Reappraisal of the diagnostic role of orthostatic hypotension in diabetes

Abstract

Given the controversial aspects of orthostatic hypotension (OH) testing in diabetes, we evaluated the diagnostic role for cardiac autonomic neuropathy (CAN) and for nondipping of OH, defined according to a fall in systolic blood pressure (BP) > or = 30 (30-OH) or > or = 20 mmHg (20-OH). 164 diabetic patients underwent 24 hours BP monitoring, three heart rate cardiovascular tests, and OH test. Compared to 30 mmHg, the 20 mmHg criterion increased the frequency of OH from 11 to 19.5%. Both 30-OH and 20-OH were associated with CAN (chi (2) = 30.5, P < 0.0001, and chi (2) = 45.1, P < 0.0001, respectively) and nondipping (chi (2) = 31.7, P < 0.0001, and chi (2) = 17.2, P = 0.0001, respectively). ROC curve for orthostatic systolic BP fall provided an AUC of 0.79 +/- 0.04 (95% CI 0.70-0.86) for diagnosing CAN and of 0.77 +/- 0.05 (95% CI 0.66-0.86) for diagnosing nondipping. Both 30-OH and 20-OH showed a low sensitivity and high specificity for CAN [sensitivity 31%, specificity 98%, Likelihood Ratio for a positive result (LR(+)) 17.1; and sensitivity 50%, specificity 95%, LR(+) 9.3, respectively], and for nondipping (sensitivity 40%, specificity 96%, LR(+) 8.9, and sensitivity 47%, specificity 87%, LR(+) 3.5, respectively), having 30-OH a higher LR(+) in both cases. OH had only moderate diagnostic accuracy, with high specificity and low sensitivity, for CAN, diagnosed on the basis of heart rate cardiovascular tests, and-as a novel finding-also for nondipping. A different definition of OH did not substantially affect its diagnostic characteristics, with just a slightly greater ability of the 30 mmHg criterion to estimate the probability of CAN and nondipping.