Site hosted by Build your free website today!

The Effectiveness of a Brief Treatment Model in a Community Mental Health Agency: Client Characteristics Associated with Self-reported Improvement

A dissertation submitted to the faculty of

Adler School of Professional Psychology

in partial fulfillment of the requirements for the degree

Doctor of Psychology

Paul J. Fitzgerald, M.A.

April 1997

©Copyright 1998, Paul J. Fitzgerald, Psy.D.

All rights reserved


Following the change to a brief treatment model at a community mental health agency, a group of 129 adult clients completed the Brief Symptom Inventory (Derogaitis & Spencer, 1982) after their first and sixth sessions of outpatient psychotherapy. Factorial analyses of variance were performed on the client group (n=125) for whom complete data were available, in order to assess the effect of client characteristics on treatment outcome after five sessions. The analysis did not reveal any characteristic or set of characteristics (including gender, prior treatment, and diagnosis at intake) that accounted for significant differences in the degree of change in BSI global severity scores between the first and fifth sessions of treatment. There were significant differences in BSI scores among the diagnostic categories, and there was an overall significant change downward in the global severity scores following five sessions of treatment. The investigation concluded that the five sessions of brief psychotherapy which clients received represented an effective intervention in terms of reducing client perceptions of their distress level, reinforcing the impression of the effectiveness of brief psychotherapy with this population. Suggestions for further research include investigation of client expectations from brief therapy, and qualitative research regarding the experiences of clients and therapists when there is reported worsening of symptoms, when symptoms remain unchanged, and when symptoms improve.


The author would like to express appreciation to the administration and staff of Family Service Centers for their cooperation and support. Special thanks go to Ronald Weglarz, Psy.D., Director of Training and Research, and Edward Marston, Ph.D., Clinical Director. The author also acknowledges the assistance, critiques, and suggestions of Lynne Deegan, Psy.D., and Chris Grunow, Psy.D. Appreciation is also due to Jerry Westermeyer, Ph.D., my chairperson, for valuable ideas and assistance in developing and focusing the scope of this project.

I also wish to express my gratitude and deep appreciation to Christine, my wife, and to my children, who have supported my efforts to complete this project with patience and understanding.

This paper is dedicated to the memory of Steve O'Brien, Ph.D., a mentor and professional without equal, a role model for social interest, and an inspiration to all who practice psychotherapy.

Chapter I - Introduction

In recent years, various models of brief psychotherapy and counseling have been described, and many authors and practitioners have given reasons for using brief treatment methods. One factor that has importance in both the public and private sectors has been the need to contain costs for mental health services; another has been the need to increase the accessibility of services and eliminate waiting lists. Other arguments for brief treatment have involved the expectation of increased effectiveness when treatment is directed toward the client's stated goals and needs. In this way of thinking, briefer therapy would be "the treatment of choice even if it were not less expensive" (Bloom, 1992, p. 8). Finally, less-apparent reasons for adopting a briefer treatment format include the fact that clients are often unwilling to continue with treatment for the length of time envisioned by the therapist, and drop out on their own.

Newer attitudes about the uses of psychotherapy have also become more prevalent in recent years. As Budman and Gurman (1988) point out, "...not everyone who requests treatment needs or can benefit from it" (p. 15). Indeed, it has been the experience of the author that many individuals and families seek psychotherapeutic services in order to find validation for their own strategies for coping with problems. Many clients will ask (if given the opportunity) whether the therapist believes that they need therapy. By this, they seem to mean the type of therapy that is generally portrayed in the media (Budman & Gurman, 1988, p.15). Budman and Gurman go on to postulate that most people would prefer to seek the help of psychotherapists as needed, much as one would consult an attorney or accountant (1988, p. 16), and that the therapeutic relationship (especially in its intense, ongoing form) is not as important to most people as therapists have assumed it was.

Family Service Centers (the community agency at which the present study was conducted) is a community mental health agency, which serves four townships of southern Cook County, Illinois. The Family and Individual Services program (which serves people who have not been psychiatrically hospitalized) uses a brief treatment model that stipulates that any length of treatment beyond twenty sessions requires administrative approval. During the time period of the study, this was modified and reduced to twelve sessions, with supervisory approval required after the first six sessions. This change was made because the waiting list had become too long, and was subsequently relaxed as the waiting list was eliminated.

The first session is the intake session; this is used for administrative case opening and assessment. During the time covered in this study, the agency administered the Brief Symptom Inventory (BSI; Derogaitis & Spencer, 1982) to clients before the second session and after the sixth session.

This study set out to investigate the following question: Given that a community agency has begun using a brief treatment approach, which clients are best suited to that approach, and are likely to experience the most benefit? This would be accomplished by examining which clients benefit most from the agency's six-session treatment model, by studying the client characteristics associated with the greatest change in the BSI. The research hypothesis was that some client characteristic (or set of client characteristics) might account for a significant difference in the improvement of BSI scores. These would be the clients who could be expected to respond quickly in brief therapy, and thus, may be the best clients to match to that modality. The client characteristics studied in the current analysis were: gender, previous therapy experience, and diagnostic category (mood disorder, anxiety disorder, adjustment disorder, and other focus of treatment).

Chapter II - Review of Literature

Although some writers have described brief treatment in terms of models that have been in existence for some time, including Adlerian therapy (Ansbacher, 1972; Dinkmeyer & Sherman, 1989), the arrival of brief treatment as a model of practice has developed for the most part since the 1960's (Burlingame & Fuhriman, 1987). In one of the earliest studies to address the impact of time limits on psychotherapeutic outcome, Shlien et al. (1962) used an Adlerian and a client-centered model with and without a time limit (of 20 sessions). The two groups actually averaged 18 and 37 interviews, respectively (treatment lengths which are comparable to those described in much of the later research on brief therapy). Their dependent measure was a Q-sort of items relating self and "ideal." These investigators found that, in general, clients showed greater improvement with time-limited therapy (either Adlerian or client-centered) than with unlimited therapy, when measured at termination and at follow-up. Unfortunately, no statistical analysis of these results was reported; however, the authors (who admit they were all experienced therapists who knew how to make effective use of limited sessions) use their results to advocate for brief therapy approaches within various theoretical orientations.

The two main lines of development of brief treatment since the early 1960's have been as an alternative to "traditional," open-ended dynamic psychotherapy (Straker, 1968); and as a method that developed independently of traditional dynamic therapy, and used for its own advantages. The latter line of development includes the influence of the family therapy field, especially the strategic therapists who built on the work of Milton Erickson (Haley, 1973). In the words of Erickson's daughter, the essence of his approach to brief treatment was that "[t]he therapist guides the client to set up a vast array of dominoes, and then reaches out and tips one. The rest fall into place by the slight impact from the first domino" (Erickson, 1988, p. 380).

This line of development has continued through the work of the Mental Research Institute and the Milwaukee Brief Treatment Center's "focused solution development" approach (DeShazer, et al., 1986; Walter & Peller, 1991). Family Service Centers' training during the period of this study included exposure to the solution-focused therapy approach, including the idea of "intermittent brief psychotherapy" for problems of longer duration.

