Breaking the fast: Meal patterns and beliefs about healthy eating style are associated with adherence to intermittent fasting diets. Appetite , 32-39.

22 Many believe that eating three meals each day is healthy and that skipping meals can be 23 detrimental. What remains unclear is whether this belief undermines attempts to restrict energy 24 intake by skipping meals. In an online survey, participants ( N =312) with experience of 25 intermittent fasting (IF) reported their beliefs about healthy meal and snack frequency, as well 26 as their non-fasting-day and fasting-day eating patterns. They also reported their level of 27 concern with fasting-day meal patterns and their concern to generate fullness when selecting 28 foods. Individuals currently following an IF diet (Current-IF dieters) and those who had 29 previously attempted an IF diet but were non-adherent (Former-IF dieters) took part. Former- 30 IF dieters were more likely to believe that it is healthy to eat three meals a day, punctuated by 31 several snacks. On fasting-days, Former-IF dieters were also more likely to eat breakfast, a 32 mid-morning snack, lunch, and a mid-afternoon snack whereas Current-IF dieters tended to 33 save their eating for dinner and a late evening snack. Former-IF dieters were also more likely 34 to be concerned about the negative consequences of missing a meal, to eat in anticipation of 35 future hunger, and to prioritise fullness over taste when selecting foods. These findings reveal 36 how beliefs about a healthy eating style can play an important role in shaping dietary patterns. 37 Interventions aimed at modifying beliefs about healthy meal patterns may promote IF diet 38 adherence. 39


Introduction 44
Perceptions of healthy eating behaviour may shape individuals' meal and snack patterns 45 and food choices. 'Breakfast is the most important meal of the day' is a common saying in both 46 lay and medical discourse. Indeed, the National Health Service (NHS, UK) promotes this idea 47 increasingly popular. A common barrier, reported by individuals following a traditional 78 continuous energy restriction (CER) diet, is difficulty adhering to the level of restriction that is 79 prescribed (Varady, 2011). It has been suggested that dieting on intermittent days leads to 80 positive cognitive side effects (e.g. improved mood, increased willpower) both because energy 81 restriction is required only intermittently and because individuals find it easy to comply with a 82 diet that permits 'normal' eating on non-fasting days ( In this study, individuals who currently follow an IF diet (Current-IF dieters) were 87 compared with those who previously followed an IF diet, but who stopped (Former-IF dieters). 88 The primary aim was to understand the characteristics that differentiate individuals who are 89 successful at adhering to an IF diet. More specifically, we explored whether underlying beliefs 90 about healthy meal and snack patterns might differentiate people who find it easy to adhere to 91 an IF diet. We hypothesised that individuals who believe it is unhealthy to skip meals or snacks 92 intermittently will be less likely to skip meals or snacks and will be less likely to adhere to an 93 IF diet. We also explored differences between the food choices of Current-IF and Former-IF 94 dieters, on both fasting and non-fasting days. In particular, we focused on the extent to which 95 different dieters select foods based on their expected satiation (ability to confer fullness) versus 96 their palatability (pleasurable taste). Many individuals express concern about feeling 97 sufficiently full  and may prioritise and select foods based on the satiety 98 they are expected to confer (Brunstrom & Rogers, 2009). We hypothesised that these 99 individuals will be less likely to adhere to an IF diet because they are likely to feel especially 100 deprived on fasting days. 101 IF diets may also require a shift from a traditional three-meals-a-day pattern of eating 102 and that this might lead to several concerns. Specifically, following an IF diet might lead to 103 feelings of missing out on the pleasure of eating, missing out on eating with family at meal 104 times, or feeling that fasting is otherwise disruptive to an individual's daily schedule. 105 Therefore, an additional aim was to assess whether non-adherence is associated with perceived 106 deprivation or inconvenience on fasting-days. Finally, the belief in a meal-to-meal model of 107 energy balancing may promote eating to avoid perceived adverse effects of future hunger 108 (Rogers, Ferriday, Jebb, & Brunstrom, 2016). Indeed, individuals might even consume a meal Accordingly, we hypothesised that the tendency to engage in "eating in anticipation of future 111 hunger" would be related to poor IF diet adherence. 112 113

