**Purpose.**:
To examine the heritability of refractive astigmatism in older women.

**Methods.**:
Astigmatism was measured with an autorefractor in 88 monozygotic and 82 dizygotic female twin pairs aged 63 to 75 years. The prevalence and distribution of astigmatism and polar values J0 and J45 were estimated by standard statistical methods. Bivariate maximum likelihood model fitting was used to estimate genetic and environmental variance components using information from both eyes.

**Results.**:
Mean astigmatism of the more astigmatic eye was 0.93 diopters (D; SD ±0.58). Astigmatism of at least 0.25 D, 0.5 D, 0.75 D, or 1.0 D in either eye was present in 99.7%, 88.5%, 66.5%, and 46.2% of cases, respectively. The main direction of astigmatism was against the rule. The age-adjusted quantitative genetic modeling revealed that additive genetic effects accounted for 33.3% (95% confidence interval [CI], 21.9%–43.8%) of the total variance of astigmatism and for 18% (95% CI, 4%–31%) of the total variance of polar value J45 of both eyes (bivariate model), with the remaining variances due to nongenetic effects. There were no significant correlations between the twin pairs for polar value J0.

**Conclusions.**:
In elderly female twins, additive genetic effects accounted for one-third of the variance of the amount of astigmatism and only a small fraction of the total variance of polar value J45.

^{ 1 }Various corneal diseases, eye surgery affecting the cornea, or other surgical procedures—for example, cataract, glaucoma, and retinal ablation surgery—may induce astigmatism. The amount and direction of astigmatism has been shown to vary with age.

^{ 2–5 }Asano and colleagues, for example, found mean astigmatism in the right eye of 0.77 D and 1.25 D among 40- to 49-year-olds and 70- to 79-year-olds, respectively.

^{ 2 }The direction of astigmatism tends to change with age, the main trend being an increase against the rule (ATR) and a decrease with the rule (WTR).

^{ 6 }Differences in the prevalence of astigmatism also seem to occur between different ethnic groups.

^{ 7 }

^{ 8 }studied the occurrence and heritability of astigmatism in a population-based sample of Norwegian twins through self-reported history of astigmatism from birth to age 31 years in 8045 twins. The best-fitting biometrical model suggested that genetic effects due to dominance explained 54% (95% CI: 20–69 years) and additive genetic effects explained 9% (95% CI: 0–40 years) of the variation in the liability to astigmatism. In the study of Hammond et al.

^{ 9 }among twin pairs aged 49 to 79 years (mean age, 62.4 years), genetic effects due to dominance accounted for 47% to 49% (95% CI, 37%–55%) and additive effects for 1% to 4% of the variance of total astigmatism (95% CI, 0%–13%). Dirani et al.

^{ 10 }studied CA in 18- to 86-year-old Australian twins (mean age, 52.11 ± 15.85 years). Heritability estimates were as high as 60% for CA. In our recent article, quantitative genetic modeling showed that heritable factors explained 83% of the variance in spherical equivalent in 63- to 76-year-old Finnish female twins.

^{ 11 }In the same population, genetic factors explained 81% of the variance in corneal refraction, additive genetic factors 62% (95% confidence interval [CI] 44%−86%), and dominant genetic factors 19% (95% CI 7%–49%).

^{ 12 }For CA, it was not possible to construct a meaningful model, although the values of the intraclass correlation coefficient (ICC) were higher for monozygotic (MZ) than dizygotic (DZ) twins.

^{ 12 }The main purpose of the present study was to determine the heredity of refractive astigmatism in the same elderly Finnish female twin subjects.

^{ 13–14 }

- The Finnish Twin Cohort Study (started in 1975): 13,888 adult twin pairs in database.
- In the year 2002, there were 1260 surviving pairs of female twins born between 1924–1937.
- Of that number, 414 pairs were invited to participate.
- A total of 217 pairs attended the study center examinations. To be included, both cotwins had to agree to participate. Reasons for nonparticipation were one or both sisters' unwillingness to participate (106 pairs), and disease or poor health status (91 pairs).
- A total of 47 pairs were excluded due to cataract or glaucoma operation.
- The final analyses comprised 170 twin pairs (88 MZ and 82 DZ).

