Veganism and body weight: An N of 1 self-experiment

The causal effect that veganism has on body weight has not been scientifically examined. An N of 1 self-experiment was conducted in which blinded body weight and additional behavioural and psychological measures were assessed during two phases of vegan vs. non-vegan lifestyle adherence. In study phase 1, body weight change was -0.8 kg over 1 month of veganism (vs. + 0.5 kg non-vegan month). In study phase 2, weight change was -1.2 kg over two veganism months (vs. + 1.6 kg non-vegan months). Behavioural and psychological measures were similar during vegan vs. non-vegan periods. Veganism appeared to reduce body weight in this N of 1 self-experiment.


Introduction
Veganism is defined by the Vegan Society in dietary terms as 'avoiding all animal foods such as meat (including fish, shellfish and insects), dairy, eggs and honey [1]. ' The adoption of veganism in many countries has grown rapidly in recent years [2]. For example, in the UK the number of self-identified vegans has grown year on year since 2014 and an estimated 3% of the UK adult population identified as being vegan in 2023 [3]. Motivations for adopting vegetarian and vegan diets can differ, but include perceived health benefits, environmental concerns and animal rights [4][5][6].
Although some research has examined relationships between vegetarianism and health outcomes, studies specifically examining the potential impact of veganism on health-related outcomes are limited [7]. This is of importance because vegetarians and vegans can have markedly different dietary motivations and patterns [8][9][10]. In particular, there is a dearth of evidence on the impact that following a vegan lifestyle has on body weight. A small number of observational studies tend to indicate that veganism is associated cross-sectionally with a lower BMI and reduced likelihood of overweight/obesity [7,11,12]. However, a major limitation of this type of observational research is that it does not allow for causal inference. Whilst it is plausible that veganism causes lower body weight, people with a lower body weight may be more likely to adopt veganism, and/or confounding unmeasured third variables associated with both body weight and veganism could explain findings [7]. For example, higher social class is associated with both lower body weight and some research suggests vegans tend to have higher education levels than meat eaters [13]. Likewise, regular physical activity is associated with lower body weight and individuals with higher physical activity levels are more likely to adopt a vegan diet than those with lower activity levels [14].
A small number of randomized control trials (RCTs) have compared the effect of prescribed diets that are low in fat (and/or energy restricted) and predominantly vegan vs. existing habitual diet (i.e., a non-active control) [7,15,16]. These studies suggest that weight loss can be achieved on prescribed low fat/energy restricted diets that are vegan [15][16][17]. However, due to their design these studies do not provide information on the causal impact that adopting a vegan lifestyle has on body weight.
An alternative approach to understanding potential causal impact of veganism on health related outcomes are N of 1 studies. N of 1 studies often involve studying the effect of early stage medical interventions in single or small numbers of patients. Although N of 1 studies have become more common in some research fields [18], they are an underutilised methodological tool in modern day nutrition research [19]. However, N of 1 studies have been used in the context of nutrition research more historically in the form of self-experimentation. Examples include John Crandon studying the relationship between vitamin C and scurvy (1939) and Victor Herbet using self-experimentation to examine the effects of folic acid deficiency in 1961 [20].
Given the increasing trend in adoption of veganism but a dearth of evidence on its causal impact on body weight, an N of 1 self-experiment of the effect of veganism on body weight was conducted. Regular measurement of body weight was made during vegan vs. non-vegan study periods, as well as measurement of other behavioural and psychological factors that could be affected by, or differ between vegan vs. E-mail address: eric.robinson@liv.ac.uk.

Design. N of 1 experiment
Participant. Physically active male aged 35 at beginning of 2022. After a largely pescatarian diet during my early 30s, in early 2020 I decided to try veganism. During 2020 to 2021 I followed a predominantly vegan lifestyle. In December 2021, I was 6′2 inches tall and weighed 76.2 kg. The experiment began on January 1st, 2022, and I weighed 77.7 kg (see Fig. 1). The 1.5 kg increase in weight during December 2021 is in line with population level holiday season weight gain phenomena [21,22,23].

