Breakfast skipping and health-compromising behaviors in adolescents ...

25 downloads 0 Views 126KB Size Report
1Department of Public Health, University of Helsinki, Finland; 2Department of ... Keywords: breakfast; smoking; exercise; adolescence; nutrition; Finland.
European Journal of Clinical Nutrition (2003) 57, 842–853

& 2003 Nature Publishing Group All rights reserved 0954-3007/03 $25.00 www.nature.com/ejcn

ORIGINAL COMMUNICATION Breakfast skipping and health-compromising behaviors in adolescents and adults A Keski-Rahkonen1,2*, J Kaprio1,3, A Rissanen2, M Virkkunen2 and RJ Rose4 1 Department of Public Health, University of Helsinki, Finland; 2Department of Psychiatry, Helsinki University Central Hospital, Finland; 3Department of Mental Health, National Public Health Institute, Helsinki, Finland; and 4Department of Psychology, Indiana University, Bloomington, Indiana, USA

Objective: To investigate which sociodemographic factors and behaviors are associated with breakfast skipping in adolescents and adults. Design: Five birth cohorts of adolescent twins and their parents received an extensive behavioral and medical self-report questionnaire that also assessed breakfast-eating frequency. Setting: Finland, 1991–1995. Subjects: A population sample of 16-y-old girls and boys (n ¼ 5448) and their parents (n ¼ 4660). Results: Parental breakfast eating was the statistically most significant factor associated with adolescent breakfast eating. Smoking, infrequent exercise, a low education level at 16, female sex, frequent alcohol use, behavioral disinhibition, and high body mass index (BMI) were significantly associated with adolescent breakfast skipping. In adults, smoking, infrequent exercise, low education level, male sex, higher BMI, and more frequent alcohol use were associated with breakfast skipping. In the adult sample, older individuals had breakfast more often than younger ones. Both adults and adolescents who frequently skipped breakfast were much more likely to exercise very little compared to those who skipped breakfast infrequently. Breakfast skipping was associated with low family socioeconomic status in adults and adolescent boys, but not in girls. Breakfast skipping clustered moderately with smoking, alcohol use, and sedentary lifestyle in both adults and adolescents. Conclusions: Breakfast skipping is associated with health-compromising behaviors in adults and adolescents. Individuals and families who skip breakfast may benefit from preventive efforts that also address risk behaviors other than eating patterns. Sponsorship: National Institute of Alcohol Abuse and Alcoholism (AA08315), Academy of Finland (44069), European Union Fifth Framework Program (QLRT-1999-00916), Yrjo¨ Jahnsson Foundation, and Jalmari and Rauha Ahokas Foundation. European Journal of Clinical Nutrition (2003) 57, 842–853. doi:10.1038/sj.ejcn.1601618 Keywords: breakfast; smoking; exercise; adolescence; nutrition; Finland

Introduction Breakfast skipping is relatively common among adolescents and adults in Western countries. Definitions of breakfast vary in different cultures and study settings. In this study, breakfast skipping is defined as not eating a morning meal at home. Breakfast skipping in adolescents has been

*Correspondence: A Keski-Rahkonen, Department of Public Health, PO Box 41, University of Helsinki, Helsinki 00014 Finland. E-mail: [email protected] Guarantors: A Keski-Rahkonen and J Kaprio. Contributors: AKR and JK were responsible for the study design, data analysis, and interpretation of results. A Rissanen, M Virkkunen, and RJ Rose contributed to the design of the study. RJ Rose and J Kaprio were responsible for the study data collection. All authors were involved in writing the manuscript. Received 8 April 2002; revised 17 July 2002; accepted 13 August 2002

associated with various health-compromising behaviors and unhealthy lifestyles, such as tobacco, alcohol, and substance use, and risk-taking in general (Revicki et al, ¨ glund et al, 1998). 1991; Isralowitz & Trostler, 1996; Ho Breakfast skipping may also be an indicator of risk to weight gain: among those who skip breakfast, increased snacking, lunch skipping, sedentary lifestyle, and obesity are more common than among breakfast eaters (Terre et al, 1990; Wolfe et al, 1994; Baumert et al, 1998; Nordlund & Jacobson, 1999; Urho & Hasunen, 1999; Serra et al, 2000). Much less is known about factors associated with breakfast skipping in adults. In industrialized countries, breakfast skipping has been linked to low family socioeconomic status (SES) (Pastore et al, 1996; Brugman et al, 1998; ¨ glund et al, 1998; Nordlund & Jacobson, 1999; O’Dea & Ho Caputi, 2001), although this finding is not consistent

Breakfast skipping and health-compromising behaviors A Keski-Rahkonen et al

843 (Walker et al, 1982). Conversely, regular breakfast eating has been associated with a health-conscious lifestyle (Baumert et al, 1998; Cavadini et al, 2000). Adolescent girls have been found to skip breakfast more often than boys (Isralowitz & Trostler, 1996; Brugman et al, 1998; ¨ glund et al, 1998; Shaw, 1998): this may be a chosen Ho method of weight control for girls, and is in some individuals associated with body dissatisfaction, dieting, or disordered eating (Melve & Baerheim, 1994; Bellisle et al, 1995; Shaw, 1998). Disordered eating is generally associated with health-compromising behaviors (Neumark-Sztainer et al, 1997). The aim of this study was to investigate whether various factors related to health-compromising and health-promoting behaviors and SES are associated with breakfast skipping in a large population sample of Finnish adolescents and adults. More specifically, we hypothesized that breakfast skipping is associated with various health-compromising factors, such as smoking, alcohol use, overweight, and also with factors more loosely related to health-compromising behaviors, such as early puberty onset, behavioral disinhibition, low level of education, low SES, and bad general health. We also hypothesized that breakfast skipping aggregates in families.

Subjects and methods Subject sample The data reported are from FinnTwin16, a population-based study of five consecutive nationwide birth cohorts of Finnish twins born between 1975 and 1979 (Rose et al, 1999). Data collection was subjected to and approved by local ethics committees. A questionnaire was mailed to twins born in 1975 through 1979 within 2 months of their 16th birthday. The questionnaire assessed personality, social relations, general health, and health habits, including breakfast eating. Follow-up questionnaires were sent to the twins at the ages of 17 and 18.5 y. From the follow-up, only data on body mass index (BMI), education, and disinhibition were used for this study. Questions on breakfast eating identical to those on the twin questionnaire were sent to the twins’ mothers and fathers as part of questionnaires that addressed parental lifestyle and health. These were mailed at the same time as the baseline twin questionnaires. At the time of the assessment, the age of the mothers ranged from 32.2 to 62 (mean 44.3, s.d. 4.9) y, and fathers, respectively, from 33.6 to 69.8 (mean 46.5, s.d. 5.7) y. A total of 3065 families were contacted: 5561 of the 6130 twins in these families (91%) returned the baseline questionnaire. Individual response rates were 93% for girls, 88% for boys, 84% for fathers, and 87% for mothers. For this study, we selected all twin pairs where both twins had responded to the breakfast eating question; the pairwise analyses are reported elsewhere (KeskiRahkonen et al, in press). Our study sample consisted of 5448

twins (2822 girls and 2626 boys) and their 4660 parents (2440 mothers and 2220 fathers).