^{ 15 }and later confirmed by applying a battery of 10 highly polymorphic gene markers at the National Public Health Institute to DNA extracted from a sample of venous blood.

^{ 16 }J0 refers to +cylinder power set orthogonally at the 0 and 90° meridians and represents WTR or ATR astigmatism. Positive values of J0 indicate WTR astigmatism, and negative values ATR astigmatism. J45 refers to a cross-cylinder set at 45 and 135°, representing oblique astigmatism. Negative values of J45 indicate astigmatism of around 45° and positive values astigmatism of around 135°.

^{2}test in the case of discrete variables (e.g., direction of astigmatism).

^{ 17 }In quantitative genetics studies, genetic effects are typically classified into additive genetic effects (A) and nonadditive genetic effects (D). Environmental effects are classified into shared environmental effects (C) and nonshared environmental effects (E). Shared environmental effects are common to both members of a pair while nonshared effects refer to external exposures affecting only one sibling, such as accidents, surgery, or measurement error. Quantitative genetic modeling is based on necessary similarities and differences in the correlations of the A, D, C, and E factors explaining the variability among MZ and DZ twins. The correlations for additive and nonadditive genetic effects are defined as 1.0 in MZ pairs, and as 0.5 and 0.25, respectively, in DZ pairs. The correlation for shared environmental effects is defined as 1.0 and for nonshared environmental effects as 0, among both MZ and DZ pairs. The aim of genetic modeling is to construct a model that explains the data well and has as few explanatory components as possible. When using bivariate modeling it is necessary to separate effects that are common for both variables included in the model and effects that are only specific for one variable in the model. To sharpen this difference small letters c (common) or s (specific) are used with letters A, D, C, and E.

^{ 18 }; Michael Neal, Virginia Commonwealth University, Richmond, VA).

*P*= 0.22) or between MZ and DZ individuals (

*P*= 0.42). The correlations between the amount of astigmatism and age were nonsignificant. The correlations between the amount of astigmatism and spherical equivalent in the whole dataset and separately among those with either positive or negative SE were nonsignificant in each eye.

**Table 1**

**Table 1**

Direction of Astigmatism | Right Eye | Left Eye | ||

Prevalence, % | Mean, D (±SD) | Prevalence, % | Mean, D (±SD) | |

WTR | 29.9 | 0.78 (±0.66) | 29.5 | 0.74 (±0.66) |

ATR | 50.9 | 0.73 (±0.52) | 48.9 | 0.68 (±0.46) |

Oblique | 19.2 | 0.61 (±0.50) | 21.6 | 0.66 (±0.42) |

*P*< 0.001) or ATR (

*P*= 0.001) directions. The correlations in the left eye did not reach significance.

^{2}test,

*P*= 0.124 for right eye, and

*P*= 0.060 for left eye).

**Table 2**

*n*= 28) and −0.145 for DZ (

*n*= 28); and for the ATR subgroup, 0.697 (

*n*= 15) for MZ and 0.027 (

*n*= 9) for DZ. The low number of cases did not permit calculations for the oblique subgroup.

^{ 19,20 }Although our study design does not permit general conclusions to be drawn on the prevalence of astigmatism in the population, the prevalence of astigmatism of ≥0.5 D (71.2% in the right and 70.0% in the left eye) found in the present study does not differ greatly from that reported by Schellini et al.,

^{ 20 }in a Brazilian population aged 70 years or older (71.7% had astigmatism of ≥0.5 D). In the National Health and Nutrition Examination Survey, the prevalence of astigmatism of ≥1.0 D was 50.1% in persons aged 60 years or older; whereas in our study, this grade of astigmatism in one eye or the other was found in 46.3% of cases.

^{ 21 }Thus the main difference between the prevalence found here and that reported in most previous studies is probably due to the fact that astigmatism in our study started at 0.25 D.

^{ 1,22 }In the present study, no significant correlations between astigmatism and SE were found. The small proportions of myopics and cases of high ametropia in this study could be one explanation for the nonsignificant relationships observed between the amount of astigmatism and spherical ametropia.

