Falls among nursing home residents: an examination of incident reports before and after restraint reduction programs

To examine changes in the rate of falling of an experimental group of restrained subjects who underwent restraint reduction, and to compare their rate of falling with a group of subjects who did not have restraint orders during the study period.

A quasi-experimental, multiple time-series study utilizing the principles of single-subject design. Each subject was followed for 25 weeks before and 25 weeks after initiation of the intervention.

Seven nursing homes.

Formal programs aimed at reducing all bed and chair restraints were initiated in all sites after staff received education and training. Multidisciplinary teams implemented the restraint reduction process on a case-by-case basis, beginning with 1 unit/floor at a time in each site. Most experimental subjects reached their optimum restraint-elimination/reduced status within 2 weeks of intervention initiation. The implementation periods ranged from 4 months to more than a year.

Subjects with chart orders for restraints at the start of the study comprised the experimental group and participated in the restraint reduction program (184 subjects). Subjects with no orders for restraints during the study period comprised the nonequivalent control group (111 subjects) and, therefore, did not undergo the intervention.

Incident reports documenting all falls during the study period were examined. Falls, the dependent variable, were classified as serious or nonserious. The independent variable (intervention) was the restraint reduction program offered to experimental subjects. Based on the principles of single subject design, the impact of the intervention on falls was calculated before and after the date the restraint reduction process was initiated for an individual (experimental group) subject. For the control group, the pre- and posttest period was calculated from the start date of the restraint reduction program on the unit on which each subject resided.

Serious falls did not increase, but nonserious falls increased significantly after restraints were removed or reduced in experimental subjects. The total mean weekly fall rate for this group increased from 1.87% of residents falling per week during preintervention to 3.01% during postintervention. The mean weekly fall rate of the control group was 3.18% at pretest and did not change statistically over time.

The increase in nonserious falls among the experimental group may be attributed to restraint reduction. The mean weekly fall rate in the experimental group postintervention (25 weeks) became comparable to the mean weekly fall rate for the control group during the entire study period (50 weeks). In light of such findings, policy makers have to confront the ethical choice between tying some frail, elderly subjects to beds and chairs versus exposing them to the risks of freedom in their old age.