Design and participants
The current investigation is a component of an extensive initiative known as the "Fatigue Management" project. Carried out by the Federal University of Ouro Preto, its primary aim is to ascertain the prevalence of cardiovascular risk factors and fatigue among employees at an iron ore extraction firm situated in the Iron Quadrangle region of central Minas Gerais, Brazil.
For this research, male participants within the age range of 26 to 60 years, engaged in rotational shifts, were extended invitations to join. The rotational shifts followed a schedule involving morning, afternoon, and night shifts, with a work-to-rest ratio of 6 hours on duty and 12 hours off. The overall weekly working hours amounted to 36, and after every four consecutive shifts, a day off was allotted. The four shift cycles spanned from 7 a.m. to 1 p.m., 1 p.m. to 7 p.m., 7 p.m. to 1 a.m., and 1 a.m. to 7 a.m.
The determination of the sample size took into account a previously recorded prevalence of 15.9% for cardiovascular risk in shift workers (Farha & Alefishat, 2018), a precision level of 5%, a design effect of 1.0, and a confidence level of 95%, resulting in a stipulated minimum sample size of 188 individuals. All employees with rotating shifts were approached to partake in the study, and initially, an assessment was made of 366 workers. However, factors such as refusal to participate, vacation periods, absences, dismissals, or incomplete food diary entries led to a final sample size of 213 individuals.
Data collect
Trained teams were responsible for conducting the anthropometric measurements and dietary survey, which took place in the company's outpatient clinics. The collected demographic data included gender, age, self-reported skin color, and education level. Age was categorized as 30 years, 30–40 years, 40–50 years, and 50–60 years; skin color was classified as black, brown, yellow, indigenous, or white; education level was categorized as completed high school, technical education, or university. Clinical evaluation included data on smoking, alcohol consumption, and physical activity. Smoking was divided into non-smokers (individuals who never smoked or quit more than six months ago) and smokers (currently smoking or quitting less than six months ago). Alcohol consumption was classified as yes or no. The International Physical Activity Questionnaire (IPAQ) version 8 - long form was used to assess physical activity levels, with high physical activity defined as > 600 MET-min/week.
Regarding anthropometric data, weight was measured using the TANITA body composition monitor model BC-558, with a maximum capacity of 150 kg and precision of 0.1 kg (Tanita Corporation of America, Inc., Arlington Heights, Illinois, USA). Height was measured using the portable AlturExata stadiometer with a centimeter scale and precision of one millimeter (AlturExata, Belo Horizonte, Minas Gerais, Brazil). Procedures were carried out with individuals barefoot, properly positioned, standing straight, with a straight and fixed gaze ahead. From the collected weight and height information, the Body Mass Index (BMI) was calculated using the formula: weight (kg) / height (m)², considering overweight when BMI values were ≥ 25.0 kg/m² (WHO, 2000).
Blood pressure was measured using a semi-automatic digital device from Microlife, model BP3AC1-1PC (Microlife, Widnau, Switzerland), following the parameters of the Brazilian Society of Cardiology (Précoma et al., 2019). Workers were divided into two groups based on the presence or absence of systemic arterial hypertension (SAH). Blood pressure values were determined by the average of three measurements. SAH was considered for individuals with an average systolic blood pressure equal to or greater than 140 mmHg or diastolic blood pressure equal to or greater than 90 mmHg.
Triglyceride, total cholesterol, and high-density lipoprotein cholesterol (HDL-c) levels were determined by enzymatic colorimetry using the Triglycerides Liquicolor Mono®, Cholesterol Liquicolor®, and Cholesterol HDL Direct-Homogeneous Direct Test® kits (Human do Brasil, Itabira, Brazil), respectively, on an automated Chemwell R6® analyzer (Awareness Technology, Palm City, FL). Low-density lipoprotein cholesterol (LDL-c) was obtained by mathematical calculation using the Friedewald formula (1972), with LDL-c (mg/dL) = Total cholesterol - HDL - (Triglycerides/5) when triglyceride concentration was less than or equal to 400 mg/dL. Participants with plasma triglyceride concentrations exceeding 400 mg/dL were evaluated for their LDL levels using a specific LDL Direct-Test kit.
Explanatory: Food consumption according to NOVA classification
The instrument used to collect data related to dietary intake was the 24-hour dietary recall (R24h), administered by trained interviewers. This method involved gathering information on the timing and location of each meal, the type of food consumed, the preparation method, the quantity in portions, and, if possible, the brand of products. To assist individuals in accurately identifying consumed portions, the book "Food Consumption: Visualizing Portions" by Monteiro (2007) was used as a reference.
Food consumption was classified according to the NOVA classification (Monteiro et al., 2019). The first group consisted of Fresh foods, which are those that have not undergone any type of alteration, being obtained directly from plants or animals, such as vegetables, fresh fruit, grains, roots, tubers, and others. The second group was made up of minimally processed foods, which are fresh foods that have undergone minimal alterations, such as dried, polished, and packaged grains, flours, washed roots and tubers, chilled or frozen meats, and pasteurized milk. The third group was culinary ingredients, which are those used to season and cook food and preparations (salt, sugar, oils, and fats). The fourth group is processed foods, which are the result of adding salt, sugar, and/or fat to fresh or minimally processed foods, such as canned goods, preserves, processed meat and fish, cheeses, and others. The fifth group is ultraprocessed products, which are industrial products that include additives, colorings, and flavor enhancers that make food more attractive to consumers, such as snacks, sweets, cookies, fatty snacks, hamburgers, ice cream (Supplementary Table 1w). In this study, we adopted a dual approach to classifying workers' food consumption, focusing on both the quantity and diversity of food consumed, according to the extent and purpose of processing.
