A school-based cluster randomized controlled trial (RCT) was conducted to test the effectiveness of the program. Contamination of the information provided across groups within the same school may have occurred. Therefore, the RCTs of schoolteachers could use schools as clusters . The schools were randomly divided into intervention and control groups. After the schools signed up for the study, the schoolteachers at these schools signed up individually.
The program, which was offered online on an individual basis, could only be watched by the intervention group. The outcome measures were evaluated using a self-administered web survey by comparing the two groups at three time points: baseline (T1), post (T2), and one month later (T3). Ethical considerations allowed the control group to participate in the same program after the final T3 response. No major changes were made to the methods after trial commencement. The trial findings were reported in a subsequent publication in accordance with the Consolidated Standards of Reporting Trials (CONSORT).
Requests for research participation were sent to 322 public elementary schools in three Japanese prefectures. Only schools that were willing to cooperate were included in the study.
A maximum of 10 schoolteachers participated from each school. The inclusion criterion was full-time schoolteachers, and the exclusion criteria were difficulty in understanding Japanese and being unable to conduct web surveys or operate video watching. The sample size was determined with reference to 50-min video programs for improving schoolteachers’ mental health literacy in Japan [21, 25]. The effect size was 0.55–0.97. Because the program of this study was shorter in duration, we assumed an effect size of 0.4. With the estimated number of schools being 30, α = 0.05, β = 0.20, and ICC = 0.02, this would yield a required study population of 221 participants. The sample size was set at 244, with a dropout rate of approximately 10%.
The program was called the WATASHI-KOKO (“We are here” in Japanese) PROGRAM. It involved a 30-min video-based e-learning program for elementary schoolteachers, which was designed to help them support children of parents with mental illness. It aimed to help schoolteachers gain basic knowledge about mental illness and children of parents with mental illness (Purpose 1), recognize children in need of support (Purpose 2), and gain confidence in supporting them (Purpose 3).
The program was created with the support of “KODOMO-PEER” (“children’s companions” in Japanese), which is the largest self-help group for adult children of parents with mental illness in Japan. The program content was created based on a childhood needs survey that received responses from adult children of parents with mental illness . First, two core members of KODOMO-PEER and the first author of this current paper created a draft program content for a year based on the survey results . Next, 13 core members of KODOMO-PEER and seven multidisciplinary professionals, including schoolteachers, a public health nurse, psychiatric nurses, and social workers, revised the draft program content through discussions held in four meetings every three weeks (8 h in total) and via email. Most of their feedback was reflected in the video content.
To gain basic knowledge about mental illness (Purpose 1), the program content was created with reference to previous research on a mental health literacy education program for schoolteachers [25, 27]. The program included types of mental illness and their main symptoms; furthermore, it reflected the fact that mental illness affects 1 in 5 people in their lifetime and that half of those affected by mental illness develop the illness by their mid-teens (No.3 of Table 1).
To gain basic knowledge about children of parents with mental illness (Purpose 1), the program content was created based on a survey that received responses from adult children of parents with mental illness . The survey found that, when in elementary school, 78% had experienced “anxious feelings,” 51% experienced “physical and mental problems of their own,” 63% experienced “constant fights between adults at home,” 51% experienced “attacks from parents,” 32% experienced “lack of laundry and cleaning,” 19% experienced “insufficient provision of food,” and 79% reported that they experienced difficulties with their parents. The program thus included children’s life situations and feelings as specific examples, along with the survey findings (No. 3 of Table 1).
To help schoolteachers recognize children in need of support (Purpose 2), the program content was developed based on the survey that received responses from adult children of parents with mental illness . In the survey, 92% of respondents indicated that they did not consult with their schoolteachers during elementary school. Thus, the following signs were highlighted as being important for recognizing a child in need of support: no parent shows up for parent-teacher conferences or class visits, a child is often late or absent from class, a child forgets many things, a child has trouble concentrating on his/her studies, or a child has poor grades. The program included these signs (No. 4 of Table 1).
