Professional Documents
Culture Documents
Abstract
Background: Growing evidence suggests that dairy products may have beneficial cardiometabolic effects. The current
Keywords: metabolic syndrome, dairy consumption, saturated fatty acids, cohort study, diabetes
Introduction
Metabolic syndrome (MetSyn)9 is a clustering of interrelated
Noncommunicable diseases are the main cause of morbidity, risk factors for cardiovascular disease (CVD) and type 2 diabetes
disability, and health care costs, accounting for nearly two-thirds mellitus, which occur together more often than by chance alone,
of deaths worldwide. Cardiovascular disease is the leading cause including high blood pressure, low fasting HDL cholesterol, high
of death in the United States and globally (1, 2). fasting TGs, high fasting blood glucose, and abdominal obesity
(3). The consumption of dairy products and associated nutrients
1
Supported by the Brazilian Ministry of Health (Science and Technology has been suggested to improve MetSyn characteristics (4, 5).
Department) and the Brazilian Ministry of Science, Technology and Innovation
Evidence that supports a beneficial effect of dairy products on
(Funding Authority for Studies and Projects and the National Council for Scientific
and Technological Development) grants 01 06 0010.00 RS, 01 06 0212.00 BA, 01 CVD risk factors came initially from the Dietary Approaches to
06 0300.00 ES, 01 06 0278.00 MG, 01 06 0115.00 SP, 01 06 0071.00 RJ, and Stop Hypertension trial (6). A meta-analysis of randomized
postdoctoral grant 249320/2013.
2
Author disclosures: M Drehmer, MA Pereira, MI Schmidt, S Alvim, PA Lotufo,
9
VC Luft, and BB Duncan, no conflicts of interest. Abbreviations used: CVD, cardiovascular disease; ELSA-Brasil, Brazilian
* To whom correspondence should be addressed. E-mail: michele.drehmer@ Longitudinal Study of Adult Health; MetScore, metabolic risk score; MetSyn,
ufrgs.br. metabolic syndrome.
Data collection and key measurements Subclinical measures. Participants arrived at the clinic after an
overnight fast of 8 to 15 h and were instructed to avoid heavy physical
Dairy intake and other diet variables. A validated FFQ with 114 items activity during the previous day. Blood was drawn by venipuncture, and
was administered to evaluate the participants usual intake over the past a standard 2-h 75-g oral glucose tolerance test was administered for all
12 mo (13). For each food item ascertained, the ELSA-Brazil FFQ participants who did not report a diagnosis or previous treatment for
included measures of portions and frequency of consumption, the latter diabetes. Glucose was measured by an ADVIA 1200 chemistry hexoki-
of which had 8 response options: >3 times/d, 23 times/d, 1 time/d, 56 nase system (Siemens). Glycated hemoglobin was measured using an
times/wk, 24 times/wk, 1 time/wk, 13 times/mo, and never/almost HPLC assay (Bio-Rad D-10 Dual Program Laboratories), which is
never. certified by the National Glycohemoglobin Standardization Program.
The following items were included as dairy products: milk (skimmed We calculated the insulin sensitivity index composite with the
milk, low-fat milk, whole milk), yogurt (regular, low-fat), cheese (regu- formula 10; 000=Offasting glucosemg=dL3fasting insulinmU=mL3
lar, low-fat), cheese spread, cream cheese base (regular, low-fat), ice mean glucose3mean insuling and used fasting and 2-h values for the
cream, desserts made with milk (pudding, mousses), and butter. Mixed means (15).
dishes and products for which the contribution of dairy products was HDL cholesterol and TGs were estimated using enzymatic pro-
deemed negligible were not included in the derivation of our dairy intake cedures (ADVIA 1200). Low HDL cholesterol was defined by sex:
estimates. Total dairy intake was calculated as the sum of all dairy pro- men <1.03 mmol/L (40 mg/dL) and women <1.29 mmol/L (50 mg/dL).
ducts reported in the dietary questionnaire. We computed servings per Hypertriglyceridemia was considered when $1.69 mmol/L (150 mg/dL).
day for total dairy intake and the following dairy subgroups: full- and
low-fat dairy, milk, cheese, yogurt, cheese spread and cream cheese base, Metabolic syndrome definition. For diagnosing MetSyn, we used the
dairy-based desserts, butter, and fermented dairy. The amount of grams joint interim statement consensus criteria (3), which require the presence
per serving for each dairy food was 240 g for milk, 120 g for yogurt, 30 g of any 3 of the following 5 risk factors: elevated waist circumference
2 of 9 Drehmer et al.
