Comparative strategies for using cluster analysis to assess dietary patterns

Dietary patterns and sleep disorders in Mexican adults from a National Health and Nutrition Survey

Given the high prevalence of multiple non-communicable chronic diseases in Mexico, the aim of the present study was to assess the association between dietary patterns and sleep disorders in a national representative sample of 5076 Mexican adults (20–59 years) from the 2016 National Health and Nutrition Survey. Through a cross-sectional study, we used the Berlin sleep symptoms questionnaire to estimate the proportion of adults with insomnia, obstructive sleep apnoea (OSA) and other related problems such as daytime symptoms and inadequate sleep duration. Dietary data were collected through a seven-day semi-quantitative food frequency questionnaire, and dietary patterns were determined through cluster analysis. Associations between dietary patterns and sleep disorders were assessed by multivariate logistic regression models adjusted for age, sex, well-being, rural/urban area type, geographical region, tobacco use, physical activity level and energy intake. Three dietary patterns were identified: traditional (high in legumes and tortilla), industrialised (high in sugar-sweetened beverages, fast foods, and alcohol, coffee or tea) and mixed (high in meat, poultry, fruits and vegetables). Multivariate logistic regression showed that the industrialised pattern yielded higher odds for daytime symptoms (OR 1⋅49; 95 % CI 1⋅12, 1⋅99) and OSA (OR 1⋅63; 95 % CI 1⋅21, 2⋅19) compared with the traditional pattern. In conclusion, dietary patterns are associated with sleep disorders in Mexican adults. Further research is required to break the vicious cycle of poor-quality diet, sleep symptoms and health.

Classifying Individuals Into a Dietary Pattern Based on Metabolomic Data

SCOPE

The objectives are to develop a metabolomic-based model capable of classifying individuals into dietary patterns and to investigate the reproducibility of the model.

METHODS AND RESULTS

K-means cluster analysis is employed to derive dietary patterns using metabolomic data. Differences across the dietary patterns are examined using nutrient biomarkers. The model is used to assign individuals to a dietary pattern in an independent cohort, A-DIET Confirm (n = 175) at four time points. The stability of participants to a dietary pattern is assessed. Four dietary patterns are derived: moderately unhealthy, convenience, moderately healthy, and prudent. The moderately unhealthy and convenience patterns has lower adherence to the alternative healthy eating index (AHEI) and the alternative mediterranean diet score (AMDS) compared to the moderately healthy and prudent patterns (AHEI = 24.5 and 22.9 vs 26.7 and 28.4, p < 0.001). The dietary patterns are replicated in A-DIET Confirm, with good reproducibility across four time points. The stability of participants' dietary pattern membership ranged from 25.0% to 61.5%.

CONCLUSION

The multivariate model classifies individuals into dietary patterns based on metabolomic data. In an independent cohort, the model classifies individuals into dietary patterns at multiple time points furthering the potential of such an approach for nutrition research.

Dietary Patterns and Health Outcomes among African American Maintenance Hemodialysis Patients

The association between dietary patterns and health outcomes, such as quality of life (QOL), in maintenance hemodialysis (MHD) patients with certain racial backgrounds has not been studied in detail. QOL is a powerful outcome measure in which dietary patterns could be a modifying factor. This study is a secondary analysis examining the association between dietary patterns and health outcomes in 101 African American (AA) maintenance hemodialysis (MHD) patients participating in the Palm Tocotrienols in Chronic Hemodialysis (PATCH) study. Quality of life (QOL) was assessed using the Kidney Disease Quality of Life 36-item survey (KDQOL-36™). Blood samples were analyzed for lipids, lipoprotein subfractions, and inflammatory markers. Food intake was measured using six non-consecutive 24-h dietary recalls over 15 months. Implausible energy intake reports were screened out by comparing reported energy intake (rEI) with predicted total energy expenditure (pTEE). Cluster analysis, using the k-means algorithm, identified two distinct dietary patterns in the study population: a high "sugar sweetened beverage" pattern (hiSSB) and a low "sugar sweetened beverage pattern" (loSSB). In the hiSSB group, consumption of SSB accounted for ~28% of energy intake, while SSB represented only 9% of energy intake in the loSSB group. The hiSSB group was characterized by a higher intake of total calories, sugar and percentage of kilocalories from carbohydrates, whereas the percentage of kilocalories from protein and fat was lower. While additional micronutrient intakes differed between groups (vitamin C, zinc, chromium), these were significantly lower than recommended values in the entire cohort. Patients in the hiSSB group presented with lower high-density lipoprotein cholesterol (HDL-C), lower large HDL particles and smaller low density lipoprotein (LDL) particle diameters. Antidepressant usage was significantly higher in the hiSSB group. Patients in the hiSSB group scored lower across all five KDQOL domains and scored significantly lower in the mental composite domain. MHD patients following a hiSSB dietary pattern had smaller dense LDL particles, lower HDL-C, and a lower QOL. Suboptimal intakes of fruits, vegetables, and grains as well as key micronutrients were evident in both patterns.

Reproducibility and Validity of A Posteriori Dietary Patterns: A Systematic Review

The effective use of dietary patterns (DPs) remains limited. There is a need to assess their consistency over multiple administrations of the same dietary source, different dietary sources, or across different studies. Similarly, their generalizability should be based on a previous assessment of DP construct validity. However, to date, no systematic reviews of reproducibility and validity of a posteriori DPs have been carried out. In addition, several methodological questions related to their identification are still open and prevent a fair comparison of epidemiological results on DPs and disease. A systematic review of the literature on the PubMed database was conducted. We identified 218 articles, 64 of which met the inclusion criteria. Of these, the 38 articles dealing with reproducibility and relative and construct validity of DPs were included. These articles (published in 1999-2017, 53% from 2010 onwards) were based on observational studies conducted worldwide. The 14 articles that assessed DP reproducibility across different statistical solutions examined different research questions. Included were: the number of food groups or subjects; input variable format (as well as adjustment for energy intake); algorithms and the number of DPs to retain in cluster analysis; rotation method; and score calculation in factor analysis. However, we identified at most 3 articles per research question on DP reproducibility across statistical solutions. From another 15 articles, reproducibility of DPs over shorter (≤1 y) time periods was generally good and higher than DP relative validity (as measured across different dietary sources). Confirmatory factor analysis was used in 15 of the included articles. It provided reassuring results in identifying valid dietary constructs characterizing the populations under consideration. Based on the available evidence, only suggestive conclusions can be derived on reproducibility across different statistical solutions. Nevertheless, most identified DPs showed good reproducibility, fair relative validity, and good construct validity.

