BMJ

Confidential: For Review Only (Over)Eating Out: An observational study of the energy content of main meals served in major UK restaurant chains

Journal: BMJ

Manuscript ID BMJ.2018.046276.R1

Article Type: Christmas research

BMJ Journal: BMJ

Date Submitted by the 18-Oct-2018 Author:

Complete List of Authors: Robinson, Eric; University of , Jones, Andrew; University of Liverpool Whitelock, Victoria; University of Liverpool Mead, Bethan; University of Liverpool Haynes, Ashleigh; University of Liverpool

Keywords: eating out, obesity, environment, restaurant food, kilocalories

https://mc.manuscriptcentral.com/bmj Page 1 of 29 BMJ

1 1 2 3 4 1 (Over)Eating Out: An observational study of the energy content 5 6 7 2 of main meals served in major UK restaurant chains 8 9 10 3 11 Confidential: For Review Only 12 13 4 14 15 5 16 17 6 18 19 7 Eric Robinson1, Andrew Jones1, Victoria Whitelock1, Bethan R. Mead1 & Ashleigh Haynes1 20 21 22 8 23 24 9 1 Institute of Psychology, Health & Society, University of Liverpool, L69 7ZA, UK 25 26 10 27 28 29 11 Correspondence: Eric Robinson, Institute of Psychology, Health & Society, University of 30 31 12 Liverpool, L69 7ZA, UK. Email: [email protected] 32 33 13 34 35 36 14 37 38 15 39 40 16 41 42 17 43 44 45 18 46 47 19 48 49 20 50 51 52 21 53 54 22 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj BMJ Page 2 of 29

2 1 2 3 23 Total Word Count: 3, 935 4 5 6 24 Key Words: food environment; eating out; restaurant food; kilocalories; obesity 7 8 25 What is already known on this topic? 9 10 26 Eating out of the home is common in the UK. 11 Confidential: For Review Only 12 27 The poor nutritional quality of ‘’ has been well documented. 13 14 15 28 The energy content of traditional ‘sit down’ restaurants has received less attention. 16 17 29 18 19 30 What this study adds? 20 21 22 31 The average kcal content of main meals served in both fast food and sit down restaurants in the 23 24 32 UK is larger than public health recommendations. 25 26 33 The proportion of main meals in UK restaurant chains that meet public health recommendations 27 28 34 for kcals is smaller than the proportion of meals that have an excessive number of kcals. 29 30 31 35 Compared to fast food restaurants, there are more significantly more kcals on average in sit down 32 33 36 restaurant meals in the UK. 34 35 37 36 37 38 38 39 40 39 41 42 40 43 44 45 41 46 47 42 48 49 43 50 51 44 52 53 54 45 55 56 46 57 58 59 60 https://mc.manuscriptcentral.com/bmj Page 3 of 29 BMJ

3 1 2 3 47 4 5 6 48 7 8 49 Abstract 9 10 50 Objectives: Eating out of the home is common, but the kilocalorie (kcal) content of main meals 11 Confidential: For Review Only 12 served by major UK restaurant chains has not been examined. The objectives of this study were 13 51 14 15 52 to examine the kcal content of main meals served in major UK restaurant chains and compare the 16 17 53 kcal content of fast food and ‘sit down’ restaurant chains. 18 19 54 Design: Observational study. 20 21 22 55 Setting: Menu and nutritional information provided by major UK restaurant chains. 23 24 56 Method: During April to July 2018, websites were accessed and visits were made to restaurant 25 26 57 chains with 50 or more outlets in the UK. Menu items that constituted lunch or evening main 27 28 29 58 meal dishes were identified and the kcal content of these meals was extracted. Beverage kcals 30 31 59 were not included. 32 33 60 Main outcome measures: The mean kcal content of meals, the proportion of meals meeting 34 35 public health recommendations for kcal consumption (600kcals or less) and the proportion of 36 61 37 38 62 meals with an excessive kcal content (1000kcals or more). 39 40 63 Results: Main meals from twenty-seven restaurant chains (21 sit down, 6 fast food) were 41 42 64 sampled. The mean kcal content of all eligible restaurant meals (13,396 in total) was 977kcals 43 44 45 65 (95% CIs: 973 to 983), the percentage of all meals that met public health recommendations for 46 47 66 kcal content was low (10%) and smaller than the percentage of meals with an excessive kcal 48 49 67 content (42%). Compared to fast food restaurants, sit down restaurants offered significantly more 50 51 52 68 excessive calorific main meals, fewer main meals meeting public health recommendations and 53 54 69 on average 268 kcals more (95% Cis: 102.6 to 433.4) in main meals. 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj BMJ Page 4 of 29

4 1 2 3 70 Conclusions: The kcal content of a large number of main meals in major UK restaurant chains is 4 5 6 71 excessive and only a minority meet public health recommendations. Although the poor 7 8 72 nutritional quality of fast food meals has been well documented, the kcal content of sit down 9 10 73 restaurant meals in the UK tends to be greater and is a cause for concern. 11 Confidential: For Review Only 12 Registration: The study protocol and analysis strategy was pre-registered on the Open Science 13 74 14 15 75 Framework (https://osf.io/w5h8q/). 16 17 76 18 19 77 Introduction 20 21 22 78 The prevalence of overweight and obesity has increased markedly across most of the 23 24 79 developed world1. Increases in energy intake caused by major changes to the food environment2 25 26 80 have been identified as a key factor explaining population level weight gain3 4. In the UK, meals 27 28 29 81 are consumed out of the home regularly; data collected from 2008-2012 showed that a quarter of 30 31 82 UK adults ate out once or more per week5. However, a more recent report from the UK Food 32 33 83 Standards Agency in 2016 indicates that eating out of the home may becoming more common, 34 35 with 39% of UK adults reporting eating out at least once per week6. A number of studies suggest 36 84 37 38 85 that people who out of the home more frequently are at increased risk of weight gain and 39 40 86 obesity7. ‘Fast food’ restaurants in particular have been highlighted as providing meals that are 41 42 87 low in nutritional quality8 9. There is also some evidence that a higher geographical density of 43 44 10 11 45 88 fast food restaurants is associated with increased risk of obesity . Because of this there have 46 47 89 been public health calls in the UK to limit where outlets can operate12 13. 48 49 90 However, more traditional ‘sit down’ restaurants also contribute substantially to the out of home 50 51 14 52 91 dining market in the UK . 53 54 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj Page 5 of 29 BMJ

5 1 2 3 92 Recent public health recommendations made by Public Health suggest that 4 5 6 93 adults should aim to consume 600 kilocalories (kcals) or less for their main lunch and dinner 7 8 94 meals to avoid excess daily energy intake and maintain a healthy body weight15. This is in part 9 10 95 motivated by Public Health England’s estimate that the average adult in the UK is consuming 11 Confidential: For Review Only 12 195 excess kcals per day15. Because the amount of energy a person consumes during a meal is 13 96 14 15 97 strongly influenced by the energy density and portion size of food served 16-19, meals provided to 16 17 98 consumers that are high in kcals promote excess energy intake and are problematic for public 18 19 99 health. However, public health action on improving the nutritional quality of food prepared 20 21 22 100 outside of the home has to date focused largely on encouraging the food industry to make 23 24 101 reductions to the kcal content of supermarket food20 and has not focused on the restaurant sector. 25 26 102 To date, there has been no examination of the number of kcals in main meals served by major 27 28 29 103 UK restaurant chains and it is therefore unclear whether consumers can adhere to public health 30 31 104 recommendations for meal kcal consumption when eating in these establishments. Moreover, 32 33 105 legislation has been passed which will result in kcal labelling of all food products sold by major 34 35 chain restaurants becoming mandatory in the US21. Similar legislation is currently being 36 106 37 38 107 considered by UK government, but mandatory kcal labelling will come at a financial cost to the 39 40 108 food industry which may cause challenges to legislation, as was the case in the US22. To address 41 42 109 such challenges it will be important to understand the extent to which major UK restaurant 43 44 45 110 chains are contributing to overconsumption by examining the typical kcal content of main meals 46 47 111 and the availability of main meals meeting public health recommendations for kcal consumption. 48 49 112 In the present study we examined the kcal content of main meals (lunch and dinner) sold 50 51 52 113 by major restaurant chains in the UK. We also compared the kcal content of main meals in fast 53 54 114 food vs. traditional ‘sit down’ restaurant chains. We reasoned that this comparison would be 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj BMJ Page 6 of 29