To help schoolteachers gain confidence in supporting children in need of support (Purpose 3), the program content was developed based on the survey . In an open-ended text response part of the survey, where the adult children of parents with mental illness were asked to state what they appreciated about their schoolteachers, the following responses were provided: staying close to the children, talking to them, listening to them, being kind and friendly, responding to their parents when they were not feeling well, allowing them to go to the school nurse’s office, and being understanding to them and their families. Next, past research has shown that many children of parents with mental illness report a lack of trustworthy adults in their childhood, which remains a barrier to building trusting relationships even in adulthood . The program, therefore, included information about how children often want their schoolteachers to treat them based on the survey responses and how they can be trustworthy adults for children. Moreover, based on the opinions of the schoolteachers who were research members, the program included the following suggestions for schoolteachers to help them deal with a parental mental illness situation at school: consult with other schoolteachers instead of worrying by themself, create an environment where it is easy to consult other schoolteachers, and persistently involve oneself in the situation without giving up—even if the relationship with the parents does not improve. The program included information about how schoolteachers could deal with the situation at school. Furthermore, it discussed life support services available for people with mental illness and provided examples (No. 5 of Table 1). For instance, the school discussed how to support, including the help of school social workers, and introduced home nursing in collaboration with other agencies.
Theoretically, the program content was also based on a four-tiered approach for delivering professional development programs related to parental mental disorders . In this approach, the foundation of professional learning lies in reducing the public stigma that professionals often experience toward persons with mental illness. Because narrative messages have been reported to be effective as anti-stigma messages , stories of adult children of parents with mental illness were presented in this program (No. 3–6 of Table 1). The next step in the approach is to raise professional awareness of the needs of both children and parents. This current program included children’s needs (No.3 of Table 1), with reference to a survey of childhood experiences in schools . The third step in the approach includes content specific to professionals. Accordingly, the current program included content focused on schoolteachers (No. 5 of Table 1).
In accordance with the learning process, the content structure of the program was based on the Transtheoretical Model (TTM) of health behavior change by Prochaska et al.  and Gagné’s instruction . The TTM model divides the process of changing behaviors into stages and uses effective approaches for each stage . The stages are pre-contemplation, contemplation, preparation, action, maintenance, and termination . Since the existence of children of parents with mental illness is not well recognized in Japanese society , we decided to focus on the stages of pre-contemplation and contemplation and included consciousness-raising, emotional arousal, and self-re-evaluation as effective approaches. To raise awareness about children of parents with mental illness, we included knowledge and objective information about the relevant illnesses and the discussed children (No.3 of Table 1); for emotional arousal, we included storytelling from a grown child (No.6 of Table 1). For self-re-evaluation, we asked the participants the following question: “Have you ever met a child like this?” (No. 1 of Table 1).
Gagné’s instruction is an effective process when teaching others . The learning process is, first, about getting attention (No. 1 of Table 1). This program informed schoolteachers about children in need of support, asking them, “Have you ever met a child like this?” and indicating the percentage of children of parents with mental illness as well as the percentage of such children who did not consult others. The next step involved informing the schoolteachers about the program goals (No. 2 of Table 1). Then, the schoolteachers were provided with new knowledge and ways to support the children (No. 3–5 of Table 1). The feedback process introduced the child’s narrative story; this included experiences the child had had with schoolteachers (No. 6 of Table 1). Finally, as part of the process of assessing learning outcomes, after watching the video, the participants were asked what they had learned and how they would apply the acquired knowledge to their future performance (No. 7 of Table 1).
The specifics of the program are shown in Table 1.
The outcomes were measured at the individual level. The questionnaire was developed by multidisciplinary experts who were research members of this project. The pre-test was conducted among 18 elementary schoolteachers. After watching the video of the program, they were asked to answer the questionnaire and also asked to indicate any points where they had found it difficult to understand sentences or words. Consequently, a few points were raised by the schoolteachers, and some of the wording was corrected.
