Abstract
Worldwide mental health services spend millions of dollars each year on training their workforce. Training programs often pay lip service to evidence, but as a recent UK review [1] showed, training for mental health professionals often lacks focus on core competencies and in the case of severe mental disorder pays little regard to the research evidence. In this paper, it is argued that training programs should be guided by evidence and should generate evidence.
In order to decide on the priorities for training, one needs to consider not only the evidence of clinical interventions, but also the priority attached to those interventions for mental health services. Therefore, training needs to target those areas most important and students need to be taught skills in delivering the effective intervention. Hence, although there is a large amount of evidence [2] that specific phobias can be treated very effectively, this topic is not a training priority as specific phobias, although very common in the population, are seldom cause for severe levels of handicap and distress.
What then are the priority areas for mental health services? Clearly, the finite numbers of mental health professionals and the huge array of mental health problems affecting very large proportions of the population mean that mental health professionals will only be able to treat a very small number of people who need and demand intervention. As stated below, there are compelling arguments to suggest that many interventions can be delivered by professionals without a mental health background, by non-professionals and by other methods such as self-help groups and computers. Thus, perhaps we need to consider the advice of Michael Shepherd and his colleagues [3] from more than 30 years ago who stated:
Administrative and medical logic alike…suggest that the cardinal requirement for the improvement of the mental health services is not a large expansion of psychiatric agencies, but rather a strengthening of the family doctor in his therapeutic role.
However, even if we succeed in our future efforts to equip primary care staff and others with mental health skills, we will still be faced with a problem of being unable to offer treatment to everyone in need. Goldberg and Gournay [4] have recently proposed a decision matrix about who to target based on four factors: diagnosis, level of handicap, likelihood of spontaneous remission, and evidence of effective treatments (drug and non-drug).
With regard to levels of evidence, Lewis et al. [5], in a discussion of evidence-based approaches, highlighted the relative lack of evidence in mental health care, compared with, for example, obstetrics and gynaecology where the systematic reviewing process (of the evidence) is virtually complete. Indeed, Lewis et al. [5] concluded that the task for completing comprehensive reviews of the evidence in mental health care was, in their words, ‘truly Herculean’.
There are, of course, models for classifying levels of evidence. For example, the levels could be:
A number of randomised control trials testifying to efficacy of an intervention which have been subject to systematic review. One or two randomised, control trials testifying to efficacy, but not enough to meet the minimum requirements for conducting a systematic review. No randomised, control trials, but trials that have used reasonable controls (e.g. quasi-experimental designs showing effectiveness). A number of single-case, experimental designs indicating effectiveness.
What then do we know about training outcomes? The answer, in brief, is very little. The best example of a well-evaluated program of training is that of Isaac Marks [6], who developed a training program for nurses in behaviour therapy in 1972 at the Institute of Psychiatry in London. The program has continued at the Maudsley and several other sites in the United Kingdom, and the Republic of Ireland over the last 25 years. Nurses enrolled in the program are given intensive full-time training (18 months) in assessment and intervention skills for the behavioural and cognitive-behavioural treatment of phobic, obsessional and sexual problems. Graduates routinely collect outcome data on patients and this has allowed Duggan et al. [7] to describe the clinical outcomes of more than 2000 patients seen by nurse therapists over a 15-year period. The audit showed sustained improvement in patient symptoms and social functioning. In addition to these very encouraging data, Marks [6] has conducted a randomised, controlled trial using nurse therapists to treat phobic and obsessional problems in primary care. The study not only provided convincing evidence that nurse therapists are very clinically effective with their patients (compared with routine general practitioner care), but it also included an economic analysis which demonstrated clear cost benefit to the patient and the healthcare system.
This research is in marked contrast to the results obtained by Gournay and Brooking [8] in a randomised, controlled trial and economic analysis of community psychiatric nurses using counselling skills with a primary care population suffering from depression, anxiety and adjustment disorders. The outcomes showed that these nurses produced no benefits to the patient or to the healthcare system. To compound these negative outcomes, Gournay and Brooking [9] also found that their interventions were, in economic terms, very expensive. It is worth pointing out that in Marks's nurse therapy training, nurses provided very focused, research-based interventions for specific diagnostic groups, whereas community psychiatric nurses in Gournay and Brooking's study were attempting to deal with an array of disorders (which probably had very high rates of spontaneous remission) using interventions with no research base. Indeed, although counselling has almost reached the status of a religion in some countries, there is to the present day, no randomised, controlled trial evidence to support its usefulness, unless it is applied in very specific fashion for very specific populations [4].
