Association between glycated hemoglobin and diet and other lifestyle factors in a nondiabetic population: cross-sectional evaluation of data from the Potsdam

¹ From the German Institute of Human Nutrition, the Department of Epidemiology, Potsdam-Rehbrücke, Germany; Unilever Research Vlaardingen, Vlaardingen, Netherlands; and the Institute of Clinical Chemistry, the Department of Medicine, Ernst-Moritz-Arndt University, Greifswald, Germany.

² Supported by grant 97 SOC 200302 05FO2 from the European Union and by contract 01 EA 9401 from the Federal Ministry of Education, Research and Technology, Germany.

³ Address reprint requests to H Boeing, German Institute of Human Nutrition, Department of Epidemiology, Arthur-Scheunert-Allee 114-116, D-14558 Bergholz-Rehbrücke, Germany. E-mail: boeing{at}www.dife.de.

ABSTRACT  
Background: Glycation reactions of proteins and other compounds, depending on blood glucose concentrations, have a detrimental effect on health.

Objective: The association of diet and other lifestyle factors with glycated hemoglobin (Hb A_1c) values was examined in a nondiabetic population.

Design: This was a cross-sectional study of 1773 middle-aged men and women. Mean Hb A_1c values were calculated for categories of diet and lifestyle factors, and odds ratios (ORs) for the highest versus lowest tertiles of Hb A_1c were determined and compared.

Results: The OR of being in the highest Hb A_1c tertile compared with the lowest increased with greater age [age 40–44 y compared with >60 y: men (OR: 2.86; 95% CI: 1.60, 5.20) and women: (6.11; 3.15, 12.30)] and greater obesity [body mass index (in kg/m²) >25 and waist-hip ratio >1.0 in men and >0.8 in women): men (2.80; 1.48, 5.45) and women (1.73; 1.15, 2.61)]. High energy and energy-adjusted saturated fat intakes were associated with increased risk of being in the highest tertile of Hb A_1c [highest compared with lowest quintile: (1.53; 1.04, 2.26; P for trend = 0.013) and (1.98; 1.33, 2.95; P for trend = 0.003), respectively]. No significant associations were observed for intakes of carbohydrates, protein, dietary fiber, or ß-carotene; however, some of the associations were nearly significant. Alcohol, vitamin C, and vitamin E intakes were inversely related to risk [highest compared with lowest quintile: (0.56; 0.38, 0.83; P for trend = 0.001), (0.50; 0.33, 0.74; P for trend = 0.003), and (0.65; 0.43, 0.96; P for trend = 0.036), respectively].

Conclusion: Hb A_1c values might be modifiable by diet and other lifestyle factors.

Key Words: Hb A_1c • glycated hemoglobin • diet • type 2 diabetes • risk factor • vitamin C • saturated fat • alcohol • European Prospective Investigation into Cancer and Nutrition Study • Potsdam • EPIC Study

INTRODUCTION  
There is substantial interest in blood glucose concentrations because glucose reacts, depending on blood glucose concentrations (1, 2), with amino groups of plasma and tissue proteins (Amadori reaction) to form glycated proteins. These glycated proteins gradually transform nonenzymatically into advanced glycation end products and have been reported to result in altered protein function of the affected molecules. Glycation of LDL, for example, was found to be associated with impaired receptor-mediated uptake and catabolism (3), and glycation reactions may cause oxidative stress through free radical generation (4, 5). Both diabetic micro- and macrovascular complications and increased atherosclerotic risk (6) were reported to be associated with advanced glycation end products (1, 6).

Furthermore, glucose concentrations play an important role in the metabolic syndrome. High serum glucose concentrations indicate the beginning of or existing glucose intolerance and insulin resistance, which may result in type 2 diabetes. The preclinical development of type 2 diabetes, however, is poorly understood and so far there is little direct evidence that the same factors influencing metabolic control in clinical diabetes might also affect the preclinical development of the disease. An increased risk of type 2 diabetes has been shown to be associated with several dietary risk factors. High saturated fat intakes have been associated with an increased risk of type 2 diabetes in various populations (7), and diets high in complex carbohydrates have been shown to protect against glucose intolerance and type 2 diabetes, mainly because of their high fiber content (8). A prospective study in Finland provided evidence to support the relation between serum vitamin E concentrations and the incidence of type 2 diabetes (9).

