On May 1st 2008 the BBC News website ran a story headlined “Keyboards ‘dirtier than a toilet’.” Perhaps inspired by an earlier 2004 news item on the same topic (Lifting the lid on computer filth“), ‘Which’ magazine had done tests on the computer keyboards in their London offices: “Out of 33 keyboards swabbed, four were regarded as a potential health hazard and one harboured five times more germs than one of the office’s toilet seats.”
On May 7th 2008 the Deputy Director of Glasgow Caledonian University emailed all the major UK universities and higher education colleges regarding recent press articles about “the problem of potentially harmful bacteria on keyboard(s) which has caused a little ripple in the student community at Glasgow Caledonian. There have been requests for cleaning materials already!” This email was forwarded to the Health & Safety officer in my office who advised all staff to read the story “If you want to know why eating at your desk is discouraged”.
The idea that ONE of the 33 keyboards tested had â€œfive times as many bacteriaâ€ as ONE toilet seat tested at the same time might explain an emotive or even visceral response but can hardly justify any policy. Aside from the paucity of the â€œsamplingâ€ the use of the relative phrase â€œfive times as many bacteriaâ€ gives no information about the absolute numbers involved: five times a tiny number is still a tiny number.
Risk statements using relative risk are notoriously misleading (one reason perhaps why they are so popular with journalists). For example information from the department of health a few years ago that oral contraceptives containing desogestral and gestodene are associated with a two-fold increase in the risk of thromboembolism led to alarming headlines, causing concern to many women and their doctors. Many women stopped taking the pill leading to an increase in unwanted pregnancies. If the same information about risks had been expressed as an absolute risk, women would have been able to see how frequent this hazardous side effect is.
The risk of thromboembolism if one takes the pill containing desogestral and gestodene is about 30 in 100000 compared to a risk of 15 in 100000 for the older, “second generation” progestogens levonorgestrel and norethisterone. Fortunately, the mortality from venous thromboembolism is low (estimated to be 1-2%), so that mortality is estimated to be no higher than 2-3 per million users. This compares favourably with all cause mortality in pregnancy and with the risks that many people accept in daily life (See British Medical Journal 1995;311 (7013):1111, 28 October).
But why compare computers with toilet seats? I tried to track down the research cited in the BBC news story which “found the average office desktop harboured 400 times more bacteria than the average office toilet seat” (again note the use of relative rather than absolute values). I found numerous reports of it â€“ see e.g. “Flushing Out The Truth”
In 2001 Charles Gerba, a microbiologist from the University of Arizona, and his team looked for five different types of bacteria (E. coli, Klebsiella pneumonia, Streptococcus, Salmonella and Staphyloccus aureus). They studied offices at four U.S. Cities. They sampled 12 different surfaces – desktop, phone, computer mouse, computer keyboard, microwave door handle, elevator button, photocopier start button, photocopier surface, toilet seat, fax machine, refrigerator handle and the water fountain handle.
In terms of bacteria per square inch, they found that the phone receiver was the filthiest – 25,000. This was followed by the desktop at 21,000, the computer keyboard at 3,300 and the computer mouse at 1,700. The least contaminated surface was the toilet seat with only 49 bacteria per square inch.
Arguably the most interesting finding from this research is that the shared toilet seat, which you might expect to be a magnet for microbial activity, was always the cleanest site. One theory is that the toilet seats are too dry to give a good home to a large population of bacteria.
So the BBC story uses relative not absolute risks, invokes emotive thoughts and conceals a surprising finding in order to create an eye-catching story. The policy implications are however not as obvious as most people’s initial reading of the story might lead them to believe. Perhaps instead of thinking anew about computer keyboards we should change our attitudes to toilet seats? Somehow though I doubt that my Health and Safety officer would recommend that we should eat lunch off the toilet seat rather than our desktops.
Interestingly the Arizona study was partly funded by a company which makes disinfecting wipes. No doubt they are very happy with the way it has been reported.