In this cohort study among 1956 Dutch men and women aged 50–75 years, a positive association was found between high-fat dairy intake and the risk of CVD mortality.
Several observational studies have evaluated the association between dairy product intake and CVD outcomes [4, 8, 9, 11–14, 24, 26, 27]. The results of these studies were inconsistent, possibly because of the heterogeneity of the population and the differences in design and dietary assessment methodology used by the various studies, as well as the exposures and outcomes studied. Dairy intake is a rather heterogeneous exposure, and little is known about specific types of dairy foods in relation to CVD.
Total dairy was not associated with CVD mortality and all-cause mortality in this study. A previous study within the Hoorn Study  showed that the consumption of high-fat dairy was significantly inversely associated with cardiovascular risk factors such as BMI and lipid levels, which may seem inconsistent with our current results. However, the cross-sectional associations are likely to be disturbed by reversed causation , a phenomenon which is excluded in longitudinal studies. Our current results on mortality are in line with an earlier prospective analysis of the Hoorn Study, where no association was found between total dairy intake and changes in cardiovascular risk factors . No distinction was made between different types of dairy in this study.
In the present study, high-fat dairy intake was associated with an increased risk of CVD mortality. Dairy foods, and especially high-fat dairy, contribute to the intake of saturated fat. Especially longer chain saturated fat has previously been associated with higher incidence of CHD . Moreover, saturated fatty acids raise the LDL-cholesterol concentration in the blood, which is one of the main risk factors for CVD. A recent study of Goldbohm et al.  found that dairy fat intake was associated with slightly increased all-cause and ischemic heart disease mortality rates. Further evidence to support our positive results on high-fat dairy and CVD mortality came from ecological studies [30, 31] and controlled feeding studies finding associations on LDL-cholesterol [32, 33]. To explore whether the positive association between high-fat dairy and CVD mortality can be attributed to fat or more sugar containing products, we split our analyses by desserts. Increased risks for CVD were only found in the “non-desserts” category. This strengthened our suggestion that it is potentially the high amount of fat that causes this high risk of CVD mortality. On the other hand, additional adjustment for saturated fat moderately reduced the association between high-fat dairy and CVD mortality, but remained statistically significant. It is therefore very well possible that besides saturated fat, there are other, unknown, components or mechanisms that could exert an effect on the risk of CVD mortality . The consumption of high-fat dairy might for example reflect more unhealthy lifestyle or dietary pattern, which was not all measured. Residual confounding cannot be completely ruled out. Furthermore, when physical activity, the confounder with the largest number of missing values (n = 255), was removed from the final model, the hazard ratio became lower (HR, 1.24; 95 % CI, 1.03, 1.49), but remained statistically significant. Caution is therefore needed with the conclusion that each SD increase in high-fat dairy intake would be associated with a 32 % higher risk of CVD mortality.
There is growing evidence for a protective effect of low-fat dairy on blood pressure [13–15]. The mechanism by which especially low-fat dairy exerts a beneficial effect on blood pressure remains to be established. Vitamin K2, naturally present in fermented dairy products, has been recently suggested to decrease aortic calcification and CHD [35, 36].
In this study, fermented dairy and low-fat dairy were not statistically significantly associated with CVD mortality or all-cause mortality. Bernstein et al.  reported inverse associations between low-fat dairy and CHD only after 26 years of follow-up. It might be possible that the beneficial effect of fermented and low-fat dairy on CVD becomes visible after longer exposure and thus longer follow-up time.
Several strengths and limitations of our study need to be addressed. In the Netherlands, the intake of dairy products is naturally high, with a wide variety of dairy products that are consumed. To put this in perspective, in the most recent European Nutrition and Health report, the highest consumption of milk- and dairy-based products was reported in countries like Norway (522 g/day), Finland (437 g/day), the Netherlands (388 g/day), and United Kingdom (320 g/day), whereas intake was low in Austria (171 g/day), Poland (181 g/day), and the Czech Republic (186 g/day) . In our data, the mean intake of dairy (466 g/day) was higher than in this report (mean dairy intake 266 g/capita/day over 16 European countries) with a large variation, which enabled us to examine the relationship between dairy food consumption and CVD over a broad range of intake.
A possible limitation is the estimation of dairy intake used in the analysis, which was derived from a semi-quantitative FFQ. The self-report of usual dietary intake could lead to misclassification. However, the FFQ was validated by comparison with a modified dietary history. Pearson correlation coefficients for estimates from the questionnaire and dietary history were on average 0.71 (range, 0.65–0.78) for macronutrients and 0.66 (range, 0.36–0.81) for vitamins and minerals . Although the FFQ was not validated for dairy intake, high correlations were found for protein and calcium (0.69 and 0.75, respectively). Moreover, in the Netherlands, dairy is consumed on typical moments during the day, often during breakfast and/or lunch and after a hot meal, making it easier for the participants to recall the usual dairy intake.
A limitation of our study is that only dairy consumption at baseline was investigated, and not during follow-up. Participants may have changed their dietary pattern during the study. Although it cannot be assumed, 50- to 75-year-old adults may be less likely to change nutritional habits.
Another limitation is that we had a limited number of CVD and all-cause mortality cases. Therefore, we could not carry out analyses for dairy consumption by tertiles and presented continuous analyses per SD. Furthermore, no information on CVD morbidity was available. The participants had to give permission to access their hospital records, which only a small number of people did. It was therefore only possible to assess the relationship with mortality, and not with morbidity.
At last, a remark should be made about the age of the population. This study population was somewhat older than other cohorts, with an age between 50 and 75 years at baseline. Caution is needed with generalizing these results to a younger population.
In this Dutch prospective cohort study, the intake of high-fat dairy products was associated with an increased risk of CVD mortality.
Based on the results of this study, it can indeed be recommended to rather consume low-fat dairy instead of high-fat dairy in order to reduce the risk of CVD. However, to draw valuable conclusions, more research is needed in larger prospective studies with sufficient cases of stroke and CHD.