Strong analytical collaboration resulted in valuable, high-quality, coherent statistics during the pandemic.
Taking this approach to other topics will help overcome existing and future problems.
A shared vision and determination to support an issue of national importance resulted in impressive analytical collaboration during the pandemic. This included collaboration between producers – and collaboration with their data suppliers, policy and communication colleagues, other parts of governments, and with academic experts. For example, NHS boards and trusts played a pivotal role in providing data to central governments and public health bodies to inform the pandemic. It was also important to have strong working relationships with policy colleagues to understand rapid changes in policy and the impact of these on the data and statistics.
We saw that the amount of collaboration between producers increased and that the way they collaborated improved. Producers were more likely to reach out to each other when working on a new piece of analysis to ask for contributions or quality assurance. More-regular meetings on specific topics also helped to facilitate collaboration between producers. This approach should be applied to other health and social care issues and would have benefits more broadly across the statistical system. It is increasingly important that producers collaborate across organisations, particularly on issues which require data sharing and linking.
A challenge to collaboration in England is the number of different producers responsible for health and social care data – a result of the fragmented nature of the health and social care system overall. During the pandemic this sometimes resulted in duplication of work among producers or delays to figures being released – for example in the case of population estimates to understand vaccination rates. The number of producers publishing statistics about health and social care in England can also make it confusing for users to know where to find information. Since the role was introduced in April 2020, we have seen the value of having a Head of Profession for Statistics at the Department of Health and Social Care. This role has played an important part in coordinating outputs across organisations and providing support and guidance within the department and to arms-length-bodies. Given the issues that can arise from the fragmented nature of the system and the creation of a new body, the UK Health Security Agency, we consider that there should be clearer analytical leadership and coordination of health and social care statistics in England. Producers in England should come together to decide how this will work and Heads of Profession for Statistics have already begun work to address this.
We encourage producers in the four nations to continue to engage on projects to provide comparable data across the UK, as they have for vaccination statistics and the COVID-19 Infection Survey. However, we also recognise that differences in policies in each country may mean that providing UK-wide comparability is not always possible – in these cases collaboration between countries remains vital to understand differences and their impact on the statistics. Producers should clearly explain to users any impact on the statistics and how they can be used, resulting from differences in policies across the UK.
We have previously highlighted the successes that result from formal collaboration with academics – for example, on the COVID-19 Infection Survey, which contributed valuable expertise and helped develop new skills in the statistical teams in ONS. In Scotland, the COVID-19 data and intelligence network was established to foster collaboration between the Scottish Government, public bodies, and academics. The network allowed people to work together to solve challenges, such as data collection and data sharing. Similarly, the One Wales approach brought together partners from the Welsh Government, the NHS, academia and public health to share datasets and expertise. Collaboration between academics and several government departments and organisations also lead to the development of the QCovid risk prediction model. The model was used to identify groups at high risk of being hospitalised or dying from coronavirus. This cross-organisational effort was awarded the 2021 Florence Nightingale healthcare data award by the Royal Statistical Society.
Producers told us that a relatively new challenge for them was cross-government working outside the analytical community. The pandemic impacted every part of governments, not just the health and social care system. Therefore, producers also had to form new relationships with multiple government departments, to share data to inform policy decisions, manage operations and inform the public about the pandemic. Producers reflected that, looking beyond the pandemic, continuing to build broader relationships across governments will help improve collaboration in future.
- There should be stronger analytical leadership and coordination of health and social care statistics.The need for this is most acute in England where the organisational landscape is most complex. Senior leaders should work together to decide how this will be implemented.
- To share best practice and tackle issues quickly and effectively, producers should maintain relationships built during the pandemic. Collaboration will be most beneficial when tackling shared goals, such as improving statistics on social care, mental health or ethnicity, and horizon scanning to anticipate future issues. Producers should proactively reach out to each other when working on a new topic or one which may be relevant to other producers or nations.
- Producers should continue to build relationships with other officials, including policy and communication colleagues, and other government analytical and scientific communities. This will mean that relationships exist to be in the best position for solving future problems.
Sharing and linking data can have life-saving impacts.
This must be prioritised by governments beyond the pandemic.
Data sharing and linking increased significantly during the pandemic. This allowed governments to develop policies and carry out rapid operational responses. The sharing of data between Public Health England, the Care Quality Commission and local authorities was vital for managing local outbreaks in care homes in England. Similarly, in Scotland the Care Inspectorate shared information about deaths, outbreaks, and staffing levels with the Scottish Government. This allowed vulnerable people in care homes to be identified. Linking data was also essential to develop clinically vulnerable patient lists in each country. The Shielded Patient List in England was used by health and care providers to support those who needed to stay at home, by other government departments to inform policies such as food parcels and statutory sick pay, and by mental health providers to support patients during and after shielding periods. In Wales, data linking was used to analyse the number of teachers and teaching assistants who were shielding, allowing the Welsh Government to understand the impact of the shielding policy on the provision of schooling during the pandemic.
In addition to being critical for the government response, data sharing and linking enabled analysis which enhanced the public’s understanding of the pandemic and impacted personal decision-making. For example, analysis by ONS linking vaccination, census, general practice (GP) and hospital data provided insights on COVID-19 vaccine take-up. A better understanding of vaccination rates by socio-demographic characteristics contributed to a push in targeted campaigns, for example to encourage vaccine take-up among ethnic minorities. We also heard that the pandemic has resulted in a better understanding across the NHS of the value of anonymised individual-level data and of providing high-quality data to be analysed centrally. Individual-level datasets provide a richer source of data than those at an aggregate level and are more flexible to adapt to new data needs. They also enable further value to be added by combining them with other data – for example, individual-level data for vaccinations and hospitalisations were linked to carry out analyses on the vaccination status of those in hospital.
While examples like these clearly demonstrate the real-life impact of sharing and linking data, public consent cannot be taken for granted. This was illustrated by the recent backlash against the sharing of GP data in England, which received largely negative media coverage. While the direct cause is not clear, during this time (between 1 June and 1 July 2021) opt-outs increased by about 1.2 million, from approximately 1.8 million to approximately 3 million. It is important that producers continue to demonstrate trustworthiness, by being transparent about plans for data sharing and linking. Producers should engage in an open and meaningful discussion with the public about the risks and benefits of it. This will support public confidence in the sharing and linking of personal information in future.
Data sharing and linking has historically been hard to achieve. Producers reflected that barriers include technical infrastructure, a tendency to be highly risk-averse and a cultural mindset where the default is not to share. A particular challenge raised by some devolved administrations and arms-length-bodies is ensuring that data sharing is two-way between themselves and central government. Some of these barriers remain, such as challenges with infrastructure. However, government data sharing notices and a clear need to share data to support the pandemic response helped to overcome many of them.
We welcome the commitments to improving data sharing and linking in the draft data strategy for NHS England and the UK Government’s Declaration on Government Reform. For this to be successful, it is important that governments sufficiently prioritise and resource work on data sharing and linking. We will continue to push for data sharing and linking, building on our previous work on joining up data.
- Governments must prioritise data sharing and linking. This should include sufficient investment, and support through legislation where necessary.
- Organisations must embed a culture of data sharing. This will be challenging,but will be supported by developing understanding of current barriers. Addressing our previous recommendations on collaboration and system-wide data infrastructures will also help to support more data sharing and linking in future.
- Producers must be transparent and engage with the public about plans for data sharing and linking. Publicly sharing positive stories about the real-life impacts of safe and secure data sharing and linking will help with this.