The Evidence Behind Deprescribing and How Screening Tools Flag Risky Medicines

The short answer

Deprescribing is the planned, supervised reduction or stopping of medicines that may be causing more harm than benefit, and the evidence for it is encouraging but incomplete. Randomized trials show that structured deprescribing can safely cut the number of pills older adults take, and in specific cases such as sedatives it can reduce clearly inappropriate use. What the trials have not consistently proven is that deprescribing lowers hard outcomes like death or hospitalization, largely because the studies are small, short, and varied in design. Screening tools such as the Beers Criteria and STOPP/START help clinicians spot risky medicines, but flagging a drug is a prompt to review, not an order to stop.

Why deprescribing became a research priority

Polypharmacy, commonly defined as taking five or more medicines, rises steeply with age and multimorbidity. Each added drug raises the chance of interactions, side effects, and prescribing that once made sense but no longer fits a person's goals or kidney function. Deprescribing is the mirror image of prescribing: a deliberate process of identifying candidates for reduction, tapering where needed, and watching for the return of symptoms. The concept is simple. Proving that it changes health, rather than just medication counts, is harder.

What the randomized trials show

A 2023 systematic review of randomized controlled trials on deprescribing in older adults with polypharmacy, published in the International Journal of Pharmacy Practice by Omuya and colleagues, gives a fair picture. Across the included trials, deprescribing interventions were generally feasible and safe, and most reduced the number of medications or doses taken. The picture on patient-centered outcomes such as quality of life and hospitalization was more mixed, and the trials differed widely in setting and in how they even defined deprescribing.

Individual trials sharpen the point. The EMPOWER trial (JAMA Internal Medicine, 2014) mailed older adults a plain-language brochure about the risks of sedative medicines; at six months, 27 percent in the intervention group had stopped their benzodiazepine compared with 5 percent under usual care, a large effect on a class widely flagged as risky in older adults. The OPTIMISE trial (JAMA, 2020) tested removing one blood-pressure medicine in adults aged 80 and older with well-controlled pressure; systolic pressure rose only modestly, serious adverse events did not differ significantly over twelve weeks, and longer follow-up showed no clear long-term harm or benefit. The lesson is a careful one: deprescribing can be done safely, yet "safe and feasible" is not the same as "improves survival."

How the screening tools flag risky medicines

Explicit criteria are lists that name specific drugs, doses, or drug-disease combinations to avoid or reconsider in older adults. They turn scattered clinical judgment into something a busy clinician, or a computer, can apply consistently.

The Beers Criteria

The American Geriatrics Society 2023 update to the Beers Criteria, published in the Journal of the American Geriatrics Society, is the most widely used tool of its kind in the United States. An expert panel reviewed well over a thousand studies from the preceding years and sorted medicines into categories: drugs older adults should generally avoid, drugs to avoid with certain conditions, drugs to use with caution, and dose adjustments for reduced kidney function. The 2023 edition revised guidance on anticoagulants and on estrogen in postmenopausal women, and streamlined criteria for several medicines that are now rarely used. The AGS states plainly that the list is a starting point for conversation, not a rule that automatically forbids a prescription.

STOPP/START

STOPP/START, updated to version 3 in 2023 in European Geriatric Medicine by O'Mahony and colleagues, is organized by body system and adds something the Beers approach does not emphasize. STOPP flags potentially inappropriate medicines to consider stopping; START flags potential prescribing omissions, meaning beneficial treatments an older patient may be missing. Version 3 expanded to 190 criteria, 133 STOPP and 57 START, reflecting new evidence accumulated since the prior version. Trials that applied earlier versions reduced potentially inappropriate prescribing, though evidence that acting on the criteria improves clinical outcomes has been mixed across larger studies.

Both tools share a design philosophy. They identify potentially inappropriate medicines, a phrase chosen with care, because a flagged drug may still be the right choice for a given person. The criteria are meant to trigger a review, weigh benefit against risk in context, and support a decision rather than replace it.

Where the evidence stops short

Three limits deserve honesty. First, many deprescribing trials measure process outcomes such as fewer pills and fewer flagged drugs more reliably than clinical ones, and reviews using formal grading often rate the certainty of evidence on mortality and hospitalization as low. Second, the very patients who stand to gain most, those who are frail or carry several conditions, are frequently excluded from trials, so generalizing the results is uncertain. Third, the screening tools were validated to detect inappropriate prescribing, not to prove that acting on every flag improves outcomes; a longer list of criteria is not the same as stronger outcome evidence.

None of this argues against deprescribing. It argues for doing it the way the tools intend: as a structured, shared decision that pairs an explicit criterion with the individual in front of the clinician, backed by a plan to monitor and, when appropriate, restart. This article is educational and is not medical advice; decisions about starting or stopping any medicine belong with a person's own clinician.