Based on key findings from our survey, it seems that the public prefers to allocate resources to individuals who would benefit the most, even if they are responsible for their illness.
These underlying beliefs are more consistent with a prospective, rather than a retrospective, model of responsibility. In our sample, people were attracted to lifestyle contracts and more willing to attribute responsibility on the basis of them, including when using mHealth technology. We will now propose a new model, then relate it to our empirical findings on relevant public intuitions in more detail.
The Prospective Intention-Based Lifestyle Contract Model
With a growing burden of lifestyle-related diseases in a climate of fiscal austerity, there is the possibility of individual responsibility playing a greater role in public healthcare rationing. Philosophy does not exist in a vacuum. It may be that philosophical opposition to the responsibilisation in healthcare will be unable to counter political momentum. An alternative method of holding individuals responsible for health may provide a path that is politically palatable as well as philosophically less objectionable than traditional retrospective conceptualisations of responsibility in healthcare. It is not our intention to minimise the significance of social inequalities which contribute to poor health [35] or detract from the need for largescale public health initiatives; rather, we hope only to offer a harm reductionist approach to responsibilisation in healthcare.
At its core, Feiring’s [20] model proposes that an individual who contributed to their condition should be given equal priority at the first instance of lifestyle-related medical need. To have equal access to future healthcare, however, they must commit to a lifestyle contract supported by medical follow-ups to achieve the contracted goal to, for example, lose weight or quit smoking. Failure to agree to, or successfully fulfil, the contract would result in deprioritised public healthcare access. Building on Feiring’s [20] lifestyle contract model, we will now outline our version of the lifestyle contract model. It retains Feiring’s core idea but differs in several key aspects. Firstly, Feiring assesses fulfillment of a contract based on the ultimate outcome, while we require only that patients make a sincere attempt to make the contracted behavioural change. Secondly, we stress that the lifestyle contract should be reasonably achievable based on the patient’s social and environmental circumstances, and that appropriate support services be made available. Finally, our model shows how mHealth technology can be used as an adjunct to facilitate the assessment of whether a lifestyle contract has been fulfilled in addition to broadening the range of reasonably achievable lifestyle contracts.
Under our proposed model, a patient whose lifestyle has contributed to their illness will have standard access at the first instance to scarce healthcare resources (Fig. 1). At this time, they will be given a lifestyle contract that commits them to certain conditions. If the contract is rejected or the requirements are not met, then their priority will be lowered for future instances of related healthcare need. The lifestyle contract should only contract behaviours that are likely to significantly improve the overall utility of the healthcare resource received. For example, the contract for a recipient of a liver transplant for alcohol-related liver disease should limit only risky behaviours that can reasonably be expected to damage the new liver, such as non-compliance with immunosuppressive medications or excessive alcohol intake.
Moreover, the lifestyle contract would assess the nature of the attempt to make a lifestyle change rather than the ultimate outcome. If the patient demonstrates a genuine attempt to change their behaviour, even if ultimately unsuccessful, then the patient is not deprioritised. For example, in the case of smoking, actively engaging with a formal cessation program may be sufficient to fulfil their contract. Bærøe and Cappelen [6] support a similar modification to Feiring’s lifestyle contract model, although they assess only whether a patient has attended relevant support services.
Requiring a sincere attempt at a lifestyle modification is less ethically problematic than Feiring’s model [20], as it represents a more realistic expectation of people. Furthermore, the contracted lifestyle modification should also be reasonably achievable for an individual’s socioeconomic circumstances which may impact a person’s ability to make the required change to varying degrees throughout a contract. For example, any resources or services required to fulfil the contract ought to be accessible and affordable, and any pre-existing social stressors which may limit the individual’s capacity to attempt to make difficult behavioural changes ought to be taken into account when drawing up the contract. These conditions are reminiscent of Savulescu’s and Davies’ concept of a Golden Opportunity [17, 44], which stipulates a set of conditions which need to apply to a health behaviour change in order for an individual to permissibly have health priority lowered based on responsibility if they choose to reject the change. These conditions include that the behavioural change must be realistically achievable for the individual and their circumstances, that the individual must be supported in making the behavioural change, and that there are clear consequences for failure to capitalise on the Golden Opportunity.
