Journal of Affective Elsevier

17

Disorders, 18 (1990) 17-28

JAD 00662

A randomized

clinical trial of inpatient

family intervention

V. Results for affective disorders John F. Clarkin I,*, Ira D. Glick 133,Gretchen L. Haas lv3, James H. Spencer Alfred B. Lewis 1,3,Joanne Peyser 4, Nancy DeMane 5, Marcie Good-Ellis Elizabeth Harris * and Veronica Lestelle 3

1,3, 3,

’ Department of Psychiatry, Cornell University Medical College, New York, NY, U.S.A., ’ New York Hospital, Westchester Dmision, White Plains, NY, U.S.A., 3 Payne Whitney Clinic oftheNew York Hospital, New York, NY, U.S.A., 4 Family Center, Greenwich, CT, U.S.A. and ’ Mount Sinai Medical Center, New York, NY, U.S.A. (Received 19 December 1988) (Accepted 3 May 1989)

Summary This paper reports the results at follow-up of a randomized clinical trial of combining family intervention with drug treatment during hospitalization for patients with affective disorder. The results suggest that female bipolar patients and their families benefited from family intervention, whereas unipolar patients and families did not. Patient outcome was positively correlated with the achievement of the goals of family intervention.

Key words: Affective

disorders;

Bipolar

disorder;

Introduction Some form of family intervention coupled with drug treatment has become a standard part of treatment for most patients hospitalized with affective disorder. But is the addition of family intervention more effective than treatment without it? And, if it is effective, for which patients and/or families?

Address for correspondence: John F. Clarkin, Ph.D., The New York Hospital, Cornell Medical Center, Westchester Division, 21 Bloomingdale Road. White Plains, NY 10605. U.S.A. 0165-0327/90/$03.50

0 1990 Elsevier Science Publishers

Family

therapy

The effectiveness of psychopharmacology in the treatment of affective disorders has been welldocumented (Prien et al., 1984) but there is growing evidence that while a ‘necessary’ component, alone it is likely to be insufficient. The impact of medication is dependent upon long-term compliance with a continuation regimen and lithium non-compliance can occur in as many as 50% of cases (Jamison and Akiskal, 1983). Even with adequate drug compliance and optimal treatment response, the impact of medication on impaired social functioning is usually limited (Friedman, 1975; DiMascio et al., 1979; Weissman et* al., 1979). In addition, psychosocial stress (e.g., family

B.V. (Biomedical

Division)

18

‘expressed emotion’ (EE)) may provoke relapse in patients with affective disorder (Vaughn and Leff, 1976; Hooley, 1986) and recent data indicate that family EE and affective style predict clinical outcome for bipolar patients at 9 months post hospital discharge (Miklowitz et al., 1988). Conversely, social support for the patient with bipolar disorder has been shown to correlate with patient outcome (O’Connell et al., 1985). Psychotherapeutic intervention to increase medication compliance and reduce social stress has been tried in a few studies involving affective disorder patients. Well-controlled clinical research on unipolar depression has demonstrated that outpatient psychotherapy used in conjunction with pharmacologic treatments can improve patient functioning in the areas of social, family, and work adjustment (DiMascio et al., 1979; Weissman et al., 1979, 1981). There have been only two studies of marital or family treatments for affective disorder (both for outpatients) (Friedman, 1975; Davenport et al., 1977). To date, there have been no investigations of family intervention with affective disorder patients during a hospitalization episode. Guided by these concerns, we have assessed the usefulness of Inpatient Family Intervention (IFI) for schizophrenic and affective disorder patients on an acute psychiatric unit (Glick et al., 1985; Haas et al., 1988; Spencer et al., 1988). In this report, we focus in greater detail on the follow-up results for patients with a major affective disorder and their families. Our hypothesis was that a brief family intervention delivered during hospitalization of a family member with an affective disorder would improve family attitudes toward the patient and enhance family attitudes towards his/her treatment (including medication compliance), and that these intervening effects would have a positive impact on patient outcome in terms of both symptomatology and functioning. Methods Treatment setting The study was conducted during the brief hospitalization of patients on the seventh floor unit of the Payne Whitney Clinic, New York Hospital. This is a 24-bed unit that admits volun-

tary patients ranging in age from adolescent to elderly. Patients of all diagnostic categories are admitted from a central evaluation service or from the emergency room at nights and on weekends. The average length of stay (mean = 36 days) is determined by the patient’s clinical need, motivation for treatment, and the availability of resources to finance hospitalization. Patients The full study of IF1 involved patients with either DSM-III schizophrenic/ schizophreniform disorder or major affective disorder who were consecutively admitted to the study unit and consented to participate in the study (Glick et al., 1985). In this report we focus on the 50 patients with DSM-III major affective disorder who met criteria for admission to the study (family available and anticipated length of stay sufficient to participate in IFI). For all patients in the study, the initial assignment to diagnostic group was made with information available at admission using DSM-III criteria. Since longitudinal course is crucial for a correct diagnosis, we used the approach advocated by the developers of RDC and by the current NIMH collaborative project on the study of depression. That is, a diagnostic ‘update’ at the end of the index episode was used to make a final research diagnosis for each patient at 18 months using all information available at that point. The data in Table 1 and all data analyses are based on these research diagnoses. Table 1 provides demographic and clinical data on the total sample of major affective disorder patients and a breakdown into two diagnostic subgroups, bipolar and unipolar types. Of the 21 bipolar patients, during the index admission 13 presented as manic, seven as depressed, and one as mixed. Seventeen of 21 were psychotic at the time of their hospitalization, and 14 of 21 had had previous episodes. Of the seven that had not had prior episodes, six presented as manic; the remaining one presented as depressed and went on to develop mania post hospitalization. The bipolar patients were predominantly female, white, and single, with a mean age of 32. Half had had previous hospitalizations. The unipolar patients were almost equally balanced with regard to sex,

