and after (lower) provocation
top dose of 320000 BU/ml The criteria for positive CPT was increased redness of the conjunctiva as diagnosed by a trained observer and itching of the eye as reported by the patient All CP Ts were performed by one person only Before CPT, all medication that could influence the results was withdrawn The patients, 8 females and 9 males were aged between 18 and 46 (mean 32 8 years) The range of pollen doses for positive reaction was from 100 to 10000 BU/ml. "Changes in eye redness" were evaluated by pairs of photographs (diapositives) taken after installation of the diluent, and after each step in the provocation procedure. Diapositives were taken with a Nikon F3-camera using a 105mm micronikkor lens mounted on bellows (Fig 1) at a distance providing a 2:1 magnification of the area photographed An automatic flash and a white reflector screen were used to provide equal amount of indirect lighting without reflexes Photographs were taken under standardized conditions and included an area defined by the lids and the pupil Professional films of the same production batch were used and pictures for comparison were taken on the same film to avoid influence of differences in film processing.
The baseline photograph and the photograph of the provoked eye was shown simultaneously on a screen The photographs were compared twice with a day or more between (i e two evaluation series) in a randomized double blind design by a five person panel. An example of such a pair is shown in Fig 2 Furthermore, in the pollen study, "adjacent" pairs of photographs were compared, i e. the baseline with the second, the second with the third and so on. All comparisons were randomized, both with respect to the order of pairs and the position on screen, i e upper or lower photo. Members of the panel made individual judgements of differences in redness of the eye on a 0-2 point scale 0 was given if no difference was observed If the upper photo had a small increase in redness it was given the score 1 If a strong and convincing increase was seen it was given score 2 The lower photo was scored similarly if this showed more redness than the upper one After breaking the code and identification of the baseline photo, this score was transformed to a score between 1 and 5 in the following way: 2 given to baseline (previous) photo becomes equivalent to 1. 1 given to baseline (previous) photo becomes equivalent to 2.
135 0 becomes equivalent to 3. 1 given to later photo becomes equivalent to 4. 2 given to later photo becomes equivalent to 5. By this procedure each pair of photographs was scored ten times By the mean (i e the subject score) and the variance estimates (m and S2) of the ten scores each subject was classified either as "most red before", "most red after" or "no difference" The following classification criteria were used: Most red before No difference Most red after
if m < 2 4 and S2 < 1 14 if 3 6-m-2 4 or S2 > 1 14 if m> 3 6 and S2 < 1 14
The classification limits are based on the 95 % confidence limits in a normal distribution with 3 as the mean value Furthermore we set criteria for the variance (s 2), which by high values indicate very different scores among the panel members These cases were interpreted as no difference However, to get a more graded score we also used the subject score (m). "Sensory irritation in the eyes" was measured each time a photograph was taken A questionnaire with a 100-mm-long visual analogue rating scale (Huskisson 1983) with the endpoints "no irritation" and "very strong irritation" was used Questions about the type of sensations (itching, pain, dryness, stinging) and about watering eyes were included. The results were evaluated statistically using correlation analysis to evaluate associations between panel members All statistical analyses were performed using SPSSpc (Norusis 1986 ; 1988) A probability less than or equal to 0 05 was considered significant.