The differences between time-limited therapy and brief psychotherapy have been categorized as "brief therapy by design and by default" (Budman & Gurman, 1988, pp. 6-7). The first idea is based upon a "contract" between therapist and client, and the latter idea reflects more general factors serving to make the therapy briefer. Obviously, an overlap between these two ways of describing brief treatment may exist. Brief psychotherapy may also make use of standards other than the number of sessions in determining when to end therapy. Most managed care organizations make use of periodic authorization of further therapy, based upon the progress in ameliorating symptoms. Family Service Centers' approach (based in part upon standards from the Medicaid program and from accrediting agencies) focuses upon the achievement of treatment goals agreed upon by the client and therapist, and reviewed regularly (Family Service Centers, 1996). This is brief therapy by design, but it is not time-limited therapy as such.

There is some suggestion that clients tend to self-limit their treatment at about ten sessions, primarily by dropping out of therapy if they feel that their problems have been adequately dealt with. This typically takes place at between eight and ten sessions, with the result that some outcome studies of brief psychotherapy have actually dealt with longer courses of treatment (e.g., 20 sessions) than studies of time-unlimited psychotherapy (Budman & Gurman, 1988, p. 5). The consensus of researchers in this area is that, despite the impression given in textbooks that the best therapy is longer-term, "therapists have been remarkably unsuccessful at convincing clients to commit to long-term psychotherapy" (Duncan, et al., 1992, p.6). In fact, one writer (Talmon, 1990) found that many of the clients he contacted for follow-up, after they kept only a single appointment, nevertheless felt that their problems had resolved significantly from the single session. Thus, whatever the schedule envisioned by the therapist, clients may stop treatment when they believe they have obtained relief from symptoms, or have otherwise reached their own goals.

The investigation of outcome in psychotherapy has a long history, marked by several "hotly debated and highly influential" studies by Rachman and Eysenck (cited in Bergin & Lambert, 1978, p.139) and others. These studies of the 1950's and 1960's dealt with courses of therapy as lengthy as two years, and (not surprisingly) it was difficult to assess the impact of the therapy, as opposed to the effects of spontaneous remission in the clients. One of the primary difficulties with many of these early studies was the inconsistency of the measures of improvement used (Bergin & Lambert, 1978, p. 140). In some of the early psychoanalytic studies, for example, the disappearance of symptoms was not enough. The researchers also looked for signs of personality changes consistent with their theory of dynamic change (p.141). Other studies have used ratings (such as "percent improved") given by the therapist, raising the problem of bias. Further, various investigators have handled the problem of dropouts in different ways. Bergin and Lambert conclude:

These figures lead to the conclusion that on the average about two thirds of neurotics who enter the broad spectrum of verbal psychotherapy as practiced over the last several decades, experience improvement in their symptoms. These surveys do not enlighten us much with regard to the effects of psychotherapy.... No single rate takes into account the responses of different diagnostic types nor the different populations served by specific clinics or therapists. (p.144)

Despite these limitations, the authors were able to suggest several conclusions from the research available up to that time. Among these were that psychotherapy seemed to have "at least modestly positive effects," that better-designed studies yielded more positive results, and that duration of therapy did not appear to be related to outcome (Bergin & Lambert, 1978, p.145). Their recommendations for further research included the use of self-report checklists and patient questionnaires, as well as therapist reports (pp. 176-178). They conclude by stating that psychotherapy seems to include strong non-specific factors or placebo effects, and that variation in outcome may result from any of several sources:

Equally important is the evidence that client characteristics are strong predictors of outcome. We believe the hypothesis is supportable that the largest proportion of variation in therapy outcome is accounted for by preexisting client factors, such as motivation for change, and the like. Therapist personal factors account for the second largest proportion of change, with technique variables coming in a distant third. (p. 180, italics added)

As research on psychotherapy outcome has gained in sophistication since the 1970's, researchers have focused on behavioral assessment and self-report, often using instruments devised for the purpose by the authors of the research. The most commonly used standardized instruments included the State-Trait Anxiety Inventory, the MMPI, and the Beck Depression Inventory (Lambert, 1983, p. 11). A number of studies also made use of physiological measures and external trained observers. Each of these approaches focused on some of the dimensions of therapeutic change, but in the late 1970's and early 1980's, researchers began to acknowledge that measuring psychotherapy outcome necessitates looking at a multidimensional process (Lambert, 1983, p. 19).

In keeping with this line of development, Howard, et. al (1986) suggested that a dose-effect relationship might exist in psychotherapy, in contradiction to Bergin and Lambert's opinion stated above. They felt that clarifying this relationship would be an important part of aggregating data across studies, in order to control for the effect of treatment length (much as medical studies control for the dosage of a medication). They favored an approach which takes the percentage of clients displaying improvement after a given number of sessions, and analyzed a number of studies to find this relationship. Their results indicated that about 10% to 18% of patients could be expected to show some improvement before the first session (just by making contact with the providers), and that about half should show improvement by the eighth session. About three-fourths of patients can be expected to show measurable improvement by the 26th session, or about six months of treatment (Howard, et. al., 1986, p. 162). This study, while providing extremely valuable information on this dose effect, remains limited by one of the problems described in earlier reviews; namely, the inconsistent application of measures of improvement. In various studies, these included reviews of progress notes, self-reports, and therapist ratings. These reviewers also raised two interesting issues: first, that they felt that clients who had less than six to eight sessions of therapy had not really received a "dose" of it, because fewer than 50% of patients in the studies showed measurable improvement at that point in time. The second point they make is that patients in time-limited therapy actually end up receiving more sessions than the median dose of unlimited therapy (Howard, et al., p. 163). This is the same point that has been noted above (Budman & Gurman, 1988, p. 5).

Outcome studies of brief psychotherapy have included investigations of the effect of therapist training and experience on brief psychotherapy outcome. In one such study (Burlingame, et al., 1989), therapists' experience level and training in brief therapy were compared using their clients' scores on the Symptom Check List-90 and the Brief Hopkins Psychiatric Rating Scale (two predecessors of the BSI) in an outpatient clinic setting. One therapist group (the "No Training" group) received only an instruction to keep their treatment within the eight-session format. Another group (the "Self-Instruction" group) received written instructions in brief treatment, while a third group (the "Intensive Instruction" group) received role-playing and didactic training in brief therapy techniques. Although they were instructed in a brief therapy procedure devised by the authors, they were told to use whatever theoretical orientation they had previously employed with clients. The therapists were matched on experience level in being assigned to these groups. The principal dependent variable was the change in self-report scores on the inventories used (although premature termination and recidivism were also studied). The authors found that changes in client scores were not significantly different across the three training conditions, nor across therapist experience (the therapists were divided into intern and experienced therapist groups). The only significant result was the finding that clients of the "No-Training" group of therapists had significantly more premature terminations than those of the other two therapist groups. The authors also found that the eight clients who displayed significant improvement in self-report scores were all clients of the trained therapists rather than the non-trained therapists.

The study described above focused on the characteristics of the therapists in measuring treatment outcome. Its authors suggested (despite the lack of significant results) that further exploration of the role of training may be needed. However, no consideration of client characteristics was made. The type of problems being brought to the clinic by clients may have an effect on their response to the type of treatment being given. Conventional therapist wisdom, for example, says that clients with personality disorders (particularly borderline personality disorder) are the ones who often require long-term psychotherapy (E. M. Marston, personal communication, July 21, 1995). A difficulty in studying this question lies in the fact that therapists' impressions about clients' Axis II diagnoses are inconsistently recorded, and may only be discussed between therapists (or between therapist and supervisor). There are usually presenting symptoms which qualify clients for Axis I diagnoses in an outpatient mental health setting, and (in the author's experience at Family Service Centers) these were often mood disorder diagnoses.

