Dietary Patterns, Meat Intake, and the Risk of Type 2 Diabetes in Women FREE

The United States is experiencing an alarming increase in the incidence of type 2 diabetes mellitus.¹ The resulting morbidity, economic costs, reduced quality of life, and risk for complications make preventive strategies indispensable. Although obesity is the most important risk factor for type 2 diabetes,¹ evidence is emerging that certain foods and dietary factors may be associated with diabetes. In particular, a diet low in whole grains and high in glycemic load² and processed meats appears to increase the risk.³ Therefore, from a preventive perspective, it would be useful to examine the association of food combinations, or self-selected dietary patterns, with risk of type 2 diabetes. Factor analysis has been commonly used to identify dietary patterns in studies of cardiovascular disease⁴ and colon cancer.⁵ Most recently, the “Western” dietary pattern has been shown to increase the risk of diabetes in a cohort of men.⁶ Therefore, the purpose of the present analysis is to examine the association between major dietary patterns and risk of type 2 diabetes among women in the ongoing Nurses’ Health Study (NHS) cohort.

The NHS began in 1976, when 121 700 female nurses aged 30 to 55 years living in 11 US states responded to a questionnaire regarding medical, lifestyle, and other health-related information.⁷ Since 1976, questionnaires have been sent biennially to update this information. Follow-up was complete for greater than 95% of the potential person time up to 1994. In 1980, the participants completed a 61-item food frequency questionnaire (FFQ). In 1984, the FFQ was expanded to 116 items. Similar FFQs were sent to the women in 1986, 1990, and 1994. We used the 1984 FFQ as baseline for this study because the expanded number of items was critical in characterizing dietary patterns. The NHS is approved by the institutional review board of the Brigham and Women’s Hospital, Boston, Mass.

For the present analysis, women were included if they completed the 1984 FFQ with fewer than 70 missing items and a total caloric range (as calculated from the FFQ) between 500 and 3500 kcal/d. We excluded women with a history of cancer, cardiovascular disease, and diabetes. We thus included in this analysis 69 554 women with follow-up for up to 14 years, from 1984 to 1998.

Dietary intake information was collected by FFQs designed to assess average food intake over the previous year. A standard portion size was given for each food item. Cohort members were asked to choose from 9 possible frequency responses, ranging from “never” to “more than 6 times a day” for each food. Total energy intake was calculated by summing up energy intakes from all foods. For this analysis, we used information from the FFQs administered in 1984, 1986, 1990, and 1994. Foods from the FFQ were classified into 36 to 38 food groups based on nutrient profiles or culinary usage. This classification follows that of a study in men with a similar dietary assessment instrument.⁴ Foods that did not fit into any of the groups or that may represent distinctive dietary behaviors were left as individual categories (eg, pizza, tea, and beer). Vitamin and mineral supplements were not included in the definition of the patterns, but they were included as covariates in the analysis. Previous validation studies among members of the NHS cohort revealed good correlations between nutrients assessed by the FFQ and multiple weeks of food records completed over the previous year.⁸ For example, correlation coefficients between 1986 FFQ and diet records obtained in 1986 were 0.68 for saturated fat, 0.76 for vitamin C, and 0.73 for dietary cholesterol. The mean correlation coefficient between frequencies of intake of 55 foods from 2 FFQs administered 12 months apart was 0.57.⁹

Our end points included incident type 2 diabetes mellitus that occurred between the return of the 1984 questionnaire and June 1, 1998. When a participant reported a diagnosis of diabetes in the biennial questionnaires, we mailed them a supplementary questionnaire that assessed symptoms, diagnostic tests, and treatment to confirm the diagnosis. Diabetes was confirmed when the participant fulfilled 1 or more of the following criteria: (1) manifestation of classic symptoms (eg, excessive thirst, polyuria, weight loss, and hunger) plus an elevated fasting glucose level (>140 mg/dL [>7.8 mmol/L]), or elevated nonfasting level (>200 mg/dL [>11.1 mmol/L]); (2) asymptomatic but plasma glucose level was elevated on at least 2 different occasions (as defined herein) or abnormal oral glucose tolerance test result (>200 mg/dL 2 hours after glucose load); and (3) receiving any hypoglycemic treatment for diabetes. These criteria for diabetes classification were consistent with those of the National Diabetes Data Group during our follow-up period.¹⁰ A validation study has shown a high level of accuracy in self-reporting of diabetes.¹¹ Medical records were successfully obtained for 62 women among a random sample of 84 with diabetes confirmed by the supplemental questionnaire. Review of records by an endocrinologist blinded to the questionnaire information confirmed 61 (98%) of the 62 women. Deaths were reported by family members, the postal service, or through searches in the National Death Index.

