Any "news" about breast cancer generates press coverage, and it is especially fun for the reporters if they can blame the victims' own unsavory behavior, like smoking. The latest such story (e.g., here) has so many interesting aspects that it is difficult to cover all of them.
First, why? The study reported a slightly elevated risk of breast cancer for some smokers. I can understand why a student might want to crunch the numbers on this as a project, but why is it news? We already know that smoking is bad for you, and even if the press release about this were accurate, it would not change that much. Who, exactly, is on the fence about smoking, thinking "if it were a bit less healthy I would quit, but I am fine with it now." Moreover, such a discovery would not really tell us it is a bit less healthy than we thought: We already know the overall longevity and health outcomes for smokers vs. nonsmokers, so all this would do is explain some of the total impact.
On the same note, this result should not be touted as news because overall the studies of smoking and breast cancer have found no measurable effect. Smoking is such an intense and complicated exposure that there are few health outcomes that it does not effect (most of the effects are bad, a few are good), so it seems reasonable to think that breast cancer is affected to some extent. But we have a lot of evidence that the effect is small and not really worth mentioning compared to the other risks from smoking and other causes of breast cancer. As with any well-studied small effect, there are some studies that show a positive association like this one (that is, it looks like the exposure causes the disease) and some that show a negative association (that is, it looks like the exposure might protect against the disease).
Second, this study, in itself, fits that description. That is, it finds that smoking is sometimes associated with breast cancer, but sometimes protective against it. Did you miss that bit in news stories? Unless you live in Los Angeles, where the paper sort of pointed that out, that is not too surprising. It turns out that the study estimated a positive association with smoking in one's youth, but a protective association with smoking post-menopause. The two estimated associations are about the same size, and they come from the same study, so if you are going to believe one, you are obliged to believe the other. That is, if you are going to say, "here is a good reason for young women to not smoke," you need to also say, "here is a reason post-menopausal women might want to start smoking" (or, perhaps more realistically, "if you are still smoking at menopause, here is a reason to not quit"). It is difficult to imagine a more patent form of intellectual dishonesty than picking only the result that supports your political views and ignoring a comparable point from the same source just because you do not like it.
Of course, it would be better to say "this does not imply it is good to smoke post-menopause because the other risks outweigh any benefit this result implies", but that would require admitting, "this really does not change the arguments against smoking either, since any effect is quite small compared to the known risks." The accurate interpretation is that both estimated effects are small, and are really just curiosities. Moreover, in the greater context of what is known about smoking and breast cancer, this one study should change our beliefs very little anyway. The study should have been treated like a new study in astrophysics – something for technical experts to use, and perhaps to be published in the news for people who like to read science news for amusement, but not run as a headline as if we should really care about it.
Third, learned readers should notice that this result came from the Nurse's Health Study. That does not make it wrong, but it should make you suspicious. That long-running cohort study is largely responsible for the opinion that many experts in epidemiology have, as I noted earlier, that nutritional epidemiology is mostly junk science. That study collects zillions of variables about its participants every year, so it is possible to study countless topics – or dredge for associations among the data and publish whatever randomly appears. A few researchers who have violated the code of omerta that surrounds the study have told me how the researchers who control it assign junior researchers particular topics and forbid publication of results that contradict claims made by previous researchers in the group. These factors, plus the fact that the data is a closely guarded secret, make this study the epitome of how epidemiology often violates the fundamental norms of proper science.
Just to mention a second story which is not actually about breast cancer, though you would probably not know that from the headlines, it was discovered that breast implants apparently increase the risk of a rare cancer. This was only discovered because the cancer is so extremely rare that just a few cases of it – there are about 60 cases reported among the perhaps 10 million women who have had implants – is enough of an increase to be noticeable. It turns out that this is not even breast cancer, but a lymphoma (a cancer of the immune system) that occurs in the scar tissue that is created.
To their credit, neither the U.S. FDA who reported this, nor any other reputable actor, is suggesting that this ought to affect anyone's decision. This cancer is easy to treat, and the risk from the anesthesia for the surgery completely swamps it. Indeed, it is down in the range of the risk from the car trips required to arrange and get the surgery.
The major thing that these stories do have in common is that they are hyped beyond the minor technical curiosities that they are, and that neither one represents a new reason to avoid the activity that causes the risk. There are plenty of arguments against those activities, of course, and those downsides must be balanced against the advantages. Not surprisingly, the news acknowledged this for cosmetic surgery, but was steadfast in avoiding any mention of why people choose to smoke. The image of a woman with a cigarette used to be considered sexy, but this was changed as a result of a social engineering campaign. Perhaps if anti-obesity campaigns decide to try to create a similar backlash against double-D-cups being considered sexy, future stories about minor health risks from breast implants will be spun as definitive reasons why we should stamp out that practice, regardless of individual preferences.
