In S. A. Kirk (Ed.) (2005), Mental disorders in the social environment: Critical perspectives. New York, NY: Columbia University Press

Tomi Gomory

Florida State University

We have now sunk to a depth where the restatement of the obvious is the first duty of intelligent men.

George Orwell

The decades following World War II heralded a dramatic refocusing of psychiatric treatment for those severely emotionally disordered. The provision of care offered this very troubled and troubling population for most of America’s earlier history was the community-based decision to confine them to suburban or rural asylums (later government funded state hospitals) to avoid the toxic stress of community living (Grob, 1994a; Rothman, 1990). The approach was to "alienate" the patient, usually forcibly, from the "hubbub" of community life, which was deemed to be primarily responsible for their mental health problems in the first place, and allow respite and healing before reintegration:

However, by the early decades of the 20^th century various exposes of harmful and coercive institutional psychiatric practices were highlighting the problems of institutional care and undermining its optimistic early promise. Here is how one prominent writer, the journalist, Albert Deutsch described a visit to a psychiatric ward in the late 1940s:

Although the actual condition of the nation’s various state mental hospital systems was apparently improving by the 1950s this improvement was ignored for the most part and more attention was paid to expanding community focused care:

These political, philosophical, and practical developments encouraged the increasing involvement of the federal government in mental health policy and supported the notion that fragmented state services could only be mended by comprehensive federal planning and funding. In addition to the evolving mental health conceptual landscape, the introduction in the 1950s of psychotropic drugs with great "scientific" promise of psychiatric symptom reduction (functionally understood as increasing socially appropriate behavior without the need for obvious coercive physical restraint), provided the justification for the idea of successful rapid community reintegration with the help of appropriate "community-based" treatment of thousands of institutionalized psychiatric patients. Not only was this seen as an attainable reality, but more importantly, as a very significant moral and social good (Grob, 1994a).

Deinstitutionalization, as this approach was named, for better or worse became the focus and the political agenda of those making mental health policy from the 1950s onward. As the historian Gerald Grob (1994b) noting this sea change states:

The passage of the Mental Health Act of 1963 provided further impetus for the development and testing of community mental health treatment alternatives to hospital care. The Act’s hoped for effect according to Dr. Robert Felix, the then influential director of NIMH (Connery, 1968), as stated in his supportive United States Senate testimony for its passage was to facilitate, "the day when the State mental hospitals as we know them today would no longer exist" (p. 51).

Assertive Community Treatment (ACT), originally called Training in Community Living (TCL), the community mental health intervention program for the severely mentally ill (SMI), which is the subject of this chapter, was developed along with other community-based efforts during the late 1960s and early 1970s specifically to address the new federal mandate for shifting the locus of care to the community (for the then available community services "smorgasbord" see, Mosher & Burti, 1989). Psychiatrist Leonard Stein and Social Work Professor Mary Ann Test (1985), two of the inventors of ACT (Psychiatrist, Arnold Marx was the other), acknowledge that this agenda was the explicit rationale for creating ACT when they ask and answer the question:

Their very first article reported treatment success:

As of June 2003, there were over 360 scholarly articles in the PsycINFO database on ACT. Thirty-four states were using ACT or an adaptation and consuming well in excess of 160 million public tax dollars annually. The National Alliance for the Mentally Ill (NAMI), an organization made up of family members of psychiatric patients who believe that mental disorders are "brain diseases" and the inventers of ACT (both originating in Madison, Wisconsin) established a national nonprofit agency in 1996 (Allness & Knoedler, 1998) with the following agenda:

