In uence of days of the week on reported food, macronutrient and alcohol intake among an adult population in south western Finland

Transcription

1 European Journal of Clinical Nutrition (1999) 52, 808±812 ß 1999 Stockton ress. All rights reserved 0954±3007/99 $ In uence of days of the week on reported food, macronutrient and alcohol intake among an adult population in south western Finland A Jula 1 *, R SeppaÈnen 1 and E Alanen 1 1 Research and Development Centre of the Social Insurance Institution, eltolantie 3, FIN Turku, Finland Objective: To study the in uence of the days of the week on calculated food and nutrient intake. Study design: Daily variation in food and nutrient intake was studied by using 4-, 5- and 7-d estimated food records, all including Friday, Saturday and Sunday. Subjects and setting: Two groups of untreated hypertensive patients (60 men and 31 women, aged 31 ± 55 y, and 132 men and 94 women, aged 35 ± 54 y) and a random population sample of 131 men and 139 women, aged 35 ± 64 y, all from the City of Turku and three neighbouring municipalities in south western Finland. Results: The intakes of meat and meat products, carbohydrate, alcohol and energy varied signi cantly across the week in all study groups with increased intakes of meat and meat products on Saturday and Sunday, of carbohydrate and alcohol on Friday and Saturday, and of energy on Friday, Saturday and Sunday. Food and nutrient intakes did not vary consistently with weekdays (Monday to Thursday) with the exception of alcohol intake being lowest on Monday. Macronutrient, alcohol and energy intake data calculated from the 5-d (Monday, Tuesday, Friday, Saturday and Sunday) and from 5 to 7 d converted food records were nearly equal to and correlated highly (r ˆ 0.96 ± 0.98) with the data calculated from the `true' 7 d records. Conclusion: The daily variation in food and nutrient intakes should be taken into consideration when food diaries are recorded. 5-d food records including two weekdays (from Monday to Thursday) and Friday, Saturday and Sunday is recommended. Sponsorship: Social Insurance Institution of Finland. Descriptors: estimated food records; food intake; nutrient intake; weekdays; weekend; Introduction Estimated food records are frequently used in the Nordic countries in the assessment of food and nutrient intakes of subjects participating in dietary surveys. The records should be kept in a reliable manner but not so intensely that they interfere with the subjects' dietary habits. While a suf ciently long recording period is necessary to re ect an individual's usual food and nutrient intakes, less days are needed for dietary assessments at group level (Bingham, 1987, Kuskowska-Wolk, 1990). Food and nutrient intakes may vary with the day of the week (Maisey et al, 1995). A food diary covering all weekdays could be expected to reduce the sampling bias due to the day-of-the-week variation of the diet. On the other hand, increasing the number of days to be recorded may in uence the accuracy of the records (Maisey et al, 1995). When the number of days to record increases, a tendency to simplify the task and non-recording may occur (Kuskowska-Wolk, 1990). Thus, Correspondence: Dr A Jula, Research and Development Centre, Social Insurance Institution, eltolantie 3, FIN Turku, Finland. Antti.Jula@Kela.Memonet.Fi Contributors: A Jula contributed to the design of the study, collection and analyses of the data, and writing of the paper. R SeppaÈnen contributed to the conception and design of the study, analyses of the data, and writing of the paper. E Alanen was responsible for statistical analyses of the data and participated in the interpretation of the ndings. A Jula and E Alanen are guarantors of the paper. Received 23 February 1999; revised 15 April 1999; accepted 25 April 1999 the representativeness of the collected data is related not only to the sampling bias but also to the response bias. Checking and coding the menu records requires less time and labour, if it is possible to reach a reliable result with fewer recording days (Black, 1988). The purpose of the present study was to investigate whether food and nutrient intakes vary across the week in subjects of working age, to compare food and nutrient intakes calculated from 4-, 5- and 7-d food records, and to evaluate whether the missing days of the week can be compensated using data derived from the existing days' records. Subjects and methods We report on the daily variation of the food and nutrient intake of subjects participating in three different studies. The subjects are characterized in Table 1. The rst group consisted of 91 untreated hypertensive men and women, aged 31 ± 55 y (Jula et al, 1990). The data on food intake were collected by means of a 7-d record before the patients were randomized for nonpharmacological treatment of hypertension. The second group consisted of 226 newly diagnosed, untreated, moderately to severely hypertensive men and women, aged 35 ± 54 y (Jula et al, 1995). Dietary data were collected by means of a 4-d food record (Thursday, Friday, Saturday, Sunday). The third group was a random sample of men and women drawn for the study from the national population register. For strati cation, 45