Dietary patterns were generated by factor analysis (principal components) based on predefined food groups and using an orthogonal rotation procedure.¹² This results in uncorrelated factors, which are easier to interpret. We determined the number of factors to retain by eigenvalue (>1), Scree test, and factor interpretability. The factor score for each pattern was calculated by summing intakes of food groups weighted by their factor loadings,¹³ and each woman received a factor score for each identified pattern. Good reproducibility of the patterns generated by this method has been demonstrated in a parallel cohort of men.¹⁴ The correlations between the scores of the 2 major patterns (“prudent” and “Western”) generated from FFQ and diet records were 0.52 for the prudent pattern and 0.74 for the Western pattern. Factor analysis was conducted using SAS PROC FACTOR.¹⁵

We used Cox proportional hazard models to examine the associations between major dietary patterns and diabetes risk. To reduce random within-person variation and best represent long-term dietary intake, we calculated cumulative averages of dietary pattern scores from our repeated dietary measurements.⁴ For example, dietary intake in 1984 was used to predict diabetes occurrence from 1984 to 1986, and the average of 1984 and 1986 intake was used to predict risk from 1986 to 1990, and so on. The regression analyses were adjusted for age (<49 y, 50-54 y, 55-59 y, 60-64 y, and ≥65 y), family history of diabetes (yes vs no), history of hypercholesterolemia (yes vs no), smoking (never, past, current with 1-14 cigarettes per day, current with 15-24 cigarettes per day, current with ≥25 cigarettes per day, and missing), hormone therapy use (premenopausal, never, past, and current hormone use), caloric intake, history of hypertension (yes vs no), physical activity (<1 h/wk, 1-1.9 h/wk, 2-3.9 h/wk, 4-6.9 h/wk, and ≥7 h/wk), alcohol intake (abstainer, 0.1-5.0 g/d, 5.1-15.0 g/d, 15.1-30.0 g/d, and >30g/d), body mass index (BMI; continuous and quadratic terms), and missing FFQ. The proportions with a missing FFQ in 1986, 1990, and 1994 were 17%, 16%, and 14%, respectively. In separate analyses, we only included symptomatic cases. In addition, we examined the association between different forms of red meat and diabetes risk, since these are the major contributors to the Western dietary pattern. We also conducted analyses jointly classifying women by processed meats and family history of diabetes, obesity status (BMI [calculated as weight in kilograms divided by the square of height in meters] ≥30), physical activity, alcohol intake, and smoking status. To minimize confounding by adiposity, we additionally adjusted for waist-hip ratio among women with the available data.

During 14 years of follow-up, we documented 2699 incident cases of type 2 diabetes mellitus, including 1604 symptomatic cases. Two major dietary patterns emerged from factor analysis. The prudent pattern was characterized by higher intake of fruits, vegetables, whole grains, fish, poultry, and low-fat dairy products, while the Western pattern was characterized by higher intakes of red and processed meats, refined grains, sweets and desserts, and high-fat dairy products. Women who scored high on the Western pattern tended to smoke more and consumed less folate, fiber, and protein (Table 1).

After multivariate adjustment, we observed a modest inverse association between the prudent pattern and type 2 diabetes (Table 2). Limiting the analysis to symptomatic cases only, a significant inverse association was observed (P for trend, .03). Women in the highest quintile of the prudent pattern score had a relative risk (RR) of 0.80 (95% confidence interval [CI], 0.67-0.95) when compared with the lowest quintile. In contrast, we observed a significant positive association with the Western pattern. The RR after multivariate adjustment was 1.49 (95% CI, 1.26-1.76; P for trend, <.001) when we compared the women in the top with the bottom quintile. When we stratified the analysis by BMI, family history of diabetes, and physical activity level, a positive association with the Western pattern score was present within all strata (Figure). Of interest is that the inverse association observed with alcohol was essentially negated by a high Western pattern score. Although the risk of diabetes was higher among individuals with the risk factor and higher Western pattern score, significant interactions between both risk factors were not observed.

To explore whether the association of diabetes with the Western pattern is independent of the latter’s major contributors, red and processed meats, we additionally adjusted for total red and processed meats. The resulting association between Western pattern score and diabetes risk was somewhat attenuated (RR, 1.28 [95% CI, 1.05-1.57] when we compared the top with the bottom quintile), but the test for trend remained statistically significant (P<.001). In secondary analysis, we also restricted our analysis to those with waist and hip circumference data and additionally adjusted for waist-hip ratio. A significant positive association was observed in each quintile, although the RR at the fifth quintile was somewhat attenuated. Significant trends remained for total processed meats and bacon.

