8,456 peer-reviewed studies, each with a short note written from its published abstract: what kind of study it was, who took part, what was measured and what the authors reported. Observational studies show associations, not proof.
Patterns in how people eat and what happens to them. Shows an association, not proof.
2000 / Cross-sectional comparative study
Thirty nondiabetic subjects (18 with BMI 21.8 +/- 3.3 kg/m2 and 12 with BMI 34.6 +/- 2.7 kg/m2) were evaluated for intramuscular triglyceride (mTG) content, fatty acid profile, and glucose uptake using euglycemic-hyperinsulinemic clamp. Glucose uptake (M) value negatively correlated with mTG levels (R2 = -.56, P < .0001), which were higher in obese subjects (11.6 +/- 2.2 vs 6.2 +/- 1.4 micromol/g wet weight muscle tissue, P < .0001). Obese subjects had a significantly lower unsaturated to saturated fat ratio (1.89 +/- 0.40 vs 2.19 +/- 0.07, P < .0001). In stepwise regression, only total mTGs and palmitic acid were significantly related to glucose uptake (R2 = .66, P < .0001).
Shows an association, not proof of cause and effect.
Manco M, Mingrone G, Greco AV, et al. Metabolism, 2000. PMID 10690948
2000 / Cross-sectional comparative study with a weight-loss intervention group
Researchers evaluated midthigh adipose tissue distribution via CT, muscle attenuation, and insulin sensitivity across lean, obese glucose-tolerant, and obese type 2 diabetic subjects, as well as obese individuals who completed a weight-loss program. Subcutaneous thigh fat was elevated in obesity but did not correlate with insulin sensitivity. Subfascial adipose tissue (~8% of total thigh fat) and intermuscular adipose tissue (~3% of thigh fat) both correlated with insulin sensitivity. Muscle attenuation reflecting lower fat content was linked to higher insulin sensitivity, and weight loss reduced total thigh fat, the proportion of intermuscular fat, and low-density muscle.
Shows an association, not proof of cause and effect.
Goodpaster BH, Thaete FL, Kelley DE Am J Clin Nutr, 2000. PMID 10731493
2000 / Ecologic study with univariate and multivariate regression
An ecologic study assessed the relationship between average per capita daily intake of major food items and childhood type 1 diabetes incidence across 40 countries in the late 1980s and early 1990s. In univariate analysis, total energy was not significantly associated with diabetes incidence (r = 0.31, NS), while energy from animal sources was positively associated (r = 0.61, P < 0.01) and energy from vegetal sources was inversely associated (r = -0.35, P < 0.05). Specifically, dairy products (r = 0.80, P < 0.0001) and meat (r = 0.55, P < 0.001) were positive predictors, whereas cereals were inverse predictors (r = -0.64, P < 0.001). In multivariate analysis, these animal and vegetal energy relations explained associations with geographic and socioeconomic covariates.
Shows an association, not proof of cause and effect.
Muntoni S, Cocco P, Aru G, et al. Am J Clin Nutr, 2000. PMID 10837294
2000 / Nested case-control study
Researchers analyzed infant feeding patterns, childhood diet, and HLA-DQB1 genotype in siblings of children with diabetes over a median follow-up of 9.7 years. Early introduction of cow's milk supplements was not significantly linked to diabetes progression (adjusted RR 1.60, 95% CI 0.5-5.1). However, childhood cow's milk consumption was associated with an increased risk of progression to type 1 diabetes (adjusted RR 5.37, 95% CI 1.6-18.4) after adjusting for sociodemographic factors, age at milk supplementation, and genetic susceptibility markers.
Shows an association, not proof of cause and effect.
Virtanen SM, Läärä E, Hyppönen E, et al. Diabetes, 2000. PMID 10866042
1999 / Cross-sectional study
Researchers evaluated the relationship between intramyocellular lipid (IMCL) content and insulin sensitivity in 23 normal-weight non-diabetic adults using 1H NMR spectroscopy of soleus muscle and a 2-hour hyperinsulinaemic-euglycaemic clamp. Simple linear regression revealed an inverse correlation between IMCL content and M-value (r = -0.579, p = 0.0037), as well as between fasting plasma non-esterified fatty acid concentration and M-value (r = -0.54, p = 0.0267). IMCL content was not related to BMI, age, or fasting plasma levels of triglycerides, non-esterified fatty acids, glucose, or insulin.
Shows an association, not proof of cause and effect.
