Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A temporal analysis of Canadian dietary choices using the Canadian Community Health Survey Cycle 2.2… Yang, Hui Wen 2013

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2013_spring_yang_huiwen.pdf [ 1.08MB ]
Metadata
JSON: 24-1.0073762.json
JSON-LD: 24-1.0073762-ld.json
RDF/XML (Pretty): 24-1.0073762-rdf.xml
RDF/JSON: 24-1.0073762-rdf.json
Turtle: 24-1.0073762-turtle.txt
N-Triples: 24-1.0073762-rdf-ntriples.txt
Original Record: 24-1.0073762-source.json
Full Text
24-1.0073762-fulltext.txt
Citation
24-1.0073762.ris

Full Text

   A TEMPORAL ANALYSIS OF CANADIAN DIETARY CHOICES USING THE CANADIAN COMMUNITY HEALTH SURVEY CYCLE 2.2: DOES NUTRIENT INTAKE AND DIET QUALITY VARY ON WEEKENDS VERSUS WEEKDAYS?  by Hui Wen Yang B.Sc., University of Toronto, 2009  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Human Nutrition)   THE UNIVERSITY OF BRITISH COLUMBIA  (Vancouver)  January 2013 © Hui Wen Yang, 2013  ii  Abstract Since dietary intake varies from day to day, research on the timing of dietary behaviours is essential for understanding the complexity of contemporary dietary patterns needed to inform nutrition-related health policies and recommendations.  Limited studies with inconsistent results have suggested that dietary intake differs on weekends versus weekdays.  Although findings from outside of Canada have previously reported that energy intake is higher on weekend days, the nature of weekday-weekend variation in dietary intake among Canadians remains unknown. In response, this study evaluated the difference in energy, nutrient intake and diet quality on weekdays versus weekend days in the Canadian population and whether temporal differences were moderated by sex, age or employment status.  Data were analyzed from participants aged >1 year, excluding pregnant or breastfeeding women (n=34,402) in the Canadian Community Health Survey Cycle 2.2, a nationally representative survey which included 24-hour dietary recall data.  Linear regression models examined the difference in energy intake, nutrient intake and diet quality (assessed using Healthy Eating Index [HEI]) between weekdays (Monday- Thursday) and weekend days (Friday-Sunday).  Caloric intake was found to be 62 kcal (SE = 23) higher on weekend days than on weekdays.  Compared to weekdays, energy-adjusted weekend intakes of carbohydrates, protein, and the majority of micronutrients were significantly lower, ranging from 2.0% to 6.6% lower, while alcohol and cholesterol intakes were 66% and 10% higher on weekends, respectively.  HEI was significantly lower on weekends than on weekdays (56.4 vs. 58.3 out of 100).  With the exception of alcohol, the magnitude of weekday-weekend differences of most of the dietary outcomes did not differ substantially by sex, age or employment status.  In conclusion, Canadians consume foods with a slightly less favorable nutrient profile and marginally poorer diet quality on weekends than on weekdays. iii  Table of Contents Abstract .......................................................................................................................................... ii Table of Contents ......................................................................................................................... iii List of Tables ................................................................................................................................ vi List of Figures .............................................................................................................................. vii List of Abbreviations ................................................................................................................. viii Acknowledgements ...................................................................................................................... ix Chapter  1: Introduction ...............................................................................................................1 Chapter  2: Literature Review ......................................................................................................3 2.1 Sources of variation in nutrient intake ............................................................................ 3 2.2 Evidence regarding differences in dietary intake on weekends versus weekdays .......... 5 2.2.1 Weekday-weekend differences in nutrient intakes in children and adolescents ......... 6 2.3 Weekday-weekend differences in adults ........................................................................ 7 2.3.1 Weekday-weekend difference in older adults ............................................................. 7 2.3.2 Findings from nationally representative datasets ........................................................ 8 2.4 The influence of weekend intake on diet quality ............................................................ 9 2.5 Determinants of weekly variation and potential implications for nutrient intakes ....... 10 2.6 Key gaps in current knowledge..................................................................................... 11 2.7 Research objectives ....................................................................................................... 12 2.8 Study hypotheses .......................................................................................................... 12 Chapter  3: Methods ....................................................................................................................14 3.1 Data source.................................................................................................................... 14 3.1.1 CCHS 2.2. Survey data ............................................................................................. 14 iv  3.1.2 Target population ...................................................................................................... 15 3.1.3 Sampling procedure .................................................................................................. 15 3.1.4 Dietary assessment procedure ................................................................................... 17 3.2 Analytical samples ........................................................................................................ 18 3.3 Variables of interest ...................................................................................................... 18 3.3.1 Independent variables ............................................................................................... 18 3.3.2 Outcome variables .................................................................................................... 20 3.3.3 Interaction terms ....................................................................................................... 25 3.3.4 Variables for sample characteristics ......................................................................... 26 3.4 Statistical analysis ......................................................................................................... 27 3.4.1 Descriptive statistics ................................................................................................. 27 3.4.2 Assessing associations between weekday-weekend intake and dietary outcomes ... 29 3.4.3 Assessing moderating effects of age, sex, and work/school status ........................... 31 3.4.4 Survey weighting and bootstrapping ........................................................................ 33 Chapter  4: Results.......................................................................................................................35 4.1 Analytical sample characteristics .................................................................................. 35 4.2 Mean nutrient intakes between weekdays and weekends ............................................. 37 4.3 Percent of change in nutrient intake on weekends compared to weekdays .................. 41 4.4 Mean Healthy Eating Index scores between weekday and weekend groups ................ 44 4.5 The moderating effects of sex, age groups, and employment status on differences in weekday-weekend dietary intake .............................................................................................. 46 4.5.1 Sex as a moderating variable .................................................................................... 49 4.5.2 Age group as a moderating variable ......................................................................... 55 v  4.5.3 Employment/student status as a moderating variable ............................................... 57 Chapter  5: Discussion and Conclusion .....................................................................................61 5.1 Weekday-weekend difference in dietary outcomes ...................................................... 61 5.1.1 Macronutrient outcomes ........................................................................................... 61 5.1.2 Micronutrient outcomes ............................................................................................ 63 5.1.3 Dietary quality and food group intakes ..................................................................... 66 5.2 The weekend effect by sex, age and employment ........................................................ 68 5.3 Strengths and limitations............................................................................................... 69 5.4 Implications on dietary assessment and future research directions .............................. 72 5.5 Conclusion .................................................................................................................... 73 References .....................................................................................................................................74 Appendices ....................................................................................................................................83 Appendix A ............................................................................................................................... 83 Appendix B ............................................................................................................................... 86  vi  List of Tables Table 1 Nutrients of interest ......................................................................................................... 21 Table 2 Scoring criteria of Canadian adaptation of healthy eating index-2005 ..........................  23 Table 3 Analytical sample characteristics .................................................................................... 36 Table 4 Linear regression analysis comparing the intake of nutrients of Canadians aged 1 and older on weekend days with weekday intake: unadjusted and energy-adjusted models. ............. 38 Table 5 Linear regression analysis comparing the healthy eating index scores of Canadians aged 1 and older on weekend days with weekday intake: unadjusted and energy-adjusted models ..... 45 Table 6 The p-values from the Wald tests determining if adding interaction terms improved the overall fit of the linear regression models, with type of day as a dummy variable and nutrient intake as outcome variables in Canadians aged 1 and older. ........................................................ 47 Table 7 The p-values from the Wald tests determining if adding interaction terms improved the overall fit of the linear regression models, with type of day as dummy variable and Healthy Eating Index as outcome variable in Canadians aged 1 and older. ............................................... 48 Table 8 Energy-adjusted mean nutrient intakes and Healthy Eating Index (HEI) scores on weekdays and weekend days among Canadians aged 1 and older, by sex, age, and employment/student status. ........................................................................................................... 50 Table 9 Comparison in weekday-weekend difference in energy and macronutrient intake between current study and other studies that used nationally representative samples. ................ 64  vii  List of Figures Figure 1 Flow chart of the analytical strategy.............................................................................. 28 Figure 2  Unadjusted and energy-adjusted percent of change in the intakes of macronutrients, dietary fibre, sugars and different types of fats of Canadians aged 1 and older on weekend days relative to weekdays ...................................................................................................................... 42 Figure 3 Unadjusted and energy-adjusted percent of difference in the intakes of micronutrients of Canadians aged 1 and older on weekend days relative to weekdays ........................................ 43 Figure 4 The intake of alcohol on weekdays and weekend days of Canadians aged 1 and older, by age groups. ............................................................................................................................... 56 Figure 5 The intake of alcohol on weekdays and weekend days of Canadians aged 1 and older, by employment/student status. ...................................................................................................... 59  viii  List of Abbreviations AMPM Automated Multiple-Pass Method CCHS 2.2 Canadian Community Health Survey Cycle 2.2 CFG Canada’s Food Guide CI Confidence interval CSFII Continuing Survey of Food Intake by Individuals DFE Dietary folate equivalents DRI Dietary Reference Intake g Gram HEI Healthy Eating Index kcal Kilocalories kg Kilogram mcg Microgram mg Milligram MUFA Monounsaturated fatty acid NE Niacin equivalents NFCS Nationwide Food Consumption Survey PUFA Polyunsaturated fatty acid R 2  Coefficient of determination RAE Retinol activity equivalents RDC Research Data Centre SE Standard error USDA United States Department of Agriculture  ix   Acknowledgements This research project would not have been possible without the support of many great people.  First and foremost, I would like to express the deepest appreciation to my supervisor, Dr. Jennifer Black, who was abundantly helpful and offered countless invaluable assistance, encouragement, and mentorship throughout the process of completing my master degree.  I would also like to thank my supervisory committee members, Dr. Susan Barr and Dr. Hassan Vatanparast, for their expertise, support and valuable insights into the research project. Additional thanks to Didier Garriguet and Jean-Michel Billette from Statistics of Canada for sharing research methodology to compute the Healthy Eating Index Score.  Sincere thanks to Lee Grenon and the staff members of British Columbia Inter-University Research Data Centre for providing great statistical support.  I would also like to thank Yumi Kondo from the Department of Statistics for the statistical consultation on survey sampling and bootstrapping analysis.  Lastly, special thanks are owed to my parents and my husband, Kevin, for their patience, endless love and support through the duration of my study. 1  Chapter  1: Introduction Diet is an important modifiable risk factor for many chronic diseases such as diabetes and cardiovascular disease (Mann, 2002).  In order to better inform public health strategies and interventions to improve nutritional status in a country such as Canada, it is essential to monitor what Canadians are eating in order to evaluate their compliance with national dietary recommendations.  The types of data which are traditionally collected to assess population-level dietary intake include measures such as mean caloric or nutrient intake for a group collected on a single day or usual intake of several days’ intake; however, when combining dietary data collected on several different days, important information relating to dietary patterns and food habits can be missed since people do not eat the same foods in the same quantity every day. Therefore, research on the timing of dietary behaviours is essential for understanding the complexity of contemporary dietary patterns needed to inform nutrition health policies and recommendations. Understanding how dietary intake varies over the course of a week is also important for identifying potential bias in estimates of usual or habitual intakes using short-term dietary measurements such as 24-hour recall.  In order to accurately measure the usual intake of a group, methodological questions must be considered: How many days of dietary assessment are required, and which days should be included?  These are important factors when designing dietary data collection strategies given the need to consider the feasibility of multiple days of collection due to cost, and the burden on participants of providing detailed data on multiple days. For population surveys, it is generally acceptable that single 24-hour recalls, with a second recall on a representative subset of sample to control for within-person variability is adequate to determine a usual intake distribution (Carriquiry 2003).  However, there have been debates as to 2  the need for collecting dietary data from different days of the week.  For example, if individuals tend to have higher caloric or nutrient intake on weekend days but data are only collected on weekdays, estimates of total or average dietary intake could be significantly underestimated.  In order to estimate the potential impact of temporal effects on intake estimation, knowledge of the nature of variation in dietary intake on different days of the week is required. Currently there is still a gap in knowledge about how specific days of the week can impact the estimates of nutrient intake in the Canadian population.  In response, this project aims to examine the differences in dietary intake between weekdays and weekend days using the Canadian Community Health Survey cycle 2.2 (CCHS 2.2), a nationally representative survey containing information on dietary intake, health status, and behavioural determinants of health and socio-demographic characteristics of 35,107 Canadians from all age groups.  CCHS 2.2 serves as a valuable source for assessing the nutritional well-being of Canadians.  Since CCHS 2.2 contains dietary information collected on all days of the week, it provides a unique opportunity to investigate how or whether dietary intake differs between weekdays and weekends in the Canadian population.  Obtaining a better understanding of the temporal variations of dietary intake may provide potential targets for intervention programs and health policies that promote better eating habits among Canadians. 3  Chapter  2: Literature Review The following section summarizes the literature available on (1) key sources of variation that shape dietary intake and their implications for assessing nutritional status of a population, (2) the role of day of the week on dietary intake in children, adults and older adults, (3) weekday- weekend differences in diet quality and eating habits, (4) the potential determinants that could influence dietary intake across the course of a week, and (5) the remaining gaps in the literature relating to temporal variation in dietary intake.  This chapter concludes with the research objectives and the hypotheses of this study. 