Ethnic differences in diet : A focus on methodology , determinants and Type 2 Diabetes Mellitus

Disclaimer/Complaints regulations If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: http://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.


Introduction
In European countries, the immigrant population is growing enormously.Differences in diet between minority groups and host populations have been observed [1][2][3][4], in which the preservation of traditional dietary habits is seen [5][6][7], but also adoption of the Western diet, high in meat and fat intake, is shown [5][6][7][8].The adoption of the Western diet has been suggested to increase the risk of adverse health effects such as obesity and type 2 diabetes [9], also among ethnic minority groups [10].To fully understand the differences in diet, studying the overall dietary pattern rather than single nutrients or foods has been suggested [11].There is a need to describe these patterns of food consumption to better understand the role of diet in the observed ethnic inequalities in health.This understanding is key in formulating how to stimulate healthy diet, not only in the majority population but also among ethnic minority groups; dietary change may be more readily achieved when recommended foods are compatible with existing patterns of food consumption [12].
Differences in dietary patterns between ethnic groups may partially be a reflection of differences in socioeconomic status.Among non-migrant populations there is strong evidence that dietary patterns differ between socio-economic strata [13][14][15][16][17].
Ethnic minorities are often overrepresented in low socio-economic status (SES) groups, which might imply that differences in their dietary patterns might simply be a reflection of differences in their socio-economic profile.Therefore, in order to understand the origins of ethnic differences in diet, there is a pressing need to address socioeconomic differences with respect to dietary behaviour.
To date, there is little evidence as to whether the well-known socioeconomic gradient in diet also applies to ethnic minority groups.The direction and strength of the association between SES and diet might not be similar across ethnic groups [18].This, because of the complex interaction between diet and ethnicity; diet is governed by deeply rooted cultural norms and values, and has particular significance for ethnic and cultural identity.Some aspects of the diet may be relinquished or adopted more readily than others [19], which may lead to a different association with SES as compared to the host population.It is therefore important to assess the socioeconomic gradient in the adherence to dietary patterns within the population(s) of interest before potential differences are wrongly assigned to ethnic origin.
Therefore, this paper aims to identify and describe the dietary patterns of ethnic minority groups in comparison with the host population and explore the role of socio-economic factors underlying these differences.More specifically, we will study the dietary patterns among the largest ethnic minority groups in the Netherlands as compared to the ethnic Dutch.These minority groups include Surinamese of South Asian and African origin of whom mostly are first generation migrants who have been living relatively long in the Netherlands.Furthermore we aim to investigate to what extent part 2 | determinants of diet socio-economic characteristics of these populations contribute to differences in dietary patterns both between and within ethnic groups.

Study design and subjects
Participants were recruited within the HEalthy LIfe in an Urban Setting (HELIUS) study as part of the HELIUS-Dietary Pattern sub-study.Detailed information about the HELIUS and the HELIUS-Dietary Pattern study can be found elsewhere [20,21].In brief: The HELIUS study was designed as a prospective cohort study, and is being carried out in Amsterdam, the Netherlands.The study has the primary aim to unravel the causes of the unequal burden of disease across ethnic groups.It includes people with an ethnic Dutch background along with individuals with African origin Surinamese, South Asian origin Surinamese, Turkish, Moroccan, and Ghanaian backgrounds.The ethnic minority groups included in the HELIUS study are the largest ethnic minority groups in Amsterdam.
Subjects in the age range of 18 to 70 years are randomly sampled, stratified by ethnic origin, through the municipality registry of Amsterdam.This registry contains data on the country of birth of residents and their parents, which are used to determine ethnic background.
In the Netherlands, country of birth has become widely accepted as a basis for identifying ethnic groups.This classification defines ethnic groups according to an individual's country of birth as well as that of his or her parents.This is the definition used in the HELIUS study.Specifically, a person is defined as of non-Dutch ethnic origin if she or he fulfils one of two criteria: he or she was born outside the Netherlands and has at least one parent who was born outside the Netherlands (first generation); or he or she was born in the Netherlands but both parents were born outside the Netherlands (second generation) [22].Within this paper we will focus on dietary differences between the South Asian Surinamese, African origin Surinamese and ethnic Dutch participants.
These two Surinamese groups are often combined as one homogeneous group in health surveys while health outcomes differ [23][24][25].As a result, risk profiles within these two different ethnic groups are largely blurred.Although African origin Surinamese ('Creole'), hereafter referred to as African Surinamese, share a common ancestry with the Africandescent populations in the West, while South Asian Surinamese ('Hindustani') share a common ancestry with the South Asian populations, these groups differ considerably in terms of diet and socioeconomic status [26).
Baseline data collection of HELIUS was still ongoing.For our study, baseline data collected until June 2013 was used.Data on population characteristics and health status was collected through a questionnaire/interview and a physical examination.Biological samples were obtained during the physical examination.The HELIUS-Dietary Pattern dietary patterns and socio-economic status | chapter 5 study includes a sub-sample of the HELIUS population.Participants of the HELIUS study that gave consent to taking part in additional studies were eligible to participate in the HELIUS-Dietary Patterns study.Of the total HELIUS sample, 90%, 74% and 79% ethnic Dutch, South Asian Surinamese and African Surinamese respectively, participated in the HELIUS-Dietary Patterns study.Among these participants, the final response rate among the ethnic Dutch, South Asian Surinamese and African Surinamese was 79%, 53% and 56% respectively.Compared to responders, non-responders or those participants that did not consented to be included in the HELIUS-Dietary Patterns study were slightly younger, showed the same sex and BMI and morbidity distribution.Smoking status and level of education did differ between the African Surinamese responders and non-responders, but the differences were marginal.The HELIUS study and the HELIUS-Dietary Patterns study were ethically approved by the AMC Medical Ethics Committee.

