Background

Suicide is a significant public health concern worldwide, with men exhibiting higher rates of suicides compared to women [1, 2]. Statistically, approximately 75% of suicides are attributed to males. Despite the prevalence of suicidal ideation and behavior among men, they are often more reluctant to seek support from their surroundings or the support system [3, 4]. Consequently, men experiencing crises often do not receive appropriate care and fail to access available support systems. This leads to a lack of timely and adequate intervention for those at risk of suicide. In response, the concept of gatekeepers became crucial within suicide prevention [5]. Gatekeepers can serve as pivotal intermediaries in recognizing warning signs and facilitating access to appropriate support. Gatekeepers are typically individuals within the familial, social, or professional circles of at-risk individuals, such as relatives, friends, or colleagues [6]. Here, people in a close social environment are particularly important, as they are in direct contact with those at risk and can detect behavioral changes more quickly.

However, a major problem is that gatekeepers themselves have little knowledge and expertise in recognizing suicidality and show great uncertainty in dealing with people at risk. A study by Hofmann and Wagner [7], who conducted interviews with relatives of men who had died by suicide, showed that the majority of relatives noticed changes in the men's behavior, but did not interpret these changes as warning signs. Overall, there was little knowledge on how to recognize men in suicidal crisis and what support was available. However, how helpful gatekeepers are depends heavily on their level of knowledge on the topic [5, 8]. Insufficient knowledge can hinder the timely identification of people at risk, resulting in a missed opportunity to intervene. Numerous studies have shown that training gatekeepers not only increases their knowledge and willingness to intervene, but also increases actual prevention behavior [9,10,11].

In response, gatekeeper training programs have emerged as a promising approach to equip individuals, particularly relatives of men with suicidal ideation and behavior, with the skills and knowledge to recognize warning signs of suicidality and how to intervene effectively [12,13,14]. Many of these gatekeeper programs are based on the QPR approach ("Question, Persuade, Refer") [12] and include the aspects of [1] addressing people at risk, [2] persuading them to accept help, and [3] referring them to support. Programs that follow the QPR approach have been evaluated many times and have been shown to lead to an increase in knowledge and improved handling of vulnerable people [5, 15, 16].

A number of studies have evaluated gatekeeper programs for professional groups, such as nurses [17], police officers [18, 19], school staff [20], or the general population [21]. In addition to the increase in knowledge, these programs usually also lead to an increase in self-efficacy, perceived preparedness, and an increased willingness to intervene. However, specific gatekeeper programs to support men are still lacking. Considering the specific symptom patterns of suicidal ideation and behavior in men, it is essential to specifically train gatekeepers in recognizing suicidal tendencies in men.

However, gatekeepers receive less attention regarding their own stress. Gatekeeper programs often focus exclusively on the person at risk and disregard the burden of the gatekeeper. Studies have already shown that close contacts of people at risk of suicide are themselves highly stressed [22,23,24]. They often suffer from depressive symptoms themselves, have trouble sleeping, can no longer pursue their hobbies and jobs or can do so to a lesser extent, and experience a drop in social contacts [25, 26]. In addition, gatekeepers often take on a multidimensional role by providing financial support, advice, transportation services, accompaniment to therapy, or help with everyday life [26]. In this context, caregivers often reported feelings of being overwhelmed, anxiety, stress, guilt, and self-blame [24, 25, 27].

Given these challenges, this randomized controlled trial evaluated the efficacy of an online gatekeeper training specifically aimed at gatekeepers in suicide prevention of men at risk, exploring both short-term outcomes and long-term impacts on knowledge, self-efficacy, and preparedness, as well as gatekeeper burden and distress.

Methods

Study design

The study was a randomized controlled trial in accordance with the CONSORT guidelines for RCTs [28]. Participants were randomly assigned to one of two conditions, either the intervention or the waitlist control group. The study was approved by the ethics committee of the Medical School Berlin on 23 November 2022 (reference number MSB-2022/108).

