Bayesian Causal Atlas · Vol. Infectious / Oncology · Pearl Structural Causal Model

Human Papillomavirus (HPV) — Cancer-Prevention Structural Causal Analysis

HPV is near-universal over the sexual lifetime and causes essentially all cervical cancer plus a rising share of oropharyngeal, anal and penile cancers — ~350,000 cervical-cancer deaths a year, ~90% in low- and middle-income countries. This oracle estimates the causal reduction in HPV-attributable cancer death across vaccination, screening and precancer-treatment strategies, all acting through the persistent-HPV → CIN → cancer axis. Effect sizes are from named trials and national linkage studies. For education, not individual medical advice.

Method. Structural Causal Model (SCM) with backdoor adjustment (Pearl). Interventions are not assumed independent: vaccination, screening and treatment all act on the HPV → CIN → cancer axis, so their overlap is removed by an eigenvalue-corrected equicorrelation model at an adjustable mean cross-correlation ρ̄ (default 0.30). Gender-neutral / male vaccination is separated as a non-cervical cancer 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 (vaccination → barrier → screening → precancer → 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 HPV 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)
HPV-axis 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). Vaccination, screening and precancer treatment act through the shared mediator persistent high-risk HPV → cervical intraepithelial neoplasia, which drives invasive HPV cancer and thence cancer death (Y). Gender-neutral / male vaccination acts outside the cervical CIN pathway (oropharyngeal / anal cancers). Named confounders — age at exposure, screening attendance, HIV — open back-door paths (adjusted). Mediator cascade: interventions attach to the node they act on (vaccination → barrier → screening → precancer), 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 at vaccination / exposure · sexual-behaviour risk · screening attendance · socioeconomic access · immune status (HIV) · smoking → back-door paths (adjusted)HPV vaccination(girls <15, pre-exposure)Nonavalent vaccine(Gardasil-9)HPV vaccination(catch-up 15–26)Precancer excision(LEEP / LLETZ)Primary HPV DNAscreeningScreen-and-treat(VIA + ablation)Cytology (Pap)screeningGender-neutral/ male vaccinationCondom useVaccination(prevent infection)BarrierScreening(detect)PrecancertreatmentPersistent HPV /precancerCervicalcancerFront-door: through the HPV → CIN → cancer axisNon-cervical HPV cancers (oropharyngeal / anal)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: vaccination (primary), screening (secondary) and excision (tertiary) act at different points on the same HPV→CIN→cancer axis, so they are partly redundant — a fully vaccinated cohort needs less screening. ρ̄ is user-adjustable because male vaccination and cervical screening address partly different cancers. 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

Vaccination is prophylactic, not therapeutic. The vaccine prevents acquisition, so its benefit collapses once a woman is already infected — pre-exposure vaccination gives IRR ~0.12, catch-up only ~0.47. It cannot clear existing infection or treat established precancer, which is why screening and excision remain essential for already-exposed cohorts.
First-generation vaccines miss ~30% of cancers. Bivalent / quadrivalent vaccines cover the types behind ~70% of cervical cancers; the nonavalent raises this to ~90% but still not all. HPV-negative and non-vaccine-type cancers mean screening cannot be abandoned even in fully vaccinated populations.
The endpoint is decades downstream. Cervical cancer develops 15–30 years after infection, so the strongest evidence is on infection, CIN2/3 and modelled cancer, with direct invasive-cancer and mortality data only now maturing (Sweden, Scotland). Effect sizes on the true endpoint carry longer-horizon uncertainty.
The burden is an access and equity problem. Nearly 90% of cervical-cancer deaths occur in low- and middle-income countries with little vaccination or screening; incidence is also rising in the poorest high-income counties. The dominant lever is delivery and equity, not choosing among efficacious tools most at-risk women never receive.
Screening carries over-diagnosis and over-treatment harms. More sensitive HPV screening transiently increases detected CIN2/3 and excisions, and loop excision modestly raises later preterm-birth risk. The cancer benefit is real but is not free of downstream harm, which the single relative risk does not net out.

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. Vaccination, screening and treatment act through one shared axis — persistent HPV → CIN. Each log-effect is split into an HPV/CIN-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. Gender-neutral / male vaccination, which prevents non-cervical HPV cancers, is NOT discounted against the cervical arms. 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% HPV / CIN control
Sum of standalone HPV reduction (naive)
Combined HPV 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%HPV↓med-fracindirect logdirect log

Which % of cross-correlation is appropriate? Not one number. The mediator overlap is fixed empirically by the HPV saturation (currently removing of the summed mediated effect when interventions are stacked). Note a domain caveat: primary, secondary and tertiary cervical arms are partly redundant on the same axis (a vaccinated cohort needs less screening), whereas male vaccination adds a distinct non-cervical benefit. HPV/CIN-mediated fractions are transparent, adjustable priors.

Front-door caveat (antithesis): the vaccine is prophylactic (benefit collapses after exposure), first-generation vaccines miss ~30% of cancers, the endpoint is decades downstream, and ~90% of deaths are in low-resource settings — so the dominant real-world lever is access and equity, not tool choice.

Executive summary

Select interventions to generate a plain-language summary.