($102 cm in men and $88 cm in women), elevated TGs or drug higher than low-fat consumption, corresponding to 189 and 155 g/d,
treatment ($150 mg/dL), reduced HDL cholesterol or drug treatment respectively. Mean dairy food subgroup intakes were 241 g/d for
(<40 mg/dL for men and <50 mg/dL for women), elevated blood milk, 38 g/d for cheese, 37.8 g/d for yogurt, 5.2 g/d for cream
pressure or drug treatment ($130 mm Hg and/or diastolic blood pressure cheese, 1.6 g/d for butter, 12 g/d for ice cream, and 8 g/d for dairy
$85 mm Hg), and elevated fasting glucose ($100 mg/dL), which included
desserts.
those classified as having diabetes as ascertained simply by laboratory
measures at baseline.
Table 1 presents baseline characteristics for the overall sample
In addition, a metabolic risk score (MetScore) was computed using and is stratified by categories of dairy consumption. Dairy intake
an approach similar to that previously reported (16) as the mean of was higher in whites than in other ethnic groups and in
z scores of the continuous metabolic risk factors (waist circumference, participants with a university degree than in those with less
systolic blood pressure, HDL cholesterol (negative z score), TGs, and formal education. Higher dairy consumption was also observed
fasting glucose as follows: among nonsmokers and those who drank less alcohol, did more
moderate and vigorous physical activity, and ate fruits and
MetScore Zwaist 1 ZSBP 2 ZlnHDL-C 1 ZlnTriglycerides 1 ZFasting glucose 5 2
vegetables every day. Values of proteins, total fat, and saturated
HDL cholesterol and TGs were log-transformed to achieve normality fat intake, when expressed as a percentage of energy intake,
before computing z scores. The z scores for waist circumference and increased according to higher dairy product consumption.
HDL cholesterol were sex-specific. We also computed a second MetScore Means and frequencies of cardiometabolic risk factors for the
using log-transformed values of 2 h of postload glucose instead of fasting overall sample and by dairy servings per day are also presented in
glucose. Table 1. Plasma glucose and blood pressure values declined with a
greater consumption of dairy products. Thus, there were less cases
Covariates: demographics, socioeconomics, and behaviors. Inter-
views and examinations were conducted by trained health professionals
of diabetes (as ascertained solely by study-determined laboratory
with strict quality control. Standardized questionnaires provided infor- values) and hypertension in participants with higher dairy con-
mation on demographics (age, sex, race, educational level, family sumption. Although TG concentrations were lower in participants
with high dairy consumption, those of HDL cholesterol were
Age, y 50.7 6 8.7 50.6 6 8.0 50.5 6 8.5 50.6 6 8.7 51.0 6 8.9
Sex, n (%)
Men 4445 (45.2) 634 (14.3) 951 (21.4) 1581 (35.6) 1279 (28.8)
Women 5390 (54.8) 404 (7.5) 1005 (18.6) 2182 (40.5) 1799 (33.4)
Self-identified skin color/race category, n (%)
White 5220 (51.7) 394 (7.8) 865 (17.1) 2004 (39.6) 1796 (35.5)
Black 1532 (15.7) 226 (14.8) 370 (24.2) 544 (35.5) 392 (25.6)
Brown (pardo or mixed) 2828 (29.0) 358 (12.7) 634 (22.4) 1048 (37.1) 788 (27.9)
Asian 218 (2.2) 29 (13.3) 46 (21.1) 95 (43.6) 48 (22.0)
Indigenous 102 (1.0) 18 (17.6) 23 (22.5) 37 (36.3) 24 (23.5)
Educational level, n (%)
Complete secondary school 3619 (35.8) 478 (13.6) 813 (23.1) 1312 (37.3) 910 (25.9)
University degree 5414 (53.5) 347 (6.6) 857 (16.2) 2123 (40.2) 1955 (37.0)
Family history of diabetes, n (%) 3577 (36.4) 415 (11.6) 721 (20.2) 1340 (37.5) 1101 (30.8)
Current smoker, n (%) 1243 (12.4) 201 (16.2) 292 (23.5) 416 (33.5) 334 (26.9)
Alcohol intake, g ethanol/d 67 6 130 107 6 202 69.7 6 136 60.9 6 113 59.4 6 112
Physical activity (moderate and vigorous), min/wk 426 6 926 304 6 783 339 6 779 448 6 946 496 6 1019
4 of 9 Drehmer et al.
