CIDP misdiagnosis leads to inappropriate treatment with economic and patient burden. Consensus guidelines assist diagnosis but are underutilized based on complexity.

We sought to create a simple clinical-electrophysiological model facilitating CIDP versus mimic neuropathy prediction, accessible as a web-based probability calculator.

2021-EAN/PNS guidelines were used for CIDP diagnosis with 26 clinical and 144 nerve-conduction variables chosen based on these criteria and publications addressing mimics. 110 CIDP and 309 mimics (IgG4-nodopathies, paraneoplastic, POEMS, anti-MAG, diabetic radiculoplexus neuropathies (DRPN), MMN, inherited, eight others) underwent data extraction. Univariate and multivariate regression analysis identified the most informative variables, validated in a CIDP subset cohort.

We analysed 9,282 clinical and 51,408 electrophysiological data points. Univariate analysis identified 11/26 clinical variables with significant odds-ratios. A multivariate regression model revealed six clinical-electrophysiologic variables—progression over 8 weeks (OR 40.66, p=0.0004), absent autonomic involvement (OR 17.82, p=0.0018), absent muscle atrophy (OR 16.65, p=0.0007), proximal weakness (OR 3.63, p=0.0024), ulnar motor conduction velocity slowing <35.7 m/s (OR 5.21, p=0.0003), ulnar motor conduction block (OR 13.37, p=0.0026) —achieved 93%-area-under-curve (95%CI 91-95). EAN/PNS electrodiagnostic criteria did not enhance the model (92% area-under-curve, 95% CI 88.6-94.8). The calculator at 92% probability cut-off showed 100% sensitivity but 67% specificity. Specificity improved to 93% after considering clinical “red-flags”, 2021-EAN/PNS electrophysiologic demyelination, and laboratory testing where indicated for NF155/Contactin1-IgG4, paraneoplastic antibodies, VEGF-gammopathy, MAG-gammopathy and inherited neuropathy gene panel.