Sociodemographic and Regional Determinants of Dietary Patterns in Russia

An empirical assessment of diets using a posteriori analysis allows us to define actual dietary patterns (DPs) in the food consumption structure of a population. This study represents an a posteriori assessment of DPs for the Russian population in general as well as their dependence on socio-demographic and regional parameters. The data were obtained from 21,923 individuals aged 25-64 years old from a Russian multicenter study of "Epidemiology of Cardiovascular Diseases in the Regions of the Russian Federation" conducted in 2013-2014. Cross-sectional study subjects were interviewed face-to-face in order to obtain data on their diet. DPs were defined using principal component analysis. Four DPs were specified as "Rational", "Salt", "Meat", and "Mixed"; all these variants together accounted for 55.9% of variance. Diets in gender and age groups corresponded to those for an all-Russian population; in several regions, the defined diets differed from the all-Russian ones. More favorable diet trends were observed among women, people with no family, people not working, and urban dwellers, and diet trends were more favorable with an increase in age, level of education, and material wealth. Thus, a posteriori DPs were defined for the Russian population, which were stable in sex/age groups and were mediated by the socio-demographic characteristics of the population.

Education and lifestyle predict change in dietary patterns and diet quality of adults 55 years and over

Background

Diet is a key risk factor for chronic disease, and an increasing concern among older adults. We aim to examine the changes in dietary patterns using principal component analysis and a diet quality index among older adults and examine the predictors of dietary change over a 4 year period.

Methods

Data was obtained via a postal survey in a prospective cohort, the Wellbeing Eating and Exercise for a Long Life (WELL) study. Australian adults aged 55 years and over (n = 1005 men and n = 1106 women) completed a food frequency at three time points and provided self-reported personal characteristics. Principal component analysis was used to assess dietary patterns and diet quality was assessed using the 2013 Revised Dietary Guideline Index. The relationships between predictors and change in dietary patterns were assessed by multiple linear regression.

Results

Two dietary patterns were consistently identified in men and women at three time points over 4 years. One was characterised by vegetables, fruit and white meat, and the other was characterised by red and processed meat and processed foods. Reduced consumption of key food groups within the principal component analysis-determined dietary patterns was observed. An increase in diet quality over 4 years was observed in men only. Reported higher education levels and favourable lifestyle characteristics, including not smoking and physical activity, at baseline predicted an increase in healthier dietary patterns over 4 years.

Conclusions

There was stability in the main dietary patterns identified over time, however participants reported an overall decrease in the frequency of consumption of key food groups. Compliance with the Australian Dietary Guidelines remained poor and therefore targeting this population in nutritional initiatives is important. Design of nutrition promotion for older adults need to consider those with lower socioeconomic status, as having a lower level of education was a predictor of poorer dietary patterns. It is important to consider how nutrition behaviours can be targeted alongside other lifestyle behaviours, such as smoking and inadequate physical activity to improve health.

Identifying heterogeneous health profiles of primary care utilizers and their differential healthcare utilization and mortality - a retrospective cohort study

BACKGROUND

Heterogeneity of population health needs and the resultant difficulty in health care resources planning are challenges faced by primary care systems globally. To address this challenge in population health management, it is critical to have a better understanding of primary care utilizers' heterogeneous health profiles. We aimed to segment a population of primary care utilizers into classes with unique disease patterns, and to report the 1 year follow up healthcare utilizations and all-cause mortality across the classes.

METHODS

Using de-identified administrative data, we included all adult Singapore citizens or permanent residents who utilized Singapore Health Services (SingHealth) primary care services in 2012. Latent class analysis was used to identify patient subgroups having unique disease patterns in the population. The models were assessed by Bayesian Information Criterion and clinical interpretability. We compared healthcare utilizations in 2013 and one-year all-cause mortality across classes and performed regression analysis to assess predictive ability of class membership on healthcare utilizations and mortality.

RESULTS

We included 100,747 patients in total. The best model (k = 6) revealed the following classes of patients: Class 1 "Relatively healthy" (n = 58,213), Class 2 "Stable metabolic disease" (n = 26,309), Class 3 "Metabolic disease with vascular complications" (n = 2964), Class 4 "High respiratory disease burden" (n = 1104), Class 5 "High metabolic disease without complication" (n = 11,122), and Class 6 "Metabolic disease with multi-organ complication" (n = 1035). The six derived classes had different disease patterns in 2012 and 1 year follow up healthcare utilizations and mortality in 2013. "Metabolic disease with multiple organ complications" class had the highest healthcare utilization (e.g. incidence rate ratio = 19.68 for hospital admissions) and highest one-year all-cause mortality (hazard ratio = 27.97).

CONCLUSIONS

Primary care utilizers are heterogeneous and can be segmented by latent class analysis into classes with unique disease patterns, healthcare utilizations and all-cause mortality. This information is critical to population level health resource planning and population health policy formulation.

A systematic review of the clinical application of data-driven population segmentation analysis

BACKGROUND

Data-driven population segmentation analysis utilizes data analytics to divide a heterogeneous population into parsimonious and relatively homogenous groups with similar healthcare characteristics. It is a promising patient-centric analysis that enables effective integrated healthcare interventions specific for each segment. Although widely applied, there is no systematic review on the clinical application of data-driven population segmentation analysis.

METHODS

We carried out a systematic literature search using PubMed, Embase and Web of Science following PRISMA criteria. We included English peer-reviewed articles that applied data-driven population segmentation analysis on empirical health data. We summarized the clinical settings in which segmentation analysis was applied, compared and contrasted strengths, limitations, and practical considerations of different segmentation methods, and assessed the segmentation outcome of all included studies. The studies were assessed by two independent reviewers.