Food frequency questionnaire
Dietary intake was collected by either self-administered semi-quantitative Food Frequency questionnaire (FFQ).An existing validated Dutch FFQ [27] formed the basis for a new comparable FFQ, specifically assessing habitual dietary intake in the Surinamese population.Detailed information about the FFQ development can be found elsewhere [28].In brief, the underlying aim in the development of this Surinamese FFQs was to conduct research into ethnic differences in the association between diet and health.Data from single 24 hour recalls, collected within a Surinamese (both of South Asian and African origin) population, formed the basis for the adaptation of this FFQ, using a standardised approach [29].Food items were selected according to their percentage contribution to, and variance in nutrient intake in the Surinamese population.A nutrient database for this FFQ was constructed consisting of data of the Dutch Food Composition table 2011 [30].Data on ethnic specific foods was based on new chemical analyses, and available international data.The FFQ included ~220 food items covering more than 90% of the intake of the main nutrients of interest.Habitual dietary intake in the previous four weeks was assessed using newly developed Surinamese FFQ.
A simple approach would be to apply a single FFQ in the two ethnic groups.
However, this option would either result in a very long questionnaire or in a compromise of its face validity.Therefore, we chose to develop a separate FFQ but strived to make both FFQs as similar as possible by applying the same rigorous methodology in their development [29].Compared to the Dutch FFQ, the resulting Surinamese FFQ has the same lay-out and consist of similar, comparable food items.However, small differences are inevitable because of differences in the diet of the groups studied.