Participants

Participants were mainly recruited via social media, online advertisements, and cooperation partners (e.g., clinics, crisis services). Recruitment took place between January 2023 and October 2023 in Germany. The end of recruitment was based on the duration of the project and that the necessary sample size was reached. Eligible participants had to meet the following inclusion criteria: (1) aged 18 years or older, (2), access to the internet, (3) being in contact with a man with suicidal ideation, (4) sufficient knowledge of German, and (5) signed informed consent. There were no exclusion criteria. In total, 313 interested individuals registered on the study website, of which 123 were included in the study. Due to the dropout rate of 31.7% from baseline to post-measurement, only completer analyses (n = 84) were performed. There were no significant differences between completers and non-completers at baseline. Figure 1 provides an overview of the participants’ flow.

Fig. 1
figure 1

Participants’ flow

Full size image

All characteristics can be seen in Table 1. Of the 84 participants, 90.5% were female and participants were on average 38.79 (SD = 12.72) years old. Over a third (34.5%) were receiving professional support at the time of participation. Most (35.7%) were worried about their partner, 17.9% about a friend, 13.1% each about their father or son and 8.3% were worried about their brother. The remaining 11.9% were concerned about other male relatives such as an uncle, grandfather, or colleague. Over a fifth (22.6%) of the male relatives had already attempted one or more suicides in the past, and over half (56.0%) had been diagnosed with a mental health disorder. Professional support was received by 48.8% of the male relatives. There were no significant differences between participants in the two conditions regarding sociodemographic or outcome characteristics at baseline. No adverse events or harms were observed in either group during the study period.

Table 1 Sociodemographic and outcome characteristics of participants in the intervention (n = 43) and waitlist (n = 41) condition and for the total sample (N = 84)
Full size table

Procedure and randomization

After registration, participants received an email with the study information and a link to an online questionnaire. The registration form also recorded the consent to participate in the study. After completing the questionnaire, participants were randomized by a third person, who was not involved in the study, using a computer-generated randomization list with a 1:1 allocation ratio. Participants in the intervention group were then given access to the online program and had 6 weeks to complete it. After the 6 weeks, they received a post-questionnaire and 3 months later a follow-up questionnaire. Participants in the waiting group had to wait 6 weeks after randomization, then received a questionnaire after the waiting period and were then given access to the online program. After a further 6 weeks, they also received a post questionnaire. Participants each were given a €20 Amazon voucher for completing the post and follow-up questionnaires.

Online program

The overall aim of the online gatekeeper program was to educate relatives of men experiencing suicidal ideation, provide information, and draw attention to risk factors and support services. The program consisted of a total of four modules: (1) Suicidal ideation and behavior in men, (2) communicating and referring to help, (3) support services, and (4) support for relatives. The individual modules comprised four films, each lasting 15–20 min, and contained psychoeducational information and videos with experts and affected men as well as audio plays with fictional dialogues between men and their relatives. In addition, a manual with the content from the modules could be downloaded. The program was completed by the participants independently and without therapeutic supervision.

The first module aimed at providing participants with background knowledge on suicidal ideation and behavior and, in particular, highlighting the specific nature of suicidality and depression in men (e.g., aggressive behavior). It also focused on risk factors and warning signs and how to identify men at risk. In the second module, participants were shown how to approach a man at risk, how to start a conversation and how to refer them to support services. The module is based on the QPR approach from Quinnett [12]. In the third module, various support services were introduced and were shown how they can be accessed. The module also focused on what can be done in cases of acute suicidality and rejection of help by the suicidal man. The last module was designed to support the relatives themselves. It was intended to encourage participants to deal with their own emotions, recognize their distress, and set boundaries. Strategies and specific support services for relatives were presented.

Measures

Sociodemographic data

The following sociodemographic data were assessed at baseline: age, gender, marital status, education level, living situation, and current professional help. In addition, sociodemographic data on the male relative was collected: age, current professional help, diagnosis, and suicide attempts.

Outcome measures

Depression

To assess depressive symptoms in the past 2 weeks, the German version of the Patient Health Questionnaire (PHQ-9) [29, 30] was used. The PHQ-9 comprised of nine items rated on a 4-point-Likert scale (0 = “not at all”; 3 = “nearly every day”) which can be added up to a total score ranging between 0 and 27. A score between 10 and 14 indicates mild depressive symptoms, a score between 15 and 19 moderate depressive symptoms, and a score above 20 is an indicator for severe depressive symptoms. The PHQ-9 has shown good internal consistency with Cronbach’s Alpha of α = 0.88 in a German clinical sample [31] and of α = 0.81 in the present sample.

