Crisis
Intervention
A.V. Andreoli, A. Muehlebach,
Response and Long-Term A Pilot Study
M. Gognalons, J. Abensur, S. Grimm, and A.J. Frances
We investigated whether process variables (therapeutic alliance and insight) measured at the termination of crisis intervention predict long-term treatment compliance and 2-year outcome. Thirty-seven consecutive depressed psychiatric patients assigned to outpatient crisis intervention (Ccl) were assessed with both questionnaires and standardized instruments at intake, 1 week, and CCI termination [mean, 6 weeks). Thirty-one subjects (64%) were also evaluated at l-year and 2-year follow-up. We found that working
T
HE PURPOSE of this study was to assess the outcome of crisis intervention (CCI) and to determine whether process measures (i.e., working alliance and development of insight at CC1 termination) predict long-term outcome. Earlier research has indicated that symptom improvement and psychosocial recovery at acute treatment termination may be unrelated to long-term outcome in hospitalized patients assigned to short-term intensive therapy.1-4 The long-term outcome of psychiatric patients referred for intensive treatment with severe DSMIII-R disorders appears to be influenced by the availability of both specialized long-term treatment in the community5 and multimodal crisis intervention.6l7 In addition, process ratings within interactive CC1 have been reported to provide valuable predictors of succesful long-term outcome in psychiatric patients.’ These observations suggest that increased alliance and insight could well be related to compliance to longterm treatment, decreased relapse risk, and enhanced symptom relief. The relationship between therapeutic alli-
From the Depattment of Psychiatry, University of Geneva Medical School, Geneva, Switzerland. Supported in part by the Scientific Research Swiss National Fund, Contract No. 3.822.1.84. Dr. Frances is a ProfessorofPsychiatry, Columbia University and College of Physicians and Surgeons. Formerly, invited Professor at the Department of Psychiatry, Geneva University Medical School. Address reprint requests to A. % Andreoli, M.D., Director Clinique de Psychiattie I, IUPG, 47 rue du XXXI Dtcembre, 1207 Geneve, Switzerland. Copyright 0 1992 by WB. Saunders Company 0010-440X/92/3306-0007$03.00/0 388
Outcome:
alliance and development of insight predicted positive global change and symptom improvement at 1 and 2 years’ follow-up. Furthermore, the observed correlation between process measures and 2-year outcome was found to be independent of age, sex, symptoms severity at intake, improvement of symptoms at CCI termination, premorbid adjustment, DSM-III-R axis l/axis II diagnosis, and therapeutic alliance at intake. Copyright 0 1992 by W. B. Saunders Company
ante and psychiatric outcome is well established. Therapeutic alliance has been reported to result in better treatment response and outcome in a number of psychiatric conditions: depressions schizophrenia,9 psychiatric inpatients,iO and psychopharmacological treatment.” There are also several recent reports indicating that process measures predict psychotherapy outcome.12J3 Alliance measures,14 working alliance,15 patient/therapist interaction ratings, i6 “focusing” process,” and core conflictual relationship patternsi were suggested to be valuable predictors of treatment response and/ or outcome in patients undergoing individual psychotherapy. Very few studies have investigated psychotherapeutic process measures as predictors of outcome of CCI. Such a research direction is in keeping with the goal to develop effective combined treatment strategies,18 especially for those patients presenting with clinical depression.19 We therefore developed a combined CC1 program,‘O which aimed to provide intensive outpatient treatment to depressed patients referred for inpatient care. Since preliminary investigations suggested that such treatment may facilitate alliance, develop insight, and reduce early treatment termination, especially in those patients with depressive symptoms,21 we designed a study intended to (1) determine whether working alliance at 1 week and development of insight at CC1 termination predict long-term treatment compliance and 2-year outcome in psychiatric subjects with depressive symptoms; and (2) evaluate a number of other potential CC1 long-term outcome predictors1-4,21 in order to determine whether the effects of the addressed process measures give an independent
Comprehensive Psychiatry, Vol. 33, No. 6 (November/December),
1992: pp 388-396
CRISIS INTERVENTION
AND OUTCOME
contribution to the variability of the observed long-term treatment compliance/outcome. METHOD
Research Design Our study had a naturalistic design and was intended to obtain a prospective investigation of CC1 long-term outcome in psychiatric patients. To determine whether CC1 response is a predictor of long-term treatment compliance and 2-year outcome, our research plan called for repeated assessments of psychosocial variables, process measures, and outcome parameters in psychiatric subjects assigned to outpatient CCI.
