Restraint use, restraint orders, and the risk of falls in hospitalized patients.

OBJECTIVES: To determine the relationship between physical restraints and falls in the acute hospital setting.
DESIGN: Matched case-control study.
SETTING: Inpatients at a 528-bed, urban, community based, acute care hospital. PARTICIPANTS: Two hundred twenty-eight patients who fell during hospitalization and 228 controls matched to cases by nursing unit and length of stay. MEASUREMENTS: Persons who fell were systematically evaluated at the time of fall by trained fall evaluators. For the cases, we sought to validate "orders for restraints" using "observed restraint use," defined as the use of restraints at the time of fall as determined through direct observation or interviews with nursing staff.
RESULTS: Patients with orders for restraints were more likely to fall than patients without orders for restraints (multivariate relative risk = 6.3, 95% confidence interval (CI) = 1.8-22.3). However, in the cases, there was poor correlation between "orders for restraints" and "observed restraint use" at the time of fall (kappa = 0.15, 95% CI-0.4-0.34).
CONCLUSION: Because orders for restraint use may not reflect actual restraint use at the time of a fall, observational studies relating use of restraints to the risk of falls should be interpreted with caution. Despite this caveat, we could find no evidence that restraints protect hospitalize patients from falling.