19 TABLE

TABLE

1

1 (continued)

A COMPARISON OF UNIPOLAR AND BIPOLAR PATIENTS ON DEMOGRAPHIC CHARACTERISTICS AND ROLE FUNCTIONING

Unipolar

Bipolar

(n = 29) Unipolar (n = 29)

Bipolar

All affectives (n = 50)

-(n=21)

n

Bn

Sex Male Female

13 16

45 55

7 14

33 67

20 30

40 60

Race White Non-white

18 11

62 38

17 4

81 19

35 15

70 30

Mean SD

38.4 12.7

Marital status Single Married Divorced Widowed

9 15 4 1

31 52 14 3

11 7 3 0

52 33 14 0

20 22 7 1

40 44 14 2

6

21

8

38

14

28

4 3

14 10

5 1

24 5

9 4

18 8

0 8

0 29

0 3

0 14

0 11

0 22

1 6 0 1

3 21 0 3

0 3 1 0

0 14 5 0

1 9 1 1

2 18 2 2

Previous hospitalizations 0 l-2 3 or more

18 7 4

62 24 14

10 8 3

48 38 14

28 15 7

56 30 14

Previous episodes 0 1-2 3 or more

14 11 4

48 38 14

7 10 4

33 48 19

21 21 8

42 42 16

Pre-hospital role functioning (RPTS) 1.0-1.9 (Good) 5 17 2.0-2.9 9 31 3.0-3.9 10 34 4.0-4.9 4 14 5.0-5.9 (Poor) 1 3

2 11 5 3 0

10 52 24 14 0

7 20 15 I 1

14 40 30 14 2

Living situation Single, living alone Single, with family of origin Single, with children Single, with children and parents Married, no children Married, children elsewhere Married, with children Married, with parents Institution

5&n

%

35.9 14.5

32.3 15.4

n

All affectives (n = 50)

(n = 21) %

n

%

Global functioning at admrssion (GAS score) Mean 32.16 28.04 SD 5.81 5.65 Maximum 40.30 42.70 Minimum 18.70 20.30

n

%

30.4 6.0 42.7 18.7

predominantly white and married, with an average age of 38 years. The majority had had no previous hospitalizations. Treatment assignment Within the first 2 days of hospitalization, patients were randomly assigned to one of two treatment groups: ZFZ (standard, multimodal hospital treatment plus IFI) or comparison treatment (standard hospital treatment without family intervention). The standard multimodal hospital treatment included a full range of diagnostic services and drug plus individual, group, and milieu therapies. The IF1 and comparison treatment groups were treated on the same unit by the same staff. Patients (and their families) assigned to IF1 received at least six family intervention sessions, lasting 45 min to 1 h. Family sessions were conducted by one of two experienced family therapists (social workers) with the patient’s individual therapist (psychiatric resident or psychology intern) serving as co-therapist. A treatment manual (Clarkin et al., 1981) described the psychoeducational family intervention.* The goals of IFI were: (1) acceptance by patient and family of the reality of the illness and development of an understanding of the current episode; (2) identification of possible precipitating stresses in the current episode; (3) identification of likely future stresses

* The manual

is available

upon request

from the first author.

20

within and outside of the family; (4) elucidation of family interactions that produce stress on the patient; (5) planning of strategies for managing and/or minimizing future stresses; and (6) acceptance by patient and family of the patient’s need for continued treatment after discharge. The families of comparison patients were interviewed at the time of admission to gather historical information, and any questions then or during the hospitalization concerning the hospital or the patient’s illness were answered factually while minimizing intervention in the family system. A partially fixed drug regimen was implemented for all patients to provide adequate dosages, and to balance the level of medication used across the two treatment conditions. Lithium was used in bipolar affective disorders in the range of 600-2100 mg daily to reach adequate blood levels of 0.661.5 mEq/l. Tricyclic antidepressants (e.g., amitriptyline, irnipramine and desipramine) were used in daily dosages of 100-400 mg with blood levels followed when necessary. Delusional depressives often received, in addition to the above, a neuroleptic in doses equivalent to 105-800 mg of chlorpromazine. Monoamine oxidase inhibitors (usually phenelzine at 45-90 mg daily) were used in cases where tricyclics failed. There were no statistically significant differences in the amount of medication received by the two groups nor in the achievement of adequate blood levels of lithium during hospitalization.