Chapter III - Methodology


Client population. The study utilized records of adult clients attending Family Service Centers' four offices, in the Individual and Family Services program. These clients have not been referred as a result of having been psychiatrically hospitalized; rather, they were all clients who came through the intake worker after calling to request services themselves. Adults under twenty-five years of age, and over fifty years of age, were excluded because they tend to have concerns which represent different stages of life. There were also too few clients from each of those groups to allow for adequate comparison. Some clients were receiving medication, either from their personal physicians or from agency psychiatrists. Almost without exception, the clients studied received some form of financial assistance with their psychotherapy costs. Some were Medicaid recipients, and most others were charged a sliding scale fee that was less than the customary cost of psychotherapy. Parents who were seen as collateral clients with their children under 18 years of age were also excluded; they were not asked to complete the self-report measures since they completed other rating scales for their children. Of the 192 matched (pre- and post-tests) sets of BSI records in the archival database at the agency, 145 fell within the age range of this study (i.e., between 25 and 50 years of age). Of those, 16 had invalid profiles due to incomplete items, leaving 129 cases in the data analysis for this study.

The cases were grouped by diagnosis according to the following categories of the DSM-IV (American Psychiatric Association, 1992): (1) Anxiety disorders, which includes all post-traumatic stress disorders, generalized anxiety disorder, panic disorder, and related disorders; (2) Mood disorders, which includes dysthymic disorder, major depressive disorders, and the bipolar disorders; (3) Adjustment disorders (in a category of their own, regardless of the presenting symptoms such as anxious or depressed mood); and (4) Other diagnoses and V-Codes. The last is a mixed group which includes relational problems (8 cases), substance use disorders (4 cases), impulse control disorders (3 cases), insomnia (1 case), bulimia (1 case), pedophilia (1 case), and schizoaffective disorder (1 case). This particular grouping resulted in the groups which were most nearly equal in number (though the mood disorders group, the largest at 57 cases, was more than three times larger than the anxiety disorders group, the smallest at 16 cases). This grouping also followed the DSM-IV organization as closely as possible, given the types of adult clients normally seen at an outpatient community mental health program such as Family Service Centers.

Tables I and II (Appendix A) list the demographic and diagnostic characteristics of the sample. Of the 129 clients whose results were analyzed, 47 (36%) were male, and 82 (64%) were female. The average age of the males was about 35 years, and the average age of the females was about 36 years. Fifty-seven clients (44%)were diagnosed with mood disorders; 37 clients (29%) were diagnosed with adjustment disorders; 19 clients (15%) had other diagnoses or V-codes; and 16 clients (12%) were diagnosed with anxiety disorders. Fifteen of the male clients (32%) reported having had prior treatment, and 26 of the female clients (also 32%) reported prior treatment. As might be expected, the group diagnosed with adjustment disorders had the fewest clients (24%) reporting prior treatment, and the group diagnosed with anxiety disorders had the most clients (44%) reporting prior treatment.


The therapists whose clients were studied in the present research included both regular employees and trainees (pre-doctoral psychology interns, doctoral practicum students in psychology, master's degree practicum students in counseling psychology, and field work students from master's degree programs in social work). The employed staff included master's level therapists (with social work and counseling psychology degrees) and those with doctoral degrees in psychology. Therapist experience varied widely, and various theoretical orientations were represented. Therapists were not required to identify their theoretical orientation, but many did so on an informal basis. The most prominently reported theoretical orientations were cognitive-behavioral and psychodynamic; however, a number of therapists described themselves as eclectic. Client consent and human subjects' rights. The clients whose records were studied in this project were given the BSI as part of an agency study of treatment effectiveness and quality assurance. The two administrations of the BSI were given in place of a single administration of the Symptom Check List-90, Revised (SCL-90-R), which was formerly (and subsequently) used as part of the intake process to help identify concerns for treatment. The clients gave consent for all of these questionnaires at the time of intake, and the results were made available to the therapists. The clients were not asked to give consent for participation in the research conducted in this study, for the following reasons: (a) the study utilized only archival data from treatment that had already been given; (b) results and client data were kept on computer disks with all identifying information removed, ensuring confidentiality and preventing any use of the clients' reports for any other purpose; (b) the clients' treatment was not affected nor influenced in any way by the current analysis of the data. The agency's Quality Assurance Committee performed a human rights review of the research proposal and offered no objections to the current use of the client data.


This study utilized the clinical scales and the Global Severity Index of the Brief Symptom Inventory (Derogaitis & Spencer, 1982). The BSI is a self-report inventory consisting of 53 items, yielding nine clinical (symptom dimension) scales, and three global indices (in addition to the above-mentioned Global Severity Index, which includes the number and severity of symptoms, there is also a Positive Symptom Distress Index, and Positive Symptom Total Score). The symptom dimension scales include the types of items listed in Appendix B. Individuals taking the BSI respond to each item on a 5-point rating scale. Most clients are able to complete the BSI within ten minutes.

The global scales all have test-retest reliabilities above .80 (Conoley & Kramer, 1989, p. 111), making repeated administrations of the BSI feasible as a measure of improvement in psychotherapy. While the clinical scales all have internal consistency coefficients in the .60 to .88 range (Conoley & Kramer, 1989, p. 111; Broday & Mason, 1991, p. 94), these scales share a large degree of variance among themselves and with the GSI, indicating poor discriminating ability (Boulet & Boss, 1991, p. 433). Thus, the difference among patterns of clinical complaints reported by clients appears to be less clearly revealed by the BSI than is the overall level of symptomatology, as expressed by the global scores. Because the Global Severity Index is most commonly regarded as the overall measure of distress, this study makes use of the GSI before the first and after the fifth therapy session, as an outcome measure.

The BSI has no validity scales, and does not attempt to identify response patterns that would indicate any bias or attempt to fake results. As a self-report measure of the client's freely revealed levels of symptoms, this has not generally been regarded as necessary for the instrument. It is generally used as a screening instrument rather than for differential diagnosis (Boulet & Boss, 1991, P. 436). Like its predecessor, the SCL-90, it is relatively well-suited for measuring clients' perception of distress and of improvement in psychological well-being, making it useful in psychotherapy outcome research.

The BSI as administered in the current study was scored by computer at a processing center located away from the agency. At the same time, information about client characteristics, demographic data, and the intake therapist's diagnoses were compiled along with the BSI scores.


The BSI was administered to all adult clients beginning therapy at the agency before the second session (the first therapy session after the initial intake interview and case opening). After the fifth therapy session (the sixth actual episode of contact with the client), the BSI was administered again. The therapist was responsible for ensuring that both administrations were completed properly, although most clients gave their answer forms to the receptionist at the agency. Once there was a match for a given client, both test forms were sent in for scoring.

