European Journal of Clinical Nutrition (2006) 60, 494–501
& 2006 Nature Publishing Group All rights reserved 0954-3007/06 $30.00 www.nature.com/ejcn
ORIGINAL ARTICLE
Snacks as an element of energy intake and food consumption M-L Ovaskainen, H Reinivuo, H Tapanainen, M-L Hannila, T Korhonen and H Pakkala Department of Epidemiology and Health Promotion, Nutrition Unit, National Public Health Institute, Helsinki, Finland
Background: An increasing frequency of snacks has been observed in meal pattern studies. Snacks can alter the diet because of their high-energy density and low-nutrient content or on the contrary. Objective: The prominence of snacks in energy intake and food consumption was assessed. Design: Dietary data were collected for 2007 adults by using a computer-assisted 48-h dietary recall in the national FINDIET 2002 survey. Energy intakes and food consumption were aggregated for snacks and for main meals. Results: Daily energy was mostly derived from main meals comprising traditional mixed dishes, milk and bread. However, a snack-dominating meal pattern was observed in 19% of men and 24% of women. This meal pattern was associated with urbanization in both genders and with physical work in men. Higher sucrose intake and lower intake of micronutrients were typical of the snack-dominating meal pattern compared to the others. Conclusions: As snacks appear to have a higher energy density and a lower content of micronutrients than main meals, a snackdominating meal pattern is inadvisable. However, further studies are needed to examine the association between meal pattern and health status.
European Journal of Clinical Nutrition (2006) 60, 494–501. doi:10.1038/sj.ejcn.1602343; published online 30 November 2005 Keywords: snack; main meal; eating frequency; energy intake; food consumption
Introduction Meals are everyday eating events structured by the frequency of food consumption, by food items appropriate to a meal and even by the order of dishes in the menu (Douglas and Gross, 1981; Ma¨kela¨, 2000; Poulain, 2002). Eating events have typically been categorized as either main meals (structured meals) or snacks (unstructured eating events between meals). These categories have been used in both cultural (Ma¨kela¨, 2000) and nutritional studies (Kearney et al., 2001; Poulain, 2002). Traditionally, and still in the 1990s, the European meal pattern has been based on three Correspondence: Dr M-L Ovaskainen, Department of Epidemiology and Health Promotion, Nutrition Unit, National Public Health Institute, Mannerheimintie 166, 00300 Helsinki, Finland. E-mail:
[email protected] Guarantor: M-L Ovaskainen. Contributors: M-LO contributed to study design, analysis and writing. HR contributed to data collection and writing. HT contributed to data analysis and writing. M-LH contributed to data analysis and writing. TK contributed to study design and data collection. HP contributed to study design, data collection and writing. Received 28 January 2005; revised 6 September 2005; accepted 3 October 2005; published online 30 November 2005
main meals, including breakfast (Bellisle et al., 1997; Roos and Pra¨tta¨la¨, 1997; Winkler et al., 1999). In Finland and Sweden, two hot meals per day have been more frequent than in the other Nordic countries (Ma¨kela¨ et al., 1999). However, snacks are becoming increasingly popular (Poulain, 2002; Devine et al., 2003) and may be added to the diet or consumed in place of traditional meals (Bellisle et al., 1997; Poulain, 2002). The concept of snacks is complex; it may comprise confectionery items or beverages only ¨ ssner, 1996), ‘a snacking food’ (like chips) (Andersson and Ro or even light meals (Poulain, 2002; Chamontin et al., 2003). The effect of snacking on dietary intakes has been examined (Longnecker et al., 1997; Roos and Pra¨tta¨la¨, 1997; Poppitt et al., 1998; Legg et al., 2000; Bellisle et al., 2003; Chamontin et al., 2003; Hampl et al., 2003), but results are contradictory. The average daily food consumption and nutrient intake have been considered the most relevant and achievable estimates of diets in epidemiological studies (Aro et al., 1998; Heitmann et al., 2000). These approaches, however, neglect the temporal distribution of eating events and the combinations of food items at each meal. Food-based guidelines for consumers may include preventive messages separately for the energy content of meals (Poppitt et al., 1998; Ma¨kela¨,
Snacks – an element of energy and food intake M-L Ovaskainen et al
495 2000; Poulain, 2002). Information on food consumption and energy intake at meals can be used to develop a dietary plan and to give examples for recommendations (Kearney et al., 2001; National Nutrition Council, 2003; NNR, 2004). In an earlier Finnish study, Roos and Pra¨tta¨la¨ (1997) found the number of meals not to be associated with either quality or quantity of fat intake. In other studies, however, increased meal frequency has been associated with decreased cholesterol concentrations (Edelstein et al., 1992), body-weight control (Kirk, 2000) and also with obesity (Berte´us Forslund et al., 2005). The increasing frequency of snacks has been shown to correlate positively with total energy intake (Hampl et al., 2003). Certain foods like potatoes may be included into the diet only by main meals (Poppitt et al., 1998; Kearney et al., 2001) and therefore the concept of a proper diet is strongly associated with main meals (Poulain, 2002). In addition, under-reporting of food consumption has increased in dietary surveys’ being more frequent with snack-type eating events (Poppitt et al., 1998; Heitmann et al., 2000; Lafay et al., 2000) and with a varying number of snacks (Tooze et al., 2004). For these reasons, the contribution of snacks to total energy intake should be ascertained. The aim of this study was to examine the prominence of snacks in the Finnish diet and to determine which foods are selected for snacks. Moreover, we wanted to establish the proportion of energy derived by a Finnish adult from snacks as opposed to main meals on an average day. The social background of persons with a snack-dominating meal pattern was also studied.