Assessment of dietary patterns
Dietary patterns allow the assessment of the whole diet, accounting for the fact that foods/nutrients are consumed in combination and are therefore highly correlated.
Dietary patterns were derived on the basis of Principal components analysis (PCA), which assesses the correlations between food groups to identify underlying patterns in the data.Intakes of food groups were obtained by collapsing food items assessed in the FFQ on the basis of similarity in nutrient profile, culinary use, or ethnic origin [supplemental table 1].As ethnic specific food groups are an important part of the corresponding ethnic group's diet we included these foods in our analysis.However, the differences for some food items are due to the fact they were not assessed in one of the ethnic groups and that we had to set the intake to zero when these items were not assessed.Additionally, some of the ethnic specific food groups were not combined within a broader food group category to prevent loss of possibly relevant details.For example, roti (Indian flat bread), which was an ethnic specific food group only measured in the Surinamese FFQ, was not combined with the low fibre bread food group (measured in the Dutch and Surinamese FFQs) because of a distinctive culinary use.This resulted in a total of 49 food groups: 35 food groups were assessed in a similar way, 11 food groups were assessed in such a way that besides similar food items, also ethnic specific food items were taken into account, 1 food groups was only assessed in the ethnic Dutch FFQ (pancakes), and 2 food groups were only assessed in the Surinamese FFQ (roti and pom).
In order to describe differences in dietary patterns between ethnic groups we performed principal component analysis for the whole sample (i.e.ethnic Dutch, South Asian Surinamese, and African Surinamese together).The number of components retained was based on the following criteria: components with an eigenvalue >1, Scree plot test, and the interpretability of the components.Food items were considered to load on a component if they had a correlation/factor loading ≥ 0.3.A larger factor loading indicates a higher correlation of the food group to the respective component.We report the percentage of variance of the food group intake explained by each pattern.This aspect, however, did not play a role in the selection of components as it depends highly on the number of variables included in the analysis.The Scree plot test clearly identified 3 major components.The components (hereafter called "dietary patterns") were labelled on the basis of those food groups that loaded highest in the respective dietary pattern.
The dietary patterns were derived on the basis of the unadjusted consumption (gr/d) of specific food groups.A factor score was calculated by summing the standardized intake of foods, weighted by the factor loadings of the foods groups for each dietary pattern; each person receives a factor score for each dietary pattern that emerged from the data.These scores rank individuals according to the degree to which they conformed or adhered to each of the derived dietary patterns.We used partial correlation coefficients, adjusted for energy intake, for correlations between nutrient intake and dietary pattern dietary patterns and socio-economic status | chapter 5 scores in order to get insight into the macronutrient composition of each of the dietary patterns.
In sensitivity analysis we assessed whether within sex and ethnic groups comparable dietary patterns were observed in to order understand the consequences of performing a pooled analysis.Furthermore, in order to examine whether the unequal distribution of ethnicities in the pooled analysis (n=1254, 425, 784 for the ethnic Dutch, South Asian Surinamese and African origin Surinamese respectively) had an effect on the derived dietary patterns, we also ran our analysis with a random sample of n=200 per ethnic group.Although small differences were found, we derived similar dietary patterns and comparable loadings as when the dietary patterns were obtained for the whole population.Therefore, we based further analysis on the patterns derived for the whole sample.

Assessment of education level and occupational status
Educational and occupational levels were used as two different proxies for SES.
Educational level was indicated by the highest education attained.Initially, four categories were used: "never been to school or elementary schooling only", "lower vocational schooling or lower secondary schooling", "intermediate vocational schooling or intermediate/higher secondary schooling (general)" and "higher vocational schooling or university".As the number of participants were small in some categories for some of the ethnic groups, we decided to aggregate the lower two educational categories in order to prevent unequal distributions across different levels of education across ethnic groups.Consequently, education was defined in the models as low, medium or high.The five different occupational levels measuring current occupational status based on the standard occupation classification 2010 [29], were also reclassified into three classes due to limited numbers within some categories.The "lowest" class represented occupations characterized by "manual labour" (skilled and unskilled manual), followed by the "middle class" characterized by "lower grade professionals and routine non-manual labour", and the "highest" occupational level, characterized by "higher grade professionals".

Other variables
Smoking was assessed as current smoker, former smoker or never smoked.Alcohol intake was assessed as currently using alcohol or not.Marital status was measured in 5 categories: "married/registered partnership", "cohabiting (living together)", "unmarried (never married)", "divorced or separated", "widow/widower".However, these categories were collapsed into 2 categories, either living together with a partner, or not (never married, divorced or separated or widow/widower).Participants were classified as having diabetes when at least one of the following variables was positive: increased fasting glucose (≥ 7 mmol/l) or use of glucose lowering medication.Hypertension was part 2 | determinants of diet based WHO-criteria (SBP ≥ 140 mmHg and DBP ≥ 90 or use of blood pressure lowering medication).Hypercholesterolemia was defined as high serum level of total cholesterol ≥ 6.2 mmol/l (240 mg/dl) or using lipid lower medication.The presence of disease "morbidity" variable included participants, which were coded as having at least one of either diabetes, hypertension and hypercholesterolemia.

Statistical analysis
Baseline characteristics were expressed as percentages (%), or means with standard deviations (SD).Linear regression analysis was used to examine the associations between ethnicity and dietary pattern scores, where one unit change of each score corresponded to 1 SD of the study population.Distribution of continuous variables was examined for normality and log transformed when necessary before entering them into the regression models.Due to a significant interaction between ethnicity and sex in the association with dietary patterns, we modelled our analysis separately for men and women.In addition to the age adjusted model (model 1), we adjusted for the following confounding variables: marital status, morbidity, smoking status, physical activity, BMI (model 2).We also tested for seasonal differences in dietary pattern scores by adjusting for the season in which the participant filled in the FFQ.To understand the role of SES within the different ethnic groups (i.e. the SES gradient in dietary patterns), we studied the association between socio-economic indicators (i.e.highest completed educational level and occupational status) and dietary pattern scores within ethnic groups.Additionally, we tested for interaction by SES in the association between ethnicity and dietary pattern scores.All analyses were performed with SPSS version 20 (Illinois, USA).

