Some difficulties inherent in the interpretation of dietary data from free-living populations

Dietary patterns linked to lower odds of malnutrition are associated with all-cause and cancer mortality in adults with cancer

OBJECTIVES

Dietary patterns, characterised by protein, polyunsaturated fatty acids, and vitamin D, reduce the odds of malnutrition in cancer survivors. However, it is unclear whether these dietary patterns also improve prognosis. This study prospectively examined associations between dietary patterns linked to lower odds of malnutrition and the risk of all-cause and cancer mortality in adult cancer survivors from the UK Biobank cohort.

DESIGN

Prospective observational study.

SETTING AND PARTICIPANTS

Cancer survivors from the UK Biobank (mean ± SD, 7.1 ± 6.3 years since diagnosis) were included (n = 2415; 59.7 ± 7.1 years; 60.7% female).

MEASUREMENTS

Dietary intake was estimated using the Oxford WebQ 24-h dietary assessment. Dietary patterns ('high oily fish and nuts', and 'low oily fish') were derived using reduced rank regression (response variables: protein (g/kg/day), polyunsaturated fatty acids (g/day) and vitamin D (μg/day)). Cox proportional hazard models estimated hazard ratios (HR) and 95% confidence intervals (CI) for all-cause and cancer mortality. Nonlinear relationships were examined using restricted cubic splines. Models were adjusted for demographic and health characteristics. Sub-group analyses investigated relationships in sub-samples of adults with i) high nutritional risk (lung, gastrointestinal, haematological, or head and neck tumours) and ii) recent cancer diagnosis (cancer diagnosis within two years prior to assessment).

RESULTS

Deaths due to all-causes (n = 305) and cancer (n = 249) were identified during a median 10.4 (IQR: 10.2-10.8) years follow-up. There were no statistically significant linear associations between the dietary patterns and all-cause or cancer mortality. However, a U-shaped association between the 'high oily fish and nuts' pattern, characterised by higher intake of oily fish and nuts and seeds, and all-cause mortality (p-non-linearity = 0.004) was identified, as well as with all-cause (p-non-linearity = 0.006) and cancer mortality (p-non-linearity = 0.035) in adults with a high nutritional risk cancer diagnosis (lung, gastrointestinal, haematological, or head and neck tumours), indicating that both above and below mean intake was associated with increased risk. The 'low oily fish' pattern, characterised by lower oily fish but higher potato intake, also had a non-linear association with all-cause mortality (p-non-linearity = 0.046) where lower but not higher than mean intake increased mortality risk. No dietary patterns were significantly associated with mortality in adults with a recent cancer diagnosis.

CONCLUSION

'High oily fish and nuts' or 'low oily fish' dietary patterns that were protective against malnutrition were associated with risk of all-cause and cancer mortality in adults with cancer. Future research should assess the efficacy of these dietary patterns in the acute treatment period when malnutrition is most prevalent.

Fiber Patterns in Young Adults Living in Different Environments (USA, Spain, and Tunisia). Anthropometric and Lifestyle Characteristics

Benefits of dietary fiber go beyond its effect on chronic diseases associated with development. Consequently, the pattern of fiber intake has been considered an indicator for diet quality. Young adults are especially vulnerable to a food environment that drives an increase in chronic diseases linked to economic development. The aim of this work was to characterize patterns of fiber intake among university students. A cross-sectional study was conducted on a sample of 730 students enrolled at the University of Castilla-La Mancha (Spain), the University of Carthage (Tunisia), and Florida International University (USA). Mean age was 21.2. Food consumption was self-reported in two 24-h recalls. Mean dietary fiber intake was 17.8 g, not reaching the adequate intake. Contrary to expectations, American participants were the highest consumers (p < 0.001), and also exhibited the highest BMI. Cereals, legumes, vegetables and fruit were the main food sources of fiber. Fiber from appetizers, prepared and precooked meals, sauces, spices and condiments accounted for 16.7% in American participants, 7.4% in Spanish participants and 2.6% in Tunisian participants. Total fiber intake increased with energy intake but did not depend on smoking habits and physical activity in any country. It is essential to improve consumers’ interpretation of guidelines on fiber intake.

