Derivation and validation of diagnostic criteria for chronic inflammatory demyelinating polyneuropathy

Abstract

To develop diagnostic criteria for chronic inflammatory demyelinating polyneuropathy (CIDP), a retrospective series of patients' records diagnosed by sexpert consensus as CIDP or other chronic polyneuropathies were analyzed. Classification and regression tree analysis was applied to 150 patients to derive a classification rule. According to the rule, diagnosis of CIDP required that a patient have a chronic non-genetic polyneuropathy, progressive for at least eight weeks, without a serum paraprotein and either 1) recordable compound muscle action potentials in > or =75% of motor nerves and either abnormal distal latency in >50% of nerves or abnormal motor conduction velocity in >50% of nerves or abnormal F wave latency in >50% of nerves; or 2) symmetrical onset of motor symptoms, symmetrical weakness of four limbs, and proximal weakness in > or =1 limb. When validated in 117 patients, the rule had 83% sensitivity (95% confidence interval 69%-93%) and 97% specificity (95% confidence interval 89%-99%) and performed better than published criteria.