The Hindu Times

Hello, hello. It's good to be back.

My source of news these past three weeks has been The Goa Herald and BBC World. Have I missed anything? I heard that tolerant, crime-free Sweden has a racist serial killer on the loose, that the French strikers in Paris have shown considerably more commitment than the French strikers at the World Cup. And my own football club finds itself managerless and second to bottom in the bloody Championship. Can all this possibly be true?

What of India? I'll spare you my banal observations. Suffice to say it's not the place to go if you don't like curry (hello Patrick Basham!), the streets are filled with delicious cows and the smoking ban is more honoured in the breach than in the observance.

More—much more—to be said later today. I have some catching up to do. Meanwhile, via Dick Puddlecote's link tank, I see The Atlantic has an interview with arch-rationalist John Ioannidis (mentioned before on this blog), whose classic paper 'Why most published research findings are false' should be on the national curriculum.

“The studies were biased,” he says. “Sometimes they were overtly biased. Sometimes it was difficult to see the bias, but it was there.” Researchers headed into their studies wanting certain results—and, lo and behold, they were getting them.

We think of the scientific process as being objective, rigorous, and even ruthless in separating out what is true from what we merely wish to be true, but in fact it’s easy to manipulate results, even unintentionally or unconsciously. “At every step in the process, there is room to distort results, a way to make a stronger claim or to select what is going to be concluded,” says Ioannidis. “There is an intellectual conflict of interest that pressures researchers to find whatever it is that is most likely to get them funded.”

And:

Though scientists and science journalists are constantly talking up the value of the peer-review process, researchers admit among themselves that biased, erroneous, and even blatantly fraudulent studies easily slip through it. Nature, the grande dame of science journals, stated in a 2006 editorial, “Scientists understand that peer review per se provides only a minimal assurance of quality, and that the public conception of peer review as a stamp of authentication is far from the truth.”

Please go read...

Share

Tweet

Nuns, prostitutes, witches and toads

[A fairly lengthy article about cervical cancer might not sound the most interesting way to spend ten minutes but it provides a good example of epidemiology confusing cause and effect.]


In 1842, the Italian physician Domenico Antonio Rigoni-Stern noticed that nuns in Verona were more susceptible to breast cancer than other women. Ruminating on what aspect of convent living might explain the phenomenon, he concluded that the nuns' corsets were too tight. This explanation was wide of the mark but the initial observation was sound; we now know that having children makes a woman less likely to develop breast cancer.

But the Italian found another association. It seemed that the nuns were less prone to cervical cancer than the rest of society. Furthermore, cervical cancer was unusually common amongst prostitutes. This observation provided the first clue to the real cause of the disease but it would take a further century and a half of confusion and dead-ends before the real truth emerged.

Early myths

The very earliest myth about cervical cancer emerged in early modern Europe where witches were believed to curse their victims with warts which ultimately led to a painful death. As we shall see, even this bizarre notion was not entirely without reason but, by the dawn of the 20th century, medicine had moved on and in 1901 The Lancet observed that cervical cancer was rare amongst Jewish women.

Suspecting that the disease was caused by excess salt, The Lancet correspondent surmised that Jews were less susceptible because they avoided bacon. An alternative explanation was offered when the subject was revisited in 1959, when a doctor speculated that circumcision reduced the risk (1). Neither man was correct about the cause but, again, the association was real.

In the 1970s, epidemiological studies showed that cervical cancer was more common amongst women who had a history of herpes. It was therefore assumed that the cancer was caused by the herpes virus. It was not, but it was becoming clear that the disease was in some way connected with sexual activity. At this point, with the answer now within reach, researchers dived down a blind alley.

Modern myths

The suggestion that cancer of the cervix was caused by smoking was first made in 1977 and, by the end of the 1980s, the belief that passive smoking was also a risk factor had reached the pages of the Journal of the American Medical Association (2). A number of epidemiological studies had shown a small but consistent relationship between tobacco smoke and cervical cancer, and yet something strange seemed to be happening. Studies frequently showed that passive smoking carried as high or higher risk than active smoking. This was completely at odds with the expected dose-response relationship. The association with passive smoking was, as one researcher commented, "almost too strong".