Phase 1 of the experiment
During January 2022, I followed a strict vegan lifestyle, including diet. This included intentionally not consuming meat, fish, dairy or any animal derived food (e.g., honey) or beverages. Then, in February 2022 I followed a non-vegan diet and actively consumed meals that were not vegan. For examples of dietary habits see Table 1. Because enjoyment of food is a known driver of energy intake and there is some anecdotal evidence that vegan diets are less palatable than non-vegan foods [24,25], during phase 1 of the experiment, immediately prior to sleep every evening I rated how enjoyable the food I ate that day was (0-100 visual analogue scale, anchors 'not at all' to 'extremely' enjoyable). To examine if eating habits that could markedly affect energy intake differed between vegan vs. non-vegan study periods I recorded how many times in total I ate each day (including vegan and non-vegan meals) [26], number of eating occasions with another person [27] and outside of the home [28], as well as whether I'd drank alcohol [29] or exercised [30] (both binary). To ensure I was blinded to body weight measurements, every week (typically at the weekend) my spouse followed a study protocol that involved weighing me on a digital scale and measuring my waist circumference (see supplementary materials for full details). In addition, because veganism tends to be associated with an increased risk of depression [13,31,32] but a reduced risk of anxiety symptoms [31,33,34] and changes to mental health may cause weight gain, on a weekly basis I reported on depressive symptoms (PHQ9, 0-10 scale) [35] and anxiety symptoms (GAD7, 0-7 scale) [36] for the past week. Daily energy intake was not measured during the study as the only existing tools to estimate free-living daily diet are based on self-report and such measures are time consuming and also argued to be grossly inaccurate in quantifying energy intake [37].
Study phase 1 results. As expected, the percentage of vegan eating occasions was higher per day during the January vegan month (100% of meals) vs. February non-vegan month (16% of meals classed as vegan). See Table 2 Fig. 3. This pattern of weight change did not appear to be explained by measured eating habits: for vegan vs. non-vegan days, number of times eating per day, number of times eating out of the home and with others were all similar, as were exercise and alcohol drinking frequency. See Table 2.

Non-study phase of the experiment
During March to July (non-study phase), I followed a largely vegan diet (albeit not strictly) and measured non-blinded body regularly (see Fig. 1). Weight of 77.4 kg (83 cm waist circumference) measured at the end of February reduced to 75.3 kg (81 cm) by the end of July. The 1.9 kg decrease in body weight could have been caused by my largely vegan diet during this period or because I was regularly self-weighing and selfweighing is associated with weight loss [38].

Phase 2 of the experiment
To address some of the limitations of phase 1 of the experiment, phase 2 differed. First, I made the measurement periods twice as long and reversed the order of the non-vegan (August -September) and vegan (October -November) periods. Weekly body weight and mental health measurements were the same as in phase 1. However, the daily ratings in phase 1 were not completed to make phase 2 more naturalistic and rule out regular self-reporting of diet related behaviour affecting body weight. During the non-vegan period, from a starting point in August of 75.3 kg my body weight increased to 76.9 kg by end of September (total weight gain of 1.6 kg). Waist circumference remained the same (82 cm). During the vegan period, from a starting point of 76.9 kg (82 cm) in October my body weight decreased to 75.7 kg (81 cm) by end of November, equating to a total weight loss of − 1.2 kg and − 1 cm waist circumference change. See Fig. 3.
Mental health. As mental health was only measured weekly and subject to minimal fluctuation, depression and anxiety symptoms for vegan vs. non-vegan weeks were examined across both phases of the experiment to maximise analytic sample size. Both anxiety (median value for both periods = 0, Z = 0.28, p = 0.78) and depression (median value for both periods = 1, Z = 0.12, p = 0.90) symptoms were close to identical during the vegan (n = 14) vs. non-vegan (n = 13) weeks.  1 As a% of meals per day, median% of non-vegan meals was 0% (vegan) and 84% (non-vegan). 2 As a% of meals per day, median% of meals eaten with others was 25% (vegan) and 33% (non-vegan). 3 As a% of meals per day, median% of meals eaten out of home was 33% (vegan) and 33% (non-vegan) Z=Mann-Whitney U test, X 2 = Chi squared test. Fig. 2. Box plots are interquartile range with median (middle).

Energy intake estimation and future body weight modelling
Self-reported daily exercise frequency was similar during the vegan vs. non-vegan month (phase 1 of study) and mobile phone pedometer data (see online supplementary materials) available for phase 1 and 2 of the study indicated that during non-vegan (13,278 steps per day) and vegan month (13,333 steps per day) activity was very similar (0.04% difference). Assuming changes in body weight observed were due to unmeasured energy intake differences, a validated energy balance calculator [39] was used to estimate daily energy intakes corresponding to weight change in each of the separate vegan and non-vegan study phases. During non-vegan periods of the study, modelling estimated − 302kcals/per day vs. vegan periods vs. non-vegan periods. Adopting non-veganism (as opposed to maintaining strict veganism) would result in a difference of +6.4 kg (8%) in my body weight over 12 months, based on these daily kcal difference. Modelling over longer time horizons produced similar weight change estimates and stabilised at a + 7.6 kg difference in my body weight by 36 months.

