Global Assessment of Functioning Following Assertive Community Treatment in Edmonton, Alberta: A Longitudinal Study Philip Tibbo, MD, FRCPC1, Ken Joffe, MSc, MD2, Pierre Chue, MRCPsych3, Andrei Metelitsa4, Evelyn Wright, RN, BscN, MS5
Objective: To examine longitudinally the effects of Assertive Community Treatment (ACT) on Global Assessment of Functioning (GAF) scores in Edmonton, Alberta. Methods: We acquired GAF scores for all clients at initial registration in the ACT program and at subsequent 18- and 36-month time points while in ACT. We analyzed both the entire ACT cohort and separate diagnostic groups. Results: We obtained baseline and follow-up GAF scores for 411 clients, of whom the largest diagnostic group suffered from schizophrenia (n = 189), followed by bipolar disorder (n = 98). Collapsed across all groups, GAF scores significantly improved at both 18 ( P < 0.0001) and 36 months ( P < 0.0001). By group, at 18-month follow-up, significant improvements were seen in patients with delusional disorder ( P < 0.05), dysthymia ( P < 0.05), schizoaffective disorder ( P < 0.05), and schizophrenia ( P < 0.001). This was also seen at 36-month follow-up, with the addition of significant improvements in those with bipolar disorder ( P < 0.05). Those patients with major affective disorder or psychosis not otherwise specified (NOS) did not show significant improvements over time. Regardless of diagnosis, those clients with baseline GAF scores of ≤ 40 significantly improved at both 18-month ( P < 0.0001) and 36-month ( P < 0.0001) follow-up, while those with baseline GAF scores above 40 did not show significant improvement. Conclusions: GAF scores improved at 18- and 36-month follow-up from enrolment in an ACT program. Groups with different diagnoses and levels of functioning at time of enrolment may not benefit to the same degree. (Can J Psychiatry 2001;46:144–148) Key Words: assertive community treatment, global assessment of functioning, chronic mentally ill
A
ssertive community treatment (ACT) is a model of psychiatric community treatment for individuals with chronic mental illness. This model is intensive, yet comprehensive and flexible where it is accepted that clients have a right to live in as normal an environment as possible. This normal environment is, however, attained with intensive supports and services tailored to clients’ needs. The aims of ACT are to prevent unnecessary and lengthy hospital stays and to improve the length and quality of the client’s tenure in the community; it endeavours to deal with problems arising in the community that may otherwise lead to a hospital admission. Therapists assist clients to connect with their community and
to make positive use of resources available to meet their housing, employment, health, and leisure needs. This model (formally known as “Training in Community Living” or “Program of Assertive Community Treatment”) was originally developed in Madison, Wisconson, in the 1970s (1,2) and has since been adapted for other urban and rural North American settings (3–5). As part of the Community Living Program (CliP), ACT was initiated in Edmonton, Alberta, on April 1, 1993. It has grown to a program employing 19 psychiatrists, 24.8 full-time equivalent (FTE) community nurses, and 2.5 FTE community support workers. The main program elements of Edmonton’s ACT model are based on Stein and Test’s original model (1,2), which includes low staff–client ratios, assertive outreach, 24-hour availability, active assertive advocacy and monitoring, continuity between community and hospital care, and flexible, individualized services.
