New research found that across 2.3 million older adults, one drug’s side effects were routinely being mistaken for a new disease and treated with another drug. This happened across 24 of the most common medication combinations.
The word polypharmacy describes what happens when a person takes multiple medications simultaneously. It is common in older adults, who are more likely to have multiple conditions that each require treatment. What polypharmacy statistics rarely capture is the specific mechanism by which the drug count climbs, how often the fourth prescription on the list exists because of the third one, or the sixth because of the fifth.
A study published in The BMJ has examined this question at population scale, tracking prescription patterns across 2,297,942 community-dwelling adults aged 66 and older in Ontario, Canada. The dataset covered complete provincial prescription records for 2022 and 2023, providing one of the most comprehensive real-world examinations of prescribing cascade patterns ever conducted.
A prescribing cascade happens when a medication’s side effect is misinterpreted as a new medical condition, prompting the prescription of a second drug to treat what is actually an adverse reaction to the first. The patient ends up on two medications when a more appropriate response would have been to reconsider or adjust the first one. The problem is not that the prescribing doctor made an error in isolation. It is that the side effect and the first drug are often managed by different clinicians, seen at different appointments, and documented in systems that do not automatically connect them.
The research team, led by Paula A. Rochon, director of research at the Weston and O’Born Centre for Mature Women’s Health at Sinai Health in Toronto and professor of medicine at the University of Toronto, started with 65 drug sequences that an international panel of 12 specialists had previously identified as potentially inappropriate prescribing cascades. Using the Ontario Drug Benefits database, which covers prescription claims for virtually all older adults in the province, they tested each sequence against three criteria.
The first drug had to be common, taken by at least 5% of the study population. The combination of first and second drug had to be observed in at least 1% of first-drug users within a year. And the sequence had to show a meaningful directional signal: the first drug had to precede the second significantly more often than the second preceded the first, measured by the adjusted sequence ratio.
Twenty-four of the 65 cascades met all three criteria. Together, they represent some of the most common medication patterns in older adult care, involving drugs prescribed for pain, cardiovascular disease, dementia, sleep, mood, and gastrointestinal problems.
The 24 chains and what they reveal
The strongest signal in the study, with an adjusted sequence ratio of 2.55, involved corticosteroids and antipsychotics. Corticosteroids, including prednisone and methylprednisolone, are widely prescribed anti-inflammatory drugs used across a range of conditions. Among their documented side effects are agitation, confusion, and mood changes, including in some cases steroid-induced psychosis or mania. When those symptoms appear, particularly in older adults whose baseline cognitive status may already be declining, they can be interpreted as a new psychiatric condition rather than as a drug effect.
An antipsychotic follows. The patient is now on a steroid for inflammation and an antipsychotic for psychiatric symptoms that the steroid caused. Antipsychotics carry significant risks in older adults, including increased risk of falls, stroke, and death in patients with dementia. The fact that steroid prescriptions were followed by antipsychotic prescriptions two and a half times more often than the reverse direction suggests this chain is not coincidental.
Laxatives and anti-diarrheal medications formed the second-strongest chain, with a ratio of 2.53. The mechanism is mechanically direct: laxatives, prescribed for constipation that may itself be caused by another medication such as an opioid, iron supplement, or calcium channel blocker, can produce loose stools or diarrhea. An anti-diarrheal drug follows. In some cases, the cascade extends further: the constipation came from iron supplements, the laxative treated it, the diarrhea came from the laxative, the anti-diarrheal treated that. Three drugs where adjusting the first might have resolved the problem.
Cholinesterase inhibitors, the standard medications for Alzheimer’s disease, appear in multiple chains. These drugs, including donepezil (Aricept), rivastigmine, and galantamine, are among the most widely prescribed medications for dementia. They work by inhibiting the breakdown of acetylcholine in the brain. Their side effects include nausea, vomiting, disturbed sleep, and vivid dreams. In 10.3% of patients who received a cholinesterase inhibitor, a sleep medication followed within a year. In others, anti-nausea medications followed, at a ratio of 2.24.
Antidepressants, specifically selective serotonin reuptake inhibitors (SSRIs) and serotonin-norepinephrine reuptake inhibitors (SNRIs), were linked to subsequent prescriptions for overactive bladder medications. These antidepressants affect bladder muscle tone as a side effect, producing urinary urgency or incontinence in some patients. The study found that antidepressant prescriptions were followed by bladder drug prescriptions at a statistically significant directional ratio, suggesting that some patients receiving bladder treatment are actually experiencing a side effect of their antidepressant.
