Hostname: page-component-848d4c4894-5nwft Total loading time: 0 Render date: 2024-05-18T02:58:32.371Z Has data issue: false hasContentIssue false

Consumption of dairy products and CVD risk: results from the French prospective cohort NutriNet-Santé

Published online by Cambridge University Press:  29 April 2021

Laury Sellem
Affiliation:
Hugh Sinclair Unit of Human Nutrition, Department of Food and Nutritional Science, University of Reading, Whiteknights, Pepper Lane, Harry Nursten Building, Reading, RG6 6DZ, UK
Bernard Srour*
Affiliation:
Sorbonne Paris Nord University, Inserm, INRAE, Cnam, Nutritional Epidemiology Research Team (EREN), Epidemiology and Statistics Research Center, University of Paris (CRESS), 93017 Bobigny, Bobigny, France
Kim G. Jackson
Affiliation:
Hugh Sinclair Unit of Human Nutrition, Department of Food and Nutritional Science, University of Reading, Whiteknights, Pepper Lane, Harry Nursten Building, Reading, RG6 6DZ, UK
Serge Hercberg
Affiliation:
Sorbonne Paris Nord University, Inserm, INRAE, Cnam, Nutritional Epidemiology Research Team (EREN), Epidemiology and Statistics Research Center, University of Paris (CRESS), 93017 Bobigny, Bobigny, France Public Health Department, Avicenne Hospital, AP-HP, Bobigny, France
Pilar Galan
Affiliation:
Sorbonne Paris Nord University, Inserm, INRAE, Cnam, Nutritional Epidemiology Research Team (EREN), Epidemiology and Statistics Research Center, University of Paris (CRESS), 93017 Bobigny, Bobigny, France
Emmanuelle Kesse-Guyot
Affiliation:
Sorbonne Paris Nord University, Inserm, INRAE, Cnam, Nutritional Epidemiology Research Team (EREN), Epidemiology and Statistics Research Center, University of Paris (CRESS), 93017 Bobigny, Bobigny, France
Chantal Julia
Affiliation:
Sorbonne Paris Nord University, Inserm, INRAE, Cnam, Nutritional Epidemiology Research Team (EREN), Epidemiology and Statistics Research Center, University of Paris (CRESS), 93017 Bobigny, Bobigny, France Public Health Department, Avicenne Hospital, AP-HP, Bobigny, France
Léopold Fezeu
Affiliation:
Sorbonne Paris Nord University, Inserm, INRAE, Cnam, Nutritional Epidemiology Research Team (EREN), Epidemiology and Statistics Research Center, University of Paris (CRESS), 93017 Bobigny, Bobigny, France
Mélanie Deschasaux-Tanguy
Affiliation:
Sorbonne Paris Nord University, Inserm, INRAE, Cnam, Nutritional Epidemiology Research Team (EREN), Epidemiology and Statistics Research Center, University of Paris (CRESS), 93017 Bobigny, Bobigny, France
Julie A. Lovegrove
Affiliation:
Hugh Sinclair Unit of Human Nutrition, Department of Food and Nutritional Science, University of Reading, Whiteknights, Pepper Lane, Harry Nursten Building, Reading, RG6 6DZ, UK
Mathilde Touvier
Affiliation:
Sorbonne Paris Nord University, Inserm, INRAE, Cnam, Nutritional Epidemiology Research Team (EREN), Epidemiology and Statistics Research Center, University of Paris (CRESS), 93017 Bobigny, Bobigny, France
*
*Corresponding author: Dr Bernard Srour, email b.srour@eren.smbh.univ-paris13.fr
Rights & Permissions [Opens in a new window]

Abstract

In France, dairy products contribute to dietary saturated fat intake, of which reduced consumption is often recommended for CVD prevention. Epidemiological evidence on the association between dairy consumption and CVD risk remains unclear, suggesting either null or inverse associations. This study aimed to investigate the associations between dairy consumption (overall and specific foods) and CVD risk in a large cohort of French adults. This prospective analysis included participants aged ≥18 years from the NutriNet-Santé cohort (2009–2019). Daily dietary intakes were collected using 24-h dietary records. Total dairy, milk, cheese, yogurts, fermented and reduced-fat dairy intakes were investigated. CVD cases (n 1952) included cerebrovascular disease (n 878 cases) and CHD (n 1219 cases). Multivariable Cox models were performed to investigate associations. This analysis included 104 805 French adults (mean age at baseline 42·8 (sd 14·6) years, mean follow-up 5·5 (sd 3·0) years, i.e. 579 155 person-years). There were no significant associations between dairy intakes and total CVD or CHD risks. However, the consumption of at least 160 g/d of fermented dairy (e.g. cheese and yogurts) was associated with a reduced risk of cerebrovascular diseases compared with intakes below 57 g/d (hazard ratio = 0·81 (95 % CI 0·66, 0·98), Ptrend = 0·01). Despite being a major dietary source of saturated fats, dairy consumption was not associated with CVD or CHD risks in this study. However, fermented dairy was associated with a lower cerebrovascular disease risk. Robust randomised controlled trials are needed to further assess the impact of consuming different dairy foods on CVD risk and potential underlying mechanisms.

Type
Full Papers
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution, and reproduction in any medium, provided the original work is properly cited.
Copyright
© The Author(s), 2021. Published by Cambridge University Press on behalf of The Nutrition Society

CVD, including CHD and cerebrovascular diseases such as strokes, are still a leading cause of mortality worldwide, causing 17·9 million deaths every year(1,Reference Vos, Lim and Abbafati2) . In France, CVD are the primary cause of death in women and the second most common cause in men(3). Beside smoking, lifestyle factors such as nutritional status and dietary habits have been identified as one of the main modifiable risk factors of CVD(4). Thus, public health guidelines around the world target the consumption of specific nutrients and food groups as a strategy for reducing CVD risk at a population level. In particular, reducing the dietary consumption of SFA is often recommended to help lower circulating levels of LDL-cholesterol, a well-established risk factor for CVD(Reference Borén, Chapman and Krauss5Reference Kaptoge, Pennells and De Bacquer6). In France, public health guidelines suggest a consumption of SFA below 12 % of dietary energy (without alcohol), with an emphasis on three fatty acids that should remain below 8 % of dietary energy (lauric C12 : 0, myristic C14 : 0 and palmitic C16 : 0 acids)(7). In parallel, the French National Health and Nutrition Programme (PNNS) focuses on recommendations related to food groups and suggests a daily consumption of two servings of dairy products per d for adults, to be chosen among milk, cheese and yogurts, but not including butter, cream or dairy desserts (e.g. custard, ice cream and cheese cake)(Reference Chaltiel, Adjibade and Deschamps8).