2.1  Sources of variation in nutrient intake An important goal of nutrition surveys is to accurately capture the distribution of usual dietary intake, which reflects habitual food intake over an extended period of time in a given population.  Estimating usual intake from short-term dietary assessment tools such as 24-hour recalls and food diaries can however, be challenging since food intake fluctuates considerably from one day to another in free-living individuals (Beaton et al., 1979; Cai et al., 2004; Jahns, Carriquiry, Arab, Mroz, & Popkin, 2004; McGee, Rhoads, Hankin, Yano, & Tillotson, 1982; Palaniappan, Cue, Payette, & Gray-Donald, 2003; Verly Junior et al., 2010).  There are two main types of variations in dietary intake in humans known as between-person variation and within- person variation (Basiotis, Thomas, Kelsay, & Mertz, 1989; Beaton, Milner, McGuire, Feather, & Little, 1983).  Between-person variation refers to the variation in dietary intake from subject to subject across a population, which is influenced by factors such as age and sex as a result of differences in energy requirements (Basiotis et al., 1989).  Within-person variation, on the other hand, refers to the day-to-day variation in dietary intake within people themselves. 4  When measuring dietary intake, it is often desirable to account for within-person variation in order to obtain a reliable estimate of the true between-person variation within a given group either by increasing the number of days of dietary record collection or by statistical adjustments (Chalmers et al., 1952; Eppright, Patton, Marlatt, & Hathaway, 1952; Palaniappan et al., 2003).  This requires knowledge of the sources of variation in dietary intake, as well as the factors that influence the contribution of within- and between-person variation to total variation in dietary intake (Cai et al., 2004; Cai et al., 2005).  There is some evidence that temporality or the timing of food intake contributes to within-person variation.  For example, ”day of the week” has been identified as a component of within-person variation in dietary intake among adolescents (Verly Junior et al., 2010), adults (Basiotis et al., 1989; Beaton et al., 1979; Beaton et al., 1983; Cai et al., 2004; Cai et al., 2005; Jackson, Byrne, Magarey, & Hills, 2008; Jula, Seppanen, & Alanen, 1999; McGee et al., 1982) and older adults (Maisey, Loughridge, Southon, & Fulcher, 1995); however, the impact of such temporal effects on the variability of  intake remains inconclusive.  Some studies have reported that that dietary intake varies significantly over the course of the week and thus, that such temporal patterns in eating behaviours must be taken into consideration when designing appropriate dietary assessment for measuring typical diets of free-living individuals (Beaton et al., 1979; Tarasuk & Beaton, 1992).  There may also be differences in the degree of temporal variation for different nutrients, with some nutrients more variable than others.  For example, intakes of energy, protein and fat have been shown to fluctuate over the course of the week while carbohydrate intake tends to remain relatively stable over time (Jackson et al., 2008; Maisey et al., 1995; Nicklas, Farris, Bao, Webber, & Berenson, 1997).  Based on these findings, proportional sampling of a mix of days of the week is 5  recommended in order to minimize the potential bias in the timing of data collection due to day of the week variation of the diet. In contrast, Cai et al. reported that day of the week contributed less than 5% of total intake variance based on the analysis of  24 days of 24-hour recalls collected from Chinese men (Cai et al., 2005) and women aged 40-70 over a 1 year period (Cai et al., 2004).  Results from study samples of Japanese female dietitians (Tokudome et al., 2002), Finnish older middle-aged men (Hartman et al., 1990), as well as Brazilian adolescents (Verly Junior et al., 2010) also suggested that daily, weekly and monthly variations are not a significant source of variation in nutrient intake.  These findings suggest that variation in the amount of nutrients consumed over the course of a week is largely random and cannot be predicted for a specific day of week or month of year (Cai et al., 2004; Cai et al., 2005; Tokudome et al., 2002; Verly Junior et al., 2010).  The authors concluded that since the temporal variation was not found to be significant, dietary data can be collected randomly on any days of the week (Verly Junior et al., 2010).  Due to the lack of agreement in which days to include for measuring reliable dietary intake, a closer examination of the temporal variation in dietary pattern is therefore important for monitoring the diet of a population. 2.2 Evidence regarding differences in dietary intake on weekends versus weekdays Studies that analyzed the day-to-day variation in nutrient intake have reported that there is a potential difference in the dietary intake between weekdays and weekend days; however, the type of nutrients that differ and the magnitude of weekday-weekend differences remain inconsistent among different age groups.  The following sections discuss the current evidence related to variation in dietary intake between weekdays and weekend days in children and 6  adolescents, adults, and older adults.  The results from studies that used nationally representative datasets are also presented. 2.2.1  Weekday-weekend differences in nutrient intakes in children and adolescents The temporal pattern of dietary intake in children has been investigated in the U.S. (Cullen, Lara, & de Moor, 2002; Nicklas et al., 1997) and European countries (Macdiarmid et al., 2009; Post, Kemper, & Storm-Van Essen, 1987; Sepp, Lennernas, Pettersson, & Abrahamsson, 2001) and the results are inconsistent.  While most of the studies reported no differences in energy or macronutrient intakes between weekdays and weekend days (Macdiarmid et al., 2009; Nicklas et al., 1997; Rockell, Parnell, Wilson, Skidmore, & Regan, 2011; Sepp et al., 2001), some studies report  a difference in the type of food consumed between weekdays and weekend days (Nicklas et al., 1997; Rockell et al., 2011).  For example, Nicklas examined the dietary intake of 281 10-year-old children who completed a 24-hour recall and noted that the intakes from egg, pork, vegetables and poultry of these children were significantly higher on Sunday, with no difference in total intakes of energy and macronutrients between weekend days and weekdays (Nicklas et al., 1997).  A study in 109 Swedish preschool children whose dietary intake was assessed using a 7-day food record revealed that micronutrient intake was significantly higher on weekdays with higher nutrient density (amount of nutrient consumed divided by total caloric intake) for calcium, zinc, selenium, retinol, riboflavin, vitamin C and dietary fibre (Sepp et al., 2001).  Results from these studies are difficult to compare as each study focused on specific age groups, with an age range of 3 to 17 years old across different studies. The temporal difference in children’s dietary intake is therefore, inconclusive in the current literature. 7  2.3 Weekday-weekend differences in adults Energy intakes have previously been found to be higher on weekends, with a parallel increase in the absolute intakes of fat and alcohol in the adult population on weekends (Beaton et al., 1979; de Castro, 1991; Gibson, Gibson, & Kitching, 1985; Jackson et al., 2008; Jula et al., 1999; McGee et al., 1982; Tarasuk & Beaton, 1992).  Beaton et al. found a 24% increase in intake of energy on Sundays compared to Tuesdays and Wednesdays in women but not men in a study of 60 Canadian adults (Beaton et al., 1979; Beaton et al., 1983).  Similarly, a study of 14 Canadian female university students revealed that intakes of energy and most nutrients were significantly higher on weekends (Gibson et al., 1985).  de Castro also found a 8% increase in energy intake on Friday, Saturday and Sunday compared to the rest of week in 323 American adults, and that the increase in fat consumption on weekends contributed to 40% of the increase in energy intake (de Castro, 1991).  A higher consumption of alcohol by 84% to 167% on weekends was found to be responsible for the increase in total energy intake in 361 hypertensive Finnish adults (Jula et al., 1999), as well as in a group of Scottish middle-aged men (Thomson et al., 1988).  Exceptions were seen in 133 Irish men and women (Harrington, 2001), 18 male graduate students (Todd, Hudes, & Calloway, 1983), and 115 college students (Chalmers et al., 1952) where no weekday-weekend differences in energy or macronutrient intakes were found. 2.3.1 Weekday-weekend difference in older adults Research on the effect of weekday-weekend variation in dietary intake in older adults is sparse. Similar to the findings from the adult population, the total energy intake was found to be the highest on weekend days in 162 Finnish men aged 55-69 (Hartman et al., 1990), as well as in 138 elderly men and women in Norwich, England (Maisey et al., 1995).  The intakes of fat and protein were also found to be higher on the weekend, while no weekday-weekend difference in 8  intake of carbohydrate was observed (Hartman et al., 1990; Maisey et al., 1995).  The magnitude of the difference in energy intake appeared to be slightly smaller than those observed in the general adult population.  Compare to weekdays, weekend energy intake was higher by 2.4-4.2% (Hartman et al., 1990; Maisey et al., 1995) and weekend fat intake was higher by 3.8% (Hartman et al., 1990) in older adults. 2.3.2 Findings from nationally representative datasets A limited number of studies have analyzed the weekday-weekend difference in dietary intake using nationally-representative nutrition datasets from the U.S (Thompson, Larkin, & Brown, 1986; Haines et al., 2003) and New Zealand (Rockell et al., 2011).  A study by Thompson et al. used the Nationwide Food Consumption Survey (NFCS) collected by the United States Department of Agriculture (USDA) in 1977-78, which sampled 15,000 respondents aged 23 to 74 in the 48 contiguous states (Thompson, Larkin, & Brown, 1986).  Haines et al., on the other hand, analyzed the data from Continuing Survey of Food Intake by Individuals (CSFII 94- 96) collected by USDA in 1994 to 1996 that surveyed 28,156 U.S. residents 2 years and older using two independent 24-hour recalls (Haines et al., 2003).  Both Haines et al. and Thompson et al. reported that on average Americans consumed 82-140 kcal more on weekend days than on weekdays.  These differences are likely the result of higher intakes from fat, protein and alcohol on weekends (Haines et al., 2003; Thompson et al., 1986).  In contrast, Rockell et al. analyzed the New Zealand 2002 Children’s Nutrition Survey and reported no differences in intakes of energy, carbohydrate, protein and fat between weekdays and weekend days among children aged 5 to 15(Rockell et al., 2011). To summarize, most (but not all) studies in the adult population suggest that intakes of energy, fat, and alcohol are higher during weekends compare to weekdays.  On the other hand, 9  weekday-weekend differences in dietary intake in children and older adults are less well established due to the limited number of studies available. 2.4 The influence of weekend intake on diet quality Currently, it remains uncertain whether there is a difference in diet quality between weekdays and weekend days, or whether people simply eat more of the same foods on weekends versus weekdays.  A number of studies have indicated that weekday-weekend differences in intakes disappeared when nutrients were expressed as a proportion of energy intake (Beaton et al., 1979; de Castro, 1991; Gibson et al., 1985; McGee et al., 1982; Nicklas et al., 1997; Nicklas et al., 1997; Tarasuk & Beaton, 1992).  Based on these findings, the variation in nutrient intake is a result of an overall increase in the amount of food consumed, rather than a significant difference in the type or nutritional quality of food consumed during weekend days.  On the other hand, other studies have shown that the proportion of energy from fat, protein, and alcohol were significantly higher on the weekends, suggesting the quality (and quantity) of the diet may be different between weekdays and weekend days (Haines et al., 2003; Jula et al., 1999; O'Dwyer, McCarthy, Burke, & Gibney, 2005; Sepp et al., 2001). It was found that people tend to make less healthy food choices on weekend days, with higher energy intake from potatoes, meat, butter, cakes, biscuits and carbonated beverages (O'Dwyer et al., 2005).  Weekend days were also reported to be associated with intakes of higher-fat foods among a sample of 520 grade five to six American students (Cullen et al., 2002). Nevertheless, most studies focused solely on intakes of individual nutrients or food groups instead of the overall diet composition.  The temporal variation in diet quality that combines different nutrients, foods and dietary habits in relation to adherence to dietary guidelines and health outcomes has not yet been investigated in the literature. 10  2.5 Determinants of weekly variation and potential implications for nutrient intakes Little is currently known about the factors that contribute to the temporal variation in dietary choices.  A study by de Castro found that larger meal size on weekends was associated with an increase in the duration of the meals in response to fewer constraints from work or class schedules (de Castro, 1991).  Other studies reported that more frequent consumption of away- from-home meals during the weekend may be responsible for the increase in energy and fat intakes seen in the adult population (O'Dwyer et al., 2005; Thompson et al., 1986).  The limited findings suggest temporal variation in dietary practices is shaped by the different socio-cultural contexts between weekdays and weekend days.  It is clear that the relationship between various social factors and weekly variation in dietary patterns needs further investigation. Obtaining a clear understanding of the factors that influence dietary practices in free- living populations has many important implications including for weight management and public health strategies.  Racette et al. reported a weight gain of 0.08 kg during weekends, but not during weekdays in a group of middle-aged men and women as a consequence of the higher food intake during weekends (Racette et al., 2008).  Rhodes et al. also reported that the weekday- weekend difference in energy intake was larger in obese participants than in normal weight subjects (Rhodes, Cleveland, Murayi, & Moshfegh, 2006).  Given that the prevalence of obesity among Canadians almost doubled over the past two decades, with over one in four Canadian adults being obese (Shields, Carroll & Ogden. 2011), it is important to understand the impact of weekend dietary patterns on long-term weight control (Gorin, Phelan, Wing, & Hill, 2004; Racette et al., 2008). 11  2.6 Key gaps in current knowledge Based on the findings reviewed here, it is found that energy, fat, and alcohol intakes tend to be higher on weekends than weekdays, although the magnitude of difference remains inconsistent among people in different age and sex groups.  It is important to note that comparisons between studies are often difficult due to the lack of agreement on the definition of “weekday” and “weekend”.  Although weekdays are commonly thought of as referring to a typical North American work/school week which includes Monday to Friday, some studies have defined Fridays as a “weekend” day.  The justification for this classification is that dietary intake on Fridays may more closely resemble intake on Saturdays and Sundays than it does of Mondays to Thursdays (de Castro, 1991; Haines et al., 2003).  In some cases, studies have collected data about intake from only one day of the weekend (usually Sunday) to represent the ”typical” intake on weekends (Beaton et al., 1979; Jackson et al., 2008; Nicklas et al., 1997).  Moreover, interpretation of results from different studies is challenging due to the differences in the ways the data are analyzed and presented.  For example, intakes can be expressed as absolute intakes, percent of total energy intake from nutrients or nutrient density as calculated by dividing the weight of the nutrient consumed over total caloric intake which make the comparisons between studies much more difficult. More importantly, there is still a lack of analyses from nationally-representative Canadian samples, therefore it remains unclear how nutrient intakes are influenced by specific days of the week among Canadians, particularly in children and older adults.  Most of findings reviewed here were limited by small sample sizes that focused on a specific age or sex group, which may lack generalizability to the Canadian population.  It is clear that temporal analysis of current nutrition data from Canadian nutrition surveys is needed to provide a better understanding of the 12  weekly variations in nutrient intakes that are specific to the Canadian population.  In addition, the temporal variation in the intakes of micronutrients has rarely been investigated in the literature.  Little is known about the magnitude and direction of the change in micronutrient intakes across different days of the week.  Similarly, while most studies examined the temporal variation in the intakes of individual nutrients, change in dietary quality in relation to dietary recommendations and health outcomes has not yet been examined in the literature.  Lastly, there is still a gap in knowledge on how weekday-weekend differences in nutrient intakes are moderated by various socio-demographic factors that shape eating behaviour such as age, sex and working status. 2.7 Research objectives The main objective of this study is therefore to examine the potential differences in dietary intake between weekdays and weekend days among a nationally representative sample of Canadians. Three specific aims are: 1. To determine whether dietary intakes of energy and nutrients differ between weekdays and weekend days. 2. To assess the differences in dietary quality between weekdays and weekend days (based on national Canadian recommendations for dietary quality). 3. To explore potential factors that may moderate the differences in nutrient intake and dietary quality between weekdays and weekend days. 2.8 Study hypotheses 1. Energy and nutrient intakes will differ significantly between weekdays and weekend days. 2. Dietary quality will differ significantly between weekdays and weekend days. 13  3. Age, sex, working status and student status will moderate differences in nutrient intake and dietary quality between weekdays and weekend days. 14  Chapter  3: Methods 3.1 Data source This study  was conducted using data collected from the Canadian Community Health Survey cycle 2.2 (CCHS 2.2), a cross sectional health survey that collected information regarding health status and dietary information of the general Canadian population in 2004 (Health Canada, 2006).  The survey contains two main components: (1) a general health questionnaire, which elicited information on respondents’ health status such as chronic health conditions and health-related behaviour as well as socio-demographic characteristics, and (2) a 24-hour dietary recall module, which assessed detailed dietary information related to all foods and beverages consumed by respondents during the previous day’s 24 hours, from midnight to midnight. 