Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Identifying patients likely to fall in hospital


Falls are dangerous for elderly people, and even if serious injury is avoided they can cause anxiety and reduce social and physical activity. A number of different things are likely to affect whether people fall, including their age, any disease they have, and any drugs they are taking. A large study from Italy [1] helps to identify patients in hospital at greater risk of falling.


Data were collected prospectively on 7900 patients in 58 hospitals during an eight-month observation period. All admitted patients were enrolled, without exception. The study lasted to discharge or death. In Italy nurses must record all falls, and doctors must review the consequences of all falls and report serious falls to the hospital administrators.

For every patient a special questionnaire was filled out by a trained physician. This included all drugs used in hospital, and drugs used in the 30 days before entering hospital. Benzodiazepines particularly were recorded and characterised by their half-life (long - greater than 24 hours; short - 12 to 24 hours; very short - less than 6 hours). Cognitive status was also assessed with a validated scale. Diseases and diagnoses were recorded, together with comorbidities.

Incidence of falling was calculated as the number of patients with one or more falls divided by the total number of patients. A multivariate regression model was used to identify factors independently associated with falling.


Of the 7900 patients 74% were older than 64 years and 34% older than 85 years. There were 1870 users of benzodiazepines. Falls occurred in 174 patients (2.2%). Multivariate analysis identified a number of factors independently associated with an increased risk of a fall (Table 1). The incidence of falls increased dramatically as combinations of these risk factors occurred together (Figure 1).

Table 1: Independent risk factors identified by multivariate analysis

Item Multivariate odds ratio (95% CI)
Age >80 years 2.7 (1.5 to 4.7)
Benzodiazepine very short half life 1.9 (1.03 to 3.3)
Benzodiazepine short half life 1.8 (1.2 to 2.8)
Other psychotropic agent 2.3 (1.6 to 3.2)
Antidiabetic drug 1.5 (1.03 to 2.2)
More than 5 drugs 1.6 (1.02 to 2.6)
Three or more diseases 1.7 (1.05 to 2.8)
Cognitive impairment 1.6 (1.08 to 2.3)
Length of stay 17 days or more 2.1 (1.4 to 3.3)

Figure 1: How the incidence of falling increases with combinations of risk factors


The study was large, which is good, and prospective, which is good, and took considerable trouble over data collection, which is also good. Despite this there was incomplete collection in 1250 patients.

What it does is to give us an insight into those facets of the patients, their condition, and their treatments that can contribute to an increased risk of falling. When several of the risk factors occur together, risk can be significant. If patients at increased risk can be identified, trouble can be taken to ensure their safety.


  1. A Passaro et al. Benzodiazepines with different half-life and falling in a hospitalized population: the GIFA study. Journal of Clinical Epidemiology 2000 53: 1222-1229.
previous or next story in this issue