The main finding of this study is that in a sample of Swedish physicians as well as the general population, more respondents were willing to offer a new, expensive treatment to a non-smoking lung cancer patient than to a smoking patient with the same disease. Hence, a statistically significantly larger proportion of the studied physicians make a priority decision that seems to be in conflict with the official values expressed in the ethical platform for Swedish priority setting. However, there was no association between perceiving that the smoking patient was responsible for her lung cancer and the inclination to offer her treatment. This is compatible with, and could even be interpreted as indicating support for, the ethical platform. It seems as if the two findings contradict each other: if physicians fully accept the ethical platform and that the principle of responsibility has no place in Swedish priority decision, they ought not discriminate between the smoker and non-smoker in this case.
Our data do not elucidate why this inconsistency of attitudes occurred at group level. One possibility is that a corresponding inconsistency also exists within individual respondents. This thought draws strength from a comparison with studies of peoples’ juridical intuitions, where it has been demonstrated that some people favour more stringent punishment while simultaneously opting for shorter sentence length in case settings [21]. This could be interpreted as a kind of mental compartmentalisation: perhaps people do not realise they are expressing responsibility-based intuition when asked about how to choose in a specific case, even if that is what they are doing. Therefore they can say they accept the official values expressed in laws, platforms, and guidelines, but still make case judgements that are at variance with the same values. Putting it in Rawlsian language: they have failed to achieve a reflective equilibrium in their value system.
The other possibility does not imply failure of reflective equilibrium on an individual level, but rather that the respondents discriminate against the smoker for some reason other than the perception that the she is responsible for her illness. Even so, this reason must be related to the fact that she is a smoker as this is the only difference between the two versions of the questionnaire, and it still contradicts the ethical platform, but not necessarily by supporting the responsibility principle. With either possibility, what remains to be explained is why some physicians harbour such forceful anti-smoker sentiments as to effectively overrule their stated disinclination to let the issue of responsibility for illness govern the decision to treat. More on this later.
In our study, oncologists were more inclined than GPs to offer the new, expensive treatment, regardless of the patient’s smoking status. This might have to do with the difference in perspectives and interests between the two groups of physicians. GPs seldom treat terminal cancer patients and, as a general rule, seldom prescribe very expensive types of medication. Indeed, among the comments from GPs in our material, many stated that tax-payers’ money ought preferably to be used for prevention or in the treatment of curable diseases.
In their comments GPs also expressed that since the patient derives no long-term benefit from the treatment, her needs-based indication for treatment is weak. In the academic discourse, this interpretation of medical need is called the capacity to benefit perspective [22]. Although not uncontested, this view is not in apparent conflict with the Swedish priority setting guidelines [5].
Oncologists, on the other hand, have quite a different professional perspective than the GPs, as they are the ones who actually work with this group of patients. Thus, they are closely familiar with the kind of treatment described in the questionnaire. Also, oncologists participate more often than GPs in the development of these kinds of new drugs and drug regimens. Indeed, some comments from oncologists in our material illustrated that they think that when a new treatment exists, it should be used. As one (oncologist) respondent rhetorically stated: why develop new drugs if we aren’t going to use them?
Interestingly, pulmonologists differed from GPs, oncologists and the general population, in that they did not discriminate between the smoking and the non-smoking patient in their inclination to treat. Perhaps this is due to the fact that pulmonologists are very used to working with smoking and ex-smoking patients from their experience with COPD patients. Indeed, several comments from pulmonologists indicated they sought to put the question of responsibility in perspective by discussing the role of the tobacco industry, tobacco addiction and genetic vulnerability.
The above findings are in apparent contrast to the study by Neuberger et al where GPs, gastroenterologists and the general population were asked to rank criteria for patient selection in the case of liver transplantation [15]. In that study only the gastroenterologists – who work just as closely with potential liver transplant patients as oncologists do with lung cancer patients – stated that alcohol consumption should be rated as one of the three most important criteria. However, this difference could very plausibly be explained with reference to consequentialist reasoning. Our study focused on a marginally life-prolonging medication for a dying lung cancer patient, unlike Neuberger’s potentially curative treatment where the gastroenterologists might be concerned with the effect of continued alcohol consumption on the transplant survival.
Regardless of which version of the questionnaire they had received, physicians were less inclined to offer the treatment than were the respondents from the general population. This might indicate physicians have a greater sense of economic awareness regarding medical treatment, or result from the fact that physicians are more used than lay persons to the thought of restricting access to treatment [16].
Interestingly, we found that smoking respondents in the general population were even less inclined to offer treatment to the smoking case patient than were the non-smoking respondents. On first thought, one might expect the smoker subgroup to have been more lenient towards smoking patients, especially since this would be in their own best interest. However, there is a body of literature suggesting a “self-blaming” attitude among smokers that could explain this observation [23].
Why, then, is there such a negative bias against smoking? Obviously, the many health risks, even outside of lung cancer, play a role. Furthermore, comments in the present material indicate that respondents perceive smoking to be a personal choice, of the kind for which the individual should bear the consequences. One further reason for anti-smoking sentiment is based upon notions of (economic) solidarity. This view was also expressed in many comments, but none put is as succinctly as American philosopher Daniel Wikler: “The person who takes risks with his own health gambles with resources which belong to others” [24].
Implications of the study
If a subgroup of Swedish physicians use smoking status as a criterion in priority setting, as indicated in the present study, this is in violation of the intentions of the Swedish guidelines for priority setting in health care. Interestingly, this violation appears to stem, not from physicians wholeheartedly embracing the responsibility principle – as no association was seen between perception of responsibility for illness and inclination to treat – but possibly from lack of reflective equilibrium or from anti-smoker sentiment as described above. In order to resolve such inconsistencies or anti-smoker sentiment, in-depth discussions among Swedish physicians about the guidelines and their implications are probably needed. Also, observations of clinical practice are needed in order to determine whether these expressed values influence real-world priority setting. Furthermore, we describe a new method of inquiry into case-based decision making that can reveal controversial opinions even in the face of clear official guidelines. It might very well prove fruitful to apply this method to other harmful activities than smoking. Broadening the perspective, we think other questions in applied medical ethics could be addressed using this method.
Strengths and limitations of the study
The obvious strength of this study lies in the fact that it was conducted within the framework of an experimental, randomised and controlled trial, placing it higher in the evidence hierarchy than previous studies in this field. It was this experimental design that enabled us to reveal the difference between case-specific and official values and norms described above. Furthermore, as the randomisation process resulted in comparable groups regarding relevant aspects, the design minimised the risk of bias.
However, due to the relatively low response-rate among physicians, we cannot know to what degree the results are generalisable. Comments made by the respondents indicate that the physicians found the case description over-simplified, which could be a reason for the low response rate in that group.
The framing of the patient case in the questionnaire, in regard to the cost and expected effect of the new treatment, is also likely to have influenced the overall proportions of both physicians’ and the general populations’ inclination to offer treatment in both cases. Had the cost of the treatment been explicitly stated (and set rather low), more responders would probably have offered treatment – and vice-versa. However, the absolute proportions of those inclined to offer the proposed treatment are not our primary focus. Rather, the core issue is the demonstrated difference across the studied groups regarding inclination to treat the patient depending upon her smoking status.
Among the general population in our sample, there were only 5% current smokers. According to healthcare reports we would have expected rates of around 12-13% [25]. One smoking respondent from the general population stated that, to her, the questions were too sensitive to answer. If that feeling is shared by many smokers, it could help to explain the low inclusion of smokers. Also, there are well known socio-economical associations between smoking and education that render smokers less inclined to answer any questionnaire.