Why one number isn't enough to say keratoconus is getting worse

"Has my keratoconus got worse?" sounds like a yes-or-no question. In practice it is one of the trickier calls in eye care, because the answer often decides whether someone is referred to discuss corneal cross-linking. For years, many clinics have leaned on a single number, the steepest point on the corneal map (Kmax). A 2025 review looked at how well different criteria actually pick up true progression, and the results suggest one number on its own may not be enough.

Please note: this article is based on the published abstract of the review. We were not able to access the full paper, so we have used only the figures reported in the abstract.

What the researchers did

The team, from the University of Coimbra in Portugal, carried out a systematic review of diagnostic test accuracy. That means they gathered studies that tested how well a particular measurement or rule identifies keratoconus that is genuinely progressing.

• Databases searched: MEDLINE/PubMed, EMBASE, Web of Science and the Cochrane CENTRAL register, from inception to 31 December 2024.
• Methods: the review followed PRISMA-DTA reporting guidelines, the protocol was registered in advance (PROSPERO CRD42023391880), and risk of bias was assessed with the QUADAS-2 tool.
• Studies included: 15 studies (2 prospective and 13 retrospective), covering 3,547 eyes from 2,654 patients.
• Comparator: each test was judged against the way the original studies defined true progression. The abstract does not describe those reference standards in detail, or whether assessors were masked.

Sensitivity and specificity, in plain words

Two terms come up again and again in this kind of research:

• Sensitivity is how good a test is at catching eyes that really are progressing. A sensitivity of 0.80 means it correctly flags about 80 out of every 100 progressing eyes, and misses the other 20.
• Specificity is how good a test is at correctly reassuring eyes that are stable. A specificity of 0.95 means about 95 out of every 100 stable eyes are correctly called stable, while 5 are wrongly flagged.

An ideal test scores high on both. A test with high specificity but low sensitivity rarely raises a false alarm, but it can miss real change. That trade-off is the heart of this review.

What they found

Composite and whole-surface measures performed best

The best-performing criteria were:

• BAD-D (Belin/Ambrósio Enhanced Ectasia Display), a composite score: maximum mean sensitivity 0.82, specificity 0.98.
• ABCD criteria (using the >95 confidence interval threshold), which look at front and back curvature, thinnest thickness and vision together: sensitivity 0.81, specificity 1.00.
• Anterior best-fit sphere radius, a measure of the overall shape of the front corneal surface: sensitivity 0.86, specificity 0.94.

These are "maximum mean" values, meaning the highest values reported across the included studies, rather than a single pooled estimate.

Kmax was good at avoiding false alarms, but missed change

In two studies, Kmax had high specificity (0.91 and 1.00) but low sensitivity (0.70 and 0.49). Put simply, when Kmax said an eye had progressed, it was usually right. But it missed a meaningful share of eyes that were in fact changing. In one study, it caught fewer than half.

Image-based criteria were weaker

Deep learning applied to colour-coded maps showed very low specificity (below 0.60), meaning many stable eyes would be wrongly flagged. Anterior segment optical coherence tomography lacked sensitivity (below 0.68).

The authors conclude that composite indices such as BAD-D and ABCD, or global measurements like best-fit sphere radius, might be better ways to define progression than single-point measurements.

What this means for you

If you are being monitored for keratoconus, this research supports what a careful keratoconus optometrist already does: looks at the whole cornea, front and back, rather than one headline number. A small rise in Kmax can be measurement noise, and a stable Kmax does not always mean a stable cornea.

Repeatability matters too. We explain why repeat and averaged scans help in our article on repeat, averaged corneal scans for tracking progression, so we won't go over that again here. Other layers of the cornea can add information as well, which we discuss in our piece on the corneal epithelium map.

The pathway stays the same. A keratoconus-focused optometrist diagnoses the condition, documents any change with sequential corneal tomography, and refers you to an ophthalmologist when progression is confirmed, because cross-linking is performed by an ophthalmologist. Afterwards, your optometrist can help with vision, including specialty and scleral lenses. At Rose Optometry in Hamilton, our optometrists, including Jagrut Lallu and Jacqueline Rowe, provide this keratoconus care.

What the studies can't tell us

• Abstract only. We have not seen the full paper, including the risk-of-bias results or exactly how each study defined true progression.
• Mostly retrospective. 13 of the 15 studies looked back at existing records, which can introduce bias in who was included and how they were measured.
• No single pooled figure. The numbers are best-case values from individual studies, not a combined estimate, so they may flatter the tests.
• Small evidence for Kmax. The Kmax figures come from only two studies.
• The reference-standard problem. There is no perfect gold standard for progression, so every test is being judged against an imperfect yardstick.
• Device differences. Results may not transfer directly between different tomography machines.

The take-home message

Deciding whether keratoconus is getting worse is more reliable when it involves looking at several measurements together, over time, on consistent equipment. Composite scores and whole-surface measures appear more balanced than Kmax alone, though the evidence is still limited. If you would like a thorough review of your corneal maps, you can book a keratoconus assessment with our optometrists in Hamilton.

References

• Gil P, d'Almeida OC, Cunha B, et al. Criteria for Keratoconus Progression: A Systematic Review of Diagnostic Test Accuracy. Cornea. 2026;45(3):387-395. PMID 41247279. https://doi.org/10.1097/ICO.0000000000004020