Confidence in supporting children
The Sense of Coping Difficulty subscale was used to assess confidence in supporting children (Purpose 3 of the program). The Sense of Coping Difficulty/Possibility Scale refers to analogous concepts of self-efficacy, which is the expectation that professionals can deal with difficult situations [32, 33]. The scale has been tested for reliability and validity and consists of two subscales: Sense of Coping Difficulty and Sense of Coping Possibility [32, 33]. In this study, the total score of the Sense of Coping Difficulty subscale at one-month post-intervention (T3) as a primary outcome measure was calculated. The Sense of Coping Difficulty subscale consists of five items. This study asked about children and parents with mental illnesses. The five items were: “It is difficult to intervene because I feel that intervening will isolate children or parents from the community,” “I do not know how to get involved if the children or parents refuse to meet with me,” “I do not know how to get involved because the children or parents have special problems,” “It is difficult to change children or parents,” and “I do not know how to respond if I want to support but the children or parents will not respond.” For each item, the total score was given as “not at all disagree” (1), “disagree” (2), “agree” (3), or “agree very much” (4). Higher scores mean that they felt more challenged in supporting the children of parents with mental illness.
Actual behaviors and attitude
Three self-developed questions were used at T1 and T3 in order to assess the effects of program Purposes 2 and 3 on behaviors and attitudes. The program introduced the life situations of children in need of support and highlighted certain signs to look out for when identifying such children. The first question asked schoolteachers whether, over the past month, they had looked at children with the perspective that they could be children in need of support. The second question asked schoolteachers whether, over the past month, they had recognized the presence of a child(ren) in need of support. The program recommended that schoolteachers who had recognized a child in need of support first consult with other teachers in the school instead of worrying about the situation by themselves. The third question asked schoolteachers whether, over the past month, they had consulted other schoolteachers within the school about a child(ren) in need of support. The schoolteachers responded to these questions using “yes” or “no” options.
Knowledge questions, with responses that could be correct or incorrect, were administered to the schoolteachers; they were developed in order to assess program Purpose 1 (gaining basic knowledge about mental illness and children of parents with mental illness). The five items related to parental mental illness were lifetime prevalence of mental illness, types of mental illness, age of onset of mental illness, difficulties of living with people with mental illness, and mental health and welfare life support services. The five questions related to children were about the percentage of children with parents having mental illness, children’s own awareness, counseling, impact on health, and the common roles of young carers. Some of the items were based on previous research on mental health literacy education [25, 27, 34], but the questionnaire items about children were developed originally based on the opinions of the research members since no previous research had been conducted on this topic.
Program goals achievement
The degree to which the participants had achieved the eight items representing the program goals was measured using a seven-point scale ranging from “not at all” (1 point) to “very well” (7 points). The following goals were set as an assessment for achieving program Purpose 1: 1) be able to describe the impact of parental mental illness on children’s lives, 2) be able to describe children’s feelings, and 3) be able to describe the significance of school for children. The following goals were set as an assessment for achieving program Purpose 2: 4) be able to recognize children in need of support. The following goals were set as an assessment for achieving program Purpose 3: 5) be able to describe how to support children, 6) be able to act appropriately toward children, and 7) be able to act appropriately in school. Professionals’ negative attitudes regarding parenting with mental illness can hamper support provision . Therefore, it is important for schoolteachers to reduce the stigma against such parents and their children and have a positive attitude toward them. The final program goal is to 8) be able to support children with hope for their future.
Process and feasibility evaluation measures
The process and feasibility evaluation measures were as follows: program satisfaction (very satisfied, somewhat satisfied, not very satisfied, not satisfied at all); whether the participants would recommend the program to other schoolteachers (would recommend, would not recommend); length of the program (long, just right, short); what the participants learned was new (yes/no); if yes, what was impressive (open-ended text response); whether their way of thinking had changed (changed/not changed); if so, how it had changed (open-ended text response); and how they planned to act (open-ended text response).