Therapists were not given any particular directions for how they were to conduct therapy in a brief treatment model. Announcements were made to all existing staff and new staff members (both employees and trainees), that any treatment beyond twelve sessions would require supervisory authorization. Inservices and staff training regarding methods of brief treatment were an ongoing part of the agency's supervision and staff development. During the period of time (July 1994 through July 1995) in which BSI's were administered, the staff training seminars covered several models of brief psychodynamic therapy (all of which are described in Bloom, 1992). In addition to the solution-focused therapy approach (De Shazer et al., 1986; Walter & Peller, 1991) already described, further training was given in "intermittent brief therapy" (Bonjean & Spector, 1994) which could be utilized with even severely disturbed clients. Particular attention was also paid to Sifneos's "Short-Term Anxiety-Provoking Psychotherapy" (Sifneos, 1992), in which the therapist helps the client to identify a "dynamic focus," or central issue (pp. 47-52). One of the assumptions made by this study is that all staff members were reasonably prepared to provide short-term psychotherapy, but that each therapist was free to make use of whatever methods or theoretical models suited her or him best. There was, moreover, no measurement made of staff training, experience, nor competence in providing brief treatment. For the purposes of this study, therefore, the author is making the assumption that the therapy being provided is reasonably representative of the type of short-term therapy being performed at any comparable community agency or practice serving a similar client population at the current time. There is a further assumption that any such typical therapist or agency will have had access to continuing staff development training in the area of brief psychotherapy practice.

Data analysis

A multivariate analysis of variance was used to investigate the difference between pre- and post-test GSI scores, and to look for significant interactions with diagnostic group, gender, and prior treatment. The test was performed twice: One analysis used the difference between GSI pretest and posttest scores as the dependent variable; this is referred to in the results as the "within subjects" analysis. The second analysis used all of the GSI scores (pre-test and post-test) together as the dependent variable; this is referred to in the results as the "between subjects" analysis. Further analysis of the data included post-hoc tests on the results of the analyses of variance, as well as individual t-tests on the between-subjects analysis to further ascertain the groups which differed significantly. Finally, correlations (using Pearson's r) were performed on the clinical subscales to analyze their relation to the global severity index.

Chapter IV - Results

Descriptive statistics

The first finding of the study was that the clients' Global Severity Index scores decreased from pretest to posttest across all the diagnostic groups. Table 1 (Appendix A) summarizes the changes in raw GSI scores on the BSI for each group of clients, and Figures 1 through 4 show the results for each diagnostic group in graphic form, including the clinical subscale results for each diagnostic group on pretest and posttest.

The overall population mean change of -.032 points was essentially identical to the means for male and female subjects (-.031 and -0.32, respectively). This indicates an improvement of 25.7 percent overall in the clients' subjective levels of distress. Table 1 also indicates the breakdown of GSI changes by diagnostic groupings and prior treatment. There is little difference between these means, except that the clients with adjustment disorders showed greater improvement (-0.46 points) than the other diagnostic groups (who averaged -.025 points change). This difference was even more pronounced for female clients than for male clients (female clients with adjustment disorders averaging -0.61 points change, while female clients with anxiety disorders averaged only -0.12 points change). This represents a 49 per cent decrease in average GSI scores for female clients with adjustment disorders, in contrast to a decrease of just under 10 per cent for female clients with anxiety disorders.

Analyses of variance

The two primary statistical analyses of the study were the repeated measures analysis of variance (Table 2), using Global Severity Index as the dependent measure for each diagnostic and demographic group, and the multivariate analysis of variance (Table 3) using the aggregated (pretest and posttest) Global Severity Index as the dependent variable, and client demographic and prior treatment data as the factors.

Pretest and posttest differences: within subjects analysis of variance

The first analysis (summarized in Table 2), failed to find significance in the effects of gender, prior treatment, or diagnostic groupings (or the interactions of these factors) on the changes in global severity index scores (within subjects) from pretest to posttest. However, the overall difference between pretest and posttest scores (the main effect) was highly significant, at greater than the .01 level.

The second analysis, which was the between-subjects multivariate analysis of variance (Table 3), found that the main effects of diagnosis and GSI scores were significant, without regard to pretest vs. posttest status. This was confirmed by a single-factor analysis of variance using diagnostic groupings with the GSI from pretest and posttest measures (Table 4).

Post-hoc t-tests were conducted, in order to further assess the occurrences of significance across the various diagnostic groups. Those results are summarized in Table 5. For the pretest Global Severity Index scores, the following comparisons yielded significant differences at the p<0.01 level: Anxiety Disorders vs. Adjustment Disorders, Mood Disorders vs. Adjustment Disorders, and Mood Disorders vs. Other Diagnoses/V-Codes. For the posttest GSI scores, the following differences were significant: Mood Disorders vs. Adjustment Disorders, and Mood Disorders vs. Other Diagnoses/V-Codes. For the combined (pretest and posttest) scores, Anxiety vs. Other, Mood Vs. Other, and Adjustment vs. Mood were significant.

Although the multivariate analysis of variance failed to find significant effects of any of the client characteristics on changes in global severity index scores in the sample, there was significance found in the overall difference between pretest and posttest scores. This suggests that, as a group, the clients seen at Family Service Centers during the period covered by the study did obtain significant benefit from the six-session treatment model in use by the agency. .

Differences between diagnostic groups: further results

Another notable result was that different diagnostic groups reported different patterns of GSI scores, and these differences were statistically significant among diagnostic groups. Further, the GSI scores of the various diagnostic groups seem to reflect different patterns of responding to the items on the instrument. This information is available from a summary of the symptom dimension scale (subscale) scores of each diagnostic group, shown graphically in Figures 1 through 4. Although (as discussed above) these subscales have not been found to have a high degree of discriminating power, they are nevertheless a potential source of useful clinical information.

As may be expected, the Anxiety Disorders group reported higher Anxiety and Phobic Anxiety scale scores than did the other groups. They also reported relatively high levels of all other types of symptoms. The clients in the Mood Disorders group reported high levels of depression, hostility, interpersonal sensitivity, paranoid ideation, and psychoticism, and relatively lower levels of phobic anxiety. The Adjustment Disorders group reported patterns of symptoms similar to the Mood Disorders group, but at a slightly lower level. However, the Adjustment Disorders group, on the whole, reported larger decreases in symptoms at post-test than did the Mood Disorders group. (This result is evident in the post-hoc tests, but was not evident in the multivariate analysis of variance due to the moderating effects of the other two diagnostic groups). The "Other" diagnostic group showed higher levels of hostility and lower levels of phobic anxiety symptoms (relative to their own group's global severity scores) than the other groups displayed in relation to their own global scores. In general, the mood disorders group reported the most distress in all areas (on both pretest and posttest); the anxiety disorders group reported the second highest levels of distress, followed by the adjustment disorders group and the "other" diagnosis group, in that order. This is the source of the significant effect of diagnostic group on GSI scores in the multivariate analysis of variance.

Changes for better and worse among client group

Table 6 summarizes the percentages of clients displaying significant decreases and increases in GSI scores from pretest to posttest. Significant change was calculated based on a 95 percent confidence level, using the variance of this sample's Global Severity Index scores on the pretest. These percentages are given as an indication of the "dose effect" of the five sessions of psychotherapy. Approximately two-thirds of the clients in each of the four diagnostic groups showed significant decreases in GSI levels. Approximately one-quarter of clients remained essentially unchanged, while about ten per cent of clients showed increases in reported distress levels. This occurred despite the fact that the mean GSI scores for each of the groupings used decreased significantly from pretest to posttest. In summary, most of the clients reported getting better over the five sessions, but a minority reported no significant change, and a smaller minority reported significantly more distress after five sessions.