ACT has been acclaimed as "A gold award program" (Test & Stein, 1976 p. 193), with "the strongest empirical support" of all community mental health treatments for the SMI population (Essock, Drake, & Burns, 1998, p. 176), based on its "repeated and unequaled success in twenty-five years of confirming research" (Allness & Knoedler, 1998, forward). The Madison ACT model’s "success" is well-documented through the originators’ published research (summarized in Stein, & Test, 1985). In addition, there is a rapidly expanding literature by ACT replicators and others who are attempting to formally institutionalize this treatment as a model of "Evidence-Based Mental Health Treatment" (i.e., Phillips, Burns, Edgar, Mueser, Linkins, Rosenheck, Drake, & McDonel Herr, 2001). Psychiatric Services the influencial American Psychiatric Association journal dedicated itself in 2001 to promoting "Evidence-Based Practice" in psychiatric treatment and published several articles arguing that ACT’s effectiveness has been well-established by the experimental evidence and should be deemed an "evidence-based" treatment. Most recently, this effort appears to be under federal auspices and supported by America’s largest health care foundation:

ACT has been subjected to at least 25 randomized controlled trials (RCTs) which have been presented in numerous articles and literature reviews (i.e. Mueser, Bond, Drake,& Resnick, 1998) all agreeing with the claim reported in a special section focusing on ACT in the April 1998 issue of the American Journal of Orthopsychiatry that:

With Phillips et al., further arguing that:

While researcher Robert Rosenheck and his team (Rosenheck, Neale, Leaf, Milstein, & Frisman, 1995; Rosenheck & Niele, 1998), claim that their 10 site RCT, the largest ever done on ACT, "Showed reduced hospital use, cost savings, greater consumer satisfaction, and, in the long term, less severe symptoms and better community functioning" (Rosenheck & Niele, 2001, pp. 1395-1396).

The balance of this chapter will examine whether ACT does causally reduce hospitalization, or improve symptomatology, or provide other treatment benefits when compared to alternate treatment without any possible adverse effects as claimed. First a brief description will be offered about the falsificationist analytic approach employed in the chapter to review the ACT research, then, eight major claims concerning ACT research methodology and successful treatment outcomes will be examined. The format will be to provide the ACT claim followed by an analysis/discussion of the available scientific evidence. The concluding section will sum up the analytic findings of the case against ACT effectiveness.

Two methods have generally been used in scientific inquiry and evaluation. One approach, the one used by ACT researchers, accumulates confirmatory scientific data about theories or interventions through "objective" observations. The more such confirmations are found the better a theory or intervention is thought to be. The more "support" found, the higher the "reliability", providing more assurance about the efficacy of the theory or intervention in the future. The methodological effort in science to accumulate discrete data for the purpose of theory building or for seeking empirical proof or confirmation is called "induction". The "inductive method" has been rigorously critiqued by the evolutionary epistemology and accompanying methodology of Critical Rationalism (Popper, 1972; Campbell, 1987). Critical Rationalists (fallibilists) assert that induction is a false scientific method and recommend using another one based on deductive logic instead. The criticism against the inductive method put somewhat crudely but succinctly, argues three key points. First, that no "objective " observation is possible (all observations are filtered through fallible biological cognitive systems and are always socially contextual), second, that similar past events cannot logically predict similar future ones (the unknown future is underdetermined by the known past and may not resemble it), and third that the summing of such past discrete events cannot be reliably and validly used for universal generalizations going forward (one future negative event logically destroys such generalization’s "evidence-based" reliable universal standing, whether theoretical or pragmatic) (Munz, 1993). Due to the logical impossibility of providing absolute proof, but the logical possibility of providing falsifying instances, the preferred approach consists of the severe testing of ideas and empirical solutions through a reiterative trial and error feedback process. It consists of rigorous efforts at falsification and subsequent elimination of falsified theories or interventions using critical debate, empirical testing or skeptical reanalysis of research "findings" as done in the present chapter (for its application to clinical practice and research see the four article debate between Gomory and Thyer and commentary by other social work academics, Gomory 2001a; 2001b; 2001c; 2002a; Thyer 2001a; 2001b; Munro, 2002; Drisko, 2001).

