Three national records, three different populations
These are complete administrative records rather than panel estimates, which is what makes them worth reading at molecule level. But they are records of three different things, and a curve is only as transferable as its denominator.
| Source | What it records | What it is not |
|---|---|---|
| NHSBSA English Prescribing Dataset | Every item dispensed in primary care in England, monthly. | Not hospital-administered medicines, and not Scotland, Wales or Northern Ireland. |
| CMS Medicaid State Drug Utilization Data | Prescriptions reimbursed by Medicaid, by state and quarter. | Not the US market. Medicaid is one payer among several, its eligibility differs by state, and the mix of people it covers is not the mix the commercial market covers. |
| PBS and RPBS date-of-supply statistics | Prescriptions supplied under the Pharmaceutical Benefits Scheme in Australia, including under co-payment supply since 2012. | Not private prescriptions outside the Scheme. The most recent three months are preliminary and revise as late claims are lodged. |
The practical consequence: compare the shape of a curve across these markets, not its level. How steep the uptake was, when it turned, how quickly a class switched and how far a price fell after exclusivity — those travel. Absolute volumes do not.
Questions about the data
Does US data here cover the whole US market?
No, and the distinction matters for a forecast. The US series is CMS Medicaid State Drug Utilization Data, which records prescriptions reimbursed by Medicaid. It is a real, complete, national record of that programme — not a sample — but Medicaid is one payer, and its population is not the commercial one. Read a Medicaid curve as a Medicaid curve.
Can I compare the three countries directly?
Compare shapes, not levels. Each series counts a different denominator, so the absolute volumes are not on a common footing. What does travel between them is the shape of the curve: how steep the uptake, when the inflection came, how fast a class switched, how far a price fell after exclusivity ended.
Is this dispensing or prescribing?
Dispensing. Every one of these sources records what was supplied and reimbursed, not what a clinician wrote or what a patient took. Items are not patients either: a molecule dispensed monthly to one person and once to twelve looks the same in the count.
How recent is the data?
Each publisher runs on its own lag, and all three revise. The most recent months in any of these series should be treated as provisional, which is why launch curves here are read on a common clock from launch rather than against the calendar.
Where these analyses come from
We maintain a molecule-level database built from national prescribing sources: the NHSBSA English Prescribing Dataset (every primary-care item since 2021, 748 million rows), CMS Medicaid State Drug Utilization Data (products mapped to molecules via the openFDA NDC directory), and Australia's PBS date-of-supply data. Launches are aligned on a common clock, coding changes are filtered from genuine launches, and every figure traces to a published public series. The same database supplies uptake analogues and erosion curves for client budget impact work.
Need the curve for your molecule?
Uptake analogues, switching dynamics and erosion benchmarks for a specific class or market, built into a payer-ready model, or delivered as a standalone analysis. Fixed fees.
Talk to us