Guide · checked October 2026
How to automate clinical audit data collection
Automate clinical audit data collection by reading the records directly from source systems through their interfaces or exports, testing that the extraction pulls the right source and sample, and keeping a person in charge of every judgement the rules can't settle.
Why manual collection struggles
Many local audits are still done by reading one system against another and typing into a spreadsheet. That makes them slow, so they are often sampled rather than complete, and hard to repeat the same way next time. HQIP's guidance on re-audit depends on doing the second round in exactly the same way, which manual methods make difficult.
Ways to get the data
- Standards-based interfaces such as FHIR, DICOMweb and HL7 v2, where your integration team can expose them.
- Scheduled exports or reports in CSV or XLSX from the source system.
- Reading the screen of systems with no interface, with strict controls on what is captured.
Whatever the route, HQIP says to test electronic record extraction to confirm the right source and the right sample are being pulled. Check the denominator against a known count before you trust the results.
What should stay human
Automation is good at counting, matching and applying explicit rules. It is not a substitute for clinical judgement on ambiguous cases. Keep the unclear cases visible, have a named person decide them with a note, and record who decided what. HQIP's data quality guide also asks for written rules on what to do when sources conflict.
Information governance still applies. Read only what the audit needs, and anonymise at the earliest stage.
What automation does not fix
- A poorly written standard: if a person can't apply it consistently, a rule can't either.
- Missing documentation: an empty field is a finding, not a technical error.
- Governance: automated access still needs approval and a record of who read what.
Where Oriel fits
Oriel connects to FHIR R4, DICOMweb PACS (including dose from RDSR) and HL7 v2 over MLLP with TLS, imports mapped CSV or XLSX files, and reads a stream from Thalix. Secrets are encrypted on the server and messages that can't be read wait as dead letters to fix and replay. Where a system has no interface, a workstation runner reads the screen with patient names masked before any screenshot is kept, and stops when staff touch the mouse.
A person approves the plan before anything runs and decides each finding. Oriel is a pilot on synthetic data only; no hospital system has been connected with real patient data.
Questions
Can audit data be collected without an interface?
Exports and screen-reading can work. Whatever the route, test that the extraction pulls the right source and sample, as HQIP advises.
Does automation mean auditing every case?
It makes complete audits practical, but the scope is still your decision and should be written into the method.
Sources
Official pages, checked October 2026. Guidance changes, so check the current version before you rely on it.
General information for healthcare managers and clinical teams, checked October 2026. It is not legal, regulatory or clinical advice; follow the current official guidance and your organisation's own policies. No patient data is used in these examples. Oriel is a working prototype shown on a fictional demo hospital with synthetic data only. It is not a medical device and is not for clinical use. It drafts and flags for people to decide; it does not diagnose or change clinical records. A clinical safety case and information-governance approval come before any use with real patient data. Oriel is a working name.