Baseline characteristics
Table 1 presents selected characteristics of the study population by ethnicity.The average age of the participants was 49 years and there were considerably more women than men in all ethnic groups; the percentage of current smokers was similar across ethnic groups (~23%).Compared to the other groups African Surinamese were more often unmarried, divorced or widowed (67.2%).More ethnic Dutch reported using alcohol (92.1%), had the lowest BMI (mean: 24.8 kg/m2) and had highest proportion of highly educated participants and participants working at the highest occupational level (60.1 and 57.8%, respectively).More Surinamese participants had either diabetes, hypertension or elevated cholesterol, with diabetes being most prevalent among South Asian Surinamese (19.1%) and hypertension most prevalent among African Surinamese (56.4%) (data not shown).dietary patterns and socio-economic status | chapter 5

Dietary patterns
Within ethnic groups we extracted comparable dietary patterns (in all 3 groups we derived a clear meat, a snack and a vegetable pattern), suggesting that the data could be pooled in order to describe differences in dietary patterns between the ethnic groups.For this pooled analysis.the factor loadings ≥0.30 of food groups for the three identified dietary patterns are shown in Table 2. Positive factor loadings indicate that the subsequent food group is highly correlated with the respective dietary pattern.The "noodle/rice dishes and white meat" pattern was characterized by high intakes of rice and noodles dishes, chicken, organ meat, fish, savoury bread filling, savoury sauces, sugar sweetened beverages, low fibre bread and bread products, and Surinamese dishes like pom (Surinamese festive dish) and roti (Indian flat bread).Most of these foods typically characterize the foods consumed in a traditional Surinamese diet.The second pattern, labelled as the "red meat, snacks, and sweets" pattern which was characterized by higher intakes of red meat and processed meat, pasta, snacks, sugar and sweets, French fries, beer, cheese, fat and oil (not olive oil) and full fat margarine, savoury sauces, cakes and cookies, potatoes and other root vegetables, pancakes and high fibre bread and bread products.Food groups in the third pattern, the so called "vegetables, fruit and nuts pattern", had high factor loadings on meat substitutes and other soy products, nuts and seeds, tomato and tomato products, brassica vegetables, other vegetables, legumes, olive oil, fruit and low fat fish.Each of these patterns explained approximately 6% of the total variation in food group intake data.We did not find any seasonal differences in dietary pattern scores across ethnic groups.
Higher scores on the noodle/rice dishes and white meat pattern and on the red meat, snacks, and sweets pattern were significantly associated with higher intakes of total energy in all 3 ethnic groups (Table 3).The red meat, snacks, and sweets pattern was negatively associated with intakes of non-haem iron, vitamin C and, particularly among ethnic Dutch; there was a strong negative correlation with dietary fibre (-0.46).Higher scores on the vegetables, fruit and nuts pattern were associated with significantly lower intakes of carbohydrates, particularly among the Surinamese groups.Strong positive associations were observed between the vegetables, fruit and nuts pattern scores and dietary fibre, iron, beta carotene and vitamin C intakes.

Ethnic differences in dietary pattern scores
Table 4 shows ethnic differences in dietary pattern scores for each of the dietary patterns.In sensitivity analysis we extracted similar patterns for men and women, however we did find significant interaction by sex with regard to ethnic differences in pattern scores (p interaction ≤ 0.001).Therefore, the results are displayed separately for men and women.Compared to the ethnic Dutch population, in the fully adjusted model, Surinamese had significantly higher scores on the noodle/rice dishes and white meat (β

