Milk and Dairy Product Consumption and Cardiovascular Diseases: An Overview of Systematic Reviews and Meta-Analyses

ABSTRACT

Milk and dairy products containing milk fat are major food sources of saturated fatty acids, which have been linked to increased risk of cardiovascular-related clinical outcomes such as cardiovascular disease (CVD), coronary heart disease (CHD), and stroke. Therefore, current recommendations by health authorities advise consumption of low-fat or fat-free milk. Today, these recommendations are seriously questioned by meta-analyses of both prospective cohort studies and randomized controlled trials (RCTs) reporting inconsistent results. The present study includes an overview of systematic reviews and meta-analyses of follow-up studies, an overview of meta-analyses involving RCTs, and an update on meta-analyses of RCTs (2013–2018) aiming to synthesize the evidence regarding the influence of dairy product consumption on the risk of major cardiovascular-related outcomes and how various doses of different dairy products affect the responses, as well as on selected biomarkers of cardiovascular disease risk, i.e., blood pressure and blood lipids. The search strategies for both designs were conducted in the MEDLINE, EMBASE, Cochrane Central Register of Controlled Trials, Cochrane Database of Systematic Reviews, and Web of Science databases from their inception to April 2018. From the 31 full-text articles retrieved for cohort studies, 17 met the eligibility criteria. The pooled risk ratio estimated for the association between the consumption of different dairy products at different dose-responses and cardiovascular outcomes (CVD, CHD, and stroke) showed a statistically significant negative association with RR values <1, or did not find evidence of significant association. The overview of 12 meta-analyses involving RCTs as well as the updated meta-analyses of RCTs did not result in significant changes on risk biomarkers such as systolic and diastolic blood pressure and total cholesterol and LDL cholesterol. Therefore, the present study states that the consumption of total dairy products, with either regular or low fat content, does not adversely affect the risk of CVD.

Introduction

Cardiovascular disease (CVD) is the leading cause of disability and premature death throughout the world. An estimated 422.7 million cases of CVD and 17.92 million CVD deaths occurred in 2015; ischemic heart disease (IHD) and stroke were the first and second leading causes of CVD globally (1). Hence, efforts to provide primary and secondary prevention worldwide are required to reduce the burden of IHD, stroke, and other related CVDs, such as peripheral arterial disease.

Risk factor modification, including improvement of dietary habits, can reduce clinical events in people with established CVD, as well as in those who are at high cardiovascular disease risk due to ≥1 risk factors. A balanced and varied diet within the context of a healthy lifestyle is considered to be the most important CVD prevention strategy (2).

Among the foods that feature in the Western diet and other industrialized countries, regular milk as well as dairy products have outstanding roles because the wide range of nutrients (proteins, fats, carbohydrates, and minerals) present in their composition play a fundamental part in the human diet (3). However, because milk and dairy products containing milk fat are major food sources of SFAs, they have been linked to an increased risk of CVD derived from higher blood concentrations of LDL cholesterol (4, 5). Therefore, current recommendations by some health authorities and governments usually advise the consumption of low-fat or fat-free milk and milk-derived products rather than regular-fat dairy foods (6). Today, these recommendations are seriously questioned by meta-analyses of both observational studies and randomized controlled trials (RCTs) reporting inconsistent results regarding the association between dairy products and CVD risk regardless of dairy fat content (7). Moreover, fermented dairy, such as yogurt or cheese, may play a special protective role, possibly due to the influence of the food matrix in the cardiometabolic response to saturated fat (8–10). Likewise, recent studies consider milk to be one of the most important sources of natural bioactive components (11), which are otherwise difficult to obtain in diets with limited use of dairy products.

In recent years, a number of reviews and meta-analyses have focused on the influence of dairy product intake on cardiovascular-related clinical outcomes, i.e., CVD, coronary heart disease (CHD), and stroke (12–17). However, the conclusions of these meta-analyses were not uniform. In addition, no meta-analysis of RCTs for major CVD events was found. However, RCTs may provide important information on causality, including the ability of dietary interventions to affect biomarkers of future disease (18). Increased blood pressure and LDL cholesterol, apart from other altered plasma lipid concentrations, namely low HDL cholesterol, high TGs, high apoB, and small, dense LDL, are considered to be key targets for the prevention and treatment of CVD (19, 20). Because atherosclerosis, vascular endothelial dysfunction, and impaired coagulation are processes usually linked to CVD, some inflammatory biomarkers, e.g., C-reactive protein (CRP) and IL-6, as well as a number of vessel adhesion molecules and coagulation factors have also been related to an increased risk of CVD (21–23). A number of meta-analyses and recent reviews have addressed how dairy product consumption affects biomarkers of CVD (24–35); the global evidence suggests that there are no harmful effects, irrespective of the content of dairy fat, on a large array of cardiometabolic variables, including lipid-related risk factors, blood pressure, inflammation, insulin resistance, and vascular function (36).

The present overview of systematic reviews and meta-analyses aimed to synthesize the evidence regarding the influence of dairy product consumption on major cardiovascular-related outcomes and how different doses of specific dairy products affect the responses. Likewise, we report an overview of meta-analyses of RCTs summarizing the effects of dairy consumption on selected cardiometabolic risk factors.

Methods

This overview of systematic reviews and meta-analyses was registered through the International Prospective Register of Systematic Reviews (PROSPERO) (CRD42018093723) and reported according to the statements for Meta-analysis of Observational Studies in Epidemiology (MOOSE) (37) and Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (38). Likewise, it was guided by previous reviews of reviews (39, 40) and the Cochrane Collaboration Handbook recommendations (41). In addition, an updated review was performed of RCTs and meta-analyses conducted to investigate the effect of milk and other dairy products on blood pressure and lipid biomarkers of cardiometabolic risk. We did not attempt to investigate the effects on inflammation, endothelial dysfunction, and coagulation factors, because this has been specifically addressed in another article of the present supplement on Role of Milk and Dairy Products in Health and Prevention of Noncommunicable Chronic Diseases: A Series of Systematic Reviews (23).

Search strategy

An electronic search strategy was conducted in the MEDLINE, EMBASE, Cochrane Central Register of Controlled Trials, Cochrane Database of Systematic Reviews, and Web of Science databases from their inception to April 2018. The following terms were combined to design the search strategy: 1) “angina,” “athero,” “cardiac arrest,” “cardiac death,” “cardiovascular,” “cardiovascular diseases,” CVD, “cardiovascular mortality,” “cerebrovascular,” “cerebrovascular accident,” “coronary artery disease,” “coronary death,” “coronary heart disease,” CHD, “death sudden,” “heart attack,” “heart attacks,” “heart disease,” “heart diseases,” “heart failure,” “ischemia,” “ischemic,” “myocardial infarction,” “myocardial infarctions,” “stroke,” “strokes,” “sudden death,” or “vascular diseases”; 2) dairy, milk, yogurt, cheese, kefir, butter, or “dairy products”; and 3) review or meta (see Table 1, Supplementary Tables 1 and 2). In addition, reference lists of identified systematic reviews and meta-analyses were reviewed.

TABLE 1.

Characteristics of the meta-analyses included for cohort studies analyzing the effects of dairy consumption on major cardiovascular events¹