2 subjects of each sex and each 10-y age group (35 ± 44, 45 ± 54 and 55 ± 64 y) were chosen. For these 270 subjects, dietary data were collected by means of a 5-d food record (Monday, Tuesday, Friday, Saturday and Sunday). All records were evenly distributed over the year, excluding holidays, to avoid seasonal effects. In addition to blank record forms, the subjects received model forms, written and illustrated instructions and guidance in small groups to ensure accurate completion of the records. The subjects were instructed by a dietician to record all consumed food and drinks in household measures, e.g. portions, spoonfuls, cups, glasses, pieces etc., or to weigh them. ortion sizes were determined using a colour photograph atlas showing portions of known weight (Haapa et al, 1985). The subjects returned the records to the dietician who checked them, asked supplemental questions if necessary, and coded the records for analysis. Food records were analysed by means of the Nutrica 1 food and nutrient calculation software (Knuts et al, 1987) and the databases on the nutrient composition of Finnish foods (Rastas et al, 1989;1993). Energy from alcohol was included in the total energy intake. The study was conducted in compliance with the Second Declaration of Helsinki and approved by the Ethical Committee of the Social Insurance Institution of Finland. Statistical methods The values are given as means s.d. Statistical analysis of the data was performed with SAS computer programmes (SAS Institute). A general linear model (GLM) with repeated measures of variance analysis was used to test the daily variation in food, macronutrient and alcohol AJulaet al intakes. For the rst stage of analysis, the GLM model was written to test for the sex and day-of-the-week effects and the sex and day interactions. If there were none, a main model of the GLM without the sex term was written to test the day-of-the-week effects. earson's product moment coef cients were used to test correlations of food, nutrient and alcohol intakes calculated from 4, 5 or 7 d data derived from the 7-d food records. Before analyses, variables with skewed distributions were moved closer to normality by using their square roots (intakes of eggs, fruit and berries in the hypertensive group (n ˆ 226) and in the random population sample (n ˆ 270), and intakes of milk and milk products and sh in the hypertensive group (n ˆ 91)). Categorization of the studied variables was used where necessary. The intake of sh was analysed as a thricotomous variable ( sh intake 0 g=d1±40g=d and > 40 g=d) in the hypertensive group (n ˆ 226) and in the random population sample (n ˆ 270). Alcohol intake was analysed as a dichotomous (yes=no) variable in all study groups. Results The daily food and macronutrient intakes of the three different study groups are presented in Tables 2 ± 7. In the hypertensive group with 7-d recording, the intakes of meat and meat products ( < 0.001), carbohydrate ( < 0.01), alcohol ( < 0.001) and total energy ( < 0.001) varied signi cantly across the week with higher intakes of meat and meat products on Saturday and Sunday, of carbohydrate and alcohol on Friday and Saturday, and of energy on Friday, Saturday and Sunday (Table 2 and 3). Across the weekdays (Monday to Thursday), intakes of meat and meat products ( < 0.05) and 809 Table 1 Characteristics of the subjects Hypertensives (n ˆ 91) Hypertensives (n ˆ 226) opulation sample (n ˆ 270) Variable Men (n ˆ 60) Women (n ˆ 31) Men (n ˆ 132) Women (n ˆ 94) Men (n ˆ 131) Women (n ˆ 139) Age (y) Body weight (kg) Height (m) Body mass index (kg/m 2 ) Socio-economic class (%) Manual workers Lower level employees Upper level employees Others ± ± Values are mean s.d. Table 2 Amount (g) of each food group consumed by hypertensive men and women (n ˆ 91) on each day of the week Weekdays Food group Mon Tue Wed Thu Fri Sat Sun (Mon to Thu) Cereals Milk and milk products Eggs Fats and oils Meat and meat products < Fish and sh products Vegetables Fruits and berries ˆ Statistical signi cance of daily variation (repeated measures of ANOVA).

