69 (1991) 799-801
ACTA O P H T H A L M O L O G I C A
Conjunctival hyperaemia in non-contact lens wearers Charles W. McMonnies and Arthur Ho Cornea and Contact Lens Research Unit, School of Optometry, University of New South Wales, Kensington,Australia
Abstract. Conjunctival hyperaemia was assessed in noncontact lens wearers using a photographic reference scale. The study was carried out on a sample of 252 females and 227 males within the age range of 1 to 89 years. The results show increasing levels of hyperaemia with age and higher levels of hyperaemia in males than in females. The differences in hyperaemia level for age and sex, however, were small and were insignificant clinically. Key words: conjunctiva - hyperaemia - statistics - epidemiology.
Conjunctival hyperaemia may be used as a measure of ocular response to contact lenses experimentally (Holden et al. 1986) and in the normal clinical environment (McMonnies 1983). Previous studies did not find a significant difference in limbal (McMonnies et al. 1982) or conjunctival hyperaemia (McMonnies & Chapman-Davies 1987a),in non-contact lens wearers between males and females, or between younger and older age groups. The age ranges examined in those studies, however, were limited. It is not known if the use of the same reference scale is valid when measuring hyperaemia levels for males and females, and for individuals of different ages. This study was conducted on males and females with a wide age range in order to determine if the level of conjunctival hyperaemia in the normal, non-contact lens wearing population is dependent on age and sex.
Method Consecutive patients presenting for routine eye examinations for the correction of refractive error
were assessed by a single observer using the grading scale for hyperaemia developed by McMonnies & Chapman-Davies (1987a,b). This is a photographic reference scale consisting of six levels or grades of conjunctival hyperaemia (0 =none, 5 = severe)which cover the range of hyperaemia responses normally encountered in contact lens practice. The scale photographs were compared subjectively with conjunctival hyperaemia levels of both eyes. All observations were made in the same clinical environment - in particular, with identical illumination levels. The ratings for the right and left eyes of each patient were recorded as their hyperaemia scores. Only subjects who had never worn contact lenses were included. There was no other basis for exclusion. A total of 479 non-contact lens wearing subjects were assessed, 252 females and 227 males. Age ranged from 1 to 82 years for the female sample and 1 to 89 years for the male sample. Results were placed into 10-year groupings. The sample frequency distribution for age and sex is shown in Table 1.
Results Because a strong correlation was found between right and left eye hyperaemia scores (Spearman rank correlations, rs= 0.73, P< O.OOOl), the data for right and left eyes of each subject could not be considered as independent observations (Ray & ODay 1985). Hence, for statistical analysis, a one-eye design was simulated by randomly selecting the hyperaemia score of either the right or left eye for 799
Females
Males
Total
0 - 10
0.59 f 0.47 (44 I 0.0)
0.72 f 0.64 (64 10.0)
0.67 f 0.58 (108 / 0.0)
11 - 20
0.63 k 0.65
0.89 f 0.60
0.76
(54 I 0.0)
(54 I 1.0)
(108 I 1.0)
21 - 30
0.90 k 0.61 (31 11.0)
1.00 f 0.49 (22 I 1.0)
0.94 f 0.56 (53 I 1.0)
31 - 40
0.68 k 0.78 (47 10.0)
1.30 f 0.80 (28 I 1.0)
0.91 f 0.84 (75 10.0)
41 - 50
1.01 f 0.81 (43 I 1.0)
1.20 k 0.78 (40 I 1.0)
1.10 f 0.80 (83 I 1.0)
51 - 60
0.82 f 0.61 (22 I 1.0)
0.86 f 0.60 (14 10.5)
0.83 f 0.60 (36 I 1.0)
61 - 70
1.31 f 0.70 (8 I 1.0)
1.67 f 0.29 (3 I 1.0)
1.41 f 0.63 (11 11.0)
71 - 80
1.00 f 0.71 (2 10.5)
0.00 f (1 10.0)
0.67 f 0.76 (3 10.0)
81 - 90
1.00 f (1 11.0)
1.50f (1 11.5)
1.25 f 0.35 (2 I 1.0)
Age groups
Combined data
+ 0.64
0.86 f 0.70 (47.9 I 1.0)
Where standard deviations could not be calculated, - is shown.