RESULTS

We retrieved 14,514 articles and included 216 articles. Data-driven population segmentation analysis was widely used in different clinical contexts. 163 studies examined the general population while 53 focused on specific population with certain diseases or conditions, including psychological, oncological, respiratory, cardiovascular, and gastrointestinal conditions. Variables used for segmentation in the studies are heterogeneous. Most studies (n = 170) utilized secondary data in community settings (n = 185). The most common segmentation method was latent class/profile/transition/growth analysis (n = 96) followed by K-means cluster analysis (n = 60) and hierarchical analysis (n = 50), each having its advantages, disadvantages, and practical considerations. We also identified key criteria to evaluate a segmentation framework: internal validity, external validity, identifiability/interpretability, substantiality, stability, actionability/accessibility, and parsimony.

CONCLUSIONS

Data-driven population segmentation has been widely applied and holds great potential in managing population health. The evaluations of segmentation outcome require the interplay of data analytics and subject matter expertise. The optimal framework for segmentation requires further research.

Application of a new dietary pattern analysis method in nutritional epidemiology

Background

Diet plays an important role in chronic disease, and the use of dietary pattern analysis has grown rapidly as a way of deconstructing the complexity of nutritional intake and its relation to health. Pattern analysis methods, such as principal component analysis (PCA), have been used to investigate various dimensions of diet. Existing analytic methods, however, do not fully utilize the predictive potential of dietary assessment data. In particular, these methods are often suboptimal at predicting clinically important variables.

Methods

We propose a new dietary pattern analysis method using the advanced LASSO (Least Absolute Shrinkage and Selection Operator) model to improve the prediction of disease-related risk factors. Despite the potential advantages of LASSO, this is the first time that the model has been adapted for dietary pattern analysis. Hence, the systematic evaluation of the LASSO model as applied to dietary data and health outcomes is highly innovative and novel. Using Food Frequency Questionnaire data from NHANES 2005–2006, we apply PCA and LASSO to identify dietary patterns related to cardiovascular disease risk factors in healthy US adults (n = 2609) after controlling for confounding variables (e.g., age and BMI). Both analyses account for the sampling weights. Model performance in terms of prediction accuracy is evaluated using an independent test set.

Results

PCA yields 10 principal components (PCs) that together account for 65% of the variation in the data set and represent distinct dietary patterns. These PCs are then used as predictors in a regression model to predict cardiovascular disease risk factors. We find that LASSO better predicts levels of triglycerides, LDL cholesterol, HDL cholesterol, and total cholesterol (adjusted R² = 0.861, 0.899, 0.890, and 0.935 respectively) than does the traditional, linear-regression-based, dietary pattern analysis method (adjusted R² = 0.163, 0.005, 0.235, and 0.024 respectively) when the latter is applied to components derived from PCA.

Conclusions

The proposed method is shown to be an appropriate and promising statistical means of deriving dietary patterns predictive of cardiovascular disease risk. Future studies, involving different diseases and risk factors, will be necessary before LASSO’s broader usefulness in nutritional epidemiology can be established.

Maternal dietary patterns during pregnancy and intelligence quotients in the offspring at 8 years of age: Findings from the ALSPAC cohort

Dietary intake during pregnancy may influence child neurodevelopment and cognitive function. This study aims to investigate the associations between dietary patterns obtained in pregnancy and intelligence quotients (IQ) among offspring at 8 years of age. Pregnant women enrolled in the Avon Longitudinal Study of Parents and Children completed a food frequency questionnaire at 32 weeks' gestation (n = 12,195). Dietary patterns were obtained by cluster analysis. Three clusters best described women's diets during pregnancy: “fruit and vegetables,” “meat and potatoes,” and “white bread and coffee.” The offspring's IQ at 8 years of age was assessed using the Wechsler Intelligence Scale for Children. Models, using variables correlated to IQ data, were performed to impute missing values. Linear regression models were employed to investigate associations between the maternal clusters and IQ in childhood. Children of women who were classified in the meat and potatoes cluster and white bread and coffee cluster during pregnancy had lower average verbal (β = −1.74; p < .001 and β = −3.05; p < .001), performance (β = −1.26; p = .011 and β = −1.75; p < .001), and full‐scale IQ (β = −1.74; p < .001 and β = −2.79; p < .001) at 8 years of age when compared to children of mothers in the fruit and vegetables cluster in imputed models of IQ and all confounders, after adjustment for a wide range of known confounders including maternal education. The pregnant women who were classified in the fruit and vegetables cluster had offspring with higher average IQ compared with offspring of mothers in the meat and potatoes cluster and white bread and coffee cluster.

Identification of chemical mixtures to which Canadian pregnant women are exposed: The MIREC Study

Depending on the chemical and the outcome, prenatal exposures to environmental chemicals can lead to adverse effects on the pregnancy and child development, especially if exposure occurs during early gestation. Instead of focusing on prenatal exposure to individual chemicals, more studies have taken into account that humans are exposed to multiple environmental chemicals on a daily basis. The objectives of this analysis were to identify the pattern of chemical mixtures to which women are exposed and to characterize women with elevated exposures to various mixtures. Statistical techniques were applied to 28 chemicals measured simultaneously in the first trimester and socio-demographic factors of 1744 participants from the Maternal-Infant Research on Environment Chemicals (MIREC) Study. Cluster analysis was implemented to categorize participants based on their socio-demographic characteristics, while principal component analysis (PCA) was used to extract the chemicals with similar patterns and to reduce the dimension of the dataset. Next, hypothesis testing determined if the mean converted concentrations of chemical substances differed significantly among women with different socio-demographic backgrounds as well as among clusters. Cluster analysis identified six main socio-demographic clusters. Eleven components, which explained approximately 70% of the variance in the data, were retained in the PCA. Persistent organic pollutants (PCB118, PCB138, PCB153, PCB180, OXYCHLOR and TRANSNONA) and phthalates (MEOHP, MEHHP and MEHP) dominated the first and second components, respectively, and the first two components explained 25.8% of the source variation. Prenatal exposure to persistent organic pollutants (first component) were positively associated with women who have lower education or higher income, were born in Canada, have BMI ≥25, or were expecting their first child in our study population. MEOHP, MEHHP and MEHP, dominating the second component, were detected in at least 98% of 1744 participants in our cohort study; however, no particular group of pregnant women was identified to be highly exposed to phthalates. While widely recognized as important to studying potential health effects, identifying the mixture of chemicals to which various segments of the population are exposed has been problematic. We present an approach using factor analysis through principal component method and cluster analysis as an attempt to determine the pregnancy exposome. Future studies should focus on how to include these matrices in examining the health effects of prenatal exposure to chemical mixtures in pregnant women and their children.