Subjects The study was conducted at the Eaux-Vives Secteur Crisis Intervention Department (EVSCID). “Secteur EauxVives” is a community mental health center affiliated with the Department of Psychiatry of the Geneva University Medical School, which delivers all of the institutional psychiatric care to the 105,000 inhabitants of the Geneva (Switzerland) catchment area. Subjects were consecutive patients referred to EVSCID for inpatient care and assigned to CC1 (see below). Additional inclusion criteria were consenting adults, aged 18 to 65, presenting with depressive symptoms. Exclusion criteria were DSM-III-R psychotic, bipolar and organic Disorder, mental retardation, and psychoactive substances dependence. Thirty-nine of the 78 subjects referred to our center for inpatient care during 2 months met criteria for inclusion in the study. Thirty-seven of them were assigned CCI. Two patients were dropped from the study because they were considered to not need this level of care and none had a standard hospitalization. Thirty-two subjects had l-year and 30 repeated 2-year follow-up interviews. We were without any information about three patients who left the Geneva area, and obtained indirect information only (showing two fair and one bad outcome) for three other subjects. One subject died of suicide during the second follow-up year and was assigned the worst score rated in the sample at second follow-up interview. A final sample of 31 subjects (83.8%) was therefore entered into the data analysis,
Treatment CC1 is a two-phase program integrating multimodal psychiatric treatment for acute psychiatric patients requiring intensive institutional care. CC1 has been extensively described in previous reports.‘“-?’ The first phase (up to 1 week) is intended to facilitate treatment acceptance and working alliance. It includes (1) daily individual sessions of supportive therapy; (2) repeated staff meeting aimed to develop general clinical hypothesis and to select specific interactional strategies; (3) drug treatment; and (4) intensive support of relatives and “significant others.” The second phase is intended to combine standard psychiatric treatment and structured, time-limited (up to 8 weeks), individual psychotherapy aimed to develop insight,
facilitate the expression of overwhelming feelings, manage interpersonal stressors, and improve long-term treatment compliance. It includes: (1) two or more individual sessions of supportive/exploratory psychotherapy per week: (2) assignment to clomipramine protocol (100 to I50 mgid according to the observed plasma levels at the drug monitoring?‘; and (3) careful negotiation of long-term treatment assignment with both patients and future therapists. Over the two phases of CCI, 15 patients (48%) received antidepressant treatment. Drug medication was prescribed by experienced psychiatrists. Supportive psychotherapy was delivered by well-trained nurses, senior residents, and clinical psychologists. Careful, intensive supervision of both psychiatric and psychotherapeutic treatments was provided by two psychoanalytically trained experienced psychiatrists.
Instruments and Assessment Baselinepredictors. Baseline predictors were assessed in five areas: sociodemographics, psychopathology and symptom severity, premorbid adjustment, interpersonal relationships, and clinical history. All evaluations were conducted by an experienced psychiatrist who was a member of the research staff. To assure careful assessment, we used a standardized Clinical Evaluation Profile (PEC). which was administrated at intake and CC1 termination.?j,?” This instrument is a 67-item inventory derivated from the Pronostic Index for Psychotherapy,zh Brief Psychiatric Rating Scale (BPRS),?’ Hamilton Anxiety Scale.‘x and MontgomeryAsberg Depression Scale.?” DSM-III-R diagnostic assessment. At CC1 termination, two experienced psychiatrists of our research staff who had access to the overall clinical records and research forms but not to follow-up results assessed each patient for DSMIII-R diagnostic criteria. Controlling for reliability. we found corrected kappa coefficients of 0.65 for major depressive episode (MDE), 0.59 for generalized anxiety disorder (GAD), and 0.67 for presence/absence of any depressive disorder. Process measures. The initial therapeutic alliance levels were rated by the clinicians on a number of scales derived from the Pronostic Index for Psychotherapy.‘h In addition, three process indices were derived by our research staff from extensive questionnaires completed by clinicians at intake, 1 week, and CC1 termination. The Adherence Index (ADHI) is a five-level scale that assesses the formal adherence of the observed treatment to the basic requirements of the proposed crisis intervention technique. The Working Alliance Index (WAI) is a fourlevel scale that rates changes in patient’s attitude toward symptoms and treatment at the termination of the first phase of CC1 (1 week). The four levels were (1) no change; (2) change in patient’s attitude toward symptoms (symptoms are now viewed as something to understand rather than to merely suppress): (3) changes in patient’s attitude toward treatment (active involvement instead of passive request for medical expertise): and (4) simultaneous change in patient‘s attitude toward symptoms and treatment. Finally. the Psychotherapeutic Attainment Index (PSAI) is a five-level scale that measures the extent of psychotherapeutic achievements at CC1 termination. The five levels were as follows: (1) no change; (2) empathetic contact; (3) expres-
