Inter-observer variability in the assessment of nerve function in leprosy patients in Ethiopia

One of the major problems in leprosy is to detect any change in nerve function early enough so as to increase the chances of recovery and prevent disability. Several tests have been developed to assess nerve function and are used in leprosy control programs worldwide, but they are frequently performed by different workers on different occasions and under variable conditions. In this study we investigated the variability between different groups of observers in the assessment of nerve function in leprosy patients in Ethiopia. Sensory function was assessed by using a set of nylon monofilaments (NF) and a ball-point pen (BP), and motor function was assessed by using voluntary motor testing (VMT). We also studied the variability between observers in the assessment of the clinical signs of neuritis. Duplicate measurements were performed in random order on 50 leprosy patients by two physio-technicians and on 50 other patients by two health assistants. The percent agreement between observers was calculated for each single nerve, and weighted kappa statistics were used to assess whether agreement was better than expected due to chance alone. Systematic differences between observers were evaluated using the Wilcoxon signed rank test. On sensory testing, inter-observer variability was found to be related to the training and experience of the observer, to the nerve tested, and to the neurological status of the patient. When tests were performed by physio-technicians, we observed 32% to 58% agreement with the NF test and 71% to 84% agreement with the BP test, measured on different scales. After weighting for the scale difference, the agreement seemed comparable with these methods but the differences in measurements with the BP test were found to be dependent upon the neurological status of the patient. The variability between observers differed according to the nerve tested, and there was some evidence of systematic differences between observers with both methods. When performed by the health assistants, agreement was between 34% and 46% with the NF and between 66% and 82% with the BP tests. After weighting for the scale difference, the agreement seemed comparable but the BP was not liable to the systematic differences seen in the NF results. These differences could be attributed to the differences in the experience of the workers with these tests. With the VMT, small variability between observers was found for all nerves tested, except the facial nerve, when performed by both the physio-technicians and by the health assistants (72% to 98% agreement).(ABSTRACT TRUNCATED AT 400 WORDS)