What 9 Million Eyeglass Prescriptions Reveal About Myopia and Astigmatism in Japan

A pair of glasses is made for one person. A database of millions of prescription records can reveal something larger: the contour of vision across a lifetime. It can show when prescriptions shift fastest, when they flatten, which forms of astigmatism dominate at different ages, and how often two eyes diverge. It can also tempt readers to mistake precision for representativeness.

JINS Inc. and researchers at the University of Osaka Graduate School of Medicine have published two peer-reviewed studies in Investigative Ophthalmology & Visual Science, the journal of the Association for Research in Vision and Ophthalmology. Ryo Kawasaki, professor of public health, led the refractive-error paper. Endowed associate professor Shizuka Koh led the astigmatism paper. Kohji Nishida of ophthalmology and JINS Medical Strategy Department researchers Hidehito Matsuoka and Kiyotaka Hori were co-authors.

Two Studies, Not One Giant Cohort

The refractive-error paper, “Distribution and Time Trends of Refractive Errors in Japanese Spectacle Wearers,” analyzed 4,525,689 people aged 6 to 35. Women accounted for 53.0 percent and men 47.0 percent. Researchers calculated spherical equivalent—the spherical lens power plus half the cylindrical power—and described its distribution by age, sex and geographic region. For customers under 15, prescriptions issued by ophthalmologists were used. Repeat purchasers supplied the observations used to estimate annual refractive change.

The astigmatism paper, “Epidemiology of Astigmatism in Japan: Analysis of More Than 9,000,000 Spectacle Prescriptions,” was a retrospective cross-sectional analysis of 9,204,993 prescription records from people aged 6 to 89. It examined cylinder power, axis orientation, age, sex, region and the difference between the two eyes. Because the two studies use different age ranges, units and analytical designs, their headline counts cannot honestly be added into a “13.7 million-person study.”

What “Progressing to 30” Actually Means

The median spherical equivalent moved in the myopic direction with age and stabilized at about 30 in men and 29 in women. At those points the median values were −3.3 diopters and −3.5 diopters, respectively. The fastest annual change occurred at age seven, with a median of −1.1 D per year. Progression decelerated between ages nine and 19, but the aggregate pattern did not become flat immediately upon adulthood.

That challenges the simple clinical story that juvenile myopia reliably settles by 18 or 21. It does not, however, discover adult progression from nothing. A 2023 International Myopia Institute review concluded that onset and progression are common between 18 and 25, particularly in intensive academic settings, and that the likelihood generally declines with age. The new paper’s contribution is the unusually high-resolution picture among Japanese spectacle purchasers.

The same analysis found high myopia, defined as spherical equivalent of −6.0 D or worse, in 1.3 percent of children under ten. The strongest median rate at age seven underscores why early refractive change matters. But the database does not identify why one child progressed, select a therapy, or test whether any intervention slowed eye growth.

Astigmatism Changes Direction as Well as Strength

Astigmatism means the eye has different refractive power along different meridians, so incoming light does not converge at a single point. A spectacle prescription expresses the amount as cylinder power, measured in diopters, and its orientation as an axis in degrees. That makes astigmatism more than a stronger-or-weaker scale: direction matters.

The study found mild and moderate astigmatism increasing from age ten, reaching roughly one person in five at age 20. Both the proportion of people with astigmatism and cylinder power rose sharply after age 50. The age distribution also shifted from with-the-rule astigmatism, more common in younger groups, toward against-the-rule astigmatism in older groups; by age 70, the latter exceeded half.

That is a population pattern, not proof that every person’s axis rotates through the same sequence. The distinction matters: comparing different age groups at one period combines biological aging with differences between generations.

Among roughly 5.06 million people with at least 0.25 D of astigmatism in one eye, 9.4 percent had a cylinder-power difference of at least 1.00 D between eyes. Most eyes were broadly symmetrical, but about one in ten astigmatic customers showed a difference too large to erase by assuming the same correction on both sides.

