Diverging trends in health at older ages in England, 2004-2024: evidence from the English Longitudinal Study of Ageing

Technical Summary: Diverging trends in health at older ages in England, 2004–2024

Problem Statement
The paper addresses a critical uncertainty in population health research: whether health at similar chronological ages is improving or deteriorating over time. This question is central to policy planning regarding population ageing, working lives, and social care eligibility, as chronological age is often used as a proxy for need and capacity. While life expectancy gains have slowed in high-income countries like the UK, and healthy life expectancy has declined, existing literature offers mixed findings. Some studies report increases in cardiometabolic risks and diseases, while others suggest stability or improvement in functional limitations and biomarkers. Furthermore, evidence is often fragmented across single health domains, relies on broad age groupings, and lacks sufficient integration of self-reported versus biomarker-based measures. Consequently, it remains unclear whether recent cohorts experience uniform health improvements or if gains in some domains coexist with deterioration in others, particularly across different sociodemographic groups.

Methodology
The study utilizes data from the English Longitudinal Study of Ageing (ELSA), a nationally representative survey of community-dwelling adults aged 50 and older in England. The analysis is cross-sectional, comparing three waves of data collected approximately a decade apart: Wave 2 (2004/05, N=8,389), Wave 6 (2012/13, N=8,549), and Wave 11 (2023/24, N=6,090).

The study examines health trends across three broad domains:

1. Self-rated health, pain, and functional limitations: Including fair/poor self-rated health (SRH), limiting long-standing illness (LLI), pain intensity, activities of daily living (ADLs), instrumental activities of daily living (IADLs), and mobility tasks.
2. Cardiometabolic and chronic conditions: Comparing self-reported diagnoses (hypertension, high cholesterol, diabetes, etc.) with observer-measured biomarkers (blood pressure, total cholesterol, glycated haemoglobin for diabetes), alongside obesity (BMI) and inflammation (C-reactive protein).
3. Mental health and cognition: Including psychiatric problems, depressive symptoms (CES-D scale), quality of life (CASP-19 scale), and memory scores.

The analytical strategy employs weighted pooled modified Poisson and linear regressions to compare outcomes over time. Models are adjusted for sex and wealth quintiles and stratified by age groups (50–59, 60–69, 70–79, 80+) and education levels (low, medium, high) to assess social patterning. Robust standard errors account for repeated observations among participants.

Key Results
The analysis reveals divergent, rather than uniformly favourable, trends in health at older ages between 2004/05 and 2023/24:

• Divergence in Self-Report vs. Biomarkers: A stark contrast exists between self-reported and biomarker-defined cardiometabolic conditions. Self-reported high cholesterol increased from 19% to 39%, while biomarker-defined high cholesterol declined significantly from 78% to 54%. Similarly, adjusted estimates show self-reported hypertension increased from 35% to 40%, whereas biomarker-defined hypertension fell from 38% to 30%. However, diabetes increased on both measures.
• Functional and Subjective Health: Adjusted estimates reveal that the probability of fair/poor self-rated health increased from 27% to 34%, and the probability of moderate or severe pain increased from 27% to 33%. Conversely, mobility impairments declined overall (58% to 52%), while I/ADL limitations remained relatively stable.
• Mental Health and Cognition: Psychiatric problems increased (6% to 10%), and quality of life scores declined. In contrast, memory scores improved (+0.58 points), particularly among those aged 70 and older.
• Social and Age Heterogeneity: Aggregate trends masked significant inequalities.
  • Education: Improvements in mobility and declines in limiting long-standing illness were concentrated among those with higher education, whereas those with lower education experienced worsening trends in these areas. Self-reported diabetes increased significantly among low-educated adults but remained stable among the highly educated.
  • Age: Deterioration in self-rated health and quality of life was more pronounced in the 80+ and 50–69 age groups, respectively. Mobility improvements were observed in the 50–79 range but not among those 80+.

Key Contributions
The paper contributes to the literature by providing a multidomain assessment of health at comparable ages over two decades within a single national dataset. Its primary contributions include:

1. Integration of Measures: It demonstrates that conclusions about population health change depend heavily on the measurement method, as self-reported conditions and biomarkers can show opposing trends (e.g., rising awareness/treatment vs. declining physiological risk).
2. Multidomain Complexity: It establishes that health at older ages is not moving in a single direction; gains in functioning, biomarkers, and cognition coexist with rising pain, poorer mental health, and increasing diabetes prevalence.
3. Inequality Analysis: It highlights that aggregate population averages conceal persistent or widening health inequalities based on age and education, particularly regarding functional limitations and cardiometabolic risks.

Significance and Claims
The authors claim that health at comparable ages in England has changed in divergent and socially patterned ways, suggesting that chronological age alone is an increasingly incomplete proxy for need, capacity, and risk in later life.

The paper argues that relying on single health indicators or population averages can misrepresent the changing profile of health in ageing populations. Specifically, the findings suggest that:

• Policy responses regarding working lives and state support eligibility must account for the fact that the health profile associated with a given age is changing.
• The coexistence of memory improvements with worsening mental health and quality of life (especially in the 50–69 age group) complicates the narrative of "healthy ageing."
• Epidemiological monitoring and policy planning for ageing populations must adopt multidomain health profiles spanning physical functioning, biomarkers, mental health, and cognition, rather than relying on chronological age or single indicators.

The study acknowledges limitations, including the inability to isolate birth-cohort effects from period effects (such as the financial crisis or the pandemic), the exclusion of residential care populations, and potential biases in self-reporting and biomarker participation. Despite these limitations, the authors conclude that multidomain health monitoring is essential for interpreting population health trends and planning effective healthy ageing, prevention, and long-term care policies.