1. Common clinic scenarios
'Doctor, my daughter's axial length is 24.8 mm now — is she already high-risk? I read online that 26 mm means maculopathy.'
'Last time the clinic only checked refraction, not axial length — was that incomplete?'
'AL 25 mm but only −2.0 D — is that serious? Compared to another child with AL 25 mm and −5.0 D, who's at higher risk?'
With myopia-control awareness rising, axial length (AL) has become a routine clinic measurement. But parental anxiety about 'AL numbers' has followed — as if crossing a threshold guarantees blindness. The March 2026 systematic review and meta-analysis in American Journal of Ophthalmology (Clark RA, Wong RK) pooled 70 population-based studies and proposed an important observation: 'AL 26 mm' is not a universal high-risk threshold; for most children, refraction (SER) is actually a more practical monitoring metric.
This article translates the paper's key points into clinic language: how the study was done, what it found, what its limitations are, how it differs from current guidelines (e.g., IMI), and how to interpret your child's AL number honestly.
2. Bottom line: what you really need to know
30-second takeaways
- 'AL 26 mm' is not a universal red line — normal AL varies by ≥ 1 mm across age, sex, height, and ethnicity, which is the same order of magnitude as the proposed 'high-risk' cutpoint
- For retinal pathology: AL is a strong structural predictor (OR 3.85 per +1 mm, 95% CI 1.70–8.72); risk escalates steeply in the AL ≥ 26 mm long-eye tail
- For cataract and POAG: SER provides a clearer risk gradient — posterior subcapsular cataract (OR 4.58), POAG (OR 2.95) both rise with deeper myopia
- The '90% vs 10%' framing: about 90% of myopic eyes stay below AL 26 mm; only 10% enter the long-eye tail
- 'Long but emmetropic' healthy eyes exist: 1–3% of adults have AL ≥ 26 mm with normal refraction, mostly taller males — proving AL 26 mm doesn't necessarily mean disease
- The authors recommend: SER as the primary metric for most children; AL interpreted as a percentile-based trajectory (adjusted for age, height, sex, ethnicity), not a single absolute number
- Evidence strength: moderate (GRADE) — this is a 'new research perspective', not yet a rewritten guideline; IMI and others still emphasize AL
- Practical approach: ophthalmology clinics still measure both, with individualized interpretation by the physician
3. Common parent Q&As
4. Why this 'AL vs SER' debate?
① Myopia is a global public health issue
Holden et al. (Ophthalmology 2016) projected that by 2050, global myopia prevalence will approach 50% and high myopia ~10%. Myopia isn't just blurry vision — it elevates the risk of myopic maculopathy, retinal detachment, glaucoma, and cataract (Tideman et al. 2016, JAMA Ophthalmol; Clark & Wong 2026 refs 2, 12-14). Myopia control is fundamentally about reducing decades of future disease risk, not just lens thickness.
② Previously: AL was the preferred 'structural endpoint'
Over the past decade, myopia-control trials and guidelines (e.g., the IMI series) have progressively positioned AL as the core monitoring metric, mainly because:
- AL is a structural measurement — unaffected by accommodation; optical biometers (e.g., IOL Master) are highly precise
- 'AL 26 mm' has been linked with uncorrectable visual impairment and myopic maculopathy in multiple adult cohorts (e.g., Tideman 2016; Hashimoto 2019)
- Pediatric AL growth rate reflects myopia progression — a common efficacy endpoint in trials
③ The question Clark & Wong 2026 raises
The 2026 meta-analysis asks: For whom, for which disease, and how strongly does 'AL 26 mm' predict outcomes?
