We are concerned that additional difficulties may interfere with the implementation of DNR orders when patients are transported to hospital locations away from their rooms. The hospital personnel temporarily responsible for a patient's care may have insufficient knowledge of DNR practices and limited information about a particular patient's DNR status to appropriately respond to life-threatening events. Hospital-based radiology departments represent a model for examining these concerns.

We administered a questionnaire to radiology department personnel to evaluate the extent to which formalized protocols are used for managing patients with DNR orders, how the DNR status of transported patients is established, and how radiology personnel respond to cardiopulmonary arrest in patients with DNR orders. The effects of formalized protocols on processes of DNR care were also examined.

Methods

Top Methods Results Discussion Author & Article Info References

Selection of Radiology Programs

We selected all 203 radiology programs in the United States that had an accredited radiology teaching program listed in the 1997-1998 edition of the Graduate Medical Education Directory [6]. An additional 204 radiology programs were included by using a random-number list to select hospitals registered in the 1996-1997 edition of the American Hospital Association roster [7], stratified by state and Washington, D.C. Radiology departments were sent an initial letter introducing them to the study, followed by a second mailing that included a cover letter and questionnaire. Nonresponding departments were sent a follow-up letter and, if necessary, a second follow-up letter that included a second copy of the questionnaire. The protocol was approved by the institutional review board of St. Joseph's Hospital and Medical Center, Phoenix, Arizona.

Questionnaire

The 10-item questionnaire assessed practices in managing patients with DNR orders (Appendix). Four general internists who did not have special expertise in DNR issues reviewed the questionnaire for clarity of language and meaning. In interpreting responses to item 3 of the questionnaire, review of the medical record for establishing a patient's DNR status was considered an unstructured method of communicating this information. Other item 3 choices and write-in responses that defined a specific method designed to promote the accuracy of DNR status communication were considered to be structured DNR communication methods.

Literature Search

A MEDLINE search was performed for 1987 to 1997 to identify articles on topics pertaining to DNR issues that were published in the core radiology journals listed in Index Medicus [8]. Exploded search terms included ethics, medicine; medical futility; advance directives; living wills; decision making; ethics committees; euthanasia, passive; life support care; terminal care; and right to die. Retrieved articles were reviewed to determine their primary content.

Statistical Analysis

Data were analyzed by using JMP software (SAS Institute, Cary, North Carolina). Descriptive statistics were generated with 95% CIs. Differences between groups with categorical variables were analyzed by using the chi-square test, a two-sided Fisher exact test, or an R x C contingency Table when appropriate. Odds ratios and adjusted odds ratios were calculated from logistic regression models. Differences with P values less than 0.05 were considered significant.

Results

Top Methods Results Discussion Author & Article Info References

We received 248 evaluable questionnaires, representing a 61% response from 50 states and Washington, D.C.; a median of 4 questionnaires (interquartile range, 2 to 6; range, 1 to 16) came from each jurisdiction. Characteristics of the respondents and their institutions are shown in Table 1. Each questionnaire item was completed by 246 to 248 respondents.

Responses to questionnaire items 1 to 6 (Appendix) are shown in Table 2. Radiology departments with a written DNR protocol (item 1) were more likely (odds ratio, 3.2 [CI, 1.5 to 6.5]; P = 0.001) to have a formal procedure for receiving DNR status information (item 2) from ward and intensive care unit nursing staff (35.6% [CI, 21.6% to 49.5%]) than departments without a DNR protocol (14.8% [CI, 9.9% to 19.7%]).

Of the 248 departments, 133 (53.6% [CI, 47.4% to 59.8%]) used one or more of the structured methods listed in Table 2 other than medical record review for establishing a patient's DNR status (item 3). Use of a structured method was more common in departments with a written protocol for managing DNR orders (73.9% [CI, 61.2% to 86.6%]) than in departments without such a protocol (49.3% [CI, 42.3% to 56.2%]) (item 1) (adjusted odds ratio, 2.25 [CI, 1.06 to 4.98]; P = 0.04) and was more common in departments with a formal procedure for receiving DNR status information (93.3% [CI, 86.0% to 100%]) than in departments without such a procedure (44.8%, [CI, 38.0% to 51.7%]) (item 2) (adjusted odds ratio, 15.34 [CI, 5.32 to 65.03]; P < 0.001).

The existence of a written protocol for managing DNR orders (item 1) or receiving DNR status information (item 2) did not predict how departments would respond to DNR orders (item 4) or whether they had previously attempted resuscitation of a patient with a DNR order (item 5). Ten of 174 respondents who indicated that cardiopulmonary resuscitation would be withheld from a patient with a DNR order (item 4) also indicated that cardiopulmonary resuscitation would be initiated if cardiopulmonary arrest occurred during an invasive procedure. This response increased the number of respondents who stated that their departments would initiate cardiopulmonary resuscitation to 60 (24.2% [CI, 18.9% to 29.5%]). Only 5 respondents (8.3% [CI, 1.3% to 15.3%]) from these 60 departments indicated that patients were informed that their DNR orders might be overruled.

Commonly cited reasons for performing cardiopulmonary resuscitation in patients with DNR orders (item 5) were unawareness of the DNR status in 61 cases (64.2% [CI, 54.6% to 73.9%]) or occurrence of cardiopulmonary arrest during an invasive procedure in 19 cases (20.0% [CI, 12.0% to 28.0%]).

In post hoc univariate analyses, hospital size, location, teaching status, and existence of a hospital ethics committee were not significant explanatory variables for any of the questionnaire items. The literature search identified two articles published from 1987 to 1997 that pertained to end-of-life ethics [9, 10].