ANDREOLI ET AL
390
sion of feelings associated with symptoms relief; (4) development of insight involving longstanding conflicts; and (5) a “focusing process” resulting in critical transformations of the conflicts underlying the patient’s complaint. Clinicians were trained to rate the presence/absence of the targeted interactional events. Weekly staff meetings allowed discussion and consensus ratings were held each week. Outcome measures. At l- and 2-year follow-up, an experienced clinician rated each of the traced subjects after direct interview on a number of clinical and psychosocial change areas (PCC) according to the additional scales of the Health-Sickness Rating Scale (HSRS).30 Changes were rated on a 5-point scale, where 3 corresponded to “stabilized,” 5 and 1 to “definitely improved or worsened.” and 4 and 2 to “some improvement” or “some worsening.” Each traced subject was also rated on a 3-point index of global outcome (3 = improved, 2 = stable, 1 = worsened) to assess “global quality of clinical evolution.” The internal consistency3’ of therapeutic alliance and outcome measures was found to be satisfactory. Cronbach alphas and intraclass correlation coefficients were obtained within the evaluation of the global cohort (n = 78) and have been reported by our research group.25 Long-term treatment. Treatment assignment at CC1 termination and history of treatment received during the land 2-year follow-up periods was rated on a four-item scale indicating the presence/absence of psychiatric treatment and/or increasing degrees of specificity of treatment: 1 = no treatment, 2 = general practitioner, 3 = aspecific psychiatric treatment, 4 = well structured psychotherapy, antidepressant medication, and/or combined treatment.
Data Analysis The relationships between baseline predictors, process measures, and outcome parameters were analyzed by Spearman’s correlations, and other nonparametric statistics. To reduce the number of comparisons, we selected an a- priori hierarchy of 10 baseline predictors (age, sex, depression severity, global anxiety, psychotic symptoms, quality and maturity of interpersonal relationships, adjustment to work, involvement/functioning in social activities) and seven process measures (initial therapeutic alliance: cooperation, optimistic v pessimistic expectations, insight, and motivation for psychotherapy; CC1 process: ADHI, WAI, PSAI) based on previous studies from our research group.21.32 As mentioned previously, the eight subscales of the HSRS were used as dependent variables. Multiple regression analysis and analysis of partial variante were used to produce partial estimates of the independent contribution of each predictive variable on outcome. In this procedure, the standardized coefficient of each variable describes its effect on the dependent variable controlling statistically for the all model. The multiple squared R (R’) indicates the total amount of the variance explained by the model, p the standardized coefficient, and the increments of R* (IR) the relevant effect of each added variable. Due to the limited power of our sample size and to the metric properties of our data, these statistical investigations must be considered as exploratory to provide hypotheses for further testing.
RESULTS
Sample Characteristics
A description of our sample is given in Table 1. The sample consisted predominantly of middle-aged women with severe depressive symptoms and anxiety. Our subjects had poor to very poor interpersonal relationships and premorbid adjustment. At intake, 24 subjects (77.4%) met DSM III-R criteria for at least one category of depressive disorders: 16 (51.6%) had MDE. Of the remaining seven patients, four had GAD, two panic disorder not meeting criteria for MDE, and one adjustment disorder with depressive features. Process Measures
Patients were found to be mostly cooperative at intake, but were rated poor to very poor on insight, motivation for psychotherapy, and expectations about treatment results (Table 2). Most of the patients we investigated had satisfactory working alliance ratings at 1 week (good to very good: 19 [61.3%]; quite enough: 5 [16.2%]; poor to very poor: 7, [22.6%]), as well as development of insight/focusing (good to very good: 16 [51.6%]; quite enough: 5 [16.2%]; poor to very poor: 10 [32.3%]) and acceptable adherence to the manual (good to very good: 21[67.7%]; poor to very poor: 10 [32.3%]) at CC1 termination. Table 1. Study Sample Characteristics Range
(n = 31)
Mean ? SD
%
Sociodemography Age Sex (male)
18-59
39.23?