- Most myopic children never reach 26 mm — adult emmetropic mean is ~23.6 mm (SD 1.1 mm); 26 mm sits at +2 SD or beyond
- 'Normal AL' varies substantially — by ≥ 1 mm across age, sex, height, ethnicity
- The common diseases myopia actually causes (cataract, glaucoma) track with SER, not just AL
5. How was the study conducted?
① Search and inclusion criteria
- Databases: PubMed + Embase (English-only)
- Time range: Jan 1, 1990 – July 31, 2025
- Inclusion: population-based observational studies (cross-sectional or cohort), ≥ 200 eyes (or 200 participants if single-eye analysis), reporting AL and/or SER with demographics or disease outcomes
- Excluded: case reports, clinic-based series, non-human studies, reports lacking relevant AL/SER + outcome data
- Results: 2432 records → 1747 deduplicated → 76 full texts → 70 studies included
② Statistical methods
- PRISMA-compliant reporting; protocol registered on PROSPERO (CRD420251123893)
- Random-effects model (DerSimonian-Laird τ²) with Hartung-Knapp confidence intervals; pooling required k ≥ 3; otherwise narrative synthesis
- Risk of bias: Newcastle-Ottawa Scale; overall low to moderate
- Certainty: GRADE; moderate for prevalence, lower for incidence and narrative
- Software: R 4.5.1 (meta, metafor packages)
- Pre-specified but infeasible: diagnostic accuracy analyses (SER/AL thresholds vs visual acuity) and Bayesian network meta-analysis (SUCRA ranking) — dropped due to incompatible reporting
③ Three main analysis components
- Normative AL variation: how do age, sex, height, ethnicity shape 'normal' AL?
- AL–SER relationship: how tightly do they couple? Do 'long but emmetropic' eyes exist?
- Which predicts which disease better: separate analyses for retinal pathology, cataract, and POAG (glaucoma)
6. Finding 1： 'Normal AL' is not one number
The first observation of Clark & Wong (2026): 'normal AL' is actually a distribution modulated by demographic factors, not a fixed value. This is the foundation of the subsequent argument.
| Factor | Effect on AL (adult or child, as reported) | Pooled estimate (k = studies) |
|---|---|---|
| Age (children) | ~0.12 mm longer per year of age | +0.12 (95% CI 0.08–0.17), k=3, I²=97% |
| Age (adults) | Essentially stable (~0.014 mm/year) | +0.014 (0.006–0.021), k=4, I²=93% |
| Height (adults) | ~0.11 mm longer per +10 cm height | +0.11 (0.08–0.15), k=8, I²=85% |
| Sex (children, female–male) | Girls' eyes ~0.38 mm shorter than boys | −0.38 (−0.47 to −0.29), k=6, I²=98% |
| Ethnicity | E.g., Han children ~1.0 mm longer than Kyrgyz peers; East Asian eyes generally longer | Narrative (population-specific) |
* Data from Clark & Wong 2026 Table 1 and narrative; I² > 75% indicates high heterogeneity.
① Why this matters
The authors make a key point: height/sex/ethnicity create AL differences of 0.3–1.0 mm — the same order of magnitude as the 'AL 26 mm' threshold itself. A tall East Asian boy with AL 25.5 mm and a short girl with AL 25.5 mm occupy completely different 'positions in the normal distribution' — the first may still be within emmetropic norms; the second is a substantial deviation.
This grounds the authors' recommendation: 'interpret AL as a percentile, not an absolute threshold' — much like pediatric growth charts (height, weight) use age- and sex-adjusted percentiles, not single numbers.
② Pediatric AL growth rate
Pediatric AL elongates ~0.08–0.17 mm/year on average (cohort means 0.25–0.30 mm/year), corresponding to ~−0.3 D/year refractive shift. But the coupling is loose: 1 mm AL change ≈ 1 D SER change in most cohorts (the schematic-eye expectation is 2.7 D/mm, but 60% of cohorts fall in the 'weak-coupling' band |α| < 1.35 D/mm). So fast AL growth doesn't guarantee fast SER progression, and vice versa.
7. Finding 2： 'long but emmetropic' healthy individuals exist
This is the finding in Clark & Wong (2026) that most directly challenges 'AL 26 mm = red line'.
