Cats, cancer and confusion

As of the 23rd May 2022 this website is archived and will receive no further updates.

was produced by the Winton programme for the public understanding of risk based in the Statistical Laboratory in the University of Cambridge. The aim was to help improve the way that uncertainty and risk are discussed in society, and show how probability and statistics can be both useful and entertaining.

Many of the animations were produced using Flash and will no longer work.

If mobile phones don't cause brain tumours, what does? Well, according to stories today in the Daily Mail and the Daily Telegraph, it might be cats. Or at any rate a parasite called Toxoplasma gondii, one of whose hosts is the domestic cat. But don't go exiling your kitties yet; the admirable Ed Yong has explained in his blog that the study in question doesn't establish anything of the sort.

Yong points out that the study (abstract here, article behind a paywall) is ecological. That is, it isn't based on an experiment, and it doesn't even involve observing individual people - instead it works at the level of entire countries. It compares the rates of human infection with Toxoplasma gondii in different countries with the incidence rates for brain cancer in adults in the same countries. The Mail article makes this clear; the Telegraph does not.

Yong explains why the ecological nature of the study makes it impossible to conclude anything about what causes what. The data are consistent with Toxoplasma gondii causing the risk of brain tumours to increase, but they are also consistent with brain tumours causing the risk of T. gondii infection to increase, or with the hypothesis that some other independent factor causes both the rate of brain tumours and the rate of T. gondii infection to increase. Yong explains why any of these is plausible.

He doesn't, however, spell out another major problem of ecological studies. All that has been established is that there's a correlation between T. gondii and brain tumour incidence at the level of entire countries. That doesn't tell us anything about correlations for individuals. That is, we don't know whether, within each of the countries, it's the people with T. gondii infections that are most likely to develop brain tumours. Actually, it could be the other way round - that is, within a country, the people with T. gondii infections could be the least likely to develop brain tumours. Assuming that a correlation at individual level has to go in the same direction as one at the level of whole populations is called the ecological fallacy and is a major pitfall of interpreting studies like this.

Given all these doubts, which are mostly explained pretty clearly in the original paper, all the original authors conclude is that "These results, though correlational, suggest that T. gondii should be investigated further as a possible oncogenic [i.e. cancer-causing] pathogen of humans."

Do the press reports reflect the major uncertainty?

Well, up to a point. The Mail does explicitly say, near the start of the article, that the researchers haven't shown anything causal. though the general thrust of the story is that you'd better watch out for cats.

The Telegraph doesn't make the lack of causal evidence explicit at all.

Further, both stories concentrate on the role of domestic cats. Actually T. gondii, like many parasites, has a complicated life-cycle. Members of the cat family do play an essential role, but during part of the parasite life-cycle, they can be hosted by pretty well any warm-blooded animal or bird. There are many routes of human infection that don't involve cats directly at all (as both articles do mention).

Neither article chooses to point out that, actually, brain tumours are fairly rare, despite the fact that this is carefully explained in the press release that, presumably, they were based on. Indeed the Mail report is written to make them sound rather common - "Brain cancer, in its various guises, claims more than 3,500 lives a year in the UK alone", which sounds quite a lot, unless one realises that this amounts to only about 2 per cent of all UK annual cancer deaths.

Is this really something we should be worrying about?

Finally, there are some very odd points in both articles. The Mail says that T. gondii infects "up to 34 per cent of Britons", despite the fact that the study they are reporting on gives the UK prevalence as under 7 per cent. I've no idea where the 34 per cent came from.

Perhaps most bizarrely of all, the Telegraph says, "A parasite spread by cats could almost double their owner’s chance of developing brain cancer, research suggests." This seems to relate to the fact that there is a ratio of 1.8 quoted somewhere in their story, which does indeed come from the original study. But it is emphatically not given as the relative risk of developing brain cancer in people infected with T. gondii compared to people who aren't infected. The authors don't even vaguely suggest that it is. Whoever wrote that sentence for the Telegraph simply got the wrong end of the stick in a very scary way.