3.1.1 CCHS 2.2. Survey data The current study required access to the confidential survey data files available at a Research Data Center (RDC). An application to gain access to these data was submitted to and approved by Statistics Canada (February, 2012).  Analyses were carried out at the British Columbia Inter-University RDC. The survey files used in the analysis included: 1. General health, vitamin and mineral supplements, and 24-hour dietary recall (HS) data file, which contained responses to the General Health questionnaire and the food records from the first 24-hour dietary recall. 2. Merged Canada food guide, food and ingredient details, and food recipe level (CFG_FID_FRL) data file which contained nutrient values for the food items, the serving size, and the food group classification according to the 1992 Canada’s Food Guide (CFG) (Health Canada, 1997) to reflect the food groups used when the survey was conducted in 2004.  The 15  1992 CFG food groups consisted of grain products, vegetables and fruits, milk products, meat and alternatives, and “other foods”, which included foods such as alcohol, soft drinks, sugary foods, and high-fat and/or high-salt snack foods.  Given the current 2007 CFG classification was not available at the time of analysis, this study used the food group classification and serving sizes based on the 1992 CFG definitions, which is the only classification that Health Canada makes available to CCHS 2.2 users.  For the most part, the definitions and serving size of the food groups were very si1milar between the 1992 and 2007 CFG.  The only minor changes made in the 2007 CGF classification were that milk products now include soy beverages and the recommended serving sizes of meat and alternatives changed from 50-100 g/serving to 75g/serving (Marchand & Lowell, 2007). 3.1.2 Target population CCHS 2.2 targeted Canadians of all age groups, aged 0 and older, living in private residences.  Persons living in the three territories, Indian reserves or Crown lands, institutions, some remote regions and full-time members of the Canadian Forces were excluded from the sampling frame. 3.1.3 Sampling procedure CCHS 2.2 used a multistage stratified cluster design in order to obtain a sample that is nationally representative in terms of age, sex, geography and socioeconomic status.  The final sample size was 35,107 respondents with an overall response rate of 76.5%.  A sample of households was selected from a variety of sampling frames including: 1. Canadian Labor Force Survey area frame 16  The primary sampling frame  in CCHS 2.2 was the Canadian Labor Force Survey area frame where the sample of household was selected using a multistage sampling scheme that consisted of three sampling strata: province level, city type (urban/rural), and socio-economic status, followed by cluster sampling (Statistics Canada, 2008).  In each stratum, six clusters, each containing 150 to 250 dwellings, were selected using random sampling with a probability proportion to size, where the size corresponds to the number of dwellings in a given cluster. Within each cluster, households were selected using systematic sampling. Finally, one respondent was selected per household using varying probabilities to ensure sufficient sample size for each age group and province. 2. A list frame of CCHS 2.1 dwellings In order to ensure minimum sample in each age/sex group was acheived, a secondary sampling frame was needed as it was challenging to obtain a sufficient number of households with young persons.  In response, the household information of respondents from CCHS 2.1 was used to generate a list of households in which there was at least one individual aged 18 and younger at the time of CCHS 2.1 data collection.  Similar to the sample procedure used in the area frame, the list was first stratified by province and urban/rural zone, followed by probability- proportional-to-size sampling (Statistics Canada, 2008). 3. The Prince Edward Island and Manitoba health care registries Extra funds were provided by Ontario, Manitoba, and Prince Edward Island (P.E.I.) to allow a large number of samples selected from these provinces.  In order to meet the provincial sample buy-ins objectives, especially in obtaining sufficient number of sample in younger age groups, the P.E.I and Manitoba health care registries were used as a 17  supplementary sampling frame to increase the probability of finding households with individuals aged 18 or less (Statistics Canada, 2008). 3.1.4 Dietary assessment procedure Dietary information was assessed using interviewer administered, computer-assisted 24- hour dietary recalls that recorded all foods and beverages consumed during the last 24 hours, from midnight to midnight, prior to the interview day (Statistics Canada, 2008).  Respondents were asked to provide information on the time the food was consumed, the eating occasion (e.g., breakfast, lunch, dinner, and snack), any additions to foods, detailed food descriptions, amounts consumed, and whether the meal was prepared at home or elsewhere.  The 24-hour recall was conducted by using the computer-assisted Automated Multiple-Pass Method (AMPM).  The AMPM is an automated system that guides the interview through a series of questions that help to minimize data collection errors and maximize respondents’ ability to call and reported food consume the previous day.  The AMPM was originally developed by the United States Department of Agriculture and was modified by Health Canada to reflect the Canadian food supply.  The AMPM has been validated against doubly-labeled water method of assessing energy expenditure in weight stable adults aged 30-69 years old and showed that the AMPM was able to assess mean energy intake within 11% of mean energy expenditure (Moshfegh et al., 2008). The dietary information of the respondents aged 0 to 5 years was provided by the parents. Children aged 6 to 11 years were asked to provide dietary information with the help of their parents.  No proxy interview was required for respondents aged 12 years and older.  The data were collected over a 12-month period to ensure that data represented intake from all seasons. All respondents completed a first 24-hour recall in person in the respondents’ home. 18  Approximately 30% of respondents completed a second 24-hour dietary recall over the phone 3 to 10 days after the first interview, which was used to adjust for within-person variability when estimating the distributions of usual nutrient intakes (Statistics Canada, 2008).  The nutritional content of the foods was estimated based on quantity of the foods reported by the respondents estimated based on the nutritional composition data provided by Health Canada’s Canadian Nutrient File 2001b, a database that contains nutrient values for foods commonly consumed in the Canadian food supply (Statistics Canada, 2008). 3.2 Analytical samples The sample used for analysis consisted of respondents aged 1 and older.  Pregnant and breastfeeding females were excluded as their nutrient requirements and dietary reference intake (DRI) thresholds are different from non-pregnant and non-breastfeeding women (Health Canada, 2006).  Respondents with invalid 24-hour dietary recalls due to either technical problems when capturing the amount of food or missing meal information were also excluded from the analysis. This analysis was based on the dietary information collected from the first 24-hour recalls only as the first recall was considered to be more reliable than second recalls.  Response bias may exit in second recall as respondents might attempt to eat “better” after the first interview (Verret, 2006). 3.3 Variables of interest 3.3.1 Independent variables The main independent variable was a dichotomous variable indicating whether the 24- hour recall pertained to a weekday or a weekend day.  According to the CCHS 2.2 user guide, weekdays were defined as Mondays to Thursdays and weekend days as Fridays to Sundays (Statistics Canada, 2008).  However, as discussed in the literature review section, there have 19  been some discrepancies in the literature on the definition of weekdays and weekend days. Given Friday can be regarded as either a weekday or a weekend day, the following weighted regression models were fitted for 30 nutrient intakes to determine whether Friday intake is more similar to Monday to Thursday (weekdays), or to that of Saturday and Sunday in the Canadian population: Unadjusted model: nutrient intake = β0 + β 1Sat-Sun + β 2Mon-Thur Energy-adjusted model: nutrient intake = β0 + β 1Sat-Sun + β 2Mon-Thur + β 3energy, where β0 denotes the intercept (Friday as reference group), and β 1, β 2 and β3 are the regression coefficients for Saturday-Sunday, weekday (Monday-Thursday) and energy intake. The results of these important preliminary analyses are presented in Appendix A. Energy intake on Saturdays and Sundays was not significantly different from Friday intake.  On the other hand, respondents reported consuming 81.1 kcal (SE = 37.7, p = 0.03) less on weekdays (Monday-Thursday) compared to Fridays.  In unadjusted model, intakes of carbohydrate, cholesterol, vitamin D and folic acid on Fridays were significantly different from Saturday and Sunday intake, while alcohol intake on Fridays was significantly higher than those on weekdays by 4.82 g (SE = 1.05, p < 0.01).  Other nutrients did not differ between Fridays and the other two weekend days. In energy-adjusted regression models, of the 30 nutrients examined as outcome variables, the intakes of 14 nutrients on weekdays (Monday-Thursday) were found to be significantly different from Fridays, while only six nutrients on Saturday and Sunday were found to be significantly different from Friday.  Some of the most noticeable differences were lower carbohydrate intake by -7.83 g (SE = 2.61, p < 0.01) and higher cholesterol intake by 41.4 mg (SE = 8.70, p < 0.01) on Saturday and Sunday compared to Friday.  On the other hand, lower 20  intake of alcohol by 4.30 g (SE = 1.00, p < 0.01), and higher intakes of calcium, sodium, and potassium by 46.9 mg (SE = 20.1, p = 0.02), 120 mg (SE = 42.9, p =0.01), and 172 mg (SE = 48.8, p < 0.01) respectively were identified on weekdays compared to Fridays. Based on these analyses, it was concluded that energy and nutrient intake reported on Fridays more closely resembled those of Saturday and Sunday, and dietary intakes remained similar between Friday, Saturday and Sunday after adjusting for energy intake.  Hence, weekdays were defined as Monday to Thursday for this study, while weekends included Fridays, Saturdays and Sundays.  Moreover, given there was an imbalance in the distribution of the number of respondents between weekdays (n = 23,489) and Saturday and Sunday (n = 7,835), including Friday (n = 3,078) as a weekend day helped to increase the sample size on weekend days (n = 10,913). 3.3.2 Outcome variables Calorie, macronutrient and micronutrient intake. As listed in Table 1, the main outcome variables included energy (in kcal), macronutrient and micronutrient intake. Macronutrients included carbohydrate (including dietary fibre and sugar), fat (including saturated fat, monounsaturated fat, polyunsaturated fat, linoleic fatty acid, linolenic fatty acid and cholesterol), protein and alcohol.  Micronutrients included vitamin A, vitamin D, vitamin C, thiamin, riboflavin, niacin, vitamin B6, vitamin B12, food folate, folic acid, total folate from food (folate and folic acid in dietary folate equivalents), calcium, phosphorus, magnesium, iron, zinc, sodium and potassium. Dietary quality. The dietary quality of respondents was assessed using the Canadian adaptation of the 2005 version of the Healthy Eating Index (HEI-2005) (Garriguet, 2009).  HEI-2005 was 21  Table 1 Nutrients of interest Nutrient of Interest Variable Name in CCHS2.2 Description Energy  FSDDDEKC Energy intake from food sources (kcal) Carbohydrate  FSDDDCAR Total carbohydrate intake from food sources (g) Dietary fibre FSDDDFI Total dietary fibre intake from food sources (g) Sugar FSDDDSUG Total sugars intake from food sources (g) Fat FSDDDFAT Total fat intake from food sources (g) Saturated fats FSDDDFAS Total saturated fatty acid intake from food sources (g) Monounsaturated fats FSDDDFAM Total monounsaturated fatty acid intake from food sources (g) Polyunsaturated fats FSDDDFAP Total polyunsaturated fatty acid intake from food sources (g) Linoleic fatty acid FSDDDFAL Linoleic fatty acid intake from food sources (g) Linolenic fatty acid FSDDDFAN Linolenic fatty acid intake from food sources (g) Cholesterol  FSDDDCHO Cholesterol intake from food sources in (mg) Protein FSDDDPRO Protein intake from food sources (g) Alcohol FSDDDALC Alcohol intake from food sources (g) Vitamin A FSDDDRAE Vitamin A intake from food sources in mcg of  retinol activity equivalents (RAE) Vitamin D FSDDDDMG Vitamin D intake from food sources (mcg) Vitamin C FSDDDC Vitamin C intake from food sources (mg) Thiamin FSDDDTHI Thiamin intake from food sources (mg) Riboflavin FSDDDRIB Riboflavin intake from food sources (mg) Niacin FSDDDNIA Niacin intake from food sources in mg of niacin equivalents (NE) Vitamin B6 FSDDDB6 Vitamin B6 intake from food sources (mg) Vitamin B12 FSDDDB12 Vitamin B12 intake from food sources (mcg) Food folate  FSDDDFON Naturally occurring folate intake from food sources (mcg) Folic acid FSDDDFOA Folic acid intake from food source fortified with folic acid (mcg) Total folate intake from food source FSDDDDFE Total folate intake (both naturally occurring folate and folic acid) from food sources in mcg of dietary folate equivalents (DFE) Calcium FSDDDCAL Calcium intake from food sources in (mg) Phosphorus FSDDDPHO Phosphorus intake from food sources in (mg) Magnesium FSDDDMAG Magnesium intake from food sources in (mg) Iron FSDDDIRO Iron intake from food sources in (mg) Zinc FSDDDZIN Zinc intake from food sources in (mg) Sodium FSDDDSOD Sodium intake from food sources in (mg) Potassium FSDDDPOT Potassium intake from food sources in (mg) 22  developed by the USDA to evaluate overall dietary quality in relation to national dietary recommendations (Guenther, Reedy, & Krebs-Smith, 2008).  The Canadian version of HEI-2005 was modified to reflect the dietary recommendations in the 2007 CFG which include 11 components, with a total maximum score of 100 (Garriguet, 2009).  Table 2 shows the scoring criteria for each of the components.  The HEI of each respondent was calculated based on the methods described by Garriguet (Garriguet, 2009).  Briefly, HEI reflects two aspects of a healthy diet: Food group components measured how well a person’s diet on a specific day followed the recommended servings for total vegetables and fruit, whole fruit, dark green and orange vegetables, total grain products, whole grain products, milk and alternatives, and meat and alternatives.  Examples of the type of food items included in each category are listed in Table 2. To compute the HEI scores for each food group component, the number of servings of items from each category was first calculated for all food items the respondent had consumed on a given day.  To do so, the reported food items were first assigned to their corresponding food groups using the supplemental data file in the CCHS 2.2 that coded food items according to the 1992 CFG food groups (Health Canada, 1997).  1992 CFG classifies foods into four groups: vegetables and fruit, milk products, meat and alternatives such as eggs and legumes, and grain products such as bread and cereals.  Since there were no variable definitions available in the supplemental data file for whole fruit, dark green and orange vegetables, and whole grain products at the time of analysis, these three food subgroups were manually coded based on the definition of the food groups by Health Canada (Health Canada, 2007b; Health Canada, 2007c). The number of food group servings was then calculated by dividing the weight of each recorded food item or ingredient by its portion size defined by the 1992 CFG (Health Canada, 1997).  For 23  Table 2 Scoring criteria of Canadian adaptation of healthy eating index-2005 Component Range of Scores Maximum Score Criteria Minimum Score Criteria Examples of Food Items that Contribute to this Component’s Score Total vegetables and fruit 0 to 10 4-10 servings a 0 serving Cabbage, banana, avocado, celery Whole fruit 0 to 5 0.8-2.1 servings a  0 serving Apple, grapefruit, kiwi, orange (excluding fruit juice) Dark green and orange vegetables 0 to 5 0.8-2.1 servings a  0 serving Broccoli, asparagus, carrots, pumpkin Total grain products 0 to 5 3-8 servings a  0 serving Bread, rice, cereal, pasta Whole grains 0 to 5 1.5-4 servings 0 serving Whole grain cereal, quinoa, rye crackers, wild rice Milk and alternatives 0 to 10 2-4 servings a  0 serving Milk, cheese, yogurt, fortified soy beverage Meat and alternatives 0 to 10 1-3 servings a  0 serving Beef, eggs, fish, tofu, legumes Unsaturated fats 0 to 10 30-45 grams  0 serving N/A Saturated fats 8 to 10 ≤7% of total energy intake 10% of total energy intake N/A 0 to 8 10% of total energy intake 15% of total energy intake Sodium 8 to 10 Adequate intake Tolerable upper intake level N/A 0 to 8 Tolerable upper intake level Twice tolerable upper intake level Other food 0 to 20 ≤5% of total energy ≥40% of total energy  Alcohol, soft drinks, jam, chips, herbs Note. a scores differ according to age and sex, as specified in 2007 Canada’s Food Guide (Health Canada, 2007a) (see Appendix B) Table adapted from Garriguet, 2007 24  example, one slice of bread is equivalent to one serving of grain products, so if a respondent reported consuming two slices of bread, he/she was estimated to consume two servings of grain products.  Scoring was assigned proportionally based on the adherence to the recommended servings, according to the respondents’ age and sex groups, as specified in 2007 CFG (Health Canada, 2007a).  Appendix B shows the recommended number of food guide servings according to 2007 CFG by age and sex groups.  For example, the recommended number of total vegetables and fruit servings is eight for teen males aged 14-18.  If a respondent in this age and sex category reported consuming three servings of total vegetables and fruits, he got a score of 3.75 out of ten (three servings consumed/ eight recommended servings * ten possible points).  On the other hand, if his/her intake was equal to or more than eight servings of total vegetables and fruits, which is the recommended servings for his age and sex, he scored a ten out of ten for this component. Dietary guideline components measured a person’s saturated and unsaturated fat consumption, the calories from the “other foods” category as a percentage of total energy intake, and sodium intake relative to adequate intake levels.  The other foods category was defined based on the 1992 CFG food group classification that included foods such as alcohol, soft drinks, sugary foods, and high-fat and/or high-salt snack foods.  The amount of unsaturated fat intake in grams was calculated by summing the intake of poly- and monounsaturated fats consumed on the same day.  