Study Population
The data reported in the present study come from NHANES collected from 2003 to 2014 (NHANES 2003-2014), which used a stratified, multistage study to recruit participants to identify a nationally representative sample of non-institutionalized civilians in the USA. Details of the NHANES have been published previously [17]. Briefly, we conducted prospective analyses on adult participants (age ≥ 18 years & age ≤ 85 years) in NHANES from 2003 to 2014. The total 27455 participants, including 13359 men and 14096 women, were included in the present study, after excluding those who had missing information on any dietary intake, mortality and other covariables; and those whose total energy intake > 5000 kcal or < 500 kcal in on the day. And before collecting this data, we have obtained approval and written informed consent.
Dietary Assessment
The dietary intake of essential micronutrients among the subjects was assessed using 24-h dietary recalls during two typical, non-consecutive days for each participant. The first data was completed in person while the second by telephone. The dietary intake components are integrated into the 33 main food groups of MyPyramid on the basis of the MyPyramid equivalent database 2.0 (MPED 2.0) of the U.S. Department of Agriculture Survey Food. And according to intake time, these main food categories are divided into breakfast, lunch, dinner and snacks (between breakfast and lunch, between lunch and dinner and after dinner). And the dietary intakes of minerals were further verified with the data collected by a validated semi-quantitative food frequency questionnaire(FFQ) developed for KOGES (Korean Genome Epidemiology Study). All dietary estimates were assigned by using the guidelines of the U.S. Department of Agriculture’s Food and Nutrient Database for Dietary Studies. In brief, the daily dietary intakes of minerals were calculated by combining serving frequency per day, the average amount of serving, and portion per unit.
Main Exposure
The exposure variable that defines the main exposure could be set as those minerals intake in breakfast, lunch and dinner, respectively. The following minerals were studied in the present study: potassium, calcium, sodium, magnesium, selenium, zinc, iron, copper and phosphorus. The sum of breakfast and snack before lunch (<12:00 p.m.), lunch and snack after lunch (12:00 p.m.-18:00 p.m.), dinner and snack after dinner (>18:00 p.m.) were identified as the intake of minerals in the morning, afternoon, and evening, respectively.
Main Outcome
This study that analyzed mortality outcomes depend on the National Death Index (NDI) to ascertain vital status. It is a highly reliable database containing information from death certificates that has achieved widespread use among health and medical investigators. The death from cancer was defined as ICD-10 codes I19-I43. A total of 2680 deaths were available for further analysis, and of these, 601 deaths due to cancer.
Confounding and Effect Modification Measurements
Information on a wide range of variables based on a structured questionnaire just as follows: (1)socio-demographic: age (continuous), sex (male or female), race/ethnicity (non-Hispanic white/non-Hispanic black/Mexican American/other), education level (< 9th grade, 9th-11th grade, high school graduate, general educational development (GED) or equivalent, some college or Associate in Arts degree, or college graduate or above), annual household income (< $20,000, $20,000 - $45,000, $45,000 - $75,000, $75,000 - $100,000, or > $100,000); (2) health behaviors: regular exercise (yes/no), current smoker (yes/no), current drinker (yes/no); (3) physical examination: body weight and height were measured with a standard medical balance and used to calculate the BMI (kg/m2); (4) dietary habits factors: total energy (kcal/day), total dietary calcium (g/day), total dietary potassium (g/day), total dietary magnesium (g/day); breakfast skipping (yes/no); dietary supplement use (yes/no); and diet quality. Diet quality was calculated by the Alternative Healthy Eating Index (AHEI).
Statistical Analysis
The data were checked for completeness, cleaned and analyzed using R 4.0.0 software according to the survey data commands, which consider the sample weights (probability weights), clustering and stratification of the survey design, and then a 5% significance level was considered. Socio-demographic, dietary nutrient intake and health-risk lifestyle factors were expressed as relative frequencies. For categorical variables, frequencies and percentages will be presented. For continuous variables, the number of subjects, mean, standard deviation will be reported. The total minerals in breakfast, lunch and dinner were divided into quintiles based on distribution, respectively. General linear models after adjusting for age and the Chi-square test were used to compare baseline characteristics by mortality status. All statistical analyses were performed by R 4.0.0, and two-sided P < 0.05 was considered to be statistically significant.
Cox Proportional Hazards Models
The Cox proportional-hazards (CPH) regression model has been widely used to analyze time-to-event data with censoring and covariates. In this study, it was used to determine cancer and all-cause mortality, and the results were presented as hazard ratios (HRs) with 99% confidence intervals (CIs). Follow-up survival time was calculated by person-months from the date of the NHANES interview date until the date of death or censoring on December 31, 2015, whichever came first. Important potential confounding variables were often included and controlled in all CPH models, such as age, sex, race, diet energy, BMI, smoking, drinking, education, income, hypertension, diabetes, dyslipidemia, dietary supplement use, AHEI, and total dietary minerals.
Predicted equivalent vitamins Substitution Models
Based on the CPH model above, we also established an equivalent mineral replacement model to evaluate the changes in mortality risk of Cardiovascular Disease (CVD) by switching the intake of one mineral at one-time point to another time point. The critical principle of substitution analysis is that when the intake of total minerals and other nutrients remains constant, the reduction of minerals provided in one period is replaced by the same minerals provided in another. In the present study, we equivalently converted 10% of the intake of potassium, calcium and magnesium from morning or afternoon to evening to observe whether and how the risk of death from cancer changes.
Sensitivity Analysis
Four sets of sensitivity analyses were additionally performed to assess the robustness of the CPH models. In set 1, the association of total dietary minerals intake with cancer and all-cause mortality was performed to examine whether the intake time could provide more information than the total amount of mineral intake. In set 2, we excluded the participants without breakfast consumption to eliminate the effect of breakfast skipping. In set 3, we removed participants with appropriate mineral supplements to minimize the impact of mineral supplements on results. In set 4, we excluded the participants who had a follow-up time of fewer than two years (including those who died within two years of follow-up) to minimize potential reverse causation caused by severe illness, such as cardiovascular disease or cancer.