Discussion

Validity aspects

To achieve the objectives of the study, we chose a vignette-based questionnaire, as it encourages participants to engage in ethical reasoning. The typical examples were designed to be easily understood, regardless of the responders' level of medical knowledge. However, the form was time consuming to fill out, which may have contributed to the somewhat low response rate. We found no significant differences in attitudes, age or sex between early and late responders, suggesting that a higher response rate would not significantly influence the response pattern.

We wanted to include some of the most frequently used arguments for/against the withdrawal of life-sustaining treatment in the questionnaire. In ethical discussions about this issue, the main arguments usually concern the right to autonomy versus the wrongness of killing. As the form became rather extensive, we reduced the number of pre-formulated arguments to leave room for responders' own arguments; a possibility that many made use of. However, another set of arguments could have given more information and possibly the selection of arguments directed the participants, thereby making them more likely to be singled out as important. That the withdrawal of treatment could be interpreted as euthanasia was used as a contra argument, although classifying an act as euthanasia does not necessarily imply that the act is considered morally wrong. However, the wording of the statement made it clear that, in this particular case, perceiving the act as euthanasia meant being against the termination of treatment. Moreover, respondents were allowed to classify the action as euthanasia in later questions, even if they prioritised a pro argument, as some did. Nevertheless, the wording of the statements may have influenced the participants' interpretations and responses. However, as it is the comparison of the cases that is in focus, the procedure might nevertheless be defensible. The alternatives for classification—a defensible act, assisted suicide, euthanasia, manslaughter, murder—were not explained, because they are widely used in public debate.

The group of physicians was randomly selected from all over Sweden. Male physicians were overrepresented (male/female 67/33%), also when compared with the demographic data of the average male/female ratio of physicians (male/female 60/40%).12 The age range of physicians (30–90 years) included 38 individuals over 67 years (pensionable age in Sweden). Whether these physicians were clinically active is unknown, but possibly some have continued to practice on a private basis. The representatives from the general public were randomly selected from the county of Stockholm, where Karolinska Institute is located. Geographical nearness can have a positive influence on the response rate, as many in the general public may have a relationship with the university, and therefore be more motivated to participate in the study. The balance of the sexes (49/51%) matched the demographic data of the average male/female ratio in the population in the county of Stockholm as well as in Sweden as a whole (49/51%, age group 18–100+ years).13 However, the population in the county of Stockholm is generally better educated compared with the national average, and therefore is not entirely representative. It could also be argued that people in major cities such as Stockholm are generally more liberal in their moral views; possibly, including representatives of the general public from a broader section of Sweden could have given more restrictive attitudes.

Factors influencing attitudes

In Sweden, a decision-competent patient has the right to refuse a life-sustaining treatment; however, not all participants recognised this right. For some, this may be because they mistakenly conceived the act as criminal (eg, euthanasia, manslaughter, murder). Others may have inferred that the patients in the vignettes were not competent due to psychiatric illness and that their requests should therefore not be respected. This may be the case especially in vignette 2, in which the patient had a history of depression. Such an interpretation is compatible with generally recognising the right of competent patients to forego life-sustaining treatments. Responders using a broad definition of suicide could also have understood the request for termination of the life-sustaining treatment as suicidal behaviour.14–16

The vignettes were appraised differently by physicians and the general public, but also by responders within the same group. This may be explained by the ethically relevant factors varied in the vignettes (ie, type of treatment, patients' age and competence), but also by background variables such as age, sex and speciality.

Type of life-sustaining treatment

Previous research has indicated that physicians have preferences regarding what type of life-sustaining treatment to withdraw; when asked to rank their preferences for the withdrawal of different treatments, haemodialysis was chosen before mechanical ventilation.17 18 A similar preference may explain why more physicians perceived the termination of dialysis in vignettes 1 and 2 as a defensible act, compared with the withdrawal of ventilator treatment in vignette 3. Why may the type of treatment to be withdrawn affect physicians' attitudes? A possible hypothesis is that death is imminent when terminating mechanical ventilation, as opposed to terminating dialysis. Also, dialysis may be terminated by the patient alone, simply by not showing up at the dialysis department. Withdrawing the ventilator treatment is different; to do so will cause severe foreseeable suffering in terms of suffocation, and therefore palliative measures such as sedation must be initiated even before symptoms arise—thereby requiring a physician's active involvement. Therefore, it may feel as if it is oneself as the physician, rather than the patient's condition, that is the direct cause of shortening the patient's life. As the Swedish Medical Association's ethical guidelines forbid doctors to participate actively in hastening a patient's death, the termination of ventilator treatment could be understood as being contrary to these guidelines.7 No matter how natural such feelings are, it is difficult to see the moral relevance of the time frame or the use of palliative sedation itself. As it is also difficult to see any other differences between terminating ventilation and dialysis that could be relevant for judging the one as more ethically defensible than the other, more exploration into the motivations underlying the differing judgements is needed in order to answer the question.

