| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Guest Access | Sign In via User Name/Password
|
|
|
|||||||
Guest Access | Sign In via User Name/Password |
|||||||||
Elkins Chest Clinic, Elkins, WV
Correspondence to: Jerome Arnett, Jr, MD, FCCP, Elkins Chest Clinic, PO Box 1926, Elkins, WV 26241
To the Editor:
I read with interest the article on ethics in end-of-life care by Kelly et al in CHEST (March 2002).1 Their objective was to determine whether the strength of do-not-resuscitate (DNR) recommendations varies with medical specialty and experience. Their conclusion, that the strength of DNR order recommendations varies with different internal medicine specialties and with different levels of experience, is not supported by their data.
None of the seven groups in the study varied significantly in the number of DNR orders recommended, so that all groups demonstrated a similar approach to this end-of-life issue. Only three findings were statistically significant, each in only one group of physicians, and these involved only strength of opinion, not the number of DNR orders recommended. One of the three significant findings, that the more senior house staff recommended DNR more strongly than did the younger interns and junior residents, may reflect increased confidence with increased length of training.
The authors’ use of statistics may have misled them. Statistics, which is concerned with correlations, can be applied to any problem but is not sufficient to show causation, which is the cardinal function of scientific research.2 The improper use of statistics often results in what has been termed "statistical malpractice."2 Findings that are not statistically significant should not be reported as a trend.
Cultural influences are important when dealing with end-of-life issues. The new field of bioethics has created an ethics that places the needs of society and third parties above those of the individual patient.3 Because of this, the attitudes and behaviors of physicians toward the terminally ill that were reported in the 1970s (references 10 and 11 in the article by Kelly et al1 ) may not be comparable to those from recent years.
Finally, how this study will "help to target educational interventions and ... ensure effective collaboration with colleagues and communication with patients," as the authors claim in their conclusions, is not explained.
References
Walter Reed Medical Center, Washington, DC
Correspondence to: William Kelly, MD, Walter Reed Medical Center, Pulmonary/CCM, Ward 77, Washington, DC 20307; e-mail: williamkellymd{at}aol.com
To the Editor:
Dr. Arnett’s interest in our publication is appreciated. We again acknowledge that the number of do-not-resuscitate recommendations was not different between subspecialist groups. However, the purpose of the study, as stated in the introduction, was to determine whether the strength of do-not-resuscitate order recommendations, not the absolute number of decisions, varied with medical subspecialty and years of training. Our results did indeed find statistically significant differences in the degree of these convictions. This is important and is of interest to internists in general and chest physicians in particular. We believe that the strength of physicians’ convictions affects their guidance to patients who are making end-of-life decisions.
Dr. Arnett charges statistical malpractice by the confusion of correlation with causation. In our study, we never claim to show any causation. We only report observations from our limited database. Our statistical significance does add greater clarity to these findings by suggesting they are not a result of chance. Additional findings that approached but did not meet significance are so disclosed with p values and statistical methods.
The specific differences that we found among medical subspecialties are consistent with the results of other reports in the medical literature. Dr. Arnett suggests that such references may be out of date (ie, references 10 and 11) but fails to note our citation of this same subspecialty bias in physician actions (reference 9) and in end-of-life publications over the subsequent 20 years (reference 18).
We propose that understanding and respecting subspecialty differences in end-of-life opinions may advance us toward more effective collaboration with colleagues and more effective communication with our patients. We hope that we have shed some light on these differences and invite readers to generate their own hypotheses about causation.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
