Setting and participants
The study was conducted in Setagaya and Suginami, two neighboring wards in downtown western Tokyo. These wards are predominantly residential. In April 2007, Setagaya had a population of 804,699 (385,768 male and 418,931 female), a population density of 13,854 people per square kilometer and 413,404 households. People aged 65 years and over (elderly) made up 17.1% of the population. In April 2009, Suginami had a population of 520,957 (251,465 male and 269,492 female), a population density of 15,313 people per square kilometer, 286,115 households and 18.7% of the population was elderly. The percentages of elderly in Setagaya and Suginami have been increasing rapidly, from 16.0% (Setagaya) and 16.8% (Suginami) in 2000, and are estimated to reach 25.8% (Setagaya) and 25.5% (Suginami) by the year 2030 [14].
Setagaya has 27 CCSCs and Suginami has 20. Participants in this program were staff of these CCSCs. Staff completed self-administered questionnaire surveys pre and post intervention.
Study design and procedure
This study adopted a cluster non-randomized design with a CCSC as a cluster unit. Figure 1 is a flow diagram of the study participants in each ward. First, we invited each CCSC to join the program. In Setagaya, nine of the 27 CCSCs applied for the program as did 11 of the 20 CCSCs in Suginami. These CCSCs were assigned to the intervention group and those that did not apply for the program were assigned to the control group; thus, the intervention group consisted of 20 CCSCs and the control group consisted of 27. Second, CCSCs in the intervention group were asked to nominate one staff member to attend all the program sessions. Most staff who participated in the program were chosen by the managers of each CCSC.
The program ran in Setagaya in 2007 and in Suginami in 2009. In Setagaya, in April 2007, a pre-intervention survey of the 103 professionals belonging to the 27 CCSCs (35 in the intervention group and 68 in the control group) was conducted after providing information about the study. A post-intervention survey was conducted in March 2008. In Suginami, in April 2009, the pre-intervention survey of the 88 professionals belonging to the 20 CCSCs (48 in the intervention group and 40 in the control group) was conducted after providing information about the study. The post-intervention survey was conducted in March 2010.
The participants were subsequently divided into three groups: program attendees (intervention group), program non-attendees (intervention group) and the control group. The first hypothesis was tested by comparing the outcomes from the three groups (1 = program attendees in the intervention group; 2 = program non-attendees in the intervention group; 0 = the control group). The second one was tested by comparing the intervention group as a whole with the control group (1 = the intervention group; 0 = the control group).
Ethical considerations
The Ethics Committee of the Faculty of Medicine at The University of Tokyo granted approval for the study. At the time of distribution of the pre-program questionnaire, potential participants were informed of the purpose and methods of this study and that their participation was optional. This statement, a guarantee of anonymity, and descriptions of other aspects of the cooperation requested were attached to the questionnaire. Return of the questionnaire was deemed to be consent to participate in the study. Individual participants were allocated an identifying code not known to the data manager.
Intervention program
The authors developed the intervention program that they delivered over 10 monthly sessions [10]. The themes and objectives of the sessions were based on the stages of the coalition development model proposed by Florin et al. [15]. The model explains the ongoing development of organizational collaborative relationships in stages. Because this study focused on interorganizational networks between CCSC and CBOs, session themes included "review of past activities," "understanding the significance of building networks with CBOs," "integrating the aims and visions of staff," and "learning ways to build networks with CBOs." Participants shared opinions and experiences about network building with CBOs in group discussions at each session.
"Review of past activities" was designed to clarify the strengths and weaknesses of past network building efforts with CBOs by the attendees' CCSCs. Those attending identified the types of CBOs they had contacted and described the nature of their connections. "Understanding the significance of building networks with CBOs" aimed to establish the basic motivation for building interorganizational networks with CBOs. Attendees discussed why they built networks with CBOs and when they considered it necessary to build networks. "Integrating the aims and visions of staff" discussed ways to build consensus, to enhance the recognition of the importance of network building and to create ideal situations in the community by building networks with CBOs. In "Learning ways to build networks with CBOs", the attendees shared their existing strategies in building networks with CBOs according to the types of CBOs.