^{ 8 }In the study of Hammond et al.,

^{ 9 }among 49- to 79-year-old twins, additive genetic effects accounted for 1% to 4% (95% CI, 0%–13%) and dominant effects for 47% to 49% (95% CI, 37%–55%) of the variance of total astigmatism. In the Hammond et al.

^{ 9 }study and our study, the MZ correlations for astigmatism were very similar (0.4 to 0.5), while the DZ correlations in Hammond et al.

^{ 9 }were 0.2 to 0.10 and in ours negative. Despite the larger sample size in Hammond et al.,

^{ 9 }neither study had the power to distinguish unambiguously between additive and nonadditive genetic sources of variation. Given the differences in sample size and in characteristics resulting in some variation in the actual variance/covariance matrices underlying the pairwise correlations, it was not unexpected that the best-fitting model differed in the two studies. Both studies point to the importance of genetic factors, but whether additive or nonadditive factors are more important cannot be stated. Thus, in Hammond et al.,

^{ 9 }an ADE model for astigmatism fit better than an AE model for the left but not the right eye in the univariate models. In the bivariate models, Hammond et al.

^{ 9 }only present results for an ADE model and no results for an AE model.

^{ 11,12 }Based on the results of this study and the results of our earlier studies on the same subjects, it is reasonable to conclude that the impact of heredity among elderly females is highest on spherical equivalent refraction and corneal refraction, while environmental factors have a stronger influence on the amount of corneal and refractive astigmatism.

^{ 2–5 }The low impact of heredity on astigmatism found in the present study with elderly females would support the theory that the influence of environmental factors on astigmatism increases at older ages. Possible changes in astigmatism should be taken into account when controlling for refraction and prescribing new glasses, and when planning refractive surgery, especially when doing costumed corneal ablations. Emmetropic nonastigmatic refraction achieved by refractive surgery at a young age may turn to astigmatic refraction later on during the life course.

**O. Pärssinen**, None;

**M. Kauppinen**, None;

**J. Kaprio**, None;

**M. Koskenvuo**, None;

**T. Rantanen**, None

*Clin Exp Optom*. 2007; 90: 5–19. [CrossRef] [PubMed]

*Jpn J Ophthalmol*. 2005; 49: 127–133. [CrossRef] [PubMed]

*Vision Res*. 1980; 20: 891–893. [CrossRef] [PubMed]

*Acta Ophthalmol (Copenh)*. 1986; 64: 487–491. [CrossRef] [PubMed]

*Invest Ophthalmol Vis Sci*. 1984; 25: 88–92. [PubMed]

*Ophthalmic Physiol Opt*. 1981; 1: 159–174. [CrossRef] [PubMed]

*J Am Optom Assoc*. 1993; 64: 215–219. [PubMed]

*Ophthalmic Epidemiol*. 2006; 13: 245–252. [CrossRef] [PubMed]

*Invest Ophthalmol Vis Sci*. 2001; 42: 1232–1236. [PubMed]

*Invest Ophthalmol Vis Sci*. 2008; 49: 1339–1344. [CrossRef] [PubMed]

*Ophthalmology*. 2010; 117: 1908–1911. [CrossRef] [PubMed]

*Acta Ophthalmol*. 2013; 91: 140–144. [CrossRef] [PubMed]

*Med Sci Sports Exerc*. 2005; 37: 72–78. [CrossRef] [PubMed]

*Twin Res. Review*. 2002; 5: 358–365. [CrossRef]

*Hum Hered*. 1978; 28: 241–254. [CrossRef] [PubMed]

*Optom Vis Sci*. 1997; 74: 367–375. [CrossRef] [PubMed]

*Twin Res*. 2003; 6: 361–376. [CrossRef] [PubMed]

*Mx: Statistical Modeling*. 6th ed. Richmond, VA: Department of Psychiatry, Virginia Commonwealth University; 2003.

*Ophthalmic Epidemiol*. 2009; 16: 231–237. [CrossRef] [PubMed]

*Ophthalmic Epidemiol*. 2009; 16: 90–97. [CrossRef] [PubMed]

*Arch Ophthalmol*. 2009; 126: 1111–1119. [CrossRef]

*Ophthalmology*. 1982; 89: 298–302. [CrossRef] [PubMed]