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.
2004 / Prospective cohort study
In the Miyagi Cohort Study, 28,333 Japanese adults aged 40-64 completed a questionnaire evaluating high-risk practices: smoking, consuming >=22.8 g alcohol/d, walking <1 hr/d, and BMI <18.5 or >=30.0. Over 11 years of follow-up, 1,200 subjects died. Compared with having no high-risk practices, multivariate RRs for 1 to 4 practices were 1.20, 1.66, 1.94, and 3.96 in men (P for trend <0.001) and 1.31, 2.14, 3.98, and 5.56 in women (P for trend <0.001). A unit increase in the number of high-risk practices corresponded to being 2.8 years older for men and 4.8 years older for women.
Shows an association, not proof of cause and effect.
Tsubono Y, Koizumi Y, Nakaya N, et al. J Epidemiol, 2004. PMID 15143877
2004 / Comparative nutrient and dietary survey analysis
Researchers analyzed nutrient data from Melbourne supermarket foods with reduced-fat (RF) claims and full-fat (FF) equivalents, along with recipes for high-fat (HF) vegetable-based dishes (>30% energy from fat, <10% saturated fat), referencing 1995 Australian National Nutrition Survey dietary intake data. Both FF and RF foods were more energy-dense than the Australian diet, whereas the HF vegetable-based dishes were less energy-dense. Although the Australian diet showed positive relationships between energy density and percent energy as fat, no such relationships were found for RF products or HF vegetable-based dishes.
Shows an association, not proof of cause and effect.
La Fontaine HA, Crowe TC, Swinburn BA, et al. Public Health Nutr, 2004. PMID 15153263
2004 / Cross-sectional analysis of 2 dietary recall samples
Researchers analyzed three nonconsecutive 24-hour dietary recalls from each child in two samples (third-graders and fifth-graders) to assess food intake during television viewing and its link to BMI. Children consumed 17-18% of total daily energy on weekdays and approximately 26% on weekend days during television viewing. Less soda, fast food, fruit, and vegetables were consumed with the television on, while food fat content did not differ significantly between television viewing and non-viewing times. Total food consumed during television viewing was not associated with BMI, but fat content of foods eaten while viewing was associated with BMI in the third-grade sample.
Shows an association, not proof of cause and effect.
Matheson DM, Killen JD, Wang Y, et al. Am J Clin Nutr, 2004. PMID 15159240
2004 / Cross-sectional analysis
Researchers analyzed baseline data from 3,302 women aged 42-52 years in the Study of Women's Health Across the Nation (SWAN) to examine lifestyle and demographic factors linked to vasomotor symptoms. After covariate adjustment, significantly more African-American and Hispanic women and fewer Chinese and Japanese women reported symptoms compared with Caucasian women. Postgraduate education was linked to fewer symptoms, while passive smoke exposure, higher BMI, premenstrual symptoms, perceived stress, and age were positively associated. No dietary nutrients were significantly associated with vasomotor symptoms.
Shows an association, not proof of cause and effect.
Gold EB, Block G, Crawford S, et al. Am J Epidemiol, 2004. PMID 15191936
2004 / Historical comparative study / repeated cross-sectional analyses
Researchers analyzed changes in BMI distribution among white non-Hispanic US men aged 40-69 years using records from 12,312 Union Army veterans (1890-1900) and 4,059 NHANES participants (1976-2000). Between 1890 and 2000, median BMI of men aged 50-59 years increased by 5.7 kg/m2 (25%), the standard deviation nearly doubled, and obesity prevalence climbed from 3.4% to 35%. While median BMI declined with age in 1890-1894, this pattern reversed by 2000. The increase in median BMI accounted for 75% of the increase in obesity prevalence, with the remainder linked to increased variance in the distribution.
Shows an association, not proof of cause and effect.
Helmchen LA, Henderson RM Ann Hum Biol, 2004. PMID 15204360
2004 / Cross-sectional study
A cross-sectional study evaluated 10,878 participants from a travel survey in the Atlanta, Georgia region to assess links between urban form, travel patterns, and obesity. Each quartile increase in land-use mix was associated with a 12.2% reduction in the likelihood of obesity across gender and ethnicity. Each additional hour spent in a car per day was linked to a 6% increase in the likelihood of obesity, whereas each additional kilometer walked per day was linked to a 4.8% reduction. Relationships between urban form, walk distance, and time in a car were stronger among white cohorts than black cohorts.
Shows an association, not proof of cause and effect.