each subject in order to avoid biased statistical results. The hyperaemia scores of 244 right eyes and 235 left eyes were selected in this process. Chisquared (R x C contingency) tests on the distributions of selected eyes against age groups (8x 2) and against sex (2 x 2) produced x2 = 7.42 (P=0.49) and x2 = 0.79 (P=0.37), respectively, demonstrating that the random selection process did not introduce significant sampling bias into the data. The set of randomly selected scores was used in all subsequent analyses. These scores are shown in Table 1, grouped by age and sex. The combined mean for all scores was 0.86 with a standard deviation of 0.70 (median = 1.0). Comparison of hyperaemia scores between pairs of age groups averaged across sex was tested using the Mann-Whitney test. Due to the 10-year group-
800
ings, there were insufficient subjects in the over 60 years of age groups to provide valid comparisons, and therefore the results of these over 60 years groups were not included in this analysis. The only comparisons which produced a type-one error probability less than 0.05 were: 1. Those involving the 1-10 years group and the three groups between 21-50 years; the 1-10 years group having the lower hyperamiea score in all comparisons, and 2. those involving a. the 11-20 years group and the 21-30 years group and b. the 11-20 years group and the 41-50 years group; the 11-20 years age group having the lower hyperaemia score in both comparisons. When the type-one error probability for the Mann-Whitney test was adjusted to account for the multiple comparisons made (Harris 1975),no comparisons were statistically signilkant at the 0.05 level. A multiple regression analysis was carried out on the results. Since the data was expected not to conform to a normal distribution, ranking transformation was applied to all data before entry into the multiple regression analysis. The ranking transformation operated by replacing the score (hyperaemia, age or sex) of each subject with their relative ranking in the sample - the lowest value being assigned a value of 1 while the highest value was assigned a value of 479. This transformation modified the multiple regression into a non-par-
Source
DF
MS
F
ametric analysis (Conover 1971). The dependent variable in the regression analysis was the ranking for hyperaemia score and the independent variables were the ranked data for age and sex. The regression summary is given in Table 2. The results may be summarized as follows: 1. Hyperaemia scores increased with age (P< 0.0001). 2. Holding age constant, males had higher hyperaemia scores than females ( P < 0.0001). 3. The rate of increase of hyperaemia score with age was not different between males and females ( P = 0.84). The standard error of estimates of hyperaemia scores calculated from the regression of ranked hyperaemia scores on age was approximately 1.25 hyperaemia scale units. Follow-up calculations to the multiple regression analysis (Conover 1971) showed that the difference in hyperaemia scores between extreme age sub-groups (0 to 10 vs 51 to 60 years) was approximately 0.8 scale units and the difference in scores between male and female subjects at the same age was approximately 0.5 scale units.
Discussion The results indicate that in a large sample, there is a detectable difference in the level of hyperaemia between males and females, and across age groups. The reasons for the differences found are not apparent from this investigation. Ocular hyperaemia levels vary with exposure to provocative factors such as allergy, infection, lack of sleep, eye strain, exposure to wind, dust, glare, smoke and smog, and sensitisingfoods (e.g. alcohol).No attempt was made to control for these potential factors in this study. Contact lens wear, a known stimulus to conjunctival hyperaemia, was excluded by including only non-contact lens wearers. Further, differences in the degree of exposure to potential hyperaemia stimuli may not be the most relevant factors in explaining the differences found if there is a physiologically increased hyperaemia response with advancing age. These differences however, are small as demonstrated by the follow-up calculations which showed that the expected difference in conjunctival hyperaemia between the extreme age groups is W3 scale units and the expected difference between males and females of the same age 51
Acta Ophthal. 69.6
is 0.5 hyperaemia scale units. On the grading scale used, a one unit change in conjunctival hyperaemia was considered just noticeable clinically. Therefore, the differences across age and sex could be considered insignificant clinically.The differences in levels of hyperaemia across age and sex may only need to be considered in experiments in which a finer grading scale than that used in this study, and a large sample size are used.
