Share this article on:

Development of Refractive Errors into Old Age


Articles: Original Article

Purpose. To evaluate refractive errors in older adults.

Methods. The distribution of refractive error components was evaluated in a sample of 569 older adults including 171 participants over the age of 80 years. The mean age was 75.2 years with a range from 59 to 106 years. Emphasis was placed on modern methods of analyzing astigmatic refractive errors, which convert cylindrical refractive errors into primary and oblique components.

Results. The known increase in hyperopia after maturity continues into old age. The primary negative astigmatic component increases dramatically in prevalence and amount after age 70 years, whereas the oblique component remains unchanged. Significant anisometropia is common in the oldest old, suggesting failure of emmetropization mechanisms with age. Substantial gender differences exist in refractive changes with age.

Conclusions. The continuing changes in all components of refractive error into old age and the surprisingly high prevalence of large amounts of astigmatism and anisometropia emphasize the importance of regular refractive evaluations among the oldest old.

Smith-Kettlewell Eye Research Institute, San Francisco, California (GHP, MES, JAB, LAL), School of Optometry, University of California at Berkeley, California (GHP, MES)

Received December 28, 2001;

revision received June 4, 2002.

Gunilla Haegerstrom-Portnoy

School of Optometry

University of California, Berkeley

Berkeley, California 94720-2020


Presented, in part, at the annual meeting of the Association for Research in Vision and Ophthalmology (ARVO), May 2001, Ft. Lauderdale, Florida.

The increasing use of refractive surgery has emphasized the need to understand normal changes in refractive error with aging. In addition, the explosive increase of people living into very old age emphasizes the importance of including this age group in studies of effects of aging on refractive error. According to the U.S. Census 2000, people over 65 years of age now comprise 12.4% of the U.S. population. The fastest growth since 1990 has been in the 85 years and over group, which increased by 38%, whereas the overall U.S. population increased by 13.2%. Now, more than 4.2 million Americans are over the age of 85 years.

It is well known that both the spherical and astigmatic errors change with age after maturity. A decrease in myopia and an increase in hyperopia with increasing age is commonly reported. 1–10 There are, however, reports of increased myopia with age. 11 The refractive changes also include an increase in the prevalence of astigmatism and a change in the type of astigmatism from with-the-rule to against-the-rule. 1–10

For example, Hirsch 1 showed that the median cylindrical error increased from −0.09 D in the late 40s to −0.91 D in the 80s in his population of 1353 patients from his own private optometric practice. His population contained 93 people over age 80 years. The prevalence of 1.00 D or more of with-the-rule astigmatism (minus cylinder axis 180) decreased from 9 to 3% in the same age groups, whereas the prevalence of against-the-rule astigmatism increased from 3 to 31.6%. The prevalence of oblique astigmatism stayed fairly stable with age varying between 10 to 15% of each age group.

Recent population studies of aging 12–16 in different parts of the world have verified these findings. The Baltimore Eye survey 12 found a significant increase in hyperopia with age as well as an increased prevalence of astigmatism. The type of astigmatism was not discussed. This study included 140 white subjects over 80 years. The Beaver Dam study 13 from Wisconsin included participants up to 84 years of age in their initial study and reported an increase in the prevalence of hyperopia defined as equivalent sphere greater than +0.50 D. Astigmatic errors were not reported. The Blue Mountains Study from Australia 14 presented equivalent sphere and astigmatic data only for their sample as a whole and not by age groups. They described spherical error as a function of age, which increased from +0.03 D in the 49 to 59 age group to +1.20 D in the over 80 age group. Astigmatic errors were not presented for the different age groups. The Blue Mountains study included 39 participants 85 years and over. The Reykjavik Eye study 16 included 1045 observers with 78 participants 80 years and over. They report an increase in prevalence of astigmatism (defined as 0.75 D or more) with age and a change from about equal prevalence of with-the-rule and against-the-rule astigmatism in 50-year-old subjects to a preponderance of against-the-rule in the oldest group. They reported that against-the-rule astigmatism is about four times more common in the over 80 age group than the with-the-rule type.

The current study extends these refractive error findings into very old ages and presents astigmatic errors using newer analysis techniques.

