Modern Community Care – What do we know that is effec‑ tive?
Modern Community Care – What do we know that is effec‑ tive? Burns, T. * Abstract: Community care has been seen a remarkable expansion in research in the last thirty years. Such research is beset with difficulties includ‑ ing fixing models long enough to get clear comparisons, the absence of consistency in description (particularly of comparator ser‑ vices) and the inevitable contamination from the ‘Pioneer’ effect of highly motivated teams. Assertive Community Treatment (ACT) teams are the most intensively researched but the evidence is contradictory. ACT is a complex intervention and a meta ‑ regression analysis is reported here that distinguished between the studies in terms of their component parts to identify effective and redundant ingredients. This analysis clari‑ fied the overwhelming impact of variation in comparator services. It also confirmed that the core ingredients in traditional generic CMHTs (multidisciplinary working, home‑based care and combined health and social care) ensured an equally effective outcome to the more in‑ tensively staffed and carefully prescribed ACT teams. Community mental health services need not follow one prescriptive model. Developing lo‑ cal services should be guided by the research into how effective aspects of care can be in‑ corporated into locally meaningful structures rather than importing complex systems from other health care cultures. Key‑Words: Community Psychiatry; Mental Health Services; Assertive Community Treat‑ ment. Introduction Modern Community care in mental health has evolved steadily since the second world war. It is traditional to date deinstitutionalisation from the introduction of the antipsychotic, chlorpromazine, in 1952. In truth, moves had been afoot for at least a decade. Querido had introduced an out‑patient service in the 1930s in Amsterdam1; the open‑door movement had been initiated in the UK in the mid 1940s cul‑ minating in Dingleton Hospital becoming a fully open‑door service by 19482; Russia had seen a rich development of day centres and dispensaries. As deinstitutionalization gathered pace with downsizing and closing of mental hospitals, sector community mental health teams be‑ gan to develop. These were predominantly in response to the complex needs of increas‑ ingly disabled patients who began to live outside hospitals. Inevitably they were multi ‑disciplinary because patients needs were var‑ ied. In the pioneer countries (France3 and the UK4) in the 1960s they not only recognised the need for breadth of input but continuity of in‑ put. 44 • Revista do Serviço de Psiquiatria do Hospital Prof. Doutor Fernando Fonseca, EPE *Professor of Social Psychiatr: University of Oxford, Department of Psychiatry ‑ The Warneford Hospital; e.mail: [email protected] Modern Community Care – What do we know that is effective? Sector teams were consolidated in the UK by the 1959 Mental Health Act5. This act required all hospitals which admitted compulsory pa‑ tients to provide them with outpatient care after discharge. It also required collaboration between health and social services in the man‑ agement of severe mental illness, in particular, involuntary care. These two conditions drove a sectorization of care. It was not possible to provide out‑patient care to all their discharged patients if hospitals did not sub‑divide respon‑ sibilities to smaller units. Similarly, it wasn’t possible to establish and maintain profession‑ al working relationships with social services across a broad canvas – local arrangements had to be made. The multi‑disciplinary sector team which evolved in France and the UK has persisted in the UK since that time and has spread. By the late 1980s nearly all inhabitants in Britain were served by such teams4 and they were in‑ troduced to considerable international interest in northern Italy after Law 1806,8. Mental Health Services Research Whilst clinical services were evolving rapidly and with increasing sophistication there was initially very little research into community mental health care. Partly this was cultural (the rise of evidence based medicine9 and health services research is a recent phenom‑ enon). Partly it reflected the problems facing clinicians. They need to respond to press‑ ing needs and can rarely afford the luxury of hypothesis driven service configuration. In addition services continued to improve gradually with few obvious “break‑points” that promote evaluation. Research in this area is further complicated by the rapidity of change. Mental health services research‑ ers are invariably aiming at a moving target as there are a number of external drivers to change which go beyond therapeutic devel‑ opments (e.g. politics, the media, social at‑ titudes). There have been a number of criticisms of the quality of early mental health services research10. These criticisms reflect several