Roberta Piroddi

In the UK, suicide is the leading cause of death for men under 50, while it is estimated that 17% of women will self-harm at some stage in their life. It is also estimated that the economic cost of suicide is £6.5B a year. Suicide and self-harm prevention are thus a national priority. 

Suicide and self-harm are preventable, and if there were ways of finding out with more precision who would be more vulnerable to harm themselves, it would be possible to help these people, their families, friends, and communities better. But so far attempts to devise ways to identify people who would need more support have been unsuccessful. Healthcare services also design interventions to prevent suicide, but there is not enough information to ascertain whether they are useful and are reaching those who need it most. 

In Cheshire and Merseyside, the NHS, local authorities, and University researchers are working with health, social care, and the public to link together different databases that collect data from hospitals, GPs, mental health care, community care and social care services on the 2.5 million people who live in the area. This is an extraordinary resource containing much more detail than most. This project will find out whether this new data resource can help to prevent suicide and self-harm. 

Firstly, this project will use linked data to identify which groups are more at risk to be in a crisis, harm themselves or die by suicide. 

Secondly, it will look at current approaches the NHS and local authorities use to prevent suicide and self-harm and assess whether they are working. Finally, these analyses will point out at changes to approaches to prevent these tragedies so that potentially more vulnerable people can be identified earlier and be provided with better support. The project will involve members of the public working alongside researchers and people working in the NHS and local government at each stage of this research. 

Since suicide and self-harm prevention is a priority for the NHS, the results of this research will be useful across the country. 

The first work-package, WP1, will develop a richer picture of risks and their mechanisms. 

WP1 will develop analysis and models of risk by using whole-population individual linked data available in CM from 2015 to the present to develop new risk segmentation models on suicide and self-harm related outcomes: deaths, admissions, and A&E attendances, to identify recurrent characteristics of population groups at high or increasing risk of these outcomes. This will provide a richer description of emerging risk groups enabled by a wide range of potential variables available in CIPHA including: demographics, ethnicity, diagnoses, disabilities , prescriptions, historical utilisation of health and social care services, informal caring roles, household and housing characteristics, welfare benefit receipt, previous substance misuse treatment, domestic violence, contact with police and neighbourhood characteristics (e.g. deprivation, access to green space etc.) 

First, longitudinal regression models will explore the relationship between outcomes and variables. Then a predictive model will generate a heat map of increased probability of harm, which will be the input for a cluster analysis to discover groups of populations at increased risk. The output of the cluster analysis will be communicated to practitioners (WP3) as a model of needs to inform intervention design. Furthermore, descriptive statistics of accumulated trajectories will serve as a basis to conduct mediation analysis to gain greater understanding of potential modifiable characteristics of risk group. 

The second work package, WP2, will evaluate existing interventions for self-harm and suicide prevention in at risk groups. 

Working with the Cheshire and Merseyside suicide prevention group, I will initial scope the existing interventions across the region and the risk groups they target. This will be used to highlight potential mismatches between the risk profile highlighted in WP1 and the basket of interventions currently resourced. Based on this, up to 3 interventions will be selected for evaluation, that potentially target groups highlighted in WP1. Using methods developed through RICE, that apply a family of quasi-experimental methods in linked data (propensity score matching, inverse probability weights, difference-in-differences and synthetic controls) I will estimate the impact of these interventions on self-harm and suicide, using CIPHA data. Examples of interventions that could be evaluated include 1) Intensive Support Pathway (Contact+), 2) Hospital Outpatient Psychotherapy Engagement (HOPE), 3) community approaches to compassionate structured contact (PRISM and CLEAR 3) including the Life Rooms.  

With these analyses, I will investigate how effectiveness varies across risk groups enabling intervention approaches to be fine-tuned to enhance effectiveness and better target at risk groups. I will place emphasis on evaluating the differences in outcomes between socioeconomic groups, to indicate potential effects of interventions on mental health inequalities. To make sure this work has the most impact, I will embed it within the CIPHA platform, enabling sharing of algorithms across the CIPHA network of university, NHS, and local government analysts, so these methods can be rapidly re-applied to evaluate the impact of future interventions as they are developed. 

The third work package WP3 will embed the use of suicide and self-harm risk segmentation and evaluation evidence in the integrated care system. 

WP3 will enable analytical outputs of prediction and evaluation in WPs 1 & 2 to inform effective preventive action. Working with my advisors at Mersey Care and other senior stakeholders, I will map local decision-making, identifying high-level system requirements and opportunities for risk segmentation and evaluative evidence to influence service design, iteratively producing a decision process map, requirement analysis and a system design. With these interactions I will understand enablers and barriers to the adoption, to co-produce and promote adoption of risk segmentation tool developed in WP1 and evaluative evidence from WP2. As I will be embedded in an NHS 

organisation, I will conduct this work organically thought daily interactions which normally also include participation to analytical and R&D networks, and clinical focus groups. 

To help senior managers and clinicians understand the value of the proposed analytical innovation, I will carry out four workshops with senior stakeholders to ensure that there is a coherence of understanding of the purpose of this work and that the perspective users co-produce the innovation and feel ownership of it.  

The evidence from this work, a generic model for risk stratification and the evaluation of interventions designed according to national priorities, have national and international relevance, so I will also disseminate results and findings beyond CM accordingly.  

The research will specifically focus on underserved groups, including people from socioeconomically disadvantaged backgrounds, ethnic groups, people with learning or physical disability, people living with multiple long-term conditions, migrants, asylum seekers, homeless people. It will also evaluate whether these groups are accessing services and whether services are effective for these groups, leading to the tailoring of services so that they better support underserved communities. 

Through dissemination and knowledge translation, uptake of the tools and evidence developed through my research will lead to wider benefits nationally and internationally.