How we can best use data for preventative services: a case study

In this article, Barking and Dagenham Council explore how joined up data approaches can help support the transition to more preventative services to help support their residents and avoiding acute services like social care and homelessness.

Introduction

In local authorities up and down the country, councillors and officers are arguing that we must move to a more preventative model of service. The current approach, where local authorities only support people once they meet a high threshold of need, is expensive and means we are waiting for residents to experience poor outcomes before reacting. By working with residents earlier, we can sometimes avoid the need for higher cost acute services such as homelessness and social care.

But if we are serious about prevention then we need to get serious about our data. Prevention requires the ability to identify people that will require support in the future, in order that resources can be used to help them earlier. The good news is that local authorities have fantastic data about their residents, spanning housing, council tax collection, social rent and social care interactions to name but a few. The next piece of good news is that local authorities have a range of statutory duties, including prevention, to use this data to support residents.

Unfortunately, as a sector, we have not invested enough in getting this data into a good place. It is fragmented across systems and sometimes in spreadsheets on hard drives. It tends to sit in silos which are hard to share across.

To really move to a preventative model, we need to bring this data together, to create an overall view of the resident and our total population. How could bringing data together support the transition to more preventative services? There are three things we have been doing in Barking and Dagenham (B&D) to achieve this.

1. Getting data to the frontline

Many frontline officers could support residents more effectively if they had access to wider data held about that resident in other parts of the council. Staff working in non-statutory roles often have very limited access to data and so work with each resident as if it is the first interaction. But if you are working in a community hub and supporting a resident with council tax arrears, it would be very helpful to know if they had wider debts to the council. This could lead to a much more holistic, and therefore preventative, discussion with the resident. The resident might be offered a different level of support, and one which addresses multiple issues.

For example, Ali came to Barking Learning Centre with his child, saying that he had no money for food. His landlord had increased the rent. Ali had sent his work coach his updated tenancy agreement a few months ago, but there were issues with the document and his Universal Credit (housing element) had not been increased. Most of Ali’s benefit was now going towards his rent. When the hub officer spoke to DWP, they said they did not want to give Ali a loan as on their system, he was shown to be in debt. The officer checked One View with Ali’s consent, and it showed that he had a small debt of approximately £200, for a previous DWP loan that he was regularly paying off. Using this data, the officer could make Ali’s case and DWP agreed for him to have another loan while he waited for his universal credit to be updated. Ali also registered with the Community Food Club to cover his and his son’s immediate needs.

It looks different if you are working in a statutory service, such as the front door of a social care service. This is because there is a lot of data that officers have the legal right to access. The issue here is that useful data can be time consuming to get hold of. For example, finding out whether someone is in social housing and their level of arrears can require emails to other services that don’t have time to instantly respond. But if this data is held in the council, why can’t we automate the process?

In B&D we have given access to wider data to a range of frontline teams; from non-statutory such as Community Hubs, to statutory such as the Multi-Agency Safeguarding Hub. We are working with staff to understand what data is helpful for decision making and how best to present it. We are also allowing staff to more effectively share each other’s work. For example, we have over 300 staff that interact with residents about their finances and often need to do budgeting exercises or a check of eligibility for benefits. This can be a lengthy process. We are saving officer time by creating a single system for this so that staff that work with finances across the council to access previous assessments done by other teams.

2. Proactive prevention

Councils can use the data they hold to identify people at risk of poor outcomes in the future. At its simplest, this is about checking the list of people that the council is about to send to court for failing to pay their council tax. Which of these people have known vulnerabilities? Which of these have never visited our debt advice service? Could an offer of support lead to improved outcomes for the resident and the council?