Discussion
Three major dietary patterns were identified, labelled as a "noodle/rice dishes and white meat", "red meat, snacks, and sweets" and "vegetables, fruit and nuts" pattern.Surinamese clearly showed greater adherence to the noodle/rice dishes and white meat, while the ethnic Dutch scored significantly higher on the other two dietary patterns.Overall these ethnic differences in pattern scores were robust within different socioeconomic groups, with the greatest ethnic differences in dietary pattern scores among men.Among Surinamese, this gradient was only observed with regard to adherence to the vegetables, fruit and nuts pattern and differed between socio-economic indicators and sex.
Only a few studies have investigated the dietary patterns within different ethnic groups living in one setting.These studies found that some patterns seem to be clearly shared between different ethnic groups (often a "Western" pattern (high in snacks and meat) and a "healthy" pattern (high in vegetables, fruits and fish), whereas other patterns are ethnic specific [1,[31][32][33].In our sample we found similar patterns, including an "ethnic specific" pattern (the noodle/rice dishes and white meat pattern), that more closely described the habitual dietary intake observed among both Surinamese groups compared to the ethnic Dutch.This study adds additional insight by considering the association between dietary patterns with SES.We found that adherence to the ethnicspecific pattern was quite robust across different occupational and educational groups.
The noodle/rice dishes and white meat pattern was characterized by foods that were typically consumed in traditional Surinamese meals.Surinamese seemed therefore to preserve their traditional eating patterns regardless of SES but did seem to incorporate more vegetables and fruits and other foods that characterize the vegetables, fruit and nuts pattern to their diet with higher SES.This implies a selective adoption, presumably a move towards more vegetables with higher SES but still fidelity to the traditional Surinamese dietary pattern.Interesting in this context, Sharma et al [34] reported that African Caribbean adults in Britain, despite their low incomes, spent more on traditional foods like yams than on potatoes, thereby maintaining cultural food preferences.This suggests the importance that is given to these traditional dietary patterns.
Both the noodle/rice dishes and white meat pattern and the red meat, snacks, and sweets pattern are characterized by high intakes of presumably "less healthy" food groups (i.e.red meat, snacks and French fries, sugar sweetened beverages and savoury sauces).This is further underscored by the highly positive correlations between these patterns and total energy intake, saturated fatty acids, and the highly negative correlations with fibre, β carotene and vitamin C. One would expect to see a decrease in adherence to these patterns with increasing SES in all ethnic groups.However, this gradient was only observed in the ethnic Dutch.Among ethnic minority groups the expected change in dietary behaviour with increasing SES, might be complicated by the value that is given to traditional foods, as described by the model of Koctürk-Runefors [35].Foods that are strongly associated with cultural identity, values and norms (i.e.
staple foods as rice and bread), may be the last to change.Whereas accessory foods (i.e. vegetables, meat and chicken) or "extras" (i.e.fruits, sweets and nuts) are less valued and change in their use seems more often related to availability or the economic situation.
In this respect, foods characterizing the vegetable, fruits and nuts pattern are clearly less associated with cultural identity, and seems therefore easier to adopt from patterns that exist in the host populations than those foods that characterize the noodle/rice dishes and white meat pattern.
To our knowledge there is only one other study that examined the extent to which socioeconomic factors are related to ethnic differences in dietary patterns [4].
According to Sommer et al. socio-economic status explained a large proportion of ethnic differences in dietary habits among pregnant women, although the fully adjusted model also pointed to significant cultural differences in dietary preferences.However, no test for effect modification by SES was performed.This insight is important as this could help us understand which subgroups would benefit most from targeted (cultural sensitive) interventions for improving dietary habits.
Within this population of mainly first generation (85.4% and 89.0% in the South Asian and African Surinamese groups, respectively) Surinamese migrants, education has been primarily completed in Suriname and might therefore not be a good proxy of current SES.Educational status has different meanings over the life course and is likely to hold different significance in different countries and cultures [18].Therefore, level of education might not be an optimal indicator of current SES in the context of dietary behaviour.On the other hand, occupational status might also act differently among ethnic minority groups [18].The rates of unemployment are higher in ethnic minority groups than in the ethnic Dutch population and, adjusted for general characteristics as educational status or work experience, ethnic minorities do not have comparable chances on the labour market as those of the ethnic Dutch [36).Additionally, income may vary by ethnic group within the same occupational class, so that occupation may not have equivalent meanings across groups [37].Unfortunately, income has not been measured in the HELIUS study; therefore we could not take into account this proxy of socio-economic status.More research is needed on useful SES indicators in ethnic minority groups [18,38].It has been recommended that researchers systematically explore the effect of their choice of SES indicator to demonstrate their cross-ethnic group validity as potential confounding variables for the specific groups and outcomes of interest [18].
Some methodological considerations should be addressed.The use of principal component analysis requires several arbitrary decisions about the selection of included variables (FFQ item definition/collapsing foods into food groups), the number of retained factors, the method of rotation, and the labels of the factors.We conducted sensitivity analyses to examine whether we observed different dietary patterns within men and dietary patterns and socio-economic status | chapter 5 women, and investigated whether the higher proportion of ethnic Dutch participants influenced the extracted patterns.However, the extracted patterns seemed to be robust and therefore applied in further analysis.The labelling of the identified patterns was subjective; this can be judged by the reader from the presented factor loadings (Table 2).Our food groupings were based on our aim of exploring ethnic differences in dietary patterns and followed general based main analytic decisions on previous scientific knowledge.In addition, there are inherent problems in dietary assessment, such as self-report bias.We used two different FFQs to measure the dietary intake within the ethnic Dutch and Surinamese populations.Therefore, differences in pattern scores might be due to differences in the FFQs.Nevertheless, the Surinamese FFQ, which was adapted from a validated Dutch FFQ [27] were developed with the aim of conducting combined analyses with the Dutch population and therefore has the same lay-out, consist of similar, comparable food items and has been developed with the same standardized approach with rigorous validated methodology [28,29].The extensive food list of the FFQs used included group-specific marker foods that may be key to elucidating differences between the dietary patterns of the ethnic groups included in this study.Important to note in this context is that the socio-economic distribution of the populations in which the 24h recalls were collected as data input for the development of the FFQs is comparable to the SES distribution within the HELIUS-dietary pattern population, implying that the FFQs are representative for the present population.Because the response rate of both Surinamese groups was lower than that of the ethnic Dutch population, we compared the Surinamese responders with those who did not complete an FFQ.No major differences were observed, except for age and education.Those who participated were slightly older and higher educated.However, this was also the case for the ethnic Dutch.
The robust ethnic differences in dietary patterns indicate that besides shared characteristics in food group intake it is also important to account for ethnic differences in the details of dietary behaviour when developing new strategies to promote healthy diets.The absence of a clear socio-economic gradient in the noodle/rice dishes and white meat pattern among both Surinamese groups suggest the importance that is given to the foods characterizing this pattern and underscore the finding of a selective change in dietary behaviour among ethnic minority groups.Thus the promotion of healthy diets should be based on existing (ethnic specific) dietary patterns, respecting the value assigned to these patterns.However, as in the host population, promotion of fruit and vegetable intake is particularly relevant for low socio-economic groups.

