What are the goals of psychotherapy, and who decides? If we apply the traditional medical ethical principles of beneficence and nonmaleficence, then psychotherapy, like any other medical treatment, should aim to help the patients with their problems, make them feel better and do them no harm. Most psychotherapists pursue this aim by working with patients to increase their capacity for self-reflection, and to help them become more aware of the links between their feelings and actions.

But this is not as simple as it seems. Although CBT can make people feel better in the short term by removing their symptoms, psychodynamic therapies may not do so. For example, patients with histories of exposure to traumatic events (whether in childhood or in adulthood) may want the therapy to take their horrible feelings away. However, psychotherapists do not take patients’ feelings and memories away, but try and help them to deal with them better. Both CBT and psychodynamic therapy seek to help regulate conscious negative feelings, and modify distorted and dysfunctional meanings of memories. Although these are perfectly reasonable goals for psychotherapists, they may not be what the patient sees as the goal of therapy. Patients may not accept that the meanings attributed to memory are distorted; they may also be unaware of the extent of their negative feelings, especially guilt, shame, and hatred. In CBT, the patient may want to remove all their negative emotion, while the therapist wants to decrease excess emotion.

There are practical techniques that all therapists employ to address these issues clinically. Ethically, however, there may still be uncertainty or conflict about what the goals of therapy are, and who should set them, which can make the issue of informed consent in psychotherapy particularly complex. The ethical principle that underpins the requirement for consent is respect for autonomy: patients should be free to choose or refuse for themselves what treatment they have. Legally, for consent to be valid it must be given voluntarily, by a patient who is competent to make that decision, and the patient must be adequately informed about what the treatment involves, including any possible negative side-effects. However, it may be difficult for therapists to advise patients exactly to what it is they are consenting. Therapists may not be able to predict what patients will experience during the course of therapy, or what the outcome will be. Should therapists inform patients about unconscious transference enactments (both positive and negative), and get specific consent for this possible ‘side-effect’? (Holmes and Lindley, 1989). My own view is that therapists need to warn patients that the process of psychotherapy is not always comfortable, and that many patients feel worse before they feel better (although this may not be true for CBT).

It must also be pointed out that some people do not get better with psychotherapy (Lambert and Bergin, 1994). Successful psychotherapy matches the right therapy and right therapist to the patient, which is why therapists need to develop good quality assessment skills, not just for their own discipline but for others. This means being able to formulate problems in different ways, and think about the ways that this particular patient is most likely to make progress.

For consent to be valid it must be given by competent patients, who are capable of exercising their autonomy in a way that expresses their values and interests, while giving weight to conflicting values and facts. Although in a general sense, most psychotherapy patients will be competent to make decisions about their own treatment, this may not be true for all. For some it is problems with the exercise of autonomy that have led them to seek therapy; for others, their psychological distress may affect their capacity to make choices about treatment. Obvious examples are those people who have very recently suffered psychological trauma or bereavement, or those who are experiencing psychotic symptoms.

There are other groups of patients who may lack competence to make treatment decisions, such as children, or patients with psychotic disorders or learning disabilities. The problem is usually not a global lack of competence, but rather that patients experience fluctuating levels of capacity, or experience rapid changes in how they make decisions. For example, it is hard to know what to make of a child's refusal to have therapy, which they agreed to only a day before. Can this be understood as an informed choice to refuse treatment? Or is this refusal merely evidence that the work has begun? There are also groups of patients are ‘coerced’ into therapy, such as children who only agree under pressure from family.

It may also be much harder to know how to think through issues of informed consent for family, marital, or group psychotherapy (Lakin, 1988). Children, or other family members may only ‘agree’ to therapy because of pressure from others in their emotional network. Although such pressures are an essential part of intimate relationships, they raise dilemmas about the nature of ‘true’ voluntariness. Voluntariness is also an obvious issue in the domain of forensic psychotherapy, as some patients may be mandated to have treatment, as an alternative to prison, or participation in therapy is expected as part of their detention. Clinically, most therapists in forensic or penal settings get to grips with this issue as part of the therapy from the start; but it does not sit easily with classical medical ethical accounts of informed consent.

