Work Synthea EHR
I pulled every patient with high blood pressure and asked what they were being treated with. The top answer was a painkiller.
Python · SQL · SQLite · Synthea
The finding
Paracetamol outranked every actual blood-pressure drug. The real antihypertensives sat further down the list, behind ordinary background medication.
Method
Synthetic patient records in SQLite; JOINs across patients, conditions, medications and encounters, with cohorts defined by subquery.
So what
You cannot read a treatment pattern off a ranked list without knowing which drug class is clinically relevant. Domain knowledge is not decoration on top of the analysis — without it, the analysis returns a confident wrong answer.
Interactive
Pick a condition to see the top prescribed medications for that cohort of synthetic patients.
Define a group. Look at their treatment.
Pick a condition and see what its patients were actually prescribed — the top 5 drugs, by how many of the cohort received each.
Full write-up
What I did
Synthea generates synthetic patient records that mimic the structure of real electronic health records. I took the CSV outputs, loaded them into SQLite, and used SQL to explore how EHR data fits together relationally.
The practical work was: writing JOINs across patients, conditions, medications, and encounters tables; building cohort definitions using subqueries (e.g. patients with a hypertension diagnosis); then querying what medications those patients were on — as a way of practising the kind of cohort-and-exposure framing used in observational pharmacoepidemiology.
Caveat: read the cohort’s treatment carefully
The hypertension cohort’s single most-prescribed drug wasn’t a first-line blood-pressure medication — it was acetaminophen, a common background painkiller. The actual antihypertensives (hydrochlorothiazide, an atenolol–chlorthalidone combination, an amlodipine–HCTZ–olmesartan combination) show up further down the list. Defining a cohort and reading off its top drug isn’t enough on its own — you have to already know which drug class is clinically relevant to the condition, or a background medicine will look like the therapy.
Why synthetic data
Real EHR data requires governance approvals I don’t yet have access to. Synthea’s records are realistic enough in structure to practise the SQL and learn the data model, without the access barriers. The limitation is that the data is not real, so any findings are illustrative rather than meaningful.
Stack
Python · SQL · SQLite · Synthea