Diagnostic properties of nerve conduction tests in population-based carpal tunnel syndrome

BACKGROUND Numerous nerve conduction tests are used for the electrodiagnosis of carpal tunnel syndrome (CTS), with a wide range of sensitivity and specificity reported for each test in clinical studies. The tests have not been assessed in population-based studies. Such information would be important when using electrodiagnosis in epidemiologic research. The purpose of this study was to compare the diagnostic accuracy of various nerve conduction tests in population-based CTS and determine the properties of the most accurate test. METHODS In a population-based study a questionnaire was mailed to a random sample of 3,000 persons. Of 2,466 responders, 262 symptomatic (numbness/tingling in the radial fingers) and 125 randomly selected asymptomatic responders underwent clinical and electrophysiologic examinations. A standardized hand diagram was administered to the symptomatic persons. At the clinical examination, the examining surgeon identified 94 symptomatic persons as having clinically certain CTS. Nerve conduction tests were then performed on the symptomatic and the asymptomatic persons by blinded examiners. Analysis with receiver operating characteristic (ROC) curves was used to compare the diagnostic accuracy of the nerve conduction tests in distinguishing the persons with clinically certain CTS from the asymptomatic persons. RESULTS No difference was shown in the diagnostic accuracy of median nerve distal motor latency, digit-wrist sensory latency, wrist-palm sensory conduction velocity, and wrist-palm/forearm sensory conduction velocity ratio (area under curve, 0.75-0.76). Median-ulnar digit-wrist sensory latency difference had a significantly higher diagnostic accuracy (area under curve, 0.80). Using the optimal cutoff value of 0.8 ms for abnormal sensory latency difference shown on the ROC curve the sensitivity was 70%, specificity 82%, positive predictive value 19% and negative predictive value 98%. Based on the clinical diagnosis among the symptomatic persons, the hand diagram (classified as classic/probable or possible/unlikely CTS) had high sensitivity but poor specificity. CONCLUSIONS Using the clinical diagnosis of CTS as the criterion standard, nerve conduction tests had moderate sensitivity and specificity and a low positive predictive value in population-based CTS. Measurement of median-ulnar sensory latency difference had the highest diagnostic accuracy. The performance of nerve conduction tests in population-based CTS does not necessarily apply to their performance in clinical settings.