Por Dawn Chen
Daniel Utter figures
Did you know that Maine's divorce rate is strongly correlated with per capita consumption of margarine? Wow, maybe abstaining from margarine will prevent a divorce! I can definitely envision a pop media article with this catchy headline. Before throwing away all the margarine to save your marriage, a smart reader like you would probably think: "how absurd, it's probably a coincidence that the trends coincide and there is no causal relationship between them after all."
These coincidental but unverified associations can also be found in scientific research, especially in recent news covering the microbiome field that explores the microbes (also known as microorganisms) that inhabit our bodies. The “good” microbes in our gut can help us better absorb nutrients and protect us against infection, while the “bad” microbes can make us sick. As researchers delve deeper into our microbiome, they have discovered that the microbes in our bodies are linked to a wide range of health and disease outcomes, includingobesity,diabetes,Alzheimer disease,depression,Multiple sclerosis,AND YES, miautism.
Ostensibly, these results suggest that a new range of therapies is on the horizon; If we change our diet, eat more probiotics like yogurt, or replace our microbiome with "good" microbes, we're on our way to alleviating these ailments, right? However, the truth is more complicated than it seems. Most of these studies only suggest that there is a relationship between the microbiome and the disease. We still don't know for sure how exactly the microbes caused the disease in the patients, or if the microbes caused the disease at all. Too often, microbiome news falls into the “correlation does not imply causation” trap, where a relationship between two variables does not imply direct cause and effect.
Correlation does not imply cause
In order to critically evaluate existing scientific discoveries, we must first understand the difference between correlation and causation. Correlation means that there is a relationship, or pattern, between two different variables, but it does not tell us the nature of the relationship between them.
Instead, causation implies that, in addition to a relationship between two events, one event causes another event to occur. For example, if we do not sleep, we will feel sleepy. The first (not sleeping) directly triggers the second (feeling sleepy).
The distinction between correlation and causation seems straightforward, but it is easy to mistakenly assume correlation causation, especially when there is a complex interplay of variables. Here are some common themes of wrongly inferring causation from correlation, or why "correlation does not imply causation":
- The relationship between the two variables is coincidental.
Correlation between unrelated variables can occur by chance. An example is the "red skin rule”, where the result of the Washington Football Team's last NFL game before the US presidential election accurately predicted all election results from 1936 to 2000. Intuitively, we know that the result of a football game has no nothing to do with the presidential election: this observation is just a coincidence. The more variables we examine, the more likely we are to find unrelated variables that are correlated by chance.
- reverse causality
Reverse causation means that there is a causal relationship between events A and B, but not in the expected order: cause and effect are reversed. For example, if we observe that the faster the windmill spins, the more wind there is, we may erroneously conclude that the rotation of windmills causes the wind. However, we know that it is the wind that makes windmills turn.
- A common (third) confounding variable causes both events
In some cases, there may be a hidden underlying variable that causes events that appear to be correlated. We might assume that event A causes event B when, in fact, there is another event C that causes events A and B. For example, many researchers have previously found that alcohol consumption is associated with an increased risk of lung cancer. . Nevertheless,to smokeit later turned out to be a confounding factor. People who consume more alcohol also smoke more, which increases the risk of lung cancer.
Observational studies cannot prove causality
Although the correlation is easily observable, determining causation is much more complicated and requires a proper experimental design. Ideally, we would like to perform experiments in a laboratory, where we tightly control all variables except the one we are interested in. However, this is almost impossible in human studies. To perform a more rigorous randomized controlled trial, we would probably need participants to live in the same place, eat the same food, exercise, and sleep at the same time, just to name a few variables. As a result, most research on the human microbiome has been largely observational.
In most large-scale studies of the human microbiome, such as theHuman Microbiome ProjectoAmerican Gut Project,Researchers recruit a group of participants, collect and sequence their stool samples, and simultaneously collect information about the participants' lifestyle, diet, and health status. By analyzing the differences in the microbiome between people with a disease and healthy people, we can find correlations between the composition of the microbiome and the disease of interest (Figure 3).
It is worth noting that the direction of causation in these relationships is often ambiguous. Specifically, scientists have discovered that patients, such as those suffering from inflammatory bowel disease, have different gut bacteria compared to healthy people. Did differences in the gut microbiome make the patient ill, or did the patient's own disease state (eg, more diarrhea or inflammation) lead to differences in the gut microbiome? We often rush to assume the first hypothesis, that the bacterium caused the disease, although the direction of this causal relationship is not so easily determined. Researchers tend to call thisChicken and egg problem🇧🇷 Also, lifestyle is a big confounding factor. Patients suffering from diseases often change their diet after diagnosis or take medication for treatment, which can alter the composition of the gut microbiome.
