We utilized previously unpublished PA content data which were obtained by chemical analysis of SBG samples collected in Greece, during two previously reported surveys in 2015 and 2021. Approximately twenty samples were obtained for each SBG category. Detailed sampling procedures during both sampling periods have already been reported [12, 16]. In summary, seven SBG categories were included: (1) cheese pies made with phyllo pastry (characterized by thin sheets of unleavened dough of crispy texture when baked, made from flour and water; typically prepared by brushing with olive oil or other types of vegetable oil between sheets of dough), (2) cheese pies made with shortcrust pastry, (3) cheese pies made with puff pastry, (4) bougatsa with cheese (samples collected only in 2021), (5) peinirli (i.e., pizza boat with cheese and vegetables and/or processed meat)/pizza, (6) vegetarian pies (spinach or leek pies without cheese typically made with phyllo pastry and olive or other types of vegetable oil), (7) meat containing pies (sausage or ham pies) typically made with puff pastry. During both sampling periods, all samples were collected from Attiki prefecture (Athens greater metropolitan area) by inspectors of the Hellenic Food Authority, from both major Greek bakery chains that are present within but also outside Athens metropolitan area, as well as from artisan bakeries (small businesses). Prices of purchase were also documented while on site, in 2021.

The procedures for the estimation of PA content in SBGs in 2015 and in 2021 were identical. The chemical analyses were performed by the same scientific personnel and have previously been described in detail [12]. Briefly, each sample was homogenized and separated into 2 sub-samples that were stored at 4 °C and analyzed within 2d. The total fat content of the samples was determined by the Soxhlet method [17]. All fat samples were converted to their methyl esters (FAME) with an in-house method based on AOAC 966.06 and Roese-Gottlieb method (AOAC 905.02) [18] to avoid any possible alteration of the FAME profile [12]. The fatty acids profile was determined by Agilent 7890 A Network Gas Chromatograph (Agilent Technologies, Santa Clara, CA, USA) coupled to a flame ionization detector (FID). In the two surveys held in 2015 and 2021, two different columns were used, an Agilent (100 m×0.25 mm×0.2 μm) and a Thermo Fisher Scientific TR-FAME (50 m × 0.22 mm, 0.25 μm), respectively. Four commercial standards purchased from Supelco (Sigma) were used for the identification of the chromatograph peaks [i.e., a FAME Mix C4–C24 (n = 37 components), a trans-9-elaidic methyl ester standard (18:1), a linoleic acid methyl ester isomer mix (18:2) (n = 4) and linolenic acid methyl ester isomer mix (18:3) (n = 8)].

The Hellenic National Nutrition and Health Survey (HNNHS) food consumption data of adult SBG consumers (N = 843, 56% females, 23.4% of all adult HNNHS participants) were used to estimate PA intakes using substitution models. The HNNHS surveyed a nationally representative sample of non-institutionalized, non-pregnant, and non-breastfeeding Greek adults. Food consumption was recorded through two non-consecutive 24-hour dietary recalls, 8–20 days apart, using the Automated Multiple Pass Method. Participants reported quantities using standardized grids, mounts, and verified food atlases. Detailed methodology and questionnaires from the HNNHS are available in previous publications [19]. The HNNHS study was approved by the Agricultural University of Athens Ethics Committee and the Hellenic Data Protection Authority, with all participants providing written informed consent. Consumption data, collected from September 2013 to May 2015, were combined with the results from the chemical analyses of SBG fatty acid profiles for further investigation.

The HNNHS food consumption dataset was updated to incorporate measured PA content of SBGs sampled for the substitution models [12, 20]. In particular, the database was updated with 2015 and 2021 PA measurements on Greek often-consumed SBGs, with adding extra columns of updated nutritional information, next to the previously analyzed values of total SFA, allowing two models generated from data obtained at two different time frames to be used. Keeping SBG consumption constant (the assumption of no changes in intake was made for substitution reasons) but modifying PA content as measured, differences were calculated by comparing 2021 to 2015 assessments. Other dietary composition and intakes were also expected to remain unchanged throughout this procedure. The daily PA intakes from these products (g/day) were calculated, by multiplying the mean PA content per SBG (g in 100 g) with the product’s daily consumption (g/day) for each consumer. The PA content was adjusted based on the individual’s average energy intake as a proportion of participants’ total daily energy intake to have comparable data. Subsequently, the estimated daily intakes were summed per individual and averaged over the number of reporting days (90.6% of the participants had two 24 h recalls). The % contribution of each SBG to the overall PA intake from SBGs was additionally estimated. The SFA intakes from SBGs and in total had been previously estimated, also as % of daily total energy intakes [12, 20].

Adult SBG consumers were classified into three age groups: 19–44 years, 45–59 years and ≥ 60 years. The marital status categories used were single, married/cohabiting, divorced/separated, and widowed and the educational levels were: up to 6 years, 12 years, and higher education (including colleges). Employment status was grouped into unemployed, employed, and pension. Physical activity level was defined according to calculation standards as low, sedentary, moderate and high based on results from the International Physical Activity Questionnaire IPAQ [21]. Smoking status was also assessed and subjects were classified as current- ex- or non- smokers. Current smokers were those that had smoked at least 100 cigarettes in their lifetime and reported smoking daily or on some days at the interview, ex-smokers had smoked at least 100 cigarettes but not in the past 30 days and never smokers had never smoked at least 100 cigarettes in total [22]. Body Mass Index (BMI; kg/m2) was calculated using participants’ weight and height. According to WHO, weight status was categorized as healthy (BMI < 25 kg/m2), overweight (25 ≤ BMI < 30 kg/m2), or obese (BMI ≥ 30 kg/m2) [23]. Disease status for hypertension, Type 2 Diabetes, hyperlipidemia, and/or CVD was assessed based on the following characteristics: (a) a self-reported positive response, (b) a documented diagnosis by a medical professional and current use of respective medications (as confirmed by the trained interviewer), or (c) an abnormal profile as determined by blood sampling (lipids, fasting glucose) or average blood pressure over 140/90 mm Hg [19]. Participants with CVD included individuals with arrhythmia, coronary heart disease, myocardial infarction, angina, heart failure, or a history of stroke, whereas those with no hypertension, dyslipidemia, or any other CVD-related conditions were categorized as healthy for further analysis. Differences between the total study sample (N = 843) and sub-totals per variables examined (e.g. sex, age, etc.), as presented in Tables and Figures, are attributed to missing data.

Baseline variables and PA intake from SBGs were stratified according to total SFA intakes (% of total daily energy intake) to identify statistically significant differences between intakes (p-value, p-trend value). Contribution of different SBGs to total PA derived energy intake was also estimated by age group and health status, as well as the correlation between mean PA content and purchase price, all using 2021 composition data. Means (Standard Deviations, SD) were used to describe normally distributed continuous variables and medians (25th, 75th percentiles) were used for skewed distributions. Categorical variables were expressed as frequencies, and between group distribution differences were examined using chi square for proportions. ANOVA or Kruskal Wallis rank sum were used for continuous data, depending on distribution. Pearson’s function was used to test the correlation between price of purchase and PA content of each type of SBG product. Level of significance was set at 5%. Statistical analysis was done using STATA 13.0 (Texas ltd, Texas, USA).