Increase Font Size Decrease Font Size View as PDF Print

Elwood PC, Givens DI, Beswick AD, Fehily AM, Pickering JE, Gallacher J. The survival advantage of milk and dairy consumption: An overview of evidence from cohort studies of vascular diseases, diabetes and cancer. J Am Coll Nutr. 2008; 27 (6): 723S-734S

PubMed ID: 19155432
Study Design:
Systematic review and meta-analysis
M - Click here for explanation of classification scheme.
POSITIVE: See Research Design and Implementation Criteria Checklist below.
Research Purpose:
  • To report the results of a literature search for prospective cohort and case-control studies of milk and dairy consumption as predictors of vascular disease and diabetes and meta-analyses of the results in the papers identified
  • To summarize the conclusions of the recent report by the World Cancer Research Fund and American Institute for Cancer Research
  • To examine the evidence related to consumption of whole vs. reduced fat milks and consider the likely effect of milk and dairy consumption on survival.
Inclusion Criteria:
  • Using Cochrane systematic review methods the computerized database MEDLINE was searched up to June 2008. Each search was limited to human/adult. The key words Milk/milk protein/dairy/dairy calcium produced 11,102 hits. Heart disease/coronary artery disease/myocardial infarction/ischaemic heart disease produced 125,572 hits and stroke produced 61,878 hits. Diabetes/metabolic syndrome gave 58,473 hits. Combined, these gave 180 papers on milk etc. and heart disease etc, 33 papers on milk and stroke and 111 on milk and diabetes
  • Papers that included the following were reviewed: Population based and prospective, gave baseline data on milk or dairy consumption, vascular disease outcome or incident diabetes
  • The references listed in each of these selected papers were also searched for other suitable reports. For heart diseases 11 papers were found to be relevant and to contain the data necessary for inclusion in a meta-analysis; for stroke seven and for diabetes four papers. Cross sectional case-control papers were also identified for the metabolic syndrome (four papers) and for myocardial infarction four.
Exclusion Criteria:

Results were reported excluding estimates.

Description of Study Protocol:
  • Recruitment: Not applicable 
  • Design: Review and meta-analysis 
  • Dietary intake/Dietary assessment methodology: Not applicable 
  • Blinding used:  Not applicable  
  • Intervention: Not applicable 
  • Statistical analysis: 
    • Reported adjusted relative risks (RR) given in each paper were noted. Pooled estimates of the RR were determined by weighting the natural logs of the reported RR in each report by the inverse of the variance. Where variance was not estimable from confidence intervals, the standard error from a study of similar size was used
    • To estimate the effect of milk or dairy consumption on survival, they relate the data on disease risks to data on mortality in a major part of the UK (England and Wales) from the various life-shortening diseases considered (vascular disease, diabetes and cancer)
    • Studies which give disease risks in relation to the type of milk, whole or fat-reduced, within the same cohort, were examined.


Data Collection Summary:
  • Timing of measurements: Not applicable   
  • Dependent variables: Not applicable 
  • Independent variables: Not applicable  
  • Control variables: Not applicable. 



Description of Actual Data Sample:
  • Initial N: The initial search found: 180 papers on milk and heart disease, 33 papers on milk and stroke and 111 on milk and diabetes
  • Attrition: For the meta-analyses, 11 papers were used for cardiovascular disease (CVD), seven papers were used for stroke and four papers were used for diabetes. Cross-sectional case-control papers were also identified for review of metabolic syndrome (four) and myocardial infarction (four) 
  • Age: Age of the study participants was not stated  
  • Ethnicity: Ethnicity of the study participants was not stated  
  • Other relevant demographics: Not applicable   
  • Anthropometrics: Were not stated  
  • Location: Location of the studies was not stated.
Summary of Results:

Metabolic Syndrome  

Overall, the data on the metabolic syndrome suggest a reduced incidence from milk and dairy consumption. There was a reduction in metabolic syndrome in subjects with the highest milk consumption (RR and 95% CI: 0.74; 0.64, 0.84).  

