Adult Myopia Progression

Abstract

Purpose: To explore evidence for myopic shift between the ages of 20 and 50 years.

Methods: Three usable sets of data with long-term adult refractive progression were identified: (1) US population–based prevalence data for those 18 to 24 years of age in 1971 and 1972 and 45 to 54 years of age from 1999 to 2004; a logit transformation of prevalence values at different refractive error thresholds allowed estimation of myopic progression in this group. (2) German clinical data describing 5- to 10-year progression for different refractive error groupings across 5-year age bands from 20 to 49 years; these were extracted, adjusted, and analyzed. (3) Five-year progression rates with similar breakdown of age and refractive error groups as the German data but in a Japanese clinical population.

Results: Estimates of progression between 20 and 50 years for the given studies were: (1) −1.1, −1.4, and −1.9 diopters (D) for baseline refractive errors of −1, −3, and −6 D, respectively; (2) a range from −1.0 to −2.9 D, increasing with degree of baseline myopia; (3) a weighted average of −1.0 D for males and −0.9 D for females but with decreasing progression with increasing myopia. In all studies, average progression rates fell with increasing age, with most progression occurring between 20 and 30 years.

Conclusions: All three studies provide evidence of around −1 D myopia progression between the ages of 20 and 50 years. This has implications for intervention to slow progression during adulthood, as well as projections of visual impairment associated with myopia.

The state of knowledge about adult refractive progression until the 1980s has been previously summarized.^1,2 It was noted that, although most young adult myopes tended to show little change in refractive error, those who did tended to do so in the direction of greater myopia. During the past half century, there has been a dramatic increase in the prevalence in myopia in certain societies, accompanying economic development and higher educational standards.³ The resulting change in the visual environment sees increased time devoted to education, greater numbers of people involved in office work, and an overall reduction in time spent outdoors. Young adult groups undertaking intensive tertiary study are known to show increases in the prevalence and magnitude of myopia (see summary in Lee et al.⁴). Nonetheless, the implications of these changing societal visual habits on adult myopia progression have not been well documented. 

A commonly held belief is that progression of primary myopia stabilizes during teenage years. This is supported by the work of the Correction of Myopia Evaluation Trial (COMET) group, who found that 77% of myopia stabilized by 18 years of age, 90% by 21 years, and 96% by 24 years. Also, the Drentse Refractive Error and Myopia (DREAM) study reported median annual refractive changes among myopes of −0.09 D and −0.08 D at ages 16 to 18 and 19 to 21 years, respectively.^5,6 More recently, Foo and colleagues⁷ reported progression data on 424 myopes from the Singapore Cohort of Risk Factors for Myopia (SCORM) from adolescence (mean age, 14.6 years) to adulthood (mean age, 28.6 years). Because of the variable follow-up, myopia progression was normalized, and the mean progression rate was −0.04 ± 0.09 D/year. These synopses may not provide sufficient detail to expose meaningful long-term population shifts in refractive error. For example, in the study of Foo et al.,⁷ 18% of the patients showed at least −1.00 D/decade progression and 39% progressed by at least −0.50 D/decade.⁷ Assessment of adult myopic progression is complicated by modest rates of annual progression and a paucity of longitudinal studies on those 25 to 40 years of age, whereas refractive changes among younger populations as well as adult seniors are well described.^5,8 

A number of papers have considered myopia progression in young adults, and these are reviewed in a recent International Myopia Institute (IMI) report.⁹ These papers generally focus on specific cohorts such as clinic-, student-, or occupation-based populations; follow patients over relatively short periods; and often report mean annual progression rates that are considered to be clinically insignificant. The IMI report does not expound upon the magnitude of myopia progression from young- to mid-adulthood. Small changes over a period of, say, a year or two, which are considered to be below the threshold recognized as clinically important progression, may accumulate to significant refractive shifts over decades. Furthermore, cross-sectional studies cannot be used to accurately assess adult progression, as the data are potentially confounded by changing prevalence with birth year.¹⁰ 

Quantification of adult myopia progression is important in projecting future morbidity from myopia-related complications. Aside from age, the degree of myopia (and its associated axial length) is the biggest risk factor for these complications.^11–13 Currently, this relationship is derived from studies of older patients, commonly in their 50s, 60s, and 70s. Projections of the disease state based on the refractive error of younger patients may not be accurate if there is a significant long-term refractive shift caused by ongoing axial elongation. In the absence of a detailed, comprehensive examination of long-term refractive progression among myopes between 20 and 50 years of age, we have identified three papers that allow some insight into progression during this period of life. 

