NEW WIC toolkit now available. Visit today!

	Contact Us
	Looking for more information? Let us help.

	Sign up for the NDC Update for the latest dairy nutrition news, fast facts, free tools for patients, expert nutrition advice and information on updated dairy nutrition resources.
	Educate your patients on the importance of 3-A-Day of Dairy: Here's a great tool (PDF: 618k) to show families how to get their 3-A-Day of Dairy every day for stronger bones. Developed in conjunction with The American Academy of Family Physicians, The American Academy of Pediatrics, The American Dietetic Association, and The National Medical Association.

Dairy Council Digest Archives

Nutrition Research: What Can Studies Tell Us?
Types of Nutrition Research Studies

Observational Epidemiological Studies. Epidemiologic research seeks to expose potential associations between diseases and diet, lifestyle factors, or other variables within populations. Most epidemiological research is observational (2,4). Observational studies obtain data measured in groups (e.g., ecologic studies) or individuals (e.g., cross-sectional, case-control, and cohort studies) (6a,7, 8,11). The following types of observational studies are used to determine potential associations between diet and disease.

• Ecologic (Correlation) Studies. In ecologic studies, disease morbidity or mortality rates in populations, often in different countries, are compared with specific dietary intakes based on food disappearance data (2,7,8). A major strength of ecologic studies is the typically large contrasts in dietary intake and disease outcome which helps to show a link between nutritional exposure and risk of disease. However, problems of correlation studies include the potential for confounding factors and the crude assessment of dietary intake by food disappearance data. Confounding factors such as other dietary components, lifestyle factors, or ethnicity can make it impossible to distinguish between a response to treatment and some other factor. Because food disappearance data do not correct for losses (e.g., in transport, storage, preparation, or spoilage), intake tends to be overestimated. Ecologic studies are useful in generating hypotheses, but they are unable to provide any conclusions regarding diet-disease relationships (2,7,8).
  
  
• Cross-sectional Epidemiological Observations (or prevalence studies). In cross-sectional studies, nutritional exposure and disease status (or a biologic measurement of disease) are measured in individuals in a defined population at one point in time (2,4,6a,7,8). Cross-sectional studies have found that individuals with low bone mineral density (12) or high blood pressure (13) are those with low calcium intakes. This type of observational study can examine a large sample size and is useful for estimating links between key risk factors and disease among individuals. However, an important weakness of cross-sectional studies is that it is unknown whether diet may have been altered in response to diagnosis or initial symptoms of disease (8). Also, limited variation in usual dietary intake among individuals in a particular population, large within-person variation in dietary intake, and inaccuracies in dietary survey methods make it difficult to establish correlations between diet and disease. Cross-sectional studies, like ecologic studies, are a relatively weak method for assessing diet-disease associations (2).
  
  
• Case-Control Epidemiological Observations. In this type of observational study, a group of individuals with a particular disease (cases) and a group free of the disease of interest (controls) are examined for differences in dietary (or other lifestyle) factors (3,7,8). For example, case-control studies are designed to answer such questions as, "Do persons with osteoporosis (case subjects) consume diets that differ from those consumed by individuals without this disease (control subjects)?" Case-control studies are relatively inexpensive, efficient, yield results fairly quickly, and require fewer individuals than do cohort investigations (2,7,8). However, similar to cross-sectional studies, a major weakness is the possibility that case subjects, because of their disease, will recall their dietary intakes differently than do control subjects (i.e., recall bias). In addition, choosing appropriate controls is always biased unless they are a random sample of the population (11).
  
  
• Cohort Studies (or long-term follow-up). Among observational study designs, cohort studies provide the most reliable information (2). Cohort studies obtain dietary (or other) information on individuals in a large population (cohort) and follow the population to observe who does and does not develop the disease of interest (2,3,6a,7,8). This type of study can help demonstrate whether individuals with differing dietary intakes vary in a risk factor for or their rates of developing a specific disease. For example, cohort studies can help answer the question, "Do persons with low calcium/dairy food intakes develop osteoporosis more frequently or sooner than those who meet their calcium needs?"
  
  Cohort studies avoid many potential sources of methodological bias associated with case-control studies (6a,7,8). Because dietary intake information is measured prior to the disease onset, the illness cannot influence dietary recall. Cohort studies also allow repeated assessments of diet over time and examination of the effects of diet on more than one disease simultaneously (7). The primary disadvantage of cohort studies is that such studies are a long, expensive, large-scale undertaking. If a disease has a low incidence and a long induction period, a large cohort must be followed over a prolonged period of time (2,3,7,8). As a result, attrition may be aproblem which can bias the results (2,11). In many cohort studies (e.g., Framingham Heart Study), dietary information is measured only once at baseline. It cannot be assumed that diet remains unchanged throughout the duration of the study (2). Also, because of the large sample size, less detailed and less accurate dietary information may be obtained than for some case-control studies.
  
  Although associations identified from observational epidemiological investigations are useful for generating hypotheses or critical questions, the researcher cannot draw conclusions regarding causality, no matter how carefully the study was conducted (3,4,6a).
  
  Confounding factors and potential biases in observational studies may cause an association between diet and disease to be coincidental (4,6a, 14-16). Confidence that a relationship in observational studies might be causal depends on the consistency of findings in different populations and with different methods and studies, the dose response relationship, biological plausibility, and consistency with corresponding animal studies (10).
  
