Therapies that set a time limit all manage the frustrations and disappointments that this can arouse, in both patients and therapists, but they do so in contrasting ways, either to facilitate them or to minimize them. Despite the polar differences between these views, there is little research evidence base for which of these two approaches leads to the best outcomes, for which clients.

The former approach sees the time limit itself as of immense therapeutic significance and potential. James Mann (1978) is the prime exemplar and most eloquent advocate of this view. The time limit of therapy is seen as a profound metaphor for the finiteness of time itself for any individual. It evokes, he argues, the reality of loss and death, but, if faced and endured, is a powerful maturational experience. Mann gives the time factor most attention, but a number of psychodynamic and relational therapies emphasize setting an exact, nonnegotiable time limit, to facilitate the experience of anxiety, disappointment, and anger. Expressing these warded-off emotions, it is argued, within a facilitating therapeutic relationship, leads to their being safely experienced, assimilated, and mastered. These therapies work on the assumption that what was perceived as catastrophic can be transformed into something both manageable and personally empowering.

Other therapies take the opposite line, reducing the significance of the time limit, either by interpreting it very flexibly, by using follow-up appointments, or by making it clear further therapy will be available in the future ‘as needed’. For example, Budman and Gurman (1998) argue that there is little empirical evidence that emphasizing termination in therapy leads to better outcomes, and assert that it is therapists rather than patients who have difficulty ending. They do not see therapy as a ‘one-shot’ operation, instead preferring to conceptualize the therapist as a ‘psychological family doctor’, available over the life span to respond to different needs in a developmental process. They also emphasize a team approach, with no one therapist being all things to all patients. Cummings (1991) describes a similar practice, remaining available to patients for return visits and intermittent brief therapy. He describes making an explicit commitment to patients. ‘I will never abandon you as long as you need me. In return for that, I want you to join me in a partnership to make me obsolete as soon as possible’ (Cummings, 1991, p. 40). Despite, or perhaps because of, this contract, he reports that one episode of therapy only lasts between one and 20 sessions.

On the whole it is those therapies rooted in psychodynamic theories that emphasize the time limit and the therapeutic value of the fixed termination, and those rooted in pragmatic eclecticism, and the cognitive-behavioral approaches, which are less concerned with this.

The use of time in brief therapy goes beyond fixing the number of sessions or setting a time limit. It can include varying the length of sessions, the frequency of sessions, and the flexibility with which therapy is delivered. For example, Mann's rigid adherence to the 12-session limit does not preclude considerable flexibility in how they are delivered. Variations to weekly sessions are mentioned, including in one instance weekly 15-minute sessions for 48 weeks. The key issue is that there is no ambiguity or uncertainty about the pattern and duration of the sessions (Mann, 1978). Budman and Gurman (1988) cite Johnson and Gelso's (1980) review of the effectiveness of time limits to argue that there is little empirical justification for traditional weekly sessions, suggesting that after weekly contact to begin the improvement, time alone is required for continued improvement. They favor more intensive work for four to eight visits, then ‘spreading out’ the sessions, to reduce dependency and enhance self-efficacy. It could be argued that there is an irresolvable tension between the therapist being responsive, adaptable, and pragmatic, and the requirement to provide a consistent, secure framework, which is not colluding with an unreflective enactment of a dysfunctional relationship pattern. Binder et al. (1987) advocate a compromise where strict time limits are replaced by a ‘time-limited attitude’ with defined but flexible duration of treatment. There is also potential for research to address these questions empirically.

The therapeutic focus is the second broad factor shared by most, if not all, brief therapies. It can relate to manifest symptoms or a presenting problem. For example, cognitive therapy was originally a brief problem-focused therapy for depression (Beck, 1979). Most brief cognitive and behavioral therapies have a problem focus, such as panic (Salkovskis and Clark, 1991), although longer-term cognitive therapy with a schema focus has also been developed (McGinn and Young, 1996). The focus for interpersonal therapy (IPT) is developed in the early sessions, relating to one of four problem areas: grief, role disputes, role transitions, and interpersonal deficits (Klerman et al., 1984). Psychodynamic, relational and some eclectic therapies often take an intrapsychic or interpersonal focus, a central emotional dilemma or an issue in personal development. Such a focus is referred to in diverse ways; the ‘dynamic focus’ (Schact et al., 1984), ‘core conflictual relationship theme’ (Luborsky and Crits-Cristoph, 1998), ‘core neurotic conflict’ (Wallerstein and Robbins, 1956), ‘nuclear conflict’ (Alexander and French, 1946), ‘central issue’ (Mann and Goldman, 1982), ‘interpersonal-developmental-existential focus’ (Budman and Gurman, 1988).

Omer (1993) describes how the focus in brief therapy has tended to be either symptom focused or person oriented, and argues for the value to the therapeutic alliance of combining the two into an integrative focus. Ryle's (1990) CAT is a good example of a therapy using an integrative focus that relates symptoms and presenting problems to underlying ‘problem procedures’ and ‘reciprocal role relationships’. This method is also unusual in the degree to which the focus is made explicit through collaborative work on successive drafts of letters and diagrams.

Safran and Muran (2000) describing their brief relational therapy (BRT) warn against a focus on content. Their conviction that the therapist is inevitably embedded in an enactment of relationship patterns makes them wary of early formulation of the central difficulty. Instead they offer a process focus, emphasizing the importance of developing mindfulness, the capacity to observe internal processes and actions in relation to other people. A focus on process rather than content is also characteristic of Hobson's (1985) psychodynamic IPT where the development of a shared language for feelings between therapist and client is one of the primary tasks.

An active therapist is a feature of working in short-term therapy. In the behavioral and cognitive methods, therapists have always been active, irrespective of the length of treatment, in collaboratively setting an agenda for the session, teaching, giving advice, using Socratic questions in guiding discovery, setting homework, suggesting structured activities and coaching. In the psychoanalytic and some humanistic traditions, the therapist is relatively less active—waiting for the client to speak at the start of the session, refraining from intervening to end a silence, following the patient's (or client's) train of thought and rarely initiating a topic or actively structuring the session. Indeed, in some forms of psychoanalytic work, the therapist is abstinent to the point of appearing personally opaque. In brief psychodynamic therapies, by contrast, the therapist is more active in interpreting the transference, unconscious conflicts and in confronting resistance. Therapists in eclectic, relational, or integrative modes are also active, for example in clarifying and collaboratively exploring the client's material, negotiating treatment goals, structuring sessions, making links between interactions in the therapist–client relationship and past relationship patterns, and possibly setting or discussing between-session tasks.

The brief therapies also differ widely in the range of difficulties to which they are considered applicable. Groves (1996) is not unusual in terming this ‘patient selection’, although of course one is not selecting patients who suit the therapy so much as selecting therapy to suit the patients. The different forms of symptom-focused CBT have intrinsic selection criteria, with separate therapy ‘packages’ developed for panic, depression, health anxieties, obsessive-compulsive disorders, eating disorders, substance abuse, anger management problems, posttraumatic stress disorders (PTSD), and suicide prevention (Salkovskis, 1996; Clark and Fairburn, 1997). Some brief psychodynamic therapies are restrictive, with long lists of exclusion criteria. For example, Sifneos (1972) considers his Short Term Anxiety Provoking Psychotherapy suitable only for people of above average intelligence, who have had at least one meaningful relationship, are able to express emotion in the assessment, have a specific chief complaint, are motivated to work hard, and have realistic expectations of treatment. Messer (2001) described brief dynamic therapists as appropriately avoiding clients who are too severely disturbed to use an insight-oriented approach or those who need more time to work through their problems, but other brief therapists take a more liberal view of suitability. For example, Wolberg (1965, p. 140) states that ‘The best strategy, in my opinion, is to assume that every patient, irrespective of diagnosis, will respond to short-term treatment unless he proves himself refractory to it.’ Garfield (1995), has only the three criteria that the client be in touch with reality, is experiencing some discomfort, and has made the effort to seek help. Malan (1976) while emphasizing the importance of thorough psychodynamic assessment in predicting suitability for brief intensive therapy, used the patient's response to trial interpretation during assessment as a guiding principle. He was also very aware that despite considerable effort, therapists are unable to predict very accurately who will do well or badly in brief therapy, a point underlined by Binder et al. (1987).