Dietary patterns by cluster analysis in pregnant women: relationship with nutrient intakes and dietary patterns in 7-year-old offspring

Little is known about how dietary patterns of mothers and their children track over time. The objectives of this study are to obtain dietary patterns in pregnancy using cluster analysis, to examine women's mean nutrient intakes in each cluster and to compare the dietary patterns of mothers to those of their children. Pregnant women (n = 12 195) from the Avon Longitudinal Study of Parents and Children reported their frequency of consumption of 47 foods and food groups. These data were used to obtain dietary patterns during pregnancy by cluster analysis. The absolute and energy‐adjusted nutrient intakes were compared between clusters. Women's dietary patterns were compared with previously derived clusters of their children at 7 years of age. Multinomial logistic regression was performed to evaluate relationships comparing maternal and offspring clusters. Three maternal clusters were identified: ‘fruit and vegetables’, ‘meat and potatoes’ and ‘white bread and coffee’. After energy adjustment women in the ‘fruit and vegetables’ cluster had the highest mean nutrient intakes. Mothers in the ‘fruit and vegetables’ cluster were more likely than mothers in ‘meat and potatoes’ (adjusted odds ratio [OR]: 2.00; 95% Confidence Interval [CI]: 1.69–2.36) or ‘white bread and coffee’ (OR: 2.18; 95% CI: 1.87–2.53) clusters to have children in a ‘plant‐based’ cluster. However the majority of children were in clusters unrelated to their mother dietary pattern. Three distinct dietary patterns were obtained in pregnancy; the ‘fruit and vegetables’ pattern being the most nutrient dense. Mothers' dietary patterns were associated with but did not dominate offspring dietary patterns.

Dietary patterns in the French adult population: a study from the second French national cross-sectional dietary survey (INCA2) (2006-2007)

Identification and characterisation of dietary patterns are needed to define public health policies to promote better food behaviours. The aim of this study was to identify the major dietary patterns in the French adult population and to determine their main demographic, socio-economic, nutritional and environmental characteristics. Dietary patterns were defined from food consumption data collected in the second French national cross-sectional dietary survey (2006-2007). Non-negative-matrix factorisation method, followed by a cluster analysis, was implemented to derive the dietary patterns. Logistic regressions were then used to determine their main demographic and socio-economic characteristics. Finally, nutritional profiles and contaminant exposure levels of dietary patterns were compared using ANOVA. Seven dietary patterns, with specific food consumption behaviours, were identified: 'Small eater', 'Health conscious', 'Mediterranean', 'Sweet and processed', 'Traditional', 'Snacker' and 'Basic consumer'. For instance, the Health-conscious pattern was characterised by a high consumption of low-fat and light products. Individuals belonging to this pattern were likely to be older and to have a better nutritional profile than the overall population, but were more exposed to many contaminants. Conversely, individuals of Snacker pattern were likely to be younger, consumed more highly processed foods, had a nutrient-poor profile but were exposed to a limited number of food contaminants. The study identified main dietary patterns in the French adult population with distinct food behaviours and specific demographic, socio-economic, nutritional and environmental features. Paradoxically, for better dietary patterns, potential health risks cannot be ruled out. Therefore, this study demonstrated the need to conduct a risk-benefit analysis to define efficient public health policies regarding diet.

A comparison of the dietary patterns derived by principal component analysis and cluster analysis in older Australians

BACKGROUND

Despite increased use of dietary pattern methods in nutritional epidemiology, there have been few direct comparisons of methods. Older adults are a particularly understudied population in the dietary pattern literature. This study aimed to compare dietary patterns derived by principal component analysis (PCA) and cluster analysis (CA) in older adults and to examine their associations with socio-demographic and health behaviours.

METHODS

Men (n = 1888) and women (n = 2071) aged 55-65 years completed a 111-item food frequency questionnaire in 2010. Food items were collapsed into 52 food groups and dietary patterns were determined by PCA and CA. Associations between dietary patterns and participant characteristics were examined using Chi-square analysis. The standardised PCA-derived dietary patterns were compared across the clusters using one-way ANOVA.

RESULTS

PCA identified four dietary patterns in men and two dietary patterns in women. CA identified three dietary patterns in both men and women. Men in cluster 1 (fruit, vegetables, wholegrains, fish and poultry) scored higher on PCA factor 1 (vegetable dishes, fruit, fish and poultry) and factor 4 (vegetables) compared to factor 2 (spreads, biscuits, cakes and confectionery) and factor 3 (red meat, processed meat, white-bread and hot chips) (mean, 95% CI; 0.92, 0.82-1.02 vs. 0.74, 0.63-0.84 vs. -0.43, -0.50- -0.35 vs. 0.60 0.46-0.74, respectively). Women in cluster 1 (fruit, vegetables and fish) scored highest on PCA factor 1 (fruit, vegetables and fish) compared to factor 2 (processed meat, hot chips cakes and confectionery) (1.05, 0.97-1.14 vs. -0.14, -0.21- -0.07, respectively). Cluster 3 (small eaters) in both men and women had negative factor scores for all the identified PCA dietary patterns. Those with dietary patterns characterised by higher consumption of red and processed meat and refined grains were more likely to be Australian-born, have a lower level of education, a higher BMI, smoke and did not meet physical activity recommendations (all P < 0.05).

CONCLUSIONS

PCA and CA identified comparable dietary patterns within older Australians. However, PCA may provide some advantages compared to CA with respect to interpretability of the resulting dietary patterns. Older adults with poor dietary patterns also displayed other negative lifestyle behaviours. Food-based dietary pattern methods may inform dietary advice that is understood by the community.

Capturing changes in dietary patterns among older adults: a latent class analysis of an ageing Irish cohort

OBJECTIVE

Data-driven approaches to dietary patterns are under-utilized; latent class analyses (LCA) are particularly rare. The present study used an LCA to identify subgroups of people with similar dietary patterns, explore changes in dietary patterns over a 10-year period and relate these dynamics to sociodemographic factors and health outcomes.

DESIGN

The 1998 baseline and 2008 follow-up of the Cork and Kerry Diabetes and Heart Disease Study. Diets were assessed with a standard FFQ. LCA, under the assumption of conditional independence, was used to identify mutually exclusive subgroups with different dietary patterns, based on food group consumption.

