Bayesian Causal Atlas · Vol. Gut Immunology · Pearl Structural Causal Model

Inflammatory Bowel Disease — Structural Causal Analysis

Inflammatory bowel disease (IBD) — Crohn’s disease and ulcerative colitis — affects roughly 3 million United States adults. Uncontrolled mucosal inflammation drives colectomy, colorectal cancer, venous thromboembolism, and hospitalisation. This oracle estimates the causal reduction in mucosal inflammation — the proximal mediator of these hard outcomes — achievable by combining advanced therapies, using effect sizes from named trials and network meta-analyses. Crucially, several of these agents carry their own serious harms, shown as competing-harm paths.

Method. Structural Causal Model (SCM) using cited, confounder-adjusted effect sizes (backdoor adjustment is the source study’s) (Pearl). Interventions are not assumed independent: their heavy mechanistic overlap (nearly all act by suppressing mucosal inflammation) is removed by an eigenvalue-corrected equicorrelation model at an adjustable mean cross-correlation ρ̄ (default 0.30). Robustness to unmeasured confounding is quantified per intervention with the E-value. Probabilities of Necessity / Sufficiency (PN / PS / PNS) are reported under a monotonicity assumption. Every hazard ratio is cited to its source trial — no effect size is invented.
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 hazard ratio, 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 inflammation-mediator overlap removed via dose-response saturation; residual direct-effect overlap removed via the eigenvalue model at ρ̄.

1.00
Combined HR
0%
Relative risk ↓
Pooled E-value
Interventions selected (k)
0
Effective independent dimensions (n_eff)
0
Redundancy discount applied
0%
Baseline 10-yr risk (illustrative anchor)
Absolute risk after intervention
Absolute risk difference (RD)
Backdoor-only HR (no front-door)
Mediator (weight) overlap removed
Number needed to treat (NNT)

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). Each therapy is a distinct node; nearly all benefit flows through the shared mediator mucosal inflammation (fecal calprotectin), which drives colorectal cancer, colectomy and venous thromboembolism — each a route to mortality. Critically, anti-TNF, JAK inhibitors and thiopurines add a competing-harm path (infection, lymphoma, drug-induced thrombosis) that the efficacy hazard ratio does NOT capture. Named confounders open back-door paths (adjusted). Illustrative of structure, not yet the identification engine.

Confounders U:age · disease extent · smoking status · prior biologic exposure · concomitant steroids · primary sclerosing cholangitis → back-door paths (adjusted)Vedolizumab (anti-integrin)Anti-IL-23 (ustek / risa)5-ASA (mesalamine)Anti-TNF (infliximab)JAK inhibitor (upa / tofa)ThiopurineMucosal inflammation(calprotectin) ↓Colorectal dysplasia→ cancerColectomy / surgeryVenous thrombo-embolismMortality (Y)Immunosuppression harms(infection / lymphoma / VTE)Front-door: through mucosal inflammationCompeting harm → mortalityBack-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 for this domain: incretin drugs, surgery, diet, and activity overwhelmingly share the weight-loss / insulin-sensitivity pathway, so ~30% of their nominal effects overlap. A mediation analysis of semaglutide found ~80% of its MACE benefit is not mediated by weight loss, which is why ρ̄ is user-adjustable rather than fixed.

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-hazard-ratio 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 HR
95% simulation interval
Standard deviation of combined HR
P(combined HR < 0.90)

Antithesis — challenging this oracle's own conclusions

Highest efficacy ≠ best choice. Upadacitinib (a JAK inhibitor) is the most effective agent for ulcerative-colitis remission (network-meta-analysis SUCRA 0.99), but the JAK class carries a boxed warning for thrombosis, major cardiovascular events and malignancy (ORAL Surveillance). The remission hazard ratio ignores this competing harm.
Active inflammation itself causes clots. IBD flares are prothrombotic, so controlling inflammation reduces disease-driven venous thromboembolism — yet JAK inhibitors ADD drug-driven thrombosis. The venous-thromboembolism node therefore receives arrows from BOTH the mediator and the competing-harm path; net effect is patient-specific.
Thiopurine benefit is bought with lymphoma risk. Azathioprine’s modest maintenance efficacy is offset by increased lymphoma and non-melanoma skin cancer, and hepatosplenic T-cell lymphoma when combined with anti-TNF in young males. Efficacy and harm must be weighed jointly.
The endpoint is a surrogate. Clinical remission and mucosal healing predict fewer colectomies and less colorectal cancer, but the mortality and cancer-prevention translation is partly observational. Colorectal-cancer risk in colitis is real but modern surveillance and inflammation control have lowered it substantially.
Indirect comparisons carry uncertainty. Most cross-drug hazard ratios come from network meta-analysis, not head-to-head trials, and placebo remission rates vary widely across studies — so the ranking is more robust than the absolute effect sizes.

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(mortality | do(∅)) — baseline
P(mortality | 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 hazard ratio would be…". Isolates each intervention's marginal causal contribution after mediator-overlap removal, so shared-pathway agents are not double-credited.

If not for…HR without itHR with full setmarginal RRR lost

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

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

Front-door (mediation) decomposition. Nearly every IBD therapy acts through one shared mediator — suppression of mucosal inflammation (fecal calprotectin / endoscopic healing). Each log-effect is split into an inflammation-mediated (indirect) and a small inflammation-independent (direct) part. Indirect parts are pooled through the mediator with dose-response saturation (you cannot suppress the same inflammation twice — combining a biologic with a second advanced therapy rarely doubles remission), removing the mediator cross-correlation; direct parts keep the residual eigenvalue correlation at ρ̄. The efficacy decomposition is SEPARATE from the competing-harm paths shown in the causal DAG.

Mediator saturation cap = 65% inflammation ↓
Sum of standalone inflammation reduction (naive)
Combined inflammation reduction after saturation
Mediator overlap removed (1 - saturation)
Direct-effect redundancy removed (1 - n_eff/k)
Front-door combined HR
Backdoor-only combined HR (comparison)
InterventionHR%inflam↓infl-medindirect logHRdirect logHR

Which % of cross-correlation is appropriate? Not one number. The mediator overlap is fixed empirically by the inflammation-suppression saturation (currently removing of the summed inflammation-mediated effect when advanced therapies are stacked). Because every agent converges on the same inflammatory pathway, overlap is large — combination advanced therapy yields far less than additive remission. Inflammation-mediated fractions are transparent, adjustable priors.

Front-door caveat (antithesis): the cited endpoint is clinical remission / mucosal healing — a surrogate. Its link to colectomy and colorectal-cancer prevention is supported but the mortality translation is partly observational. Critically, this front-door captures only the BENEFIT arm: it does NOT subtract the competing harms (JAK thrombosis, thiopurine lymphoma, anti-TNF infection) shown as separate paths in the DAG. A high-efficacy hazard ratio is therefore NOT a net-benefit estimate.

Executive summary

Select interventions to generate a plain-language summary.