One further comment needs to be added about looking for and finding support in research. It is always very easy to find supportive evidence when using a preferred theory. It has been found that our psychological predisposition is to look for agreeable evidence about our theoretical favorites while ignoring those that undermine our preferences. This inclination is called a confirmatory bias (Nickerson, 1998). So it behooves us, as rigorous researchers, to guard against confirmatory bias by the methodological commitment to constantly be vigilant and actively search for negative data, that is, experimental disconfirmation.

ACT research uses the very best measures and research methods for its evaluation studies:

Evidence. Instead of a rigorous "gold standard", we often find highly problematic methodology employed which is then down played and censored in the published literature contributing to ACT’s "success". Olfson (1990) in a draft report prepared for the National Institute of mental Health (NIMH), identified several problems with ACT research methodology and the resultant findings,

However, the article published by Olfson (1990b) based on this internal report, avoids mentioning any of these critical problems, referring to them merely as "methodological issues of interest mostly to researchers" (p. 640), providing a very positive impression of ACT effectiveness in the scholarly literature despite the contradictory empirical evidence.

NIMH and all subsequent ACT research have uniformly ignored the original recommendations of Olfson. The well known adverse effects of antipsychotic medications ( Gelman, 1999) are not addressed in any of the studies, although almost 50 per cent of ACT client contact time is spent dispensing and managing psychotropic medication (see note 1). No ACT studies have used blinded evaluators since the Marx et al. (1973) study mentioned by Olfson (Gomory, 1998). Additionally, only one ACT study has included a no-treatment control group (Solomon & Draine, 1995a; 1995b). Findings from comparison group studies only indicate if one treatment is superior to another. A no-treatment control group is required to evaluate if the experimental treatment is better than no treatment at all, an important question in evaluation research. The usual rationale for excluding a no-treatment group in psychosocial experimental research is the assumed unethical nature of withholding "effective" treatments from ill clients. But this ethical problem can only arise if the treatment withheld is known to be "effective". A trial is conducted precisely because we do not know if the experimental treatment is effective. No experimental treatment can be assumed to be effective a priori; it could turn out to be helpful, ineffective, or even harmful. In fact, placebo (pharmaceutical no-treatment) controlled research for the Federal Drug Administration is mandatory for gaining federal approval of all drugs even those for mental disorders; antidepressants for depression for example (i.e., Kirsch, Moore, Scoboria, & Nicholls, 2002).

Another serious methodological issue is the frequent use of unreliable or poorly validated research measures in ACT research. Gomory (1998) for instance, reviewed the use of the Short Clinical Rating Scale in the single Madison ACT study (Stein & Test, 1980), which claimed symptom improvement for the intervention and found the over all average "intraclass" correlations of that 13-item scale to be a troubling .59; ranging from .34 for hallucinations to .76 for thought disorder (Gomory, 1998, pp. 135-140). Additionally, the Cochrane Collaboration’s comprehensive ACT review noted:

Finally, outcome variables, some statistically significant and others non-significant, appear to be inappropriately combined to give erroneous impressions of ACT effectiveness where none was actually achieved. In a 14-year longitudinal study, the longest study ever implemented on ACT (Test, Knoedler, Allness, Burke, Brown, & Wallisch, 1991), the researchers state in part that:

The authors assert that the patients in the experimental group did significantly better in a "combined" category (time spent in hospitals/skilled nursing homes + penal settings + homelessness) than the control group over the whole experimental period (Allness & Knoedler, 1998, p. 5). This "combined" variable suggests that the experimental program not only reduces time spent hospitalized and in skilled nursing homes, (these settings are considered to be the same by the researchers, Test, Knoedler, Allness, Burke, Brown, & Wallisch, 1991, p. 243), but that it also significantly reduces, independent of this result, the amount of time spent in homelessness, and independent of either of the other components, time spent in penal settings. These are exactly the outcomes we would want such programs to impact.

