Changes in ocular astigmatism with age: A longitudinal study

To investigate changes in astigmatism with age. Are changes from with‐the‐rule (WTR) in younger ages to against‐the‐rule (ATR) in older patients mediated through oblique astigmatic axes or spherical prescriptions, and at what ages do these changes occur?

per decade from age 50 years has been reported and attributed to changes in anterior corneal astigmatism. 2It has been suggested that these changes from WTR to ATR are due to changes in the eyelid pressure 1 ; however, the evidence is not conclusive.The prevalence of oblique astigmatism is lower and more constant with age than WTR or ATR at between 10%-20% 3,4 and 25%-30%. 1 The change in prevalence from WTR to ATR with age is well documented [1][2][3][4][5] ; however, little has been published regarding how these changes are mediated.Specifically, whether small degrees of WTR astigmatism (which are very common at younger ages) rotate through oblique axes, or reduce to zero (spherical prescriptions) to re-emerge as ATR.In a small longitudinal study, Saunders 4 used prescriptions of the right eye of 122 subjects, indicating that some changes in the prevalence of WTR to ATR occur in the 30-to 39.9-year age group.However, the rate of change increased in the 40-to 49.9-year age group.For astigmatism of up to 1.00 DC, the transition from WTR to ATR was twice as likely to pass through zero (that is a sphere-only prescription) as through oblique axes, agreeing with their earlier study. 6It was suggested that some clinicians might be reluctant to prescribe small oblique cylinders, hence reducing those numbers.
Asharlous et al., 5 using autorefractor prescription data gathered from 160,608 Iranian patients aged 11-80 years, reported that 83% had the same classification of WTR, ATR or oblique (OBL) astigmatism in both the right and left eyes, and that the prevalence of WTR in that (isorule) group decreased from 21 to 30 years, ATR and OBL increased, with an apparent increase in the rate of change at around age 40 years.The proportion of bilateral oblique astigmatism was very low at just 2.7% of the total sample.However, of the 17% who had a different classification between the eyes, oblique cylinders (together with WTR or ATR in the fellow eye) accounted for 13%.It thus appears that where correlation between the eyes was found, there was a small proportion of oblique cylinders.However, where the eyes differed, the proportion of oblique axes was very much higher.Importantly, a different classification of astigmatism between the eyes might involve axes of 180° and 90° but could also involve axes of 30° and 31°, with 30° being defined as WTR and 31° as OBL.
The aim of this study was to generate a longitudinal refraction data set, which could be used to investigate changes in prescribed corrections for astigmatism with age.Is the well documented change in astigmatism from WTR to ATR with age mediated through spherical prescriptions or through oblique axes?Where prescriptions pass through the oblique, how transient are these changes?Additionally, as an oblique change in cylinder of just −0.25 DC can potentially cause problems of non-tolerance, 7 how likely are oblique cylinder changes to be prescribed and for which ages?

METHODS
Ethics approval was obtained from the University of Bradford Ethics Committee (EC27485).The patient database of a practice in West Yorkshire, belonging to a large multiple group, was made available with permission from the directors of that practice.Patient records of routine eye examinations conducted by 28 optometrists, from both electronic and older, paper patient records were examined with refraction data recorded for right and left eyes for the years that patients attended the practice.To determine the cylinder power and axis, it is highly likely a Jackson Cross Cylinder was used, either with a trial frame or with a phoropter, but these details are unknown.As changes in astigmatism could possibly occur from around 30 years of age, 1,5 although likely somewhat later, to ensure capture of any changes, patients aged 28 years and over at their first examination were included, and were recorded by patient number to ensure anonymity.Records suggesting the patient had cataract, keratoconus or chalazion, or a history of cataract or refractive surgery, were excluded.Additionally, rigid gas-permeable contact lens wearers were excluded, together with anisometropic amblyopic eyes, for which a balance or partial prescription can often be prescribed in adults.Any apparent typographical error excluded that prescription (e.g., −3.25 × 110, followed by −3.25 × 10 and at the next −3.50× 112, the middle axis clearly an error; however, the clinician could have omitted the first or the last digit).Where a recheck was conducted, the rechecked prescription was recorded and not the first prescription which caused the problem.
The sphere, cylinder and axis data were converted to power vectors, 8 and the last values of J 0 and J 45 minus the first values were calculated for each case.A simple regression analysis (Microsoft Excel, Micro soft.com) was performed to assess the correlation between the data from the right and left eyes.
Each patient's refraction history was assessed by the authors independently, and the nature of any change in their astigmatism recorded.The assessment involved consideration of changes in the classification of the astigmatism

