Pill identification: comparing the identification accuracy and error rates of Internet search engines and standard pill identification applications by entering visual characteristics

Background: Proper identification of medications is important for patient safety. Many pill identification applications are available to help identify tablets/capsules but their accuracy varies. Internet search engines could be an alternative for identification. Aim: The hypothesis was that Internet search engines would perform as well as or better than pill identification applications in determining the name and dosage of medications. Methods: A set of images of 451 pills was prepared. Visual characteristics were entered into two Internet search engines and three pill identification applications. Head to head comparisons of correct identifications were made using McNemar's test. Results: No statistically significant difference was found in correct identifications between Google (93.6%) and Drugs.com (94.0%). Both Google and Drugs.com performed better than Bing (91.4%) at the p < 0.05 level. Drugs.com, Google and Bing all performed better than Pillbox (68.3%) and Epocrates (54.1%) at the p < 0.0001 level. Conclusion: Internet search engines perform well in pill identification and better than some dedicated pill identification applications.