in‑ herent problems. Two affect mental health services research more than other health ser‑ vices research. One is the bias introduced by personal commitment. Studies are very long, the interventions are complex and are based on relationships and consequently research‑ ers are often highly committed to the service they are evaluating. A second is that research findings are high contextualised and depend‑ ent on local service configurations and policy issues. They translate poorly internationally11 and this has often been compounded by patri‑ otism. Notwithstanding these barriers, mental health services research into community care has now become a major international exer‑ cise12,13. Revista do Serviço de Psiquiatria do Hospital Prof. Doutor Fernando Fonseca, EPE • 45 Burns, T. * Phases of Community Care Devel‑ opment The development of community mental health care can be helpfully conceptualised as hav‑ ing three major phases. The first phase was that of deinstitutionalisation and the estab‑ lishment of community mental health teams by clinicians without a reflective academic framework. This phase can be considered one of ‘evolution’ and runs from the mid 1950s through to the end of the 1970s. With the pub‑ lication of Stein and Test’s landmark study14 in 1980, mental health services research es‑ tablished itself as a major endeavour, and with it the establishment of evidence‑based mental health teams. The most highly researched model for these teams was Assertive Commu‑ nity Treatment14,15. This period which ran from 1980 to 2000 can be considered the period of ‘revolution’. The third period arose when the steady stream of convincing results about evidence‑based teams was first confounded by contradictory results. In particular, two large UK studies (UK70016,17 and PRiSM17) failed to find the advantages which had been increas‑ ingly taken for granted. This “counter‑revolution” period which takes us into the present has introduced a much more rigorous and scientific evaluation of the components of care. This paper will preoccupy itself with this final phase, in particular ex‑ ploring how a careful examination of contra‑ dictions in the evidence can lead us to better understand what is effective in modern com‑ munity care. Assertive Community Treatment was first presented in a series of papers in Archives of Psychiatry in 198014,18,19. This was a seren‑ dipitous study stimulated by the closure of a ward. Stein & Test deployed the ward staff to be intensive case managers of their patients at a ratio of 10:1. They took their care out of the clinic and into patient’s homes. They were remarkably tenacious, following up patients one, two or three times a week and insisting that they took their antipsychotic medication. Their study was enormously influential. With only 126 patients randomised between the two services they demonstrated a remarkable reduction in hospitalisation (over two thirds) with significant improvements in social func‑ tioning and probably some in clinical func‑ tioning. Depending on how the study is inter‑ preted, their service was either cost neutral or actually cost‑saving. A striking finding of the study was that when the service was withdrawn (a consequence of financial constraints) all the advantages rapidly evaporated. ACT was therefore reconceptualised as an intensive multi‑disciplinary case management team providing assertive in‑vivo care for severely ill psychotic patients. It was not, as originally intended, a time limited training programme in community living but one that had to be provided continuously. 46 • Revista do Serviço de Psiquiatria do Hospital Prof. Doutor Fernando Fonseca, EPE Modern Community Care – What do we know that is effective? The Stein & Test Study has had an overwhelm‑ ing international impact. The services have been replicated and the studies have been replicated although predominantly in the US. It quickly be‑ came mandated by federal funding bodies in the US that ACT should be provided as an evidence ‑based service. This spread to Australia20 and in 1999 became mandated in the UK21 with the es‑ tablishment of over 300 new teams. The policy change in the UK was stimulated predominantly by the two influential meta‑analyses produced by Marshall and Lockwood15,22. Their conclusions from the evidence at time was that ACT reduced hospitalisation overall and that standard case management increased it. The UK700 Trial A note of caution however was founded by the UK700 Trial. This very rigorous multi‑centre trial carefully compared the care provided with intensive case‑ management (caseload sizes of 1:15) against standard case‑manage ment (caseload size 1:30). Despite this ma‑ jor difference in approach (including high fidelity ACT provision in London23) no re‑ duction in hospitalisation was found at all. The PRiSM study had similar results, indeed no European trial has every demonstrated