Results Reproducibility and sensitivity of the photographicmethod The correlation coefficients between the five different panel members within each of the two series are given in Table 1 Similar correlation coefficients for the individual subjects between the two series (not shown) varied between 0 59 and 0 73 and were not significantly different from those shown in Table 1 The within panel member correlation coefficients for the two series varied between 0 70 and 0 78 The differences in the scores on the five point scale between randomly selected panel members and within panel members showed no observable shift in either of the two possible directions, indicating that only random errors in the judgements existed. Furthermore, differences did not exceed an absolute value of 2 points except for a few cases which could be attributed to obvious mistakes during scoring The mean difference between the two series (series 2 minus series 1) was 0 023 equivalent to about 0 6% of maximal difference The coefficient of variation (CV) for the ten scores given had an average of 19% for all the comparisons performed in this study. The sensitivity of the photographic method was measured as the likelihood of detecting increased redness if the simple clinical observation (the CPT-criterium) indicated this The sensitivity of the photographic method is 0.72 when the classification based on the 95% confidence limit and variance estimate as described above (i.e less red, no differences, more red) is used If instead we defined more red by subject score values larger than 3 and less red by subject score values smaller than
Table 1 Correlation coefficients between panel members in the two series of evaluation (see text) Letters A-E designate panel members, numbers 1-2 designate first and second evaluation series Between panel member correlations in each of the two series 2
1
A A B C D E
B
C
D
1 0 67 1 0 74 0 62 1 0 67 0 61 0 73 1 0 71 0 61 0 75 0 80
E
A
B
1
1 0 63 1 0 68 0 63 0 63 0 68 0 70 0 65
C
D
E
1 0 78 0 81
1 0 73
1
INCREASED VALUE
IRRITATION CHANGE FROM BASEUNE -e REDNESS CHANGE FROM BASEUNE
0 DECREASED VALUE
20 40 60 80 100 BIRCH POLLEN DOSE STEP IN PERCENT OF MAXIMAL NEEDED STEPS
Fig 3 Standardized mean values of changes of eye irritation and redness at different provocation steps defined as percent of maximal step (cf Table 2) "Redness" is given as the mean of the subject scores (cf text) "Irritation" is measured as the change from baseline measured in mm on a visual analogue rating scale Both redness and irritation scores are transformed using the z-transformation, i e (Zi = (yi-y)/SD) Table 2 Self-reported sensory eye irritation measured in mm on a 100-mm-long visual analogue scale and compared with the clinical criteria used by the observer, i e reporting distinct itching in the eye Irritation Visual analogue rating scale (Values are in mm)
Iching No Yes a
b
Mean
SD
No of observations
12 9a 34 4b
18 2 21 4
28 21
p c 0 02 for Ho: mean = 0, Student's t-test < 0 001 for Ho: mean = 0, Student's t-test
3, we found the sensitivity to be 0 88 If on the contrary we consider the photographic scores to be true, the sensitivity of the CPT-criterium is only 0 60 for the classification (less red, no difference, more red) and 0 47 for the classification based on the subject score larger or smaller than 3 This means that by the simple clinical observation we have only a 60 % chance of detecting an increased redness as measured by the photographic method.
136 Table 3 Qualitative characterisation of the sensory irritation as reported by hayfever patients exposed to increasing doses of birch pollen extract Qualitative description or irritation
No irritation Dryness Itching Pain Other irritation
Sensory irritation measured on the visual analogue rating scale (values are in mm i e equivalent to percent of full scale) Mean
SD
No.
00 11 3 32 2 72 0 11 0
00 11 17 6 00 96
14 3 27 2 3
1.5 1.2
,"'
0.9
o CHANGEFROM BASEUNE
0.6
CHANGEFROM PREVIOUS
0.3
0
DIFFERENCES
0 "I-
-0.3 -0.6 0
20
40
60
80
100
Fig 4 Eye redness measured either as the change from baseline photography or from the photograph taken at the previous provocation step Mean changes are given for each provocation step defined as percentage of the maximal number of steps (cf Table 2). Means are based on subjects scores for each pair of photographs
Associations between eye redness and sensory irritation In the present study we found the associations shown in Fig 3 All values are standardised using the z-transformation (Zi = (yi-y)/SD) giving comparable values on a scale without dimensions The figure shows that eye redness first decreased and then increased until finally all subjects became distinctively positive according to the criteria set Contrary to this, sensory irritation slowly but significantly increased from the beginning and later, when redness increased, a steeper increase was seen. Both decreased and increased redness was significant using Mann-Whitney's test (Siegel 1956) Comparisons of mean sensory eye irritation using visual analogue rating scales to the dichotomized (i e yes/no) self-reported criterium of itching showed significantly irritated eyes even before the distinct itching was reported to the observer in the CPT-trial (Table 2) As seen from Table 3 we found that low levels of sensory irritation measured on the visual linear rating scale are associated with the qualitative description of the irritation as dryness, and the higher levels as itching or pain.
