A large-scale study published in Brain and Behavior has found that some widely used medications may have subtle but meaningful associations with cognitive performance—both positive and negative—when considered across entire populations. By analyzing data from more than half a million individuals in the United Kingdom and validating the findings in two additional cohorts, researchers developed a “cognitive footprint” framework to estimate how different medications might influence mental performance on a broad scale.
The study was designed in response to growing awareness that many medications can affect thinking and memory, even if cognitive function is not the drug’s intended target. Although these effects are often small, the widespread use of certain medications means that even slight impairments or improvements could add up to significant effects at the societal level. This has implications not only for public health but also for how drug trials and prescriptions are evaluated.
“We have been exploring the concept of a cognitive footprint of individual-level and societal-level events and interventions. This is similar to the concept of a carbon footprint, which can be positive or negative and has a cumulative effect over time,” said study author Martin Rossor, a professor at the UCL Institute of Neurology.
“Drug side effects are very common—particularly cognitive ones—but often not sufficiently recognized, especially for medications that do not specifically target the central nervous system. Even small effects may have significant societal impact if a medication is taken for a long time or by many people. So this was an opportunity to look at the positive and negative cognitive footprints of medications.”
To explore this, the researchers drew on three long-running population studies: the UK Biobank, which includes over 500,000 participants aged 37 to 73; the EPIC Norfolk study, with over 8,000 participants who completed cognitive testing; and the Caerphilly Prospective Study, a smaller cohort of nearly 3,000 older men. Each of these studies collected detailed data on participants’ medication use, health history, and cognitive performance. However, the cognitive tests used differed across studies, requiring sophisticated statistical methods to align the results.
In the UK Biobank, researchers used a statistical approach called principal component analysis to create a summary measure of overall cognitive ability from various individual tests, including reaction time, fluid intelligence, and memory tasks. They then applied Bayesian regression models to estimate the relationship between regular use of different medications and cognitive performance, accounting for a wide range of other factors such as age, income, physical health, mood, and education.
They found that many medications had measurable associations with cognition, although the size of these effects was generally small. Among those most negatively associated were medications targeting the central nervous system, including some antiepileptics and antidepressants. For example, valproic acid, a commonly prescribed drug for seizure disorders, had one of the largest negative associations with cognitive speed. Amitriptyline, a tricyclic antidepressant, was also linked to reduced performance across several cognitive measures.
“Cognitive side effects of medications are frequent and often overlooked,” Rossor told PsyPost. “Interventional trials and post-market surveillance should include structured assessments of cognition routinely, whether or not the intervention is plausibly neurotropic.”
On the other hand, some medications showed positive associations. Non-steroidal anti-inflammatory drugs, such as ibuprofen and naproxen, were linked to better scores on various cognitive tasks. Glucosamine, a supplement often used for joint pain, consistently showed small positive associations across all outcomes. Omega-3 fatty acids, known for their potential brain benefits, were also positively associated with several cognitive measures, particularly in the EPIC Norfolk and Caerphilly cohorts.
Paracetamol, also known as acetaminophen, emerged as a drug of particular interest. It was one of the most frequently used medications and consistently showed small but negative associations with cognitive performance across the three cohorts. Because of its widespread use, this resulted in a relatively large “negative cognitive footprint” when scaled up to the population level. This means that, even though the individual impact may be minor, the overall effect across millions of users could be meaningful.
“While we anticipated negative effects, we were surprised that glucosamine and ibuprofen had a significant positive cognitive footprint,” Rossor said. “It’s also noteworthy that paracetamol and ibuprofen, which are often taken for similar purposes, had among the worst and best cognitive footprints observed in the study, respectively, across several cognitive domains.”
To estimate the broader implications, the researchers extrapolated their findings to the general UK population aged 40 to 70. They used standardized effect sizes (Z-scores) to compare the cognitive impact of medications with other known factors, such as air pollution and aging. For instance, the positive footprint of ibuprofen on overall cognitive ability was roughly equivalent to reducing the average age of the population by two months or lowering nitrogen dioxide pollution by one microgram per cubic meter.
The cognitive effects were most consistent for measures of processing speed and attention, suggesting that these functions may be especially sensitive to drug-related changes. However, not all cognitive domains were equally affected, and the impact of some medications varied across different studies. For example, glucosamine showed a positive effect in the UK Biobank and EPIC Norfolk cohorts but not in the older Caerphilly sample, where it was used less frequently.
The study’s authors caution that their findings are observational and do not prove that medications directly cause changes in cognitive performance. Despite using advanced statistical models to control for a wide range of potential confounding factors, such as pain or general health, it is still possible that unmeasured variables or reverse causation played a role. People who take certain medications may differ in important ways from those who do not, and some participants may have had early signs of cognitive decline before starting medication.
“The major caveat is that association does not prove causation,” Rossor noted. “The cross-sectional nature of the study—since there was not enough data for exploring longitudinal effects—means that the relationship between medication and cognition may be difficult to interpret.”
“While we tried to isolate the effects of medication by modeling as many other factors as possible—for example, the effects of pain on cognition in individuals taking paracetamol—these influences can never be entirely ruled out. We can only provide an upper bound of the potential effect at the population level of a medication’s cognitive footprint.”
The study introduces a useful new tool—the cognitive footprint—for evaluating the broader effects of medications on mental performance. By incorporating both effect size and population prevalence, this approach highlights the importance of considering cognition in drug development, prescribing practices, and public health policy. It also underscores the need for more comprehensive research on the unintended cognitive effects of medications, including over-the-counter drugs and supplements that are widely used but rarely studied in this context.
Future research may benefit from applying the cognitive footprint concept to other areas, such as the impact of environmental exposures or lifestyle interventions on mental functioning. “We would like to explore the cognitive footprint of a variety of interventions and effects—such as, at an individual level, the footprint of shift work, and at a population level, the effect of heat waves,” Rossor explained.
The study, “The Cognitive Footprint of Medication Use,” was authored by Marta Suárez Pinilla, Charlotte R. Stoner, Martin Knapp, Parashkev Nachev, and Martin Rossor.