Psychotherapists, like any other doctor, are under an obligation to do no harm. The question then is what constitutes ‘harm’ in psychotherapy. Just as defining benefit can be difficult, because of the need to consider different perspectives and time scales, so too is defining harm. It is probably inevitable that effective psychotherapy will sometimes cause people distress, at least in the short term. Effective psychotherapy may also have unforeseen effects on patient's lives: an unhappy husband may leave his wife, a child may have to leave his family, and a person may change his/her job. These may constitute benefits in the patient's view, but may be seen as harmful by others. For example, there have been cases where patients who recover memories of abuse in therapy may sever ties with their family, causing distress to all involved. Families have sometimes claimed that the therapist has encouraged the patient in their distressing behavior, or even implanted false memories; either through incompetence, or for ideological purposes. In one case of this kind (Appelbaum and Zoltek-Jick, 1996), the patient did retract her account of their abusive childhood experience, and both she, and her family successfully sued the therapist for negligence.

Clearly, negligence and incompetence are potential risks for all clinicians, and these can (and should) be addressed by training, licensing or registration, and supervision processes. But other types of harm are also possible in therapy, which are not so common in other types of medicine: for example, threats to confidentiality and boundary violations. I will deal with boundary violations in more detail below, but at this point it is relevant to consider the issue of breaches of confidentiality, or the boundary of therapeutic privacy, as a type of harm.

My own view is that the principle of confidentiality may be better understood as the principle of informed consent to disclosure. Therapists are under an ethical duty to obtain their patient's consent before they disclose details of their psychotherapeutic treatment to anyone; including those close to the patient. Although this is undoubtedly a principle that is respectful of the patient's autonomy, it can be problematic when the patient discloses material, which indicates that someone else is (or has been) at risk of harm from them. Most professional codes and guidelines address this issue, and there is relevant case law (see below). In general psychotherapeutic practice, this is probably a rare event; more commonly dilemmas arise when the therapist perceives that it might be helpful to the patient for others to know that a patient is in therapy, and what has been discussed. Of course, the therapist can seek consent from the patient to disclose, but if the patient refuses consent, then the therapist may still face a dilemma. A nonpsychotherapeutic example occurs when an HIV-positive patient refuses to tell his or her partner (with whom they are still having a sexual relationship), and refuses to let the clinician inform the partner. In the UK, professionals are advised that it may be justified to breach confidentiality in the face of a competent refusal.

Clinicians may also come under pressure from others (family members or employers) who may contact them to discuss the patient, or to seek information about the therapy. Again, it may usually be possible for the therapist to seek consent to discuss some agreed upon material with others; ethical problems may arise when the third party asks the therapist not to tell the patient about the contact, usually because it will cause distress to the patient. In such circumstances, the therapist may have to balance the patient's claim to honesty and confidentiality against a possible harm to them.

Getting consent to any form of disclosure can be intrusive into the process of therapy, and this is a particular issue in relation to research. The ethical dilemma here is about whether therapists can use patient material without consent for teaching and research purposes. Traditionally, this has not been an issue for psychotherapists who have assumed some ownership over their experience in the therapeutic space, and who have presumably also assumed that the breach of confidentiality is justified for the public good that arises from teaching and research. The good consequences justification seems plausible enough; clinical material is essential for teaching trainees, and for research. What is different from 20 years ago is the increased social emphasis on respect for individual patient autonomy, in the form of control and ownership over anything personal, which means that therapists may be unwise to assume that it is ethically unproblematic to use patient material for teaching and research without consent, even if it is disguised.

Although some journals do not require patient's consent to publication of their details, there are others that require that the patient has not only given consent, but has read the article in which their case is mentioned or described. Getting consent after therapy has ended is not necessarily the answer, as this might be just as intrusive or distressing for the patient (Winship, 2002). Then there is the question of content. Do patients have to agree with what is written? Can they disagree only with matters of fact, rather than opinion? And to what extent is the therapist allowed to ‘own’ his or her own view of the therapy, and discuss it without the patient's permission?

