Community Mental Health Journal

, Volume 49, Issue 4, pp 419–426

Social Support Resources and Post-Acute Recovery for Older Adults with Major Depression

Authors

    • School of Social WorkUniversity of Illinois at Urbana-Champaign
  • Nancy Morrow-Howell
    • George Warren Brown School of Social WorkWashington University in St. Louis
  • Enola Proctor
    • George Warren Brown School of Social WorkWashington University in St. Louis
  • Eugene Rubin
    • Department of Psychiatry, School of MedicineWashington University in St. Louis
Original Paper

DOI: 10.1007/s10597-012-9567-1

Cite this article as:
Li, H., Morrow-Howell, N., Proctor, E. et al. Community Ment Health J (2013) 49: 419. doi:10.1007/s10597-012-9567-1

Abstract

This study assessed the relationships between older patients’ social support resources and depressive symptoms and psychosocial functioning at 6 months following a psychiatric hospital discharge. The data used in this study were extracted from a prospective study titled “Service Use of Depressed Elders after Acute Care” (National Institute of Mental Health-56208). This sample included 148 older patients who participated in the initial and the 6-month follow-up assessment. Ordinary Least Squares regression (OLS) was used to examine important social support resources in relation to older patients’ depressive symptoms and psychosocial functioning. A vast majority of patients were embedded in a social support network that consisted of acquaintances and confidants. Patients’ depressive symptoms were related to availability of a confidant and the extent to which they spent time with others. However, patients’ psychosocial functioning was not related to social support resources assessed in this study.

Keywords

Older adultsMajor depressionSocial support resources

Introduction

Major depression is a common reason for psychiatric hospitalization of older adults (Brown 2001; Colenda et al. 2002; Serby and Yu 2003). Although psychiatric hospitalization is effective in treating depressive symptoms (Zubenko et al. 1994), older patients are often discharged before completing a full course of treatment and with high levels of functional impairment (Lenze et al. 2001).

After returning home from a psychiatric hospital stay, many older patients need mental health treatment in an outpatient setting, as well as home and community-based supportive services to assist with their impaired functioning (Langa et al. 2004; Li et al. 2005; Ramana et al. 2003). Post-acute recovery marks a critical transition from inpatient care to long-term recovery (Ramana et al. 2003); and inadequate post-acute services could undermine the effectiveness of inpatient care and patient’s long-term recovery (Morrow-Howell and Proctor 1996).

Researchers have examined the relationships between social support resources and depression recovery. In an earlier study, Hinrichsen and Hernandez (1993) found that patients were less likely to recover if their caregivers were less capable of providing needed care. Of the four domains of social support resources including social network size, social interaction, availability of instrumental aid, and perceived social support, Bosworth et al. (2002) showed that perceived social support was significantly related to remission. Ezquiaga et al. (1998) found that depressed patients who received a higher level of emotional support from partners were more likely to be in remission at the 6-month follow-up. However, the influence of partners’ emotional support was no longer significant at the 12-month follow-up (Ezquiaga et al. 2004). Researchers suggested that the variation in findings may result from differences in sample characteristics, settings of treatment, recovery outcomes, and follow-up procedures (Ezquiaga et al. 1998, 2004; Bosworth et al. 2002). Studies focusing on a specific group of patients and a specific stage of recovery may provide better understanding of the relationship between social support resources and patient’s recovery.

Building on previous research, this study focused on older patients’ social support resources before hospital admission and their depression recovery outcomes during the post-acute period (i.e., the first 6 months after hospital discharge). Due to comorbidities, many depressed patients receive outpatient services from multiple health care providers including psychiatrists, psychologists, primary physicians, social workers, and clergies following their hospital discharge (Cole 2001; Li et al. 2005; Ramana et al. 2003). In order to focus on the relationship between social support resources and older patients’ recovery outcomes, we also controlled patients’ use of outpatient services at the same period of time. The specific research questions of this study were as follows:
  1. 1.

    What types of social support resources did older patient have before psychiatric hospitalization?

     
  2. 2.

