Phase angle is inversely related to the consumption of ultra-processed foods and positively related to the consumption of minimally processed foods by university students: a cross-sectional study

This is a cross-sectional study of university students. The measurements were performed in June 2022 and January 2023. The participants of the present study were undergraduate students. The vast majority of them were enrolled at the Department of Nutrition and Dietetics (> 98 %). Participants were selected by convenience sampling. Notices were electronically dispatched to the students by the Department of Nutrition and Dietetics secretariat, while advertisements were prominently displayed in frequented University areas, like the cafeteria and the restaurant zone. The eligibility criteria were (i) being a student at the University. Exclusion criteria were (i) not being a student, (ii) not fulfilling prerequisites for BIA measurement (i.e. no strenuous exercise, caffeine and alcohol intake for 8 h before measurement, no foods, liquids for 3 h before measurements, measurements not performed during menses and +/–3 d to avoid oedemas) and (iii) pregnancy. It is noted that no specific age inclusion/exclusion criteria have been applied. The STROBE-nut reporting checklist is given as an online supplementary material, Supplemental File.

A semi-quantitative FFQ was administered⁽²⁴⁾. The FFQ covered a period of one year and consisted of 69 questions, concerning the frequency of intake of several food groups: dairy products, eggs, starchy vegetables, meat and poultry, fish, legumes, vegetables, fruits, sweets, alcohol, stimulants, fats and oils⁽²⁴⁾. Most fruits and vegetables included in the FFQ were available throughout the whole year. To capture the seasonal consumption of fruits, two special questions were present, one for each season (summer fruits and winter fruits). The questionnaire, also, contained 7 questions regarding dietary habits, such as the use of oils and butter in cooking, the intake of visible fat from meat, the frequency of ordering food, the frequency of meals, the total number of meals including snacks, the number of main meals, the preference for organic or soy products and the intake of nutritional supplements⁽²⁴⁾. The possible frequencies, in multiple-choice format, were as follows: ‘never/rarely’, ‘one to three times a month’, ‘one to two times a week’, ‘three to six times a week’, ‘once a day’ and ‘more than twice a day’. For other questions on dietary habits or supplements, a ‘yes’ or ‘no’ format was used⁽²⁴⁾. The FFQ has been validated in the participant population⁽²⁴⁾.

One serving of cereals (refined or non-refined) was considered as 1 slice of bread, ½ cup rice or ½ cup pasta. One serving of potatoes was considered as 1 medium potato. One serving of fruit was considered 1 medium fruit or ¼ of a cup of dried fruits. One serving of vegetables was considered as 1 cup of raw vegetables, ½ cup of cooked vegetables, ½ plate of spinach rice/vegetable rice or 2 slices of vegetable pie (e.g. spinach pie). One serving of legumes was considered 1/2 plate of legumes. One serving of fish was considered 60 g of fish. One serving of red meat was considered as 60 g beef/pork lamb/goat, 2 slices of ham, 2 medium sausages or 4 slices of bacon. One serving of poultry was considered as 60 g of chicken or turkey. One serving of eggs was considered 1 egg. One serving of dairy (full-fat or low-fat) was considered as 1 glass of milk or 1 yogurt or 30 g of cheese/cream cheese. One serving of sweets/treats was considered as 1 piece of pastry/tart, 1 whole croissant, 1 medium wafer, 1 slice of cake, 3–4 cookies, 1 ice cream or 1 milkshake. One serving of potato chips/popcorn was considered as 70 g of potato chips/popcorn. One serving of alcohol was considered as 125 ml of wine (1 glass), 240 ml of beer (1 glass) or 1 glass of other alcoholic drinks (30 ml).

Participants’ responses to the FFQ were also used to calculate the Mediterranean Diet Score (MedDietScore)⁽²⁵⁾. MedDietScore evaluates the individual’s adherence to the Mediterranean Diet, and it is based on the frequency of consumption of the main food groups, on a scale from 0 to 55⁽²⁵⁾.

