Background
Reducing suicidal behaviour (suicide attempts and deaths) remains a critical public health issue globally. Over the period 1990 to 2016, suicide was estimated to be the leading cause of age-standardised years of life lost in high-income Asia Pacific countries, and among the top 10 leading causes in Europe, and parts of the Americas [
1,
2]. The World Health Organization’s (WHO) development of a Comprehensive Mental Health Action Plan [
3] and the inclusion of suicide mortality as an indicator for the Sustainable Development Goals (SDGs) [
4] have signalled a recognition of the role mental health, mental capital, and suicidal behaviour (SB) play in social, cultural, and economic participation that contributes to the mental wealth of nations [
5,
6] and in facilitating or undermining progress towards broader international development targets [
7]. Further significant momentum has been achieved through the World Economic Forum’s Global Shapers Community, a grassroots network of young people, who called on all countries to increase financing, public mental health education, and quality systems of care at the Annual meeting in Davos in 2020 [
8]. This elevation of mental health and wellbeing and suicide prevention in the global development agenda has led to a renewed push worldwide to strengthen mental health systems (particularly leadership and governance, and community-based care), increase service coverage and responsiveness, and set targets for reductions in suicide deaths [
3,
9,
10].
Despite this momentum, growing evidence of effective suicide prevention interventions, and the release of successive action plans, current strategies are not delivering substantial impacts [
11‐
13]. SB has a complex aetiology with a wide range of contributing factors, both individual and contextual, and is rarely a result of any single cause [
14,
15]. While there is good evidence for the significant role mental ill-health has in the aetiology of suicide [
16‐
18], discourse has turned to the role of social determinants in the causal mechanism of mental disorder and SB as targets for prevention [
19‐
21]. Factors including adverse childhood exposures, domestic and family violence, substance abuse, unemployment, and other socioeconomic factors that influence access to housing and mental health services have been found to have unidirectional or bidirectional relationships with each other and with psychological distress, mental disorder, and suicide [
22‐
27]. However, the relationship of these risk factors with mental health and suicide outcomes are often analysed using methods that assume they are independent and that their relationship with key outcomes are linear and constant through time [
28], and as such, interventions to address these risk factors are explored discretely. The complex interplay of social determinants, service system factors, population demographics, and behavioural dynamics makes it extraordinarily difficult for decision makers to determine the nature and balance of investments required to have the greatest impacts on suicidal behaviour over the short and long term. While there are numerous evidence reviews to support the case for investments aimed at addressing the social determinants of mental disorder, it is unclear whether resources should be spread across each of them or whether some are more important than others for suicide prevention in a particular context.
Systems modelling and simulation offers an important tool for systems analysis to support decision making for complex problems. Systems modelling is a robust quantitative method of complex systems science, an interdisciplinary field that studies the nature and behaviour of complex systems underpinned by well-established mathematical theory of nonlinear dynamics [
29‐
33]. It provides a robust method for mapping and quantifying the complex causal mechanism driving mental health and suicide outcomes [
11,
34‐
36]. Systems modelling is uniquely able to capture population and demographic dynamics, changes over time in social and economic drivers of psychological distress, mental disorders and suicidal behaviours (including feedback loops), workforce dynamics and the changing relationship between service supply versus demand, and the potentially non-additive (interdependencies and interacting) effects of intervention combinations, factors that bedevil traditional analytic approaches. Model development leverages disparate datasets, research evidence, and our best understanding of local system structure and behaviours of system actors in a systematic and disciplined way [
28,
37‐
42]. The process delivers an interactive decision support tool that provides a virtual environment to explore the optimal combination, targeting, timing, scale, frequency, and intensity of investments in screening, treatment, population-based mental health strategies, and social determinants required to achieve the greatest impacts within the contextual, resource, and capacity constraints of a particular region, before implementing them in the real world.
