Pediatric Allergy and Immunology
ORIGINAL ARTICLE
Asthma
Cross-sectional association of dietary patterns with asthma and atopic sensitization in childhood – in a cohort study Sumaiya Patel1, Adnan Custovic2, Jaclyn A. Smith2, Angela Simpson2, Gina Kerry2 & Clare S. Murray2 1
School of Social and Community Medicine, University of Bristol, Bristol, UK; 2Centre of Respiratory and Allergy, Institute Of Inflammation and Repair, Manchester Academic Health Science Centre, The University of Manchester, University Hospital of South Manchester NHS Foundation Trust, Manchester, UK
To cite this article: Patel S, Custovic A, Smith JA, Simpson A, Kerry G, Murray CS. Cross-sectional association of dietary patterns with asthma and atopic sensitization in childhood – in a cohort study. Pediatr Allergy Immunol 2014: 25: 565–571.
Keywords dietary patterns; children; asthma; atopy Correspondence Clare Murray, University of Manchester, Education and Research Building, University Hospital of South Manchester, 2nd Floor, Manchester M23 9LT, UK Tel.: +44 161 291 5876 Fax: +44 161 291 5730 E-mail: [email protected] Accepted for publication 6 July 2014 DOI:10.1111/pai.12276
Abstract Background: Many studies have investigated individual nutrients or foods as risk factors for allergic disease, but few have studied dietary patterns. We aimed to use principal component analysis (PCA) to determine dietary patterns in school age children and examine associations between these dietary patterns and wheeze, asthma and sensitization. Methods: Participants in a population-based birth cohort attended review clinics at ages 8 and 11 yr. A validated questionnaire was interviewer-administered to collect information on parentally reported symptoms and doctor-diagnosed asthma. Atopic sensitization was ascertained by skin-prick tests. Current asthma was defined as doctor-diagnosed asthma and wheezing in the previous 12 months. A validated semiquantitative food frequency questionnaire was completed at age 8 yr, and PCA was used to determine dietary patterns. Results: Principal component analysis identified three dietary components, which based on their characteristics we termed as Traditional (mixed meat, fish, fruit and vegetables), Western (predominantly high fat content, processed foods) and Other (predominantly grains and nuts) dietary patterns. High adherence to the Western diet pattern was significantly associated with doctor-diagnosed asthma and current asthma at age 8 yr [aOR (95% CI): 2.19 (1.20–4.01), p = 0.01; 2.59 (1.15–5.81), p = 0.02; respectively]. A similar association was found for current asthma at age 11 yr [aOR (95% CI): 2.20 (1.07–4.51), p = 0.03]. There was no evidence of an association between dietary patterns and current wheeze and allergic sensitization at either age 8 or 11 yr. Conclusion: School age children adhering strongly to a Western diet, high in fat and processed foods, had a higher risk of current asthma and doctor-diagnosed asthma.
As the hypothesis by Seaton (1) and Black and Sharpe (2), there has been considerable interest in the relationship between diet and asthma. A number of nutrients (3) and foods (4, 5) have been associated with a reduced asthma risk in children. However, dietary intervention studies (predominantly using individual nutrients), which were based on such associations, have failed to demonstrate a benefit (6, 7). Neither nutrients nor foods are consumed individually; they are eaten in combination, and the effects of the combination of nutrients/foods may be synergistic. The observed effect of diet on allergic conditions may be due to individual nutrients or specific foods, or alternatively a dietary pattern that is
associated with high or low level of certain nutrients. This may be a reason why observational studies report an association between certain nutrients and clinical outcomes, whilst intervention studies supplementing individual nutrients have failed to show an effect. A number of studies have used a priori assumptions to evaluate dietary patterns [e.g. Mediterranean dietary score (8)]. This method of assessment can be useful as it views the diet as a whole opposed to individual foods/nutrients. In recent years, there has been interest in data-driven statistical approaches such as principal component analysis (PCA) to study the overall diet in terms of dietary patterns, where the diet
Pediatric Allergy and Immunology 25 (2014) 565–571 ª 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
565
Dietary patterns and asthma
Patel et al.
is assessed in terms of eating behaviours (9). PCA is a datadriven method, which assesses the unmeasured constructs or components within the data. A few studies have used PCA to derive dietary patterns of mothers during pregnancy and evaluate its association with wheeze in children (10, 11). More recently, two studies have used PCA to examine dietary patterns and respiratory symptoms in pre-school (12) and in school age (13) children. In this study, we aimed to assess the association between dietary patterns in UK children assessed at age 8 yr with wheeze, asthma and atopic sensitization. We examined outcomes measured at the time of dietary pattern assessment and prospective associations with outcomes measured at age 11 yr.