Acknowledgments This study was reported in the International Society of Contact Lens Researchers Meeting, 1988 (Hawaii). The authors would like to thank Peter Fagan for comments on the manuscript.
References Conover W J (1971): Practical non-parametric statistics (2nd edition). pp 263-277 and 335-338.John Wiley and Sons, New York. Harris R J (1975): A primer of multivariate statistics. pp 96-101. Academic Press, New York. Holden et al. (1986): The vascular response to long-term extended contact lens wear. Clin & Exp Optom 69: 112-119. McMonnies C W (1983): Contact lens-induced corneal vascularization. Int Contact Lens Clinic 10: 12-21. McMonnies C W & Chapman-Davies A (1987a): Assessment of conjunctival hyperaemia in contact lens wearers Part I. Am J Optom Physiol Opt 6 4 246-250. McMonnies C W & Chapman-Davies A (198713): Assessment of conjunctival hyperaemia in contact lens wearers Part II. Am J Optom Physiol Opt 64: 251-255. McMonnies C W et al. (1982): The vascular response to contact lens wear. Am J Optom Physiol Opt 59: 795-799. Ray W A & O'Day D M (1985):Statistical analysis of multieye data in ophthalmic research. Invest Ophthalmol Vis Sci 26: 1186-1188.
Received on January 23rd, 1991. Author's address:
Charles W. McMonnies, MSc, FAAO, Cornea and Contact Lens Research Unit, 22-32 King Street, Randwick NSW 2031, Australia.
801
ACTA O P H T H A L M O L O G I C A
Conjunctival hyperaemia in non-contact lens wearers Charles W. McMonnies and Arthur Ho Cornea and Contact Lens Research Unit, School of Optometry, University of New South Wales, Kensington,Australia
Abstract. Conjunctival hyperaemia was assessed in noncontact lens wearers using a photographic reference scale. The study was carried out on a sample of 252 females and 227 males within the age range of 1 to 89 years. The results show increasing levels of hyperaemia with age and higher levels of hyperaemia in males than in females. The differences in hyperaemia level for age and sex, however, were small and were insignificant clinically. Key words: conjunctiva - hyperaemia - statistics - epidemiology.
Conjunctival hyperaemia may be used as a measure of ocular response to contact lenses experimentally (Holden et al. 1986) and in the normal clinical environment (McMonnies 1983). Previous studies did not find a significant difference in limbal (McMonnies et al. 1982) or conjunctival hyperaemia (McMonnies & Chapman-Davies 1987a),in non-contact lens wearers between males and females, or between younger and older age groups. The age ranges examined in those studies, however, were limited. It is not known if the use of the same reference scale is valid when measuring hyperaemia levels for males and females, and for individuals of different ages. This study was conducted on males and females with a wide age range in order to determine if the level of conjunctival hyperaemia in the normal, non-contact lens wearing population is dependent on age and sex.
Method Consecutive patients presenting for routine eye examinations for the correction of refractive error
were assessed by a single observer using the grading scale for hyperaemia developed by McMonnies & Chapman-Davies (1987a,b). This is a photographic reference scale consisting of six levels or grades of conjunctival hyperaemia (0 =none, 5 = severe)which cover the range of hyperaemia responses normally encountered in contact lens practice. The scale photographs were compared subjectively with conjunctival hyperaemia levels of both eyes. All observations were made in the same clinical environment - in particular, with identical illumination levels. The ratings for the right and left eyes of each patient were recorded as their hyperaemia scores. Only subjects who had never worn contact lenses were included. There was no other basis for exclusion. A total of 479 non-contact lens wearing subjects were assessed, 252 females and 227 males. Age ranged from 1 to 82 years for the female sample and 1 to 89 years for the male sample. Results were placed into 10-year groupings. The sample frequency distribution for age and sex is shown in Table 1.