Nevertheless, anti-smoking groups let it be known that smoking doubled a woman's risk of contracting cervical cancer and, with tobacco now firmly in the frame, the earlier observation that the disease was related to sexual activity came under fire. In 1991, the British Journal of Obstetrics and Gynecology published an article by Malcolm Griffiths which flatly stated that Rigoni-Stern's research had been sloppy. There was, said Griffiths, no reason to believe that nuns had a lower rate of cervical cancer than the rest of the female population (3). The implication was obvious: cervical cancer was not related to sexual activity.

Griffith's theory took hold and the became the orthodox view. Still, not everyone was convinced. In 2000, when the British Medical Journal reopened the debate, an exasperated professor wrote to ask:

"We now know, with lots of evidence, that cervical cancer is a venereal disease, the responsible herpes type virus being passed from female to female via male sexual partners. Why is there no comment about this in your letters and no statistics about the number of sexual contacts found in the various studies, which is clearly the most relevant fact in incidence and death from the condition?"

To which a fellow doctor replied:

"I had, until recently, believed that nuns did not get cervical cancer. I accepted the received wisdom since I found it repeated so frequently. I realized how mistaken I was when I encountered the paper on this topic by Griffiths. In his paper Griffiths clearly and persausively demonstrates that the assertion that nuns are not at risk of cervical cancer is based on a mis-reading of Rigoni-Stern's original report of 1842. I hope that if enough of us highlight Dr Griffiths' work often enough this myth will be laid to rest." (4)

But, as was already becoming clear, sexual activity was very closely related to cervical cancer risk. By the end of the millennium, scientific—as opposed to statistical—evidence conclusively proved that the sexually transmitted human papillomavirus (HPV) was implicated in all cases of cervical cancers. HPV did not guarantee the onset of cervical cancer but it was impossible to contract the disease without first being infected with it. With this knowledge, the age-old observation of the nuns and prostitutes made perfect sense.

The truth

Biological evidence indicated that cervical cancer was no more caused by herpes than breast cancer was caused by tight corsets. The rarity of cervical cancer amongst Jewish women was, scientists now realised, due to the protective P53 gene which was far more common amongst Jews than other races. Salt had been a red herring. Even the old myth that cervical cancer was connected to witchcraft had some bearing in fact, since HPV causes warts to grow. In medieval society warts were associated with toads which were, in turn, associated with witches.

Smoking, like herpes, was not the cause of cervical cancer, it was simply more common amongst women who were sexually active and, therefore, more likely to be infected with HPV. Had the Italian doctor looked at tobacco use, he would have observed that smoking was more common amongst prostitutes than amongst nuns. He might also have noted that prostitutes were more likely to drink alcohol, have children and be atheists. But none of these things caused cervical cancer; they were just more common amongst prostitutes. The statistical correlations were real, but they were meaningless until the nature of HPV was understood.

The clues had always been there. For one thing, cervical cancer mortality had fallen by 80% in the second half of the 20th century, at a time when tobacco consumption had never been higher (5). This, in itself, strongly suggested that smoking was not a cause of the disease.

That smoking was correlated with sexual activity had been known for many years. In 1994, an article in the International Journal of Epidemiology had flagged up the issue:

"The correlation between cigarette smoking and sexual activity that exists in most cultures makes evaluation of the potential additional role of smoking difficult." (6)

Indeed it did. One epidemiological study (Nischan, 1988 (7)) had found a relative risk from smoking of 1.5 (ie. a 50% increase in risk), but, as the authors noted, "the risk for smokers depended significantly on number of sexual partners"—a telling observation which warranted closer scrutiny. Indeed, when they adjusted their figures in an effort to account for this mysterious confounder, the risk—already low—fell to just 1.2 (0.8-1.6) and was no longer statistically significant.

The real giveaway was that the risks for passive smokers were as high, or higher, than those found for smokers. Given the vastly lower dose of smoke inhaled by the former, this was a biologically implausible finding which defied both common sense and the expected dose-response relationship. Another study (Scholes 1999 (8)) found a risk of 1.4 for both smokers and passive smokers. The study that attracted the most attention from the press (Slattery 1989 (9)) found a relative risk for smokers of 3.42—much higher than the association found in other studies but still not quite as high as that found in the same study for passive smokers where a 3.43 risk was reported.