Methods Study participants The Manchester Asthma and Allergy Study is a populationbased birth cohort and is described in detail elsewhere (14, 15). Briefly, the subjects were recruited in utero (weeks 8–10 of pregnancy) by screening parents in the antenatal clinic, using skin testing and a questionnaire regarding allergic diseases. All infants born at term and with no congenital cardiac or respiratory disease were eligible and were followed prospectively until age 11 yr. The study was approved by the Local Research Ethics Committee and is registered as ICRCTN72673620. Written informed consent was obtained from all parents. Clinical follow-up Participants attended review clinics at 8 and 11 yr. A validated questionnaire was interviewer-administered to collect information on parentally reported symptoms (16). Atopic sensitization was ascertained by skin-prick tests (Dermatophagoides pteronyssinus, cat, dog, grasses, moulds, milk and egg; Bayer, Elkahrt, IN, USA). At age 8 yr, parents and children completed together a validated semi-quantitative food frequency questionnaire (QBuilder, Tinuviel Software, Anglesey, UK), regarding the child’s usual dietary intake over the previous 2–3 months (see online repository). Children were weighed, without shoes or outer clothing (to nearest 0.1 kg), and measured (to nearest 1 cm) using a stadiometer. Body mass index (BMI) was calculated in kg/m2. BMI standard deviation scores (z scores) were calculated for age- and gender-specific values, using UK standards (17). Data on parental smoking, family history and socio-economic status (18) were assessed by questionnaire.
Current asthma. Defined as doctor-diagnosed asthma and current wheeze. Allergic sensitization. Defined as wheal diameter of 3 mm or greater than the negative control to at least one allergen. Statistical methods Statistical analysis was carried out using SPSS 20 (Chicago, IL, USA). Principal component analysis was used to determine a number of components, which can be considered as dietary patterns. The FFQ included 114 food and drink questions. Items in the FFQ mainly used a nine level scale. Questions, which did not use this scale (e.g. quantity of milk, cheese etc.), were coded to represent the same nine level scale. Questions with a dichotomous reply (n = 1), questions regarding types of foods (n = 2) and questions regarding specific portion sizes (e.g. size of bread slice) were excluded (n = 14). Questions regarding the frequency of consumption of these items were included (Table S1, S2). This left a total of 97 items to be included in the PCA. The scree plot showed a clear break in the curve after the third component revealing the presence of three dietary patterns. This solution also provided the most ‘understandable’ components. The three components explained 15.3% of the variance within the data. (further details in online repository). A pattern matrix was produced for the included items. Factor loadings for individual items, greater than 0.3, were retained. The individual factor loading scores represent the correlation coefficients between the food and the specific dietary pattern (Table S3). The regression method of factor score calculation was used to produce a composite score for each component for every case included in the analysis. Each component was then identified by a simple label which best described the items within it. For ease of interpretability, the population was categorized into tertiles according to their score for the dietary pattern as follows: low, moderate and high adherence to each dietary pattern (using ‘low’ score as reference category). The relationship between dietary patterns and the outcome variables was tested using logistic regression. Confounders to be included in the models were decided upon a priori. We considered the following as potential confounders: gender, parental asthma, parental smoking, allergic sensitization, BMI z-score and socio-economic status. The multivariate model was also adjusted for energy intake as this is standard practice in nutritional epidemiology. Results
Definition of clinical outcomes (age 8 and 11 yr) Current Wheeze. Defined as a positive response to the question ‘Has your child had wheezing or whistling in the chest in the last 12 months?’ Doctor-diagnosed asthma. Defined as a positive response to the question ‘Has a doctor ever given your child a diagnosis of asthma?’
566
A total of 1,211 parents were recruited during pregnancy. Of these, 1,184 term deliveries continued in the study after birth; 1,051 children were followed in the observational cohort (133 children were prenatally randomized to an environmental intervention (19) and excluded from this analysis). At age 8 yr, 899 participants completed FFQs (included in the PCA), of whom 753 completed respiratory questionnaires at both
Pediatric Allergy and Immunology 25 (2014) 565–571 ª 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
Patel et al.
Dietary patterns and asthma
age 8 and 11 yr (included in the logistic regression analysis). Children included in this analysis did not differ from those who were not included in terms of gender, gestational age, birthweight, atopic status, parental sensitization, parental asthma, social class and mean energy intake. Children included were less likely to have a parent who smoked at age 8 (28% vs. 36%, p = 0.04), but not at age 11 yr (p = 0.39). Demographics of the included 753 participants are shown in Table 1. Dietary patterns The factor loadings of the food groups for the three dietary patterns (components) are shown in Table S3. Dietary Pattern 1 included mostly fruit and vegetables with meat and oily fish loading lower down; therefore this pattern was termed the ‘Traditional’ dietary pattern. Dietary Pattern 2 contained processed foods, which are associated with a modern Western diet (chips, crisps, pizza); thus, this pattern was termed ‘Western’ dietary pattern. Dietary Pattern 3 contained items that are often eaten by people following a vegetarian diet (lentils, soya, rice, nuts) but also contained fried foods, offal and pastry dishes; this pattern was termed ‘Other’ dietary pattern. Girls showed a higher mean score on the Traditional dietary pattern compared to boys; boys had higher scores for the Western dietary pattern (Table S4). Higher scores for Traditional dietary pattern were observed for higher socio-economic groups, and the opposite was seen for Western dietary pattern. The Traditional dietary pattern was positively associated with BMI (p = 0.006). The Western diet was positively associated with parental smoking (p < 0.001) and maternal asthma (p = 0.05). Relationship between dietary patterns and clinical outcomes Age 8 yr Table 2 shows the associations between dietary patterns and clinical outcomes at age 8 yr. Participants who adhered
Table 1 Demographics of population (n = 753) at age 8 and 11 yr
Gender Age at Follow-up visit* BMI z-score* Current wheeze Doctor diagnosis of asthma by age 8/11 Current asthma Either parent smoking Positive skin prick test Energy intake (KJ/day)* *Mean (SD).
