The general references are general by design, and a defined set of questions outruns them: the complex interaction, the patient on dialysis, the injectable that needs preparing, the tablet that will not go down a feeding tube, the overdose, the pregnancy exposure. For each of these, UK practice maintains a specialist source that is not merely deeper but categorically the right instrument, and knowing this layer exists, even without holding every subscription personally, is part of safe practice, because the alternative is improvising in territory where a maintained answer exists. Here is the specialist shelf, database by database.
The shelf, by question
Use Stockley’s for detailed interaction evidence and management; the Renal Drug Database and renal pharmacy for significant impairment, dialysis or haemofiltration; Medusa for injectable preparation and administration; NEWT and SPS for swallowing difficulties or enteral tubes; TOXBASE and the National Poisons Information Service for poisoning; UKTIS and BUMPS for pregnancy; specialist paediatric sources for children; local shared-care protocols for ownership; and SPS supply tools for shortages. Check the exact SmPC where authorised product information matters and involve pharmacy early when the question is high risk or unresolved.
Worked moments, each with one right door
For apixaban in severe renal impairment, combine the exact SmPC, applicable NICE or SPS guidance, the local anticoagulation pathway and renal or pharmacy expertise. Before administering a medicine through a feeding tube, check the exact product and use NEWT or SPS with pharmacy input. Use Medusa for injectable preparation, TOXBASE for suspected overdose, and Stockley’s or specialist pharmacy for complex interaction management.
When to pick up the phone
The shelf ends with humans, by design: ward and community pharmacy for the everyday residue, medicines-information services for the genuinely hard questions, NPIS for poisoning, renal pharmacy for replacement-therapy dosing. The professional skill is not holding every answer; it is recognising, fast, which of these questions you are inside, and routing there before improvising, which is also precisely the job the orientation layer exists for: an iatroX medicine page frames the medicine and links outward, and the frame is what tells you that this particular question belongs to the specialist shelf. The complete map of all the layers, general and specialist, is the pillar at /blog?q=prescribing.
Frequently asked questions
Which of these does a hospital clinician have access to already?
Usually most: NHS access arrangements commonly cover TOXBASE, Medusa and the major references through trust routes, and the pharmacy team knows exactly what your organisation holds; the access question is answerable in one conversation, ideally before the night you need it.
What should a GP or locum keep personally?
The free layer, SPS, TOXBASE registration where eligible, UKTIS/BUMPS, plus knowledge of the routes: which questions get phoned to pharmacy, which to medicines information, which to NPIS. Ownership of subscriptions matters less than owning the routing.
Are there safe general-purpose substitutes for any of these?
For their defining questions, no, which is the point of the page: the specialist sources exist because the general ones documented their own limits. Substituting enthusiasm for the right database is the error class this whole cluster is written against.
How should teams make this shelf discoverable before it is needed?
One laminated, dated page per clinical area: the specialist sources relevant to that setting, access routes, and the three phone numbers, pharmacy, medicines information, NPIS. The shelf fails at 3am only when its existence is tribal knowledge; writing it down once is the whole fix.
What single habit best protects against specialist-territory errors?
Naming the territory out loud before answering: "this is a renal-dosing question", "this is a TOXBASE question", said or written, routes the mind to the right instrument and makes improvisation feel as irregular as it is. The databases do the knowing; the clinician's contribution is the recognition.