Material and methods 114
Participants 115 An online survey was advertised through two sources; i) our laboratory volunteer database, 116 which includes students and staff from the University of Bristol and members of the Bristol 117 (UK) public, and ii) an advertisement on the "5:2 Intermittent Fasting Diet" Facebook page. 118 All participants were aged 18 years or older and were asked to complete the survey in one 119 sitting and without interruption. Participants reported whether they were currently on a diet, 120 and if so which diet. They also indicated whether they had ever followed an IF diet. Using this 121 information, we identified a group of participants who were currently following an IF diet 122 (Current-IF dieters, n = 205) and a group who had previously tried an IF diet but had now 123 stopped (n = 117). A small number of these respondents explicitly identified that they had 124 previously tried an IF diet but stopped because they had been successful and reached their 125 target weight (n = 10), so these participants were not included in the analyses. The majority of 126 participants who did not adhere to the IF diet (Former-IF dieters, n = 107) were included in the 127 analyses. No payment was offered for participation. Instead, respondents were voluntarily 128 entered into a prize draw to win one of two shopping vouchers worth £50 Sterling. Ethical Participants were asked to report their age, sex, height, and weight. Participants' height 142 and weight was used to calculate their BMI (kg/m 2 ). All reported the duration they had been weeks", "3-6 months", "7-12 months", or "more than a year". These values were recorded for 145 all Current-IF dieters as well as any Former-IF dieters who had switched from their IF diet and 146 were currently following another, non-IF diet. Former-IF dieters were also asked to report the 147 duration of their IF diet. This was captured in a free-text response to the question: "Have you 148 ever followed an intermittent fasting diet? If so, please provide details." 149 150 Beliefs about healthy meal and snack patterns 151 To assess beliefs about healthy meal and snack patterns, two questions from the 152 Personal Theories of Hunger and Eating Questionnaire (PTHE) (Assanand, 1996) were 153 included. Participants were asked to respond to the questions "Three meals eaten at three 154 different times during the day is…?" (PTHE #9) and "Several snacks eaten throughout the day 155 is …?" (PTHE #10) with response options "very unhealthy", "somewhat unhealthy", "neutral", 156 "somewhat healthy", or "very healthy." 157 158

Meal and snack patterns on fasting-days versus non-fasting-days 159
To determine meals and snacks that were routinely consumed on a non-fasting-day, 160 participants were asked "Which of the following meals and snacks do you eat in a typical day?" 161 and were instructed to "Tick all that apply" with response options "breakfast", "mid-morning 162 snack", "lunch", "mid-afternoon snack", "dinner", and "late evening snack." To determine 163 their meal and snack patterns on fasting-days, participants were given the following scenario 164 "If you were only permitted to eat 500 calories (women) or 600 calories (men) throughout the 165 day, which of the following meals and snacks would you eat?" and again asked to "Tick all that 166 apply", with the same meal and snack response options. 167 168

Prioritisation of satiation over palatability 169
An additional aim was to understand whether priorities shift when making food 170 selections on fasting versus non-fasting-days; specifically, whether individuals select foods 171 based on satiation (fullness conferred by consuming the food) or palatability (pleasurable taste). 172 Respondents were asked "On a typical day, which of these would you prioritise if you had to 173 select one option?" and "If you were only permitted to eat 500 calories (women) or 600 calories 174 (men) of food throughout the day, which of these would you prioritise if you had to select one 175 option?". In each case, they selected either "Foods that are tasty" or "Foods that are filling". fullness on non-fasting-days to preferring taste on fasting-days (shift to taste), (iii) no shift 179 because fullness was preferred on both non-fasting-days and fasting-days (fullness always 180 preferred), or (iv) no shift because taste was preferred on both non-fasting-days and fasting-181 days (taste always preferred). 182 183 Eating in anticipation of future hunger 184 A further aim was to explore the relationship between dieter status (Current-IF or 185 Former-IF) and the tendency to eat in anticipation of future hunger. To assess these 186 relationships, participants were asked "How often do you eat a meal or snack when you aren't 187 hungry 'just in case' you might feel hungry or be without food later in the day?" with response 188 options "never", "rarely", "sometimes", "often", or "always." 189 190