Evaluation of the reliability and validity of the Brazilian Healthy Eating Index Revised

OBJECTIVE

Evaluate validity and reliability of the Brazilian Healthy Eating Index Revised according to the psychometric properties.

METHODS

Cross-sectional study of a random sample of 2,375 individuals of both sexes, aged 12 or older from the city of São Paulo, Southeastern Brazil, drawn from the Health Survey carried out in 2003. Information on the population characteristics was acquired using a questionnaire. Food intake was obtained using 24h Recall, used to calculate the Brazilian Healthy Eating Index Revised (BHEI-R). The following aspects were evaluated: (1) content validity, by comparing the components with the Dietary Guidelines for the Brazilian Population; (2) construct validity, factor analysis using principal components method and verifying whether the index can measure diet quality regardless of energy intake; (3) discriminating validity; and (4) the reliability of the BHEI-R by analyzing the internal consistency of the items using Cronbach's alpha coefficient.

RESULTS

The correlations between the component scores and energy intake were weak (r < 0.30). Principal component analysis indicated the presence of four factors with eigenvalues > 1 that represented 67% of the index variance. The discriminating validity of the BHEI-R was observed by comparing the individuals by sex and smoking habit, and identifying statistically significant differences between the means of the components of the BHEI-R and the final score. The Cronbach's alpha value (α = 0.7) indicated the presence of internal consistency between the components of the BHEI-R. The SoFAAS component followed by the total fruit component and whole fruit component presented greater correlation with the final index scores.

CONCLUSIONS

The Brazilian Healthy Eating Index Revised showed itself to be reliable and structurally valid when used to evaluate and monitor the diet quality of Brazilians.

Diet and lifestyle factors and risk of subtypes of esophageal and gastric cancers: classification tree analysis

PURPOSE

Although risk factors for squamous cell carcinoma of the esophagus and adenocarcinomas of the esophagus (EA), gastric cardia (GC), and other (noncardia) gastric (OG) sites have been identified, little is known about interactions among risk factors. We sought to examine interactions of diet, other lifestyle, and medical factors with risks of subtypes of esophageal and gastric cancers.

METHODS

We used classification tree analysis to analyze data from a population-based case-control study (1095 cases, 687 controls) conducted in Connecticut, New Jersey, and western Washington State.

RESULTS

Frequency of reported gastroesophageal reflux disease symptoms was the most important risk stratification factor for EA, GC, and OG, with dietary factors (EA, OG), smoking (EA, GC), wine intake (GC, OG), age (OG), and income (OG) appearing to modify the risk of these cancer sites. For esophageal squamous cell carcinoma, smoking was the most important risk stratification factor, with gastroesophageal reflux disease, income, race, noncitrus fruit, and energy intakes further modifying risk.

CONCLUSION

Various combinations of risk factors appear to interact to affect risk of each cancer subtype. Replication of these data mining analyses are required before suggesting causal pathways; however, the classification tree results are useful in partitioning risk and mapping multilevel interactions among risk variables.

Principal component analysis of dietary and lifestyle patterns in relation to risk of subtypes of esophageal and gastric cancer

PURPOSE

To carry out pattern analyses of dietary and lifestyle factors in relation to risk of esophageal and gastric cancers.

METHODS

We evaluated risk factors for esophageal adenocarcinoma (EA), esophageal squamous cell carcinoma (ESCC), gastric cardia adenocarcinoma (GCA), and other gastric cancers (OGA) using data from a population-based case-control study conducted in Connecticut, New Jersey, and western Washington state. Dietary/lifestyle patterns were created using principal component analysis (PCA). Impact of the resultant scores on cancer risk was estimated through logistic regression.

RESULTS

PCA identified six patterns: meat/nitrite, fruit/vegetable, smoking/alcohol, legume/meat alternate, GERD/BMI, and fish/vitamin C. Risk of each cancer under study increased with rising meat/nitrite score. Risk of EA increased with increasing GERD/BMI score, and risk of ESCC rose with increasing smoking/alcohol score and decreasing gastroesophageal reflux disease (GERD)/body mass index (BMI) score. Fruit/vegetable scores were inversely associated with EA, ESCC, and GCA.