Research on blood glucose concentrations was facilitated by the identification of glycated hemoglobin (Hb A_1c) as a biomarker of long-term glucose homeostasis that reflects blood glucose concentrations over the previous 6–8 wk (10, 11). In epidemiologic studies, this biomarker has the advantage that a single assessment of Hb A_1c is suitable to classify individuals according to their long-term blood glucose concentrations (12).

Considering the importance of blood glucose concentrations on health, the relation of lifestyle factors to Hb A_1c values in plasma was studied in a sample of nondiabetic middle-aged men and women taking part in a prospective study on diet and chronic disease—the European Prospective Investigation into Cancer and Nutrition (EPIC) Study (13).

SUBJECTS AND METHODS  
Study design
A random sample from the Potsdam, Germany, cohort of the EPIC Study (14) was taken for cross-sectional analyses of associations between long-term blood glucose concentrations, as determined on the basis of Hb A_1c values, and diet and other lifestyle factors. The cohort population in Potsdam, Germany, was recruited from a random sample drawn via the population registries of the city of Potsdam and adjacent communities; men aged 40–64 y and women aged 35–64 y were eligible for inclusion. The basic examination for the cohort study included questionnaire and interview data, medical information, and a blood sample. In 1994–1995, the participation rate was 35% of those invited to participate in this long-term cohort study on diet and chronic diseases.

Study population
Hb A_1c was measured in 2 periods during ongoing recruitment: between November 1994 and March 1995 and between February and November 1996. During these periods, 2749 and 5690 subjects, respectively, were examined for the cohort study. From those who provided blood samples (96%), random samples for Hb A_1c measurement were drawn retrospectively after exclusion of subjects who reported a previous diagnosis of diabetes mellitus or were taking antidiabetic medication. Overall, 1186 subjects from the first study period and 1000 subjects from the second study period were selected for determination of Hb A_1c values in blood.

Subjects taking ß-blockers, diuretics, or corticoids that might affect Hb A_1c values were excluded from the statistical analysis (n = 383). Another 8 subjects were excluded because values for at least one of the main exposure variables were missing. In addition, 22 subjects with an Hb A_1c value 0.07 were also excluded because they may have already had undiagnosed type 2 diabetes. Thus, the final study population consisted of 1773 subjects—745 men and 1028 women. The procedures of the EPIC-Potsdam Study were approved by the ethical committee of the state of Brandenburg and by the elected official for data protection of this state.

Collection and preparation of biological samples
Blood samples were drawn in a standardized manner by use of Monovette tubes containing citrate as anticoagulant (Sarstedt, Nuembrecht, Germany). Monovette tubes were cooled immediately at 6°C and centrifuged after 90–180 min (1500 x g, 20 min, 20°C). The erythrocyte fraction was portioned into cryotubes (Nunc, Wiesbaden, Germany) and stored at -79°C until analyzed. Hb A_1c was measured in July 1995 and September 1996 by the same laboratory.

Glycated hemoglobin
Hb A_1c was measured with use of the Dako Hb A_1c test (DAKO Diagnostics, Cambridgeshire, United Kingdom), which uses a monoclonal antibody to directly detect the structural change resulting from glycation. This monoclonal antibody is specific for Hb A_1c and does not cross-react with other glycated products or hemoglobin variants lacking the N-terminal sequence of the ß-chain. The bound conjugate is subsequently detected by the peroxidase reaction. In test series of 8 samples analyzed 10 times, a CV of 2.95% was found for the Hb A_1c analysis.

