In a recent study published in the journal Limits in nutritionresearchers developed a method to retrospectively harmonize data on the relationship between dietary patterns and cognitive health from individual studies that differed widely in methodologies and outcomes.
Review: Dietary patterns associated with cognitive decline: methods for harmonizing data from European and US cohort studies. Image credit: Elena Eremenko/Shutterstock
Background
A growing area of research is the relationship between dietary patterns and cognitive health, particularly related to understanding modifiable risk factors for dementia and other neurodegenerative diseases. Findings from various reviews and meta-analyses have found that diets rich in fruits, whole grains, and vegetables are beneficial for cognitive health.
However, findings from individual studies on dietary patterns such as the Dietary Approaches to Stop Hypertension (DASH), the Mediterranean, Anti-Inflammatory and Mediterranean DASH Diets, Intervention to Delay Neurodegenerative Disease (MIND) diets are conflicting. These inconsistencies, resulting from heterogeneity in the methods of conducting dietary and cognitive assessments, study populations, duration of follow-up, and various other study parameters, have made it challenging to conclude potential dietary guidelines for cognitive health.
A proposed approach to draw conclusions from these heterogeneous individual studies involves pooling data from individual participants from different clinical and observational trials, similar to a meta-analysis. It differs from meta-analysis in that data are searched from eligible studies rather than extracted from the publication. However, a method of data harmonization is still needed to ensure comparability and compatibility between data.
About the research
The present study describes a protocol to retrospectively harmonize individual participant data from multiple studies from the United States (US) and Europe on the relationship between dietary patterns and cognitive health. They also discussed study selection criteria to answer the research question and provided definitions of study outcomes and exposures.
Data harmonization
To determine the associations between dietary patterns and the incidence of cognitive decline and mild cognitive impairment, the selection criteria called for the inclusion of studies that examined diets representative of different populations in the US and Europe, and the study design was either clinical trial or prospective cohort study. Only studies conducted on adults over 35 years of age who were cognitively healthy were to be included in the harmonization analysis. Sensitivity analyzes are recommended to assess the interactive effects of comorbidities such as obesity, cardiovascular disease, diabetes, depression, and physical activity.
Exposures considered for analysis included dietary patterns such as adaptations of the Mediterranean Dietary Pattern, the EAT-Lancet pattern, the Healthy Eating Index, or a pattern derived from as a result from the results of principal component analysis. Dietary history, food frequency questionnaires or repeat interview data will be used to determine daily food intake results.
Outcome analyzes consisted of cognitive outcome assessments at two time points to determine changes in cognitive ability or the development of mild cognitive impairment. Data on dietary patterns will be harmonized across food and drink groups such as dairy, vegetables, fruit, meat, candy and sugar-sweetened beverages.
The researchers also recommend ensuring that the nutrient reporting units used are converted to a single unit and reported along with the number of grams in one serving of each food category. Harmonized dietary data will consist of a single exposure variable for healthy dietary patterns reported in terms of an outcome such as the Healthy Eating Index.
The proposed method of harmonizing individual participant data suggests that only those cognitive measure instruments that can potentially diagnose mild cognitive impairment or measure executive function, verbal fluency, and verbal memory are included in the analysis. In addition, cognitive data should be harmonized using pre-statistical harmonization to categorize the cognitive tests from each study according to the cognitive abilities being investigated.
Schematic study protocol to assess the relationship between dietary patterns and cognitive decline.
Data augmentation strategies were proposed to address possible heterogeneity among factors such as administration of cognitive measures, instrument adaptation, components, and instrument version. In addition, confirmatory factor analysis will be used to test the equivalence of cognitive measures.
The study also discussed methods to harmonize data for confounding factors such as lifestyle, physical activity, sociodemographic factors, anthropometry, and physical activity levels. Harmonized data on dietary intake, cognitive measures, and confounders will then be used to determine the association between diet and cognitive health through an individual participant meta-analysis of data.
Conclusions
To summarize, the study provides a method to harmonize data from multiple heterogeneous studies so that study designs can infer the relationship between dietary patterns and cognitive health. Application of this method and findings will help formulate recommendations for dietary guidelines to improve cognitive health and prevent cognitive decline.