Comparison of treatment groups at baseline The IFI and comparison groups were compared on demographic and baseline clinical measures. The two treatment groups were essentially equivalent on demographic variables. On baseline clinical ratings (Table l), patients showed equivalent levels of prehospital role functioning; IF1 patients showed less severe overall symptomatology on one of two global measures, the PEF (P < 0.05), although there were no differences on any of the specific symptom measures. Comparisons between treatment groups within the unipolar and bipolar categories showed no differences except for a trend (P < 0.09) among the bipolars for the IF1 patients to have a lower socioeconomic status than the comparison patients.

Assessment Both patient and family functioning were assessed at multiple points in time. Patient measures included those relating to global status (the Global Assessment Scale (GAS)) (Endicott et al., 1976) symptomatology and role functioning (the Psychiatric Evaluation Form (PEF)) (Endicott and Spitzer, 1972) which assess global outcome and symptoms; and the Role Performance Treatment Scale (RPTS) (Good-Ellis et al., 1987), a semistructured interview scale developed for this study to measure patient role functioning in the community. Family measures included the Family Attitude Inventory (FAI) (Levitt, 1982; Haas et al., 1986) developed for this study to measure family attitudes to treatment, social support, the patient, and family burden. Independent observers, blind to patient treatment assignment, administered the evaluation instruments. Data analysis strategies We used higher-order analyses of covariance with composite outcome measures (Beutler and Hamblin, 1986) to minimize both type I and type II errors. Two patient composite measures were derived from a principal components analysis of the residualized (adjusted for baseline) individual patient outcome measures which yielded two major factors. A similar analysis of the residualized family measures yielded two family composite outcome measures. We report the results of the three-way (treatment X sex X subdiagnosis) analyses of these composites for all affective disorder patients, and two-way (treatment X sex) analyses for bipolar and unipolar patients (Tables 2 and 3). Results of one-way analyses of the individual outcome measures are also presented as effect sizes (Tables 4). These are computed as correlation ratios (Cohen, 1965; Rosenthal and Rosnow, 1984) which express the degree of relationship between treatment assignment and outcome; r2 is the proportion of outcome variance on a particular measure which is due to treatment group membership. The demonstration of main and/or interaction effects in the experiment-wise analysis on composite measures allows us to test each of these many individual effect sizes for significance without the risk of type I error. Any significant treatment effect on an individual outcome mea-

21

sure which contributes to a significant effect in the higher-order analyses of composites can be taken as valid; it contributes to the experiment-wise result and is therefore ‘protected’ against the charge that it is a chance finding in multiple significance tests (Cohen and Cohen, 1975, pp. 162-165). Results Outcome at discharge To place the follow-up results in context, we will summarize the discharge results reported earlier (Haas et al., 1988). Female patients with affective disorder who received IFI, as compared with those in the comparison group, were doing significantly better at time of discharge from hospital on measures of global functioning and symptomatology. Males, on the other hand, were little affected by treatment assignment. Role functioning measures used at later follow-up were not completed at discharge, because the patients had not yet returned to their family and community roles. Family attitude toward treatment was significantly better in IF1 females than in comparison females, whereas family attitude toward the patient was significantly better in males who received the comparison treatment than in males who received IFI. Attrition and missing data Six (four IF1 and two comparison treatment) cases of the original 56 failed to complete the minimum amount of treatment to meet study criteria, defined arbitrarily as a minimum of two family sessions for patients in the IF1 group or 10 days of hospitalization for patients in the comparison group. The two comparison patients were female, both with a diagnosis of bipolar manic, and both left the hospital against medical advice. In the IF1 group, two were male and two were female. Diagnostically, one was psychotically depressed and the other three were bipolar. The major distinguishing characteristic of the six treatment refusers was their rejection of the hospital treatment in general, including but not specific to IFI. This group was not distinguished by other factors such as severity of illness and in no case did the clinical state of the patient worsen as far as we could tell by involvement in one or two

family sessions. Data on the remaining 50 (21 bipolar and 29 unipolar) are reported here. Of the 21 bipolar patients, 12 received IF1 and nine received the comparison treatment; of the 29 unipolar patients, 17 received IF1 and 12 received the comparison treatment. The apparent imbalance in treatment assignment resulted primarily from the diagnostic update made for research purposes at the completion of the study when more clinical data were available than at admission. At 6 months we were able to locate and interview all 50 patients, but three families refused the interview. At 18 months, we could not locate two patients and three families refused the interview. For analyses of composite outcome measures, missing items were replaced by the group mean value on the associated scale (Cohen and Cohen, 1975). For analysis of individual measures, missing values were treated as missing. Higher-order analyses of variance on composite measures Tables 2 and 3 contain results at 6 and 18 months of three-way analyses for all affective disorders combined and two-way analyses for bipolars and unipolars taken separately. Figs. 1 and 2 provide graphic illustrations of the follow-up results on the patient global/symptoms composite. Treatment effects on patients. For the combined group of patients with affective disorder, at both follow-up points, we found no main effects of treatment on the global/symptoms composite measure while the significant treatment x sex interaction found at discharge was overpowered by a treatment X subdiagnosis interaction effect. But at 18 months we found a significant treatment x diagnosis interaction (P < 0.05) indicating that bipolars did better with IFI, whereas unipolars did better with the comparison treatment (Fig. 2). This effect was anticipated by a similar trend (P < 0.09) at 6 months. Fig. 1 shows no main effect and suggests that an apparent interaction effect at 18 months (which is not shown to be statistically significant in these data) is due to IF1 males doing worse rather than females doing better. Independent analyses for bipolars and unipolars show that the effect favoring IF1 for bi-