SETTING

Republic of Ireland.

SUBJECTS

Men and women aged 50-69 years at baseline (n 923) and at 10-year follow-up (n 320).

RESULTS

Three dietary classes emerged: Western, Healthy and Low-Energy. Significant differences in demographic, lifestyle and health outcomes were associated with class membership. Between baseline and follow-up most people remained 'stable' in their dietary class. Most of those who changed class moved to the Healthy class. Higher education was associated with transition to a healthy diet; lower education was associated with stability in an unhealthy pattern. Transition to a healthy diet was associated with higher CVD risk factors at baseline: respondents were significantly more likely to be smokers, centrally obese and to have hypertension (non-significant).

CONCLUSIONS

LCA is useful for exploring dietary patterns transitions. Understanding the predictors of longitudinal stability/transitions in dietary patterns will help target public health initiatives by identifying subgroups most/least likely to change and most/least likely to sustain a change.

A primer on the use of cluster analysis or factor analysis to assess co-occurrence of risk behaviors

OBJECTIVE

The aim of this paper is to provide a guideline to a universal understanding of the analysis of co-occurrence of risk behaviors. The use of cluster analysis and factor analysis was clarified.

METHOD

A theoretical introduction to cluster analysis and factor analysis and examples from literature were provided. A representative sample (N=4395) of the Dutch population, aged 16-40 and participating from fall 2005 to spring 2006, was used to illustrate the use of both techniques in assessing the co-occurrence of risk behaviors.

RESULTS

Whereas cluster analysis techniques serve to focus on particular clusters of individuals showing the same behavioral pattern, factor analysis techniques are used to assess possible groups of interrelated health-risk behaviors that can be explained by an unknown common source. Choice between the techniques partly depends on the research question and the aim of the research, and has different implications for inferences and policy.

CONCLUSION

By integrating theory and results from an illustrative example, a guideline has been provided that contributes towards a systematic approach in the assessment of co-occurrence of risk behaviors. Following this guideline, a better comparison between outcomes from various studies is expected, leading to improved effectiveness of multiple behavior change interventions.

Dietary patterns during pregnancy and the association with sociodemographic characteristics among women attending general practices in southern Brazil: the ECCAGe Study

The assessment of the relationship between food intake and sociodemographic factors is crucial for developing effective public health policies. The present study aimed to examine dietary patterns in pregnant women and the association between these patterns and sociodemographic characteristics. Pregnant women attending general practices in southern Brazil (n = 712) answered a questionnaire and a food-frequency questionnaire with 88 items. Three dietary patterns were identified using cluster analysis. The association between the dietary patterns and sociodemographic variables was analyzed using the chi-square test and adjusted standardized residuals (p < 0,05). The restricted pattern was associated with lower maternal age, not living with a partner and being a non-working student. The varied pattern was associated with older maternal age, living with a partner, being employed and higher levels of education and income. The common-Brazilian dietary pattern included traditional Brazilian food items and was associated with lower levels of education and income, being unemployed and being a non-student.

Bitter taste perception and dietary intake patterns in irish children

BACKGROUND/AIMS

Variations in bitter receptor gene TAS2R38 affect the perception of bitter-tasting compound 6-n-propylthiouracil (PROP). The perception of PROP has been associated, in some reports, with the perception of fat and sweet tastes, and various food preferences and intakes. The aim was to investigate nutrient intakes and food patterns in a group of Irish children, using K-means cluster analysis, and compare these with TAS2R38 genotype and PROP taster status.

METHODS

Dietary intake was measured via a 3-day diet history in 483 children aged 7-13 years. Children were genotyped for TAS2R38 variation, and PROP taster status was assessed. Anthropometric and socioeconomic data were also obtained.

RESULTS

No differences were observed in macronutrient, micronutrient, or food group consumption between the TAS2R38 genotype and PROP taster groups. K-means cluster analysis identified two distinct dietary patterns, termed 'more healthful' and 'less healthful' clusters. The clusters did not differ in frequencies of TAS2R38 genotype nor PROP taster status groups, suggesting that dietary patterns are not influenced by bitter taste perception.

CONCLUSION

Bitterness perception, as measured by either TAS2R38 genotype or PROP taster status, does not appear to exert a significant effect on patterns of dietary intakes.

Baseline dietary patterns are a significant consideration in correcting dietary exposure for weight loss

BACKGROUND/OBJECTIVES

Dietary pattern studies are traditionally the domain of epidemiological research. From a clinical perspective, there is a need to explore the effects of changing food and dietary patterns of individuals. The aim was to identify patterns of food choice in the context of a clinical weight loss trial. Cluster analysis based on reported serves of food groups revealed dietary patterns informative for the clinical setting.

SUBJECTS/METHODS

Cluster analysis was conducted using diet history data from two clinical trials at baseline, and outcomes at 3 months were reviewed based on these clusters (n=231). The cluster solution was analysed using defined food groups in serves and with respect to clinical parameters and requirements for selected nutrients.

RESULTS

Two distinct dietary patterns were identified from the reported baseline dietary intakes. Subjects in Cluster 1 reported food patterns characterised by higher intakes of low-fat dairy and unsaturated oils and margarine and were generally more closely aligned to food choices encouraged in national dietary guidelines. Subjects in Cluster 2 reported a dietary pattern characterised by non-core foods and drinks, higher- and medium-fat dairy foods, fatty meats and alcohol. At 3 months, Cluster 2 subjects reported greater reductions in energy intake (-5317 kJ; P<0.001) and greater weight loss (-5.6 kg; P<0.05) compared with Cluster 1.

CONCLUSIONS

Overweight subjects with reported dietary patterns similar to dietary guidelines at baseline may have more difficulty in reducing energy intake than those with poor dietary patterns. Correcting exposure to non-core foods and drinks was key to successful weight loss.