In an attempt to solve the problem of confounding variables, it isrecent postinsideNatureby Ivan Vujkovic-Cvijin and colleagues picked out lifestyle differences that may be associated with microbiome composition. They found that gender, age, body mass index, and levels of alcohol consumption are the most important confounders associated with microbiome composition and disease status. To eliminate the effects of these confounding factors, the researchers used the individual match approach, in which a sick individual was matched with a healthy individual who was the same age, sex, and lifestyle habits. This is a common technique used in observational studies where researchers cannot control all variables under perfect experimental conditions (Figure 4). Using this technique, the researchers found that many previously found associations between gut bacterial abundance and disease state are no longer statistically significant, suggesting that some changes in the gut microbiome attributed to disease may be the result of factors of underlying confusion.
Stay healthy, stay skeptical
Despite the ambiguity surrounding causation, a growing number of commercial enterprises such asviola, uBiome (which wasraided by the FBIlast year for multiple insurance billing) orDay twoinitiated marketing interventions for the microbiome. Customers would mail in a stool sample for sequencing, and based on the types of bacteria present in the sample, companies would prescribe personalized nutritional information or provide customers with risk scores for different diseases. While these companies have good intentions of helping consumers understand their bodies, we must critically evaluate their claims.
The microbiome is undoubtedly important to our health. However, we are still not sure exactly how the microbiome does this or fits into disease progression, despite the hype surrounding the largely correlative studies. To determine whether the microbiome causes disease, some researchers are exploring the molecular mechanism of individual bacterial strains. Other researchers are working on the design of experimental studies with a larger sample size and a more rigorous methodology. With better analysis tools and data sets, we will soon be able to discover the complex functions that these small living organisms have in our bodies. In the meantime, have a cup of yogurt, just because it tastes good.
Dawn Chen is a first-year PhD student. student of Systems, Synthetic and Quantitative Biology at Harvard University.
Daniel Utter is a Ph.D. of the 6th year. organic and evolutionary biology student at Harvard University.
For more information:
- Explore other fun correlations atspurious correlations.
- Microbiologist Brett Finlay and researcher Jessica Finlay discuss ways to harness microbes in their book,The whole body microbiome.
- A book by the microbiologist Martin Blaser delves into thehow the overuse of antibiotics may be feeding our modern plagues.
- Read more about the microbes inside us atthis bookby science writer Ed Yong.
- Learn more about the need to establish cause and mechanism in microbiome studies in thisresearch work.
The first reason why correlation may not equal causation is that there is some third variable (Z) that affects both X and Y at the same time, making X and Y move together. The technical term for this missing (often unobserved) variable Z is “omitted variable”.What is one of the reasons that correlations do not indicate causation quizlet? ›
correlation does not prove causation because a correlation doesn't tell us the cause and effect relationship between two variables. We don't know if x causes y or vice versa, or if x and y are cause by a third variable. The only thing a correlation tells us is the association or link between variables.In which situation is a correlation and causation not likely? ›
Correlation is a relationship or connection between two variables where whenever one changes, the other is likely to also change. But a change in one variable doesn't cause the other to change. That's a correlation, but it's not causation. Your growth from a child to an adult is an example.Can you have correlation without causation? ›
It is well known that correlation does not prove causation. What is less well known is that causation can exist when correlation is zero. The upshot of these two facts is that, in general and without additional information, correlation reveals literally nothing about causation.What is the difference between correlation and causation give their examples? ›
Theoretically, the difference between the two types of relationships are easy to identify — an action or occurrence can cause another (e.g. smoking causes an increase in the risk of developing lung cancer), or it can correlate with another (e.g. smoking is correlated with alcoholism, but it does not cause alcoholism).What is an example of confusion between correlation and causation? ›
A more classic example of how it's easy to confuse correlation with causation is that we can statistically prove that 97% of people who got into a car accident drank at least one glass of water 24 hours before the accident.What is an example of a statement that confuses causation with correlation? ›
An increase in prey can cause an increase in predators, but an increase in predators will cause a decrease in prey. Thus, predator and prey populations can be both positively AND negatively correlated, depending on where you are in the cycle.