3 810 Table 3 A Jula et al Macronutrient and alcohol intakes of hypertensive men and women (n ˆ 91) on each day of the week Weekdays Nutrient Mon Tue Wed Thu Fri Sat Sun (Mon to Thu) Energy (MJ) < rotein (g) % of energy Fat (g) % of energy Carbohydrate (g) % of energy < Alcohol (g) < % of energy < ˆ Statistical signi cance of daily variation (repeated measures of ANOVA). Table 4 Amount (g) of each food group consumed by hypertensive men and women (n ˆ 226) on 4 d of the week Food group Thu Fri Sat Sun Cereals Milk and milk products Eggs Fats and oils Meat and meat products Fish and sh products Vegetables Fruits and berries ˆ Statistical signi cance of daily variation (repeated measures of ANOVA). Table 5 Macronutrient and alcohol intakes of hypertensive men and women (n ˆ 226) on 4 d of the week Nutrient Thu Fri Sat Sun Energy (MJ) < rotein (g) < % of energy < Fat (g) % of energy < Carbohydrate (g) % of energy < Alcohol (g) < % of energy < ˆ Statistical signi cance of daily variation (repeated measures of ANOVA). Table 6 Amount (g) of each food group consumed by a population sample of men and women (n ˆ 270) on 5 d of the week Day of Week Food group Mon Tue Fri Sat Sun Mon vs Tue Cereals Milk and milk products Eggs Fats and oils Meat and meat products Fish and sh products Vegetables * Fruits and berries ˆ Statistical signi cance by repeated measures of ANOVA. * < 0.05 for the difference between men and women.

4 Table 7 AJulaet al Macronutrient and alcohol intakes of a population sample of men and women (n ˆ 270) on 5 d of the week 811 Nutrient Mon Tue Fri Sat Sun Mon vs Tue Energy (MJ) < ** rotein (g) < % of energy < Fat (g) % of energy < *** Carbohydrate (g) % of energy < Alcohol (g) < % of energy < Values are average intakes s.d. by both sexes on each day. ˆ Statistical signi cance by repeated measures of ANOVA. ** < 0.01; *** < for the difference between men and women. Table 8 Macronutrient and alcohol intake calculated from 7, 5 and 4 d of the food records No. of days for calculations Variable 7 5 a 5 2 b 4 c 4 3 d 4 c 4 3 d 5 a 5 2 b Hypertensives (n ˆ 91) Hypertensives (n ˆ 226) opulation sample (n ˆ 270) Energy, (MJ/d) rotein, (g/d) Fat, (g/d) Carbohydrate, (g/d) Alcohol, (g/d) Values are average intakes s.d. by both sexes. a The average intake of Monday, Tuesday, Friday, Saturday and Sunday. b 5 d are converted to 7 d by replacing the missing days (Wednesday and Thursday) by the average values of Monday and Tuesday. c The average intake of Thursday, Friday, Saturday and Sunday. d 4 d are converted to 7 d by replacing the missing days (Monday, Tuesday and Wednesday) by the values of Thursday. alcohol ( < 0.05) varied slightly with lowest intakes of meat and meat products on Wednesday and of alcohol on Monday (Tables 2 and 3). In the hypertensive group with 4-d recording (Thursday, Friday, Saturday and Sunday), the intakes of meat and meat products ( < 0.01), protein ( < 0.001), fat ( < 0.05), carbohydrate ( < 0.01), alcohol ( < 0.001) and energy ( < 0.001) varied across the recording days with highest intakes of meat and meat products, protein, fat and carbohydrate on Saturday vs the lowest on Thursday, and with increased intakes of alcohol and total energy on Friday and Saturday (Tables 4 and 5). The intake of vegetables varied across the days signi cantly only in women ( < 0.05), with the highest intake on Saturday (Table 4). In the random population sample of men and women with 5-d recording (Monday, Tuesday, Friday, Saturday and Sunday), the intakes of meat and meat products ( ˆ 0.01), protein ( < 0.001), fat ( < 0.001), carbohydrate ( < 0.05), alcohol ( < 0.001) and total energy ( < 0.001) varied across the recording days, with the highest intakes of meat and meat products on Saturday and the lowest on Tuesday, with increased intakes of protein, fat, carbohydrate and total energy on Saturday and of alcohol on Friday and Saturday (Tables 6 and 7). The variation in energy intake was higher in men ( < ) than in women ( < 0.001; < 0.01 for the difference between men and women). The food and nutrient intakes did not differ between Monday and Tuesday. There was less use of alcohol on Monday than on Tuesday. Table 8 compares the energy, macronutrient and alcohol intakes calculated from the 7-, 