Detecting diabetic polyneuropathy through vibration perception threshold

Vibration perception threshold (VPT) agrees well with clinician-identified diabetic polyneuropathy (DPN) and may be used as a diagnostic indicator, reports a new China study.

A total of 421 type 2 diabetes patients participated in the study, in whom VPT was determined through a neurothesiometre. The performance of VPT was compared against three gold-standard definitions of DPN: clinician diagnosis, abnormal nerve conduction, and clinically confirmed neuropathy.

Using the clinician diagnosis as basis yielded the highest prevalence rate of DPN among the participants, at 53.9 percent. When defined according to abnormal nerve conduction or clinical confirmation, the rates of DPN were 43.2 percent and 42.0 percent, respectively.

Regardless of the definition employed, patients with DPN had significantly higher levels of median VPT than those without (p<0.001).

VPT had high specificity for detecting DPN defined by any of the three gold standards. The highest estimate was for clinician-diagnosed DPN, for which VPT had a specificity of 85.1 percent. Respective values for abnormal nerve conduction and clinical confirmation were 77.0 percent and 76.6 percent.

Sensitivity estimates were slightly lower but were nevertheless good for all three definitions: clinician diagnosis (67.0 percent), abnormal nerve conduction (66.5 percent), and clinical confirmation (67.2 percent).

Moreover, the diagnostic value of VPT for clinician-diagnosed DPN was maximized when a cut-off value of 14.9 V was used. At this value, specificity jumped to 85.6 percent, while sensitivity dipped slightly to 66.2 percent.

Receiver operating characteristic curve analysis yielded area under the curve values ranging from 0.761–0.804. VPT performed best when used to detect clinician-diagnosed DPN, and poorest when the disease was defined according to abnormal nerve conduction.