It should be noted that evidence for efficacy is only found where evidence is sought. This raises questions about who sets the research agenda, and whether we should focus our research efforts in areas where conditions are complex and research is difficult to carry out. Some examples include conditions that present in combination, such as social phobia with depression or treatment approaches which are difficult to dismantle, such as a combination of psy-chotherapeutic and pharmacological interventions. Another problem that may result from focusing too narrowly on evidence for treatment efficacy is that we may overlook components of treatments that lead to the maintenance of the status quo. A person who has recovered from an acute episode of schizophrenia but still needs to cope with some level of residual symptoms and social handicap may require support from a mental health worker to prevent further relapse. It could be argued that an approach that provides this support, although it may lack an evidence base, is nevertheless efficacious and without it the patient would deteriorate.
However, if we are to address the needs of specific populations of people with a mental illness, the decision-making matrix developed by Goldberg and Gournay [4] is useful because it provides a rational basis for targeting interventions. The first group in the matrix comprise severe mental disorders including the schizophrenias, organic disorders, bipolar disorder and life-threatening cases of eating disorder. They represent conditions which are unlikely to remit spontaneously, are associated with high levels of disability and where care will usually involve both primary care and the community mental health team (CMHT). Most of this group will require at least one hospital admission and will be referred for follow-up by a CMHT. A second group includes anxious depression, pure depression, generalised anxiety, panic disorder and obsessive-compulsive disorder. These disorders are also associated with low levels of spontaneous remission and disability but have effective pharmacological and psychological treatments available which can be offered entirely within a primary care setting provided staff have the specific training required. A third group of disorders include somatised presentations of distress, panic disorder with agoraphobia and eating disorders which have effective psychological therapies available and for which pharmacological therapies have a more limited role. Spontaneous remission can occur, but in the case of somatoform disorders and fatigue states they can become chronic with subsequent high levels of disability. In the case of panic disorder with agoraphobia and eating disorders, once again there are effective psychological treatments available and they can be managed within primary care, provided that there is access to trained staff. The final group comprise bereavement and adjustment disorder. These represent conditions that will usually resolve spontaneously and for which supportive help, rather than specific mental health intervention, is required.
It must be also be recognised that while some form of decision matrix is essential when allocating finite resources, there are difficulties inherent in this approach. The first question that arises is how do we judge disability? For example, in the case of a person suffering from schizophrenia who experiences impairment of daily living skills, their disability may be obvious to all and relatively easy to measure. However, the inner turmoil and subtle avoidance behaviours of many patients with social phobia is less easily judged in an objective fashion. In turn, if we use level of disability as a criterion for allocation to treatment, who makes the judgement? Should this be a primary care physician, a psychiatrist, or in the case of managed care in the USA, a desk-bound employee of an insurance company?
In order to address the needs of the target populations identified using the matrix, there is evidence that merely configuring mental health professionals into new case management-orientated community teams provides no benefit. Muijen et al. [10], in a randomised controlled trial conducted in South London, showed that community psychiatric nurses (CPNs) working as case managers had no better outcomes than CPNs working generically. However, Brooker et al. [11], in a quasi experimental study, provided some evidence that nurses can be trained to deliver research-based interventions, in this case family interventions for schizophrenia. Brooker's work and that of several other groups in the United Kingdom have led to the development of the Thorn Initiative. This project, which commenced at the Institute of Psychiatry and the University of Manchester in 1992, has developed a training program for nurses and other professionals and non-professionals in skills in research-based interventions in assertive community treatment, cognitive-behaviour therapy and family interventions. So far, more than 150 nurses have been trained, and although there are no controlled data available, the outcomes of patients is measured by an independent evaluator showed positive changes in clinical and social functioning in those people managed by people undertaking this course [12].
Fidelity to training is a key issue to consider. Kavanagh et al. [13] showed that workers trained in family interventions quickly stopped using their skills once they had returned to a service setting. Experience in the Thorn program supports this finding that even with intensive training, fidelity remains a key issue and that preparation of the service setting from whence the Thorn students have come for training, is critical. The work of McFarlane and his colleagues [14,15] reinforces this point. Their research demonstrates that evidence-based family interventions can be disseminated into routine clinical practice in a public hospital setting provided there is management support and that clinicians are provided with adequate training and supervision to ensure fidelity. Consequently, the involvement of service managers is crucial if investment in staff education and training is to be translated into more effective services. As the New York Demonstration Project of McFarlane et al. [14] showed, providing health services managers from the participating hospitals with information about the evidence base for family intervention, in particular its potential to reduce patient relapse rates, facilitated the implementation process. This was manifest in seemingly insignificant but fundamentally important decisions (e.g. adjusting staff rosters to enable them to attend evening meetings with family members). Clearly, service managers and clinicians must work cooperatively in the development of optimal services and training needs to be a key component of all services planning.
Areas for training
In order to consider areas for training, it is important to address the needs of those with severe mental disorders and those people who should probably always be managed by primary care team.