Frank LD, Andresen MA, Schmid TL Am J Prev Med, 2004. PMID 15261894
2004 / Standardised case-control study
A standardized case-control study (INTERHEART) in 52 countries evaluated modifiable risk factors in 15,152 myocardial infarction cases and 14,820 controls. Significant associations with acute myocardial infarction were observed for smoking (OR 2.87), raised ApoB/ApoA1 ratio (OR 3.25), hypertension (OR 1.91), diabetes (OR 2.37), abdominal obesity (OR 1.62 for middle vs lowest tertile), psychosocial factors (OR 2.67), daily fruit and vegetable intake (OR 0.70), regular physical activity (OR 0.86), and regular alcohol intake (OR 0.91). Together, these nine factors accounted for 90% of the population attributable risk in men and 94% in women globally.
Shows an association, not proof of cause and effect.
Yusuf S, Hawken S, Ounpuu S, et al. Lancet, 2004. PMID 15364185
2004 / Cross-sectional geographic analysis
Using geographic information system software, researchers mapped 155 fast-food restaurants across 156 census tracts in New Orleans in 2001 to calculate restaurant density using 1-mile and 0.5-mile buffers. After controlling for commercial activity, major highways, and median home values, fast-food restaurant density was independently correlated with median household income and percent of black residents. Predominantly black neighborhoods had 2.4 fast-food restaurants per square mile compared to 1.5 in predominantly white neighborhoods.
Shows an association, not proof of cause and effect.
Block JP, Scribner RA, DeSalvo KB Am J Prev Med, 2004. PMID 15450633
2004 / Observational dietary intake study
The food and macronutrient intake of elite Kenyan runners was assessed over a 7-day intense training period. Estimated energy intake was significantly lower than energy expenditure (2987 +/- 293 kcal vs. 3605 +/- 119 kcal, P < 0.001), and body mass decreased from 58.9 +/- 2.7 kg to 58.3 +/- 2.6 kg (P < 0.001). The diet consisted of 76.5% carbohydrate (0.4 g/kg BM per day), 13.4% fat, and 10.1% protein (1.3 g/kg BM per day). Fluid intake was modest, primarily water (1113 +/- 269 mL) and tea (1243 +/- 348 mL).
Shows an association, not proof of cause and effect.
Onywera VO, Kiplamai FK, Boit MK, et al. Int J Sport Nutr Exerc Metab, 2004. PMID 15657475
2003 / Cohort / demographic life-table modeling study
Data from US Life Tables (1999), NHANES III (1988-1994), and NHANES I/II follow-up studies were used to estimate years of life lost (YLL) across BMI categories relative to a BMI of 24. Among whites, a J- or U-shaped association was found between BMI and YLL, with an optimal BMI of approximately 23 to 25 for whites and 23 to 30 for blacks. Maximum YLL for severe obesity (BMI >45) in adults aged 20 to 30 years was 13 for white men (a 22% reduction in expected remaining life span), 8 for white women, 20 for black men, and 5 for black women. Among black men and women older than 60 years, overweight and moderate obesity were generally not linked to increased YLL, with only severe obesity resulting in YLL.
Shows an association, not proof of cause and effect.
Fontaine KR, Redden DT, Wang C, et al. JAMA, 2003. PMID 12517229
2003 / Prospective cohort study
A prospective cohort of over 900,000 U.S. adults was followed for 16 years, during which 57,145 cancer deaths occurred. Individuals with a BMI of at least 40 had significantly higher cancer mortality rates than normal-weight individuals, with a relative risk of 1.52 (95% CI, 1.13 to 2.05) in men and 1.62 (95% CI, 1.40 to 1.87) in women. Higher BMI was also significantly associated with elevated death rates from cancers of the esophagus, colon/rectum, liver, gallbladder, pancreas, kidney, non-Hodgkin's lymphoma, and multiple myeloma, as well as stomach and prostate in men, and breast, uterus, cervix, and ovary in women. Based on these estimates, overweight and obesity could account for 14% of cancer deaths in men and 20% in women in the U.S.
Shows an association, not proof of cause and effect.
Calle EE, Rodriguez C, Walker-Thurmond K, et al. N Engl J Med, 2003. PMID 12711737
2003 / Cross-sectional analysis
A cross-sectional study of 37,875 healthy adults examined BMI across meat-eaters, fish-eaters, vegetarians, and vegans. Age-adjusted mean BMI was highest in meat-eaters (24.41 kg/m² in men, 23.52 kg/m² in women) and lowest in vegans (22.49 kg/m² in men, 21.98 kg/m² in women), while fish-eaters and vegetarians had intermediate values. Lifestyle factors explained less than 5% of the BMI difference between meat-eaters and vegans, whereas macronutrient intake differences accounted for about half. High protein intake (as percent energy) and low fiber intake were the dietary factors most strongly associated with higher BMI.