Scoring was assigned proportionally based on the adherence to the recommended intake (Appendix B).  For example, if a 30-year old female respondent reported consuming 30 grams of unsaturated fat, she received a score of ten out of ten.  On the other hand, if her unsaturated fat intake was 15 grams, her score would be five out of ten since her intake was only half of the recommended intake.  In contrast to the scoring criteria for unsaturated fat component, the respondent scored higher points if he/she limited his/her energy consumption from the 25  saturated fats, other foods and sodium intake (Table 2).  The score for sodium intake was assigned based on Institute of Medicine recommendations (Institute of Medicine, 2006), where a maximum score of ten was assigned if the sodium intake was equal to or below the adequate intake (1-3 year: 1000 mg/day; 4-8 year: 1200 mg/day; 9-70+ year: 1500mg/day), eight points for consuming an intake equal to the tolerable upper intake level (1-3years: 1500 mg/day; 4-8 years: 1900 mg/day; 9-70+ years: 2300 mg/day), and zero points for intake exceeding twice the tolerable upper intake level.  Similarly, a maximum score of ten was assigned for the saturated fat component if the intake was less than 7% of total energy, and zero points for consuming more than 15% of total energy.  For the other foods component, the percent of energy from the “other foods” group was derived by dividing energy from other foods by total energy intake.  A maximum score of 20 was assigned if the intake of other foods was five percent or less of total energy intake, while zero points were assigned for intake exceeding 40% of total energy. 3.3.3 Interaction terms A moderating variable is one that modifies the relationship between predictor variables and outcomes (Vittinghoff, Glidden, Shiboski, & McCulloch, 2005).  In this study, we were interested in exploring whether the effect of weekday/weekend on dietary intake was moderated by respondents’ age, sex, or employment/student status.  These variables were selected because limited studies with inconsistent findings suggested that magnitude of weekday-weekend difference in nutrient intake may be different among people in different age and sex groups, or based on whether they are currently working/attending school or not. Age was classified into four categories according to life cycle groupings defined by Statistics Canada: children aged from 1-14 years, youths from 15-24 years, adults from 25-64 years, and older adults aged 65 and over (Statistics Canada, 2012).  Sex was classified as male or 26  female.  Employment/student status was classified into the following categories: working, which was defined as having either a full-time or part-time job and was at their work in the week prior to the interview; attending school, college or university as full-time students; retiree, which included respondents who reported retirement as their main reason for not working in the week prior to the interview; and others, which included those who were currently not a full time student, had a job but were not at their job in the week prior to the interview, or did not have a job due to reasons other than retirement. To assess these potential moderating effects, interaction terms were constructed. Interaction terms were defined as products of the moderating variables (sex, age groups, or employment status) and the independent variable (weekday versus weekend).  As a result, the following interaction terms were generated: Sex group: male*weekend, with the reference group being the dietary intake in females on weekdays. Age group: child*weekend, youth*weekend, and adult*weekend, with the reference group being the dietary intake in older adults on weekdays. Employment/student status: student*weekend, working*weekend and retiree* weekend, with the reference group being the dietary intake on weekdays in others (i.e. persons who were not a full time student, had a job but were not at job in the week prior to the interview, or did not have a job due to reasons other than retirement). 3.3.4 Variables for sample characteristics Age, sex, household education, income and food security status, residence, use of vitamin and mineral supplements, immigrant status and working/student status were used to describe characteristics of the sample.  Household education referred to the highest level of education 27  acquired by any member of the household and was categorized as following: less than secondary school, secondary school graduation, some post-secondary education, post-secondary graduate. Income adequacy was consisted of 4 income groups based on household income and total number of people living in the household: lowest ( <$15,000 for one to two people, <$20,000 for three to four people, or <$30,000 for five or more people), lower middle ($15,000 to $29,999 for one to two people,  $20,000 to $39,999 for three to four people, or $30,000 to $59,999 for five or more people), upper middle ($30,000 to $59,999 for one to two people, $40,000 to $79,999 for three to four people, or $60,000 to $79,999 for five or more people), and highest (>$60,000 for one to two person, $>$80,000 for three or more people).  Household food security status, which asked the respondents whether they were able to afford the food they needed in the previous 12 months, was categorized into food secure, food insecure without hunger, food insecure with moderate hunger, and food insecure with severe hunger.  Residence of respondents was categorized into either urban or rural.  Use of vitamins and mineral supplements referred to whether the respondents had taken any vitamins or minerals in the past month prior to the interview (yes/no).  Lastly, immigrant status referred to whether the respondent was an immigrant or not. 3.4 Statistical analysis Statistical analyses were performed using Stata version 10 statistical software (LP Stata Corp, TX, USA).  Statistical significance was defined as having a p-value less than 0.05. 3.4.1 Descriptive statistics Means and proportions were calculated for continuous and categorical variables respectively to describe the sample characteristics.  Age, sex, household education, income and 28  Figure 1 Flow chart of the analytical strategy  Note. β0 denotes the intercept, and  β 1 and β 2  are the regression coefficients for weekend and energy intake. Step 6 Testing the interaction effects of sex, age groups, and employment/student status in both unadjusted and energy adjusted models. Step 5 Examining the difference in total HEI and HEI component scores between weekdays and weekend days in energy adjusted models. Model 5: HEI = β0 + β1weekend + β2energy Step 4 Examining the difference in total HEI and HEI component scores between weekdays and weekend days in unadjusted models. Model 4: HEI = β0 + β1weekend Step 3 Examining the difference in nutrient intake between weekdays and weekend days in energy-adjusted models. Model 3: Nutrient intake = β0 + β1weekend + β2energy Step 2 Examining the difference in nutrient intake between weekdays and weekend days in unadjusted models. Model 2: Nutrient intake = β0 + β1weekend Step 1 Comparing energy intake on weekend days and weekdays. Model 1: Energy intake = β0 + β1weekend 29  food security status, residence (urban/rural), use of vitamin and mineral supplements, immigrant status and working/student status were used to describe characteristics of the sample.  In order to confirm that sample characteristics did not differ substantially between respondents who completed 24 hour recalls on weekends versus weekdays, student’s independent sample t-tests (for means) and chi-square tests (for frequencies) were carried out. 3.4.2 Assessing associations between weekday-weekend intake and dietary outcomes Given that the outcomes of interest were continuous variables (e.g. energy and nutrient intakes), a linear regression approach was used to examine the differences in dietary intakes and diet quality between weekdays and weekend days.  Type of day (weekday or weekend) was included as a dummy variable, with weekday intake as the reference group.  The coefficients of the regression were estimated using an ordinary least-squares approach.  As depicted in Figure 1 the analyses were conducted in the following steps: Step 1 Comparing energy intake on weekend days and weekdays. Unadjusted linear regression models were first carried out as a preliminary analysis to determine if energy intake was different between weekdays and weekend days. The regression equation was as following: Model 1: Energy intake = β0 + β 1weekend, where β0 denotes the intercept, and β1 is the regression coefficient for weekend. This was done to evaluate whether caloric intake differs on weekdays versus weekend days. It was found that respondents reported consuming 62 kcal (SE = 23, p <0.01) more on weekend days than on weekdays.  Energy intake was therefore taken into consideration as a control variable in the regression models to ensure that differences in nutrient intake between weekdays and weekend days were independent of the amount of food consumed.  Both 30  unadjusted and energy-adjusted regressions testing the association between type of day and dietary intake were then carried out. Step 2 Examining the differences in nutrient intake between weekdays and weekend days in unadjusted models. To test the hypothesis that nutrient intake differs significantly between weekdays and weekend days without controlling for the amount of energy/food consumed, the following unadjusted linear models were fitted for each of the 30 nutrient outcomes (as listed in Table 1): Model 2: Nutrient intake (e.g. carbohydrate) = β0 + β1weekend Step 3 Examining the differences in nutrient intake between weekdays and weekend days in energy-adjusted models. To test the hypothesis that nutrient intake differs significantly between weekdays and weekend days while controlling the difference in energy intake, the following energy-adjusted model was fitted for each of the 30 nutrient outcomes: Model 3: Nutrient intake (e.g. carbohydrate) = β0 + β1weekend + β2energy This step was carried out to examine whether weekday-weekend differences in nutrient intake remained significant after accounting for the difference in the amount of food consumed between weekdays and weekend days. Since nutrient intakes were measured in different units (e.g. carbohydrate in grams vs. vitamin C in milligrams), a unit of comparison was needed to compare the magnitude of weekday-weekend difference across 30 nutrient outcomes in a meaningful way.  To do this, percent of change on weekend days relative to weekdays was calculated by dividing weekday- weekend differences by weekday intake (reference group). 31  Step 4 Examining the differences in total HEI and HEI component scores between weekdays and weekend days in unadjusted models. To test the hypothesis that dietary quality differs significantly between weekdays and weekend days, the following unadjusted models were fitted for each of the 11 HEI component scores and total HEI: Model 4: HEI = β0 + β1weekend Step 5 Examining the difference in total HEI and HEI component scores between weekdays and weekend days in energy adjusted models Although the American HEI was designed to be independent of energy intake, it was found that there was a correlation between energy intake and the Canadian adaptation of HEI score in CCHS 2.2, likely due to under-reporting (Garriguet, 2007).  To account for this, energy- adjusted regression models were fitted to test the effect of weekday/weekend on HEI scores: Model 5: HEI = β0 + β1weekend + β2energy 3.4.3 Assessing moderating effects of age, sex, and work/school status To assess whether moderator variables were statistically significant predictors of the outcome, a linear regression with sex, age, or employment/student status moderator variables was first fitted, followed by Wald tests.  The Wald test determines if adding an interaction term significantly improves the overall fit of the model compared to models with weekend variables alone (unadjusted models), or weekend and energy intake variables (energy-adjusted models) (Hardy, 1993).  For example, a model with a sex*weekend interaction term would look like: Carbohydrate = β0 + β1 weekend + β2 sex + β3 weekend*sex where b0 denotes the intercept, and  β1, β2, β3 are the regression coefficients for weekend, sex, and their interaction  terms (weekend*sex), respectively.  The Wald test was used to test the 32  hypothesis that the coefficient of the moderator equals to zero (e.g. β3 = 0).  If the p-value of the Wald test was less than 0.05, the null hypothesis was rejected, suggesting there was evidence that the effect of weekday/weekend days on dietary outcomes may be modified by the moderator variable (Hardy, 1993).  In other words, the magnitude of weekday-weekend difference in energy intake may depend on whether respondents are men or women.  On the other hand, if the Wald test was not significant, and the individual p-value for the interaction term was also not significant, it suggested that the difference in energy intake between weekdays and weekend days was similar in both men and women. The value of adding sex, age, and work/student status as interaction terms in unadjusted and energy-adjusted models for improving model fit was assessed for each of the 42 models estimating nutrient and HEI outcomes.  The models with interaction terms were constructed as follows: 1) Sex group: Nutrient intake/HEI = β0 + β1 weekend + β2 male + β3 weekend*male Nutrient intake/HEI = β0 + β1 weekend + β2 male + β3 weekend*male + β4energy The reference group was the dietary intake in females on weekdays. 2) Age group: Nutrient intake/HEI = β0 + β1 weekend + β2 child + β3youth + β4adult + β5weekend*child+ β6weekend*youth + β7weekend*adult Nutrient intake/HEI = β0 + β1 weekend + β2 child + β3youth + β4adult + β5weekend*child+ β6weekend*youth + β7weekend*adult + β8energy The reference group was the dietary intake in older adults on weekdays 3) Employment/student status: 33  Nutrient intake/HEI = β0 + β1 weekend + β2 student + β3working + β4retiree + β5weekend*student+ β6weekend*working + β7weekend*retiree Nutrient intake/HEI = β0 + β1 weekend + β2 student + β3working + β4retiree + β5weekend*student+ β6weekend*working + β7weekend*retiree + β8energy The reference group was the dietary intake on weekdays in other respondents who were not full time students, were not at job in the week prior to the interview, or did not have a job due to reasons other than retirement. In models where the interaction term was found to be a significant predictor (p < 0.05 from the Wald test), the mean intake of the dietary outcomes on weekdays and weekend days were estimated based on the coefficients of the regressions. 3.4.4 Survey weighting and bootstrapping In order to obtain unbiased estimates of the population, survey weights provided by Statistic Canada were incorporated into all stages of analysis.  The principle of weighting was that each respondent in the sample represents several other persons in the target population (Statistics Canada, 2008).  Weighting can be regarded as an adjustment for complex sampling design and nonresponse bias of the survey to ensure the estimates are representative of the targeted population (Statistics Canada, 2008). Due to the complex sampling design in CCHS 2.2, as recommended by Statistics Canada, a bootstrap balanced repeated replication technique was used to estimate confidence intervals and standard errors for all estimates.  Bootstrapping is a nonparametric approach that estimates the sampling distribution by randomly selecting a large number of resamples from the original sample with replacement (Mooney & Duval, 1993).  Bootstrap weights that account for the 34  sampling design were provided by Statistics Canada.  The confidence intervals and standard errors of the estimates were calculated based on 500 bootstrap replicates.  The estimates at population level were obtained at this stage.  35  Chapter  4: Results 4.1 Analytical sample characteristics After excluding pregnant and breastfeeding women, infants under the age of 1 and respondents with invalid 24 hour recalls, a total of 34,402 respondents were included in the analysis.  There were 23,489 respondents whose dietary data was reported for weekdays and 10,913 for weekend days.  All results presented are at the population level. Table 3 summarizes the sample characteristics of the respondents comparing respondents whose dietary data was reported for weekdays versus weekend days.  Overall, the demographic characteristics of respondents who reported weekend intake were very similar to those with weekday intake.  No significant differences in sex, household highest level of education, food security status, income adequacy, urban/rural residency or supplement use and immigrant status was observed between the weekday and weekend respondents.  The mean age of the respondents in the weekday group was 39.1 years (SE = 0.2) and 36.9 years (SE = 0.2) in the weekend group. The respondents in the weekday group were significantly (p < 0.01) older than those in the weekend group by 2.2 years, which likely reflected the slightly but significantly higher proportion of  retirees in the weekday group than in the weekend group (11% vs. 8%).  Such a difference in age between the weekday and weekend groups was not surprising since respondents who were constrained by school or work schedules would be more likely to have their survey interviews done on weekends than those of the retirees who may have higher time availability to complete surveys on weekdays.  There is no theoretical reason to believe that dietary intake differs systematically or meaningfully with a 2-year age difference between weekday and weekend groups or that this minor difference would affect the interpretation or generalizability of other analyses.  Therefore age was not included as a control variable in final models 36  Table 3 Analytical sample characteristics  Weekday Weekend p-value a  Sample size (n) 23,489 10,913 N/A  Age (y) (SE) 39.1 (0.2) 36.9 (0.2) <0.01 Age group (%)   <0.01 Children 1-14 year 17.6 19.0 Youths 15-24 year 11.1 13.0* Adults 25-64 year 57.0 56.8 Older adults 65+ year 14.4 11.4*  Males (%) 49.6 51.2 0.19  Household highest level of education (%)   0.10 Less than secondary school education 9.8 8.4 Secondary school education 11.4 10.9 Some post-secondary education 7.1 6.8 Post-secondary graduate 69.7 71.7 Not stated 2.0 2.2  Household food security status (%) Food secure 93.0 92.2 0.27 Food insecure without hunger 4.4 4.4 Food insecure with moderate hunger 1.7 2.2 Food insecure with severe hunger 0.3 0.5 Not stated  0.6 0.7  Residency (%)   0.13 Urban 81.6 82.9 Rural 18.4 17.1  Income adequacy (%) Lowest  9.0 8.3 0.31 Lower middle 18.8 19.4 Upper middle  31.4 31.5 Highest 30.5 31.6 Not stated 10.3 9.2  Vitamin or mineral supplements used (%) Yes 40.4 39.2 0.22 No 59.6 60.8  Working Status (%) Full time student 15.7 18.1* <0.01 Working   48.3 53.0 Retiree 10.7 8.4* Other b  25.2 20.5  Immigrants (%)   0.72 Yes 18.9 20.7 No 80.1 79.1 Note. n=34,402 excluding pregnant and breast feeding women. a p-values were derived from chi-square for categorical variables and Student independent sample t-test for continuous variables. b  included those who were not a full time student, had a job but were not at their job in the week prior to the interview, or did not have a job due to reasons other than retirement. *significant difference (p<0.05) between weekdays and weekend days. 37  predicting dietary intake. 