Interpersonal violence has been the subject of much recent interest as a public health issue in developed countries.¹ According to the Swiss 2000 Health Inquiry, 1 in 10 people in Switzerland admits having been a victim of violence during the past year, and about 40 000 women could be considered as having a serious health problem related to violence.^2,3 A study conducted in Switzerland in 1994 showed that 1 in 5 women had experienced domestic physical or sexual violence at least once in their lifetime.^4,5 These numbers are very similar to those from many other countries.⁶

Being a victim of interpersonal violence may result in physical injury or death. In addition, living in a violent setting can affect health, including an increase in chronic somatic diseases, alcohol and substance misuse, depression, suicide, eating disorders, and non-compliance with medical treatment, all of which lead to increased use of medical services and resources.^7,8 Violent settings also pose a risk for children and elderly people.^9,10

Of health services, emergency departments (ED) are most frequently called upon to treat patients that are victims of violence. Patients may consult an ED specifically because of violence related injuries (regardless of their public or domestic cause), to ask for treatment, and/or to make an injury report. About 2.5 % of all patients consulting an ED in the UK are victims of assaults or violent offence occurring in the community, and this statistic is 3–10 times higher than that found in police records.¹¹ In addition, people living in a violent situation may consult an ED, regardless of the motive for consultation, more frequently than usual. Several studies have shown the general prevalence of victims of domestic violence among patients consulting an ED to be between 12 and 26%,^8,12 with fewer than 1 in 25 being spontaneously identified by health workers.^13,23 One woman in four seeking care in an ED, for any reason, is a victim of domestic violence.⁹ Professionals working in ED are often poorly equipped to manage such situations, and services do not usually address this health problem.

More knowledge about frequency of consultation, characteristics of victims, and type of violence among people attending an ED is needed in order to develop screening policies and identification strategies for patients at risk, to provide quality care, and to develop relevant educational programmes. A study was therefore conducted in 2002 in the ED of a Swiss university teaching hospital in order to collect data about the ED point prevalence and the profile of people who are victims of violence. More precisely, the study was aimed at (a) determining the proportion and characteristics of adults consulting the medical and surgical wards of the ED who had a positive result on a brief violence screening questionnaire; and (b) comparing the number of these patients with the number of patients requesting an injury report. This paper summarises the main findings of this study.

METHODS

Study design and population

The ED of the university hospital where the study took place registered 41 467 patients in 2001 for a population area of approximately 250 000 inhabitants.¹⁴ Monitoring was conducted between 1 and 31 March 2002. During that period, all consecutively admitted adult patients (>16 years old), regardless of the cause of admission, were eligible for the study. Criteria for exclusion were: patients with life threatening injuries (National Advisory Committee on Aeronautics score ⩾4), or inability to understand or speak French, to give oral informed consent, or to be questioned without the presence of a family member or accompanying person. After providing oral informed consent, the patient was asked the screening questions by the nurse in charge. Patients received written information about the study and were also provided local sources ofmedical, social, and legal support in case of violence.

Study protocol

This study was submitted to and approved by the hospital’s academic research ethics committee (positive pre-study review by the institutional review board).

The screening questionnaire was the Partner Violence Screen (PVS), developed and validated in Denver.^15,16 For the purpose of this study, the PVS was translated into French, and minor modifications were made, namely, two questions about feelings of insecurity were added to the initial questionnaire on the recommendation of the ED researchers in Denver. In addition, two other questions were added, one concerning the relationship between violence and actual consultation, and another concerning willingness to discuss the problem with a medical doctor (see fig 1 for the modified, translated questionnair). Nurses discreetly administered the questionnaire while they settled the patient into the cubicle and performed the routine examination. If the answer to the first two questions was "No", the screening test was considered negative and was stopped. Conversely, if the answer to one of the first two questions was "Yes", the test was considered positive, and the other questions were completed. When it was not possible to interview a patient, the nurse wrote the main reason for this on the questionnaire. Completed questionnaires were then put into a special box in the ED. Time to complete the questionnaire was about 20 seconds. Patients declaring having experienced violence during the previous 12 months (PDV) were those with a positive PVS whereas patients declaring having experienced no violence during the previous 12 months (PDNV) were those with a negative PVS.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

 Modified PVS in French.

Data analysis

Based on an expected prevalence of about 10%, the duration of the study needed to obtain the required sample size was 31 days. This period was also estimated to be optimum to ensure sustained attention from the care providers—that is, administering the questionnaire as part of their usual activities. In addition to the information gathered by the questionnaire, sociodemographic data, which are recorded systematically for any admitted patient, were also collected. Statistical analyses were conducted with SPSS for Windows (version 11.0).¹⁷ χ² tests were performed where applicable and p<0.05 considered statistically significant.

RESULTS

In total, 2454 adult patients were treated in the medical and surgical wards of the ED during the study period. Completed questionnaires were obtained from 1894 patients, resulting in a participation rate of 77.2%. Non-participants differed on almost all variables (excluding civil status and nationality). They were more likely to have been young, male, treated in the surgery ward, and admitted Wednesday to Sunday night between 1600 and 2400 hours. Of the 1894 patients who completed the questionnaire, 270 were ineligible and 22 refused to participate in the study (11 women and 11 men). The refusal rate was thus 1.1% and the final sample consisted of 1602 patients.