Evaluation of methodologies for assessing the overall diet: dietary quality scores and dietary pattern analysis

This paper aims to describe different approaches for studying the overall diet with advantages and limitations. Studies of the overall diet have emerged because the relationship between dietary intake and health is very complex with all kinds of interactions. These cannot be captured well by studying single dietary components. Three main approaches to study the overall diet can be distinguished. The first method is researcher-defined scores or indices of diet quality. These are usually based on guidelines for a healthy diet or on diets known to be healthy. The second approach, using principal component or cluster analysis, is driven by the underlying dietary data. In principal component analysis, scales are derived based on the underlying relationships between food groups, whereas in cluster analysis, subgroups of the population are created with people that cluster together based on their dietary intake. A third approach includes methods that are driven by a combination of biological pathways and the underlying dietary data. Reduced rank regression defines linear combinations of food intakes that maximally explain nutrient intakes or intermediate markers of disease. Decision tree analysis identifies subgroups of a population whose members share dietary characteristics that influence (intermediate markers of) disease. It is concluded that all approaches have advantages and limitations and essentially answer different questions. The third approach is still more in an exploration phase, but seems to have great potential with complementary value. More insight into the utility of conducting studies on the overall diet can be gained if more attention is given to methodological issues.

Classification of main meal patterns--a latent class approach

Relatively little examination of the meals that are prepared in households has been conducted, despite their well-defined properties and widespread community interest in their preparation. The purpose of the present study was to identify the patterns of main meal preparation among Australian adult household meal preparers aged 44 years and younger and 45 years and over, and the relationships between these patterns and likely socio-demographic and psychological predictors. An online cross-sectional survey was conducted by Meat and Livestock Australia among a representative sample of people aged 18–65 years in Australia in 2011. A total of 1076 usable questionnaires were obtained, which included categorical information about the main meal dishes that participants had prepared during the previous 6 months along with demographic information, the presence or absence of children at home, confidence in seasonal food knowledge and personal values. Latent class analysis was applied and four types of usage patterns of thirty-three popular dishes were identified for both age groups, namely, high variety, moderate variety, high protein but low beef and low variety. The meal patterns were associated differentially with the covariates between the age groups. For example, younger women were more likely to prepare a high or moderate variety of meals than younger men, while younger people who had higher levels of education were more likely to prepare high-protein but low-beef meals. Moreover, young respondents with higher BMI were less likely to prepare meals with high protein but low beef content. Among the older age group, married people were more likely to prepare a high or moderate variety of meals than people without partners. Older people who held strong universalist values were more likely to prepare a wide variety of meals with high protein but low beef content. For both age groups, people who had children living at home and those with better seasonal food knowledge were more likely to prepare a high variety of dishes. The identification of classes of meal users would enable health communication to be tailored to improve meal patterns. Moreover, the concept of meals may be useful for health promotion, because people may find it easier to change their consumption of meals rather than individual foods.

The use of cluster analysis to derive dietary patterns: methodological considerations, reproducibility, validity and the effect of energy mis-reporting

Over the last three decades, dietary pattern analysis has come to the forefront of nutritional epidemiology, where the combined effects of total diet on health can be examined. Two analytical approaches are commonly used: a priori and a posteriori. Cluster analysis is a commonly used a posteriori approach, where dietary patterns are derived based on differences in mean dietary intake separating individuals into mutually exclusive, non-overlapping groups. This review examines the literature on dietary patterns derived by cluster analysis in adult population groups, focusing, in particular, on methodological considerations, reproducibility, validity and the effect of energy mis-reporting. There is a wealth of research suggesting that the human diet can be described in terms of a limited number of eating patterns in healthy population groups using cluster analysis, where studies have accounted for differences in sex, age, socio-economic status, geographical area and weight status. Furthermore, patterns have been used to explore relationships with health and chronic diseases and more recently with nutritional biomarkers, suggesting that these patterns are biologically meaningful. Overall, it is apparent that consistent trends emerge when using cluster analysis to derive dietary patterns; however, future studies should focus on the inconsistencies in methodology and the effect of energy mis-reporting.

Exploring statistical approaches to diminish subjectivity of cluster analysis to derive dietary patterns: The Tomorrow Project

Dietary patterns derived by cluster analysis are commonly reported with little information describing how decisions are made at each step of the analytical process. Using food frequency questionnaire data obtained in 2001-2007 on Albertan men (n = 6,445) and women (n = 10,299) aged 35-69 years, the authors explored the use of statistical approaches to diminish the subjectivity inherent in cluster analysis. Reproducibility of cluster solutions, defined as agreement between 2 cluster assignments, by 3 clustering methods (Ward's minimum variance, flexible beta, K means) was evaluated. Ratios of between- versus within-cluster variances were examined, and health-related variables across clusters in the final solution were described. K means produced cluster solutions with the highest reproducibility. For men, 4 clusters were chosen on the basis of ratios of between- versus within-cluster variances, but for women, 3 clusters were chosen on the basis of interpretability of cluster labels and descriptive statistics. In comparison with those in other clusters, men and women in the "healthy" clusters by greater proportions reported normal body mass index, smaller waist circumference, and lower energy intakes. The authors' approach appeared helpful when choosing the clustering method for both sexes and the optimal number of clusters for men, but additional analyses are required to understand why it performed differently for women.

Consumer segmentation based on the level and structure of fruit and vegetable intake: an empirical evidence for US adults from the National Health and Nutrition Examination Survey (NHANES) 2005-2006

OBJECTIVE

To identify consumption patterns of fruit and vegetables within a representative sample of US adults with a focus on degree of produce processing and to explore sociodemographic, lifestyle and nutritional profiles associated with these patterns.

DESIGN

Cross-sectional analysis. Fruit and vegetable (F&V) consumption data were collected using two non-consecutive 24 h recalls. For the purpose of the study, F&V intakes were aggregated into seven subgroups indicating degree of processing, which afterwards were used as inputs into cluster analysis.

SETTING

The 2005-2006 National Health and Nutrition Examination Survey.

SUBJECTS

The sample consisted of 2444 adults aged 20-59 years.

RESULTS

Total average F&V intake of the adults was below the recommended level. Thereby, 20 % of the respondents consumed fruit only in the form of juice. Three F&V consumption patterns were identified: 'low-intake F&V consumers' (74 % of respondents), 'consumers of healthier F&V options' (13 %) and 'intensive fruit juice consumers' (13 %). These groups differed markedly in terms of their sociodemographic, lifestyle and health characteristics, such as gender, age, race/ethnicity, education, smoking, weight status, etc. Differences in nutrient profiles were also found, with the 'consumers of healthier F&V options' showing better nutritional quality compared with other clusters.

CONCLUSIONS

Only a small share of US adults combines high F&V intakes with healthier F&V options that lead to a better nutritional profile. This raises discussion about a need to deliver more specific F&V promotion messages, including advice on healthier preparation methods, especially for the specific population groups.