The issue is further complicated because the therapist's capacity to have a personal and intimate relationship with the patient is part of the therapeutic process (Klauber, 1986). The therapist makes their mind available to the patient to assist them; and this involves taking seriously their own feelings and thoughts about the patient. Keeping a record is a way to do this; and also helps with the process of thinking and supervision. However, generally speaking, the therapist does not share her thoughts and feelings about the patient with the patient (the extent to which she might do this is a matter of technique, reflection, and supervision). It is unlikely that the patient will find it helpful to discover what their therapist thinks about them from simply reading the process notes. My own practice is to make a brief note in the medical record that the session has taken place, and then keep some brief notes of the main themes of the session in a file in my office. The American Psychiatric Association recommends this practice to psychotherapists, and this would be consistent with general advice on good record keeping from the UK Royal College of Psychiatrists. Legal jurisdictions, both in the UK and in the US, do, however, make it theoretically possible for the patient to have access to those notes, unless this access would constitute harm to them or another person. The ethical issue here is that therapists cannot assume that they ‘own’ the notes of their meetings, and that only their views about the process notes need be consulted.

In CBT, it is common practice for the therapist to record the patient's experiences in therapy (thoughts, feelings, behaviors), in the same way as the patient does in ‘homework’ assignments. For both therapist and patient, sharing of these records and notes is often therapeutically helpful. What may be problematic is when the patient is recording information that may be misunderstood out of context (for example, in cases of sexual dysfunction, or violence). CBT therapists may also wish to record impressions of the patient and his/her progress that they do not wish the patient to see; and to which the patient may have legal access.

A boundary then is a construct that defines domains as separate and different. One thinks of the boundary round a cricket pitch, or the stage of the theatre, which must be set out and delineated for the play to happen. In medical ethics, the boundary is between personal and professional identities. The doctor (generally) undertakes not to bring his personal identity into the professional space. This is crucially important in medicine because, unlike other professional spaces, the patient is vulnerable as a result of their illness and disease, and may be less able to protect themselves. Professionals do have additional power that comes with knowledge, and like all power discrepancies, this can be abused. Furthermore, the success of any therapeutic relationship relies on trust; in his vulnerable state, the patient has to rely on the doctor to put the patient's interests first, and not exploit his vulnerability. If the patient cannot trust the doctor to do this, then he will not be able to use the therapeutic relationship to its full extent.

So the boundary between the personal and professional identity of doctors needs to be set and thought about as part of regular clinical practice for all doctors. Good doctors pay attention to the construction and maintenance of professional boundaries throughout their working life. But psychiatrists and psychotherapists have a particular duty to think about these issues because there are a number of cogent reasons why boundary setting and maintenance is especially significant and important in psychiatric and psychotherapeutic relationships. First, psychiatric and psychotherapeutic patients are especially vulnerable insofar as they are mentally distressed. Second, the psychotherapeutic space has to be a particularly private one, to enable the patient to explore the most delicate of feelings, especially those of a potentially shameful nature. As many commentators have noticed, the increase in the numbers of people seeking therapy and counseling mirrors the fall in the number of people who attend a church; another place that used to be associated with private and personal disclosure, and self-examination.

Thirdly, for any psychological therapy to be effective, there has to be a trusting empathic relationship between the therapist and the patient, which promotes intimacy. At both conscious and unconscious levels, patients often reenact, with the therapist, relationships they have had before with other intimates (especially common in patients with personality disorders). It is the intimacy of the therapy that makes it useful, by allowing an examination of these reenactments. However, because human intimacy is powerful, and most psychotherapy patients seek therapy because of problems of intimacy with others, it must be managed safely. Boundary setting and maintenance help to establish a secure space to look at what goes wrong with intimacy, and help to think about different ways of managing interpersonal relating. The patient has to trust that therapists will not exploit that intimacy for their own ends. The therapist has to commit to not doing so; and still balance psychological intimacy with distance in the interests of the therapy (Casement, 1985; Karasu, 1992).