Dairy products are nutrient-rich foods containing various minerals and vitamins, such as Ca, K, P and vitamins B, K and D which are associated with health benefits. However, full-fat dairy foods can also be energy-dense and may contain high levels of Na and SFA. In French adults, dairy product contributes to 24 % of total dietary SFA intake(9). Despite this high contribution to dietary SFA, epidemiological evidence on the association between dairy product consumption and CVD risk remains unclear. Recent meta-analyses of prospective cohort studies suggest either null or slightly inverse associations between total dairy foods and incident CVD(Reference Fontecha, Calvo and Juarez10,Reference Soedamah-Muthu and de Goede11) . Similar trends were also observed in more recent prospective cohort studies, such as the PURE study, which included data from twenty-one countries across five continents and observed a reduced risk of CVD associated with total dairy consumption(Reference Dehghan, Mente and Rangarajan12). In addition, epidemiological evidence on specific types of dairy products remains inconsistent, although a recent meta-analysis of nine prospective cohort studies suggested a reduced CVD risk associated with the consumption of fermented dairy foods only, which included cheese, yogurt and fermented milk, as opposed to intakes of total dairy or non-fermented milks(Reference Guo, Astrup and Lovegrove13). These results raise the question of the importance of dairy food types and potential fermentation in relation to cardiovascular health.

This study aimed to investigate the associations between the consumption of dairy foods (overall and specific types) and CVD risk in a large cohort of French adults.

Materials and methods

The NutriNet-Santé cohort

The NutriNet-Santé study is an ongoing French web-based cohort, launched in 2009 with the aim of investigating the associations between nutrition, dietary behaviours, determinants of nutrition status and health. Detail of this study has been published previously(Reference Hercberg, Castetbon and Czernichow14). Briefly, since May 2009, participants aged 18 years and over with access to the Internet have been continuously recruited among the French population using vast multimedia campaigns. All questionnaires are completed online using a dedicated website (www.etude-nutrinet-sante.fr).

Ethical approval

The NutriNet-Santé study is conducted according to the Declaration of Helsinki guidelines. It was approved by the Institutional Review Board of the French Institute for Health and Medical Research (IRB Inserm n°0000388FWA00005831) and the ‘Commission Nationale de l’Informatique et des Libertés’ (CNIL n°908450/n°909216). Each participant provides their informed consent electronically. The study is registered at clinicaltrials.gov as NCT03335644.

Data collection

Upon inclusion, participants completed a set of five questionnaires related to sociodemographic and lifestyle characteristics, such as sex, occupation, educational level (<high-school degree, <2 years after high school, ≥2 years after high school), smoking status (never smoked, former smoker and current smoker), alcohol consumption (g/d), number of children(Reference Vergnaud, Touvier and Méjean15), anthropometrics(Reference Lassale, Péneau and Touvier16,Reference Touvier, Méjean and Kesse-Guyot17) (e.g. height and weight, which were validated against a random sample of participants), dietary intakes(Reference Lassale, Castetbon and Laporte18), physical activity levels (as low, moderate or high, from validated 7-d International Physical Activity Questionnaire)(19,Reference Craig, Marshall and Sjöström20) and health status (e.g. personal and family history of diseases, prescribed medication).

Dietary data

Usual dietary intakes were assessed at inclusion and then every 6 months, using a series of three non-consecutive web-based 24-h dietary records, randomly assigned over a 2-week period (2 weekdays and 1 weekend day). The web-based questionnaires used in the study have been tested and validated against both in-person interviews by trained dietitians, and urinary and blood markers(Reference Lassale, Péneau and Touvier16Reference Touvier, Kesse-Guyot and Méjean21). In this analysis, we calculated the usual baseline dietary intakes as the average of all 24-h dietary records completed during the first 2 years of each participant’s follow-up, with a mandatory requirement of at least two 24-h dietary records during this period to be included in the analysis.

At all times throughout their assigned dietary record period, participants had access to a dedicated interface of the study website to declare all foods and beverages consumed during a 24-h period: three main meals (breakfast, lunch and dinner) and any other eating occasion. When participants could not provide weights for the food items consumed, they were invited to estimate portion sizes using validated photographs or usual containers(Reference Le Moullec, Deheeger and Preziosi22). A French food composition database (>3500 items)(Reference Arnault, Caillot and Castetbon23) was used to estimate mean daily energy, alcohol, macro- and micro-nutrient intakes. These estimates included contributions from composite dishes using French recipes validated by food and nutrition professionals. Individual dietary data were not communicated to participants to avoid any changes in dietary behaviours. Finally, those who under-reported total energy intake were identified and excluded based on the method proposed by Black(Reference Black24), using the BMR, Goldberg cut-off and a physical activity level (PAL) cut-off of 1·55 which corresponds to the WHO value for ‘light’ activity. For this calculation, intra-individual CV for BMR and PAL were fixed using the values recommended by Black, i.e. 8·5 and 15%, respectively. About 20·0% of participants of the cohort were considered as under-reporters of energy intake and were excluded from the analyses. There was no sign of over-reporters as the highest energy intakes ranged within plausible values (99th percentile = 13,761 kJ/d).

Dairy products classification

Trained dietitians categorised all dairy foods of the NutriNet-Santé composition table into one of the five dairy groups defined in the PNNS: milk, cheese, yogurts, curd cheese and ‘petit-suisses’. As per the PNNS guidelines, milk-based products containing more than 12 % of sugars were classified into a separate ‘dairy desserts’ category, while creams and butter were considered as fat sources. Thus, these three food groups were not included in this analysis. We calculated a larger ‘yogurt-like dairy products’ category which included the consumption of yogurt along with curd cheese and petit-suisses, as those were not consumed frequently enough to be analysed separately. Total dairy intakes were calculated by combining the consumption of each of the five dairy groups (milk, cheese, yogurt, curd cheese and ‘petit-suisses’). In addition, we defined fermented dairy foods as cheese, curd cheese, petit-suisses, yogurts and fermented milk, whereas non-fermented dairy included all milks (UHT, pasteurised, concentrated and flavoured) except fermented ones. Finally, reduced-fat dairy products included skimmed and semi-skimmed milk, low-fat yogurts, curd cheese, petit-suisses and cheese containing <20 g of fat per 100 g final product.