The question of who is more likely to benefit from brief or from longer-term therapy is an empirical one. Although there is still inadequate research evidence to inform many practice decisions, findings on the relationship between treatment length and outcome for clients with different levels of disturbance are beginning to have an impact on service planning and delivery. After describing the range of approaches to brief therapy, we review research on these issues.

From this overview of common features of brief therapies from a range of theoretical backgrounds and practice methods, we can discern some general working assumptions for brief therapies.

• This way of working tends to see therapy as catalyst for change in a complex system rather than as a ‘one-shot’ curative method.

• Therapists aim to maximize the therapeutic alliance and avoid regression.

Other more specific techniques include:

• intensifying therapy processes through use of a strict time limit, to maximize the therapeutic impact of working through anxiety aroused by termination issues, and

• psychoeducation, collaborative empiricism, and skills in self-observation to foster the alliance and avoid regression.

Although brief and time-limited therapies have demonstrated their worth, there are no grounds for complacency. Despite some good results for briefer therapies in randomized trials, relapse rates are high, particularly in depression. There also remains the difficulty of generalizing good results obtained in randomized controlled trials to routine service settings, where outcomes depend on many other factors, including treatment milieu, skills of staff, referral practices, availability, and access. For example, despite growth in evidence-based clinical practice guidelines (Department of Health, 2001; Parry et al., 2003a) service configurations are not delivering the therapies likely to be effective for those who need them. For example, a UK survey showed that in the year 2000 very few people with phobic anxiety were receiving CBT (Office of National Statistics, 2000).

Nor can we assume brief work is a panacea. There are cogent arguments for ‘more is better’, mainly arising from naturalistic studies of dose–response (Hansen et al., 2002) and from consumer surveys of psychotherapy recipients (Seligman, 1995). The former show that, although there are ‘diminishing returns’ from longer-term therapy, improvement rates continue to rise up to 2 years. The latter suggests that among individuals who choose to reply to a consumer survey, those who have received longer-term therapy are on the whole more satisfied with their progress and rate their own improvement more highly than those in brief therapy encounters.

There is an argument that a distinction needs to be made between therapies given under private contractual arrangements and those funded by third parties, either in health insurance schemes or public sector provision. In the first case, the individual tries to make an informed judgment on what therapy length and style they will find most beneficial and negotiates this with their chosen therapist. In the second, issues of efficient use of finite resources and equity of access also come into play, at the level of whole-system provision. A balance is inevitably struck between individual benefit and overall benefit at the level of the population served. Unfortunately the types of evidence needed for such policy decisions to be well-informed is thin on the ground, with most trials designed to address clinical questions rather than service ones (Halpern, 1999). The growth in health services research methods in psychotherapy is likely to continue to develop, to provide information on relative costs and effectiveness using pragmatic trial designs with economic evaluation. Brief therapies are potentially cost efficient, although as cost-offset may be most marked for severe presentations, longer-term therapy too has the potential for cost-effectiveness (Gabbard et al., 1997).

Early psychoanalytic therapies were much briefer than their successors—some of Freud's early therapies were very brief indeed, famously no longer than a walk in the woods. Modern brief psychodynamic therapies have their roots in the pioneering work of Ferenczi (1920) and Alexander and French (1946). The latter authors felt that, although psychoanalysts knew ‘there is no simple correlation between therapeutic results and the length and intensity of treatment’, they clung to a belief that quick therapeutic results could not be genuine. They believed they must be either transitory outcomes due to suggestion or an escape into ‘pseudo-health’—a view that many still hold today. In the 1960s, Malan, influenced by Balint, both British psychoanalytic therapists practicing in London, and Sifneos in Boston USA, developed methods for shorter-term psychodynamic therapy. The influence of these methods has been far reaching. The result has been, for selected patients, the widespread application of psychoanalytic principles over 10–25 sessions, where therapists reflect on, clarify, interpret, and confront interpersonal patterns, wishes, conflicts, and defenses (Messer and Warren, 1995; Messer, 2001).

Balint was a psychoanalyst whose considerable innovations were met with some distrust and skepticism from the analytic establishment. His experimental work with shorter-term therapy depended on the idea of establishing a focus for the therapy and working persistently with this, rather than being distracted by other aspects of the patient's difficulties (Balint et al., 1972). While orthodox in his use of structural psychodynamic theory and technique, he was a trailblazer both for developing the focus in shorter-term therapy and the value of psychotherapeutic consultancy to family physicians.

David Malan, a colleague of Balint, developed an influential approach to time-limited therapy he first termed ‘radical’ and later ‘intensive’ (Malan, 1963, 1976, 1979). The implication was that for some carefully assessed and selected patients, the time limit of a shorter therapy could accelerate the process of resolution of the central problem, or at least an important aspect of psychopathology. Unlike most, this approach favors a time limit (i.e., an agreed end date) rather than a predetermined number of sessions, to avoid the common difficulty of deciding when or whether sessions missed for any reason will count towards the total. However, an upper limit of sessions was set at 30, although most people were seen in fewer. Malan placed great store by a careful psychodynamic assessment of the patient's family and medical history, past and current relationships, to understand how events precipitating the current difficulty had emotional significance in the light of early experience. The therapist also attends carefully to the quality of the interaction. The assessment allows the therapist to judge whether to attempt a trial interpretation and the patient's response to this is an important factor in deciding whether this form of brief dynamic therapy is likely to be of benefit. The method itself is psychoanalytic, interpreting the transference, linking experience in the therapy relationship with childhood.

Malan described this in terms of two triangles—the ‘triangle of conflict’ (impulse–anxiety–defense) and the ‘triangle of persons’ (current relationship–therapist–parent). The ‘two triangles’ formulation is an economical and clear way for therapists to think about the focal conflict. Holmes (2000) gives the example of someone suffering from agoraphobia defending against anxiety by avoidance and dependency. Underlying this there may be hidden feelings of dissatisfaction and aggression, immediately towards a spouse, and in the past towards a controlling but unaffectionate mother. The therapist makes links between the anxiety, the defense, and the hidden impulse and between past relationships (usually with a parent), current relationships with others and the therapeutic relationship. In such a way, the patient is helped to tolerate anxiety and express hidden feelings, so that the ‘triangle of conflict’ is no longer enacted in current relationships.

The concept of a therapeutic plan was relatively new when Malan undertook his pioneering work in the 1960s and 1970s. Malan also had a profound commitment to research, at a time when respect for empirical evidence, particularly derived from quantitative methods, was unusual in psychoanalytic circles. From the 1980s, Malan espoused the methods of Davanloo, seeing in them a radical fulfillment of his own work.

Davanloo's (1978, 1990) method relies at heart on an orthodox psychodynamic drive/conflict model, derived from early Freud. He attracted controversy because his method involves pressurizing the patient in a relentless pursuit of any prevarication, vagueness, avoidance, or withdrawal, all seen as signs that important anxieties are being warded off. Repeated confrontation elicits anger, which is interpreted in terms of the ‘triangle of persons’ (i.e., a transference interpretation). This can lead to the powerful reexperiencing of warded-off anger from the past. Gustafson (1986) remarks on Davanloo's invariable focus on the patient's passivity as a way to deal with anger, noting that ‘all interviews of Davanloo discover this passivity’ (p. 175). Groves describes his method as Davanloo forcing the frigid patient to feel and thus creating mastery experience (Groves, 1996, p. 7).