The Madison ACT inventors report this combined variable for the first time in 1994, some 16 years after the inception of the long-term study in 1978. They introduced it in a paper presented at the 1994 annual meeting of the American psychiatric Association (APA) (Test, Knoedler, Allness, Kameshima, Senn Burke, & Rounds, 1994). Their previous published articles on this long-term study analyzed each of these variables separately, as was also done in their earlier studies (i.e. Stein, Test, & Marx, 1975). The "summing" of the three independent variables yielded statistically significant measurements favoring the experimental group at certain measurement periods. What this combining of previously discretely measured variables camouflages, is that the only significant difference between the experimental and control groups among these "component variables" was on the variable "time hospitalized/time in skilled nursing homes". Neither the "homelessness" nor the "penal settings" variable was statistically significant between experimental and control treatment for the first two years of the study, so any claimed statistical significance for the combined variable had to have been driven by the statistically significant difference found for the "time in hospital/nursing home" component. The only way one could have know this was by reviewing an earlier article (Test, Knoedler, Allness, Senn Burke, Brown & Wallisch, 1991), which gave the two year results of the long-term study, but did not mention any "combined" variable. Subjects in the experimental group spent significantly (p=. 001) less mean time, from study entry through 24 months, in hospital/skilled nursing home settings than the controls (Test et al., 1991, p. 243). But when it came to the two-year findings on homelessness and penal settings, which constitute two-thirds of the currently "combined" variable, the researchers state that,

Although the unpublished paper introducing this new "combined" variable is cited in the publication, The ACT Model: A Manual for ACT Start-Up (Allness & Knoedler, 1998) and elsewhere, to support the claim that ACT is an effective long-term treatment, the present author was refused permission by Dr. Test, the principle investigator of the long term ACT study, to quote or use the 7-year data directly from the 1994 paper. It is clear however, from the present author’s unpublished review of that paper that neither homeless conditions nor penal settings differ significantly between the two groups, not only for the first two years, but for the balance of the seven years as well. It is difficult to see why this combined variable was created so late in the analysis of the long-term study other then to suggest that there was a program effect, in reducing homelessness and jail time, when none in fact existed. Due to restrictions on researchers from quoting the unpublished text directly, we will have to await the ACT inventors long promised data and results in these domains for full clarification. Drs. Test and Stein admitted in a recent response to the present author’s published criticism of the lack of published data available for critical review 24 years after beginning the long-term ACT study (Gomory, 2001c) that:

Two years later we are still waiting for these published results.

"The assertive community treatment approach never was, and is not now, based on coercion" (Test & Stein, 2001, p. 1396).

Evidence. One of the primary rationales for shifting from psychiatric institutional to community care was the recognition that better social functioning required less coercion and more autonomous behavior (Marx et al., 1973). Another was the desire to reduce the harmful effects of institutional living. Coercion, "treatment" forced on a person against his explicit wishes, is seen, even by those who believe it useful, as the treatment of very last resort (Dennis & Monahan, 1996). Does ACT offer autonomy and coercion free treatment to its clients? What does the ACT program look like in practice? Stein (1990) explains:

This methodology appears highly intrusive. ACT activities may include such coercive moves as becoming the representative "financial payee" of the client, providing opportunities to blackmail the clients by enforcing medication compliance or threatening to withhold monies belonging to the client (Stein & Test, 1985, pp.88-89). Forcing treatment on ACT clients who do not want it is also routinely done (pp. 91-92). Even bribery may be appropriate ACT treatment: "it might be necessary to pay a socially withdrawn patient for going to the movies in addition to buying his ticket" (Test & Stein, 1976, p. 78).