This study describes the application of systems modelling and simulation undertaken as a research-practice partnership between a regional Primary Health Network (PHN) in New South Wales, Australia, their stakeholders, and several academic institutions. The study aimed to leverage a range of national, state, and local datasets to (i) identify the likely impact over time of a range of locally prioritised mental health and suicide prevention interventions being considered for investment, (ii) determine the value and balance of investments across the social determinants of mental health in the region, and (iii) determine the best combination of strategies to deliver the greatest impacts on suicidal behaviour.
Discussion
This study used systems modelling and simulation to leverage a range of national, state, and local data sets and best evidence and undertake a priori testing of scenarios that explore the impact on suicide of investing across social determinants of mental health and specific mental health and suicide prevention initiatives. Of the intervention scenarios examined, improving social connectedness was the single most effective intervention in reducing SB across the youth and total populations over the long term. This finding is consistent with studies highlighting social isolation as an important contributing factor to suicide [
52,
53]. While only the broad strategy of community support programs aimed at increasing social connectedness and reducing social isolation was modelled, the specific features of these programs were not prescribed in recognition of the vital importance of engaging local communities in their design and delivery within a cultural framework of community development [
54,
55]. However, locally designed and implemented programs should be evaluated to facilitate future model refinement.
Findings of this study also demonstrated that improvements to the social determinants of mental health do not contribute equally to reductions in suicidal behaviour. Investments in reducing childhood adversity and increasing youth employment initiation together represent best targets, not only for their impact over the selected time horizon, but also for their ongoing amplifying effects in reducing youth suicide over the longer term. These findings are consistent with reviews of the literature of individual and ecological studies highlighting the strong associations that youth unemployment and exposure to early life adversity (particularly sexual abuse and accumulation of adversities) have on youth suicidal behaviour [
56,
57] and point to the importance of looking to broader social, educational and vocational targets for the prevention of SB [
58]. While these findings highlight early life exposures and youth unemployment as important potential targets for investments for youth suicide prevention, they do not suggest a lack of importance for addressing adult unemployment, domestic violence, and homelessness on broader moral, social, and economic grounds.
While no specific programs were modelled, reducing childhood adversity by between 20 and 50% was projected to deliver the single greatest impact on suicide rates in young people among the social determinants, suggesting it to be a worthwhile target for investment and action. The scenario of a reduction in childhood adversity was modelled by multiplying the rates at which children (aged 0–14 years) at low and moderate risk of developing a mental disorder transition to moderate and high levels of risk. Therefore, programs to reduce childhood adversity may be targeted at either primary prevention (i.e. strategies to reduce exposure to domestic and family violence, abuse, neglect, poverty, war, and natural disasters) [
59‐
61] and/or at programs aimed at harm minimisation to reduce the risk of emergence of mental disorders among children and adolescence that have experienced adversity. Harm minimisation may include screening, referral, and intervention carried out in clinical settings [
62] and/or the implementation of universal, school- or community-based, resilience-focused interventions to provide more generalised fostering of mental health [
63]. Determining the feasibility and nature, targeting, timing, scale, and duration of programs needed to achieve a 20% or 50% reduction in childhood adversity are best explored in partnership with regional communities through extension of the existing systems model to test alternative strategies prior to implementation, monitoring, and evaluation of programs.
Finally, there are two important insights that this systems modelling study highlights. Firstly, that the greatest impacts on suicidal behaviour in young people are likely to be achieved with a mix of specific mental health and suicide prevention initiatives (with more immediate impacts that plateau over time) and improvements to key social determinants (with delayed impacts but amplifying effects over the longer term) as highlighted in Fig.
4. Secondly, this study highlights that more is not necessarily better. The simulated impacts of implementing all mental health and suicide prevention initiatives included in the model were little better than the impacts of the targeted combination of social connectedness programs, post suicide attempt assertive aftercare, and technology-enabled coordinated mental health care. This highlights the importance of the advanced decision support capability provided by systems modelling to facilitate a more strategic approach to the allocation of limited resources.