Key points
• Changes from with-the-rule astigmatism in younger ages were three times more likely (56 vs. 16) to pass through oblique astigmatism than sphere-only prescriptions towards against-therule in older ages.• Oblique cylinders are prescribed more commonly than previous prevalence figures would suggest, with 36% of eyes (232/640) having an oblique cylinder at least once in their refractive history.• Changes in ocular astigmatism occur after the mid-40s (55%) with many (41%) showing no changes until after their mid-50s.
(WTR, ATR, OBL or spherical [SPH]) through the patient history, for example, noting whether WTR at the first eye examination and ATR at the last had SPH or OBL as interim prescriptions and also noting any increases or decreases in cylinder power.Different definitions of WTR, ATR and OBL astigmatism have been used.However, the definitions used here are consistent with Read et al. 1 and Asharlous et al. 5 ; using negative cylinder format, WTR is 0°-30° and 150°-180°, ATR is 60°-120° and OBL is 31°-59° and 121°-149°, giving approximately equal ranges for each.

Statistical analysis
Statistical tests and analyses assume independence of observations in a sample.Where data are correlated, as with the right and left eyes, 5 treating each eye as an independent variable results in exaggerated levels of significance, 9 with Armstrong 10 adding that using data from both eyes where there is a correlation would increase Type I errors (false positives).However, performing separate analyses using just the right or left eye, while valid, may result in loss of statistical power and hence possible loss of statistical significance. 9Where data from one eye are used, the selection of which eye should be random, in contrast to many studies which favoured the right eye, or what was estimated as the dominant eye. 10 Since the prevalence of oblique cylinders appears to be higher when the right and left eyes have a different classification than when the eyes are the same, 5 if we considered either the right or left eye randomly, valuable data regarding oblique cylinders could be lost.Consequently, in this study, we used data from both eyes for analyses that do not use statistical tests and in this way avoid causing Type 1 errors.

A Summary
Most rejections of patient records were due to either insufficient years of examinations (fewer than 18 years), young patients or a history of cataract.To estimate the proportion of records included versus those rejected from all those examined, a sample was taken over several days of data collection.Of 1135 records examined, just 86 (7.6%) were included, with the overwhelming majority of rejections due to insufficient years of prescription history.Further details are shown in Table 1.

B Changes from WTR or SPH to ATR, through OBL or SPH prescriptions
There were 56 eyes, with either WTR or SPH at their first prescription which passed through OBL to ATR, of which nine patients had the same change in both the right and left eyes.A total of 16 eyes had WTR at their first prescription and passed through SPH to ATR, of which no patient had the same change in both the right and left eyes.Statistically this was highly significant ( 2 1 = 22.2, p < 0.0001).
C Age at which any cylinder changes start to occur (these include all changes, e.g., increase in ATR or decrease in WTR, in addition to part B, above) The 28-to 30-year and 31-to 40-year patient groups were combined due to low numbers of each group and assessed as one group of 91 younger patients at their first recorded eye examination.Of these patients, five were monocular (the fellow eye being amblyopic); data from nine eyes could not be assessed as their cylinder (≤0.50 DC) was too variable between successive prescriptions and one eye could not be assessed due to a 10-year period between examinations during which time their prescription had changed.Thus, a total of 167 eyes were derived from the 91 patients.These can be summarised as follows: (i) 7/167 eyes (4%).The cylinder power and/or axis appeared to have been changing at or from their first eye examination, with a mean age of these six patients at their first examination of 33 years (one patient had both eyes included).(ii) 68/167 eyes (41%).No change in cylinder throughout their recorded history.The greatest final age of any patient was 64 years and the mean final age was 54 years (SD 4.6).(iii) 92/167 eyes (55%).The cylinder was relatively constant, before starting to change, with a mean age at which the changes started of 44 years (SD 6.2).
Thirty-three patients (38% of the 86 patients with data for both eyes) had either a different progression of cylinder (i, ii or iii, above) between their right and left eyes or a difference >5 years in the onset of changes between the right and left eyes.The 40+ age groups were not assessed, as from the results shown above, more than 50% would likely have been changing by the time of their first eye examination.