Quantitative classification of food consumption
This classification was based on the daily intake of food from each food group, following the NOVA classification. Therefore, for fresh foods, the intake of fruit and vegetables (FV) was considered, and classified according to the guidelines of the World Health Organization, which recommends the consumption of 400g per day of these foods (WHO, 2020), as they act in the protection and prevention of chronic non-communicable diseases, cardiovascular diseases, and neoplasms (X. Wang et al., 2014). This recommendation serves as a reference for assessing the adequacy of FVL consumption among the study participants. As for ultraprocessed foods, due to the lack of a specific minimum recommendation for UPF consumption, we opted for an analysis based on tertiles of daily caloric intake. The proportion of daily caloric value coming from UPFs was calculated using the formula [(total kcal from UPFs)×100], and the participants were classified into three groups: tercile 1 (T1), with the lowest percentage of caloric intake coming from UPFs; tercile 2 (T2), with intermediate values; and tercile 3 (T3), with the highest percentage of UPF consumption. This classification allows for a differentiated analysis of the possible impact of UPF consumption on the participants.
NOVA dietary diversity score (DDS-NOVA)
The DDS-NOVA is an indicator developed to assess the diversity of workers' diets, focusing on the variety of food items consumed. This score is based on the premise that greater dietary diversity is associated with a more balanced diet and, potentially, a better state of health. Diversity is measured by the presence of different types of food within each food group, as defined by the NOVA classification.
To construct the DDS-NOVA, we used data from 24-hour recall records. Each type of food consumed within a specific food group gives a point to the score. For example, if a worker consumes two different types of fruit, one type of vegetable, and two types of vegetable throughout the day, they will receive a total of five points for the fresh food group. This methodology is applied in the same way to the other food groups: minimally processed, culinary ingredients, processed, and ultraprocessed.
The DDS-NOVA does not take into account the frequency or quantity of consumption, but rather the variety of food items. This allows for a qualitative analysis of the diet, complementing other quantitative measures of food consumption. The choice of this method is justified by the literature, which suggests an association between dietary diversity and a reduced risk of chronic diseases. In addition, the DDS-NOVA is inspired by a study carried out in the city of São Paulo, which validated the use of a similar score (Nova24h screener) to assess the variety of unprocessed or minimally processed whole plant foods (WPF, 33 food items) and ultraprocessed foods (UPF, 23 food items) consumed by a sample of 812 adults. Two scores are obtained from this tool by summing the number of items checked, the Nova-WPF and the Nova-UPF. The scores obtained reflected the number of subgroups of whole plant foods and ultraprocessed foods reported, providing insights into food consumption patterns in the population studied (Costa et al., 2023; dos Santos Costa et al., 2021).
Outcome: Cardiovascular Disease Risk
To estimate cardiovascular risk, we employed the Framingham Global Risk Score (FRS), which calculates the 10-year risk of coronary events, cerebrovascular events, peripheral arterial disease, or heart failure. Recommended by the Atherosclerosis Department of the Brazilian Society of Cardiology (SBC-DA), the FRS is a scoring system where each characteristic (gender, age, total cholesterol, HDL-c, systolic blood pressure, diabetes, and smoking) is assigned a corresponding score. The sum of points for all variables yields a percentage indicating the 10-year risk of cardiovascular disease. The risk categories were further divided into two groups: low cardiovascular risk (< 5%) and intermediate to high cardiovascular risk (≥ 5%) (Précoma et al., 2019).
Statistical analysis
The statistical analysis of the dataset was conducted using Stata/MP software (version 15.0). The description of variables included presenting absolute frequency values (n) and relative percentages (%), with cardiovascular risk analyzed through the Pearson chi-square test.
To assess the association of food consumption variables with cardiovascular risk, both univariate and multivariate logistic regression analyses were conducted to determine the odds ratio (OR) values and their corresponding 95% confidence intervals (CI). The explanatory variables were classified as follows: Quantitative consumption of fruit and vegetables was classified dichotomously as low and high consumption, based on the recommendations of the World Health Organization. Quantitative consumption of UPFs is classified into three categories according to the tertiles of the daily calorie percentage from UPFs. The DDS-NOVA score was assessed ordinally, keeping the original score as a continuous ordinal variable in the models.
The multivariate model was designed to incorporate covariates considered as confounding factors in the analysis, following the literature (Martinez-Gonzalez & Bes-Rastrollo, 2014; Qu et al., 2024; Santiago et al., 2017). This encompassed sociodemographic, behavioral, and dietary factors that impact the relationships between dietary intake and cardiovascular risk. Thus, the multivariate model was adjusted for the following variables: age, skin color, schooling, time working shifts, physical activity, body mass index, and total caloric intake. Collinearity among the covariates was assessed by calculating the variance inflation factor (VIF), with the "subsetByVIF" package in Stata, considering a maximum cutoff point of 10 (VIF < 10).