Estimated Reading Time = ~ 30 - 35 minutes
What is the impact of dairy on human health and risk of chronic disease?
Recent years have seen an exponential increase in interest in exclusionary diets for ethical, moral, and/or environmental considerations. The dairy industry has come under particular scrutiny for both environmental reasons, and ethical/moral concerns related to industrialised or intensive farming practices.
Within this important conversation, however, has been a tendency to conflate the the impact of dairy on health with these wider considerations. This results in a problematic discourse where the environmental and ethical considerations become justifications for outlandish, fear-mongering, and frankly unscientific assertions about the effect of dairy produce on human health.
The purpose of this Statement is to provide an objective evaluation of the role of dairy in health, without a forensic examination of the environmental research, and ethical/moral considerations. These considerations are acknowledged to be important, and are briefly expanded upon in our conclusions, as they may guide the ultimate consideration for whether an individual chooses to consume dairy produce or not. However, the decision to do so should also come with a science-based understanding of the health effects, rather than misinformation.
Statistical Abbreviations in this Statement
CI = confidence interval
HR = hazard ratio
OR = odds ratio
RR = relative risk (risk ratio)
Click here for our Glossary of nutritional research terminology
Defining a Broad Food Group
Within the food group ‘dairy’, is a broad range of foods which differ in terms of nutritional composition, and characteristics of the food matrix, which are important for the health effects associated with different foods. For this reason, broad definitions such as this are largely uninformative for an evaluation of food groupings falling under the umbrella term ‘dairy’. Dairy produce can be differentiated along the following lines:
- Unfermented vs. fermented
- Unrefined vs. refined
- Whole-milk (aka, “full fat”) vs. nonfat/low-fat
- Solid vs. liquid
Unfermented vs. Fermented:
Fermented dairy produce includes fermented milks (e.g., kefir, buttermilk), yogurts, and cheeses. Fermentation has historically provided a means of preservation, and the process of fermentation yields particular nutritional characteristics, including provision of lactic acid bacteria, higher protein content (in the case of certain yogurts), and formation of bioactive peptides which may exert beneficial effects on blood lipids, pressure, gut immune and microbiota function.
Unrefined vs. Refined:
Milk may also be subject to various refinement processes, which alter the nutritional characteristics of the end product. Butter is produced by separating cream from whole milk, and churning the cream until the fat separates from the remaining liquid, a process which alters the nutritional composition. For example, compared to cheese, butter is low in calcium, higher in fat, and the process of churning removes the milk fat globule membrane (MFGM), a tri-layered membrane rich in bioactive phospholipids and proteins which encloses the milk fat. These alterations, as will be discussed further below specifically in relation to cardiovascular disease, are not academic, but relate to differential effects of the food products on intermediate risk factors, in particular blood cholesterol levels.
Whole-milk (aka “full fat”) vs. Nonfat/Low-fat:
A distinction can be made between whole-milk produce vs. non-fat or low-fat produce, the definitions of which related to the milk fat content: whole-milk contains 3.5% fat on average, semi-skimmed milk 2.5%, and skimmed milk 0.1% fat. The differences in fat content are achieved by mechanically separating the fat from the liquid milk.
Thus, ‘dairy’ as an umbrella term in fact encompasses multiple differentiations, which influence nutritional composition and associated health effects. Throughout this article, reference will be made to the specific foods, i.e., ‘yogurt’, ‘cheese’, ‘butter’, ‘high-fat’ or ‘low-fat’, rather than refer to ‘dairy’ per se.
Dairy and Cardiovascular Disease
Historically, dairy as a food group has been implicated in cardiovascular disease (CVD) due to the saturated fat content of dairy foods. However, as outlined above, different dairy foods have differential physiological effects, and the current literature supports a more nuanced analysis of the impact of dairy foods on CVD risk. For example, a number of interventions have found that, compared to cheese, butter will have a greater impact on raising LDL-cholesterol levels.
This cholesterol-raising effect of butter may underscore the associations between high butter consumption, high saturated fat intake, and CVD mortality, with the Finnish North Karelia public health intervention from 1972 to 2005 providing the best illustration of this relationship. In 1965 the per capita butter intake in the Finnish population was 18kg, which was reduced to less than 3kg by 2005 thanks to a prolonged public health program aimed at targeting four major risk factors for CVD in the population:
- Blood cholesterol
- Blood pressure
The most significant dietary change was a 50% reduction in population dietary saturated fat, targeted specifically through reductions in butter consumption. This correlated with significant population-wide reduction in cholesterol levels which had the greatest impact in reducing CVD mortality by 80%.
The origins of the Finnish public health intervention came during a period of post-Second World War dietary shift in many Western countries, which had experienced a dramatic rise in animal produce consumption, yielding very high dietary saturated fat content in many populations. In the UK for example, total calorie intake was largely being met with eggs, cheese, milk, cooking fats (e.g. lard), bacon and meat; leading to 20% of total calories coming from saturated fat. As there was both a) high absolute levels of saturated fat, and b) high intakes of the foods that contribute to that saturated fat level, it meant that distinguishing between the effects of each was difficult to elucidate at the time. As a consequence, dairy foods were included under the umbrella of "saturated fat", with public health advice recommending the replacement of whole-milk with low-fat milk products.
However, subsequent epidemiological research indicated that further consideration of the role of whole-milk dairy produce was warranted in relation to CVD risk. In particular, milk, yogurt and cheese were associated with reductions in risk for CVD. The Multi Ethnic Study of Atherosclerosis investigated the effects of different foods rich in saturated fat on CVD risk, over a 10-year follow-up period, and found a significant reduction in risk of CVD from dairy SFA [HR 0.62; 95% CI 0.48-0.82], but increased risk from meat SFA [HR 1.48, 95% CI 0.98-2.23]. Replacing 2% of calories from meat sources of SFA with dairy sources was associated with a significant reduction in CVD risk [HR 0.75, 95% CI 0.63-0.91], as seen in the graph below.
