The dangerous confusion of correlation with causation: the case of the use of bleach and the incidence of respiratory disease

There are media reports of a recent study on the use of bleach in Europe highlighting the confusion between correlation and causation.

Correlation is a statistical measure that indicates the extent to which two or more variables fluctuate together. In the case of this study, where both of them increased together.

After taking into account factors such as passive smoking at home, the presence of household mould and the use of bleach to clean school premises, the researchers found higher rates of infections among children whose parents regularly cleaned the home with bleach in all three countries. The risk of flu was 20 per cent higher for children whose homes were cleaned with bleach, while the risk of recurrent tonsillitis was 35 per cent higher for these children than for children whose parents did not regularly use bleach to clean. The risk of any recurrent infection was 18 per cent higher for children whose parents regularly used bleach. The findings back previous studies linking cleaning products with respiratory symptoms and inflammation.

What this study is  really saying is that the use of bleach and certain infectious diseases appear to occur together in a number of households ie. there is a correlation between the two.

What it does not say, and cannot say, is that  is the use of bleach causes the infectious diseases.

It’s an important distinction because many studies imply that the fact that two variables,  (in this case the use of bleach and infectious diseases) are occurring together they are necessarily causally linked, namely that one causes the other.

This may be the case but the fact that they occur together does not establish that.

What it really does is provide an indication of where research and causal factors should be directed. What is needed in this particular study is a causal connection between the use of bleach (presumably the presence of traces of bleach in the atmosphere) causes the infectious diseases.

In the case of passive smoking there was a correlation between lung disease and passive smoking and this was backed up by  medical evidence that demonstrated  that even the lower amounts of nicotine from the inhalation of cigarette smoke by non-smokers had similar effects to those of smoking.

It’s an important distinction.

The presence of correlation  is a marker for investigation of causal connections. It’s been done in the case of passive smoking. It does not appear to have been done in the case of the inhalation of bleach fumes.

There is another possible explanation for these findings. The families that use bleach live in unnaturally unhealthy conditions which are in themselves the cause of the higher levels of respiratory infections. The poor living conditions rather than the use of bleach are the cause of the respiratory problems.

One of the other problems with this study is that it does not indicate whether the families that had high levels of respiratory disease also had a statistically significant predispositions towards these types of infections and that this was actually the distinguishing causal variable. Or, even more likely, that the presence of bleach fumes in the atmosphere will exacerbate the predisposition towards respiratory disease.

In this example, the use of bleach and the incidence of infectious disease may fluctuate together but this is a result of an intervening variable, poor living conditions.

Reputable newspaper should take more care about publishing this kind of study without the necessary caveats

There are numerous examples the confusion between correlation and causation.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s