Newsflash: Exercise is good for your health. So is cutting down on high-fat meats, and on sugar and refined carbs such as white flour and white rice. Are you surprised? I'm not, personally, but it sounds as if a lot of people out there are finding these notions hard to believe, because we keep seeing health headlines that announce the latest study to discover that following these straightforward and generally well-accepted suggestions lowers the risks of various diseases, and of obesity.
Why do studies keep pounding these points home, when most people who actually care what studies have to say on these questions (as most people in America probably do not) have long since come to believe they've heard enough to make up their minds? Well, for one thing, that's how science is supposed to work--you keep validating theories with new tests, because the new tests may be better than the old ones, and might turn up different answers or at least new twists on them. And the accumulated results from many studies give scientists a much better handle on what's really going on. (The far bigger problem in science is that the vast majority of published research isn't validated at all, but that's a different story.)
Not only that, but the fact of the matter is that every so often an at least fairly credible study turns up evidence that suggests these standard healthy habits may not do as much for us as we'd like to think, and may even hurt. Thus there was the recent scientific review that suggested there is little solid evidence that any of the recommended lifestyle and nutritional habits lower the risk of Alzheimer's, even though we've been hearing that they do for years. There are studies that, in violation of what virtually everyone who has lost weight and kept it off knows, have found that exercise doesn't make any difference in controlling weight, a claim transmitted without doubt or qualification by former U.S. FDA commissioner David Kessler in his best-selling book on obesity. There are studies that find that eating a lot of saturated fat is just fine (these are pushed by the Atkins diet fanatics, of course), and others that conclude people would be thinner and healthier if they loaded up every morning on a high-carb breakfast. More than one study has found that being overweight may help you live longer.
So what's up with all that? Probably several things. First of all, bear in mind that scientists themselves have recognized (as I've documented at length elsewhere) that even most prestigiously published medical research is refuted within a few years. Research is hard enough when done impeccably, but it is in fact plagued with surprising levels of sloppiness, bias, fudging and other distorting shenanigans. So no results should be surprising--someone, somewhere will manage to screw a study up and get exciting but false results that some perfectly good journal will be thrilled to publish. In addition, different researchers who are supposed to be studying the same thing often end up measuring very different things in very different ways, and in very different sorts of people. Thus one study might conclude that a certain diet lowers the risk of, say, diabetes while another concludes it does not--simply because one of the studies measured blood-sugar levels in healthy patients while the other measured deaths from diabetes-related complications in very ill patients. (These measurement mismatches may be part of the problem with the studies that find obesity can be protective--it seems to prove so only for older patients and those with extensive heart disease, perhaps because old or sick people who are thin face greater risks from nutritional shortages than from the slower-acting ravages of obesity, or so it has been speculated.)
But there's an even more basic reason why we shouldn't be surprised that studies conflict on the healthfulness of nutrition and habits. There are thousands of factors that come into play when it comes to the ways various nutrients, habits and other environmental factors affect us, and they interact in a complex web that science may not yet have barely begun to unravel. Scientists try to tease the roles of individual factors out in their studies, but rarely can they honestly claim to have succeeded. (And let's not even get started here on the fact that they often base their claims on animal studies.)
All of these challenges conspire to distort study results and further confuse an already confusing picture. At least with exercise, satured fat and refined carbs there's a reasonably strong consensus of evidence, if not an unblemished one. That's why I'm going to keep exercising, and limiting saturated fat and refined carbs. I'm not fully confident any of that will do much for me, but at least I know I'm playing the odds.
Monday, June 21, 2010
Sunday, June 13, 2010
Do Blood-Pressure Pills Cause Cancer?
Blood-pressure pills appear to be linked to higher risks of cancer, according to a big review study making headlines. Should these results be at least a little worrisome to you if you're on one of these medications? In my opinion, yes--it's worth discussing with your doctor. (Please don't stop taking your pills on your own.) This study has many of the earmarks of more trustworthy research. It's a meta-study, which means it looked at the results from several previous studies, so that if some of the studies were troubled there's still some hope the results from the better studies will partly compensate for those problems. The studies were randomized, controlled trials, which means the people in them were randomly assigned to either take the medication in question or not take it, which can eliminate a lot of the problems studies run into. (I'm oversimplifying, but that's the basic idea.) And the studies were large, totaling some 60,000 patients, so there shouldn't be as big a chance for statistical flukes as there would be in small studies. None of these characteristics guarantee reliable findings, but they can all help.