TABLE 2 Estimated mean systolic and diastolic blood pressure, fasting glucose, 2-h postload glucose, TGs, HDL cholesterol, and
waist circumference according to dairy servings per day of Brazilian adults: ELSA-Brasil 200820101
SBP, mm Hg
Model 1 128 6 1.22 127 6 1.19 126 6 1.17 125 6 1.17 20.91 6 0.16 ,0.001
Model 2 127 6 1.24 126 6 1.21 124 6 1.19 123 6 1.21 21.18 6 0.17 ,0.001
DBP, mm Hg
Model 1 80.6 6 0.84 80.2 6 0.81 79.3 6 0.80 79.0 6 0.80 20.58 6 0.11 ,0.001
Model 2 80.3 6 0.83 79.7 6 0.81 78.8 6 0.79 78.2 6 0.80 20.73 6 0.11 ,0.001
Fasting glucose, mg/dL
Model 1 110 6 1.36 110 6 1.32 109 6 1.30 109 6 1.30 20.38 6 0.18 0.032
Model 2 111 6 1.38 111 6 1.35 110 6 1.33 110 6 1.35 20.21 6 0.20 0.29
2-h postload glucose, mg/dL
Model 1 134 6 3.58 135 6 3.49 134 6 3.43 131 6 3.44 21.19 6 0.47 0.012
Model 2 135 6 3.59 135 6 3.49 135 6 3.45 132 6 3.49 21.15 6 0.50 0.021
TGs, mg/dL (n = 8644)
Model 1 165 6 7.65 153 6 7.44 149 6 7.32 145 6 7.34 25.62 6 1.01 ,0.001
Model 2 178 6 7.70 167 6 7.50 165 6 7.41 160 6 7.51 25.18 6 1.11 ,0.001
HDL cholesterol, mg/dLmen (n = 4382)
Model 1 56.1 6 1.46 54.9 6 1.44 55.2 6 1.42 54.8 6 1.40 20.30 6 0.18 0.10
(20.132 6 0.04, P = 0.001), and we found a similar association We observed a strong inverse association between full-fat
between full-fat dairy and MetScore (20.123 6 0.03, P < dairy and MetScore (difference of 20.062 6 0.02 servings per
0.001). The findings were consistent across categories of sex and day, P = 0.002) when analyzing subgroups of dairy consump-
race in a stratified analysis. tion. This association was also present for butter, yogurt, and
Despite a strong association between continuous MetScore fermented dairy. For dairy-based desserts, we found a positive
and dairy products, when using the current MetSyn criteria of association with MetScore (0.169 6 0.07, P = 0.023) (Table 5).