11.18
29.03
Symptoms Depression severity
1-5
3.84 + 0.73
-
Global anxiety
1-5
3.94 + 0.85
-
Psychotic symptoms
l-5
2.00 + 0.85
-
1-5
1.60 f 1.35
-
1-5
1.79 r 1.01
-
1-5
1.80 r 0.97
-
1-5
1.80 ? 0.87
-
Social adjustment Adjustment to work Involvement in social activities Interpersonal relationships Quality of interpersonal relationship Maturity of interpersonal relationship DSM-III-R axis I MDE v other
-
51.6
-
77.4
Depressive disorder (any) Y other
CRISIS INTERVENTION
391
AND OUTCOME
Table 2. Process Measures at Intake, 1 Week, and CCI Termination (n = 31)
Range
Mean + SD
Process measures at intake Cooperation
1-5
3.1 * 0.9
Op~imismipessimism
1-5
3.0 2 1.2
Insight
l-5
1.9 * 1.1
Motivation for psychotherapy
1-5
1.6 c 1.0
1-4
2.9 + 1.3
ADHI
l-5
3.3 t 1.4
PSAI
1-5
3.4 + 1.5
Process measures at 1 week WAI Process measures at CCI termination
Process Measures and Predictors at Baseline
Increased insight at CC1 termination correlated (P < .05) with better quality of interpersonal relationships (r = .34) and better therapeutic alliance (cooperation, motivation for psychotherapy, insight and pessimismioptimism) at baseline (r = 0.33-0.40). However, we did not find significant relationships between this process measure and DSM-III-R axis I and axis II diagnosis, age, sex, symptom severity, social adjustment, or antidepressant medication during CCI. At CC1 termination, increased PSAI ratings were associated with greater improvement of both depression (P < .Ol) and anxiety (P = .05). Better working alliance at 1 week was correlated only with increased cooperation at intake (P < .05). Finally, severer symptoms predicted better adherence to the proposed treatment schedule within CC1 (P < .Ol). Outcome
Subjects were mostly improved on global change, symptom severity, tolerance to symptoms, and adjustment to family at both l- and 2-year follow-up (Fig 1). However, the improvement of adjustment and social functioning was less impressive, especially at first follow-up interview.
work and involvement in social activities and better 2-year outcome. These variables had a slight correlation with social adjustment and/or quality of interpersonal relationships at 2-year follow-up (Table 3). Presence versus absence of DSM-III-R depressive disorder (any) or MDE at intake were not found to predict long-term outcome in our sample. Does Symptom Change at CCI Termination Predict Long- Term Outcome?
Symptom improvement at CC1 termination did not predict long-term outcome. Specifically, there was no significant relationship found between the improvement of anxiety, depression, or psychotic symptoms at CC1 termination and the overall follow-up measures we investigated in the present study. Taken together, improvement of depression, anxiety, and psychotic symptoms at CC1 termination accounted for 5% and 1.8% of the variability of global change at 1 and 2 years’ follow-up, respectively. Do Process Measures Predict CCI Long- Term Outcome?
Process measures were found to be the best predictors of CC1 outcome at both l- and 2-years follow-up interviews (Tables 4 and 5). Taken together, therapeutic alliance at baseline, adherence, working alliance, and global psychoterapeutic attainments accounted for 50.1% and 62.7% of global change and symptoms severity, respectively, at l-year follow-up. Furthermore, process measures predicted 44% and 39% of the variability of the same parameters at 2 years’ follow-up. Looking separately at each subgroup of pro-
Do Subjects’ Psychosocial Profiles at Intake Predict CCI Long- Term Outcome?
Better quality of interpersonal relationships and younger age were the only baseline predictors associated with better l- and 2-year outcome. As indicated in Table 3, we also observed an association between increased adjustment at
Fig 1. One- and Z-year outcome: proved at HSRS (n = 31).
percent of patients im-
392
ANDREOLI ET AL
Table 3. Two-Year Outcome and Baseline Predictors (Spearman’s correlation coefficients; n = 31)
Age Sex (female)*
Global Change
Symptom Severity
-0.205
-0.128
0.433
Depression severity
0.046
Global anxiety
0.122
Psychotic symptoms
-0.211
0.088 -0.154 0.104 -0.190
TCllsrslXs to Symptoms -0.240 0.009 -0.116 0.155 -0.285
Adjustment Autonomy -0.358
Involvement -0.456b
Work -0.096
0.000
1.261
0.543
0.223
0.139
0.061
0.021
-0.224
-0.025
Family -0.3708
Interpersonal Relationships -0.138
GOI -0.155
0.818
3.9768
1.398
0.068
0.202
0.036
0.033
0.342*
0.089
0.041
-0.291
-0.232
-0.148
0.196 -0.272
Interpersonal relationshipst
0.4118
0.325a
0.358a
0.378
0.301
0.285
0.503c
0.457
0.36ga
Interpersonal relation0.183
0.191
0.016
0.049a
0.331a
0.350a
0.186
Work
0.149
0.190
0.285
0.321a
0.250
0.348”
0.030
0.143
0.068
Involvement
0.273
0.170
0.273
0.253
0.259
0.071
0.305”
0.341a
0.271
ships*
0.273
-0.178
Abbreviations: GOI, Global Outcome Index. *Chi-square statistics. tQuality. *Maturity. P
Intervention
A.V. Andreoli, A. Muehlebach,
Response and Long-Term A Pilot Study
M. Gognalons, J. Abensur, S. Grimm, and A.J. Frances
