Background Efforts in the USA during the 21st century to stem the ever-rising tide of suicide and risk-related premature deaths, such as those caused by drug intoxications, have failed. Based primarily on identifying individuals with heightened risk nearing the precipice of death, these initiatives face fundamental obstacles that cannot be overcome readily.
Objective This paper describes the step-by-step development of a comprehensive public health approach that seeks to integrate at the community level an array of programmatic efforts, which address upstream (distal) risk factors to alter life trajectories while also involving health systems and clinical providers who care for vulnerable, distressed individuals, many of whom have attempted suicide.
Conclusion Preventing suicide and related self-injury morbidity and mortality, and their antecedents, will require a systemic approach that builds on a societal commitment to save lives and collective actions that bring together diverse communities, service organisations, healthcare providers and governmental agencies and political leaders. This will require frank, data-based appraisals of burden that drive planning, programme development and implementation, rigorous evaluation and a willingness to try-fail-and-try-again until the tide has been turned.
- public health
Statistics from Altmetric.com
Suicide rates in the 21st century continue to rise in the USA despite the efforts of many states and localities; the formation of the National Action Alliance for Suicide Prevention, a public–private partnership and its publication of a second-generation national strategic plan in 20121; the initiatives undertaken by the US Departments of Health and Human Services, of Veterans Affairs and of Defense; the deep commitment of many non-governmental organisations, family members of persons who had killed themselves and attempt survivors; and new research into the causes of suicide and the development and dissemination of ‘evidence-based’ approaches to intervention and prevention. Suicides increased from 29 350 during 2000, in a population greater than 282 million, to 44 193 during 2015 from among more than 321 million; rates per 100 000 changed from 10.40 (2000 base year) to 13.75 (13.26, age-adjusted).2 They grew from 17.67 among men and 3.95 among women to 21.48 (21.00, age-adjusted) for men and 6.25 (6.04, age-adjusted) for women. The rise in suicide rates was driven especially by men and women in the ‘middle years’ of life, 35–64 years old, where there was an increase of 35.3% (13.9–18.8 per 100 000) from 2000 to 2015. Except for men 75 years and older, no age group showed a decrease.
While reducing suicide rates in the USA remains a knotty, vexing problem, it is a winnable battle.3 Evidence regarding the effectiveness of initiatives to promote means safety,4 5 as well as other specific methods of prevention,5 has been accumulating. However, there are scant data to suggest that any single approach will have a substantial impact on reducing rates.5 The US Air Force implemented a coordinated strategy involving multiple initiatives simultaneously,6 7 but no such comprehensive, integrated efforts that embrace all ages, sexes and identified vulnerable populations, as well as incorporating local community priorities, have been undertaken to date by any state or region in the USA. This paper will lay out a series of tasks that comprise what would be involved in such an effort, illustrating them in part based on our developing experience in Colorado, a state that has set a goal for reducing suicides by 20% by 2024.
In addition to past tendencies to implement one-by-one interventions rather than a group of interlaced programmes that constitute a mosaic of actions,8 there are inherent barriers to effective prevention,9 especially as programmes have focused on seeking to detect persons at high risk as they enter perisuicidal periods and approach the precipice of death. These include: an inability to discriminate the relatively few true cases from the large number of false positive cases, that is, the preponderance of persons suffering severe psychological distress with diagnosed psychiatric disorders who, even while suicidal, neither attempt nor die by suicide; the large number of false-negative cases that escape preventive detection, that is, persons who seemed ‘normal’ but killed themselves or whose long-standing condition appeared no worse to family, friends or providers during the days immediately before their death; the inability of clinical and social services to reach many individuals who have serious suicidal intent; a continuing paucity of knowledge about fundamental biological, psychological, social and cultural factors that contribute to fatal suicide attempts among diverse populations and groups who have been labelled at-risk; and the lack of coordinated strategies for suicide prevention that can deal effectively with myriad local, regional, state and national agencies and organisations that could, in theory, play a role in preventing suicide. By exclusively seeking to identify highly vulnerable individuals at times of crisis, past efforts have not looked upstream to discern distal common risk factors that underpin multiple aetiologically related forms of premature death, including suicide, deaths from drug self-intoxication and deaths from risk-related medical disorders.10 Upstream prevention efforts reach beyond perisuicidal periods to times when no suicidal thinking or behaviours may be apparent; while this may not seem to be suicide prevention to some observers because vulnerable persons have yet to reach symptomatic states, efforts to prevent fatal cardiac events or challenging-to-treat cancers often begin decades before their apparent onset with efforts to stop smoking, reduce high blood pressure and promote a healthy lifestyle. Added to these challenges, past prevention efforts understandably have tended to reflect the priorities of focused interest groups that have driven agendas based on their personal experiences rather than assessing the greatest potential contributions to reducing the collective suicide burden.8 The challenge now for the field is to build partnerships encompassing a broader range of interest groups to foster priority settings across the prevention spectrum. This includes establishing buy-in from both traditional and non-traditional partners in public health, including those implementing violence prevention, substance abuse prevention, healthcare, criminal justice and child welfare, and employers as well as members of the workforce.