To validate the use of assertive outreach and treatment, the original ACT researchers rely on two studies, one of which is their own (Test, 1981, p. 80). The other study is by Beard, Malamud & Rossman (1978), who describe their Fountain House outreach program as follows: " … phone calls, letters, and home and hospital visits made by both staff and members. Through such contacts, subjects who dropped out were provided with further information…. In those ... instances when an individual requested that no further contacts be made, his wishes, of course, where respected" (p. 624, emphasis added). Respect for the wishes of people who choose not to be involved in the Fountain House program contrasts with the coercive methods used by Test and Stein (1976):

Two questions come to mind: why is the patient described as ambivalent, when the patient’s reported behavior indicates a resolute opposition to going to work? Second, why is the patient’s active refusal redefined as "passivity"? The disregard of patients’ expressed wishes, and the reinterpretation of their behavior to justify programmatic interventions, appear to be the outstanding characteristics of ACT. The Fountain House model, by contrast, immediately discontinues client outreach efforts if asked. This difference leaves the ACT experts with nothing except their own self-validating research to support the particular assertive approach they advocate.

Coercion appears to be a vital part of the ACT model, according to the candid admission of Diamond (1996), a close associate of the original ACT group in Madison:

Finally, Gomory (2002) has reviewed the early professional history of the ACT inventors that helps conceptualize the present deep ethical and scientific difficulties of ACT. Using their published works the article analyzed and documented both their condescending paternalistic attitude toward mental health clients and their use of highly coercive methods to conduct "scientific" research which they claimed was based on behavioral principles, but which actually contradicted the explicit findings of the then current behavioral research literature. Coercion as "treatment" is rarely used in physiological medicine (to prevent the spread of highly infectious disease for example), but is routinely used in psychiatric medicine (to force highly toxic antipsychotics on resistant schizophrenics for example). To even be considered as ethically justified such coercive treatment must have beneficial outcomes for the recipient. Let’s review the outcome claims of ACT next.

ACT significantly reduces hospitalization when compared to standard treatment. (This claim has been primarily responsible for the enthusiastic response to ACT and the only consistent outcome found across all studies.)

Evidence. ACT methods have no direct bearing on the reduced hospital stays found in the studies. This result is due to a fairly strict administrative rule not to admit or readmit any ACT clients for hospitalization regardless of the psychiatric symptoms and to carry out all treatment in the community, while at the same time freely readmitting any troubled client in the comparison group. The ACT originators make this explicit in their first experimental trial, where they list "virtual abstention from rehospitalizing any patients being managed in the community" (Marx, Test, & Stein, 1973, p. 506) as their second treatment guideline. Similarly, in one of the acclaimed Australian ACT replications, "The project group patients were not admitted if this could be avoided: instead they were seen by members of the project team … who took them back to the community…." (Hoult, Reynolds, Charbonneau-Powis, Weekes, & Briggs 1983, p. 161). No effort was made to keep the control group from readmission and 96% were readmitted (p. 160). In sum, any decrease in hospitalization is not intervention dependent; it results from an administrative rule.

The ACT inventors realized, even in 1978 that time spent in inpatient settings was not a persuasive measure of outcome success for a mental health intervention:

ACT is more cost effective than standard interventions.

Evidence. Since hospitalization is by far the more costly treatment, the cost savings are not dependent on specific ACT interventions but on keeping people away from hospitals. Cost reduction occurs as a by-product of the ACT approach. Cost reduction could occur with any other treatment rigorously pursuing the same objective of not admitting patients to hospitals.

ACT provides significantly greater client satisfaction.

Evidence. Client satisfaction appears to be independent of distinct ACT activity. For example, in the Australian study previously mentioned, the claim of client satisfaction favoring the ACT methods is contradicted by the data. It appears that the greater autonomy provided by any community treatment, not the particular interventions of ACT cause this increased satisfaction. In this study the patients were surveyed at a 12-month follow-up: "The majority (80%) of experimental group patients who were not admitted to the hospital were pleased and grateful about it; only 30% of control group patients were pleased and grateful about being admitted to hospital, whereas 39% were upset and angry." (Hoult, 1986, p. 142). Stated differently, "Treatment preference was explored by asking all patients whether they prefer admission to Macquarie Hospital or treatment at home by a community team. The majority of the project (87%) and control (61%) patients preferred community treatment" (Hoult et al., 1983, p. 163, emphasis added). A majority (61%) of the group that did not experience the ACT treatment still preferred community treatment rather than incarceration in an institution. In fact, the experimental group reported that the most important elements of the ACT treatment were the availability of staff for frequent caring, supportive, personal contact and the enhanced freedom, therapeutic elements not specific to ACT (Hoult et al., 1983, p. 163).