Case ascertainment

Participants were invited to declare any major health event on a dedicated interface on the study website, either through the yearly health status questionnaire, through a specific health check-up questionnaire sent out every 3 months, or spontaneously. We asked participants to send their medical records (e.g. complementary examinations for diagnosis, hospitalisations and electrocardiograms) to support any health event declaration. A physician expert committee validated every major health event after reviewing the participants’ medical records and collecting additional information from the participants’ treating physicians or medical practices. In the absence of any response to the study website for more than 1 year, the physician expert committee contacted the participants’ family or physicians. In addition to this process, which constituted the main source of case ascertainment, cohort data from participants were linked to medico-administrative databases from the National Health Insurance (SNIIRAM, authorisation by the Council of State No 2013-175). Finally, deaths and potentially missed CVD events in deceased participants were identified using data from the French national cause-specific mortality registry (CépiDC).

The International Classification of Diseases-Clinical Modification codes (10th revision) were used to classify CVD cases. This study focused on first incident cases of myocardial infarction (I21), angioplasty (Z95.8), acute coronary syndrome (I20.0 and I21.4), angina pectoris (I20.1, I20.8 and I20.9), stroke (I64) and transient ischaemic attack (TIA, G45.8 and G45.9) occurring between inclusion and January 2019. CHD included all cases of myocardial infarction, angioplasty, acute coronary syndrome and angina pectoris, and cerebrovascular diseases included all cases of stroke and TIA.

Statistical analyses

As of 1 January 2019, participants with no history of CVD who had completed at least two valid 24-h dietary records were included in the analyses. Mean daily dairy intakes (overall and by type of dairy food) were coded as sex-specific quartiles, as potential distinctions in dietary patterns of French adult men and women have been previously reported(Reference Kesse-Guyot, Bertrais and Péneau25,Reference Gazan, Béchaux and Crépet26) . Missing values represented <5 % for all covariates, except for physical activity and were imputed with the modal or median value for categorical and continuous variables, respectively. Physical activity scores were only calculated when all the answers from the International Physical Activity Questionnaire were provided by the participants, which resulted in a higher percentage of missing value for this variable (13·9 %). Therefore, we introduced a missing class for this variable in the main analysis. Nonetheless, we performed additional analyses including complete cases and multiple imputation for missing values. We used the MICE method(Reference van Buuren27) to create ten imputed data sets with fully conditional specification for the outcome(Reference Sterne, White and Carlin28) and the following covariates: physical activity level, education level, smoking status and BMI. We used the SAS PROC MIANALYZE procedure(29) to combine the results from the imputed data sets, based on the combination rules proposed by Rubin(Reference Rubin30,Reference Rubin31) .

Two dietary patterns were identified using a principal component analysis based on twenty food categories derived from the fifty-eight food groups defined in the French PNNS (online Supplementary Table S1). The analysis was conducted with the SAS PROC FACTOR procedure (SAS Institute Inc.). For easier interpretation, we used the SAS ‘varimax’ option to rotate the principal components orthogonally and maximise the independence of the retained principal components. The first two principal components explained 10·6 % and 7·0 % of the variance, respectively, which was consistent with the proportion of variance observed in other nutritional epidemiology studies(Reference Santos, Gorgulho and Castro32). The first principal component was characterised by higher intakes of fruits, vegetables, soups and broths, unsweetened soft drinks and wholegrains, along with lower intakes of sweetened soft drinks, which we defined as a ‘Healthy’ dietary pattern. In contrast, the second principal component was characterised by higher consumptions of fats and sauces (which included butter and dairy cream), alcohol, meat and starchy foods, which we defined as a ‘Western’ dietary pattern. We calculated an adherence score to each dietary pattern and for each participant, using the food categories factor loadings to weigh the sum of all food categories observed consumption. Thus, the adherence score measures a participant’s diet conformity to the identified dietary pattern.

Multivariable Cox proportional hazard models with age as the primary time variable were used to characterise the associations between each type of dairy consumption and incidence of CVD, CHD and cerebrovascular diseases, and to calculate cause-specific hazard ratios (HR) and 95 % CI. In the CHD model, cerebrovascular disease cases were censored at the date of diagnosis but were considered as non-cases for CHD, and reciprocally for the cerebrovascular disease model. The Schoenfeld residuals were used to confirm risk proportionality assumptions(Reference Schoenfeld33). P-values for linear trends were obtained by coding quartiles of dairy consumption as an ordinal variable. Participants contributed person-time to the Cox model until the date of CVD diagnosis, the date of the last completed questionnaire and the date of death or 1 January 2019, whichever occurred first. Models were adjusted for age (time scale), sex, physical activity (low, moderate, high, missing, computed following International Physical Activity Questionnaire recommendations), BMI (kg/m2, continuous), education level (<high-school degree, <2 years after high-school degree, ≥2 years after high-school degree), without alcohol energy intake (kcal/d, continuous), alcohol intake (g/d, continuous), smoking status (never smoked, former smoker and current smoker), number of dietary records (continuous) and family history of CVD (yes/no) (model 1).

In exploratory analyses, we performed an additional model to account for the potential influence of the nutritional quality of the diet. This included adjusting model 1 for a healthy dietary pattern derived by principal component analysis (model 2, online Supplementary Table S1). Finally, further adjustments were added to model 1 to include the influence of baseline prevalence and treatment of self-reported type 2 diabetes, hypercholesterolaemia, hypertension and hypertriglyceridaemia (model 3).

When Cox models revealed significant associations, sensitivity analyses were performed based on model 1 by adding further adjustments for a Western dietary pattern derived from principal component analysis (online Supplementary Table S1). Finally, CVD cases diagnosed in the first 2 years of each participant’s follow-up were excluded to account for reverse causation bias in statistically significant associations. All tests were two-sided, and P-values ≤ 0·05 were considered statistically significant. All analyses were carried out with SAS software (version 9.4; SAS Institute Inc.).

Results

This analysis included 104 805 participants (see Fig. 1) with a mean age of 42·8 years at baseline (sd 14·6), among which 22 291 were men (21·3 %) and 82 517 were women (78·7 %). Participants included in this analysis had completed on average 5·7 (sd 3·1) 24-h dietary records, with 8·1 % of the participants having only completed the minimum two dietary records for inclusion in the analyses. The participants’ baseline characteristics according to sex-specific quartiles of dairy intake are shown in Table 1. Overall, there was no significant difference in baseline characteristics between low consumers (1st quartile) and high consumers of dairy foods (4th quartile). In addition, 65 % of our participants had moderate to high physical activity scores from the International Physical Activity Questionnaire, 65·4 % had ≥2 years of education after high school and 82·8 % did not smoke.

Fig. 1. Flow chart of participants included in the study, NutriNet-Santé cohort, France, 2009–2019.