We investigated whether process variables (therapeutic alliance and insight) measured at the termination of crisis intervention predict long-term treatment compliance and 2-year outcome. Thirty-seven consecutive depressed psychiatric patients assigned to outpatient crisis intervention (Ccl) were assessed with both questionnaires and standardized instruments at intake, 1 week, and CCI termination [mean, 6 weeks). Thirty-one subjects (64%) were also evaluated at l-year and 2-year follow-up. We found that working
T
HE PURPOSE of this study was to assess the outcome of crisis intervention (CCI) and to determine whether process measures (i.e., working alliance and development of insight at CC1 termination) predict long-term outcome. Earlier research has indicated that symptom improvement and psychosocial recovery at acute treatment termination may be unrelated to long-term outcome in hospitalized patients assigned to short-term intensive therapy.1-4 The long-term outcome of psychiatric patients referred for intensive treatment with severe DSMIII-R disorders appears to be influenced by the availability of both specialized long-term treatment in the community5 and multimodal crisis intervention.6l7 In addition, process ratings within interactive CC1 have been reported to provide valuable predictors of succesful long-term outcome in psychiatric patients.’ These observations suggest that increased alliance and insight could well be related to compliance to longterm treatment, decreased relapse risk, and enhanced symptom relief. The relationship between therapeutic alli-
From the Depattment of Psychiatry, University of Geneva Medical School, Geneva, Switzerland. Supported in part by the Scientific Research Swiss National Fund, Contract No. 3.822.1.84. Dr. Frances is a ProfessorofPsychiatry, Columbia University and College of Physicians and Surgeons. Formerly, invited Professor at the Department of Psychiatry, Geneva University Medical School. Address reprint requests to A. % Andreoli, M.D., Director Clinique de Psychiattie I, IUPG, 47 rue du XXXI Dtcembre, 1207 Geneve, Switzerland. Copyright 0 1992 by WB. Saunders Company 0010-440X/92/3306-0007$03.00/0 388
Outcome:
alliance and development of insight predicted positive global change and symptom improvement at 1 and 2 years’ follow-up. Furthermore, the observed correlation between process measures and 2-year outcome was found to be independent of age, sex, symptoms severity at intake, improvement of symptoms at CCI termination, premorbid adjustment, DSM-III-R axis l/axis II diagnosis, and therapeutic alliance at intake. Copyright 0 1992 by W. B. Saunders Company
ante and psychiatric outcome is well established. Therapeutic alliance has been reported to result in better treatment response and outcome in a number of psychiatric conditions: depressions schizophrenia,9 psychiatric inpatients,iO and psychopharmacological treatment.” There are also several recent reports indicating that process measures predict psychotherapy outcome.12J3 Alliance measures,14 working alliance,15 patient/therapist interaction ratings, i6 “focusing” process,” and core conflictual relationship patternsi were suggested to be valuable predictors of treatment response and/ or outcome in patients undergoing individual psychotherapy. Very few studies have investigated psychotherapeutic process measures as predictors of outcome of CCI. Such a research direction is in keeping with the goal to develop effective combined treatment strategies,18 especially for those patients presenting with clinical depression.19 We therefore developed a combined CC1 program,‘O which aimed to provide intensive outpatient treatment to depressed patients referred for inpatient care. Since preliminary investigations suggested that such treatment may facilitate alliance, develop insight, and reduce early treatment termination, especially in those patients with depressive symptoms,21 we designed a study intended to (1) determine whether working alliance at 1 week and development of insight at CC1 termination predict long-term treatment compliance and 2-year outcome in psychiatric subjects with depressive symptoms; and (2) evaluate a number of other potential CC1 long-term outcome predictors1-4,21 in order to determine whether the effects of the addressed process measures give an independent
Comprehensive Psychiatry, Vol. 33, No. 6 (November/December),
1992: pp 388-396
CRISIS INTERVENTION
AND OUTCOME
contribution to the variability of the observed long-term treatment compliance/outcome. METHOD
Research Design Our study had a naturalistic design and was intended to obtain a prospective investigation of CC1 long-term outcome in psychiatric patients. To determine whether CC1 response is a predictor of long-term treatment compliance and 2-year outcome, our research plan called for repeated assessments of psychosocial variables, process measures, and outcome parameters in psychiatric subjects assigned to outpatient CCI.
