A structured medicines review in over-65s addressed problems that usual care did not

Design

Pragmatic cluster-randomised controlled trial with process evaluation

Population

60 patients aged ≥65 on >4 long-term medicines, 6 practices in South-West Wales

Primary outcome

More than one clinical problem addressed after the intervention

Effect

22/27 (82%) intervention vs 11/30 (35%) control; adjusted OR 7.48 (95% CI 1.99–28.10)

Incomplete monitoring of repeat prescriptions is a decades-old source of avoidable harm, and one that has resisted most attempts to fix it. This pragmatic cluster-randomised trial tested the Adverse Drug Reaction Profile — a structured symptom checklist completed by the patient and then shared with the practice pharmacist — in patients aged 65 or over taking more than four long-term medicines and living at home.

Sixty patients across six practices in South-West Wales were randomised, with three lost. Patients in the intervention arm were more likely to have more than one clinical problem addressed: 22 of 27, against 11 of 30 in usual care, an adjusted odds ratio of 7.48 (95% CI 1.99–28.10). Fewer reported pain by the end. Two cases were singled out by the authors as critical — a vitamin B12 deficiency detected in a patient on metformin, and a flecainide deprescribed.

The size of this trial demands restraint. Sixty patients recruited from a single practice group is a feasibility study in all but name, and an odds ratio whose confidence interval runs from 2 to 28 tells you the direction and almost nothing about the magnitude. What it does establish is that a patient-completed symptom checklist surfaces problems that routine repeat prescribing review does not, and clinicians' own objection was about staff time rather than about whether the findings were real.

• The mechanism is a patient-completed symptom checklist, not a prescriber-led drug list review — the two find different things
• Metformin and B12 is the example worth remembering: a monitoring omission that is silent until it is not
• Target patients on more than four long-term medicines; that is the population tested
• Pair it with the pharmacist rather than the consultation, which is how this trial kept it feasible
• Staff time was the clinicians' objection and it is a real one — pilot it on a defined subset before adopting it wholesale

Why it matters

It asks the patient what they are experiencing rather than asking the record what they are taking, and those produce different problem lists.

Don't overread it

Sixty patients from one practice group, with an adjusted odds ratio spanning 1.99 to 28.10. This supports feasibility and plausibility; it does not establish effect size, and no patient outcome beyond problems addressed was measured.

The statistics, in plain English

An odds ratio of 7.48 sounds decisive until you read the interval: the true effect is somewhere between roughly twice and twenty-eight times the odds of usual care. That width is what 57 analysed patients buys. The comparison 22/27 versus 11/30 is the number to hold on to, because it is the raw observation rather than a model output — and even that rests on fewer than 60 people in six practices belonging to one group, so practice-level clustering is doing more work than the confidence interval suggests.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.