22 TABLE

2

RESULTS

OF THREE-WAY

ANALYSES

ON COMPOSITE

PATIENT

MEASURES

6 Months

All patients TA SEX DX TA x SEX TAxDX SEX x DX TAxSEXxDX Covariate Error

1 1 1 1 1 1 1

Bipolar patients TA SEX TA x SEX Covariate Error TA, treatment

Role function (PCOMP 2) P

4

F

0.27 0.20 0.15 0.79 2.99 0.12 0.39

-

1 1 1 1 1 1 1 1 41

0.52 0.34 0.07 0.03 6.03 0.80 0.06 14.26

1 1 1 1 24

2.95 0.04 0.00 6.69

1 1 1 1 16

1.94 1.17 0.15 10.60

0.53 0.21 0.08

0.09 -

-

25

3.85 0.02 1.41

1 1 1

0.07 -

17 assignment;

Global/symptoms (PCOMP 1)

F

42

Unipolar patients TA SEX TA x SEX Covariate Error

SEX, sex of patient;

P

_ _ 0.02 _ 0.001

0.10 _

All

F

1 1 1 1 1 1 1 1 42

0.43 0.01 0.62 1.10 3.97 0.07 0.64

1 1 1

3.12 0.02 0.09

P

4 1 1 1

_

1

0.05

1

1 1 1 41

_

0.09 _ _

25

_

1 1 1

_

1.61 0.14 1.91

_ _

0.005 17

DX, subdiagnosis

Patients

!f*hmo

4

Role function (PCOMP 2)

0.02

of patient

polars at 6 months is a trend at the P < 0.07 level and the effect favoring comparison treatment for unipolars at 18 months is a trend at the P < 0.09 level. Inspection of the means for males and females (Fig. 2) indicates that the bipolar treat-

(worsej

18 MONTHS

18 Months

Global/symptoms (PCOMP 1) d/

AT 6 AND

0.23 0.02 1.78 0.21 11.87 0.16 0.00 18.14

_ _ _ _ 0.001 _ 0.0001

1 1 1 1 24

4.83 0.11 0.10 13.69

0.04 _ _

1 1 1 1 16

6.46 0.12 0.03 6.12

0.02

0.001

_ 0.03

(bipolar/unipolar)

Males

All

LF

Adm

P

ment effect is exclusive to the female patients, while the negative effect on unipolars appears, to some extent, in both sexes. On the role functioning composite measure, there was no main effect of treatment, but, again,

All

j

F

;I

DIS 6mo

18mo

Females

$-3

Adm DIS 6mo lf3mo

Time Fig. 1. Example

of an outcome

over time for patients with affective disorders using mean scores on global function/symptomatology composite including an analysis by sex.

23 TABLE

3

RESULTS

OF THREE-WAY

ANALYSES

ON COMPOSITE

FAMILY

MEASURES

6 Months

AT 6 AND

18 Months

Attitude to treatment/ social support (FCOMP 1)

Attitude to patient/ burden (FCOMP 2)

Attitude to treatment/ social support (FCOMP 1)

df

F

P

df

F

P

df

F

AN patients TA SEX DX TA x SEX TAxDX SEX x DX TAxSEXxDX Covariate Error

1 1 1 1 1 1 1 1 41

0.23 0.35 0.28 2.33 0.86 0.26 0.75 14.09

_ _

1 1 1 1 1 1 1 1 41

0.00 0.26 1.10 1.75 4.53 0.00 0.40 15.23

_ _

1 1 1 1 1 1 1 1 41

1.95 0.21 1.81 0.66 0.54 0.13 0.01 13.17

(/nip&r patients TA SEX TA x SEX Covariate Error

1 1 1 1 24

0.94 0.00 3.51 8.66

1 1 1 1 24

2.63 0.05 0.37 8.21

1 1 1 1 24

2.11 0.00 0.32 6.96

Bipolar patients TA SEX TA x SEX Covariate Error

1 1 1 1 26

0.12 0.59 0.16 5.31

1 1 1 1 16

3.86 1.41 1.98 4.36

1 1 1 1 16

0.14 1.15 0.53 5.71

TA, treatment

assignment;

18 MONTHS

_ _ _ _ 0.001

_ 0.07 0.007

_ _ _ 0.03

SEX, sex of patient;

