Vol. 19, No. 3 Printed in Great Britain
International Journal of Epidemiology ©International Epidemiological Association 1990
Relationship between Adiposity and Food Intake: An Example of PseudoContradictory Results Obtained in Case-Control versus BetweenPopulations Studies Rolland-Cachera M-F (Section de Nutrition, Centre de Recherche du Vesinet, Institut National de la Sante et de la Recherche Medicale, 44Chemin de Ronde, BP34,78110 Le Vesinet, France), Bellisle F,Tichet J, Chantrel A-M, GuilloudBalaille M, Vol S and Pequignot G. Relationship between adiposity and food intake: an example of pseudo-contradictory results obtained in case-control versus between-populations studies. International Journal ofEpidemiology1990, 19: 572-577. Seemingly contradictory data support controversies concerning the relationships between food intake and illnesses. The present study of 1035 adults, aged 30-39 years, shows that (1) daily energy intake is not higher in obese than in non-obese people, (2) obesity is more prevalent in social groups where energy intake is higher. These pseudocontradictory results can be reconciled on the basis of a constitution/environment interaction. A comparison of groups based on presence or absence of illness (obesity, hypertension, coronary heart disease (CHD), cancer, etc) shows that some people can develop risk factors, even though their feeding behaviour is normal. This result (no direct relationship) underlines differences in individual susceptibility. When comparisons are made between populations with different diets the results (direct relationship) express environmental factors. The hypotheses on behavioural contribution to the aetiology of certain diseases appears more clearly in between-population comparisons than in case-control studies. Results of comparisons between populations (if the hypotheses they suggest are confirmed by intervention studies) warrant prevention at the level of populations, while results of case-control studies justify particular prevention in subjects at risk.
In two previous studies in children,'-2 we showed that although no correlation appeared between adiposity and- food intake, the prevalence of overweight and obesity was related to the socially determined level of intake. These results echoed the apparent contradiction found in the obesity literature: many studies, both field- and laboratory-based, have failed to establish any difference in food intake between lean and obese adults,3"10 adolescents""14 and infants15"17 while other studies have shown a higher prevalence of overweight in populations where intake is higher.18"22 The relation-
ship between obesity and energy intake which does not appear on an individual basis becomes apparent when populations are compared. The same 'contrast' found 'within' and 'between' groups has been described for other public health issues about the relationship between intake and risk factors or illnesses such as salt intake and hypertension,23 fat intake and coronary heart disease (CHD),4 fat intake and breast cancer.24 Our results comparing two approaches for studying the relationship between intake and corpulence on the same sample suggest a hypothesis to account for the apparent contradiction in the literature. This hypothesis rests on the interaction between constitutional and environmental factors. The respective role of these factors in the aetiology of obesity will appear different depending on how the compared groups are selected. In the present study, the relationship between food intake and corpulence were examined in adults both as
•Scction de Nutrition, Centre de Recherche du Vesinet, Institut National dc la Same' et de la Recherche Midicale, 44 Chemin dc Ronde, BP34, 78110 Le Vesinet, France. *°Laboratoire de Neurobiologie de la Nutrition, EPHE, University Pierre el Marie Curie, 4 place Jussieu, 75252 Paris, France. tl R.S.A. (Inslitut Regional pour la Santii), 45 rue de Parmenticre, BP 122, 37521 La Riche Gidex, France.
571
Downloaded from http://ije.oxfordjournals.org/ at Harvard University on June 30, 2015
MARIE-FRANCOISE ROLLAND-CACHERA,* FRANCE BELUSLE,** JEAN TICHET.t ANNE-MARIE CHANTREL,* MICHEL GUILLOUD-BATAILLE,* SYLVIANE VOLt AND GEORGES PEQUIGNOT*
572
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
a function of corpulence status and on the basis of socioeconomic level. METHOD Weight and height data from 9439 subjects aged 30 to 39 years were obtained from the I.R.S.A. (Institut Regional pour la Sant6). The subjects who were in employment voluntarily took advantage of a free medical check-up through social insurance supplied by the 'Caisse Nationale d'Assurance Maladie des Travailleurs Salaries' (CNAMTS). The Quetelet Index (wt/ hr2) was used to assess corpulence.