The heterogeneity (the variation in study outcomes between studies) of dairy as a food group presents methodological challenges for analysis. This is particularly the case in meta-analyses of prospective cohort studies, where:
- High-fat, low-fat, fermented, non-fermented, milks, cheeses, and yogurts, may all be analysed together
- They can be analysed in relation to different endpoints
This may explain why meta-analyses of prospective cohort studies investigating total dairy (i.e. dairy intake from all foods) and CVD tend to yield neutral associations. In three recent analyses, the association with total diary and CVD was either:
- a reduction in non-fatal CVD risk [RR 0·88, 95% CI 0·81-0·96],
- no association with CVD mortality [HR 0.87, 95% CI 0.62-1.20],
- no association with total CVD incidence [HR 0.85, 95% CI 0.75-1.04].
In a Swedish cohort, while there was no association with total dairy consumption, comparing the highest intake of low-fat dairy consumption to the lowest resulted in a 12% reduction in stroke risk and 13% reduction in cerebral infarction risk. Overall milk intake (i.e., no distinction between low-fat or full-fat) has been associated with a modest 6% reduction in risk for overall CVD [RR 0.95, 95% CI 0.89-0.99] per 200ml/d. In relation to low-fat dairy, however, it should be noted that these foods tend to cluster within overall healthy dietary patterns at the population level.
However, overall the data on milk consumption and CVD risk is inconsistent, demonstrating weak positive or neutral effect sizes, and it is difficult at this juncture to make an affirmative conclusion; a more appropriate conclusion in relation to milk intake is that it does not increase risk, and is overall neutral in effect on CVD.
Conversely, more consistent associations are noted for fermented dairy products, in particular cheese and yogurts. In the Malmo Diet and Cancer study, while total dairy intake was associated with a 12% reduction in CVD risk, this effect was attributable primarily to fermented milk in analysis of specific foods; no association was noted for full-fat or low-fat milk.
In relation to yogurt consumption, an underlying potential mechanism may be reduced common carotid artery-intima media thickness [CCA-IMT]. A CCA-IMT test can help diagnose the extent of atherosclerotic vascular disease by measuring the thickness of the inner two layers (i.e. intima and media) of the two main arteries which carry blood to the head and neck. Yogurt consumption has been associated with lower CCA-IMT in a prospective study of elderly women. However, the overall data on yogurt and CVD risk is neutral.
Inverse relationships between CVD risk and cheese intake, in particular reduced risk of stroke, have consistently been shown. The reduction in risk for myocardial infarction observed with higher cheese intake in a Swedish cohort was attenuated after adjusting for calcium. This suggests that calcium in the cheese matrix may be a mediating factor. There is some mechanistic plausibility to this finding. Calcium may lead to the formation of “soaps'' in the intestine, reducing fat absorption and increasing exertion of fat. This may reduce the impact of dairy foods that are rich in calcium, such as cheese (but not butter, as the calcium is depleted during churning).
Fat globules are trapped within a casein matrix, in a protective encapsulation known as the ‘milk-fat globule membrane’ (MFGM). And research has shown dairy fat within the MFGM does not impair blood lipid profiles. There are also other important potential mechanisms, in particular the effects of fermentation of cheese on the gut microbiota, and the fatty acid composition of dairy fat. A number of intervention studies have consistently found that comparing butter vs. cheese, butter will have deleterious effects on blood lipids compared to cheese, and cheese may reduce both total and LDL-cholesterol. Thus, consumption of full-fat cheese suggests that the observed effects may go beyond the fat content, to an overall synergistic effect of the mix of dairy fat, casein, and calcium, within the whole-food matrix. Nonetheless, it should be stated that low-fat dairy produce is strongly associated with reduced risk of stroke.
Thus, the nutritional characteristics of milk, yogurts, and cheeses, are important for considering the relationships with CVD, as in addition to their fat composition (and presence of the MFGM), these foods provide rich sources of calcium, phosphorous, magnesium, bioactive peptides, and are very low in sodium. These characteristics may contribute to a blood pressure-reducing effect of dairy consumption. Indeed, diet pattern interventions such as the Dietary Approaches to Stop Hypertension (DASH) diet, which significantly reduces blood pressure and CVD risk factors, emphasises low-fat dairy produce. This corroborates observational evidence of reduced risk of hypertension and stroke with high consumption of low-fat milk and yogurt.
Analysis of biomarkers of dairy consumption also corroborates the potential benefit, and highlights the distinction between dairy saturated fatty acids and other saturated fatty acids.
Side Bar: Different Types of Saturated Fatty Acids
Saturated fat is an umbrella term that accounts for a collection of many individual types of saturated fatty acids, with each having a different number of carbon atoms in their chain. Based on this, some will be referred to as “short-chain”, “medium-chain”, or “long-chain” saturated fatty acids. Each individual SFA will be assigned a name denoting the number of carbon atoms in the chain, for example palmitic acid is denoted as C16:0 (here the “0” references the fact that there are no double bonds in the chain, something common to all the saturated fatty acids).
In analysis of plasma phospholipid fatty acid levels as biomarkers of fat intake, the EPIC-Norfolk study found that phospholipid levels of odd-chain saturated fatty acids C15:0 pentadecanoic acid and C17:0 heptadecanoic acid, both biomarkers of dairy intake, were inversely associated with coronary heart disease risk, while the long and even-chain saturated fatty acids C16:0 palmitic and C18:0 stearic acid were associated with a significantly increased risk.
Similarly in a Swedish cohort, analysis of plasma levels of C15:0 and C17:0 and specific foods indicated that fermented milk and cheese were associated with reduced risk of myocardial infarction. However, this study utilised plasma biomarkers, which only reflects dietary fats intake over the previous number of days. Conversely, phospholipid markers as utilised in EPIC-Norfolk reflect dietary intake over the previous months, providing a more reliable relationship with habitual dietary intake. While the use of biomarkers is becoming more common in nutritional epidemiology, in relation to quantifying dietary fat intake, it should be noted that there are differences between the fraction measured and the time-course of dietary intake that biomarkers may reflect.
In sum, the overall evidence suggests either a neutral effect of total dairy on CVD risk, or modest reductions in risk. Individual foods reveal more consistency in effect, and cheese in particular is associated with reduced risk of stroke. Milk has been neutral or modest in effect on reductions in CVD risk. Low-fat dairy is consistently associated with reduced hypertension risk.
