On April 14, The United States Preventive Services Task Force concluded that women with an elevated risk of breast cancer – who have never been diagnosed with breast cancer but whose family history and other medical factors increase their odds of developing the disease–should consider taking one of two pills that cut that risk in half. The Task Force is an independent panel of medical experts who review the medical literature to estimate the pros and cons of preventive interventions. This is the same Task Force that in recent years raised questions about the benefits of mammograms in 40 to 50-year-old women, and PSA tests for men of all ages, tests that screen respectively for breast and prostate cancer. Despite the popularity of both of these tests, the Task Force concluded that their harms often outweigh their benefits.
The irony now is that with this report on breast cancer prevention pills, the Task Force has switched from rejecting something patients believed in to endorsing something most patients will reject.
The seemingly strange way the Task Force ping-pong’s between popular and unpopular recommendations is inevitable, because these kinds of recommendations must necessarily go beyond the medical facts – it is impossible to decide what preventive measures people need without making value judgments.
To understand the way facts and value judgments get mixed together in these kinds recommendations, let’s take a closer look at these breast cancer prevention pills.
For many years now, doctors have been prescribing tamoxifen as secondary prevention to women who have already undergone treatment for breast cancer, in an attempt to thwart any breast cancer cells remaining in their body. In women whose breast cancer cells express “estrogen receptors”, tamoxifen reduces the chance that this cancer will recur, by attaching itself to those receptors, in effect crowding out estrogen. In breast cancer cells, any estrogen landing on these receptors will spur that cell to divide and multiply. But when tamoxifen lands on these receptor sites, it does not stimulate cell growth.
Raloxifene is a close cousin of tamoxifen, which has primarily been used to treat women with osteoporosis. Like tamoxifen, it competes with estrogen for the attention of estrogen receptors. Also like tamoxifen, it slows down breast cancer by preventing estrogen from stimulating cell growth. As it turns out, both raloxifene and tamoxifen also have the strange property that when they collide with bone cells, they don’t fight against estrogen, but seem to mimic estrogen, and thereby improve bone health. (Because these drugs and sometimes act like estrogen and other times act like anti-estrogen, they are called selective estrogen blockers.)
Two very similar drugs, then. Both slow down breast cancer cells while stimulating bone cells. Each drug has been shown to cut the risk of a first breast cancer in half for women with a high risk of experiencing this disease. In what is known as the P1 trial, for example, women who faced an average five year risk of breast cancer of 6% saw that risk drop to 3%, if they took tamoxifen.
Sounds like a good deal, yes? Take a pill for five years, and cut your risk of breast cancer in half. But keep in mind, most women do not face a 6% chance of breast cancer in the next five years. Women with this kind of risk are generally retirement age or beyond, and usually have a bad combination of family history, early onset of menses, and late age of first pregnancy. It is these women the Task Force believes should talk with their doctors about whether to take these medications. And how did the Task Force reach this conclusion? By determining that for some women, at least, the benefits of these pills outweigh their harms. Before looking at this harm benefit ratio more closely, let’s put this Task Force conclusion into context.
The Task Force essentially produces three kinds recommendations, which I have taken the liberty to name.
- NADA: When it concludes that the harms of an intervention outweigh the benefits, the Task Force recommends that doctors and patients avoid the intervention. Think: ultrasound screening for pancreatic cancer.
- OUGHTA: When the Task Force concludes that the benefits outweigh the harms, it pushes to make the intervention standard of care. For example: routine screening for colon cancer in people 50 years or older.
- UP TO THE INDIVIDUAL: When the Task Force concludes that the benefits of an intervention potential outweigh the harms, depending on the patient’s individual preferences, it leaves the decision up to individual patients and their doctors to weigh. This is the recommendation the Task Force made both for mammography in 40 to 50-year-olds, and for tamoxifen and raloxifene to prevent breast cancer.
When making NADA or OUGHTA recommendations, the Task Force essentially makes its own value judgment. It looks at the risks and benefits of an intervention, and concludes that no sensible person could decide differently from the Task Force. In this third type of recommendation, however, the Task Force concludes that reasonable people could make different choices, based on how they weigh the risks and benefits of the interventions.
In the case of tamoxifen and raloxifene to prevent a first breast cancer, I expect the vast majority of women will conclude that the risks of the pills outweigh the benefits. In the past two years, I have collaborated with a team of researchers at the University of Michigan (led by Angie Fagerlin, a decision psychologist in their medical school), to help women decide whether to take either of these pills. We developed an Internet-based decision aid, a tool designed to help patients weigh the pros and cons of their medical alternatives. (I write about the history of decision aids in my book, Critical Decisions.) Our decision aid provided women with individualized estimates of their odds of developing breast cancer in the next five years. We only directed women to the decision aid whose risk was high enough to have qualified for the P1 trial.
