Guideline evidence on self-harm risk tools is described as uncertain

What the study found: The authors argue that the NICE guideline on self-harm made strong recommendations against using risk assessment tools to predict repeat self-harm or suicide, even though the evidence base was very limited. They also say the guideline did not clearly acknowledge uncertainty about model impact, acceptability, and feasibility.
Why the authors say this matters: The authors conclude that future guideline updates should be informed by higher quality evidence, and they say there is an urgent need for more rigorous primary research on model impact, feasibility, and acceptability. They also note concerns about possible harms and misuse of tools, including denial of care.
What the researchers tested: This is a perspective article about the development of the 2022 NICE self-harm guideline. The authors reviewed shortcomings in the guideline process, the evidence review behind it, and newer evidence since 2022, including international work on implementation and cost-effectiveness.
What worked and what didn't: The guideline's evidence review included very little evidence, and the recommendations were said to rely almost entirely on committee expertise and experience. The authors say there is new evidence on implementation and cost-effectiveness, but they also state that prediction models should not be put into practice before adequate validation and impact assessment.
What to keep in mind: This article is a perspective, so it does not present new clinical trial results. The abstract does not describe detailed study limitations beyond noting gaps and uncertainty in the evidence base.

Key points

• The authors say the NICE self-harm guideline relied on very limited evidence for its recommendations.
• They argue the guideline did not adequately acknowledge uncertainty about impact, acceptability, and feasibility.
• The article highlights newer evidence since 2022, including international work on implementation and cost-effectiveness.
• The authors call for more primary research on model impact, feasibility, acceptability, and possible harms such as denial of care.
• They state that prediction models should not be introduced into practice without adequate validation and impact assessment.