Our internet survey has several methodological problems. First, the publicity of the survey primarily targeted internist physicians in primary care settings. The distribution of the physician specialties shows that despite responses being obtained from a certain proportion of emergency care physicians and surgeons, the responses from internists comprised a large proportion. Therefore, the results of this study cannot be representative of the overall awareness of Japanese physicians on these issues. Additionally, this study was a web-based survey in which the responses were obtained from a homepage; this clearly indicates that the set of physicians accessing the homepage were not representative of the typical Japanese physician population . In all likelihood, the physicians who participated in the survey were largely physicians with an interest in treatment decisions concerning "life-sustaining treatment"; thus, the sample analyzed must necessarily include substantial bias.
We received a number of major suggestions from this survey despite the above-noted limitations. In the course of routine treatment, most physicians had personal experience of having to make difficult decisions like those presented in the three scenarios. Nonetheless, we observed discrepancies in judgment among physicians related to specific "life-sustaining treatment" in specific scenarios. Further, we found disparities among physicians with regard to whether such medical interventions were "life-sustaining treatment." For example, there were great differences in the respective proportions of physicians who regarded nutritional supplementation by tube feeding and respiratory assistance by an artificial respirator as interventions that "should" be undertaken for patients with a prolonged disturbance of consciousness. In such circumstances, we found that physicians demonstrated a greater resistance to the attachment of an artificial respirator than to the initiation of tube feeding, and the extent of withholding of such treatment was actually greater. As has previously been studied, we hypothesized in similar fashion that for physicians, the attachment of a respirator was an alternative to which they exhibited a greater resistance among potentially permanent treatments; this is because of the high invasiveness of this procedure [13,14]. In our case scenarios, approximately half of the physicians surveyed responded that tube feeding "should be initiated" in situations of judgment during the acute phase of an illness where the potential for recovery remained; however, approximately the same number of physicians responded that medical treatment "should not be withdrawn" in scenarios where more than half a year had passed, and the medical potential for recovery was extremely low. This result underscores the strength of the resistance to the withdrawal of treatment relative to that for the withholding of treatment [15,16].
Despite the fact that the results for descriptive analyses were virtually the same with regard to the withholding of treatment in Case 1 and the withdrawal of treatment in Case 3, the lack of high concordance in these responses suggests that there is no fixed consensus among physicians concerning the withholding or withdrawal of treatment. A greater number of affirmative opinions were obtained from physicians who possessed an experience in ethical consultations for the withdrawal of treatment for patients in whom the potential for recovery was extremely low, and the next of kin had requested the withdrawal of treatment. In other words, more affirmative opinions were obtained in Case 3; moreover, according to the general ethical principles, among our three cases, this case is understood to be the one in which the selection of withdrawal would be most valid . This result suggests that there is a need for ethical consultants, and that an experience in ethical consultation is effective for producing judgments of greater validity in end-of-life care. Simultaneously, the fact that no significant relationship was observed between judgment and preference with regard to either physician characteristics or hospital characteristics suggests that there is no decision model from which to undertake ethical instruction in the current physician environment .
The physicians participating in our internet survey took part voluntarily after encountering survey publicity, despite the lack of any financial incentive. We therefore hypothesized that this group of physicians had a higher awareness of ethical issues in medical treatment than our target population of Japanese physicians in general. However, even among this group, we found that there was negligible consultation with colleagues, conference studies, or other such activities addressing ethical issues. In particular, nearly all of the physicians had no experience of activities such as applications to specialists in clinical ethics or to ethics committees. This finding may suggest that the environment and culture that allows physicians to consult with other medical staff is currently limited.
Another finding from our research is the fact that there is more than a slight discrepancy between the "should/should not" awareness of treatment options and what is actually being done. In particular, in Case 3, very few physicians who believe that enteric nutrition "should not" be withdrawn actually withheld or withdrew medical intervention. This fact signifies that there is a high resistance to genuine action, which is distinct from the issue of whether the withdrawal of treatment is valid. Strong considerations include psychological resistance concerning the causation of death through intentional acts by the physician, and the contravention of legal norms [19,20]. The intent of the treatment providers was more distinct with respect to the withdrawal of treatment than to the withholding of treatment. Consequently, we believe that such intent resides in a perspective of physician responsibility, and that the psychological resistance engendered by responsibility creates a disconnect between judgment based on ethical validity and actual treatment decisions.
We believe that the current survey results point to a plan that should be undertaken to ensure that difficult decisions regarding life-support care in medical settings are made with greater validity. First, we discern a need for individuals or organizations to provide specific support for clinical decisions that encompass ethically complex elements. In practice, the accessibility of hospital ethics committees and clinical ethicists must also be enhanced. Currently, the matters considered by ethics review committees in Japan primarily concern research, and these bodies do not serve as organizations supporting clinical decision-making in actual clinics . Additionally, while specialists in medical ethics exist, an extremely limited number of personnel actually travel to treatment settings and are able to establish close communication with treatment staff and address the resolution of clinical problem on-site. Infrastructural investments in personnel should be made.
Second, there is a need to reach a certain degree of consensus regarding the conduct of ethical decision-making in end-of-life care taking into account of the tendency Japanese physicians' attitudes toward some different clinical situations; variations of specific treatments; withholding or withdrawal of treatment. The term "life-sustaining treatment" has generally been perceived as a negative image of a practice not commonly done; however, our research suggests that there is a great discrepancy as to whether specific medical interventions based on detailed scenarios constitute "life-sustaining treatment" even among physicians. A more detailed study is required on specific medical interventions, rather than that on the image projected by "life-sustaining treatment." In Japan, in particular, notwithstanding the presence of major confusion in treatment settings, we are currently far from a consensus of opinion on the ethical differences and equivalencies in the withholding and withdrawal of medical interventions. The withholding of treatment that should not be carried out and the withdrawal that is judged to be valid must be deliberated from a greater number of bases and perspectives.
Finally, in clinical matters where the consideration of ethical issues is strongly indicated, we look forward to clinical conferences and other efforts toward regular and active information exchange among medical personnel.