D Prevalence of types of astigmatism
To determine the prevalence of types of astigmatism, either the right eye or the left eye was randomly selected for each patient and then a single prescription was randomly selected from their history (Random.org) to allow comparison of results with previous cross-sectional studies (Table 2).

E Oblique cylinders
Table 3 summarises patients for whom the assessment of their astigmatism was either similar or dissimilar for both the right and left eyes.Examples of similar are increasing ATR in the right and left eyes, or a change from WTR to ATR through SPH in both eyes.Examples of dissimilar are increasing ATR in the right eye and unchanging ATR in the left, or unchanging WTR in the right and slightly rotating OBL in the left.
In the complete data set, there were a total of 51/640 'variable' eyes (where the cylinder power and axis difference between successive prescriptions varied and could not be summarised as part of a trend), with eight patients having variable prescriptions in both their right and left eyes.Of these 51 'variable' eyes, 44 (86%) had an oblique cylinder in at least one of the prescriptions in their refraction history.
The oblique cylinders which were present for just one or two prescriptions in the patient history, as one variable prescription or as the last prescription recorded (182/232, 78.4%), can be considered transient in nature.These were much more common than those which were either unchanging or increasing in power (50/232, 21.6%).

F Increases in ATR power and changes in WTR
A total of 150 out of 640 eyes (23% of all eyes) from 98 patients showed a progressive increase in ATR astigmatism.These 150 eyes had a mean increase of 0.98 DC over a mean of 13.8 years (corresponding to an increase of 0.71 D per decade), with a mean age of 56 years (SD 9.9) at which time the ATR astigmatism started to increase.Of these 150 eyes, 18 were 28-40 years of age at their first examination and are included in Results C (iii).
One hundred and twelve out of 640 eyes (18%) from 80 patients had WTR astigmatism which remained WTR (i.e., did not rotate to OBL) and where the power showed no change, increased or decreased.
There were:

DISCUSSION
In contrast with the results of Asharlous et al. 5 reporting a high correlation of astigmatism between the right and left eyes, in the present study, 38% of 86 patients aged 28-40 at their first test had either a different progression of cylinder or a different age at the onset of changes in astigmatism between their right and left eyes (Results C) and half of the 314 binocular patients in the full data set had a different assessment of the cylinder changes between the prescription of the two eyes (Table 3).Additionally, of the 72 eyes assessed as passing from WTR to ATR through either SPH or OBL, just nine (13%) had the same change in both the right and left eyes.Comparison between the eyes might therefore suggest a good correlation if based on prevalence; however, Abbreviations: ATR, against-the-rule; OBL, oblique axes; WTR, with-the-rule.

T A B L E 3
The incidence of oblique astigmatism in cases where the changes in the right eye were assessed as similar or dissimilar to those in the left eye (N = 314 patients for whom data were collected for both the right and left eyes).where the changes in astigmatism are assessed, the eyes often show poor associations, particularly when oblique cylinders are involved.In our assessments, if either the right or the left eye was selected randomly, then valuable data regarding cylinder changes would have been missed.

