How were the lost years counted?
The study by Sarah Schrempft and colleagues sets two national ageing cohorts side by side: 8,511 people followed in England between 2012 and 2023, and 22,605 followed in Canada between 2012 and 2021, all aged 50 to 85 at baseline. What is measured is how a person gets through the day rather than a list of diagnoses: gait speed, grip strength, rising from a chair, lung function, hearing, eyesight, mobility, and basic activities such as dressing or washing. These are the measures a clinic uses to describe a patient's independence, and they forecast a great deal, from hospital admission to a move into residential care. The dividing variable is total household wealth rather than income.[1]
What makes the result legible is that the gap is converted into years. At age 60, the years of functioning lost on gait speed for those with the least wealth come to 15 in England (95 per cent confidence interval 12.2 to 17.8) and 9 in Canada (8.6 to 9.4). On basic daily activities the same calculation gives 20 years in England against 9 in Canada. This is what a doctor sees in the notes: of two patients born in the same year, one walks into the room with the gait of someone fifteen years older. In both countries less wealth also travels with a faster decline in mobility and daily activities as age advances.[1]
What remains when behaviour is taken out?
A gap like this is usually met with smoking, alcohol, weight, chronic illness and loneliness, on the assumption that most of it comes from there. The authors put all of them into the model. The gap weakened and survived across almost every measure. Care is needed about what that means, because the study is observational: there is no experiment showing that wealth causes the loss of function, and what is measured is co-occurrence. The authors also write that they could look only at conditions in adulthood and could not examine the contribution of access to private healthcare. Even so, the residue left after adjustment points to a load that individual behaviour cannot carry.[1]
I have argued here before that the referral criterion used for breast cancer treats one woman in seventy as at risk and leaves most cancers outside that group, and that the problem there was as much whom the criterion found as how accurate it was. The two ageing cohorts, ELSA and CLSA, show another face of the same problem. The wealth gap does not point the same way in women and men: those with the least wealth have more difficulty with mobility, daily activities and hearing than men in the same group, while their lung function is better. Reducing the average effect to a single number erases that distinction, and the patient who arrives at the clinic is no average.[1], [3]
The gain lands wherever the tax revenue goes
A second study published the same day takes up the question with a simulation rather than a measured outcome. Nina Escher and colleagues built a model of stunting in children and overweight in adults in Peru between 2000 and 2040, split across two socioeconomic groups. The result runs two ways. Raising taxes on unhealthy food and drink cuts the cumulative person-years lived with adult overweight between 2024 and 2040 by 58.6 million (95 per cent uncertainty interval 37.9 million to 77.7 million), but the gain gathers among the wealthiest. Directing the revenue from the same tax into raising the amount paid by an existing conditional cash transfer instead cuts person-years lived with child stunting by up to 876,000 by 2040, with 82 per cent of that reduction falling among the poorest.[2]
What holds the two studies together is a shared constraint: in both, the size of a health gain and its destination are separate questions, and the second stays invisible unless it is measured. In the ageing cohorts, the gap that remains after individual risk factors are accounted for says it will not close through behaviour alone; in the Peru model, the same fiscal instrument delivers wherever its revenue is sent. There is a plainer reading available: both findings could reflect wealth-related differences in measurement and in health service use, since wealthier people are both measured more often and treated earlier. That is why the interesting thing is whether the distribution itself gets reported as an endpoint. In most studies it currently does not.[1], [2]