  A recent review of observational studies examining calcium, dairy products, and osteoporosis found a significant positive association between calcium intake and bone health in approximately three-quarters of 86 observational trials examined (12). The investigator attributed the failure of the remaining one-quarter of studies to find a positive relationship to the inability to control for potential confounding factors, biases, and methodological weaknesses in accurately estimating calcium intake (12,17). As discussed below, stronger support for calcium's beneficial role in bone health comes from investigator controlled studies (12).
  
  
• Experimental Studies Experimental studiesinclude randomized controlled clinical trials and basic research experiments (e.g., in vitro and laboratory animal studies) (4,6a,7,10,11).
  
  
• Randomized controlled trials (clinical trials). Randomized controlled trials (RCTs) in humans are considered to be the "gold standard" for evaluating a dietary hypothesis and providing evidence of causality between diet and disease (2,6a,7,10,11,14). In this type of experimental study, subjects are randomly assigned to either a control or an experimental group. The experimental group then receives a treatment or intervention (e.g., increased calcium intake) and the outcome (e.g., bone mineral density) is compared to that of the control or placebo group (2,4,6a,8). Ideally, RCTs are double-blind, meaning that neither the subject nor the investigator is aware of who is receiving the test substance or the placebo (4). However, blinding is usually difficult to achieve in food-related research. In RCTs, the investigator has direct control over the difference in level of nutrient intake. These studies can also achieve a degree of control of confounding factors that is impossible to achieve with any observational study. However, RCTs are expensive, a long duration may be necessary to achieve results, and compliance may be a problem, particularly in RCTs of diet.
  
  Human feeding trials such as some metabolic studies and multicenter feeding investigations use a randomized controlled design to establish causal relationships between diet and disease (2,6a,6b). In well-controlled metabolic studies, subjects are randomly assigned to a control diet or a test diet, or sometimes, to a random sequence of control and test diets (i.e., cross-over design). In cross-over studies, subjects serve as their own controls. Subjects are fed precisely measured diets in which one or more components are varied and the effect on a biological variable or risk factor is measured. Metabolic studies typically are carried out in a research setting. Because adherence is high, these studies can quantify the effects of a few dietary constituents on one or more outcome variables. However, these studies generally include a small number of participants (5 to 25) from a narrowly defined population and are of short duration (6a). If larger sample sizes are needed to achieve adequate statistical power, successive cohorts can be studied using the same protocol. Alternatively, a multi-center randomized feeding design can be followed.
  
  An example of a multi-center feeding study is the DASH (Dietary Approaches to Stop Hypertension) trial (6b,13,18,19). This trial demonstrated that a diet rich in low fat dairy products, fruits and vegetables has an unequivocal beneficial effect on blood pressure, particularly in individuals with established hypertension. DASH used a sufficiently large sample size (459 adults) at four centers to allow adequate representation of different groups such as women, minorities (e.g., African Americans), and individuals with and without hypertension, as well as to detect blood pressure reductions of public health relevance (18,19). Also, because the blood pressure-lowering effect of single nutrients may be too small to detect in small-scale studies, the DASH trial tested the effect of dietary patterns on blood pressure.
  
  Compared to observational studies, investigator-controlled diet studies provide stronger, more consistent evidence of a causal relationship between diet and disease (12). As mentioned above, three-quarters of 86 observational studies examined found a significant positive association between calcium intake and bone health (12). However, a critical review of 52 investigator controlled calcium intervention studies, including RCTs, by the same researcher revealed that all but two studies showed improved bone health at high calcium/dairy intakes (12). Failure of the two studies to demonstrate a beneficial effect of increased calcium/dairy is explained by the control group's relatively high calcium intake in one study, and the inclusion of early postmenopausal women, whose bone loss is mainly due to estrogen withdrawal and less to nutrition, in the second study (12).
  
  
• Basic Research Studies. Basic research studies of the effects of diet on disease include in vitro (test tube) and experimental animal investigations (4,5,6a,7,8). These types of studies help to confirm observations and provide insights into biologic mechanisms. Compared to human investigations, basic research studies are less expensive and can be carried out in a relatively short period of time (6a). However, findings from in vitro and experimental studies cannot be extrapolated to free-living humans.
  
  Meta-analysis. A meta-analysis is a formal statistical technique of systematically combining the results from separate but similar previously completed studies to yield overall conclusions about a hypothesis (4,7,20,21). Unlike narrative scientific reviews of the literature, meta-analyses provide a quantitative synthesis of available data. The technique is best used when examining studies addressing the same question and employing similar methods to measure relevant variables (4). Meta-analysis can be a valuable tool to aggregate relevant findings across studies and help to explain differences among studies. However, the strength of conclusions from meta-analyses can vary according to such factors as the type of studies being analyzed (i.e., observational versus RCTs), criteria for inclusion of studies, and statistical methods used (4,8,20-23).
  
  

• Summary
  
  

• Introduction
  
  

• Types of Nutrition Research Studies
  
  

• Discrepancies Between Observational and Randomized Controlled Trials: Dietary Assessment
  
  

• Conclusion
  
  

• References