Doctors often avoid blood pressure pills in frail older people for fear of falls. A study of 5.5 million records weighed both risks, and one came out far ahead
Blood pressure pills are among the most prescribed drugs in the world, and for older people they come with a well-known worry: they can make you dizzy and cause a fall. Some doctors stop or avoid them in frail patients for that reason. Researchers at the University of Oxford used the health records of more than 5.5 million patients in England to weigh, person by person, how many heart attacks and strokes the drugs prevent against how many serious falls they cause.
Across all patients, the balance came out clearly in favor of treatment. For every 1,000 people treated for 10 years, about 8 serious cardiovascular events such as heart attacks and strokes were prevented, while about 2 serious falls were caused.
The surprise was in the group doctors worry about most. Among men in their 80s with moderate frailty, treating 1,000 people for 10 years prevented an estimated 47 serious cardiovascular events and caused 9 serious falls.
For fit men in their 60s with no history of heart disease, the numbers were 12 events prevented and 1 fall caused. The older and frailer group had more falls, but the benefit grew even faster, because their risk of heart attack and stroke was so much higher to begin with.
“Doctors need to consider not just whether a treatment works, but how much an individual patient is likely to benefit,” said senior author James Sheppard, a professor at Oxford’s Nuffield Department of Primary Care Health Sciences.
The study, published in Nature Communications, combined a new risk calculator called STRATIFY-CVD with existing models of side-effect risk and evidence from clinical trials of blood pressure drugs. The calculator estimates a person’s chance of a heart attack or stroke over 5 to 10 years from their age, medical history, existing conditions and medications. It also accounts for the chance of dying from something else first, which matters a lot in very old patients.
The calculator predicted risk accurately when tested on a separate set of patients, and it generally gave lower estimates than QRISK2, one of the risk tools UK doctors already use.
Not every side effect came out the same way. For fainting and acute kidney injury, benefits and harms were more finely balanced. For low blood pressure and high potassium levels, the models suggested treatment could cause more of these problems than the number of heart attacks and strokes it prevented. Those side effects are generally less serious than a heart attack or stroke, and doctors can often catch them with blood tests and dose changes.
“Our research is an important step towards giving doctors and patients a clearer, more personalised picture,” Sheppard said.
It is the second recent study to push back on stopping heart medications by default in older age. A trial of 1,880 people over 70 found that stopping statins looked safe at two years, but heart attacks, strokes and deaths crept up after that. Diet matters too, and a study of 20,940 women found saturated fat was linked to more heart attacks and strokes mainly in one genetic group.
For older adults and their families, the practical point is not to stop blood pressure medication because of fear of falling without talking to a doctor. If dizziness or falls are a problem, ask whether the dose, timing or type of drug can be adjusted, and whether blood pressure should be checked standing up as well as sitting.
There are limits. This is a modelling study, so its estimates depend on the side-effect models and trial results it was built on. The press release gives only a few worked examples rather than results for every age and frailty level, and the data come from patients in England. One author sits on UK guideline committees for blood pressure, and the other authors declared no relevant conflicts.
The work was funded by the Wellcome Trust and Royal Society, and the National Institute for Health and Care Research.
Source
Sheppard JP, et al. “Benefits and harms of antihypertensive treatment: a modelling study in 5.5 million electronic health records.” Nature Communications, 2026.
DOI: 10.1038/s41467-026-79283-1