It's worth noting that some articles describing this new study report that the study found that people on the drugs have a one percent higher risk of getting cancer. But other articles say the findings indicate risks go up on average about 10 percent, and for lung cancer a whopping 25 percent. The seeming contradiction here is due to the media's ongoing, near-complete and deeply problematic failure to specify and make clear the difference between absolute risk and relative risk. When your absolute risk of contracting a disease goes up by, say, ten percent, it means that, for example, if your risk of the disease was 14 percent, it is now 24 percent (14 plus 10), a big difference. When your relative risk goes up by ten percent, then if it was 14 percent, it is now around 15 percent (14 plus ten percent of 14)--not such a big difference. So in this example, an increase in relative risk of about 10 percent translates to an increase in absolute risk of a mere 1 percent. Research institutions typically want the public to be impressed with the importance of their results, so they often emphasize to the press relative risk, which is a generally misleadingly high number. But in this case--perhaps because there is concern that people on blood-pressure pills will overreact and stop taking them--there seems to be more interest than usual in reporting the change in absolute risk, which tends to be a very low number. Assuming you don't carry any special risk factors for lung cancer (such as smoking), you only have a tiny chance of getting it, and if you add on the apparent increased risk from blood-pressure pills, you now have a slightly less-tiny chance of getting lung cancer. Still, your chances of getting cancer of any type aren't so tiny, and a one percent increased absolute risk is nothing to sneer at. A percentage here, a percentage there, and pretty soon you're taking years off your life.
If you read about the study, you might note that even though the study found an increased risk for blood-pressure-pill takers of getting cancer, it didn't find an increased risk of them dying from it. Unfortunately, you shouldn't take this as reassuring. Most cancer studies fail to convincingly link increased risk factors to mortality, usually simply because it tends to take people many years to die of cancer (if they do die from it, that is), and studies usually don't go on long enough to follow people to their graves. (That's another reason, by the way, that researchers love running studies on mice instead of people--they're conveniently short-lived. We humans can be annoyingly enduring, from a medical-research point of view.)
Talk to your doctor about it. It may be that if you were just on the borderline of high blood pressure but have special risks for cancer, she may want you off the pills. But if your doctor feels your risks from uncontrolled high blood pressure outweigh the possible slightly increased risks of cancer turned up by this study, I'd trust him. In the unlikely chance your doctor doesn't know what you're talking about when you ask about this study, I myself would tell him that I'd be happy to sit there and wait while he looked the study up.
It's worth noting that some articles describing this new study report that the study found that people on the drugs have a one percent higher risk of getting cancer. But other articles say the findings indicate risks go up on average about 10 percent, and for lung cancer a whopping 25 percent. The seeming contradiction here is due to the media's ongoing, near-complete and deeply problematic failure to specify and make clear the difference between absolute risk and relative risk. When your absolute risk of contracting a disease goes up by, say, ten percent, it means that, for example, if your risk of the disease was 14 percent, it is now 24 percent (14 plus 10), a big difference. When your relative risk goes up by ten percent, then if it was 14 percent, it is now around 15 percent (14 plus ten percent of 14)--not such a big difference. So in this example, an increase in relative risk of about 10 percent translates to an increase in absolute risk of a mere 1 percent. Research institutions typically want the public to be impressed with the importance of their results, so they often emphasize to the press relative risk, which is a generally misleadingly high number. But in this case--perhaps because there is concern that people on blood-pressure pills will overreact and stop taking them--there seems to be more interest than usual in reporting the change in absolute risk, which tends to be a very low number. Assuming you don't carry any special risk factors for lung cancer (such as smoking), you only have a tiny chance of getting it, and if you add on the apparent increased risk from blood-pressure pills, you now have a slightly less-tiny chance of getting lung cancer. Still, your chances of getting cancer of any type aren't so tiny, and a one percent increased absolute risk is nothing to sneer at. A percentage here, a percentage there, and pretty soon you're taking years off your life.