To take such findings at face value would require tearing up the toxicological text-book. It was hugely improbable that passive exposure to tobacco smoke in any setting could be as harmful of smoking. What was much more likely was that cervical cancer was not a smoking-related disease at all and that the findings had been skewed by the confounding factor of sexual activity.

The key to the riddle was the human papillomavirus. Once it was shown to be implicated in every case of cervical cancer, the long observed correlation between sexual activity and cervical cancer made sense. And since smokers—as a group—were somewhat less likely to be celibate and somewhat more likely to be promiscuous, the correlation between smoking and cervical cancer had a rational explanation. Correlation did not equal causation. Yes, there was an association between smoking and cervical cancer but the one did not in any way cause the other.

The association with passive smoking was more intriguing and had a unique explanation. Because HPV is carried by men and easily transmitted to women, a woman's risk of catching the virus depends as much on her partner's sexual history as her own. She may be a nonsmoker with no previous lovers but if her partner has slept with many women she remains at high risk. And since the smoking male tends to be have had more sexual partners than his nonsmoking counterpart, the woman who lives with him is at higher risk of contracting HPV and, therefore, cervical cancer. Yes, living with a smoker slightly raises a woman's risk of cervical cancer but not because he smokes. Correlation does not equal causation.

The passive smoking researchers who had adjusted their results for sexual history had made the mistake of only asking questions of the women. They had, of course, made every effort to find out about the man's smoking status but had neglected to enquire about his own sexual history which was far more important, considering his role as the HPV carrier.

As the role of HPV in cervical cancer became known, it became clear that smoking was not independently associated with the disease. HPV was the key and, by 2006, an effective vaccine against it was developed. Immunisation programmes were initiated by governments around the world and the possibility of wiping out cervical cancer within a few generations became a realistic proposition.

It was a rare breakthrough in the war against cancer. It came about thanks to advancements in our understanding of biological science and was facilitated by the development of new drugs. It was, if you will, an 'old-fashioned' way of dealing with a public health problem. It owed nothing to the army of epidemiologists who were dominating popular science, nor to the proponents of the fashionable belief that cancers were caused by tobacco, diet, alcohol and environment. Not only had the social theorists missed the mark, but their preoccupation with smoking had sent researchers on a wild goose chase for more than two decades.

Carry on regardless

And yet, the anti-smoking organisations still had a slew of studies showing an association between smoking and cervical cancer. To admit these studies were worthless would be to accept the failure of their methods. Not only that, but admitting that cervical cancer was not a smoking-related disease would be to lose a weapon in their campaign to dissuade women from smoking. And so they carried on as if nothing had happened.

Today, the American Cancer Society maintains that smoking doubles the risk of cervical cancer (10), based only on the statistical correlation. The fact that the organisation also claims that chlamydia, multiple pregnancies and oral contraceptives are also risk factors shows that the ACS has learnt nothing from the long history of mistaken cause-and-effect that has surrounded cervical cancer over the centuries. Chlamydia, multiple pregnancies and oral contraceptives—like herpes, warts and prostitution—are obviously associated with unprotected sex. It would be a truly extraordinary coincidence if these factors also happened to be independent risk factors from smoking.

At least the ACS no longer claims that passive smoking raises the risk. The same cannot be said of other anti-smoking groups. Rather than accepting that the studies had been flawed because they had failed to adjust for HPV infection, they continue to take them at face value.

As implausible as it is, the idea that passive smoking is more dangerous than smoking holds an appeal to such groups. Recently, a study on breast cancer showed a higher risk for passive smokers than smokers. In truth, smoking of any kind was not a risk factor for breast cancer, as even the American Cancer Society accepted. Nonetheless, ASH released a press release titled 'Secondhand tobacco smoke more dangerous than smoking itself - implications for women especially frightening' (11) and Stanton Glantz described the finding as "the most important scientific development in the last 10 years" (12).

Their motives were obvious. Having banned smoking in virtually all indoor places, their attention turned to the outdoors, but to ban smoking in the open air required persuading the public that secondhand smoke was exceptionally toxic.