At the same time as Malan was developing a coherent brief dynamic therapy in London, Peter Sifneos was working independently in Boston to develop short-term anxiety arousing therapy (STAPP) (Sifneos, 1972). He contrasted this approach with anxiety-suppressive therapy, which he advocated for severely disturbed patients, for crisis support (up to 2 months), brief therapy (from 2 months to 1 year) or in long-term supportive therapy (Sifneos, 1971). The anxiety arousing therapy could either be offered as a crisis resolution or as a time-limited therapy from 12 to 20 sessions (although sometimes longer, as there is a negotiated ending rather than a fixed time limit). The selection criteria were stringent, to exclude anyone with poor motivation, severe or complex difficulties in relationships, unrealistic expectations of treatment, and diffuse disturbances of identity. (This would exclude many of the patients seen in public sector settings or community clinics.) The therapist worked to establish an early alliance, so that the patient views the therapist as an ally and trusted teacher. The focus is on a circumscribed area of unresolved emotional conflicts, typically in terms of links between early and current Oedipal triangle themes, with repeated clarification using anxiety-provoking questions and confrontation. ‘Characterological’ problems such as excessive dependency or obsessionality were ‘bypassed’; that is, the therapist did not allow them to shift the focus. Sifneos warns against the risks of a countertransference problem where the therapist could unconsciously use the method to ‘punish the patient, see the patient suffer, or enjoy a position of superiority’ (Sifneos, 1972, p. 114).

Psychodynamic brief therapies, particularly those of Sifneos and Davanloo, could also be seen as a form of behavioral intervention, where controlled exposure to feared emotional states reduces the anxiety associated with them, a concept elaborated by McCulloch et al. (2003) in terms of ‘affect phobia’.

A structured approach to psychodynamic therapy developed by Luborsky (1984) has been applied in a brief format (Luborsky et al., 1995; Book, 1998). The brief form of ‘supportive-expressive’ therapy evolved in parallel with research on transference using the core conflictual role theme method (Luborsky and Crits-Cristoph, 1998). This method links repetitive relationship patterns in the patient's past and present, and therapy relationships in terms of central themes. Transcripts of patients’ narrative accounts of ‘relationship episodes’ are coded in terms of the psychodynamic triad of Wishes, Responses from Other, Responses of Self; for example ‘I wish to be loved and understood, others tend to dislike and reject me, I respond by feeling anxious and unloved’. In addition to its research uses, the core conflictual role theme method has been developed to guide therapists in their formulation, maintenance of the focus, and choice of interventions for a structured, manualized form of brief psychodynamic psychotherapy over 16 sessions (Book, 1998).

James Mann (1978) was working within a psychoanalytic tradition, but has profoundly influenced the field of brief psychodynamic and relational therapy with his existential method of time-limited psychotherapy. He argues coherently that time is insolubly linked to reality and there is a ubiquitous human yearning to deny time, reality, and death by regaining a lost childhood paradise of timelessness. This is achieved in adulthood by dreams, daydreams, falling in love, drinking, or using drugs, or in mystic states of ecstasy. He describes how brief therapies evoke the horror of the finiteness of time and posits that as soon as the patient learns that the amount of time for help is limited, he or she is subject to magical, timeless, omnipotent fantasies. Dismissive of eclecticism, Mann advocates one or two intake interviews to establish a formulation of the central conflict, linking current suffering to past sources, tracing the ‘chronically endured pain’. The focus for therapy is on improving the patient's self-image, but the formulation will differ according to the underlying difficulties. This formulation is given to the patient with a goal for therapy and an explicit offer of 12 sessions—no more, no less. The frequency and length of sessions within that limit seems to have been quite flexible, however. The calendar is consulted and the time for each appointment given, plus the exact date of the last (12th) meeting. He argues for as little ambiguity or evasion as possible about the time limit, and describes a typical course of therapy of early relief and improvement, a middle phase where enthusiasm wanes and ambivalence is felt (in a reenactment of earlier relationship patterns). As the patient moves towards ending, anxiety is evoked of ‘separation without resolution from the meaningful, ambivalently experienced person’. In the end phase, affects of sadness, grief, anger, and guilt are intensely experienced and relived in the disappointing ending of therapy. The therapist too feels the pressure to prevaricate and imply that the end is not the end, in order to evade the anxiety of separation without resolution. Mann emphasizes that active management of the termination will allow the patient to internalize the therapist and this time the internalization will be more positive, less anger-laden, less guilt-laden, ‘thereby making separation a genuine maturational event’. Any anger is acknowledged as normal and explored more rather than less.

Following the development of brief psychodynamic therapies, brief therapies based on behavioral, CBT, and cognitive theories began to appear. These arose from a research-based tradition and over the last 30 years have burgeoned, applied to an every-wider range of difficulties in mental health care, physical health problems, and health promotion. Many authors (see for example, Lovell and Richards, 2000) aggregate all these approaches into a common term—CBT—and in routine practice many therapists are rather eclectic in their choice of method within this broad framework. However, there are important differences between forms of CBT that integrated cognitive concepts into behavior therapy (Breger and McCaugh, 1965; Bandura, 1969; Meichenbaum, 1977) and those springing from the work of Beck, a different tradition of cognitive therapy that was not based on behavior therapy (Beck, 1979; Salkovskis, 1996). Goldfried (2003) argues that the lack of a clear distinction between cognitive therapy and CBT has arisen since CT was erroneously labeled CBT in the NIMH Depression Trial (Elkin, 1994).

Both methods were designedly brief, focusing in the first instance on depression, anxiety disorders, and obsessive-compulsive disorders—all without comorbid personality disorders. Since then the range of mental health problems addressed has grown to include PTSDs, eating disorders, and somatic problems (Salkovskis, 1996; Clark and Fairburn, 1997). Some of the newer applications are not brief, for example, CBT for personality disorders and psychosis (Perris, 1989; Linehan, 1993; Perris and McGorry, 1999).

CBT emphasizes a functional analysis of the problematic behavior or unwanted emotion in terms of antecedents, cognitions, behaviors, and consequences. This formulation then guides the choice of active techniques such as psychoeducation, relaxation, imaginal or in vivo exposure, response prevention, cognitive restructuring, and behavioral activation. Cognitive-behavioral therapists tend to emphasize the therapist's role in facilitating new experience and behavior as well as cognitive changes, maintaining clients’ awareness of their success experiences and the differences between their present and past functioning (Goldfried and Robins, 1983).

Cognitive therapy based on Beck's cognitive model of emotion (Beck, 1967; Beck et al., 1979) emphasizes that there are always alternative ways of perceiving and appraising any situation. People with mental health problems are trapped in a specific and unhelpful way of perceiving events, because of particular assumptions or beliefs they learned earlier in life. The therapist works collaboratively and empirically, inviting the client to explore whether or not there are alternative ways of appraising their situation, and empowering them to have choices over their response. The fundamental concept is of guided discovery of these alternatives, and support in testing out the consequences of new ways of thinking. Cognitive therapists tend to focus less than cognitive-behavior therapists on the role of behavioral antecedents and consequences including the impact of the patient's behavior on other people (Castonguay et al., 1995).

There is sparse discussion of treatment length in cognitive and cognitive-behavioral literature. Therapy length tends to be fixed (either for research purposes or by the constraints of the service setting) or pragmatically negotiated with the client in routine practice. Typically therapies last between 8 and 20 sessions, although the use of follow-up and ‘booster’ sessions is common, for example in relapse prevention in depression, and in clinical practice many CB therapists not wishing to terminate therapy abruptly will gradually reduce the frequency of sessions and intensity of treatment. For this reason, some CBTs are in practice long term.

There has been a tendency for CBT interventions to become increasingly complex, although whether or in what circumstances ‘multistrand’ interventions are more effective than simple ones has not yet been established (Chambless and Gillis, 1998; Tarrier et al., 1999) and some authors express skepticism (Lovell and Richards, 2000). This is probably a specific case of the more general finding of outcome equivalence in direct comparisons of different therapies (Stiles et al., 1986; Lambert and Ogles, 2003).

