When researchers asked more than 6,000 people over 50 how much medical risk they were willing to take, the answers did not cluster. They spread across the entire scale.
That finding, published July 29 in PLOS One by a team led by Nicholas Steel of the University of East Anglia, has a direct implication for anyone who has sat in an exam room while a clinician described a procedure. Recommendations built around what patients over a certain age typically want will fit some patients well and misfit a substantial number of others.
The practical stakes are ordinary and constant. Whether to accept a surgery with a meaningful complication rate, whether to add a medication that trades symptom relief for side effects, whether to pursue an aggressive option or a conservative one: these are the decisions where a mismatch between what a patient actually values and what a clinician assumes they value produces regret.
Six Questions Put to 6,098 People
The researchers added six new items to wave 8 of the English Longitudinal Study of Ageing, a nationally representative biennial survey of people aged 50 and over living in private households in England. The questions had never been asked in previous waves.
They covered risk aversion, orientation toward the future versus the present, quality of life versus length of life, body function versus appearance, openness to experimental treatments, and willingness to leave treatment decisions to a doctor. Each was scored from 0 to 10.
Of 7,133 core participants who completed the main interview, 6,257 returned the self-completion questionnaire, and 6,133 answered at least one preference item. The final weighted sample was 6,098. Mean age was 66, and 52 percent were women. Most participants, 72 percent, reported none of the five chronic conditions the study tracked.
Responses were distributed approximately evenly across the full range on four of the six questions: avoiding risks, living for the future, avoiding experimental treatments, and leaving decisions to a doctor. Only two items showed clear consensus. Participants leaned toward quality of life over length of life, and consistently toward preserving body function over appearance.
Sex and Age Moved the Needle Most
Within that spread, several patterns held up in regression analysis.
Women were more likely than men to want to avoid risks and to avoid experimental treatments, more likely to prioritize quality of life over length of life, and notably less likely to want to defer treatment decisions to a doctor. Men were more inclined to hand decisions to clinicians and less inclined to place quality of life ahead of longevity.
Participants aged 75 and over, compared with those aged 50 to 64, were more likely to want to avoid risks, avoid experimental treatments, prioritize body function over appearance, and leave treatment decisions to a doctor.
Education showed the single largest coefficient in the analysis. Lower educational attainment was associated with wanting to avoid risks, prioritizing length over quality of life, avoiding experimental treatments, and deferring to doctors. The association with delegating decisions was the strongest in the entire model, which raises a question about whether some patients defer because they prefer to or because the consultation does not give them a workable way to participate.
An earlier version of the analysis was posted as a preprint on medRxiv in January before peer review concluded. Correlations between the six measures were relatively low, and a principal component analysis found that the first component explained only about a quarter of total variance. Four components were needed to reach roughly three quarters. Preferences, in other words, are not one trait. Someone cautious about risk is not automatically someone who wants a doctor to decide.
Averages Hide the Individual in Front of You
The evidence check matters here as much as the findings.
This is a cross-sectional survey measuring stated preferences about hypothetical situations. The authors are explicit that these preferences were not tested against real healthcare decisions, and that other unmeasured dimensions may matter more when an actual choice arrives. Nothing in the study shows that people behave the way they answered.
The data are also not recent. They come from wave 8, collected across 2016 and 2017, which means the responses predate the pandemic and are close to a decade old. Preferences are known to shift with changes in health status and life circumstances, and the authors call for longitudinal work on exactly that question.
The cohort is English. Participants live in a single-payer system where cost at the point of care is not part of the calculation, which is a meaningful difference from the United States. American patients weighing a treatment are often also weighing a deductible, a prior authorization, or a network restriction, and those pressures could plausibly shift stated preferences in either direction. Findings should not be read as directly transferable to a U.S. population without testing.
A methodological caveat sits alongside that. A non-negligible share of respondents chose the scale midpoint on most items. The authors reran the analysis excluding those responses and found the pattern held, but midpoint clustering may reflect genuine ambivalence rather than a measured position.
The authors' own conclusion is the appropriate ceiling on how far to take this. The demographic patterns, they write, "cannot replace asking about individual preferences."
Bringing Preferences into a Short Appointment
For patients and caregivers, the usable takeaway is that a preference not stated is a preference not counted, and clinicians working inside a short appointment are unlikely to guess it.
People facing a treatment decision can say plainly whether they want a recommendation or a set of options with trade-offs laid out, and whether they would trade some expected length of life for better daily function. Those two statements do more work than most patients expect. Anyone who wants a family member involved can say so at the start rather than after a decision is made.
Caregivers supporting an older relative face a related trap. This research suggests that assuming a parent over 75 wants to be shielded from decisions will be right sometimes and wrong often, and the only way to find out is to ask directly rather than to infer from age.
Decision aids exist for many specific conditions, and shared decision making has been formal NICE guidance in England since 2021, though the authors note that comparatively few have been developed for older adults. Patients can ask whether one is available for their situation and can request it before, rather than during, the appointment where the decision gets made.
What happens next is straightforward. The authors call for testing these measures in other populations and against real clinical decisions, and for longitudinal work on how preferences change as health changes. Until that exists, the practical value of the finding is a prompt, not a shortcut.
Frequently Asked Questions
What did the study measure? Six general healthcare preferences among adults aged 50 and over: risk aversion, future orientation, quality versus length of life, body function versus appearance, openness to experimental treatments, and willingness to delegate decisions to a doctor.
How many people took part? The final weighted sample was 6,098 participants in the English Longitudinal Study of Ageing, with a mean age of 66.
What was the main finding? Preferences varied widely, with responses spread across the full scale on four of the six questions. Systematic patterns appeared by sex, age, and education.
How did women and men differ? Women were more risk-averse, more likely to prioritize quality of life, and less likely to want doctors making decisions for them. Men were more likely to delegate decisions.
Does this apply to American patients? Not directly. The cohort was English and the data were collected in 2016 and 2017 in a single-payer system, so cost pressures common in U.S. care were not part of the setting.
Does the study show how people actually behave? No. It measured stated preferences about hypothetical situations. The authors note these were not tested against real healthcare decisions.
What can patients do with this? State preferences explicitly rather than expecting a clinician to infer them, and ask whether a decision aid exists for the specific condition being discussed.