Finally, on the theme of intimacy, all psychotherapists have experience of situations where judicious self-disclosure is immensely helpful to the therapeutic process (Yalom, 1986, 2002). The professional skill then is to know when and how to do this, in a way that takes the therapeutic process forward, and is not exploitative or abusive to the patient. The principle of saying less rather than more is a good one; it is also helpful to develop a few stock phrases that gently re-reroute inquiries about the therapist's personal identity (‘This is space for you, not me’; ‘I wonder if it's easier to talk about me than you’). Inappropriate self-disclosure is discussed in more detail below, as a type of boundary crossing or violation, which it may be. But there is a real danger that rigidly refusing to ever say anything about oneself has a negative effect on the development of trust and intimacy, and can also be a way for the therapist to enjoy remaining in a controlling and powerful role.

Gutheil and Gabbard (1993) make a useful distinction between boundary crossings and violations (Box 40.2). For example, if the patient brings a gift to the therapist, this is a crossing of the boundary between the personal and the professional identity. The professional identity does not require, and is not entitled to, a gift. The giving of a gift is an indication of the patient's wish to relate more personally to the therapist. If the therapist accepts the gift, they are relating more personally. The balance between their professional and personal identity alters. This may or may not be a bad thing; it may be mutative moment for the patient, or it may simply be the therapist pursuing her own wishes or needs of the moment.

The other point about boundary crossings is that they may or may not be consciously intentionally initiated by either party. A chance meeting outside the therapy session, for example, is still a crossing of the boundary, which will need to be addressed technically in terms of its meaning for the patient. Although the therapist may not have initiated the boundary crossing, there is still an ethically sensitive moment when the therapist's personal and professional identity meet. Self-disclosure is another common area where therapists may say more than they consciously intend to the patient. The fact that it is done unconsciously does not make it any less ethically sensitive.

The first step, in terms of ethical analysis and practice, is to notice that the boundary crossing is happening at all. There may be many ways of responding, and the decision-making process must be both ethical and psychodynamic (Box 40.3). The therapist has to formulate an understanding of what this boundary crossing is about for the patient, in order to match their response. If the boundary crossing is aggressive in nature, then this may indicate that the patient is anxious, and needs a reassuring response. An apparently caring or affectionate type of crossing may indicate that the patient needs reassurance that the therapist can keep the boundaries firmly, and is tough enough to keep to task; this is obviously also the case for challenges to therapeutic authority. As with all ethical dilemmas, the therapist will be helped if they discuss the issue with supervisors and colleagues; they also need to review their technical and communication skills. But the key ethical issue is to notice that the boundary is under pressure and needs attention.

Perhaps the most common example of boundary crossing in therapy is inappropriate self-disclosure by the therapist. Again, the patient often initiates this. I give an example in Box 40.4, together with some ways of responding. Like the ethical dilemma about the gift (and many other ethical dilemmas in medical practice), there has to be a resolution; it must be a good quality one, and it will involve good communication skills. What is unprofessional is not to explore whether there is a special meaning of the boundary crossing at this point both for the patient, and the therapist. It cannot be assumed that these types of transaction have no meaning or significance. For those therapists working with very disturbed patients, and those who have previously been exploited in intimate relationships, supervision is highly advisable, because boundary crossings and violations are so common (Holmes and Lindley, 1989). I would argue that it is ethically unjustifiable for a therapist not to obtain supervision for this kind of work, although not all would agree.

Boundary crossings may or may not be harmful. Boundary violations are those crossings of the boundary that cause harm to the patient, usually because they involve an exploitation of the power difference and the trust between the therapist and the patient. Physical boundary violations, especially those of a sexual nature, change the relationship between the therapist and patient so profoundly that the therapy is lost. The therapist's mind is no longer available to the patient in the professional way it once was; and this means that the therapy has been harmed.

