The Ethics of Experimental Heroin Maintenance

Abstract

In response to widespread concern about illegal drug use and the associated risk of the spread of HIV/AIDS, a study was undertaken to examine whether it was, in Principle, feasible to conduct a trial providing heroin to dependent users in a controlled manner. Such a trial involves real ethical issues which are examined in this paper. The general issues examined are: should a trial be an experiment or an exercise in public policy?; acts and omissions; countermobilisation; termination of a trial, and payment for drugs and for a trial. The specific issues examined are: selection of trial participants; privacy; issues for staff working on a trial; coupling the trial with other treatment, and issues for researchers. A number of alternative approaches to the various ethical issues are presented and discussed.

In Australia, as in many other countries, there is widespread concern about illegal drug use. In recent years this has been exacerbated by fears about the spread of HIV/AIDS within the illegal drug-using community through needle-sharing and sexual practices, and from this group to the general community. There is a common perception that current drug policy is ineffective and that new policies need to be explored (1-4). These range from stronger enforcement of anti-drug laws to legalisation of these substances.

The available evidence indicates that heroin, when provided in pure form, is a relatively safe drug (5). Hence it is primarily the illegal nature of the drug, rather than its pharmacological properties, which leads to the health and social problems associated with its use. It seems then that there is a potential for a reduction in those problems if heroin is made legally available in a controlled manner. Whether or not this potential would actually be realised, however, is hotly contested.

In 1990, the Government of the Australian Capital Territory (ACT) appointed a select committee to inquire into and report on HIV, illegal drugs and prostitution in the Territory with particular reference to: '(a) the effectiveness of current legal and social controls enabling action to prevent the spread of HIV; (b) the effectiveness of current legal controls on prostitution and drug-taking; (c) alternative social, medical or legal proposals which may assist in restricting the further spread of HIV, and (d) other such matters relating to the issues of HIV in the ACT which the committee considers should be drawn to the attention of Assembly' (6). The National Centre for Epidemiology and Population Health was approached by the Select Committee early in 1991 to conduct a trial 'to assess the impact of a policy shift towards the controlled availability of heroin to people already dependent on that drug' (6). The centre agreed to conduct a feasibility study and broadened the brief to include opiates generally and to some extent other illegal drugs and also consideration of non-dependent as well as dependent users.

The feasibility study was conducted in collaboration with the Australian Institute of Criminology and its aims were to investigate in principle issues covering public health, medical, legal, political, ethical and other considerations. This paper focuses on the relevant ethical issues, exploring a broad range of issues that may (or may not, ultimately) pose ethical problems (7).

At the time of writing, political deliberations and further feasibility research are still underway. Thus it is not yet clear whether there will be a trial at all and, if so, what form it will take. Our aim has therefore been to develop a set of minimum ethical standards which must be met, regardless of the nature of any trial eventually conducted. Whether or not Canberra's own trial goes ahead, these issues will have to be considered by anyone else contemplating similar experiments.

We approached consideration of how heroin should be made available with as few preconceptions as possible. However, we judged it likely that, initially at least, there would be a short-term trial of controlled availability.. Our deliberations were, in that sense broad-ranging.

In another way however, they were tightly focused, concentrating on the ethical issues surrounding a trial, rather than upon ethical aspects of the broader policy questions. We did not discuss, for example: the advantages and disadvantages of current policy; whether heroin use is, or should be, seen as a criminal, social or medical problem; the morality of drug use and drug users or of legislation; why some drugs are illegal and others are not, and so on. Such issues have been widely discussed by others (see, for example references 1, 7-12).

We are not aware of other papers that deal with the full range and complexity,, of ethical issues on this more micro level. Many people have advocated controlled availability, in various guises (2), but they have not dealt with the ramifications, including the ethical questions posed. Specifically they have not dealt with the peculiar ethical issues posed by controlled availability in a trial setting.

There are real ethical issues in providing heroin in a controlled manner, especially, experimentally. We discuss a number of these and outline ways in which they could be addressed. Given that ethical considerations are now a standard part of research design, particularly through the work of institutional ethics committees, this study provides one possible template for how such considerations could be incorporated.

The first part of the paper deals with more general ethical issues raised by the proposed trial and the impact that these issues have on the structure of a trial. The second part of the paper deals with specific detailed problems that need to be addressed for a trial to be run in an ethical manner. Conclusions often depend on the exact nature of a trial, not yet determined at the time of our deliberations. Therefore, as wide a range of alternative approaches as possible had to be considered.

This paper is written in the spirit of setting out considerations, much more than of reaching firm conclusions. That is inevitable given the political role in which we have been cast: policy advisers necessarily advise on options, rather than trying to pre-empt decisions. But it is also inevitable, given the nature of ethical analysis.