Back to Top | Article Outline



The data come from a subsample of the Smith-Kettlewell Institute (SKI) study. 17, 18 Initially, 900 people between the ages of 58 and 102 yrs (mean, 75.5 yrs) were randomly selected from a larger random sample in Marin County through the Buck Center for Research in Aging. 19 Medical records were requested from eye care practitioners for all participants who consented to give access to their records. A total of 102 of the 900 refused to give us access to their eye medical records. Records were requested for the remaining 798 participants, but only 662 were received despite repeated requests and visits to the doctors’ offices. Refractive information was available in 621 of these records. Of those, 569 with refractive data who had not had cataract surgery on either eye were selected. Refractive data are presented for 270 males and 299 females. The average age was 75.2 years (SD = 8.9 yrs) for males and 74.8 years (SD = 9.2 yrs) for females. This population contained 171 people 80 years and over. Table 1 contains the sample characteristics for subjects included in the refractive error analysis and subjects who were not included. There is no statistically significant difference in age, gender, years of education, socioeconomic status, or high-contrast visual acuity between the two groups.

This research followed the tenets of the Declaration of Helsinki. Informed consent was obtained from the subjects after explanation of the nature and possible consequences of the study. The research was approved by the institutional review board.

Back to Top | Article Outline

Analysis of Astigmatism

In the past, the most common methods for analyzing and describing refractive errors have been spherical equivalent or spherical error and cylinder error. If the axes of the minus cylinder component were 180° ± 15°, the term with-the-rule was used, if the axes were 90° ± 15°, the term against-the-rule was used. All other axes were labeled as oblique. Each eye thus fell into one of three categories of astigmatism. Newer analysis methods 20, 21 treat refractive error as three vectors—one spherical component (the equivalent sphere) and two cross-cylinders, one at 0° and one at 45°. If an individual has 2.00 D cylinder at, for example, 23°, that astigmatic error is converted to a primary component (cross-cylinder at 0°) and an oblique component (cross-cylinder at 45°). a The formula used in this paper for converting from standard minus cylinder notation for the primary astigmatic component is minus cylinder power × cosine(2 × axis°). The formulas for the oblique astigmatic components are as follows:

Right eye: (−1) × minus cylinder power × sin(2 × axis°)

Left eye: minus cylinder power × sin(2 × axis°)

People frequently have mirror symmetric or near-symmetric axes in the two eyes rather than parallel axes in the two eyes (even though some dispute that claim 23). In our sample, of the 36 people who had at least 0.5 D of oblique component in each eye, all had axes in the right eye >90° and axes in the left eye <90° (“V-pattern” as seen from the examiner’s side, e.g., axis 110 in the right eye and axis 70 in the left eye). The addition of the minus sign for the right eye allows averaging of the oblique component for the two eyes. Without this sign convention, the oblique error disappears on averaging.

Back to Top | Article Outline


Results are presented for six different age groups and plotted at the mean age for each group (less than 65 years (mean, 62; N = 65), 65 to 69 years (mean, 67; N = 117), 70 to 74 years (mean, 72; N = 111), 75 to 79 years (mean, 77; N = 105), 80 to 84 years (mean, 82; N = 80), and 85 or more years (mean, 89; N = 91)).

Back to Top | Article Outline

Spherical Error

The spherical error is shown in Fig. 1 for the right eyes, left eyes, and the average of both eyes. The errors bars show the standard error of the mean. The mean spherical error was near zero for the youngest age group (less than 65 years), increased in the hyperopic direction for those near 70 years, stayed constant for about a decade, and then increased further in the hyperopic direction in the oldest age group (85 or more). The standard deviations were quite large, averaging about 2 D (the standard deviations for men and women are shown in Table 2). Nevertheless, the mean spherical error was statistically significantly different between the 62- and 67-year-old groups as well as the 82- and 89-year-old groups.

Back to Top | Article Outline


The percentage of people in each age group with significant astigmatic components (1.00 D or more) is shown in Fig. 2. The percentages, which are averages for the two eyes, increased with age for both the primary and oblique components by a factor of about 3.5. The prevalence of primary astigmatism was on average also 3.5 times higher than that for the oblique component. The mean primary astigmatism increased from −0.02 D for the youngest group to −0.98 D for the oldest, whereas the oblique component means changed from −0.01 to 0.09 D (data not shown).