Oblique cylinders and changes from WTR or SPH to ATR, through OBL or SPH prescriptions
Our results suggest that changes from WTR to ATR are over three times more likely to pass through oblique axes than a spherical only prescription.This contrasts markedly with Saunders, 4 who (measuring the right eye only) found the change from WTR to ATR was twice as likely to pass through SPH as OBL.While our prevalence of oblique astigmatism was assessed to range from just 11% in the 31-to 40-year age group to 19% in the over 60s, 36% of eyes (232/640) had oblique astigmatism occurring in at least one prescription in their history, while 78% of oblique astigmatism was transient in nature and hence would be more likely not to show in the prevalence results.Saunders' 4 analysis was based on 634 separate data points generated from the right eye of 122 patients and therefore did not appear to have independence of data.He then compared frequencies of axes within age groups and based on the absence of any increase or steady change in oblique frequency in middle ages, concluded that the WTR to ATR was mediated through spherical prescriptions.No analysis based on individual prescription histories appears to have been made, although with just 122 patients, some of whom were younger than 30 years, any such conclusions might not have been significant.Saunders made the important point that some clinicians might have been reluctant to prescribe small oblique cylinders, further reducing those numbers.Grosvenor 11 suggested that cylindrical corrections <0.50 DC are seldom prescribed unless also prescribing for myopia or hyperopia while Benjamin 12 recommended not prescribing a small cylinder unless subsequent refractions suggest that it is required, adding that the acceptance of the prescription may depend on the habitual prescription.It is likely that at least some clinicians follow this advice, particularly for small oblique cylinders and hence some patient refractive histories might have yielded more small oblique cylinders had a different clinician performed the refraction.However, in recent years with the increased use of autorefractors, it appears that many clinicians do appear happy to prescribe such cylinders. 7This is further evidenced in the present study by the number of patients (44/51, 86% of those with variable prescriptions) who had an oblique cylinder prescribed in a variable cylinder history.The prescription of oblique cylinders has been particularly implicated in patient complaints with new spectacles, with 23% of all complaints (rechecks) following the dispensing of new spectacles having an oblique cylinder noted as a contributory cause in our recent recheck study. 7ome patient histories started with spherical prescriptions for their first one or two recorded prescriptions (N = 25), and consequently it could be argued that these could have changed from WTR in earlier unrecorded prescriptions and so should be added to the 'through SPH' group.However, it is also likely these might have had an earlier OBL cylinder and hence it is impossible to state whether they should be added to the through SPH or OBL groups.Therefore, what we can say with certainty is of the 72 eyes for which we observed the change from WTR to ATR, more than three times as many passed through oblique cylinders than spherical prescriptions.
Small changes in oblique astigmatism have been shown to cause patient problems 7 and so in the short term, the temporary presence of oblique cylinders in many prescriptions could increase the likelihood of patient non-tolerance compared with more stable oblique astigmatism that the patient has had time to adapt to.Additionally, out of 51 eyes which showed no clear trend and for whom their astigmatism appeared to vary markedly through their history, 44 (86%) had oblique astigmatism at some point, suggesting that clinicians who appeared happy to make big changes to a prescription, perhaps without reconciling their prescription with the patient's habitual prescription and presenting symptoms, were equally happy to prescribe oblique cylinders which might further increase the non-acceptance rate.

Age at which cylinders start to change
Cylinder changes (ATR, WTR or OBL) of either power and/ or axis appear rare in patients <40 years of age (7/167 eyes, 4%) and 41% showed no change until at least their mid-to late 50s, or after.The majority of eyes (55%) experienced no or little change until their mid-40s, after which their cylinder started to change.These onset of changes in astigmatism with age appear consistent with some studies 2,3 and a little later than others, 1,5 although the direct comparison of these longitudinal results with previous cross-sectional studies based on prevalence is not straightforward.
A clinical implication of these shifts with age is that if a large cylinder change is found for a young patient, which would not be expected from these results, then conditions such as keratoconus should be excluded.

Prevalence of types of astigmatism
Using random selection of our longitudinal data, we generated prevalence figures in an attempt to compare with previous cross-sectional studies, finding a change in the prevalence of WTR to ATR with age (Figure 1) that was largely consistent with Read et al. 1 Apparent differences were our greater prevalence of WTR in the 31-to 40-year age group (64% vs. approximately 45% for Read et al. 1 ) and our oblique astigmatism prevalence was somewhat lower at 11%-19% compared with 25%-30%, but is consistent with Saunders 4 and Namba et al. 3 As 78% of oblique astigmatism was found to be transient in nature (Table 4) and thus might not show in prevalence data (36% of all eyes had an oblique cylinder occurring at least once in their prescription history), comparison of the prevalence of oblique astigmatism with previous studies cannot be made reliably, and is further complicated as definitions of the axes corresponding to oblique astigmatism can vary. 1,4

