"Illusion of Certainty" and why you should be skeptical of everything you hear or read about health care

Can a book make you smarter? Perhaps not, but "Illusion of Certainty" by Erik Rifkin and Edward Bouwer can certainly help you make smarter decisions--especially when it comes to personal health decisions.

The basic premise is that people crave certainty when it comes to matters of personal health and the established players (particularly doctors and pharmaceutical companies) have strong motivations to downplay uncertainty when it comes to their products and practices--particularly screenings and drug treatments. The result is that health benefits and risks tend to be stated as if they were "authoritative, definitive, and based on clear and unequivocal evidence." The authors call this an "illusion of certainty." When a doctor or press article touts a new drug or therapy, the message is often one that overplays its benefits and fails to properly communicate the associated risks. The result can be unnecessary procedures, over-medication and generally less-than-optimal decision-making by all parties. I should note that in addition to discussing human health "Illusion of Certainty" covers environmental concerns such as radon in homes, nuclear waste, invasive species, etc., as the same miscommunication of risk and uncertainty exists there as well. In this article we have focused on the book's applications to the medical world since this is the aspect that is most relevant to LifeTwo's readership.

Why is there so much uncertainty in health care?

Issues involving human health are complex. Establishing the causes of something like cancer is problematic because it doesn't appear immediately after exposure to a cancer-causing substance. This delay makes it difficult to know what really caused it. Complicating matters even further, the exposure to these substances is often at very low levels for long periods of time.

To try and conclude what is causing a cancer researchers generally take a statistical look at a large population (a so-called "epidemiological study"). But these populations are far from homogeneous and it is hard to track what they are being exposed to and to accurately measure the levels of such exposure. The statistics may be accurate but there is still a lot of guess work. The studies attempt to detect very small differences between the affected population and the control group. For example, you might have a 5,000 study group and another 5,000 in a control group with the difference might be just that a few more people of one group developed cancer than the other group. Even if the results are deemed to be statistically significant there remains the uncertainty whether all of the possible variables that could have affected the outcome were properly controlled for (smoking, drinking, age, etc.). Every step of the way, from the design of the study, to how it is conducted and the interpretation of the results, are areas of intrinsic uncertainty. The authors note that such studies are also time consuming and expensive making them difficult to repeat.

Animal studies are another commonly used tool in scientific research. However, like epidemiological studies, animal studies have built-in uncertainty--a point that isn't always made clear when the findings are touted in the media. The two biggest areas of uncertainty with animal studies are: 1) animals and humans are fundamentally different; and 2) animal tests are often set at artificially high exposure levels (that is high enough to have an effect but not so high as to be lethal to the test animal). Though it was not used as an example in the book, many will remember the saccharin ban in the 1970s based on the controversial high-dosage rat experiments. In this case "high-dosage" was the equivalent of "hundreds of cans of diet soft drink a day for a lifetime." The extremely high dosage in the study had the affect of skewing the results to make saccharin look like a cancer causing substance. Subsequent tests on saccharin failed to support this contention and ultimately a moratorium was placed on the ban. In short, animal testing is useful and relevant but the risks inherent in the extrapolations can't be ignored (as they were with the saccharin ban) and conclusions must be seen in the light as containing uncertainty.

Because of the difficulty that often exists in establishing a cause and effect relationship in the medical world, when researchers find a statistical association between a disease and a suspected agent, they call the agent a "risk factor." But there is a big difference between a cause and a risk factor and this distinction is often not properly communicated. The authors spend a great deal of time on the relationship between high cholesterol and heart disease. High cholesterol is a risk factor and not a cause of coronary heart disease. This distinction is important when it comes to possible interventions. Billions of dollars are spent on drugs to reduce cholesterol yet studies have shown that 99.9% of people who take such drugs will not benefit in terms of mortality. Stated another way, for every 1000 people who take the cholesterol-lowering statin and modify their diets accordingly, 999 will not see any change in whether or not they develop coronary heart disease.

One other complicating factor is that some uncertainties turn out to be beneficial. It appears, and completely unexpected, that the takers of some anti-cholesterol drugs might in fact be lowering their chance of developing Alzheimer's disease due to the drug's anti-inflammatory effects. This is a case where an uncertainty benefits the taker. The point isn't whether it is good or not but that the drugs interact with our bodies in ways that the makers hadn't expected. Sometimes we end up discovering unexpected beneficial effects (as in the case with Alzheimer's) and sometimes we discover unexpected detrimental effects, as was the case with Vioxx (discussed below).

The problems with relative risk versus absolute risk

If you are trying to make a significant health care decision, the proper information you would want to know is the absolute risk of the proposed action. For example, let's say you are making a choice concerning taking a certain drug. You should want to know is the likelihood you will prolong your life by taking the drug versus not taking it. Unfortunately, absolute risks are rarely communicated and this is a travesty. Instead the relative risks are communicated. This might seem like a semantic difference but unfortunately relative risk profiles often distort potential benefits as the following example will illustrate.

Let's say a doctor does a blood test and determines that you should take an expensive drug with some side effects but it will reduce your chances of dying by 50%. Should you take the drug? Of course, who wouldn't want to reduce their chances of dying by 50%? But what if your chances of dying had been 1 in 1,000, and now they were going to drop to 1 in 2,000. In other words, you had had a 99.99% chance of living before and now that's been increased by .005? How about if this drug was expensive and had potential side-effects? Even though the quoted relative risk (50%) and the absolute risk (.005% improvement) were both stated accurately, they certainly tell a very different story. Less people would take an expensive drug with possible side-effects if their doctor told them that there was a 99.995% chance that it would not benefit them in any manner. Perhaps this is one reason that drug therapies are not framed in this way.

You might consider the numbers that used in the above example to be extreme, however let's take the case of mammograms, which have been heavily studied by the science and medical community. According to the authors, the absolute risk reduction of having a mammogram is .1%, that is a .1% reduction in deaths among women who have had biennial mammograms. Not much of a benefit when considering the other risks of intervention including "biopsies, radiation, false positives, and false negatives."

Every drug has risks from simple aspirin to the more current example of Vioxx, a topic covered extensively in the book. While this is a still-controversial subject, the fact is that this FDA-approved, doctor prescribed drug carried risks that may not have been fully factored into the decision making process. (I would like to note that I took Vioxx myself for a month for pain associated with an injury and at no point did my doctor ever share with me the potential risks associated with it. Granted the box carried numerous legalese warnings, I only got to these long after the decision had been made to start taking the medication. While it's certain that my doctor was not aware of all of Vioxx's risks, it would have been easy for him to have had the discussion with me about possible side effects of any prescribed drug (particularly a relatively new drug) versus the potential benefits. In this case since the benefits was relief of mild pain, I might very well have decided to have forgone taking Vioxx. The point isn't whether or not I should have taken Vioxx but that the conversation about risk was not raised by either of us.)

Where does this leave us?

This final point is the key message of the book. It is not that doctors are bad (they are not) or that pharmaceutical companies are evil (they are not). There is no conspiracy or malfeasance underlying the medical profession to drive up costs or make you undergo unnecessary or harmful procedures. The authors simply believe that the self-interests of all parties (including patients) leads to a bias towards over-stating benefits and far too often the risks and uncertainties are not adequately factored into the decision-making process. Stated another way, we know far less about human health than we like to admit and this leads to overconfidence in what we think we know. "The Illusion of Uncertainty" tells you how to overcome this making it a highly recommended read.

Amazon link:The Illusion of Certainty: Health Benefits and Risks