strongly to the Western dietary pattern had greater odds of current wheeze, but this association did not remain significant in the multivariate analysis. Participants who adhered strongly to the Western dietary pattern had significantly greater odds of doctor-diagnosed asthma and current asthma at age 8 yr, and this association remained significant after adjusting for potential confounders. There was no evidence of an association between adherence to either the Traditional or Other dietary patterns to any of the outcomes (Table 2). There were no associations between the dietary patterns and atopic sensitization at age 8 yr (Table 2). Age 11 yr Table 3 shows the prospective associations between the dietary patterns at age 8 yr and clinical outcomes at age 11 yr. After adjusting for confounders, participants adhering strongly to the Western dietary pattern at age 8 yr were at increased risk of current asthma at age 11 yr. There were no associations between the dietary patterns and atopic sensitization at age 11 yr (Table 3).
Discussion Principal findings In this analysis, we have used a data reduction method, PCA, to determine dietary patterns. The dietary components extracted were labelled Traditional, Western and Other dietary patterns. In the univariate analysis, strong adherence to the Western dietary pattern was associated with greater odds of doctor-diagnosed asthma and current asthma at age 8 yr. After adjusting for potential confounders, the odds of current asthma in those participants adhering highly to the Western diet was approximately two and a half times that of those with a low adherence. Participants who strongly adhered to the Western dietary pattern at age 8 remained at significantly greater odds of current asthma at age 11 yr. No evidence of associations was seen with the Traditional and Other dietary patterns and respiratory symptoms, or with any of the three dietary patterns with allergic sensitization.
Age 8
Age 11
Interpretation
416 7.98 0.44 121 176
416 11.5 0.47 132 198
The dietary components we identified have face validity (i.e. intuitively correct) and construct validity (i.e. observed significant relationships between dietary components and sociodemographic factors). In addition, these components are comparable to those reported by other studies (20, 21). We have demonstrated significant relationships of the dietary patterns with gender, socio-economic status, maternal asthma and smoking. In general, the Traditional and Other dietary patterns (‘healthier’ dietary patterns) were associated with higher socio-economic status, and the opposite relationship was seen for the Western dietary pattern. Similar relationships have been found by other studies looking at dietary patterns in British school children (20).
(55.2%) male (0.17) (1.00) (16.1%) (23.4%)
87 (11.6%) 207 (27.8%) n = 745 216 (31.3%) n = 690 7829 (1547)
(55.2%) male (0.54) (1.12) (17.5%) (26.3%)
95 (12.6%) 174 (23.7%) n = 733 218 (34.1%) n = 640 –
Pediatric Allergy and Immunology 25 (2014) 565–571 ª 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
567
Dietary patterns and asthma
Patel et al.
Table 2 Univariate and multivariate associations between dietary patterns and clinical outcomes at age 8 yr. [multivariate model adjusted for gender, parental asthma, parental smoking, BMI z-score, socio-economic status, energy intake and allergic sensitization (where appropriate)]
Dietary pattern
Adherence
Current wheeze age 8 Western
Low Intermediate High Traditional Low Intermediate High Other Low Intermediate High Doctor diagnosis of asthma by age 8 Western Low Intermediate High Traditional Low Intermediate High Other Low Intermediate High Current Asthma Age 8 Western Low Intermediate High Traditional Low Intermediate High Other Low Intermediate High Allergic sensitization Age 8 Western Low Intermediate High Traditional Low Intermediate High Other Low Intermediate High
Univariate OR (95% CI) n = 753
