A medication list is one of the clearest fall-risk signals a family can actually see. This page reviews what the published research shows about medications and falls in adults 65 and older: which drug classes are associated with falling, what the number of medications signals, when risk concentrates, and what is and is not proven about fixing it.
This is a plain-language review of the published research on medications and fall risk. It was prepared by the research desk of Bedford Medical Alert®, a Canadian company that sells monitored personal emergency response systems. We state that interest plainly so readers can weigh it. Bedford does not prescribe, review, or adjust medications, and nothing on this page describes a Bedford service. The review was written under a standing rule that null and negative findings are reported with the same prominence as positive ones.
Prepared by the Bedford Medical Alert research desk, August 2026; every source is cited in full below. Every citation was resolved live against PubMed, ClinicalTrials.gov, or the named official source on July 6, 2026, and the literature is current to that date; no figure is quoted from memory. This page is part of our research section.
One rule sits above everything else on this page: no one should stop or change a medication on their own. These medications treat real conditions, and stopping abruptly can be more dangerous than continuing. Medication decisions belong with a doctor or pharmacist. This page is educational; it is not medical advice.
About one in four Canadian seniors was prescribed ten or more drug classes in 2021, according to the Canadian Institute for Health Information [1]. That matters for falls for two reasons: a consistent short list of drug classes is reliably associated with falling, and the sheer number of medications is itself a signal.
The strongest and most consistent associations belong to psychotropic medications. In the current meta-analyses, antidepressants, antipsychotics, and benzodiazepines are each associated with roughly 40 to 100 percent higher odds of falling, with SSRIs at the top of the range; opioids and antiepileptics are also significantly associated [2, 3, 4, 5]. Cardiovascular drugs are not uniform: loop diuretics show a modest increase, most other heart and blood-pressure drug classes are inconsistent, and beta-blockers appear slightly protective [3]. Taking four or more prescription medications daily is linked to roughly 1.5 to 2 times the risk of recurrent falls in a Canadian systematic review [6], a figure independently corroborated by a pooled estimate of 1.75 elsewhere [4].
Timing is real, and the cleanest evidence is from Ontario. Injurious-fall risk nearly doubles in the first two weeks after an older adult starts a blood-pressure medication [7], and starting an opioid carries a 62 percent higher hazard of a fall or fracture within 14 days than starting an anti-inflammatory [8].
The weak spot is intervention, and this page reports it plainly. Only one small, older randomized trial showed that withdrawing psychotropic medication cut falls, and the authors reported that permanent withdrawal was very difficult to sustain [9]. Trials of reducing anticholinergic burden did not significantly reduce falls, and the definitive deprescribing trials with a falls endpoint are still running. The defensible position: certain medications are associated with falls, the risk is highest right after a change, a pharmacist review is a sensible and often no-cost step, and no one should stop a medication on their own.
In 2021, about one in four Canadian seniors was prescribed ten or more drug classes; cardiovascular drugs made up five of the seven most common classes, and 23.4% of seniors used an antidepressant [1]. Falls are the leading cause of injury-related hospitalization among Canadian seniors, and about 20% of community-dwelling seniors report a fall each year [10]. The two facts overlap in the same population.
The modern evidence base is a series of large systematic reviews and meta-analyses covering hundreds of observational studies [2, 3, 4], which refined an earlier landmark synthesis [5]. Table 1 shows the pooled adjusted odds ratios. Psychotropics are the most consistent signal: antipsychotics, antidepressants (SSRIs highest), and benzodiazepines. Opioids and antiepileptics are significantly associated. Cardiovascular classes are heterogeneous: loop diuretics show a modest increase while beta-blockers appear slightly protective, which undercuts any blanket claim about “heart pills” [3]. A European consensus panel (STOPPFall) lists 14 medication classes as fall-risk-increasing drugs, most of them psychotropic, and explicitly notes that the fall-prevention effect of withdrawing them has not yet been tested [11]. The anticholinergic link is agent-specific rather than a clean class effect [12]. Null findings deserve equal billing: non-opioid analgesics, NSAIDs, and anti-Parkinson drugs did not reach significance after adjustment [4].
