Skip to content

Should PREDICTED Smokers Get Transplants?

By Tom Douglas

Jack has smoked a packet a day since he was 22. Now, at 52, he needs a heart and lung transplant.

Should he be refused a transplant to allow a non-smoker with a similar medical need to receive one? More generally: does his history of smoking reduce his claim to scarce medical resources?

If it does, then what should we say about Jill, who has never touched a cigarette, but is predicted to become a smoker in the future? Perhaps Jill is 20 years old and from an ethnic group with very high rates of smoking uptake in their 20s. Or perhaps a machine-learning tool has analysed her past facebook posts and google searches and identified her as a ‘high risk’ for taking up smoking—she has an appetite for risk, an unusual susceptibility to peer pressure, and a large number of smokers among her friends. Should Jill’s predicted smoking count against her, were she to need a transplant? Intuitively, it shouldn’t. But why not?

Before tackling the case of Jill, let us return to the more straightforward example of Jack.

Most would be prepared to give Jack lower priority for a transplant if his past smoking is itself a negative prognostic indicator—if the damage already caused by his smoking makes the transplant less likely to succeed. Some would also accept that his history of smoking could be relevant in a further way: if Jack’s smoking caused his medical condition, then perhaps we can hold that Jack is morally responsible for his illness, and perhaps this weakens his claim to a transplant.

But there is also a third and less well accepted way in which Jack’s past smoking might justify giving him a lower priority for a transplant. Jack’s past smoking may suggest that he will continue to smoke in the future. Suppose this is correct: past smokers, including Jack, are more likely than others to smoke in the future, even after a serious medical diagnosis, and even after receiving a transplant. Suppose also that future smoking would make the transplant more likely to fail. Should these facts also figure in decisions about the allocation of organs for transplantation? Should we sometimes refuse people scarce medical resources on the basis of predictions about their future behaviour?

There are at least two reasons to do so.

First, future behaviour can clearly be relevant to the likelihood of treatment success. It may be indicative of a worse prognosis, and, as we have seen, genuine negative prognostic indicators are well accepted as a basis for rationing medical resources.

Second, future behaviour, like past behaviour, can be relevant to responsibility. Not only might Jack’s future smoking cause his transplant to fail, it might make him morally responsible for that failure. True, Jack is not yet responsible for the negative consequences of his future behaviour. But we can reasonably expect that he will be so-responsible, and, insofar as we take past responsibility into account, it’s not clear why we should ignore future responsibility.

Still, refusing transplants on the basis of behavioural predictions would be very controversial, not least because it might seem to commit us to rationing resources to people like Jill—people who are predicted to engage in some unhealthy behaviour, but have not yet done so.

What could justify qualms about using behavioural predictions to ration scarce medical resources?

One suggestion would be that using behavioural predictions to ration organs is problematic because we should always take an optimistic view of a person’s future choices—we should always assume that they will choose well, not badly, even if we have evidence to suggest otherwise.

I think there’s something to this suggestion. Yet there are limits on how far it can take us. Perhaps we should err on the side of assuming the best, when it comes to a person’s future choices, but at some point, if a person has repeatedly demonstrated a tendency to make poor choices, we will be justified in taking a less rosy view. This is certainly consistent with how we regard behavioural predictions in other spheres, such as forensic psychiatry and criminal justice. We might be inclined to think that a one-time offender should be treated as if they will refrain from offending in the future. But we would think it reckless to take the same approach to a serial recidivist. In that case, we might think that the evidence that the person will make bad choices in the future has simply become too strong to ignore. Alternatively, we might think that the offender has waived his right to being given further chances to show that he will choose well: he’s been given a fair opportunity to show that he is capable of reform, but he has not taken it.

Where does all this leave us when it comes to Jack and Jill? I think it suggests that it may indeed be problematic to refuse a transplant to Jill on the basis of her predicted smoking. Perhaps we should assume the best, when it comes to Jill’s future choices.

Things are less clear with Jack, however. Perhaps Jack’s long history of smoking gives us such strong evidence that he will continue to smoke that we can no longer ignore it. Or perhaps, by engaging in an unhealthy behaviour over such a long period, he has waived his right to being treated on an ‘optimistic’ basis.

I’m not suggesting that the case of Jack closely resembles the case of a recidivist criminal offender. Choosing to smoke can’t be compared with committing serious crimes, and there’s no reason to think Jack should be punished for his smoking. But there are some structural similarities between the two cases, and in both cases it’s doubtful that we should simply continue to ‘assume the best’ regarding the person’s future choices.