Comparison of clients to the BSI criteria for treatment

Table 7 shows the number and percentage of clients who met each of the criteria given by the authors of the Brief Symptom Inventory for needing treatment. These two criteria are: A Global Severity Index of 63 (T score) or higher, or two symptom dimension scales with T scores of 63 or higher (Derogaitis & Spencer, 1982) . These are listed separately in Table 7 for this sample. The criterion of GSI being over 63 was met by about three-fourths of the clients in the sample at pretest, and by only about half at posttest. The criterion of two elevated symptom scales was met at pretest by 12 percent of the clients, and by 15 percent at posttest. Combining both criteria, about four-fifths of the sample met the authors' standard for needing treatment at pretest; this declined to about two-thirds of the clients at posttest.

Correlations of Symptom Dimension Scales with Global Scale

Tables 8 and 9 list the correlations of each clinical symptom scale score with the GSI score, both at the pretest and in terms of the difference between pretest and posttest scores. At pretest (Table 8), all these correlations are significant and all exceed .70, indicating that the clinical scales are all highly correlated with the Global Severity Index, in this sample. The differences between pretest and posttest scores on all the scales are less highly correlated, but still suggest that improvement (as well as initial distress) is reflected highly in the clinical scales as well as the Global Severity Index.

Chapter V - Discussion

Interpretation of results in light of research literature

Treatment effectiveness

Assuming that this sample is representative of community mental health center (non-aftercare) clients, it may be stated that the type of psychotherapy provided to this client population appears to be highly effective in achieving relief from distress, as evidenced by the fact that Global Severity Index scale scores were lowered from pretest to posttest. Moreover, the lowering of scores was highly statistically significant, suggesting that treatment provided in the first five sessions can have a significant effect on client distress. This result goes beyond the somewhat more equivocal results that have been accumulating in the literature over the past two to three decades, demonstrating that a more reliable measure of client progress should tend to strengthen the outcome literature beyond what had been found when various less-reliable measures were mixed in various studies.

Initial severity and degree of improvement

The clients in the Adjustment Disorders group had the lowest scores at pretest (as a group), and those in the Mood Disorders group had the highest scores at pretest. This may need to be taken into account in making any interpretation of these results. The clients with more distress initially may appear to improve more dramatically due to several factors, including statistical ones (particularly regression to the mean) and subjective client perceptions (including higher motivation for treatment among those with more initial distress).

That being said, the percentage of clients who improved, remained the same, and worsened (Table 6) was relatively constant across diagnostic groups. This suggests once again that the type of brief treatment being studied here is equally effective for all types of clients being seen at the agency. Interestingly, the adjustment disorders group showed the highest percentage of clients improved, as well as the highest percentage of clients who were subjectively worse. The mood disorders group also showed slightly more clients reporting worse symptoms after treatment than did the remaining two groups. It may be hypothesized that different reasons may underlie the worsening of symptoms among these two groups. The clients with adjustment disorders (who may be presumed to have better premorbid adjustment than the clients with mood disorders or anxiety disorders) may be sensitive to the level of stress in their environment, or have expectations that are not met in psychotherapy. The clients with mood disorders may be experiencing a breaking down of their previous defense patterns, allowing their poorer coping ability or maladaptive patterns of functioning to become more evident. The whole area of clients' expectations of therapy will be addressed further in the discussion of suggestions for further research.

Dose-effect and the need for a prescription of psychotherapy

Using the definition of a "dose" of therapy given by Howard, et al. (1986), the five-session course of psychotherapy used in the present study appears to meet the criteria for effectiveness. Over fifty percent of the clients showed significant improvement in raw GSI scores after treatment. However, using the criteria given by Derogaitis in the BSI manual (Derogaitis & Spencer, 1982), only about 19 percent of the sample moved from needing treatment to not needing treatment after the five sessions (see Table 7). These criteria make use of the T-scores on the global severity index and the clinical scales, which means that individuals are being compared with normative groups rather than their own sample group.

Future research in the area of psychotherapy outcome will need to specify which criterion group (i.e., clinic samples, national norms, or something in between) will be used to measure improvement, and the reasons which would justify such choices. Certainly, if one were to assert that five sessions of therapy constitutes a complete course of therapy (rather than simply a measurably effective dose of therapy), then using the T-score approach suggested in the test manual would make more sense. Using the present results, it appears that less than one-fifth of the sample received a "complete" course of therapy (as defined by movement outside the clinical range on the instrument).

However, it should also be noted that 14 per cent of the sample began treatment with less than clinical-level scores on the BSI. These clients (who did not need treatment at the outset, by the BSI definition) presumably entered treatment and stayed in treatment for the five sessions because they felt a need to do so. Obviously, other factors that self-reported distress figure into a client's decision to enter therapy, and to remain in treatment for even a period of five sessions. In the emerging world of managed mental health care, instruments such as the BSI may be used to make decisions about eligibility (or authorization) for beginning or continuing treatment. Using such a measure alone might exclude individuals who would meet other criteria for treatment, such as impairment in functioning, interpersonal difficulties, or symptom patterns. The use of clinician ratings (even one as simple as the Global Assessment of Functioning from the DSM-IV) would help to balance the judgments that might be used in the "rationing" of mental health services. Therefore, client perception of the need for treatment, as well as more objective measures of disturbance, must be considered along with self-report scores such as the BSI, in any policy decisions that might make use of studies of treatment effectiveness.

Utility of the BSI symptom dimension scales

Tables 8 and 9 address the question raised in the BSI literature regarding the utility of the symptom dimension (clinical scale) scores. These tables compare each clinical symptom scale score with the GSI score, both at the pretest and in terms of the difference between pretest and posttest scores. These results strongly support those reported by Boulet and Boss (1991) and discussed above, in which the symptom dimension scale scores are all highly correlated with the global scores on the BSI. These results support the contention that the individual symptom scales should not be used in differential diagnosis. They may, however, provide useful clinical information concerning the focus of a client's complaints. As self-report items, the contents of each scale may be used much as the content scales of other personality inventories are used. They provide insights into an individual's subjective experience of symptoms, and the discussion of specific items may be useful in treatment planning with clients. Further, changes in the clinical scales after treatment may provide a measure of the clients' self-reported progress with respect to symptoms targeted by the therapist and client in psychotherapy.

Addressing this question further, correlations were performed between the pretest and posttest differences on the symptom dimension scales, and pretest and posttest GSI score differences (Table 9). There was more variability among the correlations when differences were examined, than when the raw (pretest) scores were examined. However, the correlations with GSI differences were lower, in general, than the correlations between pretest scale scores and GSI scores. Anxiety, Depression, Psychoticism, and Interpersonal Sensitivity were the most closely correlated with GSI both in terms of pretest scores and in terms of changes in scores from pretest to posttest. Examining the item content of these scales, one can see that they contain items (such as those dealing with worry, pessimism, and the opinions of others) that may be expected to "pull" highly for general distress in an outpatient population. The Phobic Anxiety and Paranoid Ideation scales, on the other hand, seem to be more independent of general distress. They deal with more narrowly described symptoms, and may represent better tools for differential diagnosis. The responses of this client sample bear out the impression that the Phobic Anxiety scale, in particular, is elevated in clients with diagnosed anxiety disorders, more so than in clients with other disorders.