6 1 2 3 115 important for a number of reasons. First, although the kcal content of sit down restaurants has 4 5 6 116 received little attention in the UK, a small number of studies of North American dining suggests 7 8 117 that kcal content for these restaurant types can be excessive23. Second, we speculate that over 9 10 118 time the negative publicity concerning the poor nutritional quality of fast food may have caused 11 Confidential: For Review Only 12 this sector to provide lower kcal meal options on their menus and/or reformulate existing 13 119 14 15 120 meals24, whereas the sit down restaurant sector has presumably not faced this pressure. Thus, we 16 17 121 hypothesised that, somewhat counterintuitively, the kcal content of main meals in sit down 18 19 122 restaurant chains would be more excessive than that of fast food restaurant chains in the UK. 20 21 22 123 23 24 124 Methods 25 26 125 The study protocol and analysis strategy was pre-registered on the Open Science Framework 27 28 29 126 (https://osf.io/w5h8q/). 30 31 127 32 33 128 Restaurant sampling. As our aim was to examine major UK restaurant chains we included all 34 35 chains with 501 or more outlets in the UK. To identify major restaurant chains we consulted 36 129 37 38 130 market reports listing restaurants with the largest number of UK outlets and market research 39 40 131 ranking UK restaurant chains by annual turnover, popularity, number of users, and numbers of 41 42 132 outlets 14 25-27 . To confirm eligibility, during March-April 2018 one researcher accessed the UK 43 44 45 133 website of each restaurant chain to identify those with ≥50 outlets and this was independently 46 47 134 verified by another researcher. If the number of UK outlets was not provided on a restaurant 48 49 50 51 52 1 We were aware of no formal classification of what determines a ‘major’ restaurant chain. We chose 50 outlets or 53 more as this criteria resulted in us being able to include all restaurant chains that were consistently high in annual 54 turnover and popularity according to market reports we accessed. Our scoping research also indicated that chains 55 with fewer than 50 outlets were less likely to provide online nutrition information for menu items. 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj Page 7 of 29 BMJ

7 1 2 3 135 website this information was requested by email. For a full list of all restaurant chains identified 4 5 6 136 with ≥50 outlets and the number of outlets per chain, see online supplementary materials table 1. 7 8 137 9 10 138 Characterising restaurant types. To categorise restaurant chains as ‘fast food’ or ‘sit down’, 11 Confidential: For Review Only 12 based on previous research, e.g.28 we used the following definition of fast food restaurants: 13 139 14 15 140 ‘Restaurants that primarily provide consumers with largely pre-prepared ‘quick’ meals with 16 17 141 little or no table service, with in-store seating and in which take-away orders are likely to 18 19 142 account for a significant proportion of orders’. We did not include coffee shop chains or chains 20 21 22 143 that only provided take-away food (i.e. no physical restaurant). Two researchers independently 23 24 144 coded each eligible restaurant as fast food or sit down and any disagreements were resolved 25 26 145 through discussion. 27 28 29 146 Data sources. During April-July 2018 we accessed the UK websites of all eligible restaurant 30 31 147 chains and identified current menus and nutritional information. Restaurant chains that did not 32 33 148 provide nutritional information on their UK website were contacted by email and this 34 35 information was requested. 36 149 37 38 150 39 40 151 Public/patient involvement. There was no public or patient involvement in the design of the 41 42 152 study. 43 44 45 153 46 47 154 Identification of main meal menu options. We aimed to examine the kcal content of all ‘main 48 49 155 meal’ menu options. We defined a main meal as being a menu option that would normally be the 50 51 52 156 primary dish in a lunch or dinner meal and typically be found in the ‘main course’ part of a 53 54 157 restaurant menu. Examples of main meal items using this definition include burger and chips, 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj BMJ Page 8 of 29

8 1 2 3 158 chicken Caesar salad, spaghetti Bolognese, jacket potato with a filling. We did not include 4 5 6 159 individually sold food items (e.g. sides) or sharing menu options (e.g. tapas) as it was unclear 7 8 160 what combination of individual items or sharing menu options would constitute a main meal. 9 10 161 Starters and dessert menu options were not included as they are not typically consumed as a main 11 Confidential: For Review Only 12 meal. We did not include breakfast menu options as during a scoping exercise we found that a 13 162 14 15 163 large number of eligible restaurants did not offer breakfast menu options. We included main 16 17 164 meals that could be purchased by any member of the general public and menu options for 18 19 165 specialist consumer groups were ineligible (e.g. pensioners menu, child menu). In order to 20 21 22 166 minimize effects of season, we included main meal options that appeared to be available all year 23 24 167 round only. In instances in which a main meal menu option could be customized at the explicit 25 26 168 request of the customer (e.g. swap default side dish for a different side dish), we selected the 27 28 29 169 default composition of the meal. In instances in which a meal menu option required a customer 30 31 170 to make an explicit choice (e.g. choice of salad or fries) we extracted all possible configurations 32 33 171 of the meal and recorded each variant as an individual meal. During May-June 2018 two 34 35 researchers independently accessed each restaurant menu and identified eligible meal menu 36 172 37 38 173 options. For more detailed information on coding instructions used by researchers see 39 40 174 supplementary materials. Discrepancies were resolved by discussion and if required, a third 41 42 175 researcher. If a restaurant only provided individual food item information on their online menu 43 44 45 176 with no information on which combinations of items constituted a meal option in store, two 46 47 177 researchers visited a local outlet (Liverpool city centre) of that restaurant and recorded eligible 48 49 178 menu meal options. 50 51 52 179 53 54 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj Page 9 of 29 BMJ

9 1 2 3 180 Extraction of meal kcal content. Although nutritional information tends not to be displayed on 4 5 6 181 UK restaurant menus in store, some restaurant chains provide this information on their websites. 7 8 182 A researcher accessed online nutritional information for each restaurant and extracted the 9 10 183 number of kcals for each eligible meal. A second researcher checked extraction for accuracy. If 11 Confidential: For Review Only 12 kcal information was missing from a restaurant’s nutritional information we attempted to locate 13 184 14 15 185 it from elsewhere on the website and if we were not able to, we used kcal information from a 16 17 186 close to identical meal option (e.g. sandwich with brown vs. wholegrain bread), if available. 18 19 187 Because drinks were not routinely provided with meals in restaurants, drinks were not included 20 21 22 188 when extracting kcal information for any meal options that included a drink. 23 24 189 25 26 190 Inter-coder consistency. We examined percentage agreement (i.e. proportion of restaurants 27 28 29 191 identified by both researchers as fast food or sit down) as an indicator of inter-coder consistency 30 31 192 for classification of restaurant type. We had planned to adopt a similar approach for inter-coder 32 33 193 consistency of identification of eligible main meal options, but due to there being a very large 34 35 number of menu items that were clearly not eligible (e.g. starters, sides, desserts, drinks, 36 194 37 38 195 children’s menu) coders did not record a classification (eligible vs. not eligible) for every menu 39 40 196 item and recorded eligible menu options only. We therefore approximated inter-coder 41 42 197 consistency by examining the number of menu options identified that were deemed eligible by 43 44 45 198 both coders independently vs. the number of menu options identified by one of the coders, but 46 47 199 not the other. 48 49 200 50 51 52 201 Planned analyses. To estimate the mean number of kcals in meal options we used multi-level 53 54 202 modelling, with individual meal options nested within restaurants and restaurants categorised as 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj BMJ Page 10 of 29