5- and 4-d food records, as well as from the 7-d records converted from the 4- and 5-d Table 9 Univariate correlations (earsons's r) between macronutrient and alcohol intakes calculated from 7, and from 5 and 4 d of the 7 d food records of hypertensive men and women (n ˆ 91) No. of days for the calculated variable Variable e 5 a 5 2 b 4 c 4 3 d Energy (MJ/d) rotein (g/d) Fat (g/d) Carbohydrate (g/d) Alcohol (g/d) a The average intake of Monday, Tuesday, Friday, Saturday and Sunday. b 5 d are converted to 7 d by replacing the missing days (Wednesday and Thursday) by the average values of Monday and Tuesday. c The average intake of Thursday, Friday, Saturday and Sunday. d 4 d are converted to 7 d by replacing the missing days (Monday, Tuesday and Wednesday) by the values of Thursday. e Calculated from 7 d. records. Each subject's 5-d records were converted to 7-d records by replacing the missing weekdays (Wednesday and Thursday) by the average values of the corresponding data for Monday and Tuesday. Correspondingly, the 4-d records were converted to 7-d records by replacing each of the three missing weekdays (Monday, Tuesday and Wednesday) by the values recorded on Thursday. The conversion from 4 to 7 d moved the data closer to that calculated from the `true' 7-d food records. Energy, macronutrient and alcohol intake data calculated from the 5-d (Monday, Tuesday, Friday, Saturday and Sunday) and from 5 to 7 d converted food records were almost equal to the data calculated from the `true' 7-d records.

5 812 A Jula et al The intakes of energy, protein, fat, carbohydrate and alcohol calculated either from the 5-d or from 5 to 7 d converted records correlated highly signi cantly with the data calculated from the `true' 7-d food records (r ˆ 0.96 ± 0.98, Table 9). Discussion The men and women of the three study groups represented a normal socio-economic distribution of an urban Finnish population of working age. Two study groups consisted of newly diagnosed, untreated hypertensive patients and the third group of a random population sample of men and women. The random sample can be considered representative of the target population, because the participation rate was over 80% (Evans, 1991). In all study groups, the intakes of meat and meat products, carbohydrate and alcohol varied across the week. The intake of meat and meat products was highest on Saturday and Sunday, and the intakes of carbohydrate and alcohol highest on Friday and Saturday. The energy intake was increased on Friday, Saturday and Sunday. When averaged and weighed for the number of subjects participating in the three studies, approximately 80% of the increase in total energy intake is accounted for by increased alcohol intake on Friday, 50% on Saturday and 30% on Sunday. Our ndings of increased alcohol intake at weekends are in line with several earlier studies. In the MONICA dietary survey project in Germany (Winkler et al, 1991), alcohol consumption was 26% higher on Saturday and Sunday than during weekdays.thomson et al (1988) found increased intakes of alcohol and energy at weekends in a group of Scottish men. Maisey et al (1995) reported that in an elderly British population the alcohol intake was higher on Saturdays and Sundays than during weekdays. In the current survey, the intake of alcohol was most abundant on Fridays and Saturdays which can be explained by cultural differences in drinking habits and by the fact that on Sundays most of our subjects had to anticipate for the next working day. In the study by Maisey et al (1995), the intakes of meat, meat products, sh and vegetables varied across the week. The intakes of protein, fat and total energy were higher at weekends. These results support our ndings of higher intakes of meat and meat products and total energy at weekends than during weekdays. The variation in food and nutrient intake may also be dependent of culture, religious teaching and age. In the Bogalusa heart study (Nicklas et al, 1997), the intakes of protein, fat and carbohydrate of 10-y old children did not vary across the week. However, signi cant variation was noticed in the intakes of different food groups. ercentage of energy from fruits, milk and meat dishes were higher on weekdays than at weekends and from