Subjects and methods A cross-sectional population survey (FINRISK 2002) was carried out in spring 2002 to assess risk factors of cardiovascular diseases (Laatikainen et al., 2003). A random sample (n ¼ 12 000) of adults aged 25–64 years stratified by sex, region and 10-year age group was drawn from the population register for six regions in Finland. Subjects were invited to a health examination (including body weight measurement) at the local health care centre and asked to complete a questionnaire covering such background information as general habits, habitual eating frequency, education and professional status. Subjects in a random subsample (n ¼ 3181) from five regions were asked to participate in a ¨ et al., 48-h dietary interview (FINDIET 2002 by Ma¨nnisto 2003). Of those invited, 64% participated, and 98% of these interviews were subsequently accepted (n ¼ 2007). The dietary interview covered all weekdays except Fridays. The FINDIET 2002 data for all subjects, including pregnant (n ¼ 17) and lactating women, were used in analyses.
Dietary interview protocol Details of the dietary interview have been published earlier ¨ et al., 2003). Dietary data were collected by a (Ma¨nnisto
computerized multiphased 48-h recall by trained nutritionists. A list of eating events with their times and places was compiled for each day. A detailed inventory of all food items and in each eating event followed, starting from the previous day and continuing to the day before yesterday. The name, food code, portion size and additional comments on each food and beverage were coded immediately in the dietary database according to standardized coding rules using in¨ et al., 2003). The food composihouse software (Ma¨nnisto tion database included 2100 foods and composite dishes. Portion sizes were estimated using food models, regular packages, household utensils and a picture booklet (Pietinen et al., 1988). A written copy of each interview was produced, with participants checking all information at the end of the interview. The Finnish food composition database (KTL, 2002) was used to calculate energy and nutrient intakes and aggregate food consumption. The grouping of individual foods and beverages was modified according to earlier meal pattern studies (Roos and Pra¨tta¨la¨, 1997; Lafay et al., 2000; Bellisle et al., 2003; Chamontin et al., 2003).
Concepts of meal type and meal pattern Identification of each eating event was based either on the time, the name or another description of the eating event given by the subject. According to coding rules, each eating event, even if it consisted of only a cup of coffee, was identified separately according to time. In ambivalent cases, the interviewer would clarify the name during the discussion. The interviewer selected the meal name from the following alternatives: breakfast, lunch, dinner, drink, evening snack, other snack and other eating event. To simplify the variety of meals, meal type was then divided into two categories: main meals and snacks. Main meals comprised breakfast, lunch and dinner. All eating events between main meals were considered snacks. Energy intake and food consumption were aggregated for the two meal types (snacks and main meals), and the average intake over two days was calculated. The distribution of mean energy intake from snacks and main meals was described graphically over 24 h in 2-h intervals. To separate daytime and evening intakes, cut-off times of 0500 and 1700 were used (Stroebele and de Castro, 2003). Energy density was defined as energy content per unit weight of food (kJ/100 g, Cuco´ et al., 2001). The average energy density for snacks and main meals was calculated in two ways: for all foods and beverages and for foods and energy-containing beverages. The difference in mean energy intake (kJ) between main meals and snacks was calculated to enable subjects to be grouped according to meal pattern. A negative difference (energy difference ¼ mean energy intake from main mealsmean energy intake from snacks) was coded as a snack-dominating meal pattern. Daily nutrient intake was used to describe the diet by meal pattern. Background factors and also the habitual frequency European Journal of Clinical Nutrition
Snacks – an element of energy and food intake M-L Ovaskainen et al
496 of eating events from the background questionnaire were used as explanatory variables for a snack-dominating meal pattern.
Statistical methods Food items were aggregated into 17 food groups to describe proportion of consumers by food group and to calculate the mean food consumption (g/day). In addition, the mean contribution of food groups to the energy content of snacks and main meals was determined. Analysis of variance was used to assess the difference in dietary intake between genders, the difference in energy density between meal types (within-subject effect) and the difference in nutrient intake between meal patterns for both genders. The day-to-day correlation coefficients for energy intakes were produced with the mixed linear model. A logistic regression model was used to analyse the association between background factors and the snackdominating meal pattern. The regions were aggregated as Southern Finland (including the capital region of Helsinki and Turku with its surroundings) and Northern Finland (including North-Karelia, Oulu and Northern Savo). The data were analysed using the SAS statistical package (SAS, 1999). All analyses were performed separately for men and women.