the joint interim statement consensus (3), we found statistically Several nutrients were evaluated as possible mediators of
significant associations only in the model adjusted by demo- dairy consumption and MetScore. These nutrients included
graphics, caloric intake, smoking, alcohol, and physical activity animal protein, total FAs, and SFAs from dairy products
(consumption of >4 dairy servings per day represented a 27% (including specific SFAs) and butyric, caproic, caprylic, capric,
lower odds of MetSyn; 95% CI: 0.62, 0.87). In the fully adjusted lauric, myristic, palmitic, stearic, palmitoleic/monounsaturated,
model, we did not observe a statistically significant association and polyunsaturated acids when expressed as a percentage of
(OR: 0.90; 95% CI: 0.75, 1.08). However, when we analyzed energy intake. We also considered as possible mediators choles-
the relation between total dairy intake and a high MetScore terol, lactose, sucrose, vitamins A and D, calcium, phosphorus,
(above the 75th percentile), participants with a reported intake magnesium, sodium, and potassium. After adjusting addition-
of >4 dairy servings per day had 25% lower odds (OR: 0.75; ally for saturated fat from dairy products (Table 6), associations
95% CI: 0.61, 0.92) of MetSyn in the fully adjusted model of total dairy (20.023 6 0.03, P = 0.37) and full-fat dairy
(Table 4). (20.025 6 0.02, P = 0.33) intake with MetScore were no longer
Dairy intake and metabolic syndrome in ELSA-Brasil 5 of 9
TABLE 3 Estimated mean MetScore according to type and intake of dairy servings per day of Brazilian adults: ELSA-Brasil
200820101
Total dairy
MetScore5
Model 1 0.99 6 0.24 0.88 6 0.23 0.69 6 0.23 0.58 6 0.23 20.053 6 0.01 ,0.001
Model 2 1.03 6 0.24 0.94 6 0.23 0.82 6 0.23 0.69 6 0.23 20.044 6 0.01 0.009
Full-fat dairy
MetScore
Model 1 0.95 6 0.23 0.83 6 0.23 0.69 6 0.23 0.43 6 0.24 20.155 6 0.03 ,0.001
Model 2 0.98 6 0.23 0.86 6 0.23 0.76 6 0.23 0.57 6 0.24 20.126 6 0.03 ,0.001
Low-fat dairy
MetScore
Model 1 0.83 6 0.23 0.86 6 0.23 0.71 6 0.24 0.59 6 0.25 20.071 6 0.03 0.025
Model 2 0.87 6 0.23 0.94 6 0.23 0.82 6 0.24 0.79 6 0.26 20.022 6 0.03 0.49
1
ELSA-Brasil, Brazilian Longitudinal Study of Adult Health; MetScore, metabolic risk score; sv, serving.
2
Adjusted through ANCOVA for each of the following variables: model 1: demographics [age (y), sex, race, occupational status, education, family income [amount], study center,
menopausal status, and family history of diabetes]; model 2: model 1 plus behaviors [current and previous smoking status, alcohol intake (grams of ethanol per day), physical
activity (metabolic equivalent min/wk)], and nondairy variables] (nondairy dietary variables included in models vary across dairy type because only variables with P , 0.20 in the final
model were maintained); total dairy: calorie intake (kcal/d) and nondairy food groups [fruits, vegetables, whole and refined grains, tea, sodas, juice, unprocessed red and white
present. In each of these models, greater SFA intake from dairy Men and women who consumed >4 servings of dairy products
products was associated with a lower MetScore (P < 0.001). These per day had lower levels of systolic and diastolic blood pressure,
findings suggest potential mediation through SFAs consumed in 2-h postload glucose, and TGs. There was a positive monotonic
dairy products. No other potential mediators were found. trend between the intake of dairy products and adjusted mean
HDL cholesterol in women only. We also found evidence that
SFAs from dairy products may mediate these associations.