Reference 0.74 (0.44–1.25) 1.55 (0.97–2.47) Reference 0.94 (0.57–1.54) 1.10 (0.68–1.78) Reference 1.26 (0.76–2.09) 1.55 (0.94–2.54) Reference 1.32 (0.85–2.06) 2.00 (1.31–3.07 Reference 1.07 (0.70–1.63) 1.00 (0.66–1.53) Reference 1.01 (0.67–1.53) 0.97 (0.64–1.48) Reference 0.84 (0.45–1.58) 2.03 (1.18–3.50) Reference 1.20 (0.69–2.10) 1.05 (0.59–1.86) Reference 1.14 (0.64–2.04) 1.50 (0.85–2.63) Reference 0.81 (0.54–1.20) 0.89 (0.60–1.33) Reference 0.99 (0.66–1.48) 1.17 (0.79–1.75) Reference 1.15 (0.77–1.71) 1.29 (0.86–1.93)
p value
0.26 0.06 0.81 0.69 0.36 0.09
0.21 0.001 0.75 0.99 0.97 0.89
0.60 0.01 0.52 0.86 0.66 0.16
0.29 0.58 0.95 0.44 0.49 0.22
Multivariate OR (95%CI) n = 636
Reference 0.83 (0.44–1.58) 1.54 (0.77–3.07) Reference 0.98 (0.53–1.79) 0.94 (0.50–1.77) Reference 1.17 (0.63–2.18) 1.65 (0.88–3.08) Reference 1.51 (0.87–2.59) 2.19 (1.20–4.01) Reference 1.10 (0.66–1.83) 0.97 (0.56–1.66) Reference 0.96 (0.58–1.57) 1.15 (0.69–1.94) Reference 1.05 (0.48–2.27) 2.59 (1.15–5.81) Reference 1.34 (0.67–2.67) 0.78 (0.36–1.67) Reference 1.09 (0.52–2.26) 1.85 (0.90–3.80) Reference 0.82 (0.53–1.27) 0.90 (0.55–1.49) Reference 1.11 (0.72–1.71) 1.27 (0.81–2.00) Reference 1.12 (0.74–1.71) 1.38 (0.89–2.14)
p value
0.58 0.22 0.94 0.84 0.61 0.12
0.14 0.01 0.71 0.90 0.86 0.60
0.91 0.02 0.41 0.51 0.82 0.10
0.38 0.69 0.64 0.30 0.59 0.15
P values 75th percentile) was significantly associated with wheezing in the previous 12 months. Although the results of these studies
Pediatric Allergy and Immunology 25 (2014) 565–571 ª 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
Patel et al.
Dietary patterns and asthma
Table 3 Univariate and multivariate associations between dietary patterns and clinical outcomes at age 11 yr. [multivariate model adjusted for gender, parental asthma, parental smoking, BMI z-score, socio-economic status, energy intake and allergic sensitization (where appropriate)]
Dietary pattern
Adherence
Current wheeze age 11 Western
Low Intermediate High Traditional Low Intermediate High Other Low Intermediate High Doctor diagnosis of asthma by age 11 Western Low Intermediate High Traditional Low Intermediate High Other Low Intermediate High Current Asthma Age 11 Western Low Intermediate High Traditional Low Intermediate High Other Low Intermediate High Allergic sensitization Age 11 Western Low Intermediate High Traditional Low Intermediate High Other Low Intermediate High
Univariate OR (95% CI) n = 753
Reference 1.01 (0.63–1.63) 1.35 (0.85–2.14) Reference 1.36 (0.85–2.18) 1.20 (0.75–1.94) Reference 1.54 (0.96–2.45) 1.20 (0.74–1.96) Reference 1.09 (0.73–1.64) 1.42 (0.95–2.12) Reference 0.93 (0.62–1.40) 0.99 (0.68–1.48) Reference 1.11 (0.74–1.65) 1.06 (0.71–1.60) Reference 0.94 (0.54–1.65) 1.49 (0.88–2.52) Reference 1.69 (0.98–2.90) 1.14 (0.65–2.03) Reference 1.26 (0.74–2.13) 1.00 (0.57–1.74) Reference 0.98 (0.66–1.46 0.89 (0.60–1.34) Reference 1.01 (0.67–1.51) 1.12 (0.74–1.67) Reference 0.96 (0.64–1.44) 1.22 (0.82–1.82)
p value
0.97 0.21 0.20 0.45 0.07 0.47
0.67 0.09 0.73 0.96 0.62 0.77
0.84 0.14 0.06 0.64 0.39 0.99
0.92 0.59 0.97 0.60 0.84 0.33
Multivariate OR (95% CI) n = 575
Reference 1.32 (0.73–2.40) 1.81 (0.98–3.37) Reference 1.64 (0.91–2.95) 1.03 (0.55–1.90) Reference 1.66 (0.94–2.91) 1.12 (0.61–2.07) Reference 0.94 (0.55–1.62) 1.30 (0.75–2.25) Reference 1.12 (0.72–2.04) 0.84 (0.49–1.45) Reference 1.04 (0.64–1.70) 1.30 (0.78–2.18) Reference 1.20 (0.58–2.49) 2.20 (1.07–4.51) Reference 1.70 (0.92–3.16) 0.79 (0.40–1.57) Reference 1.16 (0.61–2.20) 0.88 (0.44–1.78) Reference 0.88 (0.57–1.37) 0.96 (0.58–1.59) Reference 1.10 (0.71–1.70) 1.27 (0.80–1.99) Reference 1.09 (0.71–1.67) 1.28 (0.82–1.99)
p value
0.36 0.06 0.10 0.94 0.08 0.71
0.82 0.36 0.47 0.53 0.87 0.32
0.62 0.03 0.09 0.51 0.65 0.73
0.57 0.86 0.68 0.31 0.69 0.28
p values 0.3) for variables included in the analysis. Table S4. Mean (SD) of individual factor loadings of the three dietary patterns by demographic characteristics of study population.