One clinical symptom dimension scale which may merit interest in the present study (and attention in future research) is the BSI's Hostility scale, since this type of symptom has implications for the therapeutic alliance as well as for adjustment and functioning in life tasks. In this sample, twenty-eight clients (22%) showed increases in Hostility scores after five therapy sessions, a percentage which was roughly consistent across diagnostic groups. Those clients in the anxiety disorders and adjustment disorders groups who reported increased hostility showed a much smaller degree of increase than the other two groups, even though about the same percentage of clients in all the groups reported increases in hostility symptoms.

It is difficult to assess meaning for the increases in Hostility scale scores among some clients.This is especially true of the heterogeneous "Other/V-codes" group. It is possible that at least some of those individuals were mandated into treatment, or were involved in relationship conflicts which treatment may have exacerbated in the early sessions.

The increases in hostile feelings among the Mood Disorders group may be easier to understand. Clinicians with a psychodynamic orientation might suggest that this tendency is connected with the issue of anger and hostility in depression. Adlerian therapists, for example, describe some depressed patients as highly critical of themselves and of others. The early stages of therapy may encourage the expression of these feelings that may have previously been kept hidden. Increases in experienced hostile feelings may also represent loosening of defensive structures and strategies, as described above. This may allow previously felt hostility to be expressed more directly.

Suggestions for further research

Client response to brief therapy

The present research affirms the effectiveness of the brief treatment methods used by the staff at Family Service Centers, and does not support the hypothesis that any of the client characteristics studied account for differential levels of improvement. However, there was sufficient variability in the response patterns of the individuals to raise some intriguing possibilities for further research.

Clients' expectations for therapy may have an impact on their experience of the first few sessions of psychotherapy. This may be especially important with respect to the reasons for their referrals. Court-mandated clients, and those seeking improvement in relationships, have a different set of expectations and require different approaches than clients who are seeking relief from distressing symptoms. Therefore, categorizing clients according to their reason for referral, referral source, or even degree of voluntariness, may yield useful differences in self-reported symptoms.

Information concerning the clients' and therapists' rating of progress in therapy may be collected after the fifth session of therapy; this information may then be used to compare the individual's overall rating of the effectiveness of therapy with his or her BSI scores. Likewise, the therapist's rating of improvement may be compared with the symptom reports provided by the BSI. This would be helpful in connecting the present research with earlier research that made use of such unidimensional self-ratings of treatment effectiveness. This would also provide additional validation of he BSI itself, with this type of outpatient population.

Possibilities for grounded theory research

Qualitative research, or grounded theory (Strauss & Corbin, 1994) represents a relatively new area of investigation in the field of psychology. This approach makes use of information collected through structured interviews with the persons involved in a process (such as brief therapy), in order to generate themes for further exploration or investigation. Rather than providing hypotheses to be tested empirically, this method focuses on generating ideas that are consistent, sound and useful in practice. The expectations of clients may constitute one area for such qualitative research. The expectations, comfort level and preparation of therapists for brief treatment may constitute another area for such qualitative research .This would follow from some qualitative research (for example, Frontman & Kunkel, 1994) which has already been undertaken. Certain core ideas (such as clients' behavior indicating progress, and therapists' experience of fulfilling their role) have been derived from interviews conducted by these authors, and linked to the therapists' construal of "success." Similar research with clients has focused on their perception of the therapist's empathy (Bachelor, 1988). Comparing qualitatively obtained perceptions, expectations, and satisfaction with the type of objectively obtained measures used in the present study would be especially fruitful. This kind of comparison and integration might help in understanding some of the results discussed above (such as those involving clients reporting increases in hostility after treatment). Further research could also tie in client's expectations for treatment, ratings of their progress, and self-reported symptom levels with some measure of therapist behavior or attitudes (including counter-transference issues). Although the present study did not include sufficient data to assess the effect of Axis II disorders (or traits) on therapeutic progress, it may be that clients with these Axis II features would show a more variable response to treatment. They may also be the clients who experience more hostility after treatment (as well as evoking more hostility in those around them).

Further research on the BSI

Additional research could be undertaken with any of the symptom dimension scales of the BSI. The Hostility scale has already been discussed above; the same type of analysis may be applied to any or all of the other symptom dimensions. Phobic Anxiety, Somatization, and Paranoia all appear to be scales that may be more sensitive to diagnosis than the reemaining scales, due to their slightly lower correlations (in this study) with the Global Severity Index. Their item content also appears to be more symptom-specific and less descriptive of general distress than are the items on scales such as Anxiety, Depression, and Interpersonal Sensitivity. The last three are scales which reflect general perceptions of many people when under stress (and, indeed, were elevated in the present sample's Adjustment Disorders group).

Less obvious is the explanation for the elevated Psychoticism scale scores among all the diagnostic groups in this study. Like the items in some of the scales of the MMPI, the BSI's psychoticism items may reflect clients' perceptions of their coping abilities breaking down under stress, rather than actual psychotic symptoms.

A further type of research that may be fruitful, and which has not been addressed in the literature available for the current review, is any type of item analysis of the BSI. This could be undertaken in order to get an idea of the items which are more easily answered (i.e., answered more frequently by subjects who are not showing clinical symptoms or seeking relief), and which items are more rarely answered. Alternative analyses of the test in order to try to refine its discriminating ability might be fruitful; these might include item weightings or other scoring systems that attempt to control for the effect of some items being more sensitive to psychopathology than other items. In the standard BSI scoring, all items are weighted equally acording to the reported frequency/severity reported by the client. Certain items might be more telling simply by their presence or absence, rather than by their inclusion in the rating system as applied currently.


American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author.

Ansbacher, H. (1972). Adlerian psychology: The tradition of brief psychotherapy. Journal of Individual Psychology, 28, 137-151.

Bachelor, A. (1988). How clients perceive counselor empathy: A content analysis of "received" empathy. Psychotherapy, 25, 227-240.

Barkham, M., Rees, A., Stiles, W. B., Shapiro, D. A., Hardy, G. E., & Reynolds, S. (1996). Dose-effect relations in time-limited therapy for depression. Journal of Consulting and Clinical Psychology, 64(5), 927-935.

Bergin, A. E., & Lambert, M. J. (1978). The evaluation of therapeutic outcomes. In S. L. Garfield & A. E. Bergin (Eds.), Handbook of psychotherapy and behavior change: An Empirical Analysis (2nd ed.). New York: John Wiley.

Bloom, B. L. (1992). Planned short-term psychotherapy: A Clinical Handbook. Boston: Allyn & Bacon.

Bonjean, M. J., & Spector, R. (1994). Intermittent Brief Therapy (Unpublished manuscript). Milwaukee, WI: Milwaukee Brief Therapy Center.

Boulet, J., & Boss, M. W. (1991). Reliability and validity of the Brief Symptom Inventory. Psychological Assessment, 3(3), 433-437.

Budman, S. H., & Gurman, A. S. (1988). Theory and practice of brief therapy. New York: Guilford Press.

Burlingame, G. M., & Fuhriman, A. (1987). Conceptualizing short-term treatment: A comparative review. Counseling Psychologist, 15(4), 557-595.