Dietary data
Dietary intake was measured by use of a self-administered food-frequency questionnaire (FFQ) suitable for optical scanning. This FFQ asked for the average frequency and portion size of 146 food items eaten during the 12 mo before examination. The resulting food consumption profile was converted into nutrient intakes on the basis of a German food-composition table (15). Validity and reliability of the FFQ were assessed in validation studies (16–19). Correlation coefficients between the FFQ and the reference instrument (repeated 24-h dietary recalls), for the nutrients under investigation in this study, ranged from 0.44 for carotenoids to 0.95 for alcohol and indicated a reasonably good assessment of nutrient intake.

Physical activity data
As part of a personal computer–assisted interview, participants were asked to estimate their weekly physical activity over the previous 12 mo. The interview included questions on walking, bicycling sports, gardening, do-it-yourself and household activities, the number of hours of television watched daily, and the number of hours slept daily. In addition, information on the type of physical activity performed during work was assessed on the basis of 4 categories (sedentary, light, medium, and vigorous). The daily duration of each activity was calculated and multiplied by metabolic equivalents (METs) derived from a compendium by Ainsworth et al (20). The durations of all activities were weighted by their corresponding MET values and summed to form the physical activity level (PAL). An activity score for recreational activity only was calculated separately in the same way.

Anthropometric and other data
Body weight was measured, with participants wearing underwear only, to the nearest 0.1 kg; body height was measured to the nearest 0.1 cm. Waist circumference was measured midway between the lower rib margin and the superior iliac spine; hip circumference was measured at the widest point over the greater trochanters. Both circumferences were assessed while subjects were in a standing position and were recorded to the nearest millimeter. Body mass index (BMI) was calculated as body weight (kg) divided by height (m) squared and the waist-hip ratio (WHR) as waist circumference divided by hip circumference. The CV for inter- and intraobserver effects for most anthropometric measures was <5% (21). Sociodemographic and lifestyle variables and medical data were obtained by trained interviewers using computer-assisted interviews and a self-administered questionnaire. Questions on current medication use referred to the 4 wk before the interview.

Statistical analysis
A first analysis of variance indicated that the variable batch number, 100 randomly selected samples of the immunoassay, was significantly related to Hb A_1c values even after adjustment for other variables such as age, sex, and BMI, indicating systematic analytic differences between batches. Therefore, we standardized the data for each batch to the overall mean and SD by using a z transformation. This standardization did not change the ranking of values within batches; however, it prevented variations in results due to the laboratory methods.

In univariate analyses the mean Hb A_1c value was calculated by lifestyle category separately for men and women. For ordinal-scaled lifestyle factors with 3 categories, a P value for trend across categories was computed. To study the association between the lifestyle factors and Hb A_1c values in a multivariate analysis, we divided the population of men and women separately into tertiles of Hb A_1c using the sex-specific 33rd and 66th percentiles as cutoffs. The lowest tertile of the distribution was labeled as the low-glycation group (control subjects) and the highest tertile as the high-glycation group (case subjects) and the odds ratio (OR) for exposure categories was computed by using unconditional logistic regression analysis. Associations with Hb A_1c were considered statistically significant at < 0.05 for the test of trend across quintiles and a 95% CI of the OR for the extreme category that did not include 1. The logistic regression analysis of associations between nutrients and Hb A_1c status was supplemented by calculating unadjusted mean Hb A_1c values in each nutrient quintile. All nutrients were converted into energy-adjusted values by using the residual method (22) before quintiles were formed. Statistical analyses were performed by using SAS (version 6.10; SAS Institute Inc, Cary, NC).

RESULTS  
General characteristics of the 1773 subjects included in this analysis are shown in Table 1. About one-third of the men and one-fifth of the women were current smokers. Vitamin supplements were taken by 12% of men and 14% of women, with multivitamin preparations being the most frequently used. Most of the study participants were of the higher educational level. The mean Hb A_1c value was 0.049 for men and 0.047 for women. Tertile thresholds were at 0.046 and 0.052 for men and at 0.044 and 0.050 for women. The corresponding mean Hb A_1c values for the first, second, and third tertiles were 0.042, 0.049, and 0.058 in men and 0.040, 0.047, and 0.055 in women, respectively. Mean dietary intakes of energy, macronutrients, and specific micronutrients were assessed by the FFQ in the range typically observed in Western populations (Table 2).

