Screening an older patient's regimen for medications to stop and indicated therapies to start
STOPP/START begins not with a drug list but with a person: an older adult carrying several chronic conditions and a medication list assembled over years, often by different prescribers who never saw the whole picture. Structured review starts by reconciling exactly what is being taken and why.
STOPP (Screening Tool of Older Persons' Prescriptions) and START (Screening Tool to Alert to Right Treatment) were developed by Gallagher, O'Mahony and colleagues at University College Cork, Ireland, first published in 2008. Unlike earlier implicit judgment-based approaches, STOPP/START gave clinicians explicit, criterion-based rules — a medication either meets a defined criterion or it does not, which makes the review reproducible between different prescribers and pharmacists.
Version 2, published in 2015, was developed through a Delphi consensus process with 19 experts from 8 European countries, expanding the tool to 80 STOPP criteria and 34 START criteria and reorganizing it around physiological systems.
STOPP/START organizes its criteria by physiological system — cardiovascular, central nervous system and psychotropics, gastrointestinal, respiratory, musculoskeletal, urogenital, endocrine, and analgesics — mirroring how clinicians actually think through a case: system by system, condition by condition.
This is a deliberate contrast to list-based tools that simply enumerate drugs or drug classes to avoid without reference to the patient's diagnoses. A system-based, diagnosis-aware structure lets STOPP/START catch both prescribing errors of commission (STOPP) and errors of omission (START) — the same review pass looks in both directions.
The defining innovation of STOPP/START is that it is bidirectional: it is one of the first widely-used tools to formally screen for undertreatment (START) with the same rigor traditionally reserved for screening potentially inappropriate prescribing (STOPP).
Roughly 40% of adults over 65 in high-income countries take five or more regular medications, and the figure rises further with multimorbidity. Each additional prescriber and each additional medication increases the chance of an unrecognized drug interaction, duplicate therapy, or a "prescribing cascade" — where a side effect of one drug is mistaken for a new condition and treated with another drug.
The patient profiled here carries four chronic conditions typical of the population STOPP/START targets: type 2 diabetes, atrial fibrillation, osteoporosis, and COPD — each with its own guideline-recommended therapies, and each medication with its own potential for age-related risk.
STOPP contains 80 explicit criteria identifying medications, doses, or combinations that carry a higher risk of harm than benefit in older adults — organized by the physiological system each risk acts on, from CNS sedatives to nephrotoxic NSAIDs.
Each STOPP criterion pairs a specific medication (or class, dose, or duration) with the mechanism of harm it poses in older adults — for example: "Benzodiazepines for ≥4 weeks (no indication for longer treatment) — risk of prolonged sedation, confusion, impaired balance, falls, road traffic accidents." This explicit if-then structure is what allows the tool to be applied consistently by physicians, pharmacists, and — increasingly — embedded directly into electronic prescribing systems.
In this simulation, medications like long-acting benzodiazepines, sulfonylureas, and alpha-blocker monotherapy for hypertension are flagged first because they carry the highest STOPP risk scores; low-risk, guideline-appropriate medications like metformin or a statin are flagged last, if at all.
Across its 13 sections, several risk patterns recur:
• CNS/psychotropic — benzodiazepines, tricyclic antidepressants, first-generation antihistamines: sedation, anticholinergic burden, falls • Cardiovascular — digoxin >125mcg/day with reduced renal clearance, alpha-blocker monotherapy, aspirin without a vascular indication • Endocrine — long-acting sulfonylureas (hypoglycemia), sliding-scale insulin without basal coverage • Gastrointestinal — full-dose PPIs continued beyond 8 weeks without a clear ongoing indication • Musculoskeletal/renal — NSAIDs in chronic kidney disease, hypertension, or heart failure
The common thread is that each of these is often appropriate in a younger, healthier patient — the STOPP framework flags them specifically in the context of older age and reduced physiological reserve.
STOPP does not mean "never prescribe." It flags a risk-benefit balance that shifts with age, renal function, and co-prescribed drugs — the clinician still applies judgment about the individual patient before stopping anything.
Randomized and observational studies applying STOPP screening to hospitalized and community-dwelling older adults have reported reductions in adverse drug reactions of roughly 30%, along with reductions in prescribing errors identified at admission and discharge. STOPP-guided review has also been shown to shorten the average time spent reviewing a complex medication list compared to unstructured, implicit review, because the criteria focus attention on the highest-yield targets first.
While STOPP hunts for medications doing more harm than good, START looks the other direction: 34 criteria that flag guideline-indicated therapies missing from an older patient's regimen — the undertreatment that ageism and prescribing caution can quietly produce.
Each START criterion links a documented condition to a therapy with strong evidence of benefit that is disproportionately withheld from older patients — for example: "Anticoagulant therapy in the presence of chronic atrial fibrillation where CHA₂DS₂-VASc score suggests stroke risk, and no contraindication to anticoagulation exists." In this simulation, the patient's atrial fibrillation, osteoporosis, COPD, and diabetes are each cross-checked against their START-indicated therapies — a DOAC, a bisphosphonate, an inhaled bronchodilator, a statin or ACE inhibitor — and any missing therapy glows and moves to the start bin.
Prescribing omissions in older patients are frequently driven by "therapeutic nihilism" — an assumption that an intervention is not worth the risk or effort given the patient's age — rather than by an actual contraindication. Contributing factors include:
• Underrepresentation of older, multimorbid adults in the clinical trials that generated the original guidance • Diffusion of responsibility across multiple specialists, none of whom owns the full medication list • Clinician concern about polypharmacy that paradoxically prevents adding a beneficial drug • Shorter visit times limiting the chance to revisit chronic disease management holistically
START criteria counter this by making the omission explicit and attributable, the same way STOPP does for inappropriate prescribing.
