The causal relationships between leisure-time physical activity and body mass index in adulthood: A triangulation study

Using a triangulation framework with longitudinal twin data and three complementary analytical methods, this study provides robust evidence that higher body mass index causally reduces leisure-time physical activity in adulthood, particularly with advancing age, while offering less convincing support for the reverse causal effect.

The Gist

Imagine you are trying to figure out a classic "chicken or egg" mystery, but instead of food, the mystery is between how much you move (Leisure-Time Physical Activity) and how heavy you are (Body Mass Index or BMI).

Does being heavy make you move less? Or does moving less make you heavy?

For years, scientists have been arguing about this. Some say, "If you exercise, you'll lose weight!" Others say, "If you're heavy, it's harder to exercise, so you stop." This new study tries to solve the mystery by using a special detective toolkit called "Triangulation."

Think of triangulation like trying to find a lost treasure. If you only have one map, you might be wrong. But if you have three different maps (three different scientific methods) that all point to the same spot, you can be pretty sure you've found the treasure.

Here is how the researchers used their three "maps" to solve the case:

The Three Detective Tools

1. The Time-Travel Map (Longitudinal Data):
   The researchers looked at a huge group of Finnish twins over 36 years. They checked their weight and activity levels four times (in 1975, 1981, 1990, and 2011).

   • The Analogy: Imagine watching a movie in fast-forward. They didn't just look at a snapshot; they watched the whole story unfold to see what happened first.

2. The Genetic Lottery (Mendelian Randomization):
   This method uses your DNA like a natural lottery ticket. Since your genes are randomly assigned at birth (like a lottery), they can't be changed by your lifestyle. If people with "heavy genes" end up moving less, it suggests the weight is causing the inactivity, not the other way around.

   • The Analogy: It's like checking if people born with blue eyes are more likely to be good at chess. If they are, it's because of the genes, not because they decided to play chess to grow blue eyes.

3. The Twin Test (Twin Models):
   They compared identical twins (who share 100% of their DNA) with fraternal twins (who share 50%). Since twins grow up in the same house and eat the same food, this helps rule out "family secrets" or shared environments as the cause.

   • The Analogy: If one twin is a couch potato and gains weight while the other twin is an athlete and stays thin, but they have the same genes and childhood, the difference must be due to their choices and how their bodies react to each other.

The Verdict: What Did They Find?

After combining all three maps, the story became very clear, but with a twist depending on when you look at it.

1. The Heavy Weight Wins (BMI ➔ Less Activity)
All three methods agreed on one thing: Being heavier makes it harder to move.

• The Metaphor: Think of your body like a car. If you load the car with extra heavy furniture (high BMI), the engine has to work harder, and eventually, you just stop driving as much because it feels too exhausting.
• The Twist: This effect gets stronger as you get older. In the early years of the study, the link was weak. But by the time the participants were in their 60s, carrying extra weight was a major reason they stopped exercising.

2. The Exercise Myth (Activity ➔ Less Weight)
This is where the story gets tricky. The researchers did not find strong proof that exercising causes you to lose weight in the long run.

• The Metaphor: Imagine trying to empty a bathtub by scooping out water with a teaspoon (exercise) while the faucet is still running (your body's natural metabolism and diet). You might scoop out some water, but the faucet keeps filling it back up.
• The Result: While exercise is great for your heart and muscles, this study suggests that simply "moving more" doesn't automatically guarantee you will get thinner, especially over decades. The body has a way of compensating.

The Big Takeaway

The study concludes that high weight is a major barrier to staying active, especially as we age. It's like a heavy backpack that gets heavier the longer you wear it, eventually making you want to sit down.

What does this mean for you?

• Don't blame yourself: If you are heavy and find it hard to exercise, it's not just a lack of willpower; the extra weight physically makes movement harder.
• Focus on weight management first: To stay active as you get older, managing your weight might be the key to unlocking your ability to move.
• Exercise is still good: Even if it doesn't magically melt away pounds, moving is still vital for your health. But don't expect exercise alone to be a magic weight-loss wand if other factors (like diet and genetics) aren't addressed.

In short: Weight limits movement more than movement limits weight. The heavier you get, the harder it becomes to stay active, creating a cycle that is tough to break without help.

Technical Summary

1. Problem Statement

The causal relationship between Leisure-Time Physical Activity (LTPA) and Body Mass Index (BMI) remains controversial. While observational studies generally suggest an inverse association (higher activity leads to lower BMI), conflicting evidence exists regarding:

• Directionality: Does high BMI cause reduced activity, or does low activity cause weight gain?
• Confounding: Unmeasured confounders (e.g., diet, socioeconomic status) and reverse causality often bias observational findings.
• Genetic Confounding: Shared genetic factors (pleiotropy) may influence both traits, creating spurious correlations.
• Temporal Dynamics: Most studies fail to distinguish between between-person differences and within-person changes over time, or they lack the granularity to detect age-specific effects.

The study aims to resolve these ambiguities by applying an evidence triangulation framework, utilizing three distinct methodological approaches to test bidirectional causal hypotheses.

2. Methodology

The study utilized data from the Finnish Twin Cohort (FTC), a longitudinal dataset spanning 36 years with four measurement points (1975, 1981, 1990, 2011).