Subjects The study was conducted at the Eaux-Vives Secteur Crisis Intervention Department (EVSCID). “Secteur EauxVives” is a community mental health center affiliated with the Department of Psychiatry of the Geneva University Medical School, which delivers all of the institutional psychiatric care to the 105,000 inhabitants of the Geneva (Switzerland) catchment area. Subjects were consecutive patients referred to EVSCID for inpatient care and assigned to CC1 (see below). Additional inclusion criteria were consenting adults, aged 18 to 65, presenting with depressive symptoms. Exclusion criteria were DSM-III-R psychotic, bipolar and organic Disorder, mental retardation, and psychoactive substances dependence. Thirty-nine of the 78 subjects referred to our center for inpatient care during 2 months met criteria for inclusion in the study. Thirty-seven of them were assigned CCI. Two patients were dropped from the study because they were considered to not need this level of care and none had a standard hospitalization. Thirty-two subjects had l-year and 30 repeated 2-year follow-up interviews. We were without any information about three patients who left the Geneva area, and obtained indirect information only (showing two fair and one bad outcome) for three other subjects. One subject died of suicide during the second follow-up year and was assigned the worst score rated in the sample at second follow-up interview. A final sample of 31 subjects (83.8%) was therefore entered into the data analysis,
Treatment CC1 is a two-phase program integrating multimodal psychiatric treatment for acute psychiatric patients requiring intensive institutional care. CC1 has been extensively described in previous reports.‘“-?’ The first phase (up to 1 week) is intended to facilitate treatment acceptance and working alliance. It includes (1) daily individual sessions of supportive therapy; (2) repeated staff meeting aimed to develop general clinical hypothesis and to select specific interactional strategies; (3) drug treatment; and (4) intensive support of relatives and “significant others.” The second phase is intended to combine standard psychiatric treatment and structured, time-limited (up to 8 weeks), individual psychotherapy aimed to develop insight,
facilitate the expression of overwhelming feelings, manage interpersonal stressors, and improve long-term treatment compliance. It includes: (1) two or more individual sessions of supportive/exploratory psychotherapy per week: (2) assignment to clomipramine protocol (100 to I50 mgid according to the observed plasma levels at the drug monitoring?‘; and (3) careful negotiation of long-term treatment assignment with both patients and future therapists. Over the two phases of CCI, 15 patients (48%) received antidepressant treatment. Drug medication was prescribed by experienced psychiatrists. Supportive psychotherapy was delivered by well-trained nurses, senior residents, and clinical psychologists. Careful, intensive supervision of both psychiatric and psychotherapeutic treatments was provided by two psychoanalytically trained experienced psychiatrists.
Instruments and Assessment Baselinepredictors. Baseline predictors were assessed in five areas: sociodemographics, psychopathology and symptom severity, premorbid adjustment, interpersonal relationships, and clinical history. All evaluations were conducted by an experienced psychiatrist who was a member of the research staff. To assure careful assessment, we used a standardized Clinical Evaluation Profile (PEC). which was administrated at intake and CC1 termination.?j,?” This instrument is a 67-item inventory derivated from the Pronostic Index for Psychotherapy,zh Brief Psychiatric Rating Scale (BPRS),?’ Hamilton Anxiety Scale.‘x and MontgomeryAsberg Depression Scale.?” DSM-III-R diagnostic assessment. At CC1 termination, two experienced psychiatrists of our research staff who had access to the overall clinical records and research forms but not to follow-up results assessed each patient for DSMIII-R diagnostic criteria. Controlling for reliability. we found corrected kappa coefficients of 0.65 for major depressive episode (MDE), 0.59 for generalized anxiety disorder (GAD), and 0.67 for presence/absence of any depressive disorder. Process measures. The initial therapeutic alliance levels were rated by the clinicians on a number of scales derived from the Pronostic Index for Psychotherapy.‘h In addition, three process indices were derived by our research staff from extensive questionnaires completed by clinicians at intake, 1 week, and CC1 termination. The Adherence Index (ADHI) is a five-level scale that assesses the formal adherence of the observed treatment to the basic requirements of the proposed crisis intervention technique. The Working Alliance Index (WAI) is a fourlevel scale that rates changes in patient’s attitude toward symptoms and treatment at the termination of the first phase of CC1 (1 week). The four levels were (1) no change; (2) change in patient’s attitude toward symptoms (symptoms are now viewed as something to understand rather than to merely suppress): (3) changes in patient’s attitude toward treatment (active involvement instead of passive request for medical expertise): and (4) simultaneous change in patient‘s attitude toward symptoms and treatment. Finally. the Psychotherapeutic Attainment Index (PSAI) is a five-level scale that measures the extent of psychotherapeutic achievements at CC1 termination. The five levels were as follows: (1) no change; (2) empathetic contact; (3) expres-