10 1 2 3 203 being fast food or sit down. We examined model fit by comparing the loglikelihood ratio statistic 4 5 6 204 (loglikelihood of the multilevel model - loglikelihood of a single-level model) to a Chi-squared 7 8 205 distribution with 1 degree of freedom. We used bootstrapping (500 samples) as this improves the 9 10 206 accuracy of parameter values and reduces bias in parameter estimates. Because any difference in 11 Confidential: For Review Only 12 meal kcal content between fast food vs. sit down restaurants may be in part explained by the two 13 207 14 15 208 types of restaurant serving different types of meals, we also planned to repeat our analyses for 16 17 209 any individual meal types that were provided by the majority of both fast food and sit down 18 19 210 restaurant chains. We identified two meal types (burger and fries/chips meals, salad meals) that 20 21 22 211 met this requirement and used the same multilevel modelling approach described above. 23 24 212 We also examined the proportion of meals that met UK public health guidelines for 25 26 213 recommended kcal consumption (≤600kcals) for a main meal15 and the proportion of all meals 27 28 29 214 that were excessively high in kcals. We defined ‘excessive’ as being meals that were ≥1000kcals, 30 31 215 as this single meal constitutes 50% and 40% of the recommended total number of daily kcals for 32 33 216 women (2000kcals) and men (2500kcals) respectively and also constitutes the majority of the 34 35 number of kcals that a man or women attempting weight loss is recommended to eat in a day by 36 217 37 38 218 the UK National Health Service29. To examine whether fast food vs. sit down restaurants differed 39 40 219 in the proportion of main meals that were ≤600kcals and ≥1000kcals we conducted two 41 42 220 multilevel logistic regressions. We used 1st order marginal quasi-likelihood models. To examine 43 44 45 221 whether two level models (meals within restaurants) were more appropriate than single level 46 47 222 models we examined whether residual variance at the restaurant level was significantly different 48 49 223 from 0, by computing the Wald statistic (variance/standard error)2 and comparing this with a 50 51 52 224 Chi-Square distribution with 1 degree of freedom. We planned in the pre-registered protocol to 53 54 225 examine the above using only Chi-Squares and for completeness we report these results in the 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj Page 11 of 29 BMJ

11 1 2 3 226 online supplementary material. In all analyses α was set at .05. Multi-level analyses were 4 5 6 227 conducted in MLWiN v.3 (2017). All other analyses were conducted in SPSS 22. 7 8 228 9 10 229 Results 11 Confidential: For Review Only 12 Restaurants. We identified 52 restaurant chains with 50 or more outlets in the UK. Of these, we 13 230 14 15 231 were able to access menu and nutritional information for 30 restaurants and requested this from 16 17 232 the remaining chains, 1 of which provided information. Four of these 31 restaurants did not have 18 19 233 items on their menus that constituted meals and instead only sold individual food items (e.g. 20 21 22 234 individual pieces of chicken, sushi) so were not eligible for inclusion. Inter-coder consistency for 23 24 235 classification of eligible restaurants as fast food vs. sit down was high (96.3%, 26/27 restaurants) 25 26 236 and the one discrepancy () was resolved by discussion between the two coders after 27 28 29 237 accessing the chain’s website. The final number of eligible restaurant chains was 27 (n=6 for fast 30 31 238 food, n=21 for sit down restaurants). See Table 1 for restaurants included. 32 33 239 34 35 Meals. Of the meals identified by the first coder (13,422 meals) 98.9% were also identified by 36 240 37 38 241 the second coder and of the meals identified by the second coder (13,444 meals) 98.7% were 39 40 242 identified by the first coder, indicating consistency between the two coders when identifying 41 42 243 eligible meals. After discrepancies between the two coders (i.e. meals identified by one coder 43 44 45 244 only) were resolved, the final number of eligible meals was 13,507. The large number of eligible 46 47 245 meals was mainly attributable to the relatively large number of meals contributed by two 48 49 246 restaurant chains (, N=2,436 and Nando’s, N = 9,298) due to these restaurants offering 50 51 52 247 meals with multiple variants (e.g. chicken with a choice of any two sides, sandwich meal with a 53 54 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj BMJ Page 12 of 29

12 1 2 3 248 choice of multiple sides). With these restaurant chains removed, inter-coder consistency as 4 5 6 249 described above remained high (90.4% and 89.6%). 7 8 250 9 10 251 Kcal information. Of the 13,507 eligible meals identified, we were able to extract kcal 11 Confidential: For Review Only 12 information for 13,396 meals (99.2%) and the remaining meals were treated as missing data and 13 252 14 15 253 not included in analyses. See Table 1 for number of eligible meals per restaurant and raw kcal 16 17 254 data per meal for each restaurant. 18 19 255 20 21 22 256 Mean kcal content of all meals. Across all meals the average number of kcals per meal was 977.3 23 24 257 (95% CIs: 973 to 983, Standard error (SE) = 2.48). A two level model structure (meals within 25 26 258 restaurants) was a better fit of the data than a single level structure, χ2(1) = 2918, p < .001, 27 28 29 259 indicating that multi-level modelling was appropriate. The variance partition coefficient; the total 30 31 260 residual variance which is attributable to restaurants rather than individual meals was 36.5%. 32 33 261 Type of restaurant (sit down vs. fast food) was a significant predictor, β = 268.0, SE = 84.4 (95% 34 35 CIs 102.6 to 433.4), p < .001, explaining 35.7 % of variance at the restaurant level. These results 36 262 37 38 263 indicate that meals from sit down restaurants had 268 kcals more energy than meals from fast 39 40 264 food restaurants, on average. 41 42 265 43 44 45 266 Mean kcal content of specific meal types. Across burger and fries/chips meals (N = 1,904; 1,010 46 47 267 sit down, 894 fast food) the average number of kcals was 1170.9 (SE = 7.0). The weighted 48 49 268 multilevel model demonstrated a two level structure was a better fit than a single level structure, 50 51 52 269 χ2(1) = 411, p < .001, and the variance partition coefficient was 68.37%. Type of restaurant (sit 53 54 270 down vs. fast food) was a significant predictor, β = 414.3, SE = 141.1 (95% CIs 137.7 to 690.8), 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj Page 13 of 29 BMJ

13 1 2 3 271 p < .001, explaining 29.0% of variance at the restaurant level and indicating that burger meals in 4 5 6 272 sit down restaurants had 414kcals more energy than in fast food restaurants, on average. Across 7 8 273 salad meals (N = 304; 92 sit down, 212 fast food) the average number of kcals was 446.2 (SE = 9 10 274 10.4). A two level structure was a better fit than a single level structure, χ2(1) = 885, p < .001, 11 Confidential: For Review Only 12 and the variance partition coefficient was 69.4%. Type of restaurant (sit down vs. fast food) 13 275 14 15 276 explained 8.2% of variance at the restaurant level and sit down restaurant salad meals had on 16 17 277 average 146kcals more than fast food meals, although restaurant type was not a statistically 18 19 278 significant predictor in the model, β = 145.9, SE = 106.7 (95% CIs -49.3 to 451.0), p = .082. See 20 21 22 279 Table 2 for average kcals for burger and salad meals by restaurant. 23 24 280 25 26 281 Because there was an unexpectedly large amount of variability in the number of meals that 27 28 29 282 individual restaurants contributed to the analyses for ‘all meals’ and ‘specific meal types’ we 30 31 283 also conducted weighted multi-level analyses (see online supplementary materials). 32 33 284 34 35 Meals ≤600kcals and ≥1000kcals.Of the 13,396 possible meals identified, 1,226 (9.5%) met UK 36 285 37 38 286 public health recommendations of ≤600kcals. The total number of meals which contained 39 40 287 ≥1000kcals was 6,251 (46.7%). See Table 1. Logistic models examining proportion of meals < 41 42 288 600kcals demonstrated significant variance at the restaurant level (Wald Test Statistic = 9.0, p = 43 44 45 289 .002) suggesting a two level model was appropriate. The odds ratio for restaurant type was OR = 46 47 290 3.2 (95% CIs 1.4 to 7.4, p = .003), suggesting that fast food restaurants were approximately three 48 49 291 times more likely to offer meals < 600kcals than sit down restaurants. For proportion of meals 50 51 52 292 >1000kcals there was significant variance at the restaurant level (Wald Test Statistic = 6.0, p = 53 54 293 .014), suggesting a two level model was appropriate. The odds ratio for restaurant type was OR = 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj BMJ Page 14 of 29