Burlingame, G. M., Fuhriman, A., Paul, S., & Ogles, B. M. (1989). Implementing a time-limited therapy program: Differential effects of training and experience. Psychotherapy, 26(3), 303-313.

Conoley, J. C., & Kramer, J. J. (1989) The tenth mental measurements yearbook. Lincoln, NE: University of Nebraska Press.

Derogaitis, L. R., & Spencer, M.S. (1982). The Brief Symptom Inventory (BSI): Administration, scoring, and procedures manual-1. Baltimore: Johns Hopkins School of Medicine, Clinical Psychometrics Research Unit.

DeShazer, S., Berg, I. K., Lipchik, E., Nunnally, E., Molnar, A., Gingerich, W., & Weiner-Davis, M. (1986). Brief therapy: Focused solution development. Family Process, 25(2), 207-222.

Dinkmeyer, D., & Sherman, R. (1989). Brief Adlerian family therapy. Journal of Individual Psychology, 45, 148-158.

Duncan, B. L., Solovey, A. D., & Rusk, G. R. (1992). Changing the rules: A client-directed approach to therapy. New York: Guilford Press.

Erickson, K. K. (1988). One method for designing short-term intervention-oriented Ericksonian therapy. In Zeig, J. K., & Lankton, S. R. (Eds.), Developing Ericksonian therapy: State of the art. New York: Brunner/Mazel.

Family Service Centers (1996). Employee Handbook: Policy and Procedure Manual (1996 revision). Flossmoor, IL: Author.

Frontman, K. C., & Kunkel, M. A. (1994). A grounded theory of counselors' construal of success in the initial session. Journal of Counseling Psychology, 41(4), 429-499.

Haley, J. (1973). Uncommon therapy: The psychiatric techniques of Milton H. Erickson, M. D. New York: W. W. Norton.

Hayes, A. M., Castonguay, L. G., & Goldfried, M. R. (1996). The study of change in psychotherapy: A reexamination of the process-outcome paradigm. Comment on Stiles and Shapiro (1994). Journal of Consulting and Clinical Psychology, 64(5), 909-914.

Howard, K. I., Kopta, S. M., Krause, M. S., & Orlinsky, D. E. (1986). The dose-effect relationship in psychotherapy. American Psychologist, 41(2), 159-164.

Howard, K. I., Moras, K., Brill, P. L., Martinovich, Z., & Lutz, W. (1996). Evaluation of psychotherapy: Efficacy, effectiveness, and patient progress. American Psychologist, 51(10), 1059-1064.

Miller, I. J. (1996a). Some "short-term therapy values" are a formula for invisible rationing. Professional Psychology: Research and Practice, 27(6), 577-582.

Miller, I. J. (1996b). Time-limited brief therapy has gone too far: The result is invisible rationing. Professional Psychology: Research and Practice, 27(6), 567-576.

Schefler, G., Dasberg, H., & Ben-Shakhar, G. (1996). A randomized controlled outcome and follow-up study of Mann's time-limited psychotherapy. Journal of Consulting and Clinical Psychology, 63(4), 585-593.

Piper, W. E., Boroto, D. R., Joyce, A. S., McCallum, M., & Azim, H. F. (1995). Pattern of alliance and outcome in short-term individual psychotherapy. Psychotherapy, 32(4), 639-647.

Sechrest, L. (1994). Recipes for psychotherapy. Journal of Consulting and Clinical Psychology, 62(5), 952-954.

Sechrest, L., McKnight, P., & McKnight, K. (1996). Calibration of measures for psychotherapy outcome studies. American Psychologist, 51(10), 1065-1071.

Shlien, J. M., Mosak, H. H., & Dreikurs, R. (1962). Effect of time limits: A comparison of two psychotherapies. Journal of Counseling Psychology, 9(1), 31-34.

Sifneos, P. E. (1992). Short-term anxiety-provoking psychotherapy: A treatment manual. New York: Basic Books.

Silberschatz, G. (1994). Abuse and disabuse of the drug metaphor in psychotherapy research: Hold on to the baby as you throw out the bath. Journal of Consulting and Clinical Psychology, 62(5), 949-951.

Sledge, W. M., Moras, K., Hartley, D., & Levine, M. (1990). Effect of time-limited psychotherapy on patient dropout rates. American Journal of Psychiatry, 147(10), 1341-1347.

Stiles, W. B. (1994). Drugs, recipes, babies, bathwater, and psychotherapy process-outcome relations. Journal of Consulting and Clinical Psychology, 62(5), 955-959.

Stiles, W. B. (1996). When more of a good thing is better: Reply to Hayes et al. (1996). Journal of Consulting and Clinical Psychology, 64(5), 915-918.

Stiles, W. B., & Shapiro, D. A. (1994). Disabuse of the drug metaphor: Psychotherapy process-outcome correlations. Journal of Consulting and Clinical Psychology, 62(5), 942-948.

Straker, M. (1968). Brief psychotherapy in an outpatient clinic: Evolution and evaluation. American Journal of Psychiatry, 124(9), 1219-1226.

Strauss, A., & Corbin, J. (1994). Grounded theory methodology. In N. K. Denzin & Y. S. Lincoln (Eds.), Handbook of qualitative research. Thousand Oaks, CA: Sage.

Strupp, H. H. (1996). The tripartite model and the Consumer Reports study. American Psychologist, 51(10), 1017-1024.

Strupp, H. H., & Hadley, S. W. (1977). A tripartite model of mental health and therapeutic outcomes. American Psychologist, 32(2), 187-196.

Talmon, M. (1990). Single-session therapy: Maximizing the effect of the first (and often only) therapeutic encounter. San Francisco: Jossey-Bass.

Walter, J. L., & Peller, J. E. (1992). Becoming solution-focused in brief therapy. New York: Brunner/Mazel.

Appendix A

Table I - Demographic characteristics of sample, by diagnostic category
Male Female Total



Avg. Age


% Avg. Age N % Avg. Age
Anxiety 6 38 33.3 10 62 38.3 16 12 36.4
Mood 15 26 38.1 42 74 35.7 57 44 36.4
Adjust 14 38 33.6 23 62 33.9 37 29 33.8
Other 12 63 33.5 7 37 36.7 19 15 34.7
Total 47 36 34.6 82 64 36.2 129 100 35.3

Table II - Number of clients reporting previous treatment, by diagnostic category
Male Female Total
N %* N %* N %*
Anxiety 4 66 3 30 7 44
Mood 5 33 13 31 18 32
Adjust 3 21 6 26 9 24
Other 3 25 4 57 7 37
Total 15 32 26 32 41 32

*Percent of number of clients (male, female, and total) in each diagnostic group and total sample