DX, subdiagnosis

we found a significant interaction of treatment assignment with subdiagnosis at both 6 and 18 months follow-up (P < 0.02 and P < 0.001 respectively). Consistent with measures of global functioning and symptomatology, we found that on measures of role functioning, the bipolar patients did better with IFI, and the unipolar patients with the comparison treatment. Independent analyses for the two diagnostic subtypes show that: (1) among unipolar patients, outcome is superior with the comparison treatment at 18 months (P < 0.04) and (2) among bipolar patients outcome is superior with IF1 achieving a statistically significant difference at 18 months (P < 0.02). Treatment effects on families. There were no main effects of treatment or treatment X sex effects on either family composite for the overall

_ 0.04 _ _ 0.0001

_ _ _ 0.009

0.07 _ _ 0.05

of patient

P

_ _ _ _ 0.001

_

Attitude to patient/ burden (FCOMP 2) df

F

1 1 1 1 1 1 1 1 41

0.63 1.35 0.06 2.70 0.64 0.12 1.05 4.44

1 1 1

1.38 0.48 3.11

P

0.01 25

_ _ 0.03

1 1 1 1 16

0.18 1.78 0.01 6.70

(bipolar/unipolar).

group of affective disorders. But there was a treatment X subdiagnosis interaction effect (P < 0.04) on the attitude to patient/burden family composite at 6 months. Bipolar families have better outcome on this composite with IFI, unipolars with comparison treatment. Independent analyses show the treatment effect for bipolars to be a trend at the P -C0.07 level. One-way analyses on individual outcome measures Based on the treatment X subdiagnosis interaction seen in the higher-order analyses, independent analyses were performed on individual outcome measures for bipolar and unipolar patients. For bipolars (Table 4) we found a trend for better global functioning associated with IF1 at 6 months (P < 0.10) but no other differences on symptom

24 Bipolors

All

(worse)

-

Pahents

Males

All

All Females

‘r

(better) Adm Dis 6mo l8mo

Adm DIS 6mo lf3mo

Adm

DIS 6mo 18mo

Unipolars

(worse.)

-

All

Patients

\ IL 11 All Females

All Males

lr

lt0 -1

It-

0

‘1\

-1

\

-2

-3-

St---O

-2

Adm

-31

DIS 6mo l8mo

\ \ \

--cl

'ki--

Adm DIS 6mo 18mo

Time Fig. 2. Example

TABLE

of an outcome

over time for bipolar and unipolar composite including

patients using mean scores on global an analysis by sex.

function/symptomatology

4

FOLLOW-UP Domain

Global

RESULTS

(EFFECT

SIZES) a FOR BIPOLAR

Measure

GAS PEF-OS

AND

UNIPOLAR

PATIENTS

Bipolar

b

Unipolar

6 months

18 months

6 months

18 months

+0.41 * + 0.31

+ 0.36 0.00

- 0.09 -0.18

-0.32 -0.32

* *

Symptomatology

Grandiosity Disorganization Withdrawal Subjective distress

+ 0.03 +0.31 + 0.21 + 0.20

+0.19 + 0.03 + 0.09 +0.06

- 0.06 - 0.23 - 0.04 -0.33 *

- 0.03 -0.31 - 0.21 - 0.20

Role functioning

RPTS RPTS RPTS RPTS RPTS

+ 0.51 * * + 0.48 * * +0.17 - 0.11 +0.16

+ 0.51 * * + 0.34 + 0.08 + 0.42 * +0.49 **

-0.33 * - 0.25 -0.34 * - 0.01 -0.31 *

-0.35 -0.43 - 0.37 - 0.38 -0.29

Family

Attitude toward patient Attitude toward treatment Burden Openness to social support

+ 0.28 + 0.20 +0.28 -0.18

- 0.36 +0.40 * + 0.09 - 0.37

-0.41 ** + 0.28 -0.11 -0.31

-0.37 * + 0.22 - 0.09 -0.03

overall social family work leisure

a All results are reported in terms of effect sizes and described according to the following: (i) ‘+‘. better better outcome with comparison treatment. (ii) Values represent analysis of variance on residual change scores are computed as the deviation from the regression line describing scores at discharge as a function (iii) Eta is a measure of effect size (in terms of the F statistic) computed as a correlation ratio that relationship between treatment assignment and outcome. The value of r* represents the amount of particular measure that is associated with treatment group membership. b PEF-OS, PEF overall severity score. * * * P < 0.01; * * P < 0.05; * P < 0.10.

* ** ** **

outcome with IFI; ‘-‘, scores. Residual change of scores at admission. expresses the degree of outcome variance on a

25

measures at either 6 or 18 months. Superior role functioning in areas of work, leisure and overall role functioning were observed at 18 months with a shorter-term effect of IF1 on social functioning and overall functioning at 6 months only. Among unipolar patients we found a trend (P < 0.10) for superior outcome for the comparison patients on both global and role functioning measures at 18 months (Table 4). Families of females affective patients in IF1 had a significantly better attitude toward the patient’s treatment at 6 months and at 18 months, as they did at discharge. However, at both follow-up points, there is improvement in family attitude toward the patient and family burden among comparison males. This was also true at discharge for attitude to the patient, though not for family burden.