The authors examined multiple adult cohorts (European-ancestry Australian, Northern Irish, Beijing, Singaporean, etc.) and found: 1–3% of adults have AL ≥ 26 mm with refraction still in the emmetropic range (SER −0.50 to +0.50 D). In other words:
- Their eyes are longer (AL ≥ 26 mm)
- But corneal and/or lens power is correspondingly weaker — exactly compensating
- So vision is normal, refraction is normal, no myopia
- They are mostly taller males
💡 Why this matters
'AL ≥ 26 mm' does not necessarily mean 'myopic' or 'high-risk'. For these 'long but emmetropic' individuals, 26 mm is their emmetropic baseline — a natural extension of body proportions, not a disease signal. This is why the authors emphasize: interpret AL relative to the individual's emmetropic baseline AL — a shorter female may baseline near 23 mm, a taller East Asian male may baseline near 24.5 mm, and reaching 26 mm represents very different 'distance from baseline' for each.
The authors illustrate with a conceptual calculation: under different 'emmetropic baseline' assumptions, reaching AL 26 mm carries very different relative risks. With emmetropic baseline 23.2 mm (short-eye baseline), RR at 26 mm ≈ 44; with baseline 23.6 mm (pooled mean), RR ≈ 25; with baseline 24.6 mm (long-eye baseline), RR ≈ 7 (Clark & Wong 2026 Figure 4, conceptual). So a fixed absolute cutpoint over-flags some groups and under-flags others.
8. Finding 3： AL vs SER — which predicts which disease?
This is the paper's main result. The authors analyzed three disease families separately:
① Retinal pathology (any myopic retinopathy)
| Predictor | Pooled estimate (95% CI) | k / I² | Clinical interpretation |
|---|---|---|---|
| AL per +1 mm | OR 3.85 (1.70–8.72) | k=4, I²=69% | ~4-fold higher odds per +1 mm |
| AL ≥ 26 mm vs < 26 mm | OR 121.62 (2.31–6394.29) | k=3, I²=88% | Huge OR but very wide CI (sparse outcomes) |
| SER per −1 D (descriptive) | OR 1.5–1.8 | k<3（敘述性） | 50-80% higher odds per −1 D |
| SER high myopia vs emmetropia (descriptive) | OR 2–12（單一離群值 54） | k<3（敘述性） | High myopia clearly elevated; wide range |
* Data from Clark & Wong 2026 Table 2. 'Descriptive' means k < 3, no pooling possible; only trend summary.
5-year absolute incidence (Hisayama cohort)
The authors cite Ueda et al. 2020 (JAMA Ophthalmol) Hisayama 5-year incidence to translate relative risk into absolute terms:
| AL range | 5-year myopic maculopathy incidence |
|---|---|
| AL ≤ 24.9 mm | 0.4% |
| AL 25.0–25.9 mm | 1.8% |
| AL 26.0–26.9 mm | 6.4% |
| AL ≥ 27.0 mm | 12.0% |
The key takeaway: absolute risk is very low below AL 25 mm (0.4%) but rises steeply at ≥ 26 mm. So AL is a strong red-flag indicator for retinal pathology — but only when entering the long-eye tail.
② Cataract (by subtype)
The cataract analysis matters because it represents the common myopia-related disease most patients will face. Clark & Wong (2026) stratified cataract into three subtypes:
| Subtype | Predictor | Pooled (95% CI) | Clinical interpretation |
|---|---|---|---|
| Nuclear (most common age-related) | SER moderate/high vs emmetropia | OR 3.09 (1.85–5.16) HK CI 1.38–6.91, k=4 |
~3-fold prevalence |
| Nuclear | AL per +1 mm (descriptive) | < 1.0（k<3） | Slightly protective or null |
| Posterior subcapsular (PSC) | SER moderate/high vs emmetropia | OR 4.58 (2.18–9.64) HK CI 1.35–15.55, k=4 |
~4-5 fold prevalence |
| PSC | AL per +1 mm (descriptive) | 1.1–1.3（k<3） | Modest but imprecise |
| Cortical | SER moderate/high vs emmetropia | OR 1.07 (0.80–1.43) HK CI 0.67–1.71, k=4 |
Near-null |
* Data from Clark & Wong 2026 Table 3 (prevalence analyses; HK CI = Hartung-Knapp adjusted CI). Incidence analyses had fewer studies (k); most were borderline or null.
