1. *We Very Much Enjoyed Reading the Paper and Appreciate
Total Page:16
File Type:pdf, Size:1020Kb
1 Detailed comments from the meeting: 1. *We very much enjoyed reading the paper and appreciate that while this the research question is not novel, the UK perspective is unusual given that much of this research has been undertaken in the US. We would like to thank the committee for their positive comments on our manuscript. 2. *We found the definition of a main meal difficult to appreciate and think this could do with better description and also a few examples to help readers appreciate this. We have now expanded on our definition of a main meal and provided examples, see page 7. We defined a main meal as being a menu option that would normally be the primary dish in a lunch or dinner meal and typically be found in the ‘main course’ part of a typical restaurant menu. Examples of main meal items using this definition would include burger and chips, chicken Caesar salad, spaghetti Bolognese, jacket potato with filling. Throughout the section of the main meal eligibility section of the method we have provided examples to make it clear what was and was not eligible. See page 7-8. 3. *The statistical analysis seems overcomplicated and inconsistent in places. Multilevel model is appropriate, but the goodness of fit test shows only that restaurants differ in mean energy which is unsurprising. Why use bootstrap for standard errors when the linear model provides them? Please comment We bootstrapped the data to improve accuracy and reduce bias of parameter estimations. In MLWiN bootstrapping the standard errors ensures they are properly scaled along with coefficients. We report this now on page 9. However, we note that running non-bootstrapped analyses did not alter the findings reported. 4. *The weighted multilevel model is unnecessary. Please also use logistic multilevel model to analyse proportions <600 and >1000 kcals. We no longer report the weighted multilevel model for our main analyses. Our rationale for its inclusion was that there were a small number of restaurants contributing a very large number of meals. For this reason we have retained the analysis in the supplementary materials. We now provide logistic multilevel models for the analysis of proportions. See page 10 and 12-13. 5. *We noticed that the results claim difference in mean intake between sit-down and fast- food, yet it is not significant (page 12), please reconcile this. All of the comparisons between sit-down and fast food are statistically significant at p = .05 aside from the analysis on page 12 (p = .08) examining the kcal content of salad meals in the two types of restaurant. In this instance we are reliant on a much smaller sample size as the analysis is limited to salad meals and we believe it is probably a power issue (the mean difference is 201kcals, p = 0.08) given the reduced sample size. We have now clearly highlighted this in the Discussion on page 15-16. This analysis has no bearing on the main results reported abstract, as the analyses focusing on specific food types were additional secondary analyses because from the outset we reasoned that sample size would be a lot smaller (as outlined in pre-registered plan). 6. *What proportion of adults are estimated eat <600 kcal for main meal? This information would be helpful as background if available. This is an interesting question, but one we do not know the answer to. We have searched literature and re-read PHE’s reports relating to the 400-600-600kcal public health campaign and calorie reduction scheme. The only viable data source to address this question we are aware of (in the UK) is the National Diet and Nutrition Survey (NDNS). 2 One problem with the NDNS is that it is only a 1 day snapshot of a person’s dietary intake and information on main meal kcal content is not readily available. We would most likely need to code the data ourselves based on our definition of a main meal and we do not have sufficient time to be able to do so. However, we believe that there is a bigger problem with the NDNS: it is based entirely on self-reported dietary intake. Based on the NDNS the UK should have been losing weight, but we know this isn’t the case and it’s because self-reported energy intake is very inaccurate. Taken directly from PHE’s 2018 Calorie Reduction Report: ‘Current estimates of UK energy intakes from the NDNS show that mean reported energy intakes are 18-20% below the SACN estimated average requirements (EAR) in all age and sex groups, except for children aged under ten years when reported figures tend to be close to the EAR’. So any estimate of the % of adults meeting the 600kcal recommendation will be an underestimate and because reporting bias is associated with gender, BMI and the amount of food one is eating and also the types of foods being reported on, it would be difficult to adjust for this bias accurately. We appreciate that giving some numbers for background context is helpful to the reader and we have now provided info on PHE’s projection that the average adult in the UK is consuming approximately 195 kcals a day in excess. See page 4. 7. *Is it more appropriate to refer to energy intake rather than kcalorie throughout? We have reviewed the manuscript and prefer to use kcal predominantly because this is the unit of measurement for the study variable and meals high in kcals contribute to excess daily energy intake (so the two are distinct concepts in this context). We have now made sure that when referring to kcals and energy intake the differentiation is clear to the reader, e.g. on page 4 ‘Recent public health recommendations made by Public Health England suggest that adults should aim to consume 600 kilocalories or less for their main lunch and dinner meals to avoid excess daily energy intake and maintain a healthy body weight.’ 8. *We wonder if it is important to comment on the fact that the lines between sit down meals and fast food outlets are becoming more blurred with the introduction of services that pick up and deliver food from traditionally sit down outlets? We have now commented on this in the Discussion. See page 16. 9. *For readers outside the UK, please can you clarify if the information on calorie content is freely available on line and on menus as some of our international editors were confused by this. Most restaurants do not have kcal info on their in store menus (based on a different research project we estimate about 15% of chains to provide this information), but many do provide nutritional info online. We have now noted this in the Method (8) and Discussion (page 15). 10. *We wondered to what extent are chain restaurants with 50 or more branches different from fast food outlets? Are they similar to establishments with one or two branches? Why was the threshold of 50 branches chosen? Are these all in a similar price range? Or are some "fancier" the others? The 50 outlet threshold was chosen because to make the work feasible we had to set a threshold (e.g. there are chains with 2 outlets but many are regional and therefore not what we would consider ‘major chains’ one would find in most high streets). Our scoping research suggested that few restaurants with less than 50 chains provided nutritional information, as it appears to be the larger chains who have the resources to do this. Thus, we reasoned that 50 was a reasonable number. We have made this clearer in the manuscript. See page6. 3 Regarding what the chain restaurants are like, this is variable, but we think they are distinct to fast food outlets. There are some that are cheap and position themselves as value for money (e.g. Hungry Horse, JD Wetherspoons). Others are mid-range (e.g. Pizza Express, Nando’s). There are no restaurants that we would consider ‘high end’ but this is a matter of perspective. All of the included restaurants and their meal kcal information is presented in Table 1 and in the supplementary materials we present the number of outlets for restaurants that did vs. did not provide nutritional information online. A more general point is that there is variability in kcal content of meals across individual restaurants and understanding why would be of interest (we agree with the committee’s comment that price range or market positioning may come into play), so we now mention this in the Discussion. See page 16. 11. *We think there should be greater discussion around the fact that this is based on restaurant reported calorie content and that this is a potential weakness of this study. We agree and have expanded on this section of the Discussion, see page 13-14. Reviewer: 1 12. Thank you for the opportunity to peer review this article. It presents an observational study of energy content of meals in sit-down restaurant chains in the UK. The text was easy to follow and the study has the potential to enrich knowledge in the field of nutritional epidemiology. Please find below my questions/comments/suggestions regarding the methods and presentation of the work. We would like to thank the reviewer for their positive comments and helpful suggestions. 13. Abstract: please include the number of studied restaurant chains by type (27 6 and 21) and the number of studied eligible meals (13,396); also please clearly state that breakfast meals and beverages were excluded from the kcal count.