From idea
to project

Start with a problem you have noticed in practice. Specify the setting, population and period you can realistically study. A broad concern can become a useful, answerable question when you narrow its scope.

In the urine MCS example, the team focused on one test in one ED over one year. Existing pathology data described the tests, while the record audit supplied the clinical context. Those sources answered different parts of the question.

Take a rough question to research support early. You can revise it together once you know what data and time are available.

You do not need a finished proposal to start the conversation. Bring the problem, a rough question and an honest estimate of the time you can contribute.

• Research support or a mentor: discuss feasibility, design and the work involved.
• Local ethics and governance office: clarify the review and authorisation pathway for your project.
• Data custodian: establish whether the information exists, whether you may access it and how to obtain a usable extract.

Data recorded in a system may still take time to retrieve, interpret and prepare. Check the current contacts and allow time for that work. Confirm the approval pathway locally before accessing records or collecting project data.

For each part of your question, ask what information you need to answer it. A culture result alone cannot show whether a person had symptoms when the test was ordered.

Write a simple data dictionary. For each field, record its meaning, source, allowed values, recording instructions and how to handle missing or unclear information. Give every auditor the same instructions.

As an extra practical step, try the form on a small set of records before the main collection. Compare how different auditors interpret the fields, resolve disagreements and update the instructions. Agree on any methodological changes with your research-support team.

The tools below have different jobs and some overlap. Choose what fits the project and what your organisation approves for that information. The software name alone does not establish that a particular account, device or storage location is appropriate.

AI can help you sketch an outline or suggest fields. Check every suggestion against the question, the source material and your team's definitions. Use the organisation's approved AI pathway and keep sensitive information within approved systems.

Use the project worksheet below to prepare five short statements: the problem, the question, the resources available, the help needed and the person you will speak to next.

A useful opening is: “I've noticed this problem in our ED. This is the question I'd like to answer. I can contribute this much time. Can we discuss whether the project is feasible and what the next step should be?”