COVID-19 : predicting future adult social care demand and capacity planning

Text removed is struck through; text added is underlined. Unchanged paragraphs are summarised rather than repeated.

Benefits reported

A comprehensive list of benefits cannot be achieved currently as Worcestershire County Council only obtained the data in Autumn 2020. • Development of a more collaborative culture between health and social care partners, based on the principle of joining up health and social care responses. The initial model has been introduced to our local ICS and partners as a potential example of what analysis can be provided with combined health and social care data, this advancing conversations about current and future analytical projects. However, the model that is currently still being worked on has shown a relationship between positive Covid tests and Council services. The aim now is to explore this further in terms of service demand. • Increased knowledge of data sharing, Information Governance, DARS applications, pseudonymisation etc allowing for faster design of future analytical solutions. • Significant learning and experience gained through this project, used to inform future Business Intelligence Strategy. • Influencing senior managers to adopt a more data driven culture, moving away from task and finish approaches and more coproduced solutions. • We have used the tool locally to broker different conversations with senior managers, including the Director for Adult Social Care, about creating a more Research & Development approach to health and social care analysis that is based on opportunities from joined health and social care data rather than just trying to answer existing questions.

Unchanged: Objective for processing, Processing activities, Expected output, Expected measurable benefits.

Text removed is struck through; text added is underlined. Unchanged paragraphs are summarised rather than repeated.