View this table:
TABLE 1.. General characteristics and glycated hemoglobin (Hb A_1c) values in EPIC-Potsdam Study participants¹  

View this table:
TABLE 2.. Dietary intake of selected nutrients obtained with a food-frequency questionnaire in EPIC-Potsdam Study participants¹  
Mean Hb A_1c values by categories (Table 3) of age, obesity, and PAL indicated significantly greater Hb A_1c values with greater age and obesity. Small differences in mean values between male smokers and nonsmokers as well as between women with and without a high school education were observed. However, these differences were not significant in the multivariate regression analysis when the study population was divided into tertiles of Hb A_1c (Table 4). These regression analyses showed a clear trend of greater hemoglobin glycation across greater age and obesity categories after adjustment for PAL, smoking status, and educational attainment. Age was the strongest independent factor positively related to high Hb A_1c values in this multivariate analysis. For men, the OR of being in the highest tertile of glycation was significantly greater by 3-fold. For women, the OR associated with a high Hb A_1c value was greater in those aged 55 y, ie, after menopause. Being in the highest category of obesity, eg, having a high BMI and a high WHR, was as an additional factor associated with higher Hb A_1c values. There was no significant variation in the risk of hemoglobin glycation associated with different PALs.

View this table:
TABLE 3.. Mean glycated hemoglobin values (Hb A_1c) by age, BMI, physical activity level (PAL), and educational attainment in EPIC-Potsdam Study participants¹  

View this table:
TABLE 4.. Odds ratios (ORs) and 95% CIs for being a case subject (high Hb A_1c: third tertile) compared with being a control subject (low Hb A_1c: first tertile) by age, obesity, physical activity level (PAL), smoking status, and educational attainment in EPIC-Potsdam Study participants¹  
OR estimates for nutrients obtained from a multivariate model with adjustment for age, obesity, PAL, educational attainment, and smoking status are shown in Table 5. For each of the investigated nutrients, mean Hb A_1c values per quintile were also presented. Energy intake in the highest quintile increased the OR of having high Hb A_1c values compared with the lowest category of energy intake. Of the energy-providing nutrients, the risk for the highest quintile of energy-adjusted intake of saturated fat was 2-fold greater when compared with the lowest quintile and there was a significant trend across quintiles. Carbohydrate (with the subgroups disaccharide and polysaccharide), protein, and fiber intakes showed no significant associations with Hb A_1c. For alcohol intake, there was a significant inverse association with Hb A_1c values in the highest quintile compared with the lowest quintile and a statistically significant trend across quintiles. In the analysis of antioxidant vitamins, supplement use of the corresponding vitamin was included in the multivariate analysis as a binary variable. A significantly lower OR was observed for vitamin C intake in the highest compared with the lowest quintile and there was a significant trend across quintiles. The OR related to vitamin E intake was significant for the highest quintile compared with the lowest and the test of trend was also significant. High - and ß-carotene intakes, expressed as ß-carotene equivalents, did not show a significant association with Hb A_1c values. More complex statistical models investigating the interrelation of dietary variables and substitution and addition effects were also investigated. However, the conclusions resulting from this exercise were practically the same as those from the simple models; therefore, specific data were not presented in this article.

View this table:
TABLE 5.. Odds ratios (ORs) and 95% CIs for being a case subject (high Hb A_1c: third tertile) compared with being a control subject (low Hb A_1c: first tertile) by quintiles of intake of energy and energy-adjusted carbohydrate, fat, protein, alcohol, fiber, and vitamins with the lowest quintile as the reference category in EPIC-Potsdam Study participants¹  

DISCUSSION  
In this population of nondiabetic, middle-aged men and women, the OR of having high Hb A_1c values was progressively greater with age. In addition to age, obesity was also a significant predictor of Hb A_1c value. Beyond age and obesity, high total energy and saturated fat intakes were significantly associated with high Hb A_1c values, whereas high intakes of alcohol, vitamin C, and vitamin E were inversely associated with high Hb A_1c values.