Pediatric Allergy and Immunology 25 (2014) 565–571 ª 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
571
ORIGINAL ARTICLE
Asthma
Cross-sectional association of dietary patterns with asthma and atopic sensitization in childhood – in a cohort study Sumaiya Patel1, Adnan Custovic2, Jaclyn A. Smith2, Angela Simpson2, Gina Kerry2 & Clare S. Murray2 1
School of Social and Community Medicine, University of Bristol, Bristol, UK; 2Centre of Respiratory and Allergy, Institute Of Inflammation and Repair, Manchester Academic Health Science Centre, The University of Manchester, University Hospital of South Manchester NHS Foundation Trust, Manchester, UK
To cite this article: Patel S, Custovic A, Smith JA, Simpson A, Kerry G, Murray CS. Cross-sectional association of dietary patterns with asthma and atopic sensitization in childhood – in a cohort study. Pediatr Allergy Immunol 2014: 25: 565–571.
Keywords dietary patterns; children; asthma; atopy Correspondence Clare Murray, University of Manchester, Education and Research Building, University Hospital of South Manchester, 2nd Floor, Manchester M23 9LT, UK Tel.: +44 161 291 5876 Fax: +44 161 291 5730 E-mail: [email protected] Accepted for publication 6 July 2014 DOI:10.1111/pai.12276
Abstract Background: Many studies have investigated individual nutrients or foods as risk factors for allergic disease, but few have studied dietary patterns. We aimed to use principal component analysis (PCA) to determine dietary patterns in school age children and examine associations between these dietary patterns and wheeze, asthma and sensitization. Methods: Participants in a population-based birth cohort attended review clinics at ages 8 and 11 yr. A validated questionnaire was interviewer-administered to collect information on parentally reported symptoms and doctor-diagnosed asthma. Atopic sensitization was ascertained by skin-prick tests. Current asthma was defined as doctor-diagnosed asthma and wheezing in the previous 12 months. A validated semiquantitative food frequency questionnaire was completed at age 8 yr, and PCA was used to determine dietary patterns. Results: Principal component analysis identified three dietary components, which based on their characteristics we termed as Traditional (mixed meat, fish, fruit and vegetables), Western (predominantly high fat content, processed foods) and Other (predominantly grains and nuts) dietary patterns. High adherence to the Western diet pattern was significantly associated with doctor-diagnosed asthma and current asthma at age 8 yr [aOR (95% CI): 2.19 (1.20–4.01), p = 0.01; 2.59 (1.15–5.81), p = 0.02; respectively]. A similar association was found for current asthma at age 11 yr [aOR (95% CI): 2.20 (1.07–4.51), p = 0.03]. There was no evidence of an association between dietary patterns and current wheeze and allergic sensitization at either age 8 or 11 yr. Conclusion: School age children adhering strongly to a Western diet, high in fat and processed foods, had a higher risk of current asthma and doctor-diagnosed asthma.
As the hypothesis by Seaton (1) and Black and Sharpe (2), there has been considerable interest in the relationship between diet and asthma. A number of nutrients (3) and foods (4, 5) have been associated with a reduced asthma risk in children. However, dietary intervention studies (predominantly using individual nutrients), which were based on such associations, have failed to demonstrate a benefit (6, 7). Neither nutrients nor foods are consumed individually; they are eaten in combination, and the effects of the combination of nutrients/foods may be synergistic. The observed effect of diet on allergic conditions may be due to individual nutrients or specific foods, or alternatively a dietary pattern that is
associated with high or low level of certain nutrients. This may be a reason why observational studies report an association between certain nutrients and clinical outcomes, whilst intervention studies supplementing individual nutrients have failed to show an effect. A number of studies have used a priori assumptions to evaluate dietary patterns [e.g. Mediterranean dietary score (8)]. This method of assessment can be useful as it views the diet as a whole opposed to individual foods/nutrients. In recent years, there has been interest in data-driven statistical approaches such as principal component analysis (PCA) to study the overall diet in terms of dietary patterns, where the diet
Pediatric Allergy and Immunology 25 (2014) 565–571 ª 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
565
Dietary patterns and asthma
Patel et al.
is assessed in terms of eating behaviours (9). PCA is a datadriven method, which assesses the unmeasured constructs or components within the data. A few studies have used PCA to derive dietary patterns of mothers during pregnancy and evaluate its association with wheeze in children (10, 11). More recently, two studies have used PCA to examine dietary patterns and respiratory symptoms in pre-school (12) and in school age (13) children. In this study, we aimed to assess the association between dietary patterns in UK children assessed at age 8 yr with wheeze, asthma and atopic sensitization. We examined outcomes measured at the time of dietary pattern assessment and prospective associations with outcomes measured at age 11 yr.