Shows an association, not proof of cause and effect.
Spencer EA, Appleby PN, Davey GK, et al. Int J Obes Relat Metab Disord, 2003. PMID 12833118
2003 / Prospective cohort study
Researchers prospectively evaluated daily energy metabolism and subsequent weight change in free-living, nondiabetic adult Pima Indians. Baseline energy metabolism was measured via doubly labeled water and resting metabolic rate (RMR) via indirect calorimetry, with weight changes (mean 5.8+/-6.5 kg) assessed over 4+/-3 years in 74 participants. Baseline calculated total energy intake (r=0.25, P=0.028) and RMR (r=-0.28, P=0.016) were significantly associated with changes in body weight. Baseline energy expenditure from physical activity was not associated with body weight changes.
Shows an association, not proof of cause and effect.
Tataranni PA, Harper IT, Snitker S, et al. Int J Obes Relat Metab Disord, 2003. PMID 12975636
2003 / Cross-sectional survey
A questionnaire was administered to two national random samples of 5,000 primary care physicians (2,500 defined obesity as BMI 30-40 kg/m², 2,500 as BMI > 40), yielding 620 respondents. Physicians rated physical inactivity as significantly more important than other causes (p < 0.0009), followed by overeating and a high-fat diet. Over 50% viewed obese patients as awkward, unattractive, ugly, and noncompliant, and obesity treatment was rated significantly less effective than therapies for 9 of 10 chronic conditions (p < 0.001). However, 75% agreed a 10% weight reduction improves complications, viewing a 14% loss as acceptable, and 54% would spend more time on weight management if reimbursed appropriately.
Shows an association, not proof of cause and effect.
Foster GD, Wadden TA, Makris AP, et al. Obes Res, 2003. PMID 14569041
2003 / Community survey
A community survey assessed cardiovascular risk factors in 5079 Australian adults aged 40-44 and 60-64 years randomly selected from the electoral roll. Pet owners and non-pet owners had similar systolic blood pressure, but pet owners had significantly higher diastolic blood pressure, higher BMI, and were more likely to smoke. Pet owners undertook more mild physical activity, but their diastolic blood pressure remained significantly higher after controlling for hypertensive risk factors.
Shows an association, not proof of cause and effect.
Parslow RA, Jorm AF Med J Aust, 2003. PMID 14583076
2003 / Cross-sectional questionnaire study
The study investigated the relationship between food intake and self-reported television viewing among 74 overweight women seeking obesity treatment. Participants completed questionnaires assessing their dietary habits and TV viewing behaviors. Results indicated that snacking, but not necessarily eating meals, while watching TV was associated with increased overall caloric intake and calories from fat.
Shows an association, not proof of cause and effect.
Gore SA, Foster JA, DiLillo VG, et al. Eat Behav, 2003. PMID 15000965
2002 / Cross-sectional study
A cross-sectional study of 4,057 apparently healthy, nonsmoking men examined age-adjusted resting levels and risk of elevated fibrinogen, white blood cell count, uric acid, and metabolic syndrome score across nine fitness-body fatness combinations. Fitness was inversely and BMI was directly related to the age-adjusted values of all four variables (P for trend < 0.0001 for each). Across all four variables, the greatest age-adjusted risk was in the low fitness-obese category and the lowest risk was in the high fitness-normal weight group. Within levels of fatness, the risk for significant elevations was lower for higher fitness groups.
Shows an association, not proof of cause and effect.
Church TS, Finley CE, Earnest CP, et al. Int J Obes Relat Metab Disord, 2002. PMID 12037651
2002 / Pretest/posttest study
Employees (n = 442) from six worksites self-selected into an 8-week facilitator-based video version of the Coronary Health Improvement Project (CHIP) evaluated via a pretest/posttest design. Demographic and biometric data were assessed at baseline and 8 weeks. All sites demonstrated significant and meaningful reductions in body weight, body mass index, total cholesterol, low-density lipoprotein cholesterol, triglycerides, and fasting blood glucose. Men demonstrated greater improvement than women, and those with higher baseline risks experienced the greatest reductions; the abstract notes future research is needed to determine persistence.
