Predictors of continued physical restraint use in nursing home residents following restraint reduction efforts

To examine predictors of continued restraint use in nursing home residents following efforts aimed at restraint reduction.

Secondary analysis of data from a clinical trial using a one-group, pre-test post-test design.

Three nonprofit, religion-affiliated nursing homes in a metropolitan area.

The sample consisted of 201 physically restrained nursing home residents. Following restraint reduction efforts, 135 of the sample were still restrained. Mean age of participants was 83.9 years.

Physical restraint use was measured by observation and included any chest/vest, wrist, mitt, belt, crotch, suit, or harness restraint plus any sheet used as restraint or a geriatric chair with fixed tray table. Nursing home residents were subjected to any one of three conditions aimed at restraint reduction, including adherence to the mandate of the Omnibus Budget Reconciliation Act of 1987 (OBRA '87), staff education, and education with consultation from a gerontological clinical nurse specialist. Resident characteristics including dependency, health status, mental status, depression, behavior, fall risk; presence of treatment devices and institutional factors were determined.

Physical dependency, lower cognitive status, behavior, presence of treatment devices, presence of psychiatric disorders, fall risk, and fall risk as staff rationale for restraint were associated (P < .10) with continued restraint use. Nursing hours, staff mix, prevalence of restraint use by unit, and site were also associated (P < .10) with continued use of physical restraints. Following bivariate analysis, associated resident characteristics were subjected to logistic regression. Lower cognitive status (OR = 2.4 (for every 7-point decrease in MMSE), 95% CI, 1.7, 3.3) and fall risk as staff rationale for restraint (OR = 3.5, 95% CI., 1.5, 8.0) were predictive of continued restraint use. Adding nursing hours, staff mix, and prevalence of restraint use by unit to the logistic regression model was not statistically significant (partial chi-square = 2.79, df = 6, P = .834). Nursing home site was added to the model without changing the significance (P < .05) of cognitive status or fall risk as a staff rationale for restraint use.

Continued restraint use in nursing home residents in this study most often occurred with severe cognitive impairment and/or when fall risk was considered by staff as a rationale for restraint. Efforts to reduce or eliminate physical restraint use with these groups will require greater efforts to educate staff in the assessment and analysis of fall risk, along with targeted interventions, particularly when cognition is also impaired.