The basic characteristics included age, gender, years of experience, job title, experience in special needs teaching, experience of supporting parents with mental illness, experience of supporting students with mental illness, experience of learning about mental illness, and the history of someone close, who has, or is suspected of having, a mental illness.
Randomization and blinding
When school-based applications were received by the research office, a random number table was prepared by a person who was unrelated to the implementation process, and independent allocations were made to either the intervention or control group in the order of arrival of the applications. After the school applications were received, the same research descriptions, which were designed for individuals, were sent to both the intervention and control groups. Because of ethical considerations and the need to ensure that the participants understood the purpose of the study, we simply stated, in the research description, that the children of parents with mental illness might face challenges. After the allocation of the schools, data managers were no longer blind.
The next time any schoolteacher’s individual application form arrived at the research office, the schoolteacher was assigned to the group to which the school was assigned. The participants were guided to their allocation group after the application, and they were no longer blinded. The web surveys and the video were sent at pre-set times, using a computer research electronic data capture accumulation and management system. The quantitative data analyses were blinded to the analysis.
As the primary population, the full analysis set (FAS), excluding the scores of those who did not respond to T1 from all randomized participants, was used in all analyses, except process and feasibility evaluation. The per-protocol set (PPS), which excluded participants who did not respond on time, those who did not complete watching video on time in the intervention group, and those who withdrew from the study, was also defined for sensitivity analysis of the primary outcome.
Baseline characteristics are summarized as mean and standard deviation for continuous variables and frequency and proportion for categorical variables. Student’s t-test, chi-square test, and Fisher’s exact test were performed for comparisons between the intervention and control groups.
The difference in the Sense of Coping Difficulty subscale at T3 between groups was evaluated by a t-test on cluster-specific means. As a sub-analysis, a multilevel analysis with a mixed model was performed, with group as fixed effects, and school as random effects, adjusting for some baseline characteristics. We fitted a mixed model for repeated measures (MMRM), with group, time points (T1, T2, and T3) and their interactions as fixed effects, and schools and study participants as random effects.
The three actual behavioral and attitude items were compared between the groups using chi-square analysis. A multilevel analysis was then conducted using a mixed-effects logistic model.
For the 10 knowledge items, the percentage of correct answers for each item was compared between the groups using chi-square analysis or Fisher’s exact test. The total number of correct answers was compared using a Wilcoxon rank-sum test. Mixed-effect logistic models were also fitted.
For program goal achievement, the medians and interquartile ranges of the scores and the total scores for each item were calculated and compared, using the Wilcoxon rank-sum test. MMRM were also fitted.
In an exploratory manner, we examined the interaction between the baseline score and intervention effects on the Sense of Coping Difficulty subscale.
The open-ended text response was analyzed qualitatively as part of a process evaluation. We created classification axes based on similarity and separated them into the smallest units whose meanings could be understood . First, the first author divided the smallest units into classification axes. Then, the other researcher placed the units on the same axes. The classification of units that did not match was 5 units out of 256 units. Units that did not match were discussed by the two researchers until they matched. The number of applicable individuals—not the number of units—was counted for each classification axis.
The missing data were not imputed. A p-value of 0.05 or less was considered significant. All analyses were performed using SAS9.4 (SAS Institute Inc., Cary, NC, USA).
This study was approved by the Ethics Committee for the Intervention Study of Osaka University Hospital (approval no. 21144; 5/10/2021). Written informed consent was obtained from all the participants. This study was conducted in accordance with the principles of the Declaration of Helsinki. This project has been registered in the Clinical Trial Registry (UMIN000045483; 14/09/2021). No adverse events were reported.
While filming one individual in the video, we asked her to decide whether she would like to have her face and name included in the video. She chose to show her face but not disclose her name. We notified the person in writing and obtained her consent for the study participants to watch the video.