In addition, the administered FFQ was used to calculate the percentage of energy derived from MPF and UPF, as defined by the NOVA system⁽²⁶⁾. Based on the NOVA classification system, foods are divided into the following 4 categories according to their degree of processing: MPF or non-processed foods, processed cooking ingredients, processed foods and, finally, UPF⁽²⁶⁾. MPF or non-processed foods are eaten ‘as they are’ or have undergone negligible industrial processing, such as fresh or dried or frozen fruits and vegetables, fresh, chilled or frozen meat, fish, eggs and dairy products. They are categorised as class ‘1’ foods. UPF, rated as class ‘4’, are those that have undergone a series of industrial processes, such as carbonated soft drinks, most breakfast cereals and pastries⁽²⁶⁾. Then, the energy contribution of each category was calculated (as % of total energy intake). The USDA database was used to determine the energy values of foods⁽²⁷⁾.

Participants were administered the Hellenic Physical Activity Questionnaire⁽²⁸⁾. It is a self-completed one-page questionnaire that includes data about the physical activity of the previous seven days and has been adapted to the habits of the particular population⁽²⁸⁾. The questions are divided into three sub-categories: physical activity at work, physical activity at home and physical activity for leisure, to increase the examinee’s recall⁽²⁸⁾. Also, for a better assessment of total energy expenditure, habits related to sleep and sedentary life are recorded⁽²⁸⁾. For each activity, the examinee recorded the amount of time spent on it per day⁽²⁸⁾. Metabolic equivalents for each of the physical activities were calculated by the examiner⁽²⁸⁾. The metabolic equivalents were added, and finally, the individual’s energy expenditure was calculated, which was expressed as kilocalories per week and kilocalories per day⁽²⁸⁾.

Body weight was measured with a precision scale (Tanita MC-780, Japan) to the nearest 0·1 kg, and height was measured to the nearest 0·1 cm with a stadiometer (Seca, Hamburg, Germany). The BMI was calculated as weight (in kilograms) divided by height squared (in metres squared) and the students were categorised as underweight, normal weight, or overweight/obese according to WHO criteria. More specifically, underweight students were those with BMI < 18·5 kg/m², normal weight were those with BMI of 18·5–24·99 kg/m², overweight were those with BMI of 25–29·99 kg/m², and obese were those with BMI > 30 kg/m². Waist and hip circumferences were measured with a non-stretchable measuring tape (Seca, Hamburg, Germany) to the nearest 0·1 cm, in a standing position⁽²⁹⁾. When measuring the waist circumference, the measuring tape was placed around the narrowest part of the waist, that is between the last rib and above the level of the navel, while, when measuring the hip circumference, it was placed at a level that ensures that maximum hip circumference is measured, which is at the buttock area.

Body composition and PhA were assessed via multi-frequency BIA (Tanita MC-780, Japan). The device used for BIA measurements is a multi-frequency segmental body composition model, measuring at 5 kHz, 50 kHz and 250 kHz. In the present work, we have used PhA automatically calculated by the device at 50 kHz for all participants, while resistance (R) and reactance (Xc) were also provided. The day before BIA measurements subjects refrained from strenuous exercise (for 8 h), and they did not consume beverages high in caffeine (tea, coffee and energy drinks) nor alcohol⁽²⁹⁾. On the day of measurements, subjects had fasted for at least 3 h (no foods and no liquids)⁽²⁹⁾. Measurements were not performed during menses and +/–3 d to avoid water retention, which could potentially affect measurements. In order to perform measurements, shoes and socks were removed⁽²⁹⁾. Measurements were performed in a standing position. More particularly, subjects stepped on the device with the feet touching the electrodes and the weight of clothes was recorded and automatically subtracted from the total weight measured by the device. When indicated, the subject put his hands on the electrodes, slightly separated from the trunk⁽²⁹⁾, and the measurement was taken.