Objective for processing

Worcestershire County Council have been working with their partner AT Provider and PI Ltd (trading as PredictX) over the past year to look at how analysing trends can help predict future adult social care and A&E admissions through the way someone uses their assistive technology equipment. In light of COVID-19 and the changes to the way people are living and the support they are receiving, the council have needed to review this project. History The council are looking to develop a Predictive Model that will answer the following COVID-19 response related questions to enable Worcestershire County Council to work better with its' population and target communications and support. During 2019/20, funded through NHS Digital Social Care Pathfinders, Worcestershire County Council was working with their partner AT Provider and PI Ltd (trading as PredictX) to look at how analysing trends in assistive technology equipment usage and monitoring data can help predict future adult social care and A&E admissions. The COVID-19 pandemic lockdown in March 2020 meant vulnerable people were not behaving the same as before, with many self-isolating. The way the Adult Social Care team and Care Providers were supporting these individuals also changed. This meant the machine learning which had developed would not be as relevant and the model could not be fine tuned with additional data at this time. Question 1: Is there an identifiable population in Worcestershire that are more likely to be impacted by covid pandemic through either social care need or acute medical assistance? As it became clear the pandemic was a long-term issue the decision was made that this scope was no longer viable. However, there was potential that the learnings and linking of adult social care and health data could be used to help Worcestershire’s response to COVID-19. If so, what are the characteristics of this group? Worcestershire acknowledged there was already a lot of information on risks/characteristics around Covid. However, these were changing as more data became available e.g. shielded cohort conditions and symptoms of the disease have changed as more data becomes available. Characteristic examples: If the council could understand, and reduce the spread of, the infection in the most demand causing cohorts, it would reduce pressures on the Acute Hospitals and allow the Council to plan for social care support. * geographic location (short postcode, LSOA) It was acknowledged that the potential for this project to provide actionable insight to support the response efforts of the Council and partner agencies was probably time limited, assuming the pandemic will ultimately be controlled through herd immunity and/or full roll out of vaccinations. But the benefits of the insight meant there was appetite to progress. * age Current scope * gender NHS Digital Social Care Pathfinders agreed to the change of scope, which allowed Worcestershire County Council, Midlands and Lancashire Commissioning Support Unit and PI Ltd to progress a potential Covid-19 response model. * multiple deprivation The organisations worked together to identify datasets which could be used, at Lower Layer Super Output Area (LSOA), with the potential to provide actionable insight to support the response efforts of the Council and partner agencies. Determine which areas are more likely to see significantly increased demand for services such as health and social care services and support services through the pandemic. * ethnicity 1) Provide risk stratification to support clinical and extremely vulnerable residents of Worcestershire. * long term conditions 2) Potentially determine the demand profile of different segments of Worcestershire’s population who contract COVID-19 (for example – for every 100 positive tests in a deprived area what is the corresponding impact on support services compared to a similar level in an affluent area). This was aimed at correlation rather than causation. Worcestershire acknowledges there is already a lot of information on risks/characteristics around Covid. However, these are changing as more data becomes available e.g. shielded cohort conditions have changed, symptoms of the disease have changed as more data becomes available. The closure of the Social Care Digital Pathfinders programme at end of March 2021 meant the partners had to work fast in accessing datasets, model development and potential benefits. One of the conditions of the Social Care Digital Pathfinders programme grant is that Worcestershire continue with the project to delivery, reporting deliverable benefits in September/October 2021. The continuation of this Data Sharing Agreement is required in order for Worcestershire County Council to continue using the model and demonstrate its usefulness to other Councils who may then look to set up a similar situation with their own Agreement. If the model is successful Worcestershire County Council will then need to keep using the data as part of Worcestershire County Council’s Covid response. Worcestershire wants to take this knowledge and link to Adult Social Care data so they can establish for Worcestershire the additional/different risks that being in receipt of Adult Social Care support makes. Locally trying to identify how these manifest in the Worcestershire population particularly in relation to adult social care. The 1st iteration of this agreement was agreed in August 2020 with an expiry of 31st March 2021 of which aligned with COPI legislation at the time. For clarity the data disseminated by NHS Digital under this agreement does not rely on COPI (but the Health and Social Care Act 2012) but however is linked to COVID-19 related data the Council holds that can only be held and processed via COPI. The prediction is about using current risk factors and identify those more likely to contract COVID-19 or need COVID-19 related care within ‘x’ amount of time (To be confirmed by the model) and understanding these risks in terms of prioritising resource and support. By combining what Worcestershire already know about characteristics and combining with Adult Social Care data to prioritise intervention activity that reduces the likelihood of spread of infection; providing the greatest opportunity to do something about it as a system. Due to the extension of COPI, Worcestershire County Council seeks to extend, renew and amend this Agreement to continue obtaining more of the same datasets as previously approved and to include linkage to the Council’s COVID-19 testing data they have become data controller for, through an agreement with Public Health England. Question 2: For patients who leave hospital and enter social care, how many will require no further care, how many will need short term care, and how many will need long term care packages? The aim being: Characteristic examples: • To establish if the distribution of potential demand differs from the distribution of potential spread of infection. * geographic location (short postcode, LSOA) • The potential to influence and actively manage cohorts of the population based upon the risk of high service demand, as well as risk to public health. * age • Better informed and targeted community interventions at sub County level, to limit the impact of outbreaks on services. * gender • Ability to design the Council’s response activity more efficiently and effectively. * multiple deprivation • Insight to manage council resources more efficiently to deliver optimal support to those in receipt of services as a result of COVID-19 * ethnicity Whilst not incorporating vaccination data explicitly into the model, the product will provide insight into the effects of vaccination on local COVID-19 profile and resulting service demand, allowing system leaders to understand how best to manage resources as the pandemic starts to come under control. Worcestershire County Council might expect service demand to be increasingly driven by the non-vaccinated cohort of the population for example. * long term conditions The results so far have shown how COVID-19 