There are many other types of harm done by sexual boundary violations, particularly. First, the patients most likely to be exploited by their therapists in this way are those who have already been victims of sexual exploitation by previous caregivers; this is often the reason that they sought therapy in the first place (Kluft, 1993). The abuse by the therapist is a reenactment of their prior experience, and they are usually placed in exactly the same position as they were before: they have to keep the relationship secret, in order to protect both the abuser and other family members, and they are made to feel responsible for their therapist's comfort, pleasure, and wrong-doing. Research on the effects of sexual abuse by therapists shows that patients are likely to relapse and deteriorate, especially when the relationship ends (Jehu, 1994).

Studies of therapists who sexually abuse their patients have found them to be a heterogeneous group, which includes people who are young, old, experienced and inexperienced, male and female (Gabbard, 1989; Schoener, 1995). Some engage in this behavior repeatedly, as a means of getting a sexual partner; for some it will represent a one-off response to external stress. It is probably safest for therapists to assume that everyone (including themselves) is capable of boundary violations; that no one is immune to the risk. As suggested above, this is why supervision is necessary at times for all therapists, even the most experienced, especially for work with difficult and complex patients.

Boundary violations are not only harmful; they also represent a wrong done to the patient. Boundary violations may therefore have legal repercussions. Therapists may be sued for negligence or malpractice; rarely, they may be subject to criminal charges of assault (Strasburger et al., 1991). In some states in the USA, it is a criminal offense to have a sexual relationship with a patient, even after the therapy is ended. Professional sanctions are also likely: therapists who have sexual relationships with their patients usually have their professional registration or license revoked, in recognition of the fact that they gave up their professional identity when they began a personal relationship with the patient.

There are many reasons why therapists cross or violate the boundary between the personal and the professional domain. As we have seen, boundary crossings are commonplace in any setting where the patient and the therapist are involved in a long-term therapeutic relationship, presumably because it is hard for therapists to exclude their personal identity on an indefinite basis. Thus we should not be surprised to find that boundary crossings are common in long-term residential care. Boundary crossings and violations are also more common in relationships of intimacy combined with a power differential; particularly where that power should be used for therapeutic purposes. There is a similarity between abuse by therapists and abuse by parents (Gabbard, 1989): both are in roles of power involving care, trust, and intimacy over time.

Another particular difficulty for boundary setting and maintenance in psychotherapy is that the therapist's personal identity is part of her professional identity; the boundary is opaque and semipermeable, rather than hard and clear. Sarkar (2004) makes a nice distinction between the ‘who-ness’ of a person and the ‘what-ness’ of a person, in terms of identity. Thus, for surgeons, it may be possible for what you are (a good surgeon) to be different from who you are (e.g., a bad man). A group of surgeons will have similar professional identities and practices, regardless of how different their personal identities are; for example, the fact that they cheat at cards or are dishonest in other ways will not affect how they carry out a splenectomy. To some extent, this is also true for CBT therapists; their personal identity is less interwoven with their professional identity.

However, for the psychodynamic psychotherapist, her who-ness is intimately connected with her what-ness. Her personal identity is part of the professional identity; indeed, the long and expensive psychotherapy trainings are designed to help the trainee explore how their personal identity influences their professional identity. This aspect of training is essential precisely because the boundary between identities in psychotherapy is not always hard and clear. Therefore the therapist must pay constant attention to when, how, and why the boundary is being pushed or crossed; remembering that it is not just the therapist who is doing the pushing or crossing (Joannidis, 2002). In terms of reenactments, the patient also pushes and crosses the border, inviting the therapist to relate more as a personal figure than a professional. Lastly, if one considers that much of the pushing and crossing of boundaries in psychotherapy is done unconsciously as well as consciously, it is clear that boundary setting and maintenance occupies much of the therapist's thinking time.

Within general psychotherapeutic practice, there may be different ways of understanding and responding to boundary crossings as ethical dilemmas.

Indeed, they would hardly be dilemmas if the answer were so very clear and obvious. But there are some absolutes in relation to boundary setting and maintenance. Sexual relationships with patients do both harm and wrong to the patient and are therefore unethical. Financial exploitation is not only unethical, it is likely to be illegal. In a recent American case, a therapist was charged with insider dealing when he used information obtained in therapy sessions to make money on the stock market. Any physical touching of a patient needs to be thought about carefully, before and after it happens: although it may be therapeutically justified, it is a significant boundary crossing and should be treated as such. I reiterate that supervision is an ethical necessity; while it cannot prevent boundary violations taking place altogether, it can offer containment for the therapist's feelings as they are stirred up in the psychotherapeutic process.

