In the past, researchers looked at the relationship between individual foods and nutrients and health. Today, there is interest in looking at how combinations of foods and beverages intake, or dietary patterns, influence health by applying different scientific methods. A statistical method called reduced rank regression analysis can be used to describe the patterns of foods and beverages people eat based on a set of “response variables” that are known to be related to the health outcome of interest. This summary of a NEL review presents what research evidence currently exist when reduced rank regression analysis is the method used to study the dietary patterns of groups of people and their likelihood of developing type 2 diabetes.

Conclusion

There is insufficient evidence, due to a small number of studies, to examine the relationship between dietary patterns derived using reduced rank regression and risk of type 2 diabetes. The differences in the methods used and populations studied made it difficult to compare results, and therefore no conclusions were drawn.

What the Research Says

Three studies looked at dietary patterns found using reduced rank regression analysis and the risk of getting type 2 diabetes. However, these studies had some key issues that make it hard to make any recommendations:

• There were few studies available.
• There were many differences in how the studies were done.
• The populations studied were different between studies.

Background

The goal of this systematic review project is to identify patterns of food and beverage intake that promote health and prevent disease. Historically, most dietary guidance has been based on research conducted on individual food components or nutrients. Dietary patterns are defined as the quantities, proportions, variety, or combination of different foods, drinks, and nutrients (when available) in diets, and the frequency with which they are habitually consumed. Reduced rank regression (RRR) is a statistical method that determines dietary patterns (combinations of food intake) that explain as much variation as possible among a set of response variables related to a health outcome of interest. It is an a posteriori method since it uses both existing evidence and exploratory statistics. The objective of this systematic review was to assess the relationship between patterns of food and beverage intake identified using reduced rank regression, and risk for type 2 diabetes.

Conclusion Statement

There is insufficient evidence, due to a small number of studies, to examine the relationship between dietary patterns derived using reduced rank regression and risk of type 2 diabetes. The differences in the methods used and populations studied made it difficult to compare results, and therefore no conclusions were drawn. (Grade: Not Assignable)

Methods

Literature searches were conducted using PubMed, Embase, Navigator (BIOSIS, CAB Abstracts, and Food Science and Technology Abstracts), and Cochrane databases to identify studies that evaluated the association between dietary patterns derived using reduced rank regression analysis and risk of type 2 diabetes. Studies that met the following criteria were included in the review: Human subjects; Ages: 2 years and older; Populations: Healthy and those with elevated chronic disease risk; individuals with chronic disease; published in English in a peer-reviewed journal; Sample size: Minimum of 30 subjects per study arm; Dropout rate Less than 20 percent; Study assesses dietary intake using reduced rank regression analysis; study considered diabetes and risks of diabetes.; subjects from countries with high or very high human development (based on the 2011 Human Development Index). The date range for the conduct of the studies was unlimited.

The results of each included study were summarized in evidence worksheets (including a study quality rating) and evidence table. A group of subject matter experts were involved in a qualitative synthesis of the body of evidence, development of a conclusion statement, and assessment of the strength of the evidence (grade) using pre-established criteria including evaluation of the quality, quantity and consistency, magnitude of effect and generalizability of available evidence.

Findings

Three prospective cohort studies examined dietary patterns derived using reduced rank regression analysis and their association with T2D incidence. The studies ranged in size from 880 to 6,699 subjects, and two of the studies were conducted in the United States and one in the United Kingdom and ranged in duration from 5.2 to 11.6 years. • Comparison across studies was limited by the small number of studies, differences in methodology, and in the populations studied. Therefore no conclusions were drawn.

Discussion

The ability to draw a gradable conclusion was limited due to the following issues:

• All three studies used different types of biomarkers as response variables, such as PAI-1 and fibrinogen; HOMA-IR index; and BMI, fasting glucose, TG, HDL, and hypertension, making it difficult to make comparisons across these studies.
• The dietary patterns derived in each of these studies were directly linked to the response variables selected; therefore, the variation in the response variables used suggest that the resulting dietary patterns may not be comparable.
• There were variations in dietary assessment methods used to assess dietary intake, as well as the food groupings using in the analyses across the studies. These methodological differences make it difficult to compare the resulting dietary patterns across studies and to determine how these differences may have contributed to differences in relationships between the patterns and type 2 diabetes risks.
• The studies were not consistent in their use of confounders in the analyses. For example, alcohol intake was not included as a confounder in one study, and alcohol, BMI, and smoking status were not controlled for in another study.

There was a positive association between derived dietary patterns that included meat intake and incident T2D in the two studies that used biomarkers as response variables, though the definitions of meat differed. However, because there were so few studies available, variability in the methodology used and different populations considered, there was insufficient information from which to assess consistency or draw conclusions about the relationship between dietary patterns derived using reduced rank regression and risk of type 2 diabetes.