Table 1
Characteristics of FINDIET 2002 participants Men n (%)
Women n (%)
912 (100)
1095 (100)
190 (21) 722 (79)
263 (24) 832 (76)
Region Southern Finland Northern Finland
334 (37) 578 (63)
416 (38) 679 (62)
Professional status Nonmanual worker, student Manual worker, farmer Retired, unemployed, housewife Missing
417 274 213 8
715 95 269 16
Body mass index (BMI kg/m2) o25 (normal weight) 25–30 (overweight) 430 (obese)
286 (31) 426 (47) 200 (22)
Total n Age 25–34 years 35–64 years
(46) (30) (23) (1)
(65) (9) (25) (1)
531 (48) 354 (32) 210 (19)
Persons according to different eating events on at least 1 of 2 days in the 48-h dietary recall Breakfast 846 (93) 1057 (97) Lunch 791 (87) 975 (89) Dinner 716 (79) 810 (74) Drink 765 (84) 851 (78) Evening snack 770 (84) 972 (89) Other snack 833 (91) 1062 (97) Other eating event 40 (4) 85 (8)
Results Background factors and degree of obesity of FINDIET 2002 participants are presented in Table 1. The proportions of persons according to different eating event per day are also presented (Table 1). Almost all participants had breakfast on at least one day (93% of men, 97% of women). Based on 48-h dietary recalls, Finnish men had an average of 6.0 (s.e. 0.05) and women 6.1 (s.e. 0.05) eating events per day. The mean (s.e.) number of snacks was 3.7 (0.05) in men and 3.8 (0.05) in women. The average daily energy intake and fat content (as energy %) of the diet were higher in men than in women (Table 2). Sucrose content was, however, higher in the women’s diet. The 24-h energy intake is displayed graphically in Figure 1. The energy intake from main meals had two peaks, at lunch and at dinner. Energy from snacks was mainly obtained during the evenings and nights. During the daytime energy was mostly derived from main meals, both for men and women (Figures 1 and 2). Participants reported consumption of 1554 different food items, and the mean number of daily food items consumed was 25 (range 3–49). The following food groups were represented in snacks for at least half of the Finnish adults: coffee or tea, milk products, bread, water, sweets, meat or fish, fats, sweet bakery goods and fruit or fresh vegetables (Table 3). European Journal of Clinical Nutrition
Table 2
Dietary intakes in FINDIET 2002 study populationa Men (n ¼ 912)
Women F-value(df1, (n ¼ 1095)
df2)
Mean (s.e.) Mean (s.e.) Energy intake, MJ/day 9.16 (0.09) 6.62 (0.06) 557.9(1,1994) b Fat content of the diet, E% 34.9 (0.3) 32.4 (0.2) 47.8(1,1991) b Sucrose content of the diet, E% 9.1 (0.2) 10.8 (0.1) 51.5(1,1991) b a
Detailed information published earlier (Ma¨nnisto¨ et al., 2003). Po0.0001 for the difference between genders in analysis of variance models; other explanatory variables are age and region.
b
Most energy intake at snacks was derived from sweet bakery goods (14%), bread (14%), alcoholic beverages (13%) and milk products (12%) in men, and sweet bakery goods (18%), bread (14%), milk products (12%) and sweets and chocolate (19%) in women (Table 4). At main meals, meat and fish dishes, potatoes and cooked vegetables contributed to energy intake. The energy density (kJ/100 g) of foods and energy-containing beverages was higher in snacks than in main meals (Table 5). The day-to-day correlation coefficient for snack energy intakes was 0.19 in men and 0.37 in women, while for total energy intakes it was 0.25 in both genders.
Snacks – an element of energy and food intake M-L Ovaskainen et al
497 Table 3 Description of snacks in Finnish adults (FINDIET 2002): proportions of consumers and mean consumption of selected food groups Food groupa
Men (n ¼ 912)
Women (n ¼ 1095)
Consumers Mean Consumers Mean % consumption % consumption g/day (s.e.) g/day (s.e.)
Figure 1 Mean energy intake in 2-h time blocks contributed by snacks and main meals in Finnish adults.
MEN, total
Coffee and Tea Milk products Bread Water Sweets, chocolate Meat or fish Fats and dressing Sweet bakery goods Fruit, fresh vegetables Juice Alcoholic beverages Puddings, berry soups Pizza, burger, pasty Soft drinks Pasta, rice, porridge Potato, cooked vegetables Salty snacking foods a
Daytime
91 81 74 73 71 63 61 57 57 35 34 27
409 123 52 319 17 36 9 33 91 82 197 24
(11.8) (5.4) (1.8) (13.4) (0.9) (1.9) (0.5) (1.5) (4.3) (5.3) (18.3) (2.1)
91 84 79 87 66 53 57 69 77 39 19 39
321 114 41 503 17 18 6 32 129 74 47 31
(7.7) (4.0) (1.2) (13.9) (0.9) (1.0) (0.3) (1.1) (4.0) (4.4) (5.0) (1.8)
19 18 14 12
16 53 12 7
(1.3) (5.4) (1.4) (0.8)
21 12 20 14
12 26 14 6
(0.9) (3.1) (1.3) (0.7)
2 (0.4)
11
7
2 (0.3)
Order dictated by the proportion of men as consumers.
