Worcestershire County Council saw a need to develop improved foresight relating to demands on services. One approach the council took was to predict longer-term care needs of individuals in the area. Risk stratification was employed to identify households in Worcestershire that were likely to require support from adult social care in the future. The aim was to to identify potential target groups or locations in the County for preventative activity, communication, information and advice. This project was building on the NHS Digital Project to predict social care self-funder pick-ups.
To Identify data with indicators that might predict need for social care the following data was used: Household Acorn data, energy efficiency (EPC, thermal imaging), hospital admissions and social care. The council was unable to include some datasets due to the need for data sharing agreements e.g., Council Tax single occupancy data. Once the data was gathered it was matched using Unique Property Reference Number (UPRN) where possible. UPRNs played a key role in matching the data sets and were used as the main source of data matching. Some datasets matched by address were difficult to match due to differing address formats e.g., fl ats. Therefore, some matching had to be done manually as oppose to automatically. This allowed Worcestershire County Council to test by how well indicators identify households currently receiving older adult social care services at home or in the community.
Through matching the data using UPRNs it was possible to identify that thermal imaging showed that 0.8% of households receive social care. Furthermore, the project revealed that households in certain Acorn types were 6 times more likely to have someone aged 65+ with social care. It was also found that households in certain Acorn types and in areas with high hospital admission rates were 14 times more likely to have someone aged 65+ with social care.
Strengths and Limitations
Worcestershire County Council found that it was possible to combine data to identify households likely to be at higher risk of needing social care. It was found that 11.2% of households with a combination of indicators had someone with social care aged 65+ compared with 0.8% for all households.
However, it became apparent that some indicators were not good predictors of risk. For example, households with someone aged 65+ with social care were less likely to have a high EPC / poor thermal rating than all households. It was also identifi ed that adding indicators could increase the proportion likely to be at risk, but also lowers the size of the target group.
This approach identifi ed around 1,000 households not currently receiving social care that could be targeted for preventative activity. The project identifi ed those at risk, but the council has plans to work on intervention to reduce risk. The information gathered from the project was provided to the Social Care team at the council who plan to use the data gathered in their responses to predicting long-term care needs. The UPRN played a vital role in the prediction of these needs. The UPRN was the key data set which allowed further data sets to be linked together.