Post-hospital treatment With data from both the patient and patient’s therapist, we conducted a thorough examination of post-hospital treatment including medication, individual, family and group therapy, vocational rehabilitation, and self-help support groups. With two exceptions, IF1 and comparison groups received similar types and amounts of treatment following discharge from the hospital. However, unipolar patients treated with IF1 stayed in outpatient individual psychotherapy significantly longer than did those in the comparison treatment. Among the bipolar patients, the only difference between IF1 and comparison treatment patients was the finding that at 6 months the average lithium dosage was higher for the comparison patients (1493 mg/day) than for the IF1 patients (1008 mg/day).

Achievement of the treatment goals of IFI At hospital discharge, family therapists rated the degree of family achievement of the IF1 treatment goals initially targeted for each family (Haas et al., 1988). For bipolar patients, ratings of overall achievement of IF1 goals were positively correlated with patient outcome on the global/ symptoms composite at 6 months (r[9] = 0.60, P < 0.04). There was a trend for achievement of IF1 goals to be associated with family outcome on the attitude to patient/burden composite at 6 months (r[9] = 0.50, P < 0.08) and attitude to treatment/ social support composite at 18 months (r[9] = 0.48, P < 0.10). In contrast, achievement of IF1 goals for unipolar patients did not correlate with patient outcome or family attitudes at either 6 or 18 months. We hypothesized that achievement of a specific IF1 treatment goal, ‘family support for patient compliance’, would correlate with the patient’s actual compliance with post-hospital treatments. For bipolar patients/families, there was a trend for achievement of this goal to correlate with overall post-hospital compliance at 6 months (r[9] = 0.78, P < 0.006), and with medication compliance at 18 months (r[9] = 0.52, P < 0.09). For unipolar patients/families, there was a trend for achievement of this goal to be associated with overall post-hospital compliance (r[9] = 0.55, P < 0.06) at 6 months.

Summary of results The main findings for patients with major affective disorder and their families were as follows. (1) At both 6 and 18 months, bipolar patients showed better outcome with IFI while unipolar patients did better without it. The treatment effect in bipolars is limited to females. ( (2) A similar treatment x diagnosis interaction was seen at 6 months on family attitude to the patient. (3) The positive effect of IF1 on females in the overall group of affective disorders at discharge is attenuated over time; a negative effect on males becomes evident at 18 months. Discussion Family intervention is recommended (Anderson, 1977; Lansky, 1981; Harbin, 1982; Anderson and Reiss, 1982; McFarlane, 1982; Group for the Advancement of Psychiatry, 1985) and used for inpatients very commonly. To date, there has been very limited evidence from controlled research (Wellisch et al., 1976; Wallace and Liberman, 1985) that it is effective at all, and no evidence that it affects one type of patient/family more than another. To begin to answer these questions and identify more specific directions for future research, we designed an exploratory outcome study of IF1

26

which included patients with a variety of major psychiatric disorders including schizophrenia and affective disorder. While it was originally hypothesized that IFI would be of help to most seriously disturbed inpatients and their families (irrespective of sex or diagnosis) our data suggest more differentiated outcomes. At discharge from the hospital the most striking effect of IF1 was on fern&e patients with major affective disorder. These follow-up data show that this finding for female affectives diminishes with time. But, by 18 months mules are showing a significant negative effect of IFI. The differential effect of IF1 on females and males was an unexpected, but consistent, finding. We can only speculate at this point as to possible explanations. Among certain major psychiatric disorders (e.g., schizophrenia) males show a hypersensitivity to interpersonal stimuli and family intervention may be experienced as overstimulating and stressful. Sex-related societal expectations for independence and greater performance in work roles in males may predispose them to reject an intervention that requires involvement with the family and the raising of family expectations for performance. The most striking finding for affective disorder patients at follow-up was that IF1 benefits bipolar patients but has a negative effect on unipolar patients. We can only speculate as to why this may be. Family treatments which emphasize psychoeducation and stress management such as IF1 were initially developed for use with schizophrenic patients whose illness has an important biological component and where family education about a chronic or recurring illness is particularly helpful in preventing relapse (Goldstein et al., 1978; Leff et al., 1982; Falloon et al., 1985; Hogarty et al., 1986). Bipolar illness may fit this paradigm better than unipolar illness; in our sample 67% of the bipolars had had previous episodes as compared to 52% of the unipolars; and 52% had had previous hospitalizations as compared to 38% of the unipolars. With regard to the unipolar patients, it is unclear why they should have a worse outcome when IF1 is added to standard hospital treatment. The bipolar group contained a higher percentage of females (67% as compared to 55% in the comparison group). Sex and subdiagnosis are,

