A Bitter Pill?

The risks of third-generation contraceptive pills have been much in the news. But assessing risk can be a tricky business.

Twenty-nine years ago, I was about a week into my first job as a doctor, as a House surgeon in orthopaedics at Guy’s Hospital in London. I had not had time to get to know the patients under my inexpert care and was on a very steep and stressful learning curve. Just before three in the afternoon as I was doing my post-operative round, my bleep went mad, warning me of some dire emergency. I hurried to the men’s ward to find an anaesthetist and another doctor working hard to revive a man who had suddenly called out and then fallen back pulseless. He had had a knee operation the week previously, before I had arrived on the job and I scarcely knew his face, let alone his name. My puny contribution to the efforts of the experts were to no avail and his circulation could not be restored. His wife was waiting outside and it fell to me to tell her that he had had some sort of heart attack and had died. You will not be surprised that it is her face rather than his that I remember.

Twenty-five years later, another patient in my charge, a young student of twenty, had puzzled two other doctors by her sudden attacks of loin pain over several weeks, first on the right side, then on the left and then on both sides. By the time she came to see me, she had had numerous blood tests, an emergency kidney x-ray and a chest x-ray. They had given no clues as to the cause of the pain. She had been seen in the Accident and Emergency department of the local hospital in the middle of the night. She had been seen by a colleague of mine at the weekend. The attacks continued, but in between them, she had been well enough to go out on Territorial Army manoeuvres. When she saw me, the pain was bad enough to make her catch her breath. Apart from severe muscle spasm and a raised pulse rate I could find no abnormality. I noted that “something strange is going on here.” I arranged for her to see a medical specialist urgently. Before she could keep the appointment, while walking from the library to the cafeteria, she fell pulseless to the ground and her circulation could not be restored.

Venous Thromboembolism

Post-mortem examinations showed that what both these unfortunate people had in common was deep venous thrombosis and massive pulmonary embolism, first elucidated by the great German pathologist Virchow, well over a hundred years ago. Venous thromboembolism, as it is often called, VTE for short, has vexed doctors ever since.

Most of you will know that blood outside the body clots. It is fortunate that while in the blood vessels it does not usually clot unless the vessel is damaged and then clotting is indispensable. The damage results in the release of substances that initiate a cascade of biochemical reactions that result in a tangle of a fibrous protein called fibrin, mixed up with platelets and red blood cells. This plugs the hole in the vessels and may plug the whole vessel. Virchow’s triad has stood the test of time as an analysis of what happens with abnormal clotting. He observed that the main influences are disturbances of the vessel wall, things that change the dynamics of the blood flow and things that change the components in the blood that initiate clotting. Let us return to my two unfortunate patients.

Contributing factors

In the case of the first, he had had an operation on a lower limb and his limb had been immobilised in plaster. The stress of an operation in itself increases the clottiness of the blood, muscle action would have been absent during the operation and reduced after it, leading to sluggishness of flow. We can imagine what happened in the veins of his legs with clotting extending from a vein, often starting in a valve pocket, and eventually extending into the main vein of the leg and thence even into the main abdominal veins. Eventually, a large piece broke off, was pumped through the right side of the heart, blocked the pulmonary trunk and brought circulation suddenly and permanently to a halt.

The second person was a fit and healthy young woman. Were there any known predisposing factors? Had she perhaps an inherited predisposition for her blood to clot easily? Her grandfather had had an uncomplicated deep vein thrombosis after an operation on his leg, but this is a known risk. Had she had any injury? Well, she had sprained her ankle on Army manoeuvres four weeks before she had started to get symptoms. Following her death, her Lt Colonel investigated this and there is nothing in his detailed account to suggest that she had anything other than a minor sprain.

She had not sought any medical attention for it. And three months previous to her death, she had started to take the contraceptive pill.

Popular Pills

The first major trials of the contraceptive pill took place in the late 50s and it quickly became very popular because of its ease of use and near 100 percent efficacy. It contains two hormones, progestogen to fool the pituitary gland into thinking the taker is pregnant, so that it has no need to send signals to an ovary to release an egg; and oestrogen to give cycle control so the taker can have a monthly pseudo period. The oestrogen component also reinforces in some way the contraceptive efficacy of the pill. The first pills used about six or seven times as much oestrogen hormone as modern pills and the first case histories suggesting an association between the pill and VTE appeared in 1961. A report to the British Medical Research Council in 1967 showed a clear link between pill use and VTE and further papers from Britain, Sweden and Denmark in 1970 concluded that the risk of thrombosis was linked to the oestrogen dose. By this time, the oestrogen dose was down to about 80 micrograms from an initial 180 to 200 micrograms and it was then recommended that the level at which risk became unacceptably high was about 50 micrograms of oestrogen.

