Suicide safety plans provide a structured framework that enable people at risk of suicide to document warning signs, coping strategies, and external support persons to be accessed in the event of a suicide crisis (Stanley and Brown, 2012). They have become a standard component of suicide risk treatment over the past decade (Grant and Lusk, 2015; Labouliere et al., 2018; National Institute for Health and Care Excellence [NICE], 2022; Pettit et al., 2018), and a growing body of research has associated their use with reductions in suicidal behaviour, including two systematic reviews and a meta-analysis spanning 36 studies (Ferguson et al., 2021; Marshall et al., 2023; Nuij et al., 2021). While originally conceived as a paper-based intervention (Stanley and Brown, 2012), self-guided smartphone apps for safety planning are becoming more widely used (Larsen et al., 2016). These digital safety planning interventions are less evaluated, encompassing only 19 % of studies incorporated into the earlier reviews (Ferguson et al., 2021; Marshall et al., 2023; Nuij et al., 2021).

An indicative example that has undergone scientific evaluation is Australia's Beyond Now app, used by over 50,000 people each year (Beyond Blue, 2022). The app guides users to create a safety plan in seven steps, as outlined in Fig. 1, with steps primarily based on the Safety Planning Intervention (Stanley and Brown, 2012). While some safety planning apps also include additional therapeutic elements such as psychoeducation, symptom scales and self-help exercises (Buus et al., 2019; Dimeff et al., 2021; O'Grady et al., 2020; O'Toole et al., 2019), Beyond Now is solely focused on safety planning. Users of Beyond Now often create their plans without any assistance from a health professional and report their plans to be useful and effective in managing suicidal ideation (Melvin and Gresham, 2017; Melvin et al., 2021). A pilot open-trial study of Beyond Now in a tertiary mental health clinic reported reductions in suicidal ideation and increases in ability to cope with suicidal ideation after two months (Melvin et al., 2019), but small numbers in the pilot (N = 36) precluded an analysis of these outcomes with respect to plan-related variables such as use and perceived usefulness, content, or clinician involvement in authoring the plan. Results such as these have presented difficulties in determining whether improvement in suicide-related outcomes are uniquely predicted by the digital safety planning process, clinician involvement with safety planning, or concurrent treatment.

While safety plans are ultimately designed to reduce the risk of a future suicide attempt or fatality, focusing solely on these outcomes is challenging from a research perspective due to their low base rate in the population (Wenzel et al., 2011). The presence of suicidal ideation (particularly at higher intensity levels) can be predictive of future suicide attempts and fatalities (Simon et al., 2013; Law et al., 2018; McHugh et al., 2019), making it a useful risk factor to measure when sample sizes are too small to detect meaningful changes in suicidal behaviour. A second, strengths-based outcome that could be used to measure the impact of safety planning is suicide-related coping. This construct refers to knowledge and confidence to use internal and external coping strategies specifically to counter suicidal ideation (Stanley et al., 2017). Suicide-related coping has been negatively associated with both suicidal ideation and future suicide attempts (Interian et al., 2021; Rainbow et al., 2023; Stanley et al., 2017).

Some prior digital safety plan studies have recorded significant reductions in suicidal ideation (Boudreaux et al., 2017; Melvin et al., 2019), and/or increases in suicide-related coping (Dimeff et al., 2021; Melvin et al., 2019), but others have not (Bush et al., 2017; Pauwels et al., 2017). Many of these prior studies have been limited by either short follow-up timeframes of hours (Dimeff et al., 2021) or weeks (Boudreaux et al., 2017), or comprised small pilot evaluations of 3–36 participants (Jeong et al., 2020; Melvin et al., 2019; Pauwels et al., 2017). A larger-scale, observational study of self-guided digital safety planning over a longer period may help to address these issues and allow exploratory measurement of other potential mechanisms of plan effectiveness.

Safety planning is interactive by nature, with its effectiveness likely to be reliant on a series of processes. Prior literature has suggested potential factors that might explain why and how safety plans work (Rogers et al., 2021), but no studies conducted with digital safety plans have attempted to examine whether these factors are associated with changes in suicide-related outcomes. For example, collaboration with a mental health professional to author a safety plan is proposed to be a necessary condition for safety plan effectiveness in much prior safety plan literature (e.g., Boudreaux et al., 2017; Rogers et al., 2021; Spangler et al., 2020). It is assumed that collaboration brings a multitude of benefits, such as the opportunity to troubleshoot and co-design coping strategies (Stanley and Brown, 2012), to regularly review and improve the plan (Stanley et al., 2016), and simply to be heard and understood (Kayman et al., 2015). Such benefits may be weaker with self-guided safety plan apps, which allow users to write their own entries and also incorporate app-generated suggestions that are often generic in nature (Boudreaux et al., 2017; Melvin et al., 2019). Paper-based safety plan users in concurrent treatment were found to have a lower likelihood of future hospital admissions when their plans contained more personalised, specific content (Gamarra et al., 2015; Green et al., 2018), although clinician guidance in formulating this content was not measured.

Another unresolved question regarding safety plans is the extent to which their effectiveness is contingent upon usage. In studies involving paper-based plans, a variety of use patterns have been reported from daily (Kayman et al., 2015; Stanley et al., 2016), to intermittent (Gamarra et al., 2015). App-based plans allow for more granular tracking of use, but to our knowledge no studies have directly measured app usage in connection with suicide-related outcomes. Perceived usefulness may also increase the likelihood that a safety plan is used (Davis, 1989), and while perceived usefulness has been extensively studied in prior digital safety plan studies (e.g., Dimeff et al., 2021; Jeong et al., 2020; Melvin et al., 2019; Pauwels et al., 2017), again no studies have examined whether perceived usefulness is associated with suicide-related outcomes. To get closer to an understanding of the circumstances under which digital safety planning best works, clinical outcomes of safety plan use should be measured alongside these overlapping contextual factors. Measuring individual safety plan use, personalised content, perceived usefulness and practitioner collaboration is therefore warranted to determine whether these can uniquely predict changes in suicidal ideation and suicide-related coping over time.

The current study aimed to examine changes in suicidal ideation and suicide-related coping over time among users of a digital suicide safety planning app (Beyond Now), and explore whether any changes observed could be associated with plan-related factors: plan use, perceived usefulness, user-generated plan content and whether collaboration occurred in plan authoring. An additional aim was to describe how often and when the plan is used, and to determine whether user-generated plan content appeared more often in co-authored plans. It was hypothesised that:

Significant reductions in suicidal ideation and significant increases in suicide-related coping would be observed over three months.

Lower suicidal ideation at one and three months would be predicted by higher suicide-related coping at one and three months.

Lower suicidal ideation and higher suicide-related coping at one and three months would be predicted by the following plan-related variables: higher perceived usefulness (at one and three months), greater app use time (at one and three months), greater user-generated plan content and plan co-authoring (both collected at baseline).