Nut Consumption and Prevention of Non-communicable Diseases: A Global Individual Participant Data Meta-analysis

Overview

The relationship between nut consumption and the risk of developing non-communicable diseases (NCDs) has been a subject of growing interest. However, the findings from previous studies have been conflicting for some health outcomes, such as type-2 diabetes, and have been underexplored for other outcomes, such as neurodegenerative diseases. One contributing factor to these inconsistencies lies in the different analytical approaches and confounding factors used across studies. Furthermore, the majority of previous studies have primarily focused on populations in Europe or the United States, potentially limiting the generalizability of the findings to other global regions. The NUTPOOL project aims to address these gaps by conducting an extensive individual participant data (IPD) meta-analysis. This study will evaluate the association between total and specific types of nut consumption and the future risk of NCDs.

NCDs account for 74% of global deaths, with 41% of these occurring prematurely in individuals aged 70 years and older. Type 2 diabetes, cardiovascular diseases (CVD), and cancer are among the major chronic diseases that affect people across countries and contribute to premature deaths. Notably, 77% of all NCD deaths and 86% of premature deaths occur in low- and middle-income countries. Reducing premature mortality from NCDs through preventive strategies, especially in low- and middle-income countries, was identified as a target in the Sustainable Development Goals from the World Health Organization. Additionally, as populations age, dementia is gaining recognition as a major public health concern. Dementia is among the leading causes of death and disability, with significant economic implications. Therefore, implementing preventive strategies to maintain cognitive function and delay cognitive decline is essential. Adhering to a healthy diet is an effective and cost-efficient strategy to reduce NCDs. In fact, unhealthy diets have been linked to over 30% of NCDs and cognitive decline. Nuts are a common food component of several plant-based diets that have shown beneficial effects in preventing NCDs. Nuts are a diverse food group comprising a wide variety of oily kernels with a unique nutritional profile, including high amounts of protein, fiber, unsaturated fatty acids, vitamins, polyphenols, and other phytochemicals. These nutrients may contribute to their beneficial effects, such as antioxidant, anti-inflammatory, antihypertensive, anti-lipidemic, and anti-aging properties, as well as their potential positive influence on gut microbiota modulation. Prior research has examined the impact of nuts on various health outcomes; however, many of these studies have focused on intermediate outcomes, obtained controversial results, and have been conducted primarily in European countries or the United States. Furthermore, the data has been analyzed using a variety of approaches, the studies were designed in different ways, and different covariates were added to the statistical models, potentially leading to inconsistencies and compromising the reliability of the comparison between studies. These inconsistencies emphasize the need for further research to elucidate the role of total nuts and specific types of nut consumption on the development of highly prevalent NCDs, as well as to allow the generalizability of the results across different countries and ethnicities. In the NUTPOOL project, the objective is to address these knowledge gaps and provide a thorough assessment of the associations between total nuts and specific types of nuts in relation to a wide range of NCDs and mortality. The results from this global collaborative project will be representative of the worldwide populations, as they include prospective cohort studies from different countries, including not only American and European countries, but also Asian and Oceania. Additionally, the focus is on total and specific types of nuts, along with highly prevalent and relevant outcomes such as type 2 diabetes, cardiovascular disease (CVD), cancer, dementia, and mortality. To ensure consistency and reliability, the analyses will be standardized using Individual Participant Data (IPD) meta-analyses. The selection of studies will be conducted after the identification of cohorts with data available on nut consumption. This identification process will be based on information comprehensively extracted from recently published systematic reviews and meta-analysis (SRMAs) on the subject, as well as through extensive research network. Invitations will be extended to as many eligible cohorts as possible to mitigate selection bias and ensure comprehensive participation in the study. Studies will be eligible for inclusion if they reported reliable nutrition data from a food frequency questionnaire (FFQ) regarding total nuts and/or specific nuts, along with well-defined data on at least two of the health outcomes. The study population had to be at least 1,000 adults, and the study had to span a minimum of five years or more of median follow-up, enabling time-dependent analyses. The assessment of nut consumption involves standardizing the amount consumed to a standard serving size (1 serving = 28 g or 1 oz.). The cumulative average or baseline nut consumption will be used according to the data collection methods employed for each cohort. Nut consumption will be analyzed as both a continuous and a categorical variable. The methods used to ascertain high-prevalence NCDs include medical diagnosis, official national or regional death registers, self-reported cases, and medication checks. The statistical analysis plan will entail the subsequent two steps: First, each cohort will follow a standardized and harmonized protocol to analyze the data. Secondly, centralized meta-analyses will be performed. Multivariable Cox proportional hazards models will be used to estimate the risk of NCDs. Separate models will be conducted for categories of total nuts and specific types of nuts. A dose-response analysis will also be conducted. The models will include different levels of adjustment: * Model 1: age and sex * Model 2: M1 + race, family history of NCDs of interest, personal history of hypercholesterolemia, education level, physical activity, alcohol consumption, smoking, total energy intake, cohort specific adjustements (e.g. allocation to intervention group for intervention cohorts; recruitment centre for multi-centre cohorts; socio-economic status and urban/rural residence, as deemed appropriate) * Model 3: M2 + red and processed meat, fruit and vegetable, potatoes, fish, legume, egg, cereal, whole grain, coffee, tea, sugar- and artificially-sweetened beverages and cooking fats. For analysis by subtype of nuts, model 2 will also be mutually adjusted for each other types of nut. * Model 4: M3 + body mass index Subgroups analyses will be performed based on age, sex, race, educational level, physical activity level, smoking status, body weight status, hypercholesterolemia and diabetes status at baseline. All analyses will be restricted to participants with complete data on age, sex, nut consumption, or the disease of interest. The missing data will be handled as follows: a) If there is more than 20% missing data for a covariate, the variable will not be included in the adjustment models; b) If there is 20% or less missing data, for continuous variables, the media should be imputed, and for categorical variables, an additional missing category should be created and used it in the models to avoid losing count or power. Once all the cohorts have provided the summary statistics, and the results have been verified and harmonized, they will be combined by inverse-variance weighted random effects meta-analyses (DerSimonian and Laird procedure). The between-study variance will be used to assess heterogeneity. To ensure the reliability of the results, additional meta-analyses will be conducted by systematically excluding one cohort at a time. Statistical meta-analyses will be performed using Stata/SE software, version 18.0 (StataCorp, College Station, TX).