Lending further support, a survey of "client perspectives" on ACT "ingredients" (McGrew, Wilson, & Bond, 1996) identified in order of preference "helping relationship, attributes of therapist, availability of staff, and non-specific assistance" as what clients liked most (p. 16, table 1). Again, these attributes are not ACT specific and are applicable to all forms of "helping". The least liked of the 25 elements associated with ACT treatment was "intensity of service", the component most representative of ACT’s philosophy. The survey's authors, themselves longtime ACT experts, admit, "Somewhat surprisingly, non-specific features of the helping relationship emerged as the aspects of [ACT] most frequently mentioned as helpful" (McGrew et al. 1996, p. 190).

Claim

ACT significantly improves symptomatology and client functioning, "Our study showed … in the long term, less severe symptoms and better community functioning" (Rosenheck & Niele, 2001, pp. 1395-1396).

Evidence. ACT specific treatment does not achieve significantly superior client functional or symptomatological improvement over an alternate treatment. This can best be explained by briefly reviewing the largest controlled trial (n=873), done on ACT (Rosenheck & Niele, 1997; 1998). Rosenheck and Niele’s quote is misleading. The study actually claimed to find this outcome only at one of the two types of settings where ACT was tested. Only "at the 6 General Medical and Surgical Hospital sites (GMS) (n=528) [was] ACT associated with greater improvement in long-term (2-year) clinical outcomes". At the 4 Neuro-Psychiatric Hospital sites (n=345) ACT did not differ from the standard treatment on clinical outcomes (Rosenheck & Neale, 1998, p. 459).

Methodological Difficulties. The claim of clinical effectiveness at the GMS sites rests on finding significantly higher community living skills favoring ACT patients in the four follow-up periods while finding only at "the final interview", significantly lower symptoms, higher functioning and increased satisfaction with services (p. 459).

The "finding" of significantly higher community living skills across treatment periods favoring the ACT group is belied by the data. At 6 and 12 months the "community living skills competence" scores favor the control treatment. The control group outperformed the experimental group for well over 12 months, but an impressive difference favors the experimental treatment at the exit interview (Rosenheck & Neale, 1998, p. 463, table 2 and Rosenheck & Neale, 1997, figure 3). The seemingly positive result in the exit interview could have been caused by many factors including (1) relief at being free of a coercive program, (2) fear of offending a potentially dangerous authority in a coercive program, or (3) a desire to please the interviewer. Meanwhile, an expensive, long-term, and potentially abusive program should not be justified on the basis of an exit interview that contradicts data gathered during the treatment period.

Unintended Research Results. The researchers identified post facto, two GMS study sites (no. 2 and 5) that did not fully implement the ACT treatment: Site 5 "developed a low-intensity patient tracking program rather than [ACT] services." (Rosenheck, Neale, Leaf, Milstein, & Frisman, 1995, p. 134); and site 2 provided substantially fewer community based services and under performed in most ACT categories when evaluated for program fidelity (Rosenheck & Neale, 1997, p. 11). Attempting to show that the increased costs of ACT were the result of these 2 sites’ ineffective ACT implementation, the researchers decided to reanalyze the data with these two sites excluded. They thus eliminated 34% of the original sample. This proved fruitful because the statistically significant difference found during the original analysis of costs was reduced to a non-significant difference, (Rosenheck & Neale, 1998, p. 463). More to the point, eliminating these two sites created an unintended experimental situation to reanalyze clinical outcomes.