Absolute Risk

Early case-control studies suggested that the risk of VTE was between two and eleven times greater in pill-taking women and the absolute risk was between three and six episodes per ten thousand women per year. A large study of 65,000 women in Seattle in the early eighties suggested a relative risk of 2.8. Healthy women not on the pill seem to have an absolute risk of about one per thirty thousand women per year, so the risk in pill-taking women is about one per ten thousand per year. I should make it very clear at this point that we are not talking about risk of death here, but only of deep vein thrombosis. If we take the worst figure, about one in fifty of people who get deep vein thrombosis will have a fatal pulmonary embolus, so the risk of death from this per year of pill use is about one in a quarter of a million per year. However, a further proportion of people who get DVT will have permanent damage to the veins of their legs and in some, multiple small clots breaking away will cause permanent damage to the circulation of the lungs. Pills containing progestogen on its own do not seem to have an increased risk of VTE, but are less effective and periods are irregular, so they are less popular.

Late in 1995 media reports began to appear that so-called third generation oral contraceptive pills carried a greater risk for VTE than the older pills. The third generation pills contain the progestogen hormones gestogene or desogestrel, which can be thought of as designer hormones. The state of the art of drug synthesis has advanced to the point where the properties of the hormones can be to some extent predicted from their structure, and vice versa, and these two hormones have fewer male-hormone like effects (such as causing acne etc) and less effect on fat and carbohydrate metabolism. They were promoted as being safer for the arteries, where blood clots also occur, as in heart attacks and strokes, and better for the skin. It was difficult for doctors to advise their patients as the papers on which the media reports were based had not been published, but three eventually appeared in the scientific medical journal, the Lancet, of December 16, 1995 and another one in the British Medical Journal in January 1996. They are not easy reading and I think it is safe to say that those most likely to prescribe the contraceptive pill, general practitioners, do not as a rule read the Lancet.

Literature Reviews

Fortunately for us, there was no shortage of secondary articles, and one appeared in the Ministry of Health’s Prescriber Update in February 1996. I have read the original papers and can say that the article is an excellent and balanced summary that accurately reports the findings of the originals and correctly reflects the views of their authors. The risk of VTE in second generation pills is less than previously reported. A healthy woman who is not a current user of the pill has an annual risk of VTE of about one in 26,000. A woman who takes the modern second generation pill has an annual risk of about one in 6 to 10,000. Someone who takes a third generation pill has an annual risk of between one in 3,570 and one in 5,000, so the risk of VTE in third generation pills is roughly twice that of second generation pills. The authors echo a Lancet Leading Article in stressing “that further independent study is necessary. The interpretation of the small increase in risk of VTE must be weighed against a possible decrease in the risk of other cardiovascular endpoints. Until the relative risk of other important health outcomes such as stroke or coronary artery disease.. is clarified, there is no sound basis for recommending any change to current contraceptive practice.”

British Response

In Britain, the response of the Ministry of Health was to advise that the third generation pills should not be used by women with additional risk factors for VTE and that doctors should prescribe them only for women who were prepared to accept the increased risk and who were intolerant of other combined pills. This led instead to widespread flight, not only from third generation pills but from contraceptive pills in general, with at least anecdotal reports of many accidental pregnancies and an increase in abortions. This may seem very strange to aliens like us who habitually think logically, but you will not be quite so surprised if I said that following a recent total solar eclipse in Britain, people sought advice as to whether viewing the eclipse on television could have caused damage to their eyes…

In New Zealand, the response was more muted. Doctors took the advice from the Health Department at its face value and received it as a reminder to check for risk indicators when prescribing the pill. In December 1998 we were told between January 1993 and June 1998 there had been six deaths from VTE in women taking the third generation pills whereas between 2.2 and 3.7 deaths in this time would have been expected. Of course, with such small numbers the figures could readily be accounted for by random variation and the article pointed out results from epidemiological studies are more reliable than Committee for Adverse Reactions Monitoring data. The waters were by now quite muddied and Sandra Coney jumped into them last year to further stir them up when the headline of her column in the Sunday Star Times read “Who’s to Blame for Pill Deaths?”