14 1 2 3 294 5.1 (95% CIs 1.7 to 15.0, p = .002), suggesting sit down restaurants were approximately five 4 5 6 295 times more likely to offer meals ≥1000kcals than fast food restaurants 7 8 296 9 10 297 Discussion 11 Confidential: For Review Only 12 Statement of principal findings 13 298 14 15 299 Across the major UK chain restaurant meals included, the mean number of kcals in main meals 16 17 300 was 977kcals, a sizeable proportion (42%) were ‘excessive’ in kcal content (≥1000kcals) and 18 19 301 only a small minority (10%) were in line with public health recommendations for main meal kcal 20 21 22 302 consumption (≤600kcals). On average, the number of kcal in main meals served by sit down 23 24 303 restaurants was 268kcals larger than main meals served by fast food restaurants. Sit down 25 26 304 restaurants also tended to serve more highly calorific main meals and provide fewer main meals 27 28 29 305 meeting public health recommendations for kcal consumption. 30 31 306 32 33 307 Strengths and weakness of the study 34 35 We were able to sample a large number of restaurant chains and main meals. However, our 36 308 37 38 309 analyses were limited to restaurants that provided nutritional information and sold products 39 40 310 consistent with our inclusion criteria (27/52 identified chains.) This is a weakness of the study, as 41 42 311 it is possible that main meals in restaurants not providing online nutritional information differ in 43 44 45 312 kcal content to those that do. Reliance on self-reported kcal information from restaurant chains is 46 47 313 a weakness of the study and objectively calculated kcal content (using laboratory methods) 48 49 314 would have been preferable, but was not feasible. However, it is important to note that previous 50 51 30 52 315 research suggests that commercially provided nutritional information tends to be accurate , but 53 54 316 may underestimate kcal content of some products31. Our findings are therefore more likely to 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj Page 15 of 29 BMJ

15 1 2 3 317 underestimate than overestimate kcal content of main meals and this means that the kcal content 4 5 6 318 of UK restaurant food may be more problematic than our data suggest. Our focus here was on the 7 8 319 energy content of main meals. Although energy intake is of most relevance to population-level 9 10 320 body weight and obesity, other aspects of diet (e.g. salt, saturated fat) also shape health and 11 Confidential: For Review Only 12 disease. For example, the amount of salt in most UK supermarket ready meal products does not 13 321 14 15 322 meet nutritional guidelines32 and the salt content of UK restaurant main meals may be similarly 16 17 323 high. In the present study we did not include larger main meal items that are typically shared by 18 19 324 consumers due to uncertainty over what would constitute one portion. In addition, we examined 20 21 22 325 the number of kcals served and this does not permit us to make conclusions about consumption. 23 24 326 Although consumers will not always finish all of a meal served, ‘plate clearing’ is a fairly 25 26 327 common behaviour 33 34. Because some customers will order a main meal as well as a drink, 27 28 29 328 starter and/or dessert, we assume that on average the number of kcals consumed in both sit down 30 31 329 and fast food restaurants will be larger still. 32 33 330 34 35 Strengths and weaknesses in relation to other studies 36 331 37 38 332 Although research has examined the nutritional quality of restaurant food, this has tended to be 39 40 333 conducted in North America23 35. This is the first study we are aware of to characterise the kcal 41 42 334 content of main meals served in UK chain restaurants. A limitation of our study was that we did 43 44 45 335 not examine smaller chains or independent restaurants, although both chain and non-chain US 46 47 336 restaurants serve excess amounts of energy35. Previous studies examining nutritional quality of 48 49 337 other food products (e.g. supermarket food) have made use of WHO guidelines to examine the 50 51 52 338 proportion of products meeting specific nutrient recommendations (e.g. energy from saturated 53 54 339 fat)32. There are no international guidelines concerning kcal consumption per meal. We therefore 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj BMJ Page 16 of 29

16 1 2 3 340 examined the proportion of main meals meeting Public Health England’s recommendation of 4 5 6 341 600 kcals or less per meal (lunch and dinner). We also quantified the proportion of main meals 7 8 342 that not only failed to meet this recommendation but could be considered as ‘excessive’ in kcals. 9 10 343 Given there is no consensus on what is an excessive number of kcals we defined this as main 11 Confidential: For Review Only 12 meals that were 1000 kcals or more, as in a single course this would constitute 50% and 40% of 13 344 14 15 345 the recommended total number of daily kcals for women and men respectively, or viewed in 16 17 346 another way; the majority of kcals that a man or women attempting weight loss is recommended 18 19 347 to eat in a day. Although it is important to note that the amount of energy that any person 20 21 22 348 requires to maintain a healthy body weight varies, few people are likely to require more than 23 24 349 1000 kcals from a single main meal in order to maintain a healthy body weight. By using 25 26 350 1000kcals or more as a threshold for an excessive number of main meal kcals in the present 27 28 29 351 study we are not suggesting this should become a default threshold used by other researchers, but 30 31 352 we believe it is useful for descriptive purposes here given that there were more main meals that 32 33 353 were ‘excessive’ in kcal content than adhered to the public health recommendation of 600kcals. 34 35 36 354 37 38 355 Meaning of the study 39 40 356 There was a sizeable proportion of fast food and sit down main meals that were excessive in 41 42 357 kcals and we note that there tended to be little or no information provided on menus that would 43 44 45 358 allow consumers to identify menu options that were high in calories vs. those that were lower. 46 47 359 Consumers tend to underestimate the number of kcals in large meals36 37 and this in combination 48 49 360 with the present findings make recent calls to mandate kcal labelling of restaurant food in the UK 50 51 52 361 appropriate. Although kcal labelling is not common in UK restaurants, the best available 53 54 362 evidence suggests that kcal labelling is only likely to have a modest impact on consumer 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj Page 17 of 29 BMJ

17 1 2 3 363 behaviour38, so other public health measures to address energy intake out of the home will be 4 5 39 40 6 364 required. Because portion sizes of many food products have increased over time and 7 8 365 reductions made to the portion size and energy density of food products are unlikely to be 9 10 366 compensated for by consumers41 42, policy levers that result in the food industry reducing the 11 Confidential: For Review Only 12 number of kcals being sold to consumers are needed. This proposition is in line with the 13 367 14 15 368 observation that changes to the food environment have played a key role in the emergence of the 16 17 369 obesity problem and measures are now required to ‘renormalize’ the food environment (e.g. 18 19 370 downsizing food product portions)43. 20 21 22 371 23 24 372 Unanswered questions and future research 25 26 373 The reason why sit down restaurant main meals tended to be higher in kcals than fast food 27 28 29 374 restaurant meals in the present study is unclear, but likely to be caused by multiple factors, 30 31 375 including the type of food sold. However, when we isolated our analyses to meal types that were 32 33 376 routinely sold by both fast food and sit down restaurants we found that sit down restaurant main 34 35 meals still tended to be markedly higher in kcals. Although these analyses were reduced in 36 377 37 38 378 sample size, there was a statistically significant larger number of kcals in burger and chips main 39 40 379 meals served in sit down than fast food restaurants (414kcal difference, p < .001) and more kcals 41 42 380 in salad meals (146kcal difference, p = .08), but the latter was not statistically significantly. 43 44 45 381 Thus, decisions regarding portion size, energy density of ingredients and cooking methods are 46 47 382 also likely to explain differences in meal kcal content between sit down and fast food restaurants. 48 49 383 A further explanation is that because the fast food sector has received negative press because of 50 51 44 52 384 poor nutritional quality of products , this may have caused restaurant chains in this sector to 53 54 385 offer more lower kcal meal options and/or reformulate existing products to reduce kcal content24, 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj BMJ Page 18 of 29