If the dropped programs were less or completely ineffective, the reanalyzed clinical outcome measures of those sites effectively implementing ACT should have increased the statistically significant impact of ACT originally found. However, after " … excluding the 2 general medical and surgical sites that did not … implement the [ACT] program … clinical outcome results did not change" (Rosenheck & Neale, 1998, p. 463). In other words, with over a third of the original sample removed, about half of whom were essentially in a no treatment group, no change occurred in "the clinical outcome data". Being or not being in ACT made no difference to clinical outcome. Dr. Rosenheck confirmed that in the original analysis the clinical results of the two excluded sites were in the same direction and with similar significance as the results found at the other sites (personal communication, October 1997).

Claim

Occasionally, positive significant vocational effects are observed as a result of ACT (Marx et al., 1973; Stein & Test 1980).

Evidence. There are three ACT studies, which report positive effects on vocational outcomes and five, which do not. Two, which found a positive effect, were by the Madison group (Marx et al 1973; Stein & Test, 1980). However, their Long-term RCT building on these earlier suggestive studies found that ACT needs an "intensive vocational component" (Mueser et al. 1998, p. 56) to show any effect. The only other study with positive findings (Chandler, Meisel, McGowen, & Minz, 1996) corroborated the need for intense additional vocational efforts in such programs for positive vocational effects. ACT like other mental health interventions cannot significantly impact employment alone.

ACT has no negative effects, "The ACT literature has been very consistent in suggesting the absence of negative outcomes." (Bond, Drake, Mueser, & Lattimer, 2001, p. 149).

Evidence. Solomon and Draine found one of two key negative effects. The study was part of a larger randomized controlled trial of 200 homeless SMI's leaving an urban jail system (Solomon & Draine, 1995a) which aimed to test the effectiveness of ACT compared to individual case management and to a no-intervention control group. Solomon and Draine (1995b) noticed a high recidivism rate (56%) among the ACT group, compared to 22% among case managed individuals and 36% among the controls (p. 168). The researchers subsequently compared 22 clients in the ACT to 29 clients in individual case management in order to explain this unexpected finding. The significant findings of this second study were that clients of case managers (CM) who sought legal stipulations were more likely to return to jail, case managers were more likely to initiate a violation of probation process as an intervention strategy with clients for whom they sought legal stipulations and these clients returned to jail faster (p. 170). ACT case managers primarily pursued these activities. ACT clients due to ACT’s highly intrusive methods, were more likely to have their activities observed than in the other two control alternatives, resulting in their return to penal institutions more frequently (p. 172). Solomon and Draine (1995b) note, "These findings raise provocative questions regarding the possibility of deleterious consequences of intensive case management services for seriously mentally ill people" (p. 171). They also note the difficulties inherent in using coercion as treatment, as is done in ACT:

In addition, Solomon and Draine (1995a) found no differences in any domain between the three treatment groups. This result argues strongly for the use of no-treatment control groups in every ACT trial to determine whether ACT is even superior to minimum or no treatment.

A second negative effect, possibly related to the coercive elements of ACT, is the increased incidence of suicide in ACT settings. Cohen, Test, and Brown (1990, p. 603) report eight clear-cut and one possible suicides among the subjects of the long-term study conducted by the ACT originators, Test, Knoedler, Allness, Burke (1985). There may have been one additional suicide in this study. Test et al. (1985) report that the subjects in the study were given the structured interview from which the baseline data was obtained after three months of participation. Reporting on clients who were excluded from this interview they state, "It was not possible to interview five subjects: one committed suicide during the first three months" (p. 854). Since Cohen et al. (1990) reported only the data collected on those suicides that were given at least one structured interview; they may have left out the one suicide that occurred in the first three months.

Another study by Hoult et al. (1983) reports that "during the eight months after presenting at Macquire Hospital 10% of the project but none of the control patients were reported by relatives as having attempted suicide. These were … project patients, who prior to and during the study period made repeated suicide attempts" (p. 165).

