Thymus and SIDS; Errors of `Sampling Bias`

Recently, I encountered an example of data bias in a book¹ I’m reading.

While summarizing this case from the book, I also referenced an article². Afterward, I will further analyze what kind of data bias this case involves.

Some states in the USA passed a law modeled after the British "Bone Law", which allowed unclaimed bodies from hospitals and detention centers to be used for medical research and anatomical studies. This law was introduced because, in both the UK and USA, obtaining bodies for anatomical research had historically been difficult. However, this law not only had ethical implications but also scientific consequences. This is because most of these unclaimed bodies came from poor individuals, and their socioeconomic status was often linked to unique anatomical and health conditions.

In general, poorer individuals tend to experience more health issues and have fewer resources to treat them. The human body responds to chronic stress by releasing various hormones, such as adrenaline, and this prolonged stress can affect the size and function of glands like the thymus.

In the 19th century, there were cases of infants dying suddenly and unexpectedly, a phenomenon later termed Sudden Infant Death Syndrome (SIDS). Doctors conducting autopsies on these infants observed that many had unusually large thymus glands. However, the size of these glands was actually normal. As previously mentioned, the thymus is part of the body’s immune defense system and is sensitive to prolonged stress. Individuals under chronic stress, such as those from impoverished backgrounds, often had smaller, atrophied thymus glands.

As a result, the thymus glands of infants from impoverished families who were autopsied by doctors were often shrunken. However, doctors assumed this was a typical feature of all infants, not recognizing it as a characteristic specific to those under chronic stress.

In other words, infants from impoverished families often died from conditions related to chronic stress, such as diarrhea or malnutrition, which caused their thymus glands to atrophy over time. By contrast, infants who died from SIDS passed away suddenly, without experiencing the prolonged stress that would have led to thymus shrinkage. However, doctors incorrectly concluded that the cause of SIDS was an enlarged thymus compressing the infant’s airway, leading to suffocation.

As a result, in the early 20th century, doctors began irradiating the thymus to reduce its size, leading to severe consequences. Many children suffered burns or atrophied thymus glands, and later in life, around 10,000 of them developed cancer, with many dying prematurely.

Statistical Errors

This case demonstrates two major statistical errors: sampling bias and confounding variables.

1. Sampling Bias

The study population was skewed toward a specific group (impoverished families), which introduced sampling bias. The conclusions were based on data from children with atrophied thymus glands due to malnutrition and stress, rather than from a representative sample of the general population.

2. Confounding Variables

Confounding variables refer to external factors that were not accounted for, which distort the relationship between cause and effect. In this case, the size of the thymus was influenced by factors like chronic stress and nutritional deficiencies, but researchers only focused on the thymus size as the cause of death, failing to consider these other important factors.