Manuscript received November 1999, revised, and accepted June 2000. 1Clinical Fellow, Alberta Heritage Foundation for Medical Research, Department of Psychiatry, University of Alberta, Edmonton, Alberta. 2Resident in Psychiatry, University of Alberta, Edmonton, Alberta. 3Assistant Clinical Professor, Department of Psychiatry, University of Alberta; Clinical Coordinating Psychiatrist, Community Living Program, Edmonton, Alberta. 4Undergraduate Student, University of Alberta, Edmonton, Alberta. 5Program Manager, Community Living Program, Edmonton, Alberta. Address for correspondence: Dr P Tibbo, Department of Psychiatry, University of Alberta Hospital, 8440 112 Street, Edmonton, AB T6G 2B7 E-mail:
[email protected]
Can J Psychiatry, Vol 46, March 2001
Although there is substantial research on the effects of ACT on several outcome variables (for example, hospitalization rates and emergency-room visits) in the US (reviewed in 6,7), there is little research on this model in Canada, despite its implementation in several Canadian settings (8–11). We recently reported on hospital outcome measures for a cohort of 144
March 2001
Global Assessment of Functioning Following Assertive Community Treatment in Edmonton, Alberta
145
Table 1. Number of subjects, mean age and global assessment of functioning (GAF) scores for each diagnostic group Group
n
Age (years)
GAF baseline
GAF 1.5 years b
GAF 3 years d
GAF averagea
Schizophrenia
189
37.8
51.4
54.9
57.9
55.9b
Bipolar disorder
98
45.0
56.1
59.8
64.2c
60.5
Major affective disorder
46
39.4
60.0
62.8
60.9
62.8
c
c
58.8
57.3b
Schizoaffective disorder
41
39.9
49.1
54.6
Dysthymia
12
40.2
49.3
57.9c
61.0c
58.3b
Psychosis not otherwise specified
10
40.7
50.5
60.3
59.5
58.7
55.3
c
c
63.6
63.0
Delusional disorder
10
39.8
62.6
Adjustment disorder
4
46.8
52.5
57.5
61.0
59.0
Borderline personality disorder
1
26.0
50.0
58.3
—
58.3
295 individuals, following 1 year of ACT in Edmonton (12). Compared with the year prior to their enrolment in ACT, hospital separations, average length of stays when admitted, and emergency-room visits were reduced. Nonetheless, while service use reductions are an extremely important outcome in evaluating an ACT program, an individual’s level of functioning is also important. One might assume that if service use is reduced the group is functioning well enough not to need these services. The possibility exists, however, that these individuals do not change in functional status but that, with the intensive supports in the community, their reliance on traditional services is decreased. This paper examines this possibility by evaluating the long-term effects of ACT on the global functioning of individuals enrolled in an ACT program in Edmonton, Alberta. Methods Clients can be referred to the ACT program from the 4 general hospitals in Edmonton, from the provincial psychiatric hospital, from provincial mental health services, and from the offices of general practitioners. Individuals are eligible for registration if they are between the ages of 18 and 65 years, experience persistent and severe mental illness, and fulfill at least 3 of the following criteria: a history of multiple admissions to a psychiatric facility, difficulty functioning in the community, noncompliance with medication therapy, inability or unwillingness to receive follow-up services in an office setting, and a high risk for returning to hospital without assertive outreach. Once the referral has been accepted, the client is assigned a primary therapist who provides services tailored to individual client needs and makes most of the client contacts in a “natural” environment (that is, not in an office or clinic). The psychiatrist also maintains frequent direct contacts that include domicilary visits, crisis response, clinical supervision of treatment planning, and consultation. On registration to the ACT program, clients are further interviewed by the assigned ACT psychiatrist for DSM-IV primary and comorbid diagnosis, as well as for complete
demographic data. The Global Assessment of Functioning (GAF) scale is recorded for each client at intake into the program and reviewed at intervals of 4 to 6 months The GAF is completed with input from all members of the community treatment team, including the psychiatrist, and recorded in the Comprehensive Client Review (CCR). Thus, the same treatment team completes the initial and follow-up GAFs for each client at the designated time periods. The CCR is completed every 6 months and placed in the client’s file. For the purpose of this study, GAFs were analyzed at baseline, 18 months, and 36 months after enrolment in the ACT program. Although every effort is made to acquire GAFs on a regular basis, this does not always occur: clients may be in hospital or in jail or, due to their itinerant nature, the treatment team may have difficulty meeting with certain clients at the specific follow-up time. As a result, there was a discrepancy in sample size among groups, and examination of the data revealed that assumptions of normalcy and homogeneity were not met, precluding the use of analyses of variance (ANOVAs) as a statistical measure. For these reasons, Kruskal-Wallis 1-way analyses were used to detect general differences among groups, and Mann-Whitney follow-up was used to test for spe cific group com pari sons. Ad di tion ally, Wil coxon matched-pairs signed-ranks tests were used to detect withingroup differences over time. Results Baseline and follow-up GAF scores spanning a period of 36 months were obtained for 411 patients whose initial assessment and enrolment in ACT was from 1994 to 1996 (which allowed capture of data from 1994 to 1999). Although 497 subjects were initially eligible, 52 (10.5%) did not meet the primary diagnosis we wished to capture (for example, primary substance abuse, or not otherwise specified [NOS] diagnosis other than for psychosis), and 34 (6.8%) were lost to follow-up. Mean age across all diagnostic groups was 40.1 years (range = 18 to 64 years), with approximately equal numbers of male (n = 204) and female (n = 207) patients. The
146
The Canadian Journal of Psychiatry
70 60 50 40
GAF < 40 GAF > 40
30 20 10 0 GAF at Baseline
GAF at 1.5 yrs
GAF at 3 yrs Average GAF at follow -up
Vol 46, No 2
follow- up, sig nifi cant im prove ments were seen in patients with delusional disorder (P < 0.05), dysthymia (P < 0.05), schizoaffective disorder (P < 0.05), and schizophrenia (P < 0.001). This pattern was again evident when comparing baseline GAF with scores obtained at 36 months’ follow-up, with the addition of significant improvements in those with bipolar disorder (P < 0.05). Those patients with MAD or psychosis NOS did not show significant improvements over time (P > 0.05). While patients as a whole showed continued improvement at 36 months, compared with 18 months, only in those with schizophrenia did this reach significance (P < 0.0001).