Krssak M, Falk Petersen K, Dresner A, et al. Diabetologia, 1999. PMID 10027589
1999 / Ecological study
Type 1 diabetes incidence in children aged 0 to 14 years from 10 countries or areas was compared with national annual cow milk protein consumption. Total protein consumption did not correlate with diabetes incidence (r = +0.402), but beta-casein A1 variant consumption did (r = +0.726). A stronger relationship was found between beta-casein (A1+B) consumption and diabetes incidence (r = +0.982).
Shows an association, not proof of cause and effect.
Elliott RB, Harris DP, Hill JP, et al. Diabetologia, 1999. PMID 10096780
1999 / Prospective cohort study
Investigators evaluated 200 Finnish infants carrying HLA-DQB1*0302 from a birth cohort, comparing 100 exposed to cow's milk formula before 12 weeks with 100 exclusively breast-fed past 12 weeks. At age 3 months, formula-exposed infants had higher IgG antibodies binding bovine insulin than exclusively breast-fed infants (median 0.521 vs. 0.190; P < 0.0001). These antibodies cross-reacted with human insulin, although no infants tested positive for insulin autoantibodies at that age. Antibody levels declined by 12 and 18 months in both groups, but increased over follow-up in 11 children who developed at least two diabetes-associated autoantibodies (P < 0.0001).
Shows an association, not proof of cause and effect.
Vaarala O, Knip M, Paronen J, et al. Diabetes, 1999. PMID 10389843
1999 / Case-control study
Researchers examined 14 young lean offspring of type 2 diabetic parents and 14 matched healthy control subjects using euglycemic-hyperinsulinemic clamps, 1H/13C NMR spectroscopy, and DEXA scans. Offspring had normal fat content and distribution, but showed insulin resistance and increased intramyocellular triglyceride content in the soleus muscle (P < 0.01), with no significant difference in the tibialis anterior (P = 0.19). Stepwise regression identified soleus intramyocellular triglyceride content and plasma free fatty acid levels as the main predictors of whole-body insulin sensitivity.
Shows an association, not proof of cause and effect.
Perseghin G, Scifo P, De Cobelli F, et al. Diabetes, 1999. PMID 10426379
1999 / Cohort study (with cross-sectional data)
Dietary associations with chronic disease were summarized in a cohort of 34,192 California Seventh-day Adventists. Beef intake was associated with fatal IHD in men (RR = 2.31 for >=3 times/wk vs. vegetarians), whereas consuming nuts >=5 times/wk was tied to lower fatal and nonfatal IHD risk (RR approximately 0.5 vs. <1 time/wk). Colon and prostate cancers were significantly more likely in nonvegetarians (RR of 1.88 and 1.54, respectively), and legumes and fruit were negatively associated with certain cancers. Cross-sectional data suggest vegetarian Adventists have lower risks of diabetes mellitus, hypertension, and arthritis than nonvegetarians.
Shows an association, not proof of cause and effect.
Fraser GE Am J Clin Nutr, 1999. PMID 10479227
1999 / Multicenter prospective cohort study
In the CARDIA study, researchers tracked 2,909 young adults over 10 years to examine associations between dietary fiber intake and cardiovascular risk factors. From lowest to highest quintiles of fiber intake, significant linear reductions were observed in body weight (whites: 78.3 to 75.0 kg, P<.001; blacks: 83.5 to 79.9 kg, P=.001), waist-to-hip ratio, BMI-adjusted fasting insulin, and BMI-adjusted 2-hour postglucose insulin. Fiber was also associated with blood pressure, triglycerides, HDL and LDL cholesterol, and fibrinogen, but these associations attenuated after adjusting for fasting insulin. Intakes of fat, carbohydrate, and protein had inconsistent or weak associations with CVD risk factors compared with fiber.
Shows an association, not proof of cause and effect.
Ludwig DS, Pereira MA, Kroenke CH, et al. JAMA, 1999. PMID 10546693
1998 / Comparative cohort study
Researchers measured IgG antibodies to bovine insulin (BI) and human insulin (HI) in children receiving whole cow milk (CM) formula, hydrolyzed casein (HC) formula, or exclusive breastfeeding during the first 9 months of life. At 6 months of age, children receiving CM formula had higher median IgG antibodies to BI than those receiving HC formula (0.480 vs 0.185, P = 0.04) or exclusively breast-fed children (0.480 vs 0.160, P = 0.04). At 9 months, levels remained higher in the CM group than the HC group (0.403 vs 0.230, P = 0.02). Antibodies to BI and HI were positively correlated and cross-reacted in inhibition studies.
