No effect of education on Telomere Length: a natural experiment in aging individuals

Using a fuzzy regression discontinuity design on UK Biobank data, this study finds no causal evidence that an additional year of education affects telomere length decades later, challenging the prevailing view that education has long-lasting protective effects on biological aging.

The Gist

The Big Question: Can Schooling Slow Down Aging?

Imagine your body is a car. Every time you drive it (every time your cells divide), the tires wear down a little bit. In biology, these "tires" are called telomeres. They are the protective caps on the ends of your DNA. As you age, these caps get shorter. When they get too short, the car (your body) starts to break down, leading to aging and disease.

Scientists have long noticed a pattern: People with more education tend to have longer telomeres (better "tires") than people with less education. It looks like going to school acts like a premium car wax, protecting you from the wear and tear of life.

The Big Question: Does going to school actually cause your telomeres to stay longer? Or is it just that people who are naturally healthier, wealthier, or smarter tend to stay in school longer?

The Problem: The "Chicken and Egg" Dilemma

Usually, we can't answer this because we can't run a fair experiment. We can't randomly force some kids to stay in school for an extra year and send others home, just to see what happens to their DNA. That would be unethical.

Because we can't randomize it, we usually just look at the data and guess. But that's like looking at a photo of a car crash and trying to figure out who hit whom without knowing the speed or the road conditions. There are too many hidden factors (like money, stress, or genetics) messing up the picture.

The Solution: A "Natural Experiment"

The authors of this paper found a clever workaround. They used a real-life event that acted like a random coin flip.

In 1972, the UK government passed a law called ROSLA (Raising of the School Leaving Age). Before this law, kids could leave school at 15. After this law, they had to stay until 16.

Here is the magic part:

• If you were born in August 1957, you could leave school at 15.
• If you were born in September 1957 (just one month later), you were forced to stay until 16.

Because these two groups of people are almost identical in every way (same country, same era, similar backgrounds), the only real difference between them is that tiny bit of extra schooling. It's like a natural experiment where the universe randomly assigned the "extra year of school" treatment.

The Investigation: Checking the "Tires"

The researchers took a massive group of people from the UK Biobank (a giant database of health information). They looked at the "tires" (telomeres) of people born just before the cutoff (August) and just after the cutoff (September).

They used advanced math (called Regression Discontinuity) to compare these two groups. Think of it like a referee looking at two runners who started at the exact same line, but one was told to run an extra lap. Did that extra lap change how worn out their shoes were at the finish line?

The Result: The "No Effect" Surprise

The answer was surprising: No.

The extra year of school made zero difference to the length of their telomeres.

• The "August" kids and the "September" kids had the exact same wear and tear on their DNA decades later.
• The researchers checked this in many different ways (using different statistical tools, checking for errors, and even using Bayesian math) and the result was the same every time.

What Does This Mean?

This is a big deal for two reasons:

1. Correlation is not Causation: Just because educated people have longer telomeres doesn't mean school caused it. It's likely that educated people also have better diets, less stress, and better healthcare. The "schooling" wasn't the magic wax; it was just a signpost pointing to a healthier lifestyle.
2. A Warning for Science: For years, scientists have been trying to find "lifestyle hacks" to stop aging (like "just eat this" or "do this exercise"). This paper suggests we need to be very careful. Just because two things happen together doesn't mean one causes the other.

The Takeaway

Think of it like this: If you see a person with a shiny car and a full tank of gas, you might think the shiny paint caused the gas to be full. But in reality, the person probably just has a lot of money, which allowed them to buy both the paint and the gas.

This study shows that education is like the shiny paint. It looks great and is associated with good health, but forcing someone to stay in school for one extra year doesn't magically fix their biological "tires." To truly slow down aging, we might need to look deeper than just the classroom.

Technical Summary

1. Problem Statement

Telomere length (TL) is widely used as a biomarker for the "aging exposome," representing the cumulative impact of non-genetic environmental and behavioral exposures on aging rates. Observational studies consistently show a positive correlation between higher educational attainment and longer telomeres, suggesting that education may causally protect against biological aging.

However, establishing causality is difficult due to:

• Confounding: Education is correlated with socioeconomic status, genetics, and early-life environment.
• Gene-Environment Correlation: Genetic factors may influence both educational attainment and telomere length, masquerading as environmental effects.
• Lack of Randomization: Education cannot be ethically or practically randomized in humans.