On the other hand, there has also been a significant drive towards distilling the ‘essence’ of an effective intervention and finding more efficient ways of delivering it in briefer therapies. Often this is done with the support of written materials for clients to read between sessions. For example, Clark et al. (1999) describe a seven session cognitive-behavioral treatment for panic disorder, Wells and Papageorgiou (2001) outline a brief cognitive therapy for social phobia where patients received a mean of 5.5 sessions.

Interpersonal therapy (IPT) (Klerman et al., 1984) was developed by psychiatrists as an adjunct to medication in the treatment of depression, and was brought to international attention through the NIMH collaborative research program. It was based on the interpersonal psychiatry of Harry Stack Sullivan and others, and research findings showing the intense impact of the formation, disruption, and renewal of attachment bonds, and the link between neurosis and deficits in social bonds. Theoretically grounded in social risk factors for depression as an illness, practically the method avoids an intrapsychic emphasis, whether psychodynamic or object relations, and has been shown to have much in common with CBT in using active techniques to ameliorate present difficulties (Ablon and Jones, 2002).

IPT explores which of four problem areas are salient for a given patient—grief, role disputes, role transitions, or interpersonal deficits. In the early phase, assessment and negotiation of the treatment contract includes review of symptoms, confirmation of the diagnosis and legitimization of the sick role, assessment of interpersonal relationships, and choice of problem area, and medication plan. Within a medical model of depression, there is a psychoeducational emphasis in promoting understanding of the effects of depressive illness, hence reducing self-blame. Therapy continues using specific techniques depending on which of the four foci are agreed. For example, the therapist could aim to facilitate mourning, to identify issues in disputes and alternative actions, could encourage the patient to view role transitions in a positive way, or could work on remediating interpersonal deficits. Therapy is time limited but not constrained to a fixed number of sessions. Typically it lasts between 9 and 12 months.

Problem-solving therapy (PST) is a brief psychological treatment for depression based on cognitive-behavioral principles (D'Zurilla and Goldfried, 1971; Nezu et al., 1989). It has also been used extensively as a form of crisis intervention following deliberate self-harm or attempted suicide (Hawton and Kirk, 1989). Like CBT it is structured, collaborative and focuses on generating solutions to current problems. Problem solving is seen as having five stages: adopting a problem-solving orientation; defining the problem and selecting goals; generating alternative solutions; choosing the best solution; and implementing the best solution and evaluating its effects. Methods used include cognitive modeling, prompting, self-instructions, and reinforcement.

It is usually delivered in about six treatment sessions. PST has been used to train nonspecialist health workers as part of primary care provision in a stepped care model. Meta-analytic review of randomized trials was unable to establish its effectiveness at reducing the repetition of deliberate self-harm (Hawton et al., 1998), although in this population a further meta-analysis of six RCTs in terms depression, hopeless, and improvement in problems, found it effective (Townsend et al., 2001).

Computerized CBT and guided self-help have also been developed as brief therapy approaches to anxiety and depression, particularly to reduce the time spent in therapist contact, so that CBT can become more accessible to the large numbers of individuals who may benefit from it. The principles of ‘stepped care’ (Katon et al., 1999; Haaga, 2000), suggest that briefer, simpler, and most accessible therapies should first be offered, and more complex, expensive, and effortful therapies only if the patient has not responded to the simpler approach. A research review of self-help interventions in mental health reported that almost all are based on CBT principles, and that computers may best be seen as another way of providing access to self-help materials (Lewis et al., 2003). A systematic review of 16 studies of computerized CBT, of which 11 were randomized controlled trials, suggested that for mild to moderate anxiety and depression, CCBT may be as effective as therapist-led CBT and better than standard care, although the evidence was by no means conclusive (Kalenthaler et al., 2003).

A third broad grouping of focal brief therapies can be termed ‘relational’ in that they see mental health difficulties as fundamentally interpersonal and they explicitly link the interpersonal to the intrapsychic in a ‘two-person’ psychology. Although these approaches have been influenced to a greater or lesser extent by psychoanalytic theory, they all emphasize relational rather than drive or structural aspects (Greenberg and Mitchell, 1983). Some have been influenced by cognitive psychology. These therapies pay close attention to the unfolding process within the psychotherapeutic relationship as a metaphor for, or an enactment of, the patient's problematic and repetitive interpersonal and intrapsychic patterns. They tend to use collaborative methods to guide discovery of these links and are wary of any notion that the therapist can stand aside from ‘the transference’ in order to interpret it authoritatively.

Time-limited dynamic psychotherapy (TLDP) (Schact et al., 1984; Binder and Strupp, 1991) is a collaborative method that avoids the therapist imposing the focus by ‘overtly pushing, manipulating, seducing, coercing, badgering, controlling, extorting or indoctrinating’ the patient. The aim is to develop a ‘working model’ (Peterfreund, 1983) of interpersonal roles into which patients unconsciously cast themselves, the complementary roles into which they cast others, and the maladaptive interaction sequences, self-defeating expectations, and negative self-appraisals that result. The TDLP focus is a structure for interpersonal narratives, describing human actions, embedded in a context of interpersonal transactions, organized in a cyclical maladaptive pattern, that have been both a current and recurrent source of problems in living. The time limit is not rigid, depending on the clarity with which a treatment focus can be established, but a ‘time-limited attitude’ is maintained (Binder et al., 1987).

Psychodynamic-interpersonal therapy (PIT) (Hobson, 1985) uses the ‘here-and-now’ relationship as a vehicle for learning about oneself in relation to others. Hobson has a process focus on the therapist and patient collaboratively developing a shared language for feelings. The therapist does not ‘interpret’ transference, but offers tentative exploratory links, making use of metaphor, and seeking to offer his or her own understanding of the patient's unarticulated emotions in the context of an authentic human relationship. Also known as the ‘conversational model’ of therapy, because of its emphasis on the therapeutic dialogue, a training manual and other materials have been systematically developed and evaluated in the UK. It has been extensively researched in relation to depression (in both eight-session and 16-session formats) (Shapiro et al., 1984; Shapiro and Firth, 1987), psychosomatic difficulties (Guthrie et al., 1991; Hamilton et al., 2000), with treatment-resistant problems in psychiatric outpatient settings (Guthrie et al., 1999) and as a brief intervention following self-poisoning (Guthrie et al., 2001). For example, Guthrie et al. (1999) identified 110 patients with a range of long-standing nonpsychotic disorders who had not responded to psychiatric interventions—an unusual sample both in terms of its mix of diagnoses and selection for their challenge to standard care. Patients were randomized to receive eight sessions of PIT or to continue their usual care from a psychiatrist. There was evidence of significant advantage to the active intervention on some measures, in terms of patients’ levels of functioning and in their use of health-service resources in the 6 months following treatment. Cost–benefit analysis suggested that this reduction in demand resulted in a cost-offset for psychotherapy provision.

Brief relational therapy (BRT) is a thoroughgoing relational approach developed in the USA by Jeremy Safran and Christopher Muran (2000), based on a ‘dialectical constructivist perspective’ (Hoffman, 1998). As with Hobson's method, there is an intense focus on the ‘here-and-now’ of the psychotherapeutic relationship, where the therapist urges collaborative exploration of both the patient's and the therapist's contributions to the interaction. The therapist is urged to be cautious about making interpretations based on generalized relationship patterns, but to explore the nuances of the patient's experience and the relational meaning of this experience, through unfolding therapeutic ‘enactments’. There is extensive use of metacommunication about the meaning of what is happening between the therapist and patient, with disclosure of the countertransference.