If you read about the study, you might note that even though the study found an increased risk for blood-pressure-pill takers of getting cancer, it didn't find an increased risk of them dying from it. Unfortunately, you shouldn't take this as reassuring. Most cancer studies fail to convincingly link increased risk factors to mortality, usually simply because it tends to take people many years to die of cancer (if they do die from it, that is), and studies usually don't go on long enough to follow people to their graves. (That's another reason, by the way, that researchers love running studies on mice instead of people--they're conveniently short-lived. We humans can be annoyingly enduring, from a medical-research point of view.)
Talk to your doctor about it. It may be that if you were just on the borderline of high blood pressure but have special risks for cancer, she may want you off the pills. But if your doctor feels your risks from uncontrolled high blood pressure outweigh the possible slightly increased risks of cancer turned up by this study, I'd trust him. In the unlikely chance your doctor doesn't know what you're talking about when you ask about this study, I myself would tell him that I'd be happy to sit there and wait while he looked the study up.
Tuesday, June 8, 2010
WHO Knew If Flu Was True?
Two recently released European reports have raised the possibility that scientists who helped fuel world fear of an H1N1 flu crisis were motivated by ties to drug companies enriched by the ensuing rush to stockpile medications. The reports, which were from credible parties, including a Council of Europe committee and the British Medical Journal, noted that several of the World Health Organization scientists behind that agency's pressuring of governments to take dramatic action in response to the flu had received consulting or other fees from the drug companies, but that WHO hadn't bothered to identify these potential conflicts of interest. (Though the scientists in question had previously identified these ties in research-journal papers). With H1N1 turning out to fall far short of the devastating pandemic WHO and others predicted, many of the world's governments have ended up with vast unused stores of antiviral and other medications snatched up at WHO's urging.
There's no evidence that the scientists ginned up the warnings in order to channel sales to drug companies that had been good to them, and it seems unlikely that's the case. Scientists widely agreed there was plenty of reason to fear that H1N1 could become widespread and deadly. And we surely want our scientists, institutions and governments to err on the side of caution in these matters. It may even be that the rapid action taken by governments at WHO scientists' prodding was at least partly why H1N1 appears to have been something of a dud, so that the wasted drugs are in a way a sign of WHO's competence rather than its failure. (WHO actually continues to warn there's still a real danger of H1N1 getting out of hand.)
Still, science is supposed to be tough on biases, and especially blatant conflicts of interest. It has long been standard to insist that scientists openly and clearly state any ties they have to any corporations when there is any chance that their work could be seen as benefiting those corporations. But there's little question that conflict of interest remains a big problem in medical research. A 2003 Journal of the American Medical Association review of conflict-of-interest meta-studies involving some 67 conflict-of-interest studies and 398 other research reports confirmed a strong correlation between industry sponsorship and positive findings. (I provide sources in Wrong for that last and the following points). And the problem may be worse than it looks, because companies often disguise that they are behind certain findings by paying university researchers to put their names to studies actually conducted and written up by the companies--so-called "ghost authorship," a problem that infects about three-quarters of industry-backed drug trials, one study found According to another study, at least 30 percent of published medical researchers have been in a potential conflict-of-interest situation, and as few as 2 percent of researchers fully disclosed that fact. Every single one of the 170 psychiatric experts contributing to the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM IV) used to diagnose psychiatric disorders had financial ties to manufacturers of psychiatric drugs.
Remember this situation every time you hear of research that concludes a drug works well, or a new type of test is effective, or, yes, an infectious disease is about to wreak havoc on the world unless governments buy up all the treatments they can get their hands on. That's not to say the specific research is likely to be wrong--but it's another reason why we need to be wary consumers of research conclusions.