The training needs of those working with people with severe mental disorders
As stated, the evidence for effectiveness of interventions in mental health care is sparse. However, there are effective models of case management which can be classified under umbrella of assertive community treatment (ACT) techniques. There are tremendous difficulties in defining exactly what ACT and case management are, and indeed, these terms are used interchangeably. However, as several reviews [10, 16–18] have mentioned, there are clear differences in effective and ineffective models of case management. Effective models of case management are delivered by workers who have a range of clinical skills in assessment and therapy areas, case load sizes are small and the case manager has a central stable and therapeutic relationship with the patient. Conversely, ineffective models of case management are characterised by approaches which include brokering and networking as central interventions and case load sizes are very large.
Assertive community treatment, of course, comprises a number of components which must include brokering and networking skills to ensure that the person is availed of a range of services to provide them with the most effective social support. However, it is increasingly recognised that there are three therapeutic strands which should be pursued within this model. First, the use of medication is central to the management of severe and enduring illnesses and there is significant evidence to testify to the efficacy of drugs for the treatment of schizophrenia. Treatment options have now been enlarged with the addition of the new atypical compounds and these promise a great deal in terms of increased efficacy and reduced side effects. However, non-adherence with medication is, and probably will be for the foreseeable future, a substantial issue [19]. Recently, Kemp et al. [20] have demonstrated that a cognitive-behavioural package of education and motivational interviewing can lead to increased levels of adherence, and hence the reduction of relapse. A research team from the Institute of Psychiatry in London is currently investigating whether it is possible to train nurses in this approach by measuring both skill acquisition and patient outcome. Furthermore, these issues need to be investigated within the context of the rising tide of populations with a dual diagnosis of serious mental illness and substance/alcohol abuse [21].
Regarding the second therapeutic strand, a recent review by Mari et al. [22] conducted for the Cochrane Collaboration showed that family interventions in schizophrenia are an effective treatment modality. However, there is some debate regarding the relative efficacy of the components of this approach, with Solomon et al. [23] making a strong case for the use of more widespread educational methods rather than the use of programs that focus on expressed emotion [24]. It has already been stated that there are some data which suggest that family interventions can be taught to, and delivered by, mental health nurses [13]. However, as noted above, Kavanagh et al. [13] showed that fidelity to training in family interventions may be a major issue. Clearly, institutions that develop training programs need to attend carefully to this variable.
Regarding the third therapeutic strand of ACT, there is an increasing awareness of the potential usefulness of behavioural and cognitive-behavioural interventions with people with severe mental disorders. Smith et al. [25] have pointed to the substantial, randomised, controlled trial data, supporting the use of social skills training. This treatment modality is generally overlooked in favour of the newer, cognitively based approaches that have much less evidence to support efficacy. However, these cognitive-behavioural approaches are demonstrating some promise and, provided that they are delivered in juxtaposition to other therapeutic approaches, may have much to offer in the amelioration of the distress caused by hallucinations and delusions [26–28].
Therefore, the training priorities for the workforce dealing with the seriously mental ill should centre on skills in ACT, with a core of skills in assessment and therapeutic areas which should include behavioural and cognitive-behavioural interventions and psy-choeducational family approaches. At the same time, case managers need to have sophisticated skills in medication management as, de facto, they are in a pivotal position to report side effects and clinical status to the prescribing physician, while at the same time being ideally placed to deal with adherence issues.
The training needs of those working with primary care populations
The quotation from Shepherd et al. [3] cited earlier, concerning the need to strengthen the therapeutic role of the general practitioner, encapsulates the issue for the provision of services. Those in our mental health services probably need to focus on people with severe mental disorders, while at the same time support their colleagues in primary care to deal with the vast array (a great majority) of mental health problems. Within primary care, the numbers of people with mental health problems are so large that for the foreseeable future there will always be a need to make decisions regarding who should and who should not receive mental health services. It has been stated that the decision matrix can provide a rational basis for providing services. Goldberg and Gournay [4] believe that there is a great deal of evidence for effective non-drug interventions for some, but by no means all, problems. Using the decision matrix, they argue that conditions such as adjustment disorder and bereavement should seldom be targeted by specialist interventions, but this very large group of patients may need supportive care from voluntary agencies, self-help groups and other non-specialist workers. If one examines the effectiveness literature, the strongest evidence for effectiveness is attached to a cluster of conditions that also have low spontaneous remission rates and which, if untreated, cause significant handicap. Notable among these conditions are: panic disorder with agoraphobia, obsessive-compulsive disorder, some somataform disorders, and a few other problem categories. The overwhelming evidence suggests that behavioural and cognitive-behavioural therapies are successful, although there is of course some evidence to support other approaches (e.g. interpersonal therapy in the management of depression). However, the treatment of depression itself, and of generalised anxiety disorder in primary care, remains somewhat contentious as arguably these groups which may show a great waxing and waning, if not complete remission over time, may not show the magnitude of improvement sustained at follow-up attributable to, for example, behavioural treatments for agoraphobia [29,30].
