Pass 01
Audit and map the records
Count each record type, understand how they connect and find gaps before changing anything.
Case study · Healthcare data migration
The day I moved a whole clinic into a new system.
Years of patient records, appointments, notes, documents and referrals moved into Halaxy, with the relationships between them checked before and after the live move. Uncertain matches stopped for review, so the clinic received a usable clinical history instead of old problems in a new system.
Why importing the files wasn’t enough
The records existed.
Whether they belonged together was another question.
A bulk import would have carried the duplicates, missing information and broken links into Halaxy along with everything else.
The same patient appeared more than once across the source records.
Contact and personal details changed depending on which source we checked.
Some records and referrals were missing information they needed.
Appointments, clinicians and documents didn’t always link to the right record.
Some notes and attachments couldn’t move safely in their original format.
Check first. Move second.
Before anything moved, I counted each record type, traced how patients, clinicians, appointments and documents were connected, and separated reliable matches from the ones that needed a person.
How a patient was matched
A name alone wasn’t enough.
Existing IDs, dates of birth, contact details and record relationships were compared before a match was accepted.
Several match points10-patient pilot
The migration control path
Four passes. No guessing.
Pass 02
Clean and match safely
Use several patient details to resolve duplicates and separate clear matches from uncertain ones.
Pass 03
Test before the live move
Run a 10-patient pilot, inspect the links in Halaxy and agree which cases need a person.
Pass 04
Move, verify and repair
Prioritise upcoming appointments, compare the result with the source and retry only what failed.
Verified in Halaxy
Compared with the source before sign-off.
What happened when something didn’t add up?
The clear records kept moving.
The uncertain ones stopped.
Nothing was guessed or quietly skipped. Clear records moved automatically; anything uncertain, incomplete or unsuccessful went to a person with the source information still attached.
Moved automatically
- Several details identified the same patient
- Required information was present
- Patient, clinician and document links matched
- The record arrived complete in Halaxy
Stopped for review or repair
- More than one patient could match
- Important details were missing or conflicting
- A relationship between records was unclear
- A note, referral or attachment arrived incomplete
What reached the other side
01
The clinic could open Halaxy and get on with the job. Patient history, appointments and clinical records were connected and usable.
02
A green tick wasn’t treated as proof. Truncated notes and lost paragraph breaks were found and restored from the source.
03
Problem records didn’t disappear into a failed-import pile. 702 incomplete referrals were corrected, while 258 unsupported attachments were converted or sent for review.
04
The live move wasn’t a black box. Upcoming appointments moved first, and every warning, retry and correction could be followed back to the source.
Put the Data in Order
Sort out the duplicates, gaps and odd formatting before they get a new home.
Make Your Systems Work
Find the screen, setting or handoff making the job harder than it needs to be.
Build What’s Missing
If the standard options stop short, build the missing connection.
Make Sure It Keeps Working
Test the result, deal with the weird cases and leave people knowing what to do next.
Another project
Found the Right Clinician Faster
Routine checks happen automatically. Judgement calls stay with people.
Planning a migration?
Move the records. Keep the relationships. Show me what’s moving before somebody presses Import and carries the old problems into the new system.