The other absolute prohibition in terms of boundaries is the prohibition on gossip; specifically, talking about patients and their stories without their consent, and for no therapeutic purpose. The use of the term ‘gossip’ in this context may seem trivial, but it has been described as a subtle form of social aggression, and it can do enormous harm because information is not contained. Gossip is the antithesis of therapeutic discussions of patients; it is the use of individual's private stories for the gossiper's enjoyment, and the entertainment of others. The pleasure and excitement of having and disclosing secrets about others is very powerful, and very tempting; as can be seen daily in the tabloid press and popular magazines. The principle of confidentiality, as well as being respectful of autonomy and promoting trust, is valuable because it is the patient's strongest defense against gossip.

There are, however, circumstances where the therapist will want to breach confidentiality for purposes that are not to do with their own pleasure or entertainment, or even for the enlightening of others. I want now to turn to violations of the boundary of privacy in the public interest.

Traditionally, psychotherapists have privileged their therapeutic duties over their duties as citizens. However, there have been cultural changes over the last 30 years that have challenged this position. These cultural changes are reflected in legal cases addressing the degree to which therapists must honor their duties to society as well as their patients.

The most cited case in the context of risk and psychotherapist disclosure is that of Tarasoff. Appelbaum (1984) provides a detailed account and there is a useful review by Herbert and Young (2002). An ethical dilemma arose for a therapist when a patient disclosed in therapy sessions that he was thinking of killing a young woman (Ms Tarasoff). The therapist informed the local university police who interviewed the patient and let him go. The patient never returned to therapy, and several months later, he killed Ms Tarasoff. Her family successfully sued the therapist and his employers, on the grounds that the therapist had a duty to both warn and protect Ms Tarasoff, and he had failed to do either. The court's legal response to the therapist's dilemma was to find that the duty to public safety outweighed the duty to preserve the patient's confidentiality, and that the therapist should disclose information that indicates risk to others, even in the face of patient refusal.

Twenty-seven US states have imposed a duty on psychotherapists to breach confidentiality, when a patient makes an explicit threat to physical harm to an identifiable person; either by warning the intended victim or involving the police. Nine states leave it up to the therapist to decide; 13 have no position at all. The American Psychiatric Association Code of ethics makes it clear that it is sometimes necessary for psychiatrists to breach confidentiality to protect others from ‘imminent danger’. Therapists in the USA are also mandated to disclose information that indicates risk of abuse to a child.

Therapists in the UK are not so mandated, and there have not (yet) been any comparable cases with that of Tarasoff. There have been relevant cases about confidentiality in therapeutic relationships. In W.v.Egdell, the court found that a psychiatrist would be justified in breaching patient confidentiality in the interest of public safety, and might have a duty to do so. In Palmer. v. Tees HA, the court found that a psychiatrist would have a duty of care to an identifiable person at risk from a patient, and that therefore the psychiatrist would be justified in warning the victim and breaching confidentiality. The UK General Medical Council (GMC, 2000), which provides professional ethical guidelines for psychiatrists, also supports the argument that confidentiality may be breached in order to prevent harm to others.

None of the English cases require the therapist to breach confidentiality in the public interest; nor is there any problem with disclosure of material where the patient has consented. If a therapist is concerned about risk of harm to others, there is no reason why she cannot discuss the risk with the patient, and disclose relevant information with the patient's consent. The ethical dilemma arises when the patient refuses to consent, or the therapist does not wish to ask the patient for their consent, but wishes still to disclose information to third parties.