Finally, we were interested in whether improvement in GAF score was related to severity of illness at baseline. These Figure 1. GAF scores of all ACT clients, sub-divided by severity of GAF (< or > than 40), followed results are presented in Figure 1. Those up at 1.5 and 3 years as well as compared to averaged follow-up scores. with baseline GAF scores of ≤ 40 (n = 107), regardless of diagnosis, signifimean age and GAF scores across all time points for each diagcantly improved when compared with their averaged follownostic group of patients are presented in Table 1. While data up score (31.8 vs 51.3, P < 0.0001). This was also evident at from all diagnostic groups were included in the analyses that both 18 months’ (31.8 vs 49.9, P < 0.0001) and 36 months’ collapsed across groups, within-group differences over time (31.8 vs 54.4, P < 0.0001) follow-up. Those with baseline were not analyzed for those with borderline personality or adGAF scores above 40 (n = 304) did not show significant imjustment disorder, due to limited sample size. provement (P > 0.05) when compared with averaged followup score (60.9 vs 60.9) and with GAF scores at 18 months’ At baseline, a significant difference in GAF scores among (60.9 vs 60.1) and 36 months’ (60.9 vs 62.4) follow-up. groups was present (P < 0.01). Patients with bipolar or major affective disorder (MAD) had significantly higher scores, Discussion compared with those patients with schizoaffective disorder or schizophrenia (P < 0.01). Additionally, patients with MAD To date, this is the largest longitudinal study examining the had significantly higher scores than those with dysthymia level of functioning of individuals enrolled in ACT in Can(P < 0.05). ada. Overall, we are able to show that ACT appears to improve GAF scores in a group of clients with chronic mental Collapsed across all diagnostic groups, GAF scores signifiillness over a 3-year time span. Interestingly, there appears to cantly increased from 53.3 at baseline to an averaged followbe a differential of this effect that is based on the primary diup score of 58.3 (P < 0.0001). Significant improvements were agnosis at time of enrolment: individuals with schizophrenia noted in those with dysthymia (P < 0.01), schizoaffective disappear to increase their level of functioning to a greater deorder (P < 0.001), and schizophrenia (P < 0.0001). Signifigree and for a consistently longer period of time. At baseline, cant trends were also seen in those with psychotic disorder however, this group had the lowest GAF scores when comNOS (P = 0.09), delusional disorder (P = 0.09), depression (P pared with other diagnostic groups, and perhaps it could = 0.09), and bipolar disorder (P = 0.12). therefore be argued that this group had the greatest possibility of improvement with respect to GAF scores. Similarly, the To assess these improvements further, we compared baseline cohort of individuals with MAD had the highest GAFs at GAF scores with those obtained at both 18- and 36-months’ baseline but did not appear to benefit from ACT to the same follow-up. Again, collapsed across all groups, GAF scores degree. This may imply possible ceiling effects for ACT on significantly improved at both 18 (53.5 to 57.4, P < 0.0001) GAF scores. A further examination of the MAD cohort reand 36 months (53.5 to 60.3, P < 0.0001). Additionally, overvealed that 30% had a secondary diagnosis of borderline perall scores at 36 months were significantly higher than those sonality disorder not present in the other diagnostic groups, obtained at 18 months (P < 0.001). By group, at 18 months’ which may also have influenced the above results.