therefore, to some extent confounded in our affective sample. Because we did not (in this first experiment) randomly assign patients to treatment using subdiagnoses and gender, we have no controlled test of the independent contribution of these two variables. However, statistical analyses show the sex x treatment assignment interaction to be outweighed by the diagnosis X treatment interaction at follow-up. Table 2 reveals a consistent and statistically significant interaction between IFI and subdiagnosis and does not show a treatment X sex interaction. On the other hand, within the bipolar group, the positive effect of IF1 is seen dramatically in female patients and not at all in males. The higher-order analyses on composite outcome measures assure that the results presented here are not chance findings among multiple significance tests. Results are even more impressive when consideration is given to the apparent stability of effects on progression in magnitude from 6 to 18 months. Apparent effects of IF1 could be due to a baseline confounding factor. However, the IF1 and comparison groups for both the bipolar and unipolar groups were not significantly different at baseline on such potentially important variables as socioeconomic status, prior functioning, indices of severity of illness, or incidence of severe personality disorder. Moreover, statistical methods (e.g., analysis of covariance) were used to adjust for baseline clinical differences, A previous analysis (Haas et al., 1988) suggested that no potentially confounding factors entered during the treatment phase. In evaluating outcome data at a time point long after a treatment has ended, the potential confounding variable of intervening treatment must be considered. This study left control of outpatient aftercare to the patients and their families so as to provide a clinically realistic test of the effects of an inpatient family treatment. As noted, however, with one exception post-hospital treatment did not differ between the two treatment groups in ways that could explain our results. Differences in amount of lithium were found, but the amount seems of questionable clinical significance. This leads the investigators to conclude that exposure to an inpatient family intervention during the acute phase of illness and hospitalization is associ-

27

ated with enhanced long-term outcome on clinical, but most impressive, role functioning measures. The negative results for unipolar patients are less well understood, except to note that achievement of the goals of IF1 was not linked to outcome for these patients and families. Acknowledgements We wish to thank our statistical consultant, Professor William A. Hargreaves, Ph.D., of the University of California at San Francisco. This study was supported in part by a Biomedical Research Support grant (507-RR05396) to Cornell University Medical Center, New York, a grant from the National Institute of Mental Health (MH34466), and a grant from the Norman and Rosita Winston Foundation. References Anderson, CM. (1977) Family intervention with severely disturbed inpatients. Arch. Gen. Psychiatry 34, 697-702. Anderson, CM. and Reiss, D.J. (1982) Family treatment of patients with chronic schizophrenia: the inpatient phase. In: H.T. Harbin (Ed.), The Psychiatric Hospital and the Family, Spectrum, Jamaica, NY, pp. 79-101. Beutler, L.E. and Hamblin, D.L. (1986) Individualized outcome measures of internal change: methodological considerations. J. Consult. Clin. Psychiatry 54, 48-53. Clarkin, J.F., Spencer, J.H., Peyser, J., DeMane, N., Lestelle, V., Lewis, A.B., Haas, G.L. and Glick, I.D. (1981) IF1 for Psychotic Disorders: A Manual of Inpatient Family Intervention. Unpublished manuscript. Cohen, J. (1965) Some statistical issues in psychological research. In: B.B. Wolman (Ed.), Handbook of Clinical Psychology. McGraw-Hill, New York, NY. Cohen. J. and Cohen, P. (1975) Applied Multiple Regression/ Correlation Analysis for the Behavioral Sciences. Erlbaum, Hillsdale, NJ. Davenport, Y.B., Ebert, M.H., Adland, M.L. and Goodwin, F.K. (1977) Couples group therapy as an adjunct to lithium maintenance of the manic patient. Am. J. Orthopsychiatry 47, 495-502. DiMascio, A., Weissman, M.M., Prusoff, B.A., Neu, C., Zwilling, M. and Klerman, G.L. (1979) Differential symptom reduction by drugs and psychotherapy in acute depression. Arch. Gen. Psychiatry 36, 1450-1456. Endicott, J. and Spitzer, R.L. (1972) What! Another rating scale? The Psychiatric Evaluation Form. J. Nerv. Ment. Dis. 154, 88-104. Endicott, J., Spitzer, R., Fleiss, J. and Cohen, J. (1976) The global assessment scale. Arch. Gen. Psychiatry 33, 766-771.

