This study investigated individual differences in the use of risk factors when making risk assessments for self-harm. Clinical judgement analysis using 35 hypothetical case scenarios was used to determine how case characteristics influence risk assessment for self-harm by mental health professionals. Seven practitioners made four risk/priority assessment judgements for each case, and individual models of judgement for risk and priority were constructed by regressing 10 case variables onto these sets of judgements. All but two of the variables (potential risk factors) examined were related to risk and/or priority judgements. Risk assessors applied cues to make assessments consistently—broadly consistent with practice guidelines—although precise cue application varied between professionals. The findings demonstrate the potential for ambiguity in risk assessment (terms such as ‘low’ or ‘high’ risk had variable interpretation) and that it can be important to specify whether risk is to be assessed for the initiation, continuation, re-occurrence, or escalation of an event or condition. The study shows the importance of clear practice guidelines (not all risk factors were used as might be anticipated from practice guidelines) and illustrates how clinical judgement analysis can be used to understand and enhance the reliability of professional judgement. Copyright © 2011 John Wiley & Sons, Ltd.
Key Practitioner Message
- The process by which risk assessment is undertaken can be reliably statistically modelled to allow an in-depth examination of an individual practitioner or comparisons between practitioners, which can be used to support efforts to improve assessment reliability.
- Senior practitioners may be more consistent in applying information to clinical judgements of self-harm.
- Apparent agreement over a categorical judgement of risk may mask an underlying disagreement over numerical probabilities. The term ‘low risk’ may be particularly subject to variability in its interpretation.
- The term ‘risk’ can be ambiguous, reflecting the probability of occurrence for some practitioners and the severity of its occurrence for others, a problem that may be compounded if it is unclear exactly which event is being assessed (e.g., risk of initiating the behaviour versus risk of re-occurrence).