Data and improvement · Spreadsheet / supervised analysis
Use practice data to answer one useful question
Start with a question the team can act on, then work out what information would answer it safely.
What you’ll have at the end
A short summary of the question, the data used, what it suggests and what to check next.
How to begin
- Write a question in plain language. For example: are requests for an interpreter reaching the right team before an appointment? Avoid starting with ‘what data do we have?’
- Agree the time period, population and definitions with a service lead. Ask an authorised analyst or information-governance colleague what data can be used and how it should be handled.
- Use approved aggregate or de-identified data. Check missing values, duplicate records and changes in how information was recorded before calculating anything.
- Show counts as well as percentages, and describe who or what is missing from the data. Ask the people who run the service whether the pattern matches their experience.
- Share a brief finding with its limits and a practical next question. If a change is made, agree how you will check its effect.
What to measure
- Outcome: can the team answer the original question well enough to act?
- Process: data completeness and agreement on definitions.
- Balancing: time to prepare the analysis and any risk that small groups could identify someone.
Before using real data
Only use data you are authorised to access, in approved systems. Small or combined groups may still identify people. Do not upload patient information to this website, a personal device or a public AI service.
Agree scope, data handling, supervision and escalation with your practice. Use approved systems. Do not upload patient information to this website or a public AI tool.