HEALTH
The ankles that swelled, and the water pill that followed
Picture a woman of 78. Her blood pressure creeps up at a routine check, so her doctor starts her on a calcium channel blocker. Sensible. Evidence-based. Exactly what the guidelines say to do.
Six weeks later her ankles are puffy by teatime and by evening her shoes will not go on. She mentions it at her next appointment. Her doctor — reasonably, kindly, in a ten-minute slot — hears "swollen legs" and thinks fluid. She goes home with a water pill.
The water pill works, after a fashion. It also sends her to the lavatory three times a night, which wrecks her sleep, which leaves her light-headed and unsteady at six in the morning on a cold bathroom floor. Which is roughly where and when a great many hip fractures in people her age begin.
Nobody in that story did anything wrong. Nobody was lazy or negligent. And yet the second drug was never treating a disease at all. It was treating the first drug.
This is what doctors call a prescribing cascade. And as of this week it has stopped being a textbook curiosity and become a measured, named, counted feature of modern medicine.
An entire province, put under the microscope
In a study published in The BMJ on 10 September 2026, a team led by Dr Paula Rochon of Sinai Health in Toronto did something nobody had done at this scale: they took the prescription records of an entire Canadian province and went hunting for the fingerprints of drugs treating drugs.
The findings, which landed in circulation across the weekend, identify 24 specific medication chains that are both common enough to matter at a population level and plausible enough to cause real harm. The drugs involved are not exotic. They are the ordinary furniture of an older person's medicine cabinet: statins for cholesterol, iron supplements for anaemia, anti-inflammatories for an aching knee.
Rochon is not a fringe voice. She is Director of Research at the Weston and O'Born Centre for Mature Women's Health at Sinai Health, holds the Barry J. Goldlist Chair in Aging and Health, and is a professor of medicine at the University of Toronto's Temerty Faculty of Medicine. She has spent a career on the unglamorous question of what happens when medicine simply keeps adding.
"These sequences of events are common but often missed in clinical practice," Rochon said of the findings. "Knowing what medications you are taking, when they were started and for what indication is important in order to identify possible prescribing cascades that may be problematic."
Sixty-five suspects. Twenty-four convictions.
The method was deliberately conservative, and that is what gives the result its teeth.
In earlier work, the team assembled a panel of 12 international specialists in internal medicine, geriatric medicine and clinical pharmacology and asked them to name every drug pairing they believed could constitute a potentially inappropriate prescribing cascade — the mouthful the field abbreviates to PIPC. The panel produced a list of 65 suspects.
The new analysis then dragged all 65 through population-level prescription data held by ICES, Ontario's health data institute, working with analysts Lavina Matai and Zhiyin Li. Each candidate was judged against three tests: how commonly the first drug was prescribed across the population, how often a second drug followed it, and how strong the statistical link between the two actually was.
Only 24 survived all three. Those are the cascades that are simultaneously frequent, traceable and capable of hurting people — the ones, in other words, that are worth a health system's attention.
This was not a Toronto-only affair. The collaboration drew in researchers from the United States, Belgium, Italy, Israel and Ireland, alongside Sinai Health's Dr Vasily Giannakeas, Dr Nathan Stall and geriatrician Dr Christina Reppas-Rindlisbacher, with research staff Wei Wu and Joyce Li. The work was funded by the Canadian Institutes of Health Research.
The painkiller trap
The example the researchers pushed to the front is the one most likely to be sitting in your own bathroom cupboard right now.
Non-steroidal anti-inflammatory drugs — the ibuprofen-and-relatives family, handed out for everything from arthritic knees to back spasm — are known to nudge blood pressure upwards. It is not a rare or obscure effect. It is a documented, predictable property of the drug class.
But blood pressure is also the single most monitored number in the whole of primary care. So when a patient on an anti-inflammatory turns up with a reading that has drifted north, the system does what it is designed to do: it treats the number. A new antihypertensive is added. The knee tablet is never questioned.
The patient is now on two drugs where the honest answer might have been a different painkiller, a lower dose, or physiotherapy. And each new tablet brings its own catalogue of effects, any one of which can look like the beginning of a third condition.
Older adults are especially exposed to this pattern, the researchers note, because they are more likely to be taking several medicines at once for several conditions at once. Once a person is on six or eight drugs, the arithmetic of attribution becomes brutal. A new symptom could plausibly belong to any of them, to the interaction of two of them, or to age itself. Working out which is genuinely hard — and it is harder still in an appointment that has to cover four other things.
The chains medicine already half-knew about
The Ontario work does not arrive from nowhere. Over the past few years a small literature has been quietly documenting individual cascades, one pairing at a time, and the pattern it describes is unnervingly tidy.
