Call me crazy, but I enjoy emergency psychiatry. How else can I explain taking on the role of director of emergency psychiatry at a busy university teaching hospital after 22 years as a residency director? During those years my clinical work was almost entirely in an outpatient setting, much of it devoted to psychoanalytically informed psychotherapy. Perhaps it's not surprising that many of my colleagues expressed a mixture of shock and bemusement when they learned that I had taken this job.

To be fair, I have to admit that emergency psychiatry isn't all that I do. I still see outpatients, and I continue to treat medical students and residents as director of the medical school's mental health service for trainees. But I spend the better part of every day at the hospital in our emergency psychiatry service.

After 16 months, several aspects of this work strike me. First, I make use of my skills as a psychotherapist every day. In the comparatively uncontrolled environment of the emergency psychiatry service, it is useful to be aware of process issues, of multiple possible meanings, and of transference and countertransference. Empathy and the timing and "dosing" of verbal interventions can be as critical in this setting as in the psychotherapist's office.

Another striking thing about this work is how emotionally exhausting it can be. Despite the feelings of satisfaction I experience when a diagnosis has been nailed down or an especially good referral or disposition has been arranged, the continuous lack of continuity is draining. Every interview combines the excitement and anxiety of the first encounter with the knowledge that the first session with the patient is also the last.

The other thing that has struck me—pun intended—is the risk of injury in this work. Although the stereotype of the psychiatric patient as dangerous and violent is one of many examples of stigma, it is true that some patients really are agitated, volatile, and, at times, assaultive. Not surprisingly, a psychiatric emergency facility sees a disproportionate share of these patients. Lately I've been reflecting on this reality after experiencing an assault by a patient that resulted in a laceration and stitches, but which could very easily have been far more serious. I've been asking myself why and how this happened to me.

This was not the first time I was attacked by a patient. I have experienced three assaults over the past 28 years, all of which occurred in emergency settings. Another thing these occurrences had in common was that in each case the patient was a woman. However, I believe that the key link among the three assaults was a lapse in my application of one or more psychotherapeutic skills or principles.

My first assault—and the only one for the next 25 years or so—occurred when I was an intern at a county hospital in Los Angeles. It was a classic case of poor judgment born of misplaced confidence. After six to eight months of internship I was in my second three-month stint of psychiatry. The patient, a middle-aged African-American woman, was accompanied by two ex-husbands, one on each side of her. I failed to appreciate the significance of this arrangement and asked the former husbands to have a seat while I took the patient into an office to conduct my interview. It wasn't long before the patient had incorporated me into her paranoid psychosis. More quickly than I would have thought possible, she was out of her chair, across the space that had separated us, and clawing at my face. Fortunately, I came away with nothing more than superficial scratches—and a more humble sense of how much I had yet to learn. (Psychotherapeutic principle: therapeutic zeal alone cannot overcome psychosis; it must yield to common sense and prudence.)

The other two assaults both occurred during the past year. In both cases I let the interview continue too long: the patients were showing signs of increased agitation before they assaulted me. Although the two cases had this common characteristic, they were quite different in other respects.

The first patient had been attending a psychiatric clinic intermittently and was referred to the emergency psychiatry service to be evaluated for hospitalization because of suicidal ideation. She was enrolled in a master's-level social work program and spoke with some insight about the stresses in her life. Notably, she said nothing about feeling suicidal over the course of a 30- to 40-minute interview despite describing other symptoms—most prominently, anxiety—in considerable detail. Thinking that hospitalization might be avoidable and that she would be better served by being able to continue attending her university classes, I began to discuss alternatives to admission. At this point she began to allude to "what might happen" if she was not admitted.

It is not uncommon for interviews in the emergency psychiatry service to take such a turn, and ordinarily I respond with seriousness and concern, even if or when I feel I am being maneuvered by the patient. (I don't use the word "manipulated," because I take that to imply getting what you want by subtle means, and subtlety is most often lacking in these circumstances.) However, in this instance I responded in a manner that was, at least in retrospect, provocative. I smiled in response to the game-like quality of her rather blatant upping of the ante and stated that I felt it would be unfortunate if she harmed herself but that she needed to take responsibility for her actions. (Psychotherapeutic principle: maintain a neutral, nonjudgmental stance.)