Table 4 Contribution (%) of selected food groups to energy intakes at snacks and main meals (FINDIET 2002) Food groupa
Evening
Men (n ¼ 912)
Women (n ¼ 1095)
Snacks Main meals Snacks Main meals Energy (%) Energy (%) Energy (%) Energy (%)
WOMEN, total
Daytime
Evening
0
1
2
3
4
5
6
7
Energy intake (MJ/day) Figure 2 Mean energy intake (95% confidence interval) contributed by snacks and main meals in the daytime and in the evening in Finnish adults (filled circles – main meals, open circles – snacks).
Sweet bakery goods Bread Alcoholic beverages Milk products Meat or fish Sweets, chocolate Fats and dressings Pizza, burger, pasty Puddings, berry soup Fruit, fresh vegetables Juice Pasta, rice, porridge Soft drinks Salty snacking foods Coffee and tea Potato, cooked vegetables Water Total a
Meal patterns were identified by the difference in mean energy intake (kJ) between main meals and snacks. A negative difference indicated a higher energy intake from snacks and was recorded as a snack-dominating meal pattern (Figure 3). One fifth of men and one fourth of women were
14 14 13 12 9 8 6 5 4 4 4 2 3 1 1 1 0
2 17 1 14 28 2 9 5 2 2 2 6 0 0 0 10 0
18 14 4 14 5 10 5 4 6 7 4 3 1 2 1 1 0
3 16 1 15 25 2 7 5 4 3 3 8 0 0 0 10 0
100
100
100
100
Food group contributing to energy intake at snacks.
found to have a snack-dominating meal pattern. In men, daily energy intake was higher in those with a snackdominating meal pattern than in the others (Table 6). Daily European Journal of Clinical Nutrition
Snacks – an element of energy and food intake M-L Ovaskainen et al
498 Table 5 Energy intake and energy densities of snacks and main meals Men (n ¼ 912) Snacks Mean (s.e.) Total energy intake, kJ Energy density of foods and energy-containing beverages, kJ/100 g Energy density of foods and beverages, kJ/100 g
Main meals Mean (s.e.)
Women (n ¼ 1095)
F-value(df1,
df2) a
Snacks Mean (s.e.)
Main meals Mean (s.e.)
F-value(df1,
df2)
3305 (68) 532 (9)
5854 (74) 483 (4)
569(1,904) 24(1,886) a
2632 (45) 522 (8)
3987 (46) 436 (3)
413(1,1081) a 131(1,1082) a
231 (3)
359 (3)
994(1,889) a
203 (3)
293 (3)
693(1,1082) a
a
Po0.0001 in repeated measures analysis of variance models. Meal type is a within-subject effect, and region and age group are between-subject effects.
changed the determinants of the snack-dominating meal pattern.
0.20 Males Females
Discussion
0.15
0.10
0.05
0.00 -15
-10
-5 0 5 10 Difference in energy intake (MJ)
15
Figure 3 Empirical density function of Finnish adults for the difference in energy intake between main meals and snacks (energy difference (MJ) ¼ mean energy from main meals (MJ) – mean energy from snacks (MJ)). Negative value for the difference in energy intake ¼ a snack-dominating meal pattern.
