Case registry:  what it is

This registry documents people who chose to report their own myopia-reduction results, most of them working from the free method on their own rather than through the guided program. It is only a small fraction of everyone who has applied the method, and it deliberately leaves out the much larger BackTo20/20 program-participant dataset, which is held separately. What it captures is a conservative floor: the kind of results people reach largely self-directed. Each case carries start and end diopter numbers as stated by the person, the time window, and a verification class describing how strong the supporting evidence is. Every case links back to its original report, and the full dataset is downloadable below.

It is not a clinical trial and not a prospective cohort. Cases are self-selected success reports: members who improved and chose to write in. Dropout and non-response rates cannot be derived from this collection, and no claim of a population success rate is made from it. A cohort-level dataset with denominators, drawn from program member data, is planned as a separate release.

How this registry was built

EndMyopia started in 2008 and has spread almost entirely by word of mouth, well beyond what its own channels can count. Those channels alone include a private community of tens of thousands, a comparably sized email following, a daily blog readership built up over more than a decade, and a YouTube channel. A portion of the people who apply the method report their measured results back, and this registry is the documented part of those outcomes, spanning 2013 to 2026. As far as we can establish, no comparable decade-long, self-directed adult myopia-reduction record exists anywhere else.

The reports are collected the slow way. Participants receive periodic check-in emails asking how their eyesight is progressing, and their replies, together with posts to the EndMyopia forum, the private Facebook group, and blog and video comments, are the source material. This has run continuously for years, and several hundred further responses remain in the email archive, not yet entered.

A case is one person with a documented starting and current refraction, or a documented move to no correction. Every case links back to its original, dated public report, whether an email, a forum post, a Facebook post, or a video, so any number here can be traced to where it came from.

Cases are graded by strength of evidence, from self-measured up to independent confirmation (see the verification classes below). Of the 363 documented cases, 61 carry a third-party or official-record signal: an optometrist or ophthalmologist prescription, a passed DMV or driving-licence vision test, or a removed corrective-lens restriction on a licence.

The public dataset is de-identified, with the name column removed while source-post links are kept so the data stays auditable. The named master file and the underlying email correspondence are held privately and are not published. Corrections or removal requests go to [email protected].

The honest limits sit alongside the scale. This is a self-reported registry, subject to selection and survivorship bias, since people who improve are more likely to write back. It does not prove that the method reliably reverses established adult myopia. What it does provide, in the open and over more than a decade, is a large and unusually well-sourced body of individual reductions, a real share of them independently confirmed.

Verification classes

Each case is assigned the strongest class its published text actually supports — assigned conservatively, with the supporting quote preserved in the dataset.

How the numbers are computed

• People, not posts. 55 members have two or more progress posts; each series is merged into one case using the earliest starting numbers and the latest reported numbers.
• Diopter values are the worse eye’s spherical value as stated by the member; cylinder is excluded from deltas.
• Annualized rates are published only for cases with a 12-month-or-longer window (n=57, median 1.5 D/yr). Annualizing a two-month report overstates a sustained rate, so shorter cases show only the raw change and its window.
• Exclusions (16 cases): any case involving prior laser surgery, child cases (this registry is adult-only in v1), and cases whose numbers are internally inconsistent pending re-review. Exclusions are listed in the dataset with reasons, not silently removed.

Verified-signal cases (classes 1–2)

The 42 cases below carry the strongest verification available in published reports. Numbers appear exactly as the member stated them.

Download the dataset

The complete registry — all 363 cases including classes 3–5, exclusions with reasons, verification quotes, and source links — is available for download and independent recomputation:

• em-case-registry-v2.0.csv — flat table, one row per case (73 KB)
• em-case-registry-v2.0.json — full records including verification quotes and per-case source lists (198 KB)
• em-case-registry-v2.0.zip — both files (39 KB)

Dataset version 2.0 · generated 2026-07-09 · CC BY 4.0 with attribution to endmyopia.org. Corrections welcome: if any case is misclassified, the registry is revised and the change is logged here.

Version history

• v1.0 (2026-06-10) — initial release from the 2013–2026 published-report archive: 466 reports, 289 cases, verification-classed and person-merged.
• v2.0 (2026-07-09): added 74 self-reported cases documented since v1.0 (three child cases excluded from the adult set, two duplicates merged). 363 cases, 344 in the adult analysis set.

Citable deposit & author record

The registry is deposited as a permanent, versioned, citable research object and described in an accompanying preprint. Author: Jake Steiner, Research Scientist, Steiner Research (ORCID 0009-0004-6064-135X · Wikidata).

• Dataset: 344 adult cases (v2.0), CC BY 4.0. DOI 10.5281/zenodo.21016339. Mirrored on GitHub and Hugging Face.
• Preprint: DOI 10.5281/zenodo.21230819 (Wikidata).