Bayesian Causal Atlas · Vol. Ophthalmology · Pearl Structural Causal Model

Myopia & Refractive Error — Structural Causal Analysis

Uncorrected refractive error affects roughly 1.1 billion people and is the leading cause of visual impairment worldwide; myopia is projected to reach ~4.8 billion people by 2050. This oracle separates two problems on two causal paths: immediate correction (spectacles, surgery) of the vast uncorrected burden, and childhood progression control (atropine, orthokeratology, defocus optics, red-light, outdoor time) to prevent future high / pathologic myopia. Effect sizes are from named trials. For education, not individual medical advice.

Method. Structural Causal Model (SCM) with backdoor adjustment (Pearl). Interventions are not assumed independent: the optical/pharmacologic control arms share the axial-elongation pathway, so their overlap is removed by an eigenvalue-corrected equicorrelation model at an adjustable mean cross-correlation ρ̄ (default 0.30). Correction (spectacles / surgery) is separated as a direct acuity-restoring path. Robustness to unmeasured confounding is quantified with the E-value. PN / PS / PNS under monotonicity. Every relative risk is cited — no effect size is invented. The front door is resolved through an EXPLICIT mediator cascade (refractive → environmental → pharmacologic → optical → disease state), not one lumped node: each intervention acts on a specific node, so same-node interventions are substitutes that saturate against each other, while different-node interventions are d-separated given the intermediate node and compose in series. The cross-correlation removal thus follows from the graph structure; the residual ρ̄ cleans up only the mediator-independent (direct) effects.
ρ̄ = 0.30
A–D (all)
A high (RCT/meta) · B cohort · C case-series/modelled · D consensus/provisional. Lower-grade interventions are excluded from the DAG, front-door pooling, Pareto, Monte‑Carlo & sensitivity.

Interventions

Tick the interventions to combine. Each shows its trial effect estimate, 95% confidence interval (CI), E-value, mechanism, and citation. ★ = in the current Pareto effective set but not yet ticked.

Combined causal estimate

Headline is the front-door estimate: shared axial overlap removed via dose-response saturation; residual direct-effect overlap removed via the eigenvalue model at ρ̄.

1.00
Combined RR
0%
Relative risk ↓
Pooled E-value
Interventions selected (k)
0
Effective independent dimensions (n_eff)
0
Redundancy discount applied
0%
Baseline risk (illustrative anchor)
Absolute risk after intervention
Absolute risk difference (RD)
Backdoor-only RR (no front-door)
Axial-control overlap removed
Number needed to treat (NNT)
ρ-sensitivity band (ρ 0 → 0.6)
Interpretation

Causal attribution

Under monotonicity + exogeneity (E-value bounds the exogeneity assumption).

Probability of Necessity (PN)
Probability of Sufficiency (PS)
Prob. of Necessity & Sufficiency (PNS, lower bound)
Causal DAG
Cross-correlation
Pareto (threshold)
Monte Carlo
Front-door mediation
What-if / If-not-for
Sensitivity
Antithesis

Faithful causal directed acyclic graph (DAG). Control therapies act through the shared mediator axial elongation / refractive-error progression, which drives high / pathologic myopia and thence visual impairment (Y). Correction (spectacles, surgery) acts directly on current acuity, bypassing progression. Named confounders — onset age, parental myopia, near-work — open back-door paths (adjusted). Mediator cascade: interventions attach to the node they act on (refractive → environmental → pharmacologic → optical), which converge on the disease state and thence the endpoint — drawing the intermediate mediators explicitly is what exposes d-separation and per-channel saturation.

Confounders U:age of onset · parental myopia / genetics · near-work · ethnicity · baseline refractive error · access to care → back-door paths (adjusted)Spectacle / contactcorrectionRefractive surgery(LASIK / SMILE)Increased outdoortimeLow-dose atropine0.05%OrthokeratologyDIMS defocusspectaclesDefocus softcontacts (MiSight)Red-light therapy(RLRL)Atropine +orthokeratologyRefractivecorrectionEnvironmental(outdoor time)Pharmacologic(atropine)Opticalmyopia controlAxialelongationHigh-myopiacomplicationsFront-door: through slowing axial elongationDirect correction (restores current acuity)Back-door confounding (adjusted)

Eigenvalue diagnostics for the selected interventions under an equicorrelation matrix (off-diagonal ρ̄). A large λmax relative to k signals redundancy; n_eff is the effective number of independent interventions actually contributing.

k (selected)
0
λmax
λmin
n_eff = (Σλ)² / Σλ²
Condition number

On mechanistic grounds ρ̄ ≈ 0.30 is defensible: the optical control arms (ortho-k, DIMS, MiSight) all exploit peripheral myopic defocus, so combining two of them is largely redundant. ρ̄ is user-adjustable because atropine (pharmacologic) and red-light act through partly distinct mechanisms, which is why atropine+ortho-k is additive. Most of the overlap is now handled structurally by the mediator nodes (same-node substitutes saturate); ρ̄ governs only the residual correlation among direct effects.