Table 1. Baseline characteristics of study population according to sex-specific quartiles of dairy consumption, NutriNet-Santé cohort, France, 2009–2019

(Mean values and standard deviations; numbers and percentages, n 104 805)

* Sex-specific quartiles of total dairy consumption. Cut-offs were 112, 190 and 303 g/d for males and 112, 191 and 301 g/d for females.

P-value comparing quartiles of total dairy consumption, using two-sided χ 2 tests or Fisher tests as appropriate.

Physical activity categories according to the International Physical Activity Questionnaire (IPAQ)(19) IPAQ data were missing for 14 600 participants (13·9 %).

§ Amongst first-degree relatives.

|| Standard French serving sizes defined as 150 ml for milk, 30 g for cheese and 125 g for yogurts.

Participants consumed on average 222 g/d of dairy foods (sd 151), including 110 g/d of milk (sd 127), 37·7 g/d of cheese (sd 28·3) and 79·1 g/d of yogurt (sd 84·9), which was similar to the consumption levels observed in the general French population(9). In addition, dairy foods contributed to 18·3 % of total fat intakes (sd 13·7) and 28·9 % of SFA intakes (sd 24·4) (online Supplementary Table S2).

Associations between dairy consumption and CVD risk

Between 2009 and 2019 and a mean follow-up of 5·5 years (sd 3·0, 579 155 person-years), 2098 cases of CVD were diagnosed, among which there were 1220 cases of CHD (eighty-two acute coronary syndromes, 318 angina pectoris, 148 myocardial infarctions and 672 angioplasties) and 878 cases of cerebrovascular diseases (118 strokes and 760 TIA).

Schoenfeld residuals were not significantly associated with time, which supported the proportional hazard assumption (online Supplementary Table S3). The associations between the consumption of dairy foods and the risks of overall CVD, CHD and cerebrovascular diseases are presented in Table 2. The basic multivariable Cox proportional hazard (model 1) did not reveal any statistically significant association between the consumption of any dairy type and overall or CHD. These associations remained statistically non-significant in models 2, 3 and 4 (data not shown). However, high consumption (≥161·6 g/d for males and ≥160·9 g/d for females) of fermented dairy foods (yogurt, cheese and fermented milk) compared with low consumption (<57·3 g/d for males, <54·3 g/d for females) was associated with a 19 % decreased risk of cerebrovascular disease (HR = 0·81, 95 % CI 0·66, 0·98, P trend = 0·01). This association was borderline significant when considering the continuous intake of fermented dairy with an increment of 100 g/d (HR = 0·98, 95 % CI 0·97, 1·00, P-value = 0·05). In addition, the restricted cubic spline presented in Fig. 2 verified the linearity assumption between the consumption of fermented dairy foods and the risk of cerebrovascular disease (P-value for non-linear association = 0·23)(Reference Desquilbet and Mariotti34).

Table 2. Associations between dairy consumption and CVD risk from multivariable Cox proportional hazard models*, NutriNet-Santé cohort, France, 2009–2019

(Hazard ratios (HR) and 95 % confidence intervals, n 104 805)

* Cox models were adjusted for age (time scale), sex, physical activity (low, moderate, high, computed following IPAQ recommendations), BMI (kg/m2, continuous), education level (<high-school degree, <2 years after high-school degree, ≥2 years after high-school degree), without alcohol energy intake (kcal/d, continuous), alcohol intake (g/d, continuous), smoking status (never smoked, former smoker, current smoker), number of dietary records (continuous), family history of CVD (yes/no) (model 1).

Sex-specific cut-offs for milk were 16·1, 51·6 and 153·8 g/d in males and 16·3, 50·8 and 153·7 g/d in females. Cut-offs for cheese were 17·7, 32·9 and 51·4 g/d in males and 17·8, 33·0 and 51·6 g/d in females. Cut-offs for yogurt-like dairy were 0·04, 44·7 and 109·8 g/d in males and 11·9, 60·2 and 125·0 g/d in females. Cut-offs for high-fat dairy were 26·2, 50·5 and 84·6 g/d in males and 26·3, 50·0 and 83·9 g/d in females. Cut-offs for reduced-fat dairy were 46·7, 117·7 and 232·9 g/d for males and 47·1, 117·4 and 232·3 g/d in females. Cut-offs for fermented dairy were 57·3, 102·8 and 161·6 g/d for males and 54·3, 100·6 and 160·9 g/d in females. Cut-offs for non-fermented dairy were 16·8, 54·3 and 168·6 g/d in males and 15·9, 48·8 and 147·5 g/d in females. Cut-offs for total dairy were 112·0, 186·8 and 303·0 g/d in males and 111·7, 190·8 and 301·5 g/d in females.

HR for an absolute increment of 150 g/d of milk, 30 g/d of cheese, and 100 g/d of yogurts, high-fat, reduced-fat, fermented, non-fermented and total dairy.

§ Includes myocardial infarction, angioplasty, acute coronary syndrome and angina pectoris.

|| Includes stroke and transient ischaemic attack.

Fig. 2. Spline plot for the linearity assumption of the association between the consumption of dairy foods and risk of CVD, NutriNet-Santé cohort, France, 2009–2019 (n 104 805)a. Consumption of fermented dairy and risk of cerebrovascular diseases. Consumption of total dairy and risk of total CVD. aRestricted cubic spline SAS macro developed by Desquilbet & Mariotti(Reference Gille, Schmid and Walther35) Estimation;  Upper and lower confidence limit;  Knots.

Exploratory and sensitivity analyses

The association between fermented dairy consumption and cerebrovascular risk remained statistically significant when comparing the highest intake (4th quartile) with the lowest intake (1st quartile), after further adjustments in models 2 and 3 (Table 3). Similarly, this association remained stable in further exploratory analyses adjusting for a Western dietary pattern derived from principal component analysis (HR = 0·81, 95 % CI 0·66, 0·98, P trend = 0·01). Furthermore, the use of multiple imputation with the MICE method to manage missing values strengthened the inverse association between continuous consumption of fermented foods and cerebrovascular disease risk (HR = 0·91, 95 % CI 0·84, 0·99, P-value = 0·02), but did not significantly impact other associations between continuous dairy consumption and overall, CHD or cerebrovascular disease risk. When excluding CVD cases that required more extensive documentation to ascertain diagnosis (i.e. TIA and angina), all associations between dairy consumption and disease risk were non-significant, likely due to a loss of statistical power.

Table 3. Associations between fermented dairy foods and cerebrovascular disease risk from multivariable Cox proportional hazard models, NutriNet-Santé cohort, France, 2009–2019

(Hazard ratios (HR) and 95 % confidence intervals, n 104 805)*

* Cerebrovascular disease included incident events of strokes and transient ischaemic attacks.

