Dairy Consumption, Systolic Blood Pressure and Risk of ... · Hernandez 41, Luigi Ferrucci 42,...

Confidential: For Review O

nly

Dairy Consumption, Systolic Blood Pressure and Risk of

Hypertension: a Mendelian Randomization from 32 Studies

with 197,332 Participants

Journal: BMJ

Manuscript ID BMJ.2016.035062

Article Type: Research

BMJ Journal: BMJ

Date Submitted by the Author: 19-Aug-2016

Complete List of Authors: Ding, Ming; Harvard School of Public Health, Qi, Lu; Harvard T.H. Chan School of Public Health,

Keywords: dairy, blood pressure, hypertension, Mendelian randomization

https://mc.manuscriptcentral.com/bmj

BMJ


nly

1

Dairy Consumption, Systolic Blood Pressure and Risk of Hypertension: a

Mendelian Randomization from 32 Studies with 197,332 Participants

Ming Ding 1*, Tao Huang

1, 2, 3, 4*, Helle K.M. Bergholdt

5, 6, 7*, Alexis C. Frazier-Wood

8,

Stella Aslibekyan 9, Kari E. North

10, 11, Trudy Voortman

12, Mariaelisa Graff

10, Caren E

Smith 13

, Chao-Qiang Lai 13

, Anette Varbo 7, 14

, Rozenn N Lemaitre 15

, Ester A.L. de

Jonge 12, 16

, Frédéric Fumeron 17, 18, 19, 20

, Dolores Corella

21, 22, Carol A. Wang

23, Anne

Tjønneland 24

, Kim Overvad 25, 26

, Thorkild IA Sørensen 27, 28

, Mary F Feitosa 29

, Mary K

Wojczynski 29

, Mika Kähönen 30, 31

, Shafqat Ahmad 1, 32

, Frida Renström 32, 33

, Bruce M

Psaty 15, 20, 34, 35

, David S Siscovick 36

, Inês Barroso 37, 38, 39

, Ingegerd Johansson 40

, Dena

Hernandez 41

, Luigi Ferrucci 42

, Stefania Bandinelli 43

, Allan Linneberg 44, 45

, Camilla

Helene Sandholt 27

, Oluf Pedersen 27, 46

, Torben Hansen 27, 47

, Christina-Alexandra Schulz 48

, Emily Sonestedt 48

, Marju Orho-Melander 48

, Tzu-An Chen 8, Jerome I. Rotter

49,

Mathew A. Allison 50

, Stephen S. Rich 51

, Jose V. Sorlí 21, 22

, Oscar Coltell 22, 52

, Craig E.

Pennell 23

, Peter Eastwood 53

, Albert Hofman 12, 54

, Andre G. Uitterlinden 16

, M.Carola

Zillikens 16

, Frank J.A. van Rooij 12

, Audrey Y. Chu 55

, Lynda M. Rose 55

, Paul M Ridker 55, 56

, Jorma Viikari 57, 58

, Olli Raitakari 59, 60

, Terho Lehtimäki 61, 62

, Vera Mikkilä 60, 63

,

Walter C. Willett 1, 54, 64

, Yujie Wang 10

, Katherine L Tucker 65

, Jose M Ordovas 13, 66, 67

,

Tuomas O. Kilpeläinen 27

, Michael A Province 29

, Paul W. Franks 1, 32, 68

, Donna K Arnett 9, Toshiko Tanaka

42, Ulla Toft

44, Ulrika Ericson

48, Oscar H. Franco

12, CHARGE

consortium, Dariush Mozaffarian 69

, Frank B. Hu 1, 54, 64

, Daniel I. Chasman 55, 70, 71

,

Børge G Nordestgaard 7, 14, 72

*, Christina Ellervik 73, 74

*, and Lu Qi 1, 4

*

*Contribute equally to this work

1.Department of Nutrition, Harvard School of Public Health, Boston, MA 02115, USA.

2.Epidemiology Domain, Saw Swee Hock School of Public Health, National University

of Singapore, 117549, Singapore

3.Department of Medicine, Yong Loo Lin School of Medicine, National University of

Singapore, 117549, Singapore

4.Department of Epidemiology, School of Public Health and Tropical Medicine, Tulane

University, New Orleans, LA 70112, USA

5.Department of Clinical Biochemistry, Naestved Hospital, Denmark.

6.Department of Clinical Pharmacology, Copenhagen University Hospital Bispebjerg

Frederiksberg, Copenhagen, Denmark

7.Faculty of Health and Medical Sciences, University of Copenhagen, Denmark.

8.USDA/ARS Children's Nutrition Research Center, Baylor College of Medicine, 1100

Bates Street, Houston, TX 77071, USA

9. Department of Epidemiology, University of Alabama at Birmingham, Birmingham,

AL 35205, USA

10.Department of Epidemiology, University of North Carolina, Chapel Hill, NC, 27514,

USA

11.Carolina Center for Genome, Sciences University of North Carolina, Chapel Hill, NC,

27514, USA

Page 1 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

2

12.Department of Epidemiology Erasmus MC, University Medical Center, Rotterdam,

the Netherlands

13.Jean Mayer USDA Human Nutrition Research Center on Aging, Tufts University

Boston, MA 02111, USA

14.Department of Clinical Biochemistry and the Copenhagen General Population Study,

Herlev and Gentofte Hospital, Copenhagen University Hospital, Denmark

15.Department of Medicine, University of Washington, WA 98101, USA

16.Department of Internal Medicine Erasmus MC, University Medical Center, Rotterdam,

the Netherlands

17.INSERM UMR_S 1138, Centre de Recherche des Cordeliers, F-75006, Paris, France.

18.Univ Paris Diderot Sorbonne Paris Cité, UMR_S 1138, Centre de Recherche des

Cordeliers, F-75006, Paris, France

19.Sorbonne Universités UPMC Univ Paris 06, UMR_S 1138, Centre de Recherche des

Cordeliers, F-75006, Paris, France

20.Group Health Research Institute, Group Health Cooperative, Seattle, WA

21.Department of Preventive Medicine and Public Health, University of Valencia, 46010-

Valencia, Spain

22.CIBER Fisiopatología de la Obesidad y Nutrición, Instituto de Salud Carlos III,

Madrid Spain

23. School of Women's and Infants' Health, University of Western Australia, Australia

24.Danish Cancer Society Research Center, Copenhagen 2100, Denmark

25.Department of Public Health Section for Epidemiology, Aarhus University, DK-8000

Aarhus C, Denmark

26.Aalborg University Hospital, DK-9000 Aalborg, Denmark

27. The Novo Nordisk Foundation Center for Basic Metabolic Research, Section of

Metabolic Genetics, Faculty of Health and Medical Sciences, University of Copenhagen,

DK-2100 Copenhagen, Denmark

28.Institute of Preventive Medicine Bispebjerg and Frederiksberg Hospitals, The Capital

Region, Copenhagen 2000, Denmark

29.Department of Genetics Washington University School of Medicine, Saint Louis, MO

63108, USA

30.Department of Clinical Physiology, Tampere University Hospital, Finland

31.Department of Clinical Physiology, University of Tampere School of Medicine

32.Department of Clinical Sciences, Genetic and Molecular Epidemiology Unit, Lund

University, 20502 Malmö, Sweden

33.Department of Biobank Research, Umeå University, 90187 Umeå, Sweden

34.Department of Epidemiology, University of Washington, WA 98101, USA

35.Department of Health Sciences, University of Washington, WA 98101, USA

36.New York Academy of Medicine, New York, NY 10029, USA

37.NIHR Cambridge Biomedical Research Centre, Institute of Metabolic Science,

Addenbrooke's Hospital, Cambridge CB2 0QQ, United Kingdom

38.University of Cambridge, Metabolic Research Laboratories Institute of Metabolic

Science, Addenbrooke's Hospital, Cambridge CB2 0QQ United Kingdom

39.Wellcome Trust Sanger Institute, Wellcome Trust Genome Campus, Hinxton,

Cambridge, CB10 1SA United Kingdom

40.Department of Biobank Research, Umeå University, 90187 Umeå, Sweden

Page 2 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

3

41.Laboratory of Neurogenetics National Institute on Aging, Bethesda MD 20892

42.Translational Gerontology Branch, Baltimore, MD, 21225

43.Geriatric Unit Azienda Sanitaria Firenze (ASF), Florence, Italy

44.Research Centre for Prevention and Health, the Capital Region of Denmark,

Copenhagen, Denmark

45.Department of Clinical Experimental Research Rigshospitalet, Glostrup, Denmark

46.Faculty of Health Sciences, University of Aarhus, Aarhus, Denmark

47.Faculty of Health Sciences, University of Southern Denmark, Odense, Denmark

48.Department of Clinical Sciences in Malmö, Lund University, Sweden

49.Institute for Translational Genomics and Population Sciences, Los Angeles

Biomedical Research Institute and Department of Pediatrics at Harbor-UCLA Medical

Center, 1124 W. Carson St, Torrance, CA 90502, USA

50.Division of Preventive Medicine, Department of Family Medicine and Public Health,

University of California San Diego, 9500 Gilman Drive, La Jolla, CA 92093, USA

51.Center for Public Health Genomics, 3232 West Complex, University of Virginia,

Charlottesville, VA, USA

52.Department of Computer Languages and Systems, University Jaume I Castellon,

Spain

53. Centre for Sleep Science, School of Anatomy Physiology and Human Biology,

University of Western Australia, Australia

54.Department of Epidemiology, Harvard School of Public Health, Boston, MA 02115,

USA

55.Division of Preventive Medicine, Brigham and Women's Hospital and Harvard

Medical School, Boston, MA 02215 USA

56.Division of Cardiovascular Medicine, Brigham and Women's Hospital and Harvard

Medical Schoo, Boston, MA 02115 USA

57.Division of Medicine, Turku University Hospital, Finland

58.Department of Medicine, University of Turku, Finland

59.Department of Clinical Physiology and Nuclear Medicine, Turku University Hospital,

Finland

60.Research Centre of Applied and Preventive Cardiovascular Medicine, University of

Turku, Finland

61.Department of Clinical Chemistry Fimlab Laboratories, Tampere University Hospital,

Finland

62.Department of Clinical Chemistry, University of Tampere School of Medicine,

Finland

63.Department of Food and Environmental Sciences, University of Helsinki, Finland

64.Channing Division of Network Medicine, Department of Medicine, Brigham and

Women’s Hospital and Harvard Medical School, Boston, MA 02115, USA

65.Clinical Laboratory & Nutritional Sciences Center for Population Health & Health

Disparities, Center for Gerontology Research & Partnerships, University of

Massachusetts, Lowell, MA, USA

66.Department of Epidemiology and Population Genetics, Centro Nacional Investigación

Cardiovasculares (CNIC), Madrid, Spain

67.Instituto Madrileño de Estudios Avanzados en Alimentación Madrid, Spain

Page 3 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

4

68.Department of Public Health and Clinical Medicine Section for Medicine, Umeå

University, 90187 Umeå, Sweden

69.Friedman School of Nutrition Science & Policy, Tufts University, Boston, MA 02111,

USA

70.Division of Genetics, Brigham and Women's Hospital and Harvard Medical School,

Boston MA 02115, USA

71.Broad Institute of MIT and Harvard, Cambridge MA 02142, USA

72.The Copenhagen City Heart Study, Frederiksberg Hospital, Copenhagen University

Hospital, Denmark

73.Department of Research, Nykoebing, Falster Hospital, Nykoebing Falster, Denmark

74.Department of Laboratory Medicine, Boston Children’s Hospital, Boston, MA, USA

Correspondence and requests for reprint:

Dr. Lu Qi.