Statins, one of the most widely prescribed drug classes in the world for cholesterol management, appeared in chains involving muscle relaxants. Statins can cause myopathy, muscle pain, and weakness as side effects. When those symptoms are treated as a new musculoskeletal condition rather than identified as a statin effect, a muscle relaxant or related pain medication follows.
Iron supplements, used to treat iron-deficiency anemia, consistently cause gastrointestinal side effects including constipation, nausea, and cramping. The study found clear sequential associations between iron supplement prescriptions and subsequent prescriptions for constipation treatment.
Why the problem persists despite being well documented
The prescribing cascade is not a new concept in pharmacology. It was named and described in clinical literature in 1995. The mechanism is understood. The solution, systematic medication reviews that consider all of a patient’s drugs as a connected system rather than individual prescriptions, is widely advocated. And yet the Ontario data shows the cascades occurring at large scale in 2022 and 2023, suggesting that understanding the problem has not reliably translated into preventing it.
Several structural features of healthcare delivery sustain the problem. Older adults with multiple conditions frequently see multiple specialists, each of whom is appropriately focused on their domain. A rheumatologist manages the steroid. A psychiatrist manages the antipsychotic. Neither may know the full medication list or have responsibility for reviewing it as a whole.
Even within a single primary care relationship, the connection between a side effect and its originating drug requires recognizing that a symptom that emerged weeks or months after a medication change is causally related to that change, not to the patient’s underlying conditions or to normal aging. The older the patient and the more complex their condition history, the harder that attribution becomes.
Electronic health records could theoretically flag these patterns, and automated alerts have been proposed as one solution. The study’s authors suggest that pharmacist-led medication reviews, which assess the full drug burden as an integrated system, offer one practical intervention at the clinical level. Population-level surveillance of the kind conducted in this study offers another lever, allowing health systems to identify the most common and harmful cascades and build targeted safeguards around them.
“These sequences of events are common but often missed in clinical practice,” lead researcher Rochon said. “The good news is that once identified, prescribing cascades can often be prevented or reversed through careful medication review and patient communication.”
What the patient perspective reveals
The prescribing cascade problem has a dimension that clinical data cannot fully capture: the patient experience of accumulating medications without knowing why.
Older adults and their families frequently report that medication lists grew over time without clear explanations of what each drug was for, whether it was still necessary, or whether any of the drugs were interacting with others. Medications added at one appointment for a symptom that appeared after another appointment are rarely connected retrospectively, either by the clinician or by the patient.
The result is that people take drugs they believe are treating diseases they have, when some of those drugs are actually treating side effects of other drugs, and the original condition causing the side effect remains untreated or inadequately managed.
For patients, the most actionable implication of the study is straightforward. At any medication review appointment, each drug on the list deserves two questions: what is this for, and when did I start taking it. For any drug where the answer to the first question involves a symptom, the second question that follows is whether that symptom appeared before or after starting another medication. That is not a diagnosis. It is a flag for a conversation that the data suggests is not currently happening often enough.
What the study does not establish
The study identifies population-level associations between sequential prescriptions. It documents that drug A is followed by drug B significantly more often than the reverse, which is consistent with A causing a side effect that gets treated with B. It does not prove, in any individual case, that the first drug caused the symptom that led to the second prescription.
Some patients in the dataset who received both drugs in a cascade sequence may have independently needed both. Some may have already been experiencing the eventual symptom before starting the first drug. The directional signal and the magnitude of the adjusted sequence ratios provide strong population-level evidence that the cascade is real and systematic, but they cannot adjudicate individual cases.
The study also focused on Ontario, a Canadian province with a specific healthcare system structure, prescribing culture, and population demographics. How well the 24 specific cascades generalize to other healthcare systems depends on how similar prescribing practices are across settings. The mechanisms are pharmacological and therefore likely universal, but the frequencies may vary.
The study, “Exploring high priority potentially inappropriate prescribing cascades in older adults: population level retrospective cohort study”, was authored by Mina Tadrous, Tara Gomes, and colleagues led by Paula A. Rochon at Sinai Health, the University of Toronto, and ICES, and published September 10, 2026 in The BMJ.
Source: Sinai Health / University of Toronto / ICES. DOI: 10.1136/bmj-2026-100499