This paper does not focus on a conceptual discussion that emphasises the contextual nature of suicide, attempted suicide and other risk-related adverse outcomes (see ref 8). Rather, we highlight practical steps necessary to initiate, develop, implement and evaluate a community-based, systemic strategy that integrates the efforts of diverse stakeholders and approaches to build comprehensive, sustained, state-wide public health prevention programmes that engage diverse populations.
With this framework in mind, colleagues from the CDC-funded Injury Control Research Center for Suicide Prevention began working with leadership and staff from the Colorado Department of Public Health and Environment (CDPHE) to develop a systemic (far reaching, all-inclusive) approach to prevention that would cover the entire age range, with the intent of focusing on broad populations of highest burden while also attending to the specific needs of especially vulnerable groups. The goal has been to develop a comprehensive, upstream–downstream approach to prevention that combines tailored community-level efforts with those of established health systems, where the latter addresses the needs of highly vulnerable or severely distressed persons, appreciating that ‘community’ may involve populations that share common characteristics as well as geographically defined localities.
Why Colorado? It ranks among the leading states in terms of relative suicide burden, with 1093 suicides in 2015 among a population of nearly 5.5 million—the ninth highest rate in the USA at 20.0 per 100 000 residents. Colorado sits amid the western mountain states where suicide rates are highest nationally. Most important, there have been abiding commitments shown by the state’s political and community leaders, with roots reaching back to the 1950s. They established the Suicide Prevention Commission and funded the Office of Suicide Prevention (OSP) as part of the CDPHE. Colorado already had begun several important initiatives by 2015 when we initiated our collaborative discussions and shared a philosophy that public health approaches to prevention should embrace multiple community and health system resources, with a recognition of the importance of a common risk model and the need for context-embedded approaches to preventing premature death from suicide and other risk-related causes, such as fatal drug intoxication.8
Like other collaborative partnership processes, efforts to prevent suicide, attempted suicide and risk-related adverse outcomes depend on building trust, sharing perspectives, defining common goals and objectives and understanding areas of potential disagreement. Our process has developed in a fashion consistent with the guiding principles recently published by the National Action Alliance for Suicide Prevention (NAASP).11 These include: unity: attainment and maintenance of broad-based momentum around a shared vision; planning: use of a strategic planning process that lays out stakeholder roles and intended outcomes; integration: use of multiple, integrated suicide prevention strategies; fit: alignment of activities with context, culture and readiness; communication: clear, open and consistent communication; data: use of surveillance and evaluation data to guide action, assess progress and make changes; and sustainability: a focus on long-lasting change.
In 2015, a team of national-level researchers and Colorado prevention leadership identified a shared goal of reducing suicide by 20% by 2024 through a collaborative research-practice model. Because strong leadership and group efficiency have been shown to improve implementation of collaboratively run prevention initiatives,12 the team founded a formal structure—the Colorado National Collaborative (CNC). CNC developers strategically approached local and national stakeholders with diverse public health expertise for participation in the CNC. This included upstream prevention practitioners, mental health treatment providers, state and local government representatives, hospital association representatives, local and national suicide prevention researchers and evaluators and other stakeholders in suicide prevention (see figure 1).
As early team functioning is a critical indicator for future sustainability of collaborative efforts,13 invited participants were brought together repeatedly and consistently, in person and virtually, to solidify a shared vision and identify the unique role that each sector representative might play. Assuring the centrality of state and local leaders rather than external academics, the CNC developed a clarified governance structure; it serves as a collaborating element to the Colorado Office of Suicide Prevention. The latter houses the coordinating centre for developing activities and, in turn, is accountable to the Governor’s Health Cabinet (see figure 1). As the CNC has grown, membership has evolved. Drawing on the NAASP principles of unity and planning, CNC members regularly review the list of active local and national members to explore gaps in sector representation.
