Using a large general population sample, we found that people aged 16 and over with one or more disabilities including mental illness had relative odds of 3.0 (2.3–3.8) and those with one or more disabilities excluding mental illness had relative odds of 1.8 (1.5–2.2) of being a victim of past-year violence compared with the non-disabled after adjusting for socio-demographic and behavioural factors (with similar relative odds across violence subtypes). Compared with non-disabled victims, victims with disability were more likely to experience mental health problems following violent incidents, especially those with pre-existing mental illness. We estimated that around 8% of violence in the general population and half of violence among those with disability could be attributed to the independent effects of disability, and that this resulted in an estimated additional 116,000 people with disability experiencing actual violence in England and Wales in 2009, at an excess cost of £1.51 billion.
Overall, prevalence and risk estimates are consistent with studies from other countries. 
In the USA, one national and one statewide household survey found that women with disability had four-fold the odds of being a victim of sexual assault than non-disabled women. 
Both studies found no association between disability and physical assaults, but this may be due to limited study power. In our much larger study, we found a clear association between disability and both physical and sexual assaults. In Taiwan, national data on sexual assaults found that people with disability were more likely to experience sexual assaults than those without, particularly those with learning difficulty and chronic psychosis. 
This is consistent with our finding of high risk among those with mental illness. However, we failed to find an association between violence and either learning disability or sensory impairment. This may be due to limited power (only 170 people with learning disability participated in this study). It could also be due to participation bias; the survey was designed for the general population, and people with significant intellectual impairment or communication problems may have found it difficult to participate in the lengthy and detailed study interview.
We found that the relative odds of violence outside and within the home were equally high, with the former being more prevalent. However, the prevalence of domestic violence may have been underestimated due to response or disclosure bias. Disabled victims were less likely to complete the sensitive self-reported measure of domestic violence than non-disabled victims, and it is possible that non-completers were at higher risk. Disclosure of domestic violence may be particularly difficult for disabled victims, as they may be dependent on perpetrators, and fear increased violence or independence loss and institutionalization following disclosure. 
Nonetheless, this study suggests that interventions for both non-domestic and domestic violence are required in this population. We found that social deprivation and substance misuse did not account for the excess risk of non-domestic violence, but did account for some of the excess domestic violence risk (especially amongst those with mental illness), suggesting that these factors could be appropriate intervention targets for addressing domestic violence risk.
Past evidence suggested that those with mental illness were at particularly high risk, but this was largely based on comparisons between studies with widely differing settings and measures. 
Our study is one of the few to directly compare risks for those with self-defined mental illness versus other disability types in a community sample. We found that those with mental illness had significantly higher risks of violence victimisation, and were more likely to suffer mental ill health following violence, than those with other disabilities. This may be explained by a high concentration of intersecting risk factors at the personal, interpersonal, community and societal levels among those with mental illness 
These include high rates of exposure to childhood violence (e.g. parental domestic violence and childhood abuse), which predisposes to mental illness and personality difficulties, which in turn put people at risk of low self-esteem, interpersonal conflict, substance misuse and violence perpetration. 
This constellation of problems increases the risk of victimization, and decreases the likelihood of exiting a cycle of violence. Our findings suggest that those with mental illness would be a suitable group for targeted intervention. The factors that put them at risk may start early and require broad and complex interventions. Future research should identify which subgroups of people with mental disorder are at greatest risk of victimization. One cohort study found that people with both common mental disorders (anxiety, depression) and severe mental illness (schizophrenia) were at increased risk of physical and sexual assaults. 
Among people with severe mental illness, risk is highest for those with a history of childhood abuse, co-morbid substance misuse and social disadvantage (such as homelessness and poor social support) 
We estimated population attributable fractions (PAFs), as this “provides a bridge by which results of epidemiologic studies can be made relevant to public health policy”. 
By estimating this measure, we would classically be making the assumption that disability ‘causes’ violence, and that there are interventions that can eliminate disability and the violence risks associated with it. 
We acknowledge that the causal pathway between exposure to disability and violent victimization is poorly understood and is likely to involve a complex interplay of variables relating to victim, perpetrator and environment. However, in our analyses, we sought to estimate the proportion of violence attributable to disability that is not explained by factors shared with the general population (and measured in our study). Factors unique to those with disability may include decreased ability to understand danger, to escape from a perpetrator on whom they are dependent and to communicate experiences to health and legal authorities. 
Whilst disability-related risk accounted for a relatively small proportion of violence in the general population, the estimated number of victims with disability arising from this excess risk, and associated economic costs, are sizeable. Although we used the best available costing measures from the general population, these may not account for differences in demography, baseline health and response to violence in the disabled. A significant proportion of violent crime cost arises from its physical and emotional impact on victims. 
As we showed in this study, the psychological impact is greater among people with disability, so we are likely to have underestimated the true cost.
Strengths of this study include the large, nationally representative sample with detailed measures of disability, violence and covariates, which allowed us to generate robust estimates of violence prevalence, risk and population impact. The study has several limitations. The target population only included people living in private residential households, so findings cannot be generalized to people with disability living in residential or supported accommodation. Findings cannot be generalized to those who have significant communication or cognitive problems (of a severity that would preclude their participation in the BCS). The survey did not use a sensitive measure of sexual and domestic violence in those aged 60 and above, hence underestimating these violence subtypes in this age group. Although it is difficult to establish temporal relationships in cross-sectional surveys, those with disability had a minimum illness duration of 1 year, and the main outcome was past-year violence, so-apart from measurement errors- disability would have preceded violence. Reporting bias is possible, but its likely direction is unclear. People with mental illness or other disabilities may over-report violence since it has a greater impact on them. Conversely, they may under-report violence as they may worry more about the consequences of disclosure. Past evidence suggests that people with mental illness tend to reliably report victimisation experiences. 
Our research highlights the need for clinicians to be aware of the elevated risks of domestic and non-domestic violence among patients with all disability types; and of the increased risk of mental health problems among disabled victims. Although people with mental illness are the most vulnerable to violent victimisation, mental health professionals often fail to screen for recent abuse, and violence is rarely detected or acted upon. 
A recent review on domestic violence interventions for people with disability found that disabled victims had difficulty accessing generic services. Specialist services were rarely available and had a poor evidence base– although there were some promising approaches, including safety training and peer support. 
In the non-disabled population, there is good evidence that there are effective interventions for both primary violence prevention (e.g. parent training, life skills training for children and adolescents) and secondary violence prevention (e.g. screening tools, education programs for health professionals, advocacy support programs). 
Future research should evaluate the effectiveness of these interventions among people with disability,; as well developing interventions to address risk factors specific to this group (e.g. caregiver stress, communication barriers to disclosure). From a policy perspective, our findings strengthen the economic and public health arguments for interventions in this group, and suggest the need for greater integration between health and victim support services.
People with disability, a predominantly elderly and disadvantaged group, are at increased risk of violence both within and outside the home. The significant public health and economic burden calls for an urgent assessment of the causes of this violence, and national policies on violence prevention in this vulnerable group.