The therapist refrains from early case formulation or content focus for the sessions. Safran and Muran argue that as the therapist can never stand outside the interaction to create a formulation that is not shaped by unwitting enactment, such a therapist-derived focus is inimical to a fully relational method. Case formulation only arises from the therapist repeatedly ‘disembedding’ from whatever interpersonal pattern (‘matrix’) is being enacted. ‘Therapy thus consists of an ongoing cycle of enacting, disembedding and understanding—and this understanding is always partial at best’ (Safran and Muran, 2000, p. 178). They acknowledge that the lack of a ‘tangible’ focus can be a problem for brief therapy, linking to the therapist's anxieties about having something substantial to offer within a limited time. Instead they offer the process focus of mindfulness, modeling a capacity to observe one's internal processes and actions in relationship to other people, and thereby helping the patient develop and generalize this skill. As in cognitive analytic therapy (CAT) ruptures and repairs to the therapeutic alliance are seen as a particularly effective way to gain awareness of problematic relationship patterns. Links between the therapy relationship and relationship patterns outside therapy are made tentatively, the therapist making an effort to be aware of his or her own motivations.

Cognitive analytic therapy (CAT) is an integrative approach developed in the UK by Anthony Ryle (1990) and further extended both theoretically and clinically by others (Ryle and Kerr, 2002). Ryle aimed to integrate the effective elements of various preceding traditions—not simply at the level of therapeutic technique, but in the underpinning theory of development, personality, and psychopathology. CAT theory is rooted in Kelly's (1955) personal construct theory, cognitive and developmental psychology (stressing in particular the actively intersubjective nature of the human infant; Stern, 1984; Trevarthen and Aitken, 2001) and in psychoanalytic object relations theory. Theoretically it emphasizes repetitive aim-directed sequences of cognition, emotion, behavior and their consequences (called ‘procedures’), similar to Goldfried's (2003) ‘STAIRCASE’ (Situation, Thought, Affect, Intention, Response, Consequence, and Self Evaluation) CBT model. However, CAT theory also draws on object relations theory and Vygotsky's activity theory to assert the pervasively dialogic nature of the human world, where internalized self-other relationship patterns become the basis of reciprocal role procedures governing intrapersonal as well as interpersonal relationships. Procedures, including reciprocal role procedures, are problematic to the extent that aims are not achieved yet the maladaptive sequence is not revised. Over the past decade it has integrated Vygotskian activity theory and the Bakhtinian concept of the dialogic self (Leiman, 1992, 1997). The model has thus come to be underpinned by a radically social concept of self.

CAT, while theoretically and methodologically integrative, is therefore a fundamentally interpersonal and relational therapy. In common with BRT it requires the therapist to reflect collaboratively with the patient what reciprocal roles are being enacted in the therapy relationship, particularly at points where the therapeutic alliance is being threatened (Bennett and Parry, 2003). In contrast to BRT, however, the initial few sessions of CAT are devoted to an extended assessment leading to a jointly agreed reformulation of a patient's story, its personal meaning and the relation to it of the problem procedures they have brought with them. The narrative account is redrafted on the basis of the patient's feedback and is supplemented by a diagrammatic reformulation. Both forms of reformulation are seen from the Vygotskian perspective as psychological ‘tools’, fostering jointly focused attention and the capacity for self-reflection. The reformulation forms the basis of intervention, which often includes cognitive-behavioral methods of procedural revision.

As in psychodynamic brief therapies there is stress on the therapeutic value of the issues provoked by a fixed termination point. Ending is seen from a CAT perspective to minimize regression and avoid protracted, and usually collusive, dependency. It is also an opportunity to work through the unassimilated issues from earlier losses and to enact new reciprocal role procedures. An example might be the reciprocal role ‘appropriately withholding in relation to manageably deprived’ leading to the patient being able to feel vulnerable and to tolerate the painful feelings that the ending can provoke. The ending is formally and symbolically celebrated by the therapist writing a further letter—a ‘farewell’ letter—to the patient. This acknowledges the achievements of therapy but also anticipates loss and possible grief and anger. The patient is encouraged to write a farewell letter from his or her own perspective.

Although a 16-session format lasting approximately 4 months is used for ‘neurotic’ difficulties, CAT is one of the better-developed models for working briefly with more severely disturbed patients. Here longer contracts (usually 24 sessions) are offered and sometimes further interventions such as group work or consultation to the community mental health team (Kerr, 1999). The CAT model of borderline personality disorder (Ryle and Marlowe, 1995; Ryle, 1997) describes severe damage and disturbance of the self, characterized by a tendency (apparently secondary to chronic psychological trauma) to dissociate into different ‘self states’ (each characterized by one reciprocal role procedure). One consequence of this, apart from the tendency to enact extreme and disconnected roles, is a poor ability to reflect upon these states and an impaired capacity for empathy and executive function. The extreme role enactments in borderline personality disorder would include, for example, idealized help seeking, abusive, and vengeful anger (expressed to self or others) or dissociated, numb, ‘zombie’ states in which serious self-harm may be perpetrated. CAT thus offers a clear theoretical basis for engaging and working with the enactment of poorly integrated and maladaptive reciprocal role procedures, with similarities to that developed independently by Benjamin (2003), but applied within a brief therapy format.

A number of brief therapies draw pragmatically on a range of theories and methods to yield approaches that are eclectic.

Budman and Gurman (1988) describe a method they characterize as ‘interpersonal–developmental–existential’, which includes a range of issues in the formulation including losses, developmental dysynchronies, interpersonal conflicts, symptoms, and personality dysfunction. The focus is used to open the session, maintain a unity within the session and to close the session, drawing together the material linked to the focus in a brief summary statement. They argue for the flexible use of time on the basis of Johnson and Gelso's (1980) review of the effectiveness of time limits and Howard et al.'s (1986) dose–response findings. They suggest more intensive work for four to eight visits then ‘spreading out’ the sessions to reduce dependency and enhance self-efficacy. They criticize the traditional analytic view of ‘once and for all’ therapy and criteria for ‘completed’ therapy as unrealistic and rigid. Instead, brief therapy is available on an intermittent basis as required when facing different developmental challenges.

Cummings (1991) also argues for brief intermittent therapy throughout the life cycle as a pragmatic approach where theory and techniques from different models are synthesized. He argues that ‘termination’ is not necessarily difficult or painful and suggests it is therapists rather than patients who have difficulty ending. Instead the therapist makes a commitment to be available to the patient ‘as long as you need me—in return I want you to join me in a partnership that makes me obsolete as soon as possible’. This is a model of the therapist as a ‘psychological family doctor’ providing continuity of care over time. This way of working leads to brief treatment episodes of between one and 20 sessions, spaced flexibly.

Garfield (1989, 1995) describes an eclectic brief therapy model based on maximizing the impact of the ‘common factors’ identified in therapy research; therapists are engaged in listening, reflection, suggestion, explanation, interpretation, providing information, confrontation, reassurance, homework assignments, modeling and role play, questioning, and cautious self-disclosure. In common with Budman and Gurman and Cummings, he takes a relaxed approach to treatment length, and to selection criteria. Garfield also challenges the assumption that if people do not respond to short-term therapy, they will benefit from long-term work. He seems this as having little empirical justification, as there has been almost no research on long-term therapy.

Winston and Winston (2002) describe a pragmatic eclectic approach, which they term ‘integrated’, although it does not seem fully integrated at the theoretical level, compared with, for example, CAT. Their case formulation method uses the concept of a continuum between psychological sickness and health, according to ‘level of psychopathology, adaptive capacity, self-concept and ability to relate to others’ (p. 11). The individual treatment plan depends on the patient's position on this continuum, with cognitive-behavioral methods being used for the ‘more impaired’ and more psychodynamic, expressive techniques for the ‘least impaired’. By this means, a brief intervention can be offered for more severe and complex difficulties, such as borderline disorders (Winston et al., 1991).

Very brief therapies of up to five sessions have been developed in differing treatment modalities. Öst and colleagues have investigated the impact of single-session interventions, finding that one prolonged session of exposure has an equivalent impact to (an already brief) five sessions for a range of specific phobias: injection (Öst et al., 1992); blood injury (Hellstrom et al., 1996); flying (Öst et al., 1997); and claustrophobia (Öst et al., 2001). As other researcher groups have demonstrated the efficacy of single-session exposure (e.g., Thorpe and Salkovskis, 1997) it does seem that—at least for circumscribed behavioral goals in specific phobias—very brief interventions may be adequate.