There's no evidence that the scientists ginned up the warnings in order to channel sales to drug companies that had been good to them, and it seems unlikely that's the case. Scientists widely agreed there was plenty of reason to fear that H1N1 could become widespread and deadly. And we surely want our scientists, institutions and governments to err on the side of caution in these matters. It may even be that the rapid action taken by governments at WHO scientists' prodding was at least partly why H1N1 appears to have been something of a dud, so that the wasted drugs are in a way a sign of WHO's competence rather than its failure. (WHO actually continues to warn there's still a real danger of H1N1 getting out of hand.)
Still, science is supposed to be tough on biases, and especially blatant conflicts of interest. It has long been standard to insist that scientists openly and clearly state any ties they have to any corporations when there is any chance that their work could be seen as benefiting those corporations. But there's little question that conflict of interest remains a big problem in medical research. A 2003 Journal of the American Medical Association review of conflict-of-interest meta-studies involving some 67 conflict-of-interest studies and 398 other research reports confirmed a strong correlation between industry sponsorship and positive findings. (I provide sources in Wrong for that last and the following points). And the problem may be worse than it looks, because companies often disguise that they are behind certain findings by paying university researchers to put their names to studies actually conducted and written up by the companies--so-called "ghost authorship," a problem that infects about three-quarters of industry-backed drug trials, one study found According to another study, at least 30 percent of published medical researchers have been in a potential conflict-of-interest situation, and as few as 2 percent of researchers fully disclosed that fact. Every single one of the 170 psychiatric experts contributing to the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM IV) used to diagnose psychiatric disorders had financial ties to manufacturers of psychiatric drugs.
Remember this situation every time you hear of research that concludes a drug works well, or a new type of test is effective, or, yes, an infectious disease is about to wreak havoc on the world unless governments buy up all the treatments they can get their hands on. That's not to say the specific research is likely to be wrong--but it's another reason why we need to be wary consumers of research conclusions.
Thursday, June 3, 2010
Some Recent Headlines
Breast cancer vaccine: As cancer researchers themselves like to point out privately, they've cured cancer thousands of times--in mice. It's this simple: mice aren't people. Treatments that work on them usually don't work on people, or don't work as well, or carry serious side effects. This non-translation seems particularly problematic with cancer, given that mice appear to develop cancer a lot more readily than humans do (or at least researchers have come up with effective techniques for seeing to it that they develop it, typically through genetic engineering, as was the case with this vaccine research), and the cancers they get appear to be a great deal more treatable than the human counterparts. There's no obvious reason to believe this particular development will fare any better than its predecessors if and when it moves into human trials.
Antidepressants increase miscarriage risk: The research on the efficacy and safety of antidepressants is all over the place. The bulk of solid-seeming research seems to suggest they don't do much more for most people than a placebo does, but there are all sorts of difficulties involved in trying to pin down improvements in the states of depressed people, and the picture is further blurred by the clearly demonstrated corruption of the literature due to drug-industry influence. The safety picture for antidepressants is even murkier, because, for one thing, depressed people may be more vulnerable to a range of health problems whether or not they're on medications. This particular study doesn't seem very strong, based as it is on looking back at the medication history of women who had miscarriages--an approach that can be wide open to all sorts of statistically distorting effects. Still, it's well-recognized that fetuses can be highly vulnerable to side effects in medications, and most people would agree that it's wise to err on the side of caution when it comes to giving meds to pregnant women., so it would be wrong to simply dismiss this study. Hopefully better studies will clarify things.
Sunscreen can cause cancer: It's hard to know what to make of the ongoing flap about the possibly carcinogenic effects of sunscreen; the research is weak. The specific research behind the recent headlines that suggest the vitamin A in many sunscreens is carcinogenic was carried out on mice, and doesn't seem to be getting much respect from most of the medical community. The headlines are motivated by press releases from the Environmental Working Group, a health-advocacy group that appears to have a strong anti-chemical, anti-industry bias. (Probably the safer bias to have, if you have to have a bias, but still, bias is bias.) I can't help noting, though, that at least one other study has found that slathering skin creams on mice followed by exposure to tanning-type light seems to induce cancer. The evidence doesn't seem very compelling, but on the other hand there's nothing far-fetched about the notion that chemicals designed to penetrate and alter the skin in some way increase vulnerability to forms of sunlight known to cause cancer. Common sense suggests that staying out of the sun, or blocking it with opaque or reflective materials or ointments (like zinc oxide), is probably a better way to avoid skin cancer. Having said that, I have to point out that exposure to sunlight is linked to lower risk of almost all other forms of cancer, most likely because it raises natural production of vitamin D. It's possible that taking vitamin D supplements can provide the same protection, allowing skipping the sun exposure, but studies testing that notion have come up with conflicting findings. Hey, glad I was able to clear this whole sunscreen business up.