Measures The frequency of breakfast eating was assessed by the following question: ‘How often do you eat breakfast (for example, sandwiches, milk, hot cereal, other similar food) before going to school or going to work?’ The three alternative responses were ‘every morning’, ‘a few times a week’, ‘about once a week or less often’. The following other variables obtained from the questionnaires at the age of 16 were used in our analyses: sex, education level at 16; height and weight, from which BMI was calculated; smoking status, alcohol use; use of coffee, tea, caffeinated soda, and cocoa; types of milk and bread spread used; frequency of physical exercise, self-perceived general health, and age of puberty onset (menarche for females, voice break for males). From the questionnaires filled out at 17, we obtained BMI disinhibition, experience seeking, and susceptibility to boredom scores from the Sensation Seeking Scale (Zuckerman, 1979). Of the Sensation Seeking Subscales, behavioral disinhibition had the highest correlation with breakfast eating; in the subsequent analyses, only this subscale was used. If the behavioral disinhibition score at 17 was missing, we substituted disinhibition score at 18 (180 substitutions). Similarly, if the information about BMI was missing in the 16 y questionnaires, we used BMI at 17 instead (118 substitutions). We analyzed the education level of our adolescents at the ages of 16 and 17, because these two measurements have different implications. Mandatory schooling ends in Finland at the age of 16. Those who drop out by the age of 16 tend to have more severe problem behaviors than those not continuing in school at 17, when further education is voluntary and the choice of education reflects academic performance. Moreover, academically oriented teenagers are much less likely to engage in healthcompromising behaviors than those with a vocational ¨ m et al, 1996; Aarnio et al, 1997). orientation (Bergstro From parental questionnaires, we obtained data on father’s and mother’s breakfast eating and other variables we deemed possibly important for breakfast eating: smoking; alcohol, coffee, and tea use; height and weight, from which BMI was calculated; types of milk and bread spread used; use of vitamin and trace mineral supplements, and of natural and herbal drugs; frequency of physical exercise; highest level of parental education, unemployment in the family; shiftwork, amount of sleep, and feeling tired in the morning. Smoking and alcohol use variables were dichotomized, primarily to facilitate the assessment of possible interactions. Since questions about many key variables were formulated differently in adolescent and parental questionnaires, we were not able to directly compare the health-compromising behaviors of adults and adolescents as correlates of breakfast eating. Family SES was determined by the occupation of the father European Journal of Clinical Nutrition

Breakfast skipping and health-compromising behaviors A Keski-Rahkonen et al

844 or the mother, whichever ranked higher. If only one parent had responded, his/her occupation determined the family’s SES. If both parents’ occupations were unknown, the family was excluded from the analyses of the effect of SES on breakfast eating. The parental occupations were divided into seven categories using the Statistics Finland 1989 classification of SES (Sosioekonomisen aseman luokitus, 1989, 1990); we contrasted the two highest income categories (upperlevel white-collar workers; independent entrepreneurs and farmers) with the five lowest categories (lower-level whitecollar workers, blue-collar workers, students, pensioners, and those of unclassified or unknown occupation) to create ‘higher SES’ and ‘lower SES’ categories.

Table 1 Breakfast-eating frequencies in adolescents (N=5448) and adults (N=4660)

Statistical analyses We investigated differences between breakfast categories and explanatory variables (eg, sex, SES, education level) using cross-tabulations and the Pearson w2 test of independence, corrected for clustered sampling (expressed as an F ratio, as described by Rao and Scott (1984) in Stata, version 7.0, 2000). We assessed correlates of breakfast eating using univariate and multivariable multinomial (polychotomous) logistic regression models (Hosmer & Lemeshow, 2000), again corrected for clustered sampling. The three categories of breakfast eatingF ‘once a week or less’, ‘a few times a week’, and ‘every morning’ (reference category) Fwere used as dependent variables. All models were adjusted for sex in adolescents and for sex and age in adults. Variables that were important correlates of adolescent breakfast in previous studies or were otherwise considered relevant were entered in a multivariable model. We then inspected how the variables important for adolescents were associated with adult breakfast eating in an adult multivariable model (age of puberty onset and behavioral disinhibition were not measured in adults). Using the concepts of parsimony and goodness of fit, we chose the most appropriate models. To assess whether increased breakfast-skipping was associated with changes in the study subject’s behavior or background characteristics in the multivariable level, we tested whether the two breakfastskipping categories could be considered to be equal, and whether odds ratios of individual variables could be considered equal across the two breakfast-skipping categories. The respective fits of hierarchically nested models were assessed using the likelihood ratio test. The statistical analyses were performed with the SAS system for Windows, version 8.01 (1999) and Stata, release 7.0 (2000).

(F ¼ 86.2, Po0.00001). Among adults, breakfast skipping was more prevalent in the lower SES group than in the higher SES group (Po0.00001 for both men and women), and this was also seen in boys (F ¼ 4.37, P ¼ 0.013) (Table 2). For girls, breakfast eating did not differ by SES groups (F ¼ 2.14, P ¼ 0.426).

Results The frequencies of breakfast eating in adolescents and adults are shown in Table 1. Adults had breakfast significantly less often than adolescents (w2 ¼ 31.3, Po0.00001). Girls had breakfast significantly less often than boys (F ¼ 7.6, P ¼ 0.0005) and adult men less often than adult women European Journal of Clinical Nutrition

Daily breakfast (%) Adolescent girls Adolescent boys Adult women Adult men

68.4 73.5 71.0 59.8

Breakfast a few times a week (%) 16.0 13.4 10.0 14.1

Breakfast once a week or less often (%) 15.7 13.0 19.0 26.1

Total n 2821 2627 2440 2220

Rounding error in sums of some percentages.

Correlates of adolescent breakfast eating Sex-adjusted models of different categories of adolescent breakfast eating and its correlates are presented in Table 3. Adjustment for sex affected the odds ratios of correlated behaviors only very slightly; education level was affected the most. Breakfast skipping in adolescents was clearly associated with health-compromising behaviors and lifestyles. Individuals who rarely had breakfast were more likely to smoke, drink alcohol frequently, and use more coffee and caffeinated sodas than regular breakfast eaters. Breakfast eaters were less likely to exercise, and tended to have a higher BMI, higher behavioral disinhibition scores, and an earlier onset of puberty than breakfast eaters. Better education and higher SES were associated with more frequent breakfast eating. The education level at 16 explained a higher proportion of the variance of adolescent breakfast eating (pseudo-R2 ¼ 0.018) than family SES (pseudo-R2 ¼ 0.005). Thrill and adventure seeking, self-perceived health, type of dietary spread used for bread, and parental unemployment were not significantly associated with adolescent breakfast eating patterns. In multivariable analyses of adolescent breakfast eating detailed in Table 4, parental breakfast eating appeared as the statistically most significant factor associated with adolescent breakfast eating. Other statistically significant correlates in adolescents were smoking, exercise, education level at 17, sex, alcohol use, behavioral disinhibition, and BMI (presented in order of decreasing significance). There were no significant interactions between smoking, alcohol use, and behavioral disinhibition, or sex and age of puberty onset. These interaction terms and the age of puberty onset could be removed from the multivariable model without a