General issues for a trial

Should a trial be an experiment or an exercise in public policy?
A supervening issue is whether a trial is to be conceptualized primarily as an 'experiment' in a scientific context (13), investigating the effects of a trial on users and/or the general community or as a direct exercise in public/social policy. Pearn (14) suggests that when a new idea arises the most ethical approach is to conduct a short but well planned pilot study and then to proceed directly to a detailed controlled study which will resolve the issue; that logic argues strongly for conducting the trial as an experiment. However, the question ultimately centres on how adequate a trial, conceived as an experiment would be in terms of fulfilling the rigorous demands of scientific method. The informational demands required for the confident implementation of social policy are much less rigorous. A trial seen in that light might be nonetheless useful even if its outcomes were substantially less clear-cut.

It is unethical to use people as 'subjects' for scientific research unless it is reasonably clear that the results can be adequately evaluated and that they will have a meaningful bearing on later policy considerations (15). Whatever moral objections we have to 'using' people, we object twice over to using people to no good end. Ultimately, there is a difficult balance to be achieved between obtaining clear results with a minimum of confounding variables and producing results that are generalisable and meaningful in terms of instituting policy.

A number of issues militate against conceiving a trial as an experiment.

• It would not be possible to obtain a random sample of heroin users because the exact number and composition of the total population of heroin users is unknown. This would also make it difficult to know how representative the trial participants were of the general heroin-using population.
  
• It may be difficult to devise an ethically acceptable way of randomly assigning volunteers to a comparison group, where this assumes that controls will actively persist with criminal and health-endangering behaviours.
  
• In some ways any locale, including Canberra, which is the proposed site for a trial, is socially and demographically atypical. This may limit the generalisability of any results that a trial might produce. While this in itself would not invalidate a trial, it may limit its wider usefulness.
  
• The limitations inherent in trials, including the limited time span, might not allow all the conditions that would occur in a fully-fledged programme to emerge.

The size and scale of a trial are important for a variety of reasons. First, there should be sufficient participants to allow outcomes to be clearly demonstrated. The sample should, however, be small enough to minimise the number of people who might be adversely affected by an unsuccessful trial or the premature termination of the trial for political reasons. Second, the size of a trial may in itself influence the likelihood of its success. A small trial might be seen as relatively innocuous and therefore be less subject to pressures from local, national and international interest groups, such as illegal drug suppliers and governments opposed to any reduction in sanctions. However, the participants in a small trial could be subjected to harassment from users not in the trial, Who also wanted access to high quality heroin. Further, limiting a trial to one locale could attract users from the rest of the country. Even though they might be debarred from participation in the trial by residency criteria, they could contribute to hassling trial participants and be a source of community problems, especially crime, as well as a drain on welfare services.

There are also numerous pressure groups with strong interests both for and against a proposed trial. The result is that a trial might well run in a highly politicised atmosphere, making it difficult to arrive at a considered judgement on whether the trial should proceed to a full-scale programme. In addition, those opposed to a trial may have an interest in contriving to affect adversely the trial results: participants in a trial may behave in a manner that would not be sustained once a full programme was instituted, being very conscious that 'good behaviour' is essential for the success of a trial.

Furthermore, should a trial demonstrate the efficacy of this form of intervention in a way that is reasonably generalisable, it would be unethical not to proceed to the policy implementation stage. Not only is it unethical to 'use' people uselessly: if no action is to follow from positive results, there is no legitimate reason to under-take the experiment. Of course, the outcomes of a trial would most likely be mixed; for example it might be successful in reducing harm-related behaviours, but it might not be cost-effective. Hence, the criteria for success and failure of a trial should be determined before it begins (and these determinations should include resource considerations). A commitment to running a trial should include a commitment to implementing the resultant policy, if the trial proves successful.

If a trial is conceived as an experiment all the considerations raised above must be carefully balanced. Ethical advisers can merely raise those issues, not resolve them: experimenters must be the ones to provide the necessary assurances regarding the likely conclusiveness of any experiment; politicians must be the ones to provide the necessary assurances regarding the political will to act upon experimental results, if sufficiently conclusive. The role of ethicists here is merely to draw attention to the need for such assurances.

The alternative to treating the trial as an experiment is to make a policy implementation decision directly, based on the best information available. This option would not be hindered by the methodological constraints of the experimental paradigm. Furthermore, it ought to be done completely rather than piecemeal, since past experience (16) has shown that gradual implementation of politically sensitive policy results in an exaggerated build-up of opposition, This in turn means that the policy is never fully implemented and the rationale for the policy becomes irrelevant in the face of such public opposition (17).