How, then, should a sincere attempt at a lifestyle change be assessed? An individual’s sincerity and motivation in pursuing lifestyle change are internal states inaccessible to direct measurement, leaving external behaviours or self-reflection as proxies. The lifestyle contract should be created in consultation with the patient based on the activities or measures they feel they would realistically be able to engage in. Possible measures of sincerity of an attempt include engagement with support services offered, for example, attending drug and alcohol support groups or participating in motivational interviewing sessions. Engagement with in-person professional services, however, is not always feasible due to geographical factors or other time commitments. Participation in online behavioural modification programs or tele-health services enabled by mHealth technology may be more suitable for some individuals. Another possible lifestyle contract assessment tool is an ongoing diary detailing the ways in which the patient attempted to make the relevant health behaviour change, in addition to circumstances or thought patterns which prevented them from doing so. Here, mHealth could serve as a useful adjunct in collecting data on individual behaviours, intentions, and attempted improvements. While a physical diary may be preferable for some patients, they have the limitations of being cumbersome and conspicuous to use. In contrast, mobile phone usage is often already part of a daily routine, and logging thoughts and events digitally would be less socially conspicuous. The individual could also set periodic reminders to complete the diary. Further adding to the benefit of mHealth enabled versions of the diary, data obtained from wearable devices regarding, for example, medication adherence or blood alcohol level, could automatically be integrated with a partner app, allowing the generation of digital graphs or trends which illustrate patterns in the individual’s behaviour. Problematic behavioural patterns could then be identified and any triggers or circumstances which lead to poor health choices could be mitigated. For example, a wearable device detecting medication adherence might find that over a period of several months, a patient’s adherence is worse after night shift. Based on the identification of this pattern, the patient could trial targeted strategies such as setting timed reminders.
Data from mHealth devices alone would not be grounds for failing a lifestyle contract, as it would be insufficient to assess the sincerity of the attempt at behavioural change, however it may be used to support an overall assessment of an individual’s behavioural change efforts. mHealth devices thus serve dual functions of supporting patients by providing information to troubleshoot unhealthy habits and improve health literacy, as well as providing objective data to support the assessment of whether a patient has fulfilled their lifestyle contract. The supportive function may also help dampen the potential negative impact on the individual’s relationship with their monitoring device.
Empirically Aligned with Public Intuitions
In comparison to retrospective models of personal responsibility in healthcare, our proposed prospective model is better aligned with the public intuitions elicited in our exploratory survey. It balances efficient resource allocation with patient accountability for lifestyle-related illness, while improving the capacity for responsibility and providing a framework for individuals to take on greater agency in their health narrative [42].
Retrospective models of personal responsibility are usually based on desert-based claims, whereby if a patient foreseeably and avoidably caused their disease, then they have weaker claim to scarce public resources. Our results suggest, however, that people are more concerned with maximising the benefits derived from public resources.
A lifestyle contract represents an actionable duty to society to look after one’s health which, our findings suggest, is generally supported by the public. If that duty is unacceptably violated and scarce resources are underutilised by a patient, then consequences follow in the form of lowered healthcare priority.
Just as our study participants felt the attempt to make a healthy lifestyle change had some value, our model also allows that a sincere attempt, even if unsuccessful, to make the most of a scarce resource through behavioural change is sufficient to do due diligence to one’s duty to society.
Finally, mHealth technology acts as an efficient and publicly acceptable tool to transparently quantify adherence with lifestyle contracts.