Breakfast skipping and health-compromising behaviors A Keski-Rahkonen et al

845 Table 2 Breakfast eating by family socioeconomic statusa in adults (N=4660) and adolescents (N=5448) Family socioeconomic status Upper-level employees (%)

Independent entrepreneurs and farmers (%)

Lower-level employees (%)

Manual workers (%)

Students or pensioners (%)

Unknown occupation (%)

Total (%)

Women Adults Daily A few times a week Once a week or less

16.6 10.3 7.5

12.6 9.5 11.9

48.6 48.2 45.5

18.3 27.6 28.0

1.8 2.1 2.8

2.1 2.5 4.3

100 100 100

Adolescents Daily A few times a week Once a week or less

21.6 23.1 17.4

17.4 18.9 18.1

20.0 17.3 20.8

33.1 33.1 31.7

1.2 1.8 1.8

6.8 5.8 10.2

100 100 100

Men Adults Daily A few times a week Once a week or less

28.2 18.3 16.2

20.9 15.4 18.1

16.1 18.0 16.7

30.7 43.0 45.2

1.1 1.3 0.3

3.1 4.2 3.5

100 100 100

Adolescents Daily A few times a week Once a week or less

22.0 19.8 12.6

17.1 16.2 16.4

20.6 19.3 19.6

31.3 35.4 37.4

1.2 0.7 2.1

7.8 8.5 12.0

100 100 100

Breakfast-eating frequency

a Determined by the occupation of the father or mother, choosing the one that was ranked higher according to the Statistics Finland (1989) classification of socioeconomic status.

significant decrease in model fit. Considering the ‘breakfast a few times a week’ and ‘breakfast once a week or less often’ categories of the multinomial logistic regression model to be equal caused a very significant (Po0.00001) deterioration of model fit: thus, we present the results retaining the original three categories of breakfast eating. However, certain variablesFsmoking, education level at 17, sex, alcohol use, and BMIFcould be considered equal across the breakfast-skipping categories without a significant deterioration in model fit. This suggests that although the odds ratios of adolescent breakfast eaters and breakfast skippers clearly differed in terms of alcohol use, education level at 17, sex, and BMI, infrequent and frequent breakfast skippers were similar in these respects. However, frequent breakfast skippers exercised less, had parents who skipped breakfast more often, and were more behaviorally disinhibited than infrequent breakfast skippers.

Correlates of adult breakfast eating In age–sex-adjusted breakfast-eating models of adults (Table 5), health-compromising or potentially less healthy behaviors generally increased when frequency of breakfast skipping increased. Less healthy behaviors, like using butter (instead of margarine) and milk with a higher fat content, were more common in adult breakfast skippers than break-

fast eaters. Conversely, general health-conscious behaviors, for example, exercise and use of vitamin and trace mineral supplements, were associated in adults with breakfast eating. Less sleep was in adults associated with more breakfast skipping, as was shiftwork and feeling tired in the morning. These factors, however, explained a much smaller fraction of the variance of breakfast eating than the factors that were entered in an adult multivariable model. In the adult multivariable model (Table 6), the most statistically significant correlates of adult breakfast eating were smoking, exercise, education, sex, age, BMI, and alcohol use. None of these factors could be removed from the model without a significant deterioration in model fit, but alcohol and tobacco use did not have a significant interaction. Breakfast eating increased with age in adults. Again, considering the outcome categories ‘breakfast a few times a week’ and ‘breakfast once a week or less often’ as equal caused a significant (Po0.001) deterioration in model fit. On the level of individual variables, odds ratios of all other variables except smoking and exercise could be considered equal across the two breakfast-skipping categories without a statistically significant worsening in model fit. This suggests that an increased frequency of breakfast skipping was associated with increased smoking and decreased exercise frequency, but that in other respects infrequent and frequent breakfast skipping in adults had similar behavioral correlates. European Journal of Clinical Nutrition

Breakfast skipping and health-compromising behaviors A Keski-Rahkonen et al

846 Table 3 Adolescents who skip breakfast vs those who always eat breakfast: odds ratios of correlated behaviors and other characteristics, based on sexadjusted multinomial logistic regression models Breakfast skippers (vs eaters) Sex-adjusted odds ratios (95% confidence intervala) Correlates Parental Mother’s breakfast eating, N=4886 Every morning (71.0%) A few times a week, 10 (10.0%) Once a week or less (19.0%) Father’s breakfast eating N=4440 Every morning (59.8%) A few times a week (14.1%) Once a week or less (26.1%) Family socioeconomic status N=5024 Upper-level employee (22.6%) Self-employed, including self-employed farmers (18.8%) Lower-level employee (21.7%) Manual worker (35.5%) Student, retired, or unclassified occupation (1.4%) Individual BMI, kg/m2, N=54032 o20 (49.5%) 20–22.9 (39.5%) 23–24.9 (6.7%) 25 or more (4.2%) Smoking N=5363 Never (48.7%) Given up temporarily or permanently (20.4%) Less than once a week (9.4%) Once a week or more but less than daily (4.1%) Daily (17.4%) Behavioral disinhibition at 17 y N=5334c Low behavioral disinhibition (48.2%) High behavioral disinhibition (51.8%) Tea, cups per day, N=5398 0 (73.8%) 1 (14.9%) 2 (8.2%) 3 or more (3.1%) Coffee, cups per day, N=5410 0 (59.8%) 1–4 (32.6%) 5–8 (6.3%) 9 or more (1.3%) Age of puberty onsetd, N=5279 o13 y (44.2%) 13 y (29.5%) >13 y (26.3%) Caffeinated soda, bottles per day, N=5396 0 (90.0%) 1 or more (10.0%) Cocoa, cups per day, N=5401 0 (78.7%) 1 (12.0%) 2 or more (9.2%) Alcohol use, N=5425 Never (23.5%) Once a year or less (9.3%) A few times a year (14.4%) Once in a few months (13.7%) Once a month (11.7%) A few times a month (17.8%) Weekly or more (9.7%)

European Journal of Clinical Nutrition

Breakfast a few times a week

Breakfast once a week or less often

1.0 (reference) 1.74 (1.30–2.34)*** 2.26 (1.82–2.80)***

1.0 (reference) 2.24 (1.65–3.05)*** 3.57 (2.84–4.49)***

1.0 (reference) 1.95 (1.50–2.54)*** 2.06 (1.66–2.54)***

1.0 (reference) 1.79 (1.30–2.46)*** 2.93 (2.32–3.70)***

1.0 1.03 0.91 1.07 1.15

(reference) (0.78–1.36) (0.70–1.18) (0.84–1.36) (0.54–2.48)

1.0 1.43 1.43 1.52 2.28

(reference) (1.05–1.95)* (1.06–1.93)* (1.16–2.00)** (1.06–4.93)*

1.0 1.11 1.36 1.47

(reference) (0.94–1.32) (0.99–1.87) (0.98–2.20)