CONCLUSIONS

PCA may provide a useful approach for summarizing extensive dietary/lifestyle data into fewer interpretable combinations that discriminate between cancer cases and controls. The analyses suggest that meat/nitrite intake is associated with elevated risk of each cancer under study, whereas fruit/vegetable intake reduces risk of EA, ESCC, and GCA. GERD/obesity were confirmed as risk factors for EA and smoking/alcohol as risk factors for ESCC.

Dietary patterns and survival of older Europeans: The EPIC-Elderly Study (European Prospective Investigation into Cancer and Nutrition)

Objective

To investigate the association of a posteriori dietary patterns with overall survival of older Europeans.

Design and setting

This is a multi-centre cohort study. Cox regression analysis was used to investigate the association of the prevailing, a posteriori-derived, plant-based dietary pattern with all-cause mortality in a population of subjects who were 60 years or older at recruitment to the European Prospective Investigation into Cancer and Nutrition (EPIC-Elderly cohort). Analyses controlled for all known potential risk factors.

Subjects

In total, 74 607 men and women, 60 years or older at enrolment and without previous coronary heart disease, stroke or cancer, with complete information about dietary intakes and potentially confounding variables, and with known survival status as of December 2003, were included in the analysis.

Results

An increase in the score which measures the adherence to the plant-based diet was associated with a lower overall mortality, a one standard deviation increment corresponding to a statistically significant reduction of 14% (95% confidence interval 5–23%). In country-specific analyses the apparent association was stronger in Greece, Spain, Denmark and The Netherlands, and absent in the UK and Germany.

Conclusions

Greater adherence to the plant-based diet that was defined a posteriori in this population of European elders is associated with lower all-cause mortality. This dietary score is moderately positively correlated with the Modified Mediterranean Diet Score that has been constructed a priori and was also shown to be beneficial for the survival of the same EPIC-Elderly cohort.

Identification of dietary patterns using factor analysis in an epidemiological study in São Paulo

CONTEXT AND OBJECTIVE

Diet and nutrition are environmental factors in health/disease relationships. From the epidemiological viewpoint, diet represents a complex set of highly correlated exposures. Our objective was to identify patterns of food intake in a group of individuals living in São Paulo, and to develop objective dietary measurements for epidemiological purposes. DESIGN AND LOCAL: Exploratory factor analysis of data in a case-control study in seven teaching hospitals in São Paulo.

METHODS

The participants were 517 patients (260 oral cancer cases and 257 controls) admitted to the study hospitals between November 1998 and March 2001. The weekly intake frequencies for dairy products, cereals, meat, processed meat, vegetables, pulses, fruits and sweets were assessed by means of a semi-quantitative food frequency questionnaire. Dietary patterns were identified by factor analysis, based on the intake of the eight food groups, using principal component analysis as an extraction method followed by varimax rotation.

RESULTS

Factor analysis identified three patterns that accounted for 55% of the total variability within the sample. The first pattern ("prudent") was characterized by vegetable, fruit and meat intake; the second ("traditional") by cereals (mainly rice) and pulses (mainly beans); and the third ("snacks") by dairy products and processed meat.

CONCLUSION

This study identified food intake patterns through an a posteriori approach. Such analysis may be useful for nutritional intervention programs and, after computing scores for each individual according to the patterns identified, for establishing a relationship between diet and other epidemiological measurements of interest.

Study design and hypothesis testing: issues in the evaluation of evidence from research in nutritional epidemiology

In addressing questions about the relations of dietary factors to disease in human populations, epidemiologic studies must account for the complexity of dietary habits, the intercorrelations among dietary habits, and the correlations of those habits with other behaviors. Furthermore, for studies of chronic disease, relevant dietary exposures may occur over decades. The classic epidemiologic study designs have been used to examine the associations between diet and disease; the strengths and weaknesses of those designs must be considered. Concerns have been raised regarding the validity of the measures of diet, the differential recall of diet by diseased individuals in case-control studies, and confounding by other related factors in both case-control and cohort studies. In clinical trials there may be difficulties in effecting the necessary dietary changes, especially for macronutrients, and there are also concerns about those circumstances in which participants cannot be blinded to their treatment. For case-control and cohort studies and for some clinical trials, intercorrelations among nutrients are a concern in the identification of factors that are important in the etiology of disease. It is important to understand these considerations when interpreting nutritional epidemiologic studies for the purpose of setting public policy. No one study can be considered definitive in the understanding of a diet-disease relation. However, epidemiologic findings from multiple studies taken together can contribute significantly to our understanding of diet in relation to disease in humans.