Discussion

Top Methods Results Discussion Author & Article Info References

More than 80% of surveyed radiology departments do not have written protocols for managing patients with DNR orders. Consequently, radiology personnel respond to cardiopulmonary arrests in various ways; 24% of respondents initiated cardiopulmonary resuscitation for patients with DNR orders and almost 40% stated that they had resuscitated such patients previously.

It has been noted that no previous investigations have examined how radiology personnel interpret and respond to DNR orders [9]. Our MEDLINE search found only two articles published in the last 10 years in core radiology journals that pertained to ethical issues of DNR decisions [9, 10]. In 1996, Terry [10] found two articles published in radiology journals on subjects related to cardiopulmonary resuscitation and radiologic procedures [11, 12]. This limited dialogue within radiology journals, which has been noted to vary among medical subspecialties [13], may contribute to an insufficient awareness of DNR issues among radiology department personnel and our observation that DNR orders were frequently overruled.

Although our study is limited by the 61% return rate of the mailed questionnaires, we do not believe that nonresponder bias caused substantial underestimation of the existence of DNR protocols or overestimation of the frequency with which DNR orders were overruled. We doubt that departments with a special interest in DNR topics, as demonstrated by their adoption of DNR protocols and adherence to DNR orders, would have been less likely to respond to our multiple mailed invitations to participate than programs without such interests.

An additional limitation was our decision not to lengthen the questionnaire by adding queries about whether respondents would vary their resuscitative responses according to whether patients were undergoing invasive or noninvasive procedures. Interrogative studies indicate that 60% of anesthesiologists assume that DNR orders are suspended while patients undergo surgical procedures [14]. Radiologists may similarly assume that patients would choose resuscitation from an iatrogenic event during an invasive procedure because such events may be reversible [9]. The legitimacy of overruling DNR orders in iatrogenic settings, however, gains little support from recent discussions of this issue [5]. Moreover, consensus exists that physicians who choose to suspend DNR orders should do so only after informing patients and negotiating changes with them [15, 16]. Our study indicated that surveyed radiology personnel rarely informed patients of departmental DNR practices even though DNR orders were frequently overruled.

The reliance on the medical chart to establish DNR status that we observed is problematic because of the ambiguity of DNR orders. Previous studies indicate that DNR orders are often inadequately communicated among attending physicians, intensive care unit nurses, and housestaff physicians [1-3]. This inaccurate communication persists, albeit to a lesser degree, with the use of structured DNR order sheets and computer-based DNR communication procedures [17]. The inadequacy of the medical chart as a source of DNR information is supported by our observation that unawareness of the DNR order was the most commonly cited reason that patients with DNR orders underwent cardiopulmonary resuscitation.

We examined radiology departments only, but we believe that similar findings would be observed in other hospital areas to which patients are temporarily transported. Considering that 10% to 14% of hospitalized patients have DNR orders [18, 19], new strategies are needed to ensure the accurate communication and intended implementation of DNR orders when hospitalized patients leave their rooms. Judging by the results of our study, departmental protocols by themselves have limited effects on processes of DNR care. Hospital-wide policies written to apply explicitly to all hospital locations and structured communication procedures, such as procedure-specific DNR order forms and preprinted DNR progress note forms [1, 4, 20], may represent additional improvements in care. Further research is needed to identify the most effective processes for enhancing hospital-wide understanding of DNR orders and ensuring their appropriate implementation in all hospital settings.

Appendix: Questionnaire Items

1. Has your radiology department developed a written protocol specific to your department that guides decisions about calling codes or withholding cardiopulmonary resuscitation for patients with DNR orders who develop cardiopulmonary arrest while in your department?

Yes or No

2. Is there a formalized procedure for verbally communicating a patient's DNR status to radiology personnel? An example of a procedure might require a ward or intensive care unit nurse to call the radiology department about DNR status ahead of a patient's transport for a radiographic procedure.

Yes or No

3. How do radiology personnel most often gain an understanding of a patient's DNR status when a patient is transported to the department? Circle all choices that apply:

a) Department personnel review the medical chart that accompanies the patient to the department

b) The patient has a wrist band or other personal identifier of DNR status

c) A special DNR form accompanies the patient to the department

d) A ward or intensive care unit nurse calls the department

e) Other (please describe)

4. Consider that a patient transported to your department has a "full DNR order" in his or her medical chart requesting that all life support interventions be withheld if cardiopulmonary arrest occurs. What is usually done in your department if the patient develops cardiopulmonary arrest while undergoing a radiographic procedure? Circle one choice:

a) Cardiopulmonary resuscitation is initiated

b) Cardiopulmonary resuscitation is withheld and the physician is called

c) The code team is called

d) Other (describe actions)

5. Have patients with a "full DNR order" undergone cardiopulmonary resuscitation in your department?

Yes or No

If yes is answered, please state the reasons that life support is started.

6. Do patients with DNR orders receive any information about the written or unwritten policies and procedures of the department regarding the initiation of life support before they are transported to radiology for an imaging study?

Yes or No

7. What is your official position in the radiology department?

a) Staff radiologist

b) Physician-director of radiology department

c) Training program director but not physician-director of department (department head)

d) Other (please describe)

8. Does your hospital have an ethics committee?

Yes or No

9. Which of the following best describes your hospital?

a) State-supported university medical center

b) Tertiary care private medical center

c) Private community hospital

d) County-supported medical center

e) Veterans Administration medical center

f) Military medical center

g) Closed-staff health maintenance organization

h) Other (please specify)

10. What is the size of your hospital?

a) >1000 beds

b) 700-1000 beds

c) 400-699 beds

d) 200-399 beds

e) <200 beds

Ms. Barbieri: St. Joseph's Hospital and Medical Center, 350 West Thomas Road, Phoenix, AZ 85001.