Table 1 -- Descriptive Statistics - Raw Global Severity Index Scores - Pretest (GSI1) and Posttest (GSI2) by Gender, Prior Treatment, and Diagnosis
Grp. Dx N Mean S.D. Mean S.D.
Male - (No Tx)
Anxiety 2 .660 n/a .430 .212 -0.23
Mood 10 1.278 .679 .839 .612 -0.44
Adjustmt 10 .854 .492 .632 .280 -0.22
Other 9 .769 .651 .528 .494 -0.24
Male - (Treatment)
Anxiety 4 1.785 .623 1.250 .466 -0.54
Mood 5 1.994 1.052 1.776 1.218 -0.22
Adjustmt 3 .993 .076 .660 .147 -0.33
Other 3 .467 .452 .247 .221 -0.22
Female - (No Tx)
Anxiety 7 1.407 1.040 1.271 1.269 -0.14
Mood 28 1.398 .772 1.124 .871 -0.27
Adjustmt 12 1.309 .942 .699 .792 -0.61
Other 3 .970 .812 .540 .338 -0.43
Female - (Treatment)
Anxiety 3 .750 .151 .680 .161 -0.07
Mood 15 1.562 .764 1.305 .904 -0.26
Adjustmt 6 1.122 .943 .505 .564 -0.62
Other 4 .840 .623 .813 .940 -0.03
Total * 1.245 .794 .929 .810 -0.32

*N=124 (5 cases rejected because of missing data)

Figure 1 - Clinical Symptom Dimension Scale Scores (Anxiety Disorders Group)

Figure 2 - Clinical symptom dimension scale scores (Mood disorders group)

Figure 3 - Clinical symptom dimension scale scores (Adjustment disorders group)

Figure 4 - Clinical symptom dimension scale scores (Other disorders/V-Codes group)

Table 2 -- Multivariate Analysis of Variance (Within subjects) for GSI scores, pretest and posttest (Effect size measured by Eta squared)
Source of Variation SS DF MS F Sig of F Partial Eta2
Within Cells 23.69 108 .22 -- -- --
GSI 3.53 1 3.53 16.11 <.001 .130
Gender by GSI .00 1 .00 .00 N.S. .000
Prior Treatment by GSI .01 1 .01 .06 N.S. .001
Diagnosis by GSI .30 3 .10 .45 N.S. .012
Gender by Prior Treatment by GSI .06 1 .06 .29 N.S. .003
Gender by Diagnosis by GSI .47 3 .16 .72 N.S. .020
Prior Treatment by Diagnosis by GSI .16 3 .05 .24 N.S. .007
Gender by Prior Treatment by Diagnosis by GSI .18 3 .06 .27 N.S. .007

Table 3 -- Multivariate Analysis of Variance (Between subjects) for GSI scores (Effect size measured by Eta squared)
Source of Variation SS DF MS F Sig of F Partial Eta2
Within Cells 105.35 108 .98 -- -- --
Gender .19 1 .19 .20 N.S. .002
Prior Treatment .62 1 .62 .64 N.S. .006
Diagnosis 17.10 3 5.70 5.84 .001 .140
Gender by Prior Treatment 2.80 1 2.80 2.87 N.S. .026
Gender by Diagnosis 1.10 3 .37 .38 N.S. .010
Prior Treatment by Diagnosis 3.29 3 1.10 1.12 N.S. .030
Gender by Prior Treatment by Diagnosis 3.67 3 1.22 1.26 N.S. .034

Table 4 -- Single-Factor Analyses of Variance for GSI Scores among Diagnostic Groups
GSI1 (pre-test)
Source df SS MS F F crit
Between Groups 3 8.38624 2.79541 4.78293** 2.6771
Within Groups 125 73.057 0.58446
Total 128
GSI2 (post-test)
Source df SS MS F F crit
Between Groups 3 7.31083 2.43694 4.42362** 2.6771
Within Groups 125 68.8617 0.55089
Total 128


Table 5 -- Summary of Post-hoc multiple t-test Comparisons by Diagnostic Group

(showing only p values <.01; two-tailed)

GSI1 (pretest) Anxiety Mood Adjustment
Anxiety --- --- ---
Mood N.S. --- ---
Adjustment N.S. N.S. ---
Other N.S. 0.00017 N.S.
GSI2 (posttest) Anxiety Mood Adjustment
Anxiety --- --- ---
Mood N.S. --- ---
Adjustment N.S. 0.00387 ---
Other N.S. 0.00111 N.S.
GSI1 and GSI2 Combined Anxiety Mood Adjustment
Anxiety --- --- ---
Mood N.S. --- ---
Adjustment N.S. 0.00050 ---
Other 0.00624 <0.00001 N.S.

Table 6 - Percentages of clients improved, unchanged, and worse in each diagnostic group (95% confidence interval, based on sample variance)

Improved Unchanged Worse Total
N % N % N % N
Anxiety 10 63% 5 31% 1 6% 16
Mood 35 62% 15 26% 7 12% 57
Adjustmt 24 65% 8 22% 5 13% 37
Other 11 58% 7 37% 1 5% 19
Total 79 62% 35 27% 14 11% 129

Table 7 -- Individuals meeting BSI criteria for treatment (pretest and posttest)
Pretest (N) Pretest (%) Posttest (N) Posttest (%)
GSI of 63 or above 95 74% 67 52%
Two clinical scales at 63 or above 16 12% 20 15%
Total 111 86% 87 67%
Change (pretest-posttest) --- --- -24 -19%

Table 8 -- Correlations of raw clinical scale scores with raw GSI score (pretest)
BSI clinical symptom scale Correlation with GSI Percent shared variance
Somatization 0.71 50%
Obsessive-compulsive 0.84 71%
Interpersonal Sensitivity 0.84 71%
Depression 0.87 76%
Anxiety 0.86 74%
Hostility 0.76 58%
Phobia 0.76 58%
Paranoia 0.76 58%
Psychoticism 0.88 77%

Table 9 -- Correlations of clinical scale score differences (pretest to posttest) with GSI score difference (pretest to posttest) - total sample
BSI clinical symptom scale Correlation with GSI difference Percent shared variance
Somatization 0.65 42%
Obsessive-compulsive 0.78 61%
Interpersonal Sensitivity 0.76 58%
Depression 0.84 71%
Anxiety 0.84 71%
Hostility 0.72 52%
Phobia 0.61 37%
Paranoia 0.63 40%
Psychoticism 0.78 61%

Appendix B

Symptom Dimensions Measured by the Brief Symptom Inventory


Faintness or dizziness

Nausea or upset stomach

Trouble getting one's breath

Hot or cold spells

Numbness or tingling in parts of one's body

Weakness in parts of one's body


Trouble remembering things

Feeling blocked in getting things done

Having to check things one does repeatedly

Difficulty making decisions

Mind going blank

Trouble concentrating

Interpersonal Sensitivity:

Feelings being easily hurt

Viewing others are unfriendly

Feeling inferior to others

Feeling self-conscious with others


Thoughts of ending one's life

Feeling lonely

Having no interest in things

Hopelessness about the future

Feeling worthless


Nervousness or shakiness

Being afraid for no reason

Feeling fearful

Feeling tense or keyed up

Spells of terror or panic

Inability to sit still


Feeling easily annoyed or irritated

Uncontrollable temper outbursts

Urges to beat or hurt someone

Having urges to smash things

Frequent arguments

Phobic Anxiety:

Being afraid in the open or on the street

Being afraid to travel on buses or trains

Feeling that one must avoid certain places or activities due to fear

Feeling uneasy in crowds or public places

Feeling nervous when alone

Paranoid Ideation:

Feeling others are to blame for one's troubles

Feeling that others cannot be trusted

Feeling that one is watched or talked about by others

Feeling that one does not get credit from others for accomplishments

Believing that others will take advantage


Believing that others can control one's thoughts

Believing that one should be punished

Not feeling close to others

Feeling lonely even when with others

Believing that something is wrong with one's mind