Militating against direct policy implementation is the fact that there is little 'hard data' which can be used to decide either whether or not controlled availability of heroin is likely to be successful, or to structure the details of a policy. In addition, should the outcomes of change be detrimental, either to users or to the community at large, it is likely that more people would be affected than if a trial conceived as an experiment was instituted. Also a trial could probably be terminated more easily than a policy, should that be necessary.

Acts and Omissions

The issue of acts and omissions concerns the distinction between the blame attaching to harms which one has somehow caused to happen, as compared to those which one has merely let happen (18). The distinction is, for example, between a person's overdosing on heroin received as part of a trial as compared to overdosing on heroin bought on the streets, or between someone's being involved in a car accident after receiving heroin as part of a trial as compared to crashing after taking heroin illegally. Researchers or service providers who have supplied the heroin in question are arguably implicated in the first case in each example, in a way they would not be in the second cases.

Philosophers themselves are deeply divided on the moral importance of any distinction between such acts and omissions. Some argue for a more deontological approach, insisting that there is an enormous difference between causing harm and allowing harm to happen, so that harm resulting from intentional intervention is a major problem, even if similar or even greater harms might otherwise have happened. Other philosophers argue for a more utilitarian (or more broadly consequentialistic) approach which judges the harms caused by the intervention in light of those which would have happened had there been no intervention (19).

Taking the utilitarian approach, the following arguments could be made with regard to the initial examples given. A person who overdosed on the trial might well have overdosed on street drugs; overdosing on street drugs in fact may have been more likely, since the drug provided in any trial would be of pharmaceutical quality and of known strength and purity, unlike street drugs. Therefore the risks of overdose are lessened on a trial. Similarly, people who drove after administering their drugs in a trial might well have driven after administering their drugs on the street, probably with equal risk of having an accident. (It is, of course, also possible to implement strategies to minimise risks, for example by providing transport for trial participants.)

The non-utilitarian approach, taken to extremes, would seem to preclude any interventionist research which involved a significant level of risk, even if it were directed at a major social problem such as illegal drug use. Even in its less extreme forms, the deontological approach would set a presumption against any intentional interventions designed to alleviate such social problems. By weighting those harms which we have caused considerably more heavily than ones which happened independently of anything we have done, deontologists are in effect saying that social programmes must do very substantially more good than harm before they are prepared to regard them as morally justified. While even utilitarians might find grounds for agreeing that we should be reasonably sure that our interventions will do more good than harm before undertaking them, too heavy a bias in that direction seems difficult to defend, not least to those being asked to continue suffering needlessly.

Counter-Mobilisation

Opposition to a trial would be likely to come from two sources. The first is from people who feel such a trial is wrong or who have genuine anxiety about its likely outcomes. This opposition is legitimate and ought to be respected. The second is from people who have a vested interest in heroin remaining illegal, for example, because they benefit financially from the sale of illegal drugs. Countermobilisation from this group is less legitimate. People in the second group however, would be likely to attempt to use the legitimate concerns of the first group for their own ends; in other words legitimate concerns could be illegitimately exploited.

Countermobilisation could be avoided by instituting the programme directly as social policy rather than by first conducting a trial (16). While this might be effective in circumventing those who have a vested interest, it also stifles legitimate opposition.

It is therefore important to consider carefully the arguments and values of those who legitimately oppose a trial and to balance them against those of the trial supporters. It is important that both sides have equal access to information about a trial and its supporting and opposing arguments. In the course of a trial, the evaluation should, where possible, include measures of variables considered to be problematic and/or important by trial opponents as well as supporters.

This does not overcome the concern that manipulation of public opinion by those with vested interests might compromise the effectiveness of a trial and hence the possibility of a trial proceeding to a full-scale programme. The best ethical advisers can do here, is to flag the problem and alert evaluators of any trial to the need for a sensitive interpretation of negative experiences within it.

Termination of a Trial

A trial would probably be easier to terminate than a policy, if detrimental effects occurred. On the other hand, it might be easier to modify a policy than a trial (especially a trial conceived as an experiment) where any detrimental effects are more arguable and less obvious.

It is crucial to have in advance a list of reasons for halting, or perhaps for modifying, a trial, even before detrimental effects could be shown to be statistically significant. (In that sense, it might be necessary to compromise the scientific integrity of a trial.) To minimise undue political interference during the course of the trial itself, it is important that these termination criteria be determined, and agreed on, by both proponents and opponents of a trial, well ahead of the trial itself. When and how a trial is terminated will have important political ramifications which need to be taken into account when deciding whether or not to proceed with a trial (20).

There is also the issue of what happens to participants at the end of a trial. One view is that with this trial we can be quite confident that it will have benefited participants for its duration, so there is minimal post-trial obligation to them. In this case participants would need to provide informed consent to clean-break termination at the outset of the trial. It has, however, been recommended elsewhere (21) that new therapeutic procedures