    How did social support resources relate to older patients’ recovery from major depression after controlling for their socio-demographics, clinical descriptors, and use of outpatient medical and social services?

     

Methods

Sample

This study sample was drawn from a prospective study titled “Service Use of Depressed Elders after Acute Care” (NIMH-56208). The study was conducted at a large urban hospital and included 199 older adults who (1) were 65 years or older, (2) were hospitalized for treatment of depression, (3) met DSM-IV Axis 1 depression criteria, (4) were able to provide reliable information, and (5) were discharged to a community setting rather than nursing homes or other institutional settings.

The study included only older adults who were discharged from inpatient psychiatric treatment for major depression because geriatric patients’ symptoms (Serby and Yu 2003) and recovery courses (Alexopoulos et al. 1996; Lynch et al. 1999) are different from other age groups (Ezquiaga et al. 1998). Patients’ post-acute recovery was defined as the first 6 months after hospital discharge. The consent rate of the study was 73.5 %. The patients who consented (N = 199) did not differ significantly from the patients who refused in age, gender, race, or marital status.

The analysis was based on 148 patients who participated in the assessment at discharge and 6-month follow-up. At the 6-month follow-up, 19 of the original patients were admitted to a nursing home, 13 were deceased, and 19 were lost to follow-up for other reasons, for example, refusal to participate or inability to locate. Compared with the sample at discharge (N = 199), the sample at the 6-month follow-up (N = 148) did not differ in terms of age (t = 0.57, p = .57), race (χ2 = 2.25, df = 1, p = .13), gender (χ2 = 2.60, df = 1, p = .11), depressive symptoms (t = −0.26, p = .80), and psychosocial functioning (t = −0.19, p = .85). Data collection occurred between March 1997 and May 2000. This study was reviewed and approved by the Institutional Review Boards of Washington University in St. Louis.

Data Collection and Measurement

The information used in this analysis was gathered from multiple sources. Unit nurses administered Geriatric Depression Scale (GDS) and the Modified Global Assessment of Functioning Scale (GAF) routinely to assess patients’ depressive symptoms and psychosocial functioning during the index hospitalization. The medical director, a geropsychiatrist assessed patients’ psychotic symptoms from a review of each patient’s chart. A trained nurse practitioner interpreted the patient’s hospital chart to rate the level of physical illness. Trained research assistants with master’s degrees in social work (MSW) extracted medical information from patients’ hospital charts and conducted all of the face-to-face interviews with patients at discharge and the 6-month follow-up.

Dependent Variables

The dependent variables were patients’ recovery outcomes including depressive symptoms and psychological functioning. The depressive symptoms were assessed by the Geriatric Depression Scale (GDS), a 20-item, self-report instrument where higher ratings indicate more severe depressive symptoms. GDS has been used to assess older adult’s depressive symptoms with an excellent reliability (Cronbach’s alpha = 0.94) (Yesavage et al. 1983). In this study, Cronbach alpha of GDS at the 6-month follow-up was 0.90.

Psychosocial functioning was assessed using the Modified Global Assessment of Functioning Scale (GAF). The GAF consists of four factors, including psychological impairment, social skills, dangerousness, and ADL-occupational skills (Hall 1995). A higher rating indicates better psychosocial functioning. The concurrent reliability of GAF was assessed using the Zune Depression scale with a correlation coefficient of −0.73 (Hall 1995). The reliability of GAF at the 6-month follow-up was not assessed.

Independent Variables

The independent variables were three dimensions of social support resources including social network structure, received support, and perceived support. Social support resources were assessed by the OARS Social Resources Rating Scale (SRRS) (Fillenbaum 1988). The inter-rater reliability of the scale was tested (r = 0.82). (Fillenbaum and Smyer 1981).

Social network structure focused on the composition of the network, including marital status, living arrangements, number of acquaintances, and availability of a confidant whom can be trusted in or relied on. Received support focused on the extent to which patients had phone contacts and spent time with others. Perceived social support focused on patients’ evaluation of current support and anticipation of future support, including sufficiency of contacts, loneliness, and anticipated assistance. The information about social support resources was collected by research assistants during the index hospitalization and summarized in Table 2.

