Premature conclusions

by Jim Caryl

Something the media is very good at, and alas some scientists too, is making a conclusion about a scientific investigation before actually performing the investigation.

This is not how science works!

A recent example of this appeared in today’s Daily Mail, the popular gutter-rag that leads the way in pseudo-scientific sensationalism:

Women who drink coffee or tea during pregnancy may increase their baby’s odds of developing cancer, doctors believe.

Experts say caffeine may damage the DNA of babies in the womb, making them more susceptible to leukaemia, the most common cancer in children.

To establish the link, scientists at Leicester University will scrutinise the caffeine intake of hundreds of pregnant women and compare the results with blood samples from their babies after birth.

Researcher Dr Marcus Cooke said there was a ‘good likelihood’ the study would make a connection. Previous research has shown that caffeine damages DNA, cutting cells’ ability to fight off cancer triggers such as radiation.

Changes of this kind have been seen in the blood cells of children with leukaemia. Scientists know they occur in the womb, but do not know why.

‘Although there’s no evidence at all of a link between caffeine and cancer, we’re putting two and two together and saying: caffeine can induce these changes and it has been shown that these changes are elevated in leukaemia patients,’ added Dr Cooke.

So, they’re planning to investigate this link, though Dr Cooke is quoted as (apparently) saying there is a ‘good likelihood of making the connection‘, despite, as he is later quoted, there being ‘no evidence at all of a link between caffeine and cancer’.

Dr Cooke is also quoted as saying that ‘previous research has shown that caffeine damages DNA, cutting cells’ ability to fight off cancer triggers such as radiation‘; now, I am not going to judge Dr Cooke on the basis of such quotes, because I well know how much gutter-rags like to quote out of context, but I can’t help wondering whether this prior research was a case of caffeine being introduced to cells in a dish, rather than to an actual living and breathing mammal. Any number of chemicals can cause physiological disturbance to cell cultures, but these do not necessarily translate to their being harmful to us generally.

So what’s my problem?

My problem is that you cannot make a conclusion about a study before you have performed it; nor is is morally justifiable to create a scare on one possible, yet unfounded, pre-conclusion of such a study; not unless you are willing to write an article on the other possible outcome, which is that there is no effect of caffeine on the cancer susceptibility of cells (it is doubtful that the Mail would ever publish such an article, fond as it is of scaremongering). It is disingenuous to say “we expect to find a connection”, because the truth is, you really don’t know yet.

In science, one forms a testable hypothesis, i.e. that caffeine can induce cancer susceptibility in cells; such a hypothesis should make some predictions about what should be seen when it is tested in certain ways. The aim of experiments should be to disprove the hypothesis, not prove it, as this is the most rigorous way of pursuing evidence.

Why? Well, as is oft said, good hypotheses are those capable of being disproved; if you set out a series of experiments, none of which could result in the falsification of the hypothesis, then you can only possibly end up with a positive result (which couldn’t be trusted), or a neutral result (where you may as well not have performed the experiment in the first place). If data are collected that support the hypothesis, under circumstances where you have included scope for data that does not support the hypothesis, then the hypothesis can be tentatively accepted; but, importantly, it is not proven, as it could be falsified by future experiments.

Now, I’m not doubting that caffeine can have all sorts of unpleasant physiological effects on people, it is after all an addictive drug, but unless the study has actually been done, then any article about it is nothing but a meaningless page-fart.

Advertisements