4.2 Mean nutrient intakes between weekdays and weekends The mean caloric intake for the total analytical sample aged 1 and older was 2069 kcal (SE = 14) on weekdays and 2130 kcal (SE = 18) on weekend days.  Compared to weekdays, caloric intake on weekend was significantly higher by 62 kcal (SE = 23).  Table 4 presents the mean nutrient intakes for the total analytical sample aged 1 and older on weekdays and weekend days and their linear regression coefficients in unadjusted and energy-adjusted models.  A linear regression analysis revealed that there was a significantly higher intake of fat and alcohol on weekdays than on weekend days in both unadjusted and energy adjusted models.  There was also a significant increase in the intakes of different types of fats on weekend days compared to those on weekdays, including saturated fat, monounsaturated fatty acid (MUFA), polyunsaturated fatty acid (PUFA), linoleic fatty acid, and cholesterol. MUFA and cholesterol intakes remained significantly higher on weekend days, while saturated fat, PUFA and linoleic acid were no longer significantly different after energy adjustment. No significant differences in the intakes of carbohydrate or protein were found between weekdays and weekend days in the unadjusted models.  After adjusting for energy intake, the intakes of carbohydrate and protein were significantly lower on weekend days than on weekdays. Similarly, the intakes of most micronutrients did not differ significantly between weekday days and weekend in the unadjusted models.  Once the difference in energy intake between weekdays and weekend days was accounted for, the intakes of vitamin D and C, B vitamins, folic acid, total folate, calcium, phosphorus, magnesium, iron, zinc, sodium and potassium were significantly lower on weekend days compared to weekdays. 38  Table 4 Linear regression analysis comparing the intake of nutrients of Canadians aged 1 and older on weekend days with weekday intake: unadjusted and energy-adjusted models.   Unadjusted Models Energy-adjusted Models Nutrient Mean intake βweekend p-value R 2  Mean intake βweekend p-value R 2  Weekday (SE) Weekend (SE) (95%CI)   Weekday (SE) Weekend (SE) (95%CI)   Carbohydrate (g) 264 (1.8) 263 (2.4) -1.64 0.59 0.000 267 (0.9) 259 (1.2) -8.72 <0.01 0.75   (-7.65, 4.37)     (-11.7, -5.76) Protein (g) 83.7 (0.6) 84.3 (0.9) 0.59 0.59 0.000 84.7 (0.4) 83.0 (0.6) -1.64 0.02 0.56   (-1.58, 2.77)     (-3.06, -0.22) Alcohol (g) 6.1 (0.2) 10.6 (0.5) 4.51 <0.01 0.008 6.3 (0.2) 10.4 (0.5) 4.11 <0.01 0.08   (3.41, 5.61)     (3.07, 5.16) Fat (g) 74.3 (0.7) 77.9 (0.9) 3.64 <0.01 0.002 75.3 (0.4) 76.5 (0.4) 1.14 0.05 0.74   (1.43, 5.85)     (0.02, 2.25) Saturated FA (g) 24.6 (0.3) 25.7 (0.3) 1.02 0.02 0.001 25.0 (0.2) 25.2 (0.2) 0.21 0.46 0.59   (0.20, 1.84)     (-0.33, 0.75) MUFA (g) 29.4 (0.3) 31.3 (0.4) 1.89 <0.01 0.002 29.9 (0.2) 30.7 (0.2) 0.85 <0.01 0.67   (0.92, 2.85)     (0.30, 1.40) PUFA (g) 13.0 (0.1) 13.5 (0.2) 0.55 0.02 0.001 13.2 (0.1) 13.3 (0.1) 0.11 0.5 0.49   (0.07, 1.03)     (-0.20, 0.42) Linoleic FFA  (g) 10.4 (0.1) 10.9 (0.2) 0.44 0.03 0.001 10.6 (0.1) 10.7 (0.1) 0.073 0.59 0.47   (0.04, 0.83)     (-0.19, 0.34) Linolenic FFA (g) 1.8 (0.04) 1.8 (0.03) 0.02 0.71 0.000 1.9 (0.03) 1.8 (0.03) -0.042 0.34 0.24   (-0.08, 0.12)     (-0.13, 0.05) Cholesterol (mg) 257 (3.3) 289 (4.7) 32.7 <0.01 0.005 260 (2.8) 285 (4.3) 25.6 <0.01 0.25   (22.3, 43.1)     (16.3, 34.8) Dietary fibre (g) 17.0 (0.2) 16.2 (0.2) -0.82 <0.01 0.002 17.1 (0.1) 16.0 (0.1) -1.18 <0.01 0.32   (-1.32, -0.31)     (-1.57, -0.78)   39    Unadjusted Models Energy-adjusted Models Nutrient Mean intake βweekend p-value R 2  Mean intake βweekend p-value R 2  Weekday (SE) Weekend (SE) (95%CI)   Weekday (SE) Weekend (SE) (95%CI)   Sugar (g) 110 (1.0) 109 (1.3) -0.88 0.59 0.000 112 (0.8) 108 (0.9) -3.84 <0.01 0.41   (-4.08, 2.33)     (-6.22, -1.46) Vitamin A (mcg in RAE) 686 (9.3) 672 (22.7) -13.7 0.57 0.01 691 (8.9) 664 (22.2) -27.1 0.25 0.04   (-61.5, 34.1)     (-73.5, 19.3) Vitamin D (mcg) 5.9 (0.1) 5.7 (0.2) -0.21 0.27 0.000 5.9 (0.1) 5.6 (0.1) -0.36 0.05 0.09   (-0.59, 0.16)     (-0.71, 0.004) Vitamin C (mg) 133 (1.6) 129 (2.2) -3.86 0.15 0.000 134 (1.6) 128 (2.0) -6.21 0.02 0.09   (-9.16, 1.43)     (-11.3, -1.17) Thiamin (mg) 1.8 (0.02) 1.7 (0.02) -0.06 0.01 0.001 1.8 (0.01) 1.7 (0.01) -0.1 <0.01 0.42   (-0.01, -0.01)     (-0.14, -0.07) Riboflavin (mg) 2.0 (0.01) 2.0 (0.02) 0 0.86 0.000 2.0 (0.01) 1.9 (0.01) -0.05 <0.01 0.44   (-0.05, 0.04)     (-0.09, -0.02) Niacin (mg in NE) 38.6 (0.3) 38.8 (0.4) 0.2 0.68 0.000 39.0 (0.2) 38.3 (0.3) -0.79 0.02 0.54   (-0.78, 1.19)     (-1.46, -0.12) Vitamin B6 (mg) 1.8 (0.02) 1.8 (0.02) -0.01 0.57 0.000 1.9 (0.01) 1.8 (0.02) -0.06 <0.01 0.4   (-0.06, 0.04)     (-0.09, -0.02) Vitamin B12 (mcg) 4.3 (0.1) 4.3 (0.1) -0.04 0.77 0.000 4.4 (0.1) 4.2 (0.1) -0.16 0.25 0.05   (-0.31, 0.23)     (-0.42, 0.11) Food folate (mcg) 225 (2.3) 226 (2.9) 0.82 0.83 0.000 229 (2.1) 223 (2.5) -3.77 0.26 0.24   (-6.63, 8.28)     (-10.4, 2.82) Folic Acid (mcg) 126 (1.8) 121 (2.3) -4.91 0.09 0.000 128 (1.7) 119 (1.9) -8.41 <0.01 0.24   (-10.5, 0.67)     ( -13.3, -3.52) Folate from Food (mcg in DFE) 455 (3.8) 453 (5.2) -2.31 0.72 0.000 460 (3.0) 447 (3.8) -13.3 <0.01 0.44   (-15.1, 10.5)     (-23.1, -3.55)   40    Unadjusted Models Energy-adjusted Models Nutrient Mean intake βweekend p-value R 2  Mean intake βweekend p-value R 2  Weekday (SE) Weekend (SE) (95%CI)   Weekday (SE) Weekend (SE) (95%CI)   Calcium (mg) 921 (7.7) 896 (11.8) -25.5 0.07 0.000 931 (6.0) 883 (9.1) -47.6 <0.01 0.33   (-53.1, 2.14)   (-68.7, -26.5) Phosphorus (mg) 1350 (9.8) 1350 (13.4) -2.66 0.88 0.000 1363 (5.7) 1330 (7.7) -37.4 <0.01 0.61   (-35.8, 30.5)     (-55.9, -19.0) Magnesium (mg) 321 (2.2) 313 (2.8) -8.73 0.02 0.001 324 (1.7) 308 (1.8) -15.8 <0.01 0.51   (-16.1, -1.31)     (-20.9, -10.7) Iron (mg) 14.3 (0.1) 13.9(0.1) -0.36 0.03 0.001 14.4 (0.1) 13.7 (0.1) -0.72 <0.01 0.54   (-0.69, -0.04)     (-0.94, -0.50) Zinc (mg) 11.2 (0.1) 11.1 (0.1) -0.03 0.86 0.000 11.3 (0.1) 11.0 (0.1) -0.33 <0.01 0.43   (-0.33, 0.28)     (-0.56, -0.10) Sodium (mg) 3080 (25.9) 3090 (36.6) 13.6 0.76 0.000 3110 (18.3) 3050 (24.7) -67.9 0.02 0.50   (-74.6, 102)     (-127, 9.20) Potassium (mg) 3070 (20.5) 3020 (26.1) -56.3 0.1 0.000 3100 (13.5) 2980 (18.9) -123 <0.01 0.52   (-123, 10.4)     (-169, 76.2) Note. n=34,402 excluding pregnant and breast feeding women. βweekend=linear regression coefficient for weekend (reference group=weekday). 95% CI= 95% confidence interval.R 2 =Coefficient of determination. 41  The coefficients of determination (R 2 ) were less than 0.01 for models predicting all nutrient intake outcomes in the unadjusted models, suggesting that the type of day (weekday vs. weekend) accounted for less than 1% of the variation in nutrient intake.  Adding energy intake to the model increased the R-squared value to roughly 0.5 in most of the nutrient intake models. However, the R-squared for some nutrient outcome models, such as alcohol (R 2  = 0.08) and vitamin A (R 2  = 0.04) remained relatively low even after adjusting for energy intake. 4.3 Percent of change in nutrient intake on weekends compared to weekdays Figure 2 and Figure 3 show the percent of change in intakes of macronutrients and micronutrients respectively on weekend days relative to the intakes on weekdays.  Of the 30 nutrients examined, alcohol exhibited the greatest change in intake between weekdays and weekend days.  Alcohol intake on weekend days was 74% (SE = 8.5, p<0.01) and 66% (SE = 9.2, p < 0.01) higher than those on weekdays in unadjusted and energy-adjusted models respectively (Figure 2).  There was also a noticeably higher intake of cholesterol on weekends by 13% (SE = 2.1, p < 0.01) in the unadjusted model and 10% (SE = 1.8, p < 0.01) in the energy- adjusted model. Other than alcohol and cholesterol, the magnitude of difference between weekdays and weekend days was less than 10% for most of the macronutrients (Figure 2).  Energy adjusted intakes of carbohydrate and protein were significantly lower by 3.3% (SE = 0.6, p < 0.01) and 1.9% (SE = 0.9, p = 0.02), respectively on weekend days than on weekdays.  Of the total carbohydrate intake, the percent of change in dietary fibre was -6.9% (SE = 1.2, p < 0.01) and in sugar was -3.4% (SE = 1.1, p < 0.01).  On the other hand, there was a small but significantly higher intake of energy- adjusted fat intake by 1.5% on weekend days (SE = 0.8, p =0.05). 42  Figure 2  Unadjusted and energy-adjusted percent of change in the intakes of macronutrients, dietary fibre, sugars and different types of fats of Canadians aged 1 and older on weekend days relative to weekdays  -20.0 0.0 20.0 40.0 60.0 80.0 100.0 Carbohydrate Fats Protein Alcohol Dietary fibre Sugars Saturated fats MUFA PUFA Linoleic FFA Linolenic FFA Cholesterol % of difference on weekend relative to weekday Unadjusted Energy-adjusted bar = % of change in intake on weekend relative to weekday  error bars = 95% CI  * p  < 0.05 **p < 0.01  ** * * * ** ** ** ** * * ** ** * * ** ** 43  Figure 3 Unadjusted and energy-adjusted percent of difference in the intakes of micronutrients of Canadians aged 1 and older on weekend days relative to weekdays -15.0 -10.0 -5.0 0.0 5.0 10.0 Vitamin A Vitamin D Vitamin C Thiamin Riboflavin Niacin Vitamin B6 Vitamin B12 Food folate Folic acid Total folate Calcium Phosphorus Magnesium Iron Zinc Sodium Potassium % of difference on weekend relative to weekday Unadjusted Energy-adjusted bar = % of change in intake on weekend relative to weekday  error bars = 95% CI  * p  < 0.05 **p < 0.01  * * * * ** * ** ** ** ** ** ** * ** ** ** * ** 44  In unadjusted models, only the intakes of thiamin, magnesium and iron were found to be different by less than -3% on weekend days than on weekdays (Figure 3).  In energy adjusted models, most of the micronutrients differed but the percent of difference was equal or less than - 5% on weekend days than on weekdays.  These nutrients included vitamin C, riboflavin, niacin, vitamin B6, total folate from food (both naturally occurring folate and fortified folic acid from food), phosphorus, magnesium, iron, zinc, sodium and potassium.  The only four micronutrients that had a greater than 5% lower intake on weekends compared to weekdays were vitamin D (- 6.0%, SE = 3.1, p = 0.05), thiamin (-5.9%, SE = 1.0, p < 0.01), folic acid (-6.6%, SE = 2.0, p <0.01), and calcium (-5.1%, SE = 1.2, p < 0.01). 4.4 Mean Healthy Eating Index scores between weekday and weekend groups The total mean score of the healthy eating index (HEI) was 57.5 for the whole sample aged 1 and older, excluding pregnant and/or breast-feeding women (Table 5).  Overall, the scores for most of the components were relatively low compared to national recommendations, with mean component scores for dark green vegetable and whole grain products scoring the lowest. The unadjusted and energy-adjusted mean scores of the healthy eating index (HEI) and its 11 component scores on weekdays and weekend days are summarized in Table 5.  There was a significant difference in total HEI score between intakes reported on weekday days compared to weekdays in the unadjusted model.  Moreover, all but sodium intake of the HEI component scores were significantly different on weekend days compared to weekdays.  The HEI component scores in meat and alternatives and unsaturated fats were significantly higher on weekend days than on weekdays.  In contrast, the component scores of the food groups were significantly lower on weekends than on weekdays, ranging from -0.09 out of a 5 point scale 45  Table 5 Linear regression analysis comparing the healthy eating index scores of Canadians aged 1 and older on weekend days with weekday intake: unadjusted and energy-adjusted models  Total  Sample Mean score (SE) Unadjusted Models Energy-adjusted Models  Mean score p-value R 2  Mean score p-value R 2  Component (maximum possible score) Weekday (SE) Weekend (SE)   Weekday (SE) Weekend (SE)  Total vegetables and fruit (10) 5.89 (0.03) 5.98 (0.04) 5.75 (0.06) <0.01 0.00 6.01 (0.04) 5.73 (0.05) <0.01 0.05 Dark green or orange vegetable (5) 1.53 (0.02) 1.56 (0.02) 1.48 (0.03) 0.04 0.00 1.57 (0.02) 1.47 (0.03) 0.02 0.00 Whole fruits (5) 2.41 (0.02) 2.51 (0.03) 2.28 (0.04) <0.01 0.00 2.52 (0.03) 2.27 (0.04) <0.01 0.00 Grain products (5) 3.50 (0.02) 3.53 (0.02) 3.46 (0.03) 0.04 0.00 3.54 (0.02) 3.44 (0.02) <0.01 0.44 Whole grain products (5) 1.57 (0.02) 1.65 (0.02) 1.47 (0.03) <0.01 0.00 1.65 (0.02) 1.47 (0.03) <0.01 0.18 Milk products (10) 5.55 (0.04) 5.66 (0.05) 5.40 (0.06) <0.01 0.00 5.68 (0.04) 5.35 (0.06) <0.01 0.11 Meat and alternatives (10) 6.44 (0.03) 6.36 (0.04) 6.56 (0.06) 0.01 0.00 6.40 (0.04) 6.52 (0.06) 0.08 0.14 Unsaturated fats (10) 8.27 (0.03) 8.20 (0.03) 8.37 (0.04) <0.01 0.00 8.23 (0.03) 8.31 (0.04) 0.05 0.31 Saturated fats (10) 6.41(0.04) 6.49 (0.05) 6.31 (0.07) 0.04 0.00 6.50 (0.05) 6.35 (0.07) 0.08 0.01 Sodium (10) 5.54 (0.04) 5.56 (0.05) 5.52 (0.07) 0.61 0.00 5.51 (0.03) 5.62 (0.05) 0.07 0.44 Percentage of energy from other food (20) 10.31 (0.08) 10.76 (0.1) 9.69 (0.1) <0.01 0.01 10.7 (0.1) 9.75 (0.14) <0.01 0.04 Total score (100) 57.49 (0.15) 58.3 (0.2) 56.4 (0.3) <0.01 0.01 58.3 (0.2) 56.4 (0.3) <0.01 0.01 Note. n=34,402 excluding pregnant and breast feeding women. SE = standard error. 95% CI = 95% confidence interval.R 2 =coefficient of determination.  46  (SE = 0.04, p < 0.04) in dark green and orange vegetables to -0.26 out of a 10 point scale (SE = 0.08, p < 0.01) for milk products.  The percentage of energy from other food was higher on weekend days than on weekdays, resulting in a significantly lower HEI component score on weekend days by 1.07 out of a 20 point scale (SE = 0.1, p < 0.01).  Similarly, the HEI component score for saturated fats was significantly lower on weekend days than on weekdays, suggesting a higher intake of saturated fats on weekend. The magnitude of the weekday-weekend difference in HEI scores was very similar between the unadjusted and the energy-adjusted models.  The component scores for meat and alternatives, and percent of energy from saturated fats on weekend days were, however, no longer different from weekday scores after energy adjustment. In summary, overall diet quality was significantly poorer on weekends than on weekdays in the Canadian population.  Although the respondents had a slightly higher intake of meat and alternatives and unsaturated fats on weekend days, they had fewer servings of total vegetables and fruits, dark green and orange vegetables, whole fruits, grain products, whole grain products, and milk products on weekend days than on weekdays.  The respondents also consumed a higher percentage of energy intake from “other foods” on weekend days than on weekdays. 4.5 The moderating effects of sex, age groups, and employment status on differences in weekday-weekend dietary intake The results from the Wald test, which determined if adding the interaction terms of sex, age group or employment/student status improved the overall fit of the models, are presented in Table 6 and Table 7.  Sex, age and employment status were not significant moderators of the magnitude of weekday-weekend difference in caloric intake.  Interaction effects of sex, age group and employment status were found to be significant for the models predicting only 13 of 47  Table 6 The p-values from the Wald tests determining if adding interaction terms improved the overall fit of the linear regression models, with type of day as a dummy variable and nutrient intake as outcome variables in Canadians aged 1 and older.  Sex Interaction Terms Age Group interaction terms Working/Student Status interaction terms Nutrient Unadjusted Models a  Energy- adjusted Models b  Unadjusted Models c  Energy- adjusted Models d  Unadjusted Models e  Energy- adjusted Models f  Carbohydrate 0.12 0.01 0.19 0.01 <0.01 0.26 Protein 0.45 0.47 0.49 0.74 0.79 0.67 Alcohol 0.74 0.84 0.08 0.01 <0.01 <0.01 Fat 0.21 <0.01 0.68 0.19 0.64 0.60 Saturated FFA 0.21 0.03 0.37 0.02 0.35 0.33 MUFA 0.20 0.01 0.76 0.76 0.13 0.44 PUFA 0.75 0.49 0.41 0.70 0.22 0.50 Linoleic FFA 0.96 0.87 0.46 0.80 0.38 0.60 Linolenic FFA 0.73 0.79 0.40 0.45 0.36 0.40 Cholesterol 0.21 0.13 0.29 0.07 0.01 0.01 Dietary fibre 0.91 0.94 0.37 0.16 0.41 0.46 Sugar 0.62 0.57 <0.01 <0.01 <0.01 <0.01 Vitamin A 0.12 0.10 0.83 0.69 0.46 0.43 Vitamin D 0.61 0.55 0.27 0.38 0.35 0.25 Vitamin C 0.27 0.27 0.18 0.26 0.27 0.26 Thiamin 0.17 0.12 0.12 0.02 0.09 0.14 Riboflavin 0.47 0.45 0.58 0.46 0.31 0.31 Niacin 0.70 0.76 0.43 0.58 0.58 0.65 Vitamin B6 0.82 0.63 0.63 0.74 0.30 0.38 Vitamin B12 0.34 0.29 0.65 0.78 0.89 0.88 Food folate 0.16 0.09 0.14 0.32 0.38 0.43 Folic acid <0.01 <0.01 0.46 0.28 0.61 0.65 Total folate from food 0.35 0.32 0.15 0.27 0.41 0.30 Calcium 0.74 0.57 0.52 0.31 0.01 <0.01 Phosphorus 0.91 0.92 0.42 0.52 0.43 0.20 Magnesium 0.79 0.90 0.20 0.17 0.35 0.57 Iron 0.30 0.23 0.24 0.06 0.05 0.03 Zinc 0.78 0.88 0.16 0.05 0.15 0.15 Sodium 0.93 0.73 0.27 0.37 0.68 0.20 Potassium 0.75 0.83 0.13 0.05 0.33 0.27 Note. The Wald test was used to test hypothesis that the coefficient of interaction term equals to zero. aNutrient intake = β0 + β1 weekend + β2 male + β3 weekend*male bNutrient intake = β0 + β1 weekend + β2 male + β3 weekend*male + β4energy cNutrient intake = β0 + β1 weekend + β2 child + β3youth + β4adult + β5weekend*child+ β6weekend*youth + β7weekend*adult dNutrient intake = β0 + β1 weekend + β2 child + β3youth + β4adult + β5weekend*child+ β6weekend*youth + β7weekend*adult + β8energy eHEI = β0 + b1 weekend + β2 student + β3working + β4retiree + β5weekend*student+ β6weekend*working + β7weekend*retiree fHEI = β0 + b1 weekend + β2 student + β3working + β4retiree + β5weekend*student+ β6weekend*working + β7weekend*retiree + β8energy 48  Table 7 The p-values from the Wald tests determining if adding interaction terms improved the overall fit of the linear regression models, with type of day as dummy variable and Healthy Eating Index as outcome variable in Canadians aged 1 and older.  Sex Interaction Terms Age Group interaction terms Working/Student Status interaction terms HEI Unadjusted Models a  Energy- adjusted Models b  Unadjusted Models c  Energy- adjusted Models d  Unadjusted Models e  Energy- adjusted Models f  Total vegetables and fruit 0.29 0.31 0.56 0.60 <0.01 <0.01 Dark green or orange vegetable 0.53 0.51 0.42 0.39 <0.01 <0.01 Whole fruits  0.13 0.14 0.34 0.28 0.02 0.02 Grain products 0.69 <0.01 0.46 0.72 0.09 0.09 Whole grain products <0.01 0.15 0.18 0.17 0.07 0.07 Milk products  0.68 0.75 0.46 0.24 0.01 0.01 Meat and alternatives 0.58 0.48 0.89 0.98 0.22 0.22 Unsaturated fats 0.24 0.07 0.71 0.75 0.46 0.46 Saturated fats 0.03 0.02 0.00 0.01 0.11 0.11 Sodium 0.61 0.73 0.63 0.52 0.64 0.64 Percentage of energy from other food 0.35 0.30 0.53 0.54 0.78 0.78 Total score 0.15 0.14 0.76 0.74 0.04 0.04 Note. The Wald test was used to test hypothesis that the coefficient of interaction term equals to zero. aHEI = β0 + β1 weekend + β2 male + β3 weekend*male bHEI = β0 + β1 weekend + β2 male + β3 weekend*male + β4energy c HEI = β0 + β1 weekend + β2 child + β3youth + β4adult + β5weekend*child+ β6weekend*youth + β7weekend*adult dHEI = β0 + β1 weekend + β2 child + β3youth + β4adult + β5weekend*child+ β6weekend*youth + β7weekend*adult + β8energy eHEI = β0 + b1 weekend + β2 student + β3working + β4retiree + β5weekend*student+ β6weekend*working + β7weekend*retiree fHEI = β0 + b1 weekend + β2 student + β3working + β4retiree + β5weekend*student+ β6weekend*working + β7weekend*retiree + β8energy  49  the 30 nutrients examined, which included carbohydrate, alcohol, total fats, saturated fats, MUFA, cholesterol, sugar, folic acid, thiamin, calcium, iron, zinc and potassium.  