JFS Special: Results of the largest study on antioxidants - do they have a role in preventive health?

It’s one of the most popularly believed adages of our times: that antioxidant supplements, and lots of antioxidant-rich fruits and vegetables, can avert the damage of free radicals and prevent chronic diseases of aging — the primary causes of death, such as heart disease and cancer — and enable us to live longer. Free radicals, the unavoidable results of being alive, are natural byproducts of breathing. Decades ago, when a correlation was seen between people eating produce-rich diets and lower rates of diseases associated with free radical damage, the belief was born that the antioxidants in fruits and vegetables must be the cause. Decades of clinical trials, that have not supported the ability of antioxidant vitamins to reduce premature deaths, haven’t shaken our conviction, let alone brought most of us to even question it.

According to the National Institutes of Health, more than half of adults in the U.S. take some type of vitamin supplement, spending $23 billion each year. The benefits of antioxidants are intensely promoted not just by those with the $65 million of programs for produce and ‘healthy eating’, but the even larger vitamin and supplement industry.

As you can imagine, the just-released review of every available randomized, placebo-control, clinical trial of antioxidant vitamins conducted since 1945 has received a lot of attention. And sparked a lot of misinformation from both sides of the debate.

This study* was conducted using exemplary methodology and by the world’s most respected source of systematic reviews. It provided one answer quite impressively and we can learn a lot from it. But it still cannot overcome the fact that it was a review. Forgetting that fact has been the source of misinterpretations and over interpretations of its findings, leaving consumers unsure what to believe.

Let’s start from the top.

What they did

This review, all 191 pages, was just published online by Cochrane Database of Systematic Reviews, having been previously published in a February 2007 issue of the Journal of the American Medical Association. So, its methods have already had a year of external peer-review and corrections, in addition to JAMA’s peer-reviewers. Researchers, led by Dr. Goran Bjelakovic, M.D., DrMedSci, at Cochrane Hepato-Biliary Group, Copenhagen Trial Unit at the Center for Clinical Intervention Research at Copenhagen University Hospital in Denmark, set out to determine if antioxidant supplements can reduce premature deaths and prolong life. They wanted to see if antioxidant nutrients are effective for primary and secondary preventive health care.

They first searched through four major databases of published trials, as well as contacted all of the major manufacturers of antioxidant supplements for any unpublished trials, done from 1945 to October 2005. They looked for every randomized clinical trial that compared antioxidant supplements (including beta-carotene, vitamin A, vitamin C, vitamin E and selenium) to a placebo or no vitamins. Trials of adults who were healthy and recruited among the general population (primary prevention) were included, as well as adults who had chronic diseases in a stable phase (secondary prevention). They excluded trials involving special needs or treatments for deficiencies, such as pregnant women, children, or those with acute infectious or active malignant cancers, as there is no scientific debate about the benefits of vitamins and minerals for deficiencies of essential nutrients.

Their primary outcome was, of course all-cause mortality. It is the most objective and decisive measurement to evaluate the hypothesis that these vitamins prevent or lessen the chronic diseases that are the major causes of death, and improve health and longevity.

They identified 815 prevention trials and contacted authors to get any missing information. A total of 339 studies were excluded because they weren’t randomized trials or didn’t meet the inclusion criteria, which was straightforward. A total of 405 trials (on 40,000 people) reported no deaths at all, meaning no comparisons could be made between the vitamin and placebo groups to show an effect from the vitamins — the majority were small phase I or II trials of short duration — but the Cochrane reviewers contacted the authors to confirm they had no deaths.

They ended up with 67 clinical trials that had been conducted all over the world on a total of 232,550 people. The average age was 62 years, with the studies reporting an average of 45% female participants.

Twenty-one trials were primary prevention trials of 164,439 healthy participants; 46 trials were secondary prevention trials with 68,111 participants who had gastrointestinal, cardiovascular, neurological, ocular, dermatological, rheumatoid, kidney, cardiovascular, endocrinological or other diseases. All participants in the intervention half of each trial had been given antioxidant supplements (orally) in a range of doses, alone or in 13 various combinations, for 1 month to 12 years (average 2.8 years). The average follow-up for all of the trials was 3.4 years (28 days to 14.1 years).

It is what the Cochrane researchers did next that provides us with the most important information. They took each clinical trial, along with the protocols and a multitude of published papers for each trial, and conducted detailed analyses of the quality of the methodology, looking for bias.

