Peripheral Neuropathy — Stratified Response-Ratio Table

Bayesian Causal Atlas · longevityresearch.ca · 2026-07-21

Why this is not a Pareto-optimum reduction. Every other oracle in the atlas reports a single combined endpoint reduction. This one cannot, and the reason is in the evidence, not the code.

Only 5 of 17 interventions carry a ratio-scale estimate at all; the rest are annualised risk differences, standardised mean differences, NNTs, bare response rates, or narrative verdicts (“Improves”, “Uncertain”, “Conflicting”, “Mixed”). More decisively, the ratios that do exist are responder ratios above 1 (e.g. RR 1.73 = 1.73× as likely to reach ≥50% pain reduction), not hazard ratios below 1, and they sit on different endpoints in different populations — incident neuropathy in T1DM, pain response in diabetic neuropathy, disability-grade improvement in Guillain–Barré, relapse in CIDP.

Multiplying them together would yield a number with no referent. So the interventions are presented stratified by population, ranked within stratum, and never composed. Rows are comparable within a block and not across blocks.

CIDP (1)

Best comparable ratio in stratum: IV immunoglobulin (IVIG) RR ≈ 2.6
InterventionEndpointEffectMetricInterpretationNNTE-valueGrade
IV immunoglobulin (IVIG)≥1-pt INCAT improvement, maintained @24wkRR ≈ 2.6RRhigher = better (responder ratio)3high

Painful DPN (4)

Best comparable ratio in stratum: Capsaicin 8% patch OR 2.28
InterventionEndpointEffectMetricInterpretationNNTE-valueGrade
Capsaicin 8% patch≥30% pain reliefOR 2.28ORhigher = better (responder ratio)3.99mod
Duloxetine 60 mg≥50% pain relief @12wkRR 1.73RRhigher = better (responder ratio)52.85mod
Gabapentin ≥1200 mg≥50% pain reliefRR 1.7RRhigher = better (responder ratio)6.62.79mod
Pregabalin 300–600 mg≥50% pain reliefNNT ≈ 5NNTnot ratio-scale — not comparable5mod

GBS (severe) (2)

Best comparable ratio in stratum: Plasma exchange RR 1.64
InterventionEndpointEffectMetricInterpretationNNTE-valueGrade
Plasma exchange≥1-grade improvement @4wkRR 1.64RRhigher = better (responder ratio)high
IV immunoglobulin (IVIG)Recovery (7-grade disability)≈ Plasma exchangenarrativenot ratio-scale — not comparablemod

CIDP (IVIG-responsive) (1)

No ratio-scale estimate in this stratum — narrative or non-ratio evidence only.
InterventionEndpointEffectMetricInterpretationNNTE-valueGrade
Subcutaneous immunoglobulin (SCIg)Relapse/withdrawal over 24wk33% vs 63%response ratenot ratio-scale — not comparable3high

DPN (4)

No ratio-scale estimate in this stratum — narrative or non-ratio evidence only.
InterventionEndpointEffectMetricInterpretationNNTE-valueGrade
Structured exerciseNerve conduction, balanceImprovesnarrativenot ratio-scale — not comparablemod
Alpha-lipoic acidTotal Symptom ScoreConflictingnarrativenot ratio-scale — not comparablelow
Vitamin B12 / methylcobalaminSymptoms / conductionMixednarrativenot ratio-scale — not comparablelow
Acetyl-L-carnitinePain (SMD)SMD −0.45SMDnot ratio-scale — not comparablevlow

Neuropathic pain (1)

No ratio-scale estimate in this stratum — narrative or non-ratio evidence only.
InterventionEndpointEffectMetricInterpretationNNTE-valueGrade
AmitriptylinePain reliefUncertainnarrativenot ratio-scale — not comparablevlow

Painful CIPN (2)

No ratio-scale estimate in this stratum — narrative or non-ratio evidence only.
InterventionEndpointEffectMetricInterpretationNNTE-valueGrade
Duloxetine 60 mg (CIPN)Avg pain (BPI 0–10)MD 0.73MDnot ratio-scale — not comparablemod
Capsaicin 8% patch (CIPN)Clinician-rated relief≈44%response ratenot ratio-scale — not comparablelow

T1DM (1)

No ratio-scale estimate in this stratum — narrative or non-ratio evidence only.
InterventionEndpointEffectMetricInterpretationNNTE-valueGrade
Intensive glycaemic controlIncident neuropathy−1.84%/yrrisk differencenot ratio-scale — not comparablehigh

T2DM (1)

No ratio-scale estimate in this stratum — narrative or non-ratio evidence only.
InterventionEndpointEffectMetricInterpretationNNTE-valueGrade
Intensive glycaemic controlIncident neuropathy−0.58%/yrrisk differencenot ratio-scale — not comparablelow

Reading the table. Where the metric is RR or OR above 1, higher is better — these are the odds of reaching a responder threshold versus placebo. NNT is the number needed to treat for that same threshold. E-value is the minimum unmeasured-confounder strength needed to explain the result away. Grade is the source rating carried in the oracle.

What would change this. A combined optimum becomes computable only if the catalogue gains hazard ratios on a shared endpoint within a single population — not by parsing the existing text more cleverly.

Generated 2026-07-21 · every figure read directly from the oracle's own DATA catalogue; nothing derived or imputed