Control Variables

In order to rule out the potential impact of other factors on depressive symptoms and psychosocial functioning at the 6-month follow-up, this study controlled for patients’ socio-demographics, clinical descriptors, and the use of outpatient medical and social services. This selection of control variables follows the perspective that depression is a biopsychosocial phenomenon that is affected by biological, psychological, social and environmental factors (Morrow-Howell et al. 2006). Furthermore, these control variables were used in previous studies on depression recovery and social support (Bosworth et al. 2002; Ezquiaga et al. 1998, 2004; Paykel et al. 1996; Hinrichsen and Hernandez 1993).

Patients’ socio-demographics included age, gender, race, Medicaid enrollment, and geographic locations. Geographic locations had two response categories: urban and rural communities. An urban community was located in a Metropolitan Statistical Area (MSA) as defined by the US Office of Management and Budget (OMB) (2000), and a rural community was located outside an MSA.

Patients’ clinical descriptors included depressive symptoms and psychosocial functioning at discharge, psychotic symptoms, negative life events, physical health, and functional status. The ratings on depressive symptoms and psychosocial functioning at discharge were extracted from patients’ medical records in the form of a total score. The ratings used in this analysis were the last ones recorded in patients’ medical records prior to hospital discharge.

Psychotic symptoms were measured by the presence (1) or absence (0) of the following symptoms: auditory hallucinations, visual hallucinations, or delusions. The psychotic symptoms were coded as 1, if the patient reported at least one of the above listed psychotic symptoms.

Negative life events over the past 6 months following discharge were measured by the presence (1) or absence (0) of any negative events listed in the Duke Life Event Scale (Hughes et al. 1988). This scale included 14 events related to health, marital status, relationships, death, residential relocation, empty nest, retirement, and others. The reliability of this scale was not reported.

Patients’ physical health was assessed by the Cumulative Illness Rating Scale for Geriatrics (CIRS-G). The CIRS-G, a modified version of the Cumulative Illness Rating Scale (CIRS), is designed to estimate the severity of physical health problems (Miller et al. 1991). A lower rating indicates better physical health status. The intraclass correlation coefficients reliability of CIRS-G ranges from 0.78 to 0.81 (Hudon et al. 2005). Based on information in patients’ hospital records, a trained nurse practitioner rated and scored the conditions of each organ system.

Patients’ functional status at discharge was assessed using the Multidimensional Functional Assessment Questionnaire (MFAQ) (Fillenbaum 1988). Patients were asked to rate their abilities to perform six activities of daily living (ADL), including feeding, dressing, grooming, bathing, toileting, and getting in or out of bed. Patients were also asked to rate their abilities to perform seven instrumental activities of daily living (IADL), including walking, using the telephone, shopping for groceries, preparing meals, managing money, housekeeping and taking medications. A higher rating indicates a lower level of functional impairments. The multidimensional functional assessment is reliable (r = 0.87) and commonly used assessment for older adults’ functioning (Fillenbaum and Smyer 1981).

After hospital discharge, many older adults received medical and social services. This follow-up care included psychiatrists, primary care physicians, and/or mental health specialists, as well as social service providers. The outpatient services were categorized into five groups: mental health, physical health, home-based supportive services, community-based supportive services and hospital readmission. Mental health service providers included psychiatrists, psychologists, counselors, psychiatric social workers and primary care physicians for mental health reasons. Physical health service providers included primary care physicians for physical health reasons. Home-based supportive services included home care services and Meals-on-Wheels. Community-based supportive services included adult day services, senior centers, socialization services, and transportation services. To assess the use of the first four groups of services, patients were asked to report whether they used any services listed at the 6-month follow-up (yes/no). Finally, the fifth group, hospital readmission over the past 6 months, was assessed by asking patients whether they experienced any hospital readmission after the index hospital discharge (yes/no). The information about the control variables is summarized in Table 1.
Table 1

Patients’ demographics, clinical descriptors, and use of services (N = 148)

Variables

n (%)

Mean (Std)

