Doucet J, Senant J, Menard J F, Samson-Dollfus D, Schrub J C
Clinique thérapeutique, CHU de Rouen, Hôpital de Boisguillaume, France.
Neurophysiol Clin. 1989 Aug;19(4):327-37. doi: 10.1016/s0987-7053(89)80104-2.
The electrophysiological diagnosis of peripheral neuropathy in diabetics becomes increasingly difficult with the increase in number of parameters measured. We have examined 102 subjects: 60 diabetics, 32 control subjects and 10 with impaired glucose tolerance. From the 37 recorded parameters during 8 examinations, a linear discriminant analysis allowed us to separate the 19 diabetics with clinical neuropathy and the control subjects with only 5 parameters, with a good concordance (96%): the motor conduction velocities (MCV) of the peroneal and median nerves, the amplitude of the action potential (AP) of the sural nerve, the Hmax/Mmax ratio and the M latency of the Hoffman reflex. These parameters, weighted with a coefficient, constitute a score of discrimination. This score has been tested by the Jackknife method on the same sample (94% of well-classified patients), then validated on the other patients. Thus we propose a method to aid the diagnosis of diabetic peripheral neuropathy, without the arbitrary character of the classical single parameter method, and allowing easy repetition.
随着所测参数数量的增加，糖尿病患者周围神经病变的电生理诊断变得越来越困难。我们检查了102名受试者：60名糖尿病患者、32名对照受试者和10名糖耐量受损者。在8次检查期间记录的37个参数中，线性判别分析使我们仅用5个参数就能将19名患有临床神经病变的糖尿病患者与对照受试者区分开来，一致性良好（96%）：腓总神经和正中神经的运动传导速度（MCV）、腓肠神经动作电位（AP）的幅度、Hmax/Mmax比值以及霍夫曼反射的M潜伏期。这些参数经系数加权后构成一个判别分数。该分数已通过留一法在同一样本上进行测试（94%的患者分类正确），然后在其他患者中得到验证。因此，我们提出了一种有助于诊断糖尿病周围神经病变的方法，它没有经典单参数方法的随意性，并且易于重复。