is affecting LSOA areas differently, what adult social care services they require, and how these have changed in light of COVID-19. The aim of this stage is to explore the benefits of linking the datasets and what this would show the Council to aid its’ response to Covid. There is potential for the model to show data links which will be beneficial to the Council beyond the COPI timeframe, in which case a new DARS application would be put forward using the data from this exploration stage as evidence of why the datasets need to continue being provided. Worcestershire County Council are Data Controller for the Adult Social Care dataset and the COVID-19 Situational Awareness Explorer – Positive and Negative Tests & Cases (Public Health England). These datasets are pseudonymised and data is then uploaded to Midlands and Lancashire Commissioning Support Unit who will process on behalf of Worcestershire County Council. Worcestershire County Council requires access to; Under this Agreement Worcestershire County Council currently holds and requires; [4 paragraphs unchanged] The aim is to identify groups of individuals who are more vulnerable and likely to need greater support during Covid. Acorn segmentation socio - economic data for Worcestershire. Worcestershire County Council require a number of different datasets including the shielded patient list and any Covid data held on the Adult Social Care system. Health data analysis as already proven certain ethnicities and shielded patients are at more risk, therefore Worcestershire County Council want to link this to adult social care data to understand this cohort within Worcestershire. IMD (index of Multiple Deprivation) Decile. Shielded Patient list (SPL) - Worcestershire County Council has access to the Shielded Patient List for their [33 words unchanged] List Local Authority Dataset’, which form part of the Council’s Covid-19 response. Health data analysis as already proven certain ethnicities and shielded patients are at more risk, therefore Worcestershire County Council want to link this to adult social care data to understand this cohort within Worcestershire. Incoming Local Authority Dataset (provided to the Council on an ongoing basis) is able to be shared with the data processor in this agreement, PI Ltd, to enable support for those identified as extremely vulnerable in the Worcestershire area as there is a valid data sharing agreement in place. Worcestershire County Council will only use the Shielded Patient List data under this agreement within the permitted parameters of the directions. It is already established that individuals on the SPL are more vulnerable to COVID-19 so Worcestershire County Council want to include the information on this with the other datasets in order to look for trends in peoples need, using as much health, social care and deprivation information as possible. Exploring the links between datasets under this agreement will aid Worcestershire’s response to Covid and the best way to engage and support its residents, targeting communications through key messages cohorts identified where appropriate. SPL is just one of the many datasets the council wishes to combine. The addition of The COVID-19 Situational Awareness Explorer – Positive and Negative Tests & Cases (Public Health England) will allow the council to further enhance the data by understanding the spread of COVID-19 in the local area. SPL data will be pseudonymised and combined with the adult social care data and all data provided by NHS Digital under this Data Sharing Agreement to create one big dataset which will be sent to PI Ltd. For clarity, Worcestershire County Council are not able to share the NHS List Local Authority Dataset and therefore is not used under this agreement. The aim is to use this data as part of the model to identify if this characteristic, linked with other information such as age, ethnicity, primary care user group data can identify specific groups who are at greater risk. The information could potentially then be used as part of the strategic planning for Adult Services to understand their ‘at risk’ communities better. It is already established that individuals on the SPL are more vulnerable to COVID-19 so Worcestershire County Council want to include the information on this with the other datasets in order to look for trends in peoples need, using as much health, social care and deprivation information as possible. Data is pseudonymised at source using the same SALT key and sent to Midlands and Lancashire Commissioning Support Unit, who combine all data provided to create one big dataset. This combined dataset is pseudonymised again, using a different SALT key before being sent to PI Ltd. PI Ltd will use the SPL flag (individuals who appear on the SPL dataset), along with the other patient data, to create a model on the vulnerable cohort in Worcestershire. This will then be developed into a map of Worcestershire hotspot areas where there are communities of vulnerable individuals who will need additional support during the pandemic and/or should they get COVID-19-19. The Council will then use this when engaging communications, informing social work teams on where to focus their support. The linked data will only be processed by Midlands and Lancashire Commissioning Support Unit & PI Ltd therefore there is no possibility of Worcestershire County Council being able to re-identify. Worcestershire County Council will only use the Shielded Patient List data under this agreement within the permitted parameters of the directions. This process is covered in a valid data sharing agreement between Worcestershire County Council, Midlands and Lancashire Commissioning Support Unit and PI Ltd. For clarity, Worcestershire County Council are not able to share the NHS List Local Authority Dataset and therefore is not used under this agreement. Exploring the links between datasets under this agreement is aiding Worcestershire’s response to COVID-19. It helps identify the best way to engage and support its residents, targeting communications through key messages to cohorts identified where appropriate. The aim is to then use this data as part of the model to identify if this characteristic, linked with other information such as age, ethnicity, primary care user group data can identify specific groups who are at greater risk. PI Ltd are using the data to create a model on the vulnerable cohort in Worcestershire to provide a map of hotspot areas where there is likely to be increased demand for health and social care services during the pandemic. This change in demand could be either due to individuals contracting the virus or their care needs changing due to the impact of the virus on their usual care/support mechanism. It allows the Council to establish if the distribution of potential demand differs to the distribution of potential spread of infection. This, in turn, will be used for risk stratification to support Worcestershire residents and the ability to proactively manage cohorts of the population based upon potential demand for services. The information could potentially then be used as part of the strategic planning for Adult Services to understand their ‘at risk’ communities better. The Council could also use the information to target community interventions at specific geographies, put in preventative measures to limit the impact of local outbreaks and act in a fast, informed fashion when outbreaks occur to successfully manage the situation. The Council would also use the information to target community interventions at specific geographies, put in preventative measures to limit the impact of local outbreaks and act in a fast, informed fashion when outbreaks occur to successfully manage the situation. Worcestershire County Council already collects the Social Care data for their region. However, the linked data will only be processed by Midlands and Lancashire Commissioning Support Unit & PI Ltd therefore there is no possibility of Worcestershire County Council being able to re-identify. [2 paragraphs unchanged]