March 2001
Global Assessment of Functioning Following Assertive Community Treatment in Edmonton, Alberta
In demonstrating the effectiveness of ACT, functional outcome issues are notoriously difficult to implement and evaluate (6,7,13). No doubt, ACT has a much stronger experimental effect on hospital outcome measures than on functional outcomes. It has been argued that this discrepancy results from the limited statistical power of many studies, from problems in measuring functional outcomes, and even from the characteristics of the population studied (14). For many ACT patients who have a history of noncompliance with medications and clinical appointments, the mere fact that they attend for a regular depot injection, attend a clinic appointment or a day program, or develop a rapport with a treatment team represents the single most significant and perhaps the only major change to their health care. Some might not consider the follow-up GAF scores reported in this study to be “functional.” Moreover, even though the mean GAF scores over time may be statistically significantly different, is the recorded improvement clinically significant? The difficulty in achieving higher scores may be due to the limited gains that can be achieved by patients with severe chronic mental illness. For many patients who meet the criteria for ACT, further reductions of symptoms and continued improvement in functioning may not be attainable goals. To the extent that reducing hospital use is a common goal of ACT, improving functioning is just one of many concerns to the clinician. This study was able to show that improvement in GAF is an appropriate outcome measure, but with this difficult cohort, perhaps “maintenance” of GAF may also be considered a valid outcome measure. McRae and others reported on increased hospital use and treatment costs for clients after 5 years of intensive case management ceased (15), but unfortunately, no studies exist that examine GAF and functional outcomes after ACT. Such studies would contribute to our understanding if we wished to consider improvement or maintenance of GAF reasonable outcome measures. Studies have, however, reported reduction of hospital use as early as 6 months after enrolment in ACT (for example 15,16), whereas symptom improvement may require a much longer time frame (16,17). This study is thus important because it followed a group of patients for a long time frame (3 years), had a simple measure of functional outcome, and was able to show improvement in this variable. The additional evidence that ACT also reduces hospital use over time in an overlapping cohort (12) points to a favourable program effect. This study was carried out on a cohort of individuals with a single model of community care and no control group. To address changes in the control population over the same time period, it would be interesting to compare longitudinally the effects of other rehabilitation models on a similar cohort. The pre–post design is, however, an adequate comparative
147
Clinical Implication • ACT improves GAF in certain diagnostic groups over a 3-year time frame.
Limitation • No other cohort in another rehabilitation model was used as a control group.
measure for this study and complements the pre-post study design of a very similar cohort of individuals in whom we examined hospital outcome measures. In addition, we were not able to do an interrater reliability study because the GAF scores were given by consensus of the whole treatment team. The same team is, however, assigned to a client for the entire time he or she is in ACT, which allows for consistent GAF scores over time for each client. In conclusion, enrolment in ACT appears to do more than reduce hospital outcome measures in patients with chronic mental illness: a reasonable exposure to the ACT model appeaars to improve their overall level of functioning. References 1. Stein LI, Test MA, Marks A. Alternative to the hospital: a controlled study. Am J Psychiatry 1975;132:417–22. 2. Stein LI, Test MA. Alternative to mental hospital treatment: I. conceptual model, treatment program and clinical evaluation. Arch Gen Psychiatry 1980;37:392–7. 3. Hoult J, Reynolds I, Charbonneau-Powis M, and others. Psychiatric hospital versus community treatment: the results of a randomized trial. Aust N Z J Psychiatry 1983;17:160 –7. 