Similarly, total HEI score and component scores for grain products, saturated fats, total vegetables and fruits, dark green or orange vegetables, whole fruits, and milk products on weekday versus on weekend days were also modified by sex, age and employment/student status. The energy-adjusted intakes of these 13 nutrients and HEI scores on weekdays and weekend days by sex, age group and employment/student status are summarized in Table 8.  The following sections describe how the effect of type of day on nutrient intakes was moderated by 1) sex; 2) age group; and 3) employment/student status. 4.5.1 Sex as a moderating variable Overall, the magnitude of weekday-weekend difference in nutrient intake was similar in both males and females in unadjusted models (Table 6).  After adjusting for energy intake, the effect of type of day on carbohydrate, total fats, saturated fats, MUFA and folic acid were significantly different between males and females.  The intake of energy-adjusted carbohydrate was significantly lower on weekend than on weekday in both males and females; however, the magnitude of the difference was significantly greater in males than in females (Table 8).  Males consumed 12.3 g less carbohydrate on weekend days than on weekdays, while females consumed only 5.0 g less.  Higher weekend intakes of total fats, saturated fats, and MUFA and lower weekend intake of folic acid were only found in males.  No weekday-weekend differences were found in these nutrients among females.  There was no difference in the magnitude of weekday- weekend difference in total HEI score between males and females.  However, sex was a significant moderator on the magnitude of weekday-weekend difference in HEI component scores for grain products and saturated fats.  The component scores for grain products and 50  Table 8 Energy-adjusted mean nutrient intakes and Healthy Eating Index (HEI) scores on weekdays and weekend days among Canadians aged 1 and older, by sex, age, and employment/student status.  Carbohydrate (g) Alcohol (g)  Total Fats (g) Saturated Fats (g)  Mean Intake p-value Mean Intake p-value Mean Intake p-value Mean Intake p-value Group Weekday  Weekend  Weekday  Weekend  Weekday  Weekend  Weekday  Weekend   (SE) (SE)  (SE) (SE)  (SE) (SE)  (SE) (SE)  Sex Male 268 (1.35) 256 (1.80) <0.01 - - - 73.2 (0.50) 75.9 (0.64) <0.01 24.2 (0.21) 25.0 (0.35) 0.07 Female 267 (1.14) 262 (1.74) 0.01 - - - 77.5 (0.44) 77.1 (0.58) 0.57 25.8 (0.22) 25.4 (0.28) 0.30 Age group Children (1-14 years) 289 (1.07) 286 (1.38) 0.14 0.81 (0.12) 0.47 (0.16) 0.09 - - - 26.5 (0.17) 26.8 (0.25) 0.32 Youths (15-24 years) 277 (2.32) 270 (3.12) 0.07 3.78 (0.64) 11.52 (1.43) <0.01 - - - 24.5 (0.39) 23.6 (0.49) 0.17 Adults (25-64 years) 260 (1.39) 248 (1.89) <0.01 7.79 (0.38) 13.46 (0.76) <0.01 - - - 24.8 (0.23) 24.9 (0.36) 0.68 Older adults (65+ years) 263 (1.69) 254 (2.03) <0.01 9.03 (0.56) 9.27 (0.66) 0.77 - - - 24.6 (0.24) 26.3 (0.47) <0.01 Employment status  Students - - - 0.21 (0.26) 1.09 (0.71) 0.25 - - - - - - Working - - - 7.56 (0.41) 14.31 (0.77) <0.01 - - - - - - Retirees - - - 9.29 (0.67) 12.52 (1.17) 0.01 - - - - - - Other a  - - - 6.27 (0.39) 7.56 (0.76) 0.12 - - - - - - 51   MUFA(g) Cholesterol (mg) Sugar (g) Folic Acid (mcg)  Mean Intake p-value Mean Intake p-value Mean Intake p-value Mean Intake p-value Group Weekday  Weekend  Weekday  Weekend  Weekday  Weekend  Weekday  Weekend   (SE) (SE)  (SE) (SE)  (SE) (SE)  (SE) (SE)  Sex Male 29.1 (0.25) 30.6 (0.33) <0.01 - - - - - - 133 (2.59) 116 (2.83) <0.01 Female 30.7 (0.21) 30.8 (0.27) 0.57 - - - - - - 123 (1.82) 122 (2.63) 0.88 Age group Children (1-14 years) - - - - - - 132 (1.01) 134 (1.37) 0.51 - - - Youths (15-24 years) - - - - - - 120 (2.02) 117 (2.74) 0.47 - - - Adults (25-64 years) - - - - - - 104 (1.14) 97.2 (1.40) <0.01 - - - Older adults (65+ years) - - - - - - 107 (1.41) 105 (1.83) 0.33 - - - Employment status  Students - - - 220 (3.77) 233 (5.05) 0.04 130 (1.25) 130 (1.75) 0.78 - - - Working - - - 267 (4.57) 303 (7.17) <0.01 107 (1.25) 98.2 (1.46) <0.01 - - - Retirees - - - 267 (5.11) 313 (11.5) <0.01 106 (1.42) 100 (2.08) 0.02 - - - Other a  - - - 267 (5.19) 273 (6.76) 0.53 112 (1.31) 115 (2.01) 0.18 - - -  52   Thiamin (mg) Calcium (mg) Iron (mg) Zinc (mg)  Mean Intake p-value Mean Intake p-value Mean Intake p-value Mean Intake p-value Groups Weekday Weekend Weekday Weekend  Weekday Weekend  Weekday Weekend   (SE) (SE)  (SE) (SE)  (SE) (SE)  (SE) (SE)  Sex Male - - - - - - - - - - - - Female - - - - - - - - - - - - Age group Children (1-14 years) 1.75 (0.01) 1.71 (0.02) 0.07 - - - - - - 10.3 (0.10) 10.3 (0.15) 0.94 Youths (15-24 years) 1.75 (0.03) 1.59 (0.04) <0.01 - - - - - - 11.0 (0.18) 10.2 (0.21) <0.01 Adults (25-64 years) 1.77 (0.02) 1.66 (0.02) <0.01 - - - - - - 11.6 (0.10) 11.3 (0.15) 0.07 Older adults (65+ years) 1.81 (0.02) 1.77 (0.03) 0.24 - - - - - - 11.4 (0.13) 11.4 (0.17) 0.92 Employment status  Students - - - 1060 (13.8) 1020 (23.5) 0.13 14.3 (0.13) 13.9 (0.17) 0.04 - - - Working - - - 897 (8.9) 810 (12.3) <0.01 14.5 (0.11) 13.5 (0.13) <0.01 - - - Retirees - - - 859 (13.0) 843 (24.1) 0.55 14.7 (0.14) 14.8 (0.30) 0.68 - - - Other a  - - - 944 (11.6) 967 (21.7) 0.34 14.3 (0.15) 13.6 (0.16) <0.01 - - -   53   Potassium (mg) HEI Total vegetables and fruit HEI Dark green or orange vegetable HEI Whole fruits  Mean Intake p-value Mean Score p-value Mean Score p-value Mean Score p-value Groups Weekday Weekend Weekday Weekend  Weekday Weekend  Weekday Weekend   (SE) (SE)  (SE) (SE)  (SE) (SE)  (SE) (SE)  Sex Male - - - - - - - - - - - - Female - - - - - - - - - - - - Age group Children (1-14 years) 2820 (15.6) 2794 (22.1) 0.33 - - - - - - - - - Youths (15-24 years) 2886 (35.8) 2704 (47.9) <0.01 - - - - - - - - - Adults (25-64 years) 3189 (20.5) 3069 (30.3) <0.01 - - - - - - - - - Older adults (65+ years) 3294 (28.5) 3214 (43.9) 0.11 - - - - - - - - - Employment status  Students - - - 5.96 (0.06) 5.69 (0.11) 0.03 1.32 (0.04) 1.06 (0.05) <0.01 2.56 (0.05) 2.18 (0.07) <0.01 Working - - - 5.86 (0.06) 5.42 (0.08) <0.01 1.65 (0.04) 1.51 (0.05) 0.04 2.32 (0.05) 2.07 (0.06) <0.01 Retirees - - - 6.64 (0.09) 6.63 (0.15) 0.96 1.72 (0.14) 1.82 (0.30) 0.34 2.96 (0.07) 2.97 (0.10) 0.96 Other a  - - - 6.06 (0.08) 6.19 (0.10) 0.30 1.50 (0.05) 1.60 (0.06) 0.22 2.68 (0.06) 2.59 (0.07) 0.35 54   HEI Grain Products HEI Milk products HEI Saturated Fats HEI Total score  Mean Score p-value Mean Score p-value Mean Score p-value Mean Score p-value Group Weekday  Weekend  Weekday  Weekend  Weekday  Weekend  Weekday  Weekend   (SE) (SE)  (SE) (SE)  (SE) (SE)  (SE) (SE)  Sex Male 3.48 (0.03) 3.28 (0.03) <0.01 - - - 6.74 (0.07) 6.40 (0.10) <0.01 - - - Female 3.61 (0.02) 3.61 (0.04) 0.99 - - - 6.26 (0.07) 6.30 (0.09) 0.74 - - - Age group Children (1-14 years) - - - - - - 5.82 (0.07) 5.79 (0.09) 0.77 - - - Youths (15-24 years) - - - - - - 6.52 (0.12) 6.82 (0.15) 0.12 - - - Adults (25-64 years) - - - - - - 6.63 (0.07) 6.43 (0.10) 0.11 - - - Older adults (65+ years) - - - - - - 6.75 (0.09) 6.17 (0.16) <0.01 - - - Employment status  Students - - - 6.51 (0.08) 6.06 (0.11) <0.01 - - - 59.3 (0.29) 56.5 (0.41) <0.01 Working - - - 5.62 (0.07) 5.07 (0.09) <0.01 - - - 57.1 (0.31) 55.1 (0.42) <0.01 Retirees - - - 4.34 (0.11) 4.28 (0.14) 0.71 - - - 59.3 (0.43) 58.8 (0.72) 0.51 Other a  - - - 5.86 (0.09) 5.90 (0.13) 0.78 - - - 59.7 (0.37) 58.6 (0.45) 0.06  Note. a  included those who were currently not a full time student, had a job but were not at job in the week prior to the interview, or did not have a job due to reasons other than retirement.  – not applicable as the interaction terms was not found to be a significant predictor in the models. 55  saturated fats were significantly lower on weekends than on weekdays in males, while no difference was found in females. 4.5.2 Age group as a moderating variable In unadjusted models, the weekday-weekend differences in nutrient intake were similar among children, youth, adults and older adults.  After adjusting for energy intake, the effect of type of day on intakes of carbohydrate, alcohol, saturated fats, dietary fibre, thiamin, zinc, and potassium were significantly different among the four age groups (Table 6).  Of these seven nutrients, no weekday-weekend difference was found in children aged 1 to 14 years, suggesting there was no temporal variation in carbohydrate, alcohol, saturated fats, dietary fibre, thiamin, zinc, and potassium between weekday and weekend among children.  In youths aged 15-24, the intakes of thiamin, zinc, and potassium were less on weekend days than on weekdays (Table 8). In adults aged 25- 64 years, the intakes of thiamin and potassium were also significantly lower on weekend days than on weekdays. The magnitude of weekday-weekend differences in thiamin and potassium were similar in youths compared to in adults.  Lower carbohydrate intake on weekend days than on weekdays was found in both adults and older adults.  The magnitude of weekday-weekend differences was significantly greater in adults than in older adults.  Adults consumed 12.3 g less carbohydrate on weekends than on weekdays, while older adults consumed 8.6 g less.  The intake of saturated fats was significantly higher on weekend days than on weekdays in older adults only, which resulted in a lower HEI component score in saturated fats on weekend days in older adults. The effect of age group on the magnitude of weekday-weekend difference in nutrient intake was most noticeable for alcohol.  Figure 4 shows the magnitude of difference in alcohol intake on weekdays and on weekend days among children, youths, adults and older adults. 56  Figure 4 The intake of alcohol on weekdays and weekend days of Canadians aged 1 and older, by age groups.   0 2 4 6 8 10 12 14 16 Weekday Weekend In ta k e o f A lc o h o l (g ) Children  (1-13 years) Youths (14-24 years) Adults (25-64 years) Older adults (65+ years) 57  As expected, the mean intake of alcohol in children aged 1 to 14 years was very low and did not differ between weekdays and weekend days.  Similarly, there was no significant difference in alcohol intake between weekdays and weekend days for older adults aged 65 years and older. Compared to weekdays, alcohol intake on weekend was significantly higher in both youths and adults.  More importantly, the magnitude of weekday-weekend difference in alcohol intake was significantly greater in youths than in adults.  Weekend intake of alcohol in youths was 7.74g higher than those on weekdays, which was almost three times higher on weekends than on weekdays.  On the other hand, the difference in alcohol intake between weekday and weekend was 5.67 g in adults. 4.5.3 Employment/student status as a moderating variable The effect of type of day on alcohol, cholesterol, sugar, calcium and iron was modified by employment/student status in both unadjusted and energy-adjusted models (Table 6).  Sugar intake was significantly lower on weekend days than on weekdays in working people and retirees, but not in students; there was no difference in sugar intake between weekday and weekend in students (Table 8).  Calcium intake was lower on weekends than on weekdays in working people only.  The intake of iron on weekends was significantly lower than on weekdays in students and working people, but not in retirees.  No significant difference was observed in the magnitude of the weekday-weekend difference in iron between students and working people. There was a noticeable weekday-weekend variation in alcohol intakes between respondents who were attending school as full time students, were working, or had retired.  As shown in Figure 5, alcohol intake in students was low and did not differ between weekdays and weekend days.  On the other hand, there was a significantly higher intake of alcohol on weekends than on weekdays for respondents who were working or had retired.  Alcohol intake 58  almost doubled among working people, from 7.56 g (SE= 0.41) on weekdays to 14.3 g (SE = 0.77) on weekend days.  The magnitude of the weekday-weekend difference in alcohol intake was significantly smaller in retirees than in working people, with a higher intake of 3.23 g of alcohol on weekend than on weekday.  Similarly, the magnitude weekday-weekend difference in cholesterol intake was also different among respondents who were attending school, were working or had retired.  Intake of cholesterol was significantly higher on weekends than on weekdays by 12.7 g, 36.8 g, and 45.5 g in students, working people, and retirees respectively. Such weekday-weekend difference in cholesterol intake was significantly greater in working people and retirees than students. The weekday-weekend difference in total HEI score and component scores in total vegetables and fruit, dark green or orange vegetable, whole fruits and milk were also moderated by employment/student status.  There was no weekday-weekend difference in HEI scores in retirees, suggesting their diet quality remained unchanged over the course of a week.  On the other hand, the overall diet quality was significantly poorer on weekend days than on weekdays in students and working people as their total HEI scores were 2.79 and 2.01 lower on weekend days, respectively.  Component scores for total vegetables and fruits, dark green or orange vegetables, whole fruits, and milk products were significantly lower on weekend days than on weekdays in both students and working people, suggesting they are less likely to meet the dietary recommendations for servings of these food groups. 59  Figure 5 The intake of alcohol on weekdays and weekend days of Canadians aged 1 and older, by employment/student status.   * included those who were currently not a full time student, had a job but were not at job in the week  prior to the interview, or did not have a job due to reasons other than retirement.  0 2 4 6 8 10 12 14 16 Weekday Weekend In ta k e o f A lc o h o l (g ) Students Working Retirees Other* 60  In summary, moderating effects of sex, age, or employment/student status on weekday and weekend intakes only affect a small portion of nutrients examined.  For the majority of nutrients, interaction terms were not significant variables in regression models, suggesting that the weekday-weekend difference in nutrient intake was similar regardless of which sex and age groups respondents belonged to, or whether they were attending school, working or retired or not. Alcohol was found to be the nutrient that differed the most between weekdays and weekend days among different age groups and employment/student status.  The effect of weekday and weekend on diet quality was also similar between males and females, and between children, youths, adults, and older adults.  However, respondents who were attending school as full time students or working had a significantly poorer diet quality on weekends than on weekdays, while no difference in diet quality was seen in retirees.  61  Chapter  5: Discussion and Conclusion 5.1 Weekday-weekend difference in dietary outcomes The findings of this study revealed that there were small but significant differences in intakes of energy and some nutrients in Canadians aged one year and older between weekdays (Monday through Thursday) and weekend days (Friday to Sunday).  Respondents reported higher energy intakes on weekend days than on weekdays.  After adjusting for weekday- weekend differences in the amount of food consumed, the intakes of many nutrients became significantly different between weekdays and weekend days.  This suggests that not only the quantity of food consumed, but also the quality of dietary intake differed between weekdays and weekend days. 5.1.1 Macronutrient outcomes Compared to weekdays, intakes of calories, alcohol and fat were found to be higher, while intakes of carbohydrate and protein were found to be lower on weekend days in this nationally representative sample.  On average, energy intake was 3.0% higher (62 kcal) on weekend days compared to weekdays, which derived mainly from higher alcohol intake.  The largest relative difference in intake among all of the dietary outcomes examined was identified for alcohol intake, where consumption was 70% higher on weekend days compared to weekdays or approximately 4 grams of alcohol higher. Given that one standard alcoholic beverage (ie. one glass of wine or one can of beer) contains 13.6g of alcohol, 4 grams increase in alcohol intake is equivalent to drinking a quarter of a glass/can of wine/beer more on weekend days than on weekdays for those who drink, which is approximately 40% of the respondents in CCHS 2.2. While intakes of fat, carbohydrate and protein was also different on weekend days than 62  weekdays, the relative magnitude of difference were much smaller in these macronutrients, ranging from 1.5% in fat, -1.9% in protein and -3.3% in carbohydrate. For the most part, the findings for macronutrients were similar to previously published findings from nationally representative samples outside of Canada.  Table 9 summarizes the findings of the three previously published studies that have examined the effects of days of week on nutrient intake using nationally representative samples from the U.S. (Haines et al., 2003; Thompson et al., 1986) and New Zealand (Rockell et al., 2011) allowing for comparison of findings to the current study.  For example, using a similar methodological design applied to the current study, Haines et al. analyzed the data from1994-96 Continuing Survey of Food Intake by Individuals that collected 24-hour dietary recalls from 28,156 Americans aged two years and older and applied a similar definition of weekend days (i.e. Friday to Sunday) (Haines et al., 2003).  The study found that the percentage of energy from carbohydrate and alcohol was significantly different by -1.2% and 0.9%, respectively on weekend days than weekdays, which is only slightly lower by the -1.8% and 1.4% difference reported from the CCHS findings.  On the other hand, the weekday-weekend differences in the percentage of energy from protein (-0.3 to -0.36%) and fat (0.6% to 0.7%) were similar in both studies.  Overall, both the direction and magnitude of weekday-weekend difference in macronutrient outcomes between the two studies were consistent, likely due to the similarities in study design and sample characteristics. The study by Thompson et al., on the other hand, analyzed the difference in macronutrient intake between weekdays (Monday through Friday) and weekend days (Saturday and Sunday) of 13,215 American adults aged 23-74 years using the 1977-89 Nationwide Food Consumption Survey (Thompson et al., 1986).  