Liese AD, Schulz M, Moore CG, Mayer-Davis EJ. Dietary patterns, insulin sensitivity and adiposity in the multi-ethnic insulin resistance atherosclerosis study population. Br J Nutr 92, 973-984

Department of Epidemiology and Biostatistics, Arnold School of Public Health, University of South Carolina, Columbia 29208, USA.
British Journal Of Nutrition (Impact Factor: 3.45). 01/2005; 92(6):973-84. DOI: 10.1079/BJN20041279
Source: PubMed


Epidemiological investigations increasingly employ dietary-pattern techniques to fully integrate dietary data. The present study evaluated the relationship of dietary patterns identified by cluster analysis with measures of insulin sensitivity (SI) and adiposity in the multi-ethnic, multi-centre Insulin Resistance Atherosclerosis Study (IRAS, 1992-94). Cross-sectional data from 980 middle-aged adults, of whom 67 % had normal and 33 % had impaired glucose tolerance, were analysed. Usual dietary intake was obtained by an interviewer-administered, validated food-frequency questionnaire. Outcomes included SI, fasting insulin (FI), BMI and waist circumference. The relationship of dietary patterns to log(SI+1), log(FI), BMI and waist circumference was modelled with multivariable linear regressions. Cluster analysis identified six distinct diet patterns--'dark bread', 'wine', 'fruits', 'low-frequency eaters', 'fries' and 'white bread'. The 'white bread' and the 'fries' patterns over-represented the Hispanic IRAS population predominantly from two centres, while the 'wine' and 'dark bread' groups were dominated by non-Hispanic whites. The dietary patterns were associated significantly with each of the outcomes first at the crude, clinical level (P<0.001). Furthermore, they were significantly associated with FI, BMI and waist circumference independent of age, sex, race or ethnicity, clinic, family history of diabetes, smoking and activity (P<0.004), whereas significance was lost for SI. Studying the total dietary behaviour via a pattern approach allowed us to focus both on the qualitative and quantitative dimensions of diet. The present study identified highly consistent associations of distinct dietary patterns with measures of insulin resistance and adiposity, which are risk factors for diabetes and heart disease.