Diverse CNC membership is critical to ensuring that stakeholders from a broad range of community sectors have an opportunity to express their unique perspectives and priorities.
The CNC moved from a general discussion of principles—based on a collective vision of developing fundamental systemic changes14 to exploring the geospatial distribution of burden ‘centers’ and county-level burden drivers, based on the efforts of CDPHE epidemiology staff using 2010–2015 data collected for the National Violent Death Reporting System (NVDRS). These allow an in-depth appraisal of local variations of suicide and differential geospatial expression of contributing factors to diverse-distinctive subgroups, as well as assessing changing patterns over time. This is being coupled with an ongoing state-level and community-level inventory of prevention efforts, broadly defined, that is collecting data from state agencies, non-governmental and voluntary agencies and health systems. Taken together with NVDRS data, it becomes possible to identify those populations that are covered by preventive interventions and those that are missed—shining a light on gaps—an essential step when developing an adaptable strategy to reduce overall burden. Drawing on the NAASP guiding principle of using data for planning, these methods offer local community leaders inspection of home town data rather than a mash of less-specific state or national level statistics, allowing further identification of any missing local prevention programmes and collaborative consideration of programmes that best fit local needs.
Who, where and what
When working with communities and diverse potential stakeholders, it is especially useful to have a framework for defining who, where and what preventions to propose. The first grows from available data regarding those populations that contribute most to burden, as noted above, which naturally leads to a consideration of what to do and where to do it.8
Every community setting (eg, schools, workplaces, medical providers, the courts and residential settings) are part of an overall local fabric—a local social geography—where one can describe each site in terms of its sample bias, that is, the populations or groups that may be engaged and those that will be missed. Each sample bias must be considered when developing a comprehensive strategy that offers the potential of engaging groups and persons of interest. For example, schools and universities are settings for prevention efforts that offer access to a broad array of youth and young adults. Yet, they exclude those who have dropped out—especially, those who are vulnerable by dint of their drug use or legal problems—or who have joined the workforce. Primary care providers often are chosen to offer preventive interventions; while this may be well suited for elders who see their physicians at regular intervals, many men in the middle years of life do not participate in regular medical care.
Complementary to social geography, it is possible to track especially vulnerable individuals and groups through localities to determine the settings (including those using social media) where they may be encountered, that is, defining the social ecology for distinctive groups that have been exposed to social, family and personal adversities that may impart heightened levels of risk. By combining an understanding of social geography and social ecology, it becomes possible to start building the spatial foundation for a community-level strategy to reach those identified in the NVDRS-derived picture of local suicide burden.
Suicidal behaviour and interpersonal violence have been considered within a social-ecological framework.8 15 Used to organise the myriad factors that contribute to violent behaviour, this framework also can be used to clarify the level of action of protective factors and guide the development of interventions at individual, relationship, community and societal levels (see figure 2). The challenge that ensues relates to defining which interventions to emphasise and in which settings to deploy them for the general population and for those of special interest.
The Suicide Prevention Resource Center maintains guidance for ‘evidence based’ interventions (www.sprc.org/search/evidence/based/intervention). This, however, does not define how one might best integrate this diverse array of interventions into a comprehensive, integrated set of programmes that reach across the entire age range. The LifeSpan programme (http://www.lifespan.org.au) in New South Wales, Australia, has begun to implement a systems model much like the one being developed in Colorado that seeks to build an integrated approach to prevention. LifeSpan has identified nine overlapping elements: aftercare and crisis services; psychosocial and pharmacological treatments; guidance and support of primary care providers; training frontline staff; gatekeeper training; schools programmes; community campaigns; media guidelines; and means restriction. This effort gives major attention to ‘what’ and indications of ‘who’ and ‘where’.
Similar to the LifeSpan programme, Stone and colleagues at the CDC National Center for Injury Prevention and Control recently promulgated a technical report on strategies and related approaches to prevention—considered within an ecological framework—that offer evidence of positive impact.16 They focus on seven basic strategies: strengthening economic supports; strengthening access and delivery of ‘suicide care’; creating protective environments; promoting connectedness; teaching coping and problem-solving skills; identifying and supporting people at risk; and lessening harms and preventing future risk.
