Taking fewer medicines in later life: what deprescribing does and does not do

Deprescribing — the planned, supervised reduction or stopping of medicines that may be doing more harm than good — appears to be safe when it is done as a deliberate, patient-specific process, and it reduces the number of inappropriate prescriptions older people carry. What it has not clearly been shown to do, in randomised trials, is make people live longer or measurably improve their day-to-day health [s1][s2]. That gap between a sensible idea and proven benefit is the honest state of the evidence.

This is a topic where numbers matter and where nothing here is a prompt to change a medicine. Stopping or altering a drug is a clinical decision made with the prescriber who knows the full picture; abrupt self-discontinuation can be dangerous.

Why polypharmacy is a target at all

Taking several medicines at once — often defined as four or more — becomes common with age as conditions accumulate, and it is associated with falls, confusion, hospital admissions and drug interactions. The appeal of deprescribing is intuitive: if some of those medicines no longer serve a purpose, removing them should reduce harm without costing benefit. The question the trials try to answer is whether that intuition survives contact with data.

What the mortality evidence shows

The largest synthesis, a 2016 systematic review, pooled 132 studies covering 34,143 participants with a mean age of about 74 [s1]. In non-randomised studies, deprescribing was associated with a large reduction in death (odds ratio 0.32, 95% confidence interval 0.17 to 0.60) — the kind of result that fuels enthusiasm [s1]. But that association did not hold up in the randomised studies, where the effect on mortality was not statistically significant (OR 0.82, 95% CI 0.61 to 1.11) [s1]. Non-randomised designs are prone to confounding: healthier patients are more likely to have medicines safely withdrawn, which can make deprescribing look protective when it is partly a marker of who was well to begin with.

The review did find one signal that survived randomisation. When deprescribing was delivered as a patient-specific intervention — a tailored review of an individual's medicines — mortality fell significantly (OR 0.62, 95% CI 0.43 to 0.88) [s1]. Generalised educational programmes aimed at prescribers, by contrast, did nothing to mortality (OR 1.21, 95% CI 0.86 to 1.69) [s1]. The method matters more than the message. The review's wider purpose was to establish whether deprescribing is safe and feasible at all, alongside its secondary outcomes of adverse drug-withdrawal events, physical and psychological health, and quality of life; its overall conclusion was that a supervised, individualised approach appears both safe and workable, even where a survival benefit is not demonstrable [s1].

What the Cochrane review concluded

A Cochrane review of 32 studies took a stricter view of the same field, looking at interventions — typically structured medication reviews — intended to improve appropriate polypharmacy [s2]. Its verdict was cautious. It was uncertain whether these interventions produced clinically significant improvement; effects on the number of potentially inappropriate medications were small and rested on very low-certainty evidence (standardised mean difference −0.22, 95% CI −0.38 to −0.05) [s2]. The clearest signal was a reduction in potential prescribing omissions — medicines that should have been offered but were not, a reminder that "appropriate" prescribing means adding as well as subtracting (risk ratio for one or more omissions 0.40, 95% CI 0.18 to 0.85; very low-certainty) [s2]. The review found little or no difference in hospital admissions or quality of life [s2].

How to read this

Two things are true at once. Deprescribing done well — an individualised review by someone who knows the patient — is a reasonable, evidence-supported practice, and the safety data are reassuring. And the strong claims sometimes made for it, that trimming medicines will extend life or transform wellbeing, run ahead of what randomised trials have shown. The benefit that is best supported is narrower: fewer inappropriate and missing prescriptions, which is worth having on its own terms.

The principle recurs across specific drug classes. Guidance on tapering benzodiazepines wrestles with the same tension between a drug's diminishing value and the risk of stopping it badly. Trials of stopping beta-blockers after a heart attack test whether a once-standard medicine is still needed years on. And a statin trial in healthy over-70s shows how hard it is to demonstrate that any single medicine changes what older people care about most — independent, healthy years.

What to watch

The most useful research now is not whether deprescribing "works" in the abstract but which specific medicines, in which patients, can be reduced with net benefit — and how to build the individualised reviews that carried the only randomised mortality signal into routine care without overwhelming it.