Improper study design precludes valid effect estimates in important suicide prevention research.

The observational study design for estimating accuracy of diagnostic tests for suicide risk in clinical work is not ideal, due to the effects of directed suicide prevention to the high-risk group. This is an example of the confounding by indication and protopathic bias, which lead to misinterpretation of the accuracy terms sensitivity and specificity. The simple arithmetic mechanism presented here, forces the conclusion that the accuracy estimates sensitivity and specificity, applied in open prospective trials of surmised diagnostic tests for suicide risk, cannot be normally interpreted as accuracy estimators. Further, the related concept "prediction of suicide" is shown to be fundamentally illogical and should not be used in the present context. All these statements reveal a far-reaching problem within the suicide prevention research: Conclusions regarding the usefulness of diagnostic tests of suicide risk in the vast previous research since decades deserve reinterpretation. Diagnostic tests per se, can possess highly qualitative properties in estimating important suicidological aspects regarding a patient, but, when studied in an open design study, it cannot be demonstrated. This problem concerns rating scales and any biological and psychological tests in medicine, when confounding factors, for example, suicide prevention, influence the outcome because of the test result per se.