18 1 2 3 386 whereas similar pressures do not appear to have been experienced by the sit down restaurant 4 5 6 387 sector. There was also marked variability between individual restaurants for kcal content of main 7 8 388 meals and identifying why will be informative. In addition to the type of cuisine sold, it is 9 10 389 possible that individual restaurant market positioning and price range are associated with kcal 11 Confidential: For Review Only 12 content. The present study may be of use to future research efforts that examine whether major 13 390 14 15 391 UK restaurant chains respond to public health calls to reduce the number of kcals in their 16 17 392 products. Likewise, if legislation is passed, the results of the present study can be used to assess 18 19 393 whether the introduction of kcal labelling results in restaurant chains reformulating the 20 21 45 22 394 nutritional content of meals, as appears to have been the case in the US . In this vein it will be 23 24 395 important to characterise the nutritional quality of other parts of the UK food environment, as the 25 26 396 present study did not examine nutritional quality of other market sectors (e.g. coffee shops, 27 28 29 397 online food ordering). Online services that allow consumers to have restaurant food delivered to 30 31 398 their home are a recent development in the UK and will likely be increasing the number of meals 32 33 399 consumed that are prepared out of the home. 34 35 36 400 37 38 401 Conclusions 39 40 402 The kcal content of a large number of main meals in major UK restaurant chains is excessive and 41 42 403 only a minority meet public health recommendations. Although the poor nutritional quality of 43 44 45 404 fast food restaurant meals has been well documented, the kcal content of sit down restaurant 46 47 405 meals in the UK tends to be greater and is a cause for concern. 48 49 406 50 51 52 407 Data Sharing 53 54 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj Page 19 of 29 BMJ

19 1 2 3 408 The final data sets containing restaurant meal descriptions and number of kcals for each 4 5 6 409 restaurant used in analyses are available online at https://osf.io/cd597/ 7 8 410 9 10 411 Funding 11 Confidential: For Review Only 12 The MRC (MR/N00218/1) part fund ER’s salary. The views expressed in this publication are 13 412 14 15 413 those of the authors and not necessarily those of the MRC. 16 17 414 18 19 415 Competing interests 20 21 22 416 All authors have completed the ICMJE uniform disclosure form at 23 24 417 www.icmje.org/coi_disclosure.pdf. ER has been a named investigator on research projects 25 26 418 funded by the American Beverage Association, but does not consider this funding a conflict of 27 28 29 419 interest. 30 31 420 32 33 421 Acknowledgements and Contribution 34 35 All authors designed the study. All authors contributed to data collection. AJ and ER were 36 422 37 38 423 responsible for data analysis. ER was responsible for initial drafting of the paper and all authors 39 40 424 approved the manuscript and had full access to the data. We would like to thank Lizhi Zhang for 41 42 425 research assistance. 43 44 45 426 46 47 427 Ethical Approval 48 49 428 As the study involved no human or animal participants and made use of publicly available 50 51 52 429 information ethical approval was not required 53 54 430 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj BMJ Page 20 of 29

20 1 2 3 431 Transparency 4 5 6 432 ER acts as the guarantor for this work and confirms that the manuscript is an accurate, 7 8 433 transparent and honest account of the study, that no important aspects of the study have been 9 10 434 omitted and that any discrepancies from the study as planned have been explained. 11 Confidential: For Review Only 12 13 435 14 15 436 Open Access 16 17 437 This is an open access article distributed in accordance with the terms of the Creative Commons 18 19 438 Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon 20 21 22 439 this work, for commercial use, provided the original work is properly cited. See: 23 24 440 http://creativecommons.org/licenses/by/4.0. 25 26 441 27 28 29 442 References 30 31 443 1. Ng M, Fleming T, Robinson M, et al. Global, regional, and national prevalence of overweight 32 444 and obesity in children and adults during 1980-2013: a systematic analysis for the Global Burden 33 445 of Disease Study 2013. Lancet 2014;384(9945):766-81. 34 446 2. Swinburn BA, Sacks G, Hall KD, et al. The global obesity pandemic: shaped by global drivers 35 and local environments. Lancet 2011;378(9793):804-14. 36 447 37 448 3. Swinburn B, Sacks G, Ravussin E. Increased food energy supply is more than sufficient to 38 449 explain the US epidemic of obesity. The American Journal of Clinical Nutrition 39 450 2009;90(6):1453-56. 40 451 4. Vandevijvere S, Chow CC, Hall KD, et al. Increased food energy supply as a major driver of 41 452 the obesity epidemic: a global analysis. Bulletin of the World Health Organization 2015;93:446- 42 453 56. 43 44 454 5. Adams J, Goffe L, Brown T, et al. Frequency and socio-demographic correlates of eating 45 455 meals out and take-away meals at home: cross-sectional analysis of the UK national diet and 46 456 nutrition survey, waves 1–4 (2008–12). International Journal of Behavioral Nutrition and 47 457 Physical Activity 2015;12(1):51. 48 458 6. Food Standards Agency. Food and You - Wave Four (2016). Report accessed online 30th July, 49 459 2018 from: https://www.food.gov.uk/research/food-and-you/food-and-you-wave-four 50 460 7. Bezerra IN, Curioni C, Sichieri R. Association between eating out of home and body weight. 51 52 461 Nutrition Reviews 2012;70(2):65-79. 53 462 8. Bowman SA, Vinyard BT. Fast Food Consumption of U.S. Adults: Impact on Energy and 54 463 Nutrient Intakes and Overweight Status. Journal of the American College of Nutrition 55 464 2004;23(2):163-68. 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj Page 21 of 29 BMJ