Further 'evidence' for this wacky idea appeared in the form of the Helena heart attack study (co-authored by Glantz) which claimed that heart attack incidence fell by a whopping 40% once smoking was banned in bars and restaurants. Not accepting for a minute that this study might also be fatally flawed, anti-smoking groups used it as further evidence that secondhand smoke was far more dangerous than had been previously thought. Glantz also wrote an article titled 'Even a little secondhand smoke is dangerous' (13) and claimed that 30 seconds of exposure could kill.

In 2006, the Surgeon General told the American public that there was "no safe level" of secondhand smoke exposure. Secondhand smoke had become the most deadly entity known to man.




(1) Dunn JE, Buell P. Association of cervical cancer with circumcision of sexual partner. J Natl Cancer Inst 1959;22:7469.
(2) In January 1980, the American Journal of Epidemiology reported that there "an apparent connection, as yet unexplained, between smoking and cancer of the uterine"
(3) http://www.bmj.com/cgi/eletters/319/7210/642
(4) 'Screening and mortality from cervical cancer', Rapid responses; BMJ 1999; 319: 642 http://www.bmj.com/cgi/eletters/319/7210/642
(5) Cancer Research UK, 'UK Cervical Cancer mortality statistics'
(6) 'Cigarette smoking as a potential cause of cervical cancer; has confounding been controlled?', Andrew Phillips & George Davey Smith, International Journal of Epidemiology, 1994, Vol. 23, No. 1, pp.42-49
(7) 'Smoking and invasive cervical cancer risk: Results from a case-control study', Nischan et al. American Journal of Epidemiology, 1988, vol. 128, No. 1: pp. 74-77
(8)Scholes et al. Cancer Causes & Control, Vol, 10, No. 5, Oct. 1999 pp. 339-344 'The association between cigarette smoking and low-grade cervical abnormalities in reproductive-age women'
(9) JAMA, vol. 261, No. 11, 17/3/89, Slattery et al. 'Cigarette smoking and exposure to passive smoke are risk factors for cervical cancer'
(10) http://www.cancer.org/
(11) ASH press release, 2005 www.ash.org
(12) 'Lecture ties secondhand smoke to breast cancer', Michael Coburn, The Dartmouth News, 7.3.08
(13) www.tobaccoscam.ucsf.edu/pdf/9.1-Glantz&Parmely-EvenALittleIsDangerous.pdf


[Originally published at velvetgloveironfist.com in 2009]

Share

Tweet

Palm reading

From the once-great British Medical Journal comes another scientific breakthrough...

Firm handshake link to long life

The strength of your handshake could be a clue to how long you'll live, say scientists from University College London.

They matched older people's balance, grip strength and ability to get up from a chair with their risk of an earlier death.

Uh-huh.

The researchers found that death rates over the period of the studies were 67% higher in people with the weakest grip strength compared with the strongest.

A similar pattern was found in the other measures, with the slowest walkers almost three times more likely to die compared with the fastest.

Frail and weak people are more likely to die than strong and healthy people. Hold the front page.

Those slowest to rise from a chair had double the mortality rate compared with those quickest to their feet.

Even being able to balance on one leg appeared to be linked with a reduced risk of death.

People who can barely stand up are less healthy than those who can hop. Here's Tom with the weather.

And what are the practical applications of this doubtless well-funded piece of research?

Professor Avan Aihie Sayer, a geriatrician and co-author on the study based at Southampton University, said that she was now pushing for wider use of measures such as grip strength in hospitals as a way of spotting patients with greater problems.

That would be greeting patients by shaking their hand, wouldn't it? In these difficult economic times, it would certainly be cost-effective. And about a thousand years ago it might even have been a step forward as a method of diagnosis.

I haven't been in a hospital in a while, admittedly, but I was under the impression that we had slightly more sophisticated methods and technology these days—stethoscopes, X-rays, body-scans, blood tests etc. Or is all handshakes and leeches again these days?

Next week in the BMJ: New study shows that fiddles and butcher's dogs are fitter than the bed-ridden and terminally ill.