The calcium channel blocker followed by a diuretic — the swollen-ankles story at the top of this article — has been repeatedly documented in older adults with high blood pressure, including in earlier work by Rochon's own group examining its links to adverse events. A cholinesterase inhibitor for dementia, which commonly causes urinary urgency, followed by a bladder anticholinergic to calm the bladder down, is one of the most frequently reported chains in the field — and is particularly uncomfortable because the second drug works against the brain in exactly the way the first drug is trying to help it. Thiazide diuretics for blood pressure, which push uric acid up, followed by gout medication in people who had never had gout, have been tracked in hypertensive adults starting first-line treatment.
What the new study adds is scale and ranking. Instead of a scattering of individual papers, health systems now have a shortlist — a priority order of which chains to go after first.
Why this lands hardest on women
The researchers are explicit that the findings carry a particular weight for older women, and the reasoning is arithmetic rather than ideological.
Women live longer. Over the course of those longer lives they accumulate more chronic conditions than men. More conditions means more prescriptions; more prescriptions mean more adverse drug events; more adverse drug events mean more opportunities for one of those events to be read as a fresh diagnosis rather than traced back to its source.
Every additional tablet is another door through which a cascade can enter. Women, statistically, have more doors.
What the study does not say
Tabloid instinct says shout "YOUR DOCTOR IS POISONING YOU". Tabloid instinct, in this case, would be wrong, and it is worth being precise about why.
This is a retrospective cohort study built on dispensing records. It can show that drug B reliably follows drug A across a population. It cannot climb inside any individual consultation and prove that the prescriber mistook a side effect for a new disease. In some of these sequences the second drug will have been entirely appropriate — a patient can develop genuine hypertension while happening to take ibuprofen, and genuine gout while happening to take a thiazide.
That is precisely why the researchers use the word "potentially". A PIPC is a pattern that deserves a second look, not a verdict. The number that matters is not how many of these prescriptions were wrong — the study cannot know that — but how many were never examined at all.
There is also a boring but important caveat about geography. This is Ontario data, reflecting Ontario prescribing habits, Ontario formularies and Ontario's drug coverage rules for older adults. The chains are likely to generalise. The exact frequencies may not.
The two fixes on the table
The team's proposed remedies are notably unglamorous, which is usually a sign that somebody has thought about implementation.
The first is software. Automated clinical decision support — the prompts already built into most modern prescribing systems — could be taught to recognise a cascade as it forms. When a clinician reaches for a drug that is a known treatment for a known side effect of a drug the patient already takes, the system flags it before the prescription is issued, and asks the question a busy clinic day tends to swallow: is this new symptom actually a new disease?
The second is people, specifically pharmacists. The researchers argue for pulling pharmacists further into prescribing decisions alongside physicians rather than leaving them at the dispensing end of the pipeline. A pharmacist sees the whole list, sees the start dates, and has training aimed squarely at drug effects. It is, on the face of it, a colossal amount of underused expertise standing behind a counter.
What this means for you
If you or someone you care for is taking multiple medicines, the single most useful thing you can do costs nothing and takes twenty minutes at the kitchen table.
Write the list. Not just the drug names — every list already has those — but three columns: what it is, when it was started, and what it was started for. That third column is the one almost nobody can fill in, and it is the one that makes cascades visible. A drug whose reason nobody can remember is a drug worth asking about.
Then look at the timeline. If a new symptom appeared within days or weeks of a new tablet, say so out loud at the next appointment, in those words: "this started after I began that." Clinicians are trained to pattern-match, but they can only match patterns they are shown, and the sequence is frequently invisible from inside a single consultation — especially when the two prescriptions came from two different doctors.
Ask directly whether a new prescription is treating a side effect of an old one. It is not a rude question. It is, on the evidence of this study, one of the most clinically useful questions a patient can ask.
And then — this part is not optional — do not stop anything on your own. Several of the drugs on these lists are genuinely keeping people alive, and abrupt withdrawal of blood pressure medication, heart medication or anything affecting the brain can be considerably more dangerous than the cascade itself. The action here is a conversation, not a bin.
The gap that opens one prescription at a time
What makes this study quietly unsettling is that it does not describe a scandal. There is no villain in it. There is no negligent doctor, no reckless drug company, no suppressed safety signal. Every individual decision in a prescribing cascade looks defensible when you examine it on its own.
The harm is emergent. It lives in the joins between appointments, between specialists, between the moment a drug was started and the moment, years later, when somebody finally asks why.
"Our concern is that so often these conversations between the health care prescriber and the patient are being missed," Rochon said, "so people don't recognise the sequences of events and that they are connected to one another."
Twenty-four chains. One province. A pattern that almost certainly repeats in every ageing health system on earth. The next time somebody hands you a prescription, it is worth knowing there is now a peer-reviewed reason to ask the most obvious question in medicine — and the one asked least: what is this actually for?