21 1 2 3 465 9. O'Donnell SI, Hoerr SL, Mendoza JA, et al. Nutrient quality of fast food kids meals. The 4 466 American Journal of Clinical Nutrition 2008;88(5):1388-95. 5 6 467 10. Reitzel LR, Regan SD, Nguyen N, et al. Density and Proximity of Fast Food Restaurants and 7 468 Body Mass Index Among African Americans. American Journal of Public Health 8 469 2014;104(1):110-16. 9 470 11. Fraser LK, Edwards KL. The association between the geography of fast food outlets and 10 471 childhoodConfidential: obesity rates in , UK. Health For & Place Review 2010;16(6):1124-8. Only 11 472 12. Marsh S, 2018. The Guardian, Ministers urged to ban fast food outlets from opening near 12 schools. Accessed on 30th July, 2018. News story archived at; 13 473 14 474 https://web.archive.org/web/20180720080358/https://www.theguardian.com/society/2018/apr/23 15 475 /ministers-urged-to-ban-fast-food-outlets-from-opening-near-schools 16 476 13. Donnelly L, 2018. The Telegraph, Fast food outlets should be banned from opening near 17 477 schools to tackle obesity epidemic, say doctors. Accessed on 30th July, 2018. News story 18 478 archived at: 19 479 https://web.archive.org/web/20180610181412/https://www.telegraph.co.uk/news/2018/04/22/fast 20 21 480 -food-outlets-should-banned-opening-near-schools-tackle/amp/ 22 481 14. Mintel. Mintel Market Sizes, Restaurants in UK(2017). Report accessed via University of 23 482 Liverpool Library, Liverpool, UK. March 2018. 24 483 15. Public Health England. Calorie reduction: the scope and ambition. Accessed on 30th July, 25 484 2018 from: https://www.gov.uk/government/publications/calorie-reduction-the-scope-and- 26 485 ambition-for-action 27 486 16. Hollands GJ, Shemilt I, Marteau TM, et al. Portion, package or tableware size for changing 28 29 487 selection and consumption of food, alcohol and tobacco. The Cochrane Database of Systematic 30 488 Reviews 2015(9):CD011045. 31 489 17. Kral TV, Roe LS, Rolls BJ. Combined effects of energy density and portion size on energy 32 490 intake in women. The American Journal of Clinical Nutrition 2004;79(6):962-8. 33 491 18. Rolls BJ, Roe LS, Meengs JS. The effect of large portion sizes on energy intake is sustained 34 492 for 11 days. Obesity (Silver Spring, Md) 2007;15(6):1535-43. 35 19. Sheen F, Hardman CA, Robinson E. Plate-clearing tendencies and portion size are 36 493 37 494 independently associated with main meal food intake in women: A laboratory study. Appetite 38 495 2018;127:223-29. 39 496 20. Public Health England. Sugar reduction and wider reformulation programme: Report on 40 497 progress towards the first 5% reduction and next steps. May, 2018. Accessed online 30th July 41 498 2018 from https://www.gov.uk/government/publications/sugar-reduction-report-on-first-year- 42 499 progress 43 44 500 21. Cleveland LP, Simon D, Block JP. Compliance in 2017 With Federal Calorie Labeling in 90 45 501 Chain Restaurants and 10 Retail Food Outlets Prior to Required Implementation. American 46 502 Journal of Public Health 2018;108(8):1099-102. 47 503 22. Block JP. The Calorie-Labeling Saga—Federal Preemption and Delayed Implementation of 48 504 Public Health Law. New England Journal of Medicine 2018. 49 505 23. Scourboutakos MJ, L'Abbe MR. Restaurant menus: calories, caloric density, and serving 50 506 size. American Journal of Preventive Medicine 2012;43(3):249-55. 51 52 507 24. Hearst MO, Harnack LJ, Bauer KW, et al. Nutritional quality at eight U.S. fast-food chains: 53 508 14-year trends. American Journal of Preventive Medicine 2013;44(6):589-94. 54 509 25. Mintel. Mintel Market Sizes, Fast Food and Takeaways in UK (2016). Report accessed via 55 510 University of Liverpool Library, Liverpool, UK. Accessed 1st March 2018 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj BMJ Page 22 of 29

22 1 2 3 511 26. Statista. Leading restaurant chains ranked by number of users in the (UK) 4 512 from 2017. 2017. Report accessed 1st March from 5 6 513 https://www.statista.com/statistics/586234/restaurant-chains-usage-in-the-uk-by-number-of- 7 514 users/ 8 515 27. Statista. Fast food restaurants in the United Kingdom (UK). 2018. Report accessed 1st March 9 516 2018 from https://www.statista.com/study/39567/fast-food-restaurants-in-the-united-kingdom- 10 517 uk-statista-dossier/Confidential: For Review Only 11 518 28. Jiao J, Moudon A, Kim S, et al. Health implications of adults’ eating at and living near fast 12 food or quick service restaurants. Nutrition & Diabetes 2015;5(7):e171. 13 519 th 14 520 29. NHS, UK. Accessed on 30 July 2018, Archived at 15 521 https://web.archive.org/web/20180511083938/https://www.nhs.uk/Livewell/weight-loss- 16 522 guide/Pages/losing-weight-getting-started.aspx 17 523 30. Urban LE, McCrory MA, Dallal GE, et al. Accuracy of Stated Energy Contents of Restaurant 18 524 . JAMA 2011;306(3):287-93. 19 525 31. Urban LE, Dallal GE, Robinson LM, et al. The accuracy of stated energy contents of 20 21 526 reduced-energy, commercially prepared foods. J Am Diet Assoc 2010;110(1):116-23. 22 527 32. Howard S, Adams J, White M. Nutritional content of supermarket ready meals and recipes 23 528 by television chefs in the United Kingdom: cross sectional study. BMJ : British Medical Journal 24 529 2012;345. 25 530 33. Robinson E, Aveyard P, Jebb SA. Is Plate Clearing a Risk Factor for Obesity? A Cross- 26 531 Sectional Study of Self-Reported Data in US Adults. Obesity (Silver Spring, Md) 27 532 2015;23(2):301-04. 28 29 533 34. Hinton EC, Brunstrom JM, Fay SH, et al. Using photography in 'The Restaurant of the 30 534 Future'. A useful way to assess portion selection and plate cleaning? Appetite 2013;63:31-5. 31 535 35. Urban LE, Weber JL, Heyman MB, et al. Energy contents of frequently ordered restaurant 32 536 meals and comparison with human energy requirements and US Department of Agriculture 33 537 database information: a multisite randomized study. Journal of the Academy of Nutrition and 34 538 Dietetics 2016;116(4):590-98. e6. 35 36. Block JP, Condon SK, Kleinman K, et al. Consumers’ estimation of calorie content at fast 36 539 37 540 food restaurants: cross sectional observational study. BMJ : British Medical Journal 2013;346. 38 541 37. Chandon P, Ordabayeva N. The accuracy of less: Natural bounds explain why quantity 39 542 decreases are estimated more accurately than quantity increases. Journal of Experimental 40 543 Psychology: General 2017;146(2):250. 41 544 38. Crockett RA, King SE, Marteau TM, et al. Nutritional labelling for healthier food or 42 545 non‐alcoholic drink purchasing and consumption. The Cochrane Library 2018, Feb 27; (2): 43 44 546 CD009315 45 547 39. Young LR, Nestle M. The contribution of expanding portion sizes to the US obesity 46 548 epidemic. American Journal of Public Health 2002;92(2):246-49. 47 549 40. Benson C. Increasing portion size in Britain. Society, Biology and Human Affairs 48 550 2009;74(2):4-20. 49 551 41. Lewis HB, Ahern AL, Solis-Trapala I, et al. Effect of reducing portion size at a compulsory 50 meal on later energy intake, gut hormones, and appetite in overweight adults. Obesity (Silver 51 552 52 553 Spring, Md) 2015;23(7):1362-70. 53 554 42. Rolls BJ, Roe LS, Meengs JS. Reductions in portion size and energy density of foods are 54 555 additive and lead to sustained decreases in energy intake. The American Journal of Clinical 55 556 Nutrition 2006;83(1):11-7. 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj Page 23 of 29 BMJ