Baseline dietary intake and physical activity of Japanese American men in relation to glucose tolerance at 5-year follow-up

Japanese American men (n = 124), with normal glucose tolerance (NGT, n = 69) or impaired glucose tolerance (IGT, n = 55) at baseline, were studied for effects of baseline dietary intake and physical activity on glucose tolerance at baseline and at 5-year follow-up. At baseline, both NGT and IGT men with positive family history of diabetes (FH) showed high intakes of animal fat and protein, but only the NGT men countered this with high levels of energy expenditure. In the total sample at 5-year follow-up, 2-hour plasma glucose was significantly related to intake of animal fat (AF), partial correlation r = 0.32, P < 0.001, adjusted for total energy intake, age, self-reported health, body mass index, FH, and baseline glucose tolerance category. Energy expenditure (EE) was not related to 5-year 2-hour plasma glucose in the total sample, but displayed a relationship with 5-year 2-hour plasma glucose in those IGT (r = -0.27, P < 0.05), but not in those NGT at baseline, and in those with positive FH (r = -0.33, P < 0.05), but not in those with negative FH. Additionally, AF showed a relationship to 5-year 2-hour plasma glucose only for those in the lowest (r = 0.37, P < 0.05) and middle (r = 0.33, P < 0.05) tertiles, but not in the highest tertile of EE. For baseline IGT men, 5-year 2-hour plasma glucose was related to "high vs. low risk" categories of AF intake and EE, but only in men with a positive FH (AF ≥ 25 vs. < 25 g/day: 180.1 ± 38.6 vs. 143.6 ± 39.7 mg/dl, P = 0.048; EE ≤ 2,000 kcal/week vs. > 2,000 kcal/week, 189.9 ± 39.2 vs. 150.8 ± 37.4 mg/dl, P = 0.028; with risk categories combined, i.e., both high, mixed, both low: 192.0 ± 41.3, 165.4 ± 28.4, 139.4 ± 40.9 mg/dl, P = 0.045, linear trend, P = 0.014). Thus, high AF intake and low EE may have long-range detrimental effects on glucose tolerance, especially for those with IGT and positive FH. © 1996 Wiley-Liss, Inc.

Consumption of energy-dense, nutrient-poor foods by the US population: effect on nutrient profiles

OBJECTIVE

To examine the association of consumption of foods from the fats, sweets, and the alcohol group ("other group) with nutrient profiles.

METHODS

Using data from the NHANES II survey of 1976-80, we categorized the foods reported to be consumed by adults (n = 11,528) into six groups: meat, dairy, grain, fruit, vegetable, and "other."

RESULTS

Nearly one-third of total daily energy intake was contributed by foods from the "other" category. As the proportion of daily energy intake from "other" foods increased, total daily energy intake also increased, as did the percent energy from carbohydrate and alcohol. However, percent energy from fat and protein, intake of all examined micronutrients (except vitamin E), nutrient density, and the proportion of the population meeting the RDA of various nutrients declined with increasing intake of "other" foods. Respondents were more likely to report no servings as well as less than the recommended servings of foods from the major food groups with increasing intake of "other" foods.

CONCLUSION

The data suggest that consumption of foods from the "other" group displaced nutrient-dense foods from the diets of NHANES II respondents.