The main characteristics of the patients based upon the results of the PVS screen are presented in table 1. The general prevalence of PDV was 11.4% (n = 183). Of these, 42% of cases reported only psychological violence, 18% only physical violence, and 40% both types of violence (data not shown). In PDV group, the proportion of men was slightly higher, and they tended to be younger and single. The majority of the women were non-Swiss nationals, more often admitted to the surgical ward (because of trauma), and seen slightly more often during nights (from 2400 to 0800 hours) and the weekend.

The prevalence of PDV according to the sociodemographic characteristics of the patients is shown in table 2. There was a significantly higher prevalence of PDV among people aged 16–34 years (19.6%), among those admitted to the surgical (14.3%) rather than to the medical ward (7.8%), and among those consulting between 2400 and 0800 hours (20.7%). The highest prevalence of PDV was found among those who were separated from their partners (26.2%). The lowest prevalence (3%) was found among people aged ⩾64 years. No significant differences in sex or nationality were found.

The main type of violence occurred in public places (table 3), with men more frequently involved than women. Conversely, domestic violence, the second most reported type of violence, more commonly involved women as victims. Violence occurring in the workplace was more prevalent among men.

More than half of the PDV group were aged 16–34 years. Only 16 patients (8.7%) >65 years of age reported violence on the PVS, and no-one aged over 94 years screened positively.

Of the 183 PDV, 46 (25%) declared that violence was the reason for the current consultation (giving an overall incidence rate of 2.8%), and of these, 25 patients received an injury report, 19 had violence mentioned in their medical records, and 2 had neither. Overall, violence was identified in the medical record or by the establishment of an injury report in 95% of PDV declaring violence to be their reason for the current consultation.

DISCUSSION

Approximately 1 in 10 adult patients attending the surgical or medical units of the ED of this university hospital declared having experienced physical and/or psychological violence during the previous 12 months, resulting in an overall prevalence of reported violence of 11.4%. Of these patients, 25% declared violence as their reason for the current consultation, which was confirmed by the emergency physicians, but only half completed an injury report. Even if immediate physical violence is usually identified by emergency physicians, it is evident that injury report rates reflect neither the true prevalence of victims of violence among patients nor the incidence of immediate physical violence.

A comparison with prevalence rates obtained in other studies is difficult, owing to factors such as variations in methodology, categories of patients studied, definition and type of violence, and setting.¹⁸ For example, in our study, the PVS screen showed that 34 of 807 women (4.2%) and 12 of 195 men (1.5%) declared having experienced domestic violence, whereas in the Denver study, the overall prevalence for female victims of domestic violence admitted to a general ED ward was 29.5%.¹⁵ Our prevalence rate does not include sexual violence because women seen in the emergency gynaecology/obstetric ward were not included in our study population. Similar issues arose when we analysed the prevalence rate of PDV among elderly patients. Despite numerous reports, including from Switzerland, that violence and related behaviours are frequently encountered among the elderly, it is striking to note that in this sample, the PVS was rarely positive among patients aged ⩾65 years and never in those >95 years. This observation may indicate low sensitivity of the PVS in this specific population.

In terms of external validity, the national health survey in Switzerland shows an overall prevalence of 9% of reported violence during the previous 12 months among people aged 15–75 years and over (10% of men and 8% of women). This includes three types of violence: psychological, physical, and offence against property.² In the present study, the prevalence of physical and/or psychological violence only was 12.1% for men and 10.8% for women, which is a higher rate than in the general population.

The majority of patients admitted to the ED agreed to be questioned about violence, with a refusal rate of only 1.1%. This is in accordance with previous studies showing patients’ readiness to be asked about violence.¹⁹ Interestingly, there was no difference in refusal rates between men and women. Despite the barriers to asking patients about violence that are cited in the literature and that were mentioned by nurses and physicians during the pre-test period, the number of questionnaires completed was relatively high (22.8% missing). This confirms the results of several studies, which showed differences between the representations and beliefs of health professionals and the willingness of patients to be asked about violent behaviour.^20,21 When nurses have the sole responsibility of distributing and collecting a self administered questionnaire, response rates in ED screening studies are about 70–80%.¹³ In our study, a response rate of 77.2% was achieved despite a demanding study design. Early involvement of ED nurses in the design of the study and regular presence in the ED of the researchers may explain our high response rate.

There are some limitations to this study. A full 22.8% of patients admitted to the medical or surgical wards during the survey period were not included in our sample. They differed from participants on all main sociodemographic factors. Non-participation by these individuals is explained by increased activity and reduced staffing in the surgery ward at night and during weekends. Patient overload may have made it difficult for nurses to enrol patients systematically in an ongoing manner during those times. The sociodemographic characteristics of non-questioned patients (young men, treated in surgery ward, admitted from Wednesday to Sunday night) and the increased rate of public violence in this population suggests that we may have underestimated our overall public violence rate.

The original Denver study using the PVS screening questionnaire included only women and focused exclusively on domestic violence. In our study, the PVS questionnaire was used to screen both women and men for all types of interpersonal violence. To accomplish this, we made some slight modifications to the original questionnaire. We also translated it into French to adapt it to the study population. These modifications may have decreased the questionnaire validity, possibly diminishing the strength of our observations.

This short, well accepted, and easily administered PVS questionnaire established a high prevalence of reported violence by patients admitted to an emergency department of a general hospital (11.4%). Despite its limitations, our study clearly demonstrates that a large number of victims seeking treatment for recent violence do not receive an injury report even though immediate physical violence is generally identified by emergency staff.²² This discrepancy indicates the usefulness of this type of questionnaire.