We appreciate Dr. Mylonakis' comments and agree with the recommendation that radiologists discuss end-of-life issues with their patients before invasive procedures. Management of DNR orders in the operating room is analogous in some respects to the transport of patients with a DNR status to the radiology department. Recent guidelines emphasize the importance of honoring DNR orders or negotiating with patients temporary DNR modifications during surgical procedures (1, 2). All recommended options emphasize the importance of physician-patient discussions that clearly establish the planned responses to life-threatening events before transfer to the operating room (3, 4). Our study indicates that the surveyed radiology departments rarely informed patients of departmental resuscitative practices even though 24% of programs would overrule DNR orders. Only 8.3% of programs that overruled DNR orders advised their patients of this practice.

Unfortunately, other investigations indicate that sole reliance on physician-initiated discussions is unlikely to safeguard patients' end-of-life wishes. Most physicians, whether they are radiologists or primary care clinicians, do not discuss with their patients the likely outcomes of resuscitative interventions or explore their patients' life-sustaining care preferences (5). It appears that policies, physician prompts, local "issue champions," alterations of institutional culture, and other yet to be designed interventions are needed to facilitate patient–physician communication. Our study noted that the presence within radiology departments of a formal DNR protocol increased the likelihood that patients would be informed about resuscitative policies.

Our study, however, did not indicate that low compliance with DNR orders applied only to patients undergoing invasive procedures. Any patient transported to the radiology department was at risk for unwanted resuscitation. This conclusion was supported by the observation that unawareness of patients' DNR status was the most commonly cited reason for performing CPR.

In the final analysis, our study was performed within radiology departments for reasons of feasibility, but we suspect that its observations apply to other locations within medical centers where patients with DNR orders are transported. Clearly, physicians should anticipate life-threatening events during patient transport and discuss with their patients the appropriateness of resuscitative responses in different circumstances. It appears, however, that maintenance of patient autonomy at the end of life may depend on institutional approaches to communicate and implement patients' wishes across all areas of care. Otherwise, it may be unsafe for our patients who decline resuscitative care to leave their hospital rooms.