Concerns about fasting meal patterns 191
Participants were asked to rate their level of concern about the consequences of IF. 192 They were asked to "Imagine you were only permitted to eat a total of 500 calories (women) 193 or 600 calories (men) over a 24-hour period. Which of the following (if any) would be a 194 concern?" Participants then rated their level of concern ("No concern", "slight concern", "some 195 concern", "moderate concern", or "extreme concern") associated with variety of side effects 196 and barriers to fasting. In this study we were interested in responses that relate directly to 197 patterns of eating; "Feelings of 'missing out' on eating at habitual meal and snack times", Both groups were predominantly female; however, there was a higher proportion of 213 female respondents in the Current-IF dieter sample. Current-IF dieters were significantly older 214 and had a significantly higher BMI than Former-IF dieters. Several Former-IF dieters (n = 24) 215 did not provide information regarding the duration that they followed an IF diet. However, 216 overall, Current-IF dieters reported following an IF diet for a longer duration than Former-IF 217 dieters (p < .001). See Table 1  Additionally, Former-IF dieters were more likely to believe that several snacks eaten 224 throughout the day is healthy (PTHE #10) than did Current-IF dieters (see Table 2 Slight concern 23.4a 26.2a Some concern 3.4a 16.8b Moderate concern 5.4a 12.1b Extreme concern 1.0a 11.2b Feelings of "missing out" at meal and snack times No concern 52.7a 27.1b χ 2 (4, N = 312) = 47.82, p < .001 Slight concern 33.7a 24.3a Some concern 8.3a 22.4b

Moderate concern
2.9a 16.8b Extreme concern 2.4a 9.3b Difficulty not eating when cooking for others No concern 49.3a 31.8b χ 2 (4, N = 312) = 17.01, p = .002 Slight concern 19.5a 23.4a Some concern 17.1a 13.1a Moderate concern 8.3a 18.7b Extreme concern 5.9a 13.1b Note. All values are shown as percentages within dieter category or as means (standard deviations). Each superscript letter denotes values that do not differ significantly from one another at the α=.05 level. Table 2. Beliefs about healthy eating patterns, preference for fullness versus taste, concerns with fasting-day meal patterns, and tendency to eat in 227 anticipation of future hunger between Current-IF and Former-IF dieters.

Meal and snack patterns on fasting-and non-fasting-days 229
On non-fasting-days, both dieter groups reported similar meal and snack patterns, with 230 the only apparent difference being in the snacking behaviour of Former-IF dieters. Former-IF 231 dieters were more likely to report eating a mid-morning snack (44.9% vs. 15.6%)(χ 2 (1, N = 232 312) = 31.55, p < .001) and a mid-afternoon snack (49.5% vs. 37.6%)( χ 2 (1, N = 312) = 4.15, 233 p = .042) compared with Current-IF dieters (see Figure 1, Panel A). However, on fasting-days 234 there were significant differences between groups on all eating occasions. On fasting-days 235 Former-IF dieters were more likely to report that they would eat breakfast (32.   Table 2 shows participants' average level of concern for each potential barrier to 267 fasting, with higher values indicating a higher level of concern. Compared to Current-IF 268 dieters, Former-IF dieters reported significantly more concern that restricting their calorie 269 intake to 500 kcal (women) or 600 kcal (men), indicating that it would; i) be disruptive to their 270 daily schedule (p < .001), ii) promote feelings of "missing out" at habitual meal and snack 271 times (p < .001), and iii) produce "difficulty not eating when preparing meals for others" (p = 272 .002). 273 274

