Falls Prediction in Acute Care Units: Preliminary Results from a Prospective Cohort Study

To the Editor: Falling is common in hospitals (2.9–13 per 1,000 bed-days).1 Falls lead to long hospital stays.2 The first step in an efficient and cost-effectiveness strategy of fall prevention is screening of patients at higher risk of falls.3, 4 Recently, a brief geriatric assessment (BGA tool) composed of a few items (aged ≥85, male, ≥5 drugs taken daily, cognitive impairment, and history of falls in past 6 months) has been demonstrated to predict the risk of long hospital stays.5 Three levels of risk of long hospital stay have subsequently been identified in older inpatients: those at high risk (cognitive impairment and history of falls), those at intermediate risk (cognitive impairment or history of falls), and those at low risk (combination of the 3 other items). Because falls may extend the length of the hospital stay and because all items of the BGA tool are well-recognized risk factors for falls, it was hypothesized that the use of the BGA tool would make it possible to predict the occurrence of falls in older inpatients. There was a opportunity to test this hypothesis in a large sample of older inpatients. The objective of this analysis was to examine whether the different combinations of the BGA items predicted the occurrence of falls in older adults hospitalized in medical acute care units. Four hundred sixty-two individuals (mean age ± standard deviation 84.7 ± 6.9, 58.9% female) hospitalized in 10 medical acute care units of Angers University Hospital (Angers, France) were prospectively included in an observational cohort study between April 2013 and October 2013. Inclusion criteria were aged 65 and older, no treatment-limiting decision, and willingness to participate. Nurse teams collected information at admission in each acute care on age (≥85 or <85), sex, polypharmacy (≥5 drugs taken per day), cognitive impairment (inability to identify the month or the year (yes or no)), and a history of falls in the past 6 months (yes or no). Falls were defined as events resulting in a person coming to rest inadvertently on the ground or floor or other lower level. Nurse teams in each medical acute care unit recorded information on falls using patients' computerized files. Length of hospital stay was also calculated using the administrative registry of the University Hospital and corresponded to the delay in days between the first day of admission to the hospital and the last day of hospitalization in the acute care unit. The Angers ethics committee approved the project. A univariate Cox regression model was used to identify the association between each combination of BGA items and falls occurrence. The time to falls stratified according to the significant combinations identified using the Cox model was also examined using survival curves computed using Kaplan-Meier methods and compared using the log-rank test. P < .05 was considered statistically significant. All analyses were performed using SPSS version 19.0 (SPSS, Inc., Chicago, IL). As shown in Figure 1, two combinations of BGA items predicted the occurrence of falls: the combination of cognitive impairment and history of falls (hazard ratio (HR) = 2.34, 95% confidence interval (CI) = 1.15–4.75, P = .02) and the combination of all criteria (HR = 3.73, 95% CI = 1.14–12.16, P = .03). Kaplan-Meier distributions of occurrence of falls were significantly higher in inpatients with cognitive impairment and a history of falls (P = .02) and in those who had all BGA items (P = .02) than in the others. The BGA predicted the occurrence of falls in older inpatients hospitalized in medical acute care units. Of the combinations of BGA items, only those including cognitive impairment and history of falls were significantly associated with risk of falls. This result is in accordance with previous literature because cognitive impairment and history of falls are well-recognized independent risk factors for falls.1-4 These results also underscore that the accumulation (and probably the interaction) of several fall risk factors results in a greater risk of falls, the highest risk being reported in people with all BGA items. The main limitation of this study was the selection of older inpatients from a single hospital. Further research is needed to corroborate this finding with the objective of developing a clinically practicable tool to predict and monitor falls in medical acute care units. We are grateful to the participants for their cooperation. Conflict of Interest: The study was financially supported by the Angers University Hospital. Dr. Annweiler has served as an unpaid consultant for Ipsen Pharma company and serves as an associate editor for Gériatrie, Psychologie et Neuropsychiatrie du Vieillissement and for the Journal of Alzheimer's Disease. He has no relevant financial interest in this manuscript. Prof. Beauchet has served as an unpaid consultant for Ipsen Pharma company and serves as an associate editor for Gériatrie, Psychologie et Neuropsychiatrie du Vieillissement. He has no relevant financial interest in this manuscript. Author Contributions: Beauchet had full access to the data in the study. Study concept and design: Noublanche, Decavel, Beauchet. Acquisition of data: Noublanche, Simon. Analysis and interpretation of data: Noublanche, Simon, Annweiler, Beauchet. Drafting of the manuscript: Noublanche, Beauchet, Annweiler. Critical revision of the manuscript for important intellectual content: Simon, Decavel, Lefort. Obtained funding: Noublanche, Decavel, Lefort. Statistical expertise: Beauchet. Administrative, technical, or material support: Noublanche, Decavel, Lefort. Study supervision: Beauchet, Decavel. Sponsor's Role The sponsors had no role in the design or conduct of the study; the collection, management, analysis, or interpretation of the data; or in preparation, review, or approval of the manuscript.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.