Classification of risk in psychiatry

Psychiatric risk-assessments generally quantify risk using broad, categorical, indicators (e.g., high-risk, low-risk). We examined reliability of such indicators when applied by mental-health professionals. Four versions of a questionnaire were used, each specifying a different clinical outcome along with a range of different probabilities at which that outcome might occur. Respondents classified each probability, allowing a comparison of the level of likelihood at which different professionals would apply the terms 'high-risk', 'medium-risk' and 'low-risk'. We found little consistency among professionals who assessed risk for the same outcomes. Moreover, there were also large and unpredicted differences in response-profiles between the 4 clinical outcomes. These findings raise concerns about the communication value of current risk-assessment terminology.