Department of Epidemiology,

School of Public Health and Tropical Medicine,

Tulane University,

1440 Canal St, New Orleans, LA 70112

Telephone: 504-988-3549;

Email: [email protected]; [email protected]

Page 4 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

5

ABSTRACT

Background: Dairy intake has been inversely related to systolic blood pressure (SBP)

and risk of hypertension in observational studies. However, whether such associations are

causal remains inconclusive.

Method: We conducted a Mendelian randomization study using SNP rs4988235 related

to lactase persistence as an instrumental variable (IV). We collected data from 22 studies

with 171,213 participants and additionally included 10 published prospective studies with

26,119 participants into the observational analysis. The IV estimation was conducted

using the ratio of coefficients approach. We further summarized eight published clinical

trials (RCTs) on dairy consumption with SBP.

Results: Per T allele increase in LCT-13910 rs4988235 was associated with higher dairy

consumption (0.08 serving/day; 95% CI: 0.06, 0.11), and was associated with higher SBP

(0.17 mmHg; 95% CI: 0.01, 0.33) but not risk of hypertension (OR = 1.00; 95% CI: 0.98,

1.02). Using LCT-13910 rs4988235 as IV, we found that genetically determined dairy

consumption was associated with higher SBP (β = 2.13 mmHg per serving/day; 95% CI:

0.02, 4.23) but not risk of hypertension (OR = 1.00; 95% CI: 0.78, 1.28). Moreover, our

meta-analysis of the published RCTs showed that higher dairy intake has no significant

effect on change of SBP over 1 to 12 month interventions (comparing intervention to

control groups: β = -0.21 mmHg; 95% CI: -0.98, 0.57). In observational analysis, each

serving/day increase of dairy consumption was associated with -0.11 mmHg (95% CI: -

0.20, -0.02) lower SBP but not risk of hypertension (OR = 0.98; 95% CI: 0.97, 1.00).

Page 5 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

6

Conclusion: The weak inverse association between dairy intake and SBP in

observational studies was not supported by our comprehensive IV analysis and

systematic review of existing RCTs.

Page 6 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

7

INTRODUCTION

Elevated blood pressure (BP) is an important risk factor of cardiovascular disease (CVD)

and has been the top single contributor to the global burden of morbidity and mortality,

leading to 9.4 million deaths each year 1. In clinical trials, lowering blood pressure has

been demonstrated to be an effective strategy to reduce CVD incidence 2, with each

5mmHg reduction in BP associated with 20 % lower risk of coronary heart disease and

29 % lower risk of stroke 3.

Maintaining a healthful diet is critical for the prevention of hypertension 4, however,

whether dairy products should be incorporated into a healthful diet is controversial. In

epidemiological studies, the association of dairy consumption with blood pressure has

been inconsistent. Several observational studies have reported inverse associations of

dairy consumption with systolic blood pressure (SBP) and risk of hypertension 5-7

,

however, such associations were not observed in other studies 8-10

. Two meta-analyses of

prospective cohort studies consistently indicated that dairy consumption was associated

with lower SBP and lower risk of hypertension 11 12

. However, due to the observational

nature of the studies included, the reported associations may not indicate causality.

In recent years, Mendelian randomization analysis has been widely used to assess

potential causal estimates of various risk factors with health outcomes. This approach has

the advantage over traditional observational studies in minimizing confounding by using

genetic markers as instrumental variables (IV) of environmental risk factors. A SNP

rs4988235 upstream from the lactase persistence gene (LCT-13910) has been consistently

related to dairy intake across multiple populations 13 14

, representing a strong IV for

analyzing the causal relation between dairy intake and disease risk.

Page 7 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

8

In this study, using data collected from 32 studies with 197,332 participants, we

performed an IV analysis to examine the possible causal estimate of dairy consumption

on SBP and risk of hypertension. In addition, we conducted a meta-analysis to summarize

results of eight randomized clinical trials (RCTs) assessing dairy intake intervention on

changes of SBP.

METHODS

Study design and population

We used an IV approach to examine associations of dairy consumption with SBP and risk

of hypertension. We collected data from 22 observational studies with 171,213

participants within the CHARGE (Cohorts for Heart and Aging Research in Genomic

Epidemiology) consortium. All participants provided written informed consent, and all

participating studies received approval from local research ethics committees.

Description of all the studies in the analysis is in the appendix (supplemental pages 1 -

5).

To provide comprehensive evidence on associations of dairy intake with SBP and

risk of hypertension, we conducted a systematic review of previously published cohort

studies and RCTs. In the appendix, we describe the procedure of the systematic review

(supplemental pages 6), and show the flow chart of study selection (Supplemental

Figure 1).

Patients' involvement

No patient was involved in our study.

Dairy consumption

Page 8 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

9

Dairy product was measured by questionnaire, and included ‘skim/low fat milk’, ‘whole

milk’, ‘ice cream’, ‘yogurt’, ‘cottage/ricotta cheese’, ‘cream cheese’, ‘other cheese’, and

‘cream’. Total dairy consumption was calculated as sum of all dairy categories. The

detailed description of dairy consumption measurement of included studies is shown in

Supplemental table 1.

Outcome measures

Given that SBP is superior to diastolic blood pressure (DBP) as a major risk factor of

CVD, SBP was used as the main outcome in our analysis. The measurement of SBP is

presented in Supplemental table 1. For participants taking antihypertensive medication,

we added 15 mmHg to SBP to adjust for treatment effect. Hypertension was defined as

SBP of 140 mmHg or higher or current use of antihypertensive medication.

Genotype rs4988235

Genotyping platforms, genotype frequencies, Hardy Weinberg equilibrium P values, and

call rates for lactase persistence SNP rs4988235 are shown in Supplemental table 1. The

SNP rs4988235 was not genotyped or imputed in two studies, in which proxy SNPs

(rs309137: r2 = 0.77; rs1446585: r

2 = 1.00) were used instead.

Statistical analysis

Statistical analyses were initially conducted within each included study in accordance

with a standard analysis plan. We used additive models to examine associations of dairy

consumption with SBP and risk of hypertension adjusting for baseline age, sex, ethnicity,

and region. We examined associations of dairy consumption with SBP and risk of

hypertension using linear/logistic models adjusting for baseline age, body mass index

(BMI), BP, smoking status, physical activity, total energy intake, alcohol consumption,

Page 9 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

10

sex, ethnicity, region, and years of follow-up. After results were obtained from each

study within CHARGE, we combined the collected results with results extracted from

published cohort studies using random-effects models. Statistical heterogeneity across

studies was assessed by Cochrane Q test, with P < 0.1 indicating significant between-

study heterogeneity. In addition, we calculated the I2 statistic to evaluate the percentage

of heterogeneity that was due to between-study variation 15

.

We used instrumental variable (IV) ratio method to estimate the possible causal

relationship of dairy consumption with SBP and risk of hypertension. The variance of the

IV ratio was estimated using first-order Taylor expansion 16

.

We further conducted stratified analysis on the causal estimates of dairy intake

with SBP and risk of hypertension by frequency of CC alleles, ethnicity, country, study

design, and measurement of SBP. We used meta-regression to evaluate effect

modification by each study-level characteristics. In sensitivity analyses, we repeated our

analyses using dominant (CC vs. CT/TT) and recessive models (CC/CT vs. TT). All

meta-analyses were conducted at Harvard T.H. Chan School of Public Health using Stata

version 11.2 (STATA Corp, College Station, Texas)

RESULTS

We included 22 studies with 171,213 participants from the CHARGE consortium, and

baseline characteristics of the studies are shown in Table 1. Of the 22 studies, 9 were

conducted in the U.S. and 12 were conducted in European countries. Participants were

Caucasian in 18 studies. Dairy intake and SBP were measured prospectively in most of

the studies. The frequency of CC alleles varied across studies, with the lowest frequency

in European countries.

Page 10 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

11

By conducting a systematic review, we additionally identified 10 published cohort

studies with 26,119 participants (4-9, 23-26) and eight clinical trials with 735 participants.

The RCTs examined the effect of dairy intake on SBP with 1-12 months intervention 17-24

.

Of the 10 cohort studies, seven studies assessed SBP as the outcome 5-10 25 26

, and five

studies used hypertension as the outcome 5 25-28

. The characteristics of previous

publications were shown in Supplemental Tables 2, 3.

In the analysis including 171,213 participants in the CHARGE consortium, per T

allele increase in LCT-13910 rs4988235 was associated with higher dairy consumption

(0.08 serving/day; 95% CI: 0.06, 0.11). However, significant heterogeneity was found

across studies (P < 0.001) (Figure 1). Per T allele increase in LCT-13910 rs4988235 was

associated with higher SBP (0.17 mmHg; 95% CI: 0.01, 0.33) but not risk of

hypertension (OR = 1.00; 95% CI: 0.98, 1.02) (Figure 2). Using LCT-13910 rs4988235

as IV, we found that genetically determined dairy consumption was associated with

higher SBP (β = 2.13 mmHg per serving/day; 95% CI: 0.02, 4.23) but not risk of

hypertension (OR = 1.00; 95% CI: 0.78, 1.28).

In observational analysis combining studies within CHARGE consortium and

published observational studies, each serving/day increase in dairy consumption was

associated with lower SBP (β = -0.11 mmHg; 95% CI: -0.20, -0.02) but was not

associated with lower risk of hypertension (OR = 0.98; 95% CI: 0.97, 1.00) (Figures 3a,

3b). In addition, dairy intake did not show significant effect on changes of SBP over 1-12

months of interventions (comparing intervention to control group: β = -0.21 mmHg; 95%

CI: -0.98, 0.57) (Figure 3c). No publication bias of included RCTs was found.

Page 11 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

12

In stratified analysis, genetically determined dairy consumption was associated

with higher SBP in studies conducted in Caucasians (β = 2.38 mmHg per serving/day; 95%

CI: 0.19, 4.56) and in studies with clinical measurement of SBP (β = 2.75 mmHg per

serving/day; 95% CI: 0.09, 5.41) (Table 2). No significant effect modification on the

causal estimations was found by the stratification variables.

In sensitivity analyses, we examined the associations of dairy consumption with

SBP and risk of hypertension by modeling the LCT-13910 genotype in recessive and

dominant inheritance manner. Using recessive or dominant models, SNP rs4988235 was

not associated with SBP or risk of hypertension (Supplemental Figures 2, 3), and

genetically determined dairy consumption was not associated with SBP or risk of

hypertension (Supplemental Table 4).

DISCUSSION

In this study, using Mendelian randomization analysis in 32 studies with 197,332

participants, we examined the potential causal effect of dairy consumption on SBP and

risk of hypertension. Using the LCT-13910 variant affecting lactase persistence as the

instrumental variable, our study showed that genetically determined dairy intake slightly

increased SBP and did not effect risk of hypertension. Furthermore, a meta-analysis of

the results from published RCTs showed that dairy consumption had no effect on changes

of SBP in response to 1 to 12 months of intervention.

Dairy products are widely consumed in the U.S. and European countries, and the

association between dairy intake and blood pressure has been examined in several cross-

sectional 29-31

and prospective cohort studies (4-9, 23-26). An inverse association between

Page 12 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

13

dairy intake and SBP was found in all of the three cross-sectional studies. The results

from cohort studies have been summarized in two meta-analyses. One meta-analysis

involving approximately 45,000 subjects and 11,500 cases of elevated blood pressure

showed that dairy products were associated with lower risks of elevated pressure 11

. In

line with this, another meta-analysis, which included 9 cohort studies with a sample size

of 57,256 and a total of 15,367 incident hypertension cases, found an inverse association

between dairy foods and risk of hypertension 12

.