In the decision aid, we described the benefits of both drugs – the reduced risk of breast cancer and the strengthening of their bones. We also laid out the risks – a very small chance of endometrial cancer, a slightly larger chance of heart attack or stroke, a modest risk of cataracts, and finally a very strong likelihood of experiencing menopausal symptoms such as irregular menstrual bleeding and hot flashes. We actually provided them with precise numerical estimates of these side effects, with pictures illustrating the risks to make them easier to comprehend.
Women pondered the pros and cons and concluded, almost unanimously, that the side effects of these drugs outweighed the benefits.
The decision whether to take tamoxifen and raloxifene is no doubt a personal one, and the right choice will vary depending on how a given person weighs the respective risks and benefits of these medicines. For a woman with an extremely high risk of breast cancer over the next five years – say 10% or more – cutting that risk in half might very well be worth the hot flashes and the chance of experiencing blood clots. But very, very few women faced a five year risk is highest.
Consider, instead, a woman with a 3% chance of developing breast cancer over the next five years. That risk is much higher than average – most women face a five year risk of less than 1% – but is it high enough to justify taking one of these pills? For such a woman, tamoxifen and raloxifene only reduce that risk by 1.5%. Over five years. Five years with possible hot flashes. For a cancer women have not experienced yet. These modest benefits simply do not loom large enough to interest most women in these pills.
You might wonder at this point whether our decision aid biased women against these medications. As a physician trained in behavioral economics, I’m constantly on the lookout for decision biases. In the case of our study, however, we designed our decision aid in a manner that allowed us to test for well-known behavioral economic biases. For instance, research has shown that when people face a choice between three options and two of the options are similar, they often opt for the more different alternative even if the other options are better. In other words, a person might believe that A > B, and A’ > B, but still choose B over A and A’, because they cannot decide between A and A’.
Aware of this problem, we created several different versions of our decision aid. In one version, we presented women with a choice between tamoxifen, raloxifene or no pill. Three choices in other words, two of which – the two medications – are quite similar to each other. In another version of the decision aid, we simply presented women with the choice between pill or no pill. We thought this simpler choice would increase women’s interest in these pills, by minimizing the difficulty of choosing between the two of them. But instead, this reframing of the decision did not increased women’s interest in either of these preventive medicines.
We also designed our decision aid to take account of another well-known decision bias, what are known as recency effects. When people learn about the risks of a medication and then learn about its benefits, that order of information leads them to look favorably upon the medication, because the information they remember best, the last information they receive, is about the pill’s benefits. By contrast, people who receive the same information about this medicine, but in the opposite order, like the pill less, because the last thing they learn about are the pill’s risks, and this information sticks in their minds. To make sure this recency effect was not influencing women’s decisions, we varied the order of information across women. We discovered that this did nothing to change their willingness to take either of these medications, mainly because whichever order women received information in, they did not like the idea of taking either pill.
If these pills are so unpopular among well-informed women, why would the Task Force come out in favor of them? It comes down to judgment. The Task Force concluded that a reasonable person could look at these risks and benefits and decide that the hot flashes and blood clots are acceptable prices to pay to reduce the chance of breast cancer. The majority of women don’t have to agree with this view for the Task Force’s recommendation to be correct. Even if only a small percentage of women decide these pills are worth taking, at least they have the freedom to make that choice. And at least they know that medical experts have concluded that such a decision is a reasonable one to make.
The same goes for whether to start mammograms before the age of 50 in women at normal risk of breast cancer. The Task Force never said that women shouldn’t start mammograms at this earlier age. They just said that it was a tough judgment call, and that some women, perhaps the majority even, might conclude that the harms of early screening – the anxiety caused by false negative tests, the pain caused by unnecessary biopsies – aren’t worth the modest benefits of screening at this age.
If the Task Force is going to leave all these tough decisions up to individual patients and their doctors, why should we care about their recommendations? For starters, you will have a hard time finding a more thorough and levelheaded evaluation of the pros and cons of these kinds of interventions. These people are very good at what they do. In addition, anyone reading through Task Force reports will be forced to recognize that science alone can’t make tough decisions for us. Ultimately, science can only provide us with the facts. The rest of us eventually need to make tough judgment calls. In effect, the Task Force is doing us a huge favor, by showing us which judgments are close calls, and which ones are no-brainers.
Peter Ubel is a physician and behavioral scientist who blogs at his self-titled site, Peter Ubel and can be reached on Twitter @PeterUbel. He is the author of Critical Decisions: How You and Your Doctor Can Make the Right Medical Choices Together.