Bias in clinical trials is not a term many people recognize. Looking for bias is just a way to tell if a clinical trial has been conducted to be a “fair test” of an hypothesis. It’s a way to judge the quality of a clinical trial and determine if its results are reliable. When scientists talk about a weak trial that’s of poor quality, it isn’t an arbitrary claim, frivolous pickiness, or because there is some conspiracy to trash a study they don’t like. A weak study is one that failed to follow well-established methods to reliably test an intervention — basic standards that have been part of the scientific process for centuries — and shown bias. Bias is also used to judge a systematic review of all the relevant, credible trial evidence to help practitioners determine if a medical treatment is actually effective. We’ve talked briefly about fair test before, but for more information on biases and how trials can be manipulated to create bias, see sidebar: “What is a fair test?”

They identified 47 trials as having low risk for bias (in randomization, blinding, follow-up, etc.); and 20 trials with questionable or inadequate methodology, hence, high risks for bias. [But in the end, even considering bias didn’t change the findings.]

The money shot

Here is the greatest value of this analysis. They plotted the findings of each clinical trial, showing the effects in the intervention group taking the antioxidant supplements compared to those taking a placebo. The researchers also compared the trials in a multitude of ways: according to high and low risks of bias, primary and secondary prevention trials, excluding trials administering extra supplement doses, excluding trials with potential confounding factors and/or with extra supplements; trials of each of the vitamins separately, and each of the vitamins without selenium.

Not a single clinical trial was able to find a tenable effect for taking vitamins. There was no significant difference in mortality among those taking the antioxidant supplements compared to a placebo — regardless of the doses, antioxidants taken individually or in combination, in any population studied or after 14 years of follow-up. Every relative risk hugged either side of null (1:1 = same risk of death in those taking supplements and in those not), all within the range of random chance and margin of error. There are pages upon pages that look much like this:

The researchers were also unable to find a tenable effect for taking the antioxidant vitamins by lumping all of the studies together or by any permutation of comparisons they tried or any statistical model they used.

As the researchers noted: “In random-effects meta-analysis, antioxidant supplements had no significant effect on mortality.” Relative risk was 1.02. A null finding.

The random-effects analysis was requested by the JAMA peer-reviewers and is used when there are differences between study groups (heterogeneous trials, such as these on healthy people to patients, and differing study sizes and durations, participant ages, doses and vitamin combinations). This method yielded, as the authors noted, no statistically significant effect at all.

Cochrane Collaborate tells authors of these types of meta-analyses that there is little difference in the results regardless of which statistical analysis model is used, “and the conclusions of your review would certainly not change.” These authors tried two other statistical models and, sure enough, were still unable to yield a single tenable relative risk. With a fixed-effect analysis, they were able to derive a “statistically” significant relative risk of 1.04, but it was still an untenable relative risk for a meta-analysis. A null finding.

They tried other statistical analyses of every permutation mentioned above, using the fixed-effect model, and, still, all of the relative risks hugged either side of null: 1.17, 1.004, 1.00006, 0.998, 1.16, 0.999, 1.05, 0.92, 1.05, 0.95, 1.03, 0.89, 1.05, 1.12, 1.16, 1.02, 0.99, and on and on. None of the relative risks for mortality between those taking supplements and those taking a placebo were tenable.

Applying a Peto odds-ratio statistical method, a method used when there are little differences between a treatment and control group, they still found no tenable risk ratio (RR 1.05 for low-biased trials, for example).

The clearest and soundest take-away message from this analysis was the authors' conclusion:

“We found no convincing evidence that antioxidant supplements decrease mortality... We found no evidence to support antioxidant supplements for primary or secondary prevention.”

If there was any real effect on improving health and preventing or lessening chronic diseases that are what most people die of, then there would have been a clear, tenable effect seen. In the clinical trials that had reported the causes for deaths, cancer and heart disease were the mains causes of death — which were statistically equal between the supplement and placebo groups.

As the authors noted, antioxidant supplements are not only one of the most researched topics in the world, they’re one of the most adequately researched. By compiling the studies for us, we can better see that there is no evidence to support a benefit for taking antioxidant vitamins, believing they will help us live longer or as a preventative for chronic illnesses. Nada, zip, ziltch.