“My question is”, she wrote, “who is accountable for these deaths? Is it the drug firms who raised the spectre of legal action against the Ministry when it planned to issue warnings when the risk of these pills were first known?

“Is it the medical groups who pressure the Ministry by saying they would disassociate themselves from the advice? Or is it the various officials of the Ministry of Health who caved in under the pressure, selling New Zealand women down the river?”

She pointed out that “an astounding 75 to 80 percent of women” using the pill in New Zealand were on third generation pills. “This”, she said, “tells us something about the too-cosy relationship between doctors and drug companies in New Zealand.”

According to Coney, the Adverse Reactions Committee had advised doctors should preferentially prescribe the older second generation pills, but the pill manufacturers threatened the ministry with legal action and had “bombarded GP’s with dossiers contradicting the studies” and the Royal New Zealand College of Obstetricians and Gynaecologists said they would publicly dissociate itself from the advice. The Family Planning Association, another body that might be thought to have some expert knowledge too, “went about saying the studies…were affected by biases so that the results couldn’t be trusted.”

Rhetorical Questions

At the end of her article she asks questions that might be thought to be mildly rhetorical given the general tenor of the article. Of women using third generation pills she asks:

Are they warned of the risks?

Do they know that they could reduce their risk by using older forms of OC’s or even eliminate it by using another method?

Have their doctors explained to them the symptoms of blood clots?

Do they know they are at additional risk if they are immobilised because of illness, injury, surgery or a long plane flight?

What must we poor benighted doctors do as dossiers rain down about our ears from drug companies, as sticks labelled “informed consent” are waved at us by the Health Commissioner, as our expert bodies display their ignorance by echoing the advice given by other expert bodies throughout the world?

The publicity has had a beneficial effect in making us more careful in assessing people’s suitability for the combined pill, but it may have led us to practice a more defensive style of medicine. In a consultation I have about twenty minutes to impart quite a lot of information and know that seventy percent of what was absorbed at the time will have been forgotten by the end of the day.

Reduced Risk?

Could they indeed reduce their risk by using an older pill? One expert, Walter Spitzer, commenting in the Lancet on a World Health Organisation scientific summary writes “The summary of the conclusions plays down the controversies that have raged for the past two years about differences between second and third generations of oral contraceptives in risk of VTE. It also properly emphasises the rarity of all the three serious side effects.” He went on to point out that there is at least some evidence that third generation pills may have a smaller incidence of heart attacks in young women and that the order of risk for VTE and heart attacks is about the same. What we may gain on the swings of reduced VTE we may lose on the roundabouts of heart attacks.

If we look at risks in isolation we may reach conclusions that are both correct and yet which are absurd. Let us suppose that a sexually active woman decides that the risk of OCP is too great and so she uses no contraception at all. In a year she has a seventy percent chance of getting pregnant. During the pregnancy she has a one in 1600 chance of getting a DVT and during the week in which the baby is born a risk of about one in six hundred, roughly ten times greater than the worst risk for third generation pills – if the studies have reached a correct conclusion.


Every third year medical students knows the symptoms of blood clots. You get a painful swollen leg with tender calves. Unfortunately for us poor benighted doctors and unfortunately for our patients, most people with DVT don’t have these symptoms and most people with these symptoms don’t have DVT. Pulmonary embolus is even more difficult to diagnose without high tech help – except in the post-mortem room. Oh, if only the drug firms would distribute free retrospectoscopes instead of raining dossiers of propaganda on me! Still, I do tell patients, orally and in writing, about painful swollen legs; and chest pain with shortness of breath and spitting of blood; and about sudden loss of vision or use of limbs. If I set a test at the end of a week not many would pass. But I’ll be OK when the Health Commissioner comes calling.