The authors aim to resolve whether the observed association between education and telomere length is causal or merely correlational by leveraging a natural experiment.

2. Methodology

The study employs a Natural Experimental Design using a Fuzzy Regression Discontinuity (RD) approach.

• Natural Experiment (ROSLA): The authors utilize the 1972 "Raising of the School Leaving Age" (ROSLA) policy in England, Scotland, and Wales. This policy increased the mandatory school-leaving age from 15 to 16 for individuals born on or after September 1, 1957. This creates an arbitrary cutoff where individuals born just before the cutoff (August 1957) were unaffected, while those born just after (September 1957) were forced to stay in school for an additional year.
• Data Source: The analysis uses the UK Biobank, a large cohort study with leukocyte telomere length (LTL) measured in 474,074 participants. The final analytic sample included 256,243 individuals born between 1947 and 1967 (10 years on either side of the cutoff).
• Statistical Frameworks:
  1. Fuzzy Local-Randomization RD: Comparing individuals born in August 1957 (control) vs. September 1957 (treatment). This treats the cutoff as a randomized experiment within a narrow window.
  2. Bayesian Analysis: Using Markov Chain Monte Carlo (MCMC) sampling to calculate Bayes Factors (BF01) and Regions of Practical Equivalence (ROPE) to quantify evidence for the null hypothesis.
  3. Fuzzy Local-Linear RD: A continuity-based approach using the RDHonest package to handle discrete running variables (date of birth in months).
• Robustness Checks:
  • Varying the analysis window (1-month vs. 5-month windows around the cutoff).
  • Sensitivity testing with different Bayesian priors.
  • Removing outliers (>3 SD).
  • Controlling for visit date and clustering standard errors by UK Biobank site.
  • Placebo tests (checking if ROSLA affected visit dates, which it did not).

3. Key Contributions

• Causal Inference in Aging: This is one of the first studies to rigorously test the causal effect of education on telomere length using a natural experiment, moving beyond observational correlations.
• Methodological Rigor: The paper integrates frequentist (local-randomization) and Bayesian approaches, providing converging evidence. It addresses the "replication crisis" in social science by preregistering the design logic and conducting extensive robustness checks.
• Challenging the "Aging Exposome" Account: The study directly challenges the prevailing narrative that modifiable environmental factors like education have a direct, causal impact on cellular aging markers.

4. Results

• Observational Replication: Consistent with prior literature, the authors confirmed a significant associational link: more years of education correlated with longer telomeres, and older age correlated with shorter telomeres.
• Causal Null Effect:
  • Local-Randomization: No significant difference in telomere length was found between those forced to stay in school (Sept 1957) and those who were not (Aug 1957).
    • 1-month window: $p = 0.838$ (ITT), $p = 0.831$ (TSLS).
    • 5-month window: $p = 0.528$ (ITT), $p = 0.519$ (TSLS).
  • Bayesian Evidence: The Bayesian analysis provided "very strong" evidence in favor of the null hypothesis (no effect).
    • Bayes Factor ($BF_{01}$) = 43.48 (for the 1-month window), indicating the data is ~43 times more likely under the null hypothesis than the alternative.
    • ROPE Analysis: 100% of the 95% Highest Density Interval (HDI) fell within the Region of Practical Equivalence (0.1 SD), confirming the effect size is practically zero.
  • Robustness: The null result held across all variations, including wider time windows, different priors, outlier removal, and continuity-based linear RD models.

5. Significance and Implications

• Re-evaluating Causal Claims: The findings suggest that the strong observational link between education and telomere length is likely driven by confounding variables (e.g., genetics, early-life socioeconomic status) rather than a direct causal mechanism where schooling biologically slows aging.
• Caution for Interventions: The authors warn against interpreting observational telomere studies as evidence for intervention targets. If education does not causally lengthen telomeres, policies aimed solely at extending schooling may not yield the expected biological aging benefits.
• Contextualizing ROSLA: While previous studies using the same ROSLA dataset found causal effects of education on cognitive reserve, income, and mortality, this study demonstrates that these benefits do not necessarily translate to cellular aging markers like telomere length.
• Future Directions: The study highlights the need for more natural experiments and Mendelian randomization studies to disentangle environmental and genetic influences on biological aging, cautioning against the "aging exposome" narrative when causal evidence is lacking.

In conclusion, Judd and Kievit provide robust evidence that an additional year of mandatory education, while beneficial for many socioeconomic and cognitive outcomes, does not causally increase telomere length in aging individuals.