a significant reduction in hospitalisation from ACT. It is the exploration of these contradic‑ tory findings which helped clarify what is and isn’t effective. Understanding the Variation ACT Outcomes The initial response to the UK700 and PRiSM studies was one of polemic and debate24,25. This could not advance the subject and therefore a scientific exploration of these differences was undertaken. This exploration26 used meta ‑regression analysis. Meta‑regression analysis goes beyond routine meta‑analysis in a num‑ ber of important ways. It allows skewed data to be included; missing standard deviations can be imputed from the data; multi‑site trials can be disaggregated and patient level data can be sought and included. In addition to a powerful meta analysis we established the model fidel‑ ity of all the reported services either from data in the paper or by extra information from re‑ searchers. This was calculated using the estab‑ lished scale, IFACT27. The meta‑regression analysis was used to test the variation in outcome against four hypothesised contributory factors. These were the date of the study (do earlier studies demonstrate more reduction?) This tests of the impact of increasing sophistication of re‑ search methodology (which has undoubtedly improved over time). Secondly, the size of the study (do smaller studies report a greater ef‑ fect size?) This tests for publication bias with filtering out of negative studies by journals. Thirdly, the effect of baseline hospitalisation rates (do higher rates of routine hospitalisa‑ Revista do Serviço de Psiquiatria do Hospital Prof. Doutor Fernando Fonseca, EPE • 47 Burns, T. * tion permit greater reduction by ACT?) This is to test the hypothesis that it is the quality of the control services in the UK which had re‑ sulted in no improvement from ACT. And last‑ ly, the effect of model fidelity (do teams with greater model fidelity result in greater reduc‑ tion?). The meta‑regression analysis demon‑ strated that neither the date nor the size of the study was significantly associated with reduction hospitalisation. However, baseline hospitalisation rates and model fidelity both did have clear and statistically significant as‑ sociation with reduction hospitalisation. Figure 1: Meta‑regression of baseline hospitalisa‑ tion against reduction in inpatient days. Burns, T. et al. BMJ 2007; 335:336. Copyright ©2007 BMJ Publishing Group Ltd. The association with baseline hospitalisation rates (figure 1) does not help to distinguish ef‑ fective ingredients from redundant ones. It con‑ firms that there is something about the effective component in the control group that makes the difference but it does not help understand which these are. However the model fidelity measurements of the experimental teams and their relationship to differences of outcome do. The IFACT scale is particularly helpful in that it contains three separate domains which are inde‑ pendently measured. These domains are resourc‑ es, practices and treatments. It is not possible to assess treatment retrospectively but it is possible to accurately measure practices and resources. When these two domains are separated in the meta‑regression analysis it becomes clear that resources (i.e. staffing levels – the most expen‑ sive part of the innovation) have absolutely no effect on hospitalisation rates (figure 2). This is a second confirmation of the finding of the UK700 study which had focused particularly on staffing levels. However, team organisation (practices) is responsible for all the variation (figure 3). What this demonstrates is that it is the nature of multi ‑disciplinary work rather than staffing levels which was effective when ACT was introduced in the US and Australia. In the UK effective multidis‑ ciplinary (‘ACT‑like’) working was already pre‑ sent in the CMHT controls so there was no benefit obtained by increasing the resources by reducing the caseload. 48 • Revista do Serviço de Psiquiatria do Hospital Prof. Doutor Fernando Fonseca, EPE Modern Community Care – What do we know that is effective? Which Ingredients Make a Difference? Figure 2: Metaregression of IFACT staff scores against inpatient days. In a previous study we tested for effective in‑ gredients of home based care using cluster analysis and regression analysis28. This is a sci‑ entifically less rigorous study than the meta ‑regression analysis but does give some impor‑ tant clues to the clinical variables that make a difference. In this systematic review of home based care12 we actively included ACT and all other forms of case management providing their aim was to deliver home based care and help severely mentally ill people remain out of hospital. In this study we did not use an es‑ tablished model fidelity scale but convened an expert consensus to identify important compo‑ nents of practice. In this expert consensus we used a modified three stage Delphi