Results Because a strong correlation was found between right and left eye hyperaemia scores (Spearman rank correlations, rs= 0.73, P< O.OOOl), the data for right and left eyes of each subject could not be considered as independent observations (Ray & ODay 1985). Hence, for statistical analysis, a one-eye design was simulated by randomly selecting the hyperaemia score of either the right or left eye for 799
Females
Males
Total
0 - 10
0.59 f 0.47 (44 I 0.0)
0.72 f 0.64 (64 10.0)
0.67 f 0.58 (108 / 0.0)
11 - 20
0.63 k 0.65
0.89 f 0.60
0.76
(54 I 0.0)
(54 I 1.0)
(108 I 1.0)
21 - 30
0.90 k 0.61 (31 11.0)
1.00 f 0.49 (22 I 1.0)
0.94 f 0.56 (53 I 1.0)
31 - 40
0.68 k 0.78 (47 10.0)
1.30 f 0.80 (28 I 1.0)
0.91 f 0.84 (75 10.0)
41 - 50
1.01 f 0.81 (43 I 1.0)
1.20 k 0.78 (40 I 1.0)
1.10 f 0.80 (83 I 1.0)
51 - 60
0.82 f 0.61 (22 I 1.0)
0.86 f 0.60 (14 10.5)
0.83 f 0.60 (36 I 1.0)
61 - 70
1.31 f 0.70 (8 I 1.0)
1.67 f 0.29 (3 I 1.0)
1.41 f 0.63 (11 11.0)
71 - 80
1.00 f 0.71 (2 10.5)
0.00 f (1 10.0)
0.67 f 0.76 (3 10.0)
81 - 90
1.00 f (1 11.0)
1.50f (1 11.5)
1.25 f 0.35 (2 I 1.0)
Age groups
Combined data
+ 0.64
0.86 f 0.70 (47.9 I 1.0)
Where standard deviations could not be calculated, - is shown.
each subject in order to avoid biased statistical results. The hyperaemia scores of 244 right eyes and 235 left eyes were selected in this process. Chisquared (R x C contingency) tests on the distributions of selected eyes against age groups (8x 2) and against sex (2 x 2) produced x2 = 7.42 (P=0.49) and x2 = 0.79 (P=0.37), respectively, demonstrating that the random selection process did not introduce significant sampling bias into the data. The set of randomly selected scores was used in all subsequent analyses. These scores are shown in Table 1, grouped by age and sex. The combined mean for all scores was 0.86 with a standard deviation of 0.70 (median = 1.0). Comparison of hyperaemia scores between pairs of age groups averaged across sex was tested using the Mann-Whitney test. Due to the 10-year group-
800
ings, there were insufficient subjects in the over 60 years of age groups to provide valid comparisons, and therefore the results of these over 60 years groups were not included in this analysis. The only comparisons which produced a type-one error probability less than 0.05 were: 1. Those involving the 1-10 years group and the three groups between 21-50 years; the 1-10 years group having the lower hyperamiea score in all comparisons, and 2. those involving a. the 11-20 years group and the 21-30 years group and b. the 11-20 years group and the 41-50 years group; the 11-20 years age group having the lower hyperaemia score in both comparisons. When the type-one error probability for the Mann-Whitney test was adjusted to account for the multiple comparisons made (Harris 1975),no comparisons were statistically signilkant at the 0.05 level. A multiple regression analysis was carried out on the results. Since the data was expected not to conform to a normal distribution, ranking transformation was applied to all data before entry into the multiple regression analysis. The ranking transformation operated by replacing the score (hyperaemia, age or sex) of each subject with their relative ranking in the sample - the lowest value being assigned a value of 1 while the highest value was assigned a value of 479. This transformation modified the multiple regression into a non-par-
Source
DF
MS
F
ametric analysis (Conover 1971). The dependent variable in the regression analysis was the ranking for hyperaemia score and the independent variables were the ranked data for age and sex. The regression summary is given in Table 2. The results may be summarized as follows: 1. Hyperaemia scores increased with age (P< 0.0001). 2. Holding age constant, males had higher hyperaemia scores than females ( P < 0.0001). 3. The rate of increase of hyperaemia score with age was not different between males and females ( P = 0.84). The standard error of estimates of hyperaemia scores calculated from the regression of ranked hyperaemia scores on age was approximately 1.25 hyperaemia scale units. Follow-up calculations to the multiple regression analysis (Conover 1971) showed that the difference in hyperaemia scores between extreme age sub-groups (0 to 10 vs 51 to 60 years) was approximately 0.8 scale units and the difference in scores between male and female subjects at the same age was approximately 0.5 scale units.