4. Bond GR, Miller LD, Krumwied RD, Ward RS. Assertive case management in three CMHCs: a controlled study. Hosp Comm Psychiatry 1988;39:411–8. 5. Santos AB, Deci PA, Lachance K, and others. Providing assertive community treatment for severely mentally ill patients in a rural area. Hosp Comm Psychiatry 1993;44:34 –9. 6. Scott JE, Dixon LB. Assertive community treatment and case management for schizophrenia. Schizofr Bull 1995;21(4):657–68. 7. Mueser KT, Bond GR, Drake RE, Resnick SG. Models of community care for severe mental illness: a review of research on case management. Schizofr Bull 1998;24(1):37–74. 8. Wasylenki DA, Goering PN, Lemire D, Lindsey S, Lancee W. The hostel outreach program: assertive case management for homeless mentally ill persons. Hosp Community Psychiatry 1993;44:848–53. 9. Lafave HG, deSouza HR, Gerber GJ. Assertive community treatment of severe mental illness: a Canadian experience. Psychiatr Serv 1996;47:757–9. 10. Wilson D, Tien G, Eaves D. Increasing the community tenure of mentally disordered offenders: an assertive case management program. International Journal of Law and Psychiatry 1995;18:61–9. 11. Nelson J, Sadeler C, Cragg SM. Changes in rates of hospitalization and cost savings for psychiatric consumers participating in a case management program. Psychosocial Rehabilitation Journal 1995;18(3):113–23. 12. Tibbo P, Chue P, Wright E. Hospital outcome measures following assertive community treatment in Edmonton, Alberta. Can J Psychiatry 1999;44:276–9. 13. Sands RG, Cnaan RA. Two modes of case management: assessing their impact. Commun Mental Health J 1994;30(5):441–57. 14. Burns BJ, Santos AB. Assertive community treatment: an update of randomized trials. Psychiatr Serv 1995,46;669–75. 15. McRae J, Higgins M, Lycan C, Sherman W. What happens to patients after five years of in ten sive case man age ment stops? Hosp Comm Psy chia try 1990;41(2):175–9. 16. Test MA, Knoedler WH, Allness DJ. The long-term treatment of young schizophrenics in a community support program. New Dir Mental Health Serv 1985;26:17–27. 17. Bond GR, Witheridge TF, Dincin J, Mcrae SA, Mayes J, Ward RS. Assertive community treatment for frequent users of psychiatric hospitals in a large city: a controlled study. Hosp Comm Psychiatry 1988;39:411–8.
148
The Canadian Journal of Psychiatry
Vol 46, No 2
Résumé— Évaluation globale du fonctionnement par suite d’un suivi intensif en équipe dans la communauté, à Edmonton (Alberta) : une étude longitudinale Objectif : Procéder à un examen longitudinal des effets d’un suivi intensif en équipe dans la communauté (ACT) selon les scores de l’Évaluation globale du fonctionnement (EGF) à Edmonton (Alberta). Méthodes : Nous avons obtenu les scores de l’EGF pour tous les clients lors de l’inscription au programme ACT, puis après 18 mois et 36 mois dans le programme. Nous avons analysé la cohorte de l’ACT en entier et les groupes diagnostiques distincts. Résultats : Nous avons obtenu des scores de base et de suivi de l’EGF pour 411 clients, dont le groupe diagnostique le plus important souffrait de schizophrénie (n = 189), suivi du trouble bipolaire (n = 98). Tous groupes confondus, les scores de l’EGF s’amélioraient significativement à 18 mois ( P < 0,0001) et à 36 mois ( P < 0,0001). Par groupe, au suivi de 18 mois, des améliorations significatives ont été constatées chez les patients souffrant de trouble délirant ( P < 0,05), de dysthymie ( P < 0,05), de trouble schizo-affectif ( P < 0,05) et de schizophrénie ( P < 0,001). On a constaté la même chose au suivi de 36 mois, en plus d’améliorations significatives chez les personnes souffrant du trouble bipolaire ( P < 0,05). Les patients souffrant de trouble affectif majeur ou de psychose sans autre indication (SAI) ne montraient pas d’améliorations sensibles avec le temps. Sans égard au diagnostic, les clients dont les scores de base de l’EGF étaient 40 s’amélioraient significativement tant au suivi de 18 mois ( P < 0,0001) qu’à celui de 36 mois ( P < 0,0001), tandis que ceux dont les scores de base de l’EGF étaient supérieurs à 40 ne montraient pas d’amélioration significative. Conclusions : Les scores de l’EGF s’amélioraient aux suivis de 18 et de 36 mois, depuis l’inscription à un programme ACT, bien que les groupes de personnes ayant différents diagnostics et niveaux de fonctionnement au moment de l’inscription ne progressent pas au même degré.