Perceived stress

The perceived stress scale (PSS) [32, 33] is a self-report measure assessing to which degree situations are considered stressful and uncontrollable. The scale consists of 10 items rated on a 5-point Likert scale (0 = “never”; 4 = “very often”). A sum score as well as the subscales helplessness and self-efficacy can be calculated, with higher scores indicating higher perceived stress, an increased feeling of helplessness as well as higher perceived self-efficacy. The PSS showed good reliability with ω = 0.89 in a German sample [33] and good reliability of α = 0.89 in this sample.

Caregiver involvement

The German version of the Involvement Evaluation Questionnaire (IEQ) was used to assess the extent to which relatives are involved in the care of their relative in the past 4 weeks [34, 35]. The questionnaire consists of 31 items rated on a 5-point Likert scale (0 = “never”, 4 = “always”). A total score can be calculated from a total of 27 items as well as the subscales tension (9 items), supervision (5 items), worrying (5 items), and urging (8 items). The IEQ showed good reliability of α = 0.88 in a German sample of relatives, with the reliability of the subscales varying between α = 0.71 and α = 0.83 [35]. The IEQ also showed a good reliability in this sample with Cronbach's alpha of α = 0.85, the reliability of the subscales varied between α = 0.59 and α = 0.86. Due to the low reliability, the subscale supervision was excluded from the analyses.

Caregiver Burden

To assess the burden of the relatives, the German version of the Burden Assessment Scale (BAS) was used [36, 37]. The BAS consists of 19 items rated on a 4-point Likert scale (1 = “not at all”, 4 = “very much”) and measures subjective as well as objective distress of relatives over the past 6 months. A sum score can be calculated as well as the four subscales disrupted activities (8 items), personal distress (5 items), time perspective (3 items), and guilt (3 items). The BAS showed excellent reliability of α = 0.92 in a German sample of relatives of individuals with a mental health disorder [36], while the subscales ranged from α = 0.64 to α = 0.90. In this sample, the BAS also showed excellent reliability with Cronbach’s Alpha of α = 0.90, the reliability of the subscales varied between 0.62 and 0.88.

Gatekeeper outcomes

Proximal gatekeeper outcomes were assessed with scales originally developed in a study by Wyman et al. [38] and adapted by Holmes et al. [9] for the general population measure the outcomes of gatekeeper trainings. The scales capture perceived knowledge (7 items, 7-point Likert scale, 1 = “nothing”, 7 = “very much”), perceived preparedness (5 items, 7-point Likert scale, 1 = “not prepared at all”, 7 = “very well prepared”), efficacy (9 items, 7-point Likert scale, 1 = “strongly disagree”, 7 = “strongly agree”), and reluctance (8 items, 7-point Likert scale, 1 = “strongly disagree”, 7 = “strongly agree”). The items were translated into German by two different researchers using forward and backward translation. The scales showed poor to excellent reliability in a healthy Australian sample with Cronbach’s Alpha of α = 0.93 for perceived knowledge, α = 0.93 for perceived preparedness, α = 0.80 for efficacy, and α = 0.68 for reluctance. In this sample, the reliability also varied between poor and excellent, with Cronbach’s Alpha of α = 0.91, α = 0.87, α = 0.80, and α = 0.56, respectively. The factor reluctance was therefore excluded from all further analyses.

Knowledge

Fourteen questions on suicidal ideation and behavior in men, as well as suicide prevention, were developed by the authors specifically for this study to assess actual participant knowledge. Eight questions have a multiple-choice answer format (e.g., “Why are men particularly at risk of suicide?”), 4 items have a correct/not correct answer format (e.g., “If you talk to someone about suicide, it may that you give them the idea of a suicide.”) and 2 items have an open answer format (e.g., “Please name three risk factors for suicidal ideation and behavior in men.”). Correct answers were scored with one point and were added up to a sum score. The reliability of the questionnaire was poor with Cronbach’s Alpha of α = 0.50. Since this questionnaire is crucial for the evaluation of the program, we kept it in our analyses and discuss the limitations below.

