Performance of drug-drug interaction software for personal digital assistants.

BACKGROUND Personal digital assistants (PDAs) allow healthcare professionals to check for potential drug-drug interactions (DDIs) at the point of care, reducing the need to consult traditional references. However, PDAs can only be as effective as the software programs they use. OBJECTIVE To examine the ability of DDI software programs manufactured for Palm OS-compatible PDAs in detecting clinically important DDIs. METHODS Eight PDA software programs were assessed for sensitivity, specificity, and positive and negative predictive values for 16 well-documented DDIs contained within 6 simulated patient profiles. RESULTS Sensitivity of the software programs ranged from 0.81 to 1.0, specificity ranged from 0.52 to 1.0, positive predictive values ranged from 0.62 to 1.0, and negative predictive values ranged from 0.88 to 1.0. Five programs scored perfect sensitivity scores: DrugIx, ePocrates Rx, ePocrates Rx Pro, Lexi-Interact, and the Tarascon pocket Pharmacopoeia. Of these, the ePocrates programs scored the highest in specificity (0.9), while Lexi-Interact and the Tarascon pocket Pharmacopoeia scored considerably lower (0.52). MosbyIx was the only program to score a 1.0 in specificity; however, its sensitivity was just 0.81. CONCLUSIONS ePocrates Rx and ePocrates Rx Pro scored greater than or equal to 90% in regard to both sensitivity and specificity, making them the most reliable in detecting the clinically relevant interactions studied without the distraction of detecting those of no clinical significance. In addition, ePocrates Rx is updated regularly and is easily accessible on the Internet at no cost.