Discussion 275
Together, these findings indicate that beliefs about a healthy eating style may affect people's 276 perceived ability to restrict calorie intake in an intermittent fasting diet. Specifically, 277 participants who believed skipping meals to be unhealthy were more likely to indicate that they 278 would eat several meals a day, even on fasting-days. Current-IF dieters were more likely to 279 reserve their eating to dinner or a late evening snack on fasting days, suggesting that this approach aids diet adherence. It may be the case that breaking the fast early in the day (i.e., 281 eating breakfast) makes it more difficult for an individual to refrain from eating again, in other 282 words, promotes disinhibition . Delaying eating occasions until 283 the end of the day has the advantage that an individual can avoid contact with food in the 284 morning and early afternoon. This might serve as a simple strategy to reduce the need for 285 calorie-counting as well as reducing opportunities for reminders about eating and encounters 286 with food-related stimuli throughout the day (one just gets on with other responsibilities). Of 287 course, when avoiding meals, the pleasure of eating is missed , 288 but physical and cognitive performance is likely to be unaffected 289 Rogers et al., 2016). Additionally, people may find fullness comforting because they fear an 290 empty stomach, which is equated with hunger. However, fullness itself does not appear to be 291 intrinsically rewarding (Sclafani & Ackroff, 2004), arguably because fullness prevents further 292 energy intake . Regardless of whether individuals prioritise 293 pleasure of eating or fullness, they risk feeling deprived on fasting days, due to diminished 294 pleasure from eating, or lack of fullness, or both. 295 Additionally, individuals who are concerned about hunger may be more likely to eat in 296 order to prevent future hunger, even if they are not currently feeling a desire to eat. Our findings 297 suggest that asking prospective dieters about the biological origin of their own hunger might 298 help to inform appropriate guidancethose that express fewer concerns about hunger might 299 be particularly encouraged to adopt and IF diet Rogers et al., 300 2016). One possibility is that IF dieters might be encouraged to select low-energy-dense foods 301 on fasting-days, because these can be consumed in larger quantities and promote feelings of 302 fullness. Indeed, when only small portions are available, individuals tend to select the foods 303 that promote maximum fullness  adherence. An additional limitation is that Former-IF dieters were not asked to report their 317 reasons for quitting their IF diet, so the observed relationships between diet adherence and meal 318 patterns may be due to other side effects of fasting that were not assessed. A further concern is 319 that the two samples were not equally matched demographically. The Current-IF dieter group 320 had a higher proportion of females (93.7% versus 83.2%) and a higher average BMI (26.6 321 versus 24.3 kg/m 2 ) than the Former-IF dieter group. There was also a difference in age between 322 the Current-IF and Former-IF dieters (48.5 versus 33.6 years, respectively). Age (experience) 323 may affect beliefs about healthy eating behaviour or overall willingness to adhere to an IF diet, 324 and might help to explain some of the relationships described in this study. Future studies 325 should consider the extent to which age influences beliefs about healthy eating and willingness 326 to adhere to IF diets. 327 The importance of eating breakfast is often promoted by evidence showing an 328 association between eating later in the day (or late at night) and increased BMI (Kinsey &  329 Ormsbee, 2015). However, this evidence, similar to the previously described relationship 330 between breakfast consumption and performance in school, is not causal and the impact of 331 eating later in the day on BMI is unclear. For some, a potential benefit of consuming consistent 332 meals throughout the day is that it might reduce unplanned snacking behaviour (Berg & 333 Forslund, 2015). Individuals who snack regularly might be less likely to attempt an IF diet or 334 may be less adherent to the diet because routinely snacking prevents over-eating late at night. 335 We note that adherent (Current-IF) dieters were less likely to report snacking behaviour on 336 both fasting and non-fasting-days. 337 Many forms of IF have become popular in recent years; however, this way of eating is 338 not revolutionary. Intermittent fasting has formed a significant part of many religious practices 339 for centuries (Dietler, 2011) and meal size, frequency and timing are socio-culturally 340 determined (De Castro, 1997). Evolutionary biologists suggest that prior to the development 341 of agriculture, humans were primarily hunter-gatherers and would have regularly experienced 342 periods of food scarcity followed by periods of food abundance (Chakravarthy & Booth, 2004). 343 However, the current eating environment is characterised by continuous food availability, and 344 the overarching recommendation is that adults and children should eat regularly throughout the 345 day to maintain a "healthy" diet. As obesity continues to be a leading contributor to many major