Dairy and Type-2 Diabetes
Dairy intake has attracted attention for a potential protective effect against the development of type-2 diabetes. This may relate to the composite of:
- High protein content
- Calcium and other micronutrients (e.g. vitamin D)
- Unique factors including specific dairy fatty acids
In a meta-analysis of seven prospective cohort studies, when Tong et al. compared the highest to lowest levels of dairy intake, they found a 14% reduction in risk for type-2 diabetes [RR 0.86, 95% CI 0.79-0.92]. The magnitude of effect was greatest for low-fat dairy foods [RR 0.82, 95% CI 0.74-0.90], while high-fat dairy foods and whole milk were not associated with type-2 diabetes.
The magnitude of effect observed in the Tong study is similar to that of Elwood et al., which found a significant 15% reduction in type-2 diabetes risk [RR 0.85, 95% CI 0.75-0.96] in those with the highest total dairy intake compared with the lowest. However, neutral associations have been found for total dairy and type-2 diabetes. This highlights the methodological difficulties inherent in pooling a broad and heterogeneous food group into a meta-analysis.
On the basis of current evidence, the potential protective effect of dairy on type-2 diabetes is difficult to attribute to the dairy fat content. Multiple analyses have found no association between intake of high-fat dairy and type-2 diabetes. But analysis of biomarkers for the odd-chain C15:0 and C17:0 dairy fatty acids have suggested a reduction in risk. The EPIC-InterAct study is a prospective case-cohort study investigating the relationship between lifestyle and type-2 diabetes. The data from that study showed that even-chain saturated fatty acids were associated with an increased risk for incident type-2 diabetes, while C15:0 and C17:0 were each associated with a 21% and 33% reduction in risk, respectively.
Currently the data in relation to high-fat dairy and type-2 diabetes is ambiguous. Conversely, consistent associations have been found for low-fat dairy and incident type-2 diabetes. In addition to the previously mentioned analysis by Tong et al., a dose-response analysis by Aune et al. yielded a 27% reduction in type-2 diabetes risk [RR 0.83, 95% CI 0.76-0.90], while Gao et al. found a 19% reduction in type-2 diabetes risk [RR 0.81, 95% CI 0.74-0.89]. Thus, consistent evidence exists for a protective effect of low-fat dairy products against incident type-2 diabetes.
Arguably the most consistent relationships for specific foods and type-2 diabetes is found in relation to yogurt. With regard to yogurt, Tong et al. found a significant 17% reduction in type-2 diabetes [RR 0.83, 95% CI 0.74-0.93] comparing the highest to lower levels of intake, a similar magnitude of effect to the 15% risk reduction observed by Gao et al. [RR 0.85, 95% CI 0.75-0.97]. In analysis of data from 12 studies, Gijsbers et al. found that comparing 80g/d yogurt to 0g/d resulted in a 14% reduction in type-2 diabetes risk [RR 0.86, 95% CI 0.83-0.90]. It should be noted that these analyses do not indicate how the form of yogurt (e.g. whole-milk, low-fat, non-fat) may influence outcomes.
Cheese intake has also been associated with lower risk of type-2 diabetes, however, a limitation of the available meta-analyses is that despite being based on the same pool of included primary prospective cohort studies, inconsistent results have been obtained. The analysis by Gao et al. yielded a relative risk reduction of 18% from cheese intake [RR 0.82; 95% CI 0.77–0.87]. However Aune et al. included the same primary studies as Gao but found a 9% reduction in type-2 diabetes risk [RR 0.91, 95% CI 0.84-9.98]. While the direction of effect is consistent in the meta-analyses, the divergent magnitude of effect may reflect significant differences in the primary included studies. Thus, it appears cheese may reduce type-2 diabetes risk, but the confidence in that relationship is not as strong as that observed between yogurt intake and incident type-2 diabetes.
Potential Mechanisms of Risk Reduction
A number of potential mechanisms have been proposed related to dairy consumption and reduced type-2 diabetes risk:
- Weight regulation
- Lower triglycerides
- Decreased postprandial glycaemia
- Insulin sensitivity
- Beneficial impact on the microbiome
High dietary calcium intake may inhibit fat absorption through formation of insoluble soaps during digestion, potentially aiding in weight regulation. This mechanism may also reduce the magnitude of the rise in triglyceride-rich lipoproteins in circulation after meals. This attenuation of postprandial lipaemia may result in favourable blood lipid profiles, in particular lower triglycerides, which may reduce insulin resistance.
Dairy protein appears to result in decrease postprandial glycaemia (decreased rise in blood glucose after meals) in participants with type-2 diabetes, which may relate to the increased insulin response observed following dairy protein intake.
In addition to the effects of dairy proteins, a role for calcium in insulin action and glucose disposal has been found in a number of studies, suggesting calcium may enhance insulin sensitivity.
Dairy fatty acids may also exert anti-diabetic effects, with trans-palmitoleic acid associated with reduced risk for incident type-2 diabetes in the Multi-Ethnic Study of Atherosclerosis, and mechanistic research indicating that trans-palmitoleic acid may improve hepatic and peripheral insulin sensitivity.
Finally, the influence of dairy foods on the gut microbiome may be a factor, with short-chain fatty acids hypothesised to have extra-intestinal effects on energy metabolism, blood lipids and glucose regulation.
Thus, whether the benefit of dairy in type-2 diabetes is indirect via reduced adiposity or protection against obesity, or related to specific effects of dairy foods, remains to be fully determined. However, in sum there is strong evidence for a protective effect of yogurt and low-fat dairy foods on risk for type-2 diabetes.h
Dairy and Cancer
In any discussion of the role of diet and cancer there is an unfortunate lack of context for the inherent complexity in studying the relationship. From a dietary perspective, of more than 25,000 different bioactive components in foods, at least 500 have been identified as potential modifiers, positively or negatively, of carcinogenic processes. Add to this the fact that ‘cancer’ as a singular term does not really exist; while there are unifying physiological characteristics, different tumours have very different biological behaviour, depending on site, and the influence of genetic, epigenetic, and environmental factors. This renders the study of a potential causal role of isolated food and nutrient variables inherently challenging.