Theoretical Advantages
Our prospective lifestyle contract model also has theoretical advantages over retrospective models. Using Bærøe and Cappelen’s [6] excellent summary of the major criticisms of personal responsibility in healthcare as a framework, we will consider the same objections against using responsibility in healthcare: (1) the harshness objection; (2) the avoidability objection; (3) the intrusion objection; and (4) the objection of causality. Additionally, we will consider a fifth objection, the efficiency objection, which has also been used to argue against incorporating responsibility in healthcare. On each front, we will show why our prospective model of responsibility fares better than a retrospective model, particularly when the possible impacts of health monitoring technology are considered.
The Harshness Objection and the Avoidability Objection
The related harshness and avoidability objections will be considered together.
At its core, the harshness objection argues that even if individuals are responsible for their illness, denying healthcare to them would be too harsh, punitive, and inhumane [14]. Lowering healthcare priority may result in inability to access treatment, longer periods of suffering, or even death [14]. Therefore, the policy should not be endorsed.
Meanwhile, the objection of avoidability argues that one can be responsible for an action only if it was possible to have done otherwise. If there is no genuine alternative available, then the agent cannot be held responsible.
Whether the prospective model is too harsh seems to depend in part on the nature of the lifestyle contract. If the contract is very difficult to uphold, then failing it should not be considered blameworthy, and deprioritised access to future healthcare would indeed be too harsh. If, however, the contract is reasonably achievable, then the harshness objection is weakened.
The avoidability objection now becomes relevant. Part of why lowering priority based on retrospective responsibility is harsh is because we know that the social determinants of health (SDH) are strong influences on an individual’s health behaviours and health outcomes [35]. The SDH are defined by WHO as “the conditions in which people are born, grow, live, work, and age” [56]. SDH include factors such as race, gender, disability, education, income, education, neighbourhood, and social support networks, many of which are beyond an individual’s control. Over the last two decades, evidence has emerged that SDH are closely linked with lifestyle choices, including poor diet, alcohol, smoking, and inactivity [36, 37, 48]. For example, fresh, healthy foods may prove unaffordable to those experiencing poverty, while poorer neigbourhoods have a higher density of fast food outlets and convenience stores [27], encouraging consumption of foods with poor nutritional value. Indeed, the circumstances of one’s birth and early development have a range of lasting effects on health [36], including the development of cognitive control that promotes behavioural regulation. With the complex and far-reaching effects of SDH in mind, there may be some individuals who, due to factors outside their control, would struggle to achieve a given lifestyle change in their respective circumstances. Some go as far as to argue that because external socioeconomic factors so strongly influence lifestyles and health outcomes, moral responsibility for lifestyle-related ill health cannot be attributed at all, as they may have been severely restricted in the range of choices genuinely available to them.
There are many factors which, if embedded in a lifestyle contract, would certainly render the contract not reasonably achievable and therefore too harsh a standard for allocating healthcare resources. To be considered reasonably achievable, the lifestyle contract would need to take an individual’s SDH and socioeconomic circumstances into account and attempt to recognise where barriers to success may lie. For example, it must be financially feasible relative to income, and take into account personal circumstances such as health literacy or geographic location. The individual should also be involved in determining what changes might be realistically achievable. The greater the hardship and barriers faced by the individual, the less that ought to be required in the contract, lest the profound impacts of the social determinants of health be underestimated. Input from social work and other psychosocial supports may aid in temporarily mitigating socioeconomic hardship, however efforts to actively overcome barriers to healthy lifestyle choices, such as poverty, would be limited by the scope of the healthcare system. These barriers would be better addressed by largescale public health efforts targeted at social inequalities and SDH.
Our proposed model acknowledges that behavioural changes are difficult, particularly in disadvantaged populations, and requires only an attempt to make a relevant behavioural change. If we assume there are far fewer people who could never make even a genuine attempt at behavioural change, then our model is less harsh as a standard to hold people to.