Shows an association, not proof of cause and effect.
Vaarala O, Paronen J, Otonkoski T, et al. Scand J Immunol, 1998. PMID 9496688
1998 / Multicenter autopsy study
The Pathobiological Determinants of Atherosclerosis in Youth study evaluated the relationship between adult coronary artery disease risk factors and atherosclerosis in nearly 3,000 autopsied individuals aged 15-34 years. VLDL plus LDL cholesterol was positively, and HDL cholesterol was negatively, linked with fatty streaks and raised lesions in the aorta and right coronary artery, particularly after age 25. Elevated glycohemoglobin was associated with raised lesions across ages 15-34. BMI was linked to both fatty streaks and raised lesions of the right coronary artery in men, and smoking was tied to a 3-fold increase in raised lesions of the abdominal aorta in the 25-34-year age group.
Shows an association, not proof of cause and effect.
McGill HC, McMahan CA Am J Cardiol, 1998. PMID 9860371
1997 / Cross-sectional study
Researchers evaluated cross-sectional correlates of C-reactive protein (CRP) in 400 elderly participants free of clinical cardiovascular disease. Only 2% had CRP levels >10 mg/L. Strong bivariate correlations were found between CRP and fibrinogen (r = 0.52), plasmin-antiplasmin complex (r = 0.23), pack-years of smoking (r = 0.30), and BMI (r = 0.24; all P <= 0.001). Multiple regression models showed that in never smokers, BMI, diabetes, plasmin-antiplasmin complex, and ECG abnormalities predicted 15% of CRP variance. In ever smokers, BMI, plasmin-antiplasmin complex, pack-years of smoking, HDL cholesterol (inversely), and ankle-arm blood pressure index (inversely) predicted 42% of the variance.
Shows an association, not proof of cause and effect.
Tracy RP, Psaty BM, Macy E, et al. Arterioscler Thromb Vasc Biol, 1997. PMID 9351386
1995 / Cross-sectional observational study
Researchers collected hair samples from 133 diseased volunteers in Tokyo and surrounding areas between October 1992 and June 1993 to evaluate total hair mercury levels. Total mercury concentrations among participants with atopic dermatitis, asthma, dementia, cerebral infarct, osteoporosis, hypertension, and diabetes ranged from 2.08 ppm to 36.5 ppm. These values were reported as considerably higher than those of healthy people of the same age groups. The uptake routes and metabolic mechanisms responsible for high hair mercury levels in these individuals remain unclear.
Shows an association, not proof of cause and effect.
Nakagawa R Chemosphere, 1995. PMID 7874464
1993 / Cross-sectional analysis
This study evaluated how body mass index, abdomen-hip ratio, and dietary intake relate to fasting and postprandial insulin concentrations in 652 men aged 43-85 years. Log-transformed fasting insulin correlated significantly with body mass index, abdomen-hip ratio, total fat energy, and saturated fatty acid energy (correlation coefficients from 0.14 for total fat to 0.45 for body mass index). In multivariate models adjusting for age, smoking, and physical activity, body mass index, abdomen-hip ratio, and saturated fatty acid intake independently predicted fasting and postprandial insulin. Models indicated that reducing saturated fatty acids from 14% to 8% of total energy would correspond to an 18% decrease in fasting insulin and a 25% decrease in postprandial insulin.
Shows an association, not proof of cause and effect.
Parker DR, Weiss ST, Troisi R, et al. Am J Clin Nutr, 1993. PMID 8338037
1993 / Ethnohistoric dietary reconstruction and nutrient analysis
Investigators evaluated ethnohistoric literature and traditional recipes to approximate the composition of foods available to Pima Indians roughly 100 years ago. The traditional Pima diet was estimated to be approximately 70-80% carbohydrate, 8-12% fat, and 12-18% protein, which is much higher in carbohydrate and lower in fat compared to their modern diet. The authors found that an analogous diet is largely reproducible with modern foods and suggested that changes toward this traditional diet may help decrease diabetes incidence.
Shows an association, not proof of cause and effect.
Boyce VL, Swinburn BA Diabetes Care, 1993. PMID 8422813
1992 / Case-control study
Researchers analyzed anti-BSA and anti-ABBOS antibodies in 142 children with insulin-dependent diabetes mellitus, 79 healthy children, and 300 adult blood donors. All diabetic patients had elevated serum concentrations of IgG anti-BSA antibodies, mainly specific for the ABBOS peptide (mean +/- SE: 8.5 +/- 0.2 kfU/microL vs. 1.3 +/- 0.1 kfU/microL in healthy children). IgA antibodies were also elevated, while IgM antibodies were not. Antibody concentrations declined after diagnosis, with most patients reaching normal levels within 1 to 2 years.