Certain foods and food constituents consistently emerge from both long-term cohort studies and mechanistic research as influencing cancer risk. This makes such associations biologically plausible, with the links between processed meat and colorectal cancer an example of this. Dairy is a food group for which there is significant hyperbole regarding cancer, often not reflective of the evidence. This is complex and goes both ways: dairy has been associated with increased risk for certain cancers, for example prostate, and associated with reduced risk of others, in particular gastric and breast cancers. For the purposes of this Statement, we’ll examine the most consistent associations in both directions:
- Prostate cancer
- Breast cancer
- Gastrointestinal/colorectal cancer
Consistent associations have been observed for total dairy, and specifically high calcium intake, and increased risk for prostate cancer. There are a number of potential nuances to the associations, including cohort location, stage of disease, and the food constituent implicated. There is an implication of a ‘Western diet’ pattern in prostate cancer risk, and cohort studies conducted in North America have more consistently indicated an increased risk from total dairy, milk, cheese, and calcium intake, and prostate cancer; with Qin et al. finding a 13% increased risk [RR 1.13, 95% CI 1.02-1.26].
Dairy calcium intake appears to mediate risk, with the generally accepted mechanism being suppression of 1,25-dihydroxyvitamin-D3 activation, the biologically active form of vitamin D. In a pooled analysis of 14 cohort studies (including 806,969 participants and 51,896 incident cases of diagnosed prostate cancer, 5,061 of which were advanced stage prostate cancer), comparing the highest to lowest quintile of calcium intake resulted in an 8% increase in risk for overall prostate cancer [RR 1.08, 95% CI 1.04–1.12]. The associations for dairy and calcium and advanced prostate cancer were non-significant.
Dairy protein and calcium was associated with prostate cancer risk in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort:
- Dairy protein was associated with a 22% increase in risk [HR 1.22, 95% CI 1.07–1.41]
- Total calcium was associated with a 17% increase in risk [HR 1.17, 95% CI 1.00–1.35]
- Dairy calcium associated with an 18% increase in risk [HR 1.18, 95% CI 1.03–1.36]
One putative mechanism for which there is support from a number of lines of evidence is the effect of milk on increasing insulin-like growth-factor-1 (IGF-1) levels. This effect appears to be stronger in non-Caucasian populations. A comprehensive analysis suggests milk (rather than dairy protein or other constituents) is the exposure of interest. The relationship between IGF-1 and prostate cancer has been shown to be positive, with elevated IGF-1 associated with a 7% increase in risk [OR 1.07, 95% CI 1.02-1.12]. Interestingly, the effect of milk on IGF-1 was mediated by lifestage, with milk intake in childhood inversely associated with IGF-1 levels in adulthood. Nonetheless, there is support in the literature for the putative mechanism linking milk to prostate cancer, via IGF-1 levels, although this relationship is complicated by duration of follow-up, lifestage exposure, ethnicity.
In the US, the Prostate, Lung, Colorectal and Ovarian [PLCO] Cancer Screening trial found no association between total dairy, low-fat vs. high-fat dairy, and fermented vs. unfermented dairy, and advance or non-advanced prostate cancer incidence. In the UK, the UK Dietary Cohort Consortium found no association between milk, cheese, yogurt, or dairy protein, and prostate cancer incidence.
Aune et al. conducted the most recent, comprehensive meta-analysis, finding an overall 9% relative risk increase [RR 1.09, 95% CI 1.02-1.17] from total dairy and prostate cancer. Effects were driven by:
- Location - North American cohorts, not European or Asian
- Follow-up - Less than 10 years duration of follow-up
- Stage of Cancer - Non-advanced prostate cancer
This overall direction of effect is representative of the wider literature on dairy and prostate cancer risk, which indicates an association with dairy overall (together with biological mechanisms), but a number of important nuances related to food types, population, lifestage exposure, and specific dietary consistent (calcium in particular) to further elucidate. The characterisation of the dairy-prostate cancer association by the World Cancer Research Fund as “limited-suggestive” is therefore an appropriate conclusion.
However, while the IGF-1/prostate cancer link suggests a hormonal effect of dairy on hormone-dependent cancer, this does not hold true invariantly, as evident in associations observed between dairy intake (in particular low-fat dairy products) and breast cancer risk. It is high dietary saturated fat content that has been related to breast cancer, but low-fat dairy products have reduced, or no, saturated fat content. In the Norwegian Women and Cancer [NOWAC] study, in a population with high (age-adjusted) incidence of breast cancer and dairy intake, there was a significant reduction in premenopausal breast cancer risk from white cheese intake (median 25g/d). Such effects may have been mediated by calcium.
In the US Cancer Prevention Study II Nutrition Cohort, high dietary calcium intake (>1,250mg/d) was associated with a 20% risk reduction [RR 0.80, 95% CI 0.67-0.95] for postmenopausal breast cancer compared with the lowest (<500mg/d) level of intake. Two to three servings of dairy per day, compared to less than 0.5 servings, was associated with a 19% reduction in breast cancer risk [RR 0.81, 95% CI 0.69-0.95], an effect driven by low-fat dairy consumption when low-fat and high-fat sources were analysed separately.
These associations in the CPS-II were stronger in women with oestrogen-receptor positive (ER+) tumours, with a 30% risk reduction in ER+ participants [RR 0.6, 95% CI 0.51-0.88]. Vitamin D was associated with a 26% risk reduction [RR 0.74, 95% CI 0.59-0.93] in ER+, but not ER-, participants. These dietary variables, calcium and vitamin D, may provide biological plausibility to reductions in breast cancer risk observed with dairy intake, particularly in the context of vitamin D fortification of milk and milk products, in addition to the foods as a source of calcium.
In the Women's Health Study, in relation to premenopausal breast cancer:
- Higher calcium intakes were associated with a 39% relative risk reduction [HR 0.61, 95% CI 0.40-0.92]
- Higher vitamin D intakes were associated with a 35% relative risk reduction [HR 0.65, 95% CI 0.42-1.00]
This association with premenopausal, but not postmenopausal, women was also found in the Nurses’ Health Study, with a 32% relative risk reduction [RR 0.68, 95% CI 0.55-0.86] for the highest category of intake. This effect was strongest for low-fat dairy, which is consistent with the wider literature.