Do I really have to tell them about additional risk if they are immobilised because of illness, injury, surgery or long plane flights? My elder daughter flew to Britain a few months ago. Would I as a doctor expect her doctor to suggest that she stop the pill (I don’t know whether she’s on it. It’s none of my business). First of all, long distance flying carries a risk of DVT that is independent of being on the pill, so I should also expect him also to warn her about the risk of cosmic rays at high altitude, the risk of side-stream smoke in the cabin, the risk of acquiring hepatitis A from eating airline food and so on almost ad infinitum. In any case, it’s a risk that she would run for a few days at most, so it would have to have a very high annual risk indeed to be of comparable significance to the annual risk from the pill.

To Sandra Coney and others the issues seem to be simple. One sort of pill carries twice the risk of another sort. Drug firms have bullied the Ministry of Health and have muted the voice of doctors and other experts by stopping their mouths, not with gold, the preferred substance for scoundrels down the ages, but misleading dossiers. A risk is a risk is a risk and no one should have to run it if it can be reduced. No matter that people vastly better informed and experienced in analysis of statistics comment about the “lack of clinical importance and public health significance of VTE” with its “very low absolute rate of occurrence, low morbidity and low case-fatality.” Nothing must get in the way of a good story.

The Placebo Effect

Many people will remember Dr Bill Morris’s entertaining autobiographical talk at the last Skeptics’ conference in Wellington. From his presentation, we extract this discussion of what is still the most persistent and potent medical effect known to the human race.

The word “placebo” may or may not be recognisable to you as being of Latin origin, from the verb placeo, placere, to please, and placebo is the first person singular of the future indicative tense, or “I shall please.” Its first recorded use in the English language was in 1225 in reference to Vespers in the Office for the Dead, and the word was derived from the first word of Psalm 116 v 9 “Placebo Domino in regione vivorum.” This is usually translated as “I shall walk before the lord in the land of the living,” but as the Vespers for the dead was in effect a request for the dead to intercede with God for the benefit of the living, it is better translated as “I shall please or intercede with God on behalf of those in the land of the living.”

By 1386 it had appeared in Chaucer’s Merchant’s Tale to mean a flatterer and by 1811 it had acquired one of its modern senses as “…an epithet given to a medicine adapted more to please than to benefit the patient.”

It may surprise you to learn that it was as late as 1938 before the word appeared in its other modern sense, that of a dummy medication used as a control, and probably the first well documented randomised placebo controlled trial was that of streptomycin for the treatment of tuberculosis of the lung, in 1948.

The Medical Research Council pointed out that the natural history of tuberculosis of the lung was so variable that “evidence of improvement or cure following the use of a new drug in a few cases cannot be accepted as proof of the effect of that drug” by contrast with tuberculous meningitis which was invariably fatal without treatment.

The placebo treatment was bed rest alone, whereas the streptomycin group received both bed rest and streptomycin. In this instance the patients obviously knew they were getting streptomycin as it has to be given by injection, but the progress of the disease was followed on chest X-rays which were assessed without knowledge of which treatment the patients had received.

Ethical considerations did not apply, as the only possible alternative treatment at that time was bed rest, and in any case only limited amounts of streptomycin were available. Nearly forty five years on, it is difficult to accept that bed rest alone was perceived to be an effective treatment.

By 1950 the Journal of Clinical Investigation wrote, “It is customary to control drug experiments on various clinical syndromes with placebos, especially when the data to be evaluated are chiefly subjective.” and so by this date, the use of dummy medication in drug trials was firmly established.

It soon became clear that many people reported side effects or improvement when receiving placebos in trials and it soon became possible, though perhaps not useful, to say that the placebo effect was that which all treatments have in common. Perhaps more useful is to combine the two definitions and say that it is a non-specific effect of a treatment attributable to it but not to its pharmacological properties.

Any form of treatment can act as a placebo, and the strength of the reaction varies with the supposed potency of the treatment so that a capsule is better than a tablet, an injection is better than a capsule, an injection that stings is better than one that doesn’t and an operation is even better than an injection that stings.

In 1939 it was suggested in Italy that tying off the internal mammary arteries could greatly reduce the pain of angina pectoritis. The operation eventually became fashionable in the USA with quite spectacular results. The patients said they felt better and there was objective evidence to support this in that they could walk further and their consumption of angina pills decreased.

Eventually a double blind controlled study was done in which half the patients had their internal mammary arteries tied and the other half simply had them exposed without tying them. Neither the patients nor those who assessed them knew until the study was completed who belonged to which group. It turned out that ligation had no greater effect than the dummy operation. Since ligation of the internal mammary arteries was quite a major operation with potential for harm as well as good, and there was genuine doubt that it was useful, not only was the trial ethical, but it would have been unethical not to have done it.