Statistical analysis

All data were analyzed using IBM SPSS Statistics Version 28 [39]. Sociodemographic and outcome characteristics at baseline were analyzed using frequencies for categorical variables and means and standard deviations for continuous variables. Differences between sociodemographic and outcome characteristics at baseline between the two groups were compared using t-tests and independent χ2-tests. The sample size was calculated using G*Power with the following parameters: effect size Cohen's d = 0.7; α-error = 0.05, power: 0.95, number of groups: 2. This resulted in a total sample of at least 74 participants.

Due to the dropout rate of 31.7% at post-measurement, an intention-to-treat analysis was omitted and only those participants from whom complete post-data were available were included in the analyses. There is no definitive dropout rate that precludes performing an ITT analysis. However, an excessive number of missing values can potentially distort the results. Since the required sample size was still met with the completer data, we opted to conduct a completer analysis. The effectiveness of the program was evaluated using repeated measures ANOVAs, with condition serving as the between-subject factor and time as the within-subject factor. To evaluate the effectiveness and long-term effects of the online program in the intervention group, repeated-measures ANOVAs were conducted for all outcomes. Due to the dropout rate of 49.2% at the 3-month follow-up measurement in the intervention group, only completer analyses were performed. Cohen's d [40] was calculated for within-subject effect sizes, whereas dppc2 [41] was calculated for between-subject effect sizes. We used a statistical level of α = 0.05 (two-tailed). Several hierarchical regressions were run to determine if the stepwise addition of age, marital status, and gender of participants, as well as current professional support and outcomes at baseline, had an influence on outcomes at post-measurement in the intervention group.

Results

Outcome changes from baseline to post-measurement

Table 2 shows means, standard deviations, and effect sizes of the completer data for baseline and post-training for the intervention condition or post-waitlist for the waitlist condition. Repeated-measures ANOVAs were conducted to examine differences between the two groups on symptoms of depression, distress, involvement, burden, proximal gatekeeper outcomes, and actual knowledge with time (baseline vs. post-assessment) as within-subject factor and condition (intervention vs. waitlist) as between-subject factor.

Table 2 Means, standard deviations, effect sizes, and results from repeated measures ANOVAs for the intervention and the waitlist control group and two assessment points (baseline vs. post-intervention/post-waitlist) for completer data (n = 84)
Full size table

A significant interaction effect was found for depressive symptoms, F(1,82) = 4.02, p = 0.048. No significant interaction effects could be found for burden, distress, and gatekeeper involvement. However, a significant interaction effect emerged for perceived knowledge, F(1,82) = 18.62, p < 0.001, perceived preparedness, F(1,82) = 17.22, p < 0.001, and self-efficacy, F(1,82) = 82.00, p = 0.002. A significant interaction effect was also found for actual knowledge, F(1,82) = 9.38, p = 0.003.

Maintenance of outcome changes at follow-up

Repeated measures ANOVAs revealed a significant reduction from baseline to 3-month follow-up for perceived stress with F(2,62) = 37.49, p < 0.001, but not for depression with F(2,62) = 2.41, p = 0.098 and gatekeeper burden with F(2,62) = 2.43, p = 0.096. A significant reduction was observed for gatekeeper involvement from baseline to follow-up with F(2,62) = 6.66, p = 0.002. There was also a significant increase for knowledge with F(2,62) = 11.32, p < 0.001, for perceived knowledge with F(2,62) = 16.50, p < 0.001, perceived preparedness with F(2,62) = 24.17, p < 0.001, and for self-efficacy with F(2,62) = 19.16, p < 0.001.

Predictors of outcomes

Several hierarchical multiple regression analyses were performed to evaluate sociodemographic factors, depressive symptoms, stress, burden, as well as gatekeeper outcomes at baseline, as predictors for change in all outcomes at post-level. The analyses revealed that current professional support was a significant predictor for symptom improvement at post-measurement. Levels of gatekeeper outcomes at baseline were a significant predictor of an increase in self-efficacy, perceived preparedness, and knowledge at post-level.