A major limitation in the interpretation of these results was that they were obtained from cross-sectional data. This type of study design does not allow conclusions to be drawn regarding the temporal relation between exposure and outcome. Such temporal relations can only be analyzed in prospective studies. The current analysis, therefore, describes how Hb A_1c values and lifestyle factors are associated when assessed simultaneously. We used a descriptive statistical approach by comparing mean Hb A_1c values across categories, and a multivariate approach by calculating ORs for the highest tertile compared with the lowest tertile of Hb A_1c values. The second (middle) tertile was excluded from this particular analysis because already small analytic variation in Hb A_1c values may have resulted in misclassification of long-term glucose concentrations to adjacent tertile groups. Even though the immunoassay used to determine Hb A_1c values showed a low CV in repeated measurements from the same blood samples, misclassification is still possible (23). The applied analytic approach therefore compared 2 clearly distinguishable Hb A_1c groups. However, because of the exclusion of the middle tertile of Hb A_1c values, the OR estimates reflect only differences in the occurrence of Hb A_1c values belonging to the first and the third tertiles, depending on exposure categories, and were not valid for the whole study group. This approach allowed us to elucidate associations that might have otherwise been masked and is different from usual regression methods.

The finding that Hb A_1c values were greater with greater age is consistent with the view that hemoglobin glycation is accelerated by the aging process. However, it is not clear whether this is a cause or merely a phenomenon of aging (24, 25). In addition to the age effect, we observed a greater number of women with high Hb A_1c values after menopause. This change with menopause was also reported by Simon et al (26) for a population in France. Menopause was found to be an important risk factor for type 2 diabetes in a Japanese American cohort (27). Endocrine alterations in menopause are known to increase the abdominal visceral fat content as well as blood concentrations of cholesterol, triacylglycerol, glucose, and insulin (28). Changes in visceral fat content were found to be associated with subsequent impaired glucose-insulin homeostasis (29). Similar to the relation of sex hormones with risk of coronary heart disease (30), sex hormones may act as a protective factor against type 2 diabetes and high glucose concentrations. In this context it is interesting to note that women with type 2 diabetes completely lose their sex-related protection from cardiovascular disease (31).

Together with age, obesity was an additional factor related to Hb A_1c values in this study. Obesity has long been recognized as an important risk factor for diabetes and impaired glucose tolerance, an association that was confirmed in many prospective studies (32–35). The finding of an association of exsmoking with greater glycation in women may have been related to obesity status, which was not completely controlled for by our BMI adjustment. Physical activity was implicated in many studies as having a protective effect against type 2 diabetes (34, 36, 37). However, in the present study, no direct relation of physical activity to Hb A_1c values was observed.

In the present study, several dietary factors were found to be related to Hb A_1c when other nondietary factors were accounted for. Significantly greater ORs were found for the highest category of energy and saturated fat intakes. Additionally, the trends across categories were significant. The Zutphen Study reported a positive association between fasting glucose and intake of saturated fat after age, obesity, and energy intake were controlled for (38). Saturated fat intake has also been identified as an important lifestyle factor in the development of type 2 diabetes in various populations, such as Japanese Americans (27), Pima Indians, and Americans of Mexican descent (7). The Nurses' Health Study, however, did not show type 2 diabetes to be associated with saturated fat intake, but vegetable fat intake did show a protective effect (39). Indication of the underlying mechanism is given by metabolic studies in humans suggesting that increased saturated fat consumption may increase insulin secretion and possibly lead to insulin insensitivity (40). Other macronutrient intakes, such as those of carbohydrates and protein, were also associated with Hb A_1c values. However, because these associations were not significant, they were not commented on here.

We observed no inverse association between total fiber intake and hemoglobin glycation. The fiber type that improves glucose tolerance and reduces hemoglobin glycation (41, 42) is viscous fiber (eg, guar, pectin, and psyllium). Salmeron et al (43, 44) reported that the ratio of low cereal fiber intake to high glycemic load was associated with increased risk of type 2 diabetes in the Nurses' Health Study as well as in the Health Professionals Follow-up Study. Previous evaluations of these cohorts showed no associations between fiber consumption and subsequent development of diabetes (39, 45).