1.0 1.08 0.96 2.16

(reference) (0.90–1.28) (0.68–1.37) (1.51–3.10)***

1.0 1.43 1.52 1.60 2.28

(reference) (1.15–1.77)*** (1.16–2.00)** (1.07–2.37)* (1.83–2.85)***

1.0 1.56 1.28 1.47 4.17

(reference) (1.23–1.96)*** (0.94–1.76) (0.97–2.24) (3.34–5.21)***

1.0 (reference) 1.23 (1.05–1.45)*

1.0 (reference) 1.49 (1.26–1.76)***

1.0 0.71 0.87 0.54

(reference) (0.56–0.90)** (0.65–1.17) (0.31–0.94)*

1.0 0.66 0.60 0.99

(reference) (0.51–0.85)*** (0.43–0.84)** (0.62–1.56)

1.0 1.22 1.79 1.50

(reference) (1.02–1.45)* (1.31–2.46)*** (0.74–3.01)

1.0 1.43 3.69 4.17

(reference) (1.19–1.72)*** (2.73–4.98)*** (2.42–7.17)***

1.0 (reference) 0.90 (0.74–1.10) 0.94 (0.75–1.18)

1.0 (reference) 1.10 (0.90–1.36) 1.49 (1.20–1.86)***

1.0 (reference) 1.28 (0.98–1.68)

1.0 (reference) 1.78 (1.38–2.30)***

1.0 (reference) 0.84 (0.66–1.07) 0.85 (0.65–1.12)

1.0 (reference) 0.58 (0.44–0.78)*** 0.71 (0.52–0.96)*

1.0 0.60 1.09 1.22 1.31 1.67 2.03

1.0 1.33 1.55 1.44 1.47 2.13 2.89

(reference) (0.40–0.90)* (0.82–1.46) (0.92–1.62) (0.98–1.75) (1.29–2.15)*** (1.51–2.73)***

(reference) (0.94–1.87) (1.15–2.09)** (1.05–1.96)* (1.07–2.02)* (1.61–2.81)*** (2.11–3.96)***

Breakfast skipping and health-compromising behaviors A Keski-Rahkonen et al

847

Table 3 (continued) Breakfast skippers (vs eaters) Sex-adjusted odds ratios (95% confidence intervala) Correlates Type of milk used, N=5441 No milk (11.6%) Non-fat milk (27.2%) Low-fat 1% milk (7.1%) Low-fat 1.9% milk (41.1%) 3.5% or full-fat milk (8.2%) No preference (4.8%) Education at 16 y, N=5418 Senior high school, vocational college or polytechnic school (36.1%) Vocational school (19.2%) Junior high school (43.0%) Not in school (1.7%) Education at 17 y, N=5110 Senior high school, vocational college or polytechnic school (62.1%) Vocational school or practical work-oriented course (30.6%) Junior high school (3.4%) not in school (4.0%) Exercise N=5428 Daily (15.8%) 4–5 times a week (15.2%) 2–3 times a week (28.0%) About once a week (19.1%) Once or twice a month (10.4%) Less than once a month (6.6%) Never (5.0%)

Breakfast a few times a week

Breakfast once a week or less often

1.0 0.67 0.84 0.77 0.82 0.89

(reference) (0.51–0.88)** (0.58–1.22) (0.60–0.99)* (0.57–1.16) (0.59–1.35)

1.0 0.55 0.72 0.67 0.82 0.74

(reference) (0.42–0.73)*** (0.48–1.06) (0.52–0.88)** (0.58–1.18) (0.49–1.11)

1.0 1.41 1.16 2.92

(reference) (1.13–1.77)** (0.96–1.41) (1.62–5.25)***

1.0 2.00 1.79 4.91

(reference) (1.56–2.57)*** (1.44–2.22)*** (2.79–8.64)***

1.0 1.52 1.60 2.23

(reference) (1.26–1.83)*** (0.98–2.61) (1.49–3.34)***

1.0 1.97 3.49 3.43

(reference) (1.62–2.41)*** (2.25–5.43)*** (2.36–4.97)***

1.0 1.36 1.28 1.60 1.60 1.72 1.39

(reference) (1.01–1.84)* (.98–1.69) (1.20–2.12)*** (1.15–2.22)** (1.17–2.52)** (0.90–2.15)

1.0 0.95 1.32 1.36 2.33 4.03 3.75

(reference) (0.67–1.35) (0.99–1.77) (1.00–1.85) (1.66–3.28)*** (2.84–5.72)*** (2.57–5.47)***

* Po0.05, ** Po0.01, *** Pr0.001. a Confidence intervals adjusted for intraclass correlation. b For 118 BMI values missing at 16 y, BMI values at 17 y were substituted. c For 180 behavioral disinhibition scores missing at 17 y, behavioral disinhibition scores at 18 y were substituted. d Menarche for females, voice break for males Rounding error in sums of some percentages Other variables tested but not significant at the alpha=0.05 level: thrill and adventure seeking, self-perceived health, type of bread spread used, parental unemployment.

Clustering of health-compromising behaviors Table 7 shows the co-occurrence of certain health-compromising factors (smoking, alcohol use, sedentary lifestyle, and overweight or obesity) in our sample. The co-occurrence of two or more health-compromising factors was significantly (Po0.00001) more common among breakfast skippers than breakfast eaters. We also examined how breakfast skipping and correlated behaviors were transmitted from the parents to the offspring by examining families where parents always have breakfast (N ¼ 2110) vs families where parents frequently skip breakfast (N ¼ 356). Of the children of the breakfast-eating parents, 81.7% had breakfast every morning, whereas only 47.8% of the children of the breakfast skippers did so (the difference was statistically significant, F ¼ 84.0, Po0.00001). There was a significantly (F ¼ 19.3, Po0.00001) greater proportion of families of low SES (66.7%) in families where both parents skipped breakfast compared to families where both parents always had breakfast (48.8%). There were further significant (Po0.001) differences among the offspring in the following respects: compared to the children of parents who never skip

breakfast, the children of parents who always skip breakfast were also more likely to drink a lot of coffee and choose a higher fat type of milk. Children of breakfast skippers, even after an adjustment for sex, were heavier, exercised less, had an earlier onset of puberty, and went to less academically oriented schools.

Discussion Smoking, infrequent exercise, a low level of education, frequent alcohol use, and high BMI were associated with breakfast skipping in both adults and adolescents. Frequent breakfast skippers were much more likely to exercise very little than infrequent breakfast skippers. Contrary to our hypotheses, self-perceived general health was not associated with breakfast-eating patterns, and age of puberty onset was of borderline significance. In our study, parental breakfast eating was the statistically most significant factor associated with adolescent breakfast eating. Children of breakfast-skipping parents were much European Journal of Clinical Nutrition

Breakfast skipping and health-compromising behaviors A Keski-Rahkonen et al

848 Table 4 Multivariable modela of correlates of adolescent breakfast eating, N=5085 Breakfast skippers (vs. eaters) Adjusted odds ratios (95% Confidence Interval b) Correlates Parental Mother’s breakfast eating Every morning (64.3%) A few times a week, 10 (8.9%) Once a week or less (17.0%) Data missing (9.8%) Father’s breakfast eating Every morning (49.6%) A few times a week (11.6%) Once a week or less (21.3%) Data missing (17.6%) Individual Smoking Never (30.8%) Has smoked or still smokes (69.2%) Alcohol use Less than once a week (72.2%) Once a week or more often (27.8%) Education at 17 y Senior high school, vocational college or polytechnic school (60.3%) Vocational school (28.9%) Junior high school (3.1%) Not in school (3.8%) Data missing (3.9%) Behavioral disinhibitionc at 17 y Low behavioral disinhibition (48.3%) High behavioral disinhibition (51.6%) Sex Male (46.0%) Female (54.0%) BMId, kg/m2 o20 (49.5%) 20–22.9 (39.4%) 23–24.9 (6.8%) 25 or more (4.3%) Exercise Daily (15.5%) 4–5 times a week (15.2%) 2–3 times a week (28.3%) About once a week (19.3%) Once or twice a month (10.6%) Less than once a month (6.5%) Never (4.7%)