Dependent variables

  

GDS at the 6-month follow-up

 

11.7 (7.1)

GAF at the 6-month follow-up

 

70.7 (11.3)

Control variables

Socio-demographics

  

Age

 

75.7 (7.0)

Gender

  

 Female

108 (73.0)

 

 Male

40 (27.0)

 

Race

  

 White

124 (83.8)

 

 Black

24 (16.2)

 

Medicaid enrollment

  

 Yes

11 (7.4)

 

 No

137 (92.6)

 

Geographic locations

  

 Rural

29 (19.6)

 

 Urban

119 (80.4)

 

Clinical descriptors

  

GDS at the discharge

 

12.3 (6.8)

GAF at the discharge

 

40.4 (8.3)

Psychotic symptoms

  

 Yes

45 (30.4)

 

 No

103 (69.6)

 

Age of on-set

  

 ≥ 60

79 (58.1)

 

 < 60

57 (41.9)

 

Negative life events

  

 Yes

78 (53.1)

 

 No

69 (46.9)

 

Physical health conditions

 

9.6 (3.2)

Areas of dependency needs

 

20.5 (4.7)

Use of medical and social services

  

Mental health services

  

 Yes

138 (93.2)

 

 No

10 (6.8)

 

Physical health services

  

 Yes

136 (91.9)

 

 No

12 (8.1)

 

Home-based supportive services

  

 Yes

86 (58.1)

 

 No

62 (41.9)

 

Community-based supportive services

  

 Yes

90 (60.8)

 

 No

58 (39.2)

 

Hospital readmission

  

 Yes

94 (36.5)

 

 No

54 (63.5)

 

Data Analyses

Descriptive statistics including percentages, means, and standard deviations were used to address the question related to types of social support received by the patients. Given the small sample size and a relatively large number of social support and control variables, bivariate analysis including correlation and t tests were used to select the important variables that were related to depressive symptoms and psychosocial functioning. Only those social support and control variables that were related to patients’ outcomes at a probability level of 0.10 or below were included in the regression analysis. This selection approach is important to building a parsimonious model and reducing multicollinearity, and has been used in the similar studies (Bosworth et al. 2002; Ezquiaga et al. 1998, 2004; Paykel et al. 1996). Finally, Ordinary Least Squares regression (OLS) was used to examine the relationship between patients’ social support resources and depressive symptoms and psychosocial functioning. OLS regression is a commonly used multivariate analysis that examines the effect of a unit change in the independent variable on the dependent variable after controlling for other variables. The significance level for the OLS regression analysis was set at less than or equal to 0.05. The analyses were conducted using SAS.

Missing observations were evident from hospital records at discharge and at the 6-month follow-up interviews. At discharge, missing observations occurred in the ratings of depressive symptoms (n = 34) and psychosocial functioning (n = 8). A “hot decking” procedure was used in this study to impute the missing values to patients’ depressive symptoms (GDS) and psychosocial functioning (GAF) at discharge (Allison 2002). Using hot decking imputation, variables were sorted based on correlation, and missing values were then replaced with the values taken from matching subjects. The missing data at the 6-month follow-up were not imputed.

Results

Characteristics of the Sample

Table 1 describes the characteristics of older patients in the sample. The mean age of respondents was 75.7 (SD = 7.0) years ranging from 65 to 98. A majority of the respondents were female (73.0 %) and white (83.8 %). All of the older patients had Medicare, while less than one-tenth were enrolled in Medicaid (7.4 %). In addition, more than three-quarters resided in urban communities (80.4 %).

With respect to clinical characteristics, approximately one-third of the patients were diagnosed with psychotic symptoms (30.4 %). A majority of patients reported late onset depression (58.1 %). In addition, slightly more than one half of the patients experienced more than one negative life events in the past 6 months (53.1 %).

After discharged to their home, a vast majority of older patients used medical and community-based home health care services. More specifically, 93.2 % used mental health services, and 91.9 % received medical services for their physical health conditions. More than one-third of patients were readmitted to hospital for treatment (36.5 %). In addition, more than one half received in-home (58.1 %) and out-of home supportive services (60.8 %).