Discussion In environmental and occupational studies there is a need for objective methods to evaluate small changes in mucous membrane since it is often difficult to keep exposures unknown This increase reduces the risk of psychological effects, such as mass hysteria, which is a risk in indoor air studies (Colligan 1981) Therefore we used the photographic method The method has reasonable reproducibility as observed by the correlations between and within panel members and the insignificant differences between the mean score for the first and the second series The photographic method seems to be in agreement with the simple inspection (CPT-criterium) performed by the observer in the clinical trial, and to be even more sensitive than this simple inspection, where the observer compares new observations with pictures in his memory only This may become more evident if there is a long time between the first and the final inspection, as could be the case in studies of long-term expo-
sures However, it seems as if the photographic method can detect smaller changes than the naked eye. Taking pictures is fast, easy, and non-invasive, and can be done repeatedly over longer periods of time. Furthermore comparisons can be made randomized and double blind The comparisons may be time consuming and could be avoided by computerized video methods described in the literature (Kovalcheck et al 1982 ; Tsusima et al 1982) These methods focus on single vessels under high magnification, and are developed to detect changes, occurring within short periods of time. A promising method may be computer aided photographic analyses as described by Heyden and Mattsson (1988) This method combines the ease of our method with the ease and effectiveness of computers for complicated analysis However, it is still very expensive and not fully developed The method, which has been used to evaluate changes in the mucosa of the mouth after exposure to nicotine or tobacco smoke, includes a possibility of correcting for differences due to the photographic process, i e artifactual changes in color intensity during the film processing, which is a potential error in our method if different films are used. A higher sensitivity of the method might result in detection of increased redness before or simultaneous with the clinical finding of increased redness This was found in most of our cases Furthermore, decreased redness as found by the photographic method is unexpected to the clinical observer and therefore will be missed Comparing the photographs with the preceding photographs we find an even earlier detection of eye redness than by using the baseline photograph (Fig 4) This is a further validation of the photographic method. This study surprisingly showed initially decreased redness, and a corresponding slight sensory eye irritation We cannot exclude an effect of the trial itself, since a stress effect of provocation due to the expectancy of a hayfever attack, is a possibility This could cause an increased blood adrenalin concentration, and thereby cause peripheral vasoconstriction This still needs to be investigated However, low level sensory irritation may be correlated directly to the decreased redness This could be a reflex mediated response, which is later followed by a
137
normal inflammatory response where stronger irritation is correlated with increasing redness The response observed here could be an equivalent to the so called "white response" seen in skin, where a mechanical challenge like a scratch or pressing induces vasoconstriction seen as a white line or area (Ganong 1977) Increased redness and sensory irritation is the normal reaction to allergen challenges in the eye and is most probably induced by the release of substances which can interact with the histamine responsive receptors found in conjunctival beds (Kirkegaard et al 1982) Potential substances are the prostaglandines, leukotrienes and histamine (Bisgaard and Thorsen 1986 ; Wandal 1988 ; Hammarstr 6 m 1983). The initially decreased redness may then be related to trigeminal reflexes, systematic release of vasoconstricters in the blood, or some unknown locally released substances Leukotrienes may also be potential causes of such reactions since animal studies indicate microcirculatory reduction in blood flow after application of leukotrienes (Bisgaard and Thorsen 1986) This has, however, not been shown in humans. The findings regarding dryness associated with quantitatively low values of irritation, and pain associated with high level irritation, are in agreement with the findings forming the base for the McGill pain questionnaire (Melzack 1983). The photographic method presented here has previously been used in three environmental studies investigating effects of formaldehyde, n-decane, and tobacco dust (Bach et al 1988 ; Kjargaard et al 1989 ; Kja Trgaard and Pedersen 1989) The results from these studies demonstrate significantly decreased redness during the day or study period and a simultaneous slight increase in mean sensory irritation as measured by visual analogue scales. In these three studies we also found indications of regression towards the mean, i e that high baseline redness is associated with decreasing redness during the study period and vice versa Further, we found that tobacco smoking, but not sex and age, may be a possible confounder of change in redness These findings, however, are preliminary and need to be supported by future studies. Only in the tobacco dust study did we find a slight increase in redness associated with exposure to increasing amounts of dust in the air (Kjargaard and Pedersen 1989). In conclusion, the photographic method is easy to use in environmental studies It is more sensitive than simple inspection, reproducible, and reduces the risk of bias.
