Ethiopian orthodox fasting is associated with weight reduction and body composition changes among healthy adults: a prospective cohort study

The Ethiopian Orthodox Christian (EOC) fasts, although adopted for religious purposes, combines aspects of energy restriction, time-restricted feeding, and a vegan dietary pattern, all of which have been independently associated with weight loss and healthier body composition. However, combined effect of these practices as part of EOC fast remains unknown. This longitudinal study design evaluated the effect of EOC fasting on body weight and body composition. Information on socio-demographic characteristics, physical activity level, and fasting regimen followed was captured through an interviewer-administered questionnaire. Weight and body composition measurements were taken before and at the end of major fasting seasons. Body composition parameters were measured by bioelectrical impedance (BIA), Tanita® BC-418®, Japan). Significant changes in body weight and body composition were observed for both fasts. Significant decreases in body weight (14/44 day fast: − 0.45; P = 0.004/− 0.65; P = 0.004), FFM (− 0.82; P = 0.002/− 0.41; P < 0.0001), and trunk fat mass (− 0.68; P < 0.0001/− 0.82; P < 0.0001) were observed after adjusting for covariates including age, sex, and physical activity. The EOC fasting regimen leads to significant reductions of body weight and compositions. Longer fasting duration led to much higher effects in body weight and body composition and may be non-pharmacological strategy in prevention or treatment of chronic diseases.

www.nature.com/scientificreports/ forms of fasting with a simultaneous adoption of vegan diet can have beneficial and more sustainable nutritional and health benefits. Unfortunately, long-term evaluation of this practice is scant. Natural experiment among fasting population like among the Greek Orthodox Christian have been conducted, but often targeted monks; hence, limiting the generalizability of the findings to free living adult population 10 . More recently, Sinaga et al. 11 , studied the metabolic effect of fasting among Ethiopian Orthodox followers but the study suffered inherent limitations related to retrospective cohort study design. According to most recent estimates, more than 40% of Ethiopians are Orthodox Christians of which a large majority practicing periodic fasting. Therefore, the present study aimed to evaluate the effect of Ethiopian Orthodox Christian fasting on body weight and body composition by comparing outcomes among fasters that fasted the assumption, nativity and Wednesday and Friday fasts with non-fasters.

Methods and design
Study subjects. A total of 140 subjects (70 fasting and 70-non-fasting) Orthodox Christian followers residing in Hawassa city administration and meeting the inclusion criteria were approached. The inclusion criteria were as follows: apparently healthy subjects from Orthodox Christian communities within the age range of 30-45 years, permanently residing in the area, and not smoking or drinking alcohol (1 or 2 serving/week). Subjects were permanently living in the study area and had no plans of leaving before the completion of the study. The exclusion criteria were as follows: subjects with any health abnormality, pregnant or lactating women, and those planning to leave the study site before the completion of the study.
Ethical consideration. The protocol and the questionnaires were reviewed and cleared by the ethical review board of Natural and Computational Science of Addis Ababa University, Ethiopia (IRB/035/2018). All methods were performed in accordance to the Helsinki Declaration ethical principles for medical research involving human subjects. The nature of the study was fully explained to the study participant and written informed consent was obtained from each participant prior the study. The collected data were kept confidential.
Study design, sampling, and sample size. This study was conducted during the Assumption and Christmas fasting periods from August 2018 to January, 2019 in Hawassa, Ethiopia. The sample size was calculated using GPower to allow comparison between two groups, assuming α = 0.05, power = 0.85, a medium effect size (µ1-µ2)/α = 0.5, correlation of repeated measures (ρ) = 0.6 and considering repeated measures of four time points, as follows: where σ 2 is the assumed common variance in the two groups, µ 1 − µ 2 is the difference in means of the two groups; n is the number of time points and ρ is the assumed correlation of the repeated measures. This yielded a minimum sample size of 38 subjects per group, which was augmented to 55 subjects per arm after accounting for a maximum of 30% loss to follow-up. A total of 110 subjects (55 per arm) completed the study (Fig. 1).  Outcome measures. Anthropometric and body composition measurements were taken at four-time points: one-week before the start and a day before the end of the Assumption fast. For the Nativity (Christmas) fast, measurements were taken a week before the start and at the end of the fast. All the repeated measurements were obtained at the same time of day (± 1 h).
Anthropometry. Height was measured barefoot to the nearest 0.1 cm with a stadiometer. Body weight was measured twice wearing light clothing, to the nearest 100 g using digital scale. Both height and weight were measured following standard procedures. A digital electronic adult scale (ASTO) that measure both weight and height was used for measurement. Body mass index (BMI) was calculated as weight in kilogram divided by the square of height measured in meter.
Body composition. Body composition of the subjects was measured using bio-electric impedance analyzer (Model Tanita® BC-418®, Tokyo, Japan) with a range of 0-200 kg and with an accuracy of ~ 100 g.

