Basic Dos and Don’ts in Applying Signal Detection Methods to Spontaneous Reporting Systems Databases

Abstract

“Signals are useless until they are read and possibly subjected to intensive study by appropriate investigators in WHO or in national centres”, cited from Finney’s 1974 first article on the theory of signal detection. Recently, freely available large-scale spontaneous reporting systems databases and improved computational power of personal computers have made it easier to perform signal detection in personal researches. However, adequate knowledge for the limitations of source spontaneous reports and understanding the differences with conventional clinical and epidemiological studies are imperatives for valid interpretation of signal indexes. This review describes the basic theory, aims and limitations of the signal detection method using spontaneous reporting systems databases and practical considerations.