Predictors of Residential Independence Among Outpatients With Schizophrenia

Over the last several decades, the care of patients with schizophrenia has largely shifted from the hospital to outpatient settings. However, many of these patients are not able to live on their own in the community but require support or supervision in their place of residence (1). Although a range of residential services has been developed, surveys of psychiatric patients show a consistent preference for living arrangements that are as independent as possible (2,3,4).

Previous studies have found several variables to be associated with the ability of outpatients with schizophrenia to live independently. Younger age (5), female gender (6,7), and a shorter duration of institutional care (8) have been linked with more independent community status.

Neurocognitive abilities have also been implicated in patients' degree of residential independence. Among a group of long-stay hospital patients discharged to the community, patients' skill on a complex reaction-time task predicted their degree of community independence (8). In another study, performance on visual motor and verbal processing tasks was associated with outpatients' residential status (9). Patients' insight about their illness has been associated with whether they received residential supervision (10).

The relationship between patients' daily living skills and their residential status has also been examined. Self-care skills (11) and general functional abilities (5,7) have been associated with patients' independence. An association has not been found between either symptom severity or life satisfaction and degree of independent living (5,8,9).

Relatively few studies have assessed the independent living status of stable outpatients while also obtaining detailed measures of their social and neuropsychological functioning. In the study reported here, we evaluated patients using a battery of neuropsychological tests and assessed their independent living status, symptoms, and social functioning at follow-up two years later. The purpose of the study was to identify factors that determine the degree of residential independence among outpatients with schizophrenia who receive community-based services.

Methods

Subjects

Subjects were recruited from a public-private mental health system in Baltimore County, Maryland, during the period from July 1993 to June 1995 as part of a study of cognitive deficits and social functioning among outpatients with schizophrenia (12). Exclusionary criteria were current substance abuse, mental retardation, neurological disease, and age younger than 18 years or older than 65 years. All study participants provided informed consent.

The sample comprised 72 subjects, 50 of whom were men (69 percent). Subjects had a mean±SD age of 42.1±9.6 years. Their mean age at onset of illness was 20.9±6.1 years, and they had had a mean of 5.7±4.7 psychiatric hospitalizations. The majority of subjects—59 patients, or 82 percent—were single and had never been married; six patients, or 8 percent, were married; and seven patients, or 10 percent, were separated or divorced. Fifty-seven subjects, or 79 percent, were Caucasian; 14 subjects, or 20 percent, were African American; and one subject was Asian American. The study participants had an average of 13.6±2.1 years of education.

All but one subject was receiving antipsychotic medication at the time of residential assessment. The mean±SD dose in chlorpromazine equivalents was 539±591 mg. Thirty-one patients were receiving atypical antipsychotic medications. Subjects were assessed by their treating doctor as stable in their community functioning for at least one month before the assessment of residential independence, and no subjects had been hospitalized in the two months before the assessment.

Twenty-six patients met diagnostic criteria for schizoaffective disorder. The other patients met criteria for the following subtypes of schizophrenia: paranoid for 19 patients, undifferentiated for 18 patients, residual for seven patients, and disorganized for two patients.

Subjects were classified into three groups based on their degree of residential independence. Group 1, the most independent group, included 14 patients who lived in their own residence without any supervision for at least the previous two months. To quality for this group, patients also had to participate in constructive activity outside of the home for more than 20 hours a week. Among the patients in group 1, seven were competitively employed, five attended a prevocational day program, and two were engaged in volunteer jobs.

Group 3, the most supervised group, included 12 patients who resided in settings such as group homes and highly supervised board-and-care residences where they received direct, on-site supervision for more than six hours a day.

Group 2 included 46 patients whose living arrangements were intermediate between group 1 and group 3. Among group 2 patients, 15 lived with their parents, two lived with other nonspouse extended-family caregivers, 14 lived in apartments with drop-in supervision, 13 lived independently but did not meet the group 1 criterion of daytime activity, and two had moved to their own unsupervised, independent apartment within the previous two months.

Measures and procedures

A battery of neuropsychological tests was administered to all patients at the time of their initial recruitment. The tests included the vocabulary, arithmetic, digit span, block design, digit symbol, and picture arrangement subtests of the Wechsler Adult Intelligence Scale (WAIS-R) (13); the Wechsler Memory Scale-Revised (14); the Logical Memory test; the Rey-Osterreith Complex Figure Test (15), including the copy and recall test; the Wisconsin Card Sorting Test (16); the Trail Making Test (17); the Halstead-Wepman Aphasia Screening Test (18); and the Chicago Word Fluency Test (19).

At two-year follow-up the subjects' symptoms, social functioning, and residential status were assessed. Symptoms were rated using the Positive and Negative Syndrome Scale (PANSS) (20) by raters who showed acceptable interrater reliability on the PANSS symptom factors (r=.88 for the positive symptom factor, .74 for the negative symptom factor, and .82 for the general symptom factor). Subjects were administered the Social Functioning Scale (SFS), which asks about their abilities in seven areas—activation-engagement, interpersonal communication, frequency of activities of daily living, competence at activities of daily living, participation in social activities, participation in recreational activities, and employment or occupational activity (21).