Finally, the principle of absolute confidentiality to an individual patient is hard to maintain for therapists who work with groups and families. Clearly, the group psychotherapist has multiple duties to the group members, and the ordinary prohibitions on gossip apply. But it may be difficult to balance the conflicting interests of the different members of a group; what should happen if a group member tells the therapist something, but begs the therapist not to tell the rest of the group (Yalom, 1986)? It is usual practice in group therapy to explain to patients that all information is shared with the group from the start, and that the therapist does not keep secrets. Assuming that this is the case, the group therapist will not generally wish to agree to keeping the secret (on good clinical grounds), and will want to challenge the request in various ways. Legally, however, the patients have control over their own information. Here the dilemma is reversed; it is unethical to practice therapy poorly, and keep the secret, but it is also unethical (and possibly illegal) to fail to respect the patient's wishes. Similar issues arise in family therapy, especially if the therapist discovers in the course of therapy that a child or children have been harmed, and remain at risk of harm. The duty to protect the especially vulnerable may outweigh the therapist's duty to maintain the boundary of confidentiality.

In the current cultural climate, there is a good ethical and legal case for the therapist breaching confidentiality in cases where she perceives that there is a high risk of imminent harm to identifiable others, and where the disclosure may reduce the risk of harm. Some authorities (W.v.Egdell) will argue that she has a duty to do so. The social benefit in preventing harm (probably) outweighs the harm and wrong done to the patient.

However, it is important to recognize the powerful counterarguments to this position too; the strongest being that the therapy will be damaged, the patient will be harmed and future patients will be deterred from seeking therapy if therapists disclose patient information to others. The consequences of breaching confidentiality could be worse than not doing so, in the long term. There is also an argument (based on the European Convention of Human Rights) that everyone has a right to a private life, even those with mental illnesses or dangerous thoughts.

On a more psychodynamic note, Gutheil (2001) argues that it is the healthy part of a patient's mind that ‘employs’ or contracts with the patient, and breaches of confidentiality may still be consistent with respect for that aspect of the therapeutic alliance with the patient. It is also respectful to discuss breaches of confidentiality with the patient, and explain to them what the consequences will be. They do not have to like the consequences, but this does not mean that they cannot be supported. If the patient feels so betrayed that they can no longer continue in therapy, it is still possible for this hurt and betrayal to be acknowledged, and for the therapist to assist the patient in continuing their therapy.

Just like any other form of medical treatment, psychotherapy is not practiced in a social and legal vacuum. One of the enduring myths of medicine is that practitioners use their skills and knowledge in a value-free way, for no other purpose but the patient's good. However, there is good reason to think that medical diagnosis and treatment are value-laden processes from start to finish, and mental health is no exception (Fulford, 1989). The values of the social groups and cultures to which the therapist and patient belong will infuse the dialogue that takes place in therapy. Tensions may particularly emerge in relation to ethnicity and racial identity, gender of either therapist or patient and religious or political beliefs.

For example, take the concept of personal autonomy. Many writers about the values of therapy argue that development of a sense of self-worth and integrity is a crucial goal for psychotherapy (Holmes and Lindley, 1989; Hinshelwood, 1997). Patients are encouraged to take themselves seriously, and to think of themselves as autonomous agents who can choose for themselves, and should (generally speaking) choose courses of action that will help them to flourish as individuals.

Such a vision is consistent with the values of the Enlightenment, which have dominated European thinking since the seventeenth century (Baumeister, 1987). However, such an individualistic, or indexical, view of the self is not found in every cultural group. Different cultures understand the self as referential, rather than indexical (Landrine, 1992); that is, that the experience of the self, rather than being a single orientation of all one's social interactions, is constructed with reference to relationships with others, and social roles. A common example of this is found in British-born Asian men and women, who may seek therapy when their Western experience and expectations of indexical selfhood can clash badly with their Asian parents’ experience and expectations of referential selfhood (depicted most poignantly in the film, East is East). It may not be helpful for the therapist to take a view on which position is right for the patient; equally, it may be hard to avoid doing so.