Inverse associations were seen for antioxidants, particularly for vitamin C. The mechanism for the in vitro and in vivo effects of vitamin C on protein glycation has been suggested to be a competition of ascorbic acid and dehydroascorbic acid with glucose for reaction with the protein amino group, thereby inhibiting glycation (46). Because of the observed inverse relation between vitamin C and vitamin E and hemoglobin glycation, our results indicate that oxidative stress might not only play a role in the pathogenesis of diabetic complications (47), but also in the formation of glycation products. This explanation has also been suggested by Shoff et al (48), who found vitamin C intake to be negatively associated with glycated hemoglobin values in nondiabetic subjects.

The inverse association between alcohol consumption and high Hb A_1c values may be the effect of inhibition of gluconeogenesis in the liver. Clinical experience has indicated that heavy alcohol drinkers have lower concentrations of blood glucose than do light drinkers, reflected in corresponding fructosamine and glycated hemoglobin values (49–51).

In agreement with other studies, the results of our study indicate the potential for changes in lifestyle to reduce high glucose concentrations and subsequent risk of type 2 diabetes. The interest in nonenzymatic glycation reactions and preclinical stages of type 2 diabetes, both of which have detrimental effects on health, should initiate more efforts in identifying lifestyle patterns associated with low glucose concentrations.

ACKNOWLEDGMENTS  
We are grateful to H Stiete Greifswald for helping organize the Hb A_1c analysis, to W Bernigau for conducting part of the statistical analysis, and to Erik Rausch for devoting time to this project during his training in public health.