Authors	n	Age, y	Follow-up, y	n (cases)	Outcome observed	Exposure observed	RR (95% CI)	I 2 (%)	Risk of bias (AMSTAR-2)
Alexander et al. (13)2	31 cohort studies (33 populations)	16–103	12–26	45,315 (NA)	CVD	Total dairy products3 Milk3 Cheese3 Yogurt3	0.88 (0.75, 1.04) 0.94 (0.86, 1.03) 0.89 (0.78, 1.01) 0.93 (0.78, 1.12)	52.7 38.1 13.0 43.4	Good
				183,020 (NA)	CHD	Total dairy products3High-fat dairy products3 Low-fat dairy products3 <300 g total dairy products/d4 300–600 g total dairy products/d4 >600 g total dairy products/d4 Milk3 <244 g milk/d4 244–488 g milk/d4 >488 g milk/d4 Cheese intake3 <18 g cheese/d4 18–53 g cheese/d4 >53 g cheese/d4 Yogurt intake3	0.91 (0.80, 1.04) 1.05 (0.93, 1.19) 0.90 (0.82, 0.98)* 0.88 (0.80, 0.96)* 0.93 (0.85, 1.00) 0.86 (0.79, 0.94)* 1.05 (0.95, 1.16) 0.99 (0.86, 1.13) 1.02 (0.93, 1.10) 0.98 (0.86, 1.12) 0.82 (0.72, 0.93)* 1.00 (0.92, 1.07) 0.86 (0.75, 0.97)* 0.92 (0.87, 0.97)* 1.08 (0.91, 1.28)	52.8 29.3 0.0 63.0 51.6 0.0 5.0 0.0 0.0 29.3 0.0 0.0 50.6 0.0 41.7
				286,474 (NA)	Stroke	Total dairy products3 High-fat dairy products3 Low-fat dairy products3 <300 g total dairy products/d4 >300 g total dairy products/d4 Milk3 <244 g milk/d4 244–488 g milk/d4 >488 g milk/d4 Cheese intake3 <18 g cheese/d4 18–53 g cheese/d4 >53 g cheese/d4	0.91 (0.83, 0.99)* 0.91 (0.84, 0.99)* 0.90 (0.83, 0.96)* 0.92 (0.89, 0.96)* 0.91 (0.88, 0.95)* 0.90 (0.79, 1.02) 0.95 (0.86, 1.04) 0.98 (0.90, 1.06) 1.01 (0.92, 1.11) 0.87 (0.77, 0.99)* 1.00 (0.92, 1.07) 0.86 (0.75, 0.97)* 0.92 (0.87, 0.97)*	44.5 0.0 0.0 0.0 0.0 79.6 76.7 38.2 44.9 33.5 0.0 50.6 0.0
				NA (NA)	Ischemic stroke	Total dairy products3 Milk	0.96 (0.76, 1.20) 0.93 (0.81, 1.06)	73.2 75.6
				NA (NA)	Hemorrhagic stroke	Milk3	0.93 (0.69, 1.25)	86.6
Bechthold et al. (42)	24 cohort studies(27 populations)	20–100	5–26	351,683 (14,614)	CHD	Total dairy products3 200 g total dairy products/d4 High-fat dairy products3 Low-fat dairy products3	0.99 (0.92, 1.07) 0.99 (0.96, 1.02) 1.01 (0.96, 1.06) 0.96 (0.90, 1.03)	59.0 55.0 9.0 42.0	Excellent
				419,782 (19,207)	Stroke	Total dairy products3 200 g total dairy products/d4 High-fat dairy products3 Low-fat dairy products3	0.96 (0.90, 1.01) 0.98 (0.96, 1.00) 0.93 (0.87, 0.99)* 0.97 (0.91, 1.04)	43.0 50.0 32.0 39.0
					Ischemic stroke	Total dairy products3	0.95 (0.88, 1.01)	26.0
					Hemorrhagic stroke	Total dairy products3	1.01 (0.84, 1.21)	20.0
				85,372 (4057)	Heart failure	Total dairy products3 200 g total dairy products/d4	1.00 (0.90, 1.10) 1.08 (1.01, 1.15)	67.0 NA
Chen et al. (16)	15 cohort studies (15 populations)	16–93	8–16	102,013 (8076)	CVD	Cheese3 High-fat cheese3 Low-fat cheese3 50 g cheese/d4	0.90 (0.82, 0.99)* 0.74 (0.44, 1.24) 1.00 (0.77, 1.29) 0.92 (0.83, 1.02)	0.0 67.0 NA 16.9	Very good
				121,226 (7631)	CHD	Cheese3 High-fat cheese3 Low-fat cheese3 50 g cheese/d4	0.86 (0.77, 0.96)* 0.83 (0.68, 1.01) 1.13 (0.63, 2.05) 0.90 (0.84, 0.95)*	14.9 NA 90.5 0.0
				257,069 (10,449)	Stroke	Cheese3 50 g cheese/d4	0.90 (0.84, 0.97)* 0.94 (0.84, 1.04)	0.0 63.7
Elwood et al. (43)	10 cohort studies and 2 case control studies (12 populations)	16–79	8–28	401,682 (8850) 89,598 (4820) 320,361 (4030)	CVD IHD Stroke	Milk intake3 Milk intake3 Milk intake3	0.84 (0.78, 0.90)* 0.87 (0.74, 1.03) 0.83 (0.77, 0.90)*	NA NA NA	Acceptable
Elwood et al. (44)	15 cohort studies and 4 case control studies (14 populations)	NA	8–28	239,317 (5835) 414,097 (14,358) 2350 (1011)	IHD Stroke MI	Total dairy products3 Total dairy products3 Total dairy products3	0.84 (0.76, 0.93)* 0.79 (0.75, 0.82)* 0.83 (0.66, 0.99)*	NA NA NA	Acceptable
Elwood et al. (45)	25 cohort studies (25 populations)	NA	5–68	64,322 (10,121) 116,828 (11,019)	CVD	Butter3 Cheese3	0.93 (0.84, 1.02) 1.32 (0.49, 3.56)	NA NA	Acceptable
				379,503 (10,059) 587,690 (9725) 388,371 (5946) 121,469 (484)	IHD Ischemic stroke Hemorrhagic stroke Subarachnoid bleeds	Total dairy products3 Total dairy products3 Total dairy products3 Total dairy products intake3	0.92 (0.80, 0.99)* 0.79 (0.68, 0.91)* 0.75 (0.60, 0.94)* 0.65 (0.32, 1.31)	NA NA NA NA
Gholami et al. (15)	27 cohort studies (31 populations)	8–97	5–65	140,851 (8648) 55,421 (NA) 85,430 (NA) 471,970 (11,806)	CVD (fatal and nonfatal) CVD incidence CVD mortality CHD (fatal and nonfatal)	Total dairy products3 Total dairy products3 Total dairy products3 Total dairy products3 High-fat dairy products3 Low-fat dairy products3	0.90 (0.81, 0.99)* 0.93 (0.84, 1.04) 0.87 (0.74, 1.03) 0.99 (0.92, 1.06) 0.98 (0.94, 1.01) 1.01 (0.94, 1.09)	55.9 32.5 64.6 51.6 2.4 62.6	Very good
				190,494 (7787) 281,476 (4019) 764,917 (29,300)	CHD incidence CHD mortality Stroke (fatal and nonfatal)	Total dairy products3 Total dairy products3 Total dairy products3 High-fat dairy products3 Low-fat dairy products3	0.97 (0.90, 1.04) 1.03 (0.88, 1.21) 0.88 (0.82, 0.95)* 0.95 (0.91, 1.00) 0.94 (0.90, 0.98)*	44.9 58.1 63.1 0.0 0.0
				297,446 (13,979)	Stroke incidence	Total dairy products3	0.96 (0.88, 1.04)	49.7
				467,471 (15,321)	Stroke mortality	Total dairy products3	0.80 (0.76, 0.83)*	0.0
Gholami et al. (46)	15 cohort studies (16 populations)	8–83	10–65	135,126 (8011)	CHD	Total dairy products3 Milk3 Butter3 Cheese3 Cream3	0.97 (0.93, 1.02) 1.05 (0.96, 1.15) 0.99 (0.89, 1.11) 0.90 (0.81, 1.01) 0.96 (0.87, 1.06)	20.7 0.0 21.2 47.4 0.0	Very good
				496,943 (23,477)	Stroke	Total dairy products3 Milk3 Butter3 Cheese3 Cream3	0.93 (0.88, 0.98)* 0.91 (0.81, 1.01) 0.95 (0.85, 1.07) 0.93 (0.88, 0.99)* 0.97 (0.88, 1.06)	54.2 71.4 0.0 0.0 0.0
de Goede et al. (47)	18 cohort studies (20 populations)	30–83	8–26	336,118 (12,425) 158,595 (6440) 158,595 (1237) 87,576 (1652)	Stroke Ischemic stroke Hemorrhagic stroke Stroke mortality	200 g total dairy products/d4 200 g fermented dairy products/d4 200 g high-fat dairy products/d4 200 g low-fat dairy products/d4 200 g milk/d4 200 g high-fat milk/d4 200 g low-fat milk/d4 10 g butter/d4 40 g cheese/d4 100 g yogurt/d4 200 g total dairy products/d4 200 g milk/d4 200 g total dairy products/d4 200 g milk/d4 200 g total dairy products/d4 200 g fermented dairy products/d4 200 g milk/d4	0.99 (0.96, 1.02) 0.91 (0.82, 1.01) 0.96 (0.93, 0.99)* 0.97 (0.95, 0.99)* 0.93 (0.88, 0.98)* 1.04 (1.02, 1.06) 0.96 (0.90, 1.03) 1.00 (0.99, 1.01) 0.97 (0.94, 1.01) 1.02 (0.90, 1.17) 1.00 (0.96, 1.04) 0.95 (0.89, 1.01) 1.02 (0.98, 1.06) 0.90 (0.74, 1.09) 0.97 (0.85, 1.11) 0.80 (0.67, 0.95)* 0.88 (0.81, 0.96)*	65.6 64.5 0.0 0.0 86.0 0.0 68.2 0.0 31.2 47.8 83.6 67.6 94.4 0.0 65.1 0.0 65.3	Very good
Guo et al. (14)	26 cohort studies (28 populations)	34–67	5–25	76,207 (5525) 330,350 (8298)	CVD CHD	200 g total dairy products/d4 200 g fermented dairy products/d4 200 g high-fat dairy products/d4 200 g low-fat dairy products/d4 244 g milk/d4 10 g cheese/d4 50 g yogurt/d4 200 g total dairy products/d4 200 g fermented dairy products/d4 200 g high-fat dairy products/d4 200 g low-fat dairy products/d4 244 g milk/d4 10 g cheese/d4 50 g yogurt/d4	0.97 (0.91, 1.02) 0.98 (0.97, 0.99)* 0.93 (0.84, 1.03) 0.98 (0.95, 1.01) 1.01 (0.93, 1.10) 0.98 (0.95, 1.00) 1.03 (0.97, 1.09) 0.99 (0.96, 1.02) 0.99 (0.98, 1.01) 0.99 (0.93, 1.05) 1.00 (0.97, 1.03) 1.01 (0.96, 1.06) 0.99 (0.97, 1.02) 1.03 (0.97, 1.09)	59.9 87.5 37.4 0.0 92.4 82.6 0.0 38.9 44.6 22.9 27.3 45.5 40.3 0.0	Very good
Hu et al. (48)	15 cohort studies (18 populations)	30–103	10–65	764,635 (28,138) 293,320 (13,415) 471,315 (14,723) 456,420 (12,439) 451,847 (6625)	Stroke (fatal and nonfatal) Stroke incidence Stroke mortality Ischemic stroke Hemorrhagic stroke	Total dairy products3 High-fat dairy products3 Low-fat dairy products3 Milk3 Fermented milk3 Nonfermented milk3 Butter3 Cheese3 Cream3 Total dairy products3 Total dairy products3 Total dairy products3 Total dairy products3	0.88 (0.82, 0.94)* 0.96 (0.92, 1.01) 0.91 (0.85, 0.97)* 0.91 (0.82, 1.01) 0.80 (0.71, 0.89)* 1.02 (0.89, 1.17) 0.95 (0.85, 1.07) 0.94 (0.89, 1.00) 0.97 (0.88, 1.06) 0.95 (0.87, 1.03) 0.80 (0.76, 0.84)* 0.92 (0.82, 1.03) 0.96 (0.73, 1.25)	61.8 0.0 41.6 74.4 0.0 0.0 0.0 0.0 0.0 50.9 0.0 63.3 82.7	Good
Larsson et al. (49)	4 cohort studies (6 populations)	25–79	11–25	209,046 (49,955)	CVD mortality	Nonfermented milk3	NA	93.0	Acceptable
Mullie et al. (50)	15 cohort studies (15 populations)	34–74	10–25	403,776 (37,049) 564,717 (39,352)	CHD (fatal and nonfatal) Stroke (fatal and nonfatal)	200 g milk/d4 200 g milk/d4	1.01 (0.98, 1.05) 0.91 (0.82, 1.02)	16.0 92.0	Good
Pimpin et al. (51)	4 cohort studies (5 populations)	55–70.6	10–16.2	175,612 (9783) 149,056 (4484) 173,853 (5299) NA	Any CVD CHD Stroke Total CVD	Butter3 Butter3 Butter3 Butter3	1.00 (0.98, 1.02) 0.99 (0.96, 1.03) 1.01 (0.98, 1.03) 0.99 (0.96, 1.02)	0.0 0.0 0.0 0.0	Very good
Qin et al. (12)	22 cohort studies (NA)	21–83	8–26.2	91,057 (7641) 504,803 (21,801) 253,260 (8792)	CVD Stroke CHD	Total dairy products3 Total dairy products3 High-fat dairy products3 Low-fat dairy products3 Yogurt3 Cheese3 Butter3 Total dairy products3 High-fat dairy products3 Low-fat dairy products3 Yogurt3 Cheese3 Butter3	0.88 (0.81, 0.96)* 0.87 (0.77, 0.99)* 0.95 (0.83, 1.08) 0.93 (0.88, 0.99)* 0.98 (0.92, 1.06) 0.91 (0.84, 0.98)* 0.94 (0.84, 1.06) 0.94 (0.82, 1.07) 1.08 (0.99, 1.17) 1.02 (0.92, 1.14) 1.06 (0.90, 1.34) 0.84 (0.71, 1.00) 1.02 (0.88, 1.20)	29.6 69.8 72.1 20.0 0.0 0.0 12.9 58.5 0.0 33.5 42.9 31.8 30.7	Very good
Soedamah-Muthu et al. (52)	17 cohort studies (NA)	34–80	5–25	13,518 (2283) 259,162 (4391) 375,381 (15,554)	CVD CHD Stroke	200 mL milk/d4 200 mL milk/d4 Total dairy product intake3 Total high-fat dairy3 Total low-fat dairy3 Total milk3	0.94 (0.89, 0.99)* 1.00 (0.96, 1.04) 1.02 (0.93, 1.11) 1.04 (0.89, 1.21) 0.93 (0.74, 1.17) 0.87 (0.72, 1.07)	0.0 26.9 26.2 0.0 55.7 94.6	Very good
Wu and Sun (53)	9 cohort studies (9 populations)	21–55	10.2–17.3	291,236 (14,776) 77,510 (4381) 225,141 (7875)	Developing CVD CHD Stroke	Yogurt3 Yogurt3 Yogurt3	0.99 (0.96, 1.11) 1.04 (0.95, 1.15) 1.02 (0.92, 1.13)	52.0 36.0 58.0	Very good