When possible, an informant who lived with or cared for the patient was also interviewed using an informant version of the SFS, and the average ratings were used in the data analysis. Interviews with such informants were available for 38 subjects, or 53 percent.

Patients also were administered the Quality of Life Interview (QOLI), a self-report instrument inquiring about eight life domains—living situation, daily activities, family contact, social relations, legal and safety issues, health, finances, and school or work activities (22). Subjects were also rated by the first author using the Multnomah Community Ability Scale (23); all available information was used in making these ratings. The neuropsychological, symptom, and social functioning measures were selected to provide a comprehensive assessment of patients' functioning in these domains.

Data analysis

Univariate analyses of variance were used to determine the percentage of variance that was shared between patients' residential independence and each study variable. Study variables included demographic characteristics, PANSS factor scores, SFS scores, Multnomah item scores, and QOLI objective scores. Three variables were excluded from these analyses because the ratings were confounded by patients' residential status: the Multnomah scale items about money management and independence in daily life and the SFS employment rating. In addition, gender was not analyzed because it is a dichotomous variable.

Study variables that shared a significant percentage of variance (p<.05) with the residential independence variable were included in a multivariate analysis of variance (MANOVA) (24). The MANOVA was performed to determine the relative contribution of the variables in a discriminant function. We calculated canonical loadings, which are the correlations of each variable with the linear combination of variables in its set; they index the importance of each variable in relation to the residential independence variable. We also derived a classification table from the MANOVA discriminant function. All possible pairwise post hoc contrasts were computed between the residential groups using the Tukey honest significant difference test.

Results

Results of the univariate analyses of variance are shown in Table 1. Nine variables in the univariate analyses shared a significant percentage of variance with the residential independence variable (p<.05), ranging from 20.4 percent for frequency of family contact to 8.6 percent for size of social network. Combined, these variables account for 40.83 percent of the variance in residential independence. On each of these variables, the most independent group had the best functioning and the most supervised group had the most impaired functioning.

The canonical loadings from the MANOVA discriminant function indicated one significant root, suggesting that only one combination of variables would predict residential independence (χ²=46.32, df=18, p<.001). As Table 2 shows, the root has the highest loadings on frequency of family contact and competence in activities of daily living. The discriminant function analysis correctly classified 78 percent of the patients by residential independence status.

Pairwise contrasts were performed between residential groups on the dependent variables identified in the univariate analyses of variance, which are shown in Table 3. Significant differences between the most supervised and the most independent residential groups were found for eight of the nine variables.

Discussion and conclusions

In this study, the degree of residential independence of outpatients with schizophrenia was related to their frequency of family contact, performance of activities of daily living, participation in social and recreational activities and relationships, social presentation, and relative absence of negative symptoms. Combined, these variables accounted for 41 percent of the variance in residential classification within this patient group.

The importance of family contact is consistent with the family advocacy literature describing the support often provided to persons with serious mental illness by their family members (25,26). Data from this study did not identify the process by which family contact is related to patients' residential independence. Families' reaching out to patients may have enabled the patients to function more successfully, or healthier patients may have been able to maintain contact with their families more readily than less healthy patients.

The importance of activities of daily living in patients' residential independence has also been found in previous studies involving patients from different types of settings (5,11). The contribution of activities of daily living to patients' residential status within an outpatient sample highlights the importance of basic care skills in patients' ability to manage successfully on their own.

Unlike two previous studies (8,9), our study did not identify any neurocognitive variables that predicted patients' residential status. This negative finding is noteworthy because subjects received an extensive neuropsychological assessment, the results of which showed that the cognitive impairments of the study patients, as a group, were comparable to those of other samples of outpatients with schizophrenia (12).

Although negative symptoms predicted residential status, positive symptoms did not. Our data are consistent with those of previous studies showing the lack of a significant relationship between the social functioning of patients and their residual positive psychotic symptoms (27,28,29).

The limitations of this study include the cross-sectional nature of the analysis, which makes it difficult to determine cause-effect relationships among the variables. Other limitations include the use of multiple comparisons in the univariate analyses, the relatively small number of patients who were classified in the most independent and supervised groups, and the heterogeneity of the intermediate residential group.

Residential placement is a complex phenomenon that is related to the treatment system and its ability to impose residential placements, as well as to patients' housing needs (30,31). Although other populations and settings may be different from those in our study, we believe our findings are generalizable to outpatients with schizophrenia who are relatively stable; who are without comorbid substance abuse, mental retardation, or neurological disorders; and who reside in suburban communities that provide a range of residential alternatives for persons with serious mental illness.

The future of residential rehabilitation services is uncertain in a managed care environment (32). Better identification of patients who can live independently and of those who require more intensive supervision may help in planning rehabilitation services. Results from this study suggest that frequency of family contact and performance of activities of daily living are highly associated with patients' residential status. Additional research is needed to better understand how these factors may contribute to the ability of patients to live independently.

The authors are affiliated with the Sheppard Pratt Health System, 6501 North Charles Street, Baltimore, Maryland 21204 (e-mail, [email protected]). An earlier version of this paper was presented at the annual meeting of the Association for the Advancement of Behavior Therapy held November 13-16, 1997, in Miami.

Enlarge table

Enlarge table

Enlarge table