Why Retail Records Can Become Science

A conventional population study recruits participants from selected communities, performs standardized examinations and can measure axial length, corneal curvature, health and behavior. That control makes the biological meaning clearer. It is also expensive and slow, often limiting a study to thousands or tens of thousands of people.

Retail records reverse the bargain. The measurements already exist as part of everyday care and commerce. They cover all 47 prefectures through 434 shops and contain enough observations to draw smooth age curves and find uncommon bilateral patterns. Instead of collecting a new cohort, researchers repurpose de-identified operational records as real-world data.

Scale reduces random noise. It does not automatically remove systematic error. A biased sample of nine million may describe its selected population with extraordinary precision while still missing the population outside the store. The central question is never only “How many records?” It is “Who entered the database, and who did not?”

Japan Has Counted Vision Since 1948—But Differently

Japan’s School Health Statistics have published a long national series since fiscal 1948. In the fiscal 2025 survey, the share with uncorrected visual acuity below 1.0 exceeded 30 percent in elementary school, was around 60 percent in junior high and around 70 percent in high school. Those figures have made deteriorating student vision a persistent education and public-health concern.

Uncorrected visual acuity below 1.0 is not synonymous with myopia. Acuity measures the smallest detail a person can resolve; refraction measures the optical focusing error in diopters. Reduced acuity can have causes other than myopia, and people with the same acuity can require different prescriptions. The education statistics and the JINS results illuminate different things and should not be combined as though they were one prevalence series.

What 9 Million Records Still Cannot See

The age-30 inference combines the cross-sectional distribution across ages with annual changes estimated among repeat customers. Yet a return to an optical shop is not a scheduled research visit. Replacement may be driven by broken frames, fashion, affordability, relocation or a perceived change in sight. A new prescription does not isolate axial elongation from every other component of refraction.

The astigmatism records powerfully describe prescribed cylinder and axis, but they do not directly show whether age-related change arose in the cornea, crystalline lens or both. Nor can an age-group distribution forecast the future axis of one customer. These studies are maps, not examinations.

Industry–University Research Requires Two Kinds of Scrutiny

JINS supported the research, and Matsuoka and Hori are company employees. The refractive-error paper discloses JINS funding for Kawasaki and a relationship between Koh and JINS; Nishida reported none. The astigmatism paper also identifies JINS support and the employee authors. These disclosures do not invalidate the results. They are essential context for judging design, analysis and publication.

The collaboration also illustrates a legitimate public value: a retailer can hold a geographically broad and comparatively consistent dataset that no university could cheaply recreate. Academic epidemiology and ophthalmology can turn it into testable evidence. The responsible response is neither to dismiss company data nor to treat peer review as a guarantee. Readers need the methods, exclusions, funding, author relationships and independent replication together.

JINS and the university describe the use of eyeglass-sales big data for clinical research as a Japanese first, but the release marks that statement as the company’s own research. Japan.co.jp could not independently establish the completeness of that claim and does not present it as settled fact.

The Next Step Is Better Linkage, Not a Bigger Headline

To move from description toward prevention, future work will need to connect retail prescriptions with standardized ophthalmic examinations, axial length, corneal measurements and relevant environmental data. Longer follow-up could separate aging within one person from differences between birth cohorts. Replication in other optical chains and population-based cohorts would test whether the same curves survive outside JINS.

The immediate clinical message is restrained but useful: do not assume a young adult’s prescription has frozen; identify rapid childhood change early; and in later life examine cylinder strength, axis and inter-eye difference rather than looking only at spherical power. The studies do not prescribe an examination interval or establish a particular myopia-control treatment. Those decisions require individual assessment by an eye-care professional.

What sits in a retail database looks at first like commerce: an age, a frame and a set of lens numbers. With careful de-identification, explicit limits and clinical methods, those transactions can become an observatory for population vision. The enduring result of these two papers is not a single birthday. It is proof that routine prescription records can give Japanese vision epidemiology a new—and unusually detailed—instrument.