Not convinced? Then suppose that Jack has in fact already had several transplants. In each case, he was advised to stop smoking. But in each case, he continued to smoke and, as a result, damaged the transplanted organs. Suppose further that Jack is not strongly addicted to smoking and enjoys a privileged social position in which there are few social pressures towards smoking; he could quite easily quit, he just prefers not to. In this case, it seems to me clear that it would be acceptable to move Jack down the priority list for a further transplant on the basis of his predicted further smoking. This suggests that, contrary to orthodox wisdom, behavioural predictions can be a legitimate basis for rationing medical resources. The interesting question becomes not whether we should ever ration on the basis of such predictions, but when, exactly, we should do so.


Share on

7 Comment on this post

  1. The assumption that both are considered from the descision making authority the same in terms of for example their social impacts, is not a big assumption?
    What if Jack a heavy smoker but an expert, which is very possible because of his age. and Jail a 20 year old with predictable unhealthy behaviours.
    My question, in sum, is what if we consider merits for people.

  2. I am slightly unclear about the chronology of events. Are we assuming Jill doesn’t engage in any smoking before she receives, say, a lung transplant?

    In my view, there could be four reasons why we refrain from using predictors of the kind you mention for moving people up and down the transplant list.
    One is, like you say yourself, the lack of moral responsibility – although we might be cautious, we would think it presumptuous to act on possible future transgressions similarly like we act on past actions; therefore, while it is hard to ignore predictions on future behavior, few, I think, would say that future failures to act responsibly should be treated exactly as past failures.
    Second, we avoid using such predictors because they are arguably discriminatory.
    Third, it seems we also use, as a rule of thumb, the idea that past behavior counts for a stronger predictor of future actions than other kinds of predictors, and it actually makes sense in this case; suppose we think Jack will once again engage in his past behavior because smoking is an addictive practice and a constant temptation for him, While Jack is now given a reason not to once again engage in smoking (post-transplantation), he also has to overcome that addiction. In the case of Jill, Jill can more fully engage with this new reason not to engage in smoking, without addiction as hindrance. It is also questionable whether predictors like ethnic or cultural background are as useful as predictors once persons have undergone transplants.
    Fourth, there is the justified fear that pre-emptive methods might have far-reaching implications in other domains. If we are allowed to produce policy based on predictors, this would ease the justification of other pre-emptive methods, ones we would find deeply troubling.

  3. Thanks Viktor and Fatimah. Good points. There are certainly lots of other potential justifications for treating Jack and Jill differently and you mention several of these.

    On the discrimination issue, Viktor, I guess you are thinking that this would arise if, for example, risk of future smoking were associated with membership of a protected social group, such as an ethnic group. I agree that this is a serious worry, but I think it’s important to distinguish two variants of it. On one variant, the concern is about explicitly using e.g. ethnicity as a predictor (this is a concern about what gets called direct discrimination). I think this version of the worry could be avoided simply by not using e.g. ethnicity as a risk factor for the purposes of calculating risk of future smoking (it is often excluded from the violence risk assessment tools used in psychiatry and criminal justice for precisely this reason). On another variant, the concern is that predictions of smoking risk might be correlated with e.g. ethnicity even if ethnicity is not explicitly used as a predictor. (This is most naturally understood as a concern about *indirect* discrimination or as a concern about the risk of compounding existing unjust inequalities.) I think this problem is very difficult to avoid, but it’s not clear that this worry is any more concerning in relation to the use of predicted future smoking as a basis for rationing than it is in relation to the use of *past* smoking as an independent basis for rationing. If past smoking is correlated with ethnicity, then wouldn’t similar concerns arise there?

    1. Thanks, Victor and Tom. I think I now know what I wanted to ask.

      As far as I understand the decision here is based one the follwoing elements:

      1. possible contributing factors,
      2. their nature if they are uncertain or certain.
      3. Weight of these factors

      But how you can give weight to some qualitative factors, like social impact.

      Also how you can decide when a factor for someone is certain and for the other one is probable. Obviously, there are some certainties about Jack because of his age that Jill cannot have them.

      (I think something like ethnicity can be also a contributing factor and decision based on them is easy. Because you will probably get some statistical data and give a weight to it! For instance, obviously, people from an ethnic group or even a culture who drink more alcohol are more probable to get some special diseases!)

  4. Interesting article!

    I’d find it difficult to justify treating these cases differently.

