Direct financial incentives in exchange for compliance to drug use in medical settings have been described by Giuffrida and Torgerson¹ in a recent review of the literature. Overall, financial incentives had a positive effect on outcome, especially in treatment for tuberculosis, but data are scarce in mental health settings.

In a previous article, we have summarised our positive clinical experience with direct financial incentives in an assertive outreach (AO) team in East London, together with a survey of attitudes of this practice of AO team managers in England.²

AO teams operate a lower case-load size and focus on engaging revolving-door- and high-risk patients, in whom non-adherence to drugs is a constant problem and one of the criteria for referral to AO.

In the first part of this study, we describe a sample of four patients, in whom all other attempts to enhance treatment adherence had failed and involuntary hospital admission had seemed highly likely, if non-adherence continued. The patients received £5–10 (€7, US$9.75) as payments for each depot injection over 1–2 years. We have called this approach money for medication (MfM).

This paper looks in detail at the findings of the second part of the study, a small survey conducted to find out the attitudes of other AO teams in England regarding this practice. I will specifically analyse the ethical concerns that have been voiced in the survey. Where appropriate, I have added comments in the Discussion section made by participants in several presentations and discussions on the topic in local mental health and research circles.

There are mainly two different approaches to ethical reasoning: deontology states that moral decisions should be based on certain rules (eg, in the field of medicine, the Hippocratic oath or any other code of practice for health professionals); utilitarianism does not assume that there are automatically right things to do—it is more concerned with the outcome (eg, clinical performance indicators) and aims to achieve the greatest balance of “good” over “bad”.³

Beauchamp,^4,5 in an attempt to synthesise these two approaches and derive a practical ethical grid, describes his popular four principles approach (referred to as principlism⁶) to medical ethics as including four categories with which to judge the ethical dimension of medical interventions:

(1) Beneficence (the obligation to provide benefits and balance benefits against risks)

(2) Non-maleficence (the obligation to avoid the causation of harm)

(3) Respect for autonomy (the obligation to respect the decision-making capacities of autonomous persons)

(4) Justice (the obligation of fairness in the distribution of benefits and risks).

Ethical issues pervade community mental health, especially AO.^7,8,9,10 However, this field has been neglected in the academic realm of medical ethics.^11,12

Williamson¹⁰ summarises the discussion on ethics in AO in a proposal to shift the emphasis towards clients’ values, especially towards high levels of satisfaction and practical support with housing and benefits. This is supported by a recent randomised controlled trial (RCT) on the outcome of assertive community treatment compared with standard community mental health teams, which showed higher satisfaction and engagement in the AO group, but no improved outcome in terms of hospital admissions or violent incidents.¹³

Adshead¹¹ mentions four categories of ethical dilemmas pronounced in community care: capacity to make treatment decisions and refusals, conflicting duties to patients and third parties, coercion of individuals for public benefit and allocation of resources. She argues for an evidence-based way forward.

Christensen¹² has provided a very practical approach to categorise the ethical problems in community psychiatry. He distinguishes four categories and some subcategories, namely:

1. Valid consent and refusal

2. Psychiatric paternalism

   1. Involuntary hospitalisation

   2. Forced medication

   3. Coerced outpatient treatment (including autonomy and effects on the therapeutic relationship)

3. Resource allocation

4. Organisational relationships.

Christensen’s categories will be used later to organise the respondents’ answers further.

The article aims to answer the following questions:

1. What are the major ethical concerns associated with “MfM”?

2. How can these ethical issues be evaluated within the above-mentioned framework of ethical problems and are there any teachings for clinical practice?

METHOD

In the survey (details elsewhere), we sent out questionnaires to team managers of 150 AO teams in England asking whether financial incentives were used to increase drug adherence, what the attitudes to such a practice are and whether the issue had been discussed in the team. This was accompanied by information on the background of MfM, giving a tentatively positive view of this approach as unorthodox and possibly effective. Ethical approval for this study was granted by the responsible research ethics committee.

Question 4 of the questionnaire asked specifically; “Do you, as the team manager, have any objections to use financial incentives as a method to achieve compliance?” If the answer was “yes”, people were also invited to give a reason in a free text field. In case of “no”, no further arguments were possible. After 3 weeks, teams were reminded via the telephone.

The handwritten answers of the team managers were transcribed into an MS word file and analysed descriptively and qualitatively.

To categorise and discuss answers regarding their content in the context of community psychiatry, Christensen’s four-category system (see above) was used. To highlight further ethical problems, Beaumont’s ethical grid was used.

