(Credit: Kmpzzz on Shutterstock)
Scared of Statin Side Effects? A Calculator Built From 5 Million Patient Records Has Good News for Most People
In A Nutshell
- Researchers in England built a prediction tool using data from more than 5.6 million people to calculate an individual’s personal risk of a serious muscle disorder linked to statin use.
- For 99.6% of statin-eligible people studied, the predicted 10-year risk of a serious muscle disorder was below 10%, and the actual observed rate was well under 1%.
- A prior history of muscle problems was the single strongest risk factor, far outweighing statin use itself.
- More than 62% of people who qualified for statins weren’t taking them, despite the data showing their muscle disorder risk was very low.
Many people avoid or stop taking statins because they worry about muscle damage. Now, a large study conducted in England has built a prediction tool that calculates a person’s individual risk of a serious muscle disorder, and for the vast majority of people, that number turns out to be very low.
Fear of muscle damage is one of the most cited reasons people refuse or abandon statin treatment, drugs that are among the most prescribed in the world for preventing heart attacks and strokes. A new prediction model, published in The Lancet Digital Health, was built using the medical records of millions of patients in England, and it could reshape how doctors and patients have that conversation, by putting a precise, personal number on a risk that most people have only ever guessed at.
Not all muscle trouble is created equal. Mild aches and soreness are generally self-limiting and, according to the paper, most are not attributable to statins at all. The serious version, muscle breakdown severe enough to require hospitalization or cause death, is a different matter. Statin use was associated with a higher risk of that outcome in the model, though the absolute risk remained low for most people.
Built From Millions of Patient Records
The research team drew on electronic health records from a large primary care database in England, analyzing records from more than 1.78 million individuals during the model-building phase. They then pressure-tested the model on a separate group of nearly 3.9 million people, a validation step using data it had never seen before.
Both groups included men aged 50 and older and women aged 60 and older, age ranges that roughly correspond to the population doctors typically consider for statin prescriptions based on cardiovascular risk. People were followed for up to 10 years, and the study tracked only serious muscle disorders resulting in a hospital admission or death, deliberately excluding everyday muscle aches and pains.
Serious events were rare across both groups: about 0.30% in the model-building group and 0.35% in the validation group. For context, deaths from unrelated causes were far more common, affecting roughly 24% and 18% of each group respectively over the same period. Because dying from something else before developing a muscle disorder can skew the numbers, the model was specifically designed to account for that possibility, keeping the risk estimates grounded in reality.
A 22-Factor Formula Built From Routine Medical Records
The final model incorporated 22 factors from routine patient records: age, sex, body weight, smoking history, conditions such as kidney disease or an underactive thyroid, prior muscle problems, vitamin D deficiency, and whether the patient was taking statins and which type. The tool estimates a person’s risk of serious muscle disorders while accounting for statin use alongside all these other factors.
Statin use emerged as a meaningful predictor, with different statin types carrying different levels of associated risk. Rosuvastatin showed the strongest association, followed by atorvastatin and simvastatin. But the single strongest predictor of all was a history of previous muscle problems, which was associated with a dramatically higher likelihood of a serious event. Vitamin D deficiency and use of other drugs known to stress muscles were also significant warning signs.
When tested on the external validation group, the model performed well, correctly distinguishing higher-risk individuals from lower-risk ones about 78% of the time over a 10-year window. For shorter timeframes of one and five years, accuracy was even better.
Most Statin Candidates Face Very Low Risk, Yet Many Go Untreated
What the numbers actually show may surprise a lot of people. In the validation group, 99.6% of individuals had a predicted 10-year risk of serious muscle disorders below 10%, and the actual observed 10-year incidence was well under 1%. Among those with confirmed eligibility for statin treatment, 98.1% fell below that same threshold.
Yet more than half of statin-eligible people in the study were not taking them. Among those with confirmed eligibility, 62.5% remained untreated, despite some carrying very high cardiovascular risk. The researchers suggest their tool could help address this gap by showing hesitant patients and their doctors that the personal odds of serious muscle harm are, in most cases, far lower than feared. An online risk calculator based on the model has been made freely available, and the authors suggest it could be paired with existing cardiovascular risk tools to give both sides of the equation in a single clinical conversation.
Like any tool built from routine medical records, it has gaps: genetic predispositions and physical activity levels aren’t captured, and its performance outside England still needs to be tested. The authors say real-world impact studies are needed before it becomes standard practice.
Fear is keeping many people away from drugs that could prevent a heart attack or stroke. A tool that puts an honest, personalized number on one of medicine’s most misunderstood risks could give both patients and their doctors a real starting point for that conversation.
Disclaimer: This article is not intended to serve as medical advice. Always consult a qualified healthcare provider before making any decisions about medication, including statins.
Paper Notes
Limitations
The authors acknowledge several limitations. Some important risk factors for muscle disorders, including genetic predisposition and physical activity levels, are not routinely recorded in UK primary care data and therefore could not be included in the model, which likely reduces how well it identifies all high-risk individuals. The model may overestimate risk for a very small proportion of people at the extreme high end of predicted risk, though the authors note that even the observed risk for this group is high enough to warrant clinical caution regardless. Because cardiovascular risk scores were not consistently recorded in the data, the researchers used age thresholds as a proxy for statin eligibility, which may have excluded some younger individuals with elevated cardiovascular risk due to factors other than age. The study cohorts were drawn from primary care records in England, and the generalizability of the model to secondary care patients or populations in other countries has not yet been established. Common limitations of using routine healthcare records, including missing data and the potential for misclassification of diagnoses, also apply, though the researchers used established statistical methods to address missing data and the model showed good overall performance.
Funding & Disclosures
This study was funded by a British Heart Foundation PhD Scholarship awarded to lead author Ting Cai. James P. Sheppard and Constantinos Koshiaris were funded in whole or in part by the Wellcome Trust and the Royal Society via a Sir Henry Dale fellowship, and by a National Institute for Health and Care Research (NIHR) School for Primary Care Research grant. Richard J. McManus was supported by an NIHR senior investigator award. F. D. Richard Hobbs was partially supported by the NIHR Applied Research Collaboration Oxford and Thames Valley. The authors declared no competing interests. The funders had no role in study design, data collection, analysis, interpretation, or writing.
Publication Details
Authors: Ting Cai, Jennifer A. Hirst, Brian D. Nicholson, Richard J. McManus, F. D. Richard Hobbs, James P. Sheppard, Constantinos Koshiaris, all affiliated with the Nuffield Department of Primary Care Health Sciences, University of Oxford, Oxford, UK. Richard J. McManus also holds an affiliation with Brighton and Sussex Medical School, University of Brighton and University of Sussex, Brighton, UK. Constantinos Koshiaris also holds an affiliation with the Department of Primary Care and Population Health, University of Nicosia Medical School, Nicosia, Cyprus. | Journal: The Lancet Digital Health | Paper Title: “Predicting the risk of serious muscle disorders in individuals eligible for statin treatment in England: derivation and validation of a clinical prediction model” | DOI: https://doi.org/10.1016/j.landig.2026.101024 | Published: Online ahead of print, 2026
