The UK’s power information downside
The uncooked dataset printed within the FOI response contained over 132,000 information factors, itemizing the product identify, energy, producer and quantity of prescriptions disbursed yearly from the beginning of 2022 till February 2025.
Not like Germany, Australia or Canada, the UK has no centralised monitoring system, no obligatory product registry and no standardised naming conventions. Clinics, pharmacies and importers all document info in a different way, usually counting on free-text fields reasonably than structured inputs.
In its response, the NHSBSA makes clear that the dataset is inherently inconsistent. Non-public unlicensed hashish prescriptions are ‘manually recorded from usually handwritten prescriptions,’ and there may be ‘no standardised naming conference for the product names, energy, or quantity.’
Because of this, the identical product usually seems a number of occasions underneath barely totally different descriptions, an issue the FOI workforce warns ‘might result in duplication of some product names, energy, or quantity.’
A lot of this prescribing was initially captured as an ‘unspecified drug,’ as a result of unlicensed hashish merchandise weren’t current within the NHSBSA drug database. They have been solely recognized later via ‘a further evaluation course of,’ which assigns the merchandise to the date written reasonably than the date submitted.
The NHSBSA additional cautions that these figures ‘could also be topic to vary if prescriptions are submitted to us in later months’.
Enterprise of Hashish reconstructed the dataset utilizing a canonical product dictionary derived of recognized merchandise, normalised all FOI entries, and utilized fuzzy-matching with efficiency and model constraints to assign every document to an actual, verifiable medical hashish product.
Focusing particularly on flower merchandise, protection various considerably by 12 months. The 2023 and 2024 information achieved match charges above 90%, that means the overwhelming majority of disbursed flower quantity might be reliably assigned to particular merchandise. Nonetheless, 2022 information was much less dependable, with solely 77.5% of quantity efficiently matched as a consequence of naming errors, lacking efficiency info and ambiguous entries.
The extraordinary effort wanted to ascertain even primary market metrics illuminates a foundational problem out there, making it even tougher to offer data-driven arguments for higher entry, losing a useful alternative to analyze this usually maligned business, and providing an unfair business benefit to these with sufficient personal information to derive actionable insights.