Share

Tweet

On the level about epidemiology

[This is a guest post by Rob Lyons, deputy editor of Spiked] 

Richard Wilkinson and Kate Pickett have made great play of the fact that The Spirit Level is based on what they see as a science: ‘social epidemiology’. Leaving aside the dubious nature of their conclusions, it is worth noting that since the field of epidemiology has expanded, its explanatory powers have shrunk. While the ability to examine the spread of infectious disease has been invaluable, employing epidemiology beyond this has been - for the most part - bad news for our understanding of health. It's even worse as a method of understanding society.

The most famous early example of the application of epidemiological ideas is the ingenious manner in which John Snow, a London physician, managed to identify the source of a cholera outbreak in Broad Street, Soho in 1854. Simply by mapping the cases of cholera, he found they were clustered around one particular well. Whether Snow’s intervention to disable the well by removing the handle of its pump really saved the day is a matter of some dispute, but his method of plotting cases and looking for concrete causes was inspirational.

Fast forward to the 1950s, where a classic study by Austin Bradford Hill and Richard Doll identified for the first time (at least, outside Germany) the link between smoking and lung cancer. Bradford Hill and Doll sent a questionnaire to doctors to measure their smoking habits, then followed up the questionnaire some years later. Those doctors who smoked were much more likely to have developed the disease and no one would seriously disagree now that active smoking is a very significant risk factor for lung cancer.

There is a proviso, however: even with smoking, the most famous of these epidemiological connections, most of those engaged in the risky behaviour - smoking - do not die from lung cancer while there are a small number of non-smokers who die from lung cancer, too. Other explanatory factors, like a genetic predisposition to developing certain cancers, a less-than-ideal diet and sheer luck seem to come together, too.

Despite the search for other similar environmental health risk factors, from passive smoking to bacon butties, epidemiology has proven to be a fairly crude tool. Austin Bradford Hill, despite the success of his groundbreaking study, advised that epidemiological methods should be applied cautiously. He listed a variety of tests that should be applied to any putative connection between a risk factor and a disease, summarised in John Brignell’s book The Epidemiologists (see my review here: http://www.spiked-online.com/index.php/site/article/2446/).

1. Strength: Is the association strong enough that we can rule out other factors?
2. Consistency: Have the results been replicated by different researchers, and under different conditions?
3. Specificity: Is the exposure associated with a very specific disease as opposed to a wide range of diseases?
4. Temporality: Did the exposure precede the disease?
5. Biological gradient: Are increasing exposures associated with increasing risk of disease?
6. Plausibility: Is there a credible scientific mechanism that can explain the association?
7. Coherence: Is the association consistent with the natural history of the disease?
8. Experimental evidence: Does a physical intervention show results consistent with the association?
9. Analogy: Is there a similar result to which we can draw a relationship?

When these tests are applied rigorously, nearly all the postulated links between risk factors and harmful effects fall apart. For example, in the case of passive smoking and lung cancer, the association is weak, it is not consistent, exposure is usually assumed rather than measured, and so on. In other words, an association between passive smoking and lung cancer is either non-existent or too weak to worth worrying about.

Let’s now turn to this new field of ‘social epidemiology’. We can see in The Spirit Level that Wilkinson and Pickett’s claims fall flat on their faces when Bradford Hill’s tests are applied. The associations are not strong, rarely rising much above ‘apparently random’; other explanations are usually both available and more plausible than the idea of inequality leading to status anxiety; other researchers, simply by minor tweaks in the data, have seen the associations disappear; and the wide selection of social problems deemed to arise from inequality is a long way from ‘specific’.

As someone who would broadly see themselves as left-wing, I find there is a more pressing problem with The Spirit Level: that it dumbs down the fight between different groups in society over the carving up of the wealth we produce from a genuine struggle for material betterment to a psychological disorder demanding intervention from on high. Seeing inequality in terms of status envy is thus actually disempowering rather than liberating. But regardless of our attitude to the political question of wealth distribution, The Spirit Level is well worth criticising as bad epidemiology, too.

Share

Tweet

Damned statistics

Further to the interesting discussions in these pages about statistics and epidemiology, here is a recent letter published in The Spectator (not available online). The correspondent goes a little further in his criticisms than I would, nevertheless, it bears repeating...