The only training program in cognitive-behaviour therapy which has been comprehensively evaluated is that of Marks [2]. The program specifically targets phobic and obsessional disorders, and throughout its 25 years this narrow focus has been maintained. Nurses from this program have been involved in a randomised, controlled trial and economic analysis in primary care demonstrating benefits to the patient and the healthcare system for nurse therapist treatment compared with routine general practitioner care [6]. In addition, extensive audits of this program have been conducted which show significant gains in phobic and obsessional symptoms maintained to follow-up [7]. A 20-year follow-up of nurses trained on this course was conducted by Newell and Gournay [31], who showed that large numbers of these nurse therapists remain in clinical practice even many years after completion of their training and report that they continue to target specific groups with behavioural interventions. Furthermore, many of these nurse therapists continue to use multiple reliable measures of change. A 25-year follow-up of nurse therapists is currently underway conducted by Gournay and a team from the Institute of Psychiatry.
The nurse therapy workforce in the UK is a relatively small one, and it needs to be said that if one extrapolated from the data obtained by Newell and Gournay [32], this group, who each on average complete treatment on 80 patients a year, only treat a total of 16 000 patients annually. Goldberg and Gournay [4] have estimated that the entire workforce of the UK trained in cognitive-behaviour therapy (comprising psychiatrists, psychologists, nurses and others) will complete something less than 70 000 treatments per year. Given that the point prevalence of panic disorder with agoraphobia of clinical severity in the UK is approximately 300 000, it is clear that we need to make choices about where our efforts should be placed. As noted above, mental health workers need to use the decision matrix based on disabilities, spontaneous remission and effectiveness of available treatments. Nevertheless, we must consider ways of expanding the delivery of effective treatments. There is promise in a number of areas. For example, we have known for some time that phobic disorders can be effectively treated with minimal therapist time, using computer programs [33]. Recently, this work has expanded considerably with developments being made in computer-assisted treatments for phobias, general anxiety, obsessions and depression. Work with these developments indicates that consumer satisfaction levels are excellent and clinical gains substantial. The efficacy of self-help is also not to be underestimated. Recently, Tanner and Gournay [34] have shown that patients using a telephone conference facility run by anxiety sufferers and using behavioural self-help methods achieve good clinical outcomes. It is also important to consider the training of more numerous work forces. For example, there is evidence that general trained nurses can be used in the management of depression with considerable effect [35], and health visitors are effective in the management of postnatal depression [36]. However, as a corollary of these possible developments for disseminating effective interventions, in many countries there continues to be an emphasis on the use of ineffective interventions with populations who may get better anyway. For example, in the United Kingdom, counselling in general practice is extremely widespread with some 15 000 counsellors who are largely employed in the treatment of adjustment disorders with non-specific counselling. This phenomenon seems to continue despite evidence to suggest that counselling has no measurable benefits [37–39].
Discussion
What emerges from the above review of training issues is that we need to mount targeted initiatives. In particular, we need programs to train case managers within an assertive community treatment model, ensuring that these individuals have skills in clinical interventions, such as medication management, cognitive-behaviour therapy, and family work. Amodel for this training is already being developed in England within the context of the Thorn initiative. This program was originally confined to nurses, but lately it has become multidisciplinary, accepting students from a range of professional and non-professional backgrounds. The program has been subject to an evaluation, wherein patients case managed by these workers have been assessed before and after the period of case management. Initial results have been reported [12] and are very encouraging, demonstrating clear gains in clinical and social functioning in patients cared for by this group of workers, and there is also evidence that these workers do actually develop skills in clinical interventions and knowledge. Such training programs should be subject to a properly controlled trial, and although such an endeavour would be expensive, there is little doubt that this would be an excellent investment in our future.
With regard to the primary care area, the nurse therapy model of training is obviously one that needs some extension, and as noted above mental health services need to examine how more numerous groups such as general nurses could acquire skills in specific intervention for targeted groups.
Overall, training for the workforce represents a considerable challenge for those who provide education, and it seems clear that future training initiatives need to be highly focused on particular target groups with an imperative to train workers with skills in interventions for which there is a sound research base. Although the delineation is somewhat artificial, it also seems that we need to consider two distinct areas: that is, i.e. that of people with severe mental disorders such as schizophrenia, and the more numerous populations in primary care. Across the world, training programs for the mental health workforce have developed in an idiosyncratic fashion, and perhaps the future should be characterised by attacking the challenges posed by strategic, rather than ad hoc endeavours.
Footnotes
Acknowledgements
We are grateful to Harvey Whiteford, Commonwealth Director of Mental Health, for his helpful comments and to Frank Deane and Lindsay Oades, Illawarra Institute for Mental Health for their editorial support.