Shows an association, not proof of cause and effect.
Karjalainen J, Martin JM, Knip M, et al. N Engl J Med, 1992. PMID 1377788
1991 / Ecological correlation study
An ecological study compared validated age-standardized incidence rates of childhood diabetes (ages 0-14) from 12 countries with International Dairy Federation data on fluid cows' milk consumption. The correlation between milk consumption and the incidence of insulin-dependent diabetes mellitus was 0.96. Linear regression analysis indicated that 94% of the geographic variation in incidence might be explained by differences in milk consumption.
Shows an association, not proof of cause and effect.
Dahl-Jørgensen K, Joner G, Hanssen KF Diabetes Care, 1991. PMID 1797491
1991 / Cross-sectional study
Researchers examined the relationship between dietary composition and plasma insulin concentrations in 215 nondiabetic men with coronary artery disease. After adjusting for age, saturated fatty acid and cholesterol intakes were positively correlated with BMI, waist-to-hip ratio, and fasting insulin (r = 0.26 and r = 0.23, respectively; p < 0.05). Carbohydrate intake was negatively correlated with BMI (r = -0.21), waist-to-hip ratio (r = -0.21), and fasting insulin (r = -0.16). In multivariate analysis, saturated fat intake remained significantly associated with elevated fasting insulin concentrations independently of BMI.
Shows an association, not proof of cause and effect.
Maron DJ, Fair JM, Haskell WL Circulation, 1991. PMID 1934376
1991 / Observational study / population analysis
This study evaluated obesity trends and their relationship with diabetes in the Pima Indian population. On average, members of this population are obese as estimated by BMI, with young adults having the highest BMIs and modest increases in age- and sex-specific mean BMIs over the past 25 years. Present-day Pima children are also much heavier for height compared with children studied early in the century. Obesity was found to be familial and predicted the development of diabetes, though individuals tended to lose weight once they developed diabetes.
Shows an association, not proof of cause and effect.
Knowler WC, Pettitt DJ, Saad MF, et al. Am J Clin Nutr, 1991. PMID 2031485
1988 / Prospective cohort study
A prospective study tracked 34,000 California Seventh-day Adventists between 1976 and 1983, observing 40 deaths from pancreas cancer. Meat, egg, and coffee consumption showed suggestive associations with increased risk, but were not significantly related after controlling for smoking. Conversely, increasing consumption of vegetarian protein products, beans, lentils, peas, and dried fruit showed highly significant protective relationships with pancreas cancer risk. A prior history of diabetes and surgery for peptic or duodenal ulcer were also associated with increased fatal pancreas cancer risk.
Shows an association, not proof of cause and effect.
Mills PK, Beeson WL, Abbey DE, et al. Cancer, 1988. PMID 3365678
1985 / Prospective cohort study
Researchers analyzed 25,698 adult White Seventh-day Adventists identified in 1960 and followed for 21 years. During follow-up, the risk of diabetes as an underlying cause of death in Adventists was about half the risk for all US Whites. Among male Adventists, vegetarians had a substantially lower risk than non-vegetarians of diabetes as an underlying or contributing cause of death, and self-reported diabetes prevalence was lower in vegetarians of both sexes. Associations between meat intake and diabetes were stronger in males and were apparently not due to confounding by over- or under-weight, selected dietary factors, or physical activity.
Shows an association, not proof of cause and effect.
Snowdon DA, Phillips RL Am J Public Health, 1985. PMID 3985239
1984 / Case-control and ecological analysis
Epidemiological and case-control data in Scandinavian populations were analyzed regarding breast-feeding and childhood insulin-dependent diabetes mellitus (IDDM). Epidemiological findings suggest an inverse correlation between breast-feeding frequency and childhood IDDM. Case-control data revealed that diabetic children were breast-fed for shorter periods than their healthy siblings and the general population, and a smaller proportion of diabetic children were ever breast-fed. The authors hypothesize that insufficient breast-feeding in genetically susceptible infants may lead to beta-cell infection and later IDDM.
Shows an association, not proof of cause and effect.
Borch-Johnsen K, Joner G, Mandrup-Poulsen T, et al. Lancet, 1984. PMID 6150150