Drug-food interactions in care: how CalcMenu prevents errors before the tray leaves

Warfarin and vitamin K-rich leafy greens. MAOIs and tyramine. Ciclosporin and grapefruit juice. These interactions are well documented — and they can affect any patient treated for a chronic condition. Yet in most care kitchens, no systematic check happens between the patient’s tray and their medication.

The problem is structural, not individual

It’s not that care staff don’t know. Most nurses know the main drug-food interactions. The problem is that the information doesn’t systematically flow from pharmacy to kitchen.

In a 200-resident facility, with 5 nurses per shift and 50 different treatments, it’s impossible to mentally check every tray at every meal. It’s a structural problem that requires a systemic solution.

How CalcMenu’s alert works

CalcMenu can be connected to the patient’s medication record (via export from the care software). For each meal, CalcMenu checks whether a food on the menu interacts with a patient’s medication.

If an interaction is detected:

• An alert appears in the kitchen production interface
• The floor nurse is notified
• A substitute is automatically proposed if substitution is configured

The alert is contextual: it’s not generic (“grapefruit can interact with some medications”) — it’s specific (“Mr Smith, room 204, takes X medication — the planned tray contains grapefruit juice”).

What it isn’t

This system is not a substitute for medical evaluation. It doesn’t replace consultation with the pharmacist or treating doctor. It’s an additional safety net, not the only one.

It also doesn’t cover all known interactions — the database is updated periodically and depends on the quality of the patient’s medication data in the care record.

Why it matters now

The complexity of medication regimens is increasing. Care home residents take on average around 7–9 medications simultaneously, and 60–90% take 5 or more. Interaction risk grows exponentially with the number of molecules. Kitchens can no longer ignore this dimension — and they shouldn’t have to without a tool to help.