nutrient intakes were given as absolute intakes, and in terms of intake per energy unit. Both men and women with a snack-dominating meal pattern had higher intakes (/MJ) of alcohol, sugars and sucrose and lower intake of several micronutrients (Table 6). Sodium content of the diet was also lower for those with a snack-dominating meal pattern than the others. Determinants of the snack-dominating meal pattern were region of residence in both men and women, professional status in men and age in women (Table 7). This meal pattern was more typical in the southern than in the northern regions of Finland for both genders and was more frequent among male manual workers and farmers than among male nonmanual workers. The snack-dominating meal pattern was more prevalent in younger women than in older women. Neither self-reported number of habitual daily meals (data from baseline questionnaire) nor education European Journal of Clinical Nutrition
Main meals in the 2000s are still the primary source of dietary energy, consistent with previous observations made in Finland (Roos and Pra¨tta¨la¨, 1997; Ma¨kela¨, 2000) and elsewhere (Winkler et al., 1999; Kearney et al., 2001; Poulain, 2002; Bellisle et al., 2003). More energy (36% in men, 40% in women, Table 4) is now, however, derived from snacks than previously reported (33% in both men and women, Roos and Pra¨tta¨la¨, 1997). This result implies a tendency for higher contribution of energy intake from snacks. The snack-dominating meal pattern was introduced in this study to indicate the relatively larger energy intake derived from snacks than from main meals. This meal pattern was observed in one-fifth of men and one-fourth of women, and it was more typical in urban Southern Finland than in the other study regions. New food trends have usually started in Southern Finland (Roos and Pra¨tta¨la¨, 1997), where the availability of different foods and dining places is much wider than elsewhere. Poulain (2002) has earlier reported simplified eating events to be prevalent in urban populations, particularly among females and white-collar workers. Conversely, in our study, the snack-dominating meal pattern was more frequent in male manual workers and farmers than in male nonmanual workers. A snack-dominating meal pattern may be caused among male manual workers by irregular working times, the lack of work canteens and larger energy requirements. Consistent with this, having lunch at a workplace cafeteria has been demonstrated to be associated with higher education in employed Finnish adults (Raulio et al., 2005). Food consumption and food items at main meals and at snacks may differ by country. We found the energy contribution from sweet bakery goods, other sweets and chocolate and alcoholic beverages to be higher in snacks than in main meals, a finding also reported by Bellisle et al. (2003). In France, by contrast, tea and coffee were more typical at main meals than at snacks, while milk products were as typical at meals as in Finland. Cultural traditions
Snacks – an element of energy and food intake M-L Ovaskainen et al
499 Table 6 Nutrient content of the diet by meal pattern, mean (s.e.) Men
Women
Main meal-dominating diet n ¼ 741 a
Snack-dominating diet n ¼ 171 a
Main meal-dominating diet n ¼ 829 a
Snack-dominating diet n ¼ 266 a
Energy, MJ Energy of foods, MJ
9.02 (0.10)** 8.89 (0.09)0
9.77 (0.25)** 9.02 (0.21)0
6.56 (0.07) 6.48 (0.07)0
6.80 (0.14) 6.64 (0.14)0
Fat, g Fat, g/MJ
86.8 (1.3) 9.50 (0.07)*
88.8 (2.9) 9.08 (0.19)*
58.4 (0.8) 8.78 (0.07)
60.1 (1.7) 8.71 (0.13)
Protein, g Protein, g/MJ
87.2 (1.1) 9.77 (0.07)**
83.1 (2.1) 8.76 (0.17)**
64.2 (0.7)** 9.93 (0.07)**
60.4 (1.3)** 9.07 (0.14)**
Carbohydrates, g Carbohydrates, g/MJ
237.0 (2.9)* 26.5 (0.2)
251.4 (6.7)* 26.2 (0.4)
Alcohol, g Alcohol, g/MJ
8.3 (0.7)** 0.87 (0.06)**
25.9 (4.2)** 2.24 (0.30)**
Vitamin C, mg Vitamin C, mg/MJ
95.3 (2.9)** 11.0 (0.3)**
74.3 (5.0)** 8.4 (0.6)**
Vitamin A, mg Vitamin A, mg/MJ
1082 (61) 122 (6)*
854 (50) 92 (6)*
Vitamin E, mg Vitamin E, mg/MJ
11.94 (0.21) 1.34 (0.02)**
11.26 (0.39) 1.18 (0.04)**
Vitamin D, mg Vitamin D, mg/MJ
5.84 (0.23) 0.65 (0.02)
Fibre, g Fibre, g/MJ Sugars, g Sugars, g/MJ
186.6 (2.1)* 28.5 (0.2)* 3.0 (0.3)** 0.43 (0.04)**
196.3 (4.2)* 29.1 (0.3)* 5.2 (0.8)** 0.72 (0.1)**
107.1 (2.5) 17.2 (0.4)
99.6 (4.0) 16.1 (0.7)
948 (41) 151 (7)0
858 (56) 125 (7)0
9.06 (0.16)0 1.39 (0.02)**
8.52 (0.23)0 1.27 (0.03)**
5.49 (0.43) 0.61 (0.06)
3.90 (0.15) 0.61 (0.02)
3.29 (0.19) 0.50 (0.03)
22.15 (0.38) 2.52 (0.04)**
20.12 (0.81) 2.17 (0.09)**
18.74 (0.28) 2.94 (0.04)
17.72 (0.48) 2.71 (0.07)
101.8 (1.8)** 11.3 (0.1)**
123.6 (4.8)** 12.7 (0.4)**
85.8 (1.3)** 13.0 (0.1)**
98.7 (2.7)** 14.6 (0.3)**
Sucrose, g Sucrose, g/MJ
47.3 (1.1)** 5.18 (0.10)**
58.0 (3.1)** 6.01 (0.31)**
40.8 (0.8)** 6.07 (0.10)**
49.8 (1.9)** 7.13 (0.19)**
Calcium, mg Calcium, mg/MJ
1196 (22) 133 (2)**
1148 (41) 121 (4)**
966 (14) 150 (2)
986 (27) 147 (3)
Potassium (K), mg Potassium (K), mg/MJ
4044 (44)* 462 (4)**
3794 (87)* 404 (8)**
3225 (32)* 508 (4)**
3059 (52)* 473 (8)**
Sodium (Na), mg Sodium (Na), mg/MJ
3992 (52)* 448 (4)**
3636 (92)* 385 (9)**
2764 (32)** 429 (4)**
2544 (53)** 383 (6)**
Iron, mg Iron, mg/MJ
13.39 (0.21) 1.51 (0.02)**
12.37 (0.38) 1.33 (0.04)**
10.12 (0.14) 1.57 (0.02)**
9.7 (0.2) 1.45 (0.02)**
Magnesium, mg Magnesium, mg/MJ
403.6 (4.6) 45.9 (0.4)**
408.3 (11.2) 43.0 (0.8)**
310.9 (3.2) 48.7 (0.4)
a0
304.5 (5.3) 46.7 (0.7)
/*/**P-value 0.05–0.1/0.01–0.05/o0.01 for the difference between meal patterns in analysis of variance model; other explanatory variables were age and area.