Cox models were adjusted for age (time scale), sex, physical activity (low, moderate, high, computed following IPAQ recommendations), BMI (kg/m2, continuous), education level (<high-school degree, <2 years after high-school degree, ≥2 years after high-school degree), without alcohol energy intake (kcal/d, continuous), alcohol intake (g/d, continuous), smoking status (never smoked, former smoker, current smoker), number of dietary records (continuous), family history of CVD (yes/no) (model 1). Model 2 = Model 1 + healthy dietary pattern (derived by principal component analysis). Model 2b = Model 1 + Western dietary pattern derived by principal component analysis. Model 3 = Model 1 + prevalence and treatment of type 2 diabetes, dyslipidaemia, hypertension and hypertriglyceridaemia.

HR for an absolute increment of 100 g/d of fermented dairy foods.

§ Sex-specific cut-offs for fermented dairy were 57·3, 102·8, 161·6 g/d for males and 54·3, 100·6 and 160·9 g/d in females. Fermented dairy foods included yogurt, cheese and fermented milk.

Finally, the exclusion of cerebrovascular disease cases diagnosed during the first 2 years of follow-up (n 144 cases excluded) suggested a 9 % decreased risk of cerebrovascular disease associated with each additional 100 g of fermented dairy food consumed daily (n 734 cases/103,927 non-cases, HR = 0·91, 95 % CI 0·83, 0·99, P trend = 0·03), and a 21 % decreased risk when comparing the highest (4th quartile) with the lowest (1st quartile) consumption of fermented dairy (HR = 0·79, 95 % CI 0·64, 0·98, P trend = 0·001).

Discussion

In this prospective cohort study of French adults, we did not observe a statistically significant association between the consumption of dairy food and the risk of CVD or CHD. However, our results suggest a possible lower risk of cerebrovascular disease (i.e. stroke and TIA) associated with a higher consumption of fermented dairy foods, such as cheese and yogurts.

It is recommended to limit dietary SFA intakes for CVD prevention; however, this study did not reveal any direct association between the consumption of dairy products and total CVD or CHD risk, despite contributing to 28·9 % of dietary SFA (online Supplementary Table S2b). This supports the existing epidemiological evidence on the topic, especially from meta-analyses of prospective studies which consistently reported null or weak inverse associations between the consumption of total dairy and CVD risk(Reference Fontecha, Calvo and Juarez10,Reference Soedamah-Muthu and de Goede11,Reference Gille, Schmid and Walther35) . In a 2018 meta-analysis, Soedamah-Muthu & de Goede reported a non-statistically significant association between total dairy and CHD risk when pooling the results from fifteen prospective cohort studies and reported an 8 % reduced risk of stroke associated with an increment of 200 g of milk consumption per day (risk ratio = 0·92, 95 % CI 0·88, 0·97, I 2 = 85 %)(Reference Soedamah-Muthu and de Goede11). Since then, the PURE prospective cohort study observed 5855 CVD events over 9·1 years of follow-up from both urban and rural populations in twenty-one countries(Reference Dehghan, Mente and Rangarajan12). In this large prospective study, authors reported that a total dairy consumption of > 2 servings per d, compared with no dairy consumption, was associated with a 22 % risk reduction of major CVD (i.e. MI, stroke or heart failure) (HR = 0·78, 95 % CI 0·67, 0·90, P trend = 0·0001) and a 34 % reduced risk of incident stroke (HR = 0·66, 95 % CI 0·53, 0·82, P trend = 0·0003). More recently, a small prospective cohort study from Greece, which included 2020 participants followed-up for 10 years, observed an inverse association between total dairy intake and total CVD risk (HR = 0·48, 95 % CI 0·23, 0·90) in women only. This inverse association in women was stronger when the authors looked at yogurt consumption, with a 14 % CVD risk reduction associated with a 200 g/d increment in yogurt intake (HR = 0·86, 95 % CI 0·49, 0·98)(Reference Kouvari, Panagiotakos and Chrysohoou36).

The inverse association between fermented dairy and cerebrovascular risk observed in this study may suggest a differential effect of these types of dairy foods on cardiovascular health. In a 2019 meta-analysis of randomised controlled trials and prospective studies, Companys et al. observed that the consumption of fermented milk was associated with a reduced risk of stroke and ischaemic heart disease (risk ratio = 0·96, 95 % CI 0·94, 0·98, high heterogeneity I 2 = 95·9 %)(Reference Companys, Pla-Pagà and Calderón-Pérez37). This finding was in line with those reported in an extensive review of meta-analyses conducted by Fontecha et al. in 2019, which observed a reduced risk of stroke and stroke mortality associated with the consumption of fermented dairy, including fermented milk(Reference Hu, Huang and Wang38) and cheese(Reference Hu, Huang and Wang38Reference Qin, Xu and Han42), but not yogurt(Reference Fontecha, Calvo and Juarez10).

Another potential source of variation in the health effects of dairy consumption relates to the nutrient content of specific types of dairy foods. In the large European EPIC-Netherlands study, Laursen et al. observed 884 stroke cases over a 15·2-year follow-up. They reported that the consumption of each additional daily serving of whole-fat yogurt as a substitution for any other dairy group (low-fat yogurt, cheese, butter, buttermilk or milk) was associated with a lower risk of ischaemic stroke (HR between 0·33 and 0·36)(Reference Laursen, Sluijs and Boer43). These findings were in line with previous results observed by the same authors in another European prospective study, the Danish Diet, Cancer and Health cohort(Reference Laursen, Dahm and Johnsen44). However, emerging evidence suggests that the nutrient content of dairy should be considered in relation to the dairy food matrix, which refers to the physical structure of food and may have an impact on nutrient absorption and biomarkers of CVD risk, such as blood pressure and cholesterol metabolism. In particular, one hypothesis suggests that cheese, despite being high in fat, possesses similar features to milk and yogurt, rather than to butter, due to its high Ca, protein and milk fat globule membrane content(Reference Thorning, Bertram and Bonjour45). In addition, the fermentation process involved in the production of cheese and yogurt often results in the presence of bacteria within the food matrix, which may produce SCFA and bioactive peptides(Reference Thorning, Bertram and Bonjour45). All these components of dairy, particularly present in cheese and yogurt, may interfere with the intestinal absorption and digestibility of fat (which is mostly SFA in dairy foods) and therefore attenuate the effect of SFA on cholesterol metabolism and potentially provide a protective effect on cardiovascular health(Reference Soerensen, Thorning and Astrup46Reference Crippa, Zabzuni and Bravi49). Although more well-controlled intervention studies in humans are necessary to further investigate these potential mechanisms, this would be in line with observational evidence from meta-analyses of prospective studies, which suggest that fermented dairy consumption may be associated with lower total and LDL-cholesterol levels(Reference Agerholm-Larsen, Bell and Grunwald50Reference Shimizu, Hashiguchi and Shiga53). Finally, a potential hypotensive effect of bioactive peptides found in fermented dairy foods may reduce cerebrovascular diseases risk, which would be in line with observational epidemiological studies, but still needs further research to be fully elucidated(Reference Melini, Melini and Luziatelli54).