Shows an association, not proof of cause and effect.
Aldana SG, Greenlaw R, Diehl HA, et al. J Occup Environ Med, 2002. PMID 12227675
2002 / Multicenter cross-sectional autopsy study
The Pathobiological Determinants of Atherosclerosis in Youth (PDAY) study examined arteries, blood, and tissue from 3,000 autopsied individuals aged 15-34 years who died from external causes. Raised coronary artery lesions were positively associated with non-HDL cholesterol concentration, hypertension, obesity (in men), and blood glucose concentration, and inversely associated with HDL cholesterol concentration. Smoking predominantly affected the abdominal aorta.
Shows an association, not proof of cause and effect.
McGill HC, Herderick EE, McMahan CA, et al. Minerva Pediatr, 2002. PMID 12244281
2002 / Descriptive observational study
Athletes in the 2000 South African Ironman Triathlon were monitored for body weight, blood sodium concentrations, and postrace rectal temperatures. Body weight decreased significantly during the race (p < 0.0001) and remained below prerace values 12 hours later (p < 0.0001). Percentage change in body weight was inversely related to postrace serum sodium concentrations (r = -0.45) but was unrelated to postrace rectal temperatures or time in the marathon leg. Postrace serum sodium averaged 141.8 +/- 3.1 mmol/L, with three values (0.6%) below 135 mmol/L, and was negatively correlated with overall triathlon time (r = -0.22).
Shows an association, not proof of cause and effect.
Sharwood K, Collins M, Goedecke J, et al. Clin J Sport Med, 2002. PMID 12466696
2001 / Prospective cohort study
A cohort of 34,192 California Seventh-Day Adventists was followed from 1976 to 1988 to assess how choices regarding diet, exercise, and smoking influence life expectancy. California Adventists had higher life expectancies at age 30 than other white Californians by 7.28 years (95% CI, 6.59-7.97) in men and 4.42 years (95% CI, 3.96-4.88) in women. Commonly observed combinations of diet, exercise, body mass index, past smoking habits, and hormone replacement therapy accounted for differences of up to 10 years of life expectancy among Adventists. Independent effects between high-risk and low-risk values for different individual variables varied between 1.06 and 2.74 years.
Shows an association, not proof of cause and effect.
Fraser GE, Shavlik DJ Arch Intern Med, 2001. PMID 11434797
2001 / Observational dietary record study and mathematical modeling
Researchers modeled and evaluated the effects of macronutrient composition and non-macronutrients (such as water and fiber) on dietary energy density using 4 consecutive days of weighed food records from 45 adults. Non-macronutrients (% by weight) alone explained much more variation in dietary energy density between subjects (R2 95%) and day-to-day (R2 95%) than did % energy from dietary fat (R2 5% and 6%, respectively). Omitting beverages yielded similar results. The authors noted the smaller effect of macronutrient composition occurs primarily because macronutrients constitute only one-quarter to one-third of typical dietary intake weight.
Shows an association, not proof of cause and effect.
Grunwald GK, Seagle HM, Peters JC, et al. Br J Nutr, 2001. PMID 11502241
2001 / Prospective cohort study
Researchers followed 84,941 female nurses from 1980 to 1996 to examine the combined effects of diet and lifestyle factors on type 2 diabetes risk. Over 16 years, 3,300 new cases of type 2 diabetes were documented. Overweight or obesity was the single most important predictor, but lack of exercise, poor diet, smoking, and alcohol abstinence were also associated with significantly increased risk. Women adhering to all five low-risk factors (3.4% of the cohort) had a relative risk of diabetes of 0.09 (95% CI, 0.05 to 0.17) compared with the rest of the cohort.
Shows an association, not proof of cause and effect.
Hu FB, Manson JE, Stampfer MJ, et al. N Engl J Med, 2001. PMID 11556298
2001 / Cross-sectional survey with direct anthropometry
Researchers examined vegetarianism and its correlates in a representative sample of 4,746 adolescents from 31 public schools in Minnesota. About 6% of the sample were vegetarians. Compared to nonvegetarians, vegetarians were more likely to be female, not black, body dissatisfied, and involved in healthy and unhealthy weight control behaviors. Vegetarians more frequently reported a physician diagnosis of an eating disorder and were more likely to have contemplated and attempted suicide, with vegetarian males at especially high risk for unhealthy weight control practices.
Shows an association, not proof of cause and effect.
Perry CL, Mcguire MT, Neumark-Sztainer D, et al. J Adolesc Health, 2001. PMID 11728890