Things went downhill from there. The patient took offense at what she perceived as my amusement. I found myself thinking, "How can she, as a mental health professional, act in a such a transparently immature fashion?" Of course, what I should have realized was that this "immaturity" was an important part of why she was in the emergency psychiatry service in the first place. It is clear that by that point I had lost sight of the patient as a patient. (Psychotherapeutic principle: feeling irritated with a patient may signal a loss of empathic attunement.)

The patient began to shout and grow visibly restless. I revised my plans and decided to admit her to our observation unit, but before I could explain this to her, she had come out of her chair and taken a swing at me with her forearm. Staff from the emergency psychiatry service came quickly into the room and grabbed her, but as she struggled she kicked me in the thigh. It took several security officers to restrain her. In her distress she screamed that she was being raped. I learned only the next day that she had previously worked at a local crisis facility but had been fired for unprofessional behavior.

The most recent episode, like the first one, involved a paranoid, psychotic patient. Like each of the previous attacks, part of the reason things went badly was that I misread certain aspects of the clinical presentation. Although I didn't dismiss the presence of two ex-husbands or ignore signs of borderline personality traits, I did underestimate the dangerousness of a patient who had been sent to the emergency psychiatry service for that very reason. I also paid the price for not being more insistent about the safety of my working environment.

A week earlier a computer server, about the size of a waist-high refrigerator, had been moved into a corner of one of our interview rooms to make more space in an office that two staff members had begun to share. I told our administrator that having the server in a clinical space was unacceptable, and she assured me that a work order would go in to have it moved. Nevertheless, on the Tuesday in question the server was still in the room. The effect was to reduce the space between the patient and me.

This patient, a woman in her 30s with a diagnosis of paranoid schizophrenia, had been sent to the hospital because of threatening behavior. However, when she arrived, although she spoke somewhat loudly, her demeanor was quite friendly: I had seen her being released from the ambulance gurney chatting with our staff, lying on her side like a patrician Roman woman at a banquet in the age of Tiberius or Nero. She was calm when interviewed by the social worker and sat quietly in the waiting room before I interviewed her. (Psychotherapeutic principle: initial impressions can be misleading; patients are often different from how they first appear.)

Initially I wondered what had prompted her referral to the emergency psychiatry service. She spoke calmly and with a composure that was, frankly, unexpected. However, she became increasingly emphatic that she did not have schizophrenia and certainly didn't need any of "those" medications. As I tried to explore more fully what had gone on during the previous day or two before, her animation and agitation escalated, as did the volume of her voice. The degree of volatility and emotional lability reflected the fact that she had not been taking those medications. (See earlier principle about psychosis.)

As I was concluding the interview, having decided that hospitalization was not avoidable, she kicked her leg out at me. Although she missed my leg, she connected with the clipboard sitting on my lap: it flew up and clipped me above my left eye. (Without the server in the room, she would have been just far enough from me that her leg would not have reached the clipboard.) (Psychotherapeutic principle: the therapist is responsible for "the frame," and a safe, sound frame protects both the patient and the therapist.)

My face stung, but it took a few moments before I saw blood dripping on the desk and realized that I had been cut. Meanwhile, the patient was rapidly becoming even more wildly out of control and had to be restrained, secluded, and medicated. After her safety had been secured, I finally looked in a mirror: it was quite a sight, a cut like a Nike swoosh just above and slightly left of my left eye. My colleague in the medical emergency department put in 11 tiny stitches before I went home.

As with the other instances of assault, my behavior had contributed to my being injured: I let the interview go on too long, and I should not have used space that had become less safe. However, what sticks with me most—now that my face has healed and the scars have faded behind my eyebrow—is how damaging it is when patients lose control, especially when it results in visible injury to others. For psychotic patients in particular, this loss of control weakens the boundary between reality and fantasies of omnipotent destructiveness. So the final psychotherapeutic principle is that preventing assaults by patients is not only in our self-interest, it's good patient care as well.

Dr. Strauss is professor and director of emergency psychiatry in the department of psychiatry and behavioral science at the University of Louisville School of Medicine, 601 South Floyd Street, Suite 210, Louisville, Kentucky 40202 (e-mail, gdstra01@gwise.louisville.edu). Marcia Kraft Goin, M.D., Ph.D., is editor of this column.