The distributions of primary and oblique astigmatism for right eyes for the different age groups are shown in polar plots in Figs. 3 and 4, respectively. The meridia represent the amount of astigmatism increasing from zero diopters (at 270°) to +2.50 D at 90° (counterclockwise) and −2.50 D at 90°(clockwise). Each ring represents 5% increase in prevalence. Positive primary astigmatism refers to with-the-rule astigmatism, and negative primary astigmatism refers to against-the-rule astigmatism. Inspection of Fig. 3 shows a significant change in the shape of the distributions for the primary astigmatic component with age. The age group less than 65 years had surprisingly little primary astigmatism (much less that that found in young observers). With increasing age, a shift in the negative direction occured (against-the-rule), and the proportion of people with ≥1 D of primary astigmatism increased.

A nonparametric test was used for statistical testing because the distributions were not normally distributed. Kruskal-Wallis analysis of variance showed a statistically significant change in the shape of the distributions with age. The Wilcoxon rank sum test with Bonferroni correction for all possible pairwise comparisons revealed that although the distributions for each adjacent age group were not statistically different (i.e., less than 65 vs. 65 to 69; 65 to 69 vs. 70 to 74;. . .80 to 84 vs. 85 or more), all other comparisons were highly statistically significantly different. Inspection of Fig. 4 shows little change in the distribution of the oblique component with age, and statistical testing revealed no significant differences. In addition, most people have small amounts of the oblique component.

Even though Fig. 3 shows the dramatic changes in the distributions of primary astigmatism with age, it does not clearly reveal the number of people with significant negative primary astigmatism. Fig. 5 shows the percentage of people in each age group with ≥1.00 or ≥2.00 D of negative primary astigmatism. The youngest age group (less than 65 years) had little astigmatism, whereas in the oldest group, nearly 50% had ≥1.00 D of primary negative astigmatism (67% had ≥0.75 D of primary astigmatism). A large number also had ≥2.00 D of astigmatism. In subjects 85 years and over, nearly 20% showed 2.00 D of primary negative astigmatism (against-the-rule). The proportion of people with 1.00 D or more of positive primary astigmatism (with-the-rule) was very small, ranging from ∼7% in the youngest group (less than 65) to ∼3% in the oldest group (85 or more) (data not shown).

Back to Top | Article Outline


The percentage of people in each age group with significant anisometropia (of 1 D or more) is shown in Fig. 6 for both the spherical component, primary cylinder component, and the equivalent sphere. Anisometropia was common. Using spherical equivalents, the percentage of anisometropia rose from nearly 10 to 25% over the studied age range. Differences in the primary astigmatism between the eyes of 1.00 D or more were found in 1.5% of the youngest group; this number increased to 26% in the oldest group (85 and older), whereas spherical error anisometropia increased from 20 to 43% over the same age range.

Back to Top | Article Outline

Gender Differences

Table 2 shows the mean refractive error components for males and females for right and left eyes as well as the dioptric differences between the genders and the statistical significance levels using t-tests. For the sample as a whole, males were less hyperopic using both spherical error and equivalent sphere (right and left eyes analyzed separately). Males also had larger amounts of primary astigmatism, whereas there were no differences in the oblique astigmatic component. Primary positive astigmatism of 1.00 D or more was found in 2.8% of males and 5.4% of females, whereas primary negative astigmatism of 1.00 D or more was found in 32.8% of males and 24.2% of females. No gender differences were found for anisometropia for either sphere, equivalent sphere, primary, or oblique astigmatism.

Back to Top | Article Outline

Refractive Error and Education

Other studies have reported a relationship between refractive error and education, finding higher prevalence of myopia in more educated samples. 12–14 Within our sample, we found no significant differences in mean spherical equivalent refractive error or primary cylinder between subjects who had ≤12 years of schooling, 13 to 15 years, or >15 years of schooling. However, as a whole, our sample was significantly less hyperopic (in spherical equivalents) than other samples. Fig. 7 compares our sample to the Beaver Dam, 12 Baltimore, 12 Blue Mountains, 14 and Reykjavik 16 aging studies by plotting the percentage in each age group with spherical equivalent refractive errors of more than +0.50 D and less than −0.50 D in the right eye. These cutoffs were previously chosen and published. Only the results for white participants are included in the Baltimore Eye Survey results. For all age groups, the SKI participants showed less hyperopia and more myopia than the other studies. In the SKI sample as a whole (aged 75, for right eyes), 48.7% showed hyperopia more than +0.5 D and 26.3% showed myopia more than −0.5 D, whereas 40.2% had hyperopia +1.00 D or greater and 22.2% had myopia −1.00 D or more. The nonmonotonic changes with age in spherical equivalent in the SKI sample was due to the increasing astigmatism and general flattening of the spherical component vs. age curve between the ages of 67 and 82 years.

