Prediction of incident diabetic neuropathy using the monofilament examination: a 4-year prospective study.

Abstract

To determine the specific monofilament examination score that predicts the subsequent 4-year incidence of diabetic neuropathy with the highest degree of diagnostic accuracy. Longitudinal follow-up of 175 of 197 (89%) participants in the Toronto Diabetic Neuropathy Cohort without baseline neuropathy for incident neuropathy. We examined the baseline monofilament examination score (and other simple sensory screening tests) by receiver operating characteristic (ROC) curve analysis. Incident diabetic neuropathy developed in 50 (29%) participants over a mean follow-up of 4.1 years (interquartile range 2.6-7.1 years). Although male sex, longer diabetes duration, taller height, and higher blood pressure at baseline were associated with incident neuropathy, the strongest association was with a lower baseline monofilament score (score out of 8 was 3.7 +/- 2.5 for incident neuropathy vs. 5.7 +/- 2.3 for those who did not develop neuropathy; P < 0.001). The optimal threshold score for risk of incident neuropathy was <or=5 sensate stimuli out of 8, with 72% sensitivity, 64% specificity, positive and negative likelihood ratios of 2.5 and 0.35, and positive and negative predictive values of 87 and 46%, respectively (chi(2) = 20.7, P < 0.001). Area under the ROC curve was significantly greater for the monofilament examination compared with that for other simple sensory tests. A simple threshold of <or=5 sensate stimuli out of 8 discriminates 4-year risk of diabetic neuropathy with acceptable operating characteristics. Although there are limitations in its specificity for prediction of future neuropathy onset, the monofilament examination is appropriate as a simple diabetic neuropathy screening instrument generalizable to the clinical setting.