Antidepressants increase miscarriage risk: The research on the efficacy and safety of antidepressants is all over the place. The bulk of solid-seeming research seems to suggest they don't do much more for most people than a placebo does, but there are all sorts of difficulties involved in trying to pin down improvements in the states of depressed people, and the picture is further blurred by the clearly demonstrated corruption of the literature due to drug-industry influence. The safety picture for antidepressants is even murkier, because, for one thing, depressed people may be more vulnerable to a range of health problems whether or not they're on medications. This particular study doesn't seem very strong, based as it is on looking back at the medication history of women who had miscarriages--an approach that can be wide open to all sorts of statistically distorting effects. Still, it's well-recognized that fetuses can be highly vulnerable to side effects in medications, and most people would agree that it's wise to err on the side of caution when it comes to giving meds to pregnant women., so it would be wrong to simply dismiss this study. Hopefully better studies will clarify things.
Sunscreen can cause cancer: It's hard to know what to make of the ongoing flap about the possibly carcinogenic effects of sunscreen; the research is weak. The specific research behind the recent headlines that suggest the vitamin A in many sunscreens is carcinogenic was carried out on mice, and doesn't seem to be getting much respect from most of the medical community. The headlines are motivated by press releases from the Environmental Working Group, a health-advocacy group that appears to have a strong anti-chemical, anti-industry bias. (Probably the safer bias to have, if you have to have a bias, but still, bias is bias.) I can't help noting, though, that at least one other study has found that slathering skin creams on mice followed by exposure to tanning-type light seems to induce cancer. The evidence doesn't seem very compelling, but on the other hand there's nothing far-fetched about the notion that chemicals designed to penetrate and alter the skin in some way increase vulnerability to forms of sunlight known to cause cancer. Common sense suggests that staying out of the sun, or blocking it with opaque or reflective materials or ointments (like zinc oxide), is probably a better way to avoid skin cancer. Having said that, I have to point out that exposure to sunlight is linked to lower risk of almost all other forms of cancer, most likely because it raises natural production of vitamin D. It's possible that taking vitamin D supplements can provide the same protection, allowing skipping the sun exposure, but studies testing that notion have come up with conflicting findings. Hey, glad I was able to clear this whole sunscreen business up.
Tuesday, May 25, 2010
The Real Lesson of the Vaccines-Cause-Autism Debacle
The doctor who claimed to have proven a link between autism and routine childhood vaccinations has been stripped of his license to practice medicine, the latest in a long string of humiliations. It turned out that his now-discredited study was carried out while he was quietly being well-paid by lawyers who were suing vaccine makers, and was based on blood taken from kids at a birthday party who were each slipped eight bucks for their trouble.
Let's review: A corrupt charlatan tried to foist a sloppy and possibly fraudulent piece of research onto a vulnerable community--many parents of autistic children are haunted by the lack of good explanation for the disorder--but the persistent work of mainstream scientists laid bare the scheme. And that's what we expect from science, isn't it? There may be the very occasional peddlers of junk science who sneak into reputable journals, but science surely if slowly brings their egregious transgressions to light, and expunges them and their work like so much gristle.
But here's an alternate view to consider: Even prestigious medical journals routinely run dubious studies that are tainted by bias, corruption or incompetence, and are rarely called out on it--but because this particular study got so much attention the scientific community for once actually went through the trouble of exposing it.