Methods Study participants The Manchester Asthma and Allergy Study is a populationbased birth cohort and is described in detail elsewhere (14, 15). Briefly, the subjects were recruited in utero (weeks 8–10 of pregnancy) by screening parents in the antenatal clinic, using skin testing and a questionnaire regarding allergic diseases. All infants born at term and with no congenital cardiac or respiratory disease were eligible and were followed prospectively until age 11 yr. The study was approved by the Local Research Ethics Committee and is registered as ICRCTN72673620. Written informed consent was obtained from all parents. Clinical follow-up Participants attended review clinics at 8 and 11 yr. A validated questionnaire was interviewer-administered to collect information on parentally reported symptoms (16). Atopic sensitization was ascertained by skin-prick tests (Dermatophagoides pteronyssinus, cat, dog, grasses, moulds, milk and egg; Bayer, Elkahrt, IN, USA). At age 8 yr, parents and children completed together a validated semi-quantitative food frequency questionnaire (QBuilder, Tinuviel Software, Anglesey, UK), regarding the child’s usual dietary intake over the previous 2–3 months (see online repository). Children were weighed, without shoes or outer clothing (to nearest 0.1 kg), and measured (to nearest 1 cm) using a stadiometer. Body mass index (BMI) was calculated in kg/m2. BMI standard deviation scores (z scores) were calculated for age- and gender-specific values, using UK standards (17). Data on parental smoking, family history and socio-economic status (18) were assessed by questionnaire.
Current asthma. Defined as doctor-diagnosed asthma and current wheeze. Allergic sensitization. Defined as wheal diameter of 3 mm or greater than the negative control to at least one allergen. Statistical methods Statistical analysis was carried out using SPSS 20 (Chicago, IL, USA). Principal component analysis was used to determine a number of components, which can be considered as dietary patterns. The FFQ included 114 food and drink questions. Items in the FFQ mainly used a nine level scale. Questions, which did not use this scale (e.g. quantity of milk, cheese etc.), were coded to represent the same nine level scale. Questions with a dichotomous reply (n = 1), questions regarding types of foods (n = 2) and questions regarding specific portion sizes (e.g. size of bread slice) were excluded (n = 14). Questions regarding the frequency of consumption of these items were included (Table S1, S2). This left a total of 97 items to be included in the PCA. The scree plot showed a clear break in the curve after the third component revealing the presence of three dietary patterns. This solution also provided the most ‘understandable’ components. The three components explained 15.3% of the variance within the data. (further details in online repository). A pattern matrix was produced for the included items. Factor loadings for individual items, greater than 0.3, were retained. The individual factor loading scores represent the correlation coefficients between the food and the specific dietary pattern (Table S3). The regression method of factor score calculation was used to produce a composite score for each component for every case included in the analysis. Each component was then identified by a simple label which best described the items within it. For ease of interpretability, the population was categorized into tertiles according to their score for the dietary pattern as follows: low, moderate and high adherence to each dietary pattern (using ‘low’ score as reference category). The relationship between dietary patterns and the outcome variables was tested using logistic regression. Confounders to be included in the models were decided upon a priori. We considered the following as potential confounders: gender, parental asthma, parental smoking, allergic sensitization, BMI z-score and socio-economic status. The multivariate model was also adjusted for energy intake as this is standard practice in nutritional epidemiology. Results
Definition of clinical outcomes (age 8 and 11 yr) Current Wheeze. Defined as a positive response to the question ‘Has your child had wheezing or whistling in the chest in the last 12 months?’ Doctor-diagnosed asthma. Defined as a positive response to the question ‘Has a doctor ever given your child a diagnosis of asthma?’
566
A total of 1,211 parents were recruited during pregnancy. Of these, 1,184 term deliveries continued in the study after birth; 1,051 children were followed in the observational cohort (133 children were prenatally randomized to an environmental intervention (19) and excluded from this analysis). At age 8 yr, 899 participants completed FFQs (included in the PCA), of whom 753 completed respiratory questionnaires at both
Pediatric Allergy and Immunology 25 (2014) 565–571 ª 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
Patel et al.
Dietary patterns and asthma
age 8 and 11 yr (included in the logistic regression analysis). Children included in this analysis did not differ from those who were not included in terms of gender, gestational age, birthweight, atopic status, parental sensitization, parental asthma, social class and mean energy intake. Children included were less likely to have a parent who smoked at age 8 (28% vs. 36%, p = 0.04), but not at age 11 yr (p = 0.39). Demographics of the included 753 participants are shown in Table 1. Dietary patterns The factor loadings of the food groups for the three dietary patterns (components) are shown in Table S3. Dietary Pattern 1 included mostly fruit and vegetables with meat and oily fish loading lower down; therefore this pattern was termed the ‘Traditional’ dietary pattern. Dietary Pattern 2 contained processed foods, which are associated with a modern Western diet (chips, crisps, pizza); thus, this pattern was termed ‘Western’ dietary pattern. Dietary Pattern 3 contained items that are often eaten by people following a vegetarian diet (lentils, soya, rice, nuts) but also contained fried foods, offal and pastry dishes; this pattern was termed ‘Other’ dietary pattern. Girls showed a higher mean score on the Traditional dietary pattern compared to boys; boys had higher scores for the Western dietary pattern (Table S4). Higher scores for Traditional dietary pattern were observed for higher socio-economic groups, and the opposite was seen for Western dietary pattern. The Traditional dietary pattern was positively associated with BMI (p = 0.006). The Western diet was positively associated with parental smoking (p < 0.001) and maternal asthma (p = 0.05). Relationship between dietary patterns and clinical outcomes Age 8 yr Table 2 shows the associations between dietary patterns and clinical outcomes at age 8 yr. Participants who adhered
Table 1 Demographics of population (n = 753) at age 8 and 11 yr
Gender Age at Follow-up visit* BMI z-score* Current wheeze Doctor diagnosis of asthma by age 8/11 Current asthma Either parent smoking Positive skin prick test Energy intake (KJ/day)* *Mean (SD).