The core of the reasonably achievable lifestyle contract is to provide individuals with a genuine opportunity to try and make the most of the healthcare resources allocated to them. If an individual then fails to make a sincere attempt at a reasonably achievable contract while fully informed of the consequences, it does not obviously seem too harsh to deprioritise them. Even if we want to take into account the special importance of healthcare, perhaps by granting leniency in the number of reasonably achievable lifestyle contracts that one can fail before being deprioritised, an endless number of fresh starts is not owed by society.
Moreover, mHealth technology can increase the range of potential reasonably achievable lifestyle contracts. For example, behavioural modification programs delivered by mHealth devices or wearable lifestyle trackers is one method of overcoming geographical or time constraints, while emotional support might be made available through online social networks. Furthermore, mHealth devices can deliver health information at the individual’s own pace or in their preferred language.
Unless we are willing to discard intuitions about the existence of individual agency, there will be at least some circumstances where requiring a competent patient to make a genuine attempt at improving their lifestyle in a reasonably achievable way does not seem too harsh. This is provided that the social determinants of health mitigating the individual’s locus of control are taken into account, the consequences of failing the contract are proportionate, and there is genuine scarcity of the required resource. It is notable that some of the empirical findings from our survey were derived from zero-sum cases, where resources could only be allocated to one of two patients; healthcare resource allocation is rarely so simplistic. There are, however, some resources which are clearly either scarce, such as organ transplants, or extremely costly to the public healthcare system, such as novel curative treatments for hepatitis C [39]. While it may indeed be too harsh to require lifestyle contracts for any kind of treatment of lifestyle-related illness, it applies to a lesser degree when the required resource is scarce or significantly burdens the public system. Furthermore, we do not support complete withdrawal of publicly funded healthcare should a patient refuse or fail a proposed lifestyle contract, as this would be a disproportionate consequence. A plausible midground might be requiring the patient to contribute co-payments for future healthcare, relative to their ability to pay.
The intrusion objection
The intrusion objection argues that determining patients’ responsibility for behaviours, past or future, would be too demeaning and intrusive. A great deal of personal information may be required to determine whether someone’s behaviour contributed to an illness, and may include sensitive information about sexual behaviour, recreational drug use, or mental illness. Requiring individuals to reveal such information, according to the intrusion objection, would be disrespectful and humiliating, and should thus preclude the use of responsibility as a healthcare allocation criteria [29].
Bærøe and Cappelen [6] have argued that the prospective model of responsibility can stand up to the intrusion objection, provided that a lifestyle contract was offered to anyone whose condition could be improved by a lifestyle change, thus avoiding the need for information about past behaviour. This reads as too demanding. Many, if not all, conditions would likely benefit from a healthier lifestyle, which may extend to choices such as hazardous occupations, hobbies, sleep patterns, and locations of residence. If everyone, even those with relatively healthy lifestyles, were required to enter a lifestyle contract for equal priority, the contracted changes could become supererogatory.
In the contract-based prospective model, the monitoring method of lifestyle contract adherence collects targeted information about the specific behaviour outlined in the contract, which may be facilitated by mHealth lifestyle trackers. For example, if limited alcohol consumption is part of a lifestyle contract following a liver transplant, then only blood alcohol levels should be measured. Collecting information on only the contracted behaviour limits the intrusiveness of the lifestyle contract model.
mHealth may also alter the strength of the intrusion objection in other ways. Some may find revealing embarrassing information to an electronic device less confronting. After all, an electronic device is not capable of moral judgements or disapproval. By removing interpersonal factors, using mHealth in the assessment of responsibility may lessen the degree of intrusion experienced by patients.
The objection of causality
The objection of causality argues that (1) to be morally responsible for an event or state of affairs, it is necessary to be part of the relevant causal chain leading up to the given event or state of affairs and (2) we cannot determine whether voluntary behaviours caused any given disease state in many, if not all, cases. [25] This is particularly true of diseases with multifactorial aetiologies, where genetic predispositions, random gene mutations, and bad luck contribute in varying degrees.