23 1 2 3 557 43. Robinson E, Kersbergen I. Portion size and later food intake: evidence on the “normalizing” 4 558 effect of reducing food portion sizes. The American Journal of Clinical Nutrition 5 6 Restaurant chain Restaurant Number Mean (SD) Number and % Number and % 7 name type of meals kcal per meal meals ≤ 600kcals meals ≥ 1000kcals 8 9 Fast food 50 711.3 (214.0) 17 (34.0%) 4 (8.0%) 10 KFC Confidential:Fast food 106 986.6 For (273.2) Review5 (4.7%) Only53 (50.0%) 11 Leon Fast food 14 597.1 (86.3) 8 (57.1%) 0 (0.0%) 12 Mcdonalds Fast food 127 725.7 (242.3) 35 (27.6%) 14 (11.0%) 13 Subwaya Fast food 2436 762.9 (252.1) 760 (31.2%) 490 (20.1%) 14 Wimpy Fast food 64 721.1 (221.4) 17 (26.6%) 6 (9.4%) 15 16 Fast food restaurants (N=6)b 750.8 (128.4) 30.2% (16.8%) 16.4% (17.7%) 17 18 Sit down 33 870.6 (262.7) 5 (15.2%) 11 (33.3%) 19 Ask Sit down 44 789.8 (184.3) 7 (15.9%) 7 (15.9%) 20 Bills Sit down 16 966.0 (309.8) 2 (12.5%) 7 (43.8%) 21 Chef and Brewer Sit down 95 1177.1 (389.8) 6 (6.3%) 63 (66.3%) 22 Ember Inns Sit down 75 1084.8 (334.0) 5 (6.7%) 45 (60.0%) 23 Flaming Grill Sit down 52 1231.7 (495.6) 6 (11.5%) 36 (69.2%) 24 Sit down 62 1165.9 (370.0) 5 (8.1%) 43 (69.4%) 25 Sit down 333 1357.6 (471.5) 19 (5.7%) 261 (78.4%) 26 JD Weatherspoons Sit down 114 1119.1 (428.3) 16 (14.0%) 72 (63.2%) 27 Nando’sa Sit down 9293 1019.0 (230.6) 282 (3.0%) 4911 (52.8%) 28 Old English Inns Sit down 67 1125.2 (391.6) 6 (9.0%) 45 (67.2%) 29 Pizza Express Sit down 34 853.9 (234.1) 6 (17.6%) 7 (20.6%) 30 Sit down 33 974.8 (238.0) 4 (12.1%) 19 (57.6%) 31 Sizzling Pubs Sit down 87 1268.7 (575.0) 7 (8.0%) 56 (64.4%) 32 Sit down 37 963.3 (243.2) 2 (5.4%) 15 (40.5%) 33 Stone House Sit down 23 1274.5 (323.2) 0 (0.0%) 18 (78.3%) 34 Sit down 57 869.1 (273.2) 9 (15.8%) 17 (29.8%) 35 Sit down 20 941.6 (165.1) 1 (5.0%) 8 (40.0%) 36 Vintage Inns Sit down 40 1063.7 (413.5) 6 (15.0%) 21 (52.5%) 37 Sit down 40 836.0 (259.1) 7 (17.5%) 12 (30.0%) 38 Sit down 44 734.7 (336.7) 23 (52.3%) 10 (22.7%) 39 40 Sit down restaurants (N=21)b 1032.7 (174.7) 12.2% (10.5%) 50.3% (19.4%) 41 559 2018;107(4):640-46. 42 43 560 44. Oakes ME. Bad foods: Changing attitudes about what we eat. 2004. Transaction Publishers, 44 561 New Jersey. 45 562 45. Zlatevska N, Neumann N, Dubelaar C. Mandatory Calorie Disclosure: A Comprehensive 46 563 Analysis of Its Effect on Consumers and Retailers. Journal of Retailing 2018;94(1):89-101. 47 564 48 565 49 566 50 51 567 52 568 53 569 Table 1. Kilocalorie content of meals from eligible restaurant chains included in analyses 54 55 570 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj BMJ Page 24 of 29

24 1 2 3 571 a The relatively large number of eligible meals identified in some restaurant chains was due to 4 572 there being a large number of meal variants (e.g. chicken with a choice of any two sides, 5 6 573 sandwich meal with a choice of bread type, size, and sides) in these restaurants. 7 574 8 575 b For descriptive purposes, values in this row represent the mean (SD) of individual restaurant 9 576 values for mean kcals per meal, % of ≤ 600kcal meals and % of ≥ 1000kcals meals. 10 577 11 Confidential: For Review Only 12 578 13 14 15 579 16 17 580 18 19 581 20 21 582 Table 2. Kilocalorie content of burger and fries/chips meals and salad meals from eligible 22 583 restaurant chains included in analyses 23 584 24 25 Restaurant chain Restauran Number Mean kcal (SD) Number Mean kcal (SD) of 26 name t type of salad of salad meals of burger burger meals 27 meals meals 28 29 Burger King Fast food 1 210 (-) 24 842.5 (214.4) 30 KFC Fast food 3 663.3 (120.5) 14 1220.4 (322.3) 31 Leon Fast food 4 554.5 (40.2) - - 32 Mcdonalds Fast food 8 248.0 (75.7) 24 907.3 (141.2) 33 Subwaya Fast food 192 416.3 (124.2) - - 34 Wimpy Fast food 4 371.8 (67.8) 22 897.9 (137.8) 35 36 Fast food restaurantsb 410.7 (174.8), N=6 966.8 (171.4), N=4 37 38 All Bar One Sit down 4 605.8 (294.9) 6 1054.8 (246.6) 39 Ask Sit down 4 649.5 (268.1) - - 40 Bills Sit down 1 902.0 (-) 3 1206.3 (148.4) 41 Chef and Brewer Sit down 7 558.1 (191.1) 8 1459.0 (187.9) 42 Ember Inns Sit down 3 574.7 (174.5) 7 1295.1 (294.4) 43 Flaming Grill Sit down 1 325.0 (-) 8 1431.3 (225.3) 44 Harvester Sit down 4 551.5 (13.2) 7 1414.0 (149.2) 45 Hungry Horse Sit down 4 393.5 (295.2) 14 1965.7 (770.8) 46 JD Weatherspoons Sit down 5 494.0 (138.5) 14 1564.8 (318.1) 47 Nando’sa Sit down 30 427.8 (150.4) 912 1160.8 (153.6) 48 Old English Inns Sit down 1 247.0 (-) 6 1543.2 (212.7) 49 Pizza Express Sit down 4 885.8 (296.1) - - 50 Pizza Hut Sit down - - - - 51 Sizzling Pubs Sit down 2 280.0 (103.2) 10 1521.2 (400.9) 52 Slug and Lettuce Sit down 4 735.8 (214.6) 5 1280.2 (148.7) 53 Stone House Sit down 2 1353.0 (158.4) - - 54 Table Table Sit down 3 329.0 (88.4) 6 1105.2 (166.0) 55 Toby Carvery Sit down 2 613.5 (99.7) - - 56 Vintage Inns Sit down 3 380.0 (335.1) 2 1069.0 (195.2) 57 58 59 60 https://mc.manuscriptcentral.com/bmj Page 25 of 29 BMJ

25 1 2 3 585 Wagamama Sit down 4 413.5 (47.8) - - 4 586 Zizzi Sit down 4 466.8 (212.6) - - a 5 587 The 6 Sit down restaurantsb 559.3 (260.9), N=20 1362.2 (248.5), N=14 7 588 rela 8 589 tively large number of eligible meals identified in some restaurant chains was due to there being 9 590 a large number of meal variants (e.g. chicken with a choice of any two sides, sandwich meal with 10 591 a choiceConfidential: of bread type) in these restaurants. For Review Only 11 592 12 b For descriptive purposes, values in this row represent the mean (SD) of individual restaurant 13 593 14 594 values for mean salad meal kcals and mean burger meal kcals 15 595 16 596 - indicates absence of salad or burger meal from restaurant chain menu 17 597 18 598 (-) indicates absence of SD as only one eligible meal from restaurant 19 599 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj BMJ Page 26 of 29