The general public, I suspect, doesn’t realize the degree to which clinical trials have consistently failed to support the use of antioxidant supplements and various ‘healthy’ foods, because the studies widely cited as evidence for their benefits are observational studies, looking for correlations among groups of people (or studies in test tubes). As with most diet and nutrition epidemiological studies, certain foods and behaviors are most often markers for factors such as social-economic and genetic factors that have the greatest role on health and longevity. When they’re put to the test in randomized clinical trials and given fair tests, they fail to prove to be the cause.

It went downhill from there

As is so often the case, what a study’s data actually reveals isn’t always what the authors conclude, or the spin given in the media.

The news around the world has been trumpeting, “Vitamins may shorten your life!” And media has been reporting that antioxidants could lead to premature death. Even the authors concluded that “beta-carotene, vitamin A and vitamin E seem to increase mortality.”

There is nothing to support these fears or such scary interpretations of null findings and untenable relative risks. As been noted by multiple expert bodies, derived relative risks need to be at least one to two times above null to be considered tenable.

So, just as this analysis clearly found no benefit, those very same relative risks show no credible dangers, either.**

What would relative risks in an analysis of randomized clinical trials look like that might allow healthcare practitioners and us to make a credible conclusion of a potential benefit or risk?

Considerably different. The relative risks would be strong, consistent and unambiguous. They say a picture’s worth a thousand words, so here is the plot from a meta-analysis of clinical trials of antibiotics compared to a placebo for reducing post-op deaths after bowel surgery. Except for the first small study in 1969 on 29 patients using an early antibiotic, every clinical trial showed a similar and strikingly significant reduction in deaths. [Relative risks are all well to the left 1, meaning many times fewer people died on antibiotics compared to a placebo.] There was no hugging either side of null and statistical splitting hairs trying to divine meaning. The saddest thing is that it took years for the medical profession to get it and they continued to do trials for 17 years, long after a benefit was clearly identified in repeated randomized, clinical trials.

How many clinical trials will continue to be done on antioxidants, and consumers being sold them, despite decades and at least 67 randomized controlled trials that have clearly shown no benefit of antioxidants in primary or secondary prevention?

Bringing balance and reason back is hard

Basic nutritional science has long known, and repeatedly demonstrated, the value of essential nutrients for the prevention and treatment of deficiencies. Here, the evidence is incontrovertible. A deficiency of vitamin C results in scurvy, a deficiency of folic acid in pregnant women can increase risk of spina bifida and neural tube defects in babies, a deficiency of vitamin B6 affects the neurological system, a deficiency of iron results in anemia, a deficiency of vitamin A can cause blindness and impaired immune system (and is the leading cause of preventable blindness and death among women and children in underdeveloped countries), etc. The use of vitamins and minerals in the treatment of deficiencies such as these have solid science. There are also specific medical conditions where deficiencies are known to occur due to heightened needs or difficulty meeting nutritional needs through foods, such as malabsorption after bariatric surgeries, certain illnesses, alcoholism, organ failure or surgery. For example, vitamin B6 deficiencies are seen in those taking certain medications such as cycloserie and penicillamine, or in babies with seizures due to a rare genetic pyridoxine-dependency.

But there is no credible evidence to support the belief that most of adults suffer deficiencies and need supplements, while we may fear so. Nor is it true that most people in developed countries are poorly nourished and suffer widespread vitamin deficiencies. Getting the tiny amounts of essential vitamins our bodies need (excluding certain medical conditions) isn’t difficult when food isn’t restricted. And eating and getting the nutrients our bodies need isn’t nearly as precarious as many want us to fear — eat normally and enjoy a variety of foods, which, despite popular fears, most people naturally do when they have enough to eat and are not fretting about their diets or weight and trying to control their eating. There’s no single ‘right’ way to eat.

Among countless other doctors, Dr. Sidney M. Wolfe, M.D., lead author of Worst Pills, Best Pills, has also emphasized that there is no credible medical evidence to support taking supplements when a specific deficiency doesn’t exist or for otherwise bettering our health. Claims that poor nutrition or nutritional deficiencies cause most chronic disease and today’s health problems are not supported by the evidence or biological plausibility. Similarly, claims that diseases and health problems be prevented or treated through nutrition and supplements aren’t supported, either.

The U.S. Preventive Services Task Force — charged with issuing careful, evidence-based findings that are used to develop clinical guidelines for healthcare providers — also recently reviewed randomized clinical trials on vitamin supplementation to prevent cancers and heart disease. It found insufficient evidence to recommend supplements of vitamins A, C or E; multivitamins with folic acid; or antioxidant combinations for the prevention of cancer or cardiovascular disease. Clinical trials, many covered here, continue to yield null findings for the benefits of vitamins and supplements or certain foods in preventing heart disease, cancers or other chronic diseases of aging, reducing premature deaths, or helping us live longer.