First, why? The study reported a slightly elevated risk of breast cancer for some smokers. I can understand why a student might want to crunch the numbers on this as a project, but why is it news? We already know that smoking is bad for you, and even if the press release about this were accurate, it would not change that much. Who, exactly, is on the fence about smoking, thinking "if it were a bit less healthy I would quit, but I am fine with it now." Moreover, such a discovery would not really tell us it is a bit less healthy than we thought: We already know the overall longevity and health outcomes for smokers vs. nonsmokers, so all this would do is explain some of the total impact.
On the same note, this result should not be touted as news because overall the studies of smoking and breast cancer have found no measurable effect. Smoking is such an intense and complicated exposure that there are few health outcomes that it does not effect (most of the effects are bad, a few are good), so it seems reasonable to think that breast cancer is affected to some extent. But we have a lot of evidence that the effect is small and not really worth mentioning compared to the other risks from smoking and other causes of breast cancer. As with any well-studied small effect, there are some studies that show a positive association like this one (that is, it looks like the exposure causes the disease) and some that show a negative association (that is, it looks like the exposure might protect against the disease).
Second, this study, in itself, fits that description. That is, it finds that smoking is sometimes associated with breast cancer, but sometimes protective against it. Did you miss that bit in news stories? Unless you live in Los Angeles, where the paper sort of pointed that out, that is not too surprising. It turns out that the study estimated a positive association with smoking in one's youth, but a protective association with smoking post-menopause. The two estimated associations are about the same size, and they come from the same study, so if you are going to believe one, you are obliged to believe the other. That is, if you are going to say, "here is a good reason for young women to not smoke," you need to also say, "here is a reason post-menopausal women might want to start smoking" (or, perhaps more realistically, "if you are still smoking at menopause, here is a reason to not quit"). It is difficult to imagine a more patent form of intellectual dishonesty than picking only the result that supports your political views and ignoring a comparable point from the same source just because you do not like it.
Of course, it would be better to say "this does not imply it is good to smoke post-menopause because the other risks outweigh any benefit this result implies", but that would require admitting, "this really does not change the arguments against smoking either, since any effect is quite small compared to the known risks." The accurate interpretation is that both estimated effects are small, and are really just curiosities. Moreover, in the greater context of what is known about smoking and breast cancer, this one study should change our beliefs very little anyway. The study should have been treated like a new study in astrophysics – something for technical experts to use, and perhaps to be published in the news for people who like to read science news for amusement, but not run as a headline as if we should really care about it.
Third, learned readers should notice that this result came from the Nurse's Health Study. That does not make it wrong, but it should make you suspicious. That long-running cohort study is largely responsible for the opinion that many experts in epidemiology have, as I noted earlier, that nutritional epidemiology is mostly junk science. That study collects zillions of variables about its participants every year, so it is possible to study countless topics – or dredge for associations among the data and publish whatever randomly appears. A few researchers who have violated the code of omerta that surrounds the study have told me how the researchers who control it assign junior researchers particular topics and forbid publication of results that contradict claims made by previous researchers in the group. These factors, plus the fact that the data is a closely guarded secret, make this study the epitome of how epidemiology often violates the fundamental norms of proper science.
Just to mention a second story which is not actually about breast cancer, though you would probably not know that from the headlines, it was discovered that breast implants apparently increase the risk of a rare cancer. This was only discovered because the cancer is so extremely rare that just a few cases of it – there are about 60 cases reported among the perhaps 10 million women who have had implants – is enough of an increase to be noticeable. It turns out that this is not even breast cancer, but a lymphoma (a cancer of the immune system) that occurs in the scar tissue that is created.
To their credit, neither the U.S. FDA who reported this, nor any other reputable actor, is suggesting that this ought to affect anyone's decision. This cancer is easy to treat, and the risk from the anesthesia for the surgery completely swamps it. Indeed, it is down in the range of the risk from the car trips required to arrange and get the surgery.
The major thing that these stories do have in common is that they are hyped beyond the minor technical curiosities that they are, and that neither one represents a new reason to avoid the activity that causes the risk. There are plenty of arguments against those activities, of course, and those downsides must be balanced against the advantages. Not surprisingly, the news acknowledged this for cosmetic surgery, but was steadfast in avoiding any mention of why people choose to smoke. The image of a woman with a cigarette used to be considered sexy, but this was changed as a result of a social engineering campaign. Perhaps if anti-obesity campaigns decide to try to create a similar backlash against double-D-cups being considered sexy, future stories about minor health risks from breast implants will be spun as definitive reasons why we should stamp out that practice, regardless of individual preferences.
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.