A service evaluation of restraint, seclusion and escalation in inpatient mental health care: a retrospective observational study in England

Background Restrictive practices, including physical restraint and seclusion, remain a major concern in inpatient mental health care because of their well-documented association with physical harm, psychological trauma and adverse experiences for both patients and staff. Although previous research has shown that restraint use varies across services, less is known about why some incidents escalate to seclusion while others do not. This study examined both the variation in the use of restraint and its progression to seclusion across forensic (secure) and non-forensic inpatient settings. Methods We conducted a retrospective observational study using routinely collected incident data from inpatient mental health services within a large NHS Mental Health and Community Trust in England between June 2024 and May 2025. A total of 41,685 incidents were analysed across forensic (secure) settings (including high secure services (HSS), medium secure unit (MSU) and low secure unit (LSU), and acute inpatient settings including older adult, psychiatric intensive care unit, and acute wards). Descriptive analyses were used to estimate restraint prevalence across service type, ward type, and ward gender. Regression, sensitivity, interaction, multilevel and counterfactual analyses were used to examine the association between restraint and subsequent seclusion, and the contribution of ward-level context to escalation. Results Restraint was unevenly distributed across both forensic and non-forensic settings. In forensic settings, restraint prevalence was highest in LSU (12.86%), followed by MSU (11.81%) and HSS (11.45%). In non-forensic settings, restraint prevalence was highest in older adult wards (12.93%) and was more common in mixed (11.55%) and female wards (9.47%) compared to male wards (6.73%). Escalation from restraint to seclusion was highest in medium secure units (6.22%) and acute wards (4.72%), but minimal in high secure and older adult wards. Multilevel modelling suggested that ward-level context contributed to variation in seclusion outcomes. Counterfactual simulation suggested that, in high-risk settings, reducing escalation processes may reduce seclusion rates than achieving a similar reduction in restraint incidents alone. Conclusions Restrictive practice in inpatient mental health care appears to operate as a two-stage system, with service configuration shaping where restraint occurs and ward context influencing whether incidents progress to seclusion. Reduction efforts should target both restraint frequency and ward-level processes that interrupt escalation and support safer, more therapeutic care.