By integrating data regarding suicide burden defined by age, sex and other demographic features, together with local information regarding social geography and the social ecology pertaining to groups of interest, one can create ‘maps’ that can serve as a basis for establishing age-specific deployment of preventive interventions, as illustrated in figures 3–5. These maps depict where communities need to focus when deploying preventative interventions. Circles relate to settings conducive to engaging populations of interest with intersections arising from individual mobility; bold lettering emphasises the importance of programme development in specific areas where it is more likely to encounter either larger (although with lower rates) or potentially important groups; and capital letters are indicative of settings likely to present a higher percentage of higher risk groups and individuals. Fundamental to this approach is the recognition that prevention efforts are diverse yet potentially specific depending on the demographic and social features of targeted populations—in terms of time, space and what will be needed to respond to the extraordinary heterogeneity among those who kill themselves. The maps underscore that the settings suitable for one age group—or for subgroups each having a different social ecology‚—will differ from another and point to the fundamental need for developing community-integrated prevention offerings.
Bringing process and content together, the goal is to ‘place’ evidence-based and best-bet preventive interventions with the contexts that comprise what we call ‘communities’, appreciating that communities are recognised as the basis of their common interests and life experiences, as well as their geospatial distributions.17 They potentially will have different components or areas of emphasis depending on age, sex, sexual orientation and gender identification, cultural make-up, economic factors, health status and identified population-level risks, the latter being arrayed in a social-ecological framework.8 Interventions such as the development of means safety programmes reach across communities and depend on the involvement of selected participants (eg, gun shop retailers, pharmacists, urban planners and architects designing buildings and public spaces); others, such as the careful individual tracking of persons who have attempted suicide, benefit from alliances between health and mental health providers and, ideally, family and community partners. It is foreseeable that individual-level programmes that use ‘big data’ and algorithm identification of especially vulnerable individuals can be embedded into systems of care that are comprehensive, far-reaching and linked with computer-assisted communication tools. Dispersed groups or those who may be hidden in the face of social pressure and apparent stigma (eg, LBGTQ youth and young adults; adult men in need of mental health services) may be reached via social media and other niche marketing approaches. In sum, the extraordinary geographical, cultural and ethnic and racial diversity of the USA requires that, to be effective, preventive interventions will require careful consideration of the NAASP principle of fit: alignment of activities with context, culture and readiness. Careful tailoring of preventative interventions will be based on collaborative discussions that address the challenges of time, place, group and individual differences. While top-down leadership and commitment is essential, effective actions necessarily will be ground level—locally energised, developed and owned.
Recognition of such ground-level efforts may require movement from strict adherence to research-based empiricism to the complementary paradigm of ‘community-based empiricism’ and leadership. Practice-based evidence looks to resolve the tension between evidence-based prevention and locally grown efforts by recognising that effective programmes and practices can and do emerge from community expertise.18 Such tensions can be further resolved by a commitment from the CNC to building the evaluation capacity of communities to evaluate home-grown efforts. Valuing community-derived experiences also has the potential to reinforce local commitment and sustained efforts.
Implementation and evaluation
Moving from talk to action poses multiple challenges. To name a few: implement programme elements all-at-once versus a stepwise process; implement programme elements in all target regions or selectively at first; emphasise the evaluation of process indices (eg, training-related evaluations), impact metrics (eg, practice and access to care) changes or ultimate outcome measures (eg, fewer attempts and deaths); define ‘long enough’ to assess meaningful impact; and assure longer term sustainability. Process metrics can be useful for measuring the development of community coalitions, but they do not provide data regarding changes in practice. They are less expensive to deploy and may be useful during start-up phases. They help ascertain what programme implementers are learning and how they are navigating the challenges of building necessary local prevention fabric. Impact measures do not offer a conclusive sense that programmes are ‘working’, yet they can serve as indices of continuous quality improvement, which are central to improving health service delivery and useful as a way of monitoring engagement of target populations. Outcome data reveal whether lives were saved. Communities and advocates understandably want to move quickly and mount all-out efforts to save lives; however, researchers and advocates of rigorous programme evaluation may warn against hasty action, understanding that many efforts that seem to have had ‘face validity’ have failed.8