Bivariate analyses were conducted to examine the relationship between patients’ characteristics and GDS and GAF scores at the 6-month follow-up. Patients’ GDS scores at the 6-month follow-up were related to the GDS scores at discharge (r = 0.34, p < .0001), negative life events (t = −3.41, p = .001), psychotic symptoms (t = −2.63, p = .01), functioning (r = −0.14, p = .09), and hospital readmission (t = −1.78, p = .08). Meanwhile, patients’ GAF scores at the 6-month follow-up were related to the GAF at discharge (r = 0.29, p = .0003), psychotic symptoms (t = 2.48, p = .02), functioning (r = 0.24, p = .003), use of home-based social services (t = 2.24, p = .03), and hospital readmission (t = 3.60, p = .0004).

Patients’ Social Support Resources Before Hospitalization

Table 2 profiles patients’ social support resources before hospitalization. Less than one half of patients were married (40.6 %), but over one half lived with someone (58.8 %). In addition, nearly three quarters of patients (74.3 %) had three or more acquaintances, and most patients reported having one or more confidants (93.2 %).
Table 2

Patients’ social support resources (N = 148)

Variables

n (%)

Social network structure

 

Marital status

 

 Married

60 (40.6)

 Not Married

88 (59.4)

Living arrangement

 

 Alone

61 (41.2)

 With some one

87 (58.8)

Number of acquaintances

 

 5+

89 (60.1)

 3–4

21 (14.2)

 1–3

36 (24.3)

 None

2 (1.4)

Availability of confidants

 

 Yes

138 (93.2)

 No

10 (6.8)

Received social support

 

Phone contacts

 

 7 times or more

63 (42.6)

 2–6 times

47 (31.8)

 Once

23 (15.5)

 Not at all

15 (10.1)

Spent time with others

 

 7 times or more

12 (8.1)

 2–6 times

60 (40.5)

 Once

42 (28.4)

 Not at all

34 (23.0)

Perceived social support

 

Sufficiency of the contacts

 

 As often as wants to

63 (42.6)

 Too little

85 (57.4)

Loneliness

 

 Almost never

35 (23.6)

 Sometimes or often

113 (76.4)

Anticipated informal assistance

 

 No one will help

14 (9.5)

 Someone will help now and then

14 (9.5)

 Someone will help for a short term

35 (23.6)

 Someone will help as long as needed

85 (57.4)

A vast majority of patients were in contact with others. Over three quarters of patients had phone contact with family members or friends (89.9 %). More than three quarters of patients spent time with others (77.0 %). Although, more than one half of patients stated that they did not see their family members or friends as often as they wanted to (57.4 %), and one-third reported feeling lonely quite often (39.9 %), a vast majority of patients believed their informal support networks would provide assistance when needed (90.5 %).

Bivariate analyses were conducted to examine the relationship between patients’ social support resources and GDS and GAF scores at the 6-month follow-up. Patients’ GDS scores at the 6-month follow-up were related to the number of acquaintances (r = −0.24, p = .003), number of times spent with others (r = −0.26, p = .002), having confidants (t = 2.81, p = 0.006), loneliness (t = −2.64, p = .009), and anticipated informal assistance (r = −0.24, p = 0.003). However, patients’ GAF scores at the 6-month follow-up were related to whether patients were living alone (t = −2.29, p = .02).

Relationships Between Patient’s Outcomes and Social Support Resources Before Hospitalization

Based on the results of bivariate analyses, ten variables were included in the regression test on patients’ depressive symptoms at the 6-month follow-up. As shown in Table 3, this model was statistically significant (F = 6.78, N = 140, df = 10, p < .0001) and explained 34.5 % of the variance in the rating of depressive symptoms. Of the types of social support resources, availability of a confidant and spending time with others was significantly related to depressive symptoms. Patients who had a confidant (p = .02) were more likely to report lower ratings of depressive symptoms than those who did not have a confidant. In addition, patients who spent more time with others reported lower ratings of depressive symptoms (p = .05). Of the control variables, depressive symptoms at discharge (p = .01), psychotic features (p = .008), and negative life events (p = .003) were significantly related to patients’ ratings of depressive symptoms at six months.
Table 3