Depression, stress, and gatekeeper burden

The full model to predict improvement in depressive symptoms was statistically significant, R2 = 0.35, F(5,37) = 3.90, p = 0.006, adjusted R2 = 0.26. The addition of current professional support (step 2) led to a statistically significant increase in R2 = 0.22, F(1,38) = 11.97, p = 0.001. The addition of sociodemographic factors (step 1) and depressive symptoms at baseline (step 3) was not significant. The full model for perceived stress at post-level was statistically significant, R2 = 0.33, F(5,37) = 3.71, p = 0.008, adjusted R2 = 0.24. Again, the addition of current professional support led to a statistically significant increase in R2 = 0.07, F(1,38) = 10.79, p = 0.002. The full model for gatekeeper burden was also significant, R2 = 0.47, F(5,37) = 6.61, p < 0.001, adjusted R2 = 0.40. The addition of current professional support as well as of levels of burden at baseline led to a significant increase in R2 = 0.11, F(1,38) = 5.16, p = 0.029 and R2 = 0.31, F(1,37) = 21.34, p < 0.001, respectively.

Gatekeeper outcomes

The full model to predict an increase in perceived preparedness was significant, R2 = 0.18, F(5,37) = 2.90, p = 0.026, adjusted R2 = 0.18. The addition of sociodemographic factors (step 1) led to a statistically significant increase in R2 = 0.20, F(1,39) = 3.19, p = 0.034. The full model for self-efficacy was again significant, R2 = 0.39, F(5,37) = 6.29, p < 0.001, adjusted R2 = 0.32. Only the addition of self-efficacy scores at baseline led to a significant increase in R2 = 0.32, F(1,37) = 21.99, p < 0.001. The full model for knowledge was significant, R2 = 0.29, F(5,37) = 3.00, p = 0.024, adjusted R2 = 0.19. The addition of knowledge at baseline led to a significant increase in R2 = 0.13, F(1,37) = 6.95, p = 0.012. The full model for perceived knowledge was not significant, R2 = 0.23, F(5,37) = 2.26, p = 0.068.

Discussion

The aim of this study was to evaluate the effectiveness of an online program for gatekeepers in male suicide prevention. We examined whether participants in the intervention group would show a significant increase in proximal gatekeeper outcomes and knowledge and a significant decrease in distress, burden, and depression compared to a waitlist control group. Furthermore, we expected the changes to remain stable over a 3-month follow-up.

The results of this evaluation study showed that a short gatekeeper program, which is conducted independently without therapeutic supervision, can lead to an increase in perceived knowledge, perceived preparedness, self-efficacy, and knowledge. Our findings are consistent with other studies evaluating gatekeeper programs [16, 42, 43]. Gatekeeper outcomes and knowledge are the most studied outcomes in the evaluation of such programs. In their review, Holmes et al. [9] examined the long-term effects of gatekeeper training and the outcomes measured. Of the 23 included studies, 78% measured actual or perceived knowledge, making it the most consistently measured outcome. All studies reported an increase from baseline to post-measurement, with most reporting a further increase or stabilization over the follow-up period. Self-efficacy was also examined in 70% of the studies and a significant increase was also achieved above all studies. Knowledge and self-efficacy are key components as they may influence the intervention behavior of gatekeepers. In their study, Burnette et al. [44] postulated that an increase in knowledge and self-efficacy leads to increased intervention behavior. Wolitzky-Taylor et al. [45] also stated that self-efficacy leads to more actions and thus positively influences suicide prevention behavior. Knowledge also leads to dispelling myths and overriding false assumptions about suicidal ideation and behavior. Myths about suicidality are still prevalent and prevent people from intervening, in fear of triggering suicidal behavior or doing the wrong thing [46].

We found a significant decrease in depressive symptoms from baseline to post-level, as well as a stabilization of depressive symptoms, over 3 months. Providing participants with knowledge and effective strategies can instill a sense of confidence and reduce feelings of helplessness. This can have a positive impact on mental health outcomes. Research indicates that individuals with low self-efficacy are more likely to experience elevated levels of depression [47]. However, there are still no studies investigating the long-term effects of gatekeeper training on the mental health of participants.

