The Coda Network
In 2021 the Chief Medical Officer looked at health in coastal communities, and highlighted that the available data on health and wellbeing were poor and lacked granularity. He recommended that this should be addressed to support the development of policies aimed at improving the health of coastal communities.
The Eastern Arc Coastal Data (‘Coda’) Network was established to provide a framework to address this. It is intended to be responsive to the needs of its members, to explore opportunities to collaborate and facilitate the robust collection, sharing and analysis of data, and to identify ways in which we can work together for the benefit of our coastal communities.
If you want to get involved with the Coda Network, email us with your details and interests.
Terms of Reference
Our full terms of reference are available here.
-
In July 2022 Eastern Arc hosted two workshops aimed at discussing, with key regional stakeholders, the landscape of coastal health data including strengths and weaknesses. The first workshop was run virtually while the second in-person, and over 50 participants took part, including academics, researchers, data analysts and data owners. Participants represented over 25 organisations, including universities, NHS trusts, local authorities, charities and third sector organisations.A significant range of issues were identified through this process. Importantly, participants agreed that the issues of availability, access and applicability of data were not restricted to the coast; they were common issues faced across the country.
Rather, it was the specific characteristics – and underlying determinants – of the ‘coastal excess of disease’ that were unique, and for which the common problems of data collection and management needed to be addressed.
What follows is a qualitative summary of the key points raised across the two days, sorted into the three categories specific to data, identified in Whitty’s annual report: access, availability, applicability.
Access
Restrictions to access – and delays in processing applications for access – were often the result of a risk-averse data owner. There was a culture of nervousness and fear about sharing data, particularly with regard to GDPR implications of sharing sensitive personal data, which resulted in owners rejecting requests by default rather than exploring if and how access may have been possible.
“Governance should be an enabler” (participant, Workshop 2)
It was recognised that there was a need to empower data owners and give them the confidence to consider requests and try to facilitate access. This could be engendered through training, or the development of an agreement template that made decision-making easier or mitigated issues of blame. For instance, a better understanding (and application) of the principles of anonymisation or pseudonymisation may help to allay fears about what is being made available.
However, allowing data owners more time and consideration of requests could take more resources, which local authorities and NHS trusts may not have. One way of facilitating access was to develop long-term partnerships of trust between researchers and data owners. However, it was sometimes difficult to identify who the data owner was, or have a structure by which partnerships of trust could develop.
“Data ownership needs to be more porous” (participant, Workshop 2)
Another potential solution could be the development of an agency, such as the Office for National Statistics (ONS), which could safeguard the curation of data, or negotiate a common release contract.
Such an agency would help to build public trust in the use of data. Although the public was broadly happy to share significant data through social media, it was wary of government agencies collecting it.
The number of agencies involved in a particular area was also an issue. A county may have one council but three or more NHS trusts and multiple boroughs.
“[The county of] Essex is the perfect storm” (participant, Workshop 2)
Historically, there had been little integration between datasets owned by different organisations, and access to different datasets varied.
This might change with the new integrated care systems (ICSs), and some agencies had already managed to overcome the problem, such as the Suffolk Office of Data Analytics (Soda).
Cohort studies, such as Born in Bradford, were also mentioned as good examples to gather information in a specific area, but such studies were expensive, complex and required a long-term commitment to their continuation.
Availability
Many users were unaware of the full range of datasets available. There was a need for a central registry of datasets with accurate metadata, so that analysts could understand the parameters for each set, its quality and how comprehensive it was.
“A significant amount of data is captured, but we don’t know its importance, or the importance of it when it is combined.” (participant, Workshop 2)
“We need to do more with the data we have” (participant, Workshop 1)
Some collections of datasets did exist, such as Fingertips, but none were comprehensive. Not knowing what had been done or what was available led to duplication of effort.
As well as a registry, a network of coastal data users would help share best practice, resolve problems and identify common issues. No such network currently exists.
Data at an appropriate level of granularity was rarely available. This might be, for instance, at regional or local level, or for subgroups of the population, or those that weren’t engaging with the healthcare system. In addition to more granularity there may be the need to collect more qualitative rather than quantitative data.