A three-session therapy in a ‘two plus one’ format (Barkham, 1989) was developed in a research context for ‘subsyndromal depression’ on the basis of the research evidence of dose–response (Howard et al., 1986) and Johnson and Gelso's (1980) review of the effectiveness of time limits. Therapy comprises two sessions 1 week apart followed by a follow-up session 3 months later. Barkham et al. (1992) conducted a pilot study of this approach, with reasonably positive outcomes. In a later larger-scale randomized trial, Barkham et al. (1999) allocated 116 patients to CBT or psychodynamic versions of the 2 + 1 model, either immediately or after a 4-week delay. All patients fell below diagnostic thresholds for depression, but were entered into the trial in three bands of severity—stressed (effectively within normal population limits), subclinical (mildly symptomatic), or low-level depression. Clients at all levels made gains, with some evidence from the delayed-treatment condition that this related to the intervention rather than time-effects. Patients receiving the CBT version obtained greater benefit at 1-year follow-up (Barkham et al., 1999), suggesting that Hobson's PIT method may be less well suited to this very brief format. This model was not developed—or intended—as a therapy in its own right, but as a way of offering a more rapid response to clients, and as a way of testing the appropriateness of further therapy. To that degree these studies offer support for the utility of the model.

Sheard et al. (2000) describe a one- to three-session CAT-derived method to improve the response of psychiatrists to repeated deliberate self-harm in he context of emergency hospital care. Outcome studies are not yet available.

Newman et al. used a four-session CBT intervention for panic disorder, assisted by the use of palmtop computers for self-monitoring and assessment, with similar results to a 12-session treatment.

The clinical method of motivational interviewing (Miller and Rollnick, 1991) has been used as a very brief intervention either alone or in addition to standard treatment, particularly for alcohol and substance misuse problems. It was developed on the basis of a review of ‘active ingredients’ in effective brief therapy with these client groups, which suggested the importance of giving feedback, promoting personal responsibility for change and self-efficacy, giving straightforward advice, and offering a menu of alternative strategies. The method is nondirective and avoids any confrontation with ‘resistance’ or ‘lack of motivation’, instead taking an acceptant and empathic approach to changing motivational states. The aim is to help those reluctant to change problematic behaviors move from the pre-contemplation stage, or ambivalent contemplation (as understood by Prochaska and DiClemente's, 1984, transtheoretical model of change), to preparation where change options can be explored and then action and maintenance of change. A number of randomized trials have demonstrated its value in these settings, although others have shown it is not invariably effective (Dunn et al., 2001; Miller et al., 2003). A meta-analytic review of randomized trials (Burke et al., 2003) found adaptations of motivational interviewing a promising approach to problems involving alcohol, drugs, dieting, and exercise but that the evidence did not support their efficacy in smoking or HIV-risk behaviors.

Solution-focused brief therapy (SFBP: de Shazer, 1985; Walter and Peller, 1992) developed from brief strategic therapy (Weakland et al., 1974) in work with families and individuals, and is often delivered over four to five sessions. It pays no attention whatsoever to the origin or etiology of problems and instead focuses on helping clients to change problem-maintaining behavior, to define their goals (recognizing that their own definition may or may not be congruent with problems as perceived by professionals), and to generate solutions to difficulties they face. Questions about goals are posed in such a way that the client is able to speak about what the world would be like without their current problems. It is in this sense that the method is solution focused rather than problem solving. The focus is on collaborative identification and amplification of the patient's strengths, with extremely positive feedback and an emphasis on small aspects of meaningful change. A review by Gingerich and Eisengart (2000) identified 15 controlled studies of this approach, five of which met criteria for methodological adequacy, although only one of these (Sundstrom, 1993) was directly related to mental health issues. Sundstrom contrasted one 90-minute session of solution-focused or problem-focused therapy in 40 mildly depressed college students. At 1 week follow-up scores on the Beck Depression Inventory were reduced, and outcomes were equivalent in both therapy groups. However, the nature and size of the sample and the brevity of follow-up makes it difficult to draw conclusions about the efficacy of this approach. There is in any case a conflict between the philosophy of SFBP and research that estimates efficacy using standardized instruments, as this does not reflect user-defined outcomes (which will inevitably be quite varied). However, as with a number of emergent techniques, in the absence of well-conducted research, we are unable to comment on its effectiveness, or the range of difficulties for which it is appropriate.

There is a wide range of brief interventions aimed at responding to crises (Hobbs, 1984). The theory and practice of crisis intervention developed from the work of pioneers in the 1960s, such as the community psychiatrist Caplan (1961), and the psychoanalytic crisis therapist, Jacobson (1980). A fundamental concept is that during crisis, people are unusually receptive to restructuring their psychological processes, providing a window of opportunity for a brief intervention to have a substantial positive effect. Crisis intervention uses the intense affect associated with the crisis state in order to facilitate constructive change. The personal meaning of the crisis is explored, in terms of both present and past aspects (e.g., a loss event could re-evoke feelings associated with an earlier loss), coping resources, and components of crisis that render these ineffective. The crisis may be formulated in a way that gives individuals or family members a cognitive understanding of what has happened, so that the emotional assimilation of this is facilitated, and new coping resources are mobilized. Crisis intervention is a contractual benefit of most Health Maintenance Organization prepaid plans in the USA (Chiefetz and Salloway, 1985), although Adams (1991), writing of family crisis intervention in the San Diego Kaiser Permanente Medical Care Program, argues that virtually none provide a clear definition of what this means. In the UK, although crisis intervention is linked to home treatment of severe mental illness (Joy et al., 2004) as an alternative to hospital admission, crisis theory, and practice has not been generally characteristic of brief approaches to psychotherapy. Research on the effectiveness of crisis intervention is sparse, with the exceptions of critical incident debriefing after a traumatic event and problem solving following suicide attempts or deliberate self-harm.

Critical incident debriefing was designed as a rapid response to a traumatic event, aiming to reduce vulnerability to developing PTSD or other mental health conditions, and usually delivered in a single session. Although intuitively appealing to many clinicians, it is now clear that single-session debriefing immediately after exposure to a traumatic event is ineffective, and that on the contrary there may be an adverse impact for some individuals (e.g., van Emmerik et al., 2002; Rose et al., 2003). Overall it is clear that single-session interventions cannot be recommended as part of routine practice, and the English Department of Health guideline on treatment choice in psychological therapies (2001) explicitly argues against their use. There is very little evidence about the impact of slightly longer interventions, though Bryant et al. (1999) and André et al. (1997) employed five- and one- to six-session interventions, respectively, with more positive results. This does not imply that individuals in distress should not be offered support, nor does this general conclusion contraindicate more extended psychological intervention at some remove from the initial trauma, if posttraumatic disorder were to develop.

Interventions focused on the management of suicide attempts are reviewed by van der Sande et al. (1997), who identified 15 randomized trials of varying forms of intervention, only some of which would be classified as brief interventions. Two trials considered the impact of a 3-month crisis intervention based on a problem-solving model (Gibbons et al., 1978; Hawton et al., 1987). In total, 480 patients were randomized to receive the intervention or to act as controls; overall the interventions did not lead to a reduction in suicidal behavior. In contrast, four trials employed CBT to focus on broader aspects of current and past functioning and coping mechanisms (Liberman and Eckman, 1981; Salkovskis et al., 1990; Linehan et al., 1993; McLeavey et al., 1994). Although these trials demonstrated a significant reduction in suicide attempts, they were relatively small scale, and (in the context of this chapter), some intervention periods were far from brief (for example, treatment in Linehan et al. took place over 1 year). There is some suggestion from these trials that a focus on background rather than on current problems may be more beneficial for this group of patients.