Key observation: for cataract, SER is the useful predictor — and only for nuclear and PSC subtypes; cortical is unrelated. AL's overall effect on cataract is modest, even slightly protective for nuclear (possibly because longer eyes accompany lens-shape changes that reduce nuclear opacity — a physical hypothesis with limited evidence).
③ Primary open-angle glaucoma (POAG)
| Predictor | Pooled (95% CI) | k / I² | Clinical interpretation |
|---|---|---|---|
| AL per +1 mm | OR 1.37 (1.29–1.46) | k=5, I²=0% | 37% higher odds per +1 mm |
| SER moderate/high myopia vs emmetropia | OR 2.95 (1.93–4.51) | k=7, I²=59% | ~3-fold higher risk |
| SER any myopia vs emmetropia | OR 1.62 (1.34–1.95) | k=6, I²=0% | Even mild myopia elevates risk |
* Data from Clark & Wong 2026 Table 4.
Key observation: AL is associated with POAG, but only 37% higher risk per +1 mm; whereas SER moderate/high myopia carries 2.95-fold POAG risk vs emmetropia. SER provides a more intuitive and stronger risk gradient. Clinically: even before AL reaches 26 mm, moderate/high myopia substantially raises POAG risk, requiring long-term IOP and visual-field monitoring.
9. The '90% vs 10%' framework: the authors' core recommendation
Combining the three findings, Clark & Wong (2026) propose a practical '90% vs 10%' framework:
💡 The authors' two-tier strategy
- For ~90% of myopic children: SER-first monitoring — AL typically stays below 26 mm; SER provides a more practical, demography-agnostic metric for cataract, POAG, and overall management
- For the ~10% entering the long-eye tail (AL ≥ 26 mm): AL becomes the focus — this is the retinal-pathology high-risk subgroup; needs more intensive fundus monitoring
- How to interpret AL: as a percentile (age, height, sex, ethnicity adjusted), not a single absolute cutpoint — analogous to pediatric growth charts
The core message: for the vast majority of families and children, refraction (SER) is the most direct, meaningful monitoring metric. AL measurement is not unimportant, but functions more like an 'advanced screening tool' — its value emerges when a child's AL trajectory clearly deviates toward the long-eye tail.
⚠️ Important: this is a 'new research perspective', not yet a guideline
The Clark & Wong (2026) recommendation is not yet in international guidelines. The IMI (International Myopia Institute) 2025 reports (Jones et al. on instrumentation; Tahhan et al. IMI 2025 Digest) continue to treat AL as a core metric. So:
- Clinics will continue to measure both and interpret jointly per individual
- Myopia-control trials (DIMS, low-dose atropine, ortho-K) still use AL as the primary efficacy outcome — 'rate of progression' comparison remains useful
- If you see your child's AL number, don't panic, but also don't dismiss the meaning of 'increasing diopter vs emmetropia'
- This 2026 meta-analysis is 'inviting clinical reconsideration', not 'immediate practice change'
10. Common myths
11. Advanced: for readers who want more depth
This section uses somewhat more technical language and suits medical students, residents, optometrists, and readers interested in methodology. General readers can stop at the previous section.
① Why does SER show subtype-specific cataract associations?