Breakfast a few times a week

Breakfast once a week or less often

1.0 1.43 1.86 1.37

(reference) (1.04–1.96)* (1.48–2.34)*** (0.99–1.91)

1.0 1.86 2.75 1.60

(reference) (1.33–2.59)*** (2.14–3.52)*** (1.11–2.30)*

1.0 1.70 1.70 0.98

(reference) (1.29–2.23)*** (1.36–2.13)*** (0.74–1.30)

1.0 1.42 2.12 1.73

(reference) (1.01–1.97)* (1.65–2.73)*** (1.29–2.32)***

1.0 (reference) 1.36 (1.10–1.68)**

1.0 (reference) 1.40 (1.11–1.77)*

1.0 (reference) 1.40 (1.15–1.70)***

1.0 (reference) 1.37 (1.12–1.69)**

1.0 1.34 1.54 1.87 1.81

1.0 1.43 2.51 2.04 2.47

(reference) (1.10–1.64)** (0.91–2.60) (1.23–2.84)** (1.17–2.79)**

(reference) (1.16–1.77)*** (1.56–4.06)*** (1.38–3.02)*** (1.63–3.75)***

1.0 (reference) 1.06 (0.89–1.26)

1.0 (reference) 1.25 (1.04–1.50)*

1.0 (reference) 1.37 (1.15–1.63)***

1.0 (reference) 1.47 (1.21–1.78)***

1.0 1.09 1.18 1.41

(reference) (0.91–1.30) (0.85–1.65) (0.93–2.14)

1.0 1.14 0.92 2.00

(reference) (0.94–1.38) (0.63–1.34) (1.32–3.01)***

1.0 1.37 1.17 1.52 1.34 1.28 1.05

(reference) (0.99–1.88) (0.87–1.57) (1.13–2.06)** (0.94–1.91) (0.85–1.93) (0.65–1.70)

1.0 0.85 1.16 1.23 1.79 2.38 2.46

(reference) (0.59–1.23) (0.85–1.58) (0.88–1.70) (1.25–2.57)*** (1.62–3.49)*** (1.61–3.76)***

*Po0.05, ** Po0.01, *** Pr0.001 a Age of puberty onset, interactions between alcohol use, smoking, disinhibition, and between age of puberty onset and sex were included in the original multivariable model but could be removed without a significant decrease in model fit. b Confidence intervals adjusted for clustered sampling. c For 180 disinhibition scores missing at 17 y, disinhibition scores at 18 y were substituted. Odds ratios were calculated per unit increase. d For 118 BMI values missing at 16 y, BMI values at 17 y were substituted. Odds ratios were calculated per unit increase. Rounding error in sums of some percentages.

more likely to skip breakfast than children of regular breakfast eaters. This suggests that breakfast skipping is not a problem that can be solved solely by approaching teenagers; breakfast endorsing programs that address the entire family or just parents may be more effective. However, the European Journal of Clinical Nutrition

familial transmission of breakfast eating/breakfast skipping is a very complex issue: we have explored the genetic and environmental influences on breakfast-eating patterns in more detail elsewhere (Keski-Rahkonen et al, in press). From those previous analyses, we know that the parent–offspring

Breakfast skipping and health-compromising behaviors A Keski-Rahkonen et al

849 Table 5 Correlates of breakfast eating in adults: multinomial logistic regression models adjusted for age and sex Breakfast skippers vs eaters Adjusted odds ratios (95% confidence Interval a) Correlates Family socioeconomic status, N=4523 Upper-level employee (19.3%) Self-employed, including self-employed farmers (16.1%) Lower-level employee (34.0%) Manual worker (29.1%) Student, retired, or unclassified occupation (1.5%) Smoking, N=4602 Never (40.8%) Given up temporarily or permanently (33.7%) Current smoker (25.6%) Alcohol use, N=4625 Less than twice a month (41.0%) 3–8 times a month (41.5%) Over 8 times a month (17.5%) Tea, cups per day, N=4430 0 (67.1%) 1 (13.4%) 2 (11.8%) 3 or more (7.7%) Coffee, cups per day, N=4640 0 (6.0%) 1–4 (41.6%) 5–8 (41.4%) 9 or more (11.0%) Type of milk used, N=4648 No milk (34.0%) Non-fat milk (24.0%) Low-fat 1% milk (4.9%) Low-fat 1.9% milk (28.2%) 3.5% or full-fat milk (8.9%) Type of bread spread used, N=4656 No bread spread (9.6%) Low-fat spread (13.7%) Margarine (42.9%) Butter-margarine spread (18.8%) Butter (13.1%) Other (1.9%) Highest education, N=4657 University degree (4.5%) Senior high school graduation (17.0%) Senior high dropout or vocational school (21.2%) Mandatory education only (57.3%) Unemployment, N=4570 Employed (83.2%) Unemployed (16.8%) BMI, kg/m2, N=4608 o20 (4.9%) 20–24.9 (44.9%) 25–29.9 (38.1%) 30 (12.2%) Amount of sleep, N=4651 6.5 h or less (19.7%) 7–7.5 h (47.9%) 8–8.5 h (28.2%) 9 h or more (4.2%) Feeling tired in the morning, N=4648 Less than weekly (43.1%) More than weekly (56.9%) Use of natural and herbal drugs, N=4519 No use (68.2%) Occasional use (26.5%) Regular use (5.3%)

Breakfast a few times a week

1.0 1.18 1.56 2.18 1.89

(reference) (0.83–1.67) (1.16–2.09)** (1.64–2.89)*** (0.89–4.00)

Breakfast once a week or less often

1.0 1.69 1.74 2.69 1.98

(reference) (1.29–2.21)*** (1.37–2.22)*** (2.14–3.40)*** (1.06–3.70)*

1.0 (reference) 1.11 (0.89–1.39) 2.06 (1.63–2.61)***

1.0 (reference) 1.35 (1.11–1.62)** 3.53 (2.91–4.28)***

1.0 (reference) 1.19 (0.97–1.47) 0.84 (0.63–1.12)

1.0 (reference) 1.12 (0.94–1.33) 1.19 (0.96–1.48)

