Main Menu Health Library Antioxidants and All-Cause Death
CLINICAL and OBSERVATION STUDIES: WHY DO THEIR RESULTS DIFFER?
One of the most important puzzles in the health sciences is that of the differing results from observation and clinical type studies on populations of individuals. A health factor as a vitamin is shown to produce a major reduction in disease. But another group conducts a clinical study and finds no statistical benefit. The widespread understanding and even near religion among researchers and doctors is that a clinical study - that directly measures the influence of a factor by comparing two randomized populations, one doing one thing and the other doing another thing - is better. The dogma is "We must settle the problem with a clinical study." The result too often is more confusion and uncertainty.
The Mythology of the Randomized Clinical Study. A Randomized Clinical Study is a key need for the valuation of a drug or medicine that is an agent new to users and that is expected to produce a result in short time. Yet an argument that can be made for most problems relating to population wellness is: A clinical study not only can be unreliable but can be near worthless for measuring the likely risk to a population of some habit or external agent. As a most glaring example, clinical studies of cigarette smoking near certainly would fail to find any harm for smoking cigarettes. How can a study method that could fail to find an effect of such enormous harm be considered useful? A more truthful result may be obtained by - and in some cases only by - practical observation studies. How can this heresy be?
A clinical study of a drug measures that value of an agent that is foreign to and is not included in either the active or placebo group of a study. This provides a direct test of the effect of this new agent during time of measurement. But a clinical study of a life style factor attempts to measure something quite different than this. A clinical study attempts to measure the effect of adding more of an agent to a population that already includes potentially useful amounts of this agent. For example, a test of Vitamin C or its supplements really measures the effect of adding more of say Vitamin C to the amount that already is present in a base level diet in a population.. Thus the clinical study is not a test of Vitamin C per se but a test of the benefits of different amounts of Vitamin C present. Such a clinical study is not fundamentally capable of identifying whether or not an agent is or is not effective. As a first consideration, if an amount of Vitamin C in the base or placebo level is sufficient to produce its near maximum benefit, the added amount provided in the clinical test population will produce no further benefit. This negative esult has no bearing on the possbile effect of differing values of nutrient at at differing and lower levels. present in the base or placebo population. .
In the real world the amount of the vitamin present in the some base population diets could be sufficient to produce most of the benefit potential from this particular agent in this particular population. Life Ahead explains how certain diets of foods can produce the maximum likely benefits potential from antioxidants. There has been massive advice to the public about eating fruits and vegetables during the past 4 decades that has undoubtedly improved our usual population diet. Thus as a first problem, a clinical study will show no benefit for Vitamin C if amounts of vitamin in the base diet are sufficient to produce optimum benefit.. And in no way can this result provide a valid valuation of the benefit of Vitamin C. A clinical study initiates with the fallible assumption that much more of an nutirent will produce a much more substantail benefit that has been found previously. .
Some larger health studies as those for nurses and health professionals have used populations that are significantly more healthful than average. The population in the placebo group of the WHI study of a women's risks appeared far healthier than average. Clinical studies require advertising to enlist participants, and those concerned about their health and that have adopted healthier than average lifestyles can be over-represented in this base or placebo population. Such studies are not truly randomized.
This first problem becomes compounded further when consideration that effects of all antioxidants are at least partially duplicative and complementary. . A dozen nutrients other than Vitamin C have antioxidant properties including Vitamin E, Vitamin A, Beta and Alpha Carotene, Selenium, Lutein and Lycopene. Studies show that these nutrients are heavily intercorrelated in diets. Thus as a second problem for the clinical study , the overall antioxidant capability of the base diet must be insufficient for a useful additional effect of any individual antioxidant to be measured by a clinical study. This antioxidant sufficiency is something that today is not measurable with useful accuracy. This first problem alone places any negative result of a clinical study as of dubious real meaning.