Processing activities

[23 paragraphs unchanged] 1. Identifiable SUS+, Mental Health data (MHSDS), Maternity data (MSDS), and Community Services Data Set (CSDS) and Shielded Patient List (separate data flow from NHS Digital authorised by way of an IG letter) only is pseudonymised (at source) under a SALT key the DSCRO (Data Services for Commissioners Regional Office) DSCRO creates. The pseudonymised datasets are then securely transferred from the DSCRO to Midlands and Lancashire Commissioning Support Unit. 2. The DSCRO shares the SALT key ID with Worcestershire County Council who will apply the same SALT key to the Social Care Dataset. Dataset and COVID-19 Situational Awareness Explorer – Positive and Negative Tests & Cases. The pseudonymised (at source) data is then securely transferred to Midlands and Lancashire Commissioning Support Unit. [2 paragraphs unchanged] 5. PI will aggregate the data with small numbers suppressed and then pass to Worcestershire County council. The council will only have access to aggregated small number suppressed data to prevent the ability of the council to re-identify the data through holding the identifiable social care data record. [2 paragraphs unchanged]

Expected output

The outputs will be produced around 2-3 months after the initial data has been shared. Each element of the project requests set amount of time from the one before. The model is currently being developed, the actual consumables of which are currently being defined based on requirements below.. This extended(and amended) Agreement will allow Worcestershire County Council to utilise the model, review the outputs and define report contents. The outputs expected are: The critical requirements of the products can be summarised as follows: Access to a report suite that will enable both analysis and data interrogation/export as required - this will be allowed via single user sign on for security purposes - when staff leave the Council there is no risk of them still having access to the data. • A clear visualisation of service demand pressure resulting from local COVID-19 conditions; compared with service demand associated with non-COVID patients. This can be used to demonstrate the effects of the pandemic on local services in something approaching real time, or plan for theoretical scenarios such as new variants or effects of vaccinations. • Ability to interrogate the output analysis by key socio-demographic and geographical groups to identify patterns in services resulting from COVID-19 pandemic as this evolves (for example, estimating impact of new variants, or vaccinations). • Ability to observe the changes in service demand over time (either historically through the three waves to date, or to project into the future to plan for scenarios based on possible COVID profiles) • Clear set of input data so the model can be tested and applied to other Local Authorities with ease. • Outputs from the model are aggregated and do not identify individual residents or cases The requirements above are all in development with PI Ltd and should be finalised by 31 March 2021 to form a functional model for WCC, with fully developed products available for other local authorities completed within the first quarter of 2021/22 financial year. Due to the compressed timescales for the completion of the rescoped project, the business lead and sponsor, alongside other council staff from Adult Social Care, Public Health, Corporate Transformation and Business Intelligence have ongoing involvement in the development of the products to ensure they meet required specifications. This is facilitated around a weekly development meeting with Worcestershire County Council, PI Ltd and the Commissioning Support Unit. [2 paragraphs unchanged] The reporting suite and data will be accessed by Management Information Analysts within the Council and used to enable both analysis and data interrogation/export as required to produce high level reports to senior leaders and cabinet members to inform decision making. Access to the report will be allowed via single user sign on for security purposes - when staff leave the Council there is no risk of them still having access to the data. Map of Worcestershire - highlighting hotspots of communities with vulnerable individuals Products delivered / to be delivered : Dashboard with graphs/tables - This would show the data at the lowest geographical area possible that does not pose a risk of accidentally identifying individuals. 1 Model output 1 – map describing the relative risk to communities from pandemic (based upon vulnerability and exposure to infection) Only aggregated data with small numbers suppressed will be available to staff within the council. 2 Model output 2 – Profile of service demand impact associated with local COVID-19 profile, including net changes relative to “non-COVID” patients to estimate service pressure, assessment of costs. Functionality to interrogate output to identify differences in demand arising from different cohorts of the local population. The specifics of the outputs cannot be defined until data has been run through the model and potential links identified. 3 Promotional material to describe features and benefits to engage with other local authorities