1.0 0.61 0.56 0.59

(reference) (0.45–0.82)*** (0.40–0.77)*** (0.40–0.88)**

1.0 0.46 0.51 0.52

(reference) (0.35–0.59)*** (0.39–0.65)*** (0.38–0.71)***

1.0 1.41 1.91 2.01

(reference) (0.90–2.19) (1.22–3.00)** (1.21–3.36)**

1.0 1.24 1.93 3.52

(reference) (0.87–1.789 (1.35–2.75)*** (2.38–5.20)***

1.0 1.39 1.09 1.58 1.84

(reference) (1.08–1.79)** (0.66–1.79) (1.24–2.00)*** (1.28–2.63)***

1.0 0.92 1.00 1.39 2.45

(reference) (0.74–1.13) (0.69–1.44) (1.15–1.67)*** (1.88–3.19)***

1.0 1.13 1.14 1.30 1.14 0.63

(reference) (0.76–1.68) (0.81–1.60) (0.89–1.89) (0.75–1.75) (0.26–1.54)

1.0 1.13 1.28 1.33 1.74 1.15

(reference) (0.81–1.58) (0.98–1.69) (0.98–1.80) (1.27–2.39)*** (0.65–2.06)

1.0 1.03 1.63 1.99

(reference) (0.59–1.78) (0.96–2.75) (1.21–3.29)**

1.0 1.29 1.99 2.54

(reference) (0.80–2.07) (1.27–3.12)** (1.64–3.92)***

1.0 (reference) 1.61 (1.29–2.02)***

1.0 (reference) 1.13 (0.93–1.38)

1.0 1.48 1.51 2.29

(reference) (0.87–2.54) (0.87–2.61) (1.29–4.06)**

1.0 0.74 0.90 1.03

(reference) (0.53–1.03) (0.65–1.26) (0.71–1.50)

1.0 0.88 0.85 0.77

(reference) (0.69–1.12) (0.65–1.12) (0.46–1.29)

1.0 0.61 0.64 0.70

(reference) (0.50–0.73)*** (0.52–0.78)*** (0.48–1.01)

1.0 (reference) 1.10 (0.92–1.33)

1.0 (reference) 1.17 (1.01–1.36)*

1.0 (reference) 1.08 (0.87–1.34) 0.75 (0.45–1.24)

1.0 (reference) 0.76 (0.63–0.91)** 0.74 (0.51–1.07)

European Journal of Clinical Nutrition

Breakfast skipping and health-compromising behaviors A Keski-Rahkonen et al

850

Table 5 (continued) Breakfast skippers vs eaters Adjusted odds ratios (95% confidence Interval a) Correlates

Breakfast a few times a week

Use of vitamin and trace mineral supplements, N=4576 No use (44.4%) Occasional use (42.1%) Regular use (13.6%) Exercise N=4550 Exercise 6 times a month or more (41.4%) Exercise 3–5 times a month (30.1%) Exercise 1–2 times a month (24.6%) Exercise less than once a month (3.9%)

Breakfast once a week or less often

1.0 (reference) 1.03 (0.84–1.26) 0.67 (0.49–0.94)*

1.0 (reference) 0.85 (0.72–0.99)* 0.62 (0.48–0.80)***

1.0 1.14 1.18 0.89

1.0 1.16 1.69 2.83

(reference) (0.90–1.45) (0.93–1.49) (0.50–1.56)

(reference) (0.97–1.38) (1.40–2.03)*** (2.00–4.00)***

*Po0.05, **Po0.01, ***Pr0.001 a Confidence intervals adjusted for clustered sampling. Rounding error in sums of some percentages.

Table 6 Multivariablea model of correlates of adult breakfast eating, N=4519 Breakfast skippers (vs eaters) Adjusted odds ratios (95% confidence Interval b) Correlates

Breakfast a few times a week

Breakfast once a week or less often

1.0 (reference) 1.40 (1.14–1.72)***

1.0 (reference) 2.02 (1.70–2.41)***

1.0 (reference) 1.11 (0.90–1.37)

1.0 (reference) 1.20 (1.01–1.42)*

1.0 1.05 1.62 1.99

(reference) (0.60–1.86) (0.95–2.77) (1.19–3.34)**

1.0 1.46 1.92 2.66

(reference) (0.90–2.38) (1.21–3.06)** (1.70–4.15)***

1.0 1.32 1.33 1.98

(reference) (0.74–2.35) (0.74–2.40) (1.07–3.65)*

1.0 0.91 1.08 1.11

(reference) (0.62–1.35) (0.73–1.60) (0.72–1.71)

1.0 0.77 0.62 0.39

(reference) (0.59–1.02) (0.45–0.92)* (1.15–1.06)

1.0 0.81 0.67 0.50

(reference) (0.66–1.01) (0.51–0.89)** (0.24–1.03)

1.0 1.08 1.20 0.83 0.83

(reference) (0.85–1.38) (0.92–1.55) (0.41–1.67) (0.41–1.67)

1.0 1.22 1.81 2.52 1.46

(reference) (1.02–1.48)* (1.48–2.21)*** (1.76–3.59)*** (0.91–2.33)

Smoking Has never smoked regularly (40.6%) Has smoked or still smokes regularly (59.4%) Alcohol use Less than 2 days a month (40.7%) 3 days a month or more often (59.3%) Highest education University degree (4.6%) Senior high school graduation (17.2%) Senior high school dropout or vocational school (21.4%) Mandatory education only (56.9%) BMI o20 (3.8%) 20–24.9 (45.4%) 25–29.9 (38.6%) 30 (12.3%) Age 30–39 y (14.9%) 40–49 y (67.3%) 50–59 y (16.4%) 60 y or older (1.3%) Exercise Exercise 6 times a month or more (40.3%) Exercise 3–5 times a month (29.9%) Exercise 1–2 times a month (23.9%) Exercise less than once a month (3.8%) Data missing (2.2%) Sex Male (48.0%) Female (52.0%)

1.0 (reference) 0.64 (0.53–0.77)***

1.0 (reference) 0.73 (0.63–0.85)***

*Po0.05, **Po0.01, ***Pr0.001. a The interaction between alcohol use and smoking could be dropped from the model without a significant decrease in model fit. b Confidence intervals adjusted for clustered sampling. Rounding error in sums of some percentages.

resemblance is mostly because of genes; the parental generation has little direct behavioral impact on the offspringFquite often teenage children, especially boys, act in European Journal of Clinical Nutrition

deliberate opposition to their parents. Thus, expecting teenagers to imitate their parents’ breakfast habits may not be the most successful strategy. Probably, the most effective

Breakfast skipping and health-compromising behaviors A Keski-Rahkonen et al

851 Table 7 Co-occurrence of health-compromising factors in adults and adolescents Adults

Co-occurring factors Smokinga b

Smoking, alcohol use

Smoking, lack of exercisec Smoking, alcohol use, lack of exercise Smoking, alcohol use, lack of exercise, overweightd Smoking, alcohol use, lack of exercise, obesitye