RESULTS

We received 70 of the 150 sent out questionnaires (response rate 47%). Of the 70 respondents, 53 (76%) mentioned one or more ethical reasons for refusing.

There was a broad range of different answers; some very brief (eg, “unethical”), others giving a comprehensive reasoning and covering several issues.

First, it was striking that several team managers used strong words to denote their concerns (“I was horrified at the thought of this approach!”, “This letter has really shocked me!”).

Following Christensen’s framework, issues concerning valid consent and refusal were touched in five answers (eg, “Clients should take meds because they want to and not because they are paid to”, “Issue of client’s capacity”).

Psychiatric paternalism was mentioned in 31 answers (all relating to subcategory (c), coerced outpatient treatment), making this the most important ethical concern, including concerns regarding possible coercion, autonomy and a detrimental impact on the therapeutic relationship (eg, “Reinforces sick role”, “unhealthy dependence”, “feel disempowered”, “coercion”, “we have to respect the clients choice”, “It confuses the relationship”, “damages the therapeutic relationship and eradicates trust”, “a short-cut to meaningful engagement”, “fosters a negative dimension of the relationship”).

Issues with resource allocation were mentioned in four answers (eg, “How much would this cost the department?”, “Do we offer money for diabetes or those with hypertension?”).

Organisational relationship issues were explicitly mentioned in two answers (eg, “Problems … if the service user is transferred to care where the incentive is not in place”, “clients would inform their solicitor”).

A residual category mentioned ethical concerns, but did not specify or could not be grouped under any of the above (eg, “It would have no real meaning”).

Table 1 presents the frequency of concern mentioned overall in the survey.

DISCUSSION

The forcefulness and expressive content of some of the responses indicate the strong emotions associated with the issue. This makes it all the more important to look at the major groups of concerns in a systematic and reflective way.

Valid consent and refusal

It is true that most service users who would be considered for MfM do not want to take this drug at first. However, this alone cannot be an argument against giving money, as much of what health professionals do is about convincing and persuading people having all kinds of illnesses to do something (eg, stop smoking, get blood sugar level measured, reduce weight), which they might not feel inclined to do initially. One could argue here that mental illness is more complex than physical illness in a number of ways and that people having physical problems usually have the capacity to make decisions for or against a certain treatment, whereas people having a mental illness might lack the same capacity. One of the relevant questions is, does MfM impair the service user’s capacity to make an informed decision?

If capacity is defined as the ability to understand treatment-related information, to appreciate the significance of this information, to exercise reasoning in comparing the treatment in question with other alternatives or no treatment, and finally to express a choice coming out of this process (decide and express),¹⁴ this has to be assessed on an individual basis.

However, most of our clients who might be eligible for MfM may have long-term schizophrenia or schizoaffective illness. Acute symptoms such as hallucinations or delusions, as well as negative cognitive symptoms, may seriously impede the ability to understand and digest new information. From a recent study by Cairns et al,¹⁴ we know that >43% of inpatients lacked treatment-related decisional capacity. Patients in AO are often only a step away from admission to hospital, which means that similar rates of incapacity might be prevalent here.

Where capacity exists, is full informed consent then “bypassed” by offering MfM? If the advertised treatment is so good, why does the service user only agree once MfM has been offered? Can this really be counted as consent?

In reality, consent to any treatment decision is a complex issue. Full information about the benefits and possible side effects is only one issue; insight, the therapeutic relationship, the reaction of the social environment and possible monetary incentives or disadvantages are other factors influencing this decision. The decision of someone to, for example, stop smoking, might be influenced by information about possible negative consequences (bronchial carcinoma), by the reaction of his environment (partner who does not smoke), anticipated harm to others (children) and the fact that the intended life insurance policy offers a considerable bonus for non-smoker status. Only if the sum of these reaches a critical threshold will a positive decision to stop smoking be made.

Points for clinical practice

A psychiatric assessment of capacity and an informed consent procedure with full information about the process and the drug in question is necessary, including a weighted discussion about the advantages and disadvantages of this approach with the service user.

Psychiatric paternalism

Consent should also be voluntary and non-coercive, and this concern was considered by several respondents to be best placed in the subcategory of “coerced outpatient treatment” of the second category of Christensen’s framework.

Coercion is defined as “forcing someone to do something they do not want by threatening them”.¹⁵

Wertheimer¹⁶ proposed the following practical approach to decide whether a (therapeutic) proposal is coercive or not: “The standard view is that threats coerce but offers do not. And the crux of the distinction between threats and offers is that A makes a threat when B will be worse off than in some relevant baseline position if B does not accept A’s proposal, but that A makes an offer when B will be no worse off than in some relevant baseline position if B does not accept A’s proposal.”

