Underneath every electronic prescription in the NHS sits a resource most clinicians have used thousands of times without opening once: dm+d, the NHS dictionary of medicines and devices. Maintained through the NHSBSA as the recognised NHS standard for identifying and communicating medicinal products across clinical systems, refreshed weekly, it is the reason your system, the pharmacy's system and the payment machinery all agree on exactly what was prescribed. Comparing it with the BNF is instructive precisely because they share almost nothing except the word medicines: one is a dictionary of products, the other a reference for prescribers, and confusing the layers explains a surprising number of digital-health misunderstandings.
What dm+d actually contains
The dictionary's core is a hierarchy of identity. The active ingredient sits at the top. The virtual medicinal product (VMP) describes the abstract clinical thing, the drug, form and strength, independent of who makes it. The actual medicinal product (AMP) is a specific manufacturer's version of that thing. Pack-level entries (VMPP and AMPP) descend to the exact pack: size, price, availability. Around the hierarchy hang the attributes systems need: SNOMED CT identifiers, which is how dm+d concepts travel through UK clinical records; GTIN barcode data; Drug Tariff and pricing information; and availability status, all refreshed weekly. It is, deliberately, a complete answer to "what product, which pack, what code, what price", and a complete non-answer to everything clinical.
Why dm+d cannot answer prescribing questions
No indications, no dosing guidance, no cautions, no interactions, no monitoring: the dictionary identifies, it does not advise, and this is design, not deficiency. The clinical layers exist elsewhere and the stack composes cleanly: the BNF for concise prescribing guidance at drug level; emc for the authorised text at exact-product level, the AMP's clinical documentation, in effect; the local formulary for what the system supports; and iatroX Medicines as the structured orientation layer navigating between them. dm+d is the spine those layers hang off in software: when a decision-support system fires an alert, a formulary flags a status, or an SmPC is linked from a record, a dm+d/SNOMED identifier is usually doing the pointing.
Following one medicine through the layers
Trace apixaban downward and the architecture becomes concrete. At medicine level, the iatroX apixaban page orients: what it is, which preparations exist, what distinguishes them. At product level, each preparation, the 5 mg tablets, the distinct 1 mg/ml oral suspensions, has its own emc SmPC, the authorised text for that exact product. At dictionary level, dm+d holds the VMP for apixaban tablets and suspensions, the AMPs for each manufacturer's version, and the pack entries with codes and prices, which is what the prescribing system actually writes when a clinician clicks. And at governance level, the local formulary attaches status to the choice: /blog/how-to-find-use-local-nhs-formulary. Four layers, one medicine, no redundancy: identity, authorisation, guidance and governance each held where they belong.
Who should actually use dm+d
Pharmacists reconciling products and packs; developers and digital-health teams building anything that touches prescriptions, where dm+d/SNOMED conformance is simply the entry ticket; medicines-optimisation and finance teams working with Drug Tariff and pricing data; and clinicians in the specific moments the dictionary answers best, confirming exactly which product and pack a record refers to, or why two systems disagree about a "same" medicine. For everyone else, dm+d's gift is invisible reliability: the reason the stack's clinical layers can point at products unambiguously at all.
Frequently asked questions
Is dm+d the same as SNOMED CT?
dm+d concepts carry SNOMED CT identifiers and form the UK drug extension of that terminology; SNOMED is the broader clinical terminology, dm+d the medicines-and-devices dictionary within it for NHS purposes.
Why do prescribing systems sometimes offer products a formulary excludes?
Because the dictionary lists what exists while the formulary governs what is preferred, and system configuration decides how loudly governance is surfaced; the mismatch is a configuration question, not a dm+d error.
Can the public look up dm+d?
The data is openly published through NHSBSA routes, though it is built for systems rather than reading; for human-shaped questions, the clinical layers above it are the right doors.
Why should a working clinician care about VMPs versus AMPs at all?
Because the distinction explains everyday system behaviour: generic prescribing writes at VMP level and dispensing resolves to an AMP, which is why the box the patient receives can change while the prescription does not, and why brand-specific prescribing exists as a deliberate exception for the classes where the exact product matters. Understanding the two levels turns several recurring patient conversations from mysteries into explanations.
Is there an equivalent dictionary for devices?
dm+d's scope includes devices used in NHS prescribing, packs and appliances among them, which is why the same identifiers appear across prescribing and dispensing systems for dressings and appliances as for tablets; the clinical layers for devices, though, live with manufacturers, MHRA guidance and local procurement rather than in any monograph.