Methods

Vitale et al.¹⁴ compared the prevalence of varying degrees of myopia between cohorts from the US National Health and Nutrition Examination Survey (NHANES) conducted from 1971 to 1972 with those from 1999 to 2004. Although each of these evaluations was strictly cross-sectional in nature, there is a serendipitous correspondence between groups reported at the two times. In the 1971 to 1972 paper, data are presented for a subpopulation 18 to 24 years of age. The corresponding ages of this group from 1999 to 2004 were 45 to 57 years. In the analysis for the latter time period, data were presented for a subpopulation 45 to 54 years of age. Assuming minimal difference compared to the population 54 to 57 years of age, this overlap conveniently allows longitudinal refractive shifts to be assessed in this subgeneration, approximating the change from 20 to 50 years of age. The NHANES data are presented as prevalence of myopia using thresholds of −0.5, −2, and −8 D. Transforming prevalence to a logit scale linearizes the data against refractive error threshold.¹⁵ This allows the conversion of changes in prevalence data to estimates of refractive progression for given initial refractive errors. We used this method to estimate changes in refractive error from approximately 20 to 50 years of age for baseline refractive errors of −1, −3, and −6 D (Fig. 1). 

Goldblum et al.¹⁶ examined the electronic database of a large regional ophthalmic clinic in Germany, extracting refractive error records with at least 5 years of longitudinal data. We digitized a subgroup of data representing 6599 patients with initially myopic refractive errors who were 20 to 49 years of age from figure 2 of their paper, which plotted the proportion of patients showing increments of refractive change by age group (20−24 years, 25−29 years, 30−34 years, 35−39 years, and 40−49 years) and initial refractive group (−6, −4, −2, −1.5, −1, −0.5, and 0 D). By stitching together progression across age groups, we estimated aggregate progression for different baseline refractive errors (see Supplementary Material). The analysis was conducted as follows: 

• (i) The distributions of refractive error change for those with higher levels of initial myopia (worse than −4 D) were incompatible with a unimodal distribution of progression and the findings of previous studies.⁹ For example, figure 2 of Goldblum et al.¹⁶ shows nearly 30% of myopes with −6 D or worse demonstrating a hyperopic change in refractive error of greater than +3.25 D. In contrast, such hyperopic changes were observed in fewer than 2% at lower levels of initial myopia with a unimodal distribution of progression that was, as shown below, well fit by a Gaussian cumulative frequency distribution. Explanation of these large changes could be that some patients had undergone refractive surgery, despite the authors’ claim that such patients were excluded, or major cases of “overminussing” in the initial refractive error. Considering the high probability that these observations were anomalous, these data were excluded from the analysis.
• (ii) Cumulative refractive frequency distributions were calculated from the digitized data and Gaussian curves were fit to these data to enable estimation of the mean progression rate (see Fig. 2).
• (iii) A correction for the mismatch between mean follow-up time (8.8 years) and the 5-year age groupings of the patients was required. After accounting for decreased progression across time of 15.0% per annum,¹⁷ mean values arising from the Gaussian fits were reduced by 31.7% to approximate 5-year progression. Using the same principle for the 40- to 49-year-old group, we increased those mean values by 13.5%.
• (iv) Patients showing myopia progression might change refractive error groupings across time. Taking into consideration that the magnitude of progression was relatively small compared to the magnitude of the refractive groups and that neighboring groups showed similar progression rates for a given age, in addition to accounting for a patient changing between refractive error groups as their myopia progressed, was not feasible with the available data, so we calculated cumulative progression within each refractive error group as if all patients stayed within the same group as they aged from 20 to 50 years old. Further, our analysis assumes no cohort effect across time. In other words, younger patients at baseline are assumed to progress at the same rate as the older patients when they, too, reach older age.
• (v) Finally, we determined the mean values from the Gaussian curve fits adjusted for the follow-up time discrepancy described above and then added these values across age groups (with no further weighting) to estimate the progression between 20 and 50 years of age for each of the refractive error groups.

Figure 1.

 

Plot of data from table 3 of Vitale et al.¹⁴ showing the prevalence of myopia, linearized by a logit transformation, for the published refractive thresholds among the 18- to 24-year-old group from 1971 to 1972 and the 45- to 54-year-old group from 1999 to 2004. Refractive shifts (arrows) were estimated for those initially −1, −3, and −6 D from 1971 to 1972.

Figure 2.

 

Example of the Gaussian curve fits to the cumulative frequency distributions by change in refractive error for the 20- to 24-year-old group and at six different ranges of baseline refractive errors (shown in the top left of each panel) from Goldblum et al.,¹⁶ providing visual demonstration of the accuracy with which the mean (intersection of the curve with 50% cumulative frequency) is estimated.