REFERENCES  

1. Schnider SL, Kohn RR. Glycosylation of human collagen in aging and diabetes mellitus. J Clin Invest 1980;66:1179–81.
2. Brownlee M, Cerami A, Vlassara H. Advanced glycosylation end products in tissue and the biochemical basis of diabetic complications. N Engl J Med 1988;318:1315–21.
3. Lyons TJ. Lipoprotein glycation and its metabolic consequences. Diabetes 1992;41(suppl):67–73.
4. Giugliano D, Ceriello A, Paolisso G. Oxidative stress and diabetic vascular complications. Diabetes Care 1996;19:257–67.
5. Jain SK, Palmer M. The effect of oxygen radicals metabolites and vitamin E on glycosylation of proteins. Free Radic Biol Med 1997; 22:593–6.
6. Spagnoli LG, Mauriello A, Orlandi A, Sangiorgi G, Bonanno E. Age-related changes affecting atherosclerotic risk. Potential for pharmacological intervention. Drugs Aging 1996;8:275–98.
7. Hannah JS, Howard BV. Dietary fats, insulin resistance and diabetes. J Cardiovasc Risk 1994;1:31–7.
8. Virtanen SM, Aro A. Dietary factors in the aetiology of diabetes. Ann Med 1994;26:469–78.
9. Salonen JT, Nyyssönen K, Tuomainen, et al. Increased risk of non-insulin dependent diabetes mellitus at low plasma vitamin E concentrations: a four year follow up study in men. BMJ 1995;311:1124–7.
10. Koenig RJ, Cerami A. Hemoglobin A_1c and diabetes mellitus. Annu Rev Med 1980;31:29–34.
11. Goldstein DE, Parker KM, England JD. Clinical application of glycosylated hemoglobin measurements. Diabetes 1982;31(suppl):70–8.
12. Nathan DM, Singer DE, Godine JE, Harrington CH, Perlmuter LC. Retinopathy in older type II diabetics. Association with glucose control. Diabetes 1986;35:797–801.
13. Riboli E. Nutrition and cancer. Background and rationale of the European Prospective Investigation into Cancer and Nutrition (EPIC). Ann Oncol 1992;3:783–91.
14. Voss S, Boeing H, Jeckel A, et al. European Prospective Investigation into Cancer and Nutrition (EPIC) and health, nutrition and cancer. Ernähr Umsch 1995;42:97–101.
15. Bundesinstitut für gesundheitlichen Verbraucherschutz und Veterinärmedizin. (Federal Institute for Health Protection of Consumers and Veterinary Medicine.) The German food code and nutrient data base, BLS II.2. Berlin: BgVV Journal, 1998 (in German).
16. Bohlscheid-Thomas S, Hoting I, Boeing H, Wahrendorf J. Reproducibility and validity of food group intake of a food frequency questionnaire developed for the German part of the EPIC-project. Int J Epidemiol 1997;26(suppl):59–70.
17. Bohlscheid-Thomas S, Hoting I, Boeing H, Wahrendorf J. Reproducibility and validity of energy and macronutrient intake of a food frequency questionnaire developed for the German part of the EPIC-project. Int J Epidemiol 1997;26:71–81.
18. Boeing H, Bohlscheid-Thomas S, Voss S, Schneeweiss S, Wahrendorf J. The relative validity of vitamin intakes derived from a food frequency questionnaire compared to 24-hour recalls and biological measurements: results from the EPIC pilot study in Germany. Int J Epidemiol 1997;26:S82–90.
19. Kroke A, Klipstein-Grobusch K, Voss S, et al. Validation of a self-administered food frequency questionnaire administered in the European Prospective Investigation into Cancer and Nutrition (EPIC) Study: comparison of energy, protein, and macronutrient intakes estimated with doubly labeled water, urinary nitrogen, and repeated 24-h recall methods. Am J Clin Nutr 1999;70:439–47.
20. Ainsworth BE, Haskell WL, Leon AS, et al. Compendium of physical activities: classification of energy costs of human physical activities. Med Sci Sports Exerc 1993;25:71–80.
21. Klipstein-Grobusch K, Georg T, Boeing H. Interviewer variability in anthropometric measurements and estimates of body composition. Int J Epidemiol 1997;26:174–80.
22. Willett WC, Stampfer MJ. Total energy intake: implications for epidemiologic analysis. Am J Epidemiol 1986;124:17–27.
23. Kimmerle R, Heinemann L. HbA_1C oder Nüchternplasmaglukose für die Diabetes Erkennung: Alternativen zum oralen Glukosetoleranztest. (HbA_1C or fasting plasma glucose for the detection of diabetes: alternatives to the oral glucose tolerance test.) Diabetes Stoffw 1995;4:57–70 (in German).
24. Nakashima K, Nishizaki O, Andoh Y. Acceleration of hemoglobin glycation with aging. Clin Chim Acta 1993;215:111–8.
