Our study, published recently in the IJE, looks at the relationship between experience of violence, in the form of physical assault in the previous 12 months, and premature mortality in a sample of working-age Russian men living in Izhevsk in the Southern Urals.
We did this study because, although violence is considered important at a policy and political level, empirical public health research on the subject is patchy. There has been quite a lot of research on the relationship between violence exposure and later mental health problems, particularly post-traumatic stress disorder, and more generally the impact of violence on coping and psychological characteristics, particularly in children. Recently, there has been considerable work on the descriptive epidemiology of violence against women, both from a general population perspective and in studies focusing on women with serious mental illness.
In contrast, we found that population-based research on the physical health effects of exposure to violence was limited, and we decided to focus on possible associations between assault and mortality in our study.
We analysed data from a case–control study of mortality in Izhevsk, a city in the Southern Ural region of Russia. We gathered data on a series of 1750 deceased men, who died aged 25–54 years from any cause between 2003 and 2005. These men were compared on a range of sociodemographic factors with a group of living men resident in Izhevsk in the same timeframe. Information on both cases and controls was gathered from interviews with “proxies”, defined as individuals resident in the same household as the case or control at the time of recruitment. In particular, information was gathered for both cases and controls on alcohol, smoking, socioeconomic variables, adverse life events and imprisonment. Causes of death were available for all cases.
We were able to look at the associations between experiencing assault in the previous year and death from any cause, and found a strong “crude” association — there was a stark preponderance of assault in the cases, compared with controls.
An important question was whether this association was due to a causal effect of assault on mortality, or accounted for by similarities between men who are assaulted and men at risk of premature death. Obviously, confounding by alcohol use was a strong possibility that needed to be accounted for in the analysis. Fortunately, colleagues in our group had developed a relatively sophisticated measure of alcohol use in the same data. Using statistical modelling, they summarised a complex range of alcohol-related variables, reflecting alcohol-related social and occupational dysfunction, into a single continuous variable that we were able to use for modelling. Confounding by socioeconomic variables was also a concern; only limited information on socioeconomic position is collected in Russia, so the best we could do was to include variables in our statistical models reflecting ownership of a car or central heating, and employment status.
Another issue was how to explore or further understand the association between assault and mortality in our data. It seemed reasonable that being assaulted in the previous year would increase your probability of dying by homicide — people who are subject to violence are more likely to experience it in the future than those who do not. We found that the association remained apparent in the sample even when deaths by external causes, including from homicides, suicides and accidents, were removed.
To summarise, we found evidence for associations between assault and overall mortality, and death from external, cardiovascular and alcohol-related causes. These associations were not completely explained by alcohol use or socioeconomic factors, as far as they were measured in our data. Although we may not have been able to completely account for some confounders, such as alcohol, it is also possible that violence has a causal effect on mortality; for example, through stress pathways. Given the importance of violence in societies undergoing socioeconomic transitions, like Russia, more attention should be paid to the broad health effects of violence on victims.
Bhavsar V, Cook S, Saburova L, Leon DA. Physical assault in the previous year and total and cause-specific mortality in Russia: a case–control study of men aged 25–54 years. International Journal of Epidemiology 2016: doi: 10.1093/ije/dyw301.
Vishal Bhavsar is a Wellcome Trust Clinical Research Training Fellow at the Institute of Psychiatry, Psychology and Neuroscience at King’s College, London.