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

Chronic Low Back Pain — Structural Causal Analysis

Low back pain is the single leading global cause of years-lived-with-disability, affecting hundreds of millions at any time. This oracle estimates the causal reduction in persistent disabling low back pain across active, psychological, pharmacologic, manual and interventional options — and deliberately includes two anti-targets (long-term opioids, anticonvulsants) so the frontier shows them as dominated. Effect sizes are from named guidelines and meta-analyses. For education, not individual medical advice.

Method. Structural Causal Model (SCM) with backdoor adjustment (Pearl). Interventions are not assumed independent: the active / psychological arms share the biopsychosocial pain pathway, so their overlap is removed by an eigenvalue-corrected equicorrelation model at an adjustable mean cross-correlation ρ̄ (default 0.30). Symptomatic analgesia is separated as a function-independent direct 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 (active rehab / → psychological → pharmacologic → procedural → 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 pain 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)
Functional-pathway 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). Active, psychological and manual therapies act through the shared mediator pain processing / functional capacity, which drives activity limitation and thence persistent disabling pain (Y). Symptomatic analgesics act independently of function (they damp the pain signal). Named confounders — fear-avoidance, distress, specific pathology — open back-door paths (adjusted). Mediator cascade: interventions attach to the node they act on (active rehab / → psychological → pharmacologic → procedural), 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:psychosocial distress · fear-avoidance · baseline severity · occupation / physical load · comorbidity · specific pathology (stenosis / radiculopathy) → back-door paths (adjusted)Exercise /stabilisation therapyMultidisciplinaryrehabilitationCognitive behaviouraltherapyDuloxetine(SNRI)NSAIDsSpinalmanipulationEpidural steroidinjectionLong-term opioidAnticonvulsant(gabapentin/pregabalin)Active rehab /exercisePsychological(CBT)PharmacologicProcedural(injection)BiopsychosocialpainDisability /functionFront-door: through function / central pain modulationSymptomatic analgesia (function-independent)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 active and psychological arms overlap on the biopsychosocial pathway, so stacking exercise, rehabilitation and CBT is partly redundant. ρ̄ is user-adjustable because a symptomatic analgesic and an exercise programme share little mechanism. 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

No intervention is strongly effective. Low back pain is the world’s leading cause of years-lived-with-disability, yet the best-evidenced option (exercise) reduces pain by only ~1.2–1.8 points on a 0–10 scale. Every relative risk here is modest; the honest message is that chronic low back pain is hard to treat, not that any arm is a cure.
Two of the arms are anti-targets. Long-term opioids give only a small, not-clinically-important benefit with major harm, and anticonvulsants (gabapentin / pregabalin) are simply not effective for chronic low back pain. They are included precisely so the Pareto frontier shows them as dominated — selecting them worsens the risk/benefit balance.
Symptomatic relief is not disease modification. NSAIDs, epidural injections and opioids damp the pain signal without restoring function (the gold, mediator-independent path). Relief that does not improve activity limitation tends not to last, and the disability endpoint is what actually matters.
The endpoint is subjective and regression-prone. Pain and disability scores are self-reported and strongly subject to regression to the mean and placebo response; much acute-on-chronic back pain improves regardless of treatment, which inflates apparent effect sizes in uncontrolled settings.
Specific pathology needs a different pathway. These estimates are for non-specific chronic low back pain. Radiculopathy, stenosis, fracture, infection and malignancy (the confounder path) require targeted diagnosis and management — applying the non-specific arms to specific pathology is unsupported.

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. The restorative arms act through one shared biopsychosocial mediator. Each log-effect is split into a function-mediated (indirect) and a direct symptomatic 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. Symptomatic analgesics, which relieve pain without restoring function, are NOT discounted against the active arms — and the two anti-target arms show near-null hazard ratios, exposing dominated 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% pain control
Sum of standalone pain reduction (naive)
Combined pain 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%funcmed-fracindirect logdirect log

Which % of cross-correlation is appropriate? Not one number. The mediator overlap is fixed empirically by the pain saturation (currently removing of the summed mediated effect when interventions are stacked). Note a domain caveat: the active / psychological arms are partly redundant, whereas symptomatic analgesia adds a distinct (short-lived) mechanism. Function-mediated fractions are transparent, adjustable priors.

Front-door caveat (antithesis): no arm is strongly effective (exercise, the best, moves pain ~1.2–1.8/10), the endpoint is subjective and regression-prone, and two arms are anti-targets. Specific pathology (radiculopathy, stenosis, fracture, malignancy) needs a different pathway.

Executive summary

Select interventions to generate a plain-language summary.