Healthy lifestyle programs and anti-aging diets rich in ‘healthy foods’, fruits and vegetables, and special supplements, promise to prevent chronic diseases, promote optimal “wellness,” slow aging and add years to our life. The trouble is, there’s no scientific support for such beliefs, and things have gone way beyond basic nutrition.

The free-radical theory of aging serves as a basis for the prominent role that antioxidants have in the anti-aging, ‘lifestyle medicine’ and the preventive health movement, yet it is based on misunderstandings of how cells detect and repair the damage caused by free radicals and the important role that free radicals play in normal physiological processes. Research on flavonoids and super reds, concentrated antioxidants from colorful fruits and vegetables, and other superfoods, for instance, continues to show no credible scientific evidence for any special healthful virtues or powers to heal or prolong life.

The role of food and lifestyles in causing or preventing cancers is where misinformation is especially ripe. As Dr. Barnett Kramer, deputy director in the office of disease prevention at the National Institutes of Health, said: “Over time, the messages on diet and cancer have been ratcheted up until they are almost co-equal with the smoking messages. I think a lot of the public is completely unaware that the strength of the message is not matched by the strength of the evidence.”

Many people discount the clinical trials of supplements and “healthy” foods that have failed to show significant special benefits (beyond their role in known deficiencies) for preventing cancers, heart disease or other chronic diseases of aging or on reducing mortality, believing that whole foods or some different combination of foods and supplements offer the vital essence that can promote optimal health. This is vitalism, not science. Regardless of the beliefs in special abilities of certain foods or special diets (the very same nutrients and diets are said to heal or kill us, depending on the information source!) they are grounded on a lot of nutritional misconceptions and poor science... and marketing. Look for those fair tests and randomized clinical trials on people.

People around the world have enjoyed a wide range of diets with no common relationship to lifespans or health. And humans have been trying to find the secret to longer life for all of recorded history, and haven’t found it yet. Even careful examination of people who have lived over 100 years fails to reveal any consistent secret except one: Have long-lived parents.

Science isn’t popular, though, and is a hard sell against intense marketing. Disturbed by how much nutritional and lifestyle quackery has crossed into mainstream, 51 world-recognized scientists and experts in human aging reviewed all of the scientific evidence behind such popular beliefs and issued a scientific position on the evidence. Funded by the National Institute on Aging and published in the Journal of Gerontology: Biological Sciences, they emphatically stated:

“Our language on this matter must be unambiguous: there are no lifestyle changes, surgical procedures, vitamins, antioxidants, hormones or techniques of genetic engineering available today that have been demonstrated to influence the processes of aging.”

* External Funding

This study received external funded from the Knowledge and Research Centre for Alternative Medicine (Vi-FAB) in Denmark. This is an independent institution under the Danish Ministry of Health and Prevention which promotes research into alternative medicine and promotes information on natural medicine and alternative modalities. Its board members include alternative practitioners associations and preventive health and public health organizations.

** Precaution

There is one precaution and potential risk of taking antioxidant supplements that hasn’t been mentioned. While supplements clearly don’t improve longevity or have a preventive health benefit, they are not harmless for those taking certain medications. Various supplements can interact with other medications with adverse consequences for patients. Vitamin E, for example, can inhibit the effects of tricyclic antidepressants, potentiate the effects of phenothiazine antipsychotics, inhibit beta blockers used for high blood pressure, and increase the blood thinning effect of warfarin and risk abnormal bleeding. Be sure to work with your doctor or pharmacist, if you choose to combine over-the-counter products with prescription medications.

Questions? These are a few of the main concerns that have been in the news:

1. Claim: We need to take vitamins and supplements because our diets are so crappy today and/or our foods are poor in nutrients.

Facts: Millions of people take vitamins to ‘cover their bases,’ wrongly believing their diets are inadequate and need the extra insurance of supplements to be healthy. This study used randomized controlled trials, which means people taking supplements were compared to people not taking supplements and eating typical diets with all of those supposedly “bad” foods. Yet even still, the supplements showed no benefit.