Data quality control. Data collection was conducted by experienced interviewers. Anthropometric and
body composition measurements were conducted by an experienced clinical technician following standard procedures.
Statistical analysis. All data were checked for consistency and completeness, coded and entered into Statistical Package for Social Sciences (SPSS) version 23.0 software for analyses. Normality of the continuous data distribution was checked using Shapiro-Wilk test. Descriptive statistics like frequency and percentages were used to describe dependent variable in relation to different categorical data, while mean (percent increase) and SD or standard error was used for continuous data. In addition, estimate of variances in means of study groups was assessed using student's t-test at 95% confidence interval (CI) level. A two-way ANOVA was performed to evaluate time-fast interactions. P-values < 0.05 were considered statistically significant.

Results
The baseline characteristics of the subjects in the fasting and non-fasting group was comparable, with no significant difference in socio-demographic status (Table 1). Subjects in both groups were Orthodox Christians, were mostly government employees, and were in their early thirties. The mean BMI was in the normal range (< 25 kg/ m 2 ). The physical activity level varied between subjects, but were not significantly different between the fasting and non-fasting group. Most of the subjects had a low to moderate physical activity level. At baseline, subjects in the fasting and non-fasting group had comparable body composition parameters except for trunk fat mass that was significantly higher in the non-fasting group (Tables 2 and 3). For 14 days fasting, significant decreases in weight, FFM, and trunk FFM were observed when fasting groups compared to non-fasting group (P < 0.05). However, significant increase in fat percent, fat mass and trunk fat mas were observed in the fasting compared to non-fasting group (P < 0.05). Slight, but non-significant decrease in BMI was found for 14 days fast (Table 2).
Similarly, for 44 days fasting, significant decreases in weight, FFM, and trunk FFM were observed when fasting groups compared to non-fasting group (P < 0.05). However, significant increase in fat percent, fat mass and trunk fat mass were observed for 44 days of fasting in the fasting compared to non-fasting group (P < 0.05) ( Table 3).
For both 14 days and 44 days fast, insignificant change were observed in visceral fat rating (VFR) ( Tables 2  and 3).
The two-way ANOVA analyses showed that there was as significant time x fasting interaction for fat percent and trunk fat mass in the 14 days fast, while this was only significant for trunk fat mass for the 44 days fasts. However, the effects of the 44-day fast were statistically significant for body weight, BMI, fat percentage, FFM, and trunk fat percentage ( Table 4). The linear regression model adjusting for age, sex, marital status, income, education level and physical activity showed that the same body composition parameters changed in the 14 and 44 days fast, but revealed that the effect size increased with increased duration of fasting (44 days vs. 14 days; Table 5).
Within the fasting group, significant changes were observed between baseline and endline on all of the body composition parameters assessed (Tables 2, 3).

