Sammendrag
Background Dietary exposure is complex since foods are eaten in combinations and the foods contain a combination of several nutrients. It is likely that there is an interaction and synergy between these foods and nutrients. The cumulative effect of various foods may be detectable, while the effect of a single food might be undetectable. The complexity of the individual dietary intake can be explored by dietary pattern analyses and this approach might be especially useful if many dietary components are relevant for a disease. Such insight can provide information for setting priorities for changing dietary patterns in a population by public health initiatives. Dietary patterns are meant to reflect dietary behaviour in a population and are based on information gathered using various respondent-based dietary assessment instruments that attempt to measure habitual dietary intake. These instruments are associated with measurement errors that may influence the validity of the dietary patterns. Different methods such as nutritional biomarkers and identification of misreporters of dietary intake may be applied to evaluate the dietary patterns.
Aims The current thesis aimed to (a) identify dietary patterns among Norwegian women aged 50-69 years, (b) evaluate the dietary patterns by using plasma carotenoids as biomarkers of fruits and vegetables, (c) perform sensitivity analyses by excluding the low energy reporters from the study sample to investigate whether our dietary patterns were affected by under-reporting of energy intake, (d) investigate how the under-reporting of energy intake affected the associations between dietary patterns and self-reported chronic diseases and (e) examine the associations between dietary patterns and nutrient intake, sociodemographic factors and key risk factors for NCDs.
Methods The study population was 6974 women aged 50-69 years participating in the Norwegian Breast Cancer Screening Program administered by the Cancer Registry of Norway. Dietary intake was assessed by a 253-item food frequency questionnaire. A total of 3263 women provided blood samples and plasma carotenoids were analysed in a subset of these women (n=361). We identified low energy reporters by using the revised Goldberg cutoff method. The 253 food items were categorised into 46 (Paper I) and 49 (Paper II and III) food groups and dietary patterns were identified using principal component analysis. The association between loge-transformed plasma carotenoids and dietary pattern scores were estimated by partial correlation coefficients and multiple regression analyses (Paper I). A logistic regression model was used to estimate the association between dietary pattern scores and the prevalence of self-reported noncommunicable chronic diseases among the total study population and the study population were low energy reporters were excluded (Paper II). We estimated the correlation between nutrients and dietary pattern scores by Pearson’s correlation coefficients and multiple regression analyses were used to examine the associations between risk factors for noncommunicable chronic diseases and dietary pattern scores (Paper III).
Results We identified three dietary patterns in the total study population; the healthy “Prudent” pattern and the less healthy “Western” and “Continental” patterns. In the smaller subset of 361 women, four dietary patterns were identified; the “Vegetarian”, “Western”, “Continental” and “High-protein” pattern. An increasing score for the “Vegetarian” pattern was associated with an increasing concentration of plasma carotenoids whereas an increasing score for the “Western” and “Continental” patterns were associated with a decreasing concentration. A total of 18% of the women were identified as low energy reporters, and their presence in the study sample did not appreciably affect the composition of food groups that loaded highly (=0.30) on the dietary patterns. However, we observed an attenuation of the associations between dietary pattern scores and several of the self-reported chronic diseases when low energy reporters were included in the study sample, especially among the overweight/obese women. Furthermore, women with high adherence to the “Prudent” pattern were older, more highly educated and had a generally healthy lifestyle. Women with high adherence to the “Western” pattern were older, had lower education and, except for having a low alcohol intake, had a generally unhealthy lifestyle. Finally, women with high adherence to the “Continental” pattern were younger, more highly educated, and had a generally unhealthy lifestyle.
Conclusions We found important dietary patterns among Norwegian women aged 50-69 years. The evaluation of the dietary patterns by plasma carotenoids showed that the “Vegetarian”, “Western” and “Continental” patterns were meaningful patterns. The underreporting of energy intake did not affect the composition of high loaded food groups in the patterns, but the under-reporting attenuated the associations between dietary patterns and selfreported chronic diseases, especially among overweight/obese women. Women with a high adherence to the “Prudent” pattern had a generally healthier lifestyle than women with high adherence to the “Western” and “Continental” dietary pattern.