Back to Top | Article Outline


Our results show that the well-known change in astigmatism with age after maturity (e.g., Hirsch 1) continues into old age. The primary astigmatic component changed monotonically from near zero in our youngest age group (less than 65 years) to negative primary astigmatism of increasing amount with age. Young observers have mostly positive primary refractive astigmatism 24 (with-the-rule).

Because we did not measure any components of the eye such as corneal astigmatism, we cannot draw conclusions on the cause of this substantial change in refractive astigmatism with age. The Reykjavik study 16 measured corneal astigmatism and refractive astigmatism and suggested that the corneal changes toward more against-the-rule astigmatism were responsible for the refractive astigmatic changes because both changed similarly and linearly with age. They did not present any data showing what proportion of refractive astigmatism could be accounted for by corneal astigmatism. Others have reported a steepening of corneal curvature with age. A cross-sectional study 25 of corneal curvature in Asian eyes using corneal topography reported a linear increase in corneal curvature in the horizontal meridian with age. The vertical meridian also steepened with age but with a much shallower slope. Between the ages of 60 and 70, the average cornea was reported to be fairly spherical with little or no astigmatism and after age 70, against-the-rule astigmatism became predominant. These authors did not separate their results by gender.

Goto et al. 26 reported more against-the-rule corneal astigmatism in men in their group of 93 people over 50 years of age (mean, 67.6 years). Using a location 1.5 mm eccentric from corneal topography data, 75.5% of the men showed against-the-rule astigmatism (amount not specified—we guess of any amount), whereas 72.5% of the women showed with-the-rule astigmatism (of any amount). Even though corneal astigmatism differs from refractive astigmatism, there is a linear relationship between them. 27, 28 With this correlation in mind, the data set of Goto et al. 26 is quite different from ours. In our group, 69% of men and 65% of women showed primary negative (against-the-rule) refractive astigmatism of 0.1 D or more. The males did, however, have a significantly larger amount of primary negative astigmatism than the females (−0.60 D vs. −0.38 D average of the two eyes). The population in the Goto et al. 26 study were all Asian (Japanese), whereas our population was all white. Racial differences may need to be taken into account when analyzing astigmatic errors and their change with age. Kame et al. 29 compared changes in corneal astigmatism with age in 494 Asian eyes to previous studies involving white eyes and found a greater rate of change of corneal astigmatism in the Asian eyes. Nevertheless, it is clear that the corneal changes with age contribute substantially to the increasing prevalence of negative primary astigmatism in subjects over age 70 years.

Back to Top | Article Outline

Refractive Error and Education

We examined the relation between refractive errors and education level in our group and found no significant difference. However, our results show a much higher prevalence of myopia (and less prevalence of hyperopia) than other population studies. The SKI sample is an unusually highly educated sample with high socioeconomic status. In this sample, 93% finished high school, and 50% finished college. The average number of years of education was 16 years for the three youngest groups, 15 for the next two groups, and 14 years of education for those 85 and over. (37% of those 85 and over had completed college!). This group with 56% women attended college during the 1920s. This is a remarkably high level of education compared with, for example, the Beaver Dam study, 12 where only around 13% of the sample had more than 15 years of education or the Blue Mountains Study, 14 where only about 6% had more than 15 years of schooling compared with 50% in our study. The educational level in the Baltimore Study 13 was quite low, with only 34.6% of white participants having completed high school. The Reykjavik study 16 did not report the average educational level of their participants, but it is likely that most subjects had at least finished high school.