The prospective design of this study contributed to its strengths, allowing the assessment of mid-term associations of dairy consumption with CVD risks. In addition, this study used repeated 24-h dietary records to provide detailed and up-to-date dietary intake assessment, as opposed to FFQ which are often used in nutritional epidemiology. In this study, we identified two Healthy and Western dietary patterns using a principal component analysis (online Supplementary Table S1) and these patterns did not influence the associations observed between fermented dairy consumption and cerebrovascular risk. Finally, the use of the SNIIRAM national register allowed the maximisation of CVD case ascertainment, limiting the omission of cases when participants did not report their disease to the study investigators. However, some limitations of this study also pertain to its observational design. Indeed, residual confounding cannot be ruled out, and a causal link between the consumption of fermented dairy and a decreased risk of cerebrovascular disease cannot be established from this prospective cohort study alone, although the inclusion of many potential confounders in our main analyses suggests a robust inverse association. Furthermore, a relatively limited statistical power precluded the investigation of specific types of CVD. The NutriNet-Santé cohort is volunteered-based, and as highlighted in Table 1, the participants included in this study were generally more health-conscious, younger, more highly educated, more often women, consumed more fruits and vegetables(9) and were less likely to smoke or be overweight(Reference Andreeva, Salanave and Castetbon55), compared with the general French population.

In conclusion, in this large prospective cohort study, we found that the consumption of dairy foods may not be associated with overall CVD or CHD risks in French adults. However, we observed a higher consumption of fermented dairy products (e.g. cheese and yogurt) to be associated with a lower risk of stroke and TIA. Overall, these observational findings provide insight on the potential role of specific dairy foods in cardiometabolic health. However, future randomised controlled trials are warranted to confirm these associations.

Acknowledgements

The authors warmly thank all the volunteers of the NutriNet-Santé cohort. We also thank Younes Esseddik (IT manager); Thi Hong Van Duong, Régis Gatibelza and Jagatjit Mohinder (computer scientists); Fabien Szabo de Edelenyi, PhD (data management supervisor); Julien Allègre, Nathalie Arnault and Laurent Bourhis (data-managers/biostatisticians); Sandrine Kamdem, MD (physician) and Nathalie Druesne-Pecollo (operational coordinator) for their technical contribution to the NutriNet-Santé study.

The NutriNet-Santé study was supported by the Ministère de la Santé, Santé Publique France, Institut National de la Santé et de la Recherche Médicale (INSERM), Institut National de la Recherche Agronomique (INRA), Conservatoire National des Arts et Métiers (CNAM) and Université Paris 13. Laury Sellem was funded by the Biotechnology and Biological Sciences Research Council (BBSRC) Joint Programme Initiatives (JPI) ‘HDHL Biomarkers: Fatty Acid Metabolism - Interlinking Diet with Cardiometabolic Health (FAME)’ (Project Reference: BB/P028217/1). Researchers were independent from funders. Funders had no role in the study design, the collection, analysis and interpretation of data, the writing of the report, and the decision to submit the article for publication.

L. S., M. T. and J. L. designed the research. S. H., P. G., M. T., C. J., L. K. F. and E. K.-G. conducted the research. L. S. performed the statistical analyses. B. S. supervised the statistical analyses. L. S. drafted the manuscript. M. S., J. L. and K. G. J. supervised the writing. B. S., K. G. J., S. H., P. G., E. K.-G., C. J., L. K. F., M. D., J. L. and M. T. contributed to the data interpretation and revised each draft for important intellectual content. All authors read and approved the final manuscript. M. T. and J. L. had primary responsibility for the final content and are the guarantors. The corresponding author (B. S.) attests that all listed authors meet authorship criteria and that no other authors meeting criteria have been omitted.

Julie Lovegrove is Deputy Chair of the UK Scientific Advisory Committee for Nutrition (SACN) and was an expert on SACN’s Saturated Fats working group. All others have no conflict of interest to declare.

Supplementary material

For supplementary material referred to in this article, please visit https://doi.org/10.1017/S0007114521001422

Footnotes

These authors contributed equally to this work.

The original version of this article was published with an incorrect author name. The error has now been rectified and the online HTML and PDF versions updated.

