Disease causality
Definitions & Key takeaways
Causality refers to a consequential relationship between two things. Disease causality is the relationship between a disease and its cause. A disease may have multiple causes, which can be categorized as either direct or indirect. Direct causes are those responsible for the initiation or aggravation of the disease process, while indirect causes predispose an individual to the development of a disease. Identifying the cause of a disease is essential to its treatment.
One of the main goals of epidemiology is to identify causal relationships between outcomes - like death, diseases, or injuries - and exposures - like smoking cigarettes, eating junk food, or drinking alcohol.
For example, nowadays, it’s widely known that smoking cigarettes causes lung cancer, or in other words, that smoking cigarettes leads to the development of lung cancer in many people.
But how did we figure this out? In the mid- 1950’s, an epidemiologist named Sir Austin Bradford Hill came up with nine guidelines for determining whether or not two things are causally related, and these are called the Bradford Hill Criteria.
The first criterion refers to the strength of association, and says that two things are more likely to be causally related if the strength of association between them is large.
For example, for people who smoke, the relative risk of lung cancer is around 25, meaning people that smoke have 25 times the risk of developing lung cancer compared to people that don’t smoke.
On the flip side, the relative risk of breast cancer for people who smoke is only about 1.5, which is much lower than the relative risk of lung cancer.
So, smoking is much more likely to cause lung cancer than it is to cause breast cancer. The second criterion states that causal relationships are dose- dependent, meaning a person who is has higher amounts of exposure will also have a higher risk of developing the outcome.
For example, the relative risk of lung cancer for people who smoke 10 cigarettes per day might be 8, the relative risk of lung cancer for people who smoke 15 cigarettes per day might be 13, and the relative risk of lung cancer for people who smoke 20 cigarettes per day might be 22.
On the other hand, if a person decreases or stops their exposure, their risk of the outcome also decreases. For example, in people who stop smoking, the risk of dying from lung cancer decreases by half.
In a causal relationship, the exposure has to precede or come before the outcome, and this is called temporality, which is the third criterion.
For example, typically people are more likely to develop lung cancer after they have smoked, so smoking comes before lung cancer.
Oftentimes, there is a lag- time or gap between the exposure and the outcome, and the timing of this gap also has to be consistently true.
For example, it usually takes 20 years for lung cancer to develop after a person begins to smoke, since the negative effects of smoking on the body build up over time.
So, we should see around a 20- year lag- time in all studies on smoking and lung cancer. Generally, temporality is considered the strongest evidence for causal relationships.
The fourth criterion is that causal relationships must be biologically plausible, which means the relationship can be explained by biological mechanisms.
For example, cigarettes contain around 70 different chemicals that damage DNA, the main component of cells that controls how they grow and carry out different tasks.
Sometimes when DNA gets damaged, cells can start to grow uncontrollably, which can cause cancer to develop in the form of tumors.
Knowing the biological mechanism helps to eliminate possible confounders, or external factors that can distort the relationship between the exposure and the outcome.
The fifth criterion is that causal relationships should be coherent, meaning any new piece of biological evidence that’s discovered should fit into one step of the biological mechanism.
For example, new technology has allowed researchers to look at the specific genetic changes that happen when the DNA is damaged, which solidifies that step in the pathway from smoking to lung cancer.
Causal relationships are often supported by other, similar causal relationships, and the sixth criterion is referred to as analogy.
Specifically, a causal relationship between an exposure and an outcome is supported when there is existing evidence that a similar exposure causes a similar outcome.
Knowing this, we can predict that smoking will have an analogous effect on the lungs. The seventh criterion states that causal relationships need to be supported by results from randomized controlled trials or RCTs.
In randomized controlled trials, the researcher randomly assign two groups of people - one to be exposed and one to not be exposed.
Because the groups are randomly assigned, there’s a pretty good chance that the two groups will have similar characteristics, like having the same number of males and females or a wide range of ages in each group.
If the groups are similar in every way except for the exposure, and we can be sure that any changes in the frequency of the outcome are the result of being exposed or not.
But sometimes experimental studies can be unethical - for example, you can’t force someone to smoke cigarettes if they don’t want to!
And in that situation, you might look at how the chemicals in cigarettes affect blood, tissue, or cell samples. Causal relationships should also be tested in a variety of populations, like people of all ages, sexes, and races, and the results should consistently point to the same conclusion, so the eighth criterion is consistency.
For example, you should find that smoking causes lung cancer in people in Norway and people in Singapore, even though the exact risk of lung cancer between the two populations might differ based on environmental or genetic differences.
Finally, the ninth criterion in the Bradford Hill criteria is that the causal relationship is specific, meaning there is only one exposure that causes one outcome.
Typically, this is considered the weakest criterion, since it’s possible for multiple exposures to work together to cause an outcome, and most epidemiologists use a different model - called the causal pies model - to explain this concept.
In the causal pies model, an individual factor that contributes to cause the outcome is shown as a piece of the pie, and when all the pieces of the pie are in place, the pie is complete and the outcome occurs.
Each individual piece of the pie is called a component cause or component exposure and the complete pie is called a sufficient cause or sufficient exposure.
To go through this, let’s use the example of tuberculosis, or TB. Let’s say there are three component causes of the tuberculosis pie.
The first component cause is contact with the bacterium Mycobacterium tuberculosis, which is spread through the air when an infected person coughs or sneezes.
The second component cause is being part of a high- risk group of people, like infants, elderly people, and pregnant mothers, who have weaker immune systems and are at a higher risk of getting tuberculosis.
The third component cause is having weak lungs, as a result of an anatomic defect in the lungs, like a cyst. So, coming in contact with the bacterium, being part of a high-risk group, and having weak lungs are the three component causes that make up a sufficient cause of developing tuberculosis.
Sometimes a disease has more than one sufficient cause. For example, let’s say there are two people who develop tuberculosis.
The first person comes in contact with the bacterium and is part of a high- risk group, but they don’t have weak lungs, so it’s not included in the pie.
The second person comes in contact with the bacterium and has weak lungs, but they are not part of a high- risk group. In this situation, both individuals have a sufficient cause and will develop tuberculosis.
Now, a component cause that appears in every pie is called a necessary cause, because the outcome doesn’t occur without it.
In this example, coming in contact with the bacterium is a necessary cause, because a person has to come in contact with the bacterium in order to develop tuberculosis.
Typically, taking one necessary cause out of the pie stops that pie from becoming a sufficient cause. For example, if a person is in a high- risk group but never comes in contact with the bacterium, they’ll never get tuberculosis.
On the other hand, being part of a high risk group and having weak lungs are not necessary causes, since they are not required for developing tuberculosis, and removing one of these components will not necessarily stop tuberculosis from developing.
Alright, as a quick recap, an exposure and an outcome are more likely to have a causal relationship if they fit the nine following guidelines.
First if strength of association between them is large, if the relationship is dose- dependent, temporal, biologically plausible, and coherent; if the relationship is analogous to other causal relationships, if it’s supported by results from randomized controlled trials, and if it’s consistently supported by results from many studies.
Finally, an outcome might have one specific exposure, but typically an outcome has multiple component exposures that work together to create a sufficient exposure.
No notes for this video yet
Try adding a note below