Number of the Month

July 2000

The trouble with trying to be a professional cynic is that you cannot possibly outdo the politicians. In Britain this was the month of Gordon Brown's Comprehensive Spending Revue (pre-election give-away) after three years of Prudence (post election tax hikes). Surely that nice Mr Gore would not go in for anything so blatant, or would he?

The flow continues

Junk science continues to pour out in the columns of newspapers. It would be impossible, not to say extremely boring, to keep track of it all (though the Junkman always makes a valiant effort). One example was a classic of its kind with a headline in the Daily Telegraph ‘Vitamin’ may lead to cancer. Apparently research on 29,000 smokers in Helsinki showed that taking beta-carotene supplements to reduce the risk of cancer actually increased the risk by 18%. The story had nearly all the features we have come to know and love:

The Trojan Number, 29,000, was added to give the story credibility, but we were not told the important numbers (how many contracted cancer and how many of those took beta-carotene).

The risk ratio was 1.18, way below the level least possibly acceptable to real science, which is 2.0.

The research in fact supports what the cynics have long suggested – that the whole anti-oxidant business is nothing more than a scam.

The usual suspects appeared, in the form of Gordon McVie, director general of the Cancer Research Campaign "The research is cast iron in my view" and Clive Bates, director of ASH "All Vitamins should be classed as drugs. People think these products are benign and beneficial, but there could be serious side effects."

As Langmuir’s laws predict, all sorts of fanciful theories are produced to explain the "observation".

A point was being made by attempting to disprove a hypothesis that no reasonable person would have embraced in the first place.

There was a hidden politically correct message that was the prime purpose of the exercise – smokers are doomed and nothing can save them.

  The author of the research concluded with "We don't know enough about the way vitamins work. We need to know more." Which translates as "Please keep giving me the money."

I did an epidemiological survey among smokers in the village pub. Not only did they not take beta-carotene, most of them had never heard of it.

Let there be light

76 - that's a nice number, and this month it has a bit of magic. Memo to JK: idea for a story - Harry Potter's Spotty Horrors. This is the story from The Times, July 18th:

RED and blue light may help to eliminate acne, a dermatologist believes.
The method, developed at Hammersmith Hospital in West London, involves regular sessions with blue and red light rather than traditional treatments such as antibiotics.
Trials by Dr Tony Chu, a consultant dermatologist at the hospital, published in the British Journal of Dermatology, showed a 76 per cent decrease in the number of visible spots in 12 weeks.
Mild to moderate forms of acne are often treated by antibiotics, which can take months to work. Ultraviolet radiation in natural sunlight is also known to help clear acne but at some risk of skin damage.
Dr Chu has found that blue and red lights, at higher wavelengths than ultraviolet, can be very effective. In trials, patients were sent home with portable lights and used them for 15 minutes a day.
Dr Chu believes that the blue light kills the bacteria in the skin that cause acne while the red seems to help the healing process. "Importantly, they do this without any significant side-effects," he said.

Poor old Sir Isaac. Early in the Twentieth century that revisionist Albert Einstein was tinkering with his laws of motion. Now in the Twenty First, another genius is rubbishing his great work Opticks, in which he demonstrated that white light from the sun contains all wavelengths, including red and blue. But these days, my dear, physics is so passé.

Night Terror

A number of the media picked up a story originally published in the BMJ, that infant deaths are much more likely to occur if babies are born at night. The Trojan Number was the total number of births over the relevant period in Hesse, 380 930. Looking at the original paper, however,  we find that there were only 57 such deaths. Of these 31 occurred at night and 26 during the day, which leads the authors to conclude that they have established a significant higher mortality for night time births - risk ratio1.86 (95% confidence interval 1.10 to 3.13) . A nasty old cynic, however, might expect the 57 to be divided according to the lengths of time allocated to day and night (14 an 10 hours, respectively). This would lead to the expected numbers of 33.25 and 23.75. Below are plotted the Poisson distributions for these expected numbers. Readers may wish to judge for themselves just how astonishing the number 31 is on the blue curve or 26 on the red; then reflect on the fact that, while epidemiologists might accept risk ratios of less than 2.0, real scientists do not.


Only connect

It must follow as the night the day that the CJD industry would have a cluster all of its own. The only questions was when and where. Now that it has happened it is an absolute must for the number of the month, which is four. It was all over the media, of course, and here is The Times' version, July 14:

                      Alert on village's cluster of CJD deaths

AN URGENT government investigation was underway last night into a cluster of cases of the human form of "mad cow" disease in Leicestershire, where four people have died.
The investigation follows advice that the four deaths and another probable case - out of only 75 nationwide - were unlikely to have occurred by chance. And experts are pointing to the village of Queniborough as a possible link between them.
Glenn Day, 35, lived there, Stacey Robinson, 18, was a former resident, and Pamela Beyless, 24, often visited. All three died in 1998.
A 19-year-old man who died in May and a 24-year-old who is thought to have CJD also come from the same part of Leicestershire. It was the fifth diagnosis in the area that triggered the investigation.
The junior Health Minister Lord Hunt of Kings Heath last night attempted to calm local people's fears, saying there was no cause for alarm and that the victims would have been exposed to the infective agent "many years ago".
Experts from the National CJD Surveillance Unit and London School of Hygiene and Tropical Medicine have joined the local health authority to investigate the concentration of cases and say they are keeping an open mind about the reasons behind it and whether or not they are linked by Queniborough.
They will be particularly concerned to look at whether any contaminated beef has entered the food chain in the area, particularly since the implementation of new controls in 1989.
Figures released by the CJD surveillance unit this week showed that there were 12 confirmed vCJD deaths in the first six months of this year compared with 13 for the whole of last year. In total, there have been 75 confirmed and probable cases throughout the UK.
The rise in cases is to be discussed at a meeting of the Government's main BSE advisers on Monday, but they are expected to await the August and September figures before taking any policy decision.

The question that arises is - Why do the British press so understate our great scientific achievements? See how the story should have been written here. The Director of the said surveillance unit achieved further headline fame with School meals linked to CJD deaths in The Independent, July 16th. As Matt Ridley pointed out in his cogent Acid Test column in the Daily Telegraph of  July 18th, it is not even established that the disease is caused by food at all. He favours the theory of beef serum in vaccines, which is certainly plausible. The said Director should leave the spreading of alarm and panic to those who are paid to do it, politicians and journalists.

Talking of which, the now annual farce of the hospital league tables also grabbed the headlines. NHS tables expose death rate lottery, yelled The Times. I have wasted enough words on this futile and destructive ritual in Sorry, so I will refrain from further comment.

Finally, for DIY cluster watchers, without comment here are 75 points placed on a 2x1 rectangle using the random number generator in Mathcad to create random x and y coordinates.

P.S. For those who think I can't spell, here is a definition from Chambers:
Revue, n a loosely constructed theatrical show, more or less topical.......