A second problem is that clinical study measures a quite different value for a nutrient than does an observation study. An observation study measures the difference between the lowest and highest amounts of a nutrient present in a population as for example from a 1st vs a 5th quintile of said nutrient,. A clinical study usually starts out with the average population value of X using a placebo, and adds a further quantity of X to this average in a comparison group. If the amount in the third quintile or the average value was sufficient to produce the maximum benefit of X, the adding of the further amount via the clinical study will find no further effect and conclude wrongly "That the nutrient as no beneficial effect." If the benefit of the nutrient is linear with amount, and amount in the 5th quintile produces the maximum benefit, the clinical study will measure only half of the proper value of the nutrient measured over the full 5 quintiles. But if the risk value for the nutrient is not linear as is more usual, the clinical study can be attempting to measure much less than half of the true value of the nutrient. When testing an antioxidant for this may be usual because the benefit of the antioxidant usually is not linear with its amount.. The clinical study may be attempting to measure a value much less than half of the amount measured by the observation study. The observation study measures across all 5 quintiles, and often this will include the potential value of any supplement present in the top quintile. Useful clinical studies that rarely are found would add a new amount of nutrient to placebo populations having lowest to highest levels of this nutrient. But even this approach would fail if other antioxidant nutrients were preent in sufficient amounts.
This leads to a next problem of measurement accuracy. If observation studies measure a usual effect of say 30% reduction in risk of disease, the clinical study can have a potential for measuring only half of this or a 15% reduction in benefit. The actual average accuracy of 12 clinical studies was a difference of 0.61 in 5% to 95% limits, or plus or minus 22%. The clinical studies thus did not have the resolution power to measure an effect less than this 22%. But this is only a start of the measurement problem that can become near insurmountable for practical clinical studies.
There is a another and potentially far more serious problem with a clinical type study of human lifestyle.. The major diseases that terminate most lives today are cardiovascular diseases and cancer. These diseases and some others are complex biochemical processes that proceed over an entire lifetime. Most clinical type studies are carried out for only a few years, and reveal only a snapshot of what occurs during just this narrow time window. Cardiovascular disease is produced substantially by atherosclerosis that accumulates in arteries. This develops in large part from the LDL cholesterol in the blood at a typical rate of about 6% each year, and starting at early age. Most of the health factors known as smoking, diet, exercise, antioxidant vitamins operate in substantial part to change this rate of atherosclerosis accumulation. Smoking via its creation of a pro-oxidative environment may increase this rate by 9% per year. But after 10 years the smoking can result in a cumulative increase in artery deposits and indicated risk of disease of 32%. After 20 years this increase can become an expected 75%. (See atherosclerosis- a long time process-- on this web site)
As shown elsewhere, an antioxidant can develop its risk at a rate between 1.5% to 4 % per year of duration, or at annual risk reduction of 0.985 to 0.96 each year. It requires at least 10 years of a new agent duration for a risk to develop that can be measured by a practical study of populations. Adding to this problem to the above described problem of measurement, the potential risk difference measured by the clinical study may be only a fraction of that measured by the observation study. Most clinical studies have been carried out for from 1 to 5 year periods that are far too short to develop a risk difference measurable within the resolving power of the study. The inevitable result can be "No statistically significant effect was found" This wrongly infers and is interpreted as "No effect exists."
The problem of measuring the effect of an antioxidant on cancer is even more difficult. Antioxidants may reduce risk of cancer at a rate of about 0.984 per year of duration. But it also may requires a potentially long induction period during which nothing visible happens. When producing cancer by exposing mice to carcinogens this induction period can be half of an expected lifetime The induction period for developing lung cancer from cigarettes is at least twenty years. More on this follows. .
Actual Results from Clinical Studies Appear Mostly Consistent with Those Expected from Observation Studies: Formulas relating risk of antioxidants to amounts present and duration have been developed for Vitamin E and Cardiovascular Disease and for Vitamin E and Cancer from observation studies. These formulas were applied to forecast the risk levels actually obtained from clinical studies of these same factors. The forecasts estimated values that usually were consistent with those actually measured in the clinical studies. These comparisons explained why the clinical studies found little or no benefit for use of Vitamin E. Most of the clinical studies were done for much too short a time period to produce statistically useful valuations. These valuations suggest that far longer duration and larger clinical studies than most done to date would be needed to develop a statistically useful valuation of an antioxidant. The fact that such studies now done found little or no benefit is explained by the mechanisms involved.
