Article Text

Download PDFPDF
PP22 How do individual health trajectories drive population change? A mathematical decomposition of cohort ageing
  1. DOS Gillespie1,2,
  2. TN Coulson3,
  3. SD Tuljapurkar1
  1. 1Department of Biology, Stanford University, Stanford, USA
  2. 2Department of Public Health and Policy, University of Liverpool, Liverpool, UK
  3. 3Department of Zoology,University of Oxford, Oxford, UK


Background Public health policy should ideally be built from insights obtained from linking changes in individual-health to changes at the population-level. However, the appropriate methods are still lacking. We aimed to investigate how individual health trajectories can shape the cohort dynamics of ageing.

Methods We used a method that exactly decomposes changes in the population-level distribution of health into contributions from the constituent individual-level processes that generate them. We applied this to the over-50s in the Framingham Heart Study—a US cohort that began in 1948, and contains ˜10,000 individuals examined every two-years. We investigated how age-specific distributions of systolic blood pressure (SBP) and body mass index (BMI) changed due to mortality selection and within-individual change among survivors. Because most population-level change is driven by within-individual change, we explored how within-individual trajectories preceded death at different ages. We next investigated how anti-hypertension medication influenced within-individual trajectories, and how patterns have changed over time (1948–1982 vs. 1984–2002). We quantified uncertainty using bootstrap resampling.

Results In this (mainly Caucasian) sample, selection was most evident from around age 80, after which it increased SBP by 4.9 mmHg (95% CI: 2.2–7.6) and increased BMI by 0.19 kg/m2 (-0.09–0.46), i.e., tending to remove individuals with relatively low trait values. This apparently counter-intuitive result might be explained by our finding that death was preceded, by approximately ten years, by senescent declines in SBP and BMI. Individuals with earlier ages at death had higher SBP and BMI at younger ages but an earlier onset of senescence in these traits. Anti-hypertensives decreased the change in SBP between exams by on average 3.4 (0.3–6.5) mmHg until age 70, but by only 1.8 (-6.0–10.9) mmHg thereafter. The large reductions in SBP over time had apparently no effect on the trajectories of change or the nature of selection after age 50. However, there was an increase in BMI growth over time, by 2.7 kg/m2 (2.0–3.4) up to age 70, with signs that this might have shifted the risk of death towards relatively heavier individuals.

Conclusion Senescent declines preceding death are often neglected in public health but are an important consideration in addition to longer-term risk. Health dynamics at ages above around 70 were dominated by senescent decline, while below these ages medication and temporal change had most effect. This suggests that preventive measures in the elderly should also consider an individual’s recent trajectory of health-change as a key risk factor.

  • bio-markers
  • epidemiology
  • population change

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.