In addition to considering dairy products as one food group, dairy products have

been categorized into high-fat dairy including whole milk, cream, and cream cheese and

low-fat dairy including skim milk and yogurt. Although in observational studies an

inverse association between overall dairy intake and risk of hypertension/elevated BP

was found, the associations of high-fat and low-fat dairy with blood pressure were

inconsistent. In the two meta-analyses aforementioned, the observed inverse association

was mainly due to consumption of low-fat dairy products 11 12

. Furthermore, one meta-

analysis summarized 14 RCTs involving 702 participants and found that probiotic

fermented milk including yogurt resulted in a significant reduction in blood pressure

comparing to placebo 32

. Regarding nutrients in dairy products, clinical trials showed that

milk-derived tripeptides and peptides have hypotensive effects in prehypertensive and

hypertensive subjects 33 34

.

In our study, we found that dairy consumption did not decrease SBP or risk of

hypertension using IV estimation. Moreover, the meta-analyzed results of RCTs showed

that dairy intake had no effect on changes of SBP. There could be several reasons that the

reported associations between dairy intake and blood pressure from observational studies

Page 13 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

14

were inconsistent with our IV results. First, higher low-fat dairy intake was more likely to

be associated with a healthy diet and lifestyle 35

. Therefore, the observed inverse

association of particularly low-fat dairy intake with SBP might be due to confounding of

intake of other food items and a healthy lifestyle. Second, even if yogurt and specific

nutrients in dairy such as milk peptides have antihypertensive effects, specific dairy

products such as yogurt only compose a small fraction of total dairy products. Therefore,

as shown in our IV results, dairy intake could still increase SBP when using overall dairy

products consumed by general population as main exposure. The saturated fat and D-

galactase with relatively high content in high-fat dairy products might be responsible for

the non-favorable association of dairy intake with blood pressure 36 37

. Third, due to the

methodology limitation that the instrumental variable is assumed to be associated with

the outcome only through the exposure under study 38

, we could not separate the effect of

individual dairy product in our study to further explain the inconsistency between

observational and instrumental results.

Given heterogeneity of the association between SNP rs4988235 and dairy intake

across studies, we further conducted stratified analysis by CC frequency, ethnicity, and

country. SNP rs4988235 consistently associated with higher dairy intake across

subgroups, demonstrating the robustness of our instrumental variable. We further found

that the positive effect of dairy intake on SBP was mainly from Caucasian living in

European countries. The reason might be that dairy consumption is high in Europe,

particularly in northern European countries 13

. The dose response relationship of dairy

intake on SBP is worth further investigation.

Page 14 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

15

Our study has several strengths. First, our study is a large IV analysis on the

causality of dairy intake on blood pressure and hypertension. The large sample size

provided us enough power to estimate the causal effect of dairy intake on blood pressure.

Second, the lactase persistence SNP is a well-established variant associated with dairy

intake with solid biological basis, and therefore a highly valid IV. Even though the SNP

might not be associated with Swiss cheese and mozzarella cheese which contain trivial

amounts of lactase, those products only composited a very low proportion of total dairy

products in the whole population. Therefore, the effect of those products on the validity

of the instrumental variable might be minimal. Third, most of the studies included were

genetically homogeneous, and we performed analysis within each study first. Therefore,

the effect of population stratification on the instrumental results within each study might

be minimal. Moreover, no significant effect modification by country, ethnicity, and CC

frequency of SNP rs4988235was found in stratified analysis. Fourth, we summarized

published clinical trials on dairy consumption with SBP, which provided further

supportive evidence to the IV results.

Our study has several limitations. First, SBP was self-reported in several studies,

resulting in measurement errors. This might be the reason that dairy consumption had no

effect on self-reported SBP in subgroup analysis. Second, dairy consumption was self-

reported by questionnaire, and might be affected by measurement errors. If measurement

errors were random, the observed associations would be biased to null. However, the IV

estimates results would not be biased, although the confidence interval might be larger.

Third, dairy consumption and SBP were examined cross-sectionally in several studies

and it might result in reverse causation even if using IV analysis. However, no significant

Page 15 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

16

effect modification by study design was found in stratified analysis, indicating that

reverse causation caused by study design might be minimal.

In conclusion, the weak association between dairy intake and SBP in observational

studies was not supported by our comprehensive IV analysis and systemic review of

existing RCTs.

FUNDING

The CGPS is was funded by The Danish Council for Independent Research; Medical

Sciences(FSS); Herlev Hospital, Copenhagen University Hospital; Copenhagen County

Foundation; and Chief Physician Johan Boserup and Lise Boserup’s Fund, Denmark.

The GESUS was funded by the Region Zealand Foundation, Naestved Hospital

Foundation. Edith and Henrik Henriksens Memorial Scholarship, Johan and Lise Boserup

Foundation, TrygFonden, Johannes Fog’s Foundation, Region Zealand, Naestved

Hospital, The National Board of Health, and the Local Government Denmark Foundation.

The WGHS is supported by HL043851 and HL080467 from the National Heart, Lung,

and Blood Institute and CA047988 from the National Cancer Institute, the Donald W.

Reynolds Foundation and the Fondation Leducq, with collaborative scientific support and

funding for genotyping provided by Amgen. The NHS and the HPFS was supported by

grants UM1 CA186107, P01 CA87969, R01 CA49449, R01 HL034594, R01 HL088521,

UM1 CA167552, R01 HL35464, HL126024, HL034594, DK100383, DK091718,

HL071981, HL073168, CA87969, CA49449, CA055075, HL34594, HL088521,

U01HG004399, DK080140, P30DK46200, U01CA137088, U54CA155626, DK58845,

DK098311, U01HG004728, EY015473, CA134958, DK70756 and DK46200 from the

National Institutes of Health, with additional support for genotyping from Merck

Research Laboratories, North Wales, PA. LQ is a recipient of the American Heart

Association Scientist Development Award (0730094N). LRP is supported by the Arthur

Ashley Williams Foundation and a Harvard Ophthalmology Scholar Award (Harvard

Medical School) from the Harvard Glaucoma Center of Excellence. ATC is a Damon

Runyon Cancer Foundation Clinical Investigator. The funding sources had no role in the

design or conduct of the study; collection, management, analysis, and interpretation of

the data; or preparation, review, or approval of the manuscript. ARIC is carried out as a

collaborative study supported by National Heart, Lung, and Blood Institute contracts

(HHSN268201100005C, HHSN268201100006C, HHSN268201100007C,

HHSN268201100008C, HHSN268201100009C, HHSN268201100010C,

HHSN268201100011C, and HHSN268201100012C), R01HL087641, R01HL59367 and

R01HL086694; National Human Genome Research Institute contract U01HG004402;

and National Institutes of Health contract HHSN268200625226C. The authors thank the

staff and participants of the ARIC study for their important contributions. Infrastructure

was partly supported by Grant Number UL1RR025005, a component of the National

Page 16 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

17

Institutes of Health and NIH Roadmap for Medical Research. The Inter99 study was

funded by the Danish Research Councils, Health Foundation, Danish Centre for

Evaluation and Health Technology Assessment, Copenhagen County, Danish Heart

Foundation, Ministry of Health and Prevention, Association of Danish Pharmacies,

Augustinus Foundation, Novo Nordisk, Velux Foundation, Becket Foundation, and Ib

Henriksens Foundation. The D.E.S.I.R. study has been supported by INSERM contracts

with CNAMTS, Lilly, Novartis Pharma and Sanofi-Aventis; by INSERM (Réseaux en

Santé Publique, Interactions entre les déterminants de la santé, Cohortes Santé TGIR

2008), the Association Diabète Risque Vasculaire, the Fédération Française de

Cardiologie, La Fondation de France, ALFEDIAM, CNIEL, ONIVINS, Société

Francophone du Diabète, Ardix Medical, Bayer Diagnostics, Becton Dickinson,

Cardionics, Merck Santé, Novo Nordisk, Pierre Fabre, Roche, Topcon. The funding

sources had no role in the design or conduct of the study; collection, management,

analysis, and interpretation of the data; or preparation, review, or approval of the

manuscript. The D.E.S.I.R. Study Group: INSERM CESP U1018: B Balkau, P

Ducimetière, E Eschwège; INSERM U367: F. Alhenc-Gelas; CHU d’Angers: A Girault;

Bichat Hospital: F Fumeron, M Marre, R Roussel; CHU de Rennes: F Bonnet; CNRS

UMR8090, Lille: S Cauchi P Froguel; Centres d’Examens de Santé: Alençon, Angers,

Blois, Caen, Chartres, Chateauroux, Cholet, Le Mans, Orléans, Tours; Institute de

Recherche Médecine Générale: J Cogneau; General practitioners of the region; Institute

inter-Regional pour la Santé: C Born, E Caces, N Copin, JG Moreau, O Lantieri, F

Rakotozafy, J Tichet, S Vol. The Rotterdam Study is funded by Erasmus Medical Center

and Erasmus University, Rotterdam, Netherlands Organization for the Health Research

and Development (ZonMw), the Research Institute for Diseases in the Elderly (RIDE),

the Ministry of Education, Culture and Science, the Ministry for Health, Welfare and

Sports, the European Commission (DG XII), and the Municipality of Rotterdam. The

authors are grateful to the study participants, the staff from the Rotterdam Study and the

participating general practitioners and pharmacists. The generation and management of

GWAS genotype data for the Rotterdam Study is supported by the Netherlands

Organisation of Scientific Research NWO Investments (nr. 175.010.2005.011, 911-03-

012). This study is funded by the Research Institute for Diseases in the Elderly (014-93-

015; RIDE2), the Netherlands Genomics Initiative (NGI)/Netherlands Organisation for

Scientific Research (NWO) project nr. 050-060-810. The MDCS was initiated and

planned in collaboration with the International Agency for Research on Cancer, the

Swedish Cancer Society, and Swedish Medical Research Council and the Faculty of

Medicine Lund University, Sweden. The study is also funded by Region Skåne, City of

Malmö, Påhlsson Foundation and the Swedish Heart and Lung Foundation.

The GLACIER Study was funded by project grants from the Swedish Heart-Lung

Foundation, the Swedish Diabetes Association, the Påhlsson’s Foundation, Region Skåne,

the Swedish Research Council, the Umeå Medical Research Foundation, Novo Nordisk,

and The Heart Foundation of Northern Sweden (all to PWF). The authors also

acknowledge the funding agencies supporting the Northern Sweden Diet Database and

the Västerbotten Intervention Project, including the Swedish Research Council.