To complicate matters further, we do not always understand the relationship between individual and cultural identities and belief systems; cultural stereotypes may blind the therapist into assumptions about the individual's inner world (Akhtar, 1995; Dalal, 1999). How easy is it to assume, for example, that a patient who is a refugee espouses liberal values? The psychoanalytic framework that supports our therapeutic technique is not only a scientific theory of psychological development; it is also a collection of culture-specific beliefs, which may conflict with those of the patient. If the values of the therapist and the patient clash, must this always be understood as acting out of transference and countertransference, or can there be a real political and ethical diversity also in the therapeutic encounter? The answer to the latter question is that both processes are operating, and the therapist's job is to keep both perspectives in mind.

The legal framework in which both patient and therapist function is also relevant here. Psychotherapeutic relationships are not outside the law, either civil or criminal. Existing statutes and case law that apply to medical treatments will also apply to therapy. Thus, psychotherapists acquire a legal duty of care when they work with a patient, and that duty requires them to practice in accordance with a reasonable body of medical (psychotherapeutic) opinion. This means that they must exercise reasonable professional skills, make logical treatment decisions, conform to professional ethical guidance where it exists, gain informed consent to treatment, and tell the truth when asked. They also have to abide by the laws of the land; so as described above, psychotherapists in the USA are mandated to report suspected child abuse, even when the information comes from their patient, who may be the abuser. In such circumstances, the therapist may feel a conflict between respect for the law, and the avoidance of harm to children, and concerns that disclosure that lead to harm to the patient.

Are there circumstances where the therapist should break the law out of respect for another ethical principle? One answer might be that the therapist is free to break the law at any time in pursuit of their personal values, but law breaking as a professional duty seems incoherent, when most professions regard respect for the law as a major ethical principle guiding conduct. Having said that, if the therapist is faced with patently unjust laws, she might argue that the pursuit of justice requires her to break the law; this presumably was the case for psychiatrists in Nazi Germany who did not comply with the euthanasia laws for psychiatric patients.

Legal frameworks are an issue for therapists who work with those who have committed crimes, or who may have therapy mandated in some way. For example, forensic psychotherapy cannot claim that it is operating in a nonjudgmental value-free way. The psychotherapist is not free to be nonjudgmental about the patient's actions, or to allow the patient to set his or her own goals in therapy (Adshead, 2000, 2002). In institutions set apart for offender patients, therapy that does not look at the offending behavior, or does not seek to help the patient to act differently in the future, is probably not therapy that anyone (including the patient) would think is much use. It may not be enough for the patient to simply be more reflective; as one of my patients said to me, ‘I'm still the same person I was when I came here, I'm just more aware of myself’.

Forensic psychotherapy is only one example of how the social setting may influence on the psychotherapeutic work. Murray Cox (1976) suggests that all relationships are structured by time, depth, and mutuality, and the social setting affects all three dimensions. The type of hospital one works in, as much as training and personal style, may influence how the clinician allocates time, the depth of the relationship formed with the patient and the degree of mutuality that is possible. Private practice operates as another type of framework, especially in terms of mutuality and the nature of the contract between therapist and patient. In both the institution and private practice, there is a contract between the therapist and the patient, but the contracts are different.

Private practice is the norm for most psychotherapists in the USA and Europe. However, in other countries, medical psychotherapists may work in a centralized public health system, free at the point of service. In such a system, resources have be managed and allocated; there is not enough for all to have as much as they wish. Choices have to be made about how resources are allocated; and for psychotherapists, this will involve not just the resources allocated to psychotherapy, but how resources are allocated to mental health services in general. Mental health services are generally poorly resourced, and dynamic psychotherapy is often seen as low priority, partly because of a perceived lack of an evidence base.

There is not room here to discuss this debate in detail, which is well described in a published debate by Holmes and Tarrier (Holmes, 2002; Tarrier, 2002). In essence, the ethical debate runs on utilitarian lines: if there are not enough resources to go around, how should we share them out? One way to allocate scarce resources is to give more money for treatments that are shown to be effective, using an agreed form of process for assessing that effectiveness. But what does ‘effectiveness’ mean in psychotherapy? Is the assessment process able to measure subtle forms of benefit? Where should the views of mental health service users fit in? Perhaps the most important thing to notice is that there is a debate to be had, and its fundamental nature is both political and ethical.