STOPP/START criteria explained: a practical guide (v3)
Medicine in Practice · 7 July 2026
prescribing safety STOPP START criteria deprescribing older people medication review
What are the STOPP/START criteria?
The STOPP/START criteria are an explicit, evidence-based checklist for reviewing medicines in older people. STOPP (Screening Tool of Older Persons' Prescriptions) flags drugs that are potentially inappropriate and should probably stop; START (Screening Tool to Alert to Right Treatment) flags treatments that are probably missing and should start. If you run medication reviews in general practice, the STOPP START criteria give you a structured second pair of eyes on polypharmacy: a way to catch the harm that hides in a long repeat list.
They were built for patients aged 65 and over, the group where multimorbidity, polypharmacy and age-related changes in drug handling collide. The current set is version 3, published in 2023. STOPP/START is widely used across UK primary care for reviewing prescribing in older people, and current NICE guidance on medicines optimisation supports structured medication review for patients on several medicines.
Key takeaways
- STOPP flags potentially inappropriate medicines to consider stopping; START flags evidence-based treatments that are missing.
- Version 3 (2023) is the current set, with around 190 criteria organised by physiological system.
- The tool supports clinical judgement inside a structured medication review; it is not a tick-box that overrides the conversation with the patient.
- Common STOPP flags include long-term NSAIDs without gastroprotection, long-term benzodiazepines or Z-drugs, and high anticholinergic burden.
- A flag is a prompt to think, document and discuss, and every change stays individualised.
STOPP/START version 3: what changed in 2023
Version 3 was produced by an expert European panel led by Professor Denis O'Mahony and published in European Geriatric Medicine in 2023. It grew to roughly 190 criteria, up from 114 in version 2. Most of that growth is on the STOPP side, reflecting newer drug classes and trial evidence since the previous version.
For day-to-day work the headline is simple: more of what you actually see on a modern repeat list is now covered, including newer agents and clearer wording on anticholinergics, falls-risk drugs and cardiovascular omissions. You do not need to memorise all 190. What matters is knowing the tool exists, knowing the categories, and having the full list to hand when you review. Our interactive STOPP/START v3 reference and drill keeps the criteria searchable so you can check a flag mid-review rather than trusting memory.
How the criteria are organised
Both tools are grouped by physiological system: cardiovascular, central nervous system, renal, gastrointestinal, musculoskeletal, urogenital and endocrine, plus cross-cutting sections on drugs that predictably increase falls risk and on anticholinergic burden. That structure is deliberate. It maps onto how you think during a review, working system by system rather than drug by drug.
Each criterion is written as a specific, checkable statement: a named drug or class, a clinical context, and the reason it is inappropriate or indicated. That specificity is the point. "Review the sedatives" is vague; "long-term benzodiazepine in a patient with recurrent falls" is a flag you can act on and code.
STOPP/START examples worked through
The criteria are easiest to see in a real list. Take Mr Ali, 78, on 12 regular medicines, booked in after two falls in six months. Here is how a few STOPP flags and one START omission surface.
Long-term NSAID without gastroprotection
Mr Ali takes naproxen most days for knee osteoarthritis and has been on it for over a year with no proton pump inhibitor. STOPP flags a regular NSAID used long term without gastroprotection because of the ulcer and bleeding risk, and that risk climbs in older patients and in anyone also taking an SSRI, anticoagulant or antiplatelet. He is on an SSRI too, which compounds it. The action is not simply "stop the naproxen". You weigh his pain control, consider a topical NSAID or paracetamol, and if an oral NSAID genuinely stays, you add gastroprotection and set a review. Either way, you document the reasoning.
Benzodiazepines and Z-drugs
He has taken zopiclone nightly for years. STOPP flags long-term benzodiazepines and Z-drugs, particularly where there is a falls history, because of sedation, impaired balance and next-day drowsiness. Handle it as a deprescribing conversation. Z-drugs and benzodiazepines come down through a planned taper over weeks, and the Royal College of Psychiatrists guidance on stopping is a sensible reference for the schedule. It also happens to be exactly the kind of addictive medicine the Network Contract DES asks PCNs to prioritise in structured medication reviews.
Anticholinergic burden
Mr Ali is on oxybutynin for urinary frequency. On its own it looks minor. Stacked with other drugs that carry anticholinergic activity, it adds to a burden linked with confusion, falls and, on current evidence, cognitive decline. STOPP flags strong anticholinergics in patients with a history of falls or cognitive impairment. The move is to add up his total anticholinergic load, then look for a safer alternative for his overactive bladder. Totting up cumulative burden across a list is where a good drug-interaction training reference earns its keep.