The MESA study was supported by contracts HHSN268201500003I, N01-HC-

95159,N01-HC-95160, N01-HC-95161, N01-HC-95162, N01-HC-95163, N01-HC-

95164, N01-HC-95165, N01-HC-95166, N01-HC-95167, N01-HC-95168 and N01-HC-

Page 17 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

18

95169 from the National Heart, Lung, and Blood Institute, and by grants UL1-TR-

000040 and UL1-TR-001079 from NCRR. The FamHS was supported by grants

DK089256 and HL117078 from the National Insitutes of Health. Infrastructure for the

CHARGE Consortium is supported in part by the National Heart, Lung, and Blood

Institute grant R01HL105756. This CHS research was supported by NHLBI contracts

HHSN268201200036C, HHSN268200800007C, N01HC55222, N01HC85079,

N01HC85080, N01HC85081, N01HC85082, N01HC85083, N01HC85086; and NHLBI

grants U01HL080295, R01HL087652, R01HL105756, R01HL103612, and

R01HL120393 with additional contribution from the National Institute of Neurological

Disorders and Stroke (NINDS). Additional support was provided through R01AG023629

from the National Institute on Aging (NIA). A full list of principal CHS investigators and

institutions can be found at CHS-NHLBI.org.The provision of genotyping data was

supported in part by the National Center for Advancing Translational Sciences, CTSI

grant UL1TR000124, and the National Institute of Diabetes and Digestive and Kidney

Disease Diabetes Research Center (DRC) grant DK063491 to the Southern California

Diabetes Endocrinology Research Center. The content is solely the responsibility of the

authors and does not necessarily represent the official views of the National Institutes of

Health. The YFS has been financially supported by the Academy of Finland: grants

286284 (T.L.), 134309 (Eye), 126925, 121584, 124282, 129378 (Salve), 117787 (Gendi),

and 41071 (Skidi); the Social Insurance Institution of Finland; Kuopio, Tampere and

Turku University Hospital Medical Funds (grant X51001 for T.L.); Juho Vainio

Foundation; Paavo Nurmi Foundation; Finnish Foundation of Cardiovascular Research

(T.L.); Finnish Cultural Foundation; Tampere Tuberculosis Foundation (T.L.); Emil

Aaltonen Foundation (T.L.); and Yrjö Jahnsson Foundation (T.L.). The expert technical

assistance in the statistical analyses by Ville Aalto, Irina Lisinen and Mika Helminen is

gratefully acknowledged. The DCH and the DIOGENES cohorts were a part of the

research program of the UNIK: Food, Fitness & Pharma for Health and Disease (see

www.foodfitnesspharma.ku.dk). The UNIK project was supported by the Danish

Ministry of Science, Technology and Innovation. Tuomas O. Kilpeläinen was supported

by the Danish Council for Independent Research (DFF – 1333-00124 and Sapere Aude

program grant DFF – 1331-00730B). The PREDIMED-VALENCIA study was supported

by the Spanish Ministry of Health (Instituto de Salud Carlos III) and the Ministerio de

Economía y Competitividad (projects G03/140, CIBER 06/03, RD06/0045 PI07-0954,

CNIC-06, PI11/02505, SAF2009-12304, AGL2010-22319-C03-03 and PRX14/00527),

Fondo Europeo de Desarrollo Regional, by the University Jaume I (Project P1-1B2013-

54) and by the Generalitat Valenciana (AP111/10, AP-042/11, BEST/2015/087,

GVACOMP2011-151, ACOMP/2011/145, ACOMP/2012/190 and ACOMP/2013/159).

The BPRHS was supported by the National Institutes of Health grants P01 AG023394

and P50 HL105185. The GOLDN Study was supported by National Heart, Lung, and

Blood Institute (NHLBI) grant no. U01HL072524 (Genetic and Environmental

Determinants of Triglycerides), NHLBI R01 HL091357 (Genomewide Association Study

of Lipid Response to Fenofibrate and Dietary Fat), NHLBI grant number HL54776 and

HL078885; and by contracts 53-K06-5-10 and 58-1950-9-001 from the US Department

of Agriculture, Agriculture Research Service. The Raine Study was supported by the

National Health and Medical Research Council of Australia [grant numbers 403981 and

Page 18 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

19

003209] and the Canadian Institutes of Health Research [grant number MOP-82893]. The

authors are grateful to the Raine Study participants and their families, and to the Raine

Study research staff for cohort coordinaion and data collection. The authors gratefully

acknowledge the NH&MRC for their long term funding to the study over the last 25

years and also the following institutes for providing funding for Core Management of the

Raine Study: The University of Western Australia (UWA) , Curtin University, the Raine

Medical Research Foundation, the UWA Faculty of Medicine, Dentistry and Health

Sciences, the Telethon Kids Institute, the Women's and Infant's Research Foundation

(King Edward Memorial Hospital) and Edith Cowan University). The authors gratefully

acknowledge the assistance of the Western Australian DNA Bank (National Health and

Medical Research Council of Australia National Enabling Facility). We would also like

to acknowledge the Raine Study participants for their ongoing participation in the study,

the Raine Study Team for study co-ordination and data collection, the UWA Centre for

Science for utilisation of the facility and the Sleep Study Technicians. The 22 year Raine

Study follow-up was funded by NHMRC project grants 1027449, 1044840 and 1021855.

Funding was also generously provided by Safework Australia. The InCHIANTI study

baseline (1998-2000) was supported as a "targeted project" (ICS110.1/RF97.71) by the

Italian Ministry of Health and in part by the U.S. National Institute on Aging (Contracts:

263 MD 9164 and 263 MD 821336).

Competing interests

All authors have completed the ICMJE uniform disclosure form at

www.icmje.org/coi_disclosure.pdf and declare: no support from any organisation for the

submitted work; no financial relationships with any organisations that might have an

interest in the submitted work in the previous three years; no other relationships or

activities that could appear to have influenced the submitted work.

Licence for publication

I, Lu Qi, The Corresponding Author of this article contained within the original

manuscript which includes any diagrams & photographs within and any related or stand

alone film submitted (the Contribution”) has the right to grant on behalf of all authors

and does grant on behalf of all authors, a licence to the BMJ Publishing Group Ltd and its

licencees, to permit this Contribution (if accepted) to be published in the BMJ and any

other BMJ Group products and to exploit all subsidiary rights, as set out in our licence set

out at: http://www.bmj.com/about-bmj/resources-authors/forms-policies-and-

checklists/copyright-open-access-and-permission-reuse.”

Page 19 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

20

Reference

1. Lim SS, Vos T, Flaxman AD, Danaei G, Shibuya K, Adair-Rohani H, et al. A comparative risk

assessment of burden of disease and injury attributable to 67 risk factors and risk factor

clusters in 21 regions, 1990-2010: a systematic analysis for the Global Burden of Disease

Study 2010. Lancet 2012;380(9859):2224-60.

2. James PA, Oparil S, Carter BL, Cushman WC, Dennison-Himmelfarb C, Handler J, et al. 2014

evidence-based guideline for the management of high blood pressure in adults: report

from the panel members appointed to the Eighth Joint National Committee (JNC 8).

Jama 2014;311(5):507-20.

3. Thomopoulos C, Parati G, Zanchetti A. Effects of blood pressure lowering on outcome

incidence in hypertension: 7. Effects of more vs. less intensive blood pressure lowering

and different achieved blood pressure levels - updated overview and meta-analyses of

randomized trials. Journal of hypertension 2016;34(4):613-22.

4. Mozaffarian D. Dietary and Policy Priorities for Cardiovascular Disease, Diabetes, and Obesity:

A Comprehensive Review. Circulation 2016;133(2):187-225.

5. Wang H, Fox CS, Troy LM, McKeown NM, Jacques PF. Longitudinal association of dairy

consumption with the changes in blood pressure and the risk of incident hypertension:

the Framingham Heart Study. The British journal of nutrition 2015;114(11):1887-99.

6. Zong G, Sun Q, Yu D, Zhu J, Sun L, Ye X, et al. Dairy consumption, type 2 diabetes, and changes

in cardiometabolic traits: a prospective cohort study of middle-aged and older Chinese

in Beijing and Shanghai. Diabetes care 2014;37(1):56-63.

7. Livingstone KM, Lovegrove JA, Cockcroft JR, Elwood PC, Pickering JE, Givens DI. Does dairy

food intake predict arterial stiffness and blood pressure in men?: Evidence from the

Caerphilly Prospective Study. Hypertension 2013;61(1):42-7.

8. Dauchet L, Kesse-Guyot E, Czernichow S, Bertrais S, Estaquio C, Peneau S, et al. Dietary

patterns and blood pressure change over 5-y follow-up in the SU.VI.MAX cohort. The

American journal of clinical nutrition 2007;85(6):1650-6.

9. Snijder MB, van Dam RM, Stehouwer CD, Hiddink GJ, Heine RJ, Dekker JM. A prospective

study of dairy consumption in relation to changes in metabolic risk factors: the Hoorn

Study. Obesity (Silver Spring) 2008;16(3):706-9.

10. Samara A, Herbeth B, Ndiaye NC, Fumeron F, Billod S, Siest G, et al. Dairy product

consumption, calcium intakes, and metabolic syndrome-related factors over 5 years in

the STANISLAS study. Nutrition 2013;29(3):519-24.

11. Ralston RA, Lee JH, Truby H, Palermo CE, Walker KZ. A systematic review and meta-analysis

of elevated blood pressure and consumption of dairy foods. Journal of human

hypertension 2012;26(1):3-13.

12. Soedamah-Muthu SS, Verberne LD, Ding EL, Engberink MF, Geleijnse JM. Dairy consumption

and incidence of hypertension: a dose-response meta-analysis of prospective cohort

studies. Hypertension 2012;60(5):1131-7.

13. Bergholdt HK, Nordestgaard BG, Ellervik C. Milk intake is not associated with low risk of

diabetes or overweight-obesity: a Mendelian randomization study in 97,811 Danish

individuals. The American journal of clinical nutrition 2015;102(2):487-96.

Page 20 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

21

14. Kettunen J, Silander K, Saarela O, Amin N, Muller M, Timpson N, et al. European lactase

persistence genotype shows evidence of association with increase in body mass index.

Human molecular genetics 2010;19(6):1129-36.

15. Petitti DB. Approaches to heterogeneity in meta-analysis. Statistics in medicine

2001;20(23):3625-33.

16. Thomas DC, Lawlor DA, Thompson JR. Re: Estimation of bias in nongenetic observational

studies using "Mendelian triangulation" by Bautista et al. Annals of epidemiology

2007;17(7):511-3.

17. Drouin-Chartier JP, Gagnon J, Labonte ME, Desroches S, Charest A, Grenier G, et al. Impact of

milk consumption on cardiometabolic risk in postmenopausal women with abdominal

obesity. Nutrition journal 2015;14:12.

18. Tanaka S, Uenishi K, Ishida H, Takami Y, Hosoi T, Kadowaki T, et al. A randomized

intervention trial of 24-wk dairy consumption on waist circumference, blood pressure,

and fasting blood sugar and lipids in Japanese men with metabolic syndrome. Journal of

nutritional science and vitaminology 2014;60(5):305-12.

19. Drouin-Chartier JP, Gigleux I, Tremblay AJ, Poirier L, Lamarche B, Couture P. Impact of dairy

consumption on essential hypertension: a clinical study. Nutrition journal 2014;13:83.

20. Rideout TC, Marinangeli CP, Martin H, Browne RW, Rempel CB. Consumption of low-fat dairy

foods for 6 months improves insulin resistance without adversely affecting lipids or

bodyweight in healthy adults: a randomized free-living cross-over study. Nutrition

journal 2013;12:56.

21. Crichton GE, Howe PR, Buckley JD, Coates AM, Murphy KJ. Dairy consumption and

cardiometabolic health: outcomes of a 12-month crossover trial. Nutrition & metabolism

2012;9:19.

22. van Meijl LE, Mensink RP. Low-fat dairy consumption reduces systolic blood pressure, but

does not improve other metabolic risk parameters in overweight and obese subjects.

Nutrition, metabolism, and cardiovascular diseases : NMCD 2011;21(5):355-61.

23. Wennersberg MH, Smedman A, Turpeinen AM, Retterstol K, Tengblad S, Lipre E, et al. Dairy

products and metabolic effects in overweight men and women: results from a 6-mo

intervention study. The American journal of clinical nutrition 2009;90(4):960-8.

