The Living With platform generates many kinds of data, and the word means different things to different people. Here’s a summary of how we generally categorise requests according to the type of work involved.
Data requests tend to fall into one or more of these 4 areas.
- Reports
- Patient data extracts
- Analysis
- Research
1. Reports
These are essentially aggregated stats. For patient-identifying data, see (2).
Purpose
Typical purposes include: monitoring take-up of the digital health intervention; monitoring service throughput; audits; providing KPIs for inclusion in management reports and for presentation at steering groups.
Reports often trigger insights for further analysis or strategic review of services.
Content
Reports can be simple or complex. Simple reports include data on:
- Patient volumes - numbers of patients invited / registered / discharged; by month, by clinic (e.g. across an ICS, region or clinic group)
- PROMs - numbers submitted; by type
- Average outcome measure scores
- Patient engagement - app interaction; messages sent to the clinic; period of use
- Clinic engagement - messages sent to patients; patient conversion rate
More detailed reports depend on which products and features are enabled.
- Demographic breakdown - age; gender; ethnicity; education level; deprivation index
- Top symptoms patients choose to track
- Diaries submitted
- Treatment programmes started / completed
- Service use - GP / hospital / outpatient / physio appointments
Format
Reports are often provided as slides in PDF format for onward presentation.
We can include the graphics and data tables separately for your own presentations and summary reports (i.e. data as CSV format to open in a spreadsheet).
Delivery
Files are normally emailed to you. We can set up a shared folder online and provide access to your team members for ongoing download of reports and associated files.
Frequency
Reports are generated quarterly as standard.
They can be monthly, e.g. for KPIs for onward reporting.
Effort
As well as the initial setup, there are manual steps involved in the production of each report at present, because of the evolving nature of both the data and the service needs.
2. Patient data extracts
These are datasets in which you can drill down to individual patients and correlate their key health and engagement indicators, including outcomes.
The data is usually anonymous (‘pseudonymised’) - an individual’s data can be tracked across metrics, but the individual is not identifiable. This is fine when the analysis is within a patient’s usage of the platform - for example, cohort analysis looking at correlations of different outcome measures, or breaking data outcomes by cohort; or correlating interventions with outcomes.
Or the data can be patient-identifying if the aim is to map to other patient data held in external systems.
NB Records for individual patients can be exported directly via the platform.
The service described here is for the provision of records for multiple patients at a time.
Purposes
Typically 3 purposes:
- in-house research / studies / advanced audit
- Integration with other reporting systems, e.g. for outcome measures
- integration with electronic patient records
Delivery
Sometimes the datasets are provided in 2 steps: first the pseudonymised (non-identifying) data; then later a set of keys mapping the individuals via their NHS numbers.
When the correlation is required with data from other PMS or for uploading LW records in bulk to EPR.
These are usually pseudo in the first instance - patient identifying info is …
For integration, we just need to know what to send where and how often.
Key point: because of the PID, there is more work involved in (setting up) and in each extract. Even with pseudo, (move) we need to agree on the secure channel. Prefer to provide time-limited access to online downloads, restricted to a named individual. Can email if pseudo but only to NHS.net address, zipped file with password protection, password sent by separate channel.
Where the purpose is audit or research and there is no in-house capability for data handling, (3) and (4) become relevant.
3. Analysis
This can be a deep dive into the data on your behalf.
These lead to providing one-off reports with the potential for ongoing updates as part of (1).
Projects can be small or large.
As an example of a large project, with NHSE we are currently looking at the outcomes of long covid patients across a set of measures, examining the various cohorts at baseline and then at several time points during treatment.
The analysis is additionally exploring how outcomes correlate with engagement, on both the patient and the clinic side.
Surveys
In this project, data from the app has been enhanced via a patient survey that captured date of infection (patients often wait weeks for GP referral to long covid services); dates of vaccines; prior symptoms; prior work status; as well as views on treatment including the app.
Not all services need all available data; some have ready access to this info from their own systems, and integrate data from LW per (2) above.
Others find it easier to ask patients via LW due to IT hurdles. Eg Demographic
Exciting things are possible. Could correlate patients who read specific articles attached to messages with various behaviours and outcomes.
In principle also clinic activity- numbers of logins, time spent etc. This is something we plan to look into soon.
4. Research
Access to data can be provided to support formal research. Several Universities have run grant-funded projects in this way.
However, projects could also be in collaboration with the research department of a Trust or ICS.
Small and large - basic pseudo data extracts per (2) above. Study as part of indic PhD.
Delivery
Or we set up a repository and provide direct secure access to the underlying (pseudo) data tables, for online manipulation via SQL and a basic report generation interface, or download of datasets for incorporation into packages such as SPSS, programming in R or Python, or plain old Excel.
We can provide supplementary data analysis to enhance the above. For example, the aggregations of Index of Multiple Deprivation (IMD) are associated with individual patients’ full postcode (the latter is classed as patient-identifying and therefore not provided to researchers).
Scope
As standard, this comprises non-identifying patient data, governed by research ethics. University researchers don’t normally get access to individual patient records, and the timeframes and purposes of the data are controlled.
(Subject to the remit of the research ethics).
Can be enhanced with patient opt-in, e.g. for permission to be contacted by a researcher.
Projects can involve collaboration on feature development, for example adding questionnaires; sending surveys; integrating ‘nudges’ via notifications