Socioeconomic Subgroups A partition of the sample was made according to the subjects' profession: semiskilled and unskilled workers (258 males and 191 women), executives and skilled workers (279 men and 307 women), x2 test was used to compare the percent obese in the two social classes. Analyses of variance were used to compare anthropometric measurements and food intake between the five corpulence classes and between two socioeconomic groups. Food Consumption Burke's Dietary History Method" as adapted by Cubeau26 and validated by Pequignot27 was used to
TABLE 1. Anthopomctric measurements and food consumption in subjects aged 30 to 39 years of the five corpulence classes Men 1
wt/ht percentiles
wt/ht2 (kg/m2) Energy (kcal) Nutrients (g) Proteins Fat Carbohydrates Alcohol % daily energy intake: Protein Fat Carbohydrates Alcohol
30th
5th
70th
95th
Total (n=537) m±sd
Lean (n = 102) mlsd
Slim (n-112) mlsd
Average (n-109) mlsd
Fat (n = 109) m±sd
Obese (n = 105) mlsd
24 5 ±4.3 3101 ±712
1 9 4 1 1.1 3099±644
21 611.1 3135 ± 749
23.81 1.1 31521738
26.9 ± 1.3 3091 ±738
30.912.5 3025±687
International Journal of Epidemiology ©International Epidemiological Association 1990
Relationship between Adiposity and Food Intake: An Example of PseudoContradictory Results Obtained in Case-Control versus BetweenPopulations Studies Rolland-Cachera M-F (Section de Nutrition, Centre de Recherche du Vesinet, Institut National de la Sante et de la Recherche Medicale, 44Chemin de Ronde, BP34,78110 Le Vesinet, France), Bellisle F,Tichet J, Chantrel A-M, GuilloudBalaille M, Vol S and Pequignot G. Relationship between adiposity and food intake: an example of pseudo-contradictory results obtained in case-control versus between-populations studies. International Journal ofEpidemiology1990, 19: 572-577. Seemingly contradictory data support controversies concerning the relationships between food intake and illnesses. The present study of 1035 adults, aged 30-39 years, shows that (1) daily energy intake is not higher in obese than in non-obese people, (2) obesity is more prevalent in social groups where energy intake is higher. These pseudocontradictory results can be reconciled on the basis of a constitution/environment interaction. A comparison of groups based on presence or absence of illness (obesity, hypertension, coronary heart disease (CHD), cancer, etc) shows that some people can develop risk factors, even though their feeding behaviour is normal. This result (no direct relationship) underlines differences in individual susceptibility. When comparisons are made between populations with different diets the results (direct relationship) express environmental factors. The hypotheses on behavioural contribution to the aetiology of certain diseases appears more clearly in between-population comparisons than in case-control studies. Results of comparisons between populations (if the hypotheses they suggest are confirmed by intervention studies) warrant prevention at the level of populations, while results of case-control studies justify particular prevention in subjects at risk.
In two previous studies in children,'-2 we showed that although no correlation appeared between adiposity and- food intake, the prevalence of overweight and obesity was related to the socially determined level of intake. These results echoed the apparent contradiction found in the obesity literature: many studies, both field- and laboratory-based, have failed to establish any difference in food intake between lean and obese adults,3"10 adolescents""14 and infants15"17 while other studies have shown a higher prevalence of overweight in populations where intake is higher.18"22 The relation-
ship between obesity and energy intake which does not appear on an individual basis becomes apparent when populations are compared. The same 'contrast' found 'within' and 'between' groups has been described for other public health issues about the relationship between intake and risk factors or illnesses such as salt intake and hypertension,23 fat intake and coronary heart disease (CHD),4 fat intake and breast cancer.24 Our results comparing two approaches for studying the relationship between intake and corpulence on the same sample suggest a hypothesis to account for the apparent contradiction in the literature. This hypothesis rests on the interaction between constitutional and environmental factors. The respective role of these factors in the aetiology of obesity will appear different depending on how the compared groups are selected. In the present study, the relationship between food intake and corpulence were examined in adults both as
•Scction de Nutrition, Centre de Recherche du Vesinet, Institut National dc la Same' et de la Recherche Midicale, 44 Chemin dc Ronde, BP34, 78110 Le Vesinet, France. *°Laboratoire de Neurobiologie de la Nutrition, EPHE, University Pierre el Marie Curie, 4 place Jussieu, 75252 Paris, France. tl R.S.A. (Inslitut Regional pour la Santii), 45 rue de Parmenticre, BP 122, 37521 La Riche Gidex, France.
571
Downloaded from http://ije.oxfordjournals.org/ at Harvard University on June 30, 2015
MARIE-FRANCOISE ROLLAND-CACHERA,* FRANCE BELUSLE,** JEAN TICHET.t ANNE-MARIE CHANTREL,* MICHEL GUILLOUD-BATAILLE,* SYLVIANE VOLt AND GEORGES PEQUIGNOT*
572
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
a function of corpulence status and on the basis of socioeconomic level. METHOD Weight and height data from 9439 subjects aged 30 to 39 years were obtained from the I.R.S.A. (Institut Regional pour la Sant6). The subjects who were in employment voluntarily took advantage of a free medical check-up through social insurance supplied by the 'Caisse Nationale d'Assurance Maladie des Travailleurs Salaries' (CNAMTS). The Quetelet Index (wt/ hr2) was used to assess corpulence.
Socioeconomic Subgroups A partition of the sample was made according to the subjects' profession: semiskilled and unskilled workers (258 males and 191 women), executives and skilled workers (279 men and 307 women), x2 test was used to compare the percent obese in the two social classes. Analyses of variance were used to compare anthropometric measurements and food intake between the five corpulence classes and between two socioeconomic groups. Food Consumption Burke's Dietary History Method" as adapted by Cubeau26 and validated by Pequignot27 was used to
TABLE 1. Anthopomctric measurements and food consumption in subjects aged 30 to 39 years of the five corpulence classes Men 1
wt/ht percentiles
wt/ht2 (kg/m2) Energy (kcal) Nutrients (g) Proteins Fat Carbohydrates Alcohol % daily energy intake: Protein Fat Carbohydrates Alcohol
30th
5th
70th
95th
Total (n=537) m±sd
Lean (n = 102) mlsd
Slim (n-112) mlsd
Average (n-109) mlsd
Fat (n = 109) m±sd
Obese (n = 105) mlsd
24 5 ±4.3 3101 ±712
1 9 4 1 1.1 3099±644
21 611.1 3135 ± 749
23.81 1.1 31521738
26.9 ± 1.3 3091 ±738
30.912.5 3025±687