Table 1. Pooled odds ratios for falls, adjusted unless noted. Confidence intervals as reported by the source meta-analyses.
| Medication class | Odds ratio (CI) | Source |
|---|---|---|
| SSRI antidepressants | 2.02 (1.85 to 2.20) | [2] |
| Antidepressants (all) | 1.57 (1.43 to 1.74) | [2] |
| Antipsychotics | 1.54 (1.28 to 1.85) | [2] |
| Tricyclic antidepressants | 1.41 (1.07 to 1.86) | [2] |
| Benzodiazepines (all) | 1.42 (1.22 to 1.65) | [2] |
| Benzodiazepines, long-acting | 1.81 (1.05 to 3.16) | [2] |
| Benzodiazepines, short-acting | 1.27 (1.04 to 1.56) | [2] |
| Opioids | 1.60 (1.35 to 1.91) | [4] |
| Antiepileptics | 1.55 (1.25 to 1.92) | [4] |
| Polypharmacy (as a variable) | 1.75 (1.27 to 2.41) | [4] |
| Loop diuretics | 1.36 (1.17 to 1.57) | [3] |
| Beta-blockers | 0.88 (0.80 to 0.97) | [3] |
Digoxin, 2.06 (1.56 to 2.74), and statins, 0.80 (0.65 to 0.98), are unadjusted estimates [3]. All estimates are pooled from observational studies; see the section below on what these numbers can and cannot establish.
A Canadian systematic review of 18 observational studies in community-dwelling adults 65 and older found that taking four or more prescription medications daily was associated with 1.5 to 2 times the risk of recurrent falls, meaning two or more falls in 12 months [6]. That is a descriptive range across studies rather than a single pooled number, and several included studies lost significance after full adjustment, so “roughly” belongs in any honest restatement. The range is independently corroborated: a separate meta-analysis pooled polypharmacy at an adjusted odds ratio of 1.75 [4]. There is no single agreed cut-point; four-or-more is this review’s definition, five-or-more is the more common general one, and ten-or-more is the threshold in the Canadian prevalence figure [1, 6]. The medication count also tracks how many conditions a person manages, which is part of what the association reflects.
This is where Canadian pharmacoepidemiology is strongest, because self-controlled and new-user designs remove much of the usual confounding. In Ontario data on 543,572 older adults starting an antihypertensive, the rate of falls requiring hospital care was 94% higher in the first 14 days after starting (and 69% higher over the first 45 days), consistent across thiazides, ACE inhibitors, calcium-channel blockers, and beta-blockers, but not angiotensin-receptor blockers [7]. In Ontario data on more than one million older adults, starting an opioid was associated with a 62% higher hazard of a fall or fracture within 14 days, compared with starting an NSAID [8]. A US cohort adds that serious fall injuries track antihypertensive intensity, with the strongest signal in people who already had a fall injury [13]. Consumer guidance from Mayo Clinic states the same principle plainly and notes that a new medicine can also destabilize a drug someone was previously stable on, through an interaction [14]. The direct evidence covers blood-pressure drugs and opioids; extending “the first days are the riskiest” to every medication change is reasonable caution rather than a measured fact.
Sorting out safety for a parent whose prescriptions just changed? Call 1-888-755-3055 for a complimentary, no-pressure phone consultation. We will tell you plainly what monitoring can and cannot do.
Nearly all fall-risk estimates for medications come from observational studies, where the disease being treated (depression, insomnia, pain, heart disease) can itself raise fall risk. The clearest demonstration involves blood-pressure drugs: observational studies find an association [7, 13], but a meta-analysis of 58 randomized trials covering 280,638 participants found antihypertensive treatment was not associated with falls overall (relative risk 1.05, CI 0.89 to 1.24), though it did increase hypotension and syncope [15]. When treatment is randomized, the “blood pressure pills cause falls” signal largely disappears. Within the observational data, the fact that SSRIs rank above tricyclic antidepressants is pharmacologically counterintuitive and almost certainly reflects who gets prescribed what [2]. And the anticholinergic-burden literature disagrees with itself at the measurement level: at least 19 different burden scales exist, and their associations with falls are inconsistent across validation studies [16]. Class-level signals are robust; drug-versus-drug rankings are soft.
One randomized trial stands as the clear positive: in 93 older New Zealanders, gradually withdrawing psychotropic medication cut the fall rate by 66% (relative hazard 0.34), but the authors’ own conclusion was that permanent withdrawal is very difficult to achieve [9]. The 2012 Cochrane review of community fall prevention (159 trials, 79,193 participants) carries that finding forward on the strength of that single trial, alongside one other single-trial result in which a prescribing-modification program for family physicians reduced the risk of falling; its robust multi-trial winners are exercise and home-safety modification, not medication review [17]. Two rigorous Quebec cluster trials proved that deprescribing itself is achievable: direct-to-patient education cut benzodiazepine use (27% versus 5% discontinuation at six months) [18], and a pharmacist-led program cut inappropriate prescriptions broadly (43% versus 12%) [19]. Neither measured falls as an outcome. Interventions to reduce anticholinergic burden did not significantly reduce falls in a meta-analysis of seven trials (odds ratio 0.76, CI 0.52 to 1.11) [20], and a Cochrane review found no detectable effect at very low certainty [21]. A Canadian review protocol described current practice bluntly: withdrawal of fall-risk-increasing drugs is done on presumptive benefit despite limited evidence of effectiveness [22]. The two trials designed to settle the question are still running: ADFICE_IT in the Netherlands, with time to first injurious fall as its primary outcome [23], and DeFRID in India, which tests a deprescribing tool’s effect on fall frequency and fall-risk scores [24]. Anyone stating today that medication reviews prevent falls is ahead of the completed evidence.