Another replication study (Marks, Connolly, Muijen, Audini, McNamee, & Lawrence, 1994) reports that, "In the cohort of 189 patients, five died of self-harm in the 20 month study (three [ACT], two control). As with SMI suicides in Madison [Test and Stein’s study] such deaths were unexpected and occurred despite recent contact with staff" ( Marks, Connolly, Muijen, Audini, McNamee, & Lawrence, 1994, p. 187). While the result does not directly implicate ACT as a cause of suicidal behavior, it suggests that ACT was unable to prevent these suicides. The study’s authors spend considerable article space attempting to demonstrate that the ACT treatment was carefully and comprehensively provided to these patients. Several ACT patients are judged to be improved by the ACT experts immediately before they committed suicide. This indicates the problems with psychiatric evaluations. Psychiatry cannot reliably identify who will commit nor can it prevent suicides (Gomory, 1997).

Research is needed to explore the possible harmful coercive elements in assertive treatment that may contribute to both suicidal behavior and completed suicides. We should question the scientific validity and professional ethics of using any coercive methods in working with such vulnerable patients (Gomory, 1997). Marks et al. study’s ACT patients had very close attention paid to them by the assertive treatment team: "The … three [ACT] suicidal patients had had unusually persistent care … " (Marks et al., 1994, p. 187). Can such coercive scrutiny be counter-therapeutic?

Although ACT is packaged by institutional psychiatry and its various promoters as a discrete, well-tested, "evidence-based", modality of effective treatment, this chapter after examining some of its developmental history and the controlled experimental research finds little support for that claim, but finds instead that it is extremely paternalistic and has possible harmful effects. Why ACT remains aggressively marketed may be explained by the failure of institutional and biopsychiatric treatment efforts in general (Breggin, 1997; Fisher & Greenberg, 1997; Valenstein, 1998). It is also consistent with current trends to resort to increasingly coercive approaches. As Moser and Burti (1989) note in their comprehensive review of community mental health programs, "[t]he major problem [with ACT is] the level of paternalism, the ‘doing to’ activity it generates. … Medication is … the mainstay of the treatment. …In this system you can be cased managed to death, but no one is likely to sit down and spend … time discussing your experiences, thoughts, feelings and reactions" (pp. 344-345).

The paradigm of mental illness as medical/brain disease organizes and restricts the vast majority of potential research into helping interventions for seriously troubled persons to the biomedical model, the one model asserted to be "scientific" by institutional psychiatry and its powerful political lobbies (i.e. American Psychiatric Association; NAMI; the pharmaceutical industry), despite the scarcity of well-tested evidence for this explanation for "mental disorderliness". The Surgeon General’s recent mental health report (1999) confirms the conjectural nature of claims about the biological basis of "mental disorderliness", "there is no definitive lesion, laboratory test, or abnormality in brain tissue that can identify [mental] illness" (p. 44).

NIMH’s over one billion dollar annual budget sends a powerful signal. Researchers must attempt to find solutions that support and justify mental illness as medical/brain disorder if they expect to be funded. The ACT model paradoxically fully embraces this paradigm. As one ACT originator, Leonard Stein along with his co-author freely admits:

ACT’s misinterpreted early results appeared to demonstrate treatment success (Gomory, 1998). By not examining critically the outcome claims of these early studies, which often used unreliable psychiatric measures and instruments, but instead accepting the tautological arguments of the Madison ACT group (i.e. for reduced hospital stays) and relying on them for "support", the newer research "replications" not unexpectedly confirmed ACT "success". Once research careers are established around specific, well-entrenched treatment paradigms supported by politically savvy constituencies and are anointed as "evidence-based", the need for self-justification rarely allows self-critical admissions of error. Instead, contradictory evidence is ignored leading to ever more problematic results (Popper, 1962). ACT—a long-term, expensive, coercive and potentially unethical program—continues to be promoted as "evidence-based" despite research results that negate ACT claims of effectiveness.