The three cataract subtypes (nuclear, cortical, PSC) respond very differently to myopia:
- Nuclear + PSC are particularly sensitive to myopia (OR 3.09 and 4.58); PSC most impacts reading vision clinically and presents at younger ages
- Cortical is essentially unrelated to myopia (OR 1.07) — likely because its main risk factors are sun/UV exposure, not refraction
- Mechanistic hypothesis: myopic vitreous biochemistry and lens capsular stress distribution may accelerate nuclear and PSC opacification; cortical region is more shaped by external UV exposure than internal eye structure
② Why Hartung-Knapp and random-effects matter
- Random-effects models assume each study has its own 'true effect' (unlike fixed-effects which assume one common value) — especially suitable for pooling across populations, ages, and designs
- Hartung-Knapp-Sidik-Jonkman adjustment refines random-effects CIs — when k is small, traditional DerSimonian-Laird CIs tend to be too narrow; HK gives more conservative CIs with better nominal coverage
- The study uses HK across all three disease analyses — methodologically more rigorous, which is why some CIs look wide (e.g., PSC OR 4.58, HK CI 1.35–15.55) — honest reflection of uncertainty
③ How GRADE certainty was assigned
- GRADE (Grading of Recommendations Assessment, Development and Evaluation) is an internationally recognized evidence framework: high / moderate / low / very low
- Observational studies start at 'low', but may be upgraded for large effect, dose-response, or low bias
- In Clark & Wong (2026), prevalence analyses earned 'moderate': due to large effects, cross-population consistency, and stability under leave-one-out
- Incidence and narrative analyses earned 'lower': due to k < 3, wider CIs, and fewer source studies
④ Why diagnostic accuracy and Bayesian network meta-analysis were dropped
The authors transparently report what was pre-specified but proved infeasible:
- Diagnostic accuracy (SER/AL thresholds vs visual acuity) — too few studies reported compatible VA + SER + AL data
- Bayesian network meta-analysis (SUCRA ranking) — because exposure definitions (different SER/AL thresholds) lacked common comparators
This is why the paper's conclusion relies on 'pairwise OR comparison + conceptual synthesis' rather than 'network-meta ranking'. The 'SER-first' recommendation is therefore based on OR magnitudes and clinical reasoning, not statistical ranking of metrics. Readers should keep this boundary in mind.
⑤ Contrast with IMI guidelines
IMI (International Myopia Institute) is the most influential international body in myopia management. Its 2025 instrumentation report (Jones et al., IOVS) and annual digest (Tahhan et al., IOVS) still place AL as a core tracking metric, emphasizing optical biometers. Clark & Wong's (2026) view is a 'complement-and-challenge' stance — not replacing AL, but proposing:
- SER-first for routine monitoring and long-term disease-risk counseling in most children
- AL-secondary for tracking control efficacy and identifying long-tail children
- Interpret AL by percentile, not by a single absolute cutpoint
Time and additional studies will be needed before this recommendation enters guidelines — expect 2-5 years of follow-up work in this space.
12. Summary: what you can do
If your child's most recent exam shows AL or SER changes: don't fixate on one number — look at the trajectory. How much change in the past 6-12 months? Is it reasonable for the child's age, height, and sex? Is the child's actual vision clear? These matter more than absolute numbers.
What to prepare for the ophthalmologist visit: (1) current SER and AL values; (2) rate of change over the past year (X D progression; Y mm AL elongation); (3) parental and sibling myopia history; (4) outdoor time, near-work hours, screen time; (5) any current or past myopia-control strategies (low-dose atropine, ortho-K, DIMS lenses).
What cannot be replaced: regardless of metric choice, ≥ 2 hours of outdoor activity daily, limited near-work, correct posture and lighting, regular follow-up, adequate sleep — these are universally recommended foundations of myopia prevention and control, unchanged by this new paper.
What not to do alone: don't self-interpret risk from online AL thresholds; don't stop myopia-control treatment on your own; don't fixate on absolute numbers while ignoring the child's overall situation; don't dismiss one metric's value because of another's. Every child is individual — final interpretation belongs in the ophthalmologist's office.
One final note: Clark & Wong (2026) offer an important reminder — AL 26 mm is not an absolute red line; 'normal AL' varies by individual; for most children, SER is a more practical monitoring metric. But this is a 'new perspective', not a 'new standard'; the ophthalmology field needs more time to build consensus. Hopefully this digest enables a deeper, more evidence-grounded conversation at your child's next visit.