When the US Air Force began to implement its multifaceted suicide prevention programme in 1996–1997, it had not established an evaluation strategy. The retrospective approach that eventually was used hypothesised that, given its emphasis on upstream factors that apparently contributed to increasing risk for multiple adversities (a common risk strategy), the programme’s impact would influence multiple risk-related outcomes (ie, suicide, accidental death, homicide and family violence).6 When suicide rates climbed in 2004, a retrospective examination of programme and policy implementation allowed for a second level of hypothesis-driven analyses.7
While revealing, this type of implementation–evaluation approach is less satisfactory than prospective designs. Mindful of proponents’ urgency for implementing initiatives, NOW! researchers have been developing novel designs that seek to measure the effects of multiple evidence-based interventions at the same time while measuring overall impact. These methods take advantage of spatial diversity, understanding that there typically is insufficient capacity to implement all actions in all places instantaneously, and benefit from limited training capacity, recognising that most often it is not possible to train at once all programme implementers to high levels of programme fidelity.19 20 Another methodology that may assist with these efforts involves system dynamics modelling as a way of conceptualising the dynamic forces influencing suicide, attempted suicide and risk-related premature deaths.21
When implementing a comprehensive set of initiatives, including both upstream (distal) and downstream (more proximal) intervention efforts to address the needs of acutely distressed individuals, it will be possible to examine outcomes that are socially and clinically important, even as they may be ‘outside’ the immediate realm of suicide and suicide-related thoughts or behaviours. The US Air Force programme emphasised early interventions to stem the severity of family violence, even as it sought to reduce suicides. Thus, it was possible to test for a hypothesised increase in the frequency of mild family violence coupled with a decrease in moderate and severe violence—itself a positive outcome.6 Similarly, there are multiple interventions for youth, adults and elders that can yield quantitatively defined, personally and socially meaningful ‘intermediate’ beneficial results, arrayed along a hypothesised set of pathways-to-suicide. These types of predefined outcome metrics can provide confidence that programmes are building the desired impact long before more distal outcomes—declining rates of suicide attempts and suicides—can be measured consistently year over year. To bring these many action steps into a more cohesive format, the CNC developed a logic model to organise its efforts (see figure 6). This outlines the paths that we are taking to implement our efforts in a coherent and transparent fashion.
Suicide prevention efforts have tended to be one-by-one initiatives rather than comprehensive efforts that bring together communities, state agencies, health systems and diverse stakeholders to work in a synergistic fashion that pushes forward multiple efforts simultaneously. Suicide is not a singular problem, or a specific medical diagnosis. Rather it serves as a final common pathway for an array of elements reflecting personal, family, community and societal stresses and turmoil—typically expressed one individual at a time. While the final moments of action—killing oneself—predominantly have drawn past attention from medical and mental health professionals, it is timely to integrate the person level with what can be done in both health systems and beyond their walls across entire communities, and far upstream, so that it is possible to alter life trajectories.
Such efforts require leadership and sustained commitment, a willingness to partner with diverse groups, an openness to strict scrutiny and evaluation by others, the integrity to acknowledge failures as well as proclaim successes, a willingness to go beyond what is known with certainty and the realisation that drastically reducing the burden of suicide and risk-related premature deaths depends on fundamental social changes. While these efforts must be based on best evidence and conducted with scientific rigour, they ultimately depend on an unstinting courage of convictions that saving lives depends on a collective societal willingness to make a difference in the lives of profoundly vulnerable persons, families and communities.
What is already known on the subject
To date, there have been no sustained, far-reaching state-level or regional suicide prevention programmes in the USA that comprehensively involve populations across the life span and combine community-driven initiatives with those of health systems and other potential partners.
Most suicide prevention initiatives in the USA have focused on individuals deemed to have imminent risk.
It has been very challenging to develop comprehensive public health suicide prevention programmes that integrate upstream (distal) and downstream (proximal) strategies.
What this study adds
Discusses practical steps for the development of a comprehensive, integrated community-driven suicide prevention programmes.
Considers the barriers and opportunities for successfully implementing such efforts.
Thanks to colleagues in the Injury Control Research Center for Suicide Prevention, the Colorado Department of Public Health and Environment, the Colorado Suicide Prevention Commission and the Colorado National Collaborative.
Contributors All authors contributed to the planning of this paper and the developments described for the Colorado National Collaborative (CNC). The logic model was developed by KQ based on the discussions of the authors and colleagues. EDC drafted the paper and the remaining figures, which were reviewed and approved by all authors.
Funding This work was supported in part by a grant from the US Centers for Disease Control and Prevention (CDC; R49 CE002093).
Disclaimer The content is solely the responsibility of the authors and does not necessarily represent the official views of CDC.
Competing interests None declared.
Provenance and peer review Commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.