Research into brief therapies falls into several areas. The largest source of information comes from ‘mainstream’ research into psychological therapy. In research contexts interventions are frequently delivered in the form of manualized packages of short duration—it is a matter of observation that most therapies in research trials last between 12 and 16 weeks. In these cases we are examining the impact of therapies, which their originators may or may not have intended to be brief, but which are nonetheless investigated in this form. While there may be arguments about the appropriateness of examining the impact of therapies not intended for implementation over short time-frames (e.g., some forms of psychodynamic therapy), in the present context it is relevant to ask how therapies perform under conditions where termination is clearly signaled from the outset of therapy. From this perspective (ironically) research trials become a better exemplar of brief therapy than in many clinical contexts, where duration is poorly controlled and much more likely to be dictated by patient attrition than therapist intent.

Evidence for the differential efficacy of different forms of brief therapies remains disappointingly weak. This is usually referred to as the ‘dodo bird’ conclusion (from Alice in Wonderland: ‘Everybody has won, and all must have prizes’). Where differences are found, they tend to be small. In 1975, Luborsky et al. drew attention to the apparent equivalence of outcomes across different types of psychotherapy and 27 years later, Luborsky et al. (2002) reexamined the issue in 17 meta-analyses of comparisons of active treatments with each other, drawing a similar conclusion. Lambert and Ogles (2003) reviewed results from comparative, dismantling, and components analysis studies as suggesting the general equivalence of treatments based on different theories and techniques. They believe that these findings argue against the current trend of identifying empirically supported therapies that purport to be uniquely effective and conclude in relation to the differential efficacy question, ‘Decades of research have not resulted in support for one superior treatment or set of techniques for specific disorders’.

The verdict of the Dodo bird is controversial, and many do not accept it. Norcross (1995) lists many reasons why he finds the verdict untenable—only a handful of the many therapies have been evaluated; similar symptom reduction does not mean identical outcomes; common factors do not preclude specific effects; studies disregard the person of the psychotherapist (different therapists obtain different improvement rates); almost one-half of studies are underpowered; studies ignore the quality of the therapeutic relationship, ‘horse-race’ outcome studies are insensitive to differential treatment effects (similar group means mask individual outcome differences and mask interaction effects); and most studies examine psychologically irrelevant treatment variables (e.g., in an ex post facto search for correlates between client characteristics and outcomes). He asserts that studies with an aptitude-treatment interaction design and sufficient power on psychologically relevant variables do show differential effects. Beutler (1991, 1995, 2002) also argues that the complexity of determining specific effects has been underestimated and that those who believe in outcome equivalence have largely ignored evidence of specific effects. He contends that the gross labels used to identify manual-driven therapies do not eliminate within-therapy variations, or preserve between-therapy variations. A good example of this is discussed by Ablon and Jones (2002) where the NIMH Treatment of Depression Collaborative Research Program compared two apparently different therapies that actually had many similarities.

In addition, diagnostic homogeneity does not remove important patient differences. As a result, it is vital to consider specific types of interventions (irrespective of therapy type) and specific, nondiagnostic aspects of patients. Examples include therapist directiveness interacting with patient receptivity or resistance, and a focus on symptoms versus a focus on insight differentially benefiting those with internal versus externalizing coping styles (Beutler, 1991). Rounsaville and Carroll (2002) argue that the Dodo bird verdict is based on insufficient attention to patient–treatment matching and the many design constraints on efficacy research that reduce the likelihood of detecting large outcome differences between active treatments.

For these methodological reasons, the degree to which outcome equivalence speaks to common therapeutic processes underpinning effective interventions is not yet clear. Certainly however, champions of specific therapies should be appropriately cautious when advocating one approach over another, especially because contrast of therapies in research trials should ensure that they are of equal credibility and presumed potency in relation to the condition for which they are intended. For example, while one review (Gloaguen et al., 1998) showed apparent superiority of CBT for depression, subsequent reanalysis (Wampold et al., 2002) demonstrated that this effect only stood when ‘non bone-fide’ therapies were included; once these had been removed from the analysis CBT was equivalent in efficacy to other approaches.

There is a relative hegemony of CBT trials, and while this does not diminish the evidence for the efficacy of these approaches, it limits comment on the relative efficacy of other approaches, as direct comparisons in adequately powered trials are rare. Despite this (and the discussion above), in some areas there is evidence for differential efficacy. Many anxiety disorders are effectively treated by relatively brief behavioral and cognitive-behavioral interventions, and in the absence of strong clinical indicators it would be difficult to justify the first-line use of alternative approaches to these conditions (Department of Health, 2001). The strength of this statement is justified by Schulte et al.'s (1992) trial, which examined treatment outcomes for specific phobias. In effect they contrasted standardized in vivo exposure against an individualized treatment where therapists were free to implement any therapeutic approach. The greatest benefit was found with in vivo exposure, a fact reinforced by the finding that the patients who benefited from an individualized approach were those who had been given in vivo exposure. This result is salutary—specific phobia is a condition with a straightforward treatment approach of known efficacy, and yet at least some clinicians elected to employ alternative and less effective techniques. This raises questions about the relative role of clinical judgment and clinical guidelines (Wilson, 1996; Parry et al., 2003a).

As noted above, most of the research literature on psychological therapies relates to the impact of therapies delivered in a brief form. Most research is now conducted using a manualized form of therapy, sometimes specifying the sequence of therapy, and sometimes acting as a broader guideline to indicate which types of therapeutic intervention are permissible. On this basis therapies conducted under research conditions represent as tight a test of brief interventions as we are likely to see.

A full review of this literature is beyond the scope of this section (see Roth and Fonagy, 1996, 2005; Lambert and Ogles, 2003), but it is worth noting that overall there is good evidence for efficacy of these interventions across a wide range of mental health conditions. Given the restriction on length of research-based therapies noted above, it is reasonable to conclude that clinically significant change is possible for many conditions after about 16 sessions, and people with more circumscribed problems or milder presentations may benefit from briefer interventions. This does not mean that after 16 sessions all clients can be expected to have achieved stable remission—clinicians need to have a realistic idea of what can be achieved in a relatively brief period.

As an example, outcomes from some major studies of interventions for depression (e.g., Frank et al., 1991; Hollon et al., 1992; Elkin, 1994; Shapiro et al., 1994, 1995) act as a helpful benchmark. At posttherapy about 50% of patients had made significant gain, though over the next year about half of those who improved relapsed. This means that at the end of follow-up, only about one-fourth remained well. It could be argued that this apparently poor result reflects the nature of the disorder—depression is potentially a chronic condition marked by a history or relapse and remission, and expected outcomes need to be set against this context. Recognizing this fact, some researchers have investigated the efficacy of a model of maintenance, showing that additional monthly sessions following short-term intervention significantly reduce relapse rates (e.g., Frank et al., 1991).

It is important to recognize that relapse after brief therapy is not uncommon, though there is variation in the extent to which this occurs, in part related to the condition under examination. For example, relapse appears to be less marked in anxiety disorders (such as phobias, panic, PTSD, and obsessive-compulsive disorder) than in mood disorders, and it also clear that some conditions (for example, eating disorders or generalized anxiety disorder) present particular challenge to clinicians. The degree to which a therapy achieves remission is important, as there is evidence that relapse rates are higher among patients who show residual symptoms at termination (e.g., Jarrett et al., 2001). On this basis, there is sound evidence that clinicians need to consider strategies for managing patients who remain vulnerable at the conclusion of brief episodes of therapy, the most obvious being the planned extension of therapy, or the offer of maintenance sessions.

A second area of research focuses on the incremental impact of adding more sessions in a therapy with no fixed time limit—commonly referred to as ‘dose–response’ research. Here, therapies as delivered may or may not be brief, but the aim is to determine how brief or how extensive a therapy needs to be to achieve impact. The data analyzed in dose–response research are invariably naturalistic rather than experimental, that is, therapies of different lengths are compared but length of treatment is not manipulated experimentally in a randomized design.