Inverse relationships were also evident for calcium and vitamin D. Overall, however, similar complexities are observed in the literature with regard to menopausal status, source of dairy, and stage of cancer. The overall direction of effect is toward a reduction in risk, perhaps mediated by calcium and vitamin D intake. However, this association remains to be more concretely established. What is evident from the research regarding breast cancer is that attempting to classify potential risk associated with dairy consumption according to hormone-dependence, such as breast and prostate, does not yield the same direction of effect. And the putative mechanism associated with risk in one cancer (i.e., IGF-1) may not necessarily be operable in all circumstances.
The Adventist Health Study-2 (AHS-2), a large US prospective cohort study, evaluated the independent effects of dairy and calcium on colorectal cancer risk. Total energy from dairy was associated with a 23% reduction in risk for colorectal cancer [HR 0.77, 95% CI 0.59-0.99]. And of individual food sources, milk was associated with a 37% reduction in risk [HR 0.63, 95% CI 0.43-0.89]. Total calcium (i.e., from overall diet + supplements) was associated with a 45% reduction in colorectal risk [HR 0.55, 95% CI 0.28-0.98].
One suggested mechanism for the effect of milk specifically relates to lactoferrin, a protein which (outside of human milk) has highest levels in cow's milk. Mechanistic research indicates a beneficial effect of lactoferrin on microbial composition in the gut, and the stimulation of immune and anti-inflammatory processes in the colon. In addition, conjugated linolenic acid (CLA) is a short-chain fatty acid, for which dairy is the primary dietary source, which has mechanistic research indicating an anti-carcinogenic effect. Aune et al. conducted a meta-analysis with ten studies included, and the direction of effect for nine of those studies was toward a reduction in risk. There was a 17% reduction in risk for colorectal cancer [RR 0.83, 95% CI 0.78-0.88] with a dose-response of 400g/d total dairy. The 2017 WCRF report on colorectal cancer concluded that strong evidence indicates that dairy, which umbrella term included total dairy, milk, cheese and dietary calcium intakes, reduces risk of colorectal cancer.
Dairy and Musculoskeletal Health
Childhood stunting remains a substantial global issue, with 20% of the global infant population less than five years old (144 million children) documented as stunted, with the highest prevalence in the developing world and middle-income countries. While survey studies have indicated that absolute protein intake is adequate, these findings have been questioned as they have not considered the issue of protein quality and digestibility.
A feature of stunting risk in these populations is lack of dietary diversity, and emphasis on plant proteins, with low essential amino acid digestibility and uptake in the ileum (the last part of the small intestine). Low ileal protein digestibility is significantly associated with prevalence of stunting or underweight. Recent metabolic tracer studies have indicated high ileal digestibility for animal-source proteins in the form of milk and eggs. These foods are culturally acceptable in many at-risk populations (many of whom are traditionally lacto-ovo-vegetarian). They are also the most effective proteins for improving Digestible Indispensable Amino Acid Scores (DIAAS) in children at-risk of stunting and undernutrition. Achieving a DIAAS of 100 may be achieved by the addition of 200g milk in the diets of children aged between one and three years old, who are at risk of stunting or underweight.
These findings may have emerging relevance in the developed world. Recent US data indicates 54% of children consume plant milks. Many of these plant milks are fortified with additional calcium, but no current research exists on the bioavailability of calcium from plant milks, with the exception of soy milk. Calcium absorption from soy milk, if fortified with calcium carbonate, has been shown to be relatively equivalent to cows milk. However, tricalcium phosphate, which is commonly used in other plant milks, exhibited 20% lower calcium bioavailability.
Another relevant factor from a developmental perspective is that plant milks are 5-10 times lower in protein than cow's milk. And as stated above with regard to stunting and growth, high DIAAS protein sources appear to be crucial in this life stage. The precautionary principle, which applies in situations of uncertainty, should prevail in this area, given the substantial gaps in knowledge that exist regarding the efficacy of fortified plant milks on musculoskeletal health in development life stages. This is particularly relevant given the emergence of published reports of children with severe nutritional deficiencies as a result of consuming plant milks (e.g. here and here).
While the positive effect of dairy in relation to childhood stunting is related to milk proteins, dairy contains numerous dietary constituents considered important for bone health, in particular calcium, phosphorous, zinc, manganese, and in countries which fortify milk with vitamin D (e.g., Ireland, the US, Canada), milk may be a food source of vitamin D. It is important in any analysis of dairy foods on bone health to distinguish food trials from calcium supplementation trials, which are often cited as a proxy for suggesting an effect (or generally lack thereof) of dairy. We will thus confine our analysis to food-based research, which may be more informative given that the provision of additional calcium via dairy foods may have a greater effect on bone mineral density than calcium provided in through supplements.
The effects of dairy products vary according to lifestage. The most consistent, positive associations for dairy food consumption and bone mineral content (BMC) have been observed in children and adolescents, particularly where baseline calcium intake is low. A meta-analysis of RCTs in children indicated that the mean overall increase in BMC from increased dairy and dietary calcium intake was 2g. However, if included studies were stratified based on baseline calcium status, in participants with low baseline dietary calcium, total body BMC increased by 49g. This is supported by the wider paediatric literature, in which intervention studies of the effects of dairy consumption show consistent increases in BMC in children.
In adults, the overall data is less clear. Long-term prospective cohort studies have yielded mixed results. The most consistent results have been evident in women, with higher risk of fractures with low dairy consumption, particularly in participants with lactose intolerance. Dairy intake, in particular milk and cheese, were found to result in an overall 8% reduction in hip fracture risk [RR 0.92, 95% CI 0.87–0.97] in combined analysis of the NHS and HPFS.
While many analyses have focused on milk consumption alone in relation to hip fracture risk, a meta-analysis by Bian et al. of different dairy food sources indicated 25% reduction in hip fracture risk [95% CI 0.68-0.96] from yogurt, and 32% reduction in risk [95% CI 0.61-0.77] from cheese intake. These findings were replicated in a more recent meta-analysis using three (out of four) of the same primary included studies as Bian et al., where yogurt and cheese, but not milk, were associated with significant reductions in risk for fractures at any site.