obviously play a role in the food consumption at meals and meal rules may be more stable in some food cultures (Poulain, 2002; Bellisle et al., 2003) than in the Finnish or British (Chamontin et al., 2003) food culture. Main courses with meat or fish, a staple food and vegetables were typical of main meals in Finland, similar to main meals in France (Poulain, 2002). Women (more so than men) in Finland
preferred fruit and fresh vegetables also for snacks, consistent with reports in some other cultures (Kearney et al., 2001; Bellisle et al., 2003; Devine et al., 2003). The energy density (MJ/100 g) of all foods and energycontaining beverages eaten at snacks was higher than that of main meals. However, lower intakes of micronutrients were observed in persons with a snack-dominating meal pattern European Journal of Clinical Nutrition
Snacks – an element of energy and food intake M-L Ovaskainen et al
500 Table 7 Odds ratios for a snack-dominating meal pattern in 48-h dietary recalls of Finnish adults by a logistic regression model Men (n ¼ 912) OR a (95% CI)
Women (n ¼ 1095) OR a (95% CI)
Age Older (35–64 years) Younger (25–34 years)
1.00 1.06 (0.70–1.58)
1.00 1.41 (1.03–1.93)
Region Northern Finland Southern Finland
1.00 1.60 (1.14–2.26)
1.00 1.43 (1.08–1.89)
1.00 1.52 (1.03–2.23) 1.01 (0.65–1.57)
1.00 0.96 (0.59–1.58) 0.78 (0.56–1.09)
1.04 (0.69–1.58) 1.71 (1.21–2.43) 1.68 (1.13–2.49)
1.38 (1.00–1.89) 1.46 (1.05–1.86) 1.04 (0.63–1.71)
1.12 (0.71–1.76)
0.86 (0.61–1.20)
Professional status Nonmanual worker, student Manual worker, farmer Retired, unemployed, housewife, other Total model Younger vs older Southern vs Northern Manual worker, farmer vs nonmanual worker, student Retired, unemployed, housewife vs nonmanual worker, student a
Adjusted for average daily energy intake.
compared with the others, which supports earlier results regarding the composition of snacks (Roos and Pra¨tta¨la¨, 1997; Poppitt et al., 1998; Lennerna¨s and Andersson, 1999; Stroebele and de Castro, 2003). Although, intake differences were slight, the associations between snack-dominating meal pattern and health status warrant further studies. Meal identification can be based on subjects’ descriptions of the eating event, as in this study and in some previous reports (Winkler et al., 1999; Legg et al., 2000; Bellisle et al., 2003; Berte´us Forslund et al., 2005). Other possibilities for identifying meal type have been based on the food ¨ ssner, 1996; composition of the meal (Andersson and Ro Lennerna¨s and Andersson, 1999; Poulain, 2002) or on the amounts consumed (Legg et al., 2000). In an earlier Finnish study (Roos and Pra¨tta¨la¨, 1997), for example, breakfast with a warm dish was included in main meals, but breakfast composed of beverages only was included in snacks. We did not use information about food composition of meals to define meal types in this study. Defining breakfast as a main meal may thus lead to an underestimation in the number of snacks and in the energy intake from snacks. On the other hand, foods eaten in the evenings more often being defined as an evening snack may lead to an overestimation in the energy intake from snacks because distinguishing between an evening snack and a late dinner is not always possible. The difficulty in discriminating light meals from snacks has also been reported in other studies (Berte´us Forslund et al., 2002; Poulain, 2002; Chamontin et al., 2003; Devine et al., 2003). European Journal of Clinical Nutrition
This study used quantitative data collected by 48-h dietary recalls (two 24 h recalls) for meal pattern analyses (Lafay et al., 2000; Ma¨kela¨, 2000). The number of daily eating events was higher (about six) than figures reported in earlier studies in Finland (Roos and Pra¨tta¨la¨, 1997; Ma¨kela¨ et al., 1999) or in other countries (Longnecker et al., 1997; Winkler et al., 1999; Bellisle et al., 2003). Similar daily frequencies ¨ ssner, have been reported only in Sweden (Andersson and Ro 1996; Berte´us Forslund et al., 2002). The multiphase interviewing procedure used here enabled complete recalling of eating events during the 2 days. However, habitual meal frequency cannot be estimated well with only two days of dietary recalls even for those with a regular meal pattern (Longnecker et al., 1997; Roos and Pra¨tta¨la¨, 1997; Buzzard, 1998). When we controlled for the effect of habitual meal frequency (from general habits in the baseline questionnaire) it confirmed the result of a snack-dominating meal pattern. Thus, the method of 48-h dietary recall proved to be sufficient for meal pattern analysis. The effects of weekdays and working status on meal frequency of interviewed days were not studied here. While the number of eating occasions in this study may be accurate, the amounts or the number of food items may be under-reported. Legg et al. (2000) noted that the sizes of meals (especially breakfasts) over several days correlated moderately, but total intakes varied greatly. Two consecutive study