Adolescents

All adults, N = 4660

Breakfast skippers, N = 1546

Breakfast eaters, N = 3114

n%

n%

n%

2683 57.6 1822 39.0 753 16.2 493 10.6 273

1093 70.7 747 48.3 368 24.3 241 15.8 139

1590 51.1 1075 34.5 385 12.4 252 8.1 134

5.9 72 1.5

9.1 30 2.0

4.3 42 1.3

P

*** *** *** *** ***

All adolescents, N = 5448

Breakfast skippers, N=1587

Breakfast eaters, N = 3861

n%

n%

n%

3727 68.4 1373 25.2 901 16.5 406 7.5 18

1217 76.7 524 33.3 367 23.2 183 11.6 9

2510 65.0 849 22.0 534 13.8 223 5.8 9

w

0.3 1 0.02

0.6 1 0.06

0.2 0 0

P

*** *** *** *** ** w

a

Past or present smoking. Adults: alcohol use 3–8 days a month or more often; adolescents, alcohol use once a week or more often. c Exercise less than once a week. d Defined as body mass index (BMI)>25. e Defined as BMI>30. **Po0.01, *** Po0.001, wP value could not be calculated due to small cell sizes. b

way to influence the offspring’s breakfast habits is to create a family and peer atmosphere that endorses general healthconscious behavior. It is also important to note that parental influence on the offspring’s breakfast eating is likely to be age-specific, because the eating habits of teenagers are less under parental control than those of young children. In our study, the within-individual clustering of healthcompromising factors (smoking, alcohol use, and sedentary lifestyle) was much more conspicuous among the adult and adolescent breakfast skippers than among the breakfast eaters. A previous study (Terre et al, 1990) examined agedependent clustering of different health-related behaviors, including smoking, alcohol use, and sedentary lifestyle, in adolescents aged 11–18 y. Breakfast skipping clustered with type A personality at 11 y and with sedentary lifestyle at 12– 15 y. At 16–18 y, breakfast skipping existed independently of other behaviors. Our results do not support the finding that breakfast skipping at the age of 16 or in adulthood is independent of other risk behaviors, but clustering of healthcompromising behaviors may vary with age. Unfortunately, we were not able to measure how correlates of breakfast eating vary in different stages of adolescence. During early and mid-adolescence, disordered eating patterns are very strongly associated with health-compromising behaviors (eg alcohol, tobacco, and marijuana use); individuals with early puberty onset may be particularly vulnerable (Wilson et al, 1994; Dick et al, 2000; Abraham & O’Dea, 2001; KaltialaHeino et al, 2001; Rose et al, 2001). The association of health-

compromising behaviors and disordered eating becomes much weaker on the verge of adulthood (Neumark-Sztainer et al, 1997). Our findings suggest that the association of breakfast skipping (very mildly disordered eating) with smoking, alcohol use, and sedentary lifestyle exists throughout adulthood. However, in our cross-sectional adult sample, breakfast skipping is less common among the older than the younger individuals. It may be that eating habits become more regular with age; equally well breakfast eating may become less common among the younger generations, who will continue their lower rates of breakfast consumption through adulthood. Longitudinal studies on adult breakfast eating are needed to clarify this point. In our sample, children of breakfast-skipping parents were much more likely to have a high BMI and exercise infrequently than children of regular breakfast eaters. However, alcohol use and smoking were equally common in adolescent children of breakfast-eating and breakfastskipping families. This suggests that the transgenerational clustering of health-compromising behaviors is a complex issue meriting further study. Breakfast skipping was associated with low family SES in adults and adolescent boys, but not in girls. Several earlier studies have found that breakfast skipping is particularly prevalent among individuals from low SES families (Pastore ¨ glund et al, 1998; et al, 1996; Brugman et al, 1998; Ho Nordlund & Jacobson, 1999; O’Dea & Caputi, 2001). Our findings suggest that in teenage girls, factors other than European Journal of Clinical Nutrition

Breakfast skipping and health-compromising behaviors A Keski-Rahkonen et al

852 family SES may be more influential in determining eating patterns: for instance, dieting and body shape ideals could be homogenous influences across social classes. In our sample, the breakfast eating of girls is much more influenced by environmental factors than that of boys (Keski-Rahkonen et al, in press). In our study, we were able to measure the frequency and correlates of breakfast eating, but not factors directly determining breakfast eating. Thus, the factors we studied explained only a very low proportion of the variance in breakfast eating (for the full multivariable models, the pseudo-R2 statistic was 0.07 for adolescents and 0.04 for adults). Many factors important for regular breakfast eating were overlooked: these include, for example, total daily energy intake, time allotted for breakfast each day, and dieting-related matters. Future studies should address these questions more specifically. In the adult sample, however, we studied factors that may indirectly measure time available for breakfast each morning: amount of sleep, shiftwork, and tiredness in the morning were much more weakly correlated with breakfast eating than the health-compromising behaviors already mentioned. Another limitation of this study is that the BMIs of our study population were based on selfreported height and weight. Although self-reported height and weight may be unreliable in some population subgroups, for example, the very young and the elderly (Himes & Faricy, 2001; Kuczmarski et al, 2001), and although women in particular underestimate their weight and overestimate their height (Kuskowska-Wolk et al, 1989), the correlation between measured and self-reported height and weight has commonly been more than 0.85 in the age groups relevant for our study (Rowland, 1990; Giacchi et al, 1998; Himes & Faricy, 2001); thus, we feel that self-reported data can be used. However, because of the self-reporting bias, prevalences of overweight and obesity are probably underestimated in our study population. To what extent are results of this twin study generalizable to the nontwin population? The prevalences of breakfast eating in our study sample are similar to those of Nordic nontwin populations of comparable age (Puska & Smolan¨ glund et al, 1998; Nordlund & Jacobson, 1999; der, 1980; Ho Urho & Hasunen, 1999). The generalizability of our findings is further strengthened by the large, population-based subject sample, with education level and SES comparable to that of the general population. The high response rate also ensures that the nonresponder bias is small. We also accounted for clustered sampling within families in all of our analyses. Thus, we feel that our findings fairly well represent the Finnish population of respective age groups. Skipping breakfast reflects more than simply meal timing preferences. It appears to be one component of frequently co-occurring health-compromising behaviors. Individuals who skip breakfast may care less about their health than individuals who always eat breakfast. As breakfast skippers are more likely to be overweight than breakfast eaters, increased weight may be a result of making unhealthy food European Journal of Clinical Nutrition

choices to make up for a missed breakfast. Starting the day without the first meal may also be an attempt to control weight. Sometimes smoking is used to augment dieting. Smoking, more common among breakfast skippers than breakfast eaters, may also suppress appetite in the morning, or may interfere with the time allotted for breakfast. Simple nutritional interventions aimed at increasing the frequency of breakfast eating may fail to address these more complex contextual issues. Discouraging smoking and substance use in tandem with promoting regular exercise and meals is one way of approaching this problem. As parental influences are important determinants of adolescent breakfast eating, getting the parents to eat breakfast regularly may be a step toward getting their children to eat breakfast as well. Probably the most effective strategy to influence the offspring’s breakfast habits is to create family and peer atmospheres that endorse generally health-conscious lifestyles. More detailed studies of determinants of breakfast eating are needed to shape these strategies.