📚 HsiaoEye Pediatric Myopia Series
- 8 Pediatric Myopia Control Myths — Atropine, OK lens, control lenses, outdoor activity
- Are DIMS Lenses Effective? — 2026 meta-analysis, 0.37 D slowing/12 mo, optimal age
- Monitoring: SER or Axial Length? — 2026 meta-analysis, 90/10 framework, disease-specific risk (you are here)
References
- [本文主要來源] Clark RA, Wong RK. Spherical Equivalent Refraction Versus Axial Length for Monitoring Childhood Myopia and Estimating Disease Risk: A Systematic Review and Meta-Analysis. Am J Ophthalmol. 2026;286:235-247. doi:10.1016/j.ajo.2026.03.007
- [全球近視流行病學] Holden BA, Fricke TR, Wilson DA, et al. Global prevalence of myopia and high myopia and temporal trends from 2000 through 2050. Ophthalmology. 2016;123(5):1036-1042. doi:10.1016/j.ophtha.2016.01.006
- [眼軸與不可矯正視覺障礙] Tideman JWL, Snabel MC, Tedja MS, et al. Association of axial length with risk of uncorrectable visual impairment for Europeans with myopia. JAMA Ophthalmol. 2016;134(12):1355-1363. doi:10.1001/jamaophthalmol.2016.4009
- [Hisayama AL 與黃斑部病變] Hashimoto S, Yasuda M, Fujiwara K, et al. Association between axial length and myopic maculopathy: the Hisayama Study. Ophthalmol Retina. 2019;3(10):867-873. doi:10.1016/j.oret.2019.04.023
- [Hisayama 5 年發生率] Ueda E, Yasuda M, Fujiwara K, et al. Five-year incidence of myopic maculopathy in a general Japanese population: the Hisayama Study. JAMA Ophthalmol. 2020;138(8):887-893. doi:10.1001/jamaophthalmol.2020.2211
- [「每個 diopter 都重要」] Bullimore MA, Brennan NA. Myopia control: why each diopter matters. Optom Vis Sci. 2019;96(6):463-465. doi:10.1097/OPX.0000000000001367
- [IMI 近視定義與分類] Flitcroft DI, He M, Jonas JB, et al. IMI–Defining and classifying myopia: a proposed set of standards for clinical and epidemiologic studies. Invest Ophthalmol Vis Sci. 2019;60(3):M20-M30. doi:10.1167/iovs.18-25957
- [IMI 2025 儀器報告] Jones D, Chow A, Fadel D, et al. IMI—Instrumentation for myopia management. Invest Ophthalmol Vis Sci. 2025;66(9):7. doi:10.1167/iovs.66.9.7
- [IMI 2025 Digest] Tahhan N, Bullimore MA, He X, et al. IMI—2025 Digest. Invest Ophthalmol Vis Sci. 2025;66(12):27. doi:10.1167/iovs.66.12.27
- [近視控制療效綜述] Brennan NA, Toubouti YM, Cheng X, Bullimore MA. Efficacy in myopia control. Prog Retin Eye Res. 2021;83:100923. doi:10.1016/j.preteyeres.2020.100923
- [近視臨床試驗儀器報告] Wolffsohn JS, Kollbaum PS, Berntsen DA, et al. IMI–Clinical myopia control trials and instrumentation report. Invest Ophthalmol Vis Sci. 2019;60(3):M132-M160. doi:10.1167/iovs.18-25955
- [新加坡馬來族眼科研究] Singapore Malay Eye Study (SiMES). 多篇收錄於 Clark & Wong 2026 文獻 41, 43。
- [新加坡印度族眼科研究] Singapore Indian Eye Study (SINDI). 多篇收錄於 Clark & Wong 2026 文獻 59, 63。
- [姊妹文：DIMS 兒童近視控制] 本站延伸閱讀：兒童近視控制鏡片（DIMS）有效嗎？2026 新整合分析重點解析