ANDREOLI ET AL
390
sion of feelings associated with symptoms relief; (4) development of insight involving longstanding conflicts; and (5) a “focusing process” resulting in critical transformations of the conflicts underlying the patient’s complaint. Clinicians were trained to rate the presence/absence of the targeted interactional events. Weekly staff meetings allowed discussion and consensus ratings were held each week. Outcome measures. At l- and 2-year follow-up, an experienced clinician rated each of the traced subjects after direct interview on a number of clinical and psychosocial change areas (PCC) according to the additional scales of the Health-Sickness Rating Scale (HSRS).30 Changes were rated on a 5-point scale, where 3 corresponded to “stabilized,” 5 and 1 to “definitely improved or worsened.” and 4 and 2 to “some improvement” or “some worsening.” Each traced subject was also rated on a 3-point index of global outcome (3 = improved, 2 = stable, 1 = worsened) to assess “global quality of clinical evolution.” The internal consistency3’ of therapeutic alliance and outcome measures was found to be satisfactory. Cronbach alphas and intraclass correlation coefficients were obtained within the evaluation of the global cohort (n = 78) and have been reported by our research group.25 Long-term treatment. Treatment assignment at CC1 termination and history of treatment received during the land 2-year follow-up periods was rated on a four-item scale indicating the presence/absence of psychiatric treatment and/or increasing degrees of specificity of treatment: 1 = no treatment, 2 = general practitioner, 3 = aspecific psychiatric treatment, 4 = well structured psychotherapy, antidepressant medication, and/or combined treatment.
Data Analysis The relationships between baseline predictors, process measures, and outcome parameters were analyzed by Spearman’s correlations, and other nonparametric statistics. To reduce the number of comparisons, we selected an a- priori hierarchy of 10 baseline predictors (age, sex, depression severity, global anxiety, psychotic symptoms, quality and maturity of interpersonal relationships, adjustment to work, involvement/functioning in social activities) and seven process measures (initial therapeutic alliance: cooperation, optimistic v pessimistic expectations, insight, and motivation for psychotherapy; CC1 process: ADHI, WAI, PSAI) based on previous studies from our research group.21.32 As mentioned previously, the eight subscales of the HSRS were used as dependent variables. Multiple regression analysis and analysis of partial variante were used to produce partial estimates of the independent contribution of each predictive variable on outcome. In this procedure, the standardized coefficient of each variable describes its effect on the dependent variable controlling statistically for the all model. The multiple squared R (R’) indicates the total amount of the variance explained by the model, p the standardized coefficient, and the increments of R* (IR) the relevant effect of each added variable. Due to the limited power of our sample size and to the metric properties of our data, these statistical investigations must be considered as exploratory to provide hypotheses for further testing.
RESULTS
Sample Characteristics
A description of our sample is given in Table 1. The sample consisted predominantly of middle-aged women with severe depressive symptoms and anxiety. Our subjects had poor to very poor interpersonal relationships and premorbid adjustment. At intake, 24 subjects (77.4%) met DSM III-R criteria for at least one category of depressive disorders: 16 (51.6%) had MDE. Of the remaining seven patients, four had GAD, two panic disorder not meeting criteria for MDE, and one adjustment disorder with depressive features. Process Measures
Patients were found to be mostly cooperative at intake, but were rated poor to very poor on insight, motivation for psychotherapy, and expectations about treatment results (Table 2). Most of the patients we investigated had satisfactory working alliance ratings at 1 week (good to very good: 19 [61.3%]; quite enough: 5 [16.2%]; poor to very poor: 7, [22.6%]), as well as development of insight/focusing (good to very good: 16 [51.6%]; quite enough: 5 [16.2%]; poor to very poor: 10 [32.3%]) and acceptable adherence to the manual (good to very good: 21[67.7%]; poor to very poor: 10 [32.3%]) at CC1 termination. Table 1. Study Sample Characteristics Range
(n = 31)
Mean ? SD
%
Sociodemography Age Sex (male)
18-59
39.23?