1 Follower
9 Reads
  • Source
    • "For descriptive and analytical purposes we furthermore collapsed the 114 line items of the FFQ into thirty-three food groups based on similarities in food and nutrient composition, and computed the number of daily intake servings for each of these food groups. A description of the food groups has recently been published elsewhere (Liese et al. 2004b). The IRAS FFQ has been validated in a subset of the IRAS population (Mayer-Davis et al. 1999). "
    [Show abstract] [Hide abstract]
    ABSTRACT: The role of dietary glycaemic index (GI) and glycaemic load (GL) in disease aetiology is of increasing interest. However, nutritional factors related to dietary GI and GL are not well understood from a population perspective. We aimed to investigate the relation ship between GI and GL and dietary intake at the food and nutrient level. Study subjects were 1071 non-diabetic adults from the Insulin Resistance Atherosclerosis Study, Exam I, 1992-4. Usual dietary intake was assessed with a 114-item modified Block food frequency questionnaire. Published GI values were assigned to food line items. Correlation and regression analyses were conducted. Intake of white bread, beer, meats and fries/fried potatoes was positively associated with average GI, as was fat, starch and alcohol intake (before and after energy adjustment). Intake of fruits and low-fat milk was inversely associated with GI, as were intakes of mono- and disaccharides, and fibre. GL was positively correlated with carbohydrate foods and inversely with non-carbohydrate foods. Gender-specific regression models identified eight food groups explaining 63% (men) and 55% (women) total GI variation after adjusting for demographics; 70% of variation in GL was explained by eleven (men) and nine (women) food groups, respectively. Although the GI of a food is an indicator of the ability of carbohydrates to raise blood glucose, dietary GI, unlike GL, appears to reflect more dimensions of diet than just carbohydrates, such as the combination of foods consumed. This may have implications for the interpretation of dietary GI in epidemiologic studies.
    British Journal Of Nutrition 10/2005; 94(3):397-406. DOI:10.1079/BJN20051514 · 3.45 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: L‘obésité constitue un problème de santé publique au Canada, particulièrement chez les populations autochtones où les prévalences les plus élevées ont été rapportées. D’après les écrits recensés, plusieurs méthodes ont été essayées pour étudier la relation entre l’alimentation et l’obésité, mais les résultats sont inconstants. Le but de cette thèse est d’identifier, en termes quantitatif et qualitatif, les différences dans l’alimentation des obèses et non-obèses. Pour y parvenir, nous avons développé une nouvelle méthode à l’aide d’une banque de données portant sur les enfants Mohawk de Kahnawake afin d’identifier les différences dans les choix alimentaires. Cette même méthode a été ensuite appliquée à deux autres banques de données (celle des adultes cris de la Baie James et celle des autochtones de l’enquête ESCC 2.2). Globalement, les résultats n’ont pas montré de différences significatives dans l’alimentation des participants selon les catégories d’IMC en considérant les indicateurs reliés à la quantité et à la qualité de l’alimentation comme l’apport énergétique total, l’apport énergétique en provenance des lipides, les fibres alimentaires, la densité énergétique et la diversité alimentaire. Par contre, les résultats de la nouvelle méthode fondée sur la sélection des items alimentaires fréquemment consommés par au moins 10 % des participants ont révélé que les enfants de Kahnawake à risque d’excès de poids consommaient plus fréquemment de croustilles (p=0.001) et moins fréquemment de craquelins que les enfants avec excès de poids ou ceux ayant un poids normal (p=0.015). Ensuite, en prenant la catégorie de poids normal comme référence, le rapport de côte (Odds ratio : OR) d’être à risque d’excès de poids était de 2.16 (95 % IC : 1.14 - 4.09) fois plus élevé chez les enfants de Kahnawake qui consommaient plus fréquemment de croustilles comparativement aux non-consommateurs de croustilles, et ce, après ajustement pour l’âge. Par contre, le rapport de côte d’être à risque d’excès de poids diminuait de 79 % (OR = 0.21; 95 % IC : 0.06 – 0.72) chez les enfants consommateurs de craquelins comparativement à leurs homologues non-consommateurs. Après avoir corrigé les quantités pour l’âge, on note que les enfants avec excès de poids consommaient plus de frites que les enfants à risque d’excès de poids ou ceux ayant un poids normal (p = 0.027). Chez les femmes cries, les résultats de la nouvelle méthode ont montré que le colorant à café était associé à un risque élevé d’obésité (OR = 4.64, 95 % IC : 1.04 - 0.54); alors que le lait faible en matières grasses était associé à un moindre risque d’embonpoint (OR = 0.38, 95 % IC : 0.17 - 0.82), après ajustement pour l’âge. Quant aux hommes cris, le lait entier était associé à un moindre risque d’avoir de l’embonpoint (OR ajusté pour l’âge = 0.38, 95 % IC : 0.20 - 0.71) et, en termes de quantité corrigée pour l’âge, les hommes obèses buvaient plus de boissons sucrées aux fruits comparativement aux hommes de poids normal ou ceux ayant de l’embonpoint (p=0.015). Selon les résultats de cette méthode appliquée aux données de l’enquête ESCC 2.2, les garçons à risque d’excès de poids ou avec excès de poids consommaient moins fréquemment de pain blanc que ceux de poids normal (p=0.048). En termes de quantité toutefois, ils consommaient plus de pain blanc (p=0.040), utilisaient plus de farine de blé (p=0.006) et de levure (p=0.002). Après avoir ajusté les quantités consommées pour l’âge et l’indice d’activité physique, les femmes avec embonpoint ou obèses utilisaient plus de farine de blé (p< 0.001) que leurs homologues de poids normal. Chez les hommes, il n'y avait pas de différences ni dans les fréquences de consommation ni dans les quantités consommées. Concernant les filles, leurs apports alimentaires n'étaient pas valides (facteur d'activité de Goldberg < 1.2 dans la catégorie embonpoint / obèse). Les résultats de cette méthode innovatrice pourraient d’une part, permettre d’axer la sensibilisation sur des aliments particuliers en plus des recommandations générales du Guide Alimentaire Canadien. D’autre part, ils nous renvoient aux données biologiques de laboratoire afin d’identifier les composantes des items susceptibles de contribuer au développement de l’obésité. Obesity is a public health problem in Canada, particularly among Aboriginal populations where the highest prevalences have been reported. In the literature, several methods have been tried to study the relationship between diet and obesity, but results are inconsistent. The objective of this thesis is to identify differences in selected dimensions of diet quality and quantity across body mass index (BMI) categories for Aboriginal children and adults in Canada. To achieve this goal, we developed a new method by using data from Mohawk children in Kahnawake. The same method was then applied to two other datasets (James Bay Cree adults and off-reserve adults and children from CCHS 2.2). Overall, in each dataset, the results showed no differences in the diets of participants considering indicators such as energy intake, percent fat, fiber intake, energy density and dietary diversity. On the contrary, using the new method based on the most-frequently consumed food items by at least 10% of participants, we found that Mohawk children “at risk of overweight” consume potato chips more frequently (p=0.001) and crackers less frequently (p = 0.015), compared to normal-weight or overweight children. Compared to normal-weight, and after adjusting for age, Mohawk children who consumed more frequently chips increased their probability of being at risk of overweight (Odds ratio : OR = 2.16, 95 % CI : 1.14 - 4.09), while those who consumed crackers more frequently decreased their risk (OR = 0.21; 95 % CI : 0.06 – 0.72). In terms of quantity, and after adjusting for age, overweight children consumed larger portions of french fries (p = 0.027).Among the Cree women (compared to normal-weight, and after adjusting for age), we found out that nondairy creamer was associated to increase risk of being obese (OR = 4.64, 95 % CI : 1.04 - 20.54) while low-fat milk was associated to lower risk of overweight (OR = 0.38, 95 % CI : 0.17 - 0.82). Among Cree men, (after adjusting for age), whole milk was associated to a high risk of the men being overweight (OR = 0.38, 95 % CI : 0.20 - 0.71). Finally, in terms of quantity (after adjusting for age), obese men drank more sweetened fruit drinks compared to men of normal weight or those with overweight (p = 0.015). In the third dataset CCHS2.2, results of the new method show that boys at “risk of overweight” or overweight consumed white bread less frequently (p = 0,048) but in greater quantity when they do so (p = 0,040). They also used more flour (p = 0.006) and yeast (p = 0.002). After adjusting the quantities consumed for age and physical activity, women with overweight or obese used more flour (p < 0.001) than those of normal weight. No differences were found in terms of frequency and quantity for men. The data were not analysed for girls since their Goldberg activity factor was less than 1.2 with greater body weight. The results of this new method could, on one hand, allow us to focus awareness on foods in addition to general recommendations of the Canadian Food Guide. On the other hand, they refer us back to laboratory data to identify components of items that may contribute to the development of obesity.
  • Source
Show more


9 Reads

Angela D Liese