Expected measurable benefits

Quality Benefits from the COVID-19 model are anticipated to be seen from April 2021 when the model is live. There will be considerable drop off in benefits as and when vaccination programmes, longer term strategies and herd immunity start to take effect (the project is necessarily focussed around COVID-19 response which should hopefully be relatively short term). 1) By being able to profile the County and make predictions on the impact of Covid-19 on communities will mean Worcestershire County Council can support groups of individuals to stay safe by providing advice and guidance. • Better informed decisions based on intelligence of the COVID-19 profile across the County and the resulting demand on services such as domiciliary care, residential care and specific COVID-19 services. • Target communications and engagement • Development of a more collaborative culture between health and social care partners, based on the principle of joining up health and social care responses. • Tailor social work support • Intelligence (case) for changing working practices, boosting effectiveness and efficiency of services, by actively managing cohorts of the population based upon the risk of high service demand. • Consider primary support networks and provide additional data/support to them. • Better informed and targeted community interventions at sub County level, to limit the impact of outbreaks on services. 2) A collaborative, strength-based approach means Worcestershire County Council are joining up health and social care responses. • Better and faster decisions, and improved decision cycle speed, in response to changes in nature of local infection (including impact of local outbreaks and vaccination programmes). 3) Actively managing cohorts of the population based on high risk profiles means the onset of long term intensive social care is delayed indefinitely or at least as long as possible meaning the pressure on social care services is better managed. • Optimising use of resources following insight to assist with dynamic and proactive workforce planning 4) Being able to target community interventions at specific geographies, means the Council will be able to put in preventative measures to limit the impact of local outbreaks. • Faster and more appropriate treatment 5) If/when outbreaks occur the Council will be able to act in a fast, informed fashion when to successfully manage the situation. • Intelligence to inform longer term business continuity planning around future pandemic or infection Cash benefits Once implemented the BI team will ensure 1) If Worcestershire County Council can keep people safe the Council can reduce risk of a second wave, thus reducing risk of a local lockdown. Meaning the economy can continue to rebuild. • Delivery against the Commercial Plan to roll out products to other Local Authorities on a paid for / subscription basis; 2) If the council can limit the impact of local outbreaks it will reduce pressures on the Acute Hospitals and allow the Council to plan for social care support. • Promotion of the activity where appropriate at national and regional networks and conferences. 3) Here2Help service will continue to be available to all but can be more proactive in actually contacting the more at risk communities. Work to be undertaken beyond the project closure includes: Societal • Development of model outputs (beyond the functional model to be completed by March 2021) for subscribing authorities (initially targeting Herefordshire County Council to test) 1) By keeping the Worcestershire population away from unnecessary A&E visits there should be a reduction in risk of cross-infection from environment. • Marketing of model outputs to other local authorities. 2) The avoidance of unnecessary hospital admissions will subsequently lead to less cases of hospital degeneration and subsequent costs. The Benefits lead will assess the extent to which quality benefits have been realised through an evaluation of the benefits model in Worcestershire initially, to expand to include subscribers as appropriate. The Head of BI and Public Health Consultant for Evidence at Worcestershire County Council will coordinate this activity with input from service managers. 3) By being able to predict the impact of Covid-19 on certain groups of individuals the Council can ensure they are better informed on how to keep themselves safe, potentially reducing anxiety for those vulnerable individuals. Once this model is live and its potential realised Public Health will look to see whether it could be adapted to help support seasonal flu, as well as form a basis for tools for future pandemics. 4) There could be different levels of communication – a general, blanket message for whole county and then more tailored messages based on customer segmentation, level of risk and characteristics of communities. These messages can also consider the best medium for the client groups. For the avoidance of doubt, there will be no re-identification of individuals, although characteristics that define particular cohorts will be identified.

Benefits reported

Yielded Benefits is not a requirement for new applications. A comprehensive list of benefits cannot be achieved currently as Worcestershire County Council only obtained the data in Autumn 2020. However, the model that is currently still being worked on has shown a relationship between positive Covid tests and Council services. The aim now is to explore this further in terms of service demand.