Discussion
The present study highlighted that fasting as practiced by the Ethiopian Orthodox Christians lead to significant changes in body weight and body composition. For both the 14 days (assumption) and 44 days (nativity) fasts, significant decreases in weight, FFM, and trunk fat mass were observed in the fasting than in the non-fasting group. However, the magnitude of the changes was more pronounced in the nativity (44 days) than in the assumption (14 days) fast. The decrease in fat free mass was no more significant after adjusting for sex, age, and physical activity. The Ethiopian Orthodox Fasting can be considered as a combination of intermittent fasting, time-restricted fasting, continuous energy restriction, and adoption of a vegan diet. All these practices have been separately studied and have shown significant reductions in body weight and body fat measures 6,11 , but our study has shown that the combination of these fasting regimens also lead to significant weight loss and fat free mass loss, while also leading to a slight gain in fat during the 44 days of fast. The significant reduction in trunk fat without significant changes in visceral fat rating (VFR) in this study may suggest that fasting had more effect in reducing subcutaneous fat.
Various pathways can explain our findings. The most obvious reason is the reduced caloric intake associated with the skipping of meal and the subsequent adherence of a vegan diet that is also often lower in energy 12 . Indeed, studies evaluating energy and nutrient intake of subjects adopting orthodox fasting have shown lower  www.nature.com/scientificreports/ www.nature.com/scientificreports/ energy intake, increased antioxidant and fiber intake 12,13 . The Ethiopian Orthodox fasting, with the skipping of morning meals, also led to a 16 h or more of fasting time per day. Time-restricted feeding, allowing at least 10 h of fasting time was previously found to be associated with reduced body weight and healthier body composition even in the absence of energy restriction 14 . Time-restricted feeding was reported to support robust circadian rhythms whose disruption can increase the risk of metabolic syndrome including obesity, hypertension, insulin resistance and inflammation 15,16 . Previous studies have shown that 10 h time-restricted feeding led to a reduction of − 3.3 kg body weight, − 1.09 of BMI and − 1.01% of body fat percent change in 12 weeks, but similar levels of change or more was achieved in just two weeks (assumption fast) and eight weeks (Christmas fast) with the Ethiopian orthodox fasting 14 . In addition, loss of fat free mass (FFM) was observed after adjusted for age, sex, marital status, income, education level and physical activity level in linear regression model. Although, the EOC fasts had an impact on weight and body composition, it was ineffective in modifying body weight and fat over time, probably because of compensations happening during the non-fasting period. This is explained through both "set" or a "settling" point approaches by which body weight and fat is determined and maintained 17 .
Limitation of the study. The present study has a number of limitations that need to be considered when interpreting our findings. First, this is a natural experiment that enrolled fasting and non-fasting subjects and compared body weight and composition over time. We could not control for residual effects related to previous fasting among the fasting group, but showed that the body composition of the fasting and non-fasting group was comparable at baseline. The study would have benefited from a detailed recording of the foods consumed to characterize the differences in diet quality, which could have provided more explanations to the presence or absence of some changes. Notwithstanding the above limitations, our study has highlighted that the fasting regimen adopted by Ethiopian Orthodox church followers can lead to significant reductions in weight, FFM, body fat and trunk fat mass.
A key criticism about adopting fasting regimens for health purposes is the difficulty to adopt them by the general public.

Conclusion
Fasting is practiced widely among Ethiopian Orthodox followers prevent the increasing burden of overweight/ obesity and related non-communicable diseases. More importantly in settings where such practices are not alien and are culturally accepted, they can be practiced for health purposes in addition to religious reasons. Future studies should investigate the effect of such fasting on micronutrient status, insulin resistance, and cardio metabolic markers and further explore the effect of fasting on circadian rhythms. Table 5. Magnitude of effect size of 14 days and 44 days of fasting on body composition. BMI body mass index, CI confidence interval, Fat% fat percent, FFM fat free mass, VFR visceral fat rating, T.fat percent Trunk fat percent, T.fat mass Trunk fat mass, T.FFM Trunk fat free mass. β [95% CI] and P-values are from linear regression models, adjusted for age, sex, marital status, income, education level and physical activity level; Cohen's d = 2t /√(df); effect size (r y λ) = √ (t 2 / (t 2 + df)).