Minimum-effective-set analysis. Set a target combined risk reduction; the model finds the smallest set of interventions — accounting for front-door mediator overlap — that reaches it, and highlights them. If the target exceeds what all interventions together can achieve, the full set is shown (never an empty one). "Apply" ticks exactly that set.

Target combined risk ↓ ≥ 50%

Monte Carlo propagation. Each selected intervention's log-effect is sampled from a normal distribution implied by its 95% CI; samples are combined with the same eigenvalue overlap discount. 5,000 draws.

Median combined RR
95% simulation interval
Standard deviation of combined RR
P(combined RR < 0.90)

Antithesis — challenging this oracle's own conclusions

Most of the global burden is simply lack of glasses. ~1.1 billion people have uncorrected refractive error — the dominant, trivially correctable cause of refractive visual impairment. Fancy myopia-control technology addresses future high myopia in children, but the largest, cheapest win is distributing spectacles. Ranking controls above correction inverts the public-health priority.
Axial elongation is a surrogate. The control arms are validated on millimetres of axial length and dioptres of progression, not on lifetime visual impairment or pathologic-myopia blindness. The link from “slower elongation” to “less blindness decades later” is inferred, and rebound after stopping erodes it.
Correction does not stop progression — and control does not restore today’s vision. These are two different problems on two different paths (gold vs teal). Spectacles fix current blur but not future pathology; atropine slows the eye but a child still needs correction now. Treating them as interchangeable mis-specifies the graph.
Red-light therapy is powerful but unproven long-term. RLRL shows the strongest short-term axial control, but its retinal safety over years and its rebound profile are not established. A large short-term effect on a surrogate is not the same as durable, safe vision preservation.
Every control arm carries its own harm. Atropine causes photophobia and rebound; orthokeratology and soft lenses carry microbial-keratitis risk; combinations stack these. The progression relative risks do not net out these competing harms.

What-if — the do-operator: P(Y | do(S))

Intervening on the selected set S with Pearl's do-operator (setting the interventions, not merely observing them). Contrast against do(∅) = no intervention.

P(outcome | do(∅)) — baseline
P(outcome | do(S)) — intervened
Absolute risk reduction (ARR)
Number needed to treat (NNT)

If-not-for — but-for counterfactual (leave-one-out)

For each intervention: "if not for this one, the combined front-door effect would be…". Isolates each intervention's marginal causal contribution after mediator-overlap removal, so shared-pathway agents are not double-credited.

If not for…RR without itRR with full setmarginal RRR lost

One-at-a-time sensitivity. Each intervention's effect is swung across its 95% confidence interval (others held at point estimate); the bar is the resulting swing in the combined front-door effect. A long bar means the combined estimate leans heavily on that single trial's precision.

Optimistic bound (all at CI-low)
Point estimate
Pessimistic bound (all at CI-high)
Pooled E-value (confounding robustness)

Front-door (mediation) decomposition. Control therapies act through one shared mediator — axial elongation. Each log-effect is split into an elongation-mediated (indirect) and a direct part. Indirect parts are pooled through the mediator with dose-response saturation, removing the mediator cross-correlation; direct parts keep the residual eigenvalue correlation at ρ̄. Correction (spectacles / surgery), which restores acuity without touching elongation, is NOT discounted against the control options. Here mediated effects are pooled WITHIN each cascade node (dose-response saturation of substitutes) and composed in SERIES across nodes (d-separated channels), with the per-node reductions reported so the channel structure is visible.

Mediator saturation cap = 55% axial control
Sum of standalone axial reduction (naive)
Combined axial reduction after saturation
Mediator overlap removed (1 - saturation)
Direct-effect redundancy removed (1 - n_eff/k)
Front-door combined RR
Backdoor-only combined RR (comparison)
InterventionRR%ctrlmed-fracindirect logdirect log

Which % of cross-correlation is appropriate? Not one number. The mediator overlap is fixed empirically by the axial saturation (currently removing of the summed mediated effect when interventions are stacked). Note a domain caveat: the optical defocus arms are mutually redundant (shared mechanism), whereas atropine and red-light add mechanism. Elongation-mediated fractions are transparent, adjustable priors from published axial-length data.

Front-door caveat (antithesis): the endpoint is a surrogate — millimetres of axial length, not lifetime blindness — and rebound on stopping erodes durability. The largest real-world win is simply correcting the ~1.1 billion with uncorrected refractive error; progression control addresses a smaller, future, childhood-specific burden.

Executive summary

Select interventions to generate a plain-language summary.