Limitations
There are a number of limitations that require consideration when interpreting the findings of this study. There is potential measurement bias in the range of secondary data used to parameterise the model including the population health surveys, Medicare claims data, and PHN and Local Health District (LHD) datasets. The model acknowledges these potential sources of measurement bias and a number of commonly used strategies were employed to address them, including the triangulation of multiple data sources, parameter estimation via constrained optimisation, and local verification to identify plausible estimates.
In addition, there is potentially an under-enumeration of suicide cases used to calibrate the model, due to the misclassification of suicides to ICD codes relating to unintentional injury and events of ‘undetermined intent’ [
64]. Suicide attempts identified from hospital admissions data likely only capture those cases serious enough to warrant medical intervention, and instances of self-harm where the intent was not clear may be not coded as suicide attempts. However, this under-enumeration is consistent across simulations of the baseline case and intervention scenarios and as such are unlikely to affect the forecast estimates of impact (i.e. the % reduction in suicidal behaviour) of intervention strategies or the strategic insights derived from the model. Ongoing systematic monitoring and evaluation can determine the extent to which the model forecasts are corresponding with real-world outcomes over time, allowing refinement of model parameters to improve forecasting capabilities. Finally, as the impacts of simulated scenarios are subject to the population, demographic, behavioural, and service dynamics of the modelled region, they are not necessarily generalisable to other regions; however, depending on contextual similarity to the modelled region, the qualitative insights are likely to relevant.
Conclusions
The findings of this study suggest that targeted investments in addressing the social determinants and in mental health services provides the best opportunity to reduce SB and suicide. The current prioritised set of findings are by no means intended to provide ‘the answer’, but rather to demonstrate how systems models can bring together a body of evidence and data in way that facilitates learning among system actors regarding system behaviour in response to the introduction of new initiatives and ‘solutions’. This systems model is providing regional decision makers and stakeholders the capacity to investigate alternative scenarios related to the timing of implementation of interventions, their scale and intensity, and to test alternative assumptions regarding level of intervention uptake to inform strategic decision making. The potential of systems modelling and simulation to support a more disciplined, targeted, inclusive, and transparent approach to national and regional decision-making regarding allocation of resources to reduce suicidal behaviour has been well described [
12,
29,
34]. Importantly, such interactive systems models are being used to explore the impact of possible investment decisions on outcomes other than suicidal behaviour to ensure that unintended negative impacts on other parts of the system, such as mental health ED presentations, service wait times, and capacity, do not occur as a result of efforts to specifically address suicidal behaviour.
Finally, it is important to note that this study focussed on simulations of
improvements to the social determinants of mental health in the North Coast NSW region, which may underestimate their importance in comparison to the specific mental health and suicide prevention interventions simulated to reducing suicidal behaviour. Inevitable corollaries of the COVID-19 pandemic are a
deterioration in the social determinants of mental health, which may produce greater negative impacts on suicidal behaviour than the positive impacts of improvements to those determinants simulated in the current study. As shown in other applications of systems modelling in mental health [
38], important thresholds might exist where deterioration in current levels of unemployment in the North Coast NSW population catchment may result in greater than anticipated increases in substance misuse and suicidal behaviour. This is the subject of further investigation.
Acknowledgements
Co-lead author, Dr. Adam Skinner, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. This paper is part of the work of Rockefeller Foundation-Boston University Commission on Health Determinants, Data, and Decision Making, funded by the Rockefeller Foundation. The model was developed in partnership with the North Coast Primary Health Network and intended as a contribution to the North Coast Collective (NCC) which is a regional collaboration initially between three organisations—North Coast Primary Health Network, Northern NSW Local Health District, and Mid-North Coast Local Health District—and growing to include partners a range of community stakeholders including those outside of the health sector and people with lived experience of mental ill-health and suicidal behaviour. Critically, this regional focus seeks to deliver on regionally agreed outcomes, optimising the intervention and investment portfolio to achieve the greatest gain (value is defined by the quadruple aim of healthcare) in the most efficient way. This work was made possible by generous contributions of time, local knowledge, and content area expertise by the North Coast Collective through the participatory modelling workshops.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.