With the exception of protein, which was found to be significantly higher on weekend days than weekdays, the direction of weekday-weekend 63  difference in caloric and macronutrient intakes were again in agreement with the present study. The magnitude of weekday-weekend differences in macronutrient outcomes were, however, much larger in the study by Thompson et al. (Table 9).  The discrepancy between these two studies was likely due to the difference in definition of “weekday” and “weekend”.  Friday was categorized as a weekend day in our study, but a weekday in the study by Thompson et al. Weekday/weekend mean intake could be affected depending on whether Friday intake was being included or not.  For example, in the current study it was found that intake of fat on Friday was significantly lower than those on Saturday and Sunday (Appendix A).  As a result, weekend mean intake of fat could be lower if Friday was included as part of weekend days compared to the mean intake on Saturday and Sunday alone.  Lastly, the study by Rockell et al. examined the weekday-weekend differences in nutrient intake of children aged 5 -15 years in New Zealand and reported no differences in intakes of energy, carbohydrate, protein and total fat between weekdays and weekend days (Rockell et al., 2011).  Similarly, the effect of day of a week was not found to be associated with the intake of energy or carbohydrate among respondents aged 1 to 14 years in our study.  Although total fat intake did not differ between weekdays and weekend days, the intake of cholesterol was higher by about 10% on weekends in both studies. 5.1.2 Micronutrient outcomes In the present study, lower consumption of many micronutrients was observed on weekend days.  However, most of the energy-adjusted weekday-weekend differences in mean micronutrient intakes were very small, with less than a 3% difference identified between weekdays and weekend days. Overall, these findings indicate a small but significant temporal variation in micronutrient intake.  The only exceptions were the intakes of vitamin D, thiamin, folic acid and calcium, which were lower on weekend days by approximately 6%.  Interestingly, 64  Table 9 Comparison in weekday-weekend difference in energy and macronutrient intake between current study and other studies that used nationally representative samples. Study (Country) & Data Used Analytical Sample Dietary Assessment  & Definition of "weekday" and "weekend" Weekday-weekend difference Energy intake Carbohydrate Protein Fat Alcohol Current Study (Canada)  CCHS 2.2 >1 year of age excluded pregnant and breast feeding women n = 34,402 One 24-hour recall  Weekday: Mon- Thurs Weekend: Sat- Sun 62 kcal (3%) higher on weekend days. Interaction effects of sex, age group and employment/ student status not significant. 3.3% lower on weekend days a  Magnitude of weekday- weekend difference was greater in males than in females (- 4% vs. -1.8%). 1.9% lower on weekend days a  Interaction effects of sex, age group and employment/ student status was not significant. Total fat intake was 1.5% higher on weekend days a . With the exception of MUFA and cholesterol, there was no weekday- weekend difference in different type of fat Cholesterol intake was 9.8% higher on weekend days a . Total fat was significantly higher on weekend in males, but not in females.  66% higher on weekend days a  Youths aged 15-24 years was found to have the greatest increase in alcohol intake on weekend. Haines et al. (U.S.)  1994-96 Continuing Survey of Food Intakes by Individuals >2 years of age and older n = 28,156 Two nonconsecutive 24-hour recalls  Weekday: Mon- Thurs Weekend: Sat- Sun 82 kcal higher on weekend days in total sample.  Significant weekday- weekend difference in adults aged 19-70 years. Greatest weekday- weekend difference in adults aged 19-50 years (115 kcal higher on weekend) No weekday- weekend difference in children The percentage of energy from carbohydrate was 1.2 % lower on weekend days in total sample. Subgroup analysis showed that the weekday- weekend difference in carbohydrate was found to be significant in adults aged 19- 50 years only. The percentage of energy from protein was 0.3% lower on weekend in total sample. The percentage of energy from fat was 0.7% higher on weekend in total sample.  Subgroup analysis showed that the weekday-weekend difference in fat was found to be significant only in adults aged 19- 50 years. The percentage of energy from alcohol was 0.9% higher on weekend days.  Respondents aged 19 to 50 years had the greatest increase in alcohol intake by 1.4% of energy. 65  Study (Country) & Data Used Analytical Sample Dietary Assessment  & Definition of "weekday" and "weekend" Weekday-weekend difference Energy intake Carbohydrate Protein Fat Alcohol Thompson et al. (U.S.)  1977-78 The Nationwide Food Consumption Survey Aged 23-74 years n = 13,215 Three consecutive 24- hour recalls and two dietary records  Weekday: Mon- Fri Weekend: Sat- Sun 86-188 kcal higher on weekend in different age and sex groups.  Men aged 35-50 years had the greatest increase in energy intake on weekend (188 kcal) No weekday- weekend difference in older adults aged 65-74 years.     7.4% lower in middle-aged males (51-64 years) but 5.2% higher in females aged 23-34 years on weekend days.  No weekday- weekend difference in other age and sex groups. 5.2%-7.1% higher in females aged 23-50 on weekend days. 5.6-10% higher in both males and females on weekend days.  Females aged 35-50 years had the greatest increase in fat intake on weekend days. Not measured Rockell et al. (New Zealand)  NZ 2002 Children's Nutrition Survey Aged 5-15 years n = 2,572 One24-hour recall  Weekday: Mon- Fri Weekend: Sat- Sun No weekday- weekend difference. No weekday- weekend difference. No weekday- weekend difference. No weekday-weekend difference in total fat and saturated fat Cholesterol was significantly higher by 11% on weekend days. Not measured Note . a energy-adjusted  66  the intakes of most micronutrients did not differ significantly between weekdays and weekend days in unadjusted models.  However, once the weekday-weekend difference in energy intake was adjusted for, weekend intakes of micronutrients emerged as statistically significantly lower intakes, indicating that Canadians likely consumed foods that were less nutrient-dense on weekend days than on weekdays.  A study by Maisey et al. previously reported a difference in nutrient density of vegetable-derived micronutrients such as vitamin C, folate and beta carotene between weekdays and weekend days among a sample of 138 older adults aged 68-90 years in the U.K. (Maisey et al., 1995).  This is in contrast with two other previous studies where the weekend effect diminished when nutrient intakes were expressed relative to energy intakes.  For example, a study of 60 Canadian adults aged 25-44 (Beaton et al., 1983) and another among 14 female university students (Gibson et al., 1985) indicated that there was no difference in nutrient density between weekdays and weekend days.  The authors of these studies concluded that food consumption was comparable, with no differences in nutrient density between weekdays and weekend days.  The findings of these studies were, however, limited by their small sample sizes and low power to detect the differences, which makes comparisons with the present study difficult. 5.1.3 Dietary quality and food group intakes This study examined the Canadian adaptation of the Healthy Eating Index-2005 making it possible to compare the difference in total dietary quality as well as the consumption of various food components between weekdays and weekend days in relation to national dietary guidelines. It was found that overall Canadians consumed foods that were not only less nutrient-dense, but also were poorer in diet quality on weekends than on weekdays.  In particular, Canadians consumed a slightly lower mean number of servings for vegetables and fruits, whole fruits, 67  whole grains, milk products and other foods.  These differences likely help to explain the reported variations in macro and micro-nutrient intake observed in the current study. Of 11 HEI component scores examined, the magnitude of difference in dietary quality between weekdays and weekend days was the greatest for the “other foods” category, which included alcoholic beverages and energy-dense foods that are high in sugar and/or fat such as potato chips and soft drinks.  Given that the intakes of sugar and fat did not differ substantively between weekdays and weekend days, the lower component score in the “other foods” category on weekend days was likely a result of higher alcohol consumption observed on weekend days. HEI analyses revealed that Canadians had a lower mean number of vegetables and fruit servings on weekend days.  Lower weekend vegetable and fruit intake likely contributed to lower weekend intake of vitamin C, folate, magnesium, and potassium, nutrients rich in vegetables and fruits observed in this study.  Similarly, lower intake of grain and milk products on weekends could have contributed to the lower reported intakes of dietary fibre, vitamin B6, folic acid, calcium, vitamin D, iron and zinc observed on weekends compare to weekdays.  Similar to results in the current study, Burke et al. reported lower intake of total cereal products and total dairy products on weekends in a nationally representative sample of Irish adults aged 18-64 years (Burke, McCarthy, O'Dwyer, & Gibney, 2005). It is important to note that majority of Canadians did not report consuming a “high quality” diet defined as a diet closely in line with recommendations set by the Canadian’s Food Guide, whether it was reported on a weekday or weekend day.  Although the average total HEI scores differed significantly between weekdays (58 points out of 100) and weekend days (56 points out of 100), both mean scores reflected suboptimal dietary intake on the original 1995 Healthy Eating Index quality classification (>80 points= good diet; 50-80 points =diet that 68  requires improvement; and <50 = poor diet) (Kennedy, Ohls, Carlson, & Fleming, 1995). Findings from this study showed that intake of foods from most of the food groups examined were inadequate compared to dietary guidelines for both weekdays and weekend days, especially foods from dark green or orange vegetables and whole grain products.  Overall, the differences in diet quality between weekdays and weekend days were marginal in the general Canadian population and substantive improvements are needed across the week to improve nutrition- related health outcomes at the population-level. 5.2 The weekend effect by sex, age and employment The current study examined the magnitude of weekday-weekend difference in dietary intake among various demographic groups.  For the most part, weekday-weekend differences in intakes of energy and nutrients were similar across different sex and age groups, and employment/student status.  Although a sex-weekend interaction was observed for a few nutrients such as carbohydrate and folic acid, there was no difference in the magnitude of weekday-weekend variation in dietary quality between males and females, suggesting that the type of foods consumed were very similar in both males and females. Studies from Haines et al. and Thompson et al. showed that the magnitude of weekday- weekend difference in energy intake decreased as age increased for adults aged 19 years and older (Haines et al., 2003; Thompson et al., 1986).  Such a trend was not seen in our study as there was no significant age-weekend interaction in intake of energy and most dietary outcomes, except alcohol.  The weekday-weekend difference in alcohol intake was most apparent in youths aged 15-24 years old in our sample.  Alcohol consumption almost tripled on weekdays in this age group.  This indicates that youths drank on average half of a glass/can of wine/beer more on weekends than on weekdays.  One possible explanation for such an increase in alcohol intake 69  observed among youths could be due to drinking during social nights out during weekends in this particular age group (Parker & Williams, 2003). When analyzing the difference in dietary quality between weekdays and weekends by employment/student status, it was shown that people who were working or attending school as full time students tended to consume foods with poorer dietary quality on weekend days.  Since working/school days in North America are generally regarded as Monday to Friday, it appears that the difference in dietary quality observed in the present study was associated with the organizational difference in working/school schedules between weekdays and weekend days. This is in contrast with the study by Thompson et al. where they found no association between weekend effects on macronutrient intakes and employment status, though it is unclear how the employment status was defined in their sample (Thompson et al., 1986).  There was no weekday- weekend difference in dietary quality among the retirees, which was expected given that retired Canadians are likely less constrained by either a working or school-related schedule.  It is unclear what factors are associated with working or studying that shape the differences in dietary outcomes between weekdays and weekend days.  It is possible that there is more frequent consumption of meals away from home, which are often energy-dense with poor dietary quality during non-working/non-school as they are less bound to their regular schedule. 5.3 Strengths and limitations To our knowledge, this is the first study that provided a comprehensive evaluation of the difference in energy, macronutrient and micronutrient intakes and dietary quality between weekdays and weekend days and it is the first study that examined the weekday-weekend variation in dietary intake using a nationally representative sample of Canadians.  One of the major strengths in the present study is therefore the generalizability of these findings to the 70  Canadian population.  The use of survey weights provided by Statistics Canada ensured that nonresponse bias and complex sampling design of the survey were accounted for so that the estimates were representative of the target population.  In addition, the data were analyzed by age and sex group in order to ensure that the findings were consistent across the lifespan. Furthermore, the definition of weekdays and weekends was established empirically. Based on a linear regression approach, the intakes on Fridays were examined systematically across 30 nutrient outcomes to determine if they were more similar to the intakes on Monday to Thursday (weekdays), or to that of Saturday and Sunday in the sample.  Another strength in this study was that diet quality was measured using the Canadian adaptation of Healthy Eating Index- 2005.  This was a contribution to current understanding of temporal variation because it allowed the examination of weekday-weekend difference in dietary intake in the context of national dietary recommendations, which has never been done before. There were several limitations relating to CCHS 2.2 data that should be kept in mind when interpreting findings from in this study.  First, the accuracy of dietary intake from one 24 hour recall can be subjected to different sources of errors (Health Canada, 2006).  For example, the dietary information of children six years and younger was provided by a parent or guardian, who may not be able to report meals consumed out of his/her  presence accurately (e.g. food consumed at day care) (Health Canada, 2006).  In addition, it has been reported that 43% of respondents aged 12 and over in CCHS 2.2 were identified as “non-plausible respondents” whose energy intake was either substantively higher or lower than their total predicted energy expenditure, suggesting that their dietary intake may be under- or over-reported (Garriguet, 2008).  As a result, the accuracy of nutrient intake could be compromised with the inclusion of non-plausible respondents in the analysis.  It is possible to exclude non-plausible respondents 71  based on their difference in their predicted energy requirement and actual energy intake. However, our preliminary analysis showed that there was no statistical difference in the proportion of non-plausible respondents on weekdays and weekend days, suggesting that the recall bias was randomly distributed between weekdays and weekend days.  Given the issues of under- or over-reporting in CCHS 2.2 would not systematically bias the nutrient estimates on weekdays and weekend days, non-plausible respondents were included in this study. There are two additional limitations to this study’s approach that should be kept in mind. First of all, nutrient intake included in this study was based on intake from foods and beverages only (as collected by one 24-hour recall), but did not include intake from vitamin or mineral supplements.  It was not possible to examine the weekend effect on the intakes from supplements in CCHS 2.2 because the use of supplements referred to consumption over the past 30 days and CCHS 2.2 did not collect information on whether the supplement was used on weekdays or weekend days.  The sample of this study showed that about 40% of respondents reported taking vitamin and mineral supplements, therefore intakes of micronutrients from food were likely lower than actual intakes from both foods and supplements combined.  Another limitation of the study design is that the food group classification was based on the 1992 Canada Food Guide (CFG), as it was the only food classification available in CCHS 2.2 at the time of study. As a result, some foods could be classified to different food groups between the old and current food guides.  For example, fortified soy beverages are classified as milk products in the 2007 CFG food group classification, (Marchand & Lowell, 2007).  Nevertheless, the Canadian adaptation of Health Eating Index-2005 used in our study was developed based on the 2007 CFG (Garriguet, 2009); therefore it was still the best possible approach to examine Canadians’ diet quality in relation to the current guidelines.  Given there were minor changes between the 1992 and 2007 72  CFF food group classification, it is unlikely that findings would be substantially different if the HEI scores were estimated based on the newer classification. 5.4 Implications on dietary assessment and future research directions This study’s findings suggest that ideally it would be desirable to collect dietary information on both a weekday and a weekend day to account for the temporal variation in intake between them.  However, consideration should be given to collecting data on both weekdays and weekend days since it may increase the burden on respondents and the costs associated with data collection.  It was shown in the present study that dietary intakes did not differ radically between weekdays and weekend days, with most of the differences being less than 10% across different demographic groups.  This suggests that measuring intakes only on a weekday or a weekend day will not likely substantively affect the overall accuracy of the nutrient estimates.  