Studies of AFib patients ≥75 show roughly 30% of those with a clear indication for stroke-prevention anticoagulation are never prescribed it — often due to unfounded fall-risk concerns, despite evidence that the stroke risk from omission usually outweighs the bleeding risk from treatment.
A tool that only deprescribes risks trading one harm (adverse drug reactions) for another (undertreated disease). STOPP/START's defining feature is running both screens in the same review pass on the same problem list, so that a medication stopped for one condition does not leave a therapeutic gap unnoticed, and a medication started for another condition is checked for interactions with what remains. This integrated view is what separates STOPP/START from purely deprescribing-focused tools.
Once STOPP and START have each produced their list, the two are reconciled into a single new regimen. The clinically meaningful number is not the count of flags — it is the net change, and how carefully each change is executed.
Removing a STOPP-flagged medication is rarely as simple as stopping it at the next visit. Benzodiazepines and other CNS depressants typically require gradual tapering to avoid withdrawal or rebound symptoms; beta-blockers and clonidine can cause rebound hypertension or tachycardia if stopped abruptly; and any change should be discussed with the patient as a shared decision, with a monitoring plan for the symptom the medication was originally treating. Deprescribing guidelines (e.g., the Canadian Deprescribing Network) provide drug-specific tapering schedules that complement the STOPP flag itself.
The Beers Criteria, first published in 1991 and maintained by the American Geriatrics Society, is the most widely used alternative tool in the United States. It is primarily a list: specific drugs or drug classes to avoid or use with caution in older adults, largely independent of the patient's diagnoses.
STOPP/START differs in two structural ways: it is organized mechanistically by physiological system rather than as a flat list, and it is bidirectional, formally screening for prescribing omissions (START) as well as commissions (STOPP). Head-to-head comparative studies have generally found STOPP more sensitive than Beers at identifying medications associated with actual drug-related hospital admissions, likely because its diagnosis-linked, mechanism-based criteria track more closely with real-world harm.
A 2011 comparative study found STOPP criteria identified roughly twice as many drug-related hospital admissions as the Beers Criteria applied to the same patient cohort — attributed largely to STOPP's inclusion of diagnosis-specific and drug-drug interaction rules that a pure avoid-list cannot capture.
Because STOPP/START criteria are explicit and rule-based, they translate naturally into electronic health record alerts: a system can automatically flag a benzodiazepine prescription older than 4 weeks, or flag an atrial fibrillation diagnosis without an anticoagulant on the medication list. Several European health systems and EHR vendors have embedded STOPP/START logic directly into prescribing workflows since the mid-2010s, and this integration is generally better received by clinicians than generic drug-interaction alerts, because the criteria are narrower and more clinically specific — reducing the alert fatigue that undermines less targeted decision-support tools.
The ultimate justification for structured medication review is patient outcome: fewer adverse drug reactions, fewer avoidable hospital admissions, and better control of the conditions that were undertreated. The strongest trial evidence comes from OPERAM, a multinational European study of STOPP/START-based review in multimorbid older adults.
OPERAM (Optimising thERapy to prevent Avoidable hospital admissions in the Multimorbid elderly) was a cluster-randomized clinical trial conducted across Switzerland, the Netherlands, Belgium, and Ireland, enrolling approximately 2,000 multimorbid patients aged 70 and older with polypharmacy. The intervention arm received a structured medication review combining STOPP/START criteria with the STRIP (Systematic Tool to Reduce Inappropriate Prescribing) decision-support method, delivered by a clinical pharmacist and physician team.
Results showed the intervention meaningfully reduced the number of medications per patient and the burden of potentially inappropriate prescribing; effects on the primary composite endpoint of drug-related hospital readmission were more modest, but secondary analyses and subsequent meta-analyses support real reductions in adverse drug events when the review is implemented with adequate follow-through.
Beyond OPERAM, multiple randomized and observational studies applying STOPP/START-based review report:
• 15–30% relative reductions in adverse drug reaction incidence • Reduced fall and fracture rates when sedative and orthostatic-hypotension-inducing medications are addressed • Shorter hospital length of stay in cohorts receiving structured review at admission • Improved disease-specific control (e.g., stroke-prevention anticoagulation rates in AFib, bone-protective therapy rates after fragility fracture) when START omissions are corrected
The combined STOPP+START approach — reducing harm while closing treatment gaps — consistently outperforms deprescribing-only interventions on composite outcome measures.
The clinical value of STOPP/START is not simply "fewer pills." A regimen can improve even when medication count stays flat, because inappropriate high-risk drugs are traded for indicated, guideline-supported ones — the goal is a safer and more effective regimen, not merely a shorter one.
STOPP/START review takes clinician time — typically 10–20 minutes per patient for a thorough pass — and its benefit depends on follow-through: flags that are never acted upon, or medications stopped without adequate tapering or monitoring, do not produce the outcome gains seen in trials. The criteria also require periodic revision as evidence and guidelines evolve; a version 3 update has been anticipated to reflect newer drug classes (e.g., SGLT2 inhibitors, DOACs) more fully. Ongoing work is integrating STOPP/START logic into AI-assisted clinical decision support to reduce review time while preserving the clinician's final judgment on each patient.