In Ontario, the MedsCheck program provides an annual one-on-one pharmacist medication review at no charge to anyone with a valid OHIP card who takes three or more chronic prescription medications, with a MedsCheck at Home version for people who cannot get to a pharmacy [25]. The Canadian Medication Appropriateness and Deprescribing Network, a national network of clinicians, researchers, and patient advocates co-founded in 2015, publishes free plain-language decision aids and runs the “Do I still need this medication?” public campaign [26, 27]. To keep the claim honest: these are well-designed, accessible services for reviewing medications, and the section above applies to them too; no trial has demonstrated that any specific review service reduces falls. They are sensible steps, not proven fall prevention.
A medication list is one of the clearest fall-risk signals a family can actually see. Three things on it deserve attention: how many medications there are, whether any belong to the classes in Table 1, and what changed recently.
Do not adjust anything yourselves. The point of the evidence above is narrower than it may appear: it identifies which prescriptions are worth a professional conversation, not which to stop. Abruptly stopping a benzodiazepine, an antidepressant, or a heart medication can be more dangerous than continuing it. The conversation belongs with the prescriber or a pharmacist [25].
Book the review. In Ontario, if your parent takes three or more chronic medications, the pharmacist review is free and annual, and there is an at-home version for people who cannot travel [25]. Bring the complete list, including over-the-counter medicines, sleep aids, and supplements, since interactions are part of what the review is for [14]. Our practical guide to medications that increase fall risk in seniors covers what to ask when you get there.
Give the first two weeks after any medication start or dose change extra attention. The Ontario evidence is specific about blood-pressure drugs and opioids [7, 8]; treating every change with the same care is a sensible extension, not a research finding. Extra attention can be as simple as more frequent calls or visits during the adjustment window, and more caution with nighttime trips to the bathroom.
Be clear-eyed about what monitoring can and cannot do during that two-week window. No alert system prevents a fall, and no sensor on the market catches every fall. What a monitored system such as Bedford’s Home Freedom can change is what happens after a fall, and that matters most precisely when risk is elevated and the body is still adjusting to a new prescription: the research on delayed help, and on why buttons so often go unpressed, is the subject of the companion review in this series.
The medications-and-falls literature is overwhelmingly observational. “Fall-risk-increasing drug” is a term of association, not proven causation, and this review has tried to flag that distinction wherever it matters.
Confounding by indication is the field’s central weakness, demonstrated most clearly by the antihypertensive contrast above [15]. Part of every observational association reflects the patient, not the pill.
Drug-versus-drug rankings within a class are unreliable, and the anticholinergic-burden scales disagree with one another [2, 16]. This review therefore stays at class level.
Polypharmacy is partly a proxy. The count of medications tracks comorbidity and frailty, which independently cause falls; several studies in the Canadian review lost significance after full adjustment [6].
The intervention question is open. The definitive falls-endpoint deprescribing trials have no published results yet [23, 24], and no Canadian trial links any specific review service, including MedsCheck, to fewer falls.
This review is targeted rather than a registered systematic review, and it was prepared by a company that sells monitored alert systems, a product category adjacent to the topic. The disclosure appears in the opening section. Bedford has no commercial stake in any medication, medication review service, or deprescribing program discussed here.
Searches and citation resolution were performed on July 6, 2026 using PubMed (search, metadata, full text where needed, and identifier conversion) and ClinicalTrials.gov, plus web search for grey literature from named official sources (the Canadian Institute for Health Information, the Public Health Agency of Canada, the Ontario Ministry of Health, Mayo Clinic, and the Canadian Medication Appropriateness and Deprescribing Network). Every PMID, DOI, and trial registry number was resolved live during the research session; each figure is taken from the resolved record’s abstract or full text, or from the named official source. Confidence intervals are written “X to Y” throughout. The Canadian polypharmacy source was verified in full text, not abstract only.
If the adjustment window has you worried about someone who lives alone, start with a complimentary, no-pressure phone consultation at 1-888-755-3055. No long-term contracts, and the equipment is included with the subscription.
Prepared by the Bedford Medical Alert research desk. Literature search July 6, 2026; page published August 2026. This public edition is derived from an internal evidence brief; the internal brief is a source document and is not published.