On the other hand, proportional sampling of a mix of days of the week would be recommended for alcohol since it was found that alcohol intake differed by 70% between weekdays and weekend days. Intake of alcohol will likely be underestimated if dietary assessment is done on weekdays only. In summary, whether or not to include weekend as part of dietary assessment depends on research objectives (ie. nutrients of interest) and resources available, but would be improved to a small extent if both weekend and weekday dietary data are collected. Based on results of the current study, there are several suggestions for future research. The long term impact of weekday-weekend difference in dietary intake on health outcomes requires future exploration.  For example, it could be of interest to determine if a higher intake of calories on weekend days will contribute to weight gain or not, which could provide potential insights about weight management.  Moreover, in the current study, employment/student status was associated with poorer dietary quality during weekend days.  Future studies should focus on 73  identifying the difference in socio-cultural contexts between weekdays and weekend days that are responsible for the temporal variation in dietary intakes. 5.5 Conclusion This study clearly shows that there was a small difference in nutrient intake and dietary quality between weekdays and weekends in a nationally representative sample of Canadians.  On average, Canadians consumed slightly more calories on weekends than weekdays, and that the higher intake of alcohol during weekend days contributed to 50% of the increase in energy intake. The overall dietary quality was marginally poorer on weekends than weekdays.  In particular, weekend consumption of foods from vegetable and fruit, grain products, and milk products was lower compared to serving recommendations from Canada’s Food Guide.  As a result, many of the energy-adjusted intakes of micronutrients such as calcium, folic acid and vitamin D were slightly lower on weekends than on weekdays.  With the exception of alcohol, sex and age group did not moderate the weekend effect on dietary outcomes.  Furthermore, poorer dietary quality was observed on weekend days than on weekdays in people who were working or attending school, while no difference was seen in retirees.  This suggests that the differences in dietary intakes between weekdays and weekends could be associated with work or school schedules.  In conclusion, Canadians consume foods with a slightly less favorable nutrient profile and marginally poorer diet quality on weekends than on weekdays.  74  References Basiotis, P. P., Thomas, R. G., Kelsay, J. L., & Mertz, W. (1989). Sources of variation in energy intake by men and women as determined from one year's daily dietary records. American Journal of Clinical Nutrition, 50(3), 448-453. Beaton, G. H., Milner, J., Corey, P., McGuire, V., Cousins, M., Stewart, E., & Little, J. A. (1979). Sources of variance in 24-hour dietary recall data: Implications for nutrition study design and interpretation. American Journal of Clinical Nutrition, 32(12), 2546-2559. Beaton, G. H., Milner, J., McGuire, V., Feather, T. E., & Little, J. A. (1983). Source of variance in 24-hour dietary recall data: Implications for nutrition study design and interpretation. carbohydrate sources, vitamins, and minerals. American Journal of Clinical Nutrition, 37(6), 986-995. Burke, S. J., McCarthy, S. N., O'Dwyer, N. A., & Gibney, M. J. (2005). Analysis of the temporal intake of cereal and dairy products in Irish adults: Implications for developing food-based dietary guidelines. Public Health Nutrition, 8(3), 238-248. Cai, H., Shu, X. O., Hebert, J. R., Jin, F., Yang, G., Liu, D. K., & Zheng, W. (2004). Variation in nutrient intakes among women in Shanghai, China. European Journal of Clinical Nutrition, 58(12), 1604-1611. Cai, H., Yang, G., Xiang, Y. B., Hebert, J. R., Liu, D. K., Zheng, W., & Shu, X. O. (2005). Sources of variation in nutrient intakes among men in Shanghai, China. Public Health Nutrition, 8(8), 1293-1299. 75  Carriquiry, A. L. (2003). Estimation of usual intake distributions of nutrients and foods. Journal of Nutrition, 133(2), 601S-8S. Chalmers, F. W., Clayton, M. M., Gates, L. O., Tucker, R. E., Wertz, A. W., Young, C. M., & Foster, W. D. (1952). The dietary record; how many and which days?. Journal of the American Dietetic Association, 28(8), 711-717. Cullen, K. W., Lara, K. M., & de Moor, C. (2002). Children's dietary fat intake and fat practices vary by meal and day. Journal of the American Dietetic Association, 102(12), 1773-1778. de Castro, J. M. (1991). Weekly rhythms of spontaneous nutrient intake and meal pattern of humans. Physiology & Behavior, 50(4), 729-738. Eppright, E. S., Patton, M. B., Marlatt, A. L., & Hathaway, M. L. (1952). Dietary study methods. V. Some problems in collecting dietary information about groups of children. Journal of the American Dietetic Association, 28(1), 43-48. Garriguet, D. (2008). Impact of identifying plausible respondents on the under-reporting of energy intake in the Canadian Community Health Survey. Health Reports, 19(4), 47-55. Garriguet, D. (2007). Canadians' eating habits. Health Reports, 18(2), 17-32. Garriguet, D. (2009). Diet quality in Canada. Health Reports, 20(3), 41-52. Gibson, R., Gibson, I., & Kitching, J. (1985). A study of inter- and intrasubject variability in seven-day weighed dietary intakes with particular emphasis on trace elements. Biological Trace Element Research, 8, 79-91. 76  Gorin, A. A., Phelan, S., Wing, R. R., & Hill, J. O. (2004). Promoting long-term weight control: Does dieting consistency matter?. International Journal of Obesity & Related Metabolic Disorders: Journal of the International Association for the Study of Obesity, 28(2), 278-281. Guenther, P. M., Reedy, J., & Krebs-Smith, S. M. (2008). Development of the Healthy Eating Index-2005. Journal of the American Dietetic Association, 108(11), 1896-1901. Haines, P. S., Hama, M. Y., Guilkey, D. K., & Popkin, B. M. (2003). Weekend eating in the United States is linked with greater energy, fat, and alcohol intake. Obesity Research, 11(8), 945-949. Hardy, M. (1993). Assessing group differences in effects. Regression with dummy variables. Sage university paper series on quantitative applications in the social sciences (pp. 29-33). Newbury Park, CA: Sage. Harrington, K. E. (2001). The study of periodicity of eating and public health nutrition issues. (Doctoral Dissertation, Dublin Institute of Technology), 227. Hartman, A. M., Brown, C. C., Palmgren, J., Pietinen, P., Verkasalo, M., Myer, D., & Virtamo, J. (1990). Variability in nutrient and food intakes among older middle-aged men. Implications for design of epidemiologic and validation studies using food recording. American Journal of Epidemiology, 132(5), 999-1012. Health Canada. (1997). Canada's Food Guide to Healthy Eating for people four years old and over. ( No. Catalogue H39-2521/1992E). Ottawa, ON: Minister of Public Works and Government Services Canada. 77  Health Canada. (2006). Canadian Community Health Survey, cycle 2.2, nutrition (2004) A guide to accessing and interpreting the data. Retrieved 10, 2012, from http://www.hc-sc.gc.ca/fn- an/surveill/nutrition/commun/cchs_guide_escc-eng.php Health Canada. (2007a). Eating well with Canada's food guide. ( No. Catalogue H164-38/1- 2011E). Ottawa, ON: Health Canada. Health Canada. (2007b). Eating well with Canada's Food Guide. What is a food guide serving of grain products? Retrieved 5, 2012, from http://www.hc-sc.gc.ca/fn-an/food-guide- aliment/choose-choix/grain-cereal/serving-portion-eng.php Health Canada. (2007c). Eating well with Canada's Food Guide. What is a food guide serving of vegetables and fruit? Retrieved 5, 2012, from http://www.hc-sc.gc.ca/fn-an/food-guide- aliment/choose-choix/fruit/serving-portion-eng.php Institute of Medicine. (2004). Dietary reference intakes for water, potassium, sodium, chloride, and sulfate. Washington, DC: National Academy Press. Jackson, K. A., Byrne, N. M., Magarey, A. M., & Hills, A. P. (2008). Minimizing random error in dietary intakes assessed by 24-h recall, in overweight and obese adults. European Journal of Clinical Nutrition, 62(4), 537-543. Jahns, L., Carriquiry, A., Arab, L., Mroz, T. A., & Popkin, B. M. (2004). Within- and between- person variation in nutrient intakes of Russian and U.S. children differs by sex and age. Journal of Nutrition, 134(11), 3114-3120. 78  Jula, A., Seppanen, R., & Alanen, E. (1999). Influence of days of the week on reported food, macronutrient and alcohol intake among an adult population in south western Finland. European Journal of Clinical Nutrition, 53(10), 808-812. Kennedy, E. T., Ohls, J., Carlson, S., & Fleming, K. (1995). The Healthy Eating Index: Design and applications. Journal of the American Dietetic Association, 95(10), 1103-1108. Macdiarmid, J., Loe, J., Craig, L. C., Masson, L. F., Holmes, B., & McNeill, G. (2009). Meal and snacking patterns of school-aged children in Scotland. European Journal of Clinical Nutrition, 63(11), 1297-1304. Maisey, S., Loughridge, J., Southon, S., & Fulcher, R. (1995). Variation in food group and nutrient intake with day of the week in an elderly population. British Journal of Nutrition, 73(3), 359-373. Mann, J. I. (2002). Diet and risk of coronary heart disease and type 2 diabetes. Lancet, 360(9335), 783-789. Marchand, V., & Lowell, H. (2007). The new Canada food guide. Paediatrics & Child Health, 12(4), 329-332. McGee, D., Rhoads, G., Hankin, J., Yano, K., & Tillotson, J. (1982). Within-person variability of nutrient intake in a group of Hawaiian men of Japanese ancestry. American Journal of Clinical Nutrition, 36(4), 657-663. 79  Mooney, C. Z., & Duval, R. D. (1993). Bootstrapping: A nonparametric approach to statistical inference. Sage University Paper series on quantitative application in the social science. Newbury Park, CA: Sage. Moshfegh, A. J., Rhodes, D. G., Baer, D. J., Murayi, T., Clemens, J. C., Rumpler, W. V. & Cleveland, L. E. (2008). The US Department of Agriculture automated multiple-pass method reduces bias in the collection of energy intakes. American Journal of Clinical Nutrition, 88(2), 324-332. Nicklas, T., Farris, R., Bao, W., Webber, L., & Berenson, G. (1997). Differences in reported dietary intake of 10-year-old children of weekdays compared to Sunday: The Bogalusa Heart Study. Nutrition Research, 17(1), 34-40. O'Dwyer, N. A., McCarthy, S. N., Burke, S. J., & Gibney, M. J. (2005). The temporal pattern of the contribution of fat to energy and of food groups to fat at various eating locations: Implications for developing food-based dietary guidelines. Public Health Nutrition, 8(3), 249-257. Palaniappan, U., Cue, R. I., Payette, H., & Gray-Donald, K. (2003). Implications of day-to-day variability on measurements of usual food and nutrient intakes. Journal of Nutrition, 133(1), 232-235. Parker, H., & Williams, L. (2003). Intoxicated weekends: Young adults' work hard-play hard lifestyles, public health and public disorder. Drugs: Education, Prevention & Policy, 10(4), 345-367. 80  Post, B., Kemper, H. C., & Storm-Van Essen, L. (1987). Longitudinal changes in nutritional habits of teenagers: Differences in intake between schooldays and weekend days. British Journal of Nutrition, 57(2), 161-176. Racette, S. B., Weiss, E. P., Schechtman, K. B., Steger-May, K., Villareal, D. T., Obert, K. A., & Holloszy, J. O. (2008). Influence of weekend lifestyle patterns on body weight. Obesity, 16(8), 1826-1830. Rhodes, D., Cleveland, L., Murayi, T., & Moshfegh, A. (2006). The impact of weekend eating on nutrient intakes [Abstract]. 6th International Conference on Dietary Assessment Methods, Program & Abstract 10-12. Rockell, J. E., Parnell, W. R., Wilson, N. C., Skidmore, P. M., & Regan, A. (2011). Nutrients and foods consumed by New Zealand children on schooldays and non-schooldays. Public Health Nutrition, 14(2), 203-208. Sepp, H., Lennernas, M., Pettersson, R., & Abrahamsson, L. (2001). Children's nutrient intake at preschool and at home. Acta Paediatrica, 90(5), 483-491. Shields, M., Carroll, M. D., & Ogden, C. L. (2011). Adult obesity prevalence in Canada and the United States. NCHS Data Brief, (56), 1-8. Statistics Canada. (2008). Canadian Community Health Survey (CCHS) cycle 2.2 (2004) nutrition: General health and 24-hour dietary recall components user guide. ( No. Catalogue 82M0024GPE). Ottawa, ON: Statistics Canada. 81  Statistics Canada. (2012). Definitions, data sources and methods: Age by life cycle groupings. Retrieved 10, 2012, from http://www.statcan.gc.ca/concepts/definitions/previous- anterieures/age1a-eng.htm. Tarasuk, V., & Beaton, G. H. (1992). Statistical estimation of dietary parameters: Implications of patterns in within-subject variation--a case study of sampling strategies. American Journal of Clinical Nutrition, 55(1), 22-27. Thompson, F., Larkin, F., & Brown, M. (1986). Weekend-weekday differences in reported dietary intake: The nationwide food consumption survey, 1977–78 Nutrition Research, 6(6), 647-662. Thomson, M., Eloton, R., Fulton, M., Brown, S., Wood, D., & Oliver, M. (1988). Individual variation in dietary intake of a group of Scottish men. Journal of Human Nutrition and Dietetics, 1, 47-57. Todd, K. S., Hudes, M., & Calloway, D. H. (1983). Food intake measurement: Problems and approaches. American Journal of Clinical Nutrition, 37(1), 139-146. Tokudome, Y., Imaeda, N., Nagaya, T., Ikeda, M., Fujiwara, N., Sato, J., & Tokudome, S. (2002). Daily, weekly, seasonal, within- and between-individual variation in nutrient intake according to four season consecutive 7 day weighed diet records in Japanese female dietitians. Journal of Epidemiology, 12(2), 85-92. 82  Verly Junior, E., Fisberg, R. M., Cesar, C. L., & Marchioni, D. M. (2010). Sources of variation of energy and nutrient intake among adolescents in Sao Paulo, Brazil. Cadernos De Saude Publica, 26(11), 2129-2137. Verret, F. (2006). Methodological challenges in analyzing nutrition data from the Canadian Community Health Survey - nutrition. Proceedings of Statistics Canada’s Symposium 2006 Methodological Issues in Measuring Population Health, Vittinghoff, E., Glidden, D. V., Shiboski, S. C., & McCulloch, C. E. (2005). Chapter 4 linear regression. Regression methods in biostatistics : Linear, logistic, survival, and repeated measures models (pp. 69-131). New York, NY, USA: Spinger.   83  Appendices Appendix A Linear regression analysis comparing nutrient intake of Canadians aged 1 and older on Saturday/Sunday and weekend days (Monday-Friday) with the Friday intake: unadjusted and energy-adjusted models  Unadjusted Models Energy-adjusted Models  Nutrient βSat-Sun p-value βMon-Thur p-value βSat-Sun p-value βMon-Thur p-value (95%CI) (95%CI) (95%CI) (95%CI) Energy intake (kcal) -27.80 0.49 -81.10 0.03 NA NA NA NA (-106, 50.6)  (-155, -7.03)  Carbohydrate (g) -11.02 0.03 -6.02 0.20 -7.83 <0.01 3.27 0.17 (-20.8, -1.26)  (-15.3, 3.27)  (-13.0, -2.71)  (-1.38, 7.92)  Dietary fiber (g) -0.66 0.08 0.36 0.35 -0.50 0.13 0.83 0.01 (-1.41, 0.079)  ( -0.38, 1.09)  ( -1.15, 0.15)  (0.21, 1.45)  Sugar (g) -0.43 0.88 0.57 0.83 0.90 0.70 4.47 0.03 (-6.15, 5.28)  (-4.61, 5.76)  (-3.66, 5.46)  ( 0.49, 8.44)  Fat (g) 1.62 0.41 -2.51 0.18 2.75 0.01 0.78 0.38 (-2.22, 5.47)  (-6.18, 1.15)  ( 0.79, 4.71)  (-0.96, 2.51)  Saturated FFA (g) 0.91 0.22 -0.39 0.57 1.28 0.01 0.68 0.11 ( -0.54, 2.36)  ( -1.72, 0.94)  ( 0.32, 2.23)  ( -0.14, 1.51)  MUFA (g) 0.44 0.60 -1.58 0.06 0.91 0.06 -0.22 0.63 ( -1.23, 2.12)  ( -3.20,  0.042)  ( -0.051, 1.87)  (-1.09, 0.66)  PUFA (g) 0.011 0.98 -0.54 0.18 0.21 0.49 0.04 0.89 (-0.82,  0.84)  (-1.33, 0.25)  ( -0.38, 0.80)  (-0.49, 0.56)  Linoleic FFA  (g) 0.076 0.83 -0.38 0.24 0.24 0.36 0.09 0.67 (-0.61, 0.76)  ( -1.01, 0.25)  ( -0.27, 0.75)  ( -0.34, 0.53)  Linolenic FFA (g) -0.037 0.62 -0.04 0.57 -0.009 0.88 0.04 0.59 (-0.18, 0.11)  (-0.20, 0.11)  ( -0.13, 0.11)  ( -0.094, 0.16)  Cholesterol (mg) 38.2 <0.01 -6.10 0.48 41.4 <0.01 3.25 0.68 (18.8, 57.6)  ( -23.1, 10.9)  ( 24.3, 58.6)  (-12.0, 18.54)  84   Unadjusted Models Energy-adjusted Models  Nutrient βSat-Sun p-value βMon-Thur p-value βSat-Sun p-value βMon-Thur p-value (95%CI) (95%CI) (95%CI) (95%CI) Protein (g) 1.03 0.62 0.12 0.95 2.03 0.16 3.05 0.03 ( -3.00, 5.05)  (-3.55, 3.79)  (-0.82, 4.88)  (0.37, 5.72)  Alcohol (g) -0.45 0.71 -4.82 <0.01 -0.27 0.82 -4.30 <0.01 (-2.83, 1.93)  ( -6.87, -2.76)  (-2.55, 2.01)  ( -6.26, -2.34)  Vitamin A (mcg in RAE) -2.79 0.97 11.77 0.86 3.24 0.96 29.4 0.65 ( -133,  127)  ( -116,  139)  (-125, 132)  (-97.5, 156)  Vitamin D (mg) 0.77 <0.01 0.75 <0.01 0.84 <0.01 0.94 <0.01 (0 .22, 1.32)  (0.33, 1.17)  ( 0.30, 1.37)  ( 0.55, 1.32)  Vitamin C (mg) -1.12 0.82 3.08 0.47 -0.065 0.99 6.17 0.12 ( -10.6, 8.36)  (-5.28, 11.4)  ( -9.12, 9.00)  (-1.64, 14.0)  Thiamin (mg) -0.019 0.60 0.05 0.16 0.001 0.99 0.10 <0.01 (-0.091, 0.052)  ( -0.019, 0.11)  ( -0.055, 0.056)  (0 .054, 0.16)  Riboflavin (mg) 0.035 0.40 0.03 0.45 0.056 0.07 0.09 <0.01 (-0.047, 0.12)  (-0.045, 0.10)  ( -0.0035, 0.12)  (0 .043, 0.14)  Niacin (mg in NE) -0.037 0.97 -0.23 0.79 0.41 0.58 1.08 0.12 (-1.96, 1.89)  (-1.95, 1.49)  (-1.04, 1.86)  (-0.29, 2.44)  Vitamin B6 (mg) -0.030 0.54 -0.01 0.88 -0.010 0.80 0.05 0.18 ( -0.13,  0.066)  (-0.093, 0.080)  (-0.093, 0.072)  (-0.024, 0.12)  Vitamin B12 (mg) -0.073 0.78 -0.01 0.96 -0.021 0.93 0.14 0.55 (-0.58, 0.43)  (-0.49, 0.47)  (-0.51, 0.47)  (-0.32, 0.61)  Food folate (mcg) -2.86 0.68 -2.81 0.68 -0.79 0.90 3.23 0.60 ( -16.3, 10.6)  ( -16.2, 10.6)  ( -13.0, 11.5)  ( -8.76 , 15.2)  Folic Acid (mcg) -15.36 <0.01 -5.78 0.19 -13.8 <0.01 -1.17 0.76 (-24.6, -6.13)  (-14.4, 2.86)  (-22.1, -5.50)  (-8.84, 6.50)  Total folate (mg in DFE) -16.3 0.14 -9.05 0.40 -11.4 0.18 5.43 0.50 ( -38.2,  5.50)  (-30.0, 11.9)  ( -28.2, 5.44)  (-10.3, 21.2)  85   Unadjusted Models Energy-adjusted Models  Nutrient βSat-Sun p-value βMon-Thur p-value βSat-Sun p-value βMon-Thur p-value (95%CI) (95%CI) (95%CI) (95%CI) Calcium (mg) -10.9 0.69 17.91 0.49 -0.97 0.97 46.9 0.02 (-63.8, 42.0)  ( -32.60, 68.4)  (-44.6, 42.7)  ( 7.44, 86.4)  Phosphorus (mg) 13.9 0.63 12.32 0.64 29.5 0.10 58.0 <0.01 ( -43.0,  70.7)  ( -39.64, 64.3)  (-5.96, 65.0)  (26.3, 89.6)  Magnesium (mg) -1.90 0.76 7.41 0.19 1.29 0.76 16.7 <0.01 (-13.9, 10.1)  (-3.70, 18.5)  (-7.14, 9.73)  ( 9.16, 24.3)  Iron (mg) -0.51 0.06 0.01 0.98 -0.35 0.06 0.47 0.01 (-1.04, 0.023)  (-0.48, 0.50)  (-0.71, 0.014)  (0 .14, 0.80)  Zinc (mg) 0.13 0.62 0.12 0.62 0.27 0.18 0.52 <0.01 ( -0.40, 0.66)  ( -0.35, 0.60)  (-0.12, 0.66)  (0.17, 0.86)  Sodium (mg) 38.0 0.60 12.83 0.84 74.7 0.15 120 0.01 ( -105, 181)  (-115, 141)  ( -27.4, 177)  ( 35.7, 204)  Potassium (mg) 41.4 0.52 85.03 0.16 71.3 0.18 172 <0.01  ( -84.5, 167)  ( -33.9, 204)  ( -33.3, 176)  (76.5, 268)  Note. n=34,402 excluding pregnant and breast feeding women. β represents linear regression coefficient.  86   Appendix B Recommended number of food group servings and unsaturated fat intake per day according to 2007 Canada’s Food Guide, by age group and sex  Age group (years)  2-3 4 -8 9-13 14 -18 19- 30 31-50 50+ Male        Vegetables and fruit 4 5 6 8 10 8 7 Grain products 3 4 6 7 8 8 7 Milk and alternatives 2 2 3-4 3-4 2 2 3 Meat and alternatives 1 1 2 3 3 3 3 Unsaturated fats (grams) 30 30 30 45 45 45 45  Female Vegetables and fruit 4 5 6 7 8 7 7 Grain products 3 4 6 6 7 6 6 Milk and alternatives 2 2 3-4 3-4 2 2 3 Meat and alternatives 1 1 1 2 2 2 2 Unsaturated fats (grams) 30 30 30 30 30 30 30 (Health Canada, 2007a)

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0073762/manifest

Comment

Related Items