Study design
A community-based randomised pilot trial was conducted to evaluate the feasibility and the effectiveness of a 12-week village-based intervention for late-life depression. Yeoncheon County is a rural area with a population of 44,187 (10,735 older adults), and has two towns and 96 small villages [16]. Two villages in Yeoncheon were randomly selected as the intervention group (Gungpyeong) and active control (Baekgui), respectively. These two villages contain similar numbers of older adults [16], and the geographical distance between them is 30-minute travel by car. The trial was mainly coordinated by the Yeoncheon CMHS team from June 2017 to March 2018. Our late-life depression intervention was developed for pragmatic use in a community setting, and the CMHS staff, including nurses and social workers, administered the protocol as study assistants. The research team trained the CMHS staff to follow the study procedure in a standardized protocol. Although this study was not blinded, the residents in the two villages were not informed about the group allocations. Furthermore, although most of the assessments were performed by researchers who were also blinded to the group allocation, some were done by the staff who managed the program.
Participants
Although all older residents in the two villages (n = 451) received either the intervention treatment or standard care, the effectiveness of the protocol was examined in a subset of representative older residents selected using an age- and sex-stratified random sampling method based on population data for Yeoncheon for 2015 (n = 160) [16]. At screening, subjects with significant sensory deficits or medical illnesses that would substantially affect delivery of assessments were excluded. The sample size required for measurement of the primary outcome measure, i.e., scores on the Korean version of the Geriatric Depression Scale – Short Form (SGDS-K) [17], was determined a priori based on a previous preventive study; a sample size of 142 (71 per group) was needed to detect an inter-group difference in change in SGDS-K score with a power of 0.80 and a two-sided alpha of 0.05 (Cohen’s d = 0.48) [9]. Based on our previous experience, a 10% attrition rate was projected, which increased the required sample size to 160 (80 per group).
All participants were fully informed of the study aims and methods and informed consent was obtained before the screening interview. The study was approved by the Institutional Review Board of Seoul National University Hospital (IRB No. 1807-135-961) and was registered in a clinical trial registry (registration NCT04013165).
Intervention
A 12-week village-based intervention for late-life depression was tested, with the aim of reducing depressive symptoms and suicidal risk via a multilevel strategy delivered within the existing community-based mental health care framework. The program was conducted by CMHS staff and healthy residents living in the village; 12 healthy residents voluntarily participated in the program as local committee members. The program comprised case management according to individual risk level and group-based activities; the protocol is detailed in Figure 1.
Individual case management
Case management according to individual risk level (at risk, risk, or high-risk) was conducted by both CMHS staff and local committee members. The at-risk group was defined as having one or more sociodemographic risk factors (living alone, chronic medical illness, and heavy drinking), but with an SGDS-K score below 6. The risk group included those with a score of 6 or 7 on the SGDS-K and no history of suicidality (suicidal ideation, plans, or attempts), while individuals in the high-risk group had SGDS-K scores of 8 or more and/or a history of suicidality. Risk stratification was informed by the CMHS database, including the SGDS-K and suicidal risks, which had been screened in all the elderly in Gyeonggi Province (a larger administrative district including Yeoncheon) every year as a regional suicide prevention project.
Local committee-guided care was delivered by matching a committee member with 4~5 older residents in all risk groups who lived in the close distance with the committee member. Committee members visited the older individuals as often as possible (at least once a week). As committee members and participants were old neighbors, natural conversation on the topics of daily life or asking after each other were encouraged instead of any structured interview. They used “mood stickers” depicting facial expressions corresponding to “good”, “so-so”, or “bad”. Older adults who indicated a worsening mood or suicidal ideation were quickly referred to the CMHS team, who also engaged in case management of the older adults. According to their protocol, the higher the risk of suicide, the more intensive was the management, which could include frequent face-to-face visits. High-risk individuals were also referred to the psychiatric service, if deemed necessary (Figure 1).
The local committee members and CMHS team had biweekly meetings, at which the local committee-guided care supervised by the CMHS team and difficult cases were discussed. Individuals in charge of other community services, such as community nurses, social service workers, and policemen, were also invited to the meetings, with the aim of providing additional community services in certain cases (e.g., a disabled elderly man with poor nutrition was referred to a community meal program, and a woman living alone was registered on the 911 service and an emergency alarm system was installed in her home).
Group-based program
The group-based program involved eight weekly sessions, which were mainly aimed at enhancing interpersonal networks and community cohesion; the sessions were open to all community-dwelling older adults, regardless of risk status. The CMHS staff supervised the 2-hour-long sessions, which included 15 to 20 attendees. The first half of the group program (sessions 1–4) was used to match participants each other, allow them to get to know each other, and extend their social networks; during the second half of the program (sessions 5–8), the participants acquired skills to be aware of at-risk neighbors and to encourage help-seeking (Figure 1).
Active control
In the control group, the CMHS provided standard care for high-risk older adults who had suicidal ideation or severe depressive symptoms (SGDS-K ≥ 10) based on the CMHS database; high-risk elderly participants were referred to the case management team and the CMHS staff met with them once a month on average. At the end of the study, we introduced the control group to the intervention program.
Measures
Blinded researchers evaluated the patients’ demographics, clinical data, depressive symptoms, and suicidality, at baseline and at the end of the program; some of these assessments were conducted by the CMHS team.
Primary outcomes
We assessed changes in SGDS-K scores as the primary outcome. Scores on the SGDS-K are highly correlate with those on the original GDS, such that it has high internal consistency, and high content and differential validity [17]. SGDS-K scores corresponding to severe depression (≥ 8) were also noted [17]. Major depressive disorder, based on the DSM-IV criteria [18], was also assessed in all participants during a structured clinical interview using the Korean version of the Composite International Diagnostic Interview (K-CIDI) [19]. Suicidality (suicidal ideation, plans, or attempts) was examined using the Suicide Prevention Multisite Intervention Study on Suicidal Behaviors (SUPRE-MISS) [20].
Secondary outcomes
Secondary outcomes included social networks, functional status, and global cognition. We used a validated Korean version of the Lubben Social Network Scale (LSNS-K), which was developed specifically for older adults, to assess the quantity and frequency of participants’ social interactions with their relatives and friends [21, 22]. Daily functional status was evaluated using the Seoul-Instrumental Activities of Daily Living (S-IADL) instrument [23]. and global cognitive function was assessed using the Korean version of the Mini-Mental State Examination (MMSE-KC), which is part of the neuropsychological battery of the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD-K) [24].
Data Analysis
Baseline characteristics were compared between groups using Student’s t-test for continuous variables and the chi-square test for categorical variables. Linear mixed models including the factors of intervention group (reference: control), time (reference: baseline), and their interaction, were used to examine group differences in changes in primary and secondary outcomes from baseline to follow-up. The models were adjusted for age, sex, years of education, and type of medical insurance (Medicare or Medicaid). Logistic regression models were used to analyze severe depressive symptoms (SGDS-K ≥ 8; indicative of major depression) [17], DSM-defined major or minor depression, and suicidality outcomes at the follow-up assessment by group. All models were adjusted for age, sex, years of education, type of medical insurance, and baseline outcomes. We repeated these analyses after dividing the participants into low (SGDS-K < 6) and high (SGDS-K ≥ 6) depressive symptom score groups at baseline.
All analyses were conducted using SPSS software (ver. 25.0; SPSS Inc., Chicago, IL, USA). All statistical tests were two-tailed, and p < 0.05 was deemed to indicate statistical significance.