¹ *P < 0.05. CHD, coronary heart disease; CVD, cardiovascular disease; NA, not available.

²Servings per day were converted into grams per day (a single serving of milk as 244 g, 1 serving of cheese as 35 g, and 1 serving of total dairy, high-fat dairy, and low-fat dairy as 200 g).

³High compared with low intake.

⁴Per each increment of the cited dairy products.

The literature search was independently performed by 2 reviewers (VM-V and JF), and disagreements were solved by consensus or involving a third researcher (AG).

For the investigation of RCTs published from January, 2013 to August, 2018 related to dairy product consumption and biomarkers of cardiometabolic risk we used the following search equation for PUBMED (or its equivalent for the other databases): [“dairy products”(All Fields) OR “milk”(All Fields) OR “cheese”(All Fields) OR “yogurt”(All Fields) OR “yoghurt”(All Fields) OR “fermented milk”(All Fields)] AND [“blood lipids”(All Fields) OR “blood pressure”(All Fields) OR “cholesterol”(All Fields) OR “LDL-cholesterol”(All Fields) OR “HDL-cholesterol”(All Fields) OR “triglycerides”(All Fields) OR “low-density lipoprotein”(All Fields) OR “high-density lipoprotein”(All Fields) OR “non high-density lipoprotein cholesterol”(All Fields) OR “apolipoprotein B”(All Fields) OR “apolipoprotein A-I”(All Fields) OR “small dense LDL”(All Fields) OR “LDL particle size”(All Fields) OR “Lp(a)”(All Fields) OR “glucose”(All Fields) OR “insulin”(All Fields) OR “insulin resistance”(All Fields) OR “HOMA”(All Fields) OR “vascular function”(All Fields) OR “fibrinogen”(All Fields) OR “flow mediated dilation”(All Fields) OR “peripheral arterial tone”(All Fields) OR “adhesion molecules”(All Fields)] NOT [“human milk”(All Fields) OR “infant formula”(All Fields) OR “milk formula”(All Fields)] AND {Randomized Controlled Trial[ptyp] AND [“2013/01/01”(PDAT): “2018/08/31”(PDAT)]}. A similar strategy was applied to screen the meta-analyses of RCTs for biomarkers of cardiometabolic risk associated with milk and dairy product consumption.

The literature search for RCTs and RCT meta-analyses was independently performed by 2 reviewers (JF and AG), and disagreements were solved by consensus involving a third researcher (VM-V).

Study selection

Only systematic reviews and meta-analyses addressing the relation between dairy product consumption and cardiovascular outcomes were considered. Meta-analyses were included if they: 1) included longitudinal studies or data from longitudinal studies that could be isolated or extracted (in case they also included RCTs or cross-sectional studies), and 2) were written in English or Spanish. Meta-analyses not following systematic review methodology were excluded.

For the study of RCTs related to dairy product consumption and biomarkers of cardiometabolic risk, prospective, parallel, and crossover designs were considered. Studies had to administer dietary supplementation or a specific diet containing dairy products. However, those studies that used dietary recommendations or self-reporting alone were excluded. Studies were also excluded if a supplement that could potentially confound the effects of the milk or the dairy product was administered.

Search and data extraction

Observational cohort studies and major cardiovascular outcomes

The main characteristics of the selected studies were summarized in an ad hoc table, including information regarding the 1) first author and year of publication, 2) number of studies included, 3) length of follow-up, 4) sample characteristics (age distribution, number of subjects, and number of cases), 5) cardiovascular outcome observed, 6) type of dairy product assessed, 7) risk ratio estimations, 8) heterogeneity reported [I² (%)], and 9) AMSTAR 2 risk of bias value. In addition, the longitudinal studies included in each meta-analysis and the covariates included in their analysis were extracted.

From the 31 full-text articles retrieved, 17 studies met the eligibility criteria (Figure 1). These meta-analyses quantified the risk of the association between dairy product consumption and cardiovascular outcomes. The reports were published between 2004 and 2017 and provided data from 4–31 cohort studies and from 2–4 case control studies (Table 1). From the 17 systematic reviews and meta-analyses in which risk of bias was evaluated using the AMSTAR 2 tool (Supplemental Table 2), 8 of them were assessed as NOT in item 9, which addresses whether the authors have adequately evaluated the risk of bias of the included studies. Among the causes of this negative assessment, it is noted that most meta-analyses did not describe in depth the influence of the confounders and effect modification variables.

FIGURE 1.

PRISMA flow diagram for the research of meta-analyses of cohort studies addressing the effects of the consumption of dairy products and major events of cardiovascular diseases.