days, as used here, may therefore have interdependencies with the number of eating events and the types of foods consumed (Longnecker et al., 1997; Lafay et al., 2000; Legg et al., 2000). Variation in the diet may be underestimated for this reason, as well as because information on seasonal differences was missing. Under-reporting of snacks, particularly for those having frequent snacks (Tooze et al., 2004), has been found to be more common than under-reporting of main meals (Bellisle et al., 1997; Poppitt et al., 1998; de Castro, 2000; Lafay et al., 2000). Coverage of snacks in the interviewing procedure should therefore be controlled with an adequate number of questions and memory aids (Buzzard, 1998). Differences between participants and nonparticipants have been studied in earlier FINRISK surveys (Jousilahti et al., 2005). The study population here represented 64% of persons invited, and nonparticipation has previously shown to be associated with high mortality (Jousilahti et al., 2005). Thus, the poor participation probably limits the generalization of our results. We found differences in a snackdominating meal pattern by age only for women, although irregular eating events are typical of young adults in both sexes (Kearney et al., 2001). Modern sedentary lifestyle combined with more eating events (Berte´us Forslund et al., 2002) may lead to excessive weight, but the evidence has thus far been weak (Bellisle et al., 1997) or nonexistent (Hampl et al., 2003). Restrained eating has not been noted (Legg et al., 2000) to affect energy intakes of snacks and main meals differently. In an earlier study (Hampl et al., 2003), ‘multiple-snackers’ were observed to consume more energy
Snacks – an element of energy and food intake M-L Ovaskainen et al
501 than ‘never-snackers’. The associations between frequent snacks, lifestyle, health status, obesity and dietary restraint warrant further studies. In conclusion, a snack-dominating meal pattern is incompatible with nutrition recommendations (NNR, 2004) due to selection of nutrient-poor food items.
References ¨ ssner S (1996). Meal patterns in obese and normal Andersson I, Ro weight men: The ‘Gustaf’ study. Eur J Clin Nutr 50, 639–646. Aro A, Pietinen P, Valsta L, Turpeinen A, Ehnholm C, Dougherty RM et al. (1998). Effects of reduced-fat diets with different fatty acid compositions on serum lipoprotein lipids and apolipoproteins. Public Health Nutr 1, 109–116. Bellisle F, Dalix A, Mennen L, Galan P, Hercberg S, de Castro JM et al. (2003). Contribution of snacks and meals in the diet of French adults: a diet-dairy study. Phys Behav 79, 183–189. Bellisle F, McDevitt R, Prentice A (1997). Meal frequency and energy balance. Br J Nutr 77 (Suppl 1), S57–S70. ¨ stro ¨ m L, Lissner L (2002). Meal Berte´us Forslund H, Lindroos A, Sjo pattern and obesity in Swedish women – simple instrument describing usual meal types, frequency and temporal distribution. Eur J Clin Nutr 56, 740–747. ¨ stro ¨ m L, Lindroos A (2005). Berte´us Forslund H, Torgerson J, Sjo Snacking frequency in relation to energy intake and food choices in obese men and women compared to a reference population. Int J Obes Relat Metab Disord 29, 711–719. Buzzard M (1998). 24-h dietary recall and food record methods. In: W Willett (ed.), Nutritional epidemiology. New York: Oxford University Press, pp. 50–73. Chamontin A, Pretzer G, Booth D (2003). Ambiguity of ‘snack’ in British usage. Appetite 41, 21–29. Cuco´ G, Arija V, Martı´-Heneberg C, Ferna´ndez-Ballart J (2001). Food and nutritional profile of high energy density consumers in an adult Mediterranean population. Eur J Clin Nutr 55, 192–199. de Castro J (2000). Eating behavior: lessons from the real world of humans. Nutrition 16, 800–813. Devine C, Connors M, Sobal J, Bisogni C (2003). Sandwiching it in: spillover of work onto food choices and family roles in low- and moderate-income urban households. Soc Sci Med 56, 617–630. Douglas M, Gross J (1981). Food and culture: Measuring the intricacy of rule systems. Soc Sci Inform 20, 1–35. Edelstein S, Barrett-Connor E, Wingard D, Cohn B (1992). Increased meal frequency associated with decreased cholesterol concentrations; Rancho Bernardo, CA, 1984–1987. Am J Clin Nutr 55, 664–669. Hampl J, Heaton C, Taylor C (2003). Snacking patterns influence energy and nutrient intakes but not body mass index. J Human Nutr Dietet 16, 3–11. Heitmann B, Lissner L, Osler M (2000). Do we eat less fat, or just report so? Int J Obes 24, 435–442. Jousilahti P, Salomaa V, Kuulasmaa K, Niemela¨ M, Vartiainen E (2005). Total and cause specific mortality among participants and non-participants of population based health surveys: a comprehensive follow up of 54 372 Finnish men and women. J Epidemiol Community Health 59, 310–315. Kearney J, Hulshof K, Gibney M (2001). Eating patterns – temporal distribution, converging and diverging foods, meals eaten outside and inside of the home – implications for developing FBDG. Public Health Nutr 4, 693–698.
