Falls in English and Welsh hospitals: a national observational study based on retrospective analysis of 12 months of patient safety incident reports

Introduction: Falls in hospital inpatients are common, with reported rates ranging from 3 to 14 per 1000 bed days. They cause physical and psychological harm, are associated with impaired rehabilitation, increased length of stay and excess cost, and lead to complaints and litigation, making them a crucial area for risk management. A National Reporting and Learning System (NRLS) for patient safety incidents in England and Wales was utilised to examine frequency of falls in hospitals specialising in acute care, rehabilitation and mental health; related harm; timing; age and gender of patients who fell; and to draw general lessons from this which might inform fall-prevention strategies. Methods: The NRLS database was searched retrospectively for slips, trips and falls occurring between 1 September 2005 and 31 August 2006. Organisations were classified as “regularly reporting” if they returned reports at least monthly and with at least 100 patient safety incidents per month for acute trusts and 50 per month for community and mental health trusts. Falls rates were standardised as number of falls per 1000 occupied bed days. Reporting hospitals used standardised categories for degree of harm from incidents, and injury rates were calculated as the percentage of injuries by severity per fall. Key word searches combined with free text scrutiny were conducted to identify specific injuries. Specific falls rates for time of day, gender and age were also compared, with age and gender adjusted for bed occupancy rates from Hospital Episode Statistics (HES) data. Most data were used descriptively, though 95% confidence intervals were used to facilitate comparisons between groups and where samples are generalised to the data set as a whole. Results: Reports of 206 350 falls were received from a total of 472 organisations. Falls incidents accounted for 32.3% of all reported patient safety incidents. 152 069 (73.7%) reports were from acute hospitals, 28 198 (13.7%) from community hospitals, and 26 083 (12.6%) from mental health units. Only 102 of these could be classified as “regularly reporting” organisations, and in these the mean standardised rates of falls per 1000 bed days were 4.8 in acute hospitals, 2.1 in mental health units and 8.4 in community hospitals. 133 417 falls (64.7%) resulted in “no harm,” 64 144 (31.1%) in “low harm,” 7506 (3.6%) in “moderate harm,” and 1230 (0.6%) in “severe harm,” with 26 reported deaths. The proportions of falls resulting in some degree of harm varied significantly across the care settings: mental health units (44.5%; 95% CI 43.9 to 45.1), community hospitals (37.0%; 95% CI 36.4 to 37.6) and acute hospitals (33.4%; 95% CI 33.2 to 33.7). Patients aged 85–89 years had a higher-than-expected likelihood of falling relative to bed days. Males accounted for 50.8% (95% CI 50.5 to 51.1) of falls and females 49.2% (95% CI 48.9 to 49.5). (Occupied bed days were 45.5% male and 54.4% female.) The proportion of falls varied considerably with time, with a peak occurring between 10:00 and 11:59. Discussion: This paper describes the largest retrospective study of hospital falls incidents and draws on data from almost 500 institutions of varying types. It describes wide variations in falls recording and reporting, and in recorded falls rates between institutions of different types and between institutions of ostensibly similar case-mix. As falls are the commonest reported patient safety incident, there is a pressing need for improvements in local reporting, recording and focused analysis of incident data, and for these data to be used at local and national level better to inform and target falls prevention, as well as to explore the reasons for large apparent differences in falls rates between institutions.