Relatively consistent with the wider literature, overall results do not suggest a significant protective effect of milk consumption during adulthood. With regard to milk specifically, if findings are positive, milk only offers a modest potential reduction in hip fracture risk.
Dietary protein is an important mediating factor in the potential effects of calcium. Upregulation of intestinal absorption and greater effects on biomarkers of bone turnover have been observed with higher dairy protein interventions, which protected bone health in the context of a weight loss intervention. Whether such a 16-week intervention would translate to longer term bone health in adults is, however, uncertain. Cumulatively, the data does not indicate a protective effect for milk, but potentially protective effect of yogurt and cheese intake, in adults.
In the elderly, considerations may change again, with particular relevance for age-related sarcopenia (loss of skeletal muscle mass and function). Both decreased dietary protein and calcium are factors associated with sarcopenia, and dairy proteins may have particular utility in the promoting of skeletal muscle protein synthesis (MPS), in the prevention of sarcopenia in the elderly. Higher dairy protein intake has been shown cross-sectionally to be associated with greater bone strength in men over the age of 65, and in post-menopausal women. Interventions in elderly women in care homes have shown that dairy products inhibit bone resorption (i.e. inhibit the breakdown of bone tissue).
Both a higher dietary protein intake, and combination of resistance exercise, may offset sarcopenia. The advantage of dairy proteins in relation to sarcopenia may be their rich composition of essential amino acids, and in particular the branched-chain amino acid leucine, which results in activation of MPS at lower doses.
Finally, it is relevant to discuss outcomes used in studies. Bone mineral density (BMD) is commonly used in intervention studies. While fractures may provide a more ‘hard endpoint’, they are primarily the outcome in prospective cohort studies, rather than interventions. Despite the common use of BMD as an outcome, bone remodelling is characterised by the ‘bone-remodelling transient’, an alteration in the balance between bone resorption and bone formation processes. During a specific intervention targeting BMD, this requires a temporal period of several weeks in children, three months in adolescents, and 6-18 months in adults. Thus, any intervention within, or shorter than, these respective times will reflect the bone-remodelling transient phase, and BMD as an endpoint in a trial investigating foods, or nutrients like calcium, may reflect a temporary change in bone mass. This may be of more practical importance for interventions in adults, as the bone-remodelling transient period in children and adolescents is shorter. .
Cumulatively, the data indicates a life stage-dependent effect of dairy intake on musculoskeletal integrity. Dairy proteins are particularly important in childhood and adolescents, for both BMC and also in prevention of stunting. In adults, the data is less clear. Short-term interventions may reflect the bone-remodelling transient, while the longer-term observational data suggests no overall effect of milk, but potential effect of yogurt and cheese. In the elderly, dairy intake may emerge again as an important dietary factor for the prevention of age-related sarcopenia and preservation of muscle mass. However, interventions are needed to confirm cross-sectional findings and mechanistic plausibility.
Dairy and Food Allergy & Intolerance
It is important to make clear distinctions in relation to food allergy and food intolerances. An adverse reaction to food may be defined as toxic or non-toxic; non-toxic reactions are differentiated into reactions mediated by the immune system (food allergies), and reactions not involving the immune system (food intolerances).
Food allergies can be further divided into reactions mediated by the Immunoglobulin-E antibody (IgE), a type of antibody released by your immune system in response to an allergen, and reactions mediated by other immune responses. Allergy and intolerance can be distinguished as follows:
The most common IgE food allergies include eggs, milk, peanuts, tree nuts, wheat, fish, and shellfish. IgE-mediated food allergies typically develop in early childhood, and thus most adults with a food allergy have been diagnosed. In children, allergic reactions to milk (also eggs and wheat) are often “outgrown”, meaning tolerance is acquired by adulthood, and tolerance is acquired in >85% of children within the first three years of life. There is evidence that cows milk formulas may be associated with constipation in some infants. Of the dairy food matrix, it is the casein and whey proteins which are considered triggers to the allergic response, however, incidence of such cows milk allergy in adults is very rare.
The most relevant example of food intolerance for present purposes relates to lactose intolerance, which is characterized by a deficiency in the lactase enzyme required to break down milk sugars. And while overt clinical diagnosis of adult cows milk allergy is rare, the ten year prevalence of self-reported dairy intolerance has increased, without any concomitant increase in medical diagnoses of lactose intolerance. In a survey study of raw milk consumers, 60% stated they consumed raw milk due to a self-reported intolerance to pasteurised milk. However, only 3% of respondents in the study had been medically diagnosed with lactose intolerance. In an eight day intervention study investigating tolerance to raw vs. pasteurised whole milk (with soy milk as a non-dairy control) in participants with diagnosed lactose intolerance, raw milk produced significantly higher evidence of impaired lactose digestion (measured via hydrogen breath test) on the first day, but by the end of the intervention both raw and pasteurised whole milk elicited the same response. Thus, to date the hypothesis of raw milk tolerance in people with lactose intolerance, or self-reported intolerance to pasteurised milk, is unsupported.
However, independent of lactose intolerance via deficiency of the lactase enzyme, lactose is also a FODMAP: Fermentable Oligosaccharides, Disaccharides, Monosaccharides, and Polyols. The main disaccharide is the milk sugar lactose, and there are ethnic variations in expression of the lactase enzyme, with higher prevalence of reduced lactase enzyme activity in certain populations, for example East Asians. Thus, symptoms associated with lactose ingestion may indicate Irritable Bowel Syndrome (IBS). A final consideration is that fermentation of dairy may render it more digestible, as there is evidence that lactose intolerant persons may tolerate yogurts and certain low-lactose dairy products.
One hypothesis for which there is some, albeit weak, evidence, relates to the potential for dairy intolerance to be mediated by proteins, not lactose (i.e., that a negative lactose intolerance test would be uninformative for identifying perceived intolerance). The hypothesis relates to variants in the beta-casein protein, which is the major protein component of cows milk, with two major variants:
- A1-beta-casein - found in Holstein cows and commonly used in industrial milk production
- A2-beta-casein - found in African, Asian, Jersey or Guernsey cows.