Bærøe and Cappelen [6] argue that both the retrospective and prospective frameworks are vulnerable to the objection of causality, because both frameworks use the patient’s past behaviour to determine whether they are responsible for their medical need, either to lower priority or to determine whether a lifestyle contract should be offered.
Aside from diseases with only one aetiology, such as single gene disorders, it seems unlikely, at least in the foreseeable future, that we will be able to know what causally contributes to individual cases of illness and in what proportions. In a prospective model, a lower threshold of certainty could permissibly be applied to judgements of which patients were ‘responsible’ for their illness, because the consequences of doing so are less severe when offering a reasonably achievable lifestyle contract and only ensuring a sincere attempt than when directly deprioritising the patient. Although we cannot be certain of a patient’s causal role in illness in either a retrospective or prospective model, in a prospective model there is less need for absolute certainty.
The objection of efficiency
According to the objection of efficiency, determining individual responsibility for ill health on a case-by-case basis would be extremely time-consuming, and would siphon healthcare resources from other areas of need [29]. It would, therefore, not be practically viable.
mHealth technology is particularly relevant here. Although not yet sufficiently sophisticated, innovations in mHealth technology will likely make it possible to monitor a wide range of physiological and physical parameters with minimal resources, such as blood alcohol level [5], calorie intake [3], smoking [15], and medication adherence [3].
Unless dramatically greater uptake of mHealth technology occurs, however, this is more advantageous for prospective models of responsibility than for retrospective models. Retrospective models would require the devices to have collected sufficient data and patient consent for data access. It is also plausible that people would avoid using mHealth devices to prevent future deprioritisation. These practical considerations limit the usage of mHealth technology to determine retrospective responsibility.
For a prospective model, however, mHealth technology may drastically reduce the resource intensity of assessing a sincere attempt to adhere to a lifestyle contract. It would likely decrease labour costs, as devices would take over some of the role of monitoring lifestyle contracts. Continuous data collection would allow general trends to be made apparent to health professionals without in-depth history taking and provide valuable information to help the individual and their supporting healthcare team to identify problematic patterns. mHealth may also enable mass distribution of behavioural change programs as part of a lifestyle contract, costing little to create and maintain.
Limitations of the use of mHealth technology in lifestyle contracts
While the use of mHealth technology in the prospective model avoids some of the ethical concerns that arise in a retrospective model, it nonetheless poses possible risks. mHealth technology shares many of ethical difficulties beset by digital data as a whole, including risks to data privacy [12, 40, 51], murky data ownership laws [23], and poor oversight by regulative bodies [18, 32]. These issues also apply to existing health devices formally integrated into healthcare, such as continuous glucose monitors (CGMs) [12]. Ultimately, stronger, enforceable regulations on data management are needed to protect users [23].
Regarding the usage of mHealth technology in prospective lifestyle contracts, possible risks include the undermining of individual autonomy [41], intrusions into personal (as opposed to data) privacy [13], an antagonistic relationship between user and mHealth device, and inaccessibility to socially disadvantaged populations [2]. While an in-depth exploration of these issues is outside the scope of this paper, we will outline some general principles which aim to mitigate some of the ethical risks.
-
1.
Voluntary—A lifestyle contract should not necessarily involve digital monitoring, nor should there be a penalty for withdrawing from a lifestyle contract’s digital monitoring component, particularly if the patient finds using digital devices stressful or harmful.
-
2.
Minimising intrusion—The device should be limited to collecting data directly relevant to the lifestyle contract and be discussed with the patient prior to the contract’s creation. The device itself should be unobtrusive. Social functions, such as sharing with other app users, should be limited.
-
3.
Accessible—The patient should bear minimal costs from using the device. Patients should be adequately supported and educated in the use of the devices. Any health apps involved in the contract should be available in a language comprehensible to the patient.
-
4.
Accurate—The devices should be of high accuracy to avoid false representations of patient behaviours, which may be damaging to the doctor–patient relationship.