Collinearity diagnosis for a relative risk regression analysis: an application to assessment of diet-cancer relationship in epidemiological studies

In epidemiologic studies, two forms of collinear relationships between the intake of major nutrients, high correlations, and the relative homogeneity of the diet, can yield unstable and not easily interpreted regression estimates for the effect of diet on disease risk. This paper presents tools for assessing the magnitude and source of the corresponding collinear relationships among the estimated coefficients for relative risk regression models. I show how to extend three tools (condition indices, variance decomposition proportions, and standard inflation factors) for diagnosing collinearity in standard regression models to likelihood and partial likelihood estimation for logistic and proportional hazards models. This extension is based on the analogue role of the information matrix in such analyses and the cross-product matrix in the standard linear model. I apply the methodology to relative risk models that relate crude intakes (on the log scale) and nutrient densities to breast cancer cases in the NHANES-I follow-up study. The three diagnostic tools provide complementary evidence of the existence of a strong collinearity in all models that is due largely to homogeneity of the population with respect to our risk scale for the crude intakes. The analysis suggests that the non-significant relative risks for the crude intakes in these models may be due to their involvement in collinear relationships, while the nonsignificant relative risks for the nutrient densities are far less affected by multicollinearity.

A proxy approach to the determination of total caloric intake for use in cancer epidemiology

In the absence of complete dietary information on a cohort of 472 early-stage breast cancer patients a caloric prediction technique was developed. Deriving an accurate caloric denominator would enable examination of individual nutrient effects by controlling for potential confounding by calories. Surrogate measures of mean daily caloric intake were generated from estimates of basal metabolic rate (BMR) predicted from age, height, weight, and physical activity. A validation study was undertaken to test the relationship between the BMR proxy terms and mean daily energy intake obtained from four-day food diaries for 51 subjects. Pearson correlation coefficients between the diary-derived and predicted values of caloric intake were computed (r = 0.43, p = 0.001). The results are in the range of what many researchers found when comparing self-reports of total caloric intake in validation and reliability studies of various dietary assessment tools. Because of the large heterogeneity of results across the ages represented in this study (25-77 yrs), it is recommended that small validation studies of this type be carried out in the subpopulations of interest.

An assessment of caloric intake as an indicator of physical activity

Caloric intake has been suggested as a surrogate measure of physical activity, provided that energy balance is maintained. In this study we assessed caloric intake in relation to other measures of physical activity in a free-living population by a single 24-hr recall in white adult males, ages 22-79 years, examined in 1957-1960 as part of the U.S. Railroad Study. Intake was positively related to leisure time and occupational activity. Men with higher levels of caloric intake had lower exercise test heart rates indicating that they were probably more fit. Caloric intake was also significantly and inversely related to body mass index, the sum of the triceps and subscapular skinfold thickness measurements, blood pressure, and total serum cholesterol. These results suggest that caloric intake, even as measured in a 24-hr dietary recall, is an indicator of physical activity level. Therefore, caloric intake may be a useful indicator of physical activity in some population-based epidemiologic studies.

Long-term (24-year) recall reliability in cancer cases and controls using a 21-item food frequency questionnaire

The following two questions concerning diet recall were addressed when studying 117 incident cancer cases and 99 controls from the Adventist Mortality Study. Are recalls of past dietary habits reliable? Does recall ability differ between cancer cases and controls? Two sets of dietary data were compared using the American Cancer Society's food frequency questionnaire--as reported in 1960 and recalled in 1984. Ability to recall 21 key food items was evaluated both for individual foods and a combination of all foods by comparing recall scores. The comparison revealed that among food groups, 24-year recall ability varied greatly. There was no significant difference in recall ability between cancer incident cases and controls after controlling for factors that may be related to recall ability (e.g., age, education, and sex). Also, there was no significant difference in recall ability among subjects with or without other chronic diseases likely to affect diet pattern. The results revealed no significant differences in recall ability by sex and body mass index; however, significant differences by vegetarian status and diet stability were found. Significant differences by educational level were found only in univariate analysis.

Calories, fat and cholesterol: intake patterns in the US population by race, sex and age

Nutrient intakes were investigated for Blacks and Whites using data from the NHANES II survey (1976-80). Intake of energy, total fat, saturated fat, dietary cholesterol, P/S ratio, and per cent of calories derived from total and saturated fat are examined by sex and age, both in absolute terms and per unit of body weight. For most age and sex categories, Blacks are found to have a lower intake of energy and fats than Whites; however, Blacks have a consistently higher intake of dietary cholesterol. The ratio of polyunsaturated fats to saturated fats is higher in females than in males, but all age-sex groups are substantially below recommended levels. Per cent of calories from total and saturated fat are similar in most age-sex groups. Possible explanations of the observed patterns include activity level and metabolic differences.