Limited granularity could also refer to other facets of the data as well, such as their timeframe, the frequency of collection, or their depth.
Some data weren’t captured or stored because the costs of doing so (and managing it in the long-term) were prohibitive. This had resulted in the private sector taking on this responsibility, but then charging organisations to access their own data.
Basic data may have been captured but this was only a fraction of the whole; behavioural drivers, for instance, weren’t available or understood. In addition, the data were often not longitudinal, and only tell part of ‘a patient/individual’s story.’ There was a need for the capture of qualitative as well as quantitative data, and for this to be made widely available.
“We don’t know what we’re not getting” (participant, Workshop 2)
“The data we want the most is the hardest to reach and keep up to date” (participant, Workshop 2).
Applicability
Although, anecdotally, there was an understanding of what a coastal community was, there was no officially-accepted definition. Identity could come from socio-economic indicators; communities that have higher indices of multiple deprivation may identify more closely to their geographic location than those with lower indices.
Many communities were weary of new external interventions, such as academic-led projects for smoking cessation or adopting a healthier diet; it was important that those capturing data worked with the communities in co-creating projects, but also in sharing the findings of the research and engaging fully with the participants and wider public.
In addition, there was the question of selection bias and the difficulty of reaching those in the most deprived communities. This had been exacerbated by the Covid pandemic.
Where data were available, it was not always possible to fully interrogate them, due to elements being missed, such as common subject identifiers. This might be because the data had been collected for a specific purpose, and the missing elements made more general or secondary analysis difficult.
“How good is good enough?” (participant, Workshop 2)
Although local authorities captured a significant amount of data, few had the capacity to analyse it in-house. The involvement of analysts at an early stage of the data collection was important: they could help develop the necessary framework that would make the data relevant and useable.
Where a project had been undertaken and data gathered, it was difficult to get findings published if they demonstrated that an intervention wasn’t working; there was less interest from editors and journals if this was the case. As such, there was a danger of a costly duplication of effort due to a lack of awareness of the outcome of the original project.
Finally, there was a need to train those who would use the findings of data, such as politicians and policy-makers, to understand the capacity and limits of the data, and make allowances – or commission new data – accordingly.
-
Following the workshops to identify underlying issues, Coda held a workshop to build on these findings, to identify best practice, and to sketch out a way forward. Opportunities for using information to improve the health of coastal populations and reduce inequalities
Dr Tim Winters, Senior Health Intelligence Lead (Insight & Analytics), NHS Norfolk and Waveney ICB
TW outlined the work of the Norfolk and Waveney ICB Business Intelligence (BI) insight and analytics team. NWICB is responsible for about 1,032,000 people.
- 1 in 4 are 65 and over
- 53% live in urban areas
- 21% live in market towns
- 26% live in rural areas
- 1 in 5 live in a coastal community, including Hunstanton, Cromer, Great Yarmouth and Lowestoft, which are in the top 20% most deprived communities in the country.
The team had broken the coast down into three areas: West (26k residents), North (40k residents) and East (136k). The age distribution varied along the coast, and there was a significant difference in life expectancy, between 75 (M) and 81 (F) in Great Yarmouth, to 82 (M) and 86 (F) in Southwold. Cancer and CVD are responsible for more than 40% of this gap.
TW suggested that the ICB could work to address this by:
- Working with people to change health behaviours
- Ensuring better access to care
- Focusing on better quality of care and improving patient engagement.
Robust data collection and analysis would play an important part in this. Due to inequality in health behaviours (smoking, poor diet, excess alcohol consumption, inactivity, obesity), the opportunities for improving outcomes were likely to be greater in the most deprived communities. However, people from more deprived communities were less likely to recognise symptoms and are more likely to report barriers to seeking help.
The ICB has significant datasets that can help with this.
This can be used to identify and map populations most at risk. For instance, those at risk from smoking and obesity in the Great Yarmouth area.
Equally, it can be used to identify populations who have engaged less with monitoring, such as bowel cancer screening.
Such identification and targeting could have significant benefits. For example, if high blood pressure was detected and 80% was targeted, over three years the region could prevent more than 100 heart attacks, more than 150 strokes and more than 80 deaths (and save £3 million).