Food intake patterns and body mass index in older adults: a review of the epidemiological evidence

The relationship between food intake patterns and obesity remains unclear. The objective of the present review was to evaluate the current body of literature on food intake patterns of older adults and their associations with weight status, including obesity and waist circumference. Eleven observational studies were identified and reviewed. Diets characterized by more favorable dietary scores or indices were generally inversely related to body mass index. Results from data-driven approaches were inconsistent. Findings from this literature review suggest that there is no clear relationship between food intake patterns and body mass index or waist circumference in older adults. Limitations, including heterogeneity of food intake patterns and study populations, hinder the ability to make clear comparisons. Continued efforts to elucidate the relationship between food intake patterns and weight status indicators in older adults, including longitudinal analyses and use of novel statistical approaches for food intake pattern identification, are warranted.

Validity of dietary patterns to assess nutrient intake adequacy

The purpose of the present study was to conduct a systematic review of the literature on the value of the methods used to assess dietary patterns for measuring nutrient intake adequacy in the population. Systematic review on Pubmed database up to April 2008. The search included specific key words and MeSH terms. No language limit was set. Only studies that compared food patterns with nutrient intake adequacy or nutrient biomarkers were included in the analysis. The search resulted in 1504 articles. The inclusion and exclusion criteria limited the selection to thirty articles. Nineteen studies evaluated the usefulness of the dietary patterns, eithera prioridefined (thirteen studies), or defined by factor analysis (four studies) or by cluster analysis (two studies), but only nine of them tested their validity (foura prioridefined and foura posterioridefined). Diet indices showed moderate to good validity results for measuring the adequacy of intakes for α-carotene, β-carotene, vitamin C, vitamin B6, Ca, folic acid, Fe and Mg. The factor analysis approach showed moderate to good validity correlations with the adequacy of intake of α-carotene, β-carotene, lutein, lycopene, vitamin C, vitamin B6and folic acid. Vitamin B12and vitamin E are the micronutrients with less probability of being adequately assessed with dietary patternsa prioriora posterioridefined. Diet indices are tools with fair to moderate validity to assess micronutrient intake adequacy.

Mothers of young children cluster into 4 groups based on psychographic food decision influencers

This study explored how mothers grouped into clusters according to multiple psychographic food decision influencers and how the clusters differed in nutrient intake and nutrient content of their household food supply. Mothers (n = 201) completed a survey assessing basic demographic characteristics, food shopping and meal preparation activities, self and spouse employment, exposure to formal food or nutrition education, education level and occupation, weight status, nutrition and food preparation knowledge and skill, family member health and nutrition status, food decision influencer constructs, and dietary intake. In addition, an in-home inventory of 100 participants' household food supplies was conducted. Four distinct clusters presented when 26 psychographic food choice influencers were evaluated. These clusters appear to be valid and robust classifications of mothers in that they discriminated well on the psychographic variables used to construct the clusters as well as numerous other variables not used in the cluster analysis. In addition, the clusters appear to transcend demographic variables that often segment audiences (eg, race, mother's age, socioeconomic status), thereby adding a new dimension to the way in which this audience can be characterized. Furthermore, psychographically defined clusters predicted dietary quality. This study demonstrates that mothers are not a homogenous group and need to have their unique characteristics taken into consideration when designing strategies to promote health. These results can help health practitioners better understand factors affecting food decisions and tailor interventions to better meet the needs of mothers.

Dietary patterns and quality in West-African immigrants in Madrid

BACKGROUND

Eating patterns of immigrants deserve to be better documented because they may reflect the extent of acculturation and associated health risks. The study assessed dietary patterns and quality in Bubi immigrants (from Equatorial Guinea) using cluster analysis and comparing different diet quality indexes.

METHODS

A random sample of 83 Bubi men and 130 women living in Madrid were studied. A 99-item food frequency questionnaire was administered, body weights and heights were self-reported and socio-demographic and health information was collected during interviews. Usual intakes were collapsed into 19 food groups. Cluster analysis of standardized food intakes per 1000 kcalories was performed. Dietary quality was appraised using the Alternative Mediterranean Diet Score, the Alternative Healthy Eating Index and scores of micronutrient adequacy and prevention based on WHO/FAO recommendations.

RESULTS

Two dietary patterns were identified. The 'Healthier' pattern, so confirmed by two dietary quality indexes, featured a higher consumption of fish, fruits, vegetables, legumes, dairy products and bread while the 'Western' pattern included more processed meat, animal fat, and sweetened foods and drinks. One third of the subjects were in the 'Healthier' food cluster, with the same proportion of men and women. Age >or= 30 and residence in Madrid >or= 11 years were independently associated with the healthier diet. Consumption of traditional foods was unrelated to dietary pattern, however. Overall, Bubi diets were somewhat protective because of high intakes of fruits and vegetables and monounsaturated fat (olive oil), but not with respect to sugar, cholesterol, omega-3 fatty acids and fibre. Less than two thirds of subjects had adequate intakes of iron, calcium and folate in both dietary phenotypes. Body mass index, physical exercise, and self-reported health and cardiovascular disease condition showed no significant association with the dietary pattern.

CONCLUSION

Cluster analysis combined with dietary quality assessment facilitates the interpretation of dietary patterns, but choosing the appropriate quality indexes is a problem. A small number of such indexes should be standardized and validated for international use. In the group studied, younger subjects and more recent immigrants were more likely to have a 'Western' pattern and should be a priority target for nutrition communication.

Dietary patterns and health and nutrition outcomes in men living with HIV infection

BACKGROUND

Nutritional status is an important determinant of HIV outcomes.

OBJECTIVE

We assessed the association between dietary patterns identified by cluster analysis and change in body mass index (BMI; in kg/m(2)), CD4 count, and viral load (VL).

DESIGN

HIV-positive adult male subjects (n = 348) with a BMI >or= 20.5 were evaluated by biochemical, body composition, and dietary data. Cluster analysis was performed on 41 designated food groups derived from 3-d food records. Dietary clusters were compared for sociodemographic, nutrient intake, and clinical outcomes. Multivariate linear regression assessed associations between dietary clusters and change in BMI, CD4 count, and VL.