Study Total Number of Subjects Dietary Item Groups Compared Number With the Syndrome Adjustments for Possible Confounding Adjusted RR1 in High Milk Subgroups
Mennen et al [16] 2,439 Males Dairy Four or more portions per day vs. less than one per day 660 Age, energy, waist-hip ratio 0.63 (0.40-0.99)2
2,537 Females 941 0.76 (0.47-2.66)3
Azadbakht et al [17] 827 Subjects Dairy  Top and bottom quartile 97 Age, sex, activity, smoking, BMI, waist/hip ratio, energy, various foods, anti-hypertensive and estrogen therapy 0.75 (0.63-0.96)
Liu et al [18] 10,066 Females Milk Top and bottom quintile 1,731 Age, smoking, exercise, alcohol multivitamins parental MI 0.85 (0.71-1.02)4
Dairy 0.66 (0.55-0.80)5
Elwood et al [19] 2,251 Males Milk One pint per day vs. under a third per day 342 Age, smoking, social class, IHD, BMI, energy, alcohol, fasting total cholesterol HDL cholesterol and triglycerides 0.38 (0.18-0.78)


Relative risk (RR) of the metabolic syndrome in the high milk group 0.74 (95% CI 0.64-0.84) 

1 RR (95% CI).2 Males.3 Females.4 Milk.5 Dairy. 
Myocardial infarction (MI) (Table 2 and 3): Overall, the data on MI suggest a reduced incidence from milk and dairy consumption. There was a reduction in MI events in subjects with the highest milk consumption (RR and 95% CI: 0.83 (0.66-0.99).  

Table 2. Summary of Case-Control Studies of Milk and MI  

Study Dietary Item Number of Cases Number of Controls RR1
Gramenzi et al [24] Milk intake 287 649 0.90
Tavani et al [25] Milk intake 507 478 0.78 (0.54-1.12) 
Lockheart et al [26] Dairy intake 106 105 0.82 (0.58-1.16)
Biong et al [27] Dairy fat intake 111 107  0.67 (0.24-1.83)

RR of MI in the high milk group 0.83 (95% CI 0.66-0.99)
(Excluding Gramenzi et al [24] with estimated variance 0.79 (0.59-0.99) 

1 RR (95% CI). 
Table 3. Prospective Cohort Studies on Milk or Dairy Consumption and Incident Vascular Disease Events  

Study Number of Subjects  Duration of Follow-up Number of Events  Groups Compared Adjustments for Possible Confounding 
Milk and Dairy Foods          

Snowdon et al [28]

24,172 Subjects 20 years 758 male IHD deaths Two glasses of milk per day vs. none Age, smoking and other food items, weight and marital status
841 female IHD deaths

Shaper et al[29]

7,735 Males 9.5 years  608 IHD events Milk drunk and taken on cereals vs. none Age, social class, smoking, cholesterol, blood pressure and diabetes

Abbott et al [30]

3,150 Males 22 years 229 strokes 16 ounces per day milk drunk vs. non-drinkers Age, dietary K and Na, alcohol, smoking, activity, cholesterol and glucose, uric acid and haematocrit

Mann et al [31]

10,802 Vegetarian subjects 13 years  63 IHD deaths More than 1/2 pint milk per day vs. less than 1/2 pint Age, sex, smoking and social class

Bostick et al [32]

34,486 Females 8 years 387 IHD deaths Top and bottom quartile Age, energy, BMI, waist-hip ratio, diabetes, smoking, Vitamin E and saturated fat 

Kinjo [33]

223,170 Subjects 15 years  11,030 strokes Milk four or more times per week vs. less than once per week Sex, age, area, smoking, alcohol and occupation

Hu et al [34]

80,082 Females 14 years  939 vascular events More than two glasses of milk per day vs. less than one glass per week Time period, BMI, smoking, menopause, parental history, Vitamin E, alcohol, hypertension, aspirin use and exercise

Ness et al [35]

5,765 Males 25 years  892 IHD deaths More than one pint per day vs. less than one third per day Social class, health behavior and health status
892 IHD deaths

Elwood et al [36]

2,512 Males 20 years 493 IHD events One or more pint per day vs. one third of a pint or less per day Age, smoking, social class, IHD, BMI, energy, alcohol, fasting cholesterol, HDL-C and triglycerides
185 strokes

Sauvaget et al [38]

40,349 Subjects 16 years  1,462 stroke deaths Milk almost daily Smoking, alcohol, BMI, education, diabetes, hypertension and area
Dairy almost daily

Lamarche [37]

2,000 Males 13 years  217 IHD events Above and below average intake of dairy products Age, smoking, BMI, diabetes
Dairy or Total Dietary Calcium Intake     

Vijvjer et al [39]

2,606 Subjects 28 years  366 male IHD deaths Top and bottom quintile Age, smoking, BMI, SBP, cholesterol, energy and alcohol 
178 female IHD deaths