24. Barr SI, McCarron DA, Heaney RP, Dawson-Hughes B, Berga SL, Stern JS, et al. Effects of

increased consumption of fluid milk on energy and nutrient intake, body weight, and

cardiovascular risk factors in healthy older adults. Journal of the American Dietetic

Association 2000;100(7):810-7.

25. Heraclides A, Mishra GD, Hardy RJ, Geleijnse JM, Black S, Prynne CJ, et al. Dairy intake, blood

pressure and incident hypertension in a general British population: the 1946 birth

cohort. European journal of nutrition 2012;51(5):583-91.

26. Engberink MF, Geleijnse JM, de Jong N, Smit HA, Kok FJ, Verschuren WM. Dairy intake, blood

pressure, and incident hypertension in a general Dutch population. The Journal of

nutrition 2009;139(3):582-7.

27. Alonso A, Beunza JJ, Delgado-Rodriguez M, Martinez JA, Martinez-Gonzalez MA. Low-fat

dairy consumption and reduced risk of hypertension: the Seguimiento Universidad de

Navarra (SUN) cohort. The American journal of clinical nutrition 2005;82(5):972-9.

28. Steffen LM, Kroenke CH, Yu X, Pereira MA, Slattery ML, Van Horn L, et al. Associations of

plant food, dairy product, and meat intakes with 15-y incidence of elevated blood

pressure in young black and white adults: the Coronary Artery Risk Development in

Young Adults (CARDIA) Study. The American journal of clinical nutrition 2005;82(6):1169-

77; quiz 363-4.

Page 21 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

22

29. Djousse L, Pankow JS, Hunt SC, Heiss G, Province MA, Kabagambe EK, et al. Influence of

saturated fat and linolenic acid on the association between intake of dairy products and

blood pressure. Hypertension 2006;48(2):335-41.

30. Ruidavets JB, Bongard V, Simon C, Dallongeville J, Ducimetiere P, Arveiler D, et al.

Independent contribution of dairy products and calcium intake to blood pressure

variations at a population level. Journal of hypertension 2006;24(4):671-81.

31. Snijder MB, van der Heijden AA, van Dam RM, Stehouwer CD, Hiddink GJ, Nijpels G, et al. Is

higher dairy consumption associated with lower body weight and fewer metabolic

disturbances? The Hoorn Study. The American journal of clinical nutrition

2007;85(4):989-95.

32. Dong JY, Szeto IM, Makinen K, Gao Q, Wang J, Qin LQ, et al. Effect of probiotic fermented

milk on blood pressure: a meta-analysis of randomised controlled trials. The British

journal of nutrition 2013;110(7):1188-94.

33. Xu JY, Qin LQ, Wang PY, Li W, Chang C. Effect of milk tripeptides on blood pressure: a meta-

analysis of randomized controlled trials. Nutrition (Burbank, Los Angeles County, Calif.)

2008;24(10):933-40.

34. Cicero AF, Aubin F, Azais-Braesco V, Borghi C. Do the lactotripeptides isoleucine-proline-

proline and valine-proline-proline reduce systolic blood pressure in European subjects?

A meta-analysis of randomized controlled trials. American journal of hypertension

2013;26(3):442-9.

35. Chen M, Sun Q, Giovannucci E, Mozaffarian D, Manson JE, Willett WC, et al. Dairy

consumption and risk of type 2 diabetes: 3 cohorts of US adults and an updated meta-

analysis. BMC medicine 2014;12:215.

36. Bischoff-Ferrari HA, Dawson-Hughes B, Baron JA, Kanis JA, Orav EJ, Staehelin HB, et al. Milk

intake and risk of hip fracture in men and women: a meta-analysis of prospective cohort

studies. Journal of bone and mineral research : the official journal of the American

Society for Bone and Mineral Research 2011;26(4):833-9.

37. Appel LJ, Moore TJ, Obarzanek E, Vollmer WM, Svetkey LP, Sacks FM, et al. A clinical trial of

the effects of dietary patterns on blood pressure. DASH Collaborative Research Group.

The New England journal of medicine 1997;336(16):1117-24.

38. VanderWeele TJ, Tchetgen Tchetgen EJ, Cornelis M, Kraft P. Methodological challenges in

mendelian randomization. Epidemiology 2014;25(3):427-35.

Page 22 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

Confidential: For Review Only23

Table 1. Baseline characteristics of included cohorts listed by number of participants

Study name Ethnicity Country Number of

participants

Follow

-up

years

Male

, %

Age

,

year

SBP,

mmHg

Hypert

ension

, %

Antihypert

ensive

medication,

%

rs4988235,

n (%)

CC CT TT

CGPS Caucasian Denmark 74,219 0 45 57 140 6 20 6 36 58

WGHS Caucasian US 19,743 4 0 54 126 22 13 11 38 51

GESUS Caucasian Denmark 14,815 0 46 57 142 10 23 6 36 58

NHS Caucasian US 11,287 26 0 53 NA 9 NA 14 41 45

ARIC (Caucasian) Caucasian US 8,233 6 47 54 118 29 19 9 39 52

ARIC (African

American)

African

Caucasian US 1,889 6 36 53 127 55 41 74# 25

# 2

#

HPFS Caucasian US 6,914 24 100 55 NA 22 NA 18 39 43

INTER99 Caucasian Denmark 6,514 5 49 46 130 31 7 6 36 57

D.E.S.I.R. Caucasian France 3,378 9 50 47 131 35 9 22 49 30

Rotterdam Study Caucasian Netherlands 3,215 7 41 66 136 54 27 9 39 52

MDCS Caucasian Sweden 3,199 17 40 56 139 53 14 6&

33&

61&

GLACIER Caucasian Sweden 2,763 10 37 45 124 19 4 7 36 58

MESA Mixed US 2,424 10 47 61 132 35 28 20 49 30

FamHS Caucasian US 2,167 8 45 51 127 44 36 12 42 46

CHS Caucasian US 1,964 9 38 71 133 49 35 4 43 53

YFS Caucasian Finland 1,370 0 43 38 120 9 7 15 47 38

DCH Study Caucasian Denmark 1,297 0 45 56 135 47 13 5 32 63

DIOGENES-

controls Caucasian Denmark 1,002 0 51 54 135 38 7 6 36 58

DIOGENES-

weight gainers Caucasian Denmark 813 0 49 53 135 42 10 5 35 60

PREDIMED- Caucasian Spain 940 2 36 67 147 84 63 38 46 16

Page 23 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


VALENCIA

BPRHS

Puerto

Rican US 845 0 28 57 136 78 56 61 34 5

GOLDN Caucasian US 818 0 50 49 118 26 21 10 40 50

Raine Study Mixed* Australia 728 2 48 20 117 4 0 15 39 46

InCHIANTI Caucasian Italy 647 0 45 64 142 64 49 2 30 68

# Rs1446585 was used as a proxy.

& Rs309137 was used as a proxy.

*Mainly Caucasian, Caucasian-admixed (individuals with at least one Caucasian parent)

CGPS: The Copenhagen General Population Study; WGHS: Women's Genome Health Study; GESUS: The Danish General Suburban

Population Study; NHS: Nurses' Health Study; ARIC: Atherosclerosis Risk in Communities Study; HPFS: Health Professional

Follow-up Study; D.E.S.I.R.: The Data from an Epidemiological Study on the Insulin Resistance syndrome; MDCS: Malmö Diet and

Cancer Study; GLACIER: The GLatiramer Acetate low frequenCy safety and patIent ExpeRience Study; MESA: The Multi-Ethnic

Study of Atherosclerosis; FamHS: Family Heart Study; CHS: Cardiovascular Health Study; YFS: Young Finns Study; DCH Study:

Diet, Cancer and Health Cohort; Diogenes: Diet, Obesity and Genes Study; PREMED:The PREvención con DIeta MEDiterránea

Study; BPRHS: Boston Puerto Rican Health Study; GOLDN: Genetics of Lipid Lowering Drugs and Diet Network; Raine Study: The

Western Australian Pregnancy Cohort (Raine) Study; InCHIANTI: Invecchiare in Chianti Study

Page 24 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Figure 1. The association of SNP rs4988235 with dairy consumption (serving/day) using additive model

Linear regression adjusted for baseline age, sex, ethnicity, and region.

Page 25 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Figure 2. The association of SNP rs4988235 with systolic blood pressure (SBP, mmHg) and risk of hypertension using additive model

a. SNP rs4988235 (CT/TT vs. CC) and SBP b. SNP rs4988235 (CT/TT vs. CC) and risk of hypertension

Linear/logistic regression adjusted for baseline age, sex, ethnicity, and region.

Page 26 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Figure 3. The association of baseline dairy consumption (serving/day) with systolic blood pressure (SBP, mmHg) and risk of

hypertension in observational cohort studies and clinical trials

a. Dairy and SBP in cohort studies b. Dairy and hypertension in cohort studies c. Dairy and SBP in clinical trials

a & b. Linear/logistic regression was used in collaborative cohorts adjusted for sex, ethnicity, region, years of follow-up, as well as

age, body mass index, blood pressure/hypertension, smoking status, physical activity, total energy intake, and alcohol consumption at

baseline

Page 27 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Table 2. Stratified analysis on the causal estimates of dairy consumption with systolic blood pressure (SBP) and risk of hypertension

SBP (risk difference, 95% CI) Hypertension (relative risk, 95% CI)

Number

of

studies

SNP rs4988235

with dairy intake

(serving/d)

SNP rs4988235

with SBP (mmHg)

Dairy intake

(serving/d) with

SBP (mmHg),

IV estimation

SNP rs4988235

with risk of

hypertension

Dairy intake

(serving/d) with

risk of

hypertension,

IV estimation

CC frequency

< 10 % 12 0.07 (0.04, 0.10) 0.19 (-0.01, 0.38) 2.71 (-0.30, 5.73) 1.01 (0.97, 1.06) 1.15 (0.61, 2.18)

≥ 10 % 12 0.11 (0.06, 0.15) 0.14 (-0.21, 0.49) 1.27 (-1.95, 4.50) 1.01 (0.98, 1.04) 1.09 (0.83, 1.44)

Ethnicity

Caucasian 20 0.08 (0.05, 0.11) 0.19 (0.03, 0.35) 2.38 (0.19, 4.56) 1.01 (0.98, 1.03) 1.13 (0.83, 1.55)

Other races 4 0.12 (0.04, 0.21) -0.38 (-1.11, 0.36) -3.17 (-9.69, 3.36) 1.06 (0.89, 1.25) 1.63 (0.38, 6.97)

Country

U.S. 10 0.12 (0.08, 0.16) 0.11 (-0.18, 0.39) 0.92 (-1.48, 3.31) 1.01 (0.98, 1.04) 1.09 (0.85, 1.39)

Europe 13 0.05 (0.02, 0.09) 0.29 (0.02, 0.56) 5.80 (-0.96, 12.56) 1.01 (0.97, 1.07) 1.22 (0.45, 3.29)

Study design

Cross-sectional 9 0.08 (0.03, 0.12) 0.36 (-0.001, 0.73) 4.50 (-0.72, 9.72) 0.99 (0.92, 1.06) 0.88 (0.36, 2.14)

Cohort 15 0.09 (0.05, 0.12) 0.09 (-0.14, 0.32) 1.00 (-1.58, 3.58) 1.02 (0.99, 1.05) 1.25 (0.89, 1.75)

Measurement of SBP

Self-reported 5 0.10 (0.05, 0.15) 0.01 (-0.31, 0.34) 0.10 (-3.15, 3.35) 1.01 (0.97, 1.07) 1.10 (0.67, 1.81)

Clinical measurement 19 0.08 (0.04, 0.11) 0.22 (0.03, 0.41) 2.75 (0.09, 5.41) 1.01 (0.98, 1.05) 1.13 (0.73, 1.75)

Page 28 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

1

Supplemental materials

Description of collaborative cohorts

The Copenhagen General Population Study (CGPS): The CGPS is an ongoing

prospective cohort study initiated in 2003 and includes individuals residing in the greater

urban area of Copenhagen. Residence was determined by data from the national Danish

Civil Registration System. All individuals aged > 40 years were invited along with a

random selection of 25% of individuals aged 20-39 years, and the overall response rate

was 45%. Participants in the study completed a general questionnaire and a health

examination including a non-fasting blood sample. In this study we included

approximately 74,000 white participants of Danish descent who had been genotyped for

the rs4988235 genetic variant.