1 1 2 3 4 Supplementary Materials for (Over)Eating Out: An observational study of the energy content of 5 6 main meals served in major UK restaurant chains 7 8 Supplementary Table 1 .Restaurant chains with ≥50 outlets identified and eligibility decisions 9 10 Restaurant chain name N Eligible: Y/N (reason for 11 Confidential:outlets ineligibility) For Review Only 12 Subway 2559 Y 13 McDonalds 1249 Y 14 KFC 900 Y 15 Burger King 500 Y 16 130 N (No calorie info) 17 Pizza Gogo 108 N (No calorie info) 18 19 Favourite Fried Chicken 88 N (No calorie info) 20 84 N (Individual items only) 21 Wimpy 78 Y 22 Yo! Sushi 77 N (Individual items only) 23 Pepe’s Piri piri 67 N (No calorie info) 24 61 N (No calorie info) 25 Chopstix 57 N (No calorie info) 26 57 N (Individual items only) 27 Leon 51 Y 28 882 Y 29 30 Pizza express 490 Y 31 Nando's 360 Y 32 Frankie & Benny's 263 N (No calorie info) 33 Pizza Hut 261 Y 34 Hungry horse 261 Y 35 260 N (No calorie info) 36 Sizzling Pubs 231 Y 37 Harvester 204 Y 38 Vintage Inns 192 Y 39 Zizzi 174 Y 40 Toby Carvery 169 Y 41 42 Ember Inns 146 Y 43 Chef & Brewer 145 Y 44 Flaming Grill 131 Y 45 Wagamama 120 Y 46 Lounges 120 N (No calorie info) 47 Ask 114 Y 48 Carluccio's 103 N (No calorie info) 49 Old English Inns 100 Y 50 Café Rouge 95 N (No calorie info) 51 Côte 94 N (No calorie info) 52 53 Miller & Carter 93 N (No calorie info) 54 91 N (Individual items only) 55 90 N (No calorie info) 56 Stone House 90 Y 57 58 59 60 https://mc.manuscriptcentral.com/bmj Page 27 of 29 BMJ

2 1 2 3 4 TGI Fridays 82 N (No calorie info) 5 Bill's 82 Y 6 7 Chiquitos 73 N (No calorie info) 8 Slug and Lettuce 70 Y 9 Byron 69 N (No calorie info) 10 Table Table 69 Y 11 Pepe’s PiriConfidential: piri 67 N (No calorieFor info) Review Only 12 56 N (No calorie info) 13 54 N (No calorie info) 14 All Bar One 52 Y 15 Giraffe 52 N (No calorie info) 16 17 Note. Reasons for ineligibility: ‘No calorie info’ = restaurant website did not provide calorie information 18 19 for menu items, and did not provide when requested by email (n = 21), ‘Individual items only’ = 20 restaurant only sold individual items on menus (e.g., sushi, burger and fries sold separately), not meals (n 21 = 4). 22 23 24 Additional Information on Coding Instructions Used 25 We will only include meal options that are available all year round. We will therefore exclude 26 seasonal menu options (e.g. meal options from a ‘spring menu’). 27 28 29 Menu options in which a customer can ‘add’ items to the meal (e.g. ‘extra burger patty for £2’) 30 would not be included, as we are only including the default configuration of menu options 31 offered by restaurants. 32 33 Food items that are sold individually and would typically be served as part of a meal are 34 ineligible. For example, based on our eligibility criteria a burger on its own would not constitute 35 a lunch or dinner meal in full, because in most restaurants serving burgers, burgers are served 36 37 with at least one additional meal component (e.g. a burger and fries, a burger with a side salad). 38 Further examples would be an individual hot dog or soup (e.g. with no bread), an individual 39 sandwich (e.g. with no side of crisps, chips or salad garnish). Small plates ‘tapas style’ dishes are 40 shared. Therefore, these menu options are not eligible. 41 42 Menu options that could be perceived to be ‘individual items’, but are typically served as a meal 43 44 would be eligible. For example, pizza tends to be served on its own (e.g. no additional side), so a 45 pizza would be classed as a main meal menu option. 46 47 A salad or a jacket potato that is clearly marketed as a side dish would not be eligible. However, 48 based on our inclusion criteria, a multi-component salad or jacket potato that is marketed as a 49 main meal (e.g. salad with halloumi) or jacket potato (e.g. jacket potato served with beans) 50 would be eligible. 51 52 53 Main menu options that are labelled as ‘healthy eating’ or ‘smaller appetites’ and appear to be 54 main meal options are eligible. 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj BMJ Page 28 of 29

3 1 2 3 4 Carvery menu sections are meals that involve a portion of meat (or multiple portions of different 5 6 meats) being served to a patron and the patron then selecting their accompanying sides (and size 7 of portion) from a large list of menu options are not eligible (because it is not clear what 8 constitutes a standard meal configuration). 9 If a restaurant has a specified take-away section of the menu that has menu options which are not 10 available on the in-store eating menu, these menu options are not eligible. 11 Confidential: For Review Only 12 13 14 Chi-Squared Results 15 Proportion of meals ≤600kcals from total sample: The Chi-squared test was significant (χ2(1) = 16 1762.1, p <.001) demonstrating that sit down restaurants were less likely to offer meals 17 exceeding public health recommendations (4.0% of all sit down meals) than fast food restaurants 18 (30.1% of all fast food meals). 19 20 21 Proportion of meals ≥1000kcals from total sample: The Chi-squared test was significant (χ2(1) = 22 989.3, p <.001) demonstrating that sit down restaurants (53.6% of all sit down meals) were more 23 likely to offer extremely calorific meals than fast food restaurants (20.3% of all fast food meals). 24 25 Weighted Multi-Level Model 26 Because there was an unexpectedly large amount of variability in the number of meals that 27 individual restaurants contributed to analyses we also conducted weighted multi-level analyses. 28 29 We computed raw weights for level 1 (meal level) and level 2 units (restaurant level), using the 30 formulas set out in Pillinger (2011). Conditional level 1 weights were calculated as 1 / wi|j, 31 where wi|j was the probability of a meal being selected from all meals in the same restaurant. 32 Level 2 weights were calculated as 1 / wj, where wj is the probability of selection of the 33 restaurant that the meal belongs to from among all restaurants in the sample. When accounting 34 for these weights in the model the two level structure (meals within restaurants) was a better fit 35 of the data than a single level structure, χ2(1) = 4180, p < .001, indicating that multi-level 36 37 modelling was appropriate. The variance partition coefficient; the total residual variance which is 38 attributable to restaurants rather than individual meals was 27.5%. Type of restaurant (sit down 39 vs. fast food) was a significant predictor, β = 327.0, SE = 61.4 (95% CIs 206.6 to 447.3), p < 40 .001, explaining 44.5% of variance at the restaurant level. These results indicate that meals from 41 sit down restaurants had 327kcals more energy than meals from fast food restaurants, on average. 42 43 44 Mean kilocalorie content of specific meal types. Across all burger and fries/chips meals (N = 45 1,904; 1,010 sit down, 894 fast food) the average number of kcals was 1170.9 (SD = 231.3). The 46 weighted multilevel model demonstrated a two level structure was a better fit than a single level 47 structure, χ2(1) = 688, p < .001, and the variance partition coefficient was 35.7%. Type of 48 restaurant (sit down vs. fast food) was a significant predictor, β = 279.0, SE = 108.6 (95% CIs 49 66.2 to 491.7), p < .001, explaining 12.9% of variance at the restaurant level and indicating that 50 burger meals in sit down restaurants had 279kcals more energy than in fast food restaurants, on 51 52 average. Across salad meals (N = 304; 92 sit down, 212 fast food) the average number of kcals 53 was 446.2 (SD = 182.0). A two level structure was a better fit than a single level structure, χ2(1) 54 = 275, p < .001, and the variance partition coefficient was 77.6%. Type of restaurant (sit down 55 vs. fast food) explained 8.2% of variance at the restaurant level and sit down restaurant salad 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj Page 29 of 29 BMJ

4 1 2 3 4 meals had on average 201kcals more than fast food meals, although restaurant type was not a 5 6 statistically significant predictor in the model, β = 200.9, SE = 127.6 (95% CIs -49.3 to 451.0), p 7 = .058. 8 9 10 11 Confidential: For Review Only 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 https://mc.manuscriptcentral.com/bmj