Increases in ATR power and changes in WTR
Changes in WTR, ATR or OBL astigmatism start occurring by the mid-40s for many patients (92/167, 55%).However, an increase in ATR astigmatic power does not appear to occur until later, by the mid-50s (SD 9.9), although there was a wide spread of ages at which this increase occurred.Our result of 23% (150/640) of all eyes showing an increase in ATR astigmatism of 0.71 D per decade after a mean age of 56 years appears broadly consistent with the result of Naeser, 2 namely that the total corneal astigmatism of all patients from 50 years of age shows a change of 0.25 D ATR per decade.
Of the 112 WTR eyes showing an increase in power, decrease or no change, just 36/112 (32%) had decreasing powers, although as those showing no change or an increase were at a younger mean age at their first examination (40 vs. 47 years for the decreasing group), it is likely that many of these would go on to show a decrease in power at a later age.Just eyes showed a rotation of >20° while remaining within a WTR classification, three of whom had a power of ≥1.00 DC.The 17 eyes showing decreasing power of ≥1.00 DC at the first examination had a mean change of 0.34 DC per decade.This is a lesser change per decade at a lower mean age than the increase in ATR, suggesting that astigmatic changes start at this lower rate and accelerate at later years, perhaps due in part to the onset of lenticular changes.We excluded records where cataract was recorded in conjunction with reduced visual acuity.However, it is likely that some patients experienced astigmatic changes due to prodromal lenticular opacities, 13 while retaining good visual acuity and thus were included in the analysis.

Study limitations
This study collected data from prescriptions that were measured by a total of 28 optometrists over a maximum period of 30 years.MacKenzie 14 assessed the reproducibility of prescriptions, measured by 40 optometrists on a single 29-year-old, asymptomatic, male subject.The 95% reproducibility of the antistigmatic (J 0 and J 45 ) components was found to be 0.24 D, which the author stated was equivalent to approximately 0.50 D cylinder.This figure for reproducibility (repeated measurements of one subject by multiple practitioners, or the inter-variability) could be considered an upper limit of the variability of measurement of prescriptions, while a minimum limit can be considered as the figure for repeatability (multiple measurements by one practitioner, or the intra-variability), which Rosenfield and Chiu 15 reported as 0.25 D.
We were concerned with investigating the change in cylinder power from WTR to ATR, which has been quoted as 0.25 DC per decade, 2 and thus the expected 0.50 DC in two decades could be swamped by the variability in interpractitioner measurement.These variability figures 14,15 were determined by isolated refractions.However, in this study we examined longitudinal data where each refraction was not an isolated event, but part of a chain of findings.Additionally, when performing a refraction, each practitioner performed their measurements with a knowledge of the previous prescription, as well as any symptoms and the associated visual acuities (although as discussed above, it is likely that some practitioners did not consider the patient's refraction history).It might therefore be expected that the inter-practitioner variability would be lower than previously reported.Note: Total N = 232/640 eyes (36.3%) with an oblique cylinder occurring in at least one prescription in the recorded history.Fifty-six patients had an oblique cylinder occurring in both eyes and there were three monocular patients.
Abbreviations: ATR, against-the-rule; OBL, oblique axes; SPH, spherical refraction; WTR, with-the-rule.a Includes the 56 eyes where the prescription changed from SPH or WTR to ATR through OBL (Results B), together with those with a first OBL prescription before changing to ATR, or as part of another trend (e.g., mostly WTR axis 150-160, with one prescription with a 140 axis).b Of the 38 eyes with oblique astigmatism as their last recorded prescription (16.4% of eyes), 36 showed a rotation of their cylinder from WTR to OBL and two from ATR to OBL.

F I G U R E 1
Prevalence of types of astigmatism, illustrating the change between successive decades of patient age.ATR, against-therule; OBL, oblique axes; WTR, with-the-rule.T A B E 4 Assessment of the nature of oblique cylinders.Nature of occurrence N (%) N (%)Briefly present in the patient history a 100 (43.1) 182 (78.4)Variable (prescriptions which varied over the history, with no trend) Details of the included patients (total patients, N = 326; total eyes, N = 640; 12 patients were monocular, the fellow eye being amblyopic).
T A B L E 1