OLS results on factors related to patients’ depressive symptoms (GDS) at the 6 months (N = 140)

Variables

B

SE

95 % CI

Clinical descriptors

 GDS at discharge

0.21**

0.08

0.045 to 0.376

 Negative life events

3.36**

1.09

1.203 to 5.522

 Psychotic feature

3.05**

1.12

0.829 to 5.263

 ADL and IADL Functioning

−0.10

0.11

−0.322 to 0.126

Use of medical and social services

 Hospital readmission

0.79

1.13

−1.439 to 3.023

Social support resources

 Number of acquaintances

−0.48

0.65

−1.763 to 0.795

 Spending time with others

−1.22

0.61

−2.423 to −0.027

 Having confidants

−5.01*

2.19

−9.338 to −0.690

 Loneliness

1.44

1.22

−0.985 to 3.857

 Anticipated informal assistance

−3.20

1.87

−6.904 to 0.495

 Model

R2 = 34.5, F = 6.78, p < .0001

p < .05; ** p < .01

Based on the results of bivariate analyses, six variables were included in the regression test on patients’ psychosocial functioning at the 6-month follow-up. As shown in Table 4, this model was statistically significant (F = 6.88, df = 6, N = 148, p < .0001) and explained 22.6 % of the variance in psychosocial functioning. However, no individual type of social support was significantly related to psychosocial functioning. Of the control variables, patients’ psychosocial functioning at discharge (p = .02), psychotic symptoms (p = .04), and hospital readmission (p = .003) were significantly related to psychosocial functioning at six months.
Table 4

OLS results on factors related to patients’ psychosocial functioning (GAF) at 6 months (N = 148)

Variables

B

SE

95 % CI

Clinical descriptors

   

 GAF at discharge

0.26*

0.11

0.042 to .475

 Psychotic symptoms

−3.95*

1.93

−7.769 to −0.137

 Functioning

0.28

0.20

−0.105 to 0.668

Use of medical and social services

 Use of Home-based services

−2.94

1.86

−6.620 to 0.741

 Hospital readmission

−5.43**

1.80

−8.992 to −1.864

Social support resources

 Living with someone

2.33

1.80

−1.222 to 5.881

 Model

R2 = 22.6, F = 6.88, p < .0001

p < .05; ** p < .0001

Discussion

This study assessed older patients’ social support resources and the relationship between social support resources and depression recovery outcomes at the 6-month following a psychiatric hospital discharge. This study found that a vast majority of patients were embedded in a social support network that consisted of acquaintances and, more importantly, confidants. In addition, most patients believed that informal caregivers would provide needed assistance. However, there were notable areas of weakness in patients’ social support resources. For example, more than one-third of patients reported that they frequently felt lonely, and over one-half complained that they did not see members of support networks as often as they would like.

Regarding the relationship between social support resources and recovery outcomes, the study findings suggest that social support resources contribute to recovery from depression and that the influence of individual aspects of social support resources varied (Bosworth et al. 2002; Ezquiaga et al. 1998, 2004; Hinrichsen and Hernandez 1993). More specifically, spending more time with others and having confidants were significantly related to a lower rating on depressive symptoms. These findings support the importance of social interaction and engagement in promoting an individual’s psychosocial well-being (Lin et al. 1999). It is possible that spending time with others may help reinforce the social roles of the patient in the networks and provide a sense of value, bonding, and attachment, which may lead to the reduction of social isolation and distress (Thoits 1995). Having confidants might also minimize patients’ depressive symptoms by validating patients’ feelings about the situation, affirming personal worth and providing hope (Lin et al. 1999). The confidant may best provide this information because a confiding relationship is often characterized by emotional intensity, intimacy, and reciprocity (Thoits 1995).