Back to Top | Article Outline


Why does anisometropia prevalence increase from <10% to >25% (for equivalent sphere)? The prevalence of anisometropia of 1.00 D or more in children and young adults is on the order of 3%. 30, 31 The 10% prevalence found in our youngest age group (subjects between 58 and 64 years) is already significantly higher than that found in young children and young adults but similar to other aging studies (e.g., Wang et al. 13). The significant increase in anisometropia in maturity, which further worsens among the oldest old, suggests that the emmetropization mechanisms that maintain virtually identical refractive error in the two eyes during growth in children and young adults fail in old age.

Back to Top | Article Outline

Limitations of the Study

We were not able to get refractive records for all of the original random sample due to either refusal of the participant or lack of response from their eye doctors. For those with medical records, some participants were excluded due to lack of refractive information in the records or a finding that the participant had cataract surgery on either eye. However, there is no evidence of age, gender, education, or socioeconomic status bias in the included sample (Table 1). (Race is not an issue—virtually all the observers in the original sample of 900 were white [96%]). The visual acuity of the included refractive sample did not differ from those not included, allaying fears that the sample was biased toward those seeking eye care due to vision problems. This is a retrospective study using medical records. The refractions were obviously determined by different observers, which is likely to increase the variability of the findings.

Back to Top | Article Outline


The spherical refractive error changes in the direction of more hyperopia with increasing age. This is particularly true for the oldest old. In addition, the prevalence of clinically significant astigmatism and the average amount of astigmatism increase substantially with age. Many of the oldest old have surprisingly large amounts of astigmatism. The primary astigmatic component becomes negative (against-the-rule). The positive primary astigmatic component (with-the-rule) is definitely not the rule. Oblique astigmatism is fairly uncommon but also increases with age. Anisometropia (≥1 D) also increases significantly with age, suggesting failure of emmetropization mechanisms with age. Both the spherical and astigmatic components show dramatic increases in prevalence of anisometropia. The considerable changes in all refractive components, the large amounts of astigmatism, and the high prevalence of anisometropia underscore the importance of regular refractive care for the aging population because it is well known that hyperopia, astigmatism, and anisometropia of clinically significant amounts, if uncorrected, results in reduced visual function.

Back to Top | Article Outline


This study was supported by National Eye Institute, National Institutes of Health grant EY09588 to JAB and the Smith-Kettlewell Eye Research Institute. We thank the volunteers at the Buck Center for Research in Aging.

Back to Top | Article Outline


aThe formulas used in this paper are most similar to those used by Humphrey 22 and differ from those of Thibos et al. 20 by a factor of two for the cylindrical components. Cited Here...