Discussion
In this N = 1 self-experiment, during one-and two-month periods of following a strict vegan lifestyle a decrease in body weight was observed, compared to a non-vegan lifestyle. A small decrease in waist circumference (1 cm average) was observed during vegan relative to non-vegan study periods.
Frequency of physical activity, eating frequency and other eating habits were similar for vegan vs. non-vegan days when measured. An average difference of 302kcals per day was estimated to fully account for difference in body weight trajectories between vegan vs. non-vegan study periods. Diets adopted by vegans tend to be higher in fibre, fruits and vegetables and lower in fat [40,41], which may contribute to reduced energy intake and body weight. Higher dietary fibre during meals is associated with small decreases in energy absorption [42], which could have also theoretically made a small contribution to lower body weight during vegan study periods and why veganism is associated with lower body weight in other studies. A further explanation of decreased energy intake during vegan (vs. non-vegan) study periods was my anecdotal experience of their being a less broad range and variety of food and drink options when shopping or eating out. In line with this, both availability of food and food variety stimulate appetite and energy intake [43,44], and this may in part explain why periods of non-veganism (i.e., greater variety) were associated with relative weight gain. Irrespective of mechanism, the present findings are consistent with veganism being a potential cause of lower body weight, which is consistent with veganism being negatively associated with body weight in previous non-experimental cross-sectional studies [7]. Therefore, in addition to the now well acknowledged environmental benefits of plant-based diets [45], the present study in part supports recent suggestions that wider adoption of veganism may also be beneficial to weight gain prevention and reduction of overweight/obesity [46].
There was no difference in anxiety or depressive symptoms between vegan vs. non-vegan study periods. Vegans are more likely to report higher levels of depression, but lower levels of anxiety compared to nonvegans in observational studies [31]. The present study findings are not consistent with veganism having an immediate causal effect on anxiety or depressive symptoms. Rather than veganism causing or contributing to depression or anxiety, an alternate interpretation of these associations is an unmeasured third variable or that individuals with distinct mental health profiles are more likely to adopt a vegan diet (e.g., potentially in the hope of improving depression). Of further note, rated daily enjoyment of food was very similar for vegan vs. non-vegan periods in the present study, which contradicts anecdotal assumptions about vegan friendly food being less palatable [25].
3.1. Strengths, weaknesses and future research N = 1 designs are a valuable but underutilised methodological approach in health and diet research [18,47] and the present research highlights the value of adopting this approach to better understand the potential causal impact that lifestyle habits have on body weight. An important challenge of using N = 1 designs is 'carry over' effects, whereby effects of an intervention or change in lifestyle continue to affect outcomes of interest after that intervention has been removed Fig. 3. Green and red shading denotes vegan and non-vegan month. [18]. Here this was addressed by having multiple study phases, including a mid-study washout period, and alternating order of vegan vs. non-vegan study periods. The extent to which the present findings will generalise to other populations is an empirical question faced by all N = 1 studies and self-experimentation [20]. The present findings suggest that further research on the causal impact that veganism has on body weight is warranted; studies with larger sample sizes and use of randomized control designs would be preferable.
An additional consideration for future research is the impact veganism has on other diet-related health outcomes. Nutritional adequacy of diet was not assessed in the present study. Although veganism is associated with a diet that is lower in saturated fat and higher in fibre, without care and/or adequate dietary supplementation, veganism could cause nutritional deficiencies (e.g., vitamin D, B12, calcium) [48,49] which may contribute to increased risk of bone fracture [50]. Any such vegan diet induced nutritional deficiencies would take time to develop. In the present study veganism vs. non-veganism was examined over 1-2 month periods. To better understand the effect vegan lifestyles have on health-related outcomes studies of longer duration will be needed and of importance, as an increasing proportion of the population are adopting veganism [3,47]

Conclusions
Self-experimentation and N of 1 studies are underused but potentially valuable research tools to better understand the causal impacts on physical, psychological and mental health-related outcomes of lifestyle choices. Veganism appeared to reduce body weight in this N of 1 selfexperiment, but did not impact on mental health or daily enjoyment of food. Further research on the causal impact that veganism has on long-term body weight and other health-related outcomes would now be valuable.

Ethical approval
Ethical approval was deemed unnecessary due to the nature of the study.

Data sharing
The data set is available on request from the author.

Funding
My salary was in part supported by funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (Grant reference: PIDS, 803194).

Declaration of Competing Interest
I have been a named investigator on previous unrelated research projects funded by the American Beverage Association and Unilever.

Data availability
Data will be made available on request.