References

World Health Organization (2020) Cardiovascular diseases. https://www.who.int/westernpacific/health-topics/cardiovascular-diseases (accessed August 2020).Google Scholar
Vos, T, Lim, SS, Abbafati, C, et al. (2020) Global burden of 369 diseases, injuries in 204 countries, territories, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet 396, 12041222.10.1016/S0140-6736(20)30925-9CrossRefGoogle Scholar
Santé Publique France (2020) Cardiovascular diseases - Public Health France (Santé Publique France). /maladies-et-traumatismes/maladies-cardiovasculaires-et-accident-vasculaire-cerebral (accessed August 2020).Google Scholar
GBD 2017 Diet Collaborators (2019) Health effects of dietary risks in 195 countries, 1990–2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet 393, 19581972.CrossRefGoogle Scholar
Borén, J, Chapman, MJ, Krauss, RM, et al. (2020) Low-density lipoproteins cause atherosclerotic cardiovascular disease: pathophysiological, genetic, therapeutic insights: a consensus statement from the European Atherosclerosis Society Consensus Panel. Eur Heart J 41, 23132330.CrossRefGoogle ScholarPubMed
Kaptoge, S, Pennells, L, De Bacquer, D, et al. (2019) World Health Organization cardiovascular disease risk charts: revised models to estimate risk in 21 global regions. Lancet Glob Health 7, e1332e1345.CrossRefGoogle Scholar
ANSES (2020) Fats. https://www.anses.fr/en/content/fats (accessed June 2020).Google Scholar
Chaltiel, D, Adjibade, M, Deschamps, V, et al. (2019) Programme National Nutrition Santé - guidelines score 2 (PNNS-GS2): development and validation of a diet quality score reflecting the 2017 French dietary guidelines. Br J Nutr 122, 331342.CrossRefGoogle ScholarPubMed
ANSES (2017) Third Individual and National Survey on Food Consumption (INCA3). https://www.anses.fr/fr/system/files/NUT2014SA0234Ra.pdf (accessed August 2020).Google Scholar
Fontecha, J, Calvo, MV, Juarez, M, et al. (2019) Milk and dairy product consumption and cardiovascular diseases: an overview of systematic reviews and meta-analyses. Adv Nutr 10, S164S189.CrossRefGoogle ScholarPubMed
Soedamah-Muthu, SS & de Goede, J (2018) Dairy consumption and cardiometabolic diseases: systematic review and updated meta-analyses of prospective cohort studies. Curr Nutr Rep 7, 171182.CrossRefGoogle ScholarPubMed
Dehghan, M, Mente, A, Rangarajan, S, et al. (2018) Association of dairy intake with cardiovascular disease and mortality in 21 countries from five continents (PURE): a prospective cohort study. Lancet 392, 22882297.CrossRefGoogle ScholarPubMed
Guo, J, Astrup, A, Lovegrove, JA, et al. (2017) Milk and dairy consumption and risk of cardiovascular diseases and all-cause mortality: dose-response meta-analysis of prospective cohort studies. Eur J Epidemiol 32, 269287.CrossRefGoogle ScholarPubMed
Hercberg, S, Castetbon, K, Czernichow, S, et al. (2010) The Nutrinet-Santé Study: a web-based prospective study on the relationship between nutrition and health and determinants of dietary patterns and nutritional status. BMC Public Health 10, 242.CrossRefGoogle ScholarPubMed
Vergnaud, A-C, Touvier, M, Méjean, C, et al. (2011) Agreement between web-based and paper versions of a socio-demographic questionnaire in the NutriNet-Santé study. Int J Public Health 56, 407417.CrossRefGoogle ScholarPubMed
Lassale, C, Péneau, S, Touvier, M, et al. (2013) Validity of web-based self-reported weight and height: results of the Nutrinet-Santé study. J Med Internet Res 15, e152.CrossRefGoogle ScholarPubMed
Touvier, M, Méjean, C, Kesse-Guyot, E, et al. (2010) Comparison between web-based and paper versions of a self-administered anthropometric questionnaire. Eur J Epidemiol 25, 287296.10.1007/s10654-010-9433-9CrossRefGoogle ScholarPubMed
Lassale, C, Castetbon, K, Laporte, F, et al. (2015) Validation of a Web-based, self-administered, non-consecutive-day dietary record tool against urinary biomarkers. Br J Nutr 113, 953962.CrossRefGoogle ScholarPubMed
IPAQ Group (2005) Guidelines for Data Processing and Analysis of the International Physical Activity Questionnaire (IPAQ). https://sites.google.com/site/theipaq/scoring-protocol (accessed February 2021).Google Scholar
Craig, CL, Marshall, AL, Sjöström, M, et al. (2003) International physical activity questionnaire: 12-country reliability and validity. Med Sci Sports Exerc 35, 13811395.10.1249/01.MSS.0000078924.61453.FBCrossRefGoogle ScholarPubMed
Touvier, M, Kesse-Guyot, E, Méjean, C, et al. (2011) Comparison between an interactive web-based self-administered 24 h dietary record and an interview by a dietitian for large-scale epidemiological studies. Br J Nutr 105, 10551064.CrossRefGoogle Scholar
Le Moullec, N, Deheeger, M, Preziosi, P, et al. (1996) Validation du manuel photo utilisé pour l’enquête alimentaire de l’étude SU.VI.MAX. [Validation of the food portion size booklet used in the SU.VI.MAX study]. Cahiers de Nutrition et de Diététique 31, 158164.Google Scholar
Arnault, N, Caillot, L, Castetbon, K, et al. (2013) Table de composition des aliments, étude NutriNet-Santé (Food composition table, NutriNet-Santé study) [in French]. 2nd ed. [Economica, editor]. France: Economica.Google Scholar
Black, AE (2000) Critical evaluation of energy intake using the Goldberg cut-off for energy intake:basal metabolic rate. A practical guide to its calculation, use and limitations. Int J Obes Relat Metab Disord 24, 11191130.CrossRefGoogle ScholarPubMed
Kesse-Guyot, E, Bertrais, S, Péneau, S, et al. (2009) Dietary patterns and their sociodemographic and behavioural correlates in French middle-aged adults from the SU.VI.MAX cohort. Eur J Clin Nutr 63, 521528.CrossRefGoogle ScholarPubMed
Gazan, R, Béchaux, C, Crépet, A, et al. (2016) Dietary patterns in the French adult population: a study from the second French national cross-sectional dietary survey (INCA2) (2006–2007). Br J Nutr 116, 300315.CrossRefGoogle Scholar
van Buuren, S (2007) Multiple imputation of discrete and continuous data by fully conditional specification. Stat Methods Med Res 16, 219242.CrossRefGoogle ScholarPubMed
Sterne, JAC, White, IR, Carlin, JB, et al. (2009) Multiple imputation for missing data in epidemiological, clinical research: potential, pitfalls. BMJ 338 CrossRefGoogle Scholar
SAS/STAT 14.3 User’s guide (2020) SAS Help Center: PROC MI Statement. https://documentation.sas.com/?docsetId=statug&docsetTarget=statug_mi_syntax01.htm&docsetVersion=14.3&locale=en (accessed April 2020).Google Scholar
Rubin, DB (2004) Multiple Imputation for Nonresponse in Surveys. In: Multiple Imputation for Nonresponse in Surveys. John Wiley & Sons, Ltd. https://onlinelibrary.wiley.com/doi/abs/10.1002/9780470316696.fmatter (accessed April 2020).Google Scholar
Rubin, DB (1976) Inference and missing data. Biometrika 63, 581592.CrossRefGoogle Scholar
Santos, R de O, Gorgulho, BM, Castro, MA de, et al. (2019) Principal Component Analysis, Factor Analysis: differences, similarities in Nutritional Epidemiology application. Rev Bras Epidemiol 22, e190041.CrossRefGoogle ScholarPubMed
Schoenfeld, D (1982) Partial residuals for the proportional hazards regression model. Biometrika 69, 239241.CrossRefGoogle Scholar
Desquilbet, L & Mariotti, F (2010) Dose-response analyses using restricted cubic spline functions in public health research. Stat Med 29, 10371057.Google ScholarPubMed
Gille, D, Schmid, A, Walther, B, et al. (2018) Fermented food and non-communicable chronic diseases: a review. Nutrients 10, 448.CrossRefGoogle ScholarPubMed
Kouvari, M, Panagiotakos, DB, Chrysohoou, C, et al. (2020) Dairy products, surrogate markers,, cardiovascular disease; a sex-specific analysis from the ATTICA prospective study. Nutr Metab Cardiovasc Dis 30, 21942206.CrossRefGoogle ScholarPubMed
Companys, J, Pla-Pagà, L, Calderón-Pérez, L, et al. (2020) Fermented dairy products, probiotic supplementation, cardiometabolic diseases: a systematic review, meta-analysis. Adv Nutr 11, 834863.CrossRefGoogle Scholar
Hu, D, Huang, J, Wang, Y, et al. (2014) Dairy foods and risk of stroke: a meta-analysis of prospective cohort studies. Nutr Metab Cardiovasc Dis 24, 460469.CrossRefGoogle ScholarPubMed
Alexander, DD, Bylsma, LC, Vargas, AJ, et al. (2016) Dairy consumption and CVD: a systematic review and meta-analysis. Br J Nutr 115, 737750.CrossRefGoogle ScholarPubMed
Chen, G-C, Wang, Y, Tong, X, et al. (2017) Cheese consumption and risk of cardiovascular disease: a meta-analysis of prospective studies. Eur J Nutr 56, 25652575.CrossRefGoogle ScholarPubMed
Gholami, F, Khoramdad, M, Esmailnasab, N, et al. (2017) The effect of dairy consumption on the prevention of cardiovascular diseases: A meta-analysis of prospective studies. J Cardiovasc Thorac Res 9, 111.CrossRefGoogle ScholarPubMed
Qin, L-Q, Xu, J-Y, Han, S-F, et al. (2015) Dairy consumption and risk of cardiovascular disease: an updated meta-analysis of prospective cohort studies. Asia Pac J Clin Nutr 24, 90100.Google ScholarPubMed
Laursen, ASD, Sluijs, I, Boer, JMA, et al. (2019) Substitutions between dairy products and risk of stroke: results from the European Investigation into Cancer and Nutrition-Netherlands (EPIC-NL) cohort. Br J Nutr 121, 13981404.CrossRefGoogle ScholarPubMed
Laursen, ASD, Dahm, CC, Johnsen, SP, et al. (2018) Substitutions of dairy product intake and risk of stroke: a Danish cohort study. Eur J Epidemiol 33, 201212.CrossRefGoogle ScholarPubMed
Thorning, TK, Bertram, HC, Bonjour, J-P, et al. (2017) Whole dairy matrix or single nutrients in assessment of health effects: current evidence and knowledge gaps. Am J Clin Nutr 105, 10331045.CrossRefGoogle ScholarPubMed
Soerensen, KV, Thorning, TK, Astrup, A, et al. (2014) Effect of dairy calcium from cheese and milk on fecal fat excretion, blood lipids, and appetite in young men. Am J Clin Nutr 99, 984991.CrossRefGoogle ScholarPubMed
Raziani, F, Tholstrup, T, Kristensen, MD, et al. (2016) High intake of regular-fat cheese compared with reduced-fat cheese does not affect LDL cholesterol or risk markers of the metabolic syndrome: a randomized controlled trial. Am J Clin Nutr 104, 973981.CrossRefGoogle ScholarPubMed
Wolever, TM, Fernandes, J & Rao, AV (1996) Serum acetate:propionate ratio is related to serum cholesterol in men but not women. J Nutr 126, 27902797.Google Scholar
Crippa, G, Zabzuni, D, Bravi, E, et al. (2018) Randomized, double blind placebo-controlled pilot study of the antihypertensive effects of Grana Padano D.O.P. cheese consumption in mild - moderate hypertensive subjects. Eur Rev Med Pharmacol Sci 22, 75737581.Google ScholarPubMed
Agerholm-Larsen, L, Bell, ML, Grunwald, GK, et al. (2000) The effect of a probiotic milk product on plasma cholesterol: a meta-analysis of short-term intervention studies. Eur J Clin Nutr 54, 856860.CrossRefGoogle ScholarPubMed
Sun, J & Buys, N (2015) Effects of probiotics consumption on lowering lipids and CVD risk factors: a systematic review and meta-analysis of randomized controlled trials. Ann Med 47, 430440.CrossRefGoogle ScholarPubMed
de Goede, J, Geleijnse, JM, Ding, EL, et al. (2015) Effect of cheese consumption on blood lipids: a systematic review and meta-analysis of randomized controlled trials. Nutr Rev 73, 259275.CrossRefGoogle ScholarPubMed
Shimizu, M, Hashiguchi, M, Shiga, T, et al. (2015) Meta-Analysis: effects of Probiotic supplementation on lipid profiles in normal to mildly hypercholesterolemic individuals. PLoS One 10, e0139795.CrossRefGoogle ScholarPubMed
Melini, F, Melini, V, Luziatelli, F, et al. (2020) Health-Promoting components in fermented foods:an up-to-date systematic review. Nutrients 11, 1189.CrossRefGoogle Scholar
Andreeva, VA, Salanave, B, Castetbon, K, et al. (2015) Comparison of the sociodemographic characteristics of the large NutriNet-Santé e-cohort with French Census data: the issue of volunteer bias revisited. J Epidemiol Community Health 69, 893898.CrossRefGoogle ScholarPubMed
Figure 0

