The Korean Genome and Epidemiology Study (KoGES), which includes the Ansan-Ansung and Health Examinees (HEXA) cohort studies, is a large-scale population-based prospective study conducted in Korea by the National Institute of Health, Korea Disease Control and Prevention Agency, aimed at investigating the risk factors of NCDs among Koreans. The baseline investigations for the Ansan-Ansung cohort were conducted between 2001 and 2002 and focused on residents in small to medium-sized cities, examining lifestyle, dietary, and environmental impacts on chronic diseases. The baseline study for the Ansan-Ansung cohort included 10,030 participants aged 40–69 years. The HEXA cohort involved medical institutions, public health centers, and healthcare facilities across large and smaller cities, emphasizing the identification of both environmental and genetic risk factors for prevalent chronic diseases in diverse urban settings. The baseline study for the HEXA cohort was conducted between 2004 and 2013, and included 173,195 participants aged 40–79 years. The design and protocol of the KoGES are described in the literature and on the KoGES website [18].

Among the eligible individuals from the two cohort studies (n = 183,225), participants were excluded based on the following criteria: no linkage to death data (n = 45,113); absence of dietary data (n = 1,756); implausible energy intake (< 800 or ≥ 4,000 kcal/day for men, and < 500 or ≥ 3,500 kcal/day for women) (n = 1,996) [19]; lack of data on household income (n = 11,502), education (n = 400), alcohol drinking (n = 220), smoking (n = 174), physical activity (n = 93), and BMI (n = 115); and history of cancer diagnosis that could affect dietary habits (n = 7,716). Finally, 114,140 participants were included in the data analysis (Fig. 1). Before the commencement of the study, all participants voluntarily signed consent forms and the study received exemption approval from the Hannam University Institutional Review Board (approval number: 2023-E-01-09-0625).

Flowchart of participant selection

Dietary data obtained from a food frequency questionnaire (FFQ) were used to assess nut consumption. The reliability and validity of the FFQ used in the KoGES have been previously reported [20, 21]. The Ansan-Ansung cohort study used a 103-item FFQ and the HEXA cohort study used a 106-item FFQ at baseline. These FFQs were conducted through interviews with trained investigators. Both FFQs included the same question asking participants how often they had consumed peanuts, almonds, and pine nuts over the past year. The responses to this question comprised nine frequency options (rarely, 1 time/month, 2–3 times/month, 1–2 times/week, 3–4 times/week, 5–6 times/week, 1 time/day, 2 times/day, and 3 times/day), with three average serving sizes (0.5, 1, and 1.5 servings) representing the amount of nuts consumed by a participant. Peanuts, almonds, and pine nuts were included in the food list of FFQ as they were commonly consumed nuts in Koreans and the serving size of nuts was set in the FFQ (15 g) based on the median value of nut consumption determined from dietary data of the national nutrition survey [20, 21]. The daily intake of nuts for each participant was calculated by converting the reported frequency of consumption into a weekly intake based on one serving and multiplying it by the selected average serving size. Based on these responses, nut consumption was categorized into non-consumption, less than 1 serving/week, 1–2 servings/week, and 2 or more servings/week, considering the distribution of participants in each category of consumption frequency. Energy and nutrient intakes, calculated using dietary data obtained from the FFQs, were used in this study. The percentage of energy from macronutrients was calculated for carbohydrate, protein, and fat intakes.

To determine mortality outcomes, including all-cause and cause-specific mortality, this study analyzed the KoGES-linked National Death Index database provided by Statistics Korea. Participant deaths were monitored from the initial baseline survey to December 2021. The underlying causes of death were classified according to the Korean Standard Classification of Diseases, 7th edition, which is based on the 10th revision of the International Classification of Diseases. Deaths were classified as all-cause (A00-Z99), CVD (I00-I99), or cancer deaths (C00-D48) [22].

This study considered potential confounding variables, including age, sex, BMI, household income, education, alcohol drinking, smoking, physical activity, and history of disease, as covariates. Covariates were assessed using self-administered questionnaires in the baseline examination of KoGES. BMI was calculated as measured weight divided by measured height squared (kg/m2) and classified according to the guidelines of the Korean Society for the Study of Obesity as follows: non-overweight/obese (< 23 kg/m2), overweight (23–25 kg/m2), and obese (≥ 25 kg/m2) [23]. Physical activity was categorized into “yes” (regularly exercised for ≥ 30 min once a day) or “no”. Information on the participants’ history of diseases was obtained from both participants’ self-reported data and objective indicators provided by the health examinations. Previous diagnosis by a doctor or current use of medication or treatment of diabetes, CVD, hypertension, and metabolic syndrome were based on the self-reported data. The history criteria for diabetes included at least one of the following four criteria: a diagnosis of diabetes by a doctor, current consumption of anti-diabetic medication, a fasting plasma glucose ≥ 126 mg/dL, or a HbA1C ≥ 6.5% [24]. The history criteria for CVD included a diagnosis of or undergoing current treatment for conditions including myocardial infarction, congestive heart failure, coronary artery disease, peripheral vascular disease, cerebrovascular disease, stroke, transient ischemic attacks, or angina [25]. The history of hypertension was defined as a systolic blood pressure of ≥ 140 mmHg or a diastolic blood pressure of ≥ 90 mmHg, a diagnosis of hypertension, or currently under treatment for hypertension [26]. Metabolic syndrome was determined by the presence of three or more of the following conditions: abdominal obesity (waist circumference ≥ 90 cm in men and ≥ 85 cm in women), low HDL-cholesterol (< 40 mg/dL in men and < 50 mg/dL in women or medication use), elevated triglycerides (≥ 150 mg/dL or medication use), elevated blood pressure (systolic blood pressure ≥ 130 mmHg or diastolic blood pressure ≥ 85 mmHg or medication use), and elevated fasting blood glucose (≥ 100 mg/dL or medication use) [27, 28].

All statistical analyses were conducted using SPSS version 25.0 (Armonk, NY: IBM Corp). A cumulative hazard graph was employed to estimate the cumulative mortality risk of all-cause, CVD, and cancer mortality across nut consumption. Categorical and continuous variables for the general characteristics of participants according to nut consumption were compared using the chi-squared test (χ2 test) and analysis of variance (ANOVA). Energy and nutrient intake according to nut consumption is presented as means and standard errors after adjusting for energy, except for the percentage of energy from macronutrients, and differences were compared using a generalized linear regression model. A multivariable Cox proportional hazard regression analysis was performed to determine the HRs and 95% CIs for mortality according to nut consumption. To assess the association between nut consumption and mortality, this study applied four models adjusted for different potential confounders. Model 1 was unadjusted. Model 2 was adjusted for sex, age (continuous), and BMI (continuous). Model 3 was further adjusted for household income, education, alcohol drinking, smoking, physical activity, and energy intake (continuous). Model 4 was additionally adjusted for a history of disease. Stratified sub-analyses of the association between nut consumption and all-cause mortality were performed according to age group, sex, BMI, alcohol drinking, smoking, physical activity, and history of diseases. For stratified analysis, age group was categorized into ‘under 60’ and ‘60 or older’. Alcohol drinking and smoking status were categorized into ‘ever’ (combining past and current) and ‘never’. The history of disease was classified as ‘yes’ if diabetes, CVD, hypertension, or metabolic syndrome was present. HRs, 95% CIs, and p for interaction for all-cause mortality in the comparison with the highest frequency of nut consumption by a stratified covariate are presented. Statistical significance was defined as a two-sided p-value of less than 0.05.