Studies of Healthy People and Those Suffering Disease Must be Considered Separately: The development of cardiovascular diseases, diabetes, cancer, and other major diseases takes place over a lifetime. Once a disease occurs, its further prevention may require quite different actions. The large study base on antioxidants shows that each one develops it risk over time. But the simple use of large amounts of each these agents do not prevent further occurrences of disease. They are preventatives, not medicines.
The large majority of clinical studies have been done to test agents effectiveness as medicine for the secondary or further prevention of disease re-occurrence. Their benefits that slowed progress of diseases by just 2 to 4% per year of life are not sufficient to prevent further disease occurrence. In fact , it seems likely that the excessive amounts of Vitamin A could have produced negative benefits. Atherosclerosis produces the obstructions that permit other factors such as blood clotting to stop blood flow and actually produce disease. Once those obstructions are produced, other factors than atherosclerosis per se can become primary in producing further unwanted events.
Two meta analysis that contested the effect of antioxidants were done largely on clinical trials of disease re-occurrence, and not on primary disease development. Results of these secondary trials have long been known to be relatively useless in preventing disease re-occurrence. They are not relevant for the primary prevention of disease. These meta analyses have received much publicity and have mislead many people including some doctors and supposed experts to stop advising the use of antioxidants for primary prevention..
The Dogma against Observations Studies is not Justified: The complaint against use of an observation study is: Perhaps some unknown agent was interrelated with the factor of interest, and that this agent, not the one being considered, is the true causative factor. Fifty years ago and before today's massive research on nearly every conceivable diet and other health factor had been done, this speculation may have had some if very small merit. But today this argument does not hold credence. It is difficult to accept that a diet or other factor that can produce confirmed differences of 25% to 50% or more in risk from both amounts in blood and from amounts in diet plus supplements in dozens of studies could have escaped attention from thousands of researchers and analysts that have been researching effects of every known and suspected diet nutrient. Today observation studies are extensively tested for the possible effects of all known confounding factors, and results adjusted for such factors are presented. Studies have tested and found beneficial not only dietary antioxidants but the benefits of accompanying measured amounts of a nutrient in blood. It is near inconceivable that some mysterious unknown thing could have caused the same difference in same direction in more than a two hundred different observational studies on different antioxidants done in different world countries on different groups of people.
A time-dependent factor can operate over a period of 30 or more years, and even over an entire lifetime. Two very large studies confirm steadily increasing risk of heart disease from diabetes for time periods extending to at least 30 years. (See Global Analysis, an Example on this web site)
A case control study of the effect of nutrients in blood develops a snapshot measurement of conditions in different participant groups. This identifies some unknown time of duration of these different nutrient values. A prospective observation study typically measures results of 10 to 15 years follow-up, but at study start the groups studied also will have had quite different prior usages of the nutrients. Questionnaires have suggested prior usage of nutrient supplements of 10 or more years, and regression methods have developed estimates of 20 years usual usage differences of nutrients. Thus whereas a clinical study usually views a result for a 5 year or lesser time, observation studies may be viewing effects over a typical 20 year time frame. A combining of this longer duration with the potentially twice high value measured in an observation study produces a far higher expected value from an observation type measurement This permits many more studies to reach statistical significance.
The paper of Vitamin E and heart disease lists results of 32 observation study comparisons. 31 of the 32 showed a benefit for Vitamin E. 8 of 9 primary clinical comparisons of vitamin E and heart disease also showed a benefit for Vitamin E. For the effect of Vitamin E on cancer, 51 of 53 comparisons cited show a benefit for vitamin E. The fact that some of these studies did not have the resolving power to develop 95% significance is not relevant when viewing scores such as this These results were obtained from all types of populations and in different world countries. The likelihood that some unknown factor causes such results is not tenable.