Iso et al [40]

85,764 Females 14 years  690 strokes Top and bottom quintile Age, smoking, time interval, BMI, alcohol, menopause, hormone use, exercise, multivitamins, fatty acid intake, history of hypertension, diabetes and cholesterol 

Al-Delaimy et al [41]

39,800 Males 12 years  1,458 IHD events Top and bottom quintile Age, duration, energy, diabetes, hypercholesterolaemia family history, smoking, aspirin, BMI, alcohol, activity, Vitamin E and various nutrients 

Umesawa et al [42]

21,068 Males 10 years  234 IHD deaths Top and bottom quintile of dairy calcium intake Age, BMI, hypertension, diabetes, smoking, alcohol, potassium and energy
32,319 Females 566 stroke deaths

Ishcaemic Heart Disease (IHD) and Stroke (Table 4):   

  • Overall there was a reduction of about 10 to 15% in the incidence of IHD in the subjects who had reported drinking the most milk, relative to those drinking the least milk (0.91 (95% CI 0.82-1.00) and 0.84 (95% CI 0.76-0.93).
  • The meta-analysis indicates about a 20% reduction in stoke events in the subjects who had reported drinking the most milk, relative to those drinking the least milk within each cohort (0.79 (95% CI 0.75-0.82).

Table 4. Meta-analysis of Prospective Studies of Milk and Dairy Consumption, Ischaemic Heart Disease and Stroke  

Study Number of Subjects Number of Events Predictive Factor Adjusted RR
(95% CI)
Ischaemic heart disease        
Snowdon et al [28]
8,724 758 Milk 0.94
(females) 15,448 841 1.11
Shaper et al [29] 7,735 608 Milk 0.88 (0.55-1.40)
Mann et al [31] 10,802 63 Milk 1.50 (0.81-2.78) 
Bostick et al [32] 34,486 387 Milk 0.94 (0.66-1.35) 
Hu et al [34] 80,082 939 Whole milk 1.67 (1.14-1.90)
Low-fat milk 0.78 (0.63-0.96)
High-fat dairy 1.04 (0.96-1.12)
Low-fat dairy 0.93 (0.85-1.02)
Ness et al [35] 5,765  892 deaths  Milk 0.68 (0.40-1.13) 
Elwood et al [36] 2,512  493  Milk 0.71 (0.40-1.26) 
Al Delaimy et al [41] 39,800  1,458  Dairy calcium  1.03 (0.86-1.26) 
van Vijjver et al [39]
1,340 366 Dietary calcium  0.77 (0.53-1.11) 
(females) 1,265 178 0.91 (0.55-1.50)
Lamarche [37] 2,000 217  Dairy intake  0.73 (0.56-0.93) 
Umesawa [42]  53,387  234 deaths  Dairy calcium  0.80 (0.45-1.44) 
Kinjo et al [33]  223,170  11,030  Milk  0.79 (0.75-0.83)
Ness et al [35]  5,765  196 deaths  Milk  0.84 (0.31-2.30) 
Sauvaget et al [38]  40,349  1,462  Milk  0.94 (0.79-1.12) 
Dairy products 0.73 (0.57-0.94)
Elwood et al [36] 2,512 185  Milk  
Abbott et al [30]  3,150  229  Dairy calcium 0.67 (0.45-1.00) 
Iso et al [40]  85,764  690  Dairy calcium  0.83 (0.66-1.04) 
Umesawa [42]  53,387 566 deaths  Dairy calcium  0.53 (0.34-0.81) 


  • RR of IHD in the high milk group, including Hu et al [34] whole milk: 0.91 (95% CI 0.82-1.00), see note 1 below(Excluding Snowdon et al [28] with estimated variance: 0.90 (0.80-0.99))
  • RR if IHD in the high milk group, including Hu et al low-fat milk: 0.84 (95% CI 0.76-0.93) see note 2 and 3
    (Excluding Snowdon et al [28] with estimated variance: 0.83 (0.74-0.91))
  • RR of stroke in the high milk group: 0.79 (95% CI 0.75-0.82).
  1. When the estimates of Hu et al of 1.67 for whole milk is included there is considerable heterogeneity: (I2=54.1%)
  2. There is homogeneity when their estimate of 0.78 for low-fat milk is used in the meta-analysis
  3. The estimates by Hu et al for dairy foods were not included in the meta-analyses.