The relationship between the amount of therapy received and subsequent benefit is one that has received considerable attention. As a research question this is entirely appropriate, though the analogy with drug treatment implied by ‘dose–response’ is unfortunate. Howard et al. (1986) collated data from 15 outcome studies of 2431 patients who had completed therapy. The sample of therapies included in this analysis may not be representative—it included longer-than-typical therapies and did not include cognitive and behavioral therapies. Improvement rates can be examined in therapies of differing lengths. Graphical representation shows this to be a negatively accelerating curve—where the percentage of clients who have shown measurable improvement increases with number of sessions but fewer and fewer reach this criterion as therapy length increases. About half of the patients had improved after eight-session therapy, whereas it took 104 sessions before 90% of patients had improved. People with more severe difficulties (such as personality disorder) respond more slowly to therapeutic intervention (Howard et al., 1993). Replications and further studies have since been undertaken by Kopta et al. (1996), Anderson and Lambert (2001), Lambert, Hansen, and Finch (2001), who on the whole give more conservative estimates of the dose of therapy required for clinically significant improvement.

There is also evidence that the shape of the dose–response curve reflects the fact that different aspects of functioning change at different rates. Barkham et al. (1996) and Kopta et al. (1994) followed the pattern of session-by-session changes in questionnaire item-endorsement; both determined that symptoms changed most quickly, while characterological change was slowest. Though this finding is mathematically isomorphous with Howard et al., (1986) it implies that questions about dose–response are best asked in relation to the type of outcome that is intended. Though it seems obvious to state that more seriously disabled individuals will require more time to recover functioning, the implication is that treatment lengths need closely to reflect treatment aims.

In summarizing dose–response evidence, Lambert and Ogles (2003) state that ‘a sizeable portion of patients reliably improve after 10 sessions and […] 75% of patients will meet more rigorous criteria for success after about 50 sessions of treatment. Limiting treatment sessions to less than 20 will mean that about 50% of patients will not achieve a substantial benefit from therapy (as measured by standard self-report scales)’ (p. 156). Hence brief therapy cannot be recommended for all patients, but more than half can be helped substantially in this way. This figure is likely to improve as brief therapy methods are refined and developed.

Another approach is to derive dose–effect relationships for individual patients by analyzing session-by-session change from many thousands of patients, using a standardized metric to follow patterns of change. Two groups of workers have used fairly complex (e.g., Leuger et al., 2001) or relatively simple (e.g., Lambert et al., 2001) assessment systems to produce expected profiles of change for patients based (broadly) on their initial level of psychological distress. The benefit of this approach to the individual clinician is severalfold. It makes it clear that the extent and rate of change for more severely disabled clients will be slower than for the less distressed, and—crucially—quantifies this difference. It also enables clinicians to identify those clients who have already made an appropriate amount of change, with the possibility of more rapid discharge. Further, it signals patients who are deteriorating or whose rate of change is significantly slower than expected.

Lambert et al. (2002) randomized therapists either to receive or not to receive feedback on client progress based on individualized dose–response curves. Clients whose therapists received feedback had somewhat greater rates of improvement, but the more striking result was a significant reduction in the rate of deterioration. In addition, there was evidence that feedback enabled therapists more accurately to tailor the number of sessions to clinical need—those who had improved were more likely to be discharged, and those who required more help were more likely to be retained in therapy.

Consensus about the length of therapy is based on clinical judgment rather than empirical data, though the question of how much therapy is needed to make for clinically significant and lasting change is obviously an important one. A small number of trials address this issue directly, by comparing short and longer forms of the same therapy, delivered as part of the same comparative trial by the same group of clinicians.

Clark et al. (1999) contrasted the impact of 14 sessions of panic control therapy (PCT) against seven sessions, in 43 patients with panic disorder and little or no agoraphobic avoidance. The briefer intervention included self-study modules, which the patient read prior to sessions, and introduced many of the ideas used in sessions. At posttherapy and at 3- and 12-month follow-up, both forms of treatment were of equal efficacy. This study was conducted by a group of senior researchers experienced in using PCT, and it is notable that other researchers obtained poor results using such abbreviated forms of therapy. Black et al. (1993) developed an eight-session form of CBT for panic; though not contrasted against a longer form of the therapy, it was no more effective than placebo medication.

Shapiro et al. (1994) investigated two forms of therapy—CBT and PIT—delivered for either 8 or 16 weeks. The trial design ensured that the 120 patients represented varying levels depression (low, moderate, and high). At the end of the initial phase of treatment there were no differences in efficacy between treatments for those patients with low or moderate levels of depression, but more severely depressed patients appeared to do better with a longer duration of treatment. The pattern of gains at 12-month follow-up (Shapiro et al., 1995) suggested that eight sessions of exploratory therapy appeared to be too few, there was some evidence favoring 16 sessions of CBT, and overall, poorer maintenance of gains was evident in patients with greater levels of initial distress.

Hoglend and Piper (1997) review two independent studies of brief dynamic psychotherapy carried out in Canada, and Oslo, one of which used a fixed time-limit, while the other employed open-ended treatment. Although broad outcomes were similar, post hoc analysis using a measure of the maturity of patient's relationships to others (quality of object relations) suggested that patients high in this quality did equally well with therapy of either duration. However, though low-quality of object relations patients did better with fewer sessions when therapy was time-limited, they did better with more than 35–40 sessions when in open-ended therapy—a result that is hard to interpret, and suggests the potential complexity of dose–response relationships.

Although the foregoing suggests that there is good evidence for the efficacy of brief therapies, it is reasonable to ask whether their effects are lasting, or indeed are as robust over follow-up as their longer-term counterparts. The most rigorous method for investigating this would be a within-study contrast of shorter and longer versions of the same approach. Here there are too few available studies to draw conclusions, and those reviewed above suggest that there is mixed evidence on this point—for example, and perhaps unsurprisingly, outcomes for depression are more complex than for anxiety. More generally, conclusions about the longer-term outcome from briefer therapies is hampered by a lack of follow-up data, despite a general agreement that posttherapy outcomes may not be the most helpful indicator of therapeutic impact. As an example, a recent meta-analytic review for treatments of depression, panic, and generalized anxiety disorder—all common presentations in clinical practice—noted that comment on outcomes beyond 12 months was severely restricted by a lack of relevant studies and variation in reporting of these outcomes (Westen and Morrison, 2001).

Where findings on follow-up are available, there is reasonable evidence for the stability of gains in relation to some conditions such as panic disorder (e.g., Milrod and Busch, 1996) though it is harder to demonstrate enduring change in conditions characterized by relapse–remission cycles, such as depression, where a common finding at 12-month follow-up is that only around a fourth of those entering a trial of brief therapy remain well (Roth and Fonagy, 2005). This raises an important issue for interpreting follow-up, as without knowing something of the likely trajectory over that time period, expectations of continuing efficacy could be unrealistic.

A rather different issue is raised by outcome studies conducted in the context of primary care, where very brief therapies are offered to individuals whose presentations are less severe, chronic, and complex than might be common in other settings. A common contrast here is to ‘treatment-as-usual’ (TAU), and a common finding is that though active treatment shows benefit at posttherapy, over a relatively short time those receiving TAU show a similar level of gain. Two examples illustrate this phenomenon. Ward et al. (2000) contrasted treatment as usual from a GP to 12 sessions of nondirective counseling or CBT in 463 patients with mild to moderately depression. Though at 4 months both active treatments showed equivalent and significant advantage to TAU, at 1 year there were no differences in outcome. Parry et al. (2003b) randomized 94 patients with asthma-related anxiety to brief CBT or standard care. At the end of treatment, the CBT group were significantly improved compared with controls on panic fear, depression, locus of control and asthma-related quality of life. Six months later, these differences were no longer significant because the control group had also improved. Results such as these require careful interpretation, as the benefits to the individual of an accelerated rate of change may be highly significant, despite an apparent equity of outcome over time.