Table 1 .
Population characteristics of the HELIUS-Dietary Patterns study

table 1 .
*There are 177 missing cases with respect to occupational status.These missing cases are not significantly different between the ethnic groups.Number included per season for the ethnic Dutch, African Surinamese and South Asian Surinamese were respectively: winter: 735,

Table 2 .
Factor loadings of dietary patterns among Ethnic Dutch, South Asian Surinamese and African Surinamese participating in the HELIUS-Dietary Patterns study Footnote table 2. Only shown factor loadings > 0.30 in at least one of the ethnic groups dietary patterns and socio-economic status | chapter 5

Table 3 .
Partial Pearson correlation coefficients (r) between each of 3 food patterns derived in the HELIUS-Dietary Patterns study and daily energy and nutrient intakes Noodle/

table 3 .
Except for total energy intake, nutrients are adjusted for energy intake; All nutrients are log transformed to improve normality.All correlations are statistically significant, except for those correlations marked with #; bold=partial pearson correlations > 0.20; SFA=Saturated fatty acids, MUFA=Mono unsaturated fatty acids, PUFA=Poly unsaturated fatty acids dietary patterns and socio-economic status | chapter 5

Table 4 .
Ethnic differences in dietary pattern scores derived in the HELIUS-Dietary Patterns study * Footnote table 4: * = P < 0.001 | Model 1: adjust for age | Model 2: model 1 + marital status, morbidity, smoking status, physical activity and BMI.dietary patterns and socio-economic status | chapter 5

Table 5 .
The association between ethnicity and dietary pattern scores by socioeconomic status in the HELIUS- *Footnote table 5. Sex-specific linear regression models adjusted for age and BMI stratified by education or occupation.* = P < 0.001.*** There are 10 missing cases on educational level.There are 177 missing cases with respect to occupational status.These missing cases are not significantly different between the ethnic groups

Table 6 .
The association between socioeconomic status and dietary patterns within sex specific ethnic groups in the HELIUS-Dietary Patterns study

table 6 .
Sex-specific linear regression models adjusted for age and BMI stratified by education or occupation.= P < 0.001.** = P < 0.001.*** There are 10 missing cases on educational level.There are 177 missing cases with respect to occupational status.These missing cases are not significantly different between the ethnic groups dietary patterns and socio-economic status | chapter 5 *