Diabetes: RR for type 2 diabetes was estimated to be 10% lower in people who had a high milk intake (0.92; 0.86-0.97) (Table 5).  

Table 5. Prospective Studies of Milk or Dairy Consumption and Incident Diabetes 

Study Number of subjects Duration of follow-up Groups compared Number who developed diabetes Adjustments for possible confounding Adjusted RR1
Choi et al [51] 41,254 Males 12 years Top and bottom quintiles of total dairy 1,243 Age, total energy, follow-up time, family history, smoking, BMI, hypercholesterolaemia, hypertension, activity, alcohol and certain nutritional factors  0.91 (0.85-0.97) 
Liu et al [52] 37,183 Females 10 years Two or more servings of dairy foods per week vs. less than one serving per month 1,603 Age, total energy, diabetes in family, smoking, BMI, hypercholesterolaemia, hypertension, hormone therapy, activity, total fat, glycaemic load, diet Ca, Vitamin D and Mg 1.04 (0.84-1.30)2
            0.92 (0.78-1.09)3
van Damm et al [53] 41,186 Females 8 years Quintiles of dietary calcium intake 1,964 Age, total energy, BMI, smoking, physical activity, alcohol, parental diabetes, education, coffee and soft drinks, processed and red meat 0.93 (0.75-1.15)
Elwood et al [19] 640 Males 20 years Highest quartile of milk intake vs. lowest quartile 41 Age, smoking, BMI and social class  0.57 (0.20-1.63) 


RR of incident diabetes in the high milk group 0.92 (0.86-0.97) (using the estimate by Liu et al [52] for low-fat milk 0.91 (0.86-0.96)  

1 RR (95% CI).2 Whole milk.3 Skimmed milk.  

Cancer (Table 6):  An increased consumption of milk or dairy food is associated with significant reduction in colon cancer, RR attributable to milk being between 0.78 and 0.94 per serving per day in pooled cohort studies. There is a significantly increased risk of prostate cancer, risk associated with milk and dairy consumption in pooled cohort studies being 1.06 (1.01, 1.11) per serving per day. Milk and dairy consumption associated with increased risk of bladder cancer, estimate of risk from pooled cohort studies 0.82 (0.67, 0.99) per serving per day. No relationship of importance was reported for any other cancer.  

Table 6. Summary of Relationships between Milk or Dairy Consumption and Cancer Taken From the Report of the World Cancer Research Fund [5]  

Cancer Predictor Number of Studies Pooled Relative Risk1 Heterogeneity
Colorectal Milk 4 cohorts 0.94 (0.85-1.03) ‘low’
Milk 10 cohorts 0.78 (0.69-0.88) Not state
Prostate Milk 8 cohorts 1.05 (0.98-1.14)  ‘low’
Milk 6 case-control 1.08 (0.98-1.19) ‘moderate’
Milk and dairy 8 cohorts 1.06 (1.01-1.11) ‘moderate’
Milk and dairy 5 case-control 1.03 (0.99-1.07) ‘low’
Bladder Milk 4 cohorts 0.82 (0.67-0.99) ‘moderate’
Milk 3 case control 1.00 (0.87-1.14) ‘high’

 1 ñ

Whole and fat-reduced milks: The authors concluded that persons who choose to drink fat-reduced milks were more likely to adopt other "healthy behaviors" that could confound the results. Therefore, no conclusions were given for whole vs. reduced-fat milks.  

