between an intervention and an outcome. Randomization, however, may result in unequal distribution of the characteristics of the subjects between the control and treatment groups (e.g., baseline age or blood [serum or plasma] LDL cholesterol levels are significantly different). If the baseline values are significantly different, then it is difficult to determine if differences at the end of the study were due to the intervention or to differences at the beginning of the study. When the substance is provided as a supplement, a placebo should be provided to the control group. When the substance is a food, it may not be possible to provide a placebo and therefore subjects in such a study may not be blinded. Although the study may not be blinded in this case, a control group is still needed to draw conclusions from the study.
Randomized controlled trials typically have either a parallel or cross‐over design. Parallel design studies involve two groups of subjects, the test group and the control group, which simultaneously receive the substance or serve as the control, respectively. Crossover design involves all subjects crossing over from the intervention group to the control group, and vice versa, after a defined time period.
Although intervention studies are the most reliable category of studies for determining a cause‐and effect relationship, generalizing from the studies conducted on selected populations to different populations may not be scientifically valid. For example, if the evidence consists of studies showing an association between intake of a substance and reduced risk of juvenile diabetes, then such studies should not be extrapolated to the risk of diabetes in adults.
Observational Studies
Observational studies measure associations between the substance and disease. Observational studies lack the controlled setting of intervention studies. Observational studies are most reflective of free‐living populations and may be able to establish an association between the substance and the disease. In contrast to intervention studies, observational studies cannot determine whether an observed relationship represents a relationship in which the substance caused a reduction in disease risk or is a coincidence. Because the subjects are not randomized based on various disease risk factors at the beginning of the study, known confounders of disease risk need to be collected and adjusted for to minimize bias. For example, information on each subject’s risk factors, such as age, race, body weight, and smoking, should be collected and used to adjust the data so that the substance/disease relationship is accurately measured. Risk factors that need to be adjusted for are determined for each disease being studied. For example, the risk of cardiovascular disease increases with age; therefore, an adjustment for age is needed in order to eliminate potential confounding.
In determining whether the substance that is the subject of the claim has been measured appropriately, it is important to critically evaluate the method of assessment of dietary intake. Many observational studies rely on self‐reports of diet (e.g., diet records, 24‐hour recalls, diet histories, and food frequency questionnaires), which are estimates of food intake. Diet records are based on the premise that food weights provide an accurate estimation of food intake. Subjects weigh the foods they consume and record those values. The 24‐hour recall method requires that subjects describe which foods and how much of each food they consumed during the prior 24‐hour period. Diet histories use questionnaires or interviewers to estimate the typical diet of subjects over a certain period of time. A food frequency questionnaire is the most common dietary assessment tool used in large observational studies of diet and health. Validated food frequency questionnaires are more reliable in estimating “usual” intake of foods than diet records or 24‐hour recall methods. The questionnaire asks participants to report the frequency of consumption and portion size from a list of foods over a defined period of time. One problem with the dietary intake assessment methods described above is that there may be bias in the self‐reporting of certain foods. For example, individuals who are overweight tend to under‐report their portion sizes and therefore the actual amount of substances consumed is often underestimated. If there are reliable biomarkers of intake of a substance, these biomarkers are often measured rather than using self‐reported intakes….
Well‐designed observational studies can provide useful information for identifying possible associations to be tested by intervention studies. In contrast to intervention studies, even the best‐designed observational studies cannot establish cause and effect between an intervention and an outcome… . [However, observational studies] in some situations, can be support for a substance/disease relationship for an SSA or qualified health claim….
Research Synthesis Studies
Reports that discuss a number of different studies, such as review articles, do not provide sufficient information on the individual studies reviewed for FDA to determine critical elements such as the study population characteristics and the composition of the products used. Similarly, the lack of detailed information on studies summarized in review articles prevents FDA from determining whether the studies are flawed in critical elements such as design, conduct of studies, and data analysis. FDA must be able to review the critical elements of a study to determine whether any scientific conclusions can be drawn from it… . Most meta‐analyses, because they lack detailed information on the studies summarized, will only be used to identify reports of additional studies that may be useful to the health claim review and as background about the substance–disease relationship….
Animal and In Vitro Studies
FDA intends to use animal and in vitro studies as background information regarding mechanisms that might be involved in any relationship between the substance and disease. The physiology of animals is different than that of humans. In vitro studies are conducted in an artificial environment and cannot account for a multitude of normal physiological processes such as digestion, absorption, distribution, and metabolism that affect how humans respond to the consumption of foods and dietary substances. Animal and in vitro studies can be used to generate hypotheses, investigate biological plausibility of hypotheses, or to explore a mechanism of action of a specific food component through controlled animal diets; however, these studies do not provide information from which scientific conclusions can be drawn regarding a relationship between the substance and disease in humans.
C. Identifying Surrogate Endpoints of Disease Risk
Surrogate endpoints are risk biomarkers that have been shown to be valid predictors of disease risk and therefore may be used in place of clinical measurements of the onset of the disease in a clinical trial. Because a number of diseases develop over a long period of time, it may not be possible to carry out the study for a long enough period to see a statistically meaningful difference in the incidence of disease among study subjects in the treatment and control groups.
These are examples of surrogate endpoints of disease risk accepted by the National Institutes of Health and/or FDA’s Center for Drug Evaluation and Research: (1) serum low‐density lipoprotein (LDL) cholesterol concentration, total serum cholesterol concentration, and blood pressure for cardiovascular disease; (2) bone mineral density for osteoporosis; (3) adenomatous colon polyps for colon cancer; and (4) elevated blood sugar concentrations and insulin resistance for type 2 diabetes….
D. Evaluating Human Studies
Under the evidence‐based review approach set out in this guidance, FDA intends to evaluate each individual human study to determine whether any scientific conclusions about the substance/disease relationship can be drawn from the study. Certain critical elements of a study, such as design, data collection, and data analysis, may be so seriously flawed that they make it impossible to draw scientific conclusions from the study. FDA does not intend to use studies from which it cannot draw any scientific conclusions about the substance/disease relationship, and plans to eliminate such studies from further review. Below are examples of questions that the agency intends to consider whether scientific conclusions can be drawn from an intervention or observational study about the substance/disease relationship….
* * * * *
5.7.2 Substantiation of Dietary Supplement Claims
The regulation