However, the relationship between psychosocial functioning and living with somebody was no longer statistically significant after controlling for other variables. The GAF has been used to monitor and assess the effectiveness of depression treatment (APA 2000). However, researchers have found that the functioning is more closely related to clinical diagnoses than to social functioning (Moos et al. 2002), especially in the areas of social support (Jones et al. 1995; Phelan et al. 1994). Measured by GAF, psychosocial functioning was a multidimensional concept including psychological impairment, social skills, dangerousness, and ADL-occupational skills. It is possible that the total GAF score, a sum of the dimensional scores, could mask score variations in each dimension (Jones et al. 1995). It is also possible that social support resources assessed in this study may be too weak to influence patients’ GAF scores, given the severity of patients’ conditions.

In addition to social support resources, the study results showed that some control variables were also related to patients’ depressive symptoms and psychosocial functioning at the 6-month follow-up. Consistent with findings of other studies, our results indicated that patients’ clinical descriptors including depressive symptoms and psychosocial functioning at discharge and negative events were the main predictors of patients’ recovery outcomes at the 6-month follow-up (Ezquiaga et al. 1998). In general, this finding suggested that patients with more severe depression might take longer time to recover. For example, psychotic depression is a more severe type of major depression that is less responsive to treatment and results in a longer recovery process (Vythilingam et al. 2003). Consistent with this prognosis, patients with psychotic depression in our study experienced higher levels of depressive symptoms and lower levels of psychosocial functioning. The finding that patients’ depressive symptoms were related to their experiences of negative life events supported the link between negative life events and depressive symptoms (Kraaij et al. 2002; Blazer 2003). It is likely that the negative life events added additional psychological distress to the patients’ depressive symptoms (Paykel et al. 1996).

Finally, hospital readmission was also related to patients’ psychosocial functioning. Due to the considerable medical comorbidity, over one-third of the patients in this study experienced one or more hospital readmission at the 6-month follow-up. Inpatient treatment is used to stabilize patients’ acute mental and physical health conditions, and often followed by rehabilitation and continued recovery outside the hospital (Woo et al. 2006). It is possible that a hospital readmission no matter medical or psychiatric could slow down or even push back patients’ recovery from the index psychiatric hospitalization.

Several limitations must be considered in understanding the study results. First, this study was conducted at a gero-psychiatric unit of a single hospital and used relatively restricted sample inclusion criteria. Although these restrictions affect the generalizability of study findings, they enhance the internal validity of the study by limiting the impact of potential confounding factors such as variations in diagnosis, inpatient treatment, and discharge status of older patients on recovery outcomes. Second, patients’ social support resources were assessed while they were at hospital. The information reflected only the social support resources that patients had before their hospitalization and might not capture the potential change in patients’ social support resources during the post-acute recovery period. Third, this study controlled the use of outpatient medical and supportive services. However, these variables were measured dichotomously. This approach might not have captured the dynamic circumstances while using these services, given the concern about the quality of these services (Ramana et al. 2003). Fourth, although this study used a prospective design, the association between social support resources and recovery outcomes was still correlational. However, this study was strengthened by its focus on older patients, post-acute recovery, and specific social support resources; its control over patients’ use of outpatient services, and its use of multiple recovery outcomes.

Despite the limitations, this study has implications for outpatient mental health practice. Given its importance in promoting better depression recovery outcomes, the older patient’s social support resources should be carefully assessed and enhanced. The use of multidimensional assessment could help mental health specialists to develop a comprehensive understanding of patients’ social support resources. Furthermore, the association between patients’ post-acute depressive symptoms and spending time with others and having a confidant suggests treatment possibilities in psychosocial interventions implemented with older patients and their families. Patients’ clinical descriptors were the major predictors of recovery outcomes. To expand the clinical gain from inpatient treatment, outpatient treatment should be carefully planned, implemented, and monitored. A special attention should be given to the patients with more severe health and mental health conditions.

To conclude, depressed older patients in this study had relatively strong social support resources. Some social support resources are related to depression symptoms (GDS), but are less so to psychosocial functioning (GAF). Psychosocial treatment plans that address the important social support resources could lead to better recovery outcomes.

Copyright information

© Springer Science+Business Media New York 2012