Concerns over the quality of our diets and health are greatly exaggerated; the population continues to live longer than ever in our history and rates of heart disease, cancer and stroke have been dropping. It is when foods are restricted by healthy people that deficiencies can come into play, not getting enough to eat.

2. Claim: They eliminated all of the studies (405 trials) that showed no deaths, thereby excluding from the analysis all the studies that showed no increase in risks.

Facts: They did not delete the studies just showing no deaths in those taking vitamins. They deleted studies reporting no deaths at all, from either arm. Remember, these were randomized, controlled trials, which means each trial compared a group of people taking vitamins and a group not. The researchers were looking to see if vitamins help reduce premature deaths — death is the most definitive measure of all of the chronic diseases that are the major causes of death, such as cancer, heart disease, etc..

If there are no deaths among people taking vitamins and no deaths in those not receiving vitamins, there’s no difference and the vitamins could not be shown to reduce deaths or increase deaths. Zero events are still zero.

They could have included these trials — 405 zeroes. But 405 X 0 = 0. It would not have changed their findings.

The clinical trials that didn’t report any deaths among the people during the study period were also smaller, shorter trials. In fact, they represented only 40,000 people, compared to the 232,550 in the larger, longer 67 trials they did include. So, the researchers actually included about 85% of every person who’d participated in clinical trials of antioxidant supplements since 1966.

In a sensitivity analysis, their findings didn’t change when they included all of those excluded studies, even after adding a death in each study arm (RR 1.02).

3. Claim: Most of the people in their analysis already had diseases and it was too late to repair that oxidant damage from a lifetime of bad diets, but they didn’t study healthy people.

Facts: This review examined the use of vitamins for primary prevention and secondary prevention. Most of the people were healthy — 164,439 healthy participants. Another 68,111 people had various chronic diseases. Regardless of how the authors analyzed the data — including all participants, just those who already had diseases and just healthy people — there was no benefits shown in vitamin use.

4. Claim: They studied people not taking large enough doses or the right combination of antioxidants to really prevent the diseases of aging.

Facts: This analysis included studies of vitamin supplements primarily to meet the RDA but also mega-doses. The exception was vitamin A; given the known health risks of high doses of vitamin A, only one trial (of 109 people) had given >25,000 I.U. of vitamin A. Regardless of how the findings were analyzed, including separate analysis of those taking extra vitamins, their findings didn’t change. There was no benefits in taking any of the vitamins alone or together, regardless of the dose or duration.

5. Claim: The vitamins were synthetic, not ‘natural,’ and ‘natural’ vitamins are different because they contain other ‘vital’ bioactive factors in whole foods that aren’t found in chemically-created products.

Facts: The idea of natural sources of vitamins is a marketing concept, but all chemicals are natural and derived from natural chemical precursors. The molecular structure of a synthetic vitamin molecule is identical to one from natural sources and indistinguishable in all respects to our body.

Names can be confusing and lead you to believe that one type is less bioactive than another, such as vitamin E. The d-form (most often termed ‘d-alpha tocopherol’) that is derived from vegetable oils and other ‘natural’ sources is different from the dl-form (dl-tocopherol) that’s commonly called the ‘synthetic’ type of vitamin E, but biologically no better. Simply put, the dl-form is a combination of d-form and l-form (usually 1:1 ratio), but the body only uses the d-form. The l-form doesn’t offer any extra benefit as it’s just excreted by the body. So the dosage on the labels of dl-form and d-forms (half) will be different — but the bioeffective vitamin dose is the same.

The body has to digest foods and metabolically change them to absorb and utilize the vitamins, so synthesized vitamins, such as folate, can be even more readily absorbed than natural forms. Most synthesized vitamins are made to have the right sterio-isomeric configuration to be readily absorbed.

‘Natural’ vitamins are said to have hundreds of other, unknown substances found in whole foods that aren’t found in synthetic vitamins, but there are no credible clinical studies supporting such claims that these ingredients potentiate or improve the action or absorption of the vitamins or offer any special benefits. Vitamin pills extracted from foods undergo a process that changes their place in that food and it’s implausible that the miniscule amount of some unknown factor that might survive that process in a tiny pill has any special vital essences or health promoting properties.

6. Claim: They eliminated 245 studies because they didn’t meet their arbitrary criteria.

Facts: Their inclusion criteria were clearly defined. It would have provided no value at all to include flawed studies and a data dump simply because some might like what they found. See “Fair test.”

© 2008 Sandy Szwarc