Unpaid carers are performing pharmacovigilance, cross-department symptom surveillance, and clinical contradiction detection, without training, without tools, and without a backstop. This cluster lifts that surveillance work. It does not replace the professionals who must act on what it finds.
"She was told by an A&E nurse that her husband absolutely could not take Ibuprofen. A doctor later said he could. She had to become hyper-alert and constantly cross-examine all medical directions. She was doing pharmacovigilance work that a clinical pharmacist is trained and paid to do, with no tools, no training, and no backstop."Primary source: carer of a stroke and cardiac surgery survivor, sole coordinator across nine NHS departments
She was not a pharmacist. She had no access to a clinical decision support system. She had no way of knowing whether a symptom mentioned once at Cardiology had been noted at Neurology. She carried that information in her head, and when she got it right, a sleep apnoea diagnosis followed. When carers get it wrong, the consequences are clinical.
This cluster lifts the surveillance function. Scanning for potential contraindications. Tracking symptoms across departments. Noting when recorded advice from two different teams appears to conflict. Naming the right professional to contact. Making that contact as easy as possible.
Every function below ends at the same place: a concern named, a professional identified, a route made easy, a record kept open. The tool's job ends there. The professional's job begins.
This cluster activates across every stage, errors and contradictions do not respect the shape of the carer's journey. But it peaks hardest in the acute phases, when clinical complexity is highest and the carer is least able to carry it.
The interface below shows the flag panel for Michael Henderson's carer, continuing the same fictional persona across the cluster series. Three open concerns at different stages: one newly surfaced, one routed to a professional and awaiting confirmation, one closed after a GP review. Each flag shows exactly what the tool does and does not do.
A concern is open until a qualified professional closes it. The tool never closes a flag. The tool never reassures. Numbers here are not a score, they are a count of what remains unresolved.
Cardiology appointment 24 Jul, 2 open concerns relevant to this department will be surfaced in the consultation notepad 7 days before.
Every other cluster has design constraints that are important. In Cluster 03, the constraints are the design. A safety tool that violates any of these does not just underperform. It causes harm.
The tool scans, flags, and names the right professional to contact. It does not tell the carer what a concern means, how serious it might be, or what is likely to happen. That interpretation belongs to a clinician.
False reassurance is a harm. The tool does not say "this is probably fine," "this is likely a minor interaction," or "many people take both of these." Any statement that reduces the carer's likelihood of contacting a professional is a failure of the tool's purpose.
Surfacing a concern to a carer who is already at capacity and leaving them to work out what to do with it is not a safety tool. It is additional load. Every flag must carry a named professional and a direct route to them.
One test governs every decision in this cluster. It is applied before any feature is confirmed, any interface element is designed, and any piece of language is written.
"If this tool surfaces a concern to a carer who is already at capacity, does it make their situation better or worse in the next ten minutes?"Applied to every feature in Cluster 03 before it is built
The full build of Cluster 03 is reserved for funded development. Clinical governance, pharmacovigilance expertise, and NHS integration are prerequisites for this cluster, it cannot be built without the right institutional partnerships.