Cooking methods, carcinogens, and diet-cancer studies

A neglect of natural, preservative, and cooking-induced carcinogens or mutagens in food, along with a neglect of dietary patterns during the first portion of a person's lifetime, may be responsible for the many conflicting epidemiological reports dealing with dietary factors and cancer. From animal and occupational studies, we know that the two most important factors in the study of cancer are the dose of carcinogen and allowance for a long latent period. Most of the recent nutrition and cancer studies have ignored both factors. Some bile acids or other endogenous factors may be influenced by diet and may act as cancer-promoting agents, but promoting agents cannot be studied in the absence of knowledge about, or control of, the cancer-initiating events with which they must interact.

Metabolic effects of fructose as a natural sweetener in the physiologic meals of ambulatory obese patients with type II diabetes

The long-term effects of fructose as a natural sweetener in the physiologic meals of ambulatory obese patients with type II diabetes remain uncertain. An outpatient 12-week study was therefore conducted to evaluate the metabolic effects of crystalline fructose (60 g) supplementation of the diet of nine patients with type II diabetes (Group A, mean age 57 +/- 2 years, seven women and two men). Their results were compared with age-, sex-, and weight-matched patients with type II diabetes who were similarly studied but without fructose supplementation of their usual meals (Group B). The mean +/- SEM fasting serum glucose (224 +/- 24 versus 204 +/- 14 mg/dl) and glycosylated hemoglobin (11.57 +/- 0.49 versus 10.20 +/- 0.60 percent) values progressively decreased (p less than 0.05, week 12 versus week 0) in Group A. In contrast, both parameters increased in Group B when compared with the week 0 values, but differences were not statistically significant. Levels of fasting serum triglycerides, total cholesterol, high-density lipoprotein cholesterol, and low-density lipoprotein cholesterol remained unchanged at week 12 compared with week 0 in Group A. However, the mean apoprotein A1 concentrations increased significantly at weeks 4 and 12 in Group A, whereas only transient changes occurred in apoprotein B100 values. Levels of mean fasting serum triglycerides increased at weeks 4 (15 percent) and 12 (38 percent) in Group B; however, no significant changes occurred in the rest of the lipid, lipoprotein, and apoprotein A1 and B100 levels. Metabolic by-products of fructose such as serum lactic acid and uric acid levels remained essentially unchanged in both groups at the end of the study. In addition, no significant weight changes were observed in either group. The fructose supplementation was well tolerated without significant adverse effects. Thus, this study demonstrates that addition of moderate amounts of fructose as a natural sweetener in the physiologic mixed meal does not appear to have deleterious effects on glycemic control and lipid and lipoprotein metabolism in ambulatory obese patients with type II diabetes and poor metabolic control. Rather, a slight improvement in glycemic control and alterations in the apoprotein compositions in favor of decreased risk for coronary artery disease may occur.

Blood pressure and nutrient intake in the United States

A data base of the National Center for Health Statistics, Health and Nutrition Examination Survey I (HANES I), was used to perform a computer-assisted, comprehensive analysis of the relation of 17 nutrients to the blood pressure profile of adult Americans. Subjects were 10,372 individuals, 18 to 74 years of age, who denied a history of hypertension and intentional modification of their diet. Significant decreases in the consumption of calcium, potassium, vitamin A, and vitamin C were identified as the nutritional factors that distinguished hypertensive from normotensive subjects. Lower calcium intake was the most consistent factor in hypertensive individuals. Across the population, higher intakes of calcium, potassium, and sodium were associated with lower mean systolic blood pressure and lower absolute risk of hypertension. Increments of dietary calcium were also negatively correlated with body mass. Even though these correlations cannot be accepted as proof of causation, they have implications for future studies of the association of nutritional factors and dietary patterns with hypertension in America.