11.18
29.03
Symptoms Depression severity
1-5
3.84 + 0.73
-
Global anxiety
1-5
3.94 + 0.85
-
Psychotic symptoms
l-5
2.00 + 0.85
-
1-5
1.60 f 1.35
-
1-5
1.79 r 1.01
-
1-5
1.80 r 0.97
-
1-5
1.80 ? 0.87
-
Social adjustment Adjustment to work Involvement in social activities Interpersonal relationships Quality of interpersonal relationship Maturity of interpersonal relationship DSM-III-R axis I MDE v other
-
51.6
-
77.4
Depressive disorder (any) Y other
CRISIS INTERVENTION
391
AND OUTCOME
Table 2. Process Measures at Intake, 1 Week, and CCI Termination (n = 31)
Range
Mean + SD
Process measures at intake Cooperation
1-5
3.1 * 0.9
Op~imismipessimism
1-5
3.0 2 1.2
Insight
l-5
1.9 * 1.1
Motivation for psychotherapy
1-5
1.6 c 1.0
1-4
2.9 + 1.3
ADHI
l-5
3.3 t 1.4
PSAI
1-5
3.4 + 1.5
Process measures at 1 week WAI Process measures at CCI termination
Process Measures and Predictors at Baseline
Increased insight at CC1 termination correlated (P < .05) with better quality of interpersonal relationships (r = .34) and better therapeutic alliance (cooperation, motivation for psychotherapy, insight and pessimismioptimism) at baseline (r = 0.33-0.40). However, we did not find significant relationships between this process measure and DSM-III-R axis I and axis II diagnosis, age, sex, symptom severity, social adjustment, or antidepressant medication during CCI. At CC1 termination, increased PSAI ratings were associated with greater improvement of both depression (P < .Ol) and anxiety (P = .05). Better working alliance at 1 week was correlated only with increased cooperation at intake (P < .05). Finally, severer symptoms predicted better adherence to the proposed treatment schedule within CC1 (P < .Ol). Outcome
Subjects were mostly improved on global change, symptom severity, tolerance to symptoms, and adjustment to family at both l- and 2-year follow-up (Fig 1). However, the improvement of adjustment and social functioning was less impressive, especially at first follow-up interview.
work and involvement in social activities and better 2-year outcome. These variables had a slight correlation with social adjustment and/or quality of interpersonal relationships at 2-year follow-up (Table 3). Presence versus absence of DSM-III-R depressive disorder (any) or MDE at intake were not found to predict long-term outcome in our sample. Does Symptom Change at CCI Termination Predict Long- Term Outcome?
Symptom improvement at CC1 termination did not predict long-term outcome. Specifically, there was no significant relationship found between the improvement of anxiety, depression, or psychotic symptoms at CC1 termination and the overall follow-up measures we investigated in the present study. Taken together, improvement of depression, anxiety, and psychotic symptoms at CC1 termination accounted for 5% and 1.8% of the variability of global change at 1 and 2 years’ follow-up, respectively. Do Process Measures Predict CCI Long- Term Outcome?
Process measures were found to be the best predictors of CC1 outcome at both l- and 2-years follow-up interviews (Tables 4 and 5). Taken together, therapeutic alliance at baseline, adherence, working alliance, and global psychoterapeutic attainments accounted for 50.1% and 62.7% of global change and symptoms severity, respectively, at l-year follow-up. Furthermore, process measures predicted 44% and 39% of the variability of the same parameters at 2 years’ follow-up. Looking separately at each subgroup of pro-
Do Subjects’ Psychosocial Profiles at Intake Predict CCI Long- Term Outcome?
Better quality of interpersonal relationships and younger age were the only baseline predictors associated with better l- and 2-year outcome. As indicated in Table 3, we also observed an association between increased adjustment at
Fig 1. One- and Z-year outcome: proved at HSRS (n = 31).
percent of patients im-
392
ANDREOLI ET AL
Table 3. Two-Year Outcome and Baseline Predictors (Spearman’s correlation coefficients; n = 31)
Age Sex (female)*
Global Change
Symptom Severity
-0.205
-0.128
0.433
Depression severity
0.046
Global anxiety
0.122
Psychotic symptoms
-0.211
0.088 -0.154 0.104 -0.190
TCllsrslXs to Symptoms -0.240 0.009 -0.116 0.155 -0.285
Adjustment Autonomy -0.358
Involvement -0.456b
Work -0.096
0.000
1.261
0.543
0.223
0.139
0.061
0.021
-0.224
-0.025
Family -0.3708
Interpersonal Relationships -0.138
GOI -0.155
0.818
3.9768
1.398
0.068
0.202
0.036
0.033
0.342*
0.089
0.041
-0.291
-0.232
-0.148
0.196 -0.272
Interpersonal relationshipst
0.4118
0.325a
0.358a
0.378
0.301
0.285
0.503c
0.457
0.36ga
Interpersonal relation0.183
0.191
0.016
0.049a
0.331a
0.350a
0.186
Work
0.149
0.190
0.285
0.321a
0.250
0.348”
0.030
0.143
0.068
Involvement
0.273
0.170
0.273
0.253
0.259
0.071
0.305”
0.341a
0.271
ships*
0.273
-0.178
Abbreviations: GOI, Global Outcome Index. *Chi-square statistics. tQuality. *Maturity. P