The ICB had an effective data framework to help with this. It used the Johns Hopkins population segmentation model. Through this, it identified eleven mutually exclusive groups and provided a whole person approach along the life course that incorporates mental and physical health.TW demonstrated ways in which this was used, such as for those with frailty.
TW finished by asking what the desired outcome would be from using the data, and ‘who could do what differently if they had better information?’
Using Data to Drive Activity in East Suffolk
Nicole Rickard, Head of Communities and Leisure, East Suffolk Council, and Jeptepkeny Ronoh, Consultant in Public Health Medicine & Assistant Director – Place Lead (East Suffolk), Suffolk County Council
NR and JR outlined a number of different data-based interventions that were being implemented in East Suffolk. The district was home to over 250k residents.
- Over 27% are over 65.
- Over 96% of the population is white.
- Over 30k people live in ‘Core20’ areas.
- Broadly, more people in East Suffolk are affected by income deprivation than in Suffolk overall (11.5% compared to 10.1%); fewer people have level 3/4 qualifications, and almost 20% of the 16-64 population have no qualifications.
- Activity levels in children lower than average and obesity higher
Two ICBs were responsible for the district, and eight community partnerships had been set up to ensure a connection with communities. They had run a number of interventions, including:
- ‘Ease the Squeeze’: to support residents with the rising cost of living, co-designed with partners.
- ‘Well Minds East Suffolk’: to address mental health, including training delivered by Suffolk Mind for voluntary organisations and community groups.
- ‘Keep Warm and Well (Warm Homes)’: a pilot with N&W ICB/Great Yarmouth BC
- ‘Feel Good Suffolk’: to healthy lifestyle services – smoking cessation, physical activity, and healthy weight.
- Holiday Activities and Food (HAF) Programme
- Public Mental Health programme, with four pilot areas in East Suffolk
JR talked in more detail about ‘Lowestoft Healthy Hearts’. Lowestoft had been identified as an area for the hypertension intervention as it performed worse than the rest of the Waveney area on many health indicators. Hypertension is the largest single known risk factor for cardiovascular disease and related disability, but it’s also a doorway to addressing other issues. She also outlined the ‘Community Voices’ programme, which enabled face to face engagement with residents in a range of community hubs across the town.
Are data owners risk averse? Learning from the Covid-19 pandemic in Suffolk
Anna Crispe, Assistant Director, Knowledge, Intelligence & Evidence, Suffolk County Council
AC outlined what SCC had done with data to support the pandemic response in Suffolk, creating ‘CoronaWatch’ and the Suffolk Office of Data Analytics (SODA) Vulnerable Persons Dataset.
At the start of the pandemic they faced many challenges. There was no local data for Suffolk (which was home to 761,000 people), no effective data sharing in place with Public Health England (PHE) on infectious disease, no clarity from PHE on what our role was, and no experience of the data needed to handle a global pandemic (although they had the technical skills and knowledge). However they did have some innovative local data-sharing arrangements in place through SODA, but not systematically across the whole system and not with key partners.
From May 2020 onwards, they started getting access to detailed local data. Individuals had different ‘unique’ identifiers in different datasets, which required complex linking and matching across huge volumes of data, and there was an evolving demand for the data, which led to the development of the ‘CoronaWatch’ system.
Alongside this, SODA was tasked by the Collaborative Communities Board (CCB) to identify individuals and households in Suffolk who may be particularly vulnerable to the impact of the coronavirus either clinically, financially, or socially. This led to the creation of the Vulnerable Persons Dataset.
Over 40 datasets were encrypted before being sent to SODA and uploaded to a secure data warehouse. Suffolk County Council data, UK government data including the Shielding list, and food parcel delivery information were also uploaded.
They linked data on 145,000 individuals in 109,000 households and analysed the depth and width of vulnerability in the population. This was then given to their ‘Home But Not Alone’ Collaboration who used it to make proactive contact with people and families. Over 24k outbound calls were made to vulnerable people, and a specific funding stream to support them.
The experience of the pandemic on data collection and analysis in Suffolk has led to a number of outcomes.