Takeuchi et al.¹⁸ presented retrospective 5-year refractive changes in a large sample of Japanese patients attending an eye clinic between the years 2000 and 2012. We extracted data for a total of 163,574 patients between the ages of 20 and 49 with initial refractive error between +0.50 and −8.50 D from their figure 2. Mean 5-year progression rates were presented in the paper, obviating the need for calculation as per Goldblum et al.¹⁶ The exact re-examination times are not specifically stated in the paper, but we assumed that the follow-up was on average at 5 years. As with the Goldblum et al. data, we also assumed that patients stayed within their initial refractive grouping throughout the 30 years and that no cohort effect applied. Individual 5-year progression data were added together to get an estimate of progression between ages 20 and 50 years. 

Results

Figure 1 displays data from Vitale et al.¹⁴ for the 18- to 24-year-old group from 1971 to 1972 and for the corresponding 45- to 54-year-old group from 1999 to 2004. Prevalence values, transformed to a logit scale, from the two cohorts are plotted. Using best fit regression lines and determining change in refractive error rather than change in prevalence, the 30-year refractive shifts in this subgeneration were estimated at −1.1, −1.4, and −1.9 D for baseline refractive errors of −1, −3, and −6 D, respectively. 

Figure 2 plots, by way of example, the cumulative frequency of refractive change for the 20- to 24-year-old cohort by refractive error grouping from the paper of Goldblum et al.,¹⁶ illustrating the Gaussian curve fitting process. Mean 8.8-year progression was estimated from the value corresponding to 50% cumulative frequency. Table 1 presents adjusted 5-year progression by age and baseline refractive error for this cohort along with other cohorts and the total 30-year progression. Values for the 40- to 49-year-old cohort represent 10-year progression. Total progression estimates between the ages of 20 and 50 years ranged from −1.0 to −2.9 D for myopia ≤ −0.50 D. Although not shown, the standard deviations were between 0.4 and 0.9 D. The rate of progression evidently decreased with age, with most of the progression taking place during the third decade. The estimates are broadly consistent with those of Vitale et al.,¹⁴ including greater progression for higher baseline refractive errors. 

Table 1.

 

Mean Progression by Age and Refractive Error Grouping from Goldblum et al.,¹⁶ Adjusted to 5-Year Progression

 

Table 2 shows 5-year progression at different ages, along with total 30-year progression, calculated from the data of Takeuchi et al.¹⁸ Progression decreased with age, and the numerical value was consistently slightly higher in males than in females. Mean 30-year progression across all refractive groups, weighted by sample size, was −1.0 D for males and −0.9 D for females, with the majority occurring in the third decade. Although not shown, the standard deviations for each 5-year change were between 0.5 and 0.6 D, based on the authors’ Supplementary Table S2. True mean progression among the broader population may vary if weighted according to the refractive prevalence distribution, as this might be different from that in a clinical population. Such data are not available for Japanese in this age range, to our knowledge. Contrary to the observations from Vitale et al.¹⁴ and Goldblum et al.,¹⁶ progression was greater among those with a lower magnitude of myopia. Both males and females with high myopia tended to show modest mean progression (as low as −0.50 D) over the time period under consideration. 

Table 2.

 

Five-Year Mean Progression by Age and Refractive Error Grouping from Takeuchi et al.¹⁸

 

Discussion

This is the first analysis, to our knowledge, of cumulative myopic progression between the ages of 20 and 50 years. The major finding is the noteworthy similarity across studies with respect to the aggregate progression of about −1 D or greater between the ages of 20 and 50 years. We acknowledge that data from the papers considered here are of modest quality for the purpose of calculating progression during early- to mid-adulthood, and, individually, the papers should not be held as definitive. Our interpretation relies on the consistency of findings across three studies, which included very large sample sizes from three different continents. It is unclear whether there is a group of patients who show greater progression while others remain stable or whether adult progression is the product of a continuous distribution. The frequency plots of Goldblum et al.¹⁶ largely support the latter explanation. 

Our analysis of the NHANES data constitutes a novel methodology that can potentially be applied to other data to assess trends across time from cross-sectional data. We note that Vitale et al.¹⁴ described how the prevalence of myopia from 1999 to 2004 was consistent with an approximate −1 D shift in refractive error across the population from 1971 to 1972 to the population from 1999 to 2004.¹⁴ Our analysis here offers further insight. By concentrating on changes across time in one specific population—those who were 18 to 24 years old from 1971 to 1972 and 45 to 54 years old from 1999 to 2004—we demonstrated −1 D or more progression over this age range. Of course, such significant adult myopia progression may coexist with an underlying increase in overall myopia prevalence, but adult progression does not influence prevalence estimates for a criterion of −0.50 D. We recently explored the change in prevalence between 1971 and 1972 and from 1999 to 2004 for a criterion of −2 D and demonstrated that it is best explained by an increase of 9% in overall prevalence with an additional 14% due to adult myopia progression (Bullimore MA, et al. IOVS 2024;65:ARVO E-Abstract 139). 