When we compare for the measurement of lifestyle habits this nil likelihood of a unknown intercorrelation factor problem for observation studies with the serious problems for developing useful clinical studies, there seems to be little real contest. Our real evidence for the primary protection against disease must be developed from observation type studies.
The new Life Ahead Computer Model can Test Effects of Antioxidant Combinations. The Life Ahead Model uses state-of-art analyses of cardiovascular risk, cancer risk, and risk of all-cause-death from each of 23 nutrients in a diet. Each of these analyses are posted on the internet at the Life Ahead Site for public review together with all supporting research data found. The program first computes the level of each nutrient in any entered diet. It develops a risk level for the amount present of each nutrient and compares that risk with similarly computed risk level of the same diet nutrients of a typical US population. A computation for a diet without supplements can then can display how the actual amount of each nutrient compares with its most healthful value, and how much benefit could be obtained by moving each nutrient to more healthful values. This now can suggest how much of either a diet change or a nutrient supplement is needed to reach most healthful overall values for each nutrient.
Life Ahead tests not only added amounts of individual antioxidants but a further computation of the limiting amount of overall antioxidant that can be useful. An antioxidant index tests the total antioxidant value.present from combinations of Vitamins A, C, E, and selenium. If an entered diet includes sufficient antioxidants from food - and this is possible but unusual - no benefit will be computed for the use of any further antioxidant supplement. For the more usual diet, the program shows the kinds and amounts of antioxidant that are insufficient and computes the amount of supplements needed. to reach a most healthful overall health benefit.
This more concise engineering based analysis provides a far broader picture of the value of antioxidants in the real world than can be obtained from statistical risk ratios of any individual research study. There is no such thing as a universal value or lack of value of any one nutrient. The measured benefit, large or nil, of an antioxidant depends on the diet it is used with. The value of an antioxidant depends primarily on the deficiency it is required to replace. There can be alternatative choices for the use of differing nutrients to correct specific problems.
The important Life Ahead finding about antioxidants is this: Most US diets are deficient in antioxidants. This deficiency in antioxidants can be the most important single health deficiency in our usual diets. It is more important than usual differences in amounts of saturated fat or cholesterol. This deficiency extends but to a less extent to supposedly healtfuld diets such as those called Dash or Mediterranean. There has to my knowledge no way for usefully computing antioxidant sufficiency or deficiency in actual diets. Life Ahead now may not provide the only available technology that can accomplish this. Many people including doctors have been seriously confused by seriously flawed meta analyses that have inferred that antioxidants are not useful and might even by harmful. There is a serious need for a correction of this extraordinally incorrect advice.
The Clinical Study Could Fail to Identify the Most Important of all Health Factors. The most important recognized population health factor today is that of cigarette smoking. The near million person National Cancer study of cigarette smoking done in the 1980's found 1781 cases of lung cancer deaths. Only one of these deaths occurred during the first 23 years of smoking!
What this means is that a Randomized Clinical Study of Cigarette Smoking done for not only 10, but even for 15 to 20 years would fail to find any effect of cigarette smoking! If the clinical study advocates had controlled science then, smoking today probably would have been more prevalent today than it was back in the 1960's when 60% of men smoked cigarettes. And an incredible millions more of people would be dying from them. Fortunately, no useful clinical study of cigarettes was ever done, and the harm of the habit was learned and verified from far more useful and far more valid observation studies.
In conclusion, the randomized clinical study is not a useful method for measuring the effect of lifestyle habits or factors that develop over time. Serious science requires a full and clear understanding of why results from observation and clinical studies differ. This understanding should be both better achieved, endorsed and disseminated widely. The confusion caused by clinical study failures may have significantly shortened millions of population lives during the past two decades. A movement is needed away from the speculation and blind faith that "The Clinical Study Always is Best".
4/18/2008 Forrest H. Blanding