Severe mental illness (SMI), used here as a functional and service-relevant term rather than a single diagnostic category, makes clinical supervision and peer consultation unusually demanding. A patient may withdraw after a supportive comment, become suspicious when a plan is proposed, escalate after an apparently reasonable limit, or collapse into shame after progress is named. Such reactions are clinically important, but they are not self-evident indicators of therapist error, patient resistance, diagnosis, risk, or treatment effect. This Opinion article proposes that patient reactions can be used as supervisory data when they are treated as interpretive, context-dependent material for shared reflection. The article extends prior work on the Perceive-Assess-Dose-Safeguard (PAD-S) grammar and the Conflict-Square Algorithm (CSA) by positioning PAD-S/CSA as an optional, secondary notation layer for selected decision points in SMI supervision and peer consultation. Its primary didactic contribution is a five-question heuristic that helps supervisors move from one concrete patient reaction to one safer, accountable next step while holding relational, metacognitive, affective, neurobiological, social, service-context, and data-governance considerations in view. The article does not report new data, validate a measure, claim effectiveness, or propose automated decision support. Instead, it offers a bounded framework for making supervision more observable, teachable, human-final, and privacy-conscious in complex SMI care.
Eik Niederlohmann (Sat,) studied this question.