Contrary to physicians' reports, the response pattern among the general public does not suggest that the type of life-sustaining treatment is of importance. The proportion of the general public interpreting the withdrawal of life-sustaining treatment as euthanasia was constant in all three cases. Significantly fewer from the general public categorised vignette 2 as a defensible act compared with vignettes 1 and 3, indicating that attitudes are influenced by factors other than merely the type of life-sustaining treatment. This hypothesis finds some support in a study by Singer et al,19 in which end-of-life practice had only a moderate effect on the public's opinion towards end-of-life decisions.

Patient characteristics

It is likely that patient characteristics such as age, prognosis and competence influenced the attitudes. Significantly more physicians considered the withdrawal of dialysis in vignette 1 to be a defensible act compared with the other vignettes. In vignette 1 the patient was older, competent and capable of independently formulating her will. Although tired of life, she was not regarded as depressed. The patient was not a candidate for a kidney transplant and the dialysis treatment was therefore lifelong. The idea of patient characteristics affecting physicians' attitudes is supported by a Swedish study, in which the most important factors considered in the decision to withdraw life support in intensive care units were found to be the patient's likelihood to survive the current episode, patient advance directive, patient age and the likelihood of long-time survival.11

Why did fewer responders classify the termination of treatment in vignette 2 as a defensible act compared with the other scenarios? The analysis is complicated by the fact that the patient differed from the others in several ways. First, the history of depression created doubt regarding the patient's competence. Second, as a candidate for a kidney transplant, the prognosis was better than in the other vignettes. Third, the patient was much younger than the one in vignette 1. The arguments against the termination of dialysis in vignette 2 provided by the responders include: doubt regarding the patient's competence, young age, the risk of the patient changing his mind and the possibility of alternative medical treatments. However, the material does not allow us to determine whether these factors influenced attitudes.

That prognosis might be of importance is suggested by Singer et al,19 who found that patient prognosis had a major effect on the public's opinion towards end-of-life decisions. However, according to Singer et al19 the decision-making process only had a minor effect on the public's attitudes, a result that cannot be validated in the present study.

Background variables

Whereas age and sex were generally insignificant within the general public, they tended to influence physicians' responses. There was also an association between physicians' attitudes and the type of speciality. Psychiatrists classified termination of dialysis as defensible—or prioritised arguments in favour of this—less often compared with other specialities. A possible explanation may be that attitudes are influenced by previous experience of terminal care. Psychiatrists are fairly seldom directly involved in end-of-life decision-making. This hypothesis is supported by a large European study, according to which physicians who attended more than five terminal patients during 1 year agreed more with the duty to comply with a patient's request to withhold/withdraw a life-sustaining treatment than physicians who had attended fewer terminally ill patients.20 However, this does not explain why psychiatrists were more hesitant than physicians from other specialities in vignettes 1 and 2, but not regarding the ventilator-dependent patient in vignette 3. Nor is it a plausible explanation for why general practitioners—who are often involved in the care of chronically ill and/or dying patients—less often classified termination of dialysis as defensible, or prioritised arguments in favour of this, compared with other specialities. Possibly, differences in the care relationship between physician and patient may be of importance here; in general medicine and psychiatry it may be more common with long-term relationships compared with the other specialities included in the study. A more long-term relationship may influence the decision-making process, making it harder for the physician to accept a termination of life-sustaining treatment.

Hastening death and euthanasia

The statement that the withdrawal of treatment could be interpreted as euthanasia was used as a contra argument. Nevertheless, there were responders in both groups who classified the act as euthanasia, but still prioritised an argument in favour of terminating the treatment as the most important. This suggests that even a value-charged label such as ‘euthanasia’ does not necessarily mean the act being found ethically unacceptable. However, the result could also indicate that many responders are unaware of how euthanasia is usually defined in the medical field, that is as a doctor intentionally killing a person by the administration of drugs, at that person's voluntary and competent request.21 The fact that some responders in both groups classified the withdrawal of life-sustaining treatment as euthanasia in all three cases also suggests an association between the hastening of death, caused by the withdrawal of life-sustaining treatment, and the concept of euthanasia. The idea of an unawareness of how euthanasia is defined is supported by results from a Canadian study, in which opinions towards euthanasia differed significantly depending on the wording of the question.22 The implications of this explanation not only touch on the design of future studies, but also illustrate the need for clarity in public debate. With a unitary terminology, the misunderstandings frequently occurring in current discussions about end-of-life decisions may be prevented, thus enabling a constructive ethical discussion of the questions.