Lastly, skinfolds were measured at various anatomical sites (biceps, triceps, subscapular and supra-iliac skinfolds) according to standard procedures⁽²⁹⁾. The same caliper was used for all measurements (Slim Guide Caliper, HaB Essentials). The caliper was operated with the right hand, and measurements were made on the right side of the subject’s body. The examiner with the thumb and forefinger of his left hand had to grasp approximately 1 cm above and below the measurement point at the respective anatomical site and separated the skin from the underlying muscle tissue⁽²⁹⁾. After four seconds of applying pressure, the reading was recorded to the nearest 0·2 mm⁽²⁹⁾. The measurement was performed 3 times at each point to minimise measurement error, and the mean value was used. For the measurement of biceps and triceps skinfolds, the midpoint of the arm length (between the acromion and olecranon) was identified and marked⁽²⁹⁾. Then, vertical folds were taken at mid-biceps and mid-triceps levels⁽²⁹⁾. The subscapular skinfold measurement was diagonally taken at 1 cm below the inferior angle of the scapula⁽²⁹⁾, after locating the anatomical site by palpation of the area. For the supra-iliac skinfold, a diagonal fold, just above the front forward protrusion of the hip bone, was considered⁽²⁹⁾.

All measurements were performed by two members of the technical staff of the Department of Nutrition and Dietetics.

For normality testing, the Kolmogorov–Smirnoff test was used. Normally distributed variables are shown as means (sd), and non-normally distributed variables are shown as medians and interquartile range. Categorical variables are presented as frequencies (n, %). For comparisons between men and women, the t test was applied for normally distributed variables or the Mann–Whitney U test for non-normally distributed variables. For categorical variables, the chi-square test was used to compare frequencies between sexes.

Correlations between variables were performed using the Spearman correlation coefficient, since in most variables the normality criterion was not fulfilled. For the calculation of Spearman’s partial correlation coefficients, variables were ranked, and then, the Pearson partial correlation coefficients were calculated, after adjustment for age, sex, physical activity and BMI. Furthermore, Spearman’s partial correlation coefficients were reported in several groups, that is MedDietScore tertiles (low, medium and high adherence to the Mediterranean diet).

All reported P-values are based on two-sided tests, and the level of significance was set at 5 %. IBM SPSS Statistics for Windows 29.0.0.0.241(IBM Corp.) was used for analysis.

In total, 151 students participated, including 114 women and 37 men, aged between 18 and 44. The baseline characteristics of the participants are presented in Table 1. It is noted that overweight and obese subjects were grouped, due to the low frequency of obese subjects. Women had a higher percentage of body fat and higher thickness levels of several skinfolds. Extracellular and intracellular water, as well as lean mass, were higher in male participants. Waist circumference was higher in men. The PhA was 6·8 (6·15–7·35) in men and 5·3 (5·1–5·9) in women (medians, interquartile ranges P < 0·001).

The dietary intake of the participants is shown in Table 2. The medians and interquartile ranges of the MedDietScore, UPF and MPF intake were 31·00 (26·00–34·00), 13·72 (8·18–33·39) and 59·99 (50·16–70·73), respectively, in the total sample. No sex differences were documented in these parameters, while men had higher red meat intake than women. The correlation coefficients of UPF and MPF with food groups are shown in online supplementary material, Supplemental Tables 1↗ and 2↗.

In Table 3, the Spearman correlations between PhA and dietary variables are presented. The association of UPF with PhA was significant in women, while the association of MPF with PhA was significant in both sexes. In a multivariate analysis using Pearson’s partial correlations between PhA and dietary variables (ranked variables) after adjustment for age, physical activity, BMI and sex (where applicable), the associations remained significant (Table 4). It is noted that the associations also remained significant in further models adjusted for muscle mass instead of BMI (rho = –0·253, P = 0·004 and rho = 0·228, P = 0·007 for the association of PhA with UPF and MPF, correspondingly). The associations were further tested in stratified analysis per MedDietScore tertile (Fig. 1). It was shown that the inverse association of UPF and PhA was evident in participants who had low and moderate adherence to the Mediterranean diet (i.e. first and second tertile of the MedDietScore). Interestingly, the positive association of MPF consumption and PhA was evident in those with low adherence to the Mediterranean diet (first tertile of the MedDietScore).