Care Home Data Review: workshop summary

Issues with quality / completeness of existing data

Timeliness of publication

Providers raised concerns both with annual publications being published a while after the data they refer to was collected, and shorter term ‘near real-time’ datasets (like the Safety Huddle Tool) not refreshing quickly enough.

Short term / live data collections

There was concern with systems such as Turas not being nimble enough to keep up with the constantly changing nature of care homes, so data was out of date as soon as it was released.  HSCPs commented that they had to phone care homes directly to get the latest vacancy data, increasing the burden both on them, and on care homes.

Quarterly / annual publications

The delay between quarterly / annual returns being submitted and the data and accompanying reports & executive summaries being published limits their usefulness to management. This delay is often of the order of months – a year. For example, it was noted that workforce data from Care Inspectorate (CI) / Scottish Social Services Council (SSSC) is a year out of date at publication.

Other concerns about timeliness

Social work assessments are often out of date when received. One of the underlying issues contributing to this is that data providers need to wait for systems to update before preparing returns data.

Concerns about data coverage, accuracy & completeness

Geographic coverage

Data is not always available at a low enough level (e.g. individual level) to answer questions, but is instead aggregated by board. Resident level data is either not available or not accurate.

Inconsistent coverage

Some datasets (e.g. PHS Source social care) only cover Local Authority funded individuals, rather than all residents in a care home.

Poor response rates…

…on particular variables and datasets. Missing data is often estimated, lowering data quality. One example is the Care Home Census, where only 67% of care homes open anytime during 2021/22 submitted data (so Public Health Scotland had to estimate one-third of the data for some of the data items).

Pre-population of data fields potentially lowering data quality

In the care home census some data fields are pre-populated with the previous year’s residents’ data to save providers time. However, some of these residents will have died or moved on and if the discharge date or date of death is not provided then this will reduce data quality.

Inconsistency across datasets diminishing confidence in data

Participants commented how duplication raises concerns about quality and accuracy, as there are apparent conflicts of data when data about the same thing show different values in different datasets.

Specific examples of poor data quality

Specific examples were given in the workshop of poor data quality, with lots of overlap with gaps in data, below. We have noted all these concerns together in gaps in data.

Suggested improvements to quality / completeness of data

 Timeliness

Geographic coverage level

Some attendees noted the need for individual level data, to allow more cuts of data to be possible.

Low completeness rates

• incentivise completion by making useful data available to data providers
• improve guidance to make it easier to complete returns
• make data collection mandatory. “Data much less useful if not 100% complete”
• feedback on how data is used and the difference it is making (see Communication)
• leave data collections open throughout the whole year, making it easier to update details as and when they change (e.g. recording discharges in the care home census)

 Creation of a Core Minimum Dataset (CMDS) / reduction of duplication

Many comments mentioned the advantages of a core minimum dataset (CMDS), increased standardisation and reduced duplication to help move away from current data landscape and enable collection of data collectively agreed as important. This would free up resource to focus on additional requirements.