Back to Top | Article Outline


1. Hirsch MJ. Refractive changes with age.In: Hirsch MJ, Wick R, eds. Vision of the Aging Patient. Philadelphia: Chilton, 1960: 63–82.
2. Hirsch MJ. Summary of current research on refractive anomalies. Am J Optom Arch Am Acad Optom 1966; 43: 755–62.
3. Hyams SW, Pokotilo E, Shkurko G. Prevalence of refractive errors in adults over 40: a survey of 8102 eyes. Br J Ophthalmol 1977; 61: 428–32.
4. Baldwin WR, Mills D. A longitudinal study of corneal astigmatism and total astigmatism. Am J Optom Physiol Opt 1981; 58: 206–11.
5. Fledelius HC. Prevalences of astigmatism and anisometropia in adult Danes: with reference to presbyopes’ possible use of supermarket standard glasses. Acta Ophthalmol (Copenh) 1984; 62: 391–400.
6. Aine E. Refractive errors in a Finnish rural population. Acta Ophthalmol (Copenh) 1984; 62: 944–54.
7. Fledelius HC, Stubgaard M. Changes in refraction and corneal curvature during growth and adult life: a cross-sectional study. Acta Ophthalmol (Copenh) 1986; 64: 487–91.
8. Saunders H. A longitudinal study of the age-dependence of human ocular refraction: I. Age-dependent changes in the equivalent sphere. Ophthalmic Physiol Opt 1986; 6: 39–46.
9. Saunders H. A longitudinal study of the age dependence of human ocular refraction: III. The mediation of changes from direct to inverse astigmatism examined by means of matrices of transition probabilities. Ophthalmic Physiol Opt 1987; 7: 175–86.
10. Lavery JR, Gibson JM, Shaw DE, Rosenthal AR. Refraction and refractive errors in an elderly population. Ophthalmic Physiol Opt 1988; 8: 394–6.
11. Lee KE, Klein BE, Klein R. Changes in refractive error over a 5-year interval in the Beaver Dam Eye Study. Invest Ophthalmol Vis Sci 1999; 40: 1645–9.
12. Katz J, Tielsch JM, Sommer A. Prevalence and risk factors for refractive errors in an adult inner city population. Invest Ophthalmol Vis Sci 1997; 38: 334–40.
13. Wang Q, Klein BE, Klein R, Moss SE. Refractive status in the Beaver Dam Eye Study. Invest Ophthalmol Vis Sci 1994; 35: 4344–7.
14. Attebo K, Ivers RQ, Mitchell P. Refractive errors in an older population: the Blue Mountains Eye Study. Ophthalmology 1999; 106: 1066–72.
15. Bengtsson B, Grodum K. Refractive changes in the elderly. Acta Ophthalmol Scand 1999; 77: 37–9.
16. Gudmundsdottir E, Jonasson F, Jonsson V, Stefánsson, Sasaki H, Sasaki K and the Iceland-Japan Co-Working Study Groups. “With the rule” astigmatism is not the rule in the elderly. Reykjavik Eye Study: a population based study of refraction and visual acuity in citizens of Reykjavik 50 years and older. Acta Ophthalmol Scand 2000; 78: 642–6.
17. Haegerstrom-Portnoy G, Schneck ME, Brabyn JA. Seeing into old age: vision function beyond acuity. Optom Vis Sci 1999; 76: 141–58.
18. Brabyn J, Schneck M, Haegerstrom-Portnoy G, Lott L. The Smith-Kettlewell Institute (SKI) longitudinal study of vision function and its impact among the elderly: an overview. Optom Vis Sci 2001; 78: 264–9.
19. Reed D, Satariano WA, Gildengorin G, McMahon K, Fleshman R, Schneider E. Health and functioning among the elderly of Marin County, California: a glimpse of the future. J Gerontol A Biol Sci Med Sci 1995; 50: M61–9.
20. Thibos LN, Wheeler W, Horner D. Power vectors: an application of Fourier analysis to the description and statistical analysis of refractive error. Optom Vis Sci 1997; 74: 367–75.
21. Raasch T. Clinical refraction in three-dimensional dioptric space revisited. Optom Vis Sci 1997; 74: 376–80.
22. Humphrey WE. Automatic retinoscopy: the Humphrey vision analyser. Optician 1977; 173: 7–16.
23. McKendrick AM, Brennan NA. The axis of astigmatism in right and left eye pairs. Optom Vis Sci 1997; 74: 668–75.
24. Hirsch MJ. Changes in astigmatism after the age of forty. Am J Optom Arch Am Acad Optom 1959; 36: 395–405.
25. Hayashi K, Hayashi H, Hayashi F. Topographic analysis of the changes in corneal shape due to aging. Cornea 1995; 14: 527–32.
26. Goto T, Klyce SD, Zheng X, Maeda N, Kuroda T, Ide C. Gender- and age-related differences in corneal topography. Cornea 2001; 20: 270–6.
27. Grosvenor T, Quintero S, Perrigin DM. Predicting refractive astigmatism: a suggested simplification of Javal’s rule. Am J Optom Physiol Opt 1988; 65: 292–7.
28. Grosvenor T, Ratnakaram R. Is the relation between keratometric astigmatism and refractive astigmatism linear? Optom Vis Sci 1990; 67: 606–9.
29. Kame RT, Jue TS, Shigekuni DM. A longitudinal study of corneal astigmatism changes in Asian eyes. J Am Optom Assoc 1993; 64: 215–9.
30. Hirsch MJ. Anisometropia: a preliminary report of the Ojai Longitudinal Study. Am J Optom Arch Am Acad Optom 1967; 44: 581–5.
31. Almeder LM, Peck LB, Howland HC. Prevalence of anisometropia in volunteer laboratory and school screening populations. Invest Ophthalmol Vis Sci 1990; 31: 2448–55.

SKI study; aging; astigmatism; anisometropia; hyperopia; gender differences

© 2002 American Academy of Optometry