The income, mental health and physical health triangle, unpacked

Income is important for our health, both physical and mental. This is now well established in the peer-reviewed literature, official reports and large datasets (Reed et al, 2025). Broadly speaking, the higher your income, the healthier you are likely to be (Reed et al, 2025). However, we also know that mental illness is associated with a shorter life, largely through common physical conditions (Firth et al, 2019).

Put these elements together and a triangle emerges. Income, mental health and physical health each influence one another: low income can worsen mental and physical health; poor mental and physical health compound each other; and ill health can, in turn, erode a person’s capacity to earn. The arrows run in both directions. The Mental Elf has covered sides of this triangle before, including the reciprocal relationship between low income and mental health (Bell, 2020a; Guo & Higson-Sweeney, 2025).

Less well understood is what happens at one corner when a person sits at two others at once. When someone has both a mental illness and a low income, how do those two risks combine to shape their physical health? They might act independently, each adding to the risk; or they might interact, so the combined risk is greater than the sum of its parts. The distinction has real consequences: independent risks call for parallel responses addressing each in turn, while interacting risks call for a single, integrated response aimed at the overlap.

A new nationwide Danish study (Ejlskov et al, 2026) offers the most comprehensive answer to date. Drawing on national registers covering 5.3 million people across two decades, the authors examined whether low income and mental illness interact, or simply accumulate, in the development of later physical health conditions.

Methods

This was a whole-population register study, where the authors followed everyone living in Denmark from 1 January 2000 through to the end of 2021. This meant that the authors followed 5,279,634 people across 98.7 million person-years of follow-up.

They assessed:

• Mental health conditions, taken from hospital psychiatric records (inpatient since 1969; outpatient and emergency since 1995), sorted into 11 groups.
• Income, based on the household’s disposable income (what’s left after tax and benefits), adjusted for the number of people it has to support, averaged over 1997–99, and split into five equal bands (quintiles) specific to each age and sex group. A very comprehensive way to assess income!
• Physical health, from hospital, prescription and death registers, included 31 distinct conditions in nine groups.

The authors then compared how often physical conditions occurred in different groups. They reported this in two ways: the relative scale (rate ratios which are how many times more likely a condition was) and the absolute scale (rate differences, which are how many extra cases occurred per 100,000 people each year). This is important because a risk can look the same in relative terms while producing very different numbers of actual cases. The analysis plan was pre-registered, and the code is openly available which enhances the rigour of this study.

Results

The paper has three central findings:

1. Mental illness raised the risk of almost every physical condition, regardless of income level. People with a prior mental health condition had higher rates of physical conditions ranging from COPD and heart disease to liver disease and chronic pain, and this held whether they had a high or low income.
2. On the relative scale, poverty and mental illness did not multiply. The income gradient, which is the way risk climbs as income falls, was essentially parallel for people with and without a mental health condition across most physical health condition groups, meaning that the lines run alongside each other rather than fanning apart. Simply put, being on a lower income and mentally unwell was not statistically worse than the sum of the two.
3. However, on the absolute scale the burden was significantly higher for those in the lowest income group with a mental health condition. Because the starting risk is already higher among people on lower incomes, the same relative multiplier produces very different numbers of people with a physical health condition. For COPD, the authors counted about 716 extra cases per 100,000 person-years in the lowest income quintile among people with a mental health condition, against about 337 in the highest. For chronic pain, around 1,529 versus 1,345 extra cases; for chronic liver disease, 243 versus 110. What this means is that, in this sample, the additional burden falls disproportionately on people with the fewest resources to absorb it.

There were some exceptions though. Developmental conditions ran the opposite way, with more physical illness at higher incomes. Substance use and schizophrenia-spectrum disorders showed flatter gradients. And there was a revealing mirror-image: dyslipidaemia and high blood pressure (conditions you only know about if someone checks) showed bigger excesses among people on higher incomes, while their downstream consequences like stroke and heart failure showed bigger excesses among people on lower incomes. As the authors themselves concede, some of this map may be tracking who gets diagnosed rather than who actually gets ill.

Conclusions

The authors read their results as evidence that mental illness and low income work through separate, parallel pathways rather than amplifying one another, that is, adding up, not multiplying, and that this concentrates physical disease in people who carry both disadvantages at once.

Their headline message states that tackling one without the other will not close the gap. As they put it, neither mental-health treatment nor income-based measures alone will eliminate these disparities. Integrated, joined-up care that treats the mind, the body and the bank balance together is, based on this evidence, the only thing likely to shift the burden.