The sample sizes of the included studies ranged from 2350 to 764,917 participants, and participants’ age ranged from 8 to 103 y. These participants were followed for between 5 and 83 y. The main cardiovascular outcomes included in the meta-analyses were incidence and mortality of CVD, CHD, and stroke. In addition, some studies reported risk of IHD, myocardial infarction (MI), heart failure, and ischemic and hemorrhagic stroke. The maximum number of cardiovascular events, including fatal and nonfatal outcomes, was 11,019 for CVD, 37,049 for CHD, and 39,352 for stroke.

Eleven studies reported information regarding total dairy product consumption, 9 distinguished between regular-fat and low-fat dairy products (12–16, 42, 47, 48, 52), and 2 (14, 47) included information regarding fermented dairy product consumption. Furthermore, 9 studies reported information regarding milk consumption (13, 14, 44, 46–50, 52), 2 regarding high-fat and low-fat milk consumption (48, 49), 2 regarding nonfermented milk consumption (48, 49), and 1 regarding fermented milk consumption (48). In addition, cheese, yogurt, butter, and cream consumption were included as exposures by other studies (12–14, 16, 43, 46–48, 51, 53). The heterogeneity reported in the meta-analyses, as measured by I², varied from 0% to 94.6%.

From the total of 76 original cohort and case control studies, no single one was included in all of the systematic reviews and meta-analyses included in this overview and the covariates used in their analyses varied across the original studies, with demographic variables and health behaviors the most commonly used (Supplementary Table 1).

RCTs and biomarkers of cardiometabolic risk

The studies included the administration of milk or dairy products, individually or in combination, allowing for the investigation of the effects of the milk or dairy products. There were no restrictions regarding dosage or dosing regimen. The following primary outcomes were considered for the inclusion of the studies: systolic blood pressure (SBP), diastolic blood pressure (DBP), and plasma lipids (total cholesterol, LDL cholesterol, HDL cholesterol, and TGs). Other outcomes including inflammatory, endothelial dysfunction, and coagulation factors have been reported separately in another article of the present supplement (23).

As for the observational studies, the main characteristics of the selected meta-analyses involving RCTs are summarized in Tables 2 and 3. Likewise, the characteristics of the RCTs included in the updated meta-analysis (2013–2018) are shown in Table 4.

TABLE 2.

Characteristics of the meta-analyses of RCTs evaluating the effects of dairy product consumption on blood lipids¹

Author	RCTs, n	Participants, n	Age,2 y	Exposure observed	Intervention time, wk	Blood lipids	Changes,3 mmol/L	Heterogeneity4
Agerholm-Larsen et al. (24)	5	70	39.4 ± 2.1	Fermented dairy products	4–8	Total-C	−0.23 (−0.41, −0.05)	7.22 (df = 4, P = 0.88)
						LDL-C	−0.25 (−0.48, −0.01)	8.88 (df = 4, P = 0.94)
Sun and Buys (25)	15	788	>18	Fermented milk or yogurt	>8	Total-C	−0.27 (−0.38, −0.16)	35.5%
						LDL-C	−0.23 (−0.33, −0.13)	56.6%
Benatar et al. (26)	9	702	51 ± 16	Total dairy products	26	LDL-C	0.05 (−2.89, 6.60)	64.0%
				Whole-fat dairy vs. low-fat			−0.005 (−2.10, 1.71)	0.0%
de Goede et al. (27)	5	5–49	22–56	Cheese vs. butter	2–8	Total-C	−0.28 (−0.36, −0.19)	0.0%
						LDL-C	−0.22 (−0.29, −0.14)	0.0%
Shimizu et al. (28)	11	13–152	All ages	Fermented milk products and probiotics	4–8	Total-C	−0.17 (−0.27, −0.07)	59%
						LDL-C	−0.22 (−0.30, −0.13)	41%

¹Servings per day were converted into grams per day. LDL-C, LDL cholesterol; RCT, randomized controlled trial; total-C, total cholesterol.

²Values are means ± SDs or ranges.

³Values are changes (95% CIs).

⁴Values are Q values when the heterogeneity of effect size was tested with Q statistics based on chi-square distribution (P < 0.05 was considered statistically significant), or I² (%) when the test for heterogeneity was assessed via the I² statistic, which expresses the percentage of variation attributable to between-study heterogeneity.

TABLE 3.

Characteristics of the meta-analyses of RCTs evaluating the effects of dairy consumption on blood pressure¹

Author	RCTs, n	Participants, n	Age,2 y	Intervention time	Exposure observed	Blood pressure	Changes3 (mm Hg)	Heterogeneity4
Ding et al. (29)	8	753	20–71	1–12 mo	Total dairy products	DBP	−0.21 (−0.98, 0.57)	0.0%
Hidayat et al. (30)	7	412	23.4–61.1	1–24 mo	Total dairy products	SBP DBP	−3.33 (−5.62, −1.03) −1.08  (−3.38, −0.22)	0.0% 0.0%
Benatar et al. (26)	20	1677	51 ± 16	26 wk	Total dairy products	SBP DBP	−0.41 (−1.60, 0.81) −0.45 (−1.70, 0.80)	0.0% 40.0%
Cicero et al. (31)	14	1306	40–58	4–21 wk	Total dairy products	SBP DBP	−1.28 (−2.09, −0.48) −0.59 (−1.18, −0.01)	P = 0.13
Dong et al. (32)	14	702	39–75	4–24 wk	Probiotic fermented milk	SBP DBP	−3.10 (−4.64, −1.56) −1.09 (−2.11, −0.06)	24.1% 29.0%
Turpeinen et al. (33)	19	1500	35–70	4–21 wk	Total dairy products	SBP DBP	−4.00 (−5.90, −2.10) −1.90 (−3.10, −0.80)	75.0% 70.0%
Usinger et al. (34)	15	1232	≥18	≥4 wk	Fermented milk	SBP DBP	−2.45 (−4.30, −0.60) −0.67 (−1.48, 0.14)	71.0% 39.0%
Xu et al. (35)	12	623	43–75	4–21 wk	Milk-derived tripeptides IPP-VPP	SBP DBP	−4.80 (−6.00, −3.70) −2.20 (−3.10, −1.30)	16.2, P > 0.1 11.5, P > 0.25

¹DBP, diastolic blood pressure; IPP-VPP, isoleucine, proline, proline-valine, proline, proline; RCT, randomized controlled trial; SBP, systolic blood pressure.

²Values are means ± SDs or ranges.

³Values are changes (95% CIs).

⁴Values are Q values when the heterogeneity of effect size was tested with Q statistics based on chi-square distribution (P < 0.05 was considered statistically significant), or I² (%) when the test for heterogeneity was assessed via the I² statistic, which expresses the percentage of variation attributable to between-study heterogeneity.

TABLE 4.

Characteristics of the randomized controlled trials included and published from 2013 to 2018 for blood lipids and blood pressure biomarkers¹

