Constraints on Restraints: A Signal Detection Analysis of the Use of Mechanical Restraints on Adult Psychiatric Inpatients

Abstract

The use of mechanical restraints is highly controversial in part because they are potentially dangerous. A non-trivial number of psychiatric patients die every year as a result of being placed in restraints. Yet mental health professionals use restraints regularly, proclaiming that they are a necessary part of practice. Exacerbating this dissensus are the statutes and case law that govern the use of restraints, both of which fail to meaningfully specify the conditions under which the use of restraints is permissible. This paper describes an approach to fill this void.The use of restraints, like any diagnostic decision, turns on a tradeoff between false positives (using restraints unnecessarily) and false negatives (failing to use restraints when necessary). A given tradeoff implies a decision threshold. In this context, the decision threshold reflects the point at which the perceived level of ‘violence’ occurring legitimates the use of restraints. We elicited tradeoffs from former mental patients and psychiatrists about the costs of using mechanical restraints. Patients indicated that, on average, 86 false negatives are equivalent to 1 false positive–requiring the likelihood of violence to be in excess of 99% in order to use restraints. By contrast, psychiatrists indicated that, on average, 8 false positives are equivalent to 1 false negative — implying that restraints should be used whenever the likelihood of violence exceeds 15%. Differential weights were applied to these disparate responses in order to specify a single, normative threshold. Policy and practical implications for the use of mechanical restraints are discussed in light of these findings.