Fig. 1. Flow chart of participants included in the study, NutriNet-Santé cohort, France, 2009–2019.

Figure 1

Table 1. Baseline characteristics of study population according to sex-specific quartiles of dairy consumption, NutriNet-Santé cohort, France, 2009–2019(Mean values and standard deviations; numbers and percentages, n 104 805)

Figure 2

Table 2. Associations between dairy consumption and CVD risk from multivariable Cox proportional hazard models*, NutriNet-Santé cohort, France, 2009–2019(Hazard ratios (HR) and 95 % confidence intervals, n 104 805)

Figure 3

Fig. 2. Spline plot for the linearity assumption of the association between the consumption of dairy foods and risk of CVD, NutriNet-Santé cohort, France, 2009–2019 (n 104 805)a. Consumption of fermented dairy and risk of cerebrovascular diseases. Consumption of total dairy and risk of total CVD. aRestricted cubic spline SAS macro developed by Desquilbet & Mariotti(35).  Estimation;  Upper and lower confidence limit;  Knots.

Figure 4

Table 3. Associations between fermented dairy foods and cerebrovascular disease risk from multivariable Cox proportional hazard models, NutriNet-Santé cohort, France, 2009–2019(Hazard ratios (HR) and 95 % confidence intervals, n 104 805)*

Supplementary material: File

Sellem et al. supplementary material

Sellem et al. supplementary material

Download Sellem et al. supplementary material(File)
File 42.7 KB