Authors	Type of study (P or CO)	Sample size (n)	Subjects’ age (y), BMI (kg/m2), and other characteristics2	Exposure	Dosage	Period of intervention	Changes in blood pressure (SBP and DBP) (mm Hg)3	Changes in plasma lipids (total-C and LDL-C) (mmol/L)3
Drouin-Chartier et al. (56)	CO	76	Age: 53.3 ± 12.2; BMI: 28.2 ± 3.7	Dairy products	3.4 servings dairy products/d	4 wk	SBP = −1.0 (−3.86, 1.86) DBP = 0.00 (−2.71, 2.71)	ND
Drouin-Chartier et al. (57)	CO	27	Age: 57 ± 5.0; BMI: 31.9 ± 3.5	Milk	3.2 servings 2% fat milk/d	6 wk	ND	Total-C = 0.07 (−0.37, 0.51) LDL-C = −0.01 (−0.41, 0.39)
Fathi et al. (58)	P	75	Age: 25–45; BMI: 25–34.9	Dairy products	Milk	8 wk	ND	Total-C = −0.26 (−0.31, −0.21) LDL-C = −0.24 (−0.28, −0.20)
					Kefir			Total-C = −0.41 (−0.46, −0.36) LDL-C = −0.40 (−0.44, −0.36)
Machin et al. (59)	CO	49	Age: 53 ± 2; BMI: 30.5; elevated blood pressure	Dairy products	High-dairy (+4 servings conventional nonfat dairy products/d)	4 wk	ND	Total-C = 0.10 (0.02, 0.18) LDL-C = 0.03 (−0.01, 0.07)
Tanaka et al. (60)	P	200	Age: 20–60; BMI: 25	Dairy products	400 g dairy products/d	24 wk	SBP = 2.20 (−1.68, 6.08) DBP = 2.00 (−1.04, 5.04)	Total-C = 0.20 (−0.02, 0.42) LDL-C = 0.23 (0.04, 0.42)
Maki et al. (61)	CO	62	Age: 54.5; BMI: 29.2	Dairy products	1 serving low-fat dairy/d	5 wk	ND	Total-C = 0.03 (0.01, 0.05) LDL-C = 0.02 (0.01, 0.03)
Rideout et al. (62)	CO	23	Age: 18–75; BMI: 18.5–35.0	Dairy products	Low dairy (<2 servings dairy products/d) High dairy (4 servings dairy products/d)	1 y	SBP = −1.20 (−11.07, 8.67) DBP = 0.70 (−8.02, 9.42) SBP = −3.80 (−13.24, 5.64) DBP = −1.20 (−8.04, 5.64)	Total-C = −0.33 (−0.78, 0.12) LDL-C = −0.19 (−0.59, 0.21) Total-C = −0.13 (−0.54, 0.28) LDL-C = −0.09 (−0.49, 0.31)
Conway et al. (63)	CO	34	Age: 18 and 65; BMI: ≤35	Buttermilk	45 g buttermilk/d	4 wk	ND	Total-C = −0.18 (−0.60, 0.24) LDL-C = −0.12 (−0.44, 0.20)
Conway et al. (64)	CO	34	Age: 18 and 65; BMI: ≤35	Buttermilk	45 g buttermilk/d	4 wk	SBP = −2.60 (−7.88, 2.68) DBP = −1.20 (−5.22, 2.82)	ND
Benatar et al. (65)	P	180	Age: 46.7 ± 1.7; BMI: 24.5 ± 0.3	Dairy products	Low dairy intake Medium dairy intake High dairy intake	4 wk	SBP = −1.20 (−4.91, 2.51) DBP = −1.90 (−4.78, 0.98) SBP = 0.90 (−3.09, 4.89) DBP = −2.50 (−5.49, 0.49) SBP = −0.10 (−3.69, 3.89) DBP = −0.50 (−3.39, 2.39)	LDL-C = −0.11 (−0.41, 0.19) LDL-C = −0.09 (−0.40, 0.22) LDL-C = 0.07 (−0.22, 0.36)
Soerensen et al. (66)	CO	15	Age: 18–50; BMI: 20–28	Milk Cheese	Semi-skimmed milk (1700 mg Ca/d)]. Semihard cow cheese (1700 mg Ca/d)].	3 × 2 wk	SBP = −0.80 (−6.84, 5.24) DBP = −1.40 (−6.52, 3.72) SBP = −2.00 (−6.81, 2.81) DBP = −1.20 (−6.05, 3.65)	Total-C = −0.29 (−1.03, 0.45) LDL-C = −0.33 (−1.02, 0.36) Total-C = −0.28 (−1.00, 0.44) LDL-C = −0.24 (−0.91, 0.43)
Raziani et al. (67)	P	139	Age: 18–70; BMI: 18.5–37.5	Cheese	Regular-fat cheese Reduced-fat cheese	12 wk	SBP = 2.00 (−4.16, 8.16) DBP = −0.30 (−3.94, 3.34) SBP = −1.20 (−7.29, 4.89) DBP = −1.90 (−5.56, 1.76)	Total-C = 0.16 (0.10, 0.22) LDL-C = 0.08 (0.03, 0.13) Total-C = 0.04 (−0.02, 0.10) LDL-C = 0.02 (−0.03, 0.07)

¹CO, crossover; DBP, diastolic blood pressure; LDL-C, LDL cholesterol; ND, not determined; P, prospective; SBP, systolic blood pressure; total-C, total cholesterol.

²Values are means ± SDs or ranges.

³Values are changes (95% CIs).

From the 15 full-text meta-analyses retrieved, 12 studies met the eligibility criteria (Supplemental Figure 1). Likewise, from the 30 RCT studies retrieved from 2013 to 2018, only 12 met the inclusion criteria for the updated meta-analysis study (Supplemental Figure 2). These meta-analyses quantified the risk of consumption of dairy products on blood lipids and blood pressure biomarkers.

Risk of bias and certainty of evidence

Observational cohort studies and major cardiovascular outcomes

The AMSTAR 2 tool (54) was used to evaluate the quality of the included systematic reviews and meta-analyses. This tool includes 16 criteria, each referring to a relevant methodological aspect of the study (e.g., there should be a previous study protocol, ≥2 independent data extractors, a consensus procedure for disagreements, and a search in ≥2 electronic sources).

Because the AMSTAR 2 tool does not have established categories of quality, the included systematic reviews and meta-analyses were grouped according to the number of criteria met as follows: excellent, 15–16; very good, 12–14; good, 9–11; acceptable, 6–8; and deficient, 3–5; 23.5% of studies scored as acceptable, 17.6% as good, 52.9% as very good, and 5.9% as excellent in terms of risk of bias (54) (Supplemental Table 2). When individual domains were considered, no studies reported a list of excluded studies, and 88.2% of the studies did not report the included studies’ funding information.

The certainty of evidence for the considered meta-analyses was assessed by using Grading of Recommendations Assessment, Development and Evaluation (GRADE) (Supplemental Tables 3–5).

Updated meta-analysis of RCTs and biomarkers of cardiometabolic risk

We performed an updated meta-analysis for the RCTs published from 2013 to 2018. Two authors (JF and AG) independently assessed the risk of bias of selected RCTs following the Cochrane Collaboration's methodology (55). In case of discrepancies, a third reviewer was involved in the evaluation (VM-V). The Cochrane tool includes different domains related to random assignment and allocation concealment (selection bias), blinding (performance and measurement bias), loss to follow-up and adherence to the intention-to-treat principle (attrition bias), and selective outcome publication (reporting bias). In addition, other potential sources of bias, such as private or public funding, were included. The risk of bias was tabulated for each study and was classified as low, high, or unclear, as described in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (55).

The majority of RCTs selected for the update lacked specific reporting of the allocation concealment and because of the nature of the interventions it was not possible to follow a double-blind design. The detection bias is another concern owing to the insufficient details provided by the authors in relation to the blinding of the people who assessed the clinical outcomes. However, we did not detect a potential bias in the reporting of outcome data (Supplemental Figures 3 and 4).

Display of the results

Observational cohort studies and major cardiovascular outcomes

Forest plots were used to display the pooled risk ratio estimates for the association between dairy products and cardiovascular outcomes in the included studies. Forest plots were performed considering the relation between each main cardiovascular outcome (CVD, CHD, and stroke) and 1) dairy product exposure (high compared with low intake or dose-response) and 2) the dairy product subgroup (total dairy products, milk, cheese, yogurt, and butter). To be included in the forest plot, the pooled risk ratio reported by meta-analyses should include ≥4 studies. Forest plot graphs were created using Stata SE software, version 15 (StataCorp).

For studies that reported dairy product consumption as servings per day, we considered a single serving of milk as 244 g, a serving of cheese as 35 g, and 1 serving of total dairy, high-fat dairy, and low-fat dairy as 200 g, similar to previous meta-analyses (13).

RCTs and biomarkers of cardiometabolic risk

Forest plots were also used to display the pooled risk ratio estimates for the associations between dairy products and biomarkers of cardiometabolic risk in the included meta-analyses. Forest plots were performed considering the relation with each main risk outcome, e.g., SBP and DBP, total cholesterol, and LDL cholesterol.

We performed the corresponding meta-analyses for these outcomes with RCTs published from January 2013 to 2018 to update the potential effects of milk and dairy products on some biomarkers of cardiometabolic risk, namely blood pressure and blood lipids. The main outcome variables for the meta-analyses were the absolute changes in SBP and DBP (mm Hg) and the total cholesterol and LDL-cholesterol concentrations (mmol/L) between the intervention and control groups. The variance measures for the absolute changes in the outcomes were reported as means and 95% CIs.

Pooled effect sizes were calculated for the absolute changes in the outcomes after the dairy product interventions using a random-effects model, which takes into account within- and between-study variation. The estimated results were displayed as forest plots. Between-study heterogeneity was quantified using the I² statistic. Because the presence of heterogeneity may affect the statistical validity of the summary estimate of effect, the Q statistic was used to test the null hypothesis of statistical validity; P values <0.05 were considered significant.

Sensitivity analyses were conducted to assess whether any single study exerted undue influence on the overall results. This was conducted by excluding 1 study at a time from the analyses and recalculating the effect size each time. Publication bias was visually assessed using a funnel plot (68).

Statistical analysis was conducted using the R free software environment for statistical computing and graphics, version 3.4.4 (R Foundation for Statistical Computing). The “metafor” meta-analysis package for R was used for calculations and data visualizations.

Results

Dairy product consumption and cardiovascular outcomes

Table 1 shows the characteristics of the meta-analyses included in the present study and the RRs for dairy intake and incident CVD, CHD, and stroke (high compared with low consumption, regular- compared with low-fat content, and dose–response intake).

Total dairy products

The association between high compared with low total dairy product consumption and CVD (total incidence and mortality), CHD, and stroke was analyzed in several meta-analyses. Collectively, the intake of dairy products was not associated with CVD. Moreover, 2 meta-analyses (12, 15) reported a significant negative association (RR: 0.88; 95% CI: 0.81, 0.96 and RR: 0.90; 95% CI: 0.81, 0.99, respectively) (Figure 2). Guo et al. (14) did not find evidence of a significant association between a 200 g/d increment of total or high- or low-fat dairy product consumption (dose-response) and CVD risk (Figure 3).

FIGURE 2.

Forest plot for meta-analyses evaluating the influence of high compared with low dairy product consumption on CVD. The effect size and 95% CI for fully adjusted random effects are depicted for each meta-analysis. CVD, cardiovascular disease.