References Aarnio M, Kujala UM & Kaprio J (1997): Associations of healthrelated behaviors, school type and health status to physical activity patterns in 16 year old boys and girls. Scand. J. Soc. Med. 25, 156–167 Abraham S & O’Dea JA. (2001): Body mass index, menarche, and perception of dieting among peripubertal adolescent females. Int. J. Eat. Disord. 29, 23–28. Baumert Jr PW, Henderson JM & Thompson NJ (1998): Health risk behaviors of adolescent participants in organized sports. J. Adolesc. Health 22, 460–465. Bellisle F, Monneuse MO, Steptoe A & Wardle J (1995): Weight concerns and eating patterns: a survey of university students in Europe. Int. J. Obes. Relat. Metab. Disord. 19, 723–730. ¨ m E, Hernell O & Persson LA (1996): Cardiovascular risk Bergstro indicators cluster in girls from families of low socio-economic status. Acta Paediatr. 85, 1083–1090. Brugman E, Meulmeester JF, Spee-van der Wekke A & VerlooveVanhorick SP (1998): Breakfast-skipping in children and young adolescents in The Netherlands. Eur. J. Public Health 8, 325–328. Cavadini C, Decarli B, Grin J, Narring F & Michaud PA (2000): Food habits and sport activity during adolescence: differences between athletic and non-athletic teenagers in Switzerland. Eur. J. Clin. Nutr. 54 (Suppl 1), S16–S20. Dick DM, Rose RJ, Viken RJ & Kaprio J (2000): Pubertal timing and substance use: associations between and within families across late adolescence. Dev. Psychol. 36, 180–189. Giacchi M, Mattei R & Rossi S (1998). Correction of the self-reported BMI in a teenage population. Int. J. Obes. Relat. Metab. Disord. 2, 673–677. Himes JH & Faricy A (2001). Validity and reliability of self-reported stature and weight of US adolescents. Am. J. Hum. Biol. 13, 255–260. Hosmer DW & Lemeshow S (2000): Polychotomous independent variable. In Applied Logistic Regression, 2nd Edition, pp 56–62 New York: Wiley. ¨ glund D, Samuelson G & Mark A (1998): Food habits in Swedish Ho adolescents in relation to socioeconomic conditions. Eur. J. Clin. Nutr. 52, 784–789. Isralowitz RE & Trostler N (1996): Substance use: toward an understanding of its relation to nutrition-related attitudes and behavior among Israeli high school youth. J. Adolesc. Health 19, 184–189.

Breakfast skipping and health-compromising behaviors A Keski-Rahkonen et al

853 Kaltiala-Heino R, Rimpela¨ M, Rissanen A, & Rantanen P (2001): Early puberty and early sexual activity are associated with bulimic-type eating pathology in middle adolescence. J. Adolesc. Health 28, 346– 352. Keski-Rahkonen A, Kaprio J, Viken RJ, Rissanen A & Rose RJ (in press): Genetic and environmental factors in breakfast eating patterns. Behav. Genet. Kuczmarski MF, Kuczmarski RJ & Najjar M (2001): Effects of age on validity of self-reported height, weight, and body mass index: findings from the Third National Health and Nutrition Examination Survey, 1988–1994. J. Am. Diet. Assoc. 101, 28–34. ¨ ssner S (1989): The Kuskowska-Wolk A, Karlsson P, Stolt M & Ro predictive validity of body mass index based on self-reported weight and height. Int. J. Obes. Relat. Metab. Disord. 13, 441–453. Melve KK & Baerheim A (1994): Signs of subclinical eating disorders in teenage girls. Scand. J. Prim. Health Care 12, 197–203. Neumark-Sztainer D, Story M, Toporoff E, Himes JH, Resnick MD & Blum RW (1997): Covariations of eating behaviors with other health-related behaviors among adolescents. J. Adolesc. Health 20, 450–458. Nordlund G & Jacobson T (1999): Gymnasieelevernas matvanor relaterat till hur de ma˚r och ka¨nner sig, deras fysiska aktivitet, deras skolprestationer ock sociala bakgrund [High school students’ food habits in relation to their well-being, physical activity, school performance and social background; in Swedish]. Pedagogiska rapporter, Vol 58, Umea˚: Umea˚ Universitetet. O’Dea JA & Caputi P (2001): Association between socioeconomic status, weight, age and gender, and the body image and weight control practices of 6- to 19-year-old children and adolescents. Health Educ. Res. 16, 521–532. Pastore DR, Fisher M & Friedman SB (1996): Abnormalities in weight status, eating attitudes, and eating behaviors among urban high school students: correlations with self-esteem and anxiety. J. Adolesc. Health 18, 312–319. ¨ n terveysPuska P & Smolander A (1980): Suomalaisen aikuisva¨esto ka¨ytta¨ytyminen, keva¨t 1980 [The health habits of Finnish adults, spring 1980; in Finnish]. Kansanterveystieteen laitoksen julkaisuja, B1. Rao JNK & Scott AJ (1984): On chi-squared tests for multiway contingency tables with cell proportions estimated from survey data. Ann. Stat. 12, 46–60.

Revicki D, Sobal J & DeForge B (1991): Smoking status and the practice of other unhealthy behaviors. Fam. Med. 23, 361–364. Rose RJ, Dick DM, Viken RJ, Pulkkinen L & Kaprio J (2001): Drinking and abstaining at age 14? A genetic epidemiological study. Alcohol Clin. Exp. Res. 25, 1594–1604. Rose RJ, Kaprio J, Winter T, Koskenvuo M, & Viken RJ (1999): Familial and socioregional environmental effects on abstinence from alcohol at age sixteen. J. Stud. Alcohol Suppl. 13, 63–74. Rowland ML (1990): Self-reported weight and height. Am. J. Clin. Nutr. 52, 1125–1133. SAS System for Windows (1999): Cary, NC: SAS Institute Inc. Serra Majem L, Manno S, Ribas Barba L, Perez Rodrigo C, & Aranceta Bartrina J (2000): Desayuno y obesidad [Breakfast and obesity; in Spanish]. In: Desayuno y equilibrio alimentario: Estudio enKid, eds. L Serra Majem & J Aranceta Bartrina, pp 31–44. Barcelona: Masson. Shaw ME (1998): Adolescent breakfast skipping: an Australian study. Adolescence 33, 851–861. Sosioekonomisen aseman luokitus 1989 (1990): [The classification of socioeconomic status, 1989; in Finnish]. Ka¨sikirjoja Vol 17. Helsinki: Tilastokeskus. Stata, Version. 7.0 (2000) College Station, TX: Stata Corporation. Terre L, Drabman RS & Meydrech EF (1990): Relationships among children’s health-related behaviors: a multivariate, developmental perspective. Prev. Med. 19, 134–146. Urho U-M & Hasunen K (1999): Yla¨asteen kouluruokailu 1998 [School meals in junior high school in 1998; in Finnish]. Sosiaalija terveysministerio¨n selvityksia¨ 5. Walker A, Walker B, Jones J & Ncongwane J (1982): Breakfast habits of adolescents in four South African populations. Am. J. Clin. Nutr. 36, 650–656. Wilson DM, Killen JD, Hayward C, Robinson TN, Hammer LD, Kraemer HC, Varady A & Taylor CB (1994): Timing and rate of sexual maturation and the onset of cigarette and alcohol use among teenage girls. Arch. Pediatr. Adolesc. Med. 148, 789–795. Wolfe WS, Campbell CC, Frongillo Jr EA, Haas JD & Melnik TA (1994): Overweight schoolchildren in New York State: prevalence and characteristcs. Am. J. Public Health 84, 807–813. Zuckerman M (1979): Sensation Seeking: Beyond the Optimal Level of Arousal. Hillsdale, NJ: Lawrence Erlbaum Associates.

European Journal of Clinical Nutrition