strongly to the Western dietary pattern had greater odds of current wheeze, but this association did not remain significant in the multivariate analysis. Participants who adhered strongly to the Western dietary pattern had significantly greater odds of doctor-diagnosed asthma and current asthma at age 8 yr, and this association remained significant after adjusting for potential confounders. There was no evidence of an association between adherence to either the Traditional or Other dietary patterns to any of the outcomes (Table 2). There were no associations between the dietary patterns and atopic sensitization at age 8 yr (Table 2). Age 11 yr Table 3 shows the prospective associations between the dietary patterns at age 8 yr and clinical outcomes at age 11 yr. After adjusting for confounders, participants adhering strongly to the Western dietary pattern at age 8 yr were at increased risk of current asthma at age 11 yr. There were no associations between the dietary patterns and atopic sensitization at age 11 yr (Table 3).
Discussion Principal findings In this analysis, we have used a data reduction method, PCA, to determine dietary patterns. The dietary components extracted were labelled Traditional, Western and Other dietary patterns. In the univariate analysis, strong adherence to the Western dietary pattern was associated with greater odds of doctor-diagnosed asthma and current asthma at age 8 yr. After adjusting for potential confounders, the odds of current asthma in those participants adhering highly to the Western diet was approximately two and a half times that of those with a low adherence. Participants who strongly adhered to the Western dietary pattern at age 8 remained at significantly greater odds of current asthma at age 11 yr. No evidence of associations was seen with the Traditional and Other dietary patterns and respiratory symptoms, or with any of the three dietary patterns with allergic sensitization.
Age 8
Age 11
Interpretation
416 7.98 0.44 121 176
416 11.5 0.47 132 198
The dietary components we identified have face validity (i.e. intuitively correct) and construct validity (i.e. observed significant relationships between dietary components and sociodemographic factors). In addition, these components are comparable to those reported by other studies (20, 21). We have demonstrated significant relationships of the dietary patterns with gender, socio-economic status, maternal asthma and smoking. In general, the Traditional and Other dietary patterns (‘healthier’ dietary patterns) were associated with higher socio-economic status, and the opposite relationship was seen for the Western dietary pattern. Similar relationships have been found by other studies looking at dietary patterns in British school children (20).
(55.2%) male (0.17) (1.00) (16.1%) (23.4%)
87 (11.6%) 207 (27.8%) n = 745 216 (31.3%) n = 690 7829 (1547)
(55.2%) male (0.54) (1.12) (17.5%) (26.3%)
95 (12.6%) 174 (23.7%) n = 733 218 (34.1%) n = 640 –
Pediatric Allergy and Immunology 25 (2014) 565–571 ª 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
567
Dietary patterns and asthma
Patel et al.
Table 2 Univariate and multivariate associations between dietary patterns and clinical outcomes at age 8 yr. [multivariate model adjusted for gender, parental asthma, parental smoking, BMI z-score, socio-economic status, energy intake and allergic sensitization (where appropriate)]
Dietary pattern
Adherence
Current wheeze age 8 Western
Low Intermediate High Traditional Low Intermediate High Other Low Intermediate High Doctor diagnosis of asthma by age 8 Western Low Intermediate High Traditional Low Intermediate High Other Low Intermediate High Current Asthma Age 8 Western Low Intermediate High Traditional Low Intermediate High Other Low Intermediate High Allergic sensitization Age 8 Western Low Intermediate High Traditional Low Intermediate High Other Low Intermediate High
Univariate OR (95% CI) n = 753