Sir,

As an engineer, scientist and mathematician, I am completely in agreement with Rod Liddle's view  of statistics. I have long argued, with anyone who cares to listen (and plenty who don't), that any statistic should be completely ignored if it doesn't come with a 1,000-page manual describing exactly how it was arrived at and what it means. And that manual should be read, pored over and discussed for at least ten years.

The human race has survived and flourished for hundreds of thousands of years without statistics. We survive using superstition, lust, greed and envy. All that statistics have done is make life miserable by legitimising with dubious science the personal biases of busybodies. It tells us that we shouldn't drink, shouldn't smoke and shouldn't immunise our children with the MMR vaccine. Lie, damned lies and statistics.

Dr Tom Roberts
Derby

Share

Tweet

Die another day

In a recent post, I asked why the number of 'preventable' deaths seem to add up to more than the number of actual deaths. The answer, in a nutshell, is that you can only die once but your death can be prevented many times. In the comments, Carl V. Phillips explained how it works. I repeat his comments here in case you missed them, and because I have a feeling I'll be referring to them again in the future. 

While I certainly agree with you that some or all of the "blame the victim" body count estimates are high, be careful about this criticism -- it is not legitimate. Diseases and deaths always have multiple component causes, all of which can legitimately be called the cause (which is to say a necessary -- not sufficient -- cause of that death or disease at the particular time). 

So an individual may well die from smoking AND obesity AND eating junk food, and it is perfectly legitimate to say that had any one of these conditions been eliminated the death would not have occurred so soon. Someone who was killed by a drunk driver because a medical error prevented him from being saved in the hospital is a death due to alcohol use, motorized transport, and medical errors, so the causes add up to 3 for the one death. Thus, there is no reason to expect they would add up to the total. Indeed, they should add up to well more than the total if you have a rich enough list of causes.

I agree that this may not be how the man on the street interprets it, though I suspect if pushed that man would not actually be able to clearly state what he thinks it means. That is one of the problems with reciting raw scientific information to people who do not understand the science. Most people do not understand a relative risk statistic, but are barraged with them. But even descriptive statistics -- which most people probably think they understand at first blush -- are subtle. Nothing causes a fraction of a death -- it either causes it or not (see below). There is no obvious way to assign fractions.

As for bringing a death forward by merely one day, that is a fundamentally different question. And, yes, you could argue that a death that is accelerated by just one day by a particular cause should not be attributed to that cause for purposes of assessing public health statistics.

Wouldn't the word "factor" be more appropriate than "cause"? No. Cause is exactly the right word. It is the right word in the science of epidemiology (which is the source of this information) and is also the right word based on the usual intuitive definition. The latter, which is technically translated in epidemiology and most other sciences, is that in the absence of the particular influence, the particular outcome would not have occurred.

The word "factor" is one of those that often gets used because someone does not want to admit that they are making causal claims, even though that is exactly what they are doing. It doe not really mean anything. E.g., the phrase "risk factor", as used, has at least three or four different very distinct meanings, and therefore is worse than useless.

Following the above, it is easy to see why everything has multiple causes. Every death was caused by not only some disease, but also by birth of the individual in question, the evolution of humanity, the big bang, etc. This is part of why assigning fractions would not make sense, as noted above.

Angry Exile said:

And if a smoker goes outside the pub for a fag in winter and as a result of the alcohol passes out and subsequently carks it of hypothermia who gets to put their statistics up by one? ASH, Alcohol Concern or the Met Office?

If the individual would not have gone outside absent his smoking habit, then, yes, smoking caused his death, as did the alcohol that caused him to pass out, the weather that caused those to result in hypothermia, as well as, perhaps, his failure to put on a coat, his companion's failure to look for him after he disappeared, cutbacks on foot patrols by the police, and any number of other things.

The summary point is very simple: Everything has an infinite multitude of causes. For a particular outcome (e.g., death) we typically identify a particular set of them as the causes we are interested in intervening on (e.g., drug use, diet, medical tech), but there is nothing magical about that list. There is no reason to expect that those causes will not overlap in many cases, and once that list is made rich enough, overlap is inevitable.

Dr Phillips has recently started a new blog here.

Share

Tweet