FIGURE 5.

Forest plot for meta-analyses evaluating the influence of dose-response of dairy product consumption on CHD. The effect size and 95% CI for fully adjusted random effects are depicted for each meta-analysis. CHD, coronary heart disease.

The risk of CHD (total, incidence, or mortality) was studied in 5 meta-analyses (Figure 4). Globally, total dairy product intake was neutral for CHD risk and the same results were found for consumption of high-fat dairy products. A significantly lower risk was found for low-fat dairy products (RR: 0.90; 95% CI: 0.82, 0.98) (13). Other authors (43, 45) also found significantly lower risk of IHD (RR: 0.84; 95% CI: 0.76, 0.93 and RR: 0.92; 95% CI: 0.80, 0.99, respectively). MI was studied in 1 meta-analysis (45) and the results also showed a significantly reduced risk with high consumption of total dairy products (RR: 0.83; 95% CI: 0.66, 0.99) (Figure 4). On the other hand, in 1 meta-analysis (42) the association of dairy product intake with the incidence of heart failure risk was analyzed and no significant association was found (Figure 4).

FIGURE 3.

Forest plot for meta-analyses evaluating the influence of dose-response of dairy product consumption on CVD. The effect size and 95% CI for fully adjusted random effects are depicted for each meta-analysis. CVD, cardiovascular disease.

The dose-response for CHD and total dairy products was considered in 4 meta-analyses (Figure 5). Three found no significant differences with an intake increase of 200 g/d (14, 42, 52). Alexander et al. (13) analyzed the consumption of <300 g/d, 300–600 g/d, and >600 g/d and reported significantly lower risk of CHD with increments of 300 g/d and 600 g/d (RR: 0.88; 95% CI: 0.80, 0.96 and RR: 0.90; 95% CI: 0.79, 0.94, respectively). An increment of consumption of 200 g/d of low- or high-fat dairy products was not associated with CHD (43, 52).

FIGURE 6.

Forest plot for meta-analyses evaluating the influence of high compared with low dairy product consumption on stroke. The effect size and 95% CI for fully adjusted random effects are depicted for each meta-analysis.

Seven meta-analyses analyzed the risk of stroke (total incidence or mortality) with the consumption of total dairy products (Figure 6). Six (12, 13, 15, 45, 46, 48) reported an inverse statistically significant association (RR: 0.87; 95% CI: 0.77, 0.99; RR: 0.91; 95% CI: 0.83, 0.99; RR: 0.88; 95% CI: 0.82, 0.95; RR: 0.79; 95% CI: 0.75, 0.92; RR: 0.93; 95% CI: 0.88, 0.98; and RR: 0.88; 95% CI: 0.82, 0.94, respectively). The same results were found for stroke mortality in 2 studies (16, 48) (RR: 0.80; 95% CI: 0.76, 0.84 and RR: 0.80; 95% CI: 0.76, 0.83, respectively).

FIGURE 4.

Forest plot for meta-analyses evaluating the influence of high compared with low dairy product consumption on CHD. The effect size and 95% CI for fully adjusted random effects are depicted for each meta-analysis. CHD, coronary heart disease.

The association between consumption of regular-fat and low-fat total dairy products and stroke was followed in 5 meta-analyses; 2 (13, 42) found significantly lower risk with high-fat products (RR: 0.91; 95% CI: 0.84, 0.99 and RR: 0.93; 95% CI: 0.87, 0.99, respectively). A high consumption of low-fat dairy products was also reported to have an inverse association with the risk of stroke (12, 46, 48): RR: 0.93; 95% CI: 0.88, 0.99; RR: 0.91; 95% CI: 0.85, 0.97; and RR: 0.94; 95% CI: 0.90, 0.98 (Figure 6). Moreover, similar results were found for dose-response of 300 g/d or 450 g/d of total dairy products (13) (RR: 0.92; 95% CI: 0.89, 0.96 and RR: 0.91; 95% CI: 0.88, 0.95, respectively) or for each increment of 200 g/d of high- or low-fat dairy products (47) (RR: 0.96; 95% CI: 0.93, 0.99 and RR: 0.97; 95% CI: 0.95, 0.99, respectively) (Figure 7).

FIGURE 7.

Forest plot for meta-analyses evaluating the influence of dose-response of dairy product consumption on stroke. The effect size and 95% CI for fully adjusted random effects are depicted for each meta-analysis.

The association of ischemic stroke risk with total dairy consumption was considered in 4 meta-analyses (Figure 6). Elwood et al. (43) found a statistically significant inverse association (RR: 0.79; 95% CI: 0.68, 0.91). However, in the rest of the studies (13, 42, 48), no association was observed. Hemorrhagic stroke was reported in 3 meta-analyses (Figure 6); only 1 (43) found a statistically significant inverse association (RR: 0.75; 95% CI: 0.60, 0.94). With an increment of consumption of 200 g/d, no association with ischemic or hemorrhagic stroke or stroke mortality was found (47) (Figure 7).

Milk

The association between high or low milk consumption and CVD incidence (Figure 2) was examined in 2 meta-analyses (13, 44); 1 of them found a significant protective effect (RR: 0.84; 95% CI: 0.78, 0.90) (44). Milk consumption was not significantly associated with increased CHD risk in 2 meta-analyses (13, 46) or with IHD (44) (Figure 5). Five studies analyzed fatal and nonfatal stroke events with total milk consumption (13, 44, 46, 48, 52). A significant inverse association between stroke and milk consumption (RR: 0.83; 95% CI: 0.77, 0.90) was found by Elwood et al. (44), whereas no association was reported for the rest of the meta-analyses. No association between total milk intake and ischemic or hemorrhagic stroke was found (13) (Figure 6).

Two studies performed a dose–response analysis for milk consumption and CVD (Figure 3), with increments of intake of 200 g/d (52) and 244 g/d (14), but only the first reported a significant inverse effect (RR: 0.94; 95% CI: 0.89, 0.96). However, no association was found between an increase in milk intake and CHD incidence (Figure 5) in any of the 4 research groups examining this outcome (13, 14, 50, 52). Three studies (13, 47, 50) examined the association between total stroke events and the increment of milk intake (Figure 7), but only 1 found a significant inverse association (RR: 0.93; 95% CI: 0.88, 0.98) in the case of an increment of 200 g/d (47). This study also reported a significantly lower risk of stroke mortality with milk consumption (RR: 0.88; 95% CI: 0.81, 0.96), although no association with ischemic or hemorrhagic stroke was observed. Regarding milk fat content, only 1 study observed a significant increase of the total stroke risk (RR: 1.04; 95% CI: 1.02, 1.06) with an intake increment of 200 g/d of high-fat milk, whereas no association was reported for the same increment of low-fat milk consumption (47).

Cheese

In the last few years, 3 studies investigated the association between high compared with low cheese consumption and CVD risk (Figure 2). One reported an inverse association (16) (RR: 0.90; 95% CI: 0.82, 0.99); in the other 2 meta-analyses no associations were found (13, 43). High- and low-fat cheese intake was studied and no association with CVD risk was observed (16) (Figure 2). Dose-response with 10 g/d (15) or 50 g/d (16) consumption was also reported to be neutral regarding risk of CVD (Figure 3).

For CHD (Figure 4), 2 meta-analyses (13, 16) found a significantly lower risk (RR: 0.82; 95% CI: 0.72, 0.93 and RR: 0.86; 95% CI: 0.77, 0.96, respectively) and 2 others found no statistical significance (13, 52). The intake of high- or low-fat cheese was not significantly associated with CHD (16). The dose-response for cheese intake of 50 g/d (13, 16) and 75 g/d (13) was associated with significantly lower CHD risk (RR: 0.86; 95% CI: 0.75, 0.97; RR: 0.90; 95% CI: 0.84, 0.95; and RR: 0.92; 95% CI: 0.87, 0.97, respectively) (Figure 5).

Cheese intake was inversely associated with stroke in 4 meta-analyses (12, 13, 16, 46) (Figure 6) (RR: 0.91; 95% CI: 0.84, 0.98; RR: 0.87; 95% CI: 0.77, 0.97; RR: 0.90; 95% CI: 0.84, 0.97; and RR: 0.93; 95% CI: 0.88, 0.99, respectively). In another meta-analysis (48), cheese consumption was not significantly associated with stroke.

Moreover, a dose–response study (13) reported a significantly lower risk of stroke when the cheese intake was increased by 50 g/d (RR: 0.86; 95% CI: 0.77, 0.99) or 75 g/d (RR: 0.92; 95% CI: 0.87, 0.97) (Figure 7).

Yogurt and fermented products

The relation between yogurt intake and CVD (Figure 2) was examined in 2 meta-analyses (13, 53), with no significant association reported. Similarly, no effect of yogurt consumption on CHD risk was found (12, 13, 53) (Figure 4). For risk of stroke, no significant effect was reported (12, 53) (Figure 6). Regarding yogurt intake, increments of 50 g/d (14) and 100 g/d (47) were not related to fatal and nonfatal events in CHD (Figure 5) or stroke (Figure 7).