Discussion The results indicate that in a large sample, there is a detectable difference in the level of hyperaemia between males and females, and across age groups. The reasons for the differences found are not apparent from this investigation. Ocular hyperaemia levels vary with exposure to provocative factors such as allergy, infection, lack of sleep, eye strain, exposure to wind, dust, glare, smoke and smog, and sensitisingfoods (e.g. alcohol).No attempt was made to control for these potential factors in this study. Contact lens wear, a known stimulus to conjunctival hyperaemia, was excluded by including only non-contact lens wearers. Further, differences in the degree of exposure to potential hyperaemia stimuli may not be the most relevant factors in explaining the differences found if there is a physiologically increased hyperaemia response with advancing age. These differences however, are small as demonstrated by the follow-up calculations which showed that the expected difference in conjunctival hyperaemia between the extreme age groups is W3 scale units and the expected difference between males and females of the same age 51
Acta Ophthal. 69.6
is 0.5 hyperaemia scale units. On the grading scale used, a one unit change in conjunctival hyperaemia was considered just noticeable clinically. Therefore, the differences across age and sex could be considered insignificant clinically.The differences in levels of hyperaemia across age and sex may only need to be considered in experiments in which a finer grading scale than that used in this study, and a large sample size are used.
Acknowledgments This study was reported in the International Society of Contact Lens Researchers Meeting, 1988 (Hawaii). The authors would like to thank Peter Fagan for comments on the manuscript.
References Conover W J (1971): Practical non-parametric statistics (2nd edition). pp 263-277 and 335-338.John Wiley and Sons, New York. Harris R J (1975): A primer of multivariate statistics. pp 96-101. Academic Press, New York. Holden et al. (1986): The vascular response to long-term extended contact lens wear. Clin & Exp Optom 69: 112-119. McMonnies C W (1983): Contact lens-induced corneal vascularization. Int Contact Lens Clinic 10: 12-21. McMonnies C W & Chapman-Davies A (1987a): Assessment of conjunctival hyperaemia in contact lens wearers Part I. Am J Optom Physiol Opt 6 4 246-250. McMonnies C W & Chapman-Davies A (198713): Assessment of conjunctival hyperaemia in contact lens wearers Part II. Am J Optom Physiol Opt 64: 251-255. McMonnies C W et al. (1982): The vascular response to contact lens wear. Am J Optom Physiol Opt 59: 795-799. Ray W A & O'Day D M (1985):Statistical analysis of multieye data in ophthalmic research. Invest Ophthalmol Vis Sci 26: 1186-1188.
Received on January 23rd, 1991. Author's address:
Charles W. McMonnies, MSc, FAAO, Cornea and Contact Lens Research Unit, 22-32 King Street, Randwick NSW 2031, Australia.
801