Strengths and limitations

This study has many strengths. The authors have analysed an enormous, whole-population sample with no recruitment or recall bias, up to 22 years of follow-up, and unusually careful handling of income (age- and sex-specific, adjusted for life stage). It is pre-registered, follows STROBE, shares its code, reports both scales, and holds up under sensitivity analyses. As a piece of register epidemiology, it is hard to fault.

The deepest problem is the exposure itself. ‘Mental health condition’ here means hospital-diagnosed mental disorder and does not include the much larger group of people living with mental health problems managed in primary care, privately, or not at all. So, this is a study of more severe, hospital-visible mental illness. Worse, who ends up in those registers is itself shaped by income. People on higher incomes may be diagnosed privately (and so miscounted as having no condition), while those on lower incomes are more likely to be represented in hospital data. The income exposure is distorting how the mental health condition is measured.

The physical-health outcomes are also coarse. All physical health conditions were diagnosed through either first diagnosis, first repeat prescription or death, with no severity or staging. Further, for long-diagnosed people, the baseline income measured in 1997–99 is already downstream of their illness (social drift), so income is not a cleanly upstream cause.

Implications for practice

For frontline services, the message is uncomfortable but actionable: physical health cannot stay a footnote in mental health care. Patients on lower incomes with mental illness are presenting with more advanced diseases, like stroke and heart failure, which could be prevented if detected earlier. That argues for proactive, lowered-threshold physical screening built into mental health contacts, and for genuine outreach rather than letters that assume a stable address and an easy relationship with services.

For policy, the study is a quiet but firm argument against silos. Treating mental health and poverty through separate budgets, separate departments and separate targets all but guarantees that disease keeps concentrating where the two overlap. The authors frame this in terms of the UN’s goal of reducing inequality (Sustainable Development Goal 10); closer to home it lands squarely on debates about integrated care, social prescribing and the adequacy of working-age benefits (Bell, 2020b).

On generalisability, and why this study should matter to a reader in London or Melbourne. This is Denmark: universal healthcare, comparatively compressed incomes, low out-of-pocket barriers. If physical illness still concentrates this sharply there, the gradient is very likely steeper in systems with more cost barriers and wider inequality, such as the UK or Australia. Danish registers are not a foreign curiosity here; they are close to a best-case scenario, and the picture is still stark.

For research, the obvious next steps are the ones the design could not reach: multimorbidity (the authors modelled condition pairs, not the messy clusters real patients carry), other dimensions of disadvantage beyond income (education, wealth, neighbourhood), and crucially the full clinical spectrum, including the milder, primary-care-managed conditions this study cannot see.

To return to where we began: the finding that sounds reassuring, ‘no synergy, just addition’, is, on a second look, the more sobering one. Risks that merely add together still accumulate; yet we have built systems designed to address only one of them at a time. The work many of us do upstream, on the conditions that drive poverty, isolation and ill health in the first place, is not a soft adjunct to clinical care here. On this evidence, it is half of the only intervention likely to work.

Statement of interests

Drew Meehan used AI, specifically Anthropic’s Claude (Opus 4.8), to improve the clarity and readability of this blog. All work is the author’s own, and the author takes full responsibility for the content of this blog.

Editor

Edited by Laura Hemming.

Links

Primary paper

Linda Ejlskov, Tomáš Formánek, Natalie C Momen, Danni Chen, Uffe Heide-Jørgensen, Oleguer Plana-Ripoll (2026) The interaction between income and mental health conditions for subsequent physical health conditions: a nationwide Danish cohort study from 2000 to 2021. The Lancet Public Health 11(5): e306–e317.

Other references

Andy Bell. (2020a) Poverty causes mental illness (and vice versa): how can we end this vicious cycle? The Mental Elf.

Andy Bell. (2020b) Social security? Evidence about benefits and mental health. The Mental Elf.

Guo, X. & Higson-Sweeney N. (2025) Can we reduce the global mental health burden by targeting social determinants? The Mental Elf.

Firth J, Siddiqi N, Koyanagi A, et al. (2019) The Lancet Psychiatry Commission: a blueprint for protecting physical health in people with mental illness. Lancet Psychiatry 6: 675–712.

Reed, H. R., Nettle, D, Parra-Mujica, F, Stark, G, Wilkinson, R, Johnson, M. T, & Johnson, E. A (2025). Examining the relationship between income and both mental and physical health among adults in the UK: Analysis of 12 waves (2009-2022) of Understanding Society. PloS one, 20(3), e0316792.

Photo credits

• Photo by micheile henderson on Unsplash
• Photo by Shubham Dhage on Unsplash
• Photo by Gurth Bramall on Unsplash
• Photo by Tim Mossholder on Unsplash
• Photo by Cory Mogk on Unsplash
• Photo by Bekky Bekks on Unsplash