On the other hand, consumption of fermented milk significantly reduced the risk of stroke (48) (RR: 0.80; 95% CI: 0.71, 0.89) (Figure 6). Considering the dose-response, an increment of 200 g/d in the intake of fermented dairy products significantly reduced the risk of CVD (14) (RR: 0.98; 95% CI: 0.97, 0.99) (Figure 3) and it was also inversely associated with stroke mortality risk (47) (RR: 0.80; 95% CI: 0.67, 0.95) (Figure 7). Another study (14) did not find an association between CHD risk and an intake increment of 20 g/d of fermented dairy products (Figure 5). A neutral effect on stroke was also observed (47) when the consumption of fermented dairy products was increased by 200 g/d (Figure 7).

Butter and cream

Five meta-analyses analyzed the relation between butter consumption and different vascular diseases (Figures2 and 6), and no significant effects were reported for CVD (24), CHD (12, 46), and stroke (12, 47, 52, 53). Likewise, no significant effect for stroke risk was found for cream consumption, as reported in 2 studies (46, 48).

With regard to dose–response analyses (Figures3 and 7), no evidence of a significant association between an increment in butter consumption of 10 g/d and stroke was found (47). Another study also reported no association with CVD or CHD after increasing butter intake by 14 g/d (51).

Dairy product consumption and biomarkers of cardiometabolic risk

The characteristics of the meta-analyses of RCTs evaluating the effects of dairy product consumption on blood lipid biomarkers (total cholesterol and LDL cholesterol) are shown in Table 2 and Figure 8. Fermented dairy product intake was inversely associated with total cholesterol and LDL cholesterol in 4 meta-analyses (24, 25, 27, 28); however, 1 meta-analysis did not find significant differences for LDL cholesterol when comparing the consumption of whole-fat dairy with low-fat dairy products (26). The effect of dairy product consumption on blood pressure biomarkers was evaluated in 8 studies (Table 3, Figure 9) and 6 of them (30–35) found a significant reduction of SBP, whereas 5 studies (30–33, 35) also reported a significant DBP decrease.

FIGURE 8.

Forest plot for meta-analyses of randomized controlled trials evaluating the influence of consumption of dairy products on blood lipids: total-C (A) and LDL-C (B). The ES and 95% CI for fully adjusted random effects are depicted for each meta-analysis. ES, effect size; LDL-C, LDL cholesterol; total-C, total cholesterol.

FIGURE 9.

Forest plot for meta-analyses of randomized controlled trials evaluating the influence of consumption of dairy products on SBP (A) and DBP (B). The effect size and 95% CI for fully adjusted random effects are depicted for each meta-analysis. DBP, diastolic blood pressure; IPP-VPP, isoleucine, proline, proline-valine, proline, proline; SBP, systolic blood pressure.

In addition, we identified 30 RCTs published between 2013 and 2018 quantifying the association between dairy product consumption and blood pressure, as well as plasma lipids related to risk of CVD, particularly total cholesterol and LDL cholesterol, but only 12 RCTs were selected for meeting the inclusion criteria (56–67). Table 4 presents the descriptive information of the RCTs evaluating the effects of milk and dairy product consumption. Figures 10 and 11 depict the results for the meta-analyses describing the effects of dairy product consumption on the blood lipid biomarkers total cholesterol and LDL cholesterol, and on SBP and DBP. No significant changes were found for both total cholesterol (−0.06 mmol/L; 95% CI: −0.19, 0.07 mmol/L) and LDL cholesterol (−0.06 mmol/L; 95% CI: −0.16, 0.03 mmol/L) related to the consumption of dairy products (overall effect Z = −0.96, P = 0.34, and Z = −1.26, P = 0.21, respectively). Figure 10A, B shows these changes and combined estimates along with 95% CIs. Similarly, dairy product consumption did not result in significant changes for SBP and DBP (mm Hg) (SBP = −0.41; 95% CI: −1.73, 0.91; P = 0.93 and DBP = −0.77; 95% CI: −1.81, 0.27; P = 0.87) (Figure 11A, B).

FIGURE 10.

Forest plot for the updated meta-analyses of randomized controlled trials included from 2013 to 2018 evaluating the influence of consumption of dairy products on total-C (A) and LDL cholesterol (B) plasma concentrations. The effect size and 95% CI for fully adjusted random effects are depicted for each RCT. Pooled effect estimate is represented by the black diamond. (A) Total-C: overall effect Z = −0.96, P = 0.34; heterogeneity I² = 97.0% (91.76%, 98.78%) (Q = 411.55, df = 12, P < 0.001). (B) LDL cholesterol: overall effect Z = −1.26, P = 0.21; heterogeneity I² = 96.9% (91.36%, 98.40%) (Q = 572.86, df = 15, P < 0.001). RE, random effects; Total-C, total cholesterol.

FIGURE 11.

Forest plot for the updated meta-analysis of randomized controlled trials included from 2013 to 2018 evaluating the influence of consumption of dairy products on SBP (A) and DBP (B). The effect size and 95% CI for fully adjusted random effects are depicted for each RCT. Pooled effect estimate is represented by the black diamond. (A) SBP: overall effect Z = −0.61, P = 0.54; heterogeneity I² = 0.0% (0.00%, 20.28%) (Q = 4.83, df = 11, P = 0.93); (B) DBP: overall effect Z = −1.45, P = 0.15; heterogeneity I² = 0.0% (0.00%, 30.14%) (Q = 6.11, df = 11, P = 0.87). DBP, diastolic blood pressure; RE, random effects; SBP, systolic blood pressure.

The overall average between-trial heterogeneity (I²) was high for total cholesterol: 97.0% (95% CI: 91.76%, 98.78%) (Q = 411.55, df = 12, P < 0.001), and for LDL cholesterol: 96.9% (95% CI: 91.36%, 98.40%) (Q = 572.86, df = 15, P < 0.001). Nevertheless, for blood pressure biomarkers, the heterogeneity (I²) was low for SBP: 0.0% (95% CI: 0.00%, 20.28%) (Q = 4.83, df = 11, P = 0.93), as well as for DBP: 0.0% (95% CI: 0.00%, 30.14%) (Q = 6.11, df = 11, P = 0.87).

The sensitivity study of bias for the selected RCTs is depicted in funnel plots (Supplemental Figures 5–8).

Discussion

Recently, a systematic review and meta-analysis of prospective population-based studies including people with CVD and other noncommunicable chronic diseases, namely hypertension, metabolic syndrome, and type 2 diabetes, provided evidence supporting a negative relation between intake of total dairy, low-fat dairy, cheese, and fermented dairy and the risk of stroke (17), although conclusions were not uniform.

This overview of systematic reviews and meta-analyses provides a synthesis of the state of knowledge related to the association between dairy product consumption and CVD. The main findings of the present study stated that the consumption of total dairy products, with either regular- or low-fat content or with different dose-responses, revealed no association or lower risk of different outcomes such as total incidence and mortality for CVD, CHD, and ischemic and hemorrhagic stroke.

We also have reviewed the published meta-analyses of RCTs that investigated the potential effects of the consumption of dairy products on selected biomarkers of CVD risk, i.e., total cholesterol, LDL cholesterol, and blood pressure, and made a new updated meta-analysis including the RCTs published from 2013 to 2018. The main findings of the present study add further evidence to the hypothesis that dairy product consumption does not adversely affect blood lipids and blood pressure. Besides, some types of dairy products, such as fermented milks (i.e., yogurt, kefir, and cheese), clearly decrease those CVD risk biomarkers, which is in accordance with previous reports by other authors (8, 17, 18).

The information provided in this study updates the scientific evidence published up to April 2018 and includes a high number of recent meta-analyses. In comparison with other similar studies published in recent years, this work examines, critically and in detail, all the data and potential mechanisms described in other reviews that are aimed at elucidating the effects of the consumption of milk, fermented dairy products such as yogurt and cheeses, and cream and butter at different doses, and the possible detrimental effect of their fat content.

Current dietary recommendations recognize the contribution of milk and dairy products to a healthy diet because their consumption contributes to meeting the needs for many high-quality nutrients. Nevertheless, regular milk and dairy products containing milk fat are major food sources of SFAs, and this has been linked to an increased risk of CVD. However, the results obtained in this systematic review showed inverse or no association with CVD, CHD, or stroke.

The effects of milk or dairy product consumption on health depend on the interaction of all their nutrients, going beyond the mere sum of their individual effects. Thus, considering only 1 macronutrient, such as SFA, in a complex food such as milk may give rise to mistaken interpretations because not all SFAs have the same effect on plasma cholesterol [e.g., stearic acid (C18:0) is desaturated at the 9-position mainly in the liver to give oleic acid] (69, 70). Moreover, it has been reported that although dairy SFAs may have a negative effect on some cardiovascular indicators, such as an increase in total cholesterol and LDL cholesterol, they may give rise to increases in concentrations of HDL cholesterol that reverse cholesterol transport pathways, inhibit LDL-cholesterol oxidation, and thus prevent subsequent inflammatory processes (71). In addition, the dairy SFA increase in LDL cholesterol results in large, buoyant, and fluffy LDL particles, which are more resistant to oxidation and therefore less atherogenic than small, dense LDL particles (4, 5).