vegetables, poultry, eggs and pork highest on Sunday. The ndings were largely explained by differences in meals consumed at school versus those at home. It has been suggested that weekend days and weekdays should be proportionally represented in the calculations of food and nutrient intakes to avoid bias caused by the dayof-the-week variation (Tarasuk and Beaton, 1992). Maisey et al (1995) recommended that all days of the week should be proportionally represented in studies of the elderly. We did not nd any consistent variation in food and nutrient intakes across Monday, Tuesday, Wednesday and Thursday with the exception that alcohol intake was lowest on Monday. Friday, Saturday and Sunday differed from the other days of the week especially in the increased alcohol use and energy intake. The conversion of 4-d food records including Thursday, Friday, Saturday and Sunday to 7-d records by compensating for the missing days using data derived from the included weekday (Thursday) moved the food and nutrient intake data closer to the data calculated from 7-d records. However, macronutrient, alcohol and energy intake data calculated from the ve-day (Monday, Tuesday, Friday, Saturday and Sunday) and from 5 to 7-d converted food records were nearly equal to and correlated highly with the data calculated from the `true' 7-d records. Conclusions We suggest that 5-d food records from Friday, Saturday and Sunday and including two of weekdays from Monday to Thursday, will yield suf ciently accurate information of food and nutrient intakes in men and women of working age. Acknowledgements ÐThe authors wish to thank Merja Rastas and Vaula Karttila for their assistance in instructing, coding and computing the dietary records, and Tuula Toikka, Tuula Aaltonen and aèivi altta for assistance in statistical analyses. References Black A (1988): Economic appraisal of various study designs and methods. In Manual on methodology for food consumption studies, eds M Cameron and W van Staveren, 201 ± 222. Oxford University ress. Bingham SA (1987): The dietary assessment of individuals: methods, accuracy, new techniques and recommendations. Nutr. Abstr. Rev. 37, 705 ± 742. Evans SJW (1991): Good surveys guide. Br. Med. J. 302, 302 ± 303. Haapa E, Toponen T, ietinen & RaÈsaÈnen L (1985): Annoskuvakirja (icture book on food portions). Helsinki: Kansanterveyslaitos (National ublic Health Institute). Jula A, RoÈnnemaa T, Rastas M, Karvetti R-L & MaÈki J (1990): Long-term nonpharmacological treatment for mild to moderate hypertension. J. Intern. Med. 227, 413 ± 421. Jula A, uukka & Karanko H (1995): Standardized multiple clinic and home blood pressure measurements and ambulatory blood pressure monitoring in the classi cation of blood pressure status. Am. J. Hypertens. 8, part 2: 97A. Knuts LR, Rastas M, Haapala, SeppaÈnen R, Karvetti RL & Aro A (1987): Nutrica ± a computer program for calculation of food and nutrient intake. In: Third Eurofoods Meeting, Warsaw 24 ± 27 May rogramme and abstracts. Eurofoods. Kuskowska-Wolk A (1990) Record methods in food consumption studies: VaÊr foèda 42, (Suppl. 3), 7 ± 35. 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. Br. J. Nutr. 73, 359 ± 373. Nicklas T, Farris R, Bao W, Webber L & Berenson L (1997): Differences in reported dietary intake of 10-year-old children on weekdays compared to Sunday: The Bogalusa Heart Study. Nutr. Res. 17, 31 ± 40. Rastas M, SeppaÈnen R, Knuts L-R, Karvetti R-L & Varo (eds)(1989): Nutrient Composition of Foods, 2nd edn, 1 ± 457. HaÈmeenlinna: Karisto. Rastas M, SeppaÈnen R, Knuts L-R, Karvetti R-L & Varo (eds) (1993): Nutrient Composition of Foods, 4th edn, 1 ± 461. Vammala: Vammalan Kirjapaino. Tarasuk V & Beaton GH (1992). Statistical estimation of dietary parameters: implications of patterns in within-subject variation ± a case study of sampling strategies. Am. J. Clin. Nutr. 55, 22 ± 27. Thomson M, Elton R, Fulton M, Brown S, Wood D & Oliver M (1988): Individual variation in dietary intake of a group of Scottish men. J. Hum. Nutr. Diet. 1, 47 ± 57. Winkler G, DoÈrig A & Keil U (1991): Unterschiede in ErnaÈhrungsverhalten zwischen Wochenenden und Werktagen: Ergebnisse der ErnaÈhrungserhebung 1984=85 des MONICA-rojektes Augsburg. Z. ErnaÈhrungswiss 30, 313 ± 317.