Placebos can also cause toxic effects just like those of an active drug, and in a study of 25 patients given placebos, ten reported sleepiness, nine palpitations, eight irritability, five weakness with a fall in blood pressure of more than 20 mm of mercury, four reported diarrhoea, two collapse and two itching. Three of the patients also developed dependence on the placebo (lactose) and had withdrawal symptoms when it was stopped.

Now one suspects that if we stopped people in the street at random they might report a similar rate of these symptoms, and at the 1992 Skeptics Conference, on asking for a show of hands, I found that the proportions enjoying the symptoms listed above were greatly exceeded.

As soon as the placebo effect became clear, investigators began to look for factors which might identify the placebo responder in the hope that eliminating them from the studies would make the data much clearer.

A variety of psychological studies has been done but there are so many inconsistencies among the results that one can conclude that there is no single personality trait that characterises the positive placebo reactor, with the possible exception that stress or anxiety has been rather consistently associated with placebo reactivity. Expectation has been cited as a possible mechanism.

In one study, experimenters were told that their rats had been bred especially either for intelligence or dullness, although in fact all the rats were from the same genetic strain. The experimenters then performed learning experiments on the rats and obtained results that conformed to their expectation.

Brody cites this as possible support for the idea that if experimenters can somehow communicate their expectations of the rats’ behaviour to the rats “It seems reasonable to assume that physicians can unknowingly communicate their expectations and attitudes to the patients, altering the patients’ therapeutic outcomes as a result.” My own view is that old Procrustes is at it again, altering the accuracy of the experimenters’ observations rather than the rats’ behaviour.

Theories of placebo action have been largely psychological or psychoanalytical, but a reductionist like myself sees behaviour, feelings, thoughts and so on in terms of as yet poorly understood physico-chemical activities in the brain and peripheral nervous system, rather than as something happening somewhere in a bubble marked “psyche.” A little support for this view comes from a randomised double blind placebo controlled study by Levine and others into the mechanism of pain relief following extraction of impacted lower wisdom teeth. A third were given naloxone, a substance which is believed to inhibit the action of naturally occurring pain relieving substances in the brain called endorphins, a third were given a placebo and a third were given morphine. Those given naloxone reported significantly more pain than those given the placebo.

Levine hypothesised on this basis that placebo pain relief is mediated by endorphin release, but as Skrabanek pointed out later, they did not test their hypothesis by actually measuring endorphin levels, and in any case the results were exactly what might be expected if the naloxone were acting as a placebo itself. The paper was also severely criticised by Korczyn, but nevertheless it continues to be quoted quite extensively as “demonstrating” that pain relief by placebos is mediated by endorphins, a claim, incidentally, that is also made for pain relief by acupuncture.

We simply do not know why about thirty percent of patients experience relief of symptoms when given a therapy that cannot be expected to have any effect. In a sense, the history of medicine up to about 1950 is largely the history of placebos. We may find it amusing to look at some of the truths of yesterday which are the falsehoods of today, like the fashion for enemata in eighteenth and nineteenth century France. Sometimes enemata even of tobacco smoke were administered, and while we cannot feel entirely confident that the Tobacco Institute would disapprove of this, we can feel sure that many of our present day medical practices will appear stupid and ignorant to our great grand-children.

For the last forty five years we have had the means to set a limit to our errors, and yet colleagues tend to set store by anecdotes and case series which are in truth little better than a succession of anecdotes. The results of poorly designed case control studies continue to be accepted without proper caution. If physics is the queen of the sciences, then the randomised double blind placebo controlled study is the queen of medical investigation, though for events that occur relatively rarely, cohort studies and case control studies are inevitable second and third best choices.

People, including I am sad to say, doctors, have said to me “What does it matter whether a treatment is a placebo or not as long as it works? Surely the thing is to cure the patient and when you cannot cure, to comfort.” I can certainly agree with the aim, but not that it does not matter how we do it. If we do not make sure of the truth then we shall not be able to separate the wheat of science from the chaff of falsehood, and as Berthold Brecht put it, the aim of science is not to open a door to infinite wisdom, but to set a limit to infinite error.