Our interpretation of adult myopia progression is at odds with much of the literature, where the prevailing sentiment is that myopia stabilizes in the late teens or early 20s and therefore does not progress by substantial amounts during the adult years.^1,5,6 Exceptions to this stance clinically are considered to be outliers and comprised of students and practitioners of select vocations that require a heavy burden of near work. A more thorough examination of the literature, however, does not support this position of little adult myopia progression. The basis of the disconnect is an emphasis on the low proportion of adult myopes showing annualized progression above thresholds considered to be of clinical significance. Accumulation of subthreshold progression over many years can lead to the significant long-term progression shown in the three studies analyzed here. 

For example, Bullimore et al.⁹ reviewed the literature on adult myopia progression up to early 2022. They tabulated estimates of progression for myopic adults 18 to 25 years of age and from 25 to 40 years of age. The median annual progression rates for these groups were −0.14 and −0.06 D, respectively. Although median values may not be strictly applicable across the age range, a simple multiplication of annual progression by time would suggest a total of −1.88 D of progression between the ages of 18 and 40 years. It is noteworthy that none of the individual studies reported a mean annual progression above −0.25 D, although multiple studies have reported cumulative progression greater than this value, which is generally consistent with the tenet of our analysis.^4,19–26 More recently, Khan et al.²⁷ reported a retrospective analysis of myopia progression in a sample of racially diverse myopic adults. Mean annualized progression rates were −0.10, −0.08, and −0.04 D in their age groups of 18 to 21 years, >21 to 26 years, and >26 to 30 years, respectively. The estimated cumulative progression between 18 and 30 years thus approaches −1 D. In summary, close inspection of the many short-term studies of adult myopia progression reveals a pattern consistent with the analyses reported here. The oft-cited COMET study reported that 90% of those 21 years of age showed stabilization of myopia progression; however, their threshold for progression was −0.5 D, which is an overly liberal criterion in the context of our analysis.⁵ Furthermore, the stabilization was based on a double exponential function and thus dependent on the underlying assumption. Finally, it should be noted that age of stabilization could not be estimated in all participants. Using the COMET study criterion for progression, at least 50% of the myopic patients in Goldblum et al.¹⁶ progressed between 20 and 25 years. The threshold for discussing stabilization/progression should be lower than 0.25 D/year or, at least, should be over multiple years. A lower criterion must, of course, be placed in the context of the precision of measurement of refractive error. Measurement of axial length offers precision that is more suitable to establishing progression or stabilization of myopia in an individual.²⁸ 

Overall, the mean total progression was higher in the German study than in the Japanese study The influence of baseline refractive error on adult myopia progression remains unresolved by our analyses. Although there appeared to be greater progression among higher myopes in the NHANES¹⁴ and Goldblum et al.¹⁶ studies—indeed, around 2 D or more for high myopes (≤−6 D)—a strong trend in the opposite direction was apparent in the data of Takeuchi et al.¹⁸ One possible but unconfirmed explanation for this difference is race. In children, the influence of baseline myopia on the rate of progression is also equivocal.^29,30 Tables 1 and 2 include extracted data for emmetropic patients (−0.25 to +0.25 D and −0.50 to +0.50 D, respectively) to demonstrate that they also show, on average, myopic changes in refractive error, although the initial change in these groups represents myopia onset and not progression per se. 

Why does myopia continue to progress during adulthood? Although neither an exact mechanism nor an evolutionary advantage is known, the hypothesis is that the eye continues to elongate throughout life by an exponentially decreasing amount. In a previous analysis, we estimated that axial elongation reduced during the pediatric years by about 15% per year and that East Asians progressed by about 38% faster than their non-Asian counterparts.¹⁷ Extrapolation of that model through the adult years leads to estimates of 0.39 mm and 0.53 mm of axial elongation between the ages of 20 and 50 for non-Asians and for East Asians, respectively, with 80% of that occurring before the age of 30 years. Allowing for a ratio of refractive error to millimeter of −2.50, which is higher than the −2.04 observed in children,³¹ as compensated eye growth also decreases with age, one would project progression rates of −0.96 D and −1.33 D in non-Asians and East Asians, respectively, which are reasonably close to the values calculated from the three trials that we analyzed in this paper. 

