Objectives While the definition and importance of the phenomenon of overdiagnosis have been increasingly debated, the definition of underdiagnosis has not. This is despite underdiagnosis being less defined in the literature of evidence-based medicine (EBM) than overdiagnosis. The term underdiagnosis has been used in more than 2800 scientific papers since the 1960s. With the growing awareness of overdiagnosis, the use of underdiagnosis is growing as well, often as a counterargument to overdiagnosis. Yet, what is underdiagnosis? And how is it related to overdiagnosis? We present the preliminary results of a conceptual investigation of the definitions of underdiagnosis. We compare underdiagnosis with overdiagnosis and suggest a more coherent terminology for terms related to underdiagnosis. We further raise the suggestion to consider underdiagnosis to be ‘true negative’ just as overdiagnosis can be considered ‘true positive’ in the 2x3 diagnostic table.

Method 1) We search for explicit definitions of underdiagnosis in the literature of EBM. 2) We then searched for papers mentioning both underdiagnosis AND overdiagnosis (as well as over-diagnosis AND under-diagnosis) on PubMed. 3) In those papers from step 2, we looked for explicit definitions of underdiagnosis. If no explicit definition was stated, we looked for explicit definitions in the papers’ references related to underdiagnosis. 4) The definitions found were then analysed by comparing them to the classical 2x2 table normally used in diagnostic test accuracy. This table has also been used to analyse overdiagnosis in which the table has been expanded to 2x3 table. 5) The findings in step 4 were then used to analyse the arguments used in debates/conflict about underdiagnosis and overdiagnosis in the papers from in step 2.

Results The preliminary results show that the EBM literature rarely defines underdiagnosis. The investigated literature in general rarely states explicitly their definition of underdiagnosis. Often, underdiagnosis is understood as ‘false negative’ or lack of diagnosis – two much different situations. To the best of our knowledge, no one has suggested underdiagnosis to be ‘true negative’ just as overdiagnosis is ‘true positive’

Conclusions We present the various uses of the term underdiagnosis, as well raising the suggestion of an underdiagnosis terminology that would be more coherent with the terms used in the field of overdiagnosis. A more coherent use could improve the understanding of underdiagnosis which would improve our understanding of overdiagnosis and diagnosis in general.