When we examined major contributors to the Western pattern, positive associations with red and processed meats emerged (Table 3). Intake of total processed meats showed the strongest positive association, with an RR of 1.60 (95% CI, 1.39-1.83; P for trend, <.001) when we compared the top with bottom quintile of intake. When these meat products were analyzed as a continuous variable, we observed an RR of 1.26 (95% CI, 1.21-1.42) for each serving increase in red meat consumption, 1.38 (95% CI, 1.23-1.56) for total processed meats, 1.73 (95% CI, 1.39-2.16) for bacon, 1.49 (95% CI, 1.04-2.11) for hot dogs, and 1.43 (95% CI, 1.22-1.69) for processed meats. This suggests that processed meats intake confers a higher risk for diabetes than red meats. The associations between various meat intake and diabetes were slightly attenuated but generally remained statistically significant even after additionally adjusting for Western pattern score, suggesting that these foods are associated with diabetes risk independently of the overall Western pattern. In addition, when the analysis was restricted to those with waist-hip ratio information and controlled for this index of adiposity, the association did not change substantially.

We observed positive associations between the Western pattern, intake of several meat products, and risk of type 2 diabetes mellitus in women. Although red and processed meats are major contributors to the Western pattern, the association with the Western pattern was not fully explained by the various meats. This indicates that the remaining characteristics of the Western pattern, such as high intake of other highly processed foods, may contribute to an increased risk of diabetes. The increased risk from the different types of processed meat was similar but in general higher compared with red meat.

Few studies have prospectively investigated diet patterns or meat intake. Our findings confirm those of a cohort of men followed for up to 12 years.^3,6 A cross-sectional study of Seventh-Day Adventists in California showed a lower risk of diabetes among vegetarians, who consumed more legumes, fruits, and nuts in the absence of meat intake.¹⁶ The relationship observed with the Western pattern may be mediated through multiple mechanisms, including saturated fat, glycemic load, nitrites, and heme iron. In a group of US male health professionals, the Western pattern was correlated with fasting insulin levels.¹⁷ In a Swedish population, dietary patterns that resembled the Western pattern showed associations with hyperglycemia or hyperinsulinemia.¹⁸ One major characteristic of the Western pattern is the processed meat and most red meat intake. A higher red and processed meat intake was found to be associated with a higher risk of diabetes in a cohort of men.³ Processed meats were also associated with risk of diabetes in an earlier follow-up of this cohort.¹⁹ Nitrites are used in processed meats as a preservative, and it can be converted into nitrosamines in the gastrointestinal tract. Ecological studies suggest a relationship between nitrates and nitrite consumption with type 1 diabetes in children.^20- 21 Nitrites and some nitrous compounds have been shown to reduce insulin secretion in rats.²² However, whether they have a role in type 2 diabetes is unclear. Another possible mediator is advanced glycation end products. These are easily formed in meat and high fat products through heating and processing.²³ Advanced glycation end products has been shown to increase oxidative stress and the levels of tumor necrosis factor α in vitro,²⁴ and inflammation is believed to have a role in diabetes development.²⁵ In mice, administration of an advanced glycation end product formation inhibitor reduced the development of diabetes,²⁶ and restriction of advanced glycation end products in the diet has improved insulin sensitivity.²⁷

The prospective design of this study renders recall bias unlikely. This cohort has demonstrated accurate report of disease and because of the study participants’ access to health care, underreporting of diabetes is expected to be less than that in the general population. We used repeated measurements of various potential confounders and statistically controlled for BMI, a strong risk factor for diabetes. Dietary patterns across time have shown to be consistent in our cohort, and our use of cumulative averages of dietary pattern scores reduce the influence of random error. The principal components method of pattern analysis identifies existing patterns; therefore, the Western pattern observed in our study does not necessarily represent food choices that would pose the highest diabetes risk, nor does the prudent pattern represent the lowest risk. Although few studies have used this method to examine the association between diet and diabetes risk, dietary patterns generated with this method have been shown to be associated with other diseases.^4,28 An advantage to using dietary patterns is the potential to detect the combined effect of foods, especially if the individual components of a pattern contribute to only a small amount of risk.

We found that a diet high in red and processed meats, refined grains, and other characteristics of the Western pattern was associated with an elevated risk of type 2 diabetes mellitus in women. Red and processed meats were also independently associated with an increased risk. Therefore, it may be prudent to reduce the consumption of these food items to decrease the risk of type 2 diabetes.

Accepted for Publication: December 17, 2003.

Correspondence: Teresa T. Fung, ScD, Department of Nutrition, Simmons College, 300 The Fenway, Boston, MA 02115 (fung@simmons.edu).

Financial Disclosure: None.

Funding/Support: This study was supported by grants CA87969 from the National Institutes of Health, Bethesda, Md, and 9911DYS from the American Diabetes Association, Alexandria, Va.