RESULTS

We observed 3 dietary patterns: juice and soda; fast food and fruit drinks; and fruit, vegetable, and low-fat dairy. Subjects in the fast food and fruit drinks pattern had the lowest fiber intake, highest VL, and lowest CD4 count and had a lower income than did subjects in the other 2 clusters. Subjects in the fruit, vegetable, and low-fat dairy diet pattern had higher intakes of protein, fiber, and micronutrients and the highest BMI and CD4 count. Subjects in the juice and soda pattern had higher energy intakes and lowest BMI. On average, the fast food and fruit drinks cluster and fruit, vegetable, and low-fat dairy cluster gained 0.33 (P = 0.06) and 0.42 (P = 0.02), respectively, more in BMI than the juice and soda cluster across the study interval in a multivariate model.

CONCLUSIONS

In a cohort of HIV-positive men, we identified 3 distinct dietary patterns; each pattern was associated with specific nutrition, demographic, and HIV-related variables.

Food patterns and cardiovascular disease risk factors: the Swedish INTERGENE research program

BACKGROUND

Analyzing the impact of the intake of many foods simultaneously provides additional knowledge about analyses of nutrients and might make it easier to implement recommendations for the public.

OBJECTIVE

The objective was to examine food patterns in a Swedish population and determine how they are related to metabolic risk factors for cardiovascular disease.

DESIGN

The study is based on data from the INTERGENE population study of women and men aged 25-74 y in western Sweden. Dietary patterns were identified with cluster analysis of 93 food frequencies reported by 3452 participants. Associations with features of the metabolic syndrome, including blood lipids, blood pressure, and anthropometric measures, were analyzed.

RESULTS

Five distinct food patterns were identified, of which one was interpreted as a "healthy" reference pattern. This healthy cluster was distinguished by more frequent consumption of high-fiber and low-fat foods and lower consumption of products rich in fat and sugar. The 4 other clusters differed significantly from the reference cluster with respect to prevalence of cardiovascular disease risk factors and the metabolic syndrome. For example, body mass index and waist-to-hip ratio were significantly higher in a cluster characterized by high consumption of energy-dense drinks and white bread and low consumption of fruit and vegetables (P < 0.0001 and P = 0.004, respectively).

CONCLUSIONS

It is possible to distinguish food patterns that are related to obesity and obesity-related cardiovascular disease risk factors in contrast with a more healthy pattern conforming with current dietary guidelines. Thus, the results indicate no reason for questioning the current recommendations.

Comparison of cluster and principal component analysis techniques to derive dietary patterns in Irish adults

The aims of the present study were to examine and compare dietary patterns in adults using cluster and factor analyses and to examine the format of the dietary variables on the pattern solutions (i.e. expressed as grams/day (g/d) of each food group or as the percentage contribution to total energy intake). Food intake data were derived from the North/South Ireland Food Consumption Survey 1997-9, which was a randomised cross-sectional study of 7 d recorded food and nutrient intakes of a representative sample of 1379 Irish adults aged 18-64 years. Cluster analysis was performed using the k-means algorithm and principal component analysis (PCA) was used to extract dietary factors. Food data were reduced to thirty-three food groups. For cluster analysis, the most suitable format of the food-group variable was found to be the percentage contribution to energy intake, which produced six clusters: 'Traditional Irish'; 'Continental'; 'Unhealthy foods'; 'Light-meal foods & low-fat milk'; 'Healthy foods'; 'Wholemeal bread & desserts'. For PCA, food groups in the format of g/d were found to be the most suitable format, and this revealed four dietary patterns: 'Unhealthy foods & high alcohol'; 'Traditional Irish'; 'Healthy foods'; 'Sweet convenience foods & low alcohol'. In summary, cluster and PCA identified similar dietary patterns when presented with the same dataset. However, the two dietary pattern methods required a different format of the food-group variable, and the most appropriate format of the input variable should be considered in future studies.

A diet high in fruits and low in meats reduces the risk of colorectal adenomas

Recent evidence suggests overall dietary patterns, rather than specific dietary components, may be a better predictor of colorectal adenomas or cancers. Using cluster analysis, we aimed to assess the association between dietary patterns and colorectal adenomas and whether adjusting for total energy consumption prior to creating clusters affects this relation. Data from a case-control study of 725 individuals undergoing a colonoscopy were utilized. Cases (n = 203) had > or =1 adenoma on colonoscopy, and controls (n = 522) were those who had no adenomas. Dietary data were obtained from an FFQ. Daily intake for 18 different food groups was calculated. The values were transformed into Z-scores. Participants were first clustered without energy adjustment, then again based on their consumption per 1000 kcal (4187 kJ). There was no association between dietary patterns and colorectal adenomas without energy adjustment prior to creating dietary clusters, as clusters formed as a by-product of energy consumption. After adjusting for energy consumption, 3 distinct clusters emerged: 1) high fruit-low meat cluster; 2) high vegetable-moderate meat cluster; and 3) high meat cluster. After adjusting for potential confounders, the high vegetable-moderate meat cluster (odds ratio [OR] 2.17: [95% CI] 1.20-3.90) and high meat cluster (OR 1.70: [95% CI] 1.04-2.80) were at significantly increased odds of having had an adenoma compared with the high fruit-low meat cluster. A high-fruit, low-meat diet appears to be protective against colorectal adenomas compared with a dietary pattern of increased vegetable and meat consumption.

A dietary screening questionnaire identifies dietary patterns in older adults

Dietary patterns reflect habitual exposure of foods and nutrients and are a preferred means to assess diet and disease relations. Our objective was to design a screening tool to assess diet quality and dietary patterns among older adults and to relate the patterns to markers of general health and nutrition status. We used a population-specific data-based approach to design a diet screening instrument that was tested among subjects sampled from the Geisinger Rural Aging Study cohort (n=205). All participants attended a local clinic and had biochemical, anthropometric, and other health data collected. Dietary information was obtained via 24-h recall. We used principle components analysis to derive dietary patterns, which were then compared with nutritional outcomes using Pearson partial correlations, controlling for energy, age, BMI, and supplement use. Two dietary patterns were derived; 1 represented by more healthful foods and 1 by less optimal food choices. The healthy pattern was associated with more favorable biomarkers, more nutrient-dense diets, and lower waist circumference, whereas the converse was true for the second pattern. A screening tool can be used by older adults to identify dietary patterns that may relate to nutritional risk.