Acknowledgements This work was supported by the Danish Health Insurance Foundation.
References Bach B, M 01lhave L, Pedersen OF (1988) Human reactions to formaldehyde vapors Institute of Environmental and Occupational Medicine, Aarhus (In Danish with English summary) Bisgaard H, Thorsen S (1986) Leukotrienes in allergic and inflammatory conditions Ugeskr Laeg 148: 3435-3439 Colligan MJ (1981) The psychological effects of indoor air pollution Bull NY Acad Med 57:1014-1025 Draize JH, Woddard G, Calvery HO (1944) Methods for the study of irritation and toxicity of substances applied topically to the skin and mucous membranes J Pharmacol Exp Ther 82:377390 Ganong WF (1977) Review of medical physiology Lung medical publications, Los Angeles Hammarstrom S (1983) Leukotrienes Ann Rev Biochem 52:355377 Heyden G, Mattsson U (1988) Computer aided photograph analyses in oral medicine a pilot study Swed Dent J 12:93-99 Huskisson EC (1983) Visual analoque scales, In: Pain measurement and assessment Melzack R (ed) Raven Press, New York, pp 33-37 Kirkegaard J, Secher C, Mygind N (1982) Effect of the H 1 antihistamine chlorpheniramine maleate on histamine induced symptoms in the human conjunctiva Allergy 37:203-208 Kjargaard S, Pedersen OF (1989) Dust exposure, eye redness, eye cytology and mucous membrane irritation in a tobacco industry Int Arch Occup Environ Health 61:519-525 Kjwargaard S, Pedersen OF, Molhave L (1989) Human exposure to indoor pollutants: n-decane Environ Int 15: 473-482 Kovalcheck SW, Chen P, Wadleigh FR, Zweifach BW (1982) Characterization of conjunctival microvessels/network in diabetic and control individuals a computer based method. Microvasc R 23 :262 Melzack R (1983) The McGill pain questionnaire In: Pain measurements and assessment Melzack R (ed) Raven Press, New York, pp 41-47 Miller C, Bjorksten B, Nilsson G, Dreborg S (1984) The precision of the conjunctival provocation test Allergy 39: 37-41 Norusis M (1986) SPSS/PC+ advanced statistics SPSS Inc, Chicago Norusis M (1988) SPSS/PC basic statistics SPSS Inc, Chicaco Siegel S (1956) Nonparametric statistics for the behavioral sciences McGraw-Hill, Kogakuska, LTD, Tokyo Taudorf E, Laursen LC, Lanner A, Bj6 rksten B, Dreborg S, 50borg M, Weeke B (1987) Oral immunotherapy in birch pollen hay fever J Allergy Clin Immunol 80: 153-161 Tsusima N, Satok T, Koyama T, Kikuchi Y, Nitta J (1982) Pharmacological effects on microcirculation (MC) of human conjunctiva Microvasc R 24:231 Wandall JH (1988) Neutrophilic granulocyte function Dan Med Bull 35: 237-252