The Women’s Genome Health Study (WGHS): The WGHS is a prospective cohort of

initially healthy, female North American health care professionals at least 45 years old at

baseline representing participants in the Women’s Health Study (WHS) who provided a

blood sample at baseline and consent for blood-based analyses. The WHS was a 2x2 trial

beginning in 1992-1994 of vitamin E and low dose aspirin in prevention of cancer and

cardiovascular disease with about 10 years of follow-up. Since the end of the trial,

follow-up has continued in observational mode. Additional information related to health

and lifestyle were collected by questionnaire throughout the WHS trial and continuing

observational follow-up.

The Danish General Suburban Population Study (GESUS): The GESUS was initiated

in 2010 and concluded in 2013. The GESUS is a study of the suburban general

population in Naestved Municipality located approximately 70 km south of Copenhagen.

All individuals aged > 30 and a random selection of 25% of the younger population aged

20-30 years were invited. Participants in the study completed a general questionnaire and

a health examination including a non-fasting blood sample. In this study, we included

approximately 14,000 white participants of Danish descent, with known rs4988235

genotypes. The response rate for this subset of the study population was 50%, and the

GESUS had an overall response rate of 43% when the study was concluded in 2013.

The Nurses’ Health Study (NHS): The NHS began in 1976, when 121,700 female

registered nurses aged 30 - 55 y residing in 11 states were recruited to complete a

baseline questionnaire about their lifestyle and medical history. Questionnaires were

collected at baseline and biennially thereafter, to update information on lifestyle factors

and the occurrence of chronic diseases. In the current analysis, we used 1990 as baseline

in the NHS, when the earliest complete dietary data were collected. Our analysis included

13,000 women whose genotype data were available. All of the participants were

Caucasians and were free of diabetes, cardiovascular disease, and cancer at baseline. The

study protocol was approved by the institutional review boards of Brigham and Women’s

Hospital and Harvard School of Public Health.

The Atherosclerosis Risk in Communities Study (ARIC): The ARIC is a multi‐center

prospective investigation of atherosclerotic disease in a predominantly bi‐ racial

population conducted in four U.S. communities, involving both cohort and community

surveillance components 2. Study participants aged 45‐64 years at baseline were

recruited from 4 communities: Forsyth County, North Carolina; Jackson, Mississippi;

Page 29 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

2

suburban areas of Minneapolis, Minnesota; and Washington County, Maryland. A total

of 15,792 individuals participated in the baseline examination in 1987‐1989, with follow‐

up examinations in approximate 3‐year intervals, during 1990‐1992, 1993‐1995, and

1996‐1998. Weight and height were measured. All study participants provided written

informed consent.

The Health Professional Follow-up Study (HPFS): The HPFS was initiated in 1986,

and was composed of 51,529 male dentists, pharmacists, veterinarians, optometrists,

osteopathic physicians, and podiatrists, aged 40-75 y at baseline. The male participants

returned a baseline questionnaire about detailed medical history, lifestyle, and usual diet.

Questionnaires were collected at baseline and biennially thereafter, to update information

on lifestyle factors and the occurrence of chronic diseases. In the current analysis, we

used 1990 as baseline in the HPFS, when the earliest complete dietary data were

collected. Our analysis included 8,000 men whose genotype data were available. All of

the participants were Caucasians and were free of diabetes, cardiovascular disease, and

cancer at baseline. The study protocol was approved by the institutional review boards of

Brigham and Women’s Hospital and Harvard School of Public Health.

The Danish population-based Inter99 study: The INTER99 is a non-pharmacological

intervention study for ischemic heart disease, initiated in 1999 at the Research Centre for

Prevention and Health, Glostrup, Denmark (ClinicalTrials.gov ID-no: NCT00289237). A

random sample of 13,016 individuals living in Copenhagen County from seven different

age groups (30 - 60 years, grouped with five year intervals) was drawn from the Civil

Registration System and 6,784 of these attended the health examination.

The Data from an Epidemiological Study on the Insulin Resistance syndrome

(D.E.S.I.R.): The D.E.S.I.R. began in 1994 and included 5,212 men and women, aged

from 30 to 64 years. Participants were recruited from volunteers insured by the French

Social Security system, which offered periodic health examinations free of charge.

D.E.S.I.R. is a 9-year follow-up study with clinical and biological examinations every 3

years. A medical interview provided information about use of medication, and personal

and familial history of diseases. Our analysis included only Caucasian subjects with

lactase genotype, dietary data and examined 9 years after inclusion (n=3478). The study

was approved by the ethics committee of the Kremlin Bicêtre Hospital and by the CNIL

(Commission Nationale de l'Informatique et des Libertes).

The Rotterdam Study: The Rotterdam Study is a population-based prospective cohort

study ongoing since 1990 in the city of Rotterdam in the Netherlands. At present the The

study was designed to investigate the prevalence and incidence of and risk factors for

chronic diseases in the elderly. All inhabitants of Ommoord, a district of Rotterdam, the

Netherlands, aged 55 years and older were invited. At enrollment, participants were

interviewed at home (2h) and were examined in detail at a dedicated research facility (5h).

These measurements were repeated every 3 to 4 years. The study was approved by the

Medical Ethics Committee of Erasmus Medical Center and all participants provided

written informed consent to participate in the study and to obtain information from their

treating physicians.

Malmö Diet and Cancer Study (MDCS): The MDCS was a population based study

conducted 1991-1996. In total, 28098 individuals (45-74 years) completed all baseline

examinations. The MDC cardiovascular subcohort (n = 6,103) were invited to a follow-

Page 30 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

3

up study 2007-2012. Individuals from the follow-up study with data available on

genotypes were included in this study (n = 3,199).

Gene-Lifestyle interactions And Complex traits Involved in Elevated disease Risk

(GLACIER): The GLACIER Study (N ~ 19,000) is a population-based cohort study of

initially non-diseased adults living in the county of Västerbotten in Northern Sweden,

nested within the Northern Sweden Health and Disease Study. Clinical characteristics and

lifestyle data were obtained as part of a population-wide health screening initiative called

the Västerbotten’s Health Survey (also called the Västerbotten Intervention Program),

where habitants are invited to attend an extensive health examination the year of their

40th, 50th, and 60th birthday. The total number of GLACIER participant with genotype

and phenotype data available for the current analysis was 2,763. Ethical approval for the

GLACIER Study was obtained from the Regional Ethical Review Board in Umeå.

The Multi-Ethnic Study of Atherosclerosis Study (MESA): The MESA is a study of

the characteristics of subclinical cardiovascular disease (disease detected non-invasively

before it has produced clinical signs and symptoms) and the risk factors that predict

progression to clinically overt cardiovascular disease or progression of the subclinical

disease. MESA researchers study a diverse, population-based sample of 6,814

asymptomatic men and women aged 45 - 84. Thirty-eight percent of the recruited

participants were white, 28 percent African-American, 22 percent Hispanic, and 12

percent Asian, predominantly of Chinese descent. Participants were recruited from six

field centers across the United States and followed-up three times with an average time

period of follow-up of 2 years between each visit. Data from four visits (exam1 to exam 5)

was used for the analysis. The tenets of the Declaration of Helsinki were followed and

institutional review board approval was granted at all MESA sites. Written informed

consent was obtained from each participant.

The Family Heart Study (FamHS): The FHS began in 1992 with the ascertainment of

1,200 families (50% randomly sampled, and 50% high risk for CHD). The families (~

6,000 individuals,) were sampled on the basis of information on probands from four

population-based parent studies: the Framingham Heart Study, the Utah Family Tree

Study, and two ARIC centers (Minneapolis, and Forsyth County, NC). Approximately

eight years later, study participants belonging to the largest pedigrees were invited for a

second clinical exam. A total of 2,767 participants of European descent in 510 extended

families were examined. A total of 2,167 adults with available DNA and who provided

valid dietary information were eligible for the current study.

The Cardiovascular Health Study (CHS): The CHS is a population-based cohort study

of risk factors for coronary heart disease and stroke in adults ≥65 years conducted across

four field centers. The original predominantly European ancestry cohort of 5,201 persons

was recruited in 1989-1990 from random samples of the Medicare eligibility lists;

subsequently, an additional predominantly African-American cohort of 687 persons was

enrolled for a total sample of 5,888. The analyses included 1,964 white study participants

without cardiovascular disease at baseline, with available genotype data and with follow-

up data 9 years after baseline. CHS was approved by institutional review committees at

each site, the subjects gave informed consent, and those included in the present analysis

consented to the use of their genetic information for the study of cardiovascular disease.

The Cardiovascular Risk in Young Finns (YFS) Study: The YFS is a population-

based 27 year follow up-study. The first cross-sectional survey was conducted in 1980,

Page 31 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

4

when 3,596 Caucasian subjects aged 3-18 years participated. In adulthood, the latest 27-

year follow-up study was conducted in 2007 (ages 30-45 years) with 2,204 participants.

The study cohort for the present analysis comprised subjects who had participated in the

study in 2007 and had validated dietary data from FFQ, available genotype and other risk

factor data. The dietary intake of nutrients was assessed using a modified 131-item food

frequency questionnaire developed by the Finnish National Institute for Health and

Welfare. The study was approved by the local Ethical Committees and was performed

according to Helsinki declaration.

The Diet, Cancer and Health Cohort (DCH) and the Diet, Obesity and Genes Study

(DIOGENES): The DCH and the DIOGENES cohorts are both subsamples from the

same larger cohort, the Danish Diet, Cancer and Health cohort. The larger cohort was

composed of 57,053 individuals in the age of 50-64 years and born in Denmark. At

baseline, anthropometric measurements were taken and blood pressure was measured.

Information on usual diet and lifestyle was obtained using self-administered

questionnaires, and biological samples were collected. At 5-year follow-up, the

participants completed a repeat questionnaire and self-measured their BMI and waist

circumference. The DCH included a subsample of 1,297 randomly selected individuals

who were free of diabetes, cardiovascular disease, and cancer at baseline and at the 5-

year follow-up. The DIOGENES included 1,815 individuals, of which 813 were weight-

gainers and 1,002 were randomly selected control individuals. Weight gainers were

defined as those individuals who had experienced the greatest degree of unexplained

weight gain and were identified by using the residuals from a regression model of annual

weight change on baseline values of age, weight, height, and smoking status

(current/former/nonsmokers) and follow-up time. The participants were younger than 60

years at baseline and younger than 65 years at follow-up, with stable smoking habits,

without cancer, cardiovascular disease, or diabetes at baseline and at the 5 year follow-up,

and with a weight change not more than 5 kg/year.

