That simple assertion reflects the beginning of a serious consideration of the epidemiology of trauma. As the data come in, it is clear that a disturbing—and increasing—proportion of trauma cases are caused by interpersonal violence. In the mid-1980s, a public health surveillance approach to violence was implemented to establish the extent of the problem, identify risk groups and risk factors, and support program development [1]. Patterns of violence among different groups have been identified and are now beginning to inform intervention efforts. For instance, improved surveillance shows that violence against women is epidemic.

Females of every age face the threat of violence. Current estimates indicate that 1 in 5 girls is sexually abused and that the peak ages of such abuse are 8 to 12 years. The prevalence of rape in the United States is estimated to be 1 in every 4 women, with the peak incidence between ages 18 and 24 years. Domestic violence, or partner abuse, will involve 1 in 5 women at some time in their adult lives, generally between the ages of 24 and 32 years. Pregnancy appears to be a particularly vulnerable period, and data indicate that 1 in 6 pregnant women is physically or sexually assaulted during her pregnancy [2, 3]. Finally, elder abuse is estimated to involve as many as 1 in every 20 persons older than 65 years of age. The prevalence of elder abuse tends to increase with age, peaking among the "old-old," that is, those who are most frail and dependent.

Common imagery suggests that it is strangers in dangerous places who rape and assault women, but epidemiologic studies prove otherwise. At every age in the life span, females are more likely to be sexually or physically assaulted by father, brother, family member, neighbor, boyfriend, husband, partner, or ex-partner than by stranger or anonymous assailant; risk improves little with higher socioeconomic status. Intrafamily homicide may be preventable because it generally occurs after previous assaults and because a gun is involved in 59% of cases [4].

This issue of Annals includes four important articles that discuss interpersonal violence across the life span—childhood sexual assault, adolescent rape, and adult domestic violence—and its long-term health effects on women. Each article addresses violence from a different perspective—that of an epidemiologist, a clinical subspecialist, and a primary care provider—and thus reminds us that interdisciplinary work is key to understanding the problem.

McCauley and colleagues [5] determined the prevalence of domestic violence and identified clinical characteristics associated with current domestic violence among 1952 women seen in four community-based primary care internal medicine practices. Finding that 22% of women had been physically or sexually abused before the age of 18 years and that 21% had been abused during adulthood, the authors confirm the magnitude of the problem and its association with ongoing physical and mental health problems, as measured by standard tools such as the CAGE questionnaire for alcohol abuse and the Symptom Checklist-22. As in other recent research, McCauley and colleagues [5] find that a much smaller proportion of women, 1 in 20, have been abused within the past year. The significance of this particular time cut-off is not at all clear, and, although the cut-off is emerging as a variable in medical research, the rationale for its use is not apparent [6]. No studies have yet established that the absence of injury in the past year indicates safety for women in violent relationships. The limits of such an approach were clearly shown in a conversation with a middle-aged women in my practice who had been abused. When I asked her if the violence had subsided, her response was accompanied by a questioning look. "Well, I guess it's better. He doesn't hit me anymore ... but he sleeps with a gun under his pillow."

The earliest health care interventions on behalf of battered women were based in emergency services; more recent efforts have focused on primary care providers, including practitioners in obstetrics and gynecology. In this issue, Drossman and colleagues from the American Gastroenterological Association [7] explore the association between gastrointestinal illness and sexual abuse during childhood and adolescence. The authors conclude that a history of abuse is significantly linked to chronic medical problems, particularly functional bowel disorders, and call for increased physician involvement in the identification and more appropriate referral of patients. Although their speculation on the physiologic basis for a specific link between sexual abuse and gastrointestinal dysfunction is intriguing, it remains to be shown that the prevalence of abuse among patients seen in gastroenterology clinics differs significantly from that among women seen for atypical chest pain in cardiac clinics, for joint and muscle aches in rheumatology clinics, or for headaches in neurology settings. Research in these settings may substantiate the need for routine assessment in many clinical sites.

The essay by Weinberg in this issue [8] suggests that chronic pain is a metaphor for the sexual assault survivor's unfinished healing. The essay is a particularly eloquent demonstration of the role of the physician in promoting recovery from trauma. Other research has addressed the prevalence of abuse among patients seen in a chronic pain clinic, and the clinical dimensions of the rape trauma syndrome were elucidated in the late 1970s [9, 10]. As the physician in Weinberg's essay cares for the baker's daughter, both patient and provider are led through what have been identified as the necessary stages of recovery—establishing safety, identifying the trauma, retelling the story, and reconnecting with the community [11].

Although the survivors of sexual abuse during childhood, rape, domestic violence, or elder abuse have distinct needs, a common thread in their experience is reflected in the post-traumatic stress syndromes. In each case, violence is frequently accompanied by a sense of betrayal, vulnerability, and ambivalence about community norms that seem to accept abuse—and the abuser. Thus, the long-term clinical problems associated with abuse across the life span share a common core of hypervigilence, anxiety and phobias, somatic symptoms, dissociative disorders, sexual dysfunction, depression, substance abuse, suicide attempts, and the risk for recurrent abuse. These problems are seen as part of a post-traumatic stress disorder similar to that seen among combat veterans [12-14].

This "survivorship" framework is appropriate for understanding the difficulties faced by women who have experienced criminal victimization in the past—the survivors of rape, child abuse, or child sexual assault. In this issue, however, Alpert [15] describes the components of an effective clinical response to patients who face the continuing threat of violence in intimate relationships. Domestic violence encompasses not only physical injury but also threats, sexual abuse, emotional and psychological torment, economic control, and progressive social isolation. It is also generally ongoing. Here, the post-traumatic stress framework breaks down. Clinicians must instead recognize that, in this setting, recurrent injury and abuse are predictable and that our patients' physical and mental health problems constitute an intra-traumatic stress response. In Alpert's paper, safety planning emerges as a necessary priority.

Therefore, the physician's role in providing care to any female who is abused or sexually assaulted begins with establishing a physically and emotionally safe encounter. First and foremost, the abused patient must be able to rely on nonjudgmental and supportive interpersonal communications to enable her to participate in the complex but necessary medical and legal evaluations without further abuse and, if needed, to develop a safety plan to prepare for future threats of violence.

As physicians become more involved, challenges become clear. Do we have the resources? The introduction of new clinical practices into a medical paradigm that is not only managed but determined by cost-effectiveness requires genuine advocacy by physicians on behalf of our patients. What will we need to provide adequate care to those involved in interpersonal violence? Although formal evaluation studies have only recently been initiated, clinical violence intervention should address the following areas:

1. Access to comprehensive health care for victims of violence regardless of the relationship between the victim and the perpetrator;

2. Access to counseling and support services through managed care, insurance, and entitlement programs;

3. Specialized programs of assistance to victims and perpetrators of interpersonal violence in psychiatric and substance abuse programs;

4. Crisis intervention capacity in hospitals and ambulatory care settings that integrates nursing care, social services, safety, and medical care;

5. Employee assistance and impaired providers programs for health care personnel involved in violent relationships; and

6. Enhanced research and surveillance to contribute medical and public health knowledge to community violence prevention and intervention programs.

Clinical violence intervention is in its infancy but is growing rapidly. This issue of Annals reflects the participation of physicians in violence prevention and an emerging consensus on the importance of family and interpersonal violence at all levels of medical education. However, unless we seriously invest in the full spectrum of clinical violence prevention and intervention opportunities, we may find that victories over traditional categories of disease will turn to bitter and ironic defeat as the modern epidemic of violence takes an ever larger toll on our communities.