Two studies conducted in Chinese participants have found increased gastrointestinal symptoms in subjects consuming mixed A1:A2, milk compared to A2 milk alone. However, these symptoms were mostly significant in subjects with lactose intolerance, and thus make it difficult to differentiate from lactose intolerance. The studies’ suggested a greater effect of intolerance from the mixed A1 milk, which may reflect genetic differences in lactase expression. Another trial comparing exclusively A1 vs. A2 milk consumption found no significant difference in subjects, except for in participants with self-reported lactose intolerance. Nonetheless, while the controlled trials in humans are lacking in both quality and quantity, to date there is a body of mechanistic research which indicates differences in the metabolism of A1 vs. A2 milk, and the concentrations of metabolites are observed at levels which may potentially be biologically relevant. It is thus premature, on the basis of the limited human interventions, to make any conclusions on the practical significance of the difference in protein types for human health.
An objective examination of the health effects of dairy indicates that specific foods within the food group, in particular yogurt, cheese, and milk, are broadly associated with beneficial outcomes, which effects are evident across a number of disease endpoints, and at various life stages.
Of course, ethical and environmental considerations are increasingly important. In relation to the environmental aspect, the recent EAT-Lancet Commission recommendations for a ‘planetary health diet’ included a daily target of 250g/d dairy foods (possible range of 0-500g).
In relation to the ethical argument, this has largely been reduced to a discussion of ethics as it relates only to sustainability and animal welfare. However, ethical considerations for nutrition also extend to societal equity, food security, consideration of vulnerable population groups, and cultural factors. In this regard, it is important to consider what is ethical for these considerations.
There are perhaps two levels to think about a conclusion: individual and societal. At the individual level, the decision to include dairy in the diet is entirely within the nutritional, ethical, and environmental values of the individual. At the societal level, these considerations vary. In the developing world, the nutritional and ethical considerations of preventing prevalent issues like stunting, wasting, and underweight, indicate that dairy proteins may be important to achieve this end.
We also submit that, although adults may increasing be consciously adopting vegan diets, there are ethical considerations for the use of diets in infants and children, given the substantial gaps in our knowledge regarding the provision of specific nutrients of concern in that lifestage, and the adequacy of plant substitutes for foods like milk or yogurt. Given published reports of severe nutritional deficiencies in this regard, the precautionary principle is warranted until evidence can better inform achieving nutritional adequacy through diets exclusive of all animal-sourced protein, in particular dairy.
These concerns are less relevant for adults consciously preparing a well-planned vegan diet. For individuals wishing to continue including dairy products in their diet for nutritional (or ethical) purposes, we recommend following the recommended intake from the EAT-Lancet Commission.
Summary of Key Points
- ‘Dairy’ is an umbrella term that encompasses multiple food types that can be separated via various categories of differentiation. A distinction should be made between food types, fat content, and the overall food matix when assessing associated health effects.
- Overall evidence suggests either a neutral effect of total dairy on CVD risk, or modest reductions in risk. Cheese specifically is associated with reduced risk of stroke and low-fat dairy is consistently associated with reduced hypertension risk.
- Strong evidence exists for a protective effect of yogurt and low-fat dairy foods on risk for type-2 diabetes.
- There are associations between dairy intake and increased prostate cancer risk, but given the current gaps in the literature, the association is best thought of as “limited-suggestive” currently, with future research required.
- Dairy intake is associated with a trend towards breast cancer risk reduction, perhaps mediated by calcium and vitamin D intake. However, this association remains to be more concretely established.
- There is strong evidence that dairy reduces risk of colorectal cancer.
- Consumption of dairy protein appears to be protective against stunting and nutritional deficiencies, as well as improving bone mineral content in children.
- In adults, the data does not indicate a protective effect for milk, but potentially protective effect of yogurt and cheese intake in relation to musculoskeletal health.
- In older adults, dairy intake may be an important dietary factor for the prevention of age-related sarcopenia, preservation of muscle mass and offsetting losses of bone tissue.
- Cow’s milk allergy in adults is very rare but lactose intolerance can affect a significant proportion of people, with prevalence rates differing based on ethnicity.
- Lactose is a FODMAP, and so consideration should be given in cases of irritable bowel syndrome, where a dietitian may be utilizing a low FODMAP diet.
- Although not the target of this specific statement, it is acknowledged that ethical and environmental considerations are increasingly important. The ethics of dairy consumption not only relates to sustainability and animal welfare, but also to societal equity, food security, consideration of vulnerable population groups, and cultural factors.
Statement Author: Alan Flanagan, PhD (c)
Alan is the Research Communication Officer here at Sigma Nutrition. Alan is currently pursuing his PhD in nutrition at the University of Surrey, UK, with a research focus in chrononutrition. Alan previuosly completed a Masters in Nutritional Medicine at the same institution.
Originally a lawyer by background in Dublin, Ireland, Alan combines an investigative and logical approach to nutrition together with advocacy skills to communicate the often complicated world of nutrition science, and is dedicated to guiding healthcare professionals and the lay public in science-based nutrition.
Thank you… this is a wonderful resource to have available. Very comprehensive.
Two of my three siblings have a strong lactose intolerance and absolutely adore my yogurt and cheeses. I haven’t encountered any issues with lactose/dairy as of yet and my fingers remain crossed. I have no idea how much of that is genetic.
It might also be worth mentioning that none of my siblings do anything in regards to health; all are very sedentary and simply follow the standard western diet.
Is there any evidence on diary (negative) influence on ASD (autism)? There is a lot of speculation on the topic…
Hi Serguei, I have heard some people mention that hypothesis too but am unaware of any convincing data on this.
Thank you for what seems to be a careful & unbiased review!
We appreciate that Phil.
Thanks guys for the detailed analysis. I have questions about 3 topics:
1. CVD risk:
I’ve read some of your other writings on saturated fat and CVD. I understood:
– If you replace foods that include SFA with PUFA or whole grains, you get a reduction in CVD risk.
– If you replace foods that include SFA with refined carbohydrates, you get an increased risk, or no change.