The PREvención con DIeta MEDiterránea Study (PREDIMED-VALENCIA study):

The PREDIMED-VALENCIA study was initiated in 2003 y was composed of 1094

participants. LCT genotypes were available for 940. PREDIMED-VALENCIA is a

randomized intervention trial with Mediterranean diet versus a control diet. FFQ

questionnaires were completed at baseline and yearly. BMI and blood pressure were

directly measured yearly. We used data corresponding to the 2-y follow-up period.

The Boston Puerto Rican Health Study (BPRHS): The Boston Puerto Rican Health

Study is an ongoing longitudinal cohort study designed to examine the role of

psychosocial stress on presence and development of allostatic load and health outcomes

in Puerto Ricans, and potential modification by nutritional status, genetic variation, and

social support. Individuals who were self-identified Puerto Ricans, aged 45-75 years and

residing in the Boston, MA metro area, were recruited through door-to-door enumeration

and community approaches. Data including demographics, medical history, physical

function, cognition and dietary data were collected through a comprehensive set of

questionnaires and tests. Blood, urine and salivary samples were extracted for biomarker

and genetic analysis. Measurements were repeated at a two-year follow-up and a five-

year follow up.

Genetics of Lipid Lowering Drugs and Diet Network Study (GOLDEN): The

GOLDN Study belongs to the PROgram for GENetic Interaction (PROGENI) Network,

Page 32 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

5

which includes family studies examining gene-environment interactions via controlled

interventions. GOLDN participants were recruited from three-generational pedigrees

previously identified in the Minneapolis, MN, and Salt Lake City, UT, field centers of the

National Heart, Lung, and Blood Institute Family Heart Study. Clinical, dietary and

biochemical measurements were collected at baseline and following an intervention with

fenofibrate. A postprandial study fat-challenge was also conducted pre - and post-

fenofibrate. Individuals were >= 18 years with fasting triglycerides < 1500 mg/dl.

The Western Australian Pregnancy Cohort (Raine) Study: Raine Study is a

prospective pregnancy cohort where 2900 women were recruited from King Edward

Memorial Hospital between 1989 and 1991. Data were collected during pregnancy and

the children have been followed-up at ages 1, 2, 3, 5, 8, 10, 14, 17, 18, 20 and 22. Human

Research Ethics approval for this study was obtained from King Edward Memorial

Hospital, Princess Margaret Hospital and University of Western Australia. Parent of the

participants and participants from age 18 were consented at each follow-up. In the current

analysis, the 20 year follow-up was used as baseline. Analyses were performed at the 22

year follow-up on 728 participants who had genotyping data, food frequency data and

other co-variates.

The Invecchiare in Chianti Study (InCHIANTI): The InCHIANTI study is a

population-based epidemiological study aimed at evaluating the factors that influence

mobility in the older population living in the Chianti region in Tuscany, Italy. Overnight

fasted blood samples were for genomic DNA extraction and genotyping using Illumina

Infinium HumanHap 550K SNP arrays were used for genotyping. Dairy consumption

was assessment using a questionnaire designed for the EPIC study. The analysis was

restricted to those with data on dairy intake, BMI, blood pressure, and genotyping. The

study protocol was approved by the Italian National Institute of Research and Care of

Aging Institutional Review and Medstar Research Institute (Baltimore, MD).

Page 33 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

6

Methods of study selection

PUBMED search query

#1

"dairy products"[Mesh] OR "dairy"[tiab] OR "milk"[tiab] OR "calcium"[tiab] OR

"cheese"[tiab] OR "yogurt"[tiab]

#2

"hypertension"[Mesh] OR "blood pressure"[Mesh]

#3

(#1 AND #2))

Publication bias of included RCTs

We examined publication bias of included RCTs using Egger test, with P < 0.05 to

indicate significant asymmetry [1].

Dose-response analysis within each included cohort study

To obtain the association of per serving/day increase in dairy consumption with SBP and

risk of hypertension, we performed dose-response analysis within studies using

categorical dairy intake as main exposure.

We assigned the median dairy consumption in each category of consumption to the

corresponding outcome for each study. If the upper boundary for the highest category

was not provided, the assigned median value was 25% higher than the lower boundary of

that category. If the lower boundary for the lowest category was not provided, the

assigned median value was half of the upper boundary of that category.

To examine the association of continuous dairy consumption with risk of

hypertension, we used two-stage fixed/random-effects dose response models to combine

studies that reported results for categorized dairy consumption and studies with reported

results for continuous dairy consumption. Specifically, the RR of risk of hypertension per

unit increase of dairy consumption for each study was first estimated separately by

generalized least squares models (GLST) [2], and then the RRs from all of the studies

were pooled together by a fixed/random-effects model. We fit a fixed-effects generalized

linear model first, and changed to a random-effects generalized linear model if the p

value for the goodness of fit/heterogeneity of the previous model was < 0.05. We used

the STATA command GLST for model fitting.

To examine the association of continuous dairy consumption with blood pressure, for

studies with categorized dairy consumption, we calculated the change of blood pressure

comparing the highest category of dairy consumption to the lowest. We assumed a linear

association of continuous dairy consumption with change of blood pressure.

Reference

1. Egger M, Davey Smith G, Schneider M, Minder C: Bias in meta-analysis detected by a

simple, graphical test. BMJ 1997, 315(7109):629-634.

2. Greenland S, Longnecker MP: Methods for trend estimation from summarized dose-

response data, with applications to meta-analysis. American journal of epidemiology

1992, 135(11):1301-1309.

Page 34 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Supplemental Figure 1. Study selection process of cohort studies and randomized trials (RCT)

Articles identified initially (n = 3,755)

Articles excluded on basis of title and abstract (n = 3,723)

Articles retrieved for further evaluation (n = 32)

Articles excluded (n=13):

Cohort studies excluded due to duplicity with ours (n=3)

RCT excluded using fermented milk or soy milk as intervention group (n = 7)

RCT excluded not presenting the exact number of SBP and unable to contact the

author (n = 1)

RCT excluded presenting extreme small standard deviation (SD) of SBP (SD < 2

mmHg for both interventional and control group) (n = 1)

Articles finally included in the analysis (n = 20)

Randomized trials on dairy consumption with SBP (n = 8)

Cohort studies on dairy consumption with SBP (n = 7)

Cohort studies on dairy consumption with risk of hypertension (n = 5)

Page 35 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Supplemental table 1. Measurements of dairy products, systolic blood pressure, and hypertension and genotype method of included

cohorts

Study Dairy products Systolic blood

pressure,

hypertension

Genotype

Measur

ement

Time of

assessme

nt

Dairy products included

Mean

(servi

ng/d)

Measure

ment

Time of

assessme

nt

Genotypi

ng

method

Sample

call rate

Proxy

SNP/

R2

HWE

p-

value

CGPS

General

questio

nnaire Baseline

Milk (whole, semi-

skimmed, skimmed), cheese 1.69

Clinical

measure

ment Baseline TaqMan

>

99.9%. NA 0.001

WGHS FFQ Baseline

Skim/low fat milk, whole

milk, ice cream, yogurt,

cottage cheese, cream

cheese, other cheese,

cream, sour cream, and

sherbert 1.98

Self-

reported Baseline TaqMan 97% NA NA

GESUS

General

questio

nnaire Baseline

Milk (whole, semi-

skimmed, skimmed, butter

milk, lactose free), cheese,

fermented milk 2.38

Clinical

measure

ment

End of

follow-

up

Competit

ive

Allele-

Specific

PCR-

based

assay

(KASP)

>

99.9%. NA 0.19

NHS FFQ Baseline


milk, ice cream, yogurt, 1.37 NA

End of

follow- TaqMan 97% NA 0.35

Page 36 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


cottage/ricotta cheese,

cream cheese, other cheese,

and cream

up

ARIC

(Caucasi

an) FFQ Baseline





and cream 1.8

Clinical

measure

ment

End of

follow-

up Imputed NA NA 0.92

ARIC

(African

America

n) FFQ Baseline





and cream 1.25

Clinical

measure

ment

End of

follow-

up Imputed NA

rs144

6585/

1.00 0.96

HPFS FFQ Baseline





and cream 1.37 NA

End of

follow-

up TaqMan 97% NA 0.37

INTER9

9 FFQ Baseline


milk, chocolate milk, ice

cream, yogurt,



and cream 1.43

Clinical

measure

ment

End of

follow-

up

Illumina

HiScan

system 99.3% NA 0.16

D.E.S.I.

R. FFQ Baseline

milk, yogurt, cottage/fresh

soft cheese, custard type

desserts 1.36

Clinical

measure

ment

End of

follow-

up Kaspar 97% NA 0.36

Rotterda

m Study FFQ Baseline

Whole milk, skimmed milk,

yogurt, hard cheese, soft

cheese, whip cream, coffee

cream, ice cream, custard 3.99

Clinical

measure

ment

End of

follow-

up

Illumina

550 > 97% NA 0.13

Page 37 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


MDCS

Diet

recall Baseline




cream cheese, other cheese

and cream 4.8

Clinical

measure

ment

End of

follow-

up

Illumina

HumanO

mniExpr

ess

BeadChi

p

>

99.9%.

rs309

137/0.

77 0.02

GLACIE

R FFQ Baseline

cream, sour cream, hard

cheese 28% fat, hard cheese

10-17% fat, sour milk and

yoghurt (0.5 and 3% fat),

ice cream, milk (0.5, 1.5 and

3% fat) 3.32

Clinical

measure

ment

End of

follow-

up

Metaboc

hip array

100% NA

0.000

25

MESA FFQ Baseline

Whole milk, regular cheese,

cottage or ricotta cheese,

whole-fat yogurt, and ice

cream, 2% milk, 1% or

skim milk, and low-fat

yogurt 1.63

Self-

reported

End of

follow-

up Affy 6.0 > 95% NA NA

FamHS FFQ Baseline



cottage/ricotta cheese, and

other cheese 2.04

Clinical

measure

ment

End of

follow-

up

Illumina

GWAS

arrays 100% NA 0.019

CHS FFQ Baseline

Skim milk/buttermilk, low

fat milk/beverages with low

fat milk, whole

milk/beverages with whole


cottage cheese, other

cheese/cheese spread 1.37

Clinical

measure

ment

After 9

years of

follow-

up

Illumina

370CNV

BeadChi

p ≥ 97% NA NA

YFS FFQ

End of

follow-


milk, ice cream, yogurt, 4.3

Clinical

measure

End of

follow-

custom

Illumina 95% NA 0.19

Page 38 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


up cottage/ricotta cheese,


and cream

ment up BeadChi

p

Human6

70K

DCH

Study FFQ Baseline

Skimmed milk, Semi-

skimmed milk, Whole fat

milk, Buttermilk, Fermented

dairy (low-fat), Fermented

dairy (whole-fat), Ice cream

(low fat), Ice cream (whole

fat), Cream 1.64

Clinical

measure

ment Baseline

Illumina

HumanC

oreExom

e

BeadChi

p

genotype

s >95% NA 0.36

DIOGE

NES FFQ Baseline

Skimmed milk, Semi-

skimmed milk, Whole fat

milk, Buttermilk, Fermented

dairy (low-fat), Fermented

dairy (whole-fat), Ice cream

(low fat), Ice cream (whole

fat), Cream 1.59

Clinical

measure

ment Baseline

Illumina

Metaboc

hip >95% NA 0.64

PREDI

MED-

VALEN

CIA-

Study FFQ Baseline





and cream 1.85

Clinical

measure

ment

End of

follow-

up TaqMan 97% NA 0.221

BPRHS FFQ Baseline





and cream 2.3

Clinical

meausre

ment Baseline TaqMan >97% NA 0.5

GOLDN FFQ Baseline


milk, ice cream, yogurt, 1.97

Clinical

measure Baseline TaqMan >97% NA 0.06

Page 39 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




and cream

ment

Raine

Study FFQ Baseline

Full cream milk, Reduced

fat milk, Skim milk, hard

cheese, firm cheese, ricotta,

low fat cheese, flavoured

milk, ice cream, yoghurt,

cream cheese, soft cheese 1.9

Clinical

measure

ment

End of

follow-

up

Illumina

660W

Quad

Array;