According to these definitions, MfM could be seen as not coercive: something extra is provided in exchange for adherence, very much in the sense of an instant positive reinforcement in behaviour therapy, possibly with a generalising component. In this sense, MfM would be more of an offer, not a threat. At least, this would be the case initially. However, once the scheme is started, there is a possibility that the service user becomes used to it, and as a consequence, the intention of withdrawing this (ie, in cases of non-compliance) could turn into a threat. Still, this will be different from the practice in the US, where social benefits and entitlement to housing provision and also the freedom to stay in the community (as opposed to involuntary hospital treatment) are tied to adherence to treatment programmes and drugs through a payee system, tenancy contracts or so-called “outpatient commitment” ordered by a mental health court.^17,18

The counterargument (“not unlike third-world countries selling their kidneys”) assumes that patients with psychiatric problems, especially in AO, are vulnerable and susceptible to making decisions that are not in their best interests or selling out long-term health against short-term gains. This could be true, for example, in the case of a service user spending all his housing grant money on drugs instead of buying a washing machine. It might as well be that the service user decides in favour of MfM to achieve a short-term gain (eg, £10, to finance his drug or gambling habit), but in contrast with the “kidney” scenario, this is offered with the expectation that his health will improve in the long term and his insight and engagement increase. However, there might be an element of extortion in offering money in return for drug adherence, in that we use the severe lack of this commodity in our patients to reach our goal, albeit well intended. This would be particularly true in a scenario where the immediate needs of a patient—that is, in drug withdrawal—would be used as leverage to start the scheme.

Points for practice

Reasonable time should elapse after the offer for MfM has been made and the informed consent form signed and before the scheme started. All immediate physical and medical needs should be met before, and independent of, starting this scheme.

Following the discussion about coercion, there could also be a case of undue influence or pressure (as defined by the Belmont report¹⁹) as an offer of excessive, unwarranted, inappropriate or improper reward or other overture to obtain compliance (of research subjects). However, MfM is not a treatment as such. The treatment itself (the depot medication) is not in question and is not subject to research, a risk–benefit analysis has already been completed and these variables are known—other than in experimental research, which is undertaken to prove or validate the benefits and risk of a specific treatment.

Concerns regarding autonomy, which also come under the category of “psychiatric paternalism”, mentioned different issues, among them a possible fostering of dependence. It is valid to assume that the service user, who agrees to accept money for regular depot injections, becomes more dependent on the service to some extent, at least temporarily, until the recovery process (which would probably be impaired in the case of non-adherence) also improves the ability to access other sources of income (work, educational grants). But what if there is no recovery process as such—there might be greater social control as a consequence of the drug, yet without improvement in psychopathology or social functioning. In cases such as this, MfM could become a permanent solution to non-adherence without any clear way out.

Points for clinical practice

MfM should only be used on a temporary basis, the discontinuation of the scheme should be regularly considered (eg, in the Care Programme Approach review) and early rehabilitation support is essential to move the service user away from possible long-term dependence on this extra income. The positive nature of the reward should be clearly outlined.

Following Christensen’s categories, problems regarding the therapeutic relationship also come under psychiatric paternalism. Does the relationship between the service and the service user suffer if (s)he receives a financial incentive for adherence to a drug?

No evidence from our experience supports this view. One would not want to endanger the relationship as a whole in exchange for mere adherence to a drug, thereby possibly depriving the service user of other valuable means of recovery—for example, psychological or social support. However, considering that in many AO users, compliance to drugs is the most important single agent in the multiprofessional mix of interventions, one could argue that in these cases only adherence generates enough stability in, for example, successful psychotherapy or the ability to sustain housing.

The counterargument, made by many teams in the survey, seems to be rather intuitional: a danger is perceived as unevenness or a corruption of the normal rules of a therapeutic relationship. In the absence of hard evidence to the contrary, this cannot be neglected, but it has to be put into perspective and weighed against other possible dangers for the therapeutic relationship (in case of non-adherence), namely negligence and (at the other end of the spectrum) coercive measures under the mental health act. Future research into users’ views is needed to shed more light on these issues.

Points for clinical practice

Only where the necessity for drug adherence is clear (eg, from risk assessment, admission pattern or others) should this approach be used. An operational policy can be helpful with this to outline the approach to the team, the service users and the service environment. Thoughts should be given to a possible impact on the therapeutic relationship and a potentially detrimental effect weighed in comparison to available alternatives (ie, close follow-up with hands-on support, working through relatives, admission under mental health acts and so on).