Falloon, I.R.H., Boyd, J.L., McGill, C.W. et al. (1985) Family management in the prevention of morbidity of schizophrenia. Arch. Gen. Psychiatry 42, 887-896. Friedman, A.S. (1975) Interaction of drug therapy with marital therapy in depressive patients. Arch. Gen. Psychiatry 32, 619-637. Glass, R.M. (1984) Psychotherapy, scientific art or artistic science? Arch. Gen. Psychiatry 41, 525-526. Glick, I.D., Clarkin, J.F., Spencer, J.H., Haas, G.L., Lewis, A.B., Peyser, J., DeMane, N., Good-Ellis, M., Harris, E. and Lestelle, V. (1985) A controlled evaluation of inpatient family intervention. I. Preliminary results of the six-month follow-up. Arch. Gen. Psychiatry 42, 882-886. Goldstein, M.J., Rodnick, E.H., Evans, J.R., May, P.R.A. and Steinberg, M.R. (1978) Drug and family therapy in the aftercare of acute schizophrenics. Arch. Gen. Psychiatry 35, 1169-1177. Good-Ellis, M., Fine, S.B., Spencer, J.H. and DiVittis, A. (1987) Development of a Role Activity Performance Scale. Am. J. Occup. Ther. 41, 232-241. Group for the Advancement of Psychiatry, Committee on the Family (1985) The Family, the Patient, and the Psychiatric Hospital: Toward a New Model. Brunner/Mazel, New York, NY. Haas, G.L., DiVittis. A.T., Levitt, M., Spencer, J.H., Clarkin, J.F. and Glick, I.D. (1986) Preliminary validation of a family attitude scale. Presented at the Annual Meeting of the American Psychiatric Association, Washington, DC. Haas, G.L., Glick, I.D., Clarkin, J.F., Spencer, J.H., Lewis, A.B., Peyser, J., DeMane, N., Good-Ellis, M., Harris, E. and Lestelle, V. (1988) Inpatient family intervention: a randomized clinical trial. II: Results at hospital discharge. Arch. Gen. Psychiatry 45, 217-224. Harbin, H.T. (1982) Family treatment of the psychiatric inpatient. In: H.T. Harbin (Ed.), The Psychiatric Hospital and the Family, Spectrum, Jamaica, NY, pp. 3-25. Hogarty, C.E., Anderson, C.M., Reiss, D.J. et al. (1986) Family psychoeducation, social skills training, and maintenance chemotherapy in the aftercare of schizophrenia. Arch. Gen. Psychiatry 43, 633-642. Hooley, J.M. (1986) Expressed emotion and depression: interactions between patients and high versus low EE spouses. J. Abnorm. Psychol. 95, 237-246. Jamison, K.R. and Akiskal, H.S. (1983) Medication compliance in patients with bipolar disorder. Psychiatr. Clin. North Am. 6, 175-192. Lansky, M.R. (1981) Family psychotherapy in the hospital. In: M.R. Lansky (Ed.), Family Therapy and Major Psychopathology. Gnme and Stratton, New York, NY, pp. 395-414. Leff, J., Kuipers, L., Berkowitz, R., Eberlein-Bries, R. and Sturgeon, D. (1982) A controlled trial of social intervention in the families of schizophrenic patients. Br. J. Psychiatry 141, 121-134. Levitt, M. (1982) The Impact of Family Intervention on the Attitudes of Key Relatives of Psychiatric Inpatients. Unpublished doctoral dissertation, Columbia University. McFarlane, W.R. (1982) Multiple-family therapy in the psychiatric hospital. In: H.T. Harbin (Ed.), The Psychiatric

28 Hospital and the Family. Spectrum, Jamaica, NY, pp. 103-129. Miklowitz, D., Goldstein, M., Nuechterlein, K., Snyder, K. and Mintz, J. (1988) Family factors and the course of bipolar affective disorder. Arch. Gen. Psychiatry 45, 225-231. O’Connell, R., Mayo, J., Eng, L., Jones, J. and Gabel, R. (1985) Social support and long-term lithium outcome. Br. J. Psychiatry 147, 272-275. Prien, R., Kupfer, D., Mansky, P., Small, J., Tuason, V., Voss, C. and Johnson, W. (1984) Drug therapy in the prevention of recurrences in unipolar and bipolar affective disorders. Arch. Gen. Psychiatry 41, 1096-1104. Rosenthal, R. and Rosnow, R.L. (1984) Essentials of Behavioral Research: Methods and Data Analysis, McGraw-Hill, New York, NY. Spencer, J., Glick, I.D., Haas, G., Clarkin, J., Lewis, A., Peyser, J., DeMane, N., Good-Ellis, M., Harris, E. and Lestelle, V. (1988) A randomized clinical trial of inpatient family intervention. III: Effects at 6-month and 18-month follow-ups. Am. J. Psychiatry 145, 1115-1121.

Vaughn, C.E. and Leff, J.P. (1976) The influences of family and social factors on the course of psychiatric illness. Br. J. Psychiatry 129, 125-137. Wallace, C.J. and Liberman, R.P. (1985) Social skills training for patients with schizophrenia: a controlled clinical trial. Psychiatr. Res. 15, 239-247. Weissman, M.M., Prusoff, B.A., DiMascio, A., Neu, C.. Gorklaney, M. and Klerman, G.L. (1979) The efficacy of drugs and psychotherapy in treatment of acute depressive episodes. Am. J. Psychiatry 136, 555-558. Weissman, M.M., Myers, J.K. and Thompson, W.D. (1981) Depression and its treatment in a U.S. urban community, 1975-1976. Arch. Gen. Psychiatry 38, 417-421. Wellisch, D.D., Vincent, J. and Rotrock, G. (1976) Family therapy versus individual therapy: a study of adolescents and their parents. In: D. Olson (Ed.), Treating Relationships. Graphic, Lake Mills, IA, pp. 275-302.

A randomized clinical trial of inpatient family intervention. V. Results for affective disorders.

This paper reports the results at follow-up of a randomized clinical trial of combining family intervention with drug treatment during hospitalization...
1MB Sizes 0 Downloads 0 Views