(withdrawn) RN Fall Estimates as a Proxy Measure for Fall Counts from Incident Reports

Purpose:
To determine whether RNs' fall estimates are precise enough to be used as a proxy measure of fall counts from incident reports.

Background/Significance:
Research on fall incidence is fundamental to development of effective prevention protocols. Actual fall data are generally obtained from incident reports. Researchers sometimes use nurses' estimates of fall counts as a proxy for actual clinical fall data. Although nurses' estimates may be a more efficient measurement method than clinical record abstraction, the accuracy of nurses' proxy reports has not been examined in a large national sample.

Methods:
The sample included 1,207 NDNQI units that participated in the RN Survey and submitted patient fall data. RNs estimated total and injury falls on their units. Response options and respective analysis coding were: never, 0 falls; a few times a year or less, 6 falls; once a month or less, 12 falls; a few times a month, 24 falls; once a week, 52 falls; a few times a week, 104 falls, and every day, 365 falls. Clinical fall data for units were collected for 1 year preceding the RN survey. ICC1,K determined the relative homogeneity of nurses' estimates of fall counts within each unit. Regression analyses determined the association of nurses' estimates with actual counts total and injury falls.

Results:
The ICC1,K for nurses estimates of fall counts within each unit = 0.81, indicating high consistency of RNs' estimates of falls on their units. RNs' estimates and actual falls were significantly associated both for total falls, M = 33.5 per year, SD = 21.52, R-square = .32, p = .0001; and injury falls, M = 2.5 per year, SD = 1.98, R-square = .03, p = .0001. However, 68 percent of the variation in total falls from incident reports cannot be explained through RN estimates. Further, RN estimates of injury falls are only weakly associated with injury falls from incident reports.

Conclusions and Implications for Practice:
RNs' reports of falls differ considerably from fall counts from incident reports, which limits the appropriateness of RN estimates in research to situations where incident reports or chart reviews are not available.