7th February 2021
Bad data and ‘correlation confusion’
Misinformation and disinformation present a clear and present danger to public health. Over the last 20 years it has lead to long lasting damage, and even death within some vulnerable communities that are susceptible to the inaccurate portrayal of information. It has contributed to some of the reasons why we still see outbreaks of vaccine preventable diseases such as Measles and Mumps, and why we still see deaths from these vaccine preventable diseases.
Misinformation and disinformation surrounding coronavirus has the potential of extending the pandemic by causing vaccine hesitancy, by persuading people that they don’t need to follow the rules and potentially leading to further deaths.
Sometimes this has a purely innocent reasoning, usually due to a misreading or a misunderstanding of data, for example the claim that using a 28-day window to record COVID deaths is wrong and leading to over reporting of deaths. This has been debunked on numerous occasions; 88% of those dying within 28 days and 96% of people dying within 60 days of a positive test had COVID-19 on their death certificate. Further to this it’s also the lack of understanding of “excess deaths” which is a secondary sign of the number of deaths from the pandemic, over and above the average for the last 5 years. (Which closely follows the official death count and should reassure people that we are not over counting deaths).
However sometimes there is a more callous reason behind misinformation or disinformation, it might be where a person or company wants to promote their own therapeutics over vaccines, or from conspiracy theorists who want more people to follow them.
Public Health England created a useful set of criteria that you should follow before liking, sharing or posting information online, it’s the SHARE framework…
S – Source – Does the source of the information have reputation for accuracy?
H – Headline – Read beyond the headline! A scary headline doesn’t always mean doom and gloom
A – Analyse– Double check the facts elsewhere (For example with the NHS, in the BMJ or Lancet)
R – Retouched – Check if the picture or graph has been manipulated or changes
E – Error – Check for phony links, is it from a genuine source, email or website?
You can see more about SHARE here – www.publichealthslough.co.uk/campaigns/share
The 21st century has bought with it a huge growth of social media platforms. These are great for networking and reducing social isolation, but they can cause challenges. For example, research done at the end of 2020 showed that some of our communities that are the most vulnerable to coronavirus are also the ones most susceptible to misinformation circulating on twitter and Facebook. Why is this important? If the most vulnerable communities go unprotected then the virus will continue to circulate, continue to kill, and continue to mutate, therefore prolonging the pandemic.
A big part of misinformation and vaccine hesitancy is driven by the misinterpretation of data and what I like to call “correlation confusion”.
What is correlation confusion? It’s essentially something that looks like it’s related but isn’t. Just because something looks like it’s caused by something, doesn’t mean that it is.
For example, did you know that the numbers of people drowning in a swimming pool is directly correlated to the number of films that Nicholas Cage has stared in? The more films he was in, the more drownings in a pool. Neither related, but both look similar.
Did you know that people that use hair straighteners are more likely to get breast cancer than those that don’t use them? Are the two linked? Yes! Is the cancer caused by hair straighteners? No! What’s the causal link here? The fact that the type of people more likely to use hair straighteners are also the same type of people that are more likely than others to get breast cancer i.e. Women. The causal link here is women; they are more likely than men to use hair straighteners and are more likely than men to get breast cancer.
If I wanted to be facetious, I would have said “NEWS FLASH – Use of hair straighteners causes breast cancer” A perfect example of correlation confusion.
A more relevant example here with vaccines is as follows: If we studied 10million random people over a 2month period, around 4,000 will have a heart attack, 4,000 will get cancer and around 10,000 will die. This is called ‘all-cause mortality’ i.e. the average number of expected deaths due to things like old age, cancer, stroke etc..
Now, if we followed the same 10million people over a 2month period, but we vaccinated them all first, around 4,000 will have a heart attack, 4,000 will get cancer and around 10,000 will die. The same ‘all-cause mortality’, but some may link the death to the fact that they were recently vaccinated. They will have succumbed to ‘correlation confusion’ and not recognised that the people may have died because of old age, because they were lifetime smokers, physically inactive or obese.
It becomes increasingly important during the social media age that we don’t confuse correlation with causation, and we ensure what we are reading is genuine and not misinformation.
Vaccines bring us some light in helping end the pandemic, but we need to ensure that the largest number of people possible get vaccinated, whether vulnerable or not, to ensure that the virus can’t continue circulating or mutating.
When you get vaccinated you can help too! Give a shout out to all your friends and families that you are helping end the pandemic, that you are protecting others, protecting the NHS and helping save lives. And be proud to say #IamVaccinated