Moreover, individual SFAs possess specific properties associated with important biological functions, such as butyric acid, uniquely present in dairy products, which, in addition to being a compound with high biological activity, contributes significantly to the total SCFAs. The latter also contribute to the potential beneficial health effects of medium-chain TGs that have been shown to exert antibacterial activity (72), have a low tendency to be stored in adipose tissue, improve body composition without adversely affecting cardiometabolic risk factors, and not have an effect on the increase of cholesterol concentrations in blood (73–76).

In addition, SCFAs can interact with G-protein-coupled-receptors GPR41 and GPR43, leading to an increase in the intestinal secretion of glucagon-like peptide 1 and other incretins, which in turn can enhance satiety. Furthermore, SCFAs seem to activate AMP-activated protein kinase in muscles, increasing insulin sensitivity and fatty acid oxidation and decreasing lipid accumulation (77).

Milk fat also has appreciable amounts of SFAs with odd-numbered chains of carbon atoms (C15 and C17), which are used in clinical studies as markers of human consumption of regular dairy products. Plasma concentrations of those fatty acids were associated with a lower incidence of diabetes mellitus, and nonassociation or even a protective effect on CVD has been documented (78–80). Other SFAs, such as those that are methyl-branched, are also present, with interest for intestinal health and for which anti-inflammatory properties have been documented (80). Therefore, both groups of SFAs would not seem to have a negative impact on CVD.

It is also important to highlight the presence of other bioactive lipid components, such as conjugated linoleic acid and sphingolipids, for which benefits have been described in diseases related to the immune system and nervous system development and for potential cardioprotective effects (3, 81).

Despite the current evidence suggesting null or weak inverse association between consumption of dairy products and risk of CVD, some investigators, based on substitution analyses, continue claiming that replacing dairy fat with polyunsaturated fat, especially from plant-based foods, may confer health benefits (82, 83). Nonetheless, it is important to state that milk and dairy products are a combination of nutrients and bioactive substances, such as peptides, fatty acids, minerals, and vitamins, that interact with each other in the dairy matrix, and therefore, the overall effect on health after their consumption is not what is expected based on their nutritional content. Thus, for regular milk with its whole fat, or dairy products such as cheese with its fat and salt content, the majority of studies report that they do not increase the risk of CVD and may, in fact, be beneficial (10, 70).

In addition, possible mechanisms are suggested in randomized studies that would explain the neutral or inverse association of consumption of regular dairy products with CVD outcomes. RCTs have documented that cheese consumption was negatively associated with plasma TG and positively associated with HDL cholesterol (84, 85). The possible mechanism is related to the decrease in plasma TG synthesis owing to the presence of inhibitors of fatty acid desaturases in cheese (86).

On the other hand, the major milk proteins contain all the essential amino acids required to meet our nutritional needs, and they are defined as high-quality proteins and exhibit high availability and digestibility. Moreover, gastrointestinal digestion of milk proteins generates bioactive peptides that are reported to have numerous beneficial effects on health and are associated with a lower risk of hypertension (87, 88). Calcium is the milk mineral of greatest interest because it is involved in many vital functions and because of its high bioavailability. Recently, several prospective studies have reported the importance of calcium content in fermented dairy products such as cheese. The effect of calcium may affect the lipid profile by increasing the excretion of fat in feces and therefore exert a positive effect on serum lipid profile (89).

Limitations of the present study

There are some limitations in the present study that need to be acknowledged. First, some primary observational studies were included in most of the selected systematic reviews and meta-analyses; thus, the influence of those studies is overstated, although this overlapped evidence comes from the largest well-designed cohort studies. In addition, the amount of dairy product consumption for the dose–response analyses varied across studies; therefore, these results should be cautiously interpreted. Second, some of the included systematic reviews and meta-analyses are outdated, so that they did not include the latest cohort or case control studies, although our results did not show differences regardless of the date of publication of the studies. Third, because the search did not consider gray literature (e.g., research and project reports, annual or activity reports, theses, or conference proceedings), we cannot exclude the possibility that language restrictions and unpublished studies might modify our results to some extent. Fourth, most meta-analyses included population-based cohort studies with a large sample size and number of events, but these cohort studies were not specifically designed for assessing the influence of dairy product consumption on the incidence of CVD events; thus, in some, the identification of the type of dairy product and the assessment of consumption could have some inaccuracies. Fifth, although most meta-analyses reported inverse associations, it is possible that some people with high consumption of some dairy products (low-fat milk, yogurt) are prone to being engaged in other healthy lifestyle behaviors. Sixth, it has been repeatedly recognized that accurately quantifying dietary intake in noninstitutionalized populations is a major challenge in nutritional epidemiology. Seventh, the intake amount could differ greatly between studies included in the systematic reviews and meta-analyses, mainly considering the consumption differences between Western and Eastern countries, and this fact could threaten the robustness of our findings. In addition, although all the 17 meta-analyses included in our review described and analyzed the issues of heterogeneity (Table 1), the differences among heterogeneity reported in the included meta-analyses could affect the interpretation of results. Finally, all the 17 meta-analyses included in our study analyzed satisfactorily the heterogeneity, item 15 of AMSTAR 2, which addresses whether the meta-analysis’ authors have conducted a satisfactory investigation of publication bias (small study bias) and have discussed its likely impact on the results of the review, only 4 meta-analyses did not adequately perform this evaluation.

There is a strong debate over the validity of memory-based dietary assessment methods utilized in epidemiological research related to food group consumption and major events of disease (90, 91). Indeed, well-designed and -conducted RCTs are useful to determine the potential effects of the consumption of food products and disease risk. However, when intervention studies are not feasible, longitudinal cohorts with an appropriate control of confounding and based on the use of memory-based dietary assessment methods become an important tool (92). Anyhow, FFQs are prone to measurement error: 1) cognitively, the usual frequency of intake questions are difficult to answer; 2) the number of foods one can ask about is limited and extensive detail about food preparation is not collected; 3) FFQs generally query usual portion size, which may not be so problematic for discrete foods like pieces of fruit or packaged foods, but can be quite difficult and highly variable for foods like cheese, pasta, vegetables, beverages, fish, and meats (93).

Although RCTs provide consistent evidence about the ability of dietary interventions to affect biomarkers of CVD because they are free of most biases, we cannot ignore the fact that they also have several limitations, among which it is worth mentioning the short duration, which results in uncertainty concerning any expected change in the incidence of disease (e.g., CHD events, stroke, etc.); compliance with the intervention diet usually decreases during an extended clinical trial; allocation concealment is difficult when the groups are consuming particular foods; and patients with the control diet can change their habits to those being treated or in the intervention group (18).

Conclusions

From the results obtained in this review, it can be affirmed that the consumption of total dairy products either with regular- or low-fat content, or with different dose-responses showed no association with CVD or heart failure; the same results or lower risk of incidence or mortality were found for ischemic and hemorrhagic stroke. Moreover, evidence has been provided of an inverse association of the consumption of dairy products with the vascular diseases IHD and MI. There was no association of milk consumption, including in dose–response studies, or fat content with the different outcomes, although several relatively minor risks have been found on CVD or stroke. Cheese, fermented milk, or fermented dairy product intake showed an inverse association with fatal and nonfatal stroke. It has been reported that consumption of moderate quantities of butter showed no association with CVD, CHD, or stroke.

In consequence, current evidence from the results obtained in this review supports that consumption of dairy products does not adversely affect the risk of cardiovascular outcomes (CVD, CHD, and stroke) and may even have a subtle protective effect.

The results of the meta-analyses of the RCTs obtained in the present work confirm that dairy product consumption does not negatively affect the development of CVD, as evidenced by the nonadverse effects on risk biomarkers such as blood pressure (SBP and DBP) and plasma lipids (total cholesterol and LDL cholesterol).

However, it would be interesting to design new long-term RCT studies that help to clarify the underlying mechanisms occurring after intake of milk and dairy products, regardless of fat content, in relation to cardiovascular health at different doses and stages of life.

Notes

This supplement was sponsored by the Interprofessional Dairy Organization (INLAC), Spain.

The sponsor had no role in the design of the studies included in the supplement; in the collection, analyses, or interpretation of the data; in the writing of the manuscripts; or in the decision to publish the results. This study was partially funded by the University of Granada Plan Propio de Investigación 2016, Excellence actions: Unit of Excellence on Exercise and Health (UCEES), Plan Propio de Investigación 2018, Programa Contratos-Puente, the Junta de Andalucía, Consejería de Conocimiento, Investigación y Universidades, and European Regional Development Funds (ref. SOMM17/6107/UGR). Publication costs for this supplement were defrayed in part by the payment of page charges. The opinions expressed in this publication are those of the authors and are not attributable to the sponsors or the publisher, Editor, or Editorial Board of Advances in Nutrition.

Author disclosures: JF, MVC, MJ, AG, and VM-V, no conflicts of interest.

Supplemental Tables 1–5 and Supplemental Figures 1–8 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/advances.

AG and VM-V contributed equally to this work.

Abbreviations used: CHD, coronary heart disease; CVD, cardiovascular disease; DBP, diastolic blood pressure; IHD, ischemic heart disease; MI, myocardial infarction; RCT, randomized controlled trial; SBP, systolic blood pressure.