process29 with ten experts to agree essential components of care that could be operationalised and sent to researchers. The service questionnaire con‑ tained 20 items most of which could be iden‑ tified as ‘absent’ or ‘present’ with a handful requiring a simple categorical measurement. The questionnaire was sent to the 90 Principal Investigators of the research studies included in the systematic review and we obtained re‑ sults from 60. Figure 3: Metaregression of IFACT practice scores against inpatient days. Burns, T. et al. BMJ 2007;335:336 Copyright ©2007 BMJ Publishing Group Ltd. Revista do Serviço de Psiquiatria do Hospital Prof. Doutor Fernando Fonseca, EPE • 49 Burns, T. * finance – and medical care – psychotherapy, medication – within the team without formal referral). Some key features such as 24 hour working and a high level of medical input did not feature in this list. We also regressed the 20 characteristics against reduction in hospitalisation to see if any of them demonstrated a significant asso‑ ciation. The two that did were: regularly visit‑ ing at home and responsibility for health and social care. Figure 4: Components of care in Home based ser‑ vices. Social Psychiatry and Psychiatric Epidemiol‑ ogy (2004) 39: 789‑796. Wright, C, Catty, J, Watt, H, Burns, T. Cluster and Regression Analysis A simple count of these 20 characteristics found six frequently reported in experimen‑ tal services (figure 4). These were: smaller caseloads (1:20 or below), regularly visiting at home (a team policy for home visits beyond emergencies), high percentage of contact at home (over 60% of all contact conducted in patients homes), multi‑disciplinary team (at least three different trained professional disci‑ plines), integrated psychiatrists (psychiatrists attended all regular reviews not just those which required a medical input), responsibil‑ ity for health and social care (able to access both social care – housing, structured activity, Conclusions Most knowledge about what is effective in com‑ munity mental health care has been derived from the steady accumulation of clinical wis‑ dom. It is important not to ignore this, despite our emphasis on evidence‑based medicine. Evidence in this area, as has been pointed out earlier, is extremely hard to obtain. The ab‑ sence of experimental evidence does not mean the absence of knowledge. Research in this area is enormously time consuming and ex‑ pensive. It is also complicated by the fact that most of the interventions being researched are complex ones where it is hard to distinguish between a number of potentially confounding and effective factors. Luckily the variation in outcome has helped us get below the surface. Had the research into Assertive Community Treatment been 50 • Revista do Serviço de Psiquiatria do Hospital Prof. Doutor Fernando Fonseca, EPE Modern Community Care – What do we know that is effective? consistent in all its findings we would hard‑ ly be any further forward than we were in 1980. However the variations have forced us to deconstruct this complex intervention indirectly. By using the natural variation in outcomes and provision within these studies (both in the experimental and in the con‑ trol service) we have been able to tease out a candidate list of those components which appear to make a difference (the six factors in figure 4). This process has been remark‑ able in that it has also enabled us to dismiss the importance of a number of extremely expensive features (e.g. 24 hour availability and very small caseloads) which have long been held to be essential and whose prohibi‑ tive cost has prevented the implementation of services. Further refining of these ingredients by a regression against the variation in re‑ duction of hospitalisation has indicated that two of them (home visiting and combined health and social care) are probably the most important of the six. We are now in a position to say with a degree of confidence what effective home based care of people with severe mental illness needs to look like. It needs a multi‑disciplinary team which respects and includes both health and social care perspectives. It needs doctors who work actively within that team and are not seen as re‑ mote experts. It needs agreed case loads. The ev‑ idence does not seem to be that these case loads need to be very small. However there needs to be a cap on case loads so that case managers can predict their workload and vary it. It is worth bearing in mind those things which are so taken for granted such that they can’t be measured because they are ubiquitous. Conti‑ nuity of care, effective professional training and a respectful non‑discriminating attitude between professionals and towards patients appear to be uniform characteristics of good community mental health teams. The fact that they cannot be separated out and experimen‑ tally tested should not blind us to their core importance to effective community mental health care. References 1.Querido A. The development of socio ‑medical care in the Netherlands. London: Routledge & Kegan Paul; 1968. 