Kirk T (2000). Role of dietary carbohydrate and frequent eating in body-weight control. Proc Nutr Soc 59, 349–358. KTL (National Public Health Institute) (2002). Nutrition Unit. Finnish food composition database Finelis (version 2). Latest version at http://www.fineli.fi. Laatikainen T, Tapanainen H, Alfthan G, Salminen I, Sundvall J, Leiviska¨ J et al. (2003). FINRISK 2002. Survey implementation and results 1. Helsinki: Publications of the National Public Health Institute Report No B7/2003 http://www.ktl.fi/attachments/suomi/julkaisut/julkaisusarja_b/2003b7–1.pdf. Lafay L, Mennen L, Basdevant A, Charles M, Borys J, Eschwe`ge E, et al., the FLVS study group (2000). Does energy intake underreporting involve all kinds of food or only specific food items? Results from the Fleurbaix Laventie Ville Sante´ (FLVS) study. Int J Obes 24, 1500–1506. Legg C, Puri A, Thomas N (2000). Dietary restraint and self-reported meal sizes: diary studies with differentially informed consent. Appetite 34, 235–243. Lennerna¨s M, Andersson I (1999). Food-based classification of eating episodes (FBCE). Appetite 32, 53–65. Longnecker M, Harper J, Kim S (1997). Eating frequency in the nationwide food consumption survey (USA), 1987–1988. Appetite 29, 55–59. Ma¨kela¨ J (2000). Cultural definitions of the meal. In: H Meiselman (ed.), Dimensions of the Meal. Gaithersburg: Aspen Publishers Inc., pp. 7–18. ¨ m M, L’orange´ Fu ¨ rst E, Gronow J, Holm Ma¨kela¨ J, Kjærnaes U, Ekstro L (1999). Nordic meals: Methodological notes on a comparative survey. Appetite 32, 73–79. ¨ S, Ovaskainen M-L, Valsta L (eds) (2003). The national Ma¨nnisto Findiet 2002 study. Publications of the National Public Health Institute Report No B3/2003 (in Finnish, with English tables, figures and summaries). http://www.ktl.fi/julkaisut/julkaisusarja_b/ 2003b3. National Nutrition Council (2003). Action program for implementing national nutrition recommendations (in Finnish with an English abstract). Ministry of Agriculture and Forestry, Helsinki. NNR (Nordic Nutrition recommendations) (2004). Nord 2004;13. Copenhagen: Nordic Council of Ministers. Pietinen P, Hartman A, Haapa E, Ra¨sa¨nen L, Haapakoski J, Palmgren J et al. (1988). Reproducibility and validity of dietary assessment instruments. I. A self-administered food use questionnaire with a portion size picture booklet. Am J Epidem 128, 655–666. Poppitt S, Swann D, Black A, Prentice A (1998). Assessment of selective under-reporting of food intake by both obese and nonobese women in a metabolic facility. Int J Obes 22, 303–311. Poulain J (2002). The contemporary diet in France: ‘de-structuration’ or from commensalism to ‘vagabond feeding’. Appetite 39, 43–55. Raulio S, Roos E, Rahkonen O, Pra¨tta¨la¨ R (2005). Twenty-year trends of workplace lunches in Finland. Food Service Technol 5, (in press). Roos E, Pra¨tta¨la¨ R (1997). Meal patterns and nutrient intake among adult Finns. Appetite 29, 11–24. SAS (1999). Statistical package, Version 8.2. Gary, N.C: SAS Institute Inc. Stroebele N, de Castro J (2003). Television viewing is associated with an increase in meal frequency in human. Appetite 42, 111–113. Tooze J, Subar A, Thompson F, Troiano R, Schatzkin A, Kipnis V (2004). Psychosocial predictors of energy underreporting in a large double labeled water study. Am J Clin Nutr 79, 795–804. ¨ ring A, Keil U (1999). Meal patterns in middle-aged Winkler G, Do men in Southern Germany: results from the MONICA Augsburg dietary survey 1984/85. Appetite 32, 33–37.
European Journal of Clinical Nutrition