Research and statistics: demystifying type I and type II errors.

摘要

As highlighted in the example opening this article, the sensitivity of RIDTs for detecting novel influenza A (H1N1) can be as low as 10% (although there have been few studies to date). (1)(2) The low sensitivity results in a high likelihood of a false-negative result. Thus, any negative test result would have a high likelihood of being an erroneous result, that is, a type II error. In the example, the clinician chose not to conduct the RIDT for detecting novel influenza A (H1N1), in part, because of the likelihood of a type II error, choosing instead to treat the 22-month-old patient presumptively because he has been classified as a high-risk case due to his age. (4)