    Based on what you’ve discussed in your piece, the criteria for deciding who qualifies more for an organ donation can be divided into practical and moral criteria.

    On the practical slide:
    People engage in various risky behaviours which are looked upon more kindly than smoking, like extreme sports and even driving, which lead to more pressure on the NHS. It could be a slippery slope when you begin to consider other lifestyle choices are predictors of increased risk for needing an organ donation. And would only the behaviours that are relevant to the organ in question be taken into account? e.g. a non-smoker motorcyclist needs a lung transplant but is an above average risk for a future transplant. That’s a bit contrived, but you get the idea. From a purely practical perspective, being high risk for cancer due to genetics could be a factor that’s taken into account when determining a post-operation prognosis.

    I expect most people would only support lifestyle factors and prognosis being taken into account when they are connected with “moral responsibility”, e.g. Jack compared to Jill. So on the moral side:
    One thing that would be impossible to avoid is social class; smoking is already a large contributor to the difference in life expectancy between rich and poor. To obtain a “true” measure of moral responsibility you would have to control for various factors, and it would be impossible to consider every contributing factor (e.g. even if you have two people with similar experiences, incomes etc., one could have some genetic predisposition that makes them more likely to be a smoker).
    Also, smokers’ lungs are often used for lung transplants (~50% in recent years, if my sources are correct) – what about a non-smoker who is on the donation register vs a smoker who isn’t? I’d be curious to know public opinion on which should have priority for a transplant.

    I can’t see a strong case for an algorithm that prioritises transplants based on factors other than need and predicted short-medium term survival of the operation (without assuming the patient will return to smoking), or how it could be practically implemented (I haven’t thought out the potential issues arising from training MVA on data to produce predictors, but it would be interesting). Although I can see the case for a smoking abstention requirement which is enforced in some countries, like the 6 month alcohol abstention sometimes required for a required liver transplant.

    The only solutions I can offer are increased measures to mitigate smoking rates, increase available organs with opt-out, and more support for addiction.

  5. I work in a busy transplant center where these issues are encountered frequently.

    Another example similar to smoking and tobacco addiction is alcohol addiction and liver transplantation. Concerning alcoholic cirrhosis, a period of several months’ sobriety is required prior to liver transplantation, without which transplant is denied. In this real life example, people who need liver transplants as a consequence of their alcoholism are only offered them if their alcoholism is in remission (i.e. they are “sober”). Conversely, ongoing alcohol consumption at the time of consideration for a transplant is used as a predictor of future behavior, i.e. continued drinking following transplant, and so the transplant is denied. (Digression #1: This raises the interesting question of how we establish the length of the waiting period and if it should be the same for everyone, or if we should take other factors into consideration on a case-by-case basis.) Another variation, similar to the example of criminal justice in the main article, is how to rank people whose liver disease is a consequence of an isolated bad decision. For example, hepatitis C, the second most common cause of cirrhosis in the United States after alcoholism, is sometimes contracted from a single use of a contaminated needle. An isolated use of IV drugs that occurred decades ago followed by years of sobriety may require different moral consideration from decades of IV drug use followed by a short period of sobriety. As an extreme example, attempted suicide with acetaminophen (paracetamol) leading to fulminant hepatic failure is a frequent reason for liver transplant, despite the fact that the best predictor of suicide is a previous suicide attempt. This scenario may be viewed as an isolated event worthy of forgiveness, and the suicidal person may be viewed with pity rather than distain.

    A variation on this theme, is when the person who needs a liver transplant has a coexisting medical condition for which they are not at fault that is likely to increase the risk of transplant failure. The most common example of this is severe mental illness that would make it difficult for the person to manage the immune suppressing medications needed in order to maintain the viability of the transplant. In such cases, people who are deemed to have “strong social support” (i.e. those who reliably manage their medications or have a surrogate who performs this duty) are considered, whereas those who do not have “social support” (e.g. the homeless) are denied. (Digression #2: The process of defining and establishing “social support” obviously raises questions of both direct and indirect discrimination.) I can easily see Jill falling into this second category. Suppose she contracted hepatitis B at birth, has cirrhosis and now needs a transplant, but is homeless and actively psychotic. It is not that she has ever done anything to harm her liver or is predicted to take action that will harm her transplant, were she to get one, but that she is predicted based on her homelessness not to take action, i.e not take the medications, required to ensure the transplant’s success, that forms the basis for justifying the denial.

    Behavioral predictors can be and indeed are in practice a basis for rationing medical resouces.

Comments are closed.