Table 7. Relationships with Whole Milk and Fat Reduced Milks Compared  

Study Total Number in the Study Type of Study Outcome Disease RR in Highest 1/4 or 1/5
Whole Milk (1) High fat Dairy (3) Fat Reduced (2) Low fat Dairy (4)
Hu et al [34] 80,082 Females Prospective Ischaemic heart disease 1.67 (1.14-1.90)1  0.78 (0.63-0.96)2
1.08 (0.96-1.12)3 0.82 (0.85-1.02)4
Tavani et al [25] 985 Subjects Case-control Fatal MI 0.89 (0.57-1.38)1 0.83 (0.59-1.16)2
Lockheart et al [26] 211 Subjects Case-control MI 0.48 (0.20-1.14)3 0.96 (0.42-2.23)4 
Lui et al [18] 10,066 Females Cross sectional  Metabolic syndrome  0.71 (0.58-0.87)3 0.78 (0.64-0.95)4 
Choi et al [51] 41,254 Males 12 years prospective Diabetes  1.19 (1.00-1.43)1 0.95 (0.80-1.13)2
Liu et al [52] 37,183 Females 10 years prospective  Diabetes  1.00 (0.96-1.05)3 0.92 (0.84-1.01)4 
van Dam et al [53] 41,186 Females 8 years prospective Diabetes 1.03 (0.88-1.20)3 0.87 (0.76-1.00)4  
Mettlin et al [54] 2,561 Subjects Case control Colon cancer  1.8; 1.3-2.41 1.0 (0.7-1.4)
Rectal cancer 2.0; 1.4-2.81 0.8 (0.5-1.3)2
Prostate cancer 1.5; 1.0-2.21 1.2 (0.7-2.1)2
Bladder cancer 2.0; 1.3-3.11 0.6 (0.3-1.2)2
Veierod et al [55]  25,708 Males  9-15 years prospective Prostate cancer Set at 1.01 2.2 (1.3-3.7)2 
Sing and Frazer [56] 32,051 Subjects  6 years prospective  Colon cancer  1.04 (0.69-1.59)1 0.97 (0.66-1.42)2 
Michaud et al [57]  51,529 Males  10 year Prospective  Prostate cancer  1.12 (0.70-1.8)1 1.37 (0.90-1.5)2 
Kampman et al [58] 16,945 Subjects  Case control  Colon cancer 1.1 (0.8-1.5) 3,5  0.8 (0.6-1.0) 4,5 
0.9 (0.6-1.2) 3,6
0.7 (0.5-1.0) 4,6
Tseng et al [59] 3,612 Males  Prospective Prostate cancer 0.8 (0.5-1.3)1  1.5 (1.1-2.2)2
Gallus et al [60]  3,247 Subjects Case control  Colon cancer 0.99 (0.86-1.13)1 0.84 (0.73-0.97)2
Rectal cancer 1.22 (1.03-1.44)1 0.76 (0.64-0.91)2
Prostate cancer 1.06 (0.90-1.25)1 1.11 (0.94-1.31)2

 1 Whole milk.2 Low-fat milk.3 High fat dairy.4 Low-fat dairy.5 Males.6 Females.

Author Conclusion:
  • The analyses gives fairly clear evidence of a reduction in vascular disease and type 2 diabetes by milk and dairy consumption. Taken together with the probable reduction in colon cancer and allowing for some increase in prostate cancer there is fairly convincing overall evidence that milk and dairy consumption is associated with an increase in survival in Western communities
  • The gradient in milk intakes may contribute to health inequalities.




Reviewer Comments:

Regarding the survival advantage of milk or dairy consumption

  • The authors did not estimate an overall quantitative survival advantage and acknowledged the large number of assumptions needed
  • They state that after considering the number of death from various causes were taken into account, the data suggests that there is an overall reduction in the number of deaths and hence an increase in overall survival attributable to consumption of milk and dairy foods.

Research Design and Implementation Criteria Checklist: Review Articles
Relevance Questions
  1. Will the answer if true, have a direct bearing on the health of patients?
  2. Is the outcome or topic something that patients/clients/population groups would care about?
  3. Is the problem addressed in the review one that is relevant to nutrition or dietetics practice?
  4. Will the information, if true, require a change in practice?
Validity Questions
  1. Was the question for the review clearly focused and appropriate?
  2. Was the search strategy used to locate relevant studies comprehensive? Were the databases searched and the search terms used described?
  3. Were explicit methods used to select studies to include in the review? Were inclusion/exclusion criteria specified and appropriate? Were selection methods unbiased?
  4. Was there an appraisal of the quality and validity of studies included in the review? Were appraisal methods specified, appropriate, and reproducible?
  5. Were specific treatments/interventions/exposures described? Were treatments similar enough to be combined?
  6. Was the outcome of interest clearly indicated? Were other potential harms and benefits considered?
  7. Were processes for data abstraction, synthesis, and analysis described? Were they applied consistently across studies and groups? Was there appropriate use of qualitative and/or quantitative synthesis? Was variation in findings among studies analyzed? Were heterogeneity issued considered? If data from studies were aggregated for meta-analysis, was the procedure described?
  8. Are the results clearly presented in narrative and/or quantitative terms? If summary statistics are used, are levels of significance and/or confidence intervals included?
  9. Are conclusions supported by results with biases and limitations taken into consideration? Are limitations of the review identified and discussed?
  10. Was bias due to the review’s funding or sponsorship unlikely?