• Sample: 22,696 twin individuals (aged 18–50 at baseline; 52.4% women). A subsample of 8,527 individuals had genetic data available.
• Variables:
  • BMI: Calculated from self-reported height and weight.
  • LTPA: Measured in Metabolic Equivalent (MET) hours/day using validated questionnaires.
  • Genetics: Polygenic Scores (PGS) for BMI and LTPA derived from Genome-Wide Association Studies (GWAS).

The researchers applied three complementary analytical methods:

A. Random Intercept Cross-Lagged Panel Model (RI-CLPM)

• Purpose: To analyze longitudinal individual-level data while separating between-person (stable traits) and within-person (temporal fluctuations) variance.
• Mechanism: This model assesses whether deviations from an individual's own mean LTPA predict subsequent deviations in their BMI, and vice versa, controlling for time-invariant confounders.

B. Mendelian Randomization (MR)

• Purpose: To infer causality using genetic variants as instrumental variables (IVs), mimicking a randomized controlled trial.
• Approach:
  • One-sample MR: Conducted within the cohort.
  • Instruments: Polygenic Scores (PGS) for BMI and LTPA.
  • Methods: Instrumental Variable Regression (IVR), Inverse-Variance Weighted (IVW), Weighted Median, and MR-Egger (to correct for horizontal pleiotropy).
  • Note: The study struggled to find strong genetic instruments for LTPA (F-statistics < 10), leading to the use of a PGS for moderate-to-vigorous physical activity (MVPA) as a proxy.

C. Twin Modeling (Direction of Causation - DoC and MR-DoC)

• Purpose: To leverage the twin design to control for shared genetics and environment.
• Models:
  • DoC Models: Used cross-twin cross-trait correlations to determine the direction of causality (BMI $\to$ LTPA vs. LTPA $\to$ BMI).
  • MR-DoC Models: Integrated PGSs into the DoC framework to test for causal effects while accounting for potential horizontal pleiotropy (direct effects of genes on the outcome independent of the exposure).
• Stratification: All analyses were stratified by sex and performed at each of the four time points to detect temporal variations.

3. Key Contributions

• Triangulation Framework: This is one of the first studies to simultaneously apply RI-CLPM, MR, and Twin DoC models to the same dataset, allowing for a robust assessment of causal consistency across different assumptions.
• Temporal Resolution: Unlike standard two-sample MR studies that provide lifetime averages, this study examined time-specific causal effects across a 36-year period, revealing that causal strength varies with age.
• Sex-Specific Analysis: Detailed stratification revealed nuanced differences in causal pathways between men and women.
• Methodological Insight: The study highlights the limitations of using PGS as instruments for complex phenotypes like LTPA when F-statistics are low, and demonstrates how MR-DoC models can detect and adjust for horizontal pleiotropy.

4. Results

A. BMI $\to$ LTPA (High BMI causing Low Activity)

• Consistency: All three methods (RI-CLPM, MR, and Twin models) provided consistent evidence for a causal effect where higher BMI leads to lower LTPA.
• Temporal Trend: This effect was particularly strong in the later follow-up stages (1990 and 2011), suggesting that as individuals age, high BMI becomes a more significant barrier to physical activity.
• Sex Differences: The effect was observed in both men and women, though it strengthened over time primarily in women.

B. LTPA $\to$ BMI (High Activity causing Low Weight)

• Inconsistency: Evidence for the reverse causal path was weak and inconsistent across methods.
• RI-CLPM: Found no significant cross-lagged effect of LTPA on future BMI (except a minor positive association in men in 1990).
• MR: Due to weak genetic instruments for LTPA (low F-statistics), MR estimates had wide confidence intervals and lacked statistical significance.
• Twin Models: The DoC and MR-DoC models suggested a negative causal effect of LTPA on BMI in 1981, 1990, and 2011, but the effect sizes were substantially smaller than the BMI $\to$ LTPA path.
• Pleiotropy: In women, MR-DoC models detected horizontal pleiotropy where a higher genetic predisposition to high BMI was associated with higher LTPA (independent of actual BMI), likely due to social pressure to exercise. This suggests that failing to account for pleiotropy could underestimate the true causal effect of BMI on activity.

5. Significance and Conclusion

• Primary Conclusion: There is robust evidence for a unidirectional causal effect of higher BMI on lower LTPA, which intensifies with age. Conversely, the evidence for LTPA causally reducing BMI is less convincing and may be negligible in the long term for many individuals.
• Clinical/Public Health Implication:
  • Weight Management: The findings challenge the assumption that increasing LTPA alone is sufficient for long-term weight management in adults.
  • Intervention Strategy: Efforts to maintain physical activity in older adults should prioritize weight management to prevent the "vicious cycle" where high BMI limits mobility and further reduces activity.
• Methodological Takeaway: The study underscores the importance of triangulation. Relying on a single method (e.g., standard MR or simple regression) could lead to conflicting conclusions. The integration of longitudinal, genetic, and twin designs provides a more nuanced understanding of complex behavioral phenotypes.
• Limitations: The study relied on self-reported data (susceptible to bias), used PGS with weak instrument strength for LTPA, and noted that BMI does not distinguish between adiposity and muscle mass.

In summary, the study suggests that while high BMI actively suppresses physical activity as people age, the reverse causal pathway (activity preventing weight gain) is less definitive, highlighting the need for weight-focused interventions to sustain an active lifestyle.