Discussion
Our results are consistent with previous prospective and
After adjusting for demographics, menopausal status, family cross-sectional research (4, 1821). The protective effect of dairy
history of diabetes, dietary intake, nondietary lifestyle factors, intake in relation to MetSyn has been reported previously in
and BMI in this large cohort study of Brazilian adults, we multicenter prospective cohort studies. In the Atherosclerosis
observed a strong inverse association between total dairy intake Risk in Communities study, a cohort of 9514 participants from
and cardiometabolic risk with a linear dose-response pattern. 4 US communities, individuals in the highest quintile of dairy
Interestingly, this association was observed for full-fat dairy consumption had a 13% (CI 95%: 0.77, 0.98) lower risk of
products, butter, and yogurt but not for low-fat dairy products. developing MetSyn, and this finding was consistent across sex
TABLE 4 Adjusted associations of MetSyn with total dairy intake of Brazilian adults: ELSA-Brasil
200820101
JIS
Model 1 1.00 0.92 (0.78, 1.08) 0.82 (0.70, 0.96) 0.73 (0.62, 0.87) ,0.001
Model 2 1.00 0.95 (0.80, 1.13) 0.91 (0.77, 1.07) 0.90 (0.75, 1.08) 0.26
High MetScore2
Model 1 1.00 0.91 (0.75, 1.10) 0.76 (0.64, 0.91) 0.65 (0.53, 0.78) ,0.001
Model 2 1.00 0.93 (0.77, 1.12) 0.83 (0.69, 0.99) 0.75 (0.61, 0.92) 0.002
1
Adjusted through multiple logistic regression for the following variables: model 1: demographic characteristics, including age (y), sex, race,
occupational status, education, family income (amount), study center, menopausal status, family history of diabetes, current and previous
smoking status, alcohol intake (grams of ethanol per day), physical activity (metabolic equivalent min/wk), and calorie intake (kcal/d); model
2: model 1 + nondairy food groups [fruit, vegetables, refined grains, sodas, and processed and unprocessed red and white meat (g/d)].
ELSA-Brasil, Brazilian Longitudinal Study of Adult Health; JIS, joint interim statement definition (3); MetScore, metabolic risk score; MetSyn,
metabolic syndrome; sv, serving.
2
High MetScore = mean z score $75th percentile.
6 of 9 Drehmer et al.
TABLE 5 Adjusted difference in MetScore for a 1-unit differ- between dairy products and MetSyn and diabetes risk (25).
ence in servings per day in the intake of each dairy subgroup of Similar to our study, the Epidemiological Study on the Insulin
Brazilian adults: ELSA-Brasil 200820101 Resistance Syndrome also found no association between cheese
intake and MetSyn or diabetes. By contrast, some studies have
Dairy subgroup Difference mean 6 SE P value
suggested positive or null associations between the intake of
Full-fat dairy 20.062 6 0.02 0.002 dairy products and MetSyn or CVD risk (2628).
Low-fat dairy 0.002 6 0.02 0.92 Because of the somewhat inconsistent evidence, meta-
Milk 0.00 6 0.03 0.76 analyses can be helpful. The relative risk of MetSyn when
Cheese 20.02 6 0.02 0.29 combining results of the 4 observational studies in subjects with
Yogurt 20.15 6 0.06 0.024 high milk or dairy consumption was 0.74 (95% CI: 0.64, 0.84)
Cream cheese 20.070 6 0.08 0.41 relative to the risk in those with low consumption (29). Data
Desserts 0.169 6 0.07 0.023 from a dose-response meta-analysis that included 13,518
Butter 20.129 6 0.04 0.003 participants and 2283 CVD (fatal and nonfatal) cases were
Fermented dairy 20.05 6 0.02 0.047 analyzed in 4 other prospective cohort studies, with milk as the
main exposure. An inverse association was found between milk
1
Adjusted through multivariable linear regression for demographics variables [includ-
intake and risk of overall CVD (RR = 0.94 per 200 mL/d; 95%
ing age (y), sex, race, occupational status, education, family income (amount), and
study center], menopausal status, family history of diabetes, current and previous
CI: 0.89, 0.99 per 200 mL/d) (30). A different meta-analysis of
smoking status, alcohol intake (grams of ethanol per day), physical activity (metabolic observational studies evaluated the association between high-fat
equivalent min/wk), calorie intake (kcal/d), and nondairy food groups [fruits, vegeta- dairy products and obesity, cardiovascular disease, and MetSyn
bles, whole and refined grains, coffee, tea, sodas, juice, unprocessed and processed and found high-fat dairy intake to be inversely associated with
red and white meats, beans, eggs, and nondairy sweets and desserts (g/d)]. ELSA-
the outcomes in 11 of 16 studies (31). However, a meta-analysis
Brasil, Brazilian Longitudinal Study of Adult Health; MetScore, metabolic risk score.
of 20 mostly short-term randomized controlled studies with
1677 healthy adults found no significant effects of increased
8 of 9 Drehmer et al.
