Public health and preventing violenceBMJ 2011; 342 doi: http://dx.doi.org/10.1136/bmj.d2882 (Published 16 June 2011) Cite this as: BMJ 2011;342:d2882
- Alexander Butchart, Coordinator, prevention of violence
- 1Department of Violence and Injury Prevention and Disability, World Health Organization, 1211 Geneva 27, Switzerland
Interpersonal violence is a leading cause of death in adolescents and young adults worldwide.1 For every death caused by violence, scores of people have injuries that require urgent medical care, and in 2004 an estimated 16 million such cases received medical attention.2 In the linked study (doi:10.1136/bmj.d3313), Florence and colleagues evaluate whether sharing anonymised information obtained during the clinical care of victims with police and local government partners can help prevent violence.3 Their programme typifies the public health approach to violence prevention,4 which the programme leader, Jonathan Shepherd of Cardiff University, has pioneered in the United Kingdom.5
The study found that sharing of information was associated with a significant reduction in violence related hospital admissions (adjusted incidence rate ratio 0.58, 95% confidence interval 0.49 to 0.69) and in woundings recorded by the police (0.68, 0.61 to 0.75) in Cardiff relative to comparison cities. The study also showed that less serious assaults recorded by the police increased significantly (1.38, 1.13 to 1.70).3
The study makes several assumptions. Firstly, that at a population level the “who, what, when, where, and why” of violence can be statistically predicted. Secondly, that police recorded data on violence resulting in injuries under-ascertain the real number of cases because some victims who receive medical care do not report the incident to the police. Thirdly, that by combining data from health services and the police, predictions of future violence can be improved. Fourthly, that when patterns of violence are accurately predicted, the frequency of future acts of violence can be reduced by tackling the underlying causes and risk factors.
The statistical predictability of violent incidents and injuries has been well established in developing and developed countries, where since the 1980s epidemiological methods have been applied to understand the magnitude, characteristics, consequences, and causes of violence. Crucially, these studies show strong dose-response associations between levels of violence and potentially modifiable factors such as economic inequality; accessibility to and misuse of alcohol and firearms; poor monitoring and parental supervision of children; and behavioural problems, such as impulsivity and hyperactivity.6 7
Fewer studies have compared levels of violence shown by police recorded data versus data from health services, and most compare violence related deaths recorded in police statistics with those reported by mortuary based fatal injury surveillance systems. Such studies show that although in developed countries there is some under-ascertainment of homicides in police statistics, in developing countries the gap is larger and official police statistics substantially undercount the number of homicide victims seen in mortuaries (Shaw M, van Dijk J, Rhomberg W. Determining global crime and justice trends: an overview of results from the United Nations surveys of crime trends and operations of criminal justice systems. Unpublished background document prepared for the Expert Meeting on the World Crime and Justice report, 2004-2005, 26-28 June 2003, Turin, Italy). Such comparisons inform recommendations from the United States that to improve surveillance of violent deaths, data sharing between police, coroners, and other relevant services is essential.8
By extending a similar line of reasoning to non-fatal cases of violent injury, Florence and colleagues’ findings on the extent to which police data under-ascertain violence related injuries in the UK is important for doctors, police, and public health experts in other developed countries. In developing countries where the availability of and access to medical care is less than in developed countries, local studies are urgently needed to explore the patterns of ascertainment by health services and the police.
The assumption that by combining health data and police data, violence can be more accurately predicted raises several questions. Are police recorded cases qualitatively similar to those seen by health services? When compared with findings from population based surveys, which dataset performs best? Studies are needed to answer these questions, and to quantify the predictive value added by combining sources. It can be assumed that the combination of police data and health service data more accurately predicts violence because it produces a more complete sample of all violence related injuries than either source alone. However, because some highly prevalent forms of violence (such as child maltreatment, intimate partner violence, sexual violence, and elder maltreatment) only infrequently result in injuries that require medical care, these forms of violence are best measured through population based surveys.9 10
The prevention of interpersonal violence has been reasonably well established through outcome evaluation studies, including randomised control trials, with the strongest evidence available for programmes aimed at preventing youth violence and child maltreatment.11 However, most of these are studies of programmes that operate at the individual level and close relationship or family level, and almost all are from developed countries.12
Fewer evaluations exist of potentially more effective and more cost effective societal and community level programmes, such as the Cardiff one. In light of its large effect on preventing violence, this model will hopefully be emulated by other cities in developing and developed countries, in each case with at least the same level of monitoring using health service and police records as was applied in the original study.
Despite the practical barriers to performing a randomised controlled trial of the intervention with the city as the unit of randomisation, it should be attempted. If subsequent randomised controlled trials or even less definitive experimental studies also find the significant reductions in violence shown in Cardiff, it would increase confidence in the value of this new tool to prevent violence.
Cite this as: BMJ 2011;342:d2882
Competing interests: The author has completed the Unified Competing Interest form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declares: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work.
Provenance and peer review: Commissioned; not externally peer reviewed.