Regarding the stress of the gatekeepers themselves, we could not find a significant difference between both groups for burden, stress, and gatekeeper involvement. There may be several reasons for this. Firstly, stress does not disappear immediately after participating in such a program, but sometimes requires a certain period to be able to determine effects. Even if participants have been introduced to coping strategies and support services, these must first take effect and help must be sought. Effects are more likely to be observed over a longer period of time [18, 48]. To date, there are hardly any studies that examine gatekeeper training regarding the mental health of gatekeepers themselves. There are some programs for caregivers that were also able to achieve positive effects in terms of their stress levels [49,50,51]. However, these programs are not gatekeeper programs and can therefore only be used as a comparison to a limited extent.

Furthermore, reduction of perceived stress and gatekeeper involvement, as well as an increase in gatekeeper outcomes, was maintained from baseline to 3-month follow-up. These findings demonstrate that the acquired knowledge and the sense of preparedness and self-efficacy persist after completing a short online gatekeeper program, aligning with findings from similar evaluations of gatekeeper trainings [5]. While maintaining improvements at follow-up is crucial, it's also important to consider how the initial severity of symptoms influences treatment outcomes. Previous research has also indicated that individuals with higher symptom severity prior to an intervention often experience larger improvements [52]. Moreover, current professional support emerged as a predictor of improvement in psychopathological symptoms. This outcome is unsurprising, given that additional support plays a pivotal role in mitigating stress in such a situation.

Strengths & limitations

To our knowledge, this is the first gatekeeper program aimed exclusively at individuals who are in contact with men in suicidal crises. Considering the high suicide rate and other symptoms, it is necessary to train people on male suicidality and to impart specific knowledge. Even though this is a self-help program without therapeutic support, we recorded a relatively low dropout rate of 31.7%. Dropout rates in self-help programs vary greatly; high dropout rates are not uncommon [53, 54]. The online format also gives participants the opportunity to work through the modules at their own pace and from any location and to access them repeatedly. A program like this is ideal for people who are very involved in caring for their loved ones or who are otherwise unable to accept offers of help. Studies have already shown that there is no difference in effectiveness between online and face-to-face formats [55]. Another strength is the use of a randomized controlled design in accordance with the CONSORT 2010 statement [28] and thus the possibility of comparing two conditions with each other.

One of the biggest limitations of this study is the almost exclusively female sample. This is a recurring problem in clinical research [56]. While women are the main contact persons for men in crises [3], it would still be highly relevant to train men on suicide prevention as well. The online format also excludes certain population groups. People without access to the internet or those lacking media skills, as well as older populations, are highly unlikely to take part in the program but are nevertheless important contacts who are therefore not reached. In our sample, only close relatives such as partners, mothers, or sisters are represented. Other gatekeepers are rarely included. However, this may also be due to the chosen title of the program (“Help for Relatives”), which may not appeal to other persons. In addition, we did not measure the extent to which participants showed supportive behavior or used the strategies after completing the program. Here, too, there are only a few studies to date that examine actual behavior after participation in gatekeeper programs [10]. Additionally, we initially planned to perform an ITT analysis. Due to the high dropout rate and since the sample size remained sufficiently large, we proceeded with completer analyses, despite deviating from the originally planned approach. Importantly, no differences were observed between completers and non-completers at baseline measurement.

Conclusion

The results of the study are promising regarding the training of gatekeepers through a relatively short training program without therapeutic support. The program can be implemented at a low threshold and at low cost and made accessible to the broad community. Given the high suicide rate among men, it is essential to educate the general public about male suicidality. The program additionally resulted in a decrease in depressive symptoms among gatekeepers. These outcomes underscore the significant stress levels experienced by gatekeepers and the necessity for tailored support mechanisms. Nevertheless, they also highlight the efficacy of low-threshold support as an initial step in gatekeeper care. The program was positively evaluated by the participants, well accepted, and rated as varied and informative. Such a program can, therefore, emphasize the importance of supporting gatekeepers in suicide prevention and focus on this population group. These results can also lead to the specific burden of gatekeepers coming more to the fore and are generalizable to other populations similar to our study participants, taking into account the limitations. Further future studies in this area can also be used to develop guidelines for gatekeepers in order to increase their effectiveness.

This study is the first to specifically target gatekeepers in male suicide prevention and showed promising effects in terms of efficacy. Due to the successful evaluation, the program was made available free of charge and freely accessible after the end of the study.