Considering fat as the primary outcome, we calculated that based on a significant level of 0.05 and 80% power, we need to collect before and after Ramadan food intake data from 25 subjects to detect a significant difference of 5%. Using the “lme4” package in R (version 4.1.0), we constructed mixed-effects models adjusted for fixed (age and sex) and random variables (mosque) to estimate the transition in diet during Ramadan. In addition, we applied a False Discovery Rate (FDR) method to correct for multiple testing. The same test was applied to metabolites for men and women separately to investigate whether the effect of RIF on metabolites varies between men and women and impacts the metabolomic changes upon the observance of RIF. This analysis was performed using the “lmer” package in R. We presented all variables in the form of mean ± SD before and during Ramadan and all results as a mean difference with 95% CI. Results with an FDR-adjusted p-value of less than 0.05 were considered statistically significant.

To explore to what extent the changes in diet explain the changes in metabolites, we calculated the absolute differences in nutrients (intake during minus intake before Ramadan) and metabolites (before and after Ramadan) that changed significantly. Afterwards, each metabolite’s absolute difference was added to a basic linear regression model as a dependent variable with age, sex, number of fasting days and mosque as independent variables, and the R-squared value was reported. Then, a nutrient was added to the same model to report the change in the R-squared value (compared to the original R-squared value in the basic model), which was interpreted as the proportion of the metabolite’s change explained by the change in the nutrient intakes. The same process was repeated for SBP, DBP, weight, WC, HC, BMI, FM, F%, FFM, and TBW as dependent variables that changed after RIF (22, 25).

In a general population from different ethnic backgrounds, RIF was associated with a diet transition in which the intake of 11 dietary variables changed significantly. The main changes were increased sugar consumption and reduced consumption of starch. Despite an observed change of up to 10% in the intake of macronutrients, none of these changes were statistically significant. Observing changes in some nutritional factors and no changes in macronutrients indicates changes in food choices during Ramadan.

The diet adopted during Ramadan could explain a limited part of the changes in the metabolic profile following RIF. Consequently, a greater proportion of the changes in the metabolic profile is assumed to be due to the underlying mechanisms of fasting and other factors such as gut microbiome or lifestyle related changes (e.g., sleep, physical activity). This assumption is in agreement with the reported results in previous studies when adjusting for energy intake did not influence changes in blood pressure or body composition (22, 25). Moreover, the nutrient intakes changes did not explain much of the changes in SBP, DBP, weight, BMI, WC, HC, FM, F% and TBW.

In LORANS, we investigated earlier changes in metabolites, anthropometry and blood pressure after RIF. We reported significant changes in 14 metabolite levels (normalized between 0 and 1) after RIF (26), namely lactate, acetate, glycoprotein acetyls, triglycerides in large high-density lipoprotein (HDL), triglycerides in intermediate-density lipoprotein (IDL), triglycerides in medium HDL, triglycerides in large low-density lipoprotein (LDL), triglycerides in HDL, tyrosine, pyruvate, acetone, triglycerides in LDL, phospholipids in very small very-low-density lipoprotein (VLDL), triglycerides in very small VLDL (β = −0.31, P = <0.001; −0.22, <0.001; −0.07, 0.006; −0.06, 0.006; −0.06, 0.007; −0.08, 0.013; −0.05, 0.014; −0.07, 0.019; −0.1, 0.019; −0.09, 0.019; 0.10, 0.019; −0.05, 0.035; −0.05, 0.041; −0.05, 0.041). However, the current study found that changes in metabolites after RIF are sex-dependent. Moreover, in terms of SBP, DBP and anthropometric measurements, we observed a reduction (22, 25) after RIF in SBP (−7.3, <0.001), DBP (−3.4, <0.001), weight (−1.6, <0.001), body mass index (−0.60, <0.001), waist circumference (−1.95, 0.011), hip circumference (−2.86, <0.001), fat mass (−1.24, <0.001), fat percentage (−1.05, <0.001) and total body water (−1.45, <0.001).

Although LORANS is the first study to report the changes in diet during Ramadan in a multi-ethnic group, its results agreed with studies that recruited samples from one ethnic background. In LORANS, the caloric intake was the same before and during Ramadan. This finding is consistent with most previous studies (3, 14, 19, 20, 27). Similarly, the intake and percentage contribution of the three macronutrients (protein, fat and carbohydrates) during Ramadan was the same as before. Likewise, numerous studies observed no change in the protein (1, 14, 17, 19, 20, 27, 28), fat (1, 13, 14, 19, 28) and carbohydrates (3, 19, 20) intake. Still, some studies reported a significant increase or decrease in some of these macronutrients (3, 15, 18, 20). The observed inconsistency could potentially stem from the fact that these studies were done in various countries, given that diet is tightly linked to culture. Nevertheless, two of the former studies were conducted in the same country (Iran) and yet reported inconsistent changes in macronutrients.

In LORANS, the intakes of nutritional factors such as cholesterol, sugars and starch were investigated. In agreement with a former study by Lamri-Senhadji et al. (13), we did not observe changes in cholesterol intake. The increase in sugar consumption by more than 50% in LORANS is in accordance with all former studies that observed higher sugar intake (3, 14, 19). None of the previous studies investigated the change in starch consumption, so we could not directly compare the observed decrease in our research to other studies. However, some foods rich in starch were found to be consumed less during Ramadan in some former studies (14, 18).

Although some former studies (29, 30) had assessed the changes in metabolites and lipids after RIF, the direct comparison of observed metabolic changes in LORANS to previous studies was not possible due to the use of different metabolic profiling methods.

It is well-established that sugar intake is negatively associated with HDL and positively with total triglycerides (31); however, the increased sugar consumption in this study was associated with a non-significant reduction in HDL and no change in total triglycerides. Also, this cannot be explained by the non-significant drop in total fat intake and the different types of fat (saturated fat, monounsaturated and polyunsaturated). This might be due to the limited statistical power of our study and the effect of change in sugar intake in a month might be smaller than what our study could have detected. Moreover, the change in ketone bodies was almost independent of the changes in daily food intake. This was expected as the increased abundance of ketone bodies is due to compensating for the absence of glucose intake for more than 8 h, regardless of quantity of glucose intake before starting fasting, through converting ketogenic amino acids and fatty acids into ketones (32).

This study has several strengths. Usin a 3-day food diary, including one weekend day, is a rigorous method to observe the day-to-day variation in food intake across the different days of the week compared to other methods (33). Also, the food diary reduced the risk of recall bias compared to the 24-h dietary recalls, dietary history and food frequency questionnaires. Translating the food diaries into nutrient data using “Nutritics” may have reduced human error in calculating the nutrients. In addition, LORANS is the only study investigating the dietary change during Ramadan in a multi-ethnic sample, giving an overall picture of the changes rather than specific changes in a certain culture. Moreover, the study was community-based, which makes it generalizable. Furthermore, utilizing the collected metabolomics data alongside the dietary data emphasized the magnitude of fasting effect on the metabolic profile apart from the diet adopted during Ramadan.

Also, this study presents some limitations. Although using the 3-day food diary was optimal for this study, it may have discouraged some participants from providing the dietary data due to the time and effort required and led to a low response rate (38.5%). Still, the number of participants who completed the food diary makes it among the largest studies on this topic, especially studies in which a food diary was used to measure diet.

Having a period of 8–12 days between the end of Ramadan and the second visit may have diluted the effect on the metabolic profile. In turn, it may have weakened the explanation of the change in the metabolic profile by diet change. However, reporting a significant difference in 14 metabolites after Ramadan indicates that the effect is still detectable.