The attendees evaluated the content of the program and indicated a high level of satisfaction with it [16]. Their self-efficacy based on the SCT increased after completing the program [16].
Measures
Outcome variables
Cognitive and behavioral dimensions were the measured outcomes. The cognitive dimension included recognition of the knowledge and skills used in building networks with CBOs, the ease of working with CBOs within existing networks, and the importance of building networks with CBOs. The first cognitive dimension was assessed using the subscale of "Knowledge and Skills" from the Social Worker Empowerment Scale [17]. We revised the wording of the items to fit into the context of this study. This subscale consisted of nine items with responses on a 5-point Likert scale. We conducted a factor analysis using baseline data, and confirmed the nine items loading into a single factor. Scores were summed (ranging from 9 to 45): the higher the score, the greater the recognition of knowledge and skills by the respondents. Cronbach's alpha was 0.83 as calculated from baseline data. One item was used to assess each of the second and third cognitive dimensions. The statements were "I think that the work at the CCSC will become easier through networking with CBOs" and "I think that building a network with CBOs is important to my work at the CCSC." These items were scored on a 6-point Likert scale (6 = strongly agree; 5 = agree; 4 = somewhat agree; 3 = somewhat disagree; 2 = disagree; 1 = strongly disagree). The focus for the behavioral dimension related to the involvement in interorganizational network building with CBOs. The statement was "What is the current percentage of your entire work at the CCSC that is related to network building with CBOs?" Answers could range from 0 to 100%.
Demographic variables
Data were collected on age, sex (1 = male; 2 = female), educational level (1 = high school graduate; 2 = Junior college/vocational school graduate; 3 = college graduate), years of experience in community-based clinical practice, years of experience in clinical practice in the current catchment area of the CCSC, type of profession, and average weekly working hours (1 = under 30 hours; 2 = 30-39 hours; 3 = 40-49 hours; 4 = 50-59 hours; 5 = 60-69 hours; 6 = 70 hours or more). Type of profession was categorized as "public health nurse and registered nurse" (= 1), "certified social worker" (= 2) or "care manager" (= 3).
Statistical analysis
One of the purposes of this trial was to compare the recognition of the value of building networks with CBOs and the involvement in the work of network building with CBOs among the three groups; program attendees in the intervention group, program non-attendees in the intervention group, and the control group (at personal and behavioral level). The ratio of the number included in each group was designed as a one-three-four, on the assumption that CCSCs are equally allocated to intervention and control groups, a CCSC (a unit of cluster) composed of four staff on average, with one staff member attending all the program sessions from each CCSC in the intervention group. We postulated that a significant effect would be to detect a 15% difference in the score of the recognition of knowledge and skills, which was one of the outcomes measured in this study. As a consequence, a total sample size of 176 staff was projected to provide a power of 80% with α = 0.05 to detect a 15% difference between the three groups, assuming 0.05 of intra-class correlation coefficient and 10% dropout rate. This size was sufficient to test the other hypothesis to examine the difference between two groups of the intervention group and the control group (at organizational environmental level).
In the first stage of our analysis we tested the demographic variables and baseline scores of the outcome variables to assess comparability among the groups. Each CCSC was a cluster unit and was allocated to either the intervention or the control group. We used generalized estimating equations (GEE) taking into account the extra component of variation due to the nested design. In this study, we adopted compound symmetry as the working correlation structure.
Second we assessed the main effects of the intervention on the outcome variables using GEE after adjusting for the baseline scores of the outcome variables that differed significantly among the groups at baseline. When the GEE showed a significant difference, the intervention had had a significant effect on the outcome variable.
Third we assessed the interactions of the intervention according to the baseline scores for the outcome variables by adding interactions to the analysis of the main effect using GEE in order to examine in more detail the effects of the intervention program. A statistically significant interaction meant that the intervention had different effects according to the baseline score for the outcome variables. When we found a significant interaction, the participants were divided into two groups according to the median of the baseline score of the outcome variable and the interactions of the intervention were analyzed again.
Statistical significance was set as p < 0.05 with a two-tailed test. The statistical analyses were performed using SAS ver. 9.1.