In this issue, Heffner and colleagues [3] report the results of a survey performed to determine the procedures and practices of radiology departments in managing patients with DNR orders. A questionnaire sent to 407 radiology departments resulted in 248 (61%) replies. Only 18.5% of departments had written DNR protocols. Medical chart review was the most common method used to establish DNR status. Only 70% of departments strictly honor the DNR order, and 38% of departments had resuscitated patients with DNR orders. The authors conclude that most radiology departments do not have formal procedures to prevent unwanted resuscitation and that DNR orders are frequently overruled.

Before the survey, the authors had expected that established protocols would exist in the radiology department to ensure that DNR orders are respected. The survey results clearly show that this is not the case in many departments. The authors postulate that lack of familiarity with end-of-life ethics and lack of debate in the radiologic literature on DNR orders contribute to radiologists' willingness to disregard such orders.

It is true that debate is lacking: Only two articles have been published recently in the radiologic literature [4, 5]. This is largely because cardiorespiratory arrest is uncommon in the radiology department. Even very ill patients are usually stabilized before transfer to the radiology department. The radiologist's attitude to end-of-life ethics is a more complex question. Factors that may contribute to overriding a DNR order in the radiology department include absence of an established physician-patient relationship, fear of allegations of medical negligence, skepticism about the applicability of the DNR order to the circumstances of the cardiac arrest, and concern that the episode may be iatrogenic [4]. However, Heffner and colleagues show that the most common reason that a patient with DNR status undergoes resuscitation in a radiology department is lack of awareness of the patient's DNR status.

The article leaves several questions unanswered. The outcome of attempted resuscitation in the 38% of departments in which it occurred is unknown. Cardiopulmonary resuscitation may be briefly commenced in a patient with DNR status until the DNR status is clearly established with the patient's primary attending physician; resuscitation may then be discontinued. Given the evidence in the medical literature that the content of DNR orders is frequently miscommunicated among the primary attending physician, intensive care unit nurses, and housestaff [6-8], it is not surprising that many radiology departments initiate cardiopulmonary resuscitation until the DNR status is absolutely certain.

A recent example from my department illustrates these complex issues. The intensive care unit nurse who accompanied a patient to the radiology department was not aware that her patient's DNR status had been rescinded. When cardiorespiratory arrest occurred, resuscitation was attempted, but only because the radiology team had double-checked the patient's DNR status and discovered that it was no longer valid. Most radiologists have had to cope with failure of communication [9] and will err on the side of caution from personal experience.

The patient with DNR status who experiences cardiorespiratory arrest in the radiology department while awaiting chest radiography poses an ethical problem for the radiologist that differs greatly from that of the patient who experiences severe but treatable allergic reaction to contrast media during computed tomography. The former patient is probably experiencing a consequence of the natural course of disease; the terminal event just happens to occur away from the bedside. The latter patient is experiencing an iatrogenic complication that was not caused by the underlying illness and was clearly not the event for which the DNR status was intended. One criticism of Heffner and colleagues' study is that it does not account for the different responses that these two situations are likely to engender.

Clearly, resuscitating a patient against his or her wishes is a lapse in good patient care. However, strict adherence to the DNR policy may not always be the right approach. Some radiologic procedures have specific risks related to their invasive nature (for example, nephrostomy placement) or requirement for conscious sedation or intravascular contrast. Following a similar logic, it has been accepted practice in some hospitals for surgeons and anesthesiologists to suspend DNR orders in the operating room [10]. In fact, in a 1991 editorial in The New England Journal of Medicine, Cohen and Cohen suggested that "Whenever patients have a substantial change in therapy or enter new treatment settings with different objectives and methods, the DNR designation should be reviewed" [11].

The current position of radiologists who do invasive procedures is similar to that of anesthesiologists surveyed by Clemency and Thompson 6 years ago [10]. Sixty percent of their respondents assumed that DNR orders were suspended in the operating room and during the perioperative period. However, since publication of that article, there has been important discussion in the anesthesiologic literature [11-15], and opinion is growing that it is inconsistent to override a DNR order simply because the final precipitating factor in a cardiorespiratory arrest may be iatrogenic [12]. This debate in the anesthesiologic literature has yet to be mirrored among radiologists, but if we are to learn from the experience in the operating room, it is clear that the blanket practice of suspension of or compliance with DNR orders in the radiology department does not take into account the circumstances of the individual patient.

Thus, one can argue that DNR orders should be stratified. At one extreme is the patient who does not wish to be resuscitated, regardless of the type of event; at the other is the patient who wishes to be resuscitated from all iatrogenic events. The referring physician should include a discussion of possible events in the radiology department, endoscopy suite, or operating room with any patient considering DNR status who might spend time in any of these departments. To ensure that these orders are communicated and observed, such directives should be clearly displayed in the patient's hospital records and included on requisitions for studies or procedures that take the patient from the bedside. All hospital departments that provide patient services away from the bedside should have formal protocols for dealing with the patient's DNR status.

All physicians referring patients for examinations or procedures should appreciate the possibility that a terminal event may occur elsewhere in the hospital. The receiving department must also ensure that the wishes of a patient with DNR status are respected while making the safety of the ill patients who enter their care a paramount concern. Heffner and colleagues have shed light on a hidden corner of end-of-life ethics in the hospital setting and set a challenge for us all to meet. Whether in the coronary care unit or the radiology department, in the surgical intensive care unit or the physiotherapy department, the patient has a right to die with dignity.