Human

Omni

Express

BeadChi

p 97% NA

660W

: 1.8

E-05;

Omni:

1

InCHIA

NTI FFQ Baseline

whole milk, low fat milk,

icecream, yogurt, latte,

butter, cheese, soft cheeseer 1.09

Clinical

measure

ment Baseline

Illumina

550K 97% NA 0.23

Page 40 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Supplemental Table 2. Baseline characteristics of published randomized trials

Study

design Author Year

Sample

size Sex

Age

(year)

Health

status Intervention Control Period Results

Difference

of SBP

changes

comparing

interventio

n to

control

group

(mmHg)

Crosso

ver

Drouin-

Chartier 2015 27

Wo

men 57

Postmenopa

usal women

with

abdominal

obesity

3.2 servings/d of

2% fat milk per

2000 kcal

Diet without

milk or other

dairy

6

weeks

Intervention

group

End of trial:

109.9 mmHg

95% CI (105.1,

114.6)

Control group

End of trial:

112.1 mmHg

95% CI (107.9,

116.5)

Parallel Tanaka 2014 200 Both

20 -

60

Two or

more

components

of the

metabolic

syndrome

Both dietary

intervention and

regular home

dairy delivery of

400 g/d

Dietary

intervention

24

weeks

-1.0

(-4.0, 2.1)

Crosso

ver

Drouin-

Chartier 2014 89 Both 53

Prehyperten

sion

Three daily

servings of dairy

products

Control

products

equivalent in

4

weeks

0

(-1, 1)

Page 41 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


macronutrients

and sodium

Crosso

ver Rideout 2013 23 Both

Mean

53 Healthy

4 servings of

dairy per day

No more than

2 servings of

dairy per day

12

months

Intervention

group

End of trial:

122 ± 15

(mmHg,

mean ± SD)

Control group

End of trial:

124 ± 16

(mmHg,

mean ± SD)

Crosso

ver Crichton 2012 36 Both

18 -

75

Overweight

or obese

adults

A high dairy

diet (4 serves of

reduced fat

dairy/day)

A low dairy

control diet

(≤1 serve/day)

6

months

0.9

(-1.8, 3.6)

Crosso

ver

van

Meijl 2009 35 Both 50 Healthy

500 mL low-fat

milk and 150 g

low-fat yogurt

600 mL fruit

juice and 3

fruit biscuits

8

weeks

-2.9

(-5.5, -0.3)

Parallel

Wenners

berg 2009 121 Both

30 -

65

Metabolic

syndrome

3 - 5 portions of

dairy products

daily Habitual diet

6

months

Intervention

group (56

participants)

Baseline:

134 ± 16

(mmHg,

mean ± SD)

End of trial:

132 ± 16

Control group

(55 participants)

Page 42 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Baseline:

134 ± 16

End of trial:

133 ± 17

Parallel Barr 2000 204 Both

55 -

85 Healthy

Skim or 1%

milk intake by 3

cups per day Usual diet

12

weeks

Intervention

group (98

participants)

Baseline:

126 ± 12

(mmHg,

mean ± SD)

End of trial:

124 ± 13

Control group

(102

participants)

Baseline:

128 ± 15

End of trial:

125 ± 14

Page 43 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Supplemental Table 3. Baseline characteristics of published cohorts on dairy consumption with change of systolic blood pressure and

risk of hypertension

Systolic blood pressure

Cohort

name

Author Year Sample

size

Sex Age

(year

s)

Foll

ow-

up

year

Dairy

consump

tion

Change of

SBP (mmHg)

Confounding adjustment

Framingha

m Heart

Study Wang 2015 2,075 Both

28-

62 15

0.64

serving/d

ay

1.07

(0.95, 1.19)

Sex, age and systolic blood pressure at

the beginning of each exam interval and

average total energy intake during each

exam interval, smoking status and

physical activity at the beginning of

each exam interval and the average

caffeine coffee intake and Dietary

Guidelines Adherence Index (DGAI)

sub-score (i.e. DGAI score excluding

sub-scores for assessing the

consumption amount of milk and milk

products and the likelihood of choosing

low-fat milk and milk products) during

each exam interval, BMI at the

beginning of each exam interval.

3.53

serving/d

ay

0.47

(0.23, 0.71)

Nutrition

and Health

of Aging

Population

in China Zong 2014 2,091 Both

50-

70 6

0

serving/d

ay

-0.62

(-2.07, 0.83)

Age, sex, region, residence, smoking

status, family history of diabetes, BMI

(not for BMI and waist circumference),

dietary fiber intake, and baseline values

of SBP. 1.25

serving/d

ay

-4.15

(-6.27, -2.03)

Caerphilly

Prospective

Livingst

one 2013 2,512 Men

45-

59 5

Per

serving

-10.40

(-19.38, -

Age, alcohol consumption, smoking

habits, social class, physical activity,

Page 44 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Study increase 1.42) total energy intake, fat intake, heart rate,

mean arterial pressure, and drug use,

and protein intake.

STANISLA

S Samara 2012 588 Both

28-

60 5

Per

serving

increase

Men:

-0.68

(-2.56, 1.20)

Women:

0.81

(-0.84, 2.46)

Adjusted for age, physical activity,

alcohol and cigarette consumption,

energy intake without alcohol, education

level, mean adequacy ratio index, and

value of metabolic syndrome-related

variable at entry.

1946 Birth

Cohort

Heraclid

es 2012 1,750 Both

43-

53 10

0.93

serving/d

ay

0 (0,0) Fruit and vegetable intake, wholegrain

cereal intake, coffee and tea

consumption, total energy intake,

alcohol consumption, physical activity,

smoking status 1.29

serving/d

ay

-0.67

(-5.39,4.04)

The Hoorn

Study Snijder 2008 1,124 Both

50–

75 6.4

Per

serving

increase

0.27

(-0.22, 0.76)

Age, sex, total energy intake, baseline

value of the outcome variable, lifestyle

factors (alcohol intake, smoking,

physical activity).

SU.VI.MA

X Dauchet 2007 2,341 Both

35–

63 5.4

84 g 0 (0, 0) Adjusted for age, sex, group (treatment

vs placebo), total energy intake

(excluding alcohol), number of dietary

records completed, and SBP or DBP at

first clinical examination, tobacco,

alcohol, physical activity, education

level, BMI, dietary sodium, other

dietary variables (fruit and vegetables,

dairy products, and Keys score) if they

456 g 0.1 (-1.2, 1.5)

Page 45 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


are not part of a score included in the

model, and SBP or DBP at first clinical

examination.

Hypertension

Cohort

name

Author Year Sample

size

Sex Age

(year

s)

Foll

ow-

up

year

Dairy

consum

ption

Case

s

Obser

vation

s

Risk of

hypertensio

n

Confounding adjustment

Framingha

m Heart

Study Wang 2015 2,075 Both

28–

62 15

Per

serving

0.92

(0.86, 0.99)

Sex, age and systolic blood

pressure at the beginning

of each exam interval and

average total energy intake

during each exam interval,

smoking status and

physical activity at the

beginning of each exam

interval and the average

caffeine coffee intake and

Dietary Guidelines

Adherence Index (DGAI)

sub-score (i.e. DGAI score

excluding sub-scores for

assessing the consumption

amount of milk and milk

products and the likelihood

of choosing low-fat milk

and milk products) during

each exam interval, BMI at

the beginning of each

exam interval.

1946 Birth Heraclid 2012 1,750 Both 43- 10 224 g 208 577 Reference

Page 46 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Cohort es 53 275 g 200 571 0.88

(0.68,1.14)

Fruit and vegetable intake,

wholegrain cereal intake,

coffee and tea

consumption, total energy

intake, alcohol

consumption, physical

activity, smoking status

309 g 313 602 0.93

(0.72,1.18)

Morgan

Study

Engberi

nk 2009 3454 both

20–

65 5

206 g 185 863 Reference Adjusted for age, sex, total

energy intake (MJ/d),

socioeconomic status (5

categories), BMI (kg/m2),

smoking (yes/no), alcohol

intake (6 categories), and

daily intake of fruit (g),

vegetables (g), fish (g),

meat (g), bread (g), coffee

(mL), and tea (mL).

359 g 189 864 1.08

(0.84, 1.38)

510 g 162 864 0.95

(0.73, 1.22)

757 g 177 863 1.11

(0.85, 1.44)

CARDIA Steffen 2005 4304 both

18–

30 15

0.6

serving/

day

259 860 Reference Adjusted for baseline age,

sex, race, education,

center, and energy intake,

physical activity, alcohol

intake, baseline smoking,

vitamin supplement use,

and simultaneous

adjustment of all food

groups, explanatory

nutrients (sodium,

saturated fat, calcium,

magnesium, potassium,

and dietary fiber) and

physiological

1.4 227 861 1.02

(0.83, 1.25)

2 181 861 0.82

(0.65, 1.04)

2.9 170 861 0.85

( 0.66, 1.11)

4.25 160 861 0.82

(0.59, 1.14)

Page 47 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


measurements (baseline

systolic blood pressure,

BMI, and fasting insulin).

SUN Alonso 2005 5880 both 37 2.3

156 g 49 2709 Reference Cox regression model

adjusted for age

(continuous variable), sex,

BMI (lineal and quadratic

term), physical activity,

alcohol consumption,

sodium intake, total energy

intake, smoking (never,

former, or current),

hypercholesterolemia (yes

or no), quintiles of fruit,

vegetable, fiber, caffeine,

magnesium, potassium,

monosaturated fatty acid,

and saturated fatty acid

intakes.

292 g 38

2708 0.84

(0.54, 1.29)

386 g 39 2708 0.85

(0.54, 1.32)

530 g 24 2705 0.57

(0.34,0.95)

799 g 30 2694 0.75

(0.45,1.27)

Page 48 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Supplemental Figure 2. The associations of SNP (rs4988235) with dairy consumption (serving/day), systolic blood pressure (SBP),

and risk of hypertension using recessive models (TT vs. CC/ CT)

a. SNP (rs4988235) and dairy consumption b. SNP (rs4988235) and SBP c. SNP (rs4988235) and risk of hypertension


Page 49 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Supplemental Figure 3. The associations of SNP (rs4988235) with dairy consumption (serving/day), systolic blood pressure (SBP),

and risk of hypertension using dominant models (CT/TT vs. CC)

a. SNP (rs4988235) and dairy consumption b. SNP (rs4988235) and SBP c. SNP (rs4988235) and risk of hypertension


Page 50 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Supplemental Table 4. The causal estimate of per serving increase in dairy consumption on systolic blood pressure (SBP) and risk of

hypertension using recessive and additive models by instrumental variable method

Change of SBP (mmHg) Relative risk of hypertension

Recessive models 3.00 (-0.56, 6.56) 1.00 (0.66, 1.51)

Dominant 1.35 (-0.28, 2.97) 1.04 (0.88, 1.24)

Page 51 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

Dairy Consumption, Systolic Blood Pressure and Risk of ... · Hernandez 41, Luigi Ferrucci 42,...

Documents

Transcript of Dairy Consumption, Systolic Blood Pressure and Risk of ... · Hernandez 41, Luigi Ferrucci 42,...