- It clearly identified individuals with both wide and deep vulnerability allowing immediate proactive contact and support
- It helped the whole Suffolk ‘system’ to understand vulnerabilities in its population in new ways, which is now directly influencing resource allocation
- It will help them to mitigate the long term effects of the pandemic, including inequalities
- It made them think much harder about data ethics and put a framework and process in place to support them
AC then led a workshop in which participants considered the barriers and enablers to data sharing.
She finished by highlighting some suggested actions to arise from the workshop, and encouraged participants to commit to them.
Developing a new classification of coastal communities
Prof Sheela Agarwal and Dr Alex Gibson, University of Plymouth
The University of Plymouth’s Centre for Coastal Communities (CfCC) has been funded by UK Research and Innovation (UKRI) to co-design, implement and make publicly available an evidence-based and policy-orientated classification of English coastal communities linked to suitably granular data. This had been one of the recommendations identified in the Whitty report.
SA outlined the project. They were collaborating with a wide range of academic and policy stakeholders, including the Office for National Statistics (ONS) to analyse and classify the economic, social, cultural, historical, and geographic characteristics of English coastal communities, and develop a new interactive map of the coast.
AG demonstrated the draft Coastal England Small Area (CESA) Data Portal, and participants were asked to feedback their thoughts on it, including whether they would use it and what features they would use.
Why do adolescents in deprived coastal communities have worse mental functioning as young adults than their inland peers?
Dr Emily Murray, Director of the Essex Centre for Coastal Communities
EM had recently completed and published findings from a research project on the mental health of young adults in coastal communities, compared to their peers inland. It had used the UK Household Longitudinal Study with follow up interviews.
The results showed that there was a significant difference between the two, with a lower mental functioning on the coast.
She went on to consider possible environmental reasons for this, including economic (such as employment and housing), social (such as crime), education and infrastructure. In almost all measures the indicators were worse for those living by the coast, other than access to green space and air quality.
EM intended to explore these findings in more detail, and undertake subsequent research to better understand the reasons for the difference in mental functioning among adolescents on the coast.
How important is context in the measurement of deprivation?
Prof Andy Jones, Honorary Professor, University of Kent
The concept of rural areas being distinct and the need to map this has been long understood; Paul Cloke (Wye College) had created an index of rurality in 1976, and this had been updated by Peter Bibby and Paul Brindley for Defra in 2013.
However, there was a recognition that commonly used deprivation indices don’t necessarily capture the distinct nature of rural disadvantage. AJ and his co-investigators had attempted to develop an index of rural deprivation that better reflected this. Their conceptual framework included relative household deprivation; population characteristics; locality-related deprivation; and spatial scale.
The result was a better reflection of rural deprivation. He concluded with a number of observations and questions, including the possibility of such an index being adapted for coastal areas. If it was, what measures might be important?
Addressing the recommendations of the Whitty report
The workshop concluded with a discussion about how the Network should develop to directly address the key points of the third recommendation of the Whitty report, which relate to data. These were summarised as follows:
- Review the availability, access and applicability of data on health and wellbeing outcomes and their determinants at lower geographical levels
- Analytical capacity across the system
- Data sharing arrangements
- Undertake further multi-disciplinary research to understand the multiple drivers of poor health outcomes in coastal communities and test effective interventions and solutions.
- Review the actual, versus expected disease prevalence and service provision
- Encourage research on coastal communities in coastal universities
- Review migration patterns at lower level geographies
- Improve joint working between local authorities and academic institutions data sharing arrangements
- Develop learning networks of those leading population health in these areas should be encouraged.
Following this discussion, the following steps were agreed:
- To understand the specific strengths that Coda has to offer. These were, inter alia:
- The significant regional strengths in data collection and analysis in local authorities, ICBs and NHS trusts.
- The existing strong connections between these regional stakeholders and the EARC universities, and a desire to develop these further.
- To use these strengths to act as a ‘test-bed’ for translational research.
- To work with the stakeholders to co-develop research questions.
- To explore ways in which datasets that may not be publicly available could be shared anonymously and in confidence to better address health challenges.
- To act as a facilitator and enabler for those working elsewhere in the country on addressing coastal issues, for collaborations and comparative studies.
The Coda Network steering group would discuss these following the workshop and develop terms of reference, a full membership and programme of activity.