25. Yang YC, Lu FH, Wu JS, Chang CJ. Age and sex effects on HbA1c. A study in a healthy Chinese population. Diabetes Care 1997; 20:988–91.
26. Simon D, Senan C, Garnier P, Saint-Paul M, Papoz L. Epidemiological features of glycated haemoglobin A1c—distribution in a healthy population. The Telecom Study. Diabetologia 1989;32:864–9.
27. Fujimoto WY, Bergstrom RW, Boyko EJ, et al. Diabetes and diabetes risk factors in second- and third-generation Japanese Americans in Seattle, Washington. Diabetes Res Clin Pract 1994;24:43–52.
28. Koranyi L. Effect of menopause on carbohydrate metabolism. Orv Hetil 1995;136:457–9.
29. Lemieux S, Prud'homme D, Nadeau A, Tremblay A, Bouchard C, Despres JP. Seven-year changes in body fat and visceral adipose tissue in women. Association with indexes of plasma glucose-insulin homeostasis. Diabetes Care 1996;19:983–91.
30. Barrett-Connor E. Heart disease in women. Fertil Steril 1994; 62:127–32.
31. Miller M, Vogel RA. The practice of coronary disease prevention. London: Williams & Wilkins, 1996:225–42.
32. Warren DK, Charles MA, Hanson RL, et al. Comparison of body size measurements as predictors of NIDDM in Pima Indians. Diabetes Care 1995;18:435–9.
33. Monterrosa AE, Haffner SM, Stern MP, Hazuda HP. Sex differences in life-style factors predictive of diabetes in Mexican-Americans. Diabetes Care 1995;18:448–56.
34. Perry IJ, Wannamethee SG, Walker MK, Thomson AG, Whincup PH, Shaper AG. Prospective study of risk factors for development of non-insulin dependent diabetes in middle aged British men. BMJ 1995;310:560–4.
35. Colditz GA, Willett WC, Rotnitzky A, Manson Jo-A. Weight gain as a risk factor for clinical diabetes mellitus in women. Ann Intern Med 1995;122:481–6.
36. Manson JE, Rimm EB, Stampfer MJ, et al. Physical activity and incidence on non-insulin-dependent diabetes mellitus in women. Lancet 1991;338:774–8.
37. Helmrich SP, Ragland DR, Leung RW, Paffenbarger RS Jr. Physical activity and reduced occurrence of non-insulin-dependent diabetes mellitus. N Engl J Med 1991;325:147–52.
38. Feskens EJ, Kromhout D. Cardiovascular risk factors and the 25-year incidence of diabetes mellitus in middle-aged men. The Zutphen study. Am J Epidemiol 1989;130:1101–8.
39. Colditz GA, Manson JE, Stampfer MJ, Rosner B, Willett WC, Speizer FE. Diet and risk of clinical diabetes in women. Am J Clin Nutr 1992;55:1018–23.
40. Collier G, O'Dea K. The effect of coingestion of fat on the glucose, insulin, and gastric inhibitory polypeptide responses to carbohydrate and protein. Am J Clin Nutr 1983;37:941–4.
41. Vuorinen-Markkola H, Sinisalo M, Koivisto VA. Guar gum in insulin-dependent diabetes: effects on glycemic control and serum lipoproteins. Am J Clin Nutr 1992;56:1056–60.
42. Groop PH, Aro A, Stenman S, Groop L. Long-term effects of guar gum in subjects with non-insulin-dependent diabetes mellitus. Am J Clin Nutr 1993;58:513–8.
43. Salmeron J, Manson JE, Stampfer MJ, Colditz GA, Wing AL, Willett WC. Dietary fiber, glycemic load, and risk of non-insulin-dependent diabetes mellitus in women. JAMA 1997;277:472–7.
44. Salmeron J, Ascherio A, Rimm EB, et al. Dietary fiber, glycemic load, and risk of NIDDM in men. Diabetes Care 1997;20:545–50.
45. Marshall JA, Weiss NS, Hamman RF. The role of dietary fibre in the etiology of non-insulin-dependent diabetes mellitus. Ann Epidemiol 1993;3:18–26.
46. Davie SJ, Gould BJ, Yudkin JS. Effect of vitamin C on glycosylation of proteins. Diabetes 1992;41:167–73.
47. Dandona P, Thusu K, Cook S, et al. Oxidative damage to DNA in diabetes mellitus. Lancet 1996;347:444–5.
48. Shoff SM, Mares-Perlman JA, Cruickshanks KJ, Klein R, Klein BE, Ritter LL. Glycosylated hemoglobin concentrations and vitamin E, vitamin C, and beta-carotene intake in diabetic and nondiabetic older adults. Am J Clin Nutr 1993;58:412–6.
49. Ben G, Dal Fabbro S, Mongillo A, Pellegrini P, Fedele D. Does alcohol intake interfere with the evaluation of glycated hemoglobins? Acta Diabetol Lat 1989;26:337–43.
50. Kallner A, Blomquist L. Effect of heavy drinking and alcohol withdrawal on markers of carbohydrate metabolism. Alcohol Alcohol 1991;26:425–9.
51. Lange J, Arends J, Willms B. Alkoholinduzierte Hypoglykämie bei Patienten mit Type I diabetes. (Alcohol-induced hypoglycemia in type I diabetic patients.) Med Klin 1991;86:551–4 (in German).