In 1996, the World Health Organization declared violence a leading worldwide public health problem.2 Its subsequent publication World report on violence and health confirmed that violence had become a global phenomenon of epidemic proportions in all societies.3 In the European Union, for example, an estimated three million workers, or 2% of the labour force, have been subjected to physical aggression and violence at work.3

We already know from a number of studies conducted among health professionals that Australian health care workers frequently experience violence (Box 1). However, collection and aggregation of incident data reported by health professionals from multiple hospitals can reveal information not available from single hospital or single study reports. The purpose of such a patient safety reporting system is “to ferret out and correct vulnerabilities, not to count them”.11

The Australian Patient Safety Foundation (APSF), a non-profit research organisation, has played an important role in developing systems to collect, aggregate, monitor and analyse incidents related to patient safety since its formation in 1988, when it received its first reports of anaesthetic-related events for the Australian Incident Monitoring Study in anaesthesia. In 1998, to re-analyse data from the Quality in Australian Health Care Study,12 the APSF created a taxonomy and software which became the Australian Incident Monitoring System (AIMS). This involved developing a classification of health care incidents — both adverse events and near misses — an incident being “any event or circumstance which could have led, or did lead, to damage, loss or harm”.13

Incident reporting to AIMS has always been voluntary. Reports have been made by all types of health care professionals from a varying number of participating hospitals and other health facilities within Australia and New Zealand. Data are de-identified before aggregation and analysis, and thus cannot be used to determine incidence rates. However, their value lies in understanding the contributing, minimising and preventive factors involved, which can then be used in devising corrective strategies and action plans for incidents affecting patient safety.

Violence is one major category of incident classified within AIMS. Because of growing concern about violence in health care in Australia, we reviewed the relevant data collected using AIMS.14 Here, we present and discuss some of the summary data relating to reported incidents of violence.

The complete AIMS collection of patient-safety-related incidents from January 1998 to June 2002 contains nearly 80 000 coded reports. The data include contributing factors, action taken and outcome, when available. Among 42 338 incidents reported from 1 July 2000 to 30 June 2002, 3621 (9% of all incidents) involved patients and physical violence (for example, assaults or throwing objects) or violent verbal exchange (abuse or threats) occurred.14 Staff injury was reported in 5% of cases.Box 2 summarises the key findings.

Although 9% of reported incidents in all health units involved violence, the proportion was higher in emergency departments (16%) and higher still in mental health units (28%). Incidents in these two areas have their own characteristic patterns of contributing and precipitating factors.

In reports from 12 emergency departments, 16% of all reported incidents (190 of 1214 incidents) involved violence. Mental health problems were patient-related contributing factors for over half of the violent incidents, with alcohol or drug intoxication contributing to more than 25%. The most common staff-related contributing factors were “communication problems” and “insufficient or inadequate [numbers of] staff”. Precipitating factors included dissatisfaction with staff decisions to admit or discharge a patient from hospital, lack of support on discharge, or non-prescription of a patient-requested medication. Weapons used in incidents included hospital equipment, razor blades, scissors and blood-filled syringes.

In reports from 10 mental health services, 28% of all incidents reported (1467 of 5326 reports) involved violence. In violent incidents in which the patient’s status (voluntary or involuntary admission) was known, 79% involved patients who were legally detained. There was often a precipitating factor, such as refusal of privileges.

Health services are provided in a changing environment where new policies and programs are often needed in response to changing patient demand — such as varying demographic characteristics of patients and the nature of their presenting symptoms — but these policy changes have to be balanced against the obligation to provide a safe environment for both health professionals and patients.

AIMS has continued to evolve since an early version was evaluated in November 2002.28 It is now capable of analysing incidents from all sources in health care including investigative analyses, coroners’ reports, and mortality and morbidity reviews as well as incident reports. Importantly, it now also allows for consumer complaints.

AIMS is now used on a state-wide basis by the health departments of New South Wales, Western Australia, South Australia and the Australian Capital Territory, as well as by some health facilities in the Northern Territory, Victoria and Queensland. States collect and analyse their own data to aid state-based decision-making. Incident monitoring also has the potential to enable health services to collect local data on patient, staffing and system factors that contribute to incidents, and to develop and evaluate local management plans. However, there is currently no national system to aggregate incident monitoring data.

We recommend that a national system be developed to share and compare incident monitoring data, to monitor trends, and to facilitate learning and thinking at all levels — ward, department, hospital, state and national. At all levels, intelligent interrogation of data and imaginative initiatives will be needed to improve care.