The findings of our study have relevance to clinical practice. Anecdotally, clinicians often report myopia progression in adults, and our analysis substantiates this impression against the general doctrine that myopia stabilizes in late teenage years. This work also suggests that providing interventions to slow progression might be an appropriate strategy in young adults, as well as children, particularly those involved in studies or professions with a known risk of progression and in those with higher levels of myopia where the absolute risk of myopic maculopathy and uncorrectable visual impairment is high. Our observation may have implications with respect to estimating disease associations with myopia. Although prevalence rates of over 80% are noted among high-school seniors in East Asian countries, these rates have yet to emerge generationally among older adults.³ One may be tempted to project morbidity based on these late teenage estimates and the commonly held belief of stability of refractive error during adulthood. Adult myopic progression would render these estimates conservative, especially considering that progression is apparently linked to axial elongation. Further research is clearly needed to elucidate the role of adult progression in the prevalence of myopia-associated complications later in life. 

A known hypermetropic shift in adults around 50 to 60 years of age may reduce apparent levels of myopia, but this reduction is thought to arise from changes in the crystalline lens,³² not a reduction in axial length. This hyperopic shift appears to occur earlier in hyperopic eyes. Inspection of figure 2 in Goldblum et al.¹⁶ reveals that, in hyperopic patients between 35 and 39 years of age, the median refractive change was already in the hyperopic direction, whereas myopic patients of the same age still exhibited predominantly myopic refractive changes. Only above 50 years was the median refractive change in the hyperopic direction for myopic patients. Of course, refractive error in adults was determined without cycloplegia; thus, some of the hyperopic shift may have been due to previous latent hyperopia. 

The main limitations of our analyses are related to the sources of data. First, none of the participants in the three papers we analyzed was examined under cycloplegia, which is considered necessary to obtain a valid estimate of refractive error in younger patients.³³ Some studies have found more negative refractive error by about half a diopter in young adults without cycloplegia,^34–37 an effect that dissipates between the ages of 20 and 50 years. By this logic, an error introduced by a lack of cycloplegia would lead to our calculations underestimating the true degree of adult myopia progression. Likewise, in spite of presenting data derived from clinical populations, neither Goldblum et al.¹⁶ nor Takeuchi et al.¹⁸ reported axial length data or mentioned staphyloma or other myopia-related retinal changes. Second, two of the papers that we analyzed present data from clinical practices. It is possible that patients whose myopia progresses, or progresses faster, are more likely to return more frequently and thus may be overrepresented in the patient samples of Goldblum et al.¹⁶ and Takeuchi et al.¹⁸ This emphasizes the importance of the population-based NHANES data,¹⁴ which should be immune to such bias. Nonetheless, the NHANES data rely on a methodology for determining refractive error that is less than optimal, as acknowledged by Vitale et al.¹⁴. Third, our analysis of the study of Goldblum et al.,¹⁶ in particular, required extraction and manipulation of data, and the outcome depends on the validity of our assumptions. Fourth, none of the three studies measured axial length, a relatively more repeatable measure than refractive error. The recent IMI review of adult myopia progression concluded that adult refractive progression is due to ongoing axial elongation and discussed the potential for myopia control in adults.⁹ The magnitude of axial elongation reported in that review supports its role as the basis of the myopia progression in the three studies analyzed here. More recently, Nilagiri et al.³⁸ reported an annual elongation of 0.03 mm in myopes from 20 to 28 years. This would approximate to a 0.5 D myopic shift over this period, again consistent with the magnitude of progression we report here. 

In summary, we present data from three large studies from three different continents showing a general trend of myopia progression of about −1.00 D or greater between the ages of 20 and 50 years. This finding is contrary to the commonly held belief that myopia stabilizes in late teenage years or early adulthood and has implications for projections of the future burden of myopia.³⁹ 

Acknowledgments

Supported by Johnson & Johnson. 

This work was previously presented at the 18th International Myopia Conference, September 4–7, 2022, Rotterdam, the Netherlands. 

Disclosure: N.A. Brennan, Johnson & Johnson (E); X. Cheng, Johnson & Johnson (E); M.A. Bullimore, Alcon Research (C), Bruno Vision Care (C), CooperVision (C), Dopavision (C), EssilorLuxottica (C), Euclid Vision (C), Eyenovia (C), Genentech (C), Johnson & Johnson Vision (C), Novartis (C), Sydnexis (C), Vyluma (C), Ridgevue Publishing (O), Ridgevue Vision (O)