Resource allocation

How much and what money or funds are used? Is it fair to spend money in this way? Do we have to give money to other service users as well? How do we continue this after AO? Is this disadvantaging other service users?

As MfM is not an established treatment and no funds are allocated to this, team managers might have ethical concerns to use funds that are earmarked for other purposes, thereby depriving other service users of monetary support for their needs. Although in absolute terms, the money in question may not be much (in comparison to a possible hospital admission), this issue obviously needs to be addressed by the service as a whole.

One could also ask who decides about the actual amount of money being used. Why not pay £50 (€74.99, US$97.52) or £500, if this could increase adherence? Apart from cost-effectiveness arguments, which are not necessarily touching our issue, the “extortion” argument becomes more valid with more money being offered, as the ratio of extra income to regular income (ie, benefits) would be shifted towards the additional gratification with increased dependence on this.

Is it fair to withhold money from other service users, who have made the decision to stay adherent to their depot injection without any incentive (and who, as a consequence, then might ask for that money as well)? If we think of MfM as remuneration for the effort, pain and possible side effects that the service user has from long-term depot injections, then this is probably unfair to more compliant users. However, if we think about this as an incentive for adherence, merely tilting the balance of an individual’s informed decision-making process (much in the way as it happens with more informal gratification) towards drug adherence, person-centred incentives would not necessarily be unfair, as they (like the benefit system) merely take into account that some people need more positive reinforcement to succeed than others (see above).

Problems usually arise if a service user is discharged from the service into the care of another team or leaves the area. Problems with continuing care could arise because the patient’s expectations regarding the money received in the past might not be met by the new team. In a positive way, this could be used to introduce the service user to the idea to come off the scheme.

Points for clinical practice

Agreement on the organisational level should be sought and money identified, from which MfM could be drawn, while making sure that this is not a disadvantage to other service users. An upper limit of financial incentives has to be agreed upon and there should be a reasonable ratio of MfM to regular income. Close liaison work with the future team in case of a transfer will be necessary. The team and the service user might need to be reminded that MfM is a time-limited arrangement.

Organisational boundaries

The fact that there were only two answers relating to organisational issues as such would support the assumption that this is not a major constraining element, although legal and organisational issues tend to come up in all presentations on the subject. Although sometimes only touching ethical issues, questions as to how to organise continuity of care across boundaries and to comply with existing legislation are valid concerns and are not solved yet.

Points for clinical practice

All involved professionals should be informed, comfortable and open with this approach; the support of the management is essential and decisions should be clear and documented. Research activities and ongoing discussions are part of this effort.

To round up the arguments for and against MfM, two of Beaumont’s four ethical principles shall be considered further—namely, beneficence and non-maleficence.

Beneficence

Interestingly, the question of whether MfM is actually effective did not surface in the interviews. It seems that people are assuming that it works. This also indicates that team managers, regarding their approach to ethics, take a more rule-based (deontological) approach to the question and are not much concerned about beneficence. As we did not really ask for elaborations on the possible positive side of MfM, this might also be due to flaws in our methods.

There is no proof for this approach being effective yet,¹ but if it does not work, there is probably no argument for using it in the first place. There is an assumption that MfM will increase drug adherence and thereby positively influence the mental health of the service user and the possible outcome (hospital admissions, incidents, quality of life, social inclusion and relationships, vocational training). Further positive results could be an increased engagement, more intensive treatment (eg, cognitive behavioural therapy, see above) and less risk.

The counterargument would be to assume that all patients who are agreeable to MfM would have taken the drug anyway, perhaps at some later point of time, so that hardly any difference would be discernible if they had not been on the scheme.

If it does work and can be proven to make a difference in outcome, then this approach should not be withheld from other service users in other teams. Here, a well-designed multiteam RCT is needed to generate clear data.

Non maleficence

Can MfM harm the patient? Some concerns regarding the fostering of a possible dependence or a negative impact on the therapeutic relationship have already been discussed. These have to be taken seriously and included in further research settings—that is, the mentioned RCT or a qualitative approach to evaluate users’ attitudes and perceptions of the process. For the latter, an analysis of focus groups, conducted with the help of user organisations, could be very valuable.

In summary, careful consideration of the mentioned ethical problems needs to be incorporated into clinical practice; informed consent after thorough discussion has to be sought from the patient, guidelines or an operational policy needs to be in place to ensure that this is only used for service users with a high probability of exerting a beneficial effect deontological, as well as utilitarian principles, to be followed maximise the treatment benefit and long-term independence of the service user.