That probably seems far-fetched, but it's much closer to the truth. The evidence supporting this point of view is so plentiful and consistent that it's daunting to contemplate having to narrow it down to a mere handful of quick points so as not to make this an absurdly long post, but let me try, drawing on my research in Wrong (where these factoids are elaborated on and sourced): About two-thirds of the findings published in top medical journals turn out to be wrong or significantly exaggerated. Nearly all studies that find a particular type of food or a vitamin lowers the risk of disease fail to hold up. In spite of all the hype for gene discoveries that promise to deliver insights into our personal disease risks and customized treatment, studies have found that less than 10 percent of the published studies that identify gene links to disease hold up. Studies suggest that only about one percent of the scientific studies based on fraudulent data are identified and reported. On average roughly one-third of medical researchers surveyed in various studies admit to having committed or become personally aware of at least one act of research misconduct within the previous three years. Two-thirds of the drug-study findings that indicate a drug may cause harm are not fully reported by researchers. A third of the studies published in top medical journals contain statistical errors. And here, at least in this context, is the kicker: An estimated ninety-five percent of medical findings are never retested.
Ironically, the occasion of the rare public outing of one out of an ocean of bad studies actually serves to make medicine look good, because the story comes off as the community successfully rooting out the tiny percentage of bad work that sneaks in. Rarely does anyone openly point out in public that bad studies are the rule rather than the exception, let alone actually identify many of the vast number of studies that are not trustworthy, let further alone actually correct the record with better research.
Yes, in this case the system worked, because the findings were so inflammatory in terms of the politics of the large autism community, and because they so blatantly contradicted what scientists were already sure they knew about the subject. But it's a real exception to how the system fares, not a general validation. And that's the insight we really ought to take away from this sorry episode.
Let's review: A corrupt charlatan tried to foist a sloppy and possibly fraudulent piece of research onto a vulnerable community--many parents of autistic children are haunted by the lack of good explanation for the disorder--but the persistent work of mainstream scientists laid bare the scheme. And that's what we expect from science, isn't it? There may be the very occasional peddlers of junk science who sneak into reputable journals, but science surely if slowly brings their egregious transgressions to light, and expunges them and their work like so much gristle.
But here's an alternate view to consider: Even prestigious medical journals routinely run dubious studies that are tainted by bias, corruption or incompetence, and are rarely called out on it--but because this particular study got so much attention the scientific community for once actually went through the trouble of exposing it.
That probably seems far-fetched, but it's much closer to the truth. The evidence supporting this point of view is so plentiful and consistent that it's daunting to contemplate having to narrow it down to a mere handful of quick points so as not to make this an absurdly long post, but let me try, drawing on my research in Wrong (where these factoids are elaborated on and sourced): About two-thirds of the findings published in top medical journals turn out to be wrong or significantly exaggerated. Nearly all studies that find a particular type of food or a vitamin lowers the risk of disease fail to hold up. In spite of all the hype for gene discoveries that promise to deliver insights into our personal disease risks and customized treatment, studies have found that less than 10 percent of the published studies that identify gene links to disease hold up. Studies suggest that only about one percent of the scientific studies based on fraudulent data are identified and reported. On average roughly one-third of medical researchers surveyed in various studies admit to having committed or become personally aware of at least one act of research misconduct within the previous three years. Two-thirds of the drug-study findings that indicate a drug may cause harm are not fully reported by researchers. A third of the studies published in top medical journals contain statistical errors. And here, at least in this context, is the kicker: An estimated ninety-five percent of medical findings are never retested.
Ironically, the occasion of the rare public outing of one out of an ocean of bad studies actually serves to make medicine look good, because the story comes off as the community successfully rooting out the tiny percentage of bad work that sneaks in. Rarely does anyone openly point out in public that bad studies are the rule rather than the exception, let alone actually identify many of the vast number of studies that are not trustworthy, let further alone actually correct the record with better research.
Yes, in this case the system worked, because the findings were so inflammatory in terms of the politics of the large autism community, and because they so blatantly contradicted what scientists were already sure they knew about the subject. But it's a real exception to how the system fares, not a general validation. And that's the insight we really ought to take away from this sorry episode.