Causal Inference in Comorbidities Using Mendelian Randomization

This direction is merged into another direction: population generalizability (ancestry bias). Its last version stays readable and citable here.

Version 1 of 114 papers16 conceptsfrozen 2026-10-04

In clinical observation, diseases rarely travel alone. We see patients with both depression and gut issues, or asthma and scarred lungs. However, observing a correlation does not reveal the direction of the arrow: does condition A cause condition B, does B cause A, or are they both symptoms of a third, hidden factor?

To solve this "chicken or egg" problem, researchers are increasingly using Mendelian Randomization (MR). By using genetic variants as "natural experiments," this approach allows scientists to determine causal relationships between complex traits without the confounding factors that plague traditional observational studies.

Nature’s Randomized Controlled Trial

The intuition behind Mendelian Randomization is simple: it mimics a Randomized Controlled Trial (RCT). In a standard trial, researchers randomly assign participants to a treatment or control group to see if the treatment causes an outcome. In MR, nature does the assignment at conception.

Because alleles (genetic variants) are randomly inherited from parents to offspring, they are generally independent of lifestyle factors like diet or environment. If a specific genetic variant is known to increase a person's body mass index (BMI), and that same variant is also associated with a higher risk of depression, we can infer that the increased BMI is a causal driver of the depression, rather than the other way around.

This method has been used by Le Zhang and colleagues to show that higher BMI increases the risk of depression and bipolar disorder, while Cancan Li used it to identify a protective causal link between tea consumption and obesity.

Untangling Bidirectional Feedback Loops

One of the most powerful applications of this technique is bidirectional Mendelian Randomization. This allows researchers to test the causal arrow in both directions simultaneously.

For example, Yu‐Hang Xing and team discovered a two-way street between neuroticism and physical frailty: a genetic predisposition to neuroticism increases the risk of becoming frail, and being genetically prone to frailty increases neuroticism. Similarly, Rui Lai demonstrated a bidirectional relationship between Rheumatoid Arthritis and hypothyroidism, suggesting these conditions reinforce one another through shared inflammatory pathways.

In other cases, the arrow only points one way. Xingang Sun found that obstructive sleep apnea (OSA) causes atrial fibrillation, but heart rhythm disorders do not cause OSA. This "one-way" insight is critical for clinicians, as it suggests that treating the apnea early could prevent the heart condition from ever developing.

The Summary Data Bottleneck

While MR is a breakthrough for causality, it currently faces a significant hurdle: the granularity gap. Most MR studies rely on "summary-level data"—aggregated results from massive genome-wide association studies (GWAS) involving hundreds of thousands of people.

While this scale provides immense statistical power, it strips away individual details. As noted by Xinyu Han in their study on GERD and preterm birth, the lack of individual-level data prevents researchers from performing stratified analysis.

This means we often cannot tell if a causal relationship:

  1. Varies by Sex: Many conditions, like hypothyroidism or Alzheimer's, affect men and women differently. Rui Lai and Zihan Yin both emphasize that missing sex-stratified data obscures sex-specific biological drivers.
  2. Changes with Age: A risk factor might be causal in elderly populations but irrelevant in youth.
  3. Depends on Disease Stage: Causal drivers might fluctuate during different phases of a condition's progression, such as different trimesters of pregnancy or stages of lung scarring, as explored by Fang Zhou regarding autoimmune disorders and interstitial lung disease.

Toward Precision Causality

The next frontier in this research area is moving beyond broad population averages toward more granular, context-aware causal models.

Researchers like Cankun Wang are building massive repositories like ssREAD, which integrates single-cell and spatial RNA-seq data. By combining the causal rigor of Mendelian Randomization with this level of biological "resolution," the field is moving toward a future where we don't just know that A causes B, but exactly where, when, and in whom that causal process occurs.

Papers in this version

History & citationVersion 1 · 2026-10-04

Merged into another direction · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_causal_inference_in_comorbidities_using__2e78462b,
  title = {Causal Inference in Comorbidities Using Mendelian Randomization},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/causal-inference-in-comorbidities-using-mendelian-randomization/rdr_2e78462b6ec24e95ab70c81f353cd9b5},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 1, frozen text):

@misc{lacuna_causal_inference_in_comorbidities_using__2e78462b_v1,
  title = {Causal Inference in Comorbidities Using Mendelian Randomization},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/causal-inference-in-comorbidities-using-mendelian-randomization/rdr_2e78462b6ec24e95ab70c81f353cd9b5/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-19},
}