Reference 0.74 (0.44–1.25) 1.55 (0.97–2.47) Reference 0.94 (0.57–1.54) 1.10 (0.68–1.78) Reference 1.26 (0.76–2.09) 1.55 (0.94–2.54) Reference 1.32 (0.85–2.06) 2.00 (1.31–3.07 Reference 1.07 (0.70–1.63) 1.00 (0.66–1.53) Reference 1.01 (0.67–1.53) 0.97 (0.64–1.48) Reference 0.84 (0.45–1.58) 2.03 (1.18–3.50) Reference 1.20 (0.69–2.10) 1.05 (0.59–1.86) Reference 1.14 (0.64–2.04) 1.50 (0.85–2.63) Reference 0.81 (0.54–1.20) 0.89 (0.60–1.33) Reference 0.99 (0.66–1.48) 1.17 (0.79–1.75) Reference 1.15 (0.77–1.71) 1.29 (0.86–1.93)
p value
0.26 0.06 0.81 0.69 0.36 0.09
0.21 0.001 0.75 0.99 0.97 0.89
0.60 0.01 0.52 0.86 0.66 0.16
0.29 0.58 0.95 0.44 0.49 0.22
Multivariate OR (95%CI) n = 636
Reference 0.83 (0.44–1.58) 1.54 (0.77–3.07) Reference 0.98 (0.53–1.79) 0.94 (0.50–1.77) Reference 1.17 (0.63–2.18) 1.65 (0.88–3.08) Reference 1.51 (0.87–2.59) 2.19 (1.20–4.01) Reference 1.10 (0.66–1.83) 0.97 (0.56–1.66) Reference 0.96 (0.58–1.57) 1.15 (0.69–1.94) Reference 1.05 (0.48–2.27) 2.59 (1.15–5.81) Reference 1.34 (0.67–2.67) 0.78 (0.36–1.67) Reference 1.09 (0.52–2.26) 1.85 (0.90–3.80) Reference 0.82 (0.53–1.27) 0.90 (0.55–1.49) Reference 1.11 (0.72–1.71) 1.27 (0.81–2.00) Reference 1.12 (0.74–1.71) 1.38 (0.89–2.14)
p value
0.58 0.22 0.94 0.84 0.61 0.12
0.14 0.01 0.71 0.90 0.86 0.60
0.91 0.02 0.41 0.51 0.82 0.10
0.38 0.69 0.64 0.30 0.59 0.15
P values 75th percentile) was significantly associated with wheezing in the previous 12 months. Although the results of these studies
Pediatric Allergy and Immunology 25 (2014) 565–571 ª 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
Patel et al.
Dietary patterns and asthma
Table 3 Univariate and multivariate associations between dietary patterns and clinical outcomes at age 11 yr. [multivariate model adjusted for gender, parental asthma, parental smoking, BMI z-score, socio-economic status, energy intake and allergic sensitization (where appropriate)]
Dietary pattern
Adherence
Current wheeze age 11 Western
Low Intermediate High Traditional Low Intermediate High Other Low Intermediate High Doctor diagnosis of asthma by age 11 Western Low Intermediate High Traditional Low Intermediate High Other Low Intermediate High Current Asthma Age 11 Western Low Intermediate High Traditional Low Intermediate High Other Low Intermediate High Allergic sensitization Age 11 Western Low Intermediate High Traditional Low Intermediate High Other Low Intermediate High
Univariate OR (95% CI) n = 753
Reference 1.01 (0.63–1.63) 1.35 (0.85–2.14) Reference 1.36 (0.85–2.18) 1.20 (0.75–1.94) Reference 1.54 (0.96–2.45) 1.20 (0.74–1.96) Reference 1.09 (0.73–1.64) 1.42 (0.95–2.12) Reference 0.93 (0.62–1.40) 0.99 (0.68–1.48) Reference 1.11 (0.74–1.65) 1.06 (0.71–1.60) Reference 0.94 (0.54–1.65) 1.49 (0.88–2.52) Reference 1.69 (0.98–2.90) 1.14 (0.65–2.03) Reference 1.26 (0.74–2.13) 1.00 (0.57–1.74) Reference 0.98 (0.66–1.46 0.89 (0.60–1.34) Reference 1.01 (0.67–1.51) 1.12 (0.74–1.67) Reference 0.96 (0.64–1.44) 1.22 (0.82–1.82)
p value
0.97 0.21 0.20 0.45 0.07 0.47
0.67 0.09 0.73 0.96 0.62 0.77
0.84 0.14 0.06 0.64 0.39 0.99
0.92 0.59 0.97 0.60 0.84 0.33
Multivariate OR (95% CI) n = 575
Reference 1.32 (0.73–2.40) 1.81 (0.98–3.37) Reference 1.64 (0.91–2.95) 1.03 (0.55–1.90) Reference 1.66 (0.94–2.91) 1.12 (0.61–2.07) Reference 0.94 (0.55–1.62) 1.30 (0.75–2.25) Reference 1.12 (0.72–2.04) 0.84 (0.49–1.45) Reference 1.04 (0.64–1.70) 1.30 (0.78–2.18) Reference 1.20 (0.58–2.49) 2.20 (1.07–4.51) Reference 1.70 (0.92–3.16) 0.79 (0.40–1.57) Reference 1.16 (0.61–2.20) 0.88 (0.44–1.78) Reference 0.88 (0.57–1.37) 0.96 (0.58–1.59) Reference 1.10 (0.71–1.70) 1.27 (0.80–1.99) Reference 1.09 (0.71–1.67) 1.28 (0.82–1.99)
p value
0.36 0.06 0.10 0.94 0.08 0.71
0.82 0.36 0.47 0.53 0.87 0.32
0.62 0.03 0.09 0.51 0.65 0.73
0.57 0.86 0.68 0.31 0.69 0.28
p values 0.3) for variables included in the analysis. Table S4. Mean (SD) of individual factor loadings of the three dietary patterns by demographic characteristics of study population.
Pediatric Allergy and Immunology 25 (2014) 565–571 ª 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
571