2.Jones D. The Borders Mental Health Ser‑ vice. British Journal of Clinical & Social Psychiatry 1982;2:8‑12. 3.Kovess V, Boisguerin B, Antoine D, Rey‑ nauld M. Has the sectorization of psychi‑ atric services in France really been effec‑ tive? Soc Psychiatry Psychiatr Epidemiol 1995;30:132‑8. 4.Johnson S, Thornicroft G. The sectorisa‑ tion of psychiatric services in England and Wales. Social Psychiatry & Psychiatric Epi‑ demiology 1993 Feb;28(1):45‑47. Revista do Serviço de Psiquiatria do Hospital Prof. Doutor Fernando Fonseca, EPE • 51 Burns, T. * 5.Department of Health. The Mental Health Act: England and Wales. London: Depart‑ ment of Health; 1959. 6.Fioritti A, Lo RL, Melega V. Reform said or done? The case of Emilia‑Romagna within the Italian psychiatric context. American Journal of Psychiatry 1997 Jan;154(1):94‑8. 7.Jones K, Poletti A. Understanding the Ital‑ ian Experience. British Journal of Psychia‑ try 1985 Apr;146:341‑7. 8.Basaglia F. Breaking the circuit of control. In: Ingleby D, editor. Psychiatry and Anti ‑psychiatry.Harmondsworth: Penguin; 1981. 9.Sackett DL, Straus SE, Richardson WS, Rosenberg W, Haynes RB. Evidence Based Medicine ‑ How to practice and teach EBM. London: Churchill Livingstone, Harcourt Publishers Ltd; 1997. 10.Coid J. Failure in community care: psychiatry’s dilemma. BMJ 1994 Mar 26;308(6932):805‑6. 11.Burns T. Models of community treatments in schizophrenia: do they travel? Acta Psy‑ chiatr Scand 2000;102:11‑4. 12.Catty J, Burns T, Knapp M, Watt H, Wright C, Henderson J, et al. Home treatment for mental health problems: A systematic re‑ view. Psychological Medicine 2002;32:383 ‑401. 13.Mueser KT, Bond GR, Drake RE, Resnick SG. Models of community care for severe mental illness: a review of research on case management. Schizophrenia Bulletin 1998;24(1):37‑74. 14.Stein LI, Test MA. Alternative to mental hospital treatment. I. Conceptual model, treatment program, and clinical evalua‑ tion. Archives of General Psychiatry 1980 Apr;37(4):392‑7. 15.Marshall M, Lockwood A. Assertive Com‑ munity Treatment for people with severe mental disorders (Cochrane Review). The Cochrane Library . 25‑2‑1998. Ref Type: Journal (Full) 16.Burns T, Creed F, Fahy T, Thompson S, Tyrer P, White I. Intensive versus standard case management for severe psychotic illness: a randomised trial. The Lancet 1999;353:2185‑9. 17.Thornicroft G, Wykes T, Holloway F, John‑ son S, Szmukler G. From efficacy to ef‑ fectiveness in community mental health services. PRiSM Psychosis Study 10. British Journal of Psychiatry 1998;173:423‑7. 18.Test MA, Stein LI. Alternative to mental hos‑ pital treatment. III. Social cost. Archives of General Psychiatry 1980 Apr;37(4):409‑12. 19.Weisbrod BA, Test MA, Stein LI. Alternative to mental hospital treatment. II. Economic benefit‑cost analysis. Archives of General Psychiatry 1980 Apr;37(4):400‑5. 20.Hoult J. Community care of the acutely mentally ill. British Journal of Psychiatry 1986 Aug;149:137‑44. 21.Department of Health. Modern Stand‑ ards and Service Models: National Service Framework for Mental Health. London: De‑ partment of Health; 1999. 52 • Revista do Serviço de Psiquiatria do Hospital Prof. Doutor Fernando Fonseca, EPE Modern Community Care – What do we know that is effective? 22.Marshall M, Gray A, Lockwood A, Green R. Case management for se‑ vere mental disorders (Cochrane Re‑ view). The Cochrane Library . 2001. Ref Type: Journal (Full) 23.Fiander M, Burns T, McHugo GJ, Drake RE. Assertive community treatment across the Atlantic: comparison of model fidelity in the UK and USA. British Journal of Psychia‑ try 2003 Mar;182:248‑54. 24.Marshall M, Bond G, Stein LI, Shepherd G, McGrew J, Hoult J, et al. PRiSM Psychosis Study: Design limitations, questionable conclusions. British Journal of Psychiatry 1999;175:501‑3. 25.Gournay K, Thornicroft G. Comments on the UK700 case management trial. British Journal of Psychiatry 2000;370:370‑2. 26.Burns T, Catty J, Dash M, Roberts C, Lock‑ wood A, Marshall M. Use of intensive case management to reduce time in hospital in people with severe mental illness: system‑ atic review and meta‑regression. BMJ 2007 Aug 18;335(7615):336. 27.McGrew JH, Bond GR, Dietzen L, Salyers M. Measuring the fidelity of implementation of a mental health program model. Journal of Consulting & Clinical Psychology 1994 Aug;62(4):670‑8. 28.Wright C, Catty J, Watt H, Burns T. A sys‑ tematic review of home treatment services. Classification and sustainabiity. Soc Psy‑ chiatry Psychiatr Epidemiol 2004;39:789 ‑96. 29.Linstone HA, Turoff M. The Dephi Method: Techniques and Applications. Reading, Massachusetts: Addison‑Wesley Publishing Company; 1975. Revista do Serviço de Psiquiatria do Hospital Prof. Doutor Fernando Fonseca, EPE • 53