START: the drugs that should be there
STOPP gets the attention, but omissions cause just as much harm. Mr Ali has been on a long-term oral corticosteroid for polymyalgia rheumatica with no bone protection. START prompts you to consider bone protection in patients on long-term systemic steroids. It also covers the classics you should never miss: an anticoagulant in atrial fibrillation, a statin where there is documented cardiovascular disease and no contraindication, and an ACE inhibitor in heart failure with reduced ejection fraction. A review that only ever stops things is doing half the job.
| Tool | Common example | What to check |
|---|---|---|
| STOPP | Long-term NSAID without gastroprotection | GI bleed risk; SSRI, anticoagulant or antiplatelet co-prescribing; topical or non-drug options |
| STOPP | Long-term benzodiazepine or Z-drug | Falls and sedation; plan a taper over weeks |
| STOPP | High anticholinergic burden | Cumulative load across the whole list; cognition and falls |
| STOPP | Duplicate drug class | Two drugs from the same class with no added benefit |
| START | Anticoagulant in atrial fibrillation | Stroke prevention where there is no contraindication |
| START | Bone protection on long-term steroids | Fracture risk |
| START | Statin in documented cardiovascular disease | Secondary prevention where appropriate |
Using STOPP/START in a medication review without turning it into a tick-box
The failure mode is easy to spot once you have seen it. Someone runs the list against the criteria, generates fifteen flags, and either fires fifteen tasks at the GP or freezes. Neither helps the patient. STOPP/START is a prompt layer, not the review itself.
In practice it sits inside a structured medication review, after you have gathered the history, the bloods and, most of all, what matters to the patient. If you want the full workflow, see our guide on how to run a structured medication review; for the monitoring side, high-risk drug monitoring in primary care covers the bloods that decide whether several of these flags are even safe to action.
A workable order:
- Prepare from the record first: current list, indications, recent bloods, renal function, falls and admissions.
- Work system by system and let STOPP/START surface candidates, noting each flag rather than acting on it straight away.
- Prioritise. Which flags carry the most risk for this person right now? A falls-risk sedative in a patient who keeps falling beats a theoretical interaction.
- Decide with the patient. Every change is shared, individualised and within your competence and local shared-care arrangements.
- Document and code the decision, including the flags you deliberately left alone and why. "Reviewed, continued, patient informed" is a valid, defensible outcome.
That last point matters for governance. A flag you considered and chose not to act on is good practice when the reasoning is recorded; the same flag ignored silently is a gap. This is the difference between good, missed and unsafe, and it is exactly what you can rehearse in the scored practice labs before you meet a real patient. STOPP/START is one tool among several: the competency-mapped curriculum sets medication review and deprescribing alongside consultation skills and monitoring so the whole thing hangs together.
Frequently asked questions
What does STOPP/START stand for?
STOPP stands for Screening Tool of Older Persons' Prescriptions and lists potentially inappropriate medicines to consider stopping. START stands for Screening Tool to Alert to Right Treatment and lists evidence-based treatments that may be missing. They are used together during medication review in older people.
Is STOPP/START version 3 the current version?
Yes. Version 3 was published in 2023 in European Geriatric Medicine by an expert European panel led by Denis O'Mahony. It expanded the set to around 190 criteria, up from 114 in version 2, and is the version you should work from.
What age group are the STOPP/START criteria for?
They were developed for people aged 65 and over, where polypharmacy, multimorbidity and age-related changes in drug handling make inappropriate prescribing and prescribing omissions more likely. Clinical judgement still applies to the individual in front of you.
Does STOPP/START replace clinical judgement in a medication review?
No. The criteria are explicit prompts that surface candidates for change. Each flag is a reason to think, discuss with the patient and document a decision, not an instruction to stop or start automatically. Decisions stay individualised and within local shared-care governance.
Where can I find the full STOPP/START v3 criteria?
The full list is published in the 2023 European Geriatric Medicine paper. For day-to-day use an interactive searchable reference is faster; you can check and practise the version 3 criteria in the STOPP/START drill on this site.
You can work the v3 criteria against real cases in the interactive STOPP/START v3 drill and see where your calls land as good, missed or unsafe.