In one of our pilots, we wanted to provide additional support for residents that had outstanding debt to the council as well as low level mental health needs. One of our Link Workers worked to support Jane, who had been diagnosed with two mental health problems and had mobility issues. Previously her son had helped her with her finances, but this ended when he moved out. After this, Jane went into arrears as she was unwell. The B&D Link Worker applied for a discretionary housing payment (DHP) as well as Severe Mental Impairment discount for council tax. She was connected with the Citizens Advice Bureau to provide support to apply for Severe Disability Premium. Finally, she worked with the Housing Service to consider her options as she was affected by the bedroom tax. This support brought Jane back to a financially sustainable position, where she could again manage on her own and avoided potential poor outcomes such as being taken to court or homelessness. She was delighted with the support she received from the council.

But at the more sophisticated end, we hold a lot of historic data that can help us understand which risks are correlated with poor outcomes in the future. In B&D, working with the Centre for Homelessness Impact and four other councils, we will launch a model that predicts risk of homelessness this year and we will proactively offer support to the residents that we identify. We expect this to reduce homelessness as well as save the council money.

Colleagues in the health system tell us they have exactly the same issue. They work with people when they are in acute need but hold plenty of data about the risk factors which are well correlated with future need. Indeed, the health system presents an even starker use case for proactive support, with health leaders regularly saying that huge amounts of hospital spend is on preventable (or at least delay-able) conditions.

It’s hard to move to a preventative model while still dealing with the acute need of today. But why can’t we pick a few big drivers of demand, ones where we know prevention works, and show the value of a preventative approach? We have started by exploring whether we can prevent falls among older people, a large driver of cost for local authorities and the NHS. We are in discussions with integrated care board colleagues about building a joint model around falls risk and designing a joint pathway for the people we identify.

3. Data driven decision making

We also need to get better data to commissioners and leaders. Despite councils holding a lot of data, it is usually not structured in a useful way to understand aggregate demand for a service or how that is changing over time. Legacy casework systems are good at presenting data about a particular individual but inadequate for questions like: what are the risks in our population of children? Is that changing over time? Are there factors that are correlated to particular risks? How much more provision of service x do we need next year?

In B&D we are experimenting with this. We successfully automated large parts of the data aspects of our Supporting Families Programme: identifying eligible families and also capturing where outcomes were achieved. We are designing a dashboard to help us to understand how many care placements we are likely to need in the coming years, and also what the risks are across our children’s population.

So the data held by local authorities presents huge opportunities. We need to think carefully about how sophisticated data analytics is acted upon. Preventative services must be human-centred and relational. The nightmare scenario is that people are sorted into categories and then dealt with only according to what the data says. A lot of people are nervous about data work for this reason. We must acknowledge that the data held by councils is imperfect.

Can data help identify more vulnerable residents and support a more holistic conversation? Certainly. But this conversation should be open, and residents should lead. If you call someone who looks like they are at risk of homelessness, but they want to talk about early social care needs, then that may well be the focus of the conversation. Risk factors are linked and the routes back to self-sufficiency can be oblique.

Of course, there is a significant amount of work to do to ensure GDPR and other data sharing legislation is adhered to. But the prevention duties for local authorities are clear and the key data owners are likely to see the benefits of data sharing for preventative work. Similarly, focusing on equalities, you need to ensure you are following best practice to ensure any predictive models are not biased. But once this is done, these kinds of approach can further your equalities agenda.

After all, not everyone knows the council has a free debt advice service, or what to do if they are facing homelessness. This is especially true in a borough like B&D where we have a lot of residents that have moved here from other parts of the world. Finally, having an ethical governance process to oversee your data innovation is crucial. This is about ensuring visibility of the work and putting in place checks and balances to ensure that the way that data is being used feels like the right thing to do. This includes engaging residents directly about the approach.

Fundamentally, using data to do prevention is about ensuring that you are clear about who your most vulnerable residents are and working with them accordingly. Data is a helpful addition to all the other things that local authorities do identify and support their vulnerable residents. But it should be the start of a conversation rather than a fixed approach that reinforces the problems of service silos. This could also be the way that local authorities need to work with AI more generally. Build good data structures, be conscious of their imperfections and use this to layer on additional support in order to prevent poor outcomes for residents.