27. Snijder MB, van der Heijden AA, van Dam RM, Stehouwer CD, 35. Hu FB, Stampfer MJ, Manson JE, Ascherio A, Colditz GA, Speizer FE,
Hiddink GJ, Nijpels G, Heine RJ, Bouter LM, Dekker JM. Is higher Hennekens CH, Willett WC. Dietary saturated fats and their food
dairy consumption associated with lower body weight and fewer sources in relation to the risk of coronary heart disease in women. Am
metabolic disturbances? The Hoorn Study. Am J Clin Nutr 2007;85: J Clin Nutr 1999;70:10018.
98995. 36. Mensink RP, Zock PL, Kester AD, Katan MB. Effects of dietary fatty
28. Soedamah-Muthu SS, Verberne LD, Ding EL, Engberink MF, Geleijnse acids and carbohydrates on the ratio of serum total to HDL cholesterol
JM. Dairy consumption and incidence of hypertension. A dose-response and on serum lipids and apolipoproteins: a meta-analysis of 60
meta-analysis of prospective cohort studies. Hypertension 2012; controlled trials. Am J Clin Nutr 2003;77:114655.
60:11317.
37. Mayneris-Perxachs J, Guerendiain M, Castellote AI, Estruch R, Covas
29. Elwood PC, Givens DI, Beswick AD, Fehily AM, Pickering JE, MI, Fito M, Salas-Salvado J, Martnez-Gonzalez
MA, Aros F, Lamuela-
Gallacher J. The survival advantage of milk and dairy consumption: Raventos RM, et al. Plasma fatty acid composition, estimated desaturase
an overview of evidence from cohort studies of vascular diseases, activities, and their relation with the metabolic syndrome in a population
diabetes and cancer. J Am Coll Nutr 2008;27:723S34S.
at high risk of cardiovascular disease. Clin Nutr 2014;33:907.
30. Soedamah-Muthu SS, Ding EL, Al-Delaimy WK, Hu FB, Engberink MF,
38. Zock PL, de Vries JH, Katan MB. Impact of myristic acid versus
Willett WC, Geleijnse JM. Milk and dairy consumption and incidence
palmitic acid on serum lipid and lipoprotein levels in healthy women
of cardiovascular diseases and all-cause mortality: dose-response meta-
and men. Arterioscler Thromb 1994;14:56775.
analysis of prospective cohort studies. Am J Clin Nutr 2011;93:15871.
39. Dabadie H, Motta C, Peuchant E, LeRuyet P, Mendy F. Variations in
31. Kratz M, Baars T, Guyenet S. The relationship between high-fat dairy
consumption and obesity, cardiovascular, and metabolic disease. Eur J daily intakes of myristic and a-linolenic acids in sn-2 position modify
Nutr 2013;52:124. lipid profile and red blood cell membrane fluidity. Br J Nutr 2006;
96:2839.
32. Jakobsen MU, OReilly EJ, Heitmann BL, Pereira MA, Balter K, Fraser
GE, Goldbourt U, Hallmans G, Knekt P, Liu S, et al. Major types of 40. Samuelson G, Bratteby LE, Mohsen R, Vessby B. Dietary fat intake in
dietary fat and risk of coronary heart disease: a pooled analysis of 11 healthy adolescents: inverse relationships between the estimated intake
cohort studies. Am J Clin Nutr 2009;89:142532. of saturated fatty acids and serum cholesterol. Br J Nutr 2001;85:333
33. Siri-Tarino PW, Sun Q, Hu FB, Krauss RM. Meta-analysis of prospec- 41.
tive cohort studies evaluating the association of saturated fat with 41. Pihlanto-Leppala A, Koskinen P, Piilola K, Tupasela T, Korhonen H.