So the studies you present here are concluding that replacing SFA from meat with SFA from dairy, we get a reduction in CVD risk.
What is the expected outcome when you replace dairy with same amount of calories from foods that contain more PUFA? Would you expect a reduction in CVD risk then? Is that an irrelevant comparison because SFA content is not the same?
2. Breast cancer:
Looking at “Dairy, soy, and risk of breast cancer: those confounded milks”(Fraser, 2020): do you understand the positive correlation with dairy in that study? (I understand that breast cancer risk is rather complex to study, due to reduction by Calcium and Vitamin D)
3. Recommendations on children:
– Do we not have some studies that compare children in developed world who eat balanced vegan diets without dairy? I say developed world, because I understand that it’s harder to reach nutrition goals without animal products for children who don’t have access to diversity of foods.
I know just one called VeChi Diet Study from Germany: https://pubmed.ncbi.nlm.nih.gov/31013738/ -> They indeed report few stunted kids in vegan group, but it seems to be correlated with exclusive breast feeding for periods longer then recommended.
– Do cultures who don’t consume dairy(even as kids) have some genetic adaptations? Or by habit do they eat other sources of calcium, vitamin D, protein and fatty acids?
Thanks for your questions, Cenk.
1. CVD Risk
Yes, we’re generally talking about a spectrum of effect in terms of replacement, and while dairy SFA may benefit replacing meat, substitution of all SFA with PUFA generally yields greater relative risk reductions. This was demonstrated in the NHS and HPFS cohorts, where replacing 5% dairy fats with total PUFA was associated with:
* 24% reduction in CVD risk [RR 0.76, 95% CI 0.71-0.81]
* 26% reduction in CHD risk [RR 0.74, 95% CI 0.68-0.81]
* 22% reduction in stroke risk [RR 0.78, 95% CI 0.7-0.88]
So the general rule of PUFA>SFA still applies in the context of dairy.
2. Breast cancer.
I’m familiar with that AHS-2 study. In general, it isn’t that entirely inconsistent with the wider literature. While studying any cancer and diet relationship is difficult, what we go by is the overall direction of effect – as there will always be some null or contradictory studies. The overall direction of effect with dairy and BC is toward reduction in risk. Nonetheless, the AHS-2 cohort is very well conducted prospective study in a generally health conscious population, with a diversity of dietary habits.
This study, overall, is not entirely inconsistent: for example, in premenopausal women there were no significant associations for any dairy measure. In postmenopausal women the significant finding related to milk, but not other parameters. The strongest finding of the study was the substitution analysis with regard to isoflavones, and the wider literature does suggest a protective effect of soy isoflavones against BC, which may mediated the effects of soy milk.
I agree, the issue is a real lack of data, in addition to difficulties studying populations based on parental surveys (bias). Breastfeeding is a big effect modifier, but also total energy and protein was similar: this may reflect the fact that all children in that age group in the VeChi study had nutritional adequacy. It may be that sufficient protein and energy is enough, but right now the best data on this experimentally in fact comes from the developing world, so it is hard to make meaningful comparisons right now until further research comes out. I’m not familiar with cultures who exclude dairy in entirety, if you could point me to any relevant ones I’d be happy to take a look!
Thanks for the response Alan.
About point 1:
I understand what you say, but then I still don’t understand the wording on the summary: “Overall evidence suggests either a neutral effect of total dairy on CVD risk, or modest reductions in risk”.
If we can only talk about a relative risk compared to the replaced food, how can we have a global statement such as “Effect of food A is reduction in risk”.
Sorry if I’m being slow 🙂
About point 3:
“I’m not familiar with cultures who exclude dairy in entirety, if you could point me to any relevant ones I’d be happy to take a look!”
I may be wrong. I assumed that, due to fact that some countries have high rates of lactose intolerance, milk consumption must be low among all ages. Now taking another look at it, it doesn’t seem to be the case for children. For instance even though Vietnam was one of the lowest dairy consumers, 47% of children are classified as dairy users.
Apologies about the delay!
In relation to No.1, this is simply because we don’t only have to talk about relative risk compared to replacement: it depends on the type of analysis used. A basic analysis can examine the effects of a given food/nutrient by comparing the highest quantile of intake vs. lowest. A substitution analysis can be also used to model the effects of replacement, for example, isocalorically replacing 5% of energy from one food/nutrient with another. Does that make sense?
In relation to No.3, that is my understanding: in countries that may not traditionally have consumed dairy, prevalence has increased due to transnational trade, etc.
Any thoughts on the adverse impact on women’s hormones of dairy consumption? So many of my patients are doing dairy-free diets to address infertility and oligomenorrhea. I think this is not supported by evidence outside of anecdotal evidence. Thank you in advance
I am unaware of any evidence that shows that dairy consumption is problematic for those issues or that it disrupts hormones in women. As is the case with people generally, some people don’t tolerate dairy, which could be related to lactose, dairy proteins, etc. And intolerance differs across ethnic groups and genetics. So such people would experience some health benefit to avoidance of foods that cause symptoms. However, I have not seen evidence this relates to female hormonal issues or female-specific health issues.
Dear Team, in your great statement you wrote: ‘‘Dairy fatty acids may also exert anti-diabetic effects, with trans-palmitoleic acid associated with reduced risk for incident type-2 diabetes in the Multi-Ethnic Study of Atherosclerosis, and mechanistic research indicating that trans-palmitoleic acid may improve hepatic and peripheral insulin sensitivity.‘‘
What do you think about this new RTC of the impact of low-fat or full-fat dairy foods on glucose homeostasis. No evidence that consuming 3+ servings of either low-fat or full-fat dairy foods per day for 12 weeks improves glucose tolerance or other measures of metabolic health (increase in insulin resistance). Thank you in advance
Study:The relationship between high-fat dairy consumption and obesity, cardiovascular, and metabolic disease https://pubmed.ncbi